Gitiles
Code Review Sign In
gerrit-public.fairphone.software / fp2-dev / platform / external / llvm / 2f50083a5a30363d5e6e7ae34f7b0615d419860d / . / docs / LangRef.html
blob: db59e0d0738149cb251ac5fab67964385e75a55b [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>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]44 </ol>
45 </li>
46 <li><a href="#t_derived">Derived Types</a>
47 <ol>
Chris Lattner251ab812007-12-18 06:18:21 +0000[diff] [blame]48 <li><a href="#t_integer">Integer Type</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]49 <li><a href="#t_array">Array Type</a></li>
50 <li><a href="#t_function">Function Type</a></li>
51 <li><a href="#t_pointer">Pointer Type</a></li>
52 <li><a href="#t_struct">Structure Type</a></li>
53 <li><a href="#t_pstruct">Packed Structure Type</a></li>
54 <li><a href="#t_vector">Vector Type</a></li>
55 <li><a href="#t_opaque">Opaque Type</a></li>
56 </ol>
57 </li>
Chris Lattner515195a2009-02-02 07:32:36 +0000[diff] [blame]58 <li><a href="#t_uprefs">Type Up-references</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]59 </ol>
60 </li>
61 <li><a href="#constants">Constants</a>
62 <ol>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]63 <li><a href="#simpleconstants">Simple Constants</a></li>
Chris Lattner97063852009-02-28 18:32:25 +0000[diff] [blame]64 <li><a href="#complexconstants">Complex Constants</a></li>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]65 <li><a href="#globalconstants">Global Variable and Function Addresses</a></li>
66 <li><a href="#undefvalues">Undefined Values</a></li>
67 <li><a href="#constantexprs">Constant Expressions</a></li>
Nick Lewycky4dcf8102009-04-04 07:22:01 +0000[diff] [blame]68 <li><a href="#metadata">Embedded Metadata</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]69 </ol>
70 </li>
71 <li><a href="#othervalues">Other Values</a>
72 <ol>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]73 <li><a href="#inlineasm">Inline Assembler Expressions</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]74 </ol>
75 </li>
76 <li><a href="#instref">Instruction Reference</a>
77 <ol>
78 <li><a href="#terminators">Terminator Instructions</a>
79 <ol>
80 <li><a href="#i_ret">'<tt>ret</tt>' Instruction</a></li>
81 <li><a href="#i_br">'<tt>br</tt>' Instruction</a></li>
82 <li><a href="#i_switch">'<tt>switch</tt>' Instruction</a></li>
83 <li><a href="#i_invoke">'<tt>invoke</tt>' Instruction</a></li>
84 <li><a href="#i_unwind">'<tt>unwind</tt>' Instruction</a></li>
85 <li><a href="#i_unreachable">'<tt>unreachable</tt>' Instruction</a></li>
86 </ol>
87 </li>
88 <li><a href="#binaryops">Binary Operations</a>
89 <ol>
90 <li><a href="#i_add">'<tt>add</tt>' Instruction</a></li>
91 <li><a href="#i_sub">'<tt>sub</tt>' Instruction</a></li>
92 <li><a href="#i_mul">'<tt>mul</tt>' Instruction</a></li>
93 <li><a href="#i_udiv">'<tt>udiv</tt>' Instruction</a></li>
94 <li><a href="#i_sdiv">'<tt>sdiv</tt>' Instruction</a></li>
95 <li><a href="#i_fdiv">'<tt>fdiv</tt>' Instruction</a></li>
96 <li><a href="#i_urem">'<tt>urem</tt>' Instruction</a></li>
97 <li><a href="#i_srem">'<tt>srem</tt>' Instruction</a></li>
98 <li><a href="#i_frem">'<tt>frem</tt>' Instruction</a></li>
99 </ol>
100 </li>
101 <li><a href="#bitwiseops">Bitwise Binary Operations</a>
102 <ol>
103 <li><a href="#i_shl">'<tt>shl</tt>' Instruction</a></li>
104 <li><a href="#i_lshr">'<tt>lshr</tt>' Instruction</a></li>
105 <li><a href="#i_ashr">'<tt>ashr</tt>' Instruction</a></li>
106 <li><a href="#i_and">'<tt>and</tt>' Instruction</a></li>
107 <li><a href="#i_or">'<tt>or</tt>' Instruction</a></li>
108 <li><a href="#i_xor">'<tt>xor</tt>' Instruction</a></li>
109 </ol>
110 </li>
111 <li><a href="#vectorops">Vector Operations</a>
112 <ol>
113 <li><a href="#i_extractelement">'<tt>extractelement</tt>' Instruction</a></li>
114 <li><a href="#i_insertelement">'<tt>insertelement</tt>' Instruction</a></li>
115 <li><a href="#i_shufflevector">'<tt>shufflevector</tt>' Instruction</a></li>
116 </ol>
117 </li>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]118 <li><a href="#aggregateops">Aggregate Operations</a>
119 <ol>
120 <li><a href="#i_extractvalue">'<tt>extractvalue</tt>' Instruction</a></li>
121 <li><a href="#i_insertvalue">'<tt>insertvalue</tt>' Instruction</a></li>
122 </ol>
123 </li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]124 <li><a href="#memoryops">Memory Access and Addressing Operations</a>
125 <ol>
126 <li><a href="#i_malloc">'<tt>malloc</tt>' Instruction</a></li>
127 <li><a href="#i_free">'<tt>free</tt>' Instruction</a></li>
128 <li><a href="#i_alloca">'<tt>alloca</tt>' Instruction</a></li>
129 <li><a href="#i_load">'<tt>load</tt>' Instruction</a></li>
130 <li><a href="#i_store">'<tt>store</tt>' Instruction</a></li>
131 <li><a href="#i_getelementptr">'<tt>getelementptr</tt>' Instruction</a></li>
132 </ol>
133 </li>
134 <li><a href="#convertops">Conversion Operations</a>
135 <ol>
136 <li><a href="#i_trunc">'<tt>trunc .. to</tt>' Instruction</a></li>
137 <li><a href="#i_zext">'<tt>zext .. to</tt>' Instruction</a></li>
138 <li><a href="#i_sext">'<tt>sext .. to</tt>' Instruction</a></li>
139 <li><a href="#i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a></li>
140 <li><a href="#i_fpext">'<tt>fpext .. to</tt>' Instruction</a></li>
141 <li><a href="#i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a></li>
142 <li><a href="#i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a></li>
143 <li><a href="#i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a></li>
144 <li><a href="#i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a></li>
145 <li><a href="#i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a></li>
146 <li><a href="#i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a></li>
147 <li><a href="#i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a></li>
148 </ol>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]149 </li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]150 <li><a href="#otherops">Other Operations</a>
151 <ol>
152 <li><a href="#i_icmp">'<tt>icmp</tt>' Instruction</a></li>
153 <li><a href="#i_fcmp">'<tt>fcmp</tt>' Instruction</a></li>
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]154 <li><a href="#i_vicmp">'<tt>vicmp</tt>' Instruction</a></li>
155 <li><a href="#i_vfcmp">'<tt>vfcmp</tt>' Instruction</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]156 <li><a href="#i_phi">'<tt>phi</tt>' Instruction</a></li>
157 <li><a href="#i_select">'<tt>select</tt>' Instruction</a></li>
158 <li><a href="#i_call">'<tt>call</tt>' Instruction</a></li>
159 <li><a href="#i_va_arg">'<tt>va_arg</tt>' Instruction</a></li>
160 </ol>
161 </li>
162 </ol>
163 </li>
164 <li><a href="#intrinsics">Intrinsic Functions</a>
165 <ol>
166 <li><a href="#int_varargs">Variable Argument Handling Intrinsics</a>
167 <ol>
168 <li><a href="#int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a></li>
169 <li><a href="#int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a></li>
170 <li><a href="#int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a></li>
171 </ol>
172 </li>
173 <li><a href="#int_gc">Accurate Garbage Collection Intrinsics</a>
174 <ol>
175 <li><a href="#int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a></li>
176 <li><a href="#int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a></li>
177 <li><a href="#int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a></li>
178 </ol>
179 </li>
180 <li><a href="#int_codegen">Code Generator Intrinsics</a>
181 <ol>
182 <li><a href="#int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a></li>
183 <li><a href="#int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a></li>
184 <li><a href="#int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a></li>
185 <li><a href="#int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a></li>
186 <li><a href="#int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a></li>
187 <li><a href="#int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a></li>
188 <li><a href="#int_readcyclecounter"><tt>llvm.readcyclecounter</tt>' Intrinsic</a></li>
189 </ol>
190 </li>
191 <li><a href="#int_libc">Standard C Library Intrinsics</a>
192 <ol>
193 <li><a href="#int_memcpy">'<tt>llvm.memcpy.*</tt>' Intrinsic</a></li>
194 <li><a href="#int_memmove">'<tt>llvm.memmove.*</tt>' Intrinsic</a></li>
195 <li><a href="#int_memset">'<tt>llvm.memset.*</tt>' Intrinsic</a></li>
196 <li><a href="#int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a></li>
197 <li><a href="#int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a></li>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]198 <li><a href="#int_sin">'<tt>llvm.sin.*</tt>' Intrinsic</a></li>
199 <li><a href="#int_cos">'<tt>llvm.cos.*</tt>' Intrinsic</a></li>
200 <li><a href="#int_pow">'<tt>llvm.pow.*</tt>' Intrinsic</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]201 </ol>
202 </li>
203 <li><a href="#int_manip">Bit Manipulation Intrinsics</a>
204 <ol>
205 <li><a href="#int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a></li>
206 <li><a href="#int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic </a></li>
207 <li><a href="#int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic </a></li>
208 <li><a href="#int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic </a></li>
209 <li><a href="#int_part_select">'<tt>llvm.part.select.*</tt>' Intrinsic </a></li>
210 <li><a href="#int_part_set">'<tt>llvm.part.set.*</tt>' Intrinsic </a></li>
211 </ol>
212 </li>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]213 <li><a href="#int_overflow">Arithmetic with Overflow Intrinsics</a>
214 <ol>
Bill Wendling3e1258b2009-02-08 04:04:40 +0000[diff] [blame]215 <li><a href="#int_sadd_overflow">'<tt>llvm.sadd.with.overflow.*</tt> Intrinsics</a></li>
216 <li><a href="#int_uadd_overflow">'<tt>llvm.uadd.with.overflow.*</tt> Intrinsics</a></li>
217 <li><a href="#int_ssub_overflow">'<tt>llvm.ssub.with.overflow.*</tt> Intrinsics</a></li>
218 <li><a href="#int_usub_overflow">'<tt>llvm.usub.with.overflow.*</tt> Intrinsics</a></li>
219 <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]220 <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]221 </ol>
222 </li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]223 <li><a href="#int_debugger">Debugger intrinsics</a></li>
224 <li><a href="#int_eh">Exception Handling intrinsics</a></li>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]225 <li><a href="#int_trampoline">Trampoline Intrinsic</a>
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]226 <ol>
227 <li><a href="#int_it">'<tt>llvm.init.trampoline</tt>' Intrinsic</a></li>
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]228 </ol>
229 </li>
Bill Wendling9127adb2008-11-18 22:10:53 +0000[diff] [blame]230 <li><a href="#int_atomics">Atomic intrinsics</a>
231 <ol>
232 <li><a href="#int_memory_barrier"><tt>llvm.memory_barrier</tt></a></li>
233 <li><a href="#int_atomic_cmp_swap"><tt>llvm.atomic.cmp.swap</tt></a></li>
234 <li><a href="#int_atomic_swap"><tt>llvm.atomic.swap</tt></a></li>
235 <li><a href="#int_atomic_load_add"><tt>llvm.atomic.load.add</tt></a></li>
236 <li><a href="#int_atomic_load_sub"><tt>llvm.atomic.load.sub</tt></a></li>
237 <li><a href="#int_atomic_load_and"><tt>llvm.atomic.load.and</tt></a></li>
238 <li><a href="#int_atomic_load_nand"><tt>llvm.atomic.load.nand</tt></a></li>
239 <li><a href="#int_atomic_load_or"><tt>llvm.atomic.load.or</tt></a></li>
240 <li><a href="#int_atomic_load_xor"><tt>llvm.atomic.load.xor</tt></a></li>
241 <li><a href="#int_atomic_load_max"><tt>llvm.atomic.load.max</tt></a></li>
242 <li><a href="#int_atomic_load_min"><tt>llvm.atomic.load.min</tt></a></li>
243 <li><a href="#int_atomic_load_umax"><tt>llvm.atomic.load.umax</tt></a></li>
244 <li><a href="#int_atomic_load_umin"><tt>llvm.atomic.load.umin</tt></a></li>
245 </ol>
246 </li>
Reid Spencerb043f672007-07-20 19:59:11 +0000[diff] [blame]247 <li><a href="#int_general">General intrinsics</a>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]248 <ol>
Reid Spencerb043f672007-07-20 19:59:11 +0000[diff] [blame]249 <li><a href="#int_var_annotation">
Bill Wendlinge4164592008-11-19 05:56:17 +0000[diff] [blame]250 '<tt>llvm.var.annotation</tt>' Intrinsic</a></li>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]251 <li><a href="#int_annotation">
Bill Wendlinge4164592008-11-19 05:56:17 +0000[diff] [blame]252 '<tt>llvm.annotation.*</tt>' Intrinsic</a></li>
Anton Korobeynikove6e764f2008-01-15 22:31:34 +0000[diff] [blame]253 <li><a href="#int_trap">
Bill Wendlinge4164592008-11-19 05:56:17 +0000[diff] [blame]254 '<tt>llvm.trap</tt>' Intrinsic</a></li>
255 <li><a href="#int_stackprotector">
256 '<tt>llvm.stackprotector</tt>' Intrinsic</a></li>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]257 </ol>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]258 </li>
259 </ol>
260 </li>
261</ol>
262
263<div class="doc_author">
264 <p>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a>
265 and <a href="mailto:vadve@cs.uiuc.edu">Vikram Adve</a></p>
266</div>
267
268<!-- *********************************************************************** -->
269<div class="doc_section"> <a name="abstract">Abstract </a></div>
270<!-- *********************************************************************** -->
271
272<div class="doc_text">
273<p>This document is a reference manual for the LLVM assembly language.
Bill Wendlinge7846a52008-08-05 22:29:16 +0000[diff] [blame]274LLVM is a Static Single Assignment (SSA) based representation that provides
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]275type safety, low-level operations, flexibility, and the capability of
276representing 'all' high-level languages cleanly. It is the common code
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]277representation used throughout all phases of the LLVM compilation
278strategy.</p>
279</div>
280
281<!-- *********************************************************************** -->
282<div class="doc_section"> <a name="introduction">Introduction</a> </div>
283<!-- *********************************************************************** -->
284
285<div class="doc_text">
286
287<p>The LLVM code representation is designed to be used in three
288different forms: as an in-memory compiler IR, as an on-disk bitcode
289representation (suitable for fast loading by a Just-In-Time compiler),
290and as a human readable assembly language representation. This allows
291LLVM to provide a powerful intermediate representation for efficient
292compiler transformations and analysis, while providing a natural means
293to debug and visualize the transformations. The three different forms
294of LLVM are all equivalent. This document describes the human readable
295representation and notation.</p>
296
297<p>The LLVM representation aims to be light-weight and low-level
298while being expressive, typed, and extensible at the same time. It
299aims to be a "universal IR" of sorts, by being at a low enough level
300that high-level ideas may be cleanly mapped to it (similar to how
301microprocessors are "universal IR's", allowing many source languages to
302be mapped to them). By providing type information, LLVM can be used as
303the target of optimizations: for example, through pointer analysis, it
304can be proven that a C automatic variable is never accessed outside of
305the current function... allowing it to be promoted to a simple SSA
306value instead of a memory location.</p>
307
308</div>
309
310<!-- _______________________________________________________________________ -->
311<div class="doc_subsubsection"> <a name="wellformed">Well-Formedness</a> </div>
312
313<div class="doc_text">
314
315<p>It is important to note that this document describes 'well formed'
316LLVM assembly language. There is a difference between what the parser
317accepts and what is considered 'well formed'. For example, the
318following instruction is syntactically okay, but not well formed:</p>
319
320<div class="doc_code">
321<pre>
322%x = <a href="#i_add">add</a> i32 1, %x
323</pre>
324</div>
325
326<p>...because the definition of <tt>%x</tt> does not dominate all of
327its uses. The LLVM infrastructure provides a verification pass that may
328be used to verify that an LLVM module is well formed. This pass is
329automatically run by the parser after parsing input assembly and by
330the optimizer before it outputs bitcode. The violations pointed out
331by the verifier pass indicate bugs in transformation passes or input to
332the parser.</p>
333</div>
334
Chris Lattnera83fdc02007-10-03 17:34:29 +0000[diff] [blame]335<!-- Describe the typesetting conventions here. -->
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]336
337<!-- *********************************************************************** -->
338<div class="doc_section"> <a name="identifiers">Identifiers</a> </div>
339<!-- *********************************************************************** -->
340
341<div class="doc_text">
342
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]343 <p>LLVM identifiers come in two basic types: global and local. Global
344 identifiers (functions, global variables) begin with the @ character. Local
345 identifiers (register names, types) begin with the % character. Additionally,
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]346 there are three different formats for identifiers, for different purposes:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]347
348<ol>
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]349 <li>Named values are represented as a string of characters with their prefix.
350 For example, %foo, @DivisionByZero, %a.really.long.identifier. The actual
351 regular expression used is '<tt>[%@][a-zA-Z$._][a-zA-Z$._0-9]*</tt>'.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]352 Identifiers which require other characters in their names can be surrounded
Daniel Dunbar73f9a772008-10-14 23:51:43 +0000[diff] [blame]353 with quotes. Special characters may be escaped using "\xx" where xx is the
354 ASCII code for the character in hexadecimal. In this way, any character can
355 be used in a name value, even quotes themselves.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]356
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]357 <li>Unnamed values are represented as an unsigned numeric value with their
358 prefix. For example, %12, @2, %44.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]359
360 <li>Constants, which are described in a <a href="#constants">section about
361 constants</a>, below.</li>
362</ol>
363
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]364<p>LLVM requires that values start with a prefix for two reasons: Compilers
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]365don't need to worry about name clashes with reserved words, and the set of
366reserved words may be expanded in the future without penalty. Additionally,
367unnamed identifiers allow a compiler to quickly come up with a temporary
368variable without having to avoid symbol table conflicts.</p>
369
370<p>Reserved words in LLVM are very similar to reserved words in other
371languages. There are keywords for different opcodes
372('<tt><a href="#i_add">add</a></tt>',
373 '<tt><a href="#i_bitcast">bitcast</a></tt>',
374 '<tt><a href="#i_ret">ret</a></tt>', etc...), for primitive type names ('<tt><a
375href="#t_void">void</a></tt>', '<tt><a href="#t_primitive">i32</a></tt>', etc...),
376and others. These reserved words cannot conflict with variable names, because
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]377none of them start with a prefix character ('%' or '@').</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]378
379<p>Here is an example of LLVM code to multiply the integer variable
380'<tt>%X</tt>' by 8:</p>
381
382<p>The easy way:</p>
383
384<div class="doc_code">
385<pre>
386%result = <a href="#i_mul">mul</a> i32 %X, 8
387</pre>
388</div>
389
390<p>After strength reduction:</p>
391
392<div class="doc_code">
393<pre>
394%result = <a href="#i_shl">shl</a> i32 %X, i8 3
395</pre>
396</div>
397
398<p>And the hard way:</p>
399
400<div class="doc_code">
401<pre>
402<a href="#i_add">add</a> i32 %X, %X <i>; yields {i32}:%0</i>
403<a href="#i_add">add</a> i32 %0, %0 <i>; yields {i32}:%1</i>
404%result = <a href="#i_add">add</a> i32 %1, %1
405</pre>
406</div>
407
408<p>This last way of multiplying <tt>%X</tt> by 8 illustrates several
409important lexical features of LLVM:</p>
410
411<ol>
412
413 <li>Comments are delimited with a '<tt>;</tt>' and go until the end of
414 line.</li>
415
416 <li>Unnamed temporaries are created when the result of a computation is not
417 assigned to a named value.</li>
418
419 <li>Unnamed temporaries are numbered sequentially</li>
420
421</ol>
422
423<p>...and it also shows a convention that we follow in this document. When
424demonstrating instructions, we will follow an instruction with a comment that
425defines the type and name of value produced. Comments are shown in italic
426text.</p>
427
428</div>
429
430<!-- *********************************************************************** -->
431<div class="doc_section"> <a name="highlevel">High Level Structure</a> </div>
432<!-- *********************************************************************** -->
433
434<!-- ======================================================================= -->
435<div class="doc_subsection"> <a name="modulestructure">Module Structure</a>
436</div>
437
438<div class="doc_text">
439
440<p>LLVM programs are composed of "Module"s, each of which is a
441translation unit of the input programs. Each module consists of
442functions, global variables, and symbol table entries. Modules may be
443combined together with the LLVM linker, which merges function (and
444global variable) definitions, resolves forward declarations, and merges
445symbol table entries. Here is an example of the "hello world" module:</p>
446
447<div class="doc_code">
448<pre><i>; Declare the string constant as a global constant...</i>
449<a href="#identifiers">@.LC0</a> = <a href="#linkage_internal">internal</a> <a
450 href="#globalvars">constant</a> <a href="#t_array">[13 x i8]</a> c"hello world\0A\00" <i>; [13 x i8]*</i>
451
452<i>; External declaration of the puts function</i>
453<a href="#functionstructure">declare</a> i32 @puts(i8 *) <i>; i32(i8 *)* </i>
454
455<i>; Definition of main function</i>
456define i32 @main() { <i>; i32()* </i>
Dan Gohman01852382009-01-04 23:44:43 +0000[diff] [blame]457 <i>; Convert [13 x i8]* to i8 *...</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]458 %cast210 = <a
Dan Gohman01852382009-01-04 23:44:43 +0000[diff] [blame]459 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]460
461 <i>; Call puts function to write out the string to stdout...</i>
462 <a
463 href="#i_call">call</a> i32 @puts(i8 * %cast210) <i>; i32</i>
464 <a
465 href="#i_ret">ret</a> i32 0<br>}<br>
466</pre>
467</div>
468
469<p>This example is made up of a <a href="#globalvars">global variable</a>
470named "<tt>.LC0</tt>", an external declaration of the "<tt>puts</tt>"
471function, and a <a href="#functionstructure">function definition</a>
472for "<tt>main</tt>".</p>
473
474<p>In general, a module is made up of a list of global values,
475where both functions and global variables are global values. Global values are
476represented by a pointer to a memory location (in this case, a pointer to an
477array of char, and a pointer to a function), and have one of the following <a
478href="#linkage">linkage types</a>.</p>
479
480</div>
481
482<!-- ======================================================================= -->
483<div class="doc_subsection">
484 <a name="linkage">Linkage Types</a>
485</div>
486
487<div class="doc_text">
488
489<p>
490All Global Variables and Functions have one of the following types of linkage:
491</p>
492
493<dl>
494
Rafael Espindolaa168fc92009-01-15 20:18:42 +0000[diff] [blame]495 <dt><tt><b><a name="linkage_private">private</a></b></tt>: </dt>
496
497 <dd>Global values with private linkage are only directly accessible by
498 objects in the current module. In particular, linking code into a module with
499 an private global value may cause the private to be renamed as necessary to
500 avoid collisions. Because the symbol is private to the module, all
501 references can be updated. This doesn't show up in any symbol table in the
502 object file.
503 </dd>
504
Dale Johannesen96e7e092008-05-23 23:13:41 +0000[diff] [blame]505 <dt><tt><b><a name="linkage_internal">internal</a></b></tt>: </dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]506
Duncan Sandsa75223a2009-01-16 09:29:46 +0000[diff] [blame]507 <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]508 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]509 '<tt>static</tt>' keyword in C.
510 </dd>
511
Chris Lattner68433442009-04-13 05:44:34 +0000[diff] [blame]512 <dt><tt><b><a name="available_externally">available_externally</a></b></tt>:
513 </dt>
514
515 <dd>Globals with "<tt>available_externally</tt>" linkage are never emitted
516 into the object file corresponding to the LLVM module. They exist to
517 allow inlining and other optimizations to take place given knowledge of the
518 definition of the global, which is known to be somewhere outside the module.
519 Globals with <tt>available_externally</tt> linkage are allowed to be discarded
520 at will, and are otherwise the same as <tt>linkonce_odr</tt>. This linkage
521 type is only allowed on definitions, not declarations.</dd>
522
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]523 <dt><tt><b><a name="linkage_linkonce">linkonce</a></b></tt>: </dt>
524
525 <dd>Globals with "<tt>linkonce</tt>" linkage are merged with other globals of
526 the same name when linkage occurs. This is typically used to implement
527 inline functions, templates, or other code which must be generated in each
528 translation unit that uses it. Unreferenced <tt>linkonce</tt> globals are
529 allowed to be discarded.
530 </dd>
531
Dale Johannesen96e7e092008-05-23 23:13:41 +0000[diff] [blame]532 <dt><tt><b><a name="linkage_common">common</a></b></tt>: </dt>
533
534 <dd>"<tt>common</tt>" linkage is exactly the same as <tt>linkonce</tt>
535 linkage, except that unreferenced <tt>common</tt> globals may not be
536 discarded. This is used for globals that may be emitted in multiple
537 translation units, but that are not guaranteed to be emitted into every
538 translation unit that uses them. One example of this is tentative
539 definitions in C, such as "<tt>int X;</tt>" at global scope.
540 </dd>
541
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]542 <dt><tt><b><a name="linkage_weak">weak</a></b></tt>: </dt>
543
Dale Johannesen96e7e092008-05-23 23:13:41 +0000[diff] [blame]544 <dd>"<tt>weak</tt>" linkage is the same as <tt>common</tt> linkage, except
545 that some targets may choose to emit different assembly sequences for them
546 for target-dependent reasons. This is used for globals that are declared
547 "weak" in C source code.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]548 </dd>
549
550 <dt><tt><b><a name="linkage_appending">appending</a></b></tt>: </dt>
551
552 <dd>"<tt>appending</tt>" linkage may only be applied to global variables of
553 pointer to array type. When two global variables with appending linkage are
554 linked together, the two global arrays are appended together. This is the
555 LLVM, typesafe, equivalent of having the system linker append together
556 "sections" with identical names when .o files are linked.
557 </dd>
558
559 <dt><tt><b><a name="linkage_externweak">extern_weak</a></b></tt>: </dt>
Duncan Sands19d161f2009-03-07 15:45:40 +0000[diff] [blame]560
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]561 <dd>The semantics of this linkage follow the ELF object file model: the
562 symbol is weak until linked, if not linked, the symbol becomes null instead
563 of being an undefined reference.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]564 </dd>
565
Duncan Sands19d161f2009-03-07 15:45:40 +0000[diff] [blame]566 <dt><tt><b><a name="linkage_linkonce">linkonce_odr</a></b></tt>: </dt>
Duncan Sands19d161f2009-03-07 15:45:40 +0000[diff] [blame]567 <dt><tt><b><a name="linkage_weak">weak_odr</a></b></tt>: </dt>
Chris Lattner68433442009-04-13 05:44:34 +0000[diff] [blame]568 <dd>Some languages allow differing globals to be merged, such as two
Duncan Sands19d161f2009-03-07 15:45:40 +0000[diff] [blame]569 functions with different semantics. Other languages, such as <tt>C++</tt>,
570 ensure that only equivalent globals are ever merged (the "one definition
Chris Lattner68433442009-04-13 05:44:34 +0000[diff] [blame]571 rule" - "ODR"). Such languages can use the <tt>linkonce_odr</tt>
Duncan Sandsb95df792009-03-11 20:14:15 +0000[diff] [blame]572 and <tt>weak_odr</tt> linkage types to indicate that the global will only
573 be merged with equivalent globals. These linkage types are otherwise the
574 same as their non-<tt>odr</tt> versions.
Duncan Sands19d161f2009-03-07 15:45:40 +0000[diff] [blame]575 </dd>
576
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]577 <dt><tt><b><a name="linkage_external">externally visible</a></b></tt>:</dt>
578
579 <dd>If none of the above identifiers are used, the global is externally
580 visible, meaning that it participates in linkage and can be used to resolve
581 external symbol references.
582 </dd>
583</dl>
584
585 <p>
586 The next two types of linkage are targeted for Microsoft Windows platform
587 only. They are designed to support importing (exporting) symbols from (to)
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]588 DLLs (Dynamic Link Libraries).
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]589 </p>
590
591 <dl>
592 <dt><tt><b><a name="linkage_dllimport">dllimport</a></b></tt>: </dt>
593
594 <dd>"<tt>dllimport</tt>" linkage causes the compiler to reference a function
595 or variable via a global pointer to a pointer that is set up by the DLL
596 exporting the symbol. On Microsoft Windows targets, the pointer name is
Dan Gohman5ec99832009-01-12 21:35:55 +0000[diff] [blame]597 formed by combining <code>__imp_</code> and the function or variable name.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]598 </dd>
599
600 <dt><tt><b><a name="linkage_dllexport">dllexport</a></b></tt>: </dt>
601
602 <dd>"<tt>dllexport</tt>" linkage causes the compiler to provide a global
603 pointer to a pointer in a DLL, so that it can be referenced with the
604 <tt>dllimport</tt> attribute. On Microsoft Windows targets, the pointer
Dan Gohman5ec99832009-01-12 21:35:55 +0000[diff] [blame]605 name is formed by combining <code>__imp_</code> and the function or variable
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]606 name.
607 </dd>
608
609</dl>
610
Dan Gohman4dfac702008-11-24 17:18:39 +0000[diff] [blame]611<p>For example, since the "<tt>.LC0</tt>"
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]612variable is defined to be internal, if another module defined a "<tt>.LC0</tt>"
613variable and was linked with this one, one of the two would be renamed,
614preventing a collision. Since "<tt>main</tt>" and "<tt>puts</tt>" are
615external (i.e., lacking any linkage declarations), they are accessible
616outside of the current module.</p>
617<p>It is illegal for a function <i>declaration</i>
Duncan Sands565f65d2009-03-11 08:08:06 +0000[diff] [blame]618to have any linkage type other than "externally visible", <tt>dllimport</tt>
619or <tt>extern_weak</tt>.</p>
Duncan Sands19d161f2009-03-07 15:45:40 +0000[diff] [blame]620<p>Aliases can have only <tt>external</tt>, <tt>internal</tt>, <tt>weak</tt>
621or <tt>weak_odr</tt> linkages.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]622</div>
623
624<!-- ======================================================================= -->
625<div class="doc_subsection">
626 <a name="callingconv">Calling Conventions</a>
627</div>
628
629<div class="doc_text">
630
631<p>LLVM <a href="#functionstructure">functions</a>, <a href="#i_call">calls</a>
632and <a href="#i_invoke">invokes</a> can all have an optional calling convention
633specified for the call. The calling convention of any pair of dynamic
634caller/callee must match, or the behavior of the program is undefined. The
635following calling conventions are supported by LLVM, and more may be added in
636the future:</p>
637
638<dl>
639 <dt><b>"<tt>ccc</tt>" - The C calling convention</b>:</dt>
640
641 <dd>This calling convention (the default if no other calling convention is
642 specified) matches the target C calling conventions. This calling convention
643 supports varargs function calls and tolerates some mismatch in the declared
644 prototype and implemented declaration of the function (as does normal C).
645 </dd>
646
647 <dt><b>"<tt>fastcc</tt>" - The fast calling convention</b>:</dt>
648
649 <dd>This calling convention attempts to make calls as fast as possible
650 (e.g. by passing things in registers). This calling convention allows the
651 target to use whatever tricks it wants to produce fast code for the target,
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]652 without having to conform to an externally specified ABI (Application Binary
653 Interface). Implementations of this convention should allow arbitrary
Arnold Schwaighofer07444922008-05-14 09:17:12 +0000[diff] [blame]654 <a href="CodeGenerator.html#tailcallopt">tail call optimization</a> to be
655 supported. This calling convention does not support varargs and requires the
656 prototype of all callees to exactly match the prototype of the function
657 definition.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]658 </dd>
659
660 <dt><b>"<tt>coldcc</tt>" - The cold calling convention</b>:</dt>
661
662 <dd>This calling convention attempts to make code in the caller as efficient
663 as possible under the assumption that the call is not commonly executed. As
664 such, these calls often preserve all registers so that the call does not break
665 any live ranges in the caller side. This calling convention does not support
666 varargs and requires the prototype of all callees to exactly match the
667 prototype of the function definition.
668 </dd>
669
670 <dt><b>"<tt>cc &lt;<em>n</em>&gt;</tt>" - Numbered convention</b>:</dt>
671
672 <dd>Any calling convention may be specified by number, allowing
673 target-specific calling conventions to be used. Target specific calling
674 conventions start at 64.
675 </dd>
676</dl>
677
678<p>More calling conventions can be added/defined on an as-needed basis, to
679support pascal conventions or any other well-known target-independent
680convention.</p>
681
682</div>
683
684<!-- ======================================================================= -->
685<div class="doc_subsection">
686 <a name="visibility">Visibility Styles</a>
687</div>
688
689<div class="doc_text">
690
691<p>
692All Global Variables and Functions have one of the following visibility styles:
693</p>
694
695<dl>
696 <dt><b>"<tt>default</tt>" - Default style</b>:</dt>
697
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]698 <dd>On targets that use the ELF object file format, default visibility means
699 that the declaration is visible to other
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]700 modules and, in shared libraries, means that the declared entity may be
701 overridden. On Darwin, default visibility means that the declaration is
702 visible to other modules. Default visibility corresponds to "external
703 linkage" in the language.
704 </dd>
705
706 <dt><b>"<tt>hidden</tt>" - Hidden style</b>:</dt>
707
708 <dd>Two declarations of an object with hidden visibility refer to the same
709 object if they are in the same shared object. Usually, hidden visibility
710 indicates that the symbol will not be placed into the dynamic symbol table,
711 so no other module (executable or shared library) can reference it
712 directly.
713 </dd>
714
715 <dt><b>"<tt>protected</tt>" - Protected style</b>:</dt>
716
717 <dd>On ELF, protected visibility indicates that the symbol will be placed in
718 the dynamic symbol table, but that references within the defining module will
719 bind to the local symbol. That is, the symbol cannot be overridden by another
720 module.
721 </dd>
722</dl>
723
724</div>
725
726<!-- ======================================================================= -->
727<div class="doc_subsection">
Chris Lattner5b6dc6e2009-01-11 20:53:49 +0000[diff] [blame]728 <a name="namedtypes">Named Types</a>
729</div>
730
731<div class="doc_text">
732
733<p>LLVM IR allows you to specify name aliases for certain types. This can make
734it easier to read the IR and make the IR more condensed (particularly when
735recursive types are involved). An example of a name specification is:
736</p>
737
738<div class="doc_code">
739<pre>
740%mytype = type { %mytype*, i32 }
741</pre>
742</div>
743
744<p>You may give a name to any <a href="#typesystem">type</a> except "<a
745href="t_void">void</a>". Type name aliases may be used anywhere a type is
746expected with the syntax "%mytype".</p>
747
748<p>Note that type names are aliases for the structural type that they indicate,
749and that you can therefore specify multiple names for the same type. This often
750leads to confusing behavior when dumping out a .ll file. Since LLVM IR uses
751structural typing, the name is not part of the type. When printing out LLVM IR,
752the printer will pick <em>one name</em> to render all types of a particular
753shape. This means that if you have code where two different source types end up
754having the same LLVM type, that the dumper will sometimes print the "wrong" or
755unexpected type. This is an important design point and isn't going to
756change.</p>
757
758</div>
759
Chris Lattner5b6dc6e2009-01-11 20:53:49 +0000[diff] [blame]760<!-- ======================================================================= -->
761<div class="doc_subsection">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]762 <a name="globalvars">Global Variables</a>
763</div>
764
765<div class="doc_text">
766
767<p>Global variables define regions of memory allocated at compilation time
768instead of run-time. Global variables may optionally be initialized, may have
769an explicit section to be placed in, and may have an optional explicit alignment
770specified. A variable may be defined as "thread_local", which means that it
771will not be shared by threads (each thread will have a separated copy of the
772variable). A variable may be defined as a global "constant," which indicates
773that the contents of the variable will <b>never</b> be modified (enabling better
774optimization, allowing the global data to be placed in the read-only section of
775an executable, etc). Note that variables that need runtime initialization
776cannot be marked "constant" as there is a store to the variable.</p>
777
778<p>
779LLVM explicitly allows <em>declarations</em> of global variables to be marked
780constant, even if the final definition of the global is not. This capability
781can be used to enable slightly better optimization of the program, but requires
782the language definition to guarantee that optimizations based on the
783'constantness' are valid for the translation units that do not include the
784definition.
