Gitiles
Code Review Sign In
gerrit-public.fairphone.software / fp2-dev / platform / external / llvm / d468f1c7d7e83ebcf1631d7e528f2ad7ccacf5db / . / docs / LangRef.html
blob: 566911452731acd49e27935b3950bb21203d7d03 [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>
25 <li><a href="#globalvars">Global Variables</a></li>
26 <li><a href="#functionstructure">Functions</a></li>
27 <li><a href="#aliasstructure">Aliases</a>
28 <li><a href="#paramattrs">Parameter Attributes</a></li>
Gordon Henriksen13fe5e32007-12-10 03:18:06 +0000[diff] [blame]29 <li><a href="#gc">Garbage Collector Names</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]30 <li><a href="#moduleasm">Module-Level Inline Assembly</a></li>
31 <li><a href="#datalayout">Data Layout</a></li>
32 </ol>
33 </li>
34 <li><a href="#typesystem">Type System</a>
35 <ol>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]36 <li><a href="#t_classifications">Type Classifications</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]37 <li><a href="#t_primitive">Primitive Types</a>
38 <ol>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]39 <li><a href="#t_floating">Floating Point Types</a></li>
40 <li><a href="#t_void">Void Type</a></li>
41 <li><a href="#t_label">Label Type</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]42 </ol>
43 </li>
44 <li><a href="#t_derived">Derived Types</a>
45 <ol>
Chris Lattner251ab812007-12-18 06:18:21 +0000[diff] [blame]46 <li><a href="#t_integer">Integer Type</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]47 <li><a href="#t_array">Array Type</a></li>
48 <li><a href="#t_function">Function Type</a></li>
49 <li><a href="#t_pointer">Pointer Type</a></li>
50 <li><a href="#t_struct">Structure Type</a></li>
51 <li><a href="#t_pstruct">Packed Structure Type</a></li>
52 <li><a href="#t_vector">Vector Type</a></li>
53 <li><a href="#t_opaque">Opaque Type</a></li>
54 </ol>
55 </li>
56 </ol>
57 </li>
58 <li><a href="#constants">Constants</a>
59 <ol>
60 <li><a href="#simpleconstants">Simple Constants</a>
61 <li><a href="#aggregateconstants">Aggregate Constants</a>
62 <li><a href="#globalconstants">Global Variable and Function Addresses</a>
63 <li><a href="#undefvalues">Undefined Values</a>
64 <li><a href="#constantexprs">Constant Expressions</a>
65 </ol>
66 </li>
67 <li><a href="#othervalues">Other Values</a>
68 <ol>
69 <li><a href="#inlineasm">Inline Assembler Expressions</a>
70 </ol>
71 </li>
72 <li><a href="#instref">Instruction Reference</a>
73 <ol>
74 <li><a href="#terminators">Terminator Instructions</a>
75 <ol>
76 <li><a href="#i_ret">'<tt>ret</tt>' Instruction</a></li>
77 <li><a href="#i_br">'<tt>br</tt>' Instruction</a></li>
78 <li><a href="#i_switch">'<tt>switch</tt>' Instruction</a></li>
79 <li><a href="#i_invoke">'<tt>invoke</tt>' Instruction</a></li>
80 <li><a href="#i_unwind">'<tt>unwind</tt>' Instruction</a></li>
81 <li><a href="#i_unreachable">'<tt>unreachable</tt>' Instruction</a></li>
82 </ol>
83 </li>
84 <li><a href="#binaryops">Binary Operations</a>
85 <ol>
86 <li><a href="#i_add">'<tt>add</tt>' Instruction</a></li>
87 <li><a href="#i_sub">'<tt>sub</tt>' Instruction</a></li>
88 <li><a href="#i_mul">'<tt>mul</tt>' Instruction</a></li>
89 <li><a href="#i_udiv">'<tt>udiv</tt>' Instruction</a></li>
90 <li><a href="#i_sdiv">'<tt>sdiv</tt>' Instruction</a></li>
91 <li><a href="#i_fdiv">'<tt>fdiv</tt>' Instruction</a></li>
92 <li><a href="#i_urem">'<tt>urem</tt>' Instruction</a></li>
93 <li><a href="#i_srem">'<tt>srem</tt>' Instruction</a></li>
94 <li><a href="#i_frem">'<tt>frem</tt>' Instruction</a></li>
95 </ol>
96 </li>
97 <li><a href="#bitwiseops">Bitwise Binary Operations</a>
98 <ol>
99 <li><a href="#i_shl">'<tt>shl</tt>' Instruction</a></li>
100 <li><a href="#i_lshr">'<tt>lshr</tt>' Instruction</a></li>
101 <li><a href="#i_ashr">'<tt>ashr</tt>' Instruction</a></li>
102 <li><a href="#i_and">'<tt>and</tt>' Instruction</a></li>
103 <li><a href="#i_or">'<tt>or</tt>' Instruction</a></li>
104 <li><a href="#i_xor">'<tt>xor</tt>' Instruction</a></li>
105 </ol>
106 </li>
107 <li><a href="#vectorops">Vector Operations</a>
108 <ol>
109 <li><a href="#i_extractelement">'<tt>extractelement</tt>' Instruction</a></li>
110 <li><a href="#i_insertelement">'<tt>insertelement</tt>' Instruction</a></li>
111 <li><a href="#i_shufflevector">'<tt>shufflevector</tt>' Instruction</a></li>
112 </ol>
113 </li>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]114 <li><a href="#aggregateops">Aggregate Operations</a>
115 <ol>
116 <li><a href="#i_extractvalue">'<tt>extractvalue</tt>' Instruction</a></li>
117 <li><a href="#i_insertvalue">'<tt>insertvalue</tt>' Instruction</a></li>
118 </ol>
119 </li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]120 <li><a href="#memoryops">Memory Access and Addressing Operations</a>
121 <ol>
122 <li><a href="#i_malloc">'<tt>malloc</tt>' Instruction</a></li>
123 <li><a href="#i_free">'<tt>free</tt>' Instruction</a></li>
124 <li><a href="#i_alloca">'<tt>alloca</tt>' Instruction</a></li>
125 <li><a href="#i_load">'<tt>load</tt>' Instruction</a></li>
126 <li><a href="#i_store">'<tt>store</tt>' Instruction</a></li>
127 <li><a href="#i_getelementptr">'<tt>getelementptr</tt>' Instruction</a></li>
128 </ol>
129 </li>
130 <li><a href="#convertops">Conversion Operations</a>
131 <ol>
132 <li><a href="#i_trunc">'<tt>trunc .. to</tt>' Instruction</a></li>
133 <li><a href="#i_zext">'<tt>zext .. to</tt>' Instruction</a></li>
134 <li><a href="#i_sext">'<tt>sext .. to</tt>' Instruction</a></li>
135 <li><a href="#i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a></li>
136 <li><a href="#i_fpext">'<tt>fpext .. to</tt>' Instruction</a></li>
137 <li><a href="#i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a></li>
138 <li><a href="#i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a></li>
139 <li><a href="#i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a></li>
140 <li><a href="#i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a></li>
141 <li><a href="#i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a></li>
142 <li><a href="#i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a></li>
143 <li><a href="#i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a></li>
144 </ol>
145 <li><a href="#otherops