785</p>
786
787<p>As SSA values, global variables define pointer values that are in
788scope (i.e. they dominate) all basic blocks in the program. Global
789variables always define a pointer to their "content" type because they
790describe a region of memory, and all memory objects in LLVM are
791accessed through pointers.</p>
792
Christopher Lambdd0049d2007-12-11 09:31:00 +0000[diff] [blame]793<p>A global variable may be declared to reside in a target-specifc numbered
794address space. For targets that support them, address spaces may affect how
795optimizations are performed and/or what target instructions are used to access
Christopher Lamb20a39e92007-12-12 08:44:39 +0000[diff] [blame]796the variable. The default address space is zero. The address space qualifier
797must precede any other attributes.</p>
Christopher Lambdd0049d2007-12-11 09:31:00 +0000[diff] [blame]798
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]799<p>LLVM allows an explicit section to be specified for globals. If the target
800supports it, it will emit globals to the section specified.</p>
801
802<p>An explicit alignment may be specified for a global. If not present, or if
803the alignment is set to zero, the alignment of the global is set by the target
804to whatever it feels convenient. If an explicit alignment is specified, the
805global is forced to have at least that much alignment. All alignments must be
806a power of 2.</p>
807
Christopher Lambdd0049d2007-12-11 09:31:00 +0000[diff] [blame]808<p>For example, the following defines a global in a numbered address space with
809an initializer, section, and alignment:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]810
811<div class="doc_code">
812<pre>
Dan Gohman21ef02c2009-01-11 00:40:00 +0000[diff] [blame]813@G = addrspace(5) constant float 1.0, section "foo", align 4
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]814</pre>
815</div>
816
817</div>
818
819
820<!-- ======================================================================= -->
821<div class="doc_subsection">
822 <a name="functionstructure">Functions</a>
823</div>
824
825<div class="doc_text">
826
827<p>LLVM function definitions consist of the "<tt>define</tt>" keyord,
828an optional <a href="#linkage">linkage type</a>, an optional
829<a href="#visibility">visibility style</a>, an optional
830<a href="#callingconv">calling convention</a>, a return type, an optional
831<a href="#paramattrs">parameter attribute</a> for the return type, a function
832name, a (possibly empty) argument list (each with optional
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]833<a href="#paramattrs">parameter attributes</a>), optional
834<a href="#fnattrs">function attributes</a>, an optional section,
835an optional alignment, an optional <a href="#gc">garbage collector name</a>,
Chris Lattneree202542008-10-04 18:10:21 +0000[diff] [blame]836an opening curly brace, a list of basic blocks, and a closing curly brace.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]837
838LLVM function declarations consist of the "<tt>declare</tt>" keyword, an
839optional <a href="#linkage">linkage type</a>, an optional
840<a href="#visibility">visibility style</a>, an optional
841<a href="#callingconv">calling convention</a>, a return type, an optional
842<a href="#paramattrs">parameter attribute</a> for the return type, a function
Gordon Henriksen13fe5e32007-12-10 03:18:06 +0000[diff] [blame]843name, a possibly empty list of arguments, an optional alignment, and an optional
Gordon Henriksend606f5b2007-12-10 03:30:21 +0000[diff] [blame]844<a href="#gc">garbage collector name</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]845
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]846<p>A function definition contains a list of basic blocks, forming the CFG
847(Control Flow Graph) for
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]848the function. Each basic block may optionally start with a label (giving the
849basic block a symbol table entry), contains a list of instructions, and ends
850with a <a href="#terminators">terminator</a> instruction (such as a branch or
851function return).</p>
852
853<p>The first basic block in a function is special in two ways: it is immediately
854executed on entrance to the function, and it is not allowed to have predecessor
855basic blocks (i.e. there can not be any branches to the entry block of a
856function). Because the block can have no predecessors, it also cannot have any
857<a href="#i_phi">PHI nodes</a>.</p>
858
859<p>LLVM allows an explicit section to be specified for functions. If the target
860supports it, it will emit functions to the section specified.</p>
861
862<p>An explicit alignment may be specified for a function. If not present, or if
863the alignment is set to zero, the alignment of the function is set by the target
864to whatever it feels convenient. If an explicit alignment is specified, the
865function is forced to have at least that much alignment. All alignments must be
866a power of 2.</p>
867
Devang Pateld0bfcc72008-10-07 17:48:33 +0000[diff] [blame]868 <h5>Syntax:</h5>
869
870<div class="doc_code">
Chris Lattner1e5c5cd02008-10-13 16:55:18 +0000[diff] [blame]871<tt>
872define [<a href="#linkage">linkage</a>] [<a href="#visibility">visibility</a>]
873 [<a href="#callingconv">cconv</a>] [<a href="#paramattrs">ret attrs</a>]
874 &lt;ResultType&gt; @&lt;FunctionName&gt; ([argument list])
875 [<a href="#fnattrs">fn Attrs</a>] [section "name"] [align N]
876 [<a href="#gc">gc</a>] { ... }
877</tt>
Devang Pateld0bfcc72008-10-07 17:48:33 +0000[diff] [blame]878</div>
879
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]880</div>
881
882
883<!-- ======================================================================= -->
884<div class="doc_subsection">
885 <a name="aliasstructure">Aliases</a>
886</div>
887<div class="doc_text">
888 <p>Aliases act as "second name" for the aliasee value (which can be either
Anton Korobeynikov96822812008-03-22 08:36:14 +0000[diff] [blame]889 function, global variable, another alias or bitcast of global value). Aliases
890 may have an optional <a href="#linkage">linkage type</a>, and an
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]891 optional <a href="#visibility">visibility style</a>.</p>
892
893 <h5>Syntax:</h5>
894
895<div class="doc_code">
896<pre>
Duncan Sandsd7bfabf2008-09-12 20:48:21 +0000[diff] [blame]897@&lt;Name&gt; = alias [Linkage] [Visibility] &lt;AliaseeTy&gt; @&lt;Aliasee&gt;
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]898</pre>
899</div>
900
901</div>
902
903
904
905<!-- ======================================================================= -->
906<div class="doc_subsection"><a name="paramattrs">Parameter Attributes</a></div>
907<div class="doc_text">
908 <p>The return type and each parameter of a function type may have a set of
909 <i>parameter attributes</i> associated with them. Parameter attributes are
910 used to communicate additional information about the result or parameters of
Duncan Sandsf5588dc2007-11-27 13:23:08 +0000[diff] [blame]911 a function. Parameter attributes are considered to be part of the function,
912 not of the function type, so functions with different parameter attributes
913 can have the same function type.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]914
915 <p>Parameter attributes are simple keywords that follow the type specified. If
916 multiple parameter attributes are needed, they are space separated. For
917 example:</p>
918
919<div class="doc_code">
920<pre>
Nick Lewycky3022a742009-02-15 23:06:14 +0000[diff] [blame]921declare i32 @printf(i8* noalias nocapture, ...)
Chris Lattnerf33b8452008-10-04 18:33:34 +0000[diff] [blame]922declare i32 @atoi(i8 zeroext)
923declare signext i8 @returns_signed_char()
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]924</pre>
925</div>
926
Duncan Sandsf5588dc2007-11-27 13:23:08 +0000[diff] [blame]927 <p>Note that any attributes for the function result (<tt>nounwind</tt>,
928 <tt>readonly</tt>) come immediately after the argument list.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]929
930 <p>Currently, only the following parameter attributes are defined:</p>
931 <dl>
Reid Spencerf234bed2007-07-19 23:13:04 +0000[diff] [blame]932 <dt><tt>zeroext</tt></dt>
Chris Lattnerf33b8452008-10-04 18:33:34 +0000[diff] [blame]933 <dd>This indicates to the code generator that the parameter or return value
934 should be zero-extended to a 32-bit value by the caller (for a parameter)
935 or the callee (for a return value).</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]936
Reid Spencerf234bed2007-07-19 23:13:04 +0000[diff] [blame]937 <dt><tt>signext</tt></dt>
Chris Lattnerf33b8452008-10-04 18:33:34 +0000[diff] [blame]938 <dd>This indicates to the code generator that the parameter or return value
939 should be sign-extended to a 32-bit value by the caller (for a parameter)
940 or the callee (for a return value).</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]941
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]942 <dt><tt>inreg</tt></dt>
Dale Johannesenc08a0e22008-09-25 20:47:45 +0000[diff] [blame]943 <dd>This indicates that this parameter or return value should be treated
944 in a special target-dependent fashion during while emitting code for a
945 function call or return (usually, by putting it in a register as opposed
Chris Lattnerf33b8452008-10-04 18:33:34 +0000[diff] [blame]946 to memory, though some targets use it to distinguish between two different
947 kinds of registers). Use of this attribute is target-specific.</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]948
Duncan Sands7d94e5a2008-10-06 08:14:18 +0000[diff] [blame]949 <dt><tt><a name="byval">byval</a></tt></dt>
Chris Lattner04c86182008-01-15 04:34:22 +0000[diff] [blame]950 <dd>This indicates that the pointer parameter should really be passed by
951 value to the function. The attribute implies that a hidden copy of the
952 pointee is made between the caller and the callee, so the callee is unable
Chris Lattner6a9f3c42008-08-05 18:21:08 +0000[diff] [blame]953 to modify the value in the callee. This attribute is only valid on LLVM
Chris Lattner04c86182008-01-15 04:34:22 +0000[diff] [blame]954 pointer arguments. It is generally used to pass structs and arrays by
Duncan Sands7d94e5a2008-10-06 08:14:18 +0000[diff] [blame]955 value, but is also valid on pointers to scalars. The copy is considered to
956 belong to the caller not the callee (for example,
957 <tt><a href="#readonly">readonly</a></tt> functions should not write to
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]958 <tt>byval</tt> parameters). This is not a valid attribute for return
Chris Lattnere39c5bf2009-02-05 05:42:28 +0000[diff] [blame]959 values. The byval attribute also supports specifying an alignment with the
960 align attribute. This has a target-specific effect on the code generator
961 that usually indicates a desired alignment for the synthesized stack
962 slot.</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]963
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]964 <dt><tt>sret</tt></dt>
Duncan Sands616cc032008-02-18 04:19:38 +0000[diff] [blame]965 <dd>This indicates that the pointer parameter specifies the address of a
966 structure that is the return value of the function in the source program.
Chris Lattnerf33b8452008-10-04 18:33:34 +0000[diff] [blame]967 This pointer must be guaranteed by the caller to be valid: loads and stores
968 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]969 be applied to the first parameter. This is not a valid attribute for
970 return values. </dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]971
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]972 <dt><tt>noalias</tt></dt>
Nick Lewyckyff384472008-11-24 03:41:24 +0000[diff] [blame]973 <dd>This indicates that the pointer does not alias any global or any other
974 parameter. The caller is responsible for ensuring that this is the
Nick Lewyckya604a942008-11-24 05:00:44 +0000[diff] [blame]975 case. On a function return value, <tt>noalias</tt> additionally indicates
976 that the pointer does not alias any other pointers visible to the
Nick Lewycky40fb6f22008-12-19 06:39:12 +0000[diff] [blame]977 caller. For further details, please see the discussion of the NoAlias
978 response in
979 <a href="http://llvm.org/docs/AliasAnalysis.html#MustMayNo">alias
980 analysis</a>.</dd>
981
982 <dt><tt>nocapture</tt></dt>
983 <dd>This indicates that the callee does not make any copies of the pointer
984 that outlive the callee itself. This is not a valid attribute for return
985 values.</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]986
Duncan Sands4ee46812007-07-27 19:57:41 +0000[diff] [blame]987 <dt><tt>nest</tt></dt>
Duncan Sandsf1a7d4c2008-07-08 09:27:25 +0000[diff] [blame]988 <dd>This indicates that the pointer parameter can be excised using the
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]989 <a href="#int_trampoline">trampoline intrinsics</a>. This is not a valid
990 attribute for return values.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]991 </dl>
992
993</div>
994
995<!-- ======================================================================= -->
996<div class="doc_subsection">
Gordon Henriksen13fe5e32007-12-10 03:18:06 +0000[diff] [blame]997 <a name="gc">Garbage Collector Names</a>
998</div>
999
1000<div class="doc_text">
1001<p>Each function may specify a garbage collector name, which is simply a
1002string.</p>
1003
1004<div class="doc_code"><pre
1005>define void @f() gc "name" { ...</pre></div>
1006
1007<p>The compiler declares the supported values of <i>name</i>. Specifying a
1008collector which will cause the compiler to alter its output in order to support
1009the named garbage collection algorithm.</p>
1010</div>
1011
1012<!-- ======================================================================= -->
1013<div class="doc_subsection">
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]1014 <a name="fnattrs">Function Attributes</a>
Devang Pateld468f1c2008-09-04 23:05:13 +0000[diff] [blame]1015</div>
1016
1017<div class="doc_text">
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]1018
1019<p>Function attributes are set to communicate additional information about
1020 a function. Function attributes are considered to be part of the function,
1021 not of the function type, so functions with different parameter attributes
1022 can have the same function type.</p>
1023
1024 <p>Function attributes are simple keywords that follow the type specified. If
1025 multiple attributes are needed, they are space separated. For
1026 example:</p>
Devang Pateld468f1c2008-09-04 23:05:13 +0000[diff] [blame]1027
1028<div class="doc_code">
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]1029<pre>
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]1030define void @f() noinline { ... }
1031define void @f() alwaysinline { ... }
1032define void @f() alwaysinline optsize { ... }
1033define void @f() optsize
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]1034</pre>
Devang Pateld468f1c2008-09-04 23:05:13 +0000[diff] [blame]1035</div>
1036
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]1037<dl>
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]1038<dt><tt>alwaysinline</tt></dt>
Chris Lattner82d70f92008-10-04 18:23:17 +0000[diff] [blame]1039<dd>This attribute indicates that the inliner should attempt to inline this
1040function into callers whenever possible, ignoring any active inlining size
1041threshold for this caller.</dd>
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]1042
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]1043<dt><tt>noinline</tt></dt>
Chris Lattner82d70f92008-10-04 18:23:17 +0000[diff] [blame]1044<dd>This attribute indicates that the inliner should never inline this function
Chris Lattner377f3ae2008-10-05 17:14:59 +0000[diff] [blame]1045in any situation. This attribute may not be used together with the
Chris Lattner82d70f92008-10-04 18:23:17 +0000[diff] [blame]1046<tt>alwaysinline</tt> attribute.</dd>
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]1047
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]1048<dt><tt>optsize</tt></dt>
Devang Patelc29099b2008-09-29 18:34:44 +0000[diff] [blame]1049<dd>This attribute suggests that optimization passes and code generator passes
Chris Lattner82d70f92008-10-04 18:23:17 +0000[diff] [blame]1050make choices that keep the code size of this function low, and otherwise do
1051optimizations specifically to reduce code size.</dd>
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]1052
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]1053<dt><tt>noreturn</tt></dt>
Chris Lattner82d70f92008-10-04 18:23:17 +0000[diff] [blame]1054<dd>This function attribute indicates that the function never returns normally.
1055This produces undefined behavior at runtime if the function ever does
1056dynamically return.</dd>
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]1057
1058<dt><tt>nounwind</tt></dt>
Chris Lattner82d70f92008-10-04 18:23:17 +0000[diff] [blame]1059<dd>This function attribute indicates that the function never returns with an
1060unwind or exceptional control flow. If the function does unwind, its runtime
1061behavior is undefined.</dd>
1062
1063<dt><tt>readnone</tt></dt>
Duncan Sands2f500832009-05-06 06:49:50 +0000[diff] [blame^]1064<dd>This attribute indicates that the function computes its result (or decides to
1065unwind an exception) based strictly on its arguments, without dereferencing any
Duncan Sands7d94e5a2008-10-06 08:14:18 +0000[diff] [blame]1066pointer arguments or otherwise accessing any mutable state (e.g. memory, control
1067registers, etc) visible to caller functions. It does not write through any
1068pointer arguments (including <tt><a href="#byval">byval</a></tt> arguments) and
Duncan Sands2f500832009-05-06 06:49:50 +0000[diff] [blame^]1069never changes any state visible to callers. This means that it cannot unwind
1070exceptions by calling the <tt>C++</tt> exception throwing methods, but could
1071use the <tt>unwind</tt> instruction.</dd>
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]1072
Duncan Sands7d94e5a2008-10-06 08:14:18 +0000[diff] [blame]1073<dt><tt><a name="readonly">readonly</a></tt></dt>
1074<dd>This attribute indicates that the function does not write through any
1075pointer arguments (including <tt><a href="#byval">byval</a></tt> arguments)
1076or otherwise modify any state (e.g. memory, control registers, etc) visible to
1077caller functions. It may dereference pointer arguments and read state that may
Duncan Sands2f500832009-05-06 06:49:50 +0000[diff] [blame^]1078be set in the caller. A readonly function always returns the same value (or
1079unwinds an exception identically) when called with the same set of arguments
1080and global state. It cannot unwind an exception by calling the <tt>C++</tt>
1081exception throwing methods, but may use the <tt>unwind</tt> instruction.</dd>
Bill Wendlingdfaabba2008-11-13 01:02:51 +0000[diff] [blame]1082
1083<dt><tt><a name="ssp">ssp</a></tt></dt>
Bill Wendling34f70d82008-11-26 19:19:05 +0000[diff] [blame]1084<dd>This attribute indicates that the function should emit a stack smashing
Bill Wendlingdfaabba2008-11-13 01:02:51 +0000[diff] [blame]1085protector. It is in the form of a "canary"&mdash;a random value placed on the
1086stack before the local variables that's checked upon return from the function to
1087see if it has been overwritten. A heuristic is used to determine if a function
Bill Wendling34f70d82008-11-26 19:19:05 +0000[diff] [blame]1088needs stack protectors or not.
Bill Wendlingdfaabba2008-11-13 01:02:51 +0000[diff] [blame]1089
Bill Wendlingbe9ec3f2008-11-26 19:07:40 +0000[diff] [blame]1090<p>If a function that has an <tt>ssp</tt> attribute is inlined into a function
1091that doesn't have an <tt>ssp</tt> attribute, then the resulting function will
1092have an <tt>ssp</tt> attribute.</p></dd>
1093
1094<dt><tt>sspreq</tt></dt>
Bill Wendling34f70d82008-11-26 19:19:05 +0000[diff] [blame]1095<dd>This attribute indicates that the function should <em>always</em> emit a
Bill Wendlingdfaabba2008-11-13 01:02:51 +0000[diff] [blame]1096stack smashing protector. This overrides the <tt><a href="#ssp">ssp</a></tt>
Bill Wendling34f70d82008-11-26 19:19:05 +0000[diff] [blame]1097function attribute.
Bill Wendlingbe9ec3f2008-11-26 19:07:40 +0000[diff] [blame]1098
1099<p>If a function that has an <tt>sspreq</tt> attribute is inlined into a
1100function that doesn't have an <tt>sspreq</tt> attribute or which has
1101an <tt>ssp</tt> attribute, then the resulting function will have
1102an <tt>sspreq</tt> attribute.</p></dd>
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]1103</dl>
1104
Devang Pateld468f1c2008-09-04 23:05:13 +0000[diff] [blame]1105</div>
1106
1107<!-- ======================================================================= -->
1108<div class="doc_subsection">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1109 <a name="moduleasm">Module-Level Inline Assembly</a>
1110</div>
1111
1112<div class="doc_text">
1113<p>
1114Modules may contain "module-level inline asm" blocks, which corresponds to the
1115GCC "file scope inline asm" blocks. These blocks are internally concatenated by
1116LLVM and treated as a single unit, but may be separated in the .ll file if
1117desired. The syntax is very simple:
1118</p>
1119
1120<div class="doc_code">
1121<pre>
1122module asm "inline asm code goes here"
1123module asm "more can go here"
1124</pre>
1125</div>
1126
1127<p>The strings can contain any character by escaping non-printable characters.
1128 The escape sequence used is simply "\xx" where "xx" is the two digit hex code
1129 for the number.
1130</p>
1131
1132<p>
1133 The inline asm code is simply printed to the machine code .s file when
1134 assembly code is generated.
1135</p>
1136</div>
1137
1138<!-- ======================================================================= -->
1139<div class="doc_subsection">
1140 <a name="datalayout">Data Layout</a>
1141</div>
1142
1143<div class="doc_text">
1144<p>A module may specify a target specific data layout string that specifies how
1145data is to be laid out in memory. The syntax for the data layout is simply:</p>
1146<pre> target datalayout = "<i>layout specification</i>"</pre>
1147<p>The <i>layout specification</i> consists of a list of specifications
1148separated by the minus sign character ('-'). Each specification starts with a
1149letter and may include other information after the letter to define some
1150aspect of the data layout. The specifications accepted are as follows: </p>
1151<dl>
1152 <dt><tt>E</tt></dt>
1153 <dd>Specifies that the target lays out data in big-endian form. That is, the
1154 bits with the most significance have the lowest address location.</dd>
1155 <dt><tt>e</tt></dt>
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]1156 <dd>Specifies that the target lays out data in little-endian form. That is,
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1157 the bits with the least significance have the lowest address location.</dd>
1158 <dt><tt>p:<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1159 <dd>This specifies the <i>size</i> of a pointer and its <i>abi</i> and
1160 <i>preferred</i> alignments. All sizes are in bits. Specifying the <i>pref</i>
1161 alignment is optional. If omitted, the preceding <tt>:</tt> should be omitted
1162 too.</dd>
1163 <dt><tt>i<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1164 <dd>This specifies the alignment for an integer type of a given bit
1165 <i>size</i>. The value of <i>size</i> must be in the range [1,2^23).</dd>
1166 <dt><tt>v<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1167 <dd>This specifies the alignment for a vector type of a given bit
1168 <i>size</i>.</dd>
1169 <dt><tt>f<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1170 <dd>This specifies the alignment for a floating point type of a given bit
1171 <i>size</i>. The value of <i>size</i> must be either 32 (float) or 64
1172 (double).</dd>
1173 <dt><tt>a<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1174 <dd>This specifies the alignment for an aggregate type of a given bit
1175 <i>size</i>.</dd>
1176</dl>
1177<p>When constructing the data layout for a given target, LLVM starts with a
1178default set of specifications which are then (possibly) overriden by the
1179specifications in the <tt>datalayout</tt> keyword. The default specifications
1180are given in this list:</p>
1181<ul>
1182 <li><tt>E</tt> - big endian</li>
1183 <li><tt>p:32:64:64</tt> - 32-bit pointers with 64-bit alignment</li>
1184 <li><tt>i1:8:8</tt> - i1 is 8-bit (byte) aligned</li>
1185 <li><tt>i8:8:8</tt> - i8 is 8-bit (byte) aligned</li>
1186 <li><tt>i16:16:16</tt> - i16 is 16-bit aligned</li>
1187 <li><tt>i32:32:32</tt> - i32 is 32-bit aligned</li>
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]1188 <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]1189 alignment of 64-bits</li>
1190 <li><tt>f32:32:32</tt> - float is 32-bit aligned</li>
1191 <li><tt>f64:64:64</tt> - double is 64-bit aligned</li>
1192 <li><tt>v64:64:64</tt> - 64-bit vector is 64-bit aligned</li>
1193 <li><tt>v128:128:128</tt> - 128-bit vector is 128-bit aligned</li>
1194 <li><tt>a0:0:1</tt> - aggregates are 8-bit aligned</li>
1195</ul>
Chris Lattner6a9f3c42008-08-05 18:21:08 +0000[diff] [blame]1196<p>When LLVM is determining the alignment for a given type, it uses the
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]1197following rules:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1198<ol>
1199 <li>If the type sought is an exact match for one of the specifications, that
1200 specification is used.</li>
1201 <li>If no match is found, and the type sought is an integer type, then the
1202 smallest integer type that is larger than the bitwidth of the sought type is
1203 used. If none of the specifications are larger than the bitwidth then the the
1204 largest integer type is used. For example, given the default specifications
1205 above, the i7 type will use the alignment of i8 (next largest) while both
1206 i65 and i256 will use the alignment of i64 (largest specified).</li>
1207 <li>If no match is found, and the type sought is a vector type, then the
1208 largest vector type that is smaller than the sought vector type will be used
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]1209 as a fall back. This happens because &lt;128 x double&gt; can be implemented
1210 in terms of 64 &lt;2 x double&gt;, for example.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1211</ol>
1212</div>
1213
1214<!-- *********************************************************************** -->
1215<div class="doc_section"> <a name="typesystem">Type System</a> </div>
1216<!-- *********************************************************************** -->
1217
1218<div class="doc_text">
1219
1220<p>The LLVM type system is one of the most important features of the
1221intermediate representation. Being typed enables a number of
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]1222optimizations to be performed on the intermediate representation directly,
1223without having to do
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1224extra analyses on the side before the transformation. A strong type
1225system makes it easier to read the generated code and enables novel
1226analyses and transformations that are not feasible to perform on normal
1227three address code representations.</p>
1228
1229</div>
1230
1231<!-- ======================================================================= -->
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1232<div class="doc_subsection"> <a name="t_classifications">Type
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1233Classifications</a> </div>
1234<div class="doc_text">
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1235<p>The types fall into a few useful
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1236classifications:</p>
1237
1238<table border="1" cellspacing="0" cellpadding="4">
1239 <tbody>
1240 <tr><th>Classification</th><th>Types</th></tr>
1241 <tr>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1242 <td><a href="#t_integer">integer</a></td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1243 <td><tt>i1, i2, i3, ... i8, ... i16, ... i32, ... i64, ... </tt></td>
1244 </tr>
1245 <tr>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1246 <td><a href="#t_floating">floating point</a></td>
1247 <td><tt>float, double, x86_fp80, fp128, ppc_fp128</tt></td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1248 </tr>
1249 <tr>
1250 <td><a name="t_firstclass">first class</a></td>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1251 <td><a href="#t_integer">integer</a>,
1252 <a href="#t_floating">floating point</a>,
1253 <a href="#t_pointer">pointer</a>,
Dan Gohmanf6237db2008-06-18 18:42:13 +0000[diff] [blame]1254 <a href="#t_vector">vector</a>,
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]1255 <a href="#t_struct">structure</a>,
1256 <a href="#t_array">array</a>,
Dan Gohmand13951e2008-05-23 22:50:26 +0000[diff] [blame]1257 <a href="#t_label">label</a>.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1258 </td>
1259 </tr>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1260 <tr>
1261 <td><a href="#t_primitive">primitive</a></td>
1262 <td><a href="#t_label">label</a>,
1263 <a href="#t_void">void</a>,
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1264 <a href="#t_floating">floating point</a>.</td>
1265 </tr>
1266 <tr>
1267 <td><a href="#t_derived">derived</a></td>
1268 <td><a href="#t_integer">integer</a>,
1269 <a href="#t_array">array</a>,
1270 <a href="#t_function">function</a>,
1271 <a href="#t_pointer">pointer</a>,
1272 <a href="#t_struct">structure</a>,
1273 <a href="#t_pstruct">packed structure</a>,
1274 <a href="#t_vector">vector</a>,
1275 <a href="#t_opaque">opaque</a>.
Dan Gohman032ba852008-10-14 16:32:04 +0000[diff] [blame]1276 </td>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1277 </tr>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1278 </tbody>
1279</table>
1280
1281<p>The <a href="#t_firstclass">first class</a> types are perhaps the
1282most important. Values of these types are the only ones which can be
1283produced by instructions, passed as arguments, or used as operands to
Dan Gohman4f29e422008-05-23 21:53:15 +0000[diff] [blame]1284instructions.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1285</div>
1286
1287<!-- ======================================================================= -->
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1288<div class="doc_subsection"> <a name="t_primitive">Primitive Types</a> </div>
Chris Lattner86437612008-01-04 04:34:14 +0000[diff] [blame]1289
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1290<div class="doc_text">
1291<p>The primitive types are the fundamental building blocks of the LLVM
1292system.</p>
1293
Chris Lattner86437612008-01-04 04:34:14 +0000[diff] [blame]1294</div>
1295
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1296<!-- _______________________________________________________________________ -->
1297<div class="doc_subsubsection"> <a name="t_floating">Floating Point Types</a> </div>
1298
1299<div class="doc_text">
1300 <table>
1301 <tbody>
1302 <tr><th>Type</th><th>Description</th></tr>
1303 <tr><td><tt>float</tt></td><td>32-bit floating point value</td></tr>
1304 <tr><td><tt>double</tt></td><td>64-bit floating point value</td></tr>
1305 <tr><td><tt>fp128</tt></td><td>128-bit floating point value (112-bit mantissa)</td></tr>
1306 <tr><td><tt>x86_fp80</tt></td><td>80-bit floating point value (X87)</td></tr>
1307 <tr><td><tt>ppc_fp128</tt></td><td>128-bit floating point value (two 64-bits)</td></tr>
1308 </tbody>
1309 </table>
1310</div>
1311
1312<!-- _______________________________________________________________________ -->
1313<div class="doc_subsubsection"> <a name="t_void">Void Type</a> </div>
1314
1315<div class="doc_text">
1316<h5>Overview:</h5>
1317<p>The void type does not represent any value and has no size.</p>
1318
1319<h5>Syntax:</h5>
1320
1321<pre>
1322 void
1323</pre>
1324</div>
1325
1326<!-- _______________________________________________________________________ -->
1327<div class="doc_subsubsection"> <a name="t_label">Label Type</a> </div>
1328
1329<div class="doc_text">
1330<h5>Overview:</h5>
1331<p>The label type represents code labels.</p>
1332
1333<h5>Syntax:</h5>
1334
1335<pre>
1336 label
1337</pre>
1338</div>
1339
1340
1341<!-- ======================================================================= -->
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1342<div class="doc_subsection"> <a name="t_derived">Derived Types</a> </div>
1343
1344<div class="doc_text">
1345
1346<p>The real power in LLVM comes from the derived types in the system.
1347This is what allows a programmer to represent arrays, functions,
1348pointers, and other useful types. Note that these derived types may be
1349recursive: For example, it is possible to have a two dimensional array.</p>
1350
1351</div>
1352
1353<!-- _______________________________________________________________________ -->
1354<div class="doc_subsubsection"> <a name="t_integer">Integer Type</a> </div>
1355
1356<div class="doc_text">
1357
1358<h5>Overview:</h5>
1359<p>The integer type is a very simple derived type that simply specifies an
1360arbitrary bit width for the integer type desired. Any bit width from 1 bit to
13612^23-1 (about 8 million) can be specified.</p>
1362
1363<h5>Syntax:</h5>
1364
1365<pre>
1366 iN
1367</pre>
1368
1369<p>The number of bits the integer will occupy is specified by the <tt>N</tt>
1370value.</p>
1371
1372<h5>Examples:</h5>
1373<table class="layout">
Chris Lattner251ab812007-12-18 06:18:21 +0000[diff] [blame]1374 <tbody>
1375 <tr>
1376 <td><tt>i1</tt></td>
1377 <td>a single-bit integer.</td>
1378 </tr><tr>
1379 <td><tt>i32</tt></td>
1380 <td>a 32-bit integer.</td>
1381 </tr><tr>
1382 <td><tt>i1942652</tt></td>
1383 <td>a really big integer of over 1 million bits.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1384 </tr>
Chris Lattner251ab812007-12-18 06:18:21 +0000[diff] [blame]1385 </tbody>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1386</table>
djge93155c2009-01-24 15:58:40 +0000[diff] [blame]1387
1388<p>Note that the code generator does not yet support large integer types
1389to be used as function return types. The specific limit on how large a
1390return type the code generator can currently handle is target-dependent;
1391currently it's often 64 bits for 32-bit targets and 128 bits for 64-bit
1392targets.</p>
1393
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1394</div>
1395
1396<!-- _______________________________________________________________________ -->
1397<div class="doc_subsubsection"> <a name="t_array">Array Type</a> </div>
1398
1399<div class="doc_text">
1400
1401<h5>Overview:</h5>
1402
1403<p>The array type is a very simple derived type that arranges elements
1404sequentially in memory. The array type requires a size (number of
1405elements) and an underlying data type.</p>
1406
1407<h5>Syntax:</h5>
1408
1409<pre>
1410 [&lt;# elements&gt; x &lt;elementtype&gt;]
1411</pre>
1412
1413<p>The number of elements is a constant integer value; elementtype may
1414be any type with a size.</p>
1415
1416<h5>Examples:</h5>
1417<table class="layout">
1418 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1419 <td class="left"><tt>[40 x i32]</tt></td>
1420 <td class="left">Array of 40 32-bit integer values.</td>
1421 </tr>
1422 <tr class="layout">
1423 <td class="left"><tt>[41 x i32]</tt></td>
1424 <td class="left">Array of 41 32-bit integer values.</td>
1425 </tr>
1426 <tr class="layout">
1427 <td class="left"><tt>[4 x i8]</tt></td>
1428 <td class="left">Array of 4 8-bit integer values.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1429 </tr>
1430</table>
1431<p>Here are some examples of multidimensional arrays:</p>
1432<table class="layout">
1433 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1434 <td class="left"><tt>[3 x [4 x i32]]</tt></td>
1435 <td class="left">3x4 array of 32-bit integer values.</td>
1436 </tr>
1437 <tr class="layout">
1438 <td class="left"><tt>[12 x [10 x float]]</tt></td>
1439 <td class="left">12x10 array of single precision floating point values.</td>
1440 </tr>
1441 <tr class="layout">
1442 <td class="left"><tt>[2 x [3 x [4 x i16]]]</tt></td>
1443 <td class="left">2x3x4 array of 16-bit integer values.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1444 </tr>
1445</table>
1446
1447<p>Note that 'variable sized arrays' can be implemented in LLVM with a zero
1448length array. Normally, accesses past the end of an array are undefined in
1449LLVM (e.g. it is illegal to access the 5th element of a 3 element array).
1450As a special case, however, zero length arrays are recognized to be variable
1451length. This allows implementation of 'pascal style arrays' with the LLVM
1452type "{ i32, [0 x float]}", for example.</p>
1453
djge93155c2009-01-24 15:58:40 +0000[diff] [blame]1454<p>Note that the code generator does not yet support large aggregate types
1455to be used as function return types. The specific limit on how large an
1456aggregate return type the code generator can currently handle is
1457target-dependent, and also dependent on the aggregate element types.</p>
1458
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1459</div>
1460
1461<!-- _______________________________________________________________________ -->
1462<div class="doc_subsubsection"> <a name="t_function">Function Type</a> </div>
1463<div class="doc_text">
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1464
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1465<h5>Overview:</h5>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1466
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1467<p>The function type can be thought of as a function signature. It
Devang Pateld4ba41d2008-03-24 05:35:41 +0000[diff] [blame]1468consists of a return type and a list of formal parameter types. The
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1469return 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]1470If 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]1471class types, and the struct must have at least one element.</p>
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]1472
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1473<h5>Syntax:</h5>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1474
1475<pre>
1476 &lt;returntype list&gt; (&lt;parameter list&gt;)
1477</pre>
1478
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1479<p>...where '<tt>&lt;parameter list&gt;</tt>' is a comma-separated list of type
1480specifiers. Optionally, the parameter list may include a type <tt>...</tt>,
1481which indicates that the function takes a variable number of arguments.
1482Variable argument functions can access their arguments with the <a
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]1483 href="#int_varargs">variable argument handling intrinsic</a> functions.
1484'<tt>&lt;returntype list&gt;</tt>' is a comma-separated list of
1485<a href="#t_firstclass">first class</a> type specifiers.</p>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1486
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1487<h5>Examples:</h5>
1488<table class="layout">
1489 <tr class="layout">
1490 <td class="left"><tt>i32 (i32)</tt></td>
1491 <td class="left">function taking an <tt>i32</tt>, returning an <tt>i32</tt>
1492 </td>
1493 </tr><tr class="layout">
Reid Spencerf234bed2007-07-19 23:13:04 +0000[diff] [blame]1494 <td class="left"><tt>float&nbsp;(i16&nbsp;signext,&nbsp;i32&nbsp;*)&nbsp;*
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1495 </tt></td>
1496 <td class="left"><a href="#t_pointer">Pointer</a> to a function that takes
1497 an <tt>i16</tt> that should be sign extended and a
1498 <a href="#t_pointer">pointer</a> to <tt>i32</tt>, returning
1499 <tt>float</tt>.
1500 </td>
1501 </tr><tr class="layout">
1502 <td class="left"><tt>i32 (i8*, ...)</tt></td>
1503 <td class="left">A vararg function that takes at least one
1504 <a href="#t_pointer">pointer</a> to <tt>i8 </tt> (char in C),
1505 which returns an integer. This is the signature for <tt>printf</tt> in
1506 LLVM.
1507 </td>
Devang Pateld4ba41d2008-03-24 05:35:41 +0000[diff] [blame]1508 </tr><tr class="layout">
1509 <td class="left"><tt>{i32, i32} (i32)</tt></td>
Misha Brukmanafc88b02008-11-27 06:41:20 +0000[diff] [blame]1510 <td class="left">A function taking an <tt>i32</tt>, returning two
1511 <tt>i32</tt> values as an aggregate of type <tt>{ i32, i32 }</tt>
Devang Pateld4ba41d2008-03-24 05:35:41 +0000[diff] [blame]1512 </td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1513 </tr>
1514</table>
1515
1516</div>
1517<!-- _______________________________________________________________________ -->
1518<div class="doc_subsubsection"> <a name="t_struct">Structure Type</a> </div>
1519<div class="doc_text">
1520<h5>Overview:</h5>
1521<p>The structure type is used to represent a collection of data members
1522together in memory. The packing of the field types is defined to match
1523the ABI of the underlying processor. The elements of a structure may
1524be any type that has a size.</p>
1525<p>Structures are accessed using '<tt><a href="#i_load">load</a></tt>
1526and '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a
1527field with the '<tt><a href="#i_getelementptr">getelementptr</a></tt>'
1528instruction.</p>
1529<h5>Syntax:</h5>
1530<pre> { &lt;type list&gt; }<br></pre>
1531<h5>Examples:</h5>
1532<table class="layout">
1533 <tr class="layout">
1534 <td class="left"><tt>{ i32, i32, i32 }</tt></td>
1535 <td class="left">A triple of three <tt>i32</tt> values</td>
1536 </tr><tr class="layout">
1537 <td class="left"><tt>{&nbsp;float,&nbsp;i32&nbsp;(i32)&nbsp;*&nbsp;}</tt></td>
1538 <td class="left">A pair, where the first element is a <tt>float</tt> and the
1539 second element is a <a href="#t_pointer">pointer</a> to a
1540 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
1541 an <tt>i32</tt>.</td>
1542 </tr>
1543</table>
djge93155c2009-01-24 15:58:40 +0000[diff] [blame]1544
1545<p>Note that the code generator does not yet support large aggregate types
1546to be used as function return types. The specific limit on how large an
1547aggregate return type the code generator can currently handle is
1548target-dependent, and also dependent on the aggregate element types.</p>
1549
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1550</div>
1551
1552<!-- _______________________________________________________________________ -->
1553<div class="doc_subsubsection"> <a name="t_pstruct">Packed Structure Type</a>
1554</div>
1555<div class="doc_text">
1556<h5>Overview:</h5>
1557<p>The packed structure type is used to represent a collection of data members
1558together in memory. There is no padding between fields. Further, the alignment
1559of a packed structure is 1 byte. The elements of a packed structure may
1560be any type that has a size.</p>
1561<p>Structures are accessed using '<tt><a href="#i_load">load</a></tt>
1562and '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a
1563field with the '<tt><a href="#i_getelementptr">getelementptr</a></tt>'
1564instruction.</p>
1565<h5>Syntax:</h5>
1566<pre> &lt; { &lt;type list&gt; } &gt; <br></pre>
1567<h5>Examples:</h5>
1568<table class="layout">
1569 <tr class="layout">
1570 <td class="left"><tt>&lt; { i32, i32, i32 } &gt;</tt></td>
1571 <td class="left">A triple of three <tt>i32</tt> values</td>
1572 </tr><tr class="layout">
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]1573 <td class="left">
1574<tt>&lt;&nbsp;{&nbsp;float,&nbsp;i32&nbsp;(i32)*&nbsp;}&nbsp;&gt;</tt></td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1575 <td class="left">A pair, where the first element is a <tt>float</tt> and the
1576 second element is a <a href="#t_pointer">pointer</a> to a
1577 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
1578 an <tt>i32</tt>.</td>
1579 </tr>
1580</table>
1581</div>
1582
1583<!-- _______________________________________________________________________ -->
1584<div class="doc_subsubsection"> <a name="t_pointer">Pointer Type</a> </div>
1585<div class="doc_text">
1586<h5>Overview:</h5>
1587<p>As in many languages, the pointer type represents a pointer or
Christopher Lambdd0049d2007-12-11 09:31:00 +0000[diff] [blame]1588reference to another object, which must live in memory. Pointer types may have
1589an optional address space attribute defining the target-specific numbered
1590address space where the pointed-to object resides. The default address space is
1591zero.</p>
Chris Lattner96edbd32009-02-08 19:53:29 +0000[diff] [blame]1592
1593<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]1594it permit pointers to labels (<tt>label*</tt>). Use <tt>i8*</tt> instead.</p>
Chris Lattner96edbd32009-02-08 19:53:29 +0000[diff] [blame]1595
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1596<h5>Syntax:</h5>
1597<pre> &lt;type&gt; *<br></pre>
1598<h5>Examples:</h5>
1599<table class="layout">
1600 <tr class="layout">
Dan Gohman01852382009-01-04 23:44:43 +0000[diff] [blame]1601 <td class="left"><tt>[4 x i32]*</tt></td>
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1602 <td class="left">A <a href="#t_pointer">pointer</a> to <a
1603 href="#t_array">array</a> of four <tt>i32</tt> values.</td>
1604 </tr>
1605 <tr class="layout">
1606 <td class="left"><tt>i32 (i32 *) *</tt></td>
1607 <td class="left"> A <a href="#t_pointer">pointer</a> to a <a
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1608 href="#t_function">function</a> that takes an <tt>i32*</tt>, returning an
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1609 <tt>i32</tt>.</td>
1610 </tr>
1611 <tr class="layout">
1612 <td class="left"><tt>i32 addrspace(5)*</tt></td>
1613 <td class="left">A <a href="#t_pointer">pointer</a> to an <tt>i32</tt> value
1614 that resides in address space #5.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1615 </tr>
1616</table>
1617</div>
1618
1619<!-- _______________________________________________________________________ -->
1620<div class="doc_subsubsection"> <a name="t_vector">Vector Type</a> </div>
1621<div class="doc_text">
1622
1623<h5>Overview:</h5>
1624
1625<p>A vector type is a simple derived type that represents a vector
1626of elements. Vector types are used when multiple primitive data
1627are operated in parallel using a single instruction (SIMD).
1628A vector type requires a size (number of
1629elements) and an underlying primitive data type. Vectors must have a power
1630of two length (1, 2, 4, 8, 16 ...). Vector types are
1631considered <a href="#t_firstclass">first class</a>.</p>
1632
1633<h5>Syntax:</h5>
1634
1635<pre>
1636 &lt; &lt;# elements&gt; x &lt;elementtype&gt; &gt;
1637</pre>
1638
1639<p>The number of elements is a constant integer value; elementtype may
1640be any integer or floating point type.</p>
1641
1642<h5>Examples:</h5>
1643
1644<table class="layout">
1645 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1646 <td class="left"><tt>&lt;4 x i32&gt;</tt></td>
1647 <td class="left">Vector of 4 32-bit integer values.</td>
1648 </tr>
1649 <tr class="layout">
1650 <td class="left"><tt>&lt;8 x float&gt;</tt></td>
1651 <td class="left">Vector of 8 32-bit floating-point values.</td>
1652 </tr>
1653 <tr class="layout">
1654 <td class="left"><tt>&lt;2 x i64&gt;</tt></td>
1655 <td class="left">Vector of 2 64-bit integer values.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1656 </tr>
1657</table>
djge93155c2009-01-24 15:58:40 +0000[diff] [blame]1658
1659<p>Note that the code generator does not yet support large vector types
1660to be used as function return types. The specific limit on how large a
1661vector return type codegen can currently handle is target-dependent;
1662currently it's often a few times longer than a hardware vector register.</p>
1663
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1664</div>
1665
1666<!-- _______________________________________________________________________ -->
1667<div class="doc_subsubsection"> <a name="t_opaque">Opaque Type</a> </div>
1668<div class="doc_text">
1669
1670<h5>Overview:</h5>
1671
1672<p>Opaque types are used to represent unknown types in the system. This
Gordon Henriksenda0706e2007-10-14 00:34:53 +0000[diff] [blame]1673corresponds (for example) to the C notion of a forward declared structure type.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1674In LLVM, opaque types can eventually be resolved to any type (not just a
1675structure type).</p>
1676
1677<h5>Syntax:</h5>
1678
1679<pre>
1680 opaque
1681</pre>
1682
1683<h5>Examples:</h5>
1684
1685<table class="layout">
1686 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1687 <td class="left"><tt>opaque</tt></td>
1688 <td class="left">An opaque type.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1689 </tr>
1690</table>
1691</div>
1692
Chris Lattner515195a2009-02-02 07:32:36 +0000[diff] [blame]1693<!-- ======================================================================= -->
1694<div class="doc_subsection">
1695 <a name="t_uprefs">Type Up-references</a>
1696</div>
1697
1698<div class="doc_text">
1699<h5>Overview:</h5>
1700<p>
1701An "up reference" allows you to refer to a lexically enclosing type without
1702requiring it to have a name. For instance, a structure declaration may contain a
1703pointer to any of the types it is lexically a member of. Example of up
1704references (with their equivalent as named type declarations) include:</p>
1705
1706<pre>
Chris Lattner5ad632d2009-02-09 10:00:56 +0000[diff] [blame]1707 { \2 * } %x = type { %x* }
Chris Lattner515195a2009-02-02 07:32:36 +0000[diff] [blame]1708 { \2 }* %y = type { %y }*
1709 \1* %z = type %z*
1710</pre>
1711
1712<p>
1713An up reference is needed by the asmprinter for printing out cyclic types when
1714there is no declared name for a type in the cycle. Because the asmprinter does
1715not want to print out an infinite type string, it needs a syntax to handle
1716recursive types that have no names (all names are optional in llvm IR).
1717</p>
1718
1719<h5>Syntax:</h5>
1720<pre>
1721 \&lt;level&gt;
1722</pre>
1723
1724<p>
1725The level is the count of the lexical type that is being referred to.
1726</p>
1727
1728<h5>Examples:</h5>
1729
1730<table class="layout">
1731 <tr class="layout">
1732 <td class="left"><tt>\1*</tt></td>
1733 <td class="left">Self-referential pointer.</td>
1734 </tr>
1735 <tr class="layout">
1736 <td class="left"><tt>{ { \3*, i8 }, i32 }</tt></td>
1737 <td class="left">Recursive structure where the upref refers to the out-most
1738 structure.</td>
1739 </tr>
1740</table>
1741</div>
1742
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1743
1744<!-- *********************************************************************** -->
1745<div class="doc_section"> <a name="constants">Constants</a> </div>
1746<!-- *********************************************************************** -->
1747
1748<div class="doc_text">
1749
1750<p>LLVM has several different basic types of constants. This section describes
1751them all and their syntax.</p>
1752
1753</div>
1754
1755<!-- ======================================================================= -->
1756<div class="doc_subsection"><a name="simpleconstants">Simple Constants</a></div>
1757
1758<div class="doc_text">
1759
1760<dl>
1761 <dt><b>Boolean constants</b></dt>
1762
1763 <dd>The two strings '<tt>true</tt>' and '<tt>false</tt>' are both valid
1764 constants of the <tt><a href="#t_primitive">i1</a></tt> type.
1765 </dd>
1766
1767 <dt><b>Integer constants</b></dt>
1768
1769 <dd>Standard integers (such as '4') are constants of the <a
1770 href="#t_integer">integer</a> type. Negative numbers may be used with
1771 integer types.
1772 </dd>
1773
1774 <dt><b>Floating point constants</b></dt>
1775
1776 <dd>Floating point constants use standard decimal notation (e.g. 123.421),
1777 exponential notation (e.g. 1.23421e+2), or a more precise hexadecimal
Chris Lattnerdc15b1d2008-04-01 18:45:27 +0000[diff] [blame]1778 notation (see below). The assembler requires the exact decimal value of
1779 a floating-point constant. For example, the assembler accepts 1.25 but
1780 rejects 1.3 because 1.3 is a repeating decimal in binary. Floating point
1781 constants must have a <a href="#t_floating">floating point</a> type. </dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1782
1783 <dt><b>Null pointer constants</b></dt>
1784
1785 <dd>The identifier '<tt>null</tt>' is recognized as a null pointer constant
1786 and must be of <a href="#t_pointer">pointer type</a>.</dd>
1787
1788</dl>
1789
Dale Johannesenf82a52f2009-02-11 22:14:51 +0000[diff] [blame]1790<p>The one non-intuitive notation for constants is the hexadecimal form
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1791of floating point constants. For example, the form '<tt>double
17920x432ff973cafa8000</tt>' is equivalent to (but harder to read than) '<tt>double
17934.5e+15</tt>'. The only time hexadecimal floating point constants are required
1794(and the only time that they are generated by the disassembler) is when a
1795floating point constant must be emitted but it cannot be represented as a
Dale Johannesenf82a52f2009-02-11 22:14:51 +0000[diff] [blame]1796decimal floating point number in a reasonable number of digits. For example,
1797NaN's, infinities, and other
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1798special values are represented in their IEEE hexadecimal format so that
1799assembly and disassembly do not cause any bits to change in the constants.</p>
Dale Johannesenf82a52f2009-02-11 22:14:51 +0000[diff] [blame]1800<p>When using the hexadecimal form, constants of types float and double are
1801represented using the 16-digit form shown above (which matches the IEEE754
1802representation for double); float values must, however, be exactly representable
1803as IEE754 single precision.
1804Hexadecimal format is always used for long
1805double, and there are three forms of long double. The 80-bit
1806format used by x86 is represented as <tt>0xK</tt>
1807followed by 20 hexadecimal digits.
1808The 128-bit format used by PowerPC (two adjacent doubles) is represented
1809by <tt>0xM</tt> followed by 32 hexadecimal digits. The IEEE 128-bit
1810format is represented
1811by <tt>0xL</tt> followed by 32 hexadecimal digits; no currently supported
1812target uses this format. Long doubles will only work if they match
1813the long double format on your target. All hexadecimal formats are big-endian
1814(sign bit at the left).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1815</div>
1816
1817<!-- ======================================================================= -->
Chris Lattner97063852009-02-28 18:32:25 +0000[diff] [blame]1818<div class="doc_subsection">
1819<a name="aggregateconstants"> <!-- old anchor -->
1820<a name="complexconstants">Complex Constants</a></a>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1821</div>
1822
1823<div class="doc_text">
Chris Lattner97063852009-02-28 18:32:25 +0000[diff] [blame]1824<p>Complex constants are a (potentially recursive) combination of simple
1825constants and smaller complex constants.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1826
1827<dl>
1828 <dt><b>Structure constants</b></dt>
1829
1830 <dd>Structure constants are represented with notation similar to structure
1831 type definitions (a comma separated list of elements, surrounded by braces
Chris Lattner6c8de962007-12-25 20:34:52 +0000[diff] [blame]1832 (<tt>{}</tt>)). For example: "<tt>{ i32 4, float 17.0, i32* @G }</tt>",
1833 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]1834 must have <a href="#t_struct">structure type</a>, and the number and
1835 types of elements must match those specified by the type.
1836 </dd>
1837
1838 <dt><b>Array constants</b></dt>
1839
1840 <dd>Array constants are represented with notation similar to array type
1841 definitions (a comma separated list of elements, surrounded by square brackets
1842 (<tt>[]</tt>)). For example: "<tt>[ i32 42, i32 11, i32 74 ]</tt>". Array
1843 constants must have <a href="#t_array">array type</a>, and the number and
1844 types of elements must match those specified by the type.
1845 </dd>
1846
1847 <dt><b>Vector constants</b></dt>
1848
1849 <dd>Vector constants are represented with notation similar to vector type
1850 definitions (a comma separated list of elements, surrounded by
1851 less-than/greater-than's (<tt>&lt;&gt;</tt>)). For example: "<tt>&lt; i32 42,
1852 i32 11, i32 74, i32 100 &gt;</tt>". Vector constants must have <a
1853 href="#t_vector">vector type</a>, and the number and types of elements must
1854 match those specified by the type.
1855 </dd>
1856
1857 <dt><b>Zero initialization</b></dt>
1858
1859 <dd>The string '<tt>zeroinitializer</tt>' can be used to zero initialize a
1860 value to zero of <em>any</em> type, including scalar and aggregate types.
1861 This is often used to avoid having to print large zero initializers (e.g. for
1862 large arrays) and is always exactly equivalent to using explicit zero
1863 initializers.
1864 </dd>
Nick Lewycky4dcf8102009-04-04 07:22:01 +0000[diff] [blame]1865
1866 <dt><b>Metadata node</b></dt>
1867
1868 <dd>A metadata node is a structure-like constant with the type of an empty
1869 struct. For example: "<tt>{ } !{ i32 0, { } !"test" }</tt>". Unlike other
1870 constants that are meant to be interpreted as part of the instruction stream,
1871 metadata is a place to attach additional information such as debug info.
1872 </dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1873</dl>
1874
1875</div>
1876
1877<!-- ======================================================================= -->
1878<div class="doc_subsection">
1879 <a name="globalconstants">Global Variable and Function Addresses</a>
1880</div>
1881
1882<div class="doc_text">
1883
1884<p>The addresses of <a href="#globalvars">global variables</a> and <a
1885href="#functionstructure">functions</a> are always implicitly valid (link-time)
1886constants. These constants are explicitly referenced when the <a
1887href="#identifiers">identifier for the global</a> is used and always have <a
1888href="#t_pointer">pointer</a> type. For example, the following is a legal LLVM
1889file:</p>
1890
1891<div class="doc_code">
1892<pre>
1893@X = global i32 17
1894@Y = global i32 42
1895@Z = global [2 x i32*] [ i32* @X, i32* @Y ]
1896</pre>
1897</div>
1898
1899</div>
1900
1901<!-- ======================================================================= -->
1902<div class="doc_subsection"><a name="undefvalues">Undefined Values</a></div>
1903<div class="doc_text">
1904 <p>The string '<tt>undef</tt>' is recognized as a type-less constant that has
1905 no specific value. Undefined values may be of any type and be used anywhere
1906 a constant is permitted.</p>
1907
1908 <p>Undefined values indicate to the compiler that the program is well defined
1909 no matter what value is used, giving the compiler more freedom to optimize.
1910 </p>
1911</div>
1912
1913<!-- ======================================================================= -->
1914<div class="doc_subsection"><a name="constantexprs">Constant Expressions</a>
1915</div>
1916
1917<div class="doc_text">
1918
1919<p>Constant expressions are used to allow expressions involving other constants
1920to be used as constants. Constant expressions may be of any <a
1921href="#t_firstclass">first class</a> type and may involve any LLVM operation
1922that does not have side effects (e.g. load and call are not supported). The
1923following is the syntax for constant expressions:</p>
1924
1925<dl>
1926 <dt><b><tt>trunc ( CST to TYPE )</tt></b></dt>
1927 <dd>Truncate a constant to another type. The bit size of CST must be larger
1928 than the bit size of TYPE. Both types must be integers.</dd>
1929
1930 <dt><b><tt>zext ( CST to TYPE )</tt></b></dt>
1931 <dd>Zero extend a constant to another type. The bit size of CST must be
1932 smaller or equal to the bit size of TYPE. Both types must be integers.</dd>
1933
1934 <dt><b><tt>sext ( CST to TYPE )</tt></b></dt>
1935 <dd>Sign extend a constant to another type. The bit size of CST must be
1936 smaller or equal to the bit size of TYPE. Both types must be integers.</dd>
1937
1938 <dt><b><tt>fptrunc ( CST to TYPE )</tt></b></dt>
1939 <dd>Truncate a floating point constant to another floating point type. The
1940 size of CST must be larger than the size of TYPE. Both types must be
1941 floating point.</dd>
1942
1943 <dt><b><tt>fpext ( CST to TYPE )</tt></b></dt>
1944 <dd>Floating point extend a constant to another type. The size of CST must be
1945 smaller or equal to the size of TYPE. Both types must be floating point.</dd>
1946
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]1947 <dt><b><tt>fptoui ( CST to TYPE )</tt></b></dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1948 <dd>Convert a floating point constant to the corresponding unsigned integer
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]1949 constant. TYPE must be a scalar or vector integer type. CST must be of scalar
1950 or vector floating point type. Both CST and TYPE must be scalars, or vectors
1951 of the same number of elements. If the value won't fit in the integer type,
1952 the results are undefined.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1953
1954 <dt><b><tt>fptosi ( CST to TYPE )</tt></b></dt>
1955 <dd>Convert a floating point constant to the corresponding signed integer
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]1956 constant. TYPE must be a scalar or vector integer type. CST must be of scalar
1957 or vector floating point type. Both CST and TYPE must be scalars, or vectors
1958 of the same number of elements. If the value won't fit in the integer type,
1959 the results are undefined.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1960
1961 <dt><b><tt>uitofp ( CST to TYPE )</tt></b></dt>
1962 <dd>Convert an unsigned integer constant to the corresponding floating point
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]1963 constant. TYPE must be a scalar or vector floating point type. CST must be of
1964 scalar or vector integer type. Both CST and TYPE must be scalars, or vectors
1965 of the same number of elements. If the value won't fit in the floating point
1966 type, the results are undefined.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1967
1968 <dt><b><tt>sitofp ( CST to TYPE )</tt></b></dt>
1969 <dd>Convert a signed integer constant to the corresponding floating point
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]1970 constant. TYPE must be a scalar or vector floating point type. CST must be of
1971 scalar or vector integer type. Both CST and TYPE must be scalars, or vectors
1972 of the same number of elements. If the value won't fit in the floating point
1973 type, the results are undefined.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1974
1975 <dt><b><tt>ptrtoint ( CST to TYPE )</tt></b></dt>
1976 <dd>Convert a pointer typed constant to the corresponding integer constant
1977 TYPE must be an integer type. CST must be of pointer type. The CST value is
1978 zero extended, truncated, or unchanged to make it fit in TYPE.</dd>
1979
1980 <dt><b><tt>inttoptr ( CST to TYPE )</tt></b></dt>
1981 <dd>Convert a integer constant to a pointer constant. TYPE must be a
1982 pointer type. CST must be of integer type. The CST value is zero extended,
1983 truncated, or unchanged to make it fit in a pointer size. This one is
1984 <i>really</i> dangerous!</dd>
1985
1986 <dt><b><tt>bitcast ( CST to TYPE )</tt></b></dt>
Chris Lattner557bc5d2009-02-28 18:27:03 +0000[diff] [blame]1987 <dd>Convert a constant, CST, to another TYPE. The constraints of the operands
1988 are the same as those for the <a href="#i_bitcast">bitcast
1989 instruction</a>.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1990
1991 <dt><b><tt>getelementptr ( CSTPTR, IDX0, IDX1, ... )</tt></b></dt>
1992
1993 <dd>Perform the <a href="#i_getelementptr">getelementptr operation</a> on
1994 constants. As with the <a href="#i_getelementptr">getelementptr</a>
1995 instruction, the index list may have zero or more indexes, which are required
1996 to make sense for the type of "CSTPTR".</dd>
1997
1998 <dt><b><tt>select ( COND, VAL1, VAL2 )</tt></b></dt>
1999
2000 <dd>Perform the <a href="#i_select">select operation</a> on
2001 constants.</dd>
2002
2003 <dt><b><tt>icmp COND ( VAL1, VAL2 )</tt></b></dt>
2004 <dd>Performs the <a href="#i_icmp">icmp operation</a> on constants.</dd>
2005
2006 <dt><b><tt>fcmp COND ( VAL1, VAL2 )</tt></b></dt>
2007 <dd>Performs the <a href="#i_fcmp">fcmp operation</a> on constants.</dd>
2008
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]2009 <dt><b><tt>vicmp COND ( VAL1, VAL2 )</tt></b></dt>
2010 <dd>Performs the <a href="#i_vicmp">vicmp operation</a> on constants.</dd>
2011
2012 <dt><b><tt>vfcmp COND ( VAL1, VAL2 )</tt></b></dt>
2013 <dd>Performs the <a href="#i_vfcmp">vfcmp operation</a> on constants.</dd>
2014
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2015 <dt><b><tt>extractelement ( VAL, IDX )</tt></b></dt>
2016
2017 <dd>Perform the <a href="#i_extractelement">extractelement
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]2018 operation</a> on constants.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2019
2020 <dt><b><tt>insertelement ( VAL, ELT, IDX )</tt></b></dt>
2021
2022 <dd>Perform the <a href="#i_insertelement">insertelement
2023 operation</a> on constants.</dd>
2024
2025
2026 <dt><b><tt>shufflevector ( VEC1, VEC2, IDXMASK )</tt></b></dt>
2027
2028 <dd>Perform the <a href="#i_shufflevector">shufflevector
2029 operation</a> on constants.</dd>
2030
2031 <dt><b><tt>OPCODE ( LHS, RHS )</tt></b></dt>
2032
2033 <dd>Perform the specified operation of the LHS and RHS constants. OPCODE may
2034 be any of the <a href="#binaryops">binary</a> or <a href="#bitwiseops">bitwise
2035 binary</a> operations. The constraints on operands are the same as those for
2036 the corresponding instruction (e.g. no bitwise operations on floating point
2037 values are allowed).</dd>
2038</dl>
2039</div>
2040
Nick Lewycky4dcf8102009-04-04 07:22:01 +0000[diff] [blame]2041<!-- ======================================================================= -->
2042<div class="doc_subsection"><a name="metadata">Embedded Metadata</a>
2043</div>
2044
2045<div class="doc_text">
2046
2047<p>Embedded metadata provides a way to attach arbitrary data to the
2048instruction stream without affecting the behaviour of the program. There are
2049two metadata primitives, strings and nodes. All metadata has the type of an
2050empty struct and is identified in syntax by a preceding exclamation point
2051('<tt>!</tt>').
2052</p>
2053
2054<p>A metadata string is a string surrounded by double quotes. It can contain
2055any character by escaping non-printable characters with "\xx" where "xx" is
2056the two digit hex code. For example: "<tt>!"test\00"</tt>".
2057</p>
2058
2059<p>Metadata nodes are represented with notation similar to structure constants
2060(a comma separated list of elements, surrounded by braces and preceeded by an
2061exclamation point). For example: "<tt>!{ { } !"test\00", i32 10}</tt>".
2062</p>
2063
2064<p>Optimizations may rely on metadata to provide additional information about
2065the program that isn't available in the instructions, or that isn't easily
2066computable. Similarly, the code generator may expect a certain metadata format
2067to be used to express debugging information.</p>
2068</div>
2069
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2070<!-- *********************************************************************** -->
2071<div class="doc_section"> <a name="othervalues">Other Values</a> </div>
2072<!-- *********************************************************************** -->
2073
2074<!-- ======================================================================= -->
2075<div class="doc_subsection">
2076<a name="inlineasm">Inline Assembler Expressions</a>
2077</div>
2078
2079<div class="doc_text">
2080
2081<p>
2082LLVM supports inline assembler expressions (as opposed to <a href="#moduleasm">
2083Module-Level Inline Assembly</a>) through the use of a special value. This
2084value represents the inline assembler as a string (containing the instructions
2085to emit), a list of operand constraints (stored as a string), and a flag that
2086indicates whether or not the inline asm expression has side effects. An example
2087inline assembler expression is:
2088</p>
2089
2090<div class="doc_code">
2091<pre>
2092i32 (i32) asm "bswap $0", "=r,r"
2093</pre>
2094</div>
2095
2096<p>
2097Inline assembler expressions may <b>only</b> be used as the callee operand of
2098a <a href="#i_call"><tt>call</tt> instruction</a>. Thus, typically we have:
2099</p>
2100
2101<div class="doc_code">
2102<pre>
2103%X = call i32 asm "<a href="#int_bswap">bswap</a> $0", "=r,r"(i32 %Y)
2104</pre>
2105</div>
2106
2107<p>
2108Inline asms with side effects not visible in the constraint list must be marked
2109as having side effects. This is done through the use of the
2110'<tt>sideeffect</tt>' keyword, like so:
2111</p>
2112
2113<div class="doc_code">
2114<pre>
2115call void asm sideeffect "eieio", ""()
2116</pre>
2117</div>
2118
2119<p>TODO: The format of the asm and constraints string still need to be
2120documented here. Constraints on what can be done (e.g. duplication, moving, etc
Chris Lattner1cdd64e2008-10-04 18:36:02 +0000[diff] [blame]2121need to be documented). This is probably best done by reference to another
2122document that covers inline asm from a holistic perspective.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2123</p>
2124
2125</div>
2126
2127<!-- *********************************************************************** -->
2128<div class="doc_section"> <a name="instref">Instruction Reference</a> </div>
2129<!-- *********************************************************************** -->
2130
2131<div class="doc_text">
2132
2133<p>The LLVM instruction set consists of several different
2134classifications of instructions: <a href="#terminators">terminator
2135instructions</a>, <a href="#binaryops">binary instructions</a>,
2136<a href="#bitwiseops">bitwise binary instructions</a>, <a
2137 href="#memoryops">memory instructions</a>, and <a href="#otherops">other
2138instructions</a>.</p>
2139
2140</div>
2141
2142<!-- ======================================================================= -->
2143<div class="doc_subsection"> <a name="terminators">Terminator
2144Instructions</a> </div>
2145
2146<div class="doc_text">
2147
2148<p>As mentioned <a href="#functionstructure">previously</a>, every
2149basic block in a program ends with a "Terminator" instruction, which
2150indicates which block should be executed after the current block is
2151finished. These terminator instructions typically yield a '<tt>void</tt>'
2152value: they produce control flow, not values (the one exception being
2153the '<a href="#i_invoke"><tt>invoke</tt></a>' instruction).</p>
2154<p>There are six different terminator instructions: the '<a
2155 href="#i_ret"><tt>ret</tt></a>' instruction, the '<a href="#i_br"><tt>br</tt></a>'
2156instruction, the '<a href="#i_switch"><tt>switch</tt></a>' instruction,
2157the '<a href="#i_invoke"><tt>invoke</tt></a>' instruction, the '<a
2158 href="#i_unwind"><tt>unwind</tt></a>' instruction, and the '<a
2159 href="#i_unreachable"><tt>unreachable</tt></a>' instruction.</p>
2160
2161</div>
2162
2163<!-- _______________________________________________________________________ -->
2164<div class="doc_subsubsection"> <a name="i_ret">'<tt>ret</tt>'
2165Instruction</a> </div>
2166<div class="doc_text">
2167<h5>Syntax:</h5>
Dan Gohman3e700032008-10-04 19:00:07 +0000[diff] [blame]2168<pre>
2169 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]2170 ret void <i>; Return from void function</i>
2171</pre>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]2172
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2173<h5>Overview:</h5>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]2174
Dan Gohman3e700032008-10-04 19:00:07 +0000[diff] [blame]2175<p>The '<tt>ret</tt>' instruction is used to return control flow (and
2176optionally a value) from a function back to the caller.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2177<p>There are two forms of the '<tt>ret</tt>' instruction: one that
Dan Gohman3e700032008-10-04 19:00:07 +0000[diff] [blame]2178returns a value and then causes control flow, and one that just causes
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2179control flow to occur.</p>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]2180
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2181<h5>Arguments:</h5>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]2182
Dan Gohman3e700032008-10-04 19:00:07 +0000[diff] [blame]2183<p>The '<tt>ret</tt>' instruction optionally accepts a single argument,
2184the return value. The type of the return value must be a
2185'<a href="#t_firstclass">first class</a>' type.</p>
2186
2187<p>A function is not <a href="#wellformed">well formed</a> if
2188it it has a non-void return type and contains a '<tt>ret</tt>'
2189instruction with no return value or a return value with a type that
2190does not match its type, or if it has a void return type and contains
2191a '<tt>ret</tt>' instruction with a return value.</p>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]2192
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2193<h5>Semantics:</h5>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]2194
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2195<p>When the '<tt>ret</tt>' instruction is executed, control flow
2196returns back to the calling function's context. If the caller is a "<a
2197 href="#i_call"><tt>call</tt></a>" instruction, execution continues at
2198the instruction after the call. If the caller was an "<a
2199 href="#i_invoke"><tt>invoke</tt></a>" instruction, execution continues
2200at the beginning of the "normal" destination block. If the instruction
2201returns a value, that value shall set the call or invoke instruction's
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]2202return value.</p>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]2203
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2204<h5>Example:</h5>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]2205
2206<pre>
2207 ret i32 5 <i>; Return an integer value of 5</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2208 ret void <i>; Return from a void function</i>
Bill Wendlingd163e2d2009-02-28 22:12:54 +0000[diff] [blame]2209 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]2210</pre>
Dan Gohman60967192009-01-12 23:12:39 +0000[diff] [blame]2211
djge93155c2009-01-24 15:58:40 +0000[diff] [blame]2212<p>Note that the code generator does not yet fully support large
2213 return values. The specific sizes that are currently supported are
2214 dependent on the target. For integers, on 32-bit targets the limit
2215 is often 64 bits, and on 64-bit targets the limit is often 128 bits.
2216 For aggregate types, the current limits are dependent on the element
2217 types; for example targets are often limited to 2 total integer
2218 elements and 2 total floating-point elements.</p>
Dan Gohman60967192009-01-12 23:12:39 +0000[diff] [blame]2219
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2220</div>
2221<!-- _______________________________________________________________________ -->
2222<div class="doc_subsubsection"> <a name="i_br">'<tt>br</tt>' Instruction</a> </div>
2223<div class="doc_text">
2224<h5>Syntax:</h5>
2225<pre> br i1 &lt;cond&gt;, label &lt;iftrue&gt;, label &lt;iffalse&gt;<br> br label &lt;dest&gt; <i>; Unconditional branch</i>
2226</pre>
2227<h5>Overview:</h5>
2228<p>The '<tt>br</tt>' instruction is used to cause control flow to
2229transfer to a different basic block in the current function. There are
2230two forms of this instruction, corresponding to a conditional branch
2231and an unconditional branch.</p>
2232<h5>Arguments:</h5>
2233<p>The conditional branch form of the '<tt>br</tt>' instruction takes a
2234single '<tt>i1</tt>' value and two '<tt>label</tt>' values. The
2235unconditional form of the '<tt>br</tt>' instruction takes a single
2236'<tt>label</tt>' value as a target.</p>
2237<h5>Semantics:</h5>
2238<p>Upon execution of a conditional '<tt>br</tt>' instruction, the '<tt>i1</tt>'
2239argument is evaluated. If the value is <tt>true</tt>, control flows
2240to the '<tt>iftrue</tt>' <tt>label</tt> argument. If "cond" is <tt>false</tt>,
2241control flows to the '<tt>iffalse</tt>' <tt>label</tt> argument.</p>
2242<h5>Example:</h5>
2243<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
2244 href="#i_ret">ret</a> i32 1<br>IfUnequal:<br> <a href="#i_ret">ret</a> i32 0<br></pre>
2245</div>
2246<!-- _______________________________________________________________________ -->
2247<div class="doc_subsubsection">
2248 <a name="i_switch">'<tt>switch</tt>' Instruction</a>
2249</div>
2250
2251<div class="doc_text">
2252<h5>Syntax:</h5>
2253
2254<pre>
2255 switch &lt;intty&gt; &lt;value&gt;, label &lt;defaultdest&gt; [ &lt;intty&gt; &lt;val&gt;, label &lt;dest&gt; ... ]
2256</pre>
2257
2258<h5>Overview:</h5>
2259
2260<p>The '<tt>switch</tt>' instruction is used to transfer control flow to one of
2261several different places. It is a generalization of the '<tt>br</tt>'
2262instruction, allowing a branch to occur to one of many possible
2263destinations.</p>
2264
2265
2266<h5>Arguments:</h5>
2267
2268<p>The '<tt>switch</tt>' instruction uses three parameters: an integer
2269comparison value '<tt>value</tt>', a default '<tt>label</tt>' destination, and
2270an array of pairs of comparison value constants and '<tt>label</tt>'s. The
2271table is not allowed to contain duplicate constant entries.</p>
2272
2273<h5>Semantics:</h5>
2274
2275<p>The <tt>switch</tt> instruction specifies a table of values and
2276destinations. When the '<tt>switch</tt>' instruction is executed, this
2277table is searched for the given value. If the value is found, control flow is
2278transfered to the corresponding destination; otherwise, control flow is
2279transfered to the default destination.</p>
2280
2281<h5>Implementation:</h5>
2282
2283<p>Depending on properties of the target machine and the particular
2284<tt>switch</tt> instruction, this instruction may be code generated in different
2285ways. For example, it could be generated as a series of chained conditional
2286branches or with a lookup table.</p>
2287
2288<h5>Example:</h5>
2289
2290<pre>
2291 <i>; Emulate a conditional br instruction</i>
2292 %Val = <a href="#i_zext">zext</a> i1 %value to i32
Dan Gohman01852382009-01-04 23:44:43 +0000[diff] [blame]2293 switch i32 %Val, label %truedest [ i32 0, label %falsedest ]
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2294
2295 <i>; Emulate an unconditional br instruction</i>
2296 switch i32 0, label %dest [ ]
2297
2298 <i>; Implement a jump table:</i>
Dan Gohman01852382009-01-04 23:44:43 +0000[diff] [blame]2299 switch i32 %val, label %otherwise [ i32 0, label %onzero
2300 i32 1, label %onone
2301 i32 2, label %ontwo ]
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2302</pre>
2303</div>
2304
2305<!-- _______________________________________________________________________ -->
2306<div class="doc_subsubsection">
2307 <a name="i_invoke">'<tt>invoke</tt>' Instruction</a>
2308</div>
2309
2310<div class="doc_text">
2311
2312<h5>Syntax:</h5>
2313
2314<pre>
Devang Pateld0bfcc72008-10-07 17:48:33 +0000[diff] [blame]2315 &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]2316 to label &lt;normal label&gt; unwind label &lt;exception label&gt;
2317</pre>
2318
2319<h5>Overview:</h5>
2320
2321<p>The '<tt>invoke</tt>' instruction causes control to transfer to a specified
2322function, with the possibility of control flow transfer to either the
2323'<tt>normal</tt>' label or the
2324'<tt>exception</tt>' label. If the callee function returns with the
2325"<tt><a href="#i_ret">ret</a></tt>" instruction, control flow will return to the
2326"normal" label. If the callee (or any indirect callees) returns with the "<a
2327href="#i_unwind"><tt>unwind</tt></a>" instruction, control is interrupted and
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]2328continued at the dynamically nearest "exception" label.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2329
2330<h5>Arguments:</h5>
2331
2332<p>This instruction requires several arguments:</p>
2333
2334<ol>
2335 <li>
2336 The optional "cconv" marker indicates which <a href="#callingconv">calling
2337 convention</a> the call should use. If none is specified, the call defaults
2338 to using C calling conventions.
2339 </li>
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]2340
2341 <li>The optional <a href="#paramattrs">Parameter Attributes</a> list for
2342 return values. Only '<tt>zeroext</tt>', '<tt>signext</tt>',
2343 and '<tt>inreg</tt>' attributes are valid here.</li>
2344
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2345 <li>'<tt>ptr to function ty</tt>': shall be the signature of the pointer to
2346 function value being invoked. In most cases, this is a direct function
2347 invocation, but indirect <tt>invoke</tt>s are just as possible, branching off
2348 an arbitrary pointer to function value.
2349 </li>
2350
2351 <li>'<tt>function ptr val</tt>': An LLVM value containing a pointer to a
2352 function to be invoked. </li>
2353
2354 <li>'<tt>function args</tt>': argument list whose types match the function
2355 signature argument types. If the function signature indicates the function
2356 accepts a variable number of arguments, the extra arguments can be
2357 specified. </li>
2358
2359 <li>'<tt>normal label</tt>': the label reached when the called function
2360 executes a '<tt><a href="#i_ret">ret</a></tt>' instruction. </li>
2361
2362 <li>'<tt>exception label</tt>': the label reached when a callee returns with
2363 the <a href="#i_unwind"><tt>unwind</tt></a> instruction. </li>
2364
Devang Pateld0bfcc72008-10-07 17:48:33 +0000[diff] [blame]2365 <li>The optional <a href="#fnattrs">function attributes</a> list. Only
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]2366 '<tt>noreturn</tt>', '<tt>nounwind</tt>', '<tt>readonly</tt>' and
2367 '<tt>readnone</tt>' attributes are valid here.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2368</ol>
2369
2370<h5>Semantics:</h5>
2371
2372<p>This instruction is designed to operate as a standard '<tt><a
2373href="#i_call">call</a></tt>' instruction in most regards. The primary
2374difference is that it establishes an association with a label, which is used by
2375the runtime library to unwind the stack.</p>
2376
2377<p>This instruction is used in languages with destructors to ensure that proper
2378cleanup is performed in the case of either a <tt>longjmp</tt> or a thrown
2379exception. Additionally, this is important for implementation of
2380'<tt>catch</tt>' clauses in high-level languages that support them.</p>
2381
2382<h5>Example:</h5>
2383<pre>
Nick Lewyckya1c11a12008-03-16 07:18:12 +0000[diff] [blame]2384 %retval = invoke i32 @Test(i32 15) to label %Continue
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2385 unwind label %TestCleanup <i>; {i32}:retval set</i>
Nick Lewyckya1c11a12008-03-16 07:18:12 +0000[diff] [blame]2386 %retval = invoke <a href="#callingconv">coldcc</a> i32 %Testfnptr(i32 15) to label %Continue
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2387 unwind label %TestCleanup <i>; {i32}:retval set</i>
2388</pre>
2389</div>
2390
2391
2392<!-- _______________________________________________________________________ -->
2393
2394<div class="doc_subsubsection"> <a name="i_unwind">'<tt>unwind</tt>'
2395Instruction</a> </div>
2396
2397<div class="doc_text">
2398
2399<h5>Syntax:</h5>
2400<pre>
2401 unwind
2402</pre>
2403
2404<h5>Overview:</h5>
2405
2406<p>The '<tt>unwind</tt>' instruction unwinds the stack, continuing control flow
2407at the first callee in the dynamic call stack which used an <a
2408href="#i_invoke"><tt>invoke</tt></a> instruction to perform the call. This is
2409primarily used to implement exception handling.</p>
2410
2411<h5>Semantics:</h5>
2412
Chris Lattner8b094fc2008-04-19 21:01:16 +0000[diff] [blame]2413<p>The '<tt>unwind</tt>' instruction causes execution of the current function to
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2414immediately halt. The dynamic call stack is then searched for the first <a
2415href="#i_invoke"><tt>invoke</tt></a> instruction on the call stack. Once found,
2416execution continues at the "exceptional" destination block specified by the
2417<tt>invoke</tt> instruction. If there is no <tt>invoke</tt> instruction in the
2418dynamic call chain, undefined behavior results.</p>
2419</div>
2420
2421<!-- _______________________________________________________________________ -->
2422
2423<div class="doc_subsubsection"> <a name="i_unreachable">'<tt>unreachable</tt>'
2424Instruction</a> </div>
2425
2426<div class="doc_text">
2427
2428<h5>Syntax:</h5>
2429<pre>
2430 unreachable
2431</pre>
2432
2433<h5>Overview:</h5>
2434
2435<p>The '<tt>unreachable</tt>' instruction has no defined semantics. This
2436instruction is used to inform the optimizer that a particular portion of the
2437code is not reachable. This can be used to indicate that the code after a
2438no-return function cannot be reached, and other facts.</p>
2439
2440<h5>Semantics:</h5>
2441
2442<p>The '<tt>unreachable</tt>' instruction has no defined semantics.</p>
2443</div>
2444
2445
2446
2447<!-- ======================================================================= -->
2448<div class="doc_subsection"> <a name="binaryops">Binary Operations</a> </div>
2449<div class="doc_text">
2450<p>Binary operators are used to do most of the computation in a
Chris Lattnerab596d92008-04-01 18:47:32 +0000[diff] [blame]2451program. They require two operands of the same type, execute an operation on them, and
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2452produce a single value. The operands might represent
2453multiple 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]2454The result value has the same type as its operands.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2455<p>There are several different binary operators:</p>
2456</div>
2457<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2458<div class="doc_subsubsection">
2459 <a name="i_add">'<tt>add</tt>' Instruction</a>
2460</div>
2461
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2462<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2463
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2464<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2465
2466<pre>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2467 &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]2468</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2469
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2470<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2471
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2472<p>The '<tt>add</tt>' instruction returns the sum of its two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2473
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2474<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2475
2476<p>The two arguments to the '<tt>add</tt>' instruction must be <a
2477 href="#t_integer">integer</a>, <a href="#t_floating">floating point</a>, or
2478 <a href="#t_vector">vector</a> values. Both arguments must have identical
2479 types.</p>
2480
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2481<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2482
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2483<p>The value produced is the integer or floating point sum of the two
2484operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2485
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2486<p>If an integer sum has unsigned overflow, the result returned is the
2487mathematical result modulo 2<sup>n</sup>, where n is the bit width of
2488the result.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2489
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2490<p>Because LLVM integers use a two's complement representation, this
2491instruction is appropriate for both signed and unsigned integers.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2492
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2493<h5>Example:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2494
2495<pre>
2496 &lt;result&gt; = add i32 4, %var <i>; yields {i32}:result = 4 + %var</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2497</pre>
2498</div>
2499<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2500<div class="doc_subsubsection">
2501 <a name="i_sub">'<tt>sub</tt>' Instruction</a>
2502</div>
2503
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2504<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2505
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2506<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2507
2508<pre>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2509 &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]2510</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2511
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2512<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2513
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2514<p>The '<tt>sub</tt>' instruction returns the difference of its two
2515operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2516
2517<p>Note that the '<tt>sub</tt>' instruction is used to represent the
2518'<tt>neg</tt>' instruction present in most other intermediate
2519representations.</p>
2520
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2521<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2522
2523<p>The two arguments to the '<tt>sub</tt>' instruction must be <a
2524 href="#t_integer">integer</a>, <a href="#t_floating">floating point</a>,
2525 or <a href="#t_vector">vector</a> values. Both arguments must have identical
2526 types.</p>
2527
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2528<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2529
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2530<p>The value produced is the integer or floating point difference of
2531the two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2532
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2533<p>If an integer difference has unsigned overflow, the result returned is the
2534mathematical result modulo 2<sup>n</sup>, where n is the bit width of
2535the result.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2536
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2537<p>Because LLVM integers use a two's complement representation, this
2538instruction is appropriate for both signed and unsigned integers.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2539
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2540<h5>Example:</h5>
2541<pre>
2542 &lt;result&gt; = sub i32 4, %var <i>; yields {i32}:result = 4 - %var</i>
2543 &lt;result&gt; = sub i32 0, %val <i>; yields {i32}:result = -%var</i>
2544</pre>
2545</div>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2546
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2547<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2548<div class="doc_subsubsection">
2549 <a name="i_mul">'<tt>mul</tt>' Instruction</a>
2550</div>
2551
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2552<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2553
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2554<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2555<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]2556</pre>
2557<h5>Overview:</h5>
2558<p>The '<tt>mul</tt>' instruction returns the product of its two
2559operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2560
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2561<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2562
2563<p>The two arguments to the '<tt>mul</tt>' instruction must be <a
2564href="#t_integer">integer</a>, <a href="#t_floating">floating point</a>,
2565or <a href="#t_vector">vector</a> values. Both arguments must have identical
2566types.</p>
2567
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2568<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2569
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2570<p>The value produced is the integer or floating point product of the
2571two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2572
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2573<p>If the result of an integer multiplication has unsigned overflow,
2574the result returned is the mathematical result modulo
25752<sup>n</sup>, where n is the bit width of the result.</p>
2576<p>Because LLVM integers use a two's complement representation, and the
2577result is the same width as the operands, this instruction returns the
2578correct result for both signed and unsigned integers. If a full product
2579(e.g. <tt>i32</tt>x<tt>i32</tt>-><tt>i64</tt>) is needed, the operands
2580should be sign-extended or zero-extended as appropriate to the
2581width of the full product.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2582<h5>Example:</h5>
2583<pre> &lt;result&gt; = mul i32 4, %var <i>; yields {i32}:result = 4 * %var</i>
2584</pre>
2585</div>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2586
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2587<!-- _______________________________________________________________________ -->
2588<div class="doc_subsubsection"> <a name="i_udiv">'<tt>udiv</tt>' Instruction
2589</a></div>
2590<div class="doc_text">
2591<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2592<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]2593</pre>
2594<h5>Overview:</h5>
2595<p>The '<tt>udiv</tt>' instruction returns the quotient of its two
2596operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2597
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2598<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2599
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2600<p>The two arguments to the '<tt>udiv</tt>' instruction must be
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2601<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2602values. Both arguments must have identical types.</p>
2603
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2604<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2605
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2606<p>The value produced is the unsigned integer quotient of the two operands.</p>
2607<p>Note that unsigned integer division and signed integer division are distinct
2608operations; for signed integer division, use '<tt>sdiv</tt>'.</p>
2609<p>Division by zero leads to undefined behavior.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2610<h5>Example:</h5>
2611<pre> &lt;result&gt; = udiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
2612</pre>
2613</div>
2614<!-- _______________________________________________________________________ -->
2615<div class="doc_subsubsection"> <a name="i_sdiv">'<tt>sdiv</tt>' Instruction
2616</a> </div>
2617<div class="doc_text">
2618<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2619<pre>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2620 &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]2621</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2622
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2623<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2624
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2625<p>The '<tt>sdiv</tt>' instruction returns the quotient of its two
2626operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2627
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2628<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2629
2630<p>The two arguments to the '<tt>sdiv</tt>' instruction must be
2631<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2632values. Both arguments must have identical types.</p>
2633
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2634<h5>Semantics:</h5>
Chris Lattnerdc15b1d2008-04-01 18:45:27 +0000[diff] [blame]2635<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]2636<p>Note that signed integer division and unsigned integer division are distinct
2637operations; for unsigned integer division, use '<tt>udiv</tt>'.</p>
2638<p>Division by zero leads to undefined behavior. Overflow also leads to
2639undefined behavior; this is a rare case, but can occur, for example,
2640by doing a 32-bit division of -2147483648 by -1.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2641<h5>Example:</h5>
2642<pre> &lt;result&gt; = sdiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
2643</pre>
2644</div>
2645<!-- _______________________________________________________________________ -->
2646<div class="doc_subsubsection"> <a name="i_fdiv">'<tt>fdiv</tt>'
2647Instruction</a> </div>
2648<div class="doc_text">
2649<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2650<pre>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2651 &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]2652</pre>
2653<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2654
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2655<p>The '<tt>fdiv</tt>' instruction returns the quotient of its two
2656operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2657
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2658<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2659
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2660<p>The two arguments to the '<tt>fdiv</tt>' instruction must be
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2661<a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a>
2662of floating point values. Both arguments must have identical types.</p>
2663
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2664<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2665
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2666<p>The value produced is the floating point quotient of the two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2667
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2668<h5>Example:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2669
2670<pre>
2671 &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]2672</pre>
2673</div>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2674
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2675<!-- _______________________________________________________________________ -->
2676<div class="doc_subsubsection"> <a name="i_urem">'<tt>urem</tt>' Instruction</a>
2677</div>
2678<div class="doc_text">
2679<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2680<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]2681</pre>
2682<h5>Overview:</h5>
2683<p>The '<tt>urem</tt>' instruction returns the remainder from the
2684unsigned division of its two arguments.</p>
2685<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2686<p>The two arguments to the '<tt>urem</tt>' instruction must be
2687<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2688values. Both arguments must have identical types.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2689<h5>Semantics:</h5>
2690<p>This instruction returns the unsigned integer <i>remainder</i> of a division.
Chris Lattnerdc15b1d2008-04-01 18:45:27 +0000[diff] [blame]2691This instruction always performs an unsigned division to get the remainder.</p>
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2692<p>Note that unsigned integer remainder and signed integer remainder are
2693distinct operations; for signed integer remainder, use '<tt>srem</tt>'.</p>
2694<p>Taking the remainder of a division by zero leads to undefined behavior.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2695<h5>Example:</h5>
2696<pre> &lt;result&gt; = urem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
2697</pre>
2698
2699</div>
2700<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2701<div class="doc_subsubsection">
2702 <a name="i_srem">'<tt>srem</tt>' Instruction</a>
2703</div>
2704
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2705<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2706
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2707<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2708
2709<pre>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2710 &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]2711</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2712
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2713<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2714
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2715<p>The '<tt>srem</tt>' instruction returns the remainder from the
Dan Gohman3e3fd8c2007-11-05 23:35:22 +0000[diff] [blame]2716signed division of its two operands. This instruction can also take
2717<a href="#t_vector">vector</a> versions of the values in which case
2718the elements must be integers.</p>
Chris Lattner08497ce2008-01-04 04:33:49 +0000[diff] [blame]2719
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2720<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2721
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2722<p>The two arguments to the '<tt>srem</tt>' instruction must be
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2723<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2724values. Both arguments must have identical types.</p>
2725
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2726<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2727
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2728<p>This instruction returns the <i>remainder</i> of a division (where the result
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2729has the same sign as the dividend, <tt>op1</tt>), not the <i>modulo</i>
2730operator (where the result has the same sign as the divisor, <tt>op2</tt>) of
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2731a value. For more information about the difference, see <a
2732 href="http://mathforum.org/dr.math/problems/anne.4.28.99.html">The
2733Math Forum</a>. For a table of how this is implemented in various languages,
2734please see <a href="http://en.wikipedia.org/wiki/Modulo_operation">
2735Wikipedia: modulo operation</a>.</p>
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2736<p>Note that signed integer remainder and unsigned integer remainder are
2737distinct operations; for unsigned integer remainder, use '<tt>urem</tt>'.</p>
2738<p>Taking the remainder of a division by zero leads to undefined behavior.
2739Overflow also leads to undefined behavior; this is a rare case, but can occur,
2740for example, by taking the remainder of a 32-bit division of -2147483648 by -1.
2741(The remainder doesn't actually overflow, but this rule lets srem be
2742implemented using instructions that return both the result of the division
2743and the remainder.)</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2744<h5>Example:</h5>
2745<pre> &lt;result&gt; = srem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
2746</pre>
2747
2748</div>
2749<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2750<div class="doc_subsubsection">
2751 <a name="i_frem">'<tt>frem</tt>' Instruction</a> </div>
2752
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2753<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2754
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2755<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2756<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]2757</pre>
2758<h5>Overview:</h5>
2759<p>The '<tt>frem</tt>' instruction returns the remainder from the
2760division of its two operands.</p>
2761<h5>Arguments:</h5>
2762<p>The two arguments to the '<tt>frem</tt>' instruction must be
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2763<a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a>
2764of floating point values. Both arguments must have identical types.</p>
2765
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2766<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2767
Chris Lattnerdc15b1d2008-04-01 18:45:27 +0000[diff] [blame]2768<p>This instruction returns the <i>remainder</i> of a division.
2769The remainder has the same sign as the dividend.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2770
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2771<h5>Example:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2772
2773<pre>
2774 &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]2775</pre>
2776</div>
2777
2778<!-- ======================================================================= -->
2779<div class="doc_subsection"> <a name="bitwiseops">Bitwise Binary
2780Operations</a> </div>
2781<div class="doc_text">
2782<p>Bitwise binary operators are used to do various forms of
2783bit-twiddling in a program. They are generally very efficient
2784instructions and can commonly be strength reduced from other
Chris Lattnerdc15b1d2008-04-01 18:45:27 +0000[diff] [blame]2785instructions. They require two operands of the same type, execute an operation on them,
2786and 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]2787</div>
2788
2789<!-- _______________________________________________________________________ -->
2790<div class="doc_subsubsection"> <a name="i_shl">'<tt>shl</tt>'
2791Instruction</a> </div>
2792<div class="doc_text">
2793<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2794<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]2795</pre>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2796
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2797<h5>Overview:</h5>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2798
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2799<p>The '<tt>shl</tt>' instruction returns the first operand shifted to
2800the left a specified number of bits.</p>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2801
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2802<h5>Arguments:</h5>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2803
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2804<p>Both arguments to the '<tt>shl</tt>' instruction must be the same <a
Nate Begemanbb1ce942008-07-29 15:49:41 +0000[diff] [blame]2805 href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2806type. '<tt>op2</tt>' is treated as an unsigned value.</p>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2807
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2808<h5>Semantics:</h5>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2809
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2810<p>The value produced is <tt>op1</tt> * 2<sup><tt>op2</tt></sup> mod 2<sup>n</sup>,
2811where 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]2812equal to or larger than the number of bits in <tt>op1</tt>, the result is undefined.
2813If the arguments are vectors, each vector element of <tt>op1</tt> is shifted by the
2814corresponding shift amount in <tt>op2</tt>.</p>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2815
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2816<h5>Example:</h5><pre>
2817 &lt;result&gt; = shl i32 4, %var <i>; yields {i32}: 4 &lt;&lt; %var</i>
2818 &lt;result&gt; = shl i32 4, 2 <i>; yields {i32}: 16</i>
2819 &lt;result&gt; = shl i32 1, 10 <i>; yields {i32}: 1024</i>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2820 &lt;result&gt; = shl i32 1, 32 <i>; undefined</i>
Mon P Wang9901e732008-12-09 05:46:39 +0000[diff] [blame]2821 &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]2822</pre>
2823</div>
2824<!-- _______________________________________________________________________ -->
2825<div class="doc_subsubsection"> <a name="i_lshr">'<tt>lshr</tt>'
2826Instruction</a> </div>
2827<div class="doc_text">
2828<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2829<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]2830</pre>
2831
2832<h5>Overview:</h5>
2833<p>The '<tt>lshr</tt>' instruction (logical shift right) returns the first
2834operand shifted to the right a specified number of bits with zero fill.</p>
2835
2836<h5>Arguments:</h5>
2837<p>Both arguments to the '<tt>lshr</tt>' instruction must be the same
Nate Begemanbb1ce942008-07-29 15:49:41 +0000[diff] [blame]2838<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]2839type. '<tt>op2</tt>' is treated as an unsigned value.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2840
2841<h5>Semantics:</h5>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2842
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2843<p>This instruction always performs a logical shift right operation. The most
2844significant bits of the result will be filled with zero bits after the
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2845shift. If <tt>op2</tt> is (statically or dynamically) equal to or larger than
Mon P Wangb0f51822008-12-10 08:55:09 +0000[diff] [blame]2846the number of bits in <tt>op1</tt>, the result is undefined. If the arguments are
2847vectors, each vector element of <tt>op1</tt> is shifted by the corresponding shift
2848amount in <tt>op2</tt>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2849
2850<h5>Example:</h5>
2851<pre>
2852 &lt;result&gt; = lshr i32 4, 1 <i>; yields {i32}:result = 2</i>
2853 &lt;result&gt; = lshr i32 4, 2 <i>; yields {i32}:result = 1</i>
2854 &lt;result&gt; = lshr i8 4, 3 <i>; yields {i8}:result = 0</i>
2855 &lt;result&gt; = lshr i8 -2, 1 <i>; yields {i8}:result = 0x7FFFFFFF </i>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2856 &lt;result&gt; = lshr i32 1, 32 <i>; undefined</i>
Mon P Wang9901e732008-12-09 05:46:39 +0000[diff] [blame]2857 &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]2858</pre>
2859</div>
2860
2861<!-- _______________________________________________________________________ -->
2862<div class="doc_subsubsection"> <a name="i_ashr">'<tt>ashr</tt>'
2863Instruction</a> </div>
2864<div class="doc_text">
2865
2866<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2867<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]2868</pre>
2869
2870<h5>Overview:</h5>
2871<p>The '<tt>ashr</tt>' instruction (arithmetic shift right) returns the first
2872operand shifted to the right a specified number of bits with sign extension.</p>
2873
2874<h5>Arguments:</h5>
2875<p>Both arguments to the '<tt>ashr</tt>' instruction must be the same
Nate Begemanbb1ce942008-07-29 15:49:41 +0000[diff] [blame]2876<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]2877type. '<tt>op2</tt>' is treated as an unsigned value.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2878
2879<h5>Semantics:</h5>
2880<p>This instruction always performs an arithmetic shift right operation,
2881The most significant bits of the result will be filled with the sign bit
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2882of <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]2883larger than the number of bits in <tt>op1</tt>, the result is undefined. If the
2884arguments are vectors, each vector element of <tt>op1</tt> is shifted by the
2885corresponding shift amount in <tt>op2</tt>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2886
2887<h5>Example:</h5>
2888<pre>
2889 &lt;result&gt; = ashr i32 4, 1 <i>; yields {i32}:result = 2</i>
2890 &lt;result&gt; = ashr i32 4, 2 <i>; yields {i32}:result = 1</i>
2891 &lt;result&gt; = ashr i8 4, 3 <i>; yields {i8}:result = 0</i>
2892 &lt;result&gt; = ashr i8 -2, 1 <i>; yields {i8}:result = -1</i>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2893 &lt;result&gt; = ashr i32 1, 32 <i>; undefined</i>
Mon P Wang9901e732008-12-09 05:46:39 +0000[diff] [blame]2894 &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]2895</pre>
2896</div>
2897
2898<!-- _______________________________________________________________________ -->
2899<div class="doc_subsubsection"> <a name="i_and">'<tt>and</tt>'
2900Instruction</a> </div>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2901
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2902<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2903
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2904<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2905
2906<pre>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2907 &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]2908</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2909
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2910<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2911
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2912<p>The '<tt>and</tt>' instruction returns the bitwise logical and of
2913its two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2914
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2915<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2916
2917<p>The two arguments to the '<tt>and</tt>' instruction must be
2918<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2919values. Both arguments must have identical types.</p>
2920
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2921<h5>Semantics:</h5>
2922<p>The truth table used for the '<tt>and</tt>' instruction is:</p>
2923<p> </p>
Bill Wendling1e19eac2008-09-07 10:29:20 +0000[diff] [blame]2924<div>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2925<table border="1" cellspacing="0" cellpadding="4">
2926 <tbody>
2927 <tr>
2928 <td>In0</td>
2929 <td>In1</td>
2930 <td>Out</td>
2931 </tr>
2932 <tr>
2933 <td>0</td>
2934 <td>0</td>
2935 <td>0</td>
2936 </tr>
2937 <tr>
2938 <td>0</td>
2939 <td>1</td>
2940 <td>0</td>
2941 </tr>
2942 <tr>
2943 <td>1</td>
2944 <td>0</td>
2945 <td>0</td>
2946 </tr>
2947 <tr>
2948 <td>1</td>
2949 <td>1</td>
2950 <td>1</td>
2951 </tr>
2952 </tbody>
2953</table>
2954</div>
2955<h5>Example:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2956<pre>
2957 &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]2958 &lt;result&gt; = and i32 15, 40 <i>; yields {i32}:result = 8</i>
2959 &lt;result&gt; = and i32 4, 8 <i>; yields {i32}:result = 0</i>
2960</pre>
2961</div>
2962<!-- _______________________________________________________________________ -->
2963<div class="doc_subsubsection"> <a name="i_or">'<tt>or</tt>' Instruction</a> </div>
2964<div class="doc_text">
2965<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2966<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]2967</pre>
2968<h5>Overview:</h5>
2969<p>The '<tt>or</tt>' instruction returns the bitwise logical inclusive
2970or of its two operands.</p>
2971<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2972
2973<p>The two arguments to the '<tt>or</tt>' instruction must be
2974<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2975values. Both arguments must have identical types.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2976<h5>Semantics:</h5>
2977<p>The truth table used for the '<tt>or</tt>' instruction is:</p>
2978<p> </p>
Bill Wendling1e19eac2008-09-07 10:29:20 +0000[diff] [blame]2979<div>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2980<table border="1" cellspacing="0" cellpadding="4">
2981 <tbody>
2982 <tr>
2983 <td>In0</td>
2984 <td>In1</td>
2985 <td>Out</td>
2986 </tr>
2987 <tr>
2988 <td>0</td>
2989 <td>0</td>
2990 <td>0</td>
2991 </tr>
2992 <tr>
2993 <td>0</td>
2994 <td>1</td>
2995 <td>1</td>
2996 </tr>
2997 <tr>
2998 <td>1</td>
2999 <td>0</td>
3000 <td>1</td>
3001 </tr>
3002 <tr>
3003 <td>1</td>
3004 <td>1</td>
3005 <td>1</td>
3006 </tr>
3007 </tbody>
3008</table>
3009</div>
3010<h5>Example:</h5>
3011<pre> &lt;result&gt; = or i32 4, %var <i>; yields {i32}:result = 4 | %var</i>
3012 &lt;result&gt; = or i32 15, 40 <i>; yields {i32}:result = 47</i>
3013 &lt;result&gt; = or i32 4, 8 <i>; yields {i32}:result = 12</i>
3014</pre>
3015</div>
3016<!-- _______________________________________________________________________ -->
3017<div class="doc_subsubsection"> <a name="i_xor">'<tt>xor</tt>'
3018Instruction</a> </div>
3019<div class="doc_text">
3020<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3021<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]3022</pre>
3023<h5>Overview:</h5>
3024<p>The '<tt>xor</tt>' instruction returns the bitwise logical exclusive
3025or of its two operands. The <tt>xor</tt> is used to implement the
3026"one's complement" operation, which is the "~" operator in C.</p>
3027<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3028<p>The two arguments to the '<tt>xor</tt>' instruction must be
3029<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3030values. Both arguments must have identical types.</p>
3031
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3032<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3033
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3034<p>The truth table used for the '<tt>xor</tt>' instruction is:</p>
3035<p> </p>
Bill Wendling1e19eac2008-09-07 10:29:20 +0000[diff] [blame]3036<div>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3037<table border="1" cellspacing="0" cellpadding="4">
3038 <tbody>
3039 <tr>
3040 <td>In0</td>
3041 <td>In1</td>
3042 <td>Out</td>
3043 </tr>
3044 <tr>
3045 <td>0</td>
3046 <td>0</td>
3047 <td>0</td>
3048 </tr>
3049 <tr>
3050 <td>0</td>
3051 <td>1</td>
3052 <td>1</td>
3053 </tr>
3054 <tr>
3055 <td>1</td>
3056 <td>0</td>
3057 <td>1</td>
3058 </tr>
3059 <tr>
3060 <td>1</td>
3061 <td>1</td>
3062 <td>0</td>
3063 </tr>
3064 </tbody>
3065</table>
3066</div>
3067<p> </p>
3068<h5>Example:</h5>
3069<pre> &lt;result&gt; = xor i32 4, %var <i>; yields {i32}:result = 4 ^ %var</i>
3070 &lt;result&gt; = xor i32 15, 40 <i>; yields {i32}:result = 39</i>
3071 &lt;result&gt; = xor i32 4, 8 <i>; yields {i32}:result = 12</i>
3072 &lt;result&gt; = xor i32 %V, -1 <i>; yields {i32}:result = ~%V</i>
3073</pre>
3074</div>
3075
3076<!-- ======================================================================= -->
3077<div class="doc_subsection">
3078 <a name="vectorops">Vector Operations</a>
3079</div>
3080
3081<div class="doc_text">
3082
3083<p>LLVM supports several instructions to represent vector operations in a
3084target-independent manner. These instructions cover the element-access and
3085vector-specific operations needed to process vectors effectively. While LLVM
3086does directly support these vector operations, many sophisticated algorithms
3087will want to use target-specific intrinsics to take full advantage of a specific
3088target.</p>
3089
3090</div>
3091
3092<!-- _______________________________________________________________________ -->
3093<div class="doc_subsubsection">
3094 <a name="i_extractelement">'<tt>extractelement</tt>' Instruction</a>
3095</div>
3096
3097<div class="doc_text">
3098
3099<h5>Syntax:</h5>
3100
3101<pre>
3102 &lt;result&gt; = extractelement &lt;n x &lt;ty&gt;&gt; &lt;val&gt;, i32 &lt;idx&gt; <i>; yields &lt;ty&gt;</i>
3103</pre>
3104
3105<h5>Overview:</h5>
3106
3107<p>
3108The '<tt>extractelement</tt>' instruction extracts a single scalar
3109element from a vector at a specified index.
3110</p>
3111
3112
3113<h5>Arguments:</h5>
3114
3115<p>
3116The first operand of an '<tt>extractelement</tt>' instruction is a
3117value of <a href="#t_vector">vector</a> type. The second operand is
3118an index indicating the position from which to extract the element.
3119The index may be a variable.</p>
3120
3121<h5>Semantics:</h5>
3122
3123<p>
3124The result is a scalar of the same type as the element type of
3125<tt>val</tt>. Its value is the value at position <tt>idx</tt> of
3126<tt>val</tt>. If <tt>idx</tt> exceeds the length of <tt>val</tt>, the
3127results are undefined.
3128</p>
3129
3130<h5>Example:</h5>
3131
3132<pre>
3133 %result = extractelement &lt;4 x i32&gt; %vec, i32 0 <i>; yields i32</i>
3134</pre>
3135</div>
3136
3137
3138<!-- _______________________________________________________________________ -->
3139<div class="doc_subsubsection">
3140 <a name="i_insertelement">'<tt>insertelement</tt>' Instruction</a>
3141</div>
3142
3143<div class="doc_text">
3144
3145<h5>Syntax:</h5>
3146
3147<pre>
Dan Gohmanbcc3c502008-05-12 23:38:42 +0000[diff] [blame]3148 &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]3149</pre>
3150
3151<h5>Overview:</h5>
3152
3153<p>
3154The '<tt>insertelement</tt>' instruction inserts a scalar
3155element into a vector at a specified index.
3156</p>
3157
3158
3159<h5>Arguments:</h5>
3160
3161<p>
3162The first operand of an '<tt>insertelement</tt>' instruction is a
3163value of <a href="#t_vector">vector</a> type. The second operand is a
3164scalar value whose type must equal the element type of the first
3165operand. The third operand is an index indicating the position at
3166which to insert the value. The index may be a variable.</p>
3167
3168<h5>Semantics:</h5>
3169
3170<p>
3171The result is a vector of the same type as <tt>val</tt>. Its
3172element values are those of <tt>val</tt> except at position
3173<tt>idx</tt>, where it gets the value <tt>elt</tt>. If <tt>idx</tt>
3174exceeds the length of <tt>val</tt>, the results are undefined.
3175</p>
3176
3177<h5>Example:</h5>
3178
3179<pre>
3180 %result = insertelement &lt;4 x i32&gt; %vec, i32 1, i32 0 <i>; yields &lt;4 x i32&gt;</i>
3181</pre>
3182</div>
3183
3184<!-- _______________________________________________________________________ -->
3185<div class="doc_subsubsection">
3186 <a name="i_shufflevector">'<tt>shufflevector</tt>' Instruction</a>
3187</div>
3188
3189<div class="doc_text">
3190
3191<h5>Syntax:</h5>
3192
3193<pre>
Mon P Wangbff5d9c2008-11-10 04:46:22 +0000[diff] [blame]3194 &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]3195</pre>
3196
3197<h5>Overview:</h5>
3198
3199<p>
3200The '<tt>shufflevector</tt>' instruction constructs a permutation of elements
Mon P Wangbff5d9c2008-11-10 04:46:22 +0000[diff] [blame]3201from two input vectors, returning a vector with the same element type as
3202the input and length that is the same as the shuffle mask.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3203</p>
3204
3205<h5>Arguments:</h5>
3206
3207<p>
Mon P Wangbff5d9c2008-11-10 04:46:22 +0000[diff] [blame]3208The first two operands of a '<tt>shufflevector</tt>' instruction are vectors
3209with types that match each other. The third argument is a shuffle mask whose
3210element type is always 'i32'. The result of the instruction is a vector whose
3211length is the same as the shuffle mask and whose element type is the same as
3212the element type of the first two operands.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3213</p>
3214
3215<p>
3216The shuffle mask operand is required to be a constant vector with either
3217constant integer or undef values.
3218</p>
3219
3220<h5>Semantics:</h5>
3221
3222<p>
3223The elements of the two input vectors are numbered from left to right across
3224both of the vectors. The shuffle mask operand specifies, for each element of
Mon P Wangbff5d9c2008-11-10 04:46:22 +0000[diff] [blame]3225the result vector, which element of the two input vectors the result element
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3226gets. The element selector may be undef (meaning "don't care") and the second
3227operand may be undef if performing a shuffle from only one vector.
3228</p>
3229
3230<h5>Example:</h5>
3231
3232<pre>
3233 %result = shufflevector &lt;4 x i32&gt; %v1, &lt;4 x i32&gt; %v2,
3234 &lt;4 x i32&gt; &lt;i32 0, i32 4, i32 1, i32 5&gt; <i>; yields &lt;4 x i32&gt;</i>
3235 %result = shufflevector &lt;4 x i32&gt; %v1, &lt;4 x i32&gt; undef,
3236 &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]3237 %result = shufflevector &lt;8 x i32&gt; %v1, &lt;8 x i32&gt; undef,
3238 &lt;4 x i32&gt; &lt;i32 0, i32 1, i32 2, i32 3&gt; <i>; yields &lt;4 x i32&gt;</i>
3239 %result = shufflevector &lt;4 x i32&gt; %v1, &lt;4 x i32&gt; %v2,
3240 &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]3241</pre>
3242</div>
3243
3244
3245<!-- ======================================================================= -->
3246<div class="doc_subsection">
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3247 <a name="aggregateops">Aggregate Operations</a>
3248</div>
3249
3250<div class="doc_text">
3251
3252<p>LLVM supports several instructions for working with aggregate values.
3253</p>
3254
3255</div>
3256
3257<!-- _______________________________________________________________________ -->
3258<div class="doc_subsubsection">
3259 <a name="i_extractvalue">'<tt>extractvalue</tt>' Instruction</a>
3260</div>
3261
3262<div class="doc_text">
3263
3264<h5>Syntax:</h5>
3265
3266<pre>
3267 &lt;result&gt; = extractvalue &lt;aggregate type&gt; &lt;val&gt;, &lt;idx&gt;{, &lt;idx&gt;}*
3268</pre>
3269
3270<h5>Overview:</h5>
3271
3272<p>
Dan Gohman6c6dea02008-05-13 18:16:06 +0000[diff] [blame]3273The '<tt>extractvalue</tt>' instruction extracts the value of a struct field
3274or array element from an aggregate value.
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3275</p>
3276
3277
3278<h5>Arguments:</h5>
3279
3280<p>
3281The first operand of an '<tt>extractvalue</tt>' instruction is a
3282value of <a href="#t_struct">struct</a> or <a href="#t_array">array</a>
Dan Gohman6c6dea02008-05-13 18:16:06 +0000[diff] [blame]3283type. The operands are constant indices to specify which value to extract
Dan Gohmane5febe42008-05-31 00:58:22 +0000[diff] [blame]3284in a similar manner as indices in a
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3285'<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.
3286</p>
3287
3288<h5>Semantics:</h5>
3289
3290<p>
3291The result is the value at the position in the aggregate specified by
3292the index operands.
3293</p>
3294
3295<h5>Example:</h5>
3296
3297<pre>
Dan Gohmane5febe42008-05-31 00:58:22 +0000[diff] [blame]3298 %result = extractvalue {i32, float} %agg, 0 <i>; yields i32</i>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3299</pre>
3300</div>
3301
3302
3303<!-- _______________________________________________________________________ -->
3304<div class="doc_subsubsection">
3305 <a name="i_insertvalue">'<tt>insertvalue</tt>' Instruction</a>
3306</div>
3307
3308<div class="doc_text">
3309
3310<h5>Syntax:</h5>
3311
3312<pre>
Dan Gohmane5febe42008-05-31 00:58:22 +0000[diff] [blame]3313 &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]3314</pre>
3315
3316<h5>Overview:</h5>
3317
3318<p>
3319The '<tt>insertvalue</tt>' instruction inserts a value
Dan Gohman6c6dea02008-05-13 18:16:06 +0000[diff] [blame]3320into a struct field or array element in an aggregate.
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3321</p>
3322
3323
3324<h5>Arguments:</h5>
3325
3326<p>
3327The first operand of an '<tt>insertvalue</tt>' instruction is a
3328value of <a href="#t_struct">struct</a> or <a href="#t_array">array</a> type.
3329The second operand is a first-class value to insert.
Dan Gohman4f29e422008-05-23 21:53:15 +0000[diff] [blame]3330The following operands are constant indices
Dan Gohmane5febe42008-05-31 00:58:22 +0000[diff] [blame]3331indicating the position at which to insert the value in a similar manner as
Dan Gohman6c6dea02008-05-13 18:16:06 +0000[diff] [blame]3332indices in a
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3333'<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.
3334The value to insert must have the same type as the value identified
Dan Gohman6c6dea02008-05-13 18:16:06 +0000[diff] [blame]3335by the indices.
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]3336</p>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3337
3338<h5>Semantics:</h5>
3339
3340<p>
3341The result is an aggregate of the same type as <tt>val</tt>. Its
3342value is that of <tt>val</tt> except that the value at the position
Dan Gohman6c6dea02008-05-13 18:16:06 +0000[diff] [blame]3343specified by the indices is that of <tt>elt</tt>.
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3344</p>
3345
3346<h5>Example:</h5>
3347
3348<pre>
Dan Gohmanb1aab4e2008-06-23 15:26:37 +0000[diff] [blame]3349 %result = insertvalue {i32, float} %agg, i32 1, 0 <i>; yields {i32, float}</i>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3350</pre>
3351</div>
3352
3353
3354<!-- ======================================================================= -->
3355<div class="doc_subsection">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3356 <a name="memoryops">Memory Access and Addressing Operations</a>
3357</div>
3358
3359<div class="doc_text">
3360
3361<p>A key design point of an SSA-based representation is how it
3362represents memory. In LLVM, no memory locations are in SSA form, which
3363makes things very simple. This section describes how to read, write,
3364allocate, and free memory in LLVM.</p>
3365
3366</div>
3367
3368<!-- _______________________________________________________________________ -->
3369<div class="doc_subsubsection">
3370 <a name="i_malloc">'<tt>malloc</tt>' Instruction</a>
3371</div>
3372
3373<div class="doc_text">
3374
3375<h5>Syntax:</h5>
3376
3377<pre>
3378 &lt;result&gt; = malloc &lt;type&gt;[, i32 &lt;NumElements&gt;][, align &lt;alignment&gt;] <i>; yields {type*}:result</i>
3379</pre>
3380
3381<h5>Overview:</h5>
3382
3383<p>The '<tt>malloc</tt>' instruction allocates memory from the system
Christopher Lambcfe00962007-12-17 01:00:21 +0000[diff] [blame]3384heap and returns a pointer to it. The object is always allocated in the generic
3385address space (address space zero).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3386
3387<h5>Arguments:</h5>
3388
3389<p>The '<tt>malloc</tt>' instruction allocates
3390<tt>sizeof(&lt;type&gt;)*NumElements</tt>
3391bytes of memory from the operating system and returns a pointer of the
3392appropriate type to the program. If "NumElements" is specified, it is the
Gabor Greif5082cf42008-02-09 22:24:34 +0000[diff] [blame]3393number of elements allocated, otherwise "NumElements" is defaulted to be one.
Chris Lattner10368b62008-04-02 00:38:26 +0000[diff] [blame]3394If a constant alignment is specified, the value result of the allocation is guaranteed to
Gabor Greif5082cf42008-02-09 22:24:34 +0000[diff] [blame]3395be aligned to at least that boundary. If not specified, or if zero, the target can
3396choose to align the allocation on any convenient boundary.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3397
3398<p>'<tt>type</tt>' must be a sized type.</p>
3399
3400<h5>Semantics:</h5>
3401
3402<p>Memory is allocated using the system "<tt>malloc</tt>" function, and
Nick Lewyckyff384472008-11-24 03:41:24 +0000[diff] [blame]3403a pointer is returned. The result of a zero byte allocation is undefined. The
Chris Lattner8b094fc2008-04-19 21:01:16 +0000[diff] [blame]3404result is null if there is insufficient memory available.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3405
3406<h5>Example:</h5>
3407
3408<pre>
Dan Gohmanf54f50a2009-01-04 23:49:44 +0000[diff] [blame]3409 %array = malloc [4 x i8] <i>; yields {[%4 x i8]*}:array</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3410
3411 %size = <a href="#i_add">add</a> i32 2, 2 <i>; yields {i32}:size = i32 4</i>
3412 %array1 = malloc i8, i32 4 <i>; yields {i8*}:array1</i>
3413 %array2 = malloc [12 x i8], i32 %size <i>; yields {[12 x i8]*}:array2</i>
3414 %array3 = malloc i32, i32 4, align 1024 <i>; yields {i32*}:array3</i>
3415 %array4 = malloc i32, align 1024 <i>; yields {i32*}:array4</i>
3416</pre>
Dan Gohman60967192009-01-12 23:12:39 +0000[diff] [blame]3417
3418<p>Note that the code generator does not yet respect the
3419 alignment value.</p>
3420
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3421</div>
3422
3423<!-- _______________________________________________________________________ -->
3424<div class="doc_subsubsection">
3425 <a name="i_free">'<tt>free</tt>' Instruction</a>
3426</div>
3427
3428<div class="doc_text">
3429
3430<h5>Syntax:</h5>
3431
3432<pre>
Dan Gohmanf54f50a2009-01-04 23:49:44 +0000[diff] [blame]3433 free &lt;type&gt; &lt;value&gt; <i>; yields {void}</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3434</pre>
3435
3436<h5>Overview:</h5>
3437
3438<p>The '<tt>free</tt>' instruction returns memory back to the unused
3439memory heap to be reallocated in the future.</p>
3440
3441<h5>Arguments:</h5>
3442
3443<p>'<tt>value</tt>' shall be a pointer value that points to a value
3444that was allocated with the '<tt><a href="#i_malloc">malloc</a></tt>'
3445instruction.</p>
3446
3447<h5>Semantics:</h5>
3448
3449<p>Access to the memory pointed to by the pointer is no longer defined
Chris Lattner329d6532008-04-19 22:41:32 +0000[diff] [blame]3450after this instruction executes. If the pointer is null, the operation
3451is a noop.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3452
3453<h5>Example:</h5>
3454
3455<pre>
Dan Gohmanf54f50a2009-01-04 23:49:44 +0000[diff] [blame]3456 %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]3457 free [4 x i8]* %array
3458</pre>
3459</div>
3460
3461<!-- _______________________________________________________________________ -->
3462<div class="doc_subsubsection">
3463 <a name="i_alloca">'<tt>alloca</tt>' Instruction</a>
3464</div>
3465
3466<div class="doc_text">
3467
3468<h5>Syntax:</h5>
3469
3470<pre>
3471 &lt;result&gt; = alloca &lt;type&gt;[, i32 &lt;NumElements&gt;][, align &lt;alignment&gt;] <i>; yields {type*}:result</i>
3472</pre>
3473
3474<h5>Overview:</h5>
3475
3476<p>The '<tt>alloca</tt>' instruction allocates memory on the stack frame of the
3477currently executing function, to be automatically released when this function
Christopher Lambcfe00962007-12-17 01:00:21 +0000[diff] [blame]3478returns to its caller. The object is always allocated in the generic address
3479space (address space zero).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3480
3481<h5>Arguments:</h5>
3482
3483<p>The '<tt>alloca</tt>' instruction allocates <tt>sizeof(&lt;type&gt;)*NumElements</tt>
3484bytes of memory on the runtime stack, returning a pointer of the
Gabor Greif5082cf42008-02-09 22:24:34 +0000[diff] [blame]3485appropriate type to the program. If "NumElements" is specified, it is the
3486number of elements allocated, otherwise "NumElements" is defaulted to be one.
Chris Lattner10368b62008-04-02 00:38:26 +0000[diff] [blame]3487If a constant alignment is specified, the value result of the allocation is guaranteed
Gabor Greif5082cf42008-02-09 22:24:34 +0000[diff] [blame]3488to be aligned to at least that boundary. If not specified, or if zero, the target
3489can choose to align the allocation on any convenient boundary.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3490
3491<p>'<tt>type</tt>' may be any sized type.</p>
3492
3493<h5>Semantics:</h5>
3494
Chris Lattner8b094fc2008-04-19 21:01:16 +0000[diff] [blame]3495<p>Memory is allocated; a pointer is returned. The operation is undefiend if
3496there is insufficient stack space for the allocation. '<tt>alloca</tt>'d
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3497memory is automatically released when the function returns. The '<tt>alloca</tt>'
3498instruction is commonly used to represent automatic variables that must
3499have an address available. When the function returns (either with the <tt><a
3500 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]3501instructions), the memory is reclaimed. Allocating zero bytes
3502is legal, but the result is undefined.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3503
3504<h5>Example:</h5>
3505
3506<pre>
Dan Gohmanf54f50a2009-01-04 23:49:44 +0000[diff] [blame]3507 %ptr = alloca i32 <i>; yields {i32*}:ptr</i>
3508 %ptr = alloca i32, i32 4 <i>; yields {i32*}:ptr</i>
3509 %ptr = alloca i32, i32 4, align 1024 <i>; yields {i32*}:ptr</i>
3510 %ptr = alloca i32, align 1024 <i>; yields {i32*}:ptr</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3511</pre>
3512</div>
3513
3514<!-- _______________________________________________________________________ -->
3515<div class="doc_subsubsection"> <a name="i_load">'<tt>load</tt>'
3516Instruction</a> </div>
3517<div class="doc_text">
3518<h5>Syntax:</h5>
3519<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>
3520<h5>Overview:</h5>
3521<p>The '<tt>load</tt>' instruction is used to read from memory.</p>
3522<h5>Arguments:</h5>
3523<p>The argument to the '<tt>load</tt>' instruction specifies the memory
3524address from which to load. The pointer must point to a <a
3525 href="#t_firstclass">first class</a> type. If the <tt>load</tt> is
3526marked as <tt>volatile</tt>, then the optimizer is not allowed to modify
3527the number or order of execution of this <tt>load</tt> with other
3528volatile <tt>load</tt> and <tt><a href="#i_store">store</a></tt>
3529instructions. </p>
Chris Lattner21003c82008-01-06 21:04:43 +0000[diff] [blame]3530<p>
Chris Lattner10368b62008-04-02 00:38:26 +0000[diff] [blame]3531The optional constant "align" argument specifies the alignment of the operation
Chris Lattner21003c82008-01-06 21:04:43 +0000[diff] [blame]3532(that is, the alignment of the memory address). A value of 0 or an
3533omitted "align" argument means that the operation has the preferential
3534alignment for the target. It is the responsibility of the code emitter
3535to ensure that the alignment information is correct. Overestimating
3536the alignment results in an undefined behavior. Underestimating the
3537alignment may produce less efficient code. An alignment of 1 is always
3538safe.
3539</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3540<h5>Semantics:</h5>
Duncan Sandsfe87fff2009-03-22 11:33:16 +0000[diff] [blame]3541<p>The location of memory pointed to is loaded. If the value being loaded
3542is of scalar type then the number of bytes read does not exceed the minimum
3543number of bytes needed to hold all bits of the type. For example, loading an
3544<tt>i24</tt> reads at most three bytes. When loading a value of a type like
3545<tt>i20</tt> with a size that is not an integral number of bytes, the result
3546is undefined if the value was not originally written using a store of the
3547same type.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3548<h5>Examples:</h5>
3549<pre> %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
3550 <a
3551 href="#i_store">store</a> i32 3, i32* %ptr <i>; yields {void}</i>
3552 %val = load i32* %ptr <i>; yields {i32}:val = i32 3</i>
3553</pre>
3554</div>
3555<!-- _______________________________________________________________________ -->
3556<div class="doc_subsubsection"> <a name="i_store">'<tt>store</tt>'
3557Instruction</a> </div>
3558<div class="doc_text">
3559<h5>Syntax:</h5>
3560<pre> store &lt;ty&gt; &lt;value&gt;, &lt;ty&gt;* &lt;pointer&gt;[, align &lt;alignment&gt;] <i>; yields {void}</i>
3561 volatile store &lt;ty&gt; &lt;value&gt;, &lt;ty&gt;* &lt;pointer&gt;[, align &lt;alignment&gt;] <i>; yields {void}</i>
3562</pre>
3563<h5>Overview:</h5>
3564<p>The '<tt>store</tt>' instruction is used to write to memory.</p>
3565<h5>Arguments:</h5>
3566<p>There are two arguments to the '<tt>store</tt>' instruction: a value
3567to 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]3568operand must be a pointer to the <a href="#t_firstclass">first class</a> type
3569of the '<tt>&lt;value&gt;</tt>'
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3570operand. If the <tt>store</tt> is marked as <tt>volatile</tt>, then the
3571optimizer is not allowed to modify the number or order of execution of
3572this <tt>store</tt> with other volatile <tt>load</tt> and <tt><a
3573 href="#i_store">store</a></tt> instructions.</p>
Chris Lattner21003c82008-01-06 21:04:43 +0000[diff] [blame]3574<p>
Chris Lattner10368b62008-04-02 00:38:26 +0000[diff] [blame]3575The optional constant "align" argument specifies the alignment of the operation
Chris Lattner21003c82008-01-06 21:04:43 +0000[diff] [blame]3576(that is, the alignment of the memory address). A value of 0 or an
3577omitted "align" argument means that the operation has the preferential
3578alignment for the target. It is the responsibility of the code emitter
3579to ensure that the alignment information is correct. Overestimating
3580the alignment results in an undefined behavior. Underestimating the
3581alignment may produce less efficient code. An alignment of 1 is always
3582safe.
3583</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3584<h5>Semantics:</h5>
3585<p>The contents of memory are updated to contain '<tt>&lt;value&gt;</tt>'
Duncan Sandsfe87fff2009-03-22 11:33:16 +0000[diff] [blame]3586at the location specified by the '<tt>&lt;pointer&gt;</tt>' operand.
3587If '<tt>&lt;value&gt;</tt>' is of scalar type then the number of bytes
3588written does not exceed the minimum number of bytes needed to hold all
3589bits of the type. For example, storing an <tt>i24</tt> writes at most
3590three bytes. When writing a value of a type like <tt>i20</tt> with a
3591size that is not an integral number of bytes, it is unspecified what
3592happens to the extra bits that do not belong to the type, but they will
3593typically be overwritten.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3594<h5>Example:</h5>
3595<pre> %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
Bill Wendling63ffa142007-10-22 05:10:05 +0000[diff] [blame]3596 store i32 3, i32* %ptr <i>; yields {void}</i>
3597 %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]3598</pre>
3599</div>
3600
3601<!-- _______________________________________________________________________ -->
3602<div class="doc_subsubsection">
3603 <a name="i_getelementptr">'<tt>getelementptr</tt>' Instruction</a>
3604</div>
3605
3606<div class="doc_text">
3607<h5>Syntax:</h5>
3608<pre>
Matthijs Kooijman32a080e2008-10-13 13:44:15 +0000[diff] [blame]3609 &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]3610</pre>
3611
3612<h5>Overview:</h5>
3613
3614<p>
3615The '<tt>getelementptr</tt>' instruction is used to get the address of a
Matthijs Kooijman32a080e2008-10-13 13:44:15 +0000[diff] [blame]3616subelement of an aggregate data structure. It performs address calculation only
3617and does not access memory.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3618
3619<h5>Arguments:</h5>
3620
Matthijs Kooijman32a080e2008-10-13 13:44:15 +0000[diff] [blame]3621<p>The first argument is always a pointer, and forms the basis of the
3622calculation. The remaining arguments are indices, that indicate which of the
3623elements of the aggregate object are indexed. The interpretation of each index
3624is dependent on the type being indexed into. The first index always indexes the
3625pointer value given as the first argument, the second index indexes a value of
3626the type pointed to (not necessarily the value directly pointed to, since the
3627first index can be non-zero), etc. The first type indexed into must be a pointer
3628value, subsequent types can be arrays, vectors and structs. Note that subsequent
3629types being indexed into can never be pointers, since that would require loading
3630the pointer before continuing calculation.</p>
3631
3632<p>The type of each index argument depends on the type it is indexing into.
3633When indexing into a (packed) structure, only <tt>i32</tt> integer
3634<b>constants</b> are allowed. When indexing into an array, pointer or vector,
Sanjiv Guptaa2e72d92009-04-27 03:21:00 +0000[diff] [blame]3635integers of any width are allowed (also non-constants).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3636
3637<p>For example, let's consider a C code fragment and how it gets
3638compiled to LLVM:</p>
3639
3640<div class="doc_code">
3641<pre>
3642struct RT {
3643 char A;
3644 int B[10][20];
3645 char C;
3646};
3647struct ST {
3648 int X;
3649 double Y;
3650 struct RT Z;
3651};
3652
3653int *foo(struct ST *s) {
3654 return &amp;s[1].Z.B[5][13];
3655}
3656</pre>
3657</div>
3658
3659<p>The LLVM code generated by the GCC frontend is:</p>
3660
3661<div class="doc_code">
3662<pre>
Chris Lattner5b6dc6e2009-01-11 20:53:49 +0000[diff] [blame]3663%RT = <a href="#namedtypes">type</a> { i8 , [10 x [20 x i32]], i8 }
3664%ST = <a href="#namedtypes">type</a> { i32, double, %RT }
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3665
3666define i32* %foo(%ST* %s) {
3667entry:
3668 %reg = getelementptr %ST* %s, i32 1, i32 2, i32 1, i32 5, i32 13
3669 ret i32* %reg
3670}
3671</pre>
3672</div>
3673
3674<h5>Semantics:</h5>
3675
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3676<p>In the example above, the first index is indexing into the '<tt>%ST*</tt>'
3677type, which is a pointer, yielding a '<tt>%ST</tt>' = '<tt>{ i32, double, %RT
3678}</tt>' type, a structure. The second index indexes into the third element of
3679the structure, yielding a '<tt>%RT</tt>' = '<tt>{ i8 , [10 x [20 x i32]],
3680i8 }</tt>' type, another structure. The third index indexes into the second
3681element of the structure, yielding a '<tt>[10 x [20 x i32]]</tt>' type, an
3682array. The two dimensions of the array are subscripted into, yielding an
3683'<tt>i32</tt>' type. The '<tt>getelementptr</tt>' instruction returns a pointer
3684to this element, thus computing a value of '<tt>i32*</tt>' type.</p>
3685
3686<p>Note that it is perfectly legal to index partially through a
3687structure, returning a pointer to an inner element. Because of this,
3688the LLVM code for the given testcase is equivalent to:</p>
3689
3690<pre>
3691 define i32* %foo(%ST* %s) {
3692 %t1 = getelementptr %ST* %s, i32 1 <i>; yields %ST*:%t1</i>
3693 %t2 = getelementptr %ST* %t1, i32 0, i32 2 <i>; yields %RT*:%t2</i>
3694 %t3 = getelementptr %RT* %t2, i32 0, i32 1 <i>; yields [10 x [20 x i32]]*:%t3</i>
3695 %t4 = getelementptr [10 x [20 x i32]]* %t3, i32 0, i32 5 <i>; yields [20 x i32]*:%t4</i>
3696 %t5 = getelementptr [20 x i32]* %t4, i32 0, i32 13 <i>; yields i32*:%t5</i>
3697 ret i32* %t5
3698 }
3699</pre>
3700
Chris Lattner50609942009-03-09 20:55:18 +0000[diff] [blame]3701<p>Note that it is undefined to access an array out of bounds: array
3702and pointer indexes must always be within the defined bounds of the
3703array type when accessed with an instruction that dereferences the
3704pointer (e.g. a load or store instruction). The one exception for
3705this rule is zero length arrays. These arrays are defined to be
3706accessible as variable length arrays, which requires access beyond the
3707zero'th element.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3708
3709<p>The getelementptr instruction is often confusing. For some more insight
3710into how it works, see <a href="GetElementPtr.html">the getelementptr
3711FAQ</a>.</p>
3712
3713<h5>Example:</h5>
3714
3715<pre>
3716 <i>; yields [12 x i8]*:aptr</i>
Matthijs Kooijman32a080e2008-10-13 13:44:15 +0000[diff] [blame]3717 %aptr = getelementptr {i32, [12 x i8]}* %saptr, i64 0, i32 1
3718 <i>; yields i8*:vptr</i>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]3719 %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]3720 <i>; yields i8*:eptr</i>
3721 %eptr = getelementptr [12 x i8]* %aptr, i64 0, i32 1
Sanjiv Gupta4f9a0dc2009-04-25 07:27:44 +0000[diff] [blame]3722 <i>; yields i32*:iptr</i>
Sanjiv Gupta1e46c582009-04-24 16:38:13 +0000[diff] [blame]3723 %iptr = getelementptr [10 x i32]* @arr, i16 0, i16 0
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3724</pre>
3725</div>
3726
3727<!-- ======================================================================= -->
3728<div class="doc_subsection"> <a name="convertops">Conversion Operations</a>
3729</div>
3730<div class="doc_text">
3731<p>The instructions in this category are the conversion instructions (casting)
3732which all take a single operand and a type. They perform various bit conversions
3733on the operand.</p>
3734</div>
3735
3736<!-- _______________________________________________________________________ -->
3737<div class="doc_subsubsection">
3738 <a name="i_trunc">'<tt>trunc .. to</tt>' Instruction</a>
3739</div>
3740<div class="doc_text">
3741
3742<h5>Syntax:</h5>
3743<pre>
3744 &lt;result&gt; = trunc &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3745</pre>
3746
3747<h5>Overview:</h5>
3748<p>
3749The '<tt>trunc</tt>' instruction truncates its operand to the type <tt>ty2</tt>.
3750</p>
3751
3752<h5>Arguments:</h5>
3753<p>
3754The '<tt>trunc</tt>' instruction takes a <tt>value</tt> to trunc, which must
3755be an <a href="#t_integer">integer</a> type, and a type that specifies the size
3756and type of the result, which must be an <a href="#t_integer">integer</a>
3757type. The bit size of <tt>value</tt> must be larger than the bit size of
3758<tt>ty2</tt>. Equal sized types are not allowed.</p>
3759
3760<h5>Semantics:</h5>
3761<p>
3762The '<tt>trunc</tt>' instruction truncates the high order bits in <tt>value</tt>
3763and converts the remaining bits to <tt>ty2</tt>. Since the source size must be
3764larger than the destination size, <tt>trunc</tt> cannot be a <i>no-op cast</i>.
3765It will always truncate bits.</p>
3766
3767<h5>Example:</h5>
3768<pre>
3769 %X = trunc i32 257 to i8 <i>; yields i8:1</i>
3770 %Y = trunc i32 123 to i1 <i>; yields i1:true</i>
3771 %Y = trunc i32 122 to i1 <i>; yields i1:false</i>
3772</pre>
3773</div>
3774
3775<!-- _______________________________________________________________________ -->
3776<div class="doc_subsubsection">
3777 <a name="i_zext">'<tt>zext .. to</tt>' Instruction</a>
3778</div>
3779<div class="doc_text">
3780
3781<h5>Syntax:</h5>
3782<pre>
3783 &lt;result&gt; = zext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3784</pre>
3785
3786<h5>Overview:</h5>
3787<p>The '<tt>zext</tt>' instruction zero extends its operand to type
3788<tt>ty2</tt>.</p>
3789
3790
3791<h5>Arguments:</h5>
3792<p>The '<tt>zext</tt>' instruction takes a value to cast, which must be of
3793<a href="#t_integer">integer</a> type, and a type to cast it to, which must
3794also be of <a href="#t_integer">integer</a> type. The bit size of the
3795<tt>value</tt> must be smaller than the bit size of the destination type,
3796<tt>ty2</tt>.</p>
3797
3798<h5>Semantics:</h5>
3799<p>The <tt>zext</tt> fills the high order bits of the <tt>value</tt> with zero
3800bits until it reaches the size of the destination type, <tt>ty2</tt>.</p>
3801
3802<p>When zero extending from i1, the result will always be either 0 or 1.</p>
3803
3804<h5>Example:</h5>
3805<pre>
3806 %X = zext i32 257 to i64 <i>; yields i64:257</i>
3807 %Y = zext i1 true to i32 <i>; yields i32:1</i>
3808</pre>
3809</div>
3810
3811<!-- _______________________________________________________________________ -->
3812<div class="doc_subsubsection">
3813 <a name="i_sext">'<tt>sext .. to</tt>' Instruction</a>
3814</div>
3815<div class="doc_text">
3816
3817<h5>Syntax:</h5>
3818<pre>
3819 &lt;result&gt; = sext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3820</pre>
3821
3822<h5>Overview:</h5>
3823<p>The '<tt>sext</tt>' sign extends <tt>value</tt> to the type <tt>ty2</tt>.</p>
3824
3825<h5>Arguments:</h5>
3826<p>
3827The '<tt>sext</tt>' instruction takes a value to cast, which must be of
3828<a href="#t_integer">integer</a> type, and a type to cast it to, which must
3829also be of <a href="#t_integer">integer</a> type. The bit size of the
3830<tt>value</tt> must be smaller than the bit size of the destination type,
3831<tt>ty2</tt>.</p>
3832
3833<h5>Semantics:</h5>
3834<p>
3835The '<tt>sext</tt>' instruction performs a sign extension by copying the sign
3836bit (highest order bit) of the <tt>value</tt> until it reaches the bit size of
3837the type <tt>ty2</tt>.</p>
3838
3839<p>When sign extending from i1, the extension always results in -1 or 0.</p>
3840
3841<h5>Example:</h5>
3842<pre>
3843 %X = sext i8 -1 to i16 <i>; yields i16 :65535</i>
3844 %Y = sext i1 true to i32 <i>; yields i32:-1</i>
3845</pre>
3846</div>
3847
3848<!-- _______________________________________________________________________ -->
3849<div class="doc_subsubsection">
3850 <a name="i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a>
3851</div>
3852
3853<div class="doc_text">
3854
3855<h5>Syntax:</h5>
3856
3857<pre>
3858 &lt;result&gt; = fptrunc &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3859</pre>
3860
3861<h5>Overview:</h5>
3862<p>The '<tt>fptrunc</tt>' instruction truncates <tt>value</tt> to type
3863<tt>ty2</tt>.</p>
3864
3865
3866<h5>Arguments:</h5>
3867<p>The '<tt>fptrunc</tt>' instruction takes a <a href="#t_floating">floating
3868 point</a> value to cast and a <a href="#t_floating">floating point</a> type to
3869cast it to. The size of <tt>value</tt> must be larger than the size of
3870<tt>ty2</tt>. This implies that <tt>fptrunc</tt> cannot be used to make a
3871<i>no-op cast</i>.</p>
3872
3873<h5>Semantics:</h5>
3874<p> The '<tt>fptrunc</tt>' instruction truncates a <tt>value</tt> from a larger
3875<a href="#t_floating">floating point</a> type to a smaller
3876<a href="#t_floating">floating point</a> type. If the value cannot fit within
3877the destination type, <tt>ty2</tt>, then the results are undefined.</p>
3878
3879<h5>Example:</h5>
3880<pre>
3881 %X = fptrunc double 123.0 to float <i>; yields float:123.0</i>
3882 %Y = fptrunc double 1.0E+300 to float <i>; yields undefined</i>
3883</pre>
3884</div>
3885
3886<!-- _______________________________________________________________________ -->
3887<div class="doc_subsubsection">
3888 <a name="i_fpext">'<tt>fpext .. to</tt>' Instruction</a>
3889</div>
3890<div class="doc_text">
3891
3892<h5>Syntax:</h5>
3893<pre>
3894 &lt;result&gt; = fpext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3895</pre>
3896
3897<h5>Overview:</h5>
3898<p>The '<tt>fpext</tt>' extends a floating point <tt>value</tt> to a larger
3899floating point value.</p>
3900
3901<h5>Arguments:</h5>
3902<p>The '<tt>fpext</tt>' instruction takes a
3903<a href="#t_floating">floating point</a> <tt>value</tt> to cast,
3904and a <a href="#t_floating">floating point</a> type to cast it to. The source
3905type must be smaller than the destination type.</p>
3906
3907<h5>Semantics:</h5>
3908<p>The '<tt>fpext</tt>' instruction extends the <tt>value</tt> from a smaller
3909<a href="#t_floating">floating point</a> type to a larger
3910<a href="#t_floating">floating point</a> type. The <tt>fpext</tt> cannot be
3911used to make a <i>no-op cast</i> because it always changes bits. Use
3912<tt>bitcast</tt> to make a <i>no-op cast</i> for a floating point cast.</p>
3913
3914<h5>Example:</h5>
3915<pre>
3916 %X = fpext float 3.1415 to double <i>; yields double:3.1415</i>
3917 %Y = fpext float 1.0 to float <i>; yields float:1.0 (no-op)</i>
3918</pre>
3919</div>
3920
3921<!-- _______________________________________________________________________ -->
3922<div class="doc_subsubsection">
3923 <a name="i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a>
3924</div>
3925<div class="doc_text">
3926
3927<h5>Syntax:</h5>
3928<pre>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3929 &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]3930</pre>
3931
3932<h5>Overview:</h5>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3933<p>The '<tt>fptoui</tt>' converts a floating point <tt>value</tt> to its
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3934unsigned integer equivalent of type <tt>ty2</tt>.
3935</p>
3936
3937<h5>Arguments:</h5>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3938<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]3939scalar or vector <a href="#t_floating">floating point</a> value, and a type
3940to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
3941type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
3942vector integer type with the same number of elements as <tt>ty</tt></p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3943
3944<h5>Semantics:</h5>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3945<p> The '<tt>fptoui</tt>' instruction converts its
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3946<a href="#t_floating">floating point</a> operand into the nearest (rounding
3947towards zero) unsigned integer value. If the value cannot fit in <tt>ty2</tt>,
3948the results are undefined.</p>
3949
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3950<h5>Example:</h5>
3951<pre>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3952 %X = fptoui double 123.0 to i32 <i>; yields i32:123</i>
Chris Lattner681f1e82007-09-22 03:17:52 +0000[diff] [blame]3953 %Y = fptoui float 1.0E+300 to i1 <i>; yields undefined:1</i>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3954 %X = fptoui float 1.04E+17 to i8 <i>; yields undefined:1</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3955</pre>
3956</div>
3957
3958<!-- _______________________________________________________________________ -->
3959<div class="doc_subsubsection">
3960 <a name="i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a>
3961</div>
3962<div class="doc_text">
3963
3964<h5>Syntax:</h5>
3965<pre>
3966 &lt;result&gt; = fptosi &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3967</pre>
3968
3969<h5>Overview:</h5>
3970<p>The '<tt>fptosi</tt>' instruction converts
3971<a href="#t_floating">floating point</a> <tt>value</tt> to type <tt>ty2</tt>.
3972</p>
3973
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3974<h5>Arguments:</h5>
3975<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]3976scalar or vector <a href="#t_floating">floating point</a> value, and a type
3977to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
3978type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
3979vector integer type with the same number of elements as <tt>ty</tt></p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3980
3981<h5>Semantics:</h5>
3982<p>The '<tt>fptosi</tt>' instruction converts its
3983<a href="#t_floating">floating point</a> operand into the nearest (rounding
3984towards zero) signed integer value. If the value cannot fit in <tt>ty2</tt>,
3985the results are undefined.</p>
3986
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3987<h5>Example:</h5>
3988<pre>
3989 %X = fptosi double -123.0 to i32 <i>; yields i32:-123</i>
Chris Lattner681f1e82007-09-22 03:17:52 +0000[diff] [blame]3990 %Y = fptosi float 1.0E-247 to i1 <i>; yields undefined:1</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3991 %X = fptosi float 1.04E+17 to i8 <i>; yields undefined:1</i>
3992</pre>
3993</div>
3994
3995<!-- _______________________________________________________________________ -->
3996<div class="doc_subsubsection">
3997 <a name="i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a>
3998</div>
3999<div class="doc_text">
4000
4001<h5>Syntax:</h5>
4002<pre>
4003 &lt;result&gt; = uitofp &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
4004</pre>
4005
4006<h5>Overview:</h5>
4007<p>The '<tt>uitofp</tt>' instruction regards <tt>value</tt> as an unsigned
4008integer and converts that value to the <tt>ty2</tt> type.</p>
4009
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4010<h5>Arguments:</h5>
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]4011<p>The '<tt>uitofp</tt>' instruction takes a value to cast, which must be a
4012scalar or vector <a href="#t_integer">integer</a> value, and a type to cast it
4013to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
4014type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
4015floating point type with the same number of elements as <tt>ty</tt></p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4016
4017<h5>Semantics:</h5>
4018<p>The '<tt>uitofp</tt>' instruction interprets its operand as an unsigned
4019integer quantity and converts it to the corresponding floating point value. If
4020the value cannot fit in the floating point value, the results are undefined.</p>
4021
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4022<h5>Example:</h5>
4023<pre>
4024 %X = uitofp i32 257 to float <i>; yields float:257.0</i>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4025 %Y = uitofp i8 -1 to double <i>; yields double:255.0</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4026</pre>
4027</div>
4028
4029<!-- _______________________________________________________________________ -->
4030<div class="doc_subsubsection">
4031 <a name="i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a>
4032</div>
4033<div class="doc_text">
4034
4035<h5>Syntax:</h5>
4036<pre>
4037 &lt;result&gt; = sitofp &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
4038</pre>
4039
4040<h5>Overview:</h5>
4041<p>The '<tt>sitofp</tt>' instruction regards <tt>value</tt> as a signed
4042integer and converts that value to the <tt>ty2</tt> type.</p>
4043
4044<h5>Arguments:</h5>
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]4045<p>The '<tt>sitofp</tt>' instruction takes a value to cast, which must be a
4046scalar or vector <a href="#t_integer">integer</a> value, and a type to cast it
4047to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
4048type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
4049floating point type with the same number of elements as <tt>ty</tt></p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4050
4051<h5>Semantics:</h5>
4052<p>The '<tt>sitofp</tt>' instruction interprets its operand as a signed
4053integer quantity and converts it to the corresponding floating point value. If
4054the value cannot fit in the floating point value, the results are undefined.</p>
4055
4056<h5>Example:</h5>
4057<pre>
4058 %X = sitofp i32 257 to float <i>; yields float:257.0</i>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4059 %Y = sitofp i8 -1 to double <i>; yields double:-1.0</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4060</pre>
4061</div>
4062
4063<!-- _______________________________________________________________________ -->
4064<div class="doc_subsubsection">
4065 <a name="i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a>
4066</div>
4067<div class="doc_text">
4068
4069<h5>Syntax:</h5>
4070<pre>
4071 &lt;result&gt; = ptrtoint &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
4072</pre>
4073
4074<h5>Overview:</h5>
4075<p>The '<tt>ptrtoint</tt>' instruction converts the pointer <tt>value</tt> to
4076the integer type <tt>ty2</tt>.</p>
4077
4078<h5>Arguments:</h5>
4079<p>The '<tt>ptrtoint</tt>' instruction takes a <tt>value</tt> to cast, which
4080must 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]4081<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]4082
4083<h5>Semantics:</h5>
4084<p>The '<tt>ptrtoint</tt>' instruction converts <tt>value</tt> to integer type
4085<tt>ty2</tt> by interpreting the pointer value as an integer and either
4086truncating or zero extending that value to the size of the integer type. If
4087<tt>value</tt> is smaller than <tt>ty2</tt> then a zero extension is done. If
4088<tt>value</tt> is larger than <tt>ty2</tt> then a truncation is done. If they
4089are the same size, then nothing is done (<i>no-op cast</i>) other than a type
4090change.</p>
4091
4092<h5>Example:</h5>
4093<pre>
4094 %X = ptrtoint i32* %X to i8 <i>; yields truncation on 32-bit architecture</i>
4095 %Y = ptrtoint i32* %x to i64 <i>; yields zero extension on 32-bit architecture</i>
4096</pre>
4097</div>
4098
4099<!-- _______________________________________________________________________ -->
4100<div class="doc_subsubsection">
4101 <a name="i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a>
4102</div>
4103<div class="doc_text">
4104
4105<h5>Syntax:</h5>
4106<pre>
4107 &lt;result&gt; = inttoptr &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
4108</pre>
4109
4110<h5>Overview:</h5>
4111<p>The '<tt>inttoptr</tt>' instruction converts an integer <tt>value</tt> to
4112a pointer type, <tt>ty2</tt>.</p>
4113
4114<h5>Arguments:</h5>
4115<p>The '<tt>inttoptr</tt>' instruction takes an <a href="#t_integer">integer</a>
4116value to cast, and a type to cast it to, which must be a
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4117<a href="#t_pointer">pointer</a> type.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4118
4119<h5>Semantics:</h5>
4120<p>The '<tt>inttoptr</tt>' instruction converts <tt>value</tt> to type
4121<tt>ty2</tt> by applying either a zero extension or a truncation depending on
4122the size of the integer <tt>value</tt>. If <tt>value</tt> is larger than the
4123size of a pointer then a truncation is done. If <tt>value</tt> is smaller than
4124the size of a pointer then a zero extension is done. If they are the same size,
4125nothing is done (<i>no-op cast</i>).</p>
4126
4127<h5>Example:</h5>
4128<pre>
4129 %X = inttoptr i32 255 to i32* <i>; yields zero extension on 64-bit architecture</i>
4130 %X = inttoptr i32 255 to i32* <i>; yields no-op on 32-bit architecture</i>
4131 %Y = inttoptr i64 0 to i32* <i>; yields truncation on 32-bit architecture</i>
4132</pre>
4133</div>
4134
4135<!-- _______________________________________________________________________ -->
4136<div class="doc_subsubsection">
4137 <a name="i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a>
4138</div>
4139<div class="doc_text">
4140
4141<h5>Syntax:</h5>
4142<pre>
4143 &lt;result&gt; = bitcast &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
4144</pre>
4145
4146<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4147
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4148<p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
4149<tt>ty2</tt> without changing any bits.</p>
4150
4151<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4152
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4153<p>The '<tt>bitcast</tt>' instruction takes a value to cast, which must be
Dan Gohman7305fa02008-09-08 16:45:59 +0000[diff] [blame]4154a non-aggregate first class value, and a type to cast it to, which must also be
4155a non-aggregate <a href="#t_firstclass">first class</a> type. The bit sizes of
4156<tt>value</tt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4157and the destination type, <tt>ty2</tt>, must be identical. If the source
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4158type is a pointer, the destination type must also be a pointer. This
4159instruction supports bitwise conversion of vectors to integers and to vectors
4160of other types (as long as they have the same size).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4161
4162<h5>Semantics:</h5>
4163<p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
4164<tt>ty2</tt>. It is always a <i>no-op cast</i> because no bits change with
4165this conversion. The conversion is done as if the <tt>value</tt> had been
4166stored to memory and read back as type <tt>ty2</tt>. Pointer types may only be
4167converted to other pointer types with this instruction. To convert pointers to
4168other types, use the <a href="#i_inttoptr">inttoptr</a> or
4169<a href="#i_ptrtoint">ptrtoint</a> instructions first.</p>
4170
4171<h5>Example:</h5>
4172<pre>
4173 %X = bitcast i8 255 to i8 <i>; yields i8 :-1</i>
4174 %Y = bitcast i32* %x to sint* <i>; yields sint*:%x</i>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4175 %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]4176</pre>
4177</div>
4178
4179<!-- ======================================================================= -->
4180<div class="doc_subsection"> <a name="otherops">Other Operations</a> </div>
4181<div class="doc_text">
4182<p>The instructions in this category are the "miscellaneous"
4183instructions, which defy better classification.</p>
4184</div>
4185
4186<!-- _______________________________________________________________________ -->
4187<div class="doc_subsubsection"><a name="i_icmp">'<tt>icmp</tt>' Instruction</a>
4188</div>
4189<div class="doc_text">
4190<h5>Syntax:</h5>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4191<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]4192</pre>
4193<h5>Overview:</h5>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4194<p>The '<tt>icmp</tt>' instruction returns a boolean value or
4195a vector of boolean values based on comparison
4196of its two integer, integer vector, or pointer operands.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4197<h5>Arguments:</h5>
4198<p>The '<tt>icmp</tt>' instruction takes three operands. The first operand is
4199the condition code indicating the kind of comparison to perform. It is not
4200a value, just a keyword. The possible condition code are:
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4201</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4202<ol>
4203 <li><tt>eq</tt>: equal</li>
4204 <li><tt>ne</tt>: not equal </li>
4205 <li><tt>ugt</tt>: unsigned greater than</li>
4206 <li><tt>uge</tt>: unsigned greater or equal</li>
4207 <li><tt>ult</tt>: unsigned less than</li>
4208 <li><tt>ule</tt>: unsigned less or equal</li>
4209 <li><tt>sgt</tt>: signed greater than</li>
4210 <li><tt>sge</tt>: signed greater or equal</li>
4211 <li><tt>slt</tt>: signed less than</li>
4212 <li><tt>sle</tt>: signed less or equal</li>
4213</ol>
4214<p>The remaining two arguments must be <a href="#t_integer">integer</a> or
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4215<a href="#t_pointer">pointer</a>
4216or integer <a href="#t_vector">vector</a> typed.
4217They must also be identical types.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4218<h5>Semantics:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4219<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]4220the condition code given as <tt>cond</tt>. The comparison performed always
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4221yields 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]4222</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4223<ol>
4224 <li><tt>eq</tt>: yields <tt>true</tt> if the operands are equal,
4225 <tt>false</tt> otherwise. No sign interpretation is necessary or performed.
4226 </li>
4227 <li><tt>ne</tt>: yields <tt>true</tt> if the operands are unequal,
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4228 <tt>false</tt> otherwise. No sign interpretation is necessary or performed.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4229 <li><tt>ugt</tt>: interprets the operands as unsigned values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4230 <tt>true</tt> if <tt>op1</tt> is greater than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4231 <li><tt>uge</tt>: interprets the operands as unsigned values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4232 <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]4233 <li><tt>ult</tt>: interprets the operands as unsigned values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4234 <tt>true</tt> if <tt>op1</tt> is less than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4235 <li><tt>ule</tt>: interprets the operands as unsigned values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4236 <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]4237 <li><tt>sgt</tt>: interprets the operands as signed values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4238 <tt>true</tt> if <tt>op1</tt> is greater than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4239 <li><tt>sge</tt>: interprets the operands as signed values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4240 <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]4241 <li><tt>slt</tt>: interprets the operands as signed values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4242 <tt>true</tt> if <tt>op1</tt> is less than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4243 <li><tt>sle</tt>: interprets the operands as signed values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4244 <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]4245</ol>
4246<p>If the operands are <a href="#t_pointer">pointer</a> typed, the pointer
4247values are compared as if they were integers.</p>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4248<p>If the operands are integer vectors, then they are compared
4249element by element. The result is an <tt>i1</tt> vector with
4250the same number of elements as the values being compared.
4251Otherwise, the result is an <tt>i1</tt>.
4252</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4253
4254<h5>Example:</h5>
4255<pre> &lt;result&gt; = icmp eq i32 4, 5 <i>; yields: result=false</i>
4256 &lt;result&gt; = icmp ne float* %X, %X <i>; yields: result=false</i>
4257 &lt;result&gt; = icmp ult i16 4, 5 <i>; yields: result=true</i>
4258 &lt;result&gt; = icmp sgt i16 4, 5 <i>; yields: result=false</i>
4259 &lt;result&gt; = icmp ule i16 -4, 5 <i>; yields: result=false</i>
4260 &lt;result&gt; = icmp sge i16 4, 5 <i>; yields: result=false</i>
4261</pre>
Dan Gohmana53eb382009-01-22 01:39:38 +0000[diff] [blame]4262
4263<p>Note that the code generator does not yet support vector types with
4264 the <tt>icmp</tt> instruction.</p>
4265
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4266</div>
4267
4268<!-- _______________________________________________________________________ -->
4269<div class="doc_subsubsection"><a name="i_fcmp">'<tt>fcmp</tt>' Instruction</a>
4270</div>
4271<div class="doc_text">
4272<h5>Syntax:</h5>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4273<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]4274</pre>
4275<h5>Overview:</h5>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4276<p>The '<tt>fcmp</tt>' instruction returns a boolean value
4277or vector of boolean values based on comparison
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4278of its operands.</p>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4279<p>
4280If the operands are floating point scalars, then the result
4281type is a boolean (<a href="#t_primitive"><tt>i1</tt></a>).
4282</p>
4283<p>If the operands are floating point vectors, then the result type
4284is a vector of boolean with the same number of elements as the
4285operands being compared.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4286<h5>Arguments:</h5>
4287<p>The '<tt>fcmp</tt>' instruction takes three operands. The first operand is
4288the condition code indicating the kind of comparison to perform. It is not
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4289a value, just a keyword. The possible condition code are:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4290<ol>
4291 <li><tt>false</tt>: no comparison, always returns false</li>
4292 <li><tt>oeq</tt>: ordered and equal</li>
4293 <li><tt>ogt</tt>: ordered and greater than </li>
4294 <li><tt>oge</tt>: ordered and greater than or equal</li>
4295 <li><tt>olt</tt>: ordered and less than </li>
4296 <li><tt>ole</tt>: ordered and less than or equal</li>
4297 <li><tt>one</tt>: ordered and not equal</li>
4298 <li><tt>ord</tt>: ordered (no nans)</li>
4299 <li><tt>ueq</tt>: unordered or equal</li>
4300 <li><tt>ugt</tt>: unordered or greater than </li>
4301 <li><tt>uge</tt>: unordered or greater than or equal</li>
4302 <li><tt>ult</tt>: unordered or less than </li>
4303 <li><tt>ule</tt>: unordered or less than or equal</li>
4304 <li><tt>une</tt>: unordered or not equal</li>
4305 <li><tt>uno</tt>: unordered (either nans)</li>
4306 <li><tt>true</tt>: no comparison, always returns true</li>
4307</ol>
4308<p><i>Ordered</i> means that neither operand is a QNAN while
4309<i>unordered</i> means that either operand may be a QNAN.</p>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4310<p>Each of <tt>val1</tt> and <tt>val2</tt> arguments must be
4311either a <a href="#t_floating">floating point</a> type
4312or a <a href="#t_vector">vector</a> of floating point type.
4313They must have identical types.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4314<h5>Semantics:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4315<p>The '<tt>fcmp</tt>' instruction compares <tt>op1</tt> and <tt>op2</tt>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4316according to the condition code given as <tt>cond</tt>.
4317If the operands are vectors, then the vectors are compared
4318element by element.
4319Each comparison performed
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4320always yields an <a href="#t_primitive">i1</a> result, as follows:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4321<ol>
4322 <li><tt>false</tt>: always yields <tt>false</tt>, regardless of operands.</li>
4323 <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]4324 <tt>op1</tt> is equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4325 <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]4326 <tt>op1</tt> is greather than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4327 <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]4328 <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4329 <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]4330 <tt>op1</tt> is less than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4331 <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]4332 <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4333 <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]4334 <tt>op1</tt> is not equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4335 <li><tt>ord</tt>: yields <tt>true</tt> if both operands are not a QNAN.</li>
4336 <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]4337 <tt>op1</tt> is equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4338 <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]4339 <tt>op1</tt> is greater than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4340 <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]4341 <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4342 <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]4343 <tt>op1</tt> is less than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4344 <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]4345 <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4346 <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]4347 <tt>op1</tt> is not equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4348 <li><tt>uno</tt>: yields <tt>true</tt> if either operand is a QNAN.</li>
4349 <li><tt>true</tt>: always yields <tt>true</tt>, regardless of operands.</li>
4350</ol>
4351
4352<h5>Example:</h5>
4353<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]4354 &lt;result&gt; = fcmp one float 4.0, 5.0 <i>; yields: result=true</i>
4355 &lt;result&gt; = fcmp olt float 4.0, 5.0 <i>; yields: result=true</i>
4356 &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]4357</pre>
Dan Gohmana53eb382009-01-22 01:39:38 +0000[diff] [blame]4358
4359<p>Note that the code generator does not yet support vector types with
4360 the <tt>fcmp</tt> instruction.</p>
4361
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4362</div>
4363
4364<!-- _______________________________________________________________________ -->
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]4365<div class="doc_subsubsection">
4366 <a name="i_vicmp">'<tt>vicmp</tt>' Instruction</a>
4367</div>
4368<div class="doc_text">
4369<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4370<pre> &lt;result&gt; = vicmp &lt;cond&gt; &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]4371</pre>
4372<h5>Overview:</h5>
4373<p>The '<tt>vicmp</tt>' instruction returns an integer vector value based on
4374element-wise comparison of its two integer vector operands.</p>
4375<h5>Arguments:</h5>
4376<p>The '<tt>vicmp</tt>' instruction takes three operands. The first operand is
4377the condition code indicating the kind of comparison to perform. It is not
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4378a value, just a keyword. The possible condition code are:</p>
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]4379<ol>
4380 <li><tt>eq</tt>: equal</li>
4381 <li><tt>ne</tt>: not equal </li>
4382 <li><tt>ugt</tt>: unsigned greater than</li>
4383 <li><tt>uge</tt>: unsigned greater or equal</li>
4384 <li><tt>ult</tt>: unsigned less than</li>
4385 <li><tt>ule</tt>: unsigned less or equal</li>
4386 <li><tt>sgt</tt>: signed greater than</li>
4387 <li><tt>sge</tt>: signed greater or equal</li>
4388 <li><tt>slt</tt>: signed less than</li>
4389 <li><tt>sle</tt>: signed less or equal</li>
4390</ol>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4391<p>The remaining two arguments must be <a href="#t_vector">vector</a> or
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]4392<a href="#t_integer">integer</a> typed. They must also be identical types.</p>
4393<h5>Semantics:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4394<p>The '<tt>vicmp</tt>' instruction compares <tt>op1</tt> and <tt>op2</tt>
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]4395according to the condition code given as <tt>cond</tt>. The comparison yields a
4396<a href="#t_vector">vector</a> of <a href="#t_integer">integer</a> result, of
4397identical type as the values being compared. The most significant bit in each
4398element is 1 if the element-wise comparison evaluates to true, and is 0
4399otherwise. All other bits of the result are undefined. The condition codes
4400are evaluated identically to the <a href="#i_icmp">'<tt>icmp</tt>'
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4401instruction</a>.</p>
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]4402
4403<h5>Example:</h5>
4404<pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4405 &lt;result&gt; = vicmp eq &lt;2 x i32&gt; &lt; i32 4, i32 0&gt;, &lt; i32 5, i32 0&gt; <i>; yields: result=&lt;2 x i32&gt; &lt; i32 0, i32 -1 &gt;</i>
4406 &lt;result&gt; = vicmp ult &lt;2 x i8 &gt; &lt; i8 1, i8 2&gt;, &lt; i8 2, i8 2 &gt; <i>; yields: result=&lt;2 x i8&gt; &lt; i8 -1, i8 0 &gt;</i>
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]4407</pre>
4408</div>
4409
4410<!-- _______________________________________________________________________ -->
4411<div class="doc_subsubsection">
4412 <a name="i_vfcmp">'<tt>vfcmp</tt>' Instruction</a>
4413</div>
4414<div class="doc_text">
4415<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4416<pre> &lt;result&gt; = vfcmp &lt;cond&gt; &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt;</pre>
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]4417<h5>Overview:</h5>
4418<p>The '<tt>vfcmp</tt>' instruction returns an integer vector value based on
4419element-wise comparison of its two floating point vector operands. The output
4420elements have the same width as the input elements.</p>
4421<h5>Arguments:</h5>
4422<p>The '<tt>vfcmp</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>
Nate Begeman646fa482008-05-12 19:01:56 +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>The remaining two arguments must be <a href="#t_vector">vector</a> of
4444<a href="#t_floating">floating point</a> typed. They must also be identical
4445types.</p>
4446<h5>Semantics:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4447<p>The '<tt>vfcmp</tt>' instruction compares <tt>op1</tt> and <tt>op2</tt>
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]4448according to the condition code given as <tt>cond</tt>. The comparison yields a
4449<a href="#t_vector">vector</a> of <a href="#t_integer">integer</a> result, with
4450an identical number of elements as the values being compared, and each element
4451having identical with to the width of the floating point elements. The most
4452significant bit in each element is 1 if the element-wise comparison evaluates to
4453true, and is 0 otherwise. All other bits of the result are undefined. The
4454condition codes are evaluated identically to the
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4455<a href="#i_fcmp">'<tt>fcmp</tt>' instruction</a>.</p>
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]4456
4457<h5>Example:</h5>
4458<pre>
Chris Lattner1e5c5cd02008-10-13 16:55:18 +0000[diff] [blame]4459 <i>; yields: result=&lt;2 x i32&gt; &lt; i32 0, i32 -1 &gt;</i>
4460 &lt;result&gt; = vfcmp oeq &lt;2 x float&gt; &lt; float 4, float 0 &gt;, &lt; float 5, float 0 &gt;
4461
4462 <i>; yields: result=&lt;2 x i64&gt; &lt; i64 -1, i64 0 &gt;</i>
4463 &lt;result&gt; = vfcmp ult &lt;2 x double&gt; &lt; double 1, double 2 &gt;, &lt; double 2, double 2&gt;
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]4464</pre>
4465</div>
4466
4467<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4468<div class="doc_subsubsection">
4469 <a name="i_phi">'<tt>phi</tt>' Instruction</a>
4470</div>
4471
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4472<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4473
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4474<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4475
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4476<pre> &lt;result&gt; = phi &lt;ty&gt; [ &lt;val0&gt;, &lt;label0&gt;], ...<br></pre>
4477<h5>Overview:</h5>
4478<p>The '<tt>phi</tt>' instruction is used to implement the &#966; node in
4479the SSA graph representing the function.</p>
4480<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4481
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4482<p>The type of the incoming values is specified with the first type
4483field. After this, the '<tt>phi</tt>' instruction takes a list of pairs
4484as arguments, with one pair for each predecessor basic block of the
4485current block. Only values of <a href="#t_firstclass">first class</a>
4486type may be used as the value arguments to the PHI node. Only labels
4487may be used as the label arguments.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4488
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4489<p>There must be no non-phi instructions between the start of a basic
4490block and the PHI instructions: i.e. PHI instructions must be first in
4491a basic block.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4492
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4493<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4494
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4495<p>At runtime, the '<tt>phi</tt>' instruction logically takes on the value
4496specified by the pair corresponding to the predecessor basic block that executed
4497just prior to the current block.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4498
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4499<h5>Example:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4500<pre>
4501Loop: ; Infinite loop that counts from 0 on up...
4502 %indvar = phi i32 [ 0, %LoopHeader ], [ %nextindvar, %Loop ]
4503 %nextindvar = add i32 %indvar, 1
4504 br label %Loop
4505</pre>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4506</div>
4507
4508<!-- _______________________________________________________________________ -->
4509<div class="doc_subsubsection">
4510 <a name="i_select">'<tt>select</tt>' Instruction</a>
4511</div>
4512
4513<div class="doc_text">
4514
4515<h5>Syntax:</h5>
4516
4517<pre>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4518 &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>
4519
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4520 <i>selty</i> is either i1 or {&lt;N x i1&gt;}
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4521</pre>
4522
4523<h5>Overview:</h5>
4524
4525<p>
4526The '<tt>select</tt>' instruction is used to choose one value based on a
4527condition, without branching.
4528</p>
4529
4530
4531<h5>Arguments:</h5>
4532
4533<p>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4534The '<tt>select</tt>' instruction requires an 'i1' value or
4535a vector of 'i1' values indicating the
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4536condition, and two values of the same <a href="#t_firstclass">first class</a>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4537type. If the val1/val2 are vectors and
4538the condition is a scalar, then entire vectors are selected, not
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4539individual elements.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4540</p>
4541
4542<h5>Semantics:</h5>
4543
4544<p>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4545If 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]4546value argument; otherwise, it returns the second value argument.
4547</p>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4548<p>
4549If the condition is a vector of i1, then the value arguments must
4550be vectors of the same size, and the selection is done element
4551by element.
4552</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4553
4554<h5>Example:</h5>
4555
4556<pre>
4557 %X = select i1 true, i8 17, i8 42 <i>; yields i8:17</i>
4558</pre>
Dan Gohmana53eb382009-01-22 01:39:38 +0000[diff] [blame]4559
4560<p>Note that the code generator does not yet support conditions
4561 with vector type.</p>
4562
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4563</div>
4564
4565
4566<!-- _______________________________________________________________________ -->
4567<div class="doc_subsubsection">
4568 <a name="i_call">'<tt>call</tt>' Instruction</a>
4569</div>
4570
4571<div class="doc_text">
4572
4573<h5>Syntax:</h5>
4574<pre>
Devang Pateld0bfcc72008-10-07 17:48:33 +0000[diff] [blame]4575 &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]4576</pre>
4577
4578<h5>Overview:</h5>
4579
4580<p>The '<tt>call</tt>' instruction represents a simple function call.</p>
4581
4582<h5>Arguments:</h5>
4583
4584<p>This instruction requires several arguments:</p>
4585
4586<ol>
4587 <li>
4588 <p>The optional "tail" marker indicates whether the callee function accesses
4589 any allocas or varargs in the caller. If the "tail" marker is present, the
4590 function call is eligible for tail call optimization. Note that calls may
4591 be marked "tail" even if they do not occur before a <a
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4592 href="#i_ret"><tt>ret</tt></a> instruction.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4593 </li>
4594 <li>
4595 <p>The optional "cconv" marker indicates which <a href="#callingconv">calling
4596 convention</a> the call should use. If none is specified, the call defaults
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4597 to using C calling conventions.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4598 </li>
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]4599
4600 <li>
4601 <p>The optional <a href="#paramattrs">Parameter Attributes</a> list for
4602 return values. Only '<tt>zeroext</tt>', '<tt>signext</tt>',
4603 and '<tt>inreg</tt>' attributes are valid here.</p>
4604 </li>
4605
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4606 <li>
Nick Lewycky93082fc2007-09-08 13:57:50 +0000[diff] [blame]4607 <p>'<tt>ty</tt>': the type of the call instruction itself which is also
4608 the type of the return value. Functions that return no value are marked
4609 <tt><a href="#t_void">void</a></tt>.</p>
4610 </li>
4611 <li>
4612 <p>'<tt>fnty</tt>': shall be the signature of the pointer to function
4613 value being invoked. The argument types must match the types implied by
4614 this signature. This type can be omitted if the function is not varargs
4615 and if the function type does not return a pointer to a function.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4616 </li>
4617 <li>
4618 <p>'<tt>fnptrval</tt>': An LLVM value containing a pointer to a function to
4619 be invoked. In most cases, this is a direct function invocation, but
4620 indirect <tt>call</tt>s are just as possible, calling an arbitrary pointer
4621 to function value.</p>
4622 </li>
4623 <li>
4624 <p>'<tt>function args</tt>': argument list whose types match the
4625 function signature argument types. All arguments must be of
4626 <a href="#t_firstclass">first class</a> type. If the function signature
4627 indicates the function accepts a variable number of arguments, the extra
4628 arguments can be specified.</p>
4629 </li>
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]4630 <li>
Devang Pateld0bfcc72008-10-07 17:48:33 +0000[diff] [blame]4631 <p>The optional <a href="#fnattrs">function attributes</a> list. Only
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]4632 '<tt>noreturn</tt>', '<tt>nounwind</tt>', '<tt>readonly</tt>' and
4633 '<tt>readnone</tt>' attributes are valid here.</p>
4634 </li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4635</ol>
4636
4637<h5>Semantics:</h5>
4638
4639<p>The '<tt>call</tt>' instruction is used to cause control flow to
4640transfer to a specified function, with its incoming arguments bound to
4641the specified values. Upon a '<tt><a href="#i_ret">ret</a></tt>'
4642instruction in the called function, control flow continues with the
4643instruction after the function call, and the return value of the
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4644function is bound to the result argument.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4645
4646<h5>Example:</h5>
4647
4648<pre>
Nick Lewycky93082fc2007-09-08 13:57:50 +0000[diff] [blame]4649 %retval = call i32 @test(i32 %argc)
Chris Lattner5e893ef2008-03-21 17:24:17 +0000[diff] [blame]4650 call i32 (i8 *, ...)* @printf(i8 * %msg, i32 12, i8 42) <i>; yields i32</i>
4651 %X = tail call i32 @foo() <i>; yields i32</i>
4652 %Y = tail call <a href="#callingconv">fastcc</a> i32 @foo() <i>; yields i32</i>
4653 call void %foo(i8 97 signext)
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]4654
4655 %struct.A = type { i32, i8 }
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]4656 %r = call %struct.A @foo() <i>; yields { 32, i8 }</i>
Dan Gohman3e700032008-10-04 19:00:07 +0000[diff] [blame]4657 %gr = extractvalue %struct.A %r, 0 <i>; yields i32</i>
4658 %gr1 = extractvalue %struct.A %r, 1 <i>; yields i8</i>
Chris Lattnerac454b32008-10-08 06:26:11 +0000[diff] [blame]4659 %Z = call void @foo() noreturn <i>; indicates that %foo never returns normally</i>
Matthijs Kooijman2c4e05a2008-10-07 10:03:45 +0000[diff] [blame]4660 %ZZ = call zeroext i32 @bar() <i>; Return value is %zero extended</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4661</pre>
4662
4663</div>
4664
4665<!-- _______________________________________________________________________ -->
4666<div class="doc_subsubsection">
4667 <a name="i_va_arg">'<tt>va_arg</tt>' Instruction</a>
4668</div>
4669
4670<div class="doc_text">
4671
4672<h5>Syntax:</h5>
4673
4674<pre>
4675 &lt;resultval&gt; = va_arg &lt;va_list*&gt; &lt;arglist&gt;, &lt;argty&gt;
4676</pre>
4677
4678<h5>Overview:</h5>
4679
4680<p>The '<tt>va_arg</tt>' instruction is used to access arguments passed through
4681the "variable argument" area of a function call. It is used to implement the
4682<tt>va_arg</tt> macro in C.</p>
4683
4684<h5>Arguments:</h5>
4685
4686<p>This instruction takes a <tt>va_list*</tt> value and the type of
4687the argument. It returns a value of the specified argument type and
4688increments the <tt>va_list</tt> to point to the next argument. The
4689actual type of <tt>va_list</tt> is target specific.</p>
4690
4691<h5>Semantics:</h5>
4692
4693<p>The '<tt>va_arg</tt>' instruction loads an argument of the specified
4694type from the specified <tt>va_list</tt> and causes the
4695<tt>va_list</tt> to point to the next argument. For more information,
4696see the variable argument handling <a href="#int_varargs">Intrinsic
4697Functions</a>.</p>
4698
4699<p>It is legal for this instruction to be called in a function which does not
4700take a variable number of arguments, for example, the <tt>vfprintf</tt>
4701function.</p>
4702
4703<p><tt>va_arg</tt> is an LLVM instruction instead of an <a
4704href="#intrinsics">intrinsic function</a> because it takes a type as an
4705argument.</p>
4706
4707<h5>Example:</h5>
4708
4709<p>See the <a href="#int_varargs">variable argument processing</a> section.</p>
4710
Dan Gohman60967192009-01-12 23:12:39 +0000[diff] [blame]4711<p>Note that the code generator does not yet fully support va_arg
4712 on many targets. Also, it does not currently support va_arg with
4713 aggregate types on any target.</p>
4714
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4715</div>
4716
4717<!-- *********************************************************************** -->
4718<div class="doc_section"> <a name="intrinsics">Intrinsic Functions</a> </div>
4719<!-- *********************************************************************** -->
4720
4721<div class="doc_text">
4722
4723<p>LLVM supports the notion of an "intrinsic function". These functions have
4724well known names and semantics and are required to follow certain restrictions.
4725Overall, these intrinsics represent an extension mechanism for the LLVM
4726language that does not require changing all of the transformations in LLVM when
4727adding to the language (or the bitcode reader/writer, the parser, etc...).</p>
4728
4729<p>Intrinsic function names must all start with an "<tt>llvm.</tt>" prefix. This
4730prefix is reserved in LLVM for intrinsic names; thus, function names may not
4731begin with this prefix. Intrinsic functions must always be external functions:
4732you cannot define the body of intrinsic functions. Intrinsic functions may
4733only be used in call or invoke instructions: it is illegal to take the address
4734of an intrinsic function. Additionally, because intrinsic functions are part
4735of the LLVM language, it is required if any are added that they be documented
4736here.</p>
4737
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]4738<p>Some intrinsic functions can be overloaded, i.e., the intrinsic represents
4739a family of functions that perform the same operation but on different data
4740types. Because LLVM can represent over 8 million different integer types,
4741overloading is used commonly to allow an intrinsic function to operate on any
4742integer type. One or more of the argument types or the result type can be
4743overloaded to accept any integer type. Argument types may also be defined as
4744exactly matching a previous argument's type or the result type. This allows an
4745intrinsic function which accepts multiple arguments, but needs all of them to
4746be of the same type, to only be overloaded with respect to a single argument or
4747the result.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4748
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]4749<p>Overloaded intrinsics will have the names of its overloaded argument types
4750encoded into its function name, each preceded by a period. Only those types
4751which are overloaded result in a name suffix. Arguments whose type is matched
4752against another type do not. For example, the <tt>llvm.ctpop</tt> function can
4753take an integer of any width and returns an integer of exactly the same integer
4754width. This leads to a family of functions such as
4755<tt>i8 @llvm.ctpop.i8(i8 %val)</tt> and <tt>i29 @llvm.ctpop.i29(i29 %val)</tt>.
4756Only one type, the return type, is overloaded, and only one type suffix is
4757required. Because the argument's type is matched against the return type, it
4758does not require its own name suffix.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4759
4760<p>To learn how to add an intrinsic function, please see the
4761<a href="ExtendingLLVM.html">Extending LLVM Guide</a>.
4762</p>
4763
4764</div>
4765
4766<!-- ======================================================================= -->
4767<div class="doc_subsection">
4768 <a name="int_varargs">Variable Argument Handling Intrinsics</a>
4769</div>
4770
4771<div class="doc_text">
4772
4773<p>Variable argument support is defined in LLVM with the <a
4774 href="#i_va_arg"><tt>va_arg</tt></a> instruction and these three
4775intrinsic functions. These functions are related to the similarly
4776named macros defined in the <tt>&lt;stdarg.h&gt;</tt> header file.</p>
4777
4778<p>All of these functions operate on arguments that use a
4779target-specific value type "<tt>va_list</tt>". The LLVM assembly
4780language reference manual does not define what this type is, so all
4781transformations should be prepared to handle these functions regardless of
4782the type used.</p>
4783
4784<p>This example shows how the <a href="#i_va_arg"><tt>va_arg</tt></a>
4785instruction and the variable argument handling intrinsic functions are
4786used.</p>
4787
4788<div class="doc_code">
4789<pre>
4790define i32 @test(i32 %X, ...) {
4791 ; Initialize variable argument processing
4792 %ap = alloca i8*
4793 %ap2 = bitcast i8** %ap to i8*
4794 call void @llvm.va_start(i8* %ap2)
4795
4796 ; Read a single integer argument
4797 %tmp = va_arg i8** %ap, i32
4798
4799 ; Demonstrate usage of llvm.va_copy and llvm.va_end
4800 %aq = alloca i8*
4801 %aq2 = bitcast i8** %aq to i8*
4802 call void @llvm.va_copy(i8* %aq2, i8* %ap2)
4803 call void @llvm.va_end(i8* %aq2)
4804
4805 ; Stop processing of arguments.
4806 call void @llvm.va_end(i8* %ap2)
4807 ret i32 %tmp
4808}
4809
4810declare void @llvm.va_start(i8*)
4811declare void @llvm.va_copy(i8*, i8*)
4812declare void @llvm.va_end(i8*)
4813</pre>
4814</div>
4815
4816</div>
4817
4818<!-- _______________________________________________________________________ -->
4819<div class="doc_subsubsection">
4820 <a name="int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a>
4821</div>
4822
4823
4824<div class="doc_text">
4825<h5>Syntax:</h5>
4826<pre> declare void %llvm.va_start(i8* &lt;arglist&gt;)<br></pre>
4827<h5>Overview:</h5>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4828<p>The '<tt>llvm.va_start</tt>' intrinsic initializes
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4829<tt>*&lt;arglist&gt;</tt> for subsequent use by <tt><a
4830href="#i_va_arg">va_arg</a></tt>.</p>
4831
4832<h5>Arguments:</h5>
4833
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4834<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]4835
4836<h5>Semantics:</h5>
4837
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4838<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]4839macro available in C. In a target-dependent way, it initializes the
4840<tt>va_list</tt> element to which the argument points, so that the next call to
4841<tt>va_arg</tt> will produce the first variable argument passed to the function.
4842Unlike the C <tt>va_start</tt> macro, this intrinsic does not need to know the
4843last argument of the function as the compiler can figure that out.</p>
4844
4845</div>
4846
4847<!-- _______________________________________________________________________ -->
4848<div class="doc_subsubsection">
4849 <a name="int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a>
4850</div>
4851
4852<div class="doc_text">
4853<h5>Syntax:</h5>
4854<pre> declare void @llvm.va_end(i8* &lt;arglist&gt;)<br></pre>
4855<h5>Overview:</h5>
4856
4857<p>The '<tt>llvm.va_end</tt>' intrinsic destroys <tt>*&lt;arglist&gt;</tt>,
4858which has been initialized previously with <tt><a href="#int_va_start">llvm.va_start</a></tt>
4859or <tt><a href="#i_va_copy">llvm.va_copy</a></tt>.</p>
4860
4861<h5>Arguments:</h5>
4862
4863<p>The argument is a pointer to a <tt>va_list</tt> to destroy.</p>
4864
4865<h5>Semantics:</h5>
4866
4867<p>The '<tt>llvm.va_end</tt>' intrinsic works just like the <tt>va_end</tt>
4868macro available in C. In a target-dependent way, it destroys the
4869<tt>va_list</tt> element to which the argument points. Calls to <a
4870href="#int_va_start"><tt>llvm.va_start</tt></a> and <a href="#int_va_copy">
4871<tt>llvm.va_copy</tt></a> must be matched exactly with calls to
4872<tt>llvm.va_end</tt>.</p>
4873
4874</div>
4875
4876<!-- _______________________________________________________________________ -->
4877<div class="doc_subsubsection">
4878 <a name="int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a>
4879</div>
4880
4881<div class="doc_text">
4882
4883<h5>Syntax:</h5>
4884
4885<pre>
4886 declare void @llvm.va_copy(i8* &lt;destarglist&gt;, i8* &lt;srcarglist&gt;)
4887</pre>
4888
4889<h5>Overview:</h5>
4890
4891<p>The '<tt>llvm.va_copy</tt>' intrinsic copies the current argument position
4892from the source argument list to the destination argument list.</p>
4893
4894<h5>Arguments:</h5>
4895
4896<p>The first argument is a pointer to a <tt>va_list</tt> element to initialize.
4897The second argument is a pointer to a <tt>va_list</tt> element to copy from.</p>
4898
4899
4900<h5>Semantics:</h5>
4901
4902<p>The '<tt>llvm.va_copy</tt>' intrinsic works just like the <tt>va_copy</tt>
4903macro available in C. In a target-dependent way, it copies the source
4904<tt>va_list</tt> element into the destination <tt>va_list</tt> element. This
4905intrinsic is necessary because the <tt><a href="#int_va_start">
4906llvm.va_start</a></tt> intrinsic may be arbitrarily complex and require, for
4907example, memory allocation.</p>
4908
4909</div>
4910
4911<!-- ======================================================================= -->
4912<div class="doc_subsection">
4913 <a name="int_gc">Accurate Garbage Collection Intrinsics</a>
4914</div>
4915
4916<div class="doc_text">
4917
4918<p>
4919LLVM support for <a href="GarbageCollection.html">Accurate Garbage
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]4920Collection</a> (GC) requires the implementation and generation of these
4921intrinsics.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4922These intrinsics allow identification of <a href="#int_gcroot">GC roots on the
4923stack</a>, as well as garbage collector implementations that require <a
4924href="#int_gcread">read</a> and <a href="#int_gcwrite">write</a> barriers.
4925Front-ends for type-safe garbage collected languages should generate these
4926intrinsics to make use of the LLVM garbage collectors. For more details, see <a
4927href="GarbageCollection.html">Accurate Garbage Collection with LLVM</a>.
4928</p>
Christopher Lambcfe00962007-12-17 01:00:21 +0000[diff] [blame]4929
4930<p>The garbage collection intrinsics only operate on objects in the generic
4931 address space (address space zero).</p>
4932
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4933</div>
4934
4935<!-- _______________________________________________________________________ -->
4936<div class="doc_subsubsection">
4937 <a name="int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a>
4938</div>
4939
4940<div class="doc_text">
4941
4942<h5>Syntax:</h5>
4943
4944<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4945 declare void @llvm.gcroot(i8** %ptrloc, i8* %metadata)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4946</pre>
4947
4948<h5>Overview:</h5>
4949
4950<p>The '<tt>llvm.gcroot</tt>' intrinsic declares the existence of a GC root to
4951the code generator, and allows some metadata to be associated with it.</p>
4952
4953<h5>Arguments:</h5>
4954
4955<p>The first argument specifies the address of a stack object that contains the
4956root pointer. The second pointer (which must be either a constant or a global
4957value address) contains the meta-data to be associated with the root.</p>
4958
4959<h5>Semantics:</h5>
4960
Chris Lattnera7d94ba2008-04-24 05:59:56 +0000[diff] [blame]4961<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]4962location. At compile-time, the code generator generates information to allow
Gordon Henriksen40393542007-12-25 02:31:26 +0000[diff] [blame]4963the runtime to find the pointer at GC safe points. The '<tt>llvm.gcroot</tt>'
4964intrinsic may only be used in a function which <a href="#gc">specifies a GC
4965algorithm</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4966
4967</div>
4968
4969
4970<!-- _______________________________________________________________________ -->
4971<div class="doc_subsubsection">
4972 <a name="int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a>
4973</div>
4974
4975<div class="doc_text">
4976
4977<h5>Syntax:</h5>
4978
4979<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4980 declare i8* @llvm.gcread(i8* %ObjPtr, i8** %Ptr)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4981</pre>
4982
4983<h5>Overview:</h5>
4984
4985<p>The '<tt>llvm.gcread</tt>' intrinsic identifies reads of references from heap
4986locations, allowing garbage collector implementations that require read
4987barriers.</p>
4988
4989<h5>Arguments:</h5>
4990
4991<p>The second argument is the address to read from, which should be an address
4992allocated from the garbage collector. The first object is a pointer to the
4993start of the referenced object, if needed by the language runtime (otherwise
4994null).</p>
4995
4996<h5>Semantics:</h5>
4997
4998<p>The '<tt>llvm.gcread</tt>' intrinsic has the same semantics as a load
4999instruction, but may be replaced with substantially more complex code by the
Gordon Henriksen40393542007-12-25 02:31:26 +0000[diff] [blame]5000garbage collector runtime, as needed. The '<tt>llvm.gcread</tt>' intrinsic
5001may 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_gcwrite">'<tt>llvm.gcwrite</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 void @llvm.gcwrite(i8* %P1, i8* %Obj, i8** %P2)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5018</pre>
5019
5020<h5>Overview:</h5>
5021
5022<p>The '<tt>llvm.gcwrite</tt>' intrinsic identifies writes of references to heap
5023locations, allowing garbage collector implementations that require write
5024barriers (such as generational or reference counting collectors).</p>
5025
5026<h5>Arguments:</h5>
5027
5028<p>The first argument is the reference to store, the second is the start of the
5029object to store it to, and the third is the address of the field of Obj to
5030store to. If the runtime does not require a pointer to the object, Obj may be
5031null.</p>
5032
5033<h5>Semantics:</h5>
5034
5035<p>The '<tt>llvm.gcwrite</tt>' intrinsic has the same semantics as a store
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.gcwrite</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<!-- ======================================================================= -->
5046<div class="doc_subsection">
5047 <a name="int_codegen">Code Generator Intrinsics</a>
5048</div>
5049
5050<div class="doc_text">
5051<p>
5052These intrinsics are provided by LLVM to expose special features that may only
5053be implemented with code generator support.
5054</p>
5055
5056</div>
5057
5058<!-- _______________________________________________________________________ -->
5059<div class="doc_subsubsection">
5060 <a name="int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a>
5061</div>
5062
5063<div class="doc_text">
5064
5065<h5>Syntax:</h5>
5066<pre>
5067 declare i8 *@llvm.returnaddress(i32 &lt;level&gt;)
5068</pre>
5069
5070<h5>Overview:</h5>
5071
5072<p>
5073The '<tt>llvm.returnaddress</tt>' intrinsic attempts to compute a
5074target-specific value indicating the return address of the current function
5075or one of its callers.
5076</p>
5077
5078<h5>Arguments:</h5>
5079
5080<p>
5081The argument to this intrinsic indicates which function to return the address
5082for. Zero indicates the calling function, one indicates its caller, etc. The
5083argument is <b>required</b> to be a constant integer value.
5084</p>
5085
5086<h5>Semantics:</h5>
5087
5088<p>
5089The '<tt>llvm.returnaddress</tt>' intrinsic either returns a pointer indicating
5090the return address of the specified call frame, or zero if it cannot be
5091identified. The value returned by this intrinsic is likely to be incorrect or 0
5092for arguments other than zero, so it should only be used for debugging purposes.
5093</p>
5094
5095<p>
5096Note that calling this intrinsic does not prevent function inlining or other
5097aggressive transformations, so the value returned may not be that of the obvious
5098source-language caller.
5099</p>
5100</div>
5101
5102
5103<!-- _______________________________________________________________________ -->
5104<div class="doc_subsubsection">
5105 <a name="int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a>
5106</div>
5107
5108<div class="doc_text">
5109
5110<h5>Syntax:</h5>
5111<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]5112 declare i8 *@llvm.frameaddress(i32 &lt;level&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5113</pre>
5114
5115<h5>Overview:</h5>
5116
5117<p>
5118The '<tt>llvm.frameaddress</tt>' intrinsic attempts to return the
5119target-specific frame pointer value for the specified stack frame.
5120</p>
5121
5122<h5>Arguments:</h5>
5123
5124<p>
5125The argument to this intrinsic indicates which function to return the frame
5126pointer for. Zero indicates the calling function, one indicates its caller,
5127etc. The argument is <b>required</b> to be a constant integer value.
5128</p>
5129
5130<h5>Semantics:</h5>
5131
5132<p>
5133The '<tt>llvm.frameaddress</tt>' intrinsic either returns a pointer indicating
5134the frame address of the specified call frame, or zero if it cannot be
5135identified. The value returned by this intrinsic is likely to be incorrect or 0
5136for arguments other than zero, so it should only be used for debugging purposes.
5137</p>
5138
5139<p>
5140Note that calling this intrinsic does not prevent function inlining or other
5141aggressive transformations, so the value returned may not be that of the obvious
5142source-language caller.
5143</p>
5144</div>
5145
5146<!-- _______________________________________________________________________ -->
5147<div class="doc_subsubsection">
5148 <a name="int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a>
5149</div>
5150
5151<div class="doc_text">
5152
5153<h5>Syntax:</h5>
5154<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]5155 declare i8 *@llvm.stacksave()
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5156</pre>
5157
5158<h5>Overview:</h5>
5159
5160<p>
5161The '<tt>llvm.stacksave</tt>' intrinsic is used to remember the current state of
5162the function stack, for use with <a href="#int_stackrestore">
5163<tt>llvm.stackrestore</tt></a>. This is useful for implementing language
5164features like scoped automatic variable sized arrays in C99.
5165</p>
5166
5167<h5>Semantics:</h5>
5168
5169<p>
5170This intrinsic returns a opaque pointer value that can be passed to <a
5171href="#int_stackrestore"><tt>llvm.stackrestore</tt></a>. When an
5172<tt>llvm.stackrestore</tt> intrinsic is executed with a value saved from
5173<tt>llvm.stacksave</tt>, it effectively restores the state of the stack to the
5174state it was in when the <tt>llvm.stacksave</tt> intrinsic executed. In
5175practice, this pops any <a href="#i_alloca">alloca</a> blocks from the stack
5176that were allocated after the <tt>llvm.stacksave</tt> was executed.
5177</p>
5178
5179</div>
5180
5181<!-- _______________________________________________________________________ -->
5182<div class="doc_subsubsection">
5183 <a name="int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a>
5184</div>
5185
5186<div class="doc_text">
5187
5188<h5>Syntax:</h5>
5189<pre>
5190 declare void @llvm.stackrestore(i8 * %ptr)
5191</pre>
5192
5193<h5>Overview:</h5>
5194
5195<p>
5196The '<tt>llvm.stackrestore</tt>' intrinsic is used to restore the state of
5197the function stack to the state it was in when the corresponding <a
5198href="#int_stacksave"><tt>llvm.stacksave</tt></a> intrinsic executed. This is
5199useful for implementing language features like scoped automatic variable sized
5200arrays in C99.
5201</p>
5202
5203<h5>Semantics:</h5>
5204
5205<p>
5206See the description for <a href="#int_stacksave"><tt>llvm.stacksave</tt></a>.
5207</p>
5208
5209</div>
5210
5211
5212<!-- _______________________________________________________________________ -->
5213<div class="doc_subsubsection">
5214 <a name="int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a>
5215</div>
5216
5217<div class="doc_text">
5218
5219<h5>Syntax:</h5>
5220<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]5221 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]5222</pre>
5223
5224<h5>Overview:</h5>
5225
5226
5227<p>
5228The '<tt>llvm.prefetch</tt>' intrinsic is a hint to the code generator to insert
5229a prefetch instruction if supported; otherwise, it is a noop. Prefetches have
5230no
5231effect on the behavior of the program but can change its performance
5232characteristics.
5233</p>
5234
5235<h5>Arguments:</h5>
5236
5237<p>
5238<tt>address</tt> is the address to be prefetched, <tt>rw</tt> is the specifier
5239determining if the fetch should be for a read (0) or write (1), and
5240<tt>locality</tt> is a temporal locality specifier ranging from (0) - no
5241locality, to (3) - extremely local keep in cache. The <tt>rw</tt> and
5242<tt>locality</tt> arguments must be constant integers.
5243</p>
5244
5245<h5>Semantics:</h5>
5246
5247<p>
5248This intrinsic does not modify the behavior of the program. In particular,
5249prefetches cannot trap and do not produce a value. On targets that support this
5250intrinsic, the prefetch can provide hints to the processor cache for better
5251performance.
5252</p>
5253
5254</div>
5255
5256<!-- _______________________________________________________________________ -->
5257<div class="doc_subsubsection">
5258 <a name="int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a>
5259</div>
5260
5261<div class="doc_text">
5262
5263<h5>Syntax:</h5>
5264<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]5265 declare void @llvm.pcmarker(i32 &lt;id&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5266</pre>
5267
5268<h5>Overview:</h5>
5269
5270
5271<p>
5272The '<tt>llvm.pcmarker</tt>' intrinsic is a method to export a Program Counter
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]5273(PC) in a region of
5274code to simulators and other tools. The method is target specific, but it is
5275expected that the marker will use exported symbols to transmit the PC of the
5276marker.
5277The marker makes no guarantees that it will remain with any specific instruction
5278after optimizations. It is possible that the presence of a marker will inhibit
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5279optimizations. The intended use is to be inserted after optimizations to allow
5280correlations of simulation runs.
5281</p>
5282
5283<h5>Arguments:</h5>
5284
5285<p>
5286<tt>id</tt> is a numerical id identifying the marker.
5287</p>
5288
5289<h5>Semantics:</h5>
5290
5291<p>
5292This intrinsic does not modify the behavior of the program. Backends that do not
5293support this intrinisic may ignore it.
5294</p>
5295
5296</div>
5297
5298<!-- _______________________________________________________________________ -->
5299<div class="doc_subsubsection">
5300 <a name="int_readcyclecounter">'<tt>llvm.readcyclecounter</tt>' Intrinsic</a>
5301</div>
5302
5303<div class="doc_text">
5304
5305<h5>Syntax:</h5>
5306<pre>
5307 declare i64 @llvm.readcyclecounter( )
5308</pre>
5309
5310<h5>Overview:</h5>
5311
5312
5313<p>
5314The '<tt>llvm.readcyclecounter</tt>' intrinsic provides access to the cycle
5315counter register (or similar low latency, high accuracy clocks) on those targets
5316that support it. On X86, it should map to RDTSC. On Alpha, it should map to RPCC.
5317As the backing counters overflow quickly (on the order of 9 seconds on alpha), this
5318should only be used for small timings.
5319</p>
5320
5321<h5>Semantics:</h5>
5322
5323<p>
5324When directly supported, reading the cycle counter should not modify any memory.
5325Implementations are allowed to either return a application specific value or a
5326system wide value. On backends without support, this is lowered to a constant 0.
5327</p>
5328
5329</div>
5330
5331<!-- ======================================================================= -->
5332<div class="doc_subsection">
5333 <a name="int_libc">Standard C Library Intrinsics</a>
5334</div>
5335
5336<div class="doc_text">
5337<p>
5338LLVM provides intrinsics for a few important standard C library functions.
5339These intrinsics allow source-language front-ends to pass information about the
5340alignment of the pointer arguments to the code generator, providing opportunity
5341for more efficient code generation.
5342</p>
5343
5344</div>
5345
5346<!-- _______________________________________________________________________ -->
5347<div class="doc_subsubsection">
5348 <a name="int_memcpy">'<tt>llvm.memcpy</tt>' Intrinsic</a>
5349</div>
5350
5351<div class="doc_text">
5352
5353<h5>Syntax:</h5>
Chris Lattner82c2e432008-11-21 16:42:48 +0000[diff] [blame]5354<p>This is an overloaded intrinsic. You can use llvm.memcpy on any integer bit
5355width. Not all targets support all bit widths however.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5356<pre>
Chris Lattner82c2e432008-11-21 16:42:48 +0000[diff] [blame]5357 declare void @llvm.memcpy.i8(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5358 i8 &lt;len&gt;, i32 &lt;align&gt;)
5359 declare void @llvm.memcpy.i16(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5360 i16 &lt;len&gt;, i32 &lt;align&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5361 declare void @llvm.memcpy.i32(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5362 i32 &lt;len&gt;, i32 &lt;align&gt;)
5363 declare void @llvm.memcpy.i64(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5364 i64 &lt;len&gt;, i32 &lt;align&gt;)
5365</pre>
5366
5367<h5>Overview:</h5>
5368
5369<p>
5370The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the source
5371location to the destination location.
5372</p>
5373
5374<p>
5375Note that, unlike the standard libc function, the <tt>llvm.memcpy.*</tt>
5376intrinsics do not return a value, and takes an extra alignment argument.
5377</p>
5378
5379<h5>Arguments:</h5>
5380
5381<p>
5382The first argument is a pointer to the destination, the second is a pointer to
5383the source. The third argument is an integer argument
5384specifying the number of bytes to copy, and the fourth argument is the alignment
5385of the source and destination locations.
5386</p>
5387
5388<p>
5389If the call to this intrinisic has an alignment value that is not 0 or 1, then
5390the caller guarantees that both the source and destination pointers are aligned
5391to that boundary.
5392</p>
5393
5394<h5>Semantics:</h5>
5395
5396<p>
5397The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the source
5398location to the destination location, which are not allowed to overlap. It
5399copies "len" bytes of memory over. If the argument is known to be aligned to
5400some boundary, this can be specified as the fourth argument, otherwise it should
5401be set to 0 or 1.
5402</p>
5403</div>
5404
5405
5406<!-- _______________________________________________________________________ -->
5407<div class="doc_subsubsection">
5408 <a name="int_memmove">'<tt>llvm.memmove</tt>' Intrinsic</a>
5409</div>
5410
5411<div class="doc_text">
5412
5413<h5>Syntax:</h5>
Chris Lattner82c2e432008-11-21 16:42:48 +0000[diff] [blame]5414<p>This is an overloaded intrinsic. You can use llvm.memmove on any integer bit
5415width. Not all targets support all bit widths however.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5416<pre>
Chris Lattner82c2e432008-11-21 16:42:48 +0000[diff] [blame]5417 declare void @llvm.memmove.i8(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5418 i8 &lt;len&gt;, i32 &lt;align&gt;)
5419 declare void @llvm.memmove.i16(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5420 i16 &lt;len&gt;, i32 &lt;align&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5421 declare void @llvm.memmove.i32(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5422 i32 &lt;len&gt;, i32 &lt;align&gt;)
5423 declare void @llvm.memmove.i64(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5424 i64 &lt;len&gt;, i32 &lt;align&gt;)
5425</pre>
5426
5427<h5>Overview:</h5>
5428
5429<p>
5430The '<tt>llvm.memmove.*</tt>' intrinsics move a block of memory from the source
5431location to the destination location. It is similar to the
Chris Lattnerdba16ea2008-01-06 19:51:52 +0000[diff] [blame]5432'<tt>llvm.memcpy</tt>' intrinsic but allows the two memory locations to overlap.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5433</p>
5434
5435<p>
5436Note that, unlike the standard libc function, the <tt>llvm.memmove.*</tt>
5437intrinsics do not return a value, and takes an extra alignment argument.
5438</p>
5439
5440<h5>Arguments:</h5>
5441
5442<p>
5443The first argument is a pointer to the destination, the second is a pointer to
5444the source. The third argument is an integer argument
5445specifying the number of bytes to copy, and the fourth argument is the alignment
5446of the source and destination locations.
5447</p>
5448
5449<p>
5450If the call to this intrinisic has an alignment value that is not 0 or 1, then
5451the caller guarantees that the source and destination pointers are aligned to
5452that boundary.
5453</p>
5454
5455<h5>Semantics:</h5>
5456
5457<p>
5458The '<tt>llvm.memmove.*</tt>' intrinsics copy a block of memory from the source
5459location to the destination location, which may overlap. It
5460copies "len" bytes of memory over. If the argument is known to be aligned to
5461some boundary, this can be specified as the fourth argument, otherwise it should
5462be set to 0 or 1.
5463</p>
5464</div>
5465
5466
5467<!-- _______________________________________________________________________ -->
5468<div class="doc_subsubsection">
5469 <a name="int_memset">'<tt>llvm.memset.*</tt>' Intrinsics</a>
5470</div>
5471
5472<div class="doc_text">
5473
5474<h5>Syntax:</h5>
Chris Lattner82c2e432008-11-21 16:42:48 +0000[diff] [blame]5475<p>This is an overloaded intrinsic. You can use llvm.memset on any integer bit
5476width. Not all targets support all bit widths however.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5477<pre>
Chris Lattner82c2e432008-11-21 16:42:48 +0000[diff] [blame]5478 declare void @llvm.memset.i8(i8 * &lt;dest&gt;, i8 &lt;val&gt;,
5479 i8 &lt;len&gt;, i32 &lt;align&gt;)
5480 declare void @llvm.memset.i16(i8 * &lt;dest&gt;, i8 &lt;val&gt;,
5481 i16 &lt;len&gt;, i32 &lt;align&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5482 declare void @llvm.memset.i32(i8 * &lt;dest&gt;, i8 &lt;val&gt;,
5483 i32 &lt;len&gt;, i32 &lt;align&gt;)
5484 declare void @llvm.memset.i64(i8 * &lt;dest&gt;, i8 &lt;val&gt;,
5485 i64 &lt;len&gt;, i32 &lt;align&gt;)
5486</pre>
5487
5488<h5>Overview:</h5>
5489
5490<p>
5491The '<tt>llvm.memset.*</tt>' intrinsics fill a block of memory with a particular
5492byte value.
5493</p>
5494
5495<p>
5496Note that, unlike the standard libc function, the <tt>llvm.memset</tt> intrinsic
5497does not return a value, and takes an extra alignment argument.
5498</p>
5499
5500<h5>Arguments:</h5>
5501
5502<p>
5503The first argument is a pointer to the destination to fill, the second is the
5504byte value to fill it with, the third argument is an integer
5505argument specifying the number of bytes to fill, and the fourth argument is the
5506known alignment of destination location.
5507</p>
5508
5509<p>
5510If the call to this intrinisic has an alignment value that is not 0 or 1, then
5511the caller guarantees that the destination pointer is aligned to that boundary.
5512</p>
5513
5514<h5>Semantics:</h5>
5515
5516<p>
5517The '<tt>llvm.memset.*</tt>' intrinsics fill "len" bytes of memory starting at
5518the
5519destination location. If the argument is known to be aligned to some boundary,
5520this can be specified as the fourth argument, otherwise it should be set to 0 or
55211.
5522</p>
5523</div>
5524
5525
5526<!-- _______________________________________________________________________ -->
5527<div class="doc_subsubsection">
5528 <a name="int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a>
5529</div>
5530
5531<div class="doc_text">
5532
5533<h5>Syntax:</h5>
Dale Johannesenf9adbb62007-10-02 17:47:38 +0000[diff] [blame]5534<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]5535floating point or vector of floating point type. Not all targets support all
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5536types however.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5537<pre>
Dale Johannesenf9adbb62007-10-02 17:47:38 +0000[diff] [blame]5538 declare float @llvm.sqrt.f32(float %Val)
5539 declare double @llvm.sqrt.f64(double %Val)
5540 declare x86_fp80 @llvm.sqrt.f80(x86_fp80 %Val)
5541 declare fp128 @llvm.sqrt.f128(fp128 %Val)
5542 declare ppc_fp128 @llvm.sqrt.ppcf128(ppc_fp128 %Val)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5543</pre>
5544
5545<h5>Overview:</h5>
5546
5547<p>
5548The '<tt>llvm.sqrt</tt>' intrinsics return the sqrt of the specified operand,
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5549returning the same value as the libm '<tt>sqrt</tt>' functions would. Unlike
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5550<tt>sqrt</tt> in libm, however, <tt>llvm.sqrt</tt> has undefined behavior for
Chris Lattnerf914a002008-01-29 07:00:44 +0000[diff] [blame]5551negative numbers other than -0.0 (which allows for better optimization, because
5552there is no need to worry about errno being set). <tt>llvm.sqrt(-0.0)</tt> is
5553defined to return -0.0 like IEEE sqrt.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5554</p>
5555
5556<h5>Arguments:</h5>
5557
5558<p>
5559The argument and return value are floating point numbers of the same type.
5560</p>
5561
5562<h5>Semantics:</h5>
5563
5564<p>
5565This function returns the sqrt of the specified operand if it is a nonnegative
5566floating point number.
5567</p>
5568</div>
5569
5570<!-- _______________________________________________________________________ -->
5571<div class="doc_subsubsection">
5572 <a name="int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a>
5573</div>
5574
5575<div class="doc_text">
5576
5577<h5>Syntax:</h5>
Dale Johannesenf9adbb62007-10-02 17:47:38 +0000[diff] [blame]5578<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]5579floating point or vector of floating point type. Not all targets support all
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5580types however.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5581<pre>
Dale Johannesenf9adbb62007-10-02 17:47:38 +0000[diff] [blame]5582 declare float @llvm.powi.f32(float %Val, i32 %power)
5583 declare double @llvm.powi.f64(double %Val, i32 %power)
5584 declare x86_fp80 @llvm.powi.f80(x86_fp80 %Val, i32 %power)
5585 declare fp128 @llvm.powi.f128(fp128 %Val, i32 %power)
5586 declare ppc_fp128 @llvm.powi.ppcf128(ppc_fp128 %Val, i32 %power)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5587</pre>
5588
5589<h5>Overview:</h5>
5590
5591<p>
5592The '<tt>llvm.powi.*</tt>' intrinsics return the first operand raised to the
5593specified (positive or negative) power. The order of evaluation of
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5594multiplications is not defined. When a vector of floating point type is
5595used, the second argument remains a scalar integer value.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5596</p>
5597
5598<h5>Arguments:</h5>
5599
5600<p>
5601The second argument is an integer power, and the first is a value to raise to
5602that power.
5603</p>
5604
5605<h5>Semantics:</h5>
5606
5607<p>
5608This function returns the first value raised to the second power with an
5609unspecified sequence of rounding operations.</p>
5610</div>
5611
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5612<!-- _______________________________________________________________________ -->
5613<div class="doc_subsubsection">
5614 <a name="int_sin">'<tt>llvm.sin.*</tt>' Intrinsic</a>
5615</div>
5616
5617<div class="doc_text">
5618
5619<h5>Syntax:</h5>
5620<p>This is an overloaded intrinsic. You can use <tt>llvm.sin</tt> on any
5621floating point or vector of floating point type. Not all targets support all
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5622types however.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5623<pre>
5624 declare float @llvm.sin.f32(float %Val)
5625 declare double @llvm.sin.f64(double %Val)
5626 declare x86_fp80 @llvm.sin.f80(x86_fp80 %Val)
5627 declare fp128 @llvm.sin.f128(fp128 %Val)
5628 declare ppc_fp128 @llvm.sin.ppcf128(ppc_fp128 %Val)
5629</pre>
5630
5631<h5>Overview:</h5>
5632
5633<p>
5634The '<tt>llvm.sin.*</tt>' intrinsics return the sine of the operand.
5635</p>
5636
5637<h5>Arguments:</h5>
5638
5639<p>
5640The argument and return value are floating point numbers of the same type.
5641</p>
5642
5643<h5>Semantics:</h5>
5644
5645<p>
5646This function returns the sine of the specified operand, returning the
5647same values as the libm <tt>sin</tt> functions would, and handles error
Dan Gohmaneaba92e2007-10-17 18:05:13 +0000[diff] [blame]5648conditions in the same way.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5649</div>
5650
5651<!-- _______________________________________________________________________ -->
5652<div class="doc_subsubsection">
5653 <a name="int_cos">'<tt>llvm.cos.*</tt>' Intrinsic</a>
5654</div>
5655
5656<div class="doc_text">
5657
5658<h5>Syntax:</h5>
5659<p>This is an overloaded intrinsic. You can use <tt>llvm.cos</tt> on any
5660floating point or vector of floating point type. Not all targets support all
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5661types however.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5662<pre>
5663 declare float @llvm.cos.f32(float %Val)
5664 declare double @llvm.cos.f64(double %Val)
5665 declare x86_fp80 @llvm.cos.f80(x86_fp80 %Val)
5666 declare fp128 @llvm.cos.f128(fp128 %Val)
5667 declare ppc_fp128 @llvm.cos.ppcf128(ppc_fp128 %Val)
5668</pre>
5669
5670<h5>Overview:</h5>
5671
5672<p>
5673The '<tt>llvm.cos.*</tt>' intrinsics return the cosine of the operand.
5674</p>
5675
5676<h5>Arguments:</h5>
5677
5678<p>
5679The argument and return value are floating point numbers of the same type.
5680</p>
5681
5682<h5>Semantics:</h5>
5683
5684<p>
5685This function returns the cosine of the specified operand, returning the
5686same values as the libm <tt>cos</tt> functions would, and handles error
Dan Gohmaneaba92e2007-10-17 18:05:13 +0000[diff] [blame]5687conditions in the same way.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5688</div>
5689
5690<!-- _______________________________________________________________________ -->
5691<div class="doc_subsubsection">
5692 <a name="int_pow">'<tt>llvm.pow.*</tt>' Intrinsic</a>
5693</div>
5694
5695<div class="doc_text">
5696
5697<h5>Syntax:</h5>
5698<p>This is an overloaded intrinsic. You can use <tt>llvm.pow</tt> on any
5699floating point or vector of floating point type. Not all targets support all
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5700types however.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5701<pre>
5702 declare float @llvm.pow.f32(float %Val, float %Power)
5703 declare double @llvm.pow.f64(double %Val, double %Power)
5704 declare x86_fp80 @llvm.pow.f80(x86_fp80 %Val, x86_fp80 %Power)
5705 declare fp128 @llvm.pow.f128(fp128 %Val, fp128 %Power)
5706 declare ppc_fp128 @llvm.pow.ppcf128(ppc_fp128 %Val, ppc_fp128 Power)
5707</pre>
5708
5709<h5>Overview:</h5>
5710
5711<p>
5712The '<tt>llvm.pow.*</tt>' intrinsics return the first operand raised to the
5713specified (positive or negative) power.
5714</p>
5715
5716<h5>Arguments:</h5>
5717
5718<p>
5719The second argument is a floating point power, and the first is a value to
5720raise to that power.
5721</p>
5722
5723<h5>Semantics:</h5>
5724
5725<p>
5726This function returns the first value raised to the second power,
5727returning the
5728same values as the libm <tt>pow</tt> functions would, and handles error
Dan Gohmaneaba92e2007-10-17 18:05:13 +0000[diff] [blame]5729conditions in the same way.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5730</div>
5731
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5732
5733<!-- ======================================================================= -->
5734<div class="doc_subsection">
5735 <a name="int_manip">Bit Manipulation Intrinsics</a>
5736</div>
5737
5738<div class="doc_text">
5739<p>
5740LLVM provides intrinsics for a few important bit manipulation operations.
5741These allow efficient code generation for some algorithms.
5742</p>
5743
5744</div>
5745
5746<!-- _______________________________________________________________________ -->
5747<div class="doc_subsubsection">
5748 <a name="int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a>
5749</div>
5750
5751<div class="doc_text">
5752
5753<h5>Syntax:</h5>
5754<p>This is an overloaded intrinsic function. You can use bswap on any integer
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5755type that is an even number of bytes (i.e. BitWidth % 16 == 0).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5756<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5757 declare i16 @llvm.bswap.i16(i16 &lt;id&gt;)
5758 declare i32 @llvm.bswap.i32(i32 &lt;id&gt;)
5759 declare i64 @llvm.bswap.i64(i64 &lt;id&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5760</pre>
5761
5762<h5>Overview:</h5>
5763
5764<p>
5765The '<tt>llvm.bswap</tt>' family of intrinsics is used to byte swap integer
5766values with an even number of bytes (positive multiple of 16 bits). These are
5767useful for performing operations on data that is not in the target's native
5768byte order.
5769</p>
5770
5771<h5>Semantics:</h5>
5772
5773<p>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5774The <tt>llvm.bswap.i16</tt> intrinsic returns an i16 value that has the high
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5775and low byte of the input i16 swapped. Similarly, the <tt>llvm.bswap.i32</tt>
5776intrinsic returns an i32 value that has the four bytes of the input i32
5777swapped, 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]5778i32 will have its bytes in 3, 2, 1, 0 order. The <tt>llvm.bswap.i48</tt>,
5779<tt>llvm.bswap.i64</tt> and other intrinsics extend this concept to
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5780additional even-byte lengths (6 bytes, 8 bytes and more, respectively).
5781</p>
5782
5783</div>
5784
5785<!-- _______________________________________________________________________ -->
5786<div class="doc_subsubsection">
5787 <a name="int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic</a>
5788</div>
5789
5790<div class="doc_text">
5791
5792<h5>Syntax:</h5>
5793<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]5794width. Not all targets support all bit widths however.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5795<pre>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]5796 declare i8 @llvm.ctpop.i8(i8 &lt;src&gt;)
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5797 declare i16 @llvm.ctpop.i16(i16 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5798 declare i32 @llvm.ctpop.i32(i32 &lt;src&gt;)
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5799 declare i64 @llvm.ctpop.i64(i64 &lt;src&gt;)
5800 declare i256 @llvm.ctpop.i256(i256 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5801</pre>
5802
5803<h5>Overview:</h5>
5804
5805<p>
5806The '<tt>llvm.ctpop</tt>' family of intrinsics counts the number of bits set in a
5807value.
5808</p>
5809
5810<h5>Arguments:</h5>
5811
5812<p>
5813The only argument is the value to be counted. The argument may be of any
5814integer type. The return type must match the argument type.
5815</p>
5816
5817<h5>Semantics:</h5>
5818
5819<p>
5820The '<tt>llvm.ctpop</tt>' intrinsic counts the 1's in a variable.
5821</p>
5822</div>
5823
5824<!-- _______________________________________________________________________ -->
5825<div class="doc_subsubsection">
5826 <a name="int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic</a>
5827</div>
5828
5829<div class="doc_text">
5830
5831<h5>Syntax:</h5>
5832<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]5833integer bit width. Not all targets support all bit widths however.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5834<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5835 declare i8 @llvm.ctlz.i8 (i8 &lt;src&gt;)
5836 declare i16 @llvm.ctlz.i16(i16 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5837 declare i32 @llvm.ctlz.i32(i32 &lt;src&gt;)
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5838 declare i64 @llvm.ctlz.i64(i64 &lt;src&gt;)
5839 declare i256 @llvm.ctlz.i256(i256 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5840</pre>
5841
5842<h5>Overview:</h5>
5843
5844<p>
5845The '<tt>llvm.ctlz</tt>' family of intrinsic functions counts the number of
5846leading zeros in a variable.
5847</p>
5848
5849<h5>Arguments:</h5>
5850
5851<p>
5852The only argument is the value to be counted. The argument may be of any
5853integer type. The return type must match the argument type.
5854</p>
5855
5856<h5>Semantics:</h5>
5857
5858<p>
5859The '<tt>llvm.ctlz</tt>' intrinsic counts the leading (most significant) zeros
5860in a variable. If the src == 0 then the result is the size in bits of the type
5861of src. For example, <tt>llvm.ctlz(i32 2) = 30</tt>.
5862</p>
5863</div>
5864
5865
5866
5867<!-- _______________________________________________________________________ -->
5868<div class="doc_subsubsection">
5869 <a name="int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic</a>
5870</div>
5871
5872<div class="doc_text">
5873
5874<h5>Syntax:</h5>
5875<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]5876integer bit width. Not all targets support all bit widths however.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5877<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5878 declare i8 @llvm.cttz.i8 (i8 &lt;src&gt;)
5879 declare i16 @llvm.cttz.i16(i16 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5880 declare i32 @llvm.cttz.i32(i32 &lt;src&gt;)
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5881 declare i64 @llvm.cttz.i64(i64 &lt;src&gt;)
5882 declare i256 @llvm.cttz.i256(i256 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5883</pre>
5884
5885<h5>Overview:</h5>
5886
5887<p>
5888The '<tt>llvm.cttz</tt>' family of intrinsic functions counts the number of
5889trailing zeros.
5890</p>
5891
5892<h5>Arguments:</h5>
5893
5894<p>
5895The only argument is the value to be counted. The argument may be of any
5896integer type. The return type must match the argument type.
5897</p>
5898
5899<h5>Semantics:</h5>
5900
5901<p>
5902The '<tt>llvm.cttz</tt>' intrinsic counts the trailing (least significant) zeros
5903in a variable. If the src == 0 then the result is the size in bits of the type
5904of src. For example, <tt>llvm.cttz(2) = 1</tt>.
5905</p>
5906</div>
5907
5908<!-- _______________________________________________________________________ -->
5909<div class="doc_subsubsection">
5910 <a name="int_part_select">'<tt>llvm.part.select.*</tt>' Intrinsic</a>
5911</div>
5912
5913<div class="doc_text">
5914
5915<h5>Syntax:</h5>
5916<p>This is an overloaded intrinsic. You can use <tt>llvm.part.select</tt>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5917on any integer bit width.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5918<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5919 declare i17 @llvm.part.select.i17 (i17 %val, i32 %loBit, i32 %hiBit)
5920 declare i29 @llvm.part.select.i29 (i29 %val, i32 %loBit, i32 %hiBit)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5921</pre>
5922
5923<h5>Overview:</h5>
5924<p>The '<tt>llvm.part.select</tt>' family of intrinsic functions selects a
5925range of bits from an integer value and returns them in the same bit width as
5926the original value.</p>
5927
5928<h5>Arguments:</h5>
5929<p>The first argument, <tt>%val</tt> and the result may be integer types of
5930any bit width but they must have the same bit width. The second and third
5931arguments must be <tt>i32</tt> type since they specify only a bit index.</p>
5932
5933<h5>Semantics:</h5>
5934<p>The operation of the '<tt>llvm.part.select</tt>' intrinsic has two modes
5935of operation: forwards and reverse. If <tt>%loBit</tt> is greater than
5936<tt>%hiBits</tt> then the intrinsic operates in reverse mode. Otherwise it
5937operates in forward mode.</p>
5938<p>In forward mode, this intrinsic is the equivalent of shifting <tt>%val</tt>
5939right by <tt>%loBit</tt> bits and then ANDing it with a mask with
5940only the <tt>%hiBit - %loBit</tt> bits set, as follows:</p>
5941<ol>
5942 <li>The <tt>%val</tt> is shifted right (LSHR) by the number of bits specified
5943 by <tt>%loBits</tt>. This normalizes the value to the low order bits.</li>
5944 <li>The <tt>%loBits</tt> value is subtracted from the <tt>%hiBits</tt> value
5945 to determine the number of bits to retain.</li>
5946 <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]5947 <li>The mask is ANDed with <tt>%val</tt> to produce the result.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5948</ol>
5949<p>In reverse mode, a similar computation is made except that the bits are
5950returned in the reverse order. So, for example, if <tt>X</tt> has the value
5951<tt>i16 0x0ACF (101011001111)</tt> and we apply
5952<tt>part.select(i16 X, 8, 3)</tt> to it, we get back the value
5953<tt>i16 0x0026 (000000100110)</tt>.</p>
5954</div>
5955
5956<div class="doc_subsubsection">
5957 <a name="int_part_set">'<tt>llvm.part.set.*</tt>' Intrinsic</a>
5958</div>
5959
5960<div class="doc_text">
5961
5962<h5>Syntax:</h5>
5963<p>This is an overloaded intrinsic. You can use <tt>llvm.part.set</tt>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5964on any integer bit width.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5965<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5966 declare i17 @llvm.part.set.i17.i9 (i17 %val, i9 %repl, i32 %lo, i32 %hi)
5967 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]5968</pre>
5969
5970<h5>Overview:</h5>
5971<p>The '<tt>llvm.part.set</tt>' family of intrinsic functions replaces a range
5972of bits in an integer value with another integer value. It returns the integer
5973with the replaced bits.</p>
5974
5975<h5>Arguments:</h5>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]5976<p>The first argument, <tt>%val</tt>, and the result may be integer types of
5977any 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]5978whose bits will be replaced. The second argument, <tt>%repl</tt> may be an
5979integer of any bit width. The third and fourth arguments must be <tt>i32</tt>
5980type since they specify only a bit index.</p>
5981
5982<h5>Semantics:</h5>
5983<p>The operation of the '<tt>llvm.part.set</tt>' intrinsic has two modes
5984of operation: forwards and reverse. If <tt>%lo</tt> is greater than
5985<tt>%hi</tt> then the intrinsic operates in reverse mode. Otherwise it
5986operates in forward mode.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]5987
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5988<p>For both modes, the <tt>%repl</tt> value is prepared for use by either
5989truncating it down to the size of the replacement area or zero extending it
5990up to that size.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]5991
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5992<p>In forward mode, the bits between <tt>%lo</tt> and <tt>%hi</tt> (inclusive)
5993are replaced with corresponding bits from <tt>%repl</tt>. That is the 0th bit
5994in <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]5995to the <tt>%hi</tt>th bit.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]5996
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5997<p>In reverse mode, a similar computation is made except that the bits are
5998reversed. That is, the <tt>0</tt>th bit in <tt>%repl</tt> replaces the
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5999<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]6000
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6001<h5>Examples:</h5>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6002
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6003<pre>
6004 llvm.part.set(0xFFFF, 0, 4, 7) -&gt; 0xFF0F
6005 llvm.part.set(0xFFFF, 0, 7, 4) -&gt; 0xFF0F
6006 llvm.part.set(0xFFFF, 1, 7, 4) -&gt; 0xFF8F
6007 llvm.part.set(0xFFFF, F, 8, 3) -&gt; 0xFFE7
6008 llvm.part.set(0xFFFF, 0, 3, 8) -&gt; 0xFE07
6009</pre>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6010
6011</div>
6012
Bill Wendling3e1258b2009-02-08 04:04:40 +0000[diff] [blame]6013<!-- ======================================================================= -->
6014<div class="doc_subsection">
6015 <a name="int_overflow">Arithmetic with Overflow Intrinsics</a>
6016</div>
6017
6018<div class="doc_text">
6019<p>
6020LLVM provides intrinsics for some arithmetic with overflow operations.
6021</p>
6022
6023</div>
6024
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6025<!-- _______________________________________________________________________ -->
6026<div class="doc_subsubsection">
Bill Wendling3e1258b2009-02-08 04:04:40 +0000[diff] [blame]6027 <a name="int_sadd_overflow">'<tt>llvm.sadd.with.overflow.*</tt>' Intrinsics</a>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6028</div>
6029
6030<div class="doc_text">
6031
6032<h5>Syntax:</h5>
6033
6034<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]6035on any integer bit width.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6036
6037<pre>
6038 declare {i16, i1} @llvm.sadd.with.overflow.i16(i16 %a, i16 %b)
6039 declare {i32, i1} @llvm.sadd.with.overflow.i32(i32 %a, i32 %b)
6040 declare {i64, i1} @llvm.sadd.with.overflow.i64(i64 %a, i64 %b)
6041</pre>
6042
6043<h5>Overview:</h5>
6044
6045<p>The '<tt>llvm.sadd.with.overflow</tt>' family of intrinsic functions perform
6046a signed addition of the two arguments, and indicate whether an overflow
6047occurred during the signed summation.</p>
6048
6049<h5>Arguments:</h5>
6050
6051<p>The arguments (%a and %b) and the first element of the result structure may
6052be of integer types of any bit width, but they must have the same bit width. The
6053second element of the result structure must be of type <tt>i1</tt>. <tt>%a</tt>
6054and <tt>%b</tt> are the two values that will undergo signed addition.</p>
6055
6056<h5>Semantics:</h5>
6057
6058<p>The '<tt>llvm.sadd.with.overflow</tt>' family of intrinsic functions perform
6059a signed addition of the two variables. They return a structure &mdash; the
6060first element of which is the signed summation, and the second element of which
6061is a bit specifying if the signed summation resulted in an overflow.</p>
6062
6063<h5>Examples:</h5>
6064<pre>
6065 %res = call {i32, i1} @llvm.sadd.with.overflow.i32(i32 %a, i32 %b)
6066 %sum = extractvalue {i32, i1} %res, 0
6067 %obit = extractvalue {i32, i1} %res, 1
6068 br i1 %obit, label %overflow, label %normal
6069</pre>
6070
6071</div>
6072
6073<!-- _______________________________________________________________________ -->
6074<div class="doc_subsubsection">
Bill Wendling3e1258b2009-02-08 04:04:40 +0000[diff] [blame]6075 <a name="int_uadd_overflow">'<tt>llvm.uadd.with.overflow.*</tt>' Intrinsics</a>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6076</div>
6077
6078<div class="doc_text">
6079
6080<h5>Syntax:</h5>
6081
6082<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]6083on any integer bit width.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6084
6085<pre>
6086 declare {i16, i1} @llvm.uadd.with.overflow.i16(i16 %a, i16 %b)
6087 declare {i32, i1} @llvm.uadd.with.overflow.i32(i32 %a, i32 %b)
6088 declare {i64, i1} @llvm.uadd.with.overflow.i64(i64 %a, i64 %b)
6089</pre>
6090
6091<h5>Overview:</h5>
6092
6093<p>The '<tt>llvm.uadd.with.overflow</tt>' family of intrinsic functions perform
6094an unsigned addition of the two arguments, and indicate whether a carry occurred
6095during the unsigned summation.</p>
6096
6097<h5>Arguments:</h5>
6098
6099<p>The arguments (%a and %b) and the first element of the result structure may
6100be of integer types of any bit width, but they must have the same bit width. The
6101second element of the result structure must be of type <tt>i1</tt>. <tt>%a</tt>
6102and <tt>%b</tt> are the two values that will undergo unsigned addition.</p>
6103
6104<h5>Semantics:</h5>
6105
6106<p>The '<tt>llvm.uadd.with.overflow</tt>' family of intrinsic functions perform
6107an unsigned addition of the two arguments. They return a structure &mdash; the
6108first element of which is the sum, and the second element of which is a bit
6109specifying if the unsigned summation resulted in a carry.</p>
6110
6111<h5>Examples:</h5>
6112<pre>
6113 %res = call {i32, i1} @llvm.uadd.with.overflow.i32(i32 %a, i32 %b)
6114 %sum = extractvalue {i32, i1} %res, 0
6115 %obit = extractvalue {i32, i1} %res, 1
6116 br i1 %obit, label %carry, label %normal
6117</pre>
6118
6119</div>
6120
6121<!-- _______________________________________________________________________ -->
6122<div class="doc_subsubsection">
Bill Wendling3e1258b2009-02-08 04:04:40 +0000[diff] [blame]6123 <a name="int_ssub_overflow">'<tt>llvm.ssub.with.overflow.*</tt>' Intrinsics</a>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6124</div>
6125
6126<div class="doc_text">
6127
6128<h5>Syntax:</h5>
6129
6130<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]6131on any integer bit width.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6132
6133<pre>
6134 declare {i16, i1} @llvm.ssub.with.overflow.i16(i16 %a, i16 %b)
6135 declare {i32, i1} @llvm.ssub.with.overflow.i32(i32 %a, i32 %b)
6136 declare {i64, i1} @llvm.ssub.with.overflow.i64(i64 %a, i64 %b)
6137</pre>
6138
6139<h5>Overview:</h5>
6140
6141<p>The '<tt>llvm.ssub.with.overflow</tt>' family of intrinsic functions perform
6142a signed subtraction of the two arguments, and indicate whether an overflow
6143occurred during the signed subtraction.</p>
6144
6145<h5>Arguments:</h5>
6146
6147<p>The arguments (%a and %b) and the first element of the result structure may
6148be of integer types of any bit width, but they must have the same bit width. The
6149second element of the result structure must be of type <tt>i1</tt>. <tt>%a</tt>
6150and <tt>%b</tt> are the two values that will undergo signed subtraction.</p>
6151
6152<h5>Semantics:</h5>
6153
6154<p>The '<tt>llvm.ssub.with.overflow</tt>' family of intrinsic functions perform
6155a signed subtraction of the two arguments. They return a structure &mdash; the
6156first element of which is the subtraction, and the second element of which is a bit
6157specifying if the signed subtraction resulted in an overflow.</p>
6158
6159<h5>Examples:</h5>
6160<pre>
6161 %res = call {i32, i1} @llvm.ssub.with.overflow.i32(i32 %a, i32 %b)
6162 %sum = extractvalue {i32, i1} %res, 0
6163 %obit = extractvalue {i32, i1} %res, 1
6164 br i1 %obit, label %overflow, label %normal
6165</pre>
6166
6167</div>
6168
6169<!-- _______________________________________________________________________ -->
6170<div class="doc_subsubsection">
Bill Wendling3e1258b2009-02-08 04:04:40 +0000[diff] [blame]6171 <a name="int_usub_overflow">'<tt>llvm.usub.with.overflow.*</tt>' Intrinsics</a>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6172</div>
6173
6174<div class="doc_text">
6175
6176<h5>Syntax:</h5>
6177
6178<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]6179on any integer bit width.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6180
6181<pre>
6182 declare {i16, i1} @llvm.usub.with.overflow.i16(i16 %a, i16 %b)
6183 declare {i32, i1} @llvm.usub.with.overflow.i32(i32 %a, i32 %b)
6184 declare {i64, i1} @llvm.usub.with.overflow.i64(i64 %a, i64 %b)
6185</pre>
6186
6187<h5>Overview:</h5>
6188
6189<p>The '<tt>llvm.usub.with.overflow</tt>' family of intrinsic functions perform
6190an unsigned subtraction of the two arguments, and indicate whether an overflow
6191occurred during the unsigned subtraction.</p>
6192
6193<h5>Arguments:</h5>
6194
6195<p>The arguments (%a and %b) and the first element of the result structure may
6196be of integer types of any bit width, but they must have the same bit width. The
6197second element of the result structure must be of type <tt>i1</tt>. <tt>%a</tt>
6198and <tt>%b</tt> are the two values that will undergo unsigned subtraction.</p>
6199
6200<h5>Semantics:</h5>
6201
6202<p>The '<tt>llvm.usub.with.overflow</tt>' family of intrinsic functions perform
6203an unsigned subtraction of the two arguments. They return a structure &mdash; the
6204first element of which is the subtraction, and the second element of which is a bit
6205specifying if the unsigned subtraction resulted in an overflow.</p>
6206
6207<h5>Examples:</h5>
6208<pre>
6209 %res = call {i32, i1} @llvm.usub.with.overflow.i32(i32 %a, i32 %b)
6210 %sum = extractvalue {i32, i1} %res, 0
6211 %obit = extractvalue {i32, i1} %res, 1
6212 br i1 %obit, label %overflow, label %normal
6213</pre>
6214
6215</div>
6216
6217<!-- _______________________________________________________________________ -->
6218<div class="doc_subsubsection">
Bill Wendling3e1258b2009-02-08 04:04:40 +0000[diff] [blame]6219 <a name="int_smul_overflow">'<tt>llvm.smul.with.overflow.*</tt>' Intrinsics</a>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6220</div>
6221
6222<div class="doc_text">
6223
6224<h5>Syntax:</h5>
6225
6226<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]6227on any integer bit width.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6228
6229<pre>
6230 declare {i16, i1} @llvm.smul.with.overflow.i16(i16 %a, i16 %b)
6231 declare {i32, i1} @llvm.smul.with.overflow.i32(i32 %a, i32 %b)
6232 declare {i64, i1} @llvm.smul.with.overflow.i64(i64 %a, i64 %b)
6233</pre>
6234
6235<h5>Overview:</h5>
6236
6237<p>The '<tt>llvm.smul.with.overflow</tt>' family of intrinsic functions perform
6238a signed multiplication of the two arguments, and indicate whether an overflow
6239occurred during the signed multiplication.</p>
6240
6241<h5>Arguments:</h5>
6242
6243<p>The arguments (%a and %b) and the first element of the result structure may
6244be of integer types of any bit width, but they must have the same bit width. The
6245second element of the result structure must be of type <tt>i1</tt>. <tt>%a</tt>
6246and <tt>%b</tt> are the two values that will undergo signed multiplication.</p>
6247
6248<h5>Semantics:</h5>
6249
6250<p>The '<tt>llvm.smul.with.overflow</tt>' family of intrinsic functions perform
6251a signed multiplication of the two arguments. They return a structure &mdash;
6252the first element of which is the multiplication, and the second element of
6253which is a bit specifying if the signed multiplication resulted in an
6254overflow.</p>
6255
6256<h5>Examples:</h5>
6257<pre>
6258 %res = call {i32, i1} @llvm.smul.with.overflow.i32(i32 %a, i32 %b)
6259 %sum = extractvalue {i32, i1} %res, 0
6260 %obit = extractvalue {i32, i1} %res, 1
6261 br i1 %obit, label %overflow, label %normal
6262</pre>
6263
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6264</div>
6265
Bill Wendlingbda98b62009-02-08 23:00:09 +0000[diff] [blame]6266<!-- _______________________________________________________________________ -->
6267<div class="doc_subsubsection">
6268 <a name="int_umul_overflow">'<tt>llvm.umul.with.overflow.*</tt>' Intrinsics</a>
6269</div>
6270
6271<div class="doc_text">
6272
6273<h5>Syntax:</h5>
6274
6275<p>This is an overloaded intrinsic. You can use <tt>llvm.umul.with.overflow</tt>
6276on any integer bit width.</p>
6277
6278<pre>
6279 declare {i16, i1} @llvm.umul.with.overflow.i16(i16 %a, i16 %b)
6280 declare {i32, i1} @llvm.umul.with.overflow.i32(i32 %a, i32 %b)
6281 declare {i64, i1} @llvm.umul.with.overflow.i64(i64 %a, i64 %b)
6282</pre>
6283
6284<h5>Overview:</h5>
6285
6286<p><i><b>Warning:</b> '<tt>llvm.umul.with.overflow</tt>' is badly broken. It is
6287actively being fixed, but it should not currently be used!</i></p>
6288
6289<p>The '<tt>llvm.umul.with.overflow</tt>' family of intrinsic functions perform
6290a unsigned multiplication of the two arguments, and indicate whether an overflow
6291occurred during the unsigned multiplication.</p>
6292
6293<h5>Arguments:</h5>
6294
6295<p>The arguments (%a and %b) and the first element of the result structure may
6296be of integer types of any bit width, but they must have the same bit width. The
6297second element of the result structure must be of type <tt>i1</tt>. <tt>%a</tt>
6298and <tt>%b</tt> are the two values that will undergo unsigned
6299multiplication.</p>
6300
6301<h5>Semantics:</h5>
6302
6303<p>The '<tt>llvm.umul.with.overflow</tt>' family of intrinsic functions perform
6304an unsigned multiplication of the two arguments. They return a structure &mdash;
6305the first element of which is the multiplication, and the second element of
6306which is a bit specifying if the unsigned multiplication resulted in an
6307overflow.</p>
6308
6309<h5>Examples:</h5>
6310<pre>
6311 %res = call {i32, i1} @llvm.umul.with.overflow.i32(i32 %a, i32 %b)
6312 %sum = extractvalue {i32, i1} %res, 0
6313 %obit = extractvalue {i32, i1} %res, 1
6314 br i1 %obit, label %overflow, label %normal
6315</pre>
6316
6317</div>
6318
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6319<!-- ======================================================================= -->
6320<div class="doc_subsection">
6321 <a name="int_debugger">Debugger Intrinsics</a>
6322</div>
6323
6324<div class="doc_text">
6325<p>
6326The LLVM debugger intrinsics (which all start with <tt>llvm.dbg.</tt> prefix),
6327are described in the <a
6328href="SourceLevelDebugging.html#format_common_intrinsics">LLVM Source Level
6329Debugging</a> document.
6330</p>
6331</div>
6332
6333
6334<!-- ======================================================================= -->
6335<div class="doc_subsection">
6336 <a name="int_eh">Exception Handling Intrinsics</a>
6337</div>
6338
6339<div class="doc_text">
6340<p> The LLVM exception handling intrinsics (which all start with
6341<tt>llvm.eh.</tt> prefix), are described in the <a
6342href="ExceptionHandling.html#format_common_intrinsics">LLVM Exception
6343Handling</a> document. </p>
6344</div>
6345
6346<!-- ======================================================================= -->
6347<div class="doc_subsection">
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]6348 <a name="int_trampoline">Trampoline Intrinsic</a>
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]6349</div>
6350
6351<div class="doc_text">
6352<p>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]6353 This intrinsic makes it possible to excise one parameter, marked with
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]6354 the <tt>nest</tt> attribute, from a function. The result is a callable
6355 function pointer lacking the nest parameter - the caller does not need
6356 to provide a value for it. Instead, the value to use is stored in
6357 advance in a "trampoline", a block of memory usually allocated
6358 on the stack, which also contains code to splice the nest value into the
6359 argument list. This is used to implement the GCC nested function address
6360 extension.
6361</p>
6362<p>
6363 For example, if the function is
6364 <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]6365 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]6366<pre>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]6367 %tramp = alloca [10 x i8], align 4 ; size and alignment only correct for X86
6368 %tramp1 = getelementptr [10 x i8]* %tramp, i32 0, i32 0
6369 %p = call i8* @llvm.init.trampoline( i8* %tramp1, i8* bitcast (i32 (i8* nest , i32, i32)* @f to i8*), i8* %nval )
6370 %fp = bitcast i8* %p to i32 (i32, i32)*
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]6371</pre>
Bill Wendlinge262b4c2007-09-22 09:23:55 +0000[diff] [blame]6372 <p>The call <tt>%val = call i32 %fp( i32 %x, i32 %y )</tt> is then equivalent
6373 to <tt>%val = call i32 %f( i8* %nval, i32 %x, i32 %y )</tt>.</p>
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]6374</div>
6375
6376<!-- _______________________________________________________________________ -->
6377<div class="doc_subsubsection">
6378 <a name="int_it">'<tt>llvm.init.trampoline</tt>' Intrinsic</a>
6379</div>
6380<div class="doc_text">
6381<h5>Syntax:</h5>
6382<pre>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]6383declare 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]6384</pre>
6385<h5>Overview:</h5>
6386<p>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]6387 This fills the memory pointed to by <tt>tramp</tt> with code
6388 and returns a function pointer suitable for executing it.
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]6389</p>
6390<h5>Arguments:</h5>
6391<p>
6392 The <tt>llvm.init.trampoline</tt> intrinsic takes three arguments, all
6393 pointers. The <tt>tramp</tt> argument must point to a sufficiently large
6394 and sufficiently aligned block of memory; this memory is written to by the
Duncan Sands35012212007-08-22 23:39:54 +0000[diff] [blame]6395 intrinsic. Note that the size and the alignment are target-specific - LLVM
6396 currently provides no portable way of determining them, so a front-end that
6397 generates this intrinsic needs to have some target-specific knowledge.
6398 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]6399</p>
6400<h5>Semantics:</h5>
6401<p>
6402 The block of memory pointed to by <tt>tramp</tt> is filled with target
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]6403 dependent code, turning it into a function. A pointer to this function is
6404 returned, but needs to be bitcast to an
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]6405 <a href="#int_trampoline">appropriate function pointer type</a>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]6406 before being called. The new function's signature is the same as that of
6407 <tt>func</tt> with any arguments marked with the <tt>nest</tt> attribute
6408 removed. At most one such <tt>nest</tt> argument is allowed, and it must be
6409 of pointer type. Calling the new function is equivalent to calling
6410 <tt>func</tt> with the same argument list, but with <tt>nval</tt> used for the
6411 missing <tt>nest</tt> argument. If, after calling
6412 <tt>llvm.init.trampoline</tt>, the memory pointed to by <tt>tramp</tt> is
6413 modified, then the effect of any later call to the returned function pointer is
6414 undefined.
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]6415</p>
6416</div>
6417
6418<!-- ======================================================================= -->
6419<div class="doc_subsection">
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]6420 <a name="int_atomics">Atomic Operations and Synchronization Intrinsics</a>
6421</div>
6422
6423<div class="doc_text">
6424<p>
6425 These intrinsic functions expand the "universal IR" of LLVM to represent
6426 hardware constructs for atomic operations and memory synchronization. This
6427 provides an interface to the hardware, not an interface to the programmer. It
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]6428 is aimed at a low enough level to allow any programming models or APIs
6429 (Application Programming Interfaces) which
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]6430 need atomic behaviors to map cleanly onto it. It is also modeled primarily on
6431 hardware behavior. Just as hardware provides a "universal IR" for source
6432 languages, it also provides a starting point for developing a "universal"
6433 atomic operation and synchronization IR.
6434</p>
6435<p>
6436 These do <em>not</em> form an API such as high-level threading libraries,
6437 software transaction memory systems, atomic primitives, and intrinsic
6438 functions as found in BSD, GNU libc, atomic_ops, APR, and other system and
6439 application libraries. The hardware interface provided by LLVM should allow
6440 a clean implementation of all of these APIs and parallel programming models.
6441 No one model or paradigm should be selected above others unless the hardware
6442 itself ubiquitously does so.
6443
6444</p>
6445</div>
6446
6447<!-- _______________________________________________________________________ -->
6448<div class="doc_subsubsection">
6449 <a name="int_memory_barrier">'<tt>llvm.memory.barrier</tt>' Intrinsic</a>
6450</div>
6451<div class="doc_text">
6452<h5>Syntax:</h5>
6453<pre>
6454declare void @llvm.memory.barrier( i1 &lt;ll&gt;, i1 &lt;ls&gt;, i1 &lt;sl&gt;, i1 &lt;ss&gt;,
6455i1 &lt;device&gt; )
6456
6457</pre>
6458<h5>Overview:</h5>
6459<p>
6460 The <tt>llvm.memory.barrier</tt> intrinsic guarantees ordering between
6461 specific pairs of memory access types.
6462</p>
6463<h5>Arguments:</h5>
6464<p>
6465 The <tt>llvm.memory.barrier</tt> intrinsic requires five boolean arguments.
6466 The first four arguments enables a specific barrier as listed below. The fith
6467 argument specifies that the barrier applies to io or device or uncached memory.
6468
6469</p>
6470 <ul>
6471 <li><tt>ll</tt>: load-load barrier</li>
6472 <li><tt>ls</tt>: load-store barrier</li>
6473 <li><tt>sl</tt>: store-load barrier</li>
6474 <li><tt>ss</tt>: store-store barrier</li>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]6475 <li><tt>device</tt>: barrier applies to device and uncached memory also.</li>
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]6476 </ul>
6477<h5>Semantics:</h5>
6478<p>
6479 This intrinsic causes the system to enforce some ordering constraints upon
6480 the loads and stores of the program. This barrier does not indicate
6481 <em>when</em> any events will occur, it only enforces an <em>order</em> in
6482 which they occur. For any of the specified pairs of load and store operations
6483 (f.ex. load-load, or store-load), all of the first operations preceding the
6484 barrier will complete before any of the second operations succeeding the
6485 barrier begin. Specifically the semantics for each pairing is as follows:
6486</p>
6487 <ul>
6488 <li><tt>ll</tt>: All loads before the barrier must complete before any load
6489 after the barrier begins.</li>
6490
6491 <li><tt>ls</tt>: All loads before the barrier must complete before any
6492 store after the barrier begins.</li>
6493 <li><tt>ss</tt>: All stores before the barrier must complete before any
6494 store after the barrier begins.</li>
6495 <li><tt>sl</tt>: All stores before the barrier must complete before any
6496 load after the barrier begins.</li>
6497 </ul>
6498<p>
6499 These semantics are applied with a logical "and" behavior when more than one
6500 is enabled in a single memory barrier intrinsic.
6501</p>
6502<p>
6503 Backends may implement stronger barriers than those requested when they do not
6504 support as fine grained a barrier as requested. Some architectures do not
6505 need all types of barriers and on such architectures, these become noops.
6506</p>
6507<h5>Example:</h5>
6508<pre>
6509%ptr = malloc i32
6510 store i32 4, %ptr
6511
6512%result1 = load i32* %ptr <i>; yields {i32}:result1 = 4</i>
6513 call void @llvm.memory.barrier( i1 false, i1 true, i1 false, i1 false )
6514 <i>; guarantee the above finishes</i>
6515 store i32 8, %ptr <i>; before this begins</i>
6516</pre>
6517</div>
6518
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6519<!-- _______________________________________________________________________ -->
6520<div class="doc_subsubsection">
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6521 <a name="int_atomic_cmp_swap">'<tt>llvm.atomic.cmp.swap.*</tt>' Intrinsic</a>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6522</div>
6523<div class="doc_text">
6524<h5>Syntax:</h5>
6525<p>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6526 This is an overloaded intrinsic. You can use <tt>llvm.atomic.cmp.swap</tt> on
6527 any integer bit width and for different address spaces. Not all targets
6528 support all bit widths however.</p>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6529
6530<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6531declare i8 @llvm.atomic.cmp.swap.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;cmp&gt;, i8 &lt;val&gt; )
6532declare i16 @llvm.atomic.cmp.swap.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;cmp&gt;, i16 &lt;val&gt; )
6533declare i32 @llvm.atomic.cmp.swap.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;cmp&gt;, i32 &lt;val&gt; )
6534declare 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]6535
6536</pre>
6537<h5>Overview:</h5>
6538<p>
6539 This loads a value in memory and compares it to a given value. If they are
6540 equal, it stores a new value into the memory.
6541</p>
6542<h5>Arguments:</h5>
6543<p>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6544 The <tt>llvm.atomic.cmp.swap</tt> intrinsic takes three arguments. The result as
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6545 well as both <tt>cmp</tt> and <tt>val</tt> must be integer values with the
6546 same bit width. The <tt>ptr</tt> argument must be a pointer to a value of
6547 this integer type. While any bit width integer may be used, targets may only
6548 lower representations they support in hardware.
6549
6550</p>
6551<h5>Semantics:</h5>
6552<p>
6553 This entire intrinsic must be executed atomically. It first loads the value
6554 in memory pointed to by <tt>ptr</tt> and compares it with the value
6555 <tt>cmp</tt>. If they are equal, <tt>val</tt> is stored into the memory. The
6556 loaded value is yielded in all cases. This provides the equivalent of an
6557 atomic compare-and-swap operation within the SSA framework.
6558</p>
6559<h5>Examples:</h5>
6560
6561<pre>
6562%ptr = malloc i32
6563 store i32 4, %ptr
6564
6565%val1 = add i32 4, 4
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6566%result1 = call i32 @llvm.atomic.cmp.swap.i32.p0i32( i32* %ptr, i32 4, %val1 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6567 <i>; yields {i32}:result1 = 4</i>
6568%stored1 = icmp eq i32 %result1, 4 <i>; yields {i1}:stored1 = true</i>
6569%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 8</i>
6570
6571%val2 = add i32 1, 1
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6572%result2 = call i32 @llvm.atomic.cmp.swap.i32.p0i32( i32* %ptr, i32 5, %val2 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6573 <i>; yields {i32}:result2 = 8</i>
6574%stored2 = icmp eq i32 %result2, 5 <i>; yields {i1}:stored2 = false</i>
6575
6576%memval2 = load i32* %ptr <i>; yields {i32}:memval2 = 8</i>
6577</pre>
6578</div>
6579
6580<!-- _______________________________________________________________________ -->
6581<div class="doc_subsubsection">
6582 <a name="int_atomic_swap">'<tt>llvm.atomic.swap.*</tt>' Intrinsic</a>
6583</div>
6584<div class="doc_text">
6585<h5>Syntax:</h5>
6586
6587<p>
6588 This is an overloaded intrinsic. You can use <tt>llvm.atomic.swap</tt> on any
6589 integer bit width. Not all targets support all bit widths however.</p>
6590<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6591declare i8 @llvm.atomic.swap.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;val&gt; )
6592declare i16 @llvm.atomic.swap.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;val&gt; )
6593declare i32 @llvm.atomic.swap.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;val&gt; )
6594declare i64 @llvm.atomic.swap.i64.p0i64( i64* &lt;ptr&gt;, i64 &lt;val&gt; )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6595
6596</pre>
6597<h5>Overview:</h5>
6598<p>
6599 This intrinsic loads the value stored in memory at <tt>ptr</tt> and yields
6600 the value from memory. It then stores the value in <tt>val</tt> in the memory
6601 at <tt>ptr</tt>.
6602</p>
6603<h5>Arguments:</h5>
6604
6605<p>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6606 The <tt>llvm.atomic.swap</tt> intrinsic takes two arguments. Both the
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6607 <tt>val</tt> argument and the result must be integers of the same bit width.
6608 The first argument, <tt>ptr</tt>, must be a pointer to a value of this
6609 integer type. The targets may only lower integer representations they
6610 support.
6611</p>
6612<h5>Semantics:</h5>
6613<p>
6614 This intrinsic loads the value pointed to by <tt>ptr</tt>, yields it, and
6615 stores <tt>val</tt> back into <tt>ptr</tt> atomically. This provides the
6616 equivalent of an atomic swap operation within the SSA framework.
6617
6618</p>
6619<h5>Examples:</h5>
6620<pre>
6621%ptr = malloc i32
6622 store i32 4, %ptr
6623
6624%val1 = add i32 4, 4
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6625%result1 = call i32 @llvm.atomic.swap.i32.p0i32( i32* %ptr, i32 %val1 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6626 <i>; yields {i32}:result1 = 4</i>
6627%stored1 = icmp eq i32 %result1, 4 <i>; yields {i1}:stored1 = true</i>
6628%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 8</i>
6629
6630%val2 = add i32 1, 1
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6631%result2 = call i32 @llvm.atomic.swap.i32.p0i32( i32* %ptr, i32 %val2 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6632 <i>; yields {i32}:result2 = 8</i>
6633
6634%stored2 = icmp eq i32 %result2, 8 <i>; yields {i1}:stored2 = true</i>
6635%memval2 = load i32* %ptr <i>; yields {i32}:memval2 = 2</i>
6636</pre>
6637</div>
6638
6639<!-- _______________________________________________________________________ -->
6640<div class="doc_subsubsection">
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6641 <a name="int_atomic_load_add">'<tt>llvm.atomic.load.add.*</tt>' Intrinsic</a>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6642
6643</div>
6644<div class="doc_text">
6645<h5>Syntax:</h5>
6646<p>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6647 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]6648 integer bit width. Not all targets support all bit widths however.</p>
6649<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6650declare i8 @llvm.atomic.load.add.i8..p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6651declare i16 @llvm.atomic.load.add.i16..p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6652declare i32 @llvm.atomic.load.add.i32..p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6653declare 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]6654
6655</pre>
6656<h5>Overview:</h5>
6657<p>
6658 This intrinsic adds <tt>delta</tt> to the value stored in memory at
6659 <tt>ptr</tt>. It yields the original value at <tt>ptr</tt>.
6660</p>
6661<h5>Arguments:</h5>
6662<p>
6663
6664 The intrinsic takes two arguments, the first a pointer to an integer value
6665 and the second an integer value. The result is also an integer value. These
6666 integer types can have any bit width, but they must all have the same bit
6667 width. The targets may only lower integer representations they support.
6668</p>
6669<h5>Semantics:</h5>
6670<p>
6671 This intrinsic does a series of operations atomically. It first loads the
6672 value stored at <tt>ptr</tt>. It then adds <tt>delta</tt>, stores the result
6673 to <tt>ptr</tt>. It yields the original value stored at <tt>ptr</tt>.
6674</p>
6675
6676<h5>Examples:</h5>
6677<pre>
6678%ptr = malloc i32
6679 store i32 4, %ptr
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6680%result1 = call i32 @llvm.atomic.load.add.i32.p0i32( i32* %ptr, i32 4 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6681 <i>; yields {i32}:result1 = 4</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6682%result2 = call i32 @llvm.atomic.load.add.i32.p0i32( i32* %ptr, i32 2 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6683 <i>; yields {i32}:result2 = 8</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6684%result3 = call i32 @llvm.atomic.load.add.i32.p0i32( i32* %ptr, i32 5 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6685 <i>; yields {i32}:result3 = 10</i>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6686%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 15</i>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6687</pre>
6688</div>
6689
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6690<!-- _______________________________________________________________________ -->
6691<div class="doc_subsubsection">
6692 <a name="int_atomic_load_sub">'<tt>llvm.atomic.load.sub.*</tt>' Intrinsic</a>
6693
6694</div>
6695<div class="doc_text">
6696<h5>Syntax:</h5>
6697<p>
6698 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]6699 any integer bit width and for different address spaces. Not all targets
6700 support all bit widths however.</p>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6701<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6702declare i8 @llvm.atomic.load.sub.i8.p0i32( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6703declare i16 @llvm.atomic.load.sub.i16.p0i32( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6704declare i32 @llvm.atomic.load.sub.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6705declare 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]6706
6707</pre>
6708<h5>Overview:</h5>
6709<p>
6710 This intrinsic subtracts <tt>delta</tt> to the value stored in memory at
6711 <tt>ptr</tt>. It yields the original value at <tt>ptr</tt>.
6712</p>
6713<h5>Arguments:</h5>
6714<p>
6715
6716 The intrinsic takes two arguments, the first a pointer to an integer value
6717 and the second an integer value. The result is also an integer value. These
6718 integer types can have any bit width, but they must all have the same bit
6719 width. The targets may only lower integer representations they support.
6720</p>
6721<h5>Semantics:</h5>
6722<p>
6723 This intrinsic does a series of operations atomically. It first loads the
6724 value stored at <tt>ptr</tt>. It then subtracts <tt>delta</tt>, stores the
6725 result to <tt>ptr</tt>. It yields the original value stored at <tt>ptr</tt>.
6726</p>
6727
6728<h5>Examples:</h5>
6729<pre>
6730%ptr = malloc i32
6731 store i32 8, %ptr
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6732%result1 = call i32 @llvm.atomic.load.sub.i32.p0i32( i32* %ptr, i32 4 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6733 <i>; yields {i32}:result1 = 8</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6734%result2 = call i32 @llvm.atomic.load.sub.i32.p0i32( i32* %ptr, i32 2 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6735 <i>; yields {i32}:result2 = 4</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6736%result3 = call i32 @llvm.atomic.load.sub.i32.p0i32( i32* %ptr, i32 5 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6737 <i>; yields {i32}:result3 = 2</i>
6738%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = -3</i>
6739</pre>
6740</div>
6741
6742<!-- _______________________________________________________________________ -->
6743<div class="doc_subsubsection">
6744 <a name="int_atomic_load_and">'<tt>llvm.atomic.load.and.*</tt>' Intrinsic</a><br>
6745 <a name="int_atomic_load_nand">'<tt>llvm.atomic.load.nand.*</tt>' Intrinsic</a><br>
6746 <a name="int_atomic_load_or">'<tt>llvm.atomic.load.or.*</tt>' Intrinsic</a><br>
6747 <a name="int_atomic_load_xor">'<tt>llvm.atomic.load.xor.*</tt>' Intrinsic</a><br>
6748
6749</div>
6750<div class="doc_text">
6751<h5>Syntax:</h5>
6752<p>
6753 These are overloaded intrinsics. You can use <tt>llvm.atomic.load_and</tt>,
6754 <tt>llvm.atomic.load_nand</tt>, <tt>llvm.atomic.load_or</tt>, and
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6755 <tt>llvm.atomic.load_xor</tt> on any integer bit width and for different
6756 address spaces. Not all targets support all bit widths however.</p>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6757<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6758declare i8 @llvm.atomic.load.and.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6759declare i16 @llvm.atomic.load.and.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6760declare i32 @llvm.atomic.load.and.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6761declare 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]6762
6763</pre>
6764
6765<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6766declare i8 @llvm.atomic.load.or.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6767declare i16 @llvm.atomic.load.or.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6768declare i32 @llvm.atomic.load.or.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6769declare 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]6770
6771</pre>
6772
6773<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6774declare i8 @llvm.atomic.load.nand.i8.p0i32( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6775declare i16 @llvm.atomic.load.nand.i16.p0i32( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6776declare i32 @llvm.atomic.load.nand.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6777declare 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]6778
6779</pre>
6780
6781<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6782declare i8 @llvm.atomic.load.xor.i8.p0i32( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6783declare i16 @llvm.atomic.load.xor.i16.p0i32( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6784declare i32 @llvm.atomic.load.xor.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6785declare 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]6786
6787</pre>
6788<h5>Overview:</h5>
6789<p>
6790 These intrinsics bitwise the operation (and, nand, or, xor) <tt>delta</tt> to
6791 the value stored in memory at <tt>ptr</tt>. It yields the original value
6792 at <tt>ptr</tt>.
6793</p>
6794<h5>Arguments:</h5>
6795<p>
6796
6797 These intrinsics take two arguments, the first a pointer to an integer value
6798 and the second an integer value. The result is also an integer value. These
6799 integer types can have any bit width, but they must all have the same bit
6800 width. The targets may only lower integer representations they support.
6801</p>
6802<h5>Semantics:</h5>
6803<p>
6804 These intrinsics does a series of operations atomically. They first load the
6805 value stored at <tt>ptr</tt>. They then do the bitwise operation
6806 <tt>delta</tt>, store the result to <tt>ptr</tt>. They yield the original
6807 value stored at <tt>ptr</tt>.
6808</p>
6809
6810<h5>Examples:</h5>
6811<pre>
6812%ptr = malloc i32
6813 store i32 0x0F0F, %ptr
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6814%result0 = call i32 @llvm.atomic.load.nand.i32.p0i32( i32* %ptr, i32 0xFF )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6815 <i>; yields {i32}:result0 = 0x0F0F</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6816%result1 = call i32 @llvm.atomic.load.and.i32.p0i32( i32* %ptr, i32 0xFF )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6817 <i>; yields {i32}:result1 = 0xFFFFFFF0</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6818%result2 = call i32 @llvm.atomic.load.or.i32.p0i32( i32* %ptr, i32 0F )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6819 <i>; yields {i32}:result2 = 0xF0</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6820%result3 = call i32 @llvm.atomic.load.xor.i32.p0i32( i32* %ptr, i32 0F )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6821 <i>; yields {i32}:result3 = FF</i>
6822%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = F0</i>
6823</pre>
6824</div>
6825
6826
6827<!-- _______________________________________________________________________ -->
6828<div class="doc_subsubsection">
6829 <a name="int_atomic_load_max">'<tt>llvm.atomic.load.max.*</tt>' Intrinsic</a><br>
6830 <a name="int_atomic_load_min">'<tt>llvm.atomic.load.min.*</tt>' Intrinsic</a><br>
6831 <a name="int_atomic_load_umax">'<tt>llvm.atomic.load.umax.*</tt>' Intrinsic</a><br>
6832 <a name="int_atomic_load_umin">'<tt>llvm.atomic.load.umin.*</tt>' Intrinsic</a><br>
6833
6834</div>
6835<div class="doc_text">
6836<h5>Syntax:</h5>
6837<p>
6838 These are overloaded intrinsics. You can use <tt>llvm.atomic.load_max</tt>,
6839 <tt>llvm.atomic.load_min</tt>, <tt>llvm.atomic.load_umax</tt>, and
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6840 <tt>llvm.atomic.load_umin</tt> on any integer bit width and for different
6841 address spaces. Not all targets
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6842 support all bit widths however.</p>
6843<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6844declare i8 @llvm.atomic.load.max.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6845declare i16 @llvm.atomic.load.max.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6846declare i32 @llvm.atomic.load.max.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6847declare 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]6848
6849</pre>
6850
6851<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6852declare i8 @llvm.atomic.load.min.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6853declare i16 @llvm.atomic.load.min.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6854declare i32 @llvm.atomic.load.min.i32..p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6855declare 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]6856
6857</pre>
6858
6859<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6860declare i8 @llvm.atomic.load.umax.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6861declare i16 @llvm.atomic.load.umax.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6862declare i32 @llvm.atomic.load.umax.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6863declare 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]6864
6865</pre>
6866
6867<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6868declare i8 @llvm.atomic.load.umin.i8..p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6869declare i16 @llvm.atomic.load.umin.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6870declare i32 @llvm.atomic.load.umin.i32..p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6871declare 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]6872
6873</pre>
6874<h5>Overview:</h5>
6875<p>
6876 These intrinsics takes the signed or unsigned minimum or maximum of
6877 <tt>delta</tt> and the value stored in memory at <tt>ptr</tt>. It yields the
6878 original value at <tt>ptr</tt>.
6879</p>
6880<h5>Arguments:</h5>
6881<p>
6882
6883 These intrinsics take two arguments, the first a pointer to an integer value
6884 and the second an integer value. The result is also an integer value. These
6885 integer types can have any bit width, but they must all have the same bit
6886 width. The targets may only lower integer representations they support.
6887</p>
6888<h5>Semantics:</h5>
6889<p>
6890 These intrinsics does a series of operations atomically. They first load the
6891 value stored at <tt>ptr</tt>. They then do the signed or unsigned min or max
6892 <tt>delta</tt> and the value, store the result to <tt>ptr</tt>. They yield
6893 the original value stored at <tt>ptr</tt>.
6894</p>
6895
6896<h5>Examples:</h5>
6897<pre>
6898%ptr = malloc i32
6899 store i32 7, %ptr
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6900%result0 = call i32 @llvm.atomic.load.min.i32.p0i32( i32* %ptr, i32 -2 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6901 <i>; yields {i32}:result0 = 7</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6902%result1 = call i32 @llvm.atomic.load.max.i32.p0i32( i32* %ptr, i32 8 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6903 <i>; yields {i32}:result1 = -2</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6904%result2 = call i32 @llvm.atomic.load.umin.i32.p0i32( i32* %ptr, i32 10 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6905 <i>; yields {i32}:result2 = 8</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6906%result3 = call i32 @llvm.atomic.load.umax.i32.p0i32( i32* %ptr, i32 30 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6907 <i>; yields {i32}:result3 = 8</i>
6908%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 30</i>
6909</pre>
6910</div>
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]6911
6912<!-- ======================================================================= -->
6913<div class="doc_subsection">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6914 <a name="int_general">General Intrinsics</a>
6915</div>
6916
6917<div class="doc_text">
6918<p> This class of intrinsics is designed to be generic and has
6919no specific purpose. </p>
6920</div>
6921
6922<!-- _______________________________________________________________________ -->
6923<div class="doc_subsubsection">
6924 <a name="int_var_annotation">'<tt>llvm.var.annotation</tt>' Intrinsic</a>
6925</div>
6926
6927<div class="doc_text">
6928
6929<h5>Syntax:</h5>
6930<pre>
6931 declare void @llvm.var.annotation(i8* &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
6932</pre>
6933
6934<h5>Overview:</h5>
6935
6936<p>
6937The '<tt>llvm.var.annotation</tt>' intrinsic
6938</p>
6939
6940<h5>Arguments:</h5>
6941
6942<p>
6943The first argument is a pointer to a value, the second is a pointer to a
6944global string, the third is a pointer to a global string which is the source
6945file name, and the last argument is the line number.
6946</p>
6947
6948<h5>Semantics:</h5>
6949
6950<p>
Anton Korobeynikove6e764f2008-01-15 22:31:34 +0000[diff] [blame]6951This intrinsic allows annotation of local variables with arbitrary strings.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6952This can be useful for special purpose optimizations that want to look for these
Anton Korobeynikove6e764f2008-01-15 22:31:34 +0000[diff] [blame]6953annotations. These have no other defined use, they are ignored by code
6954generation and optimization.
6955</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6956</div>
6957
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]6958<!-- _______________________________________________________________________ -->
6959<div class="doc_subsubsection">
Tanya Lattnerc9869b12007-09-21 23:57:59 +0000[diff] [blame]6960 <a name="int_annotation">'<tt>llvm.annotation.*</tt>' Intrinsic</a>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]6961</div>
6962
6963<div class="doc_text">
6964
6965<h5>Syntax:</h5>
Tanya Lattnere545be72007-09-21 23:56:27 +0000[diff] [blame]6966<p>This is an overloaded intrinsic. You can use '<tt>llvm.annotation</tt>' on
6967any integer bit width.
6968</p>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]6969<pre>
Tanya Lattner09161fe2007-09-22 00:03:01 +0000[diff] [blame]6970 declare i8 @llvm.annotation.i8(i8 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
6971 declare i16 @llvm.annotation.i16(i16 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
6972 declare i32 @llvm.annotation.i32(i32 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
6973 declare i64 @llvm.annotation.i64(i64 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
6974 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]6975</pre>
6976
6977<h5>Overview:</h5>
Tanya Lattnere545be72007-09-21 23:56:27 +0000[diff] [blame]6978
6979<p>
6980The '<tt>llvm.annotation</tt>' intrinsic.
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]6981</p>
6982
6983<h5>Arguments:</h5>
6984
6985<p>
6986The first argument is an integer value (result of some expression),
6987the second is a pointer to a global string, the third is a pointer to a global
6988string which is the source file name, and the last argument is the line number.
Tanya Lattnere545be72007-09-21 23:56:27 +0000[diff] [blame]6989It returns the value of the first argument.
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]6990</p>
6991
6992<h5>Semantics:</h5>
6993
6994<p>
6995This intrinsic allows annotations to be put on arbitrary expressions
6996with arbitrary strings. This can be useful for special purpose optimizations
6997that want to look for these annotations. These have no other defined use, they
6998are ignored by code generation and optimization.
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]6999</p>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]7000</div>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]7001
Anton Korobeynikove6e764f2008-01-15 22:31:34 +0000[diff] [blame]7002<!-- _______________________________________________________________________ -->
7003<div class="doc_subsubsection">
7004 <a name="int_trap">'<tt>llvm.trap</tt>' Intrinsic</a>
7005</div>
7006
7007<div class="doc_text">
7008
7009<h5>Syntax:</h5>
7010<pre>
7011 declare void @llvm.trap()
7012</pre>
7013
7014<h5>Overview:</h5>
7015
7016<p>
7017The '<tt>llvm.trap</tt>' intrinsic
7018</p>
7019
7020<h5>Arguments:</h5>
7021
7022<p>
7023None
7024</p>
7025
7026<h5>Semantics:</h5>
7027
7028<p>
7029This intrinsics is lowered to the target dependent trap instruction. If the
7030target does not have a trap instruction, this intrinsic will be lowered to the
7031call of the abort() function.
7032</p>
7033</div>
7034
Bill Wendlinge4164592008-11-19 05:56:17 +0000[diff] [blame]7035<!-- _______________________________________________________________________ -->
7036<div class="doc_subsubsection">
Misha Brukman5dd7f4d2008-11-22 23:55:29 +0000[diff] [blame]7037 <a name="int_stackprotector">'<tt>llvm.stackprotector</tt>' Intrinsic</a>
Bill Wendlinge4164592008-11-19 05:56:17 +0000[diff] [blame]7038</div>
7039<div class="doc_text">
7040<h5>Syntax:</h5>
7041<pre>
7042declare void @llvm.stackprotector( i8* &lt;guard&gt;, i8** &lt;slot&gt; )
7043
7044</pre>
7045<h5>Overview:</h5>
7046<p>
7047 The <tt>llvm.stackprotector</tt> intrinsic takes the <tt>guard</tt> and stores
7048 it onto the stack at <tt>slot</tt>. The stack slot is adjusted to ensure that
7049 it is placed on the stack before local variables.
7050</p>
7051<h5>Arguments:</h5>
7052<p>
7053 The <tt>llvm.stackprotector</tt> intrinsic requires two pointer arguments. The
7054 first argument is the value loaded from the stack guard
7055 <tt>@__stack_chk_guard</tt>. The second variable is an <tt>alloca</tt> that
7056 has enough space to hold the value of the guard.
7057</p>
7058<h5>Semantics:</h5>
7059<p>
7060 This intrinsic causes the prologue/epilogue inserter to force the position of
7061 the <tt>AllocaInst</tt> stack slot to be before local variables on the
7062 stack. This is to ensure that if a local variable on the stack is overwritten,
7063 it will destroy the value of the guard. When the function exits, the guard on
7064 the stack is checked against the original guard. If they're different, then
7065 the program aborts by calling the <tt>__stack_chk_fail()</tt> function.
7066</p>
7067</div>
7068
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]7069<!-- *********************************************************************** -->
7070<hr>
7071<address>
7072 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
Misha Brukman947321d2008-12-11 17:34:48 +0000[diff] [blame]7073 src="http://jigsaw.w3.org/css-validator/images/vcss-blue" alt="Valid CSS"></a>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]7074 <a href="http://validator.w3.org/check/referer"><img
Misha Brukman947321d2008-12-11 17:34:48 +0000[diff] [blame]7075 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]7076
7077 <a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
7078 <a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br>
7079 Last modified: $Date$
7080</address>
Chris Lattner08497ce2008-01-04 04:33:49 +0000[diff] [blame]7081
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]7082</body>
7083</html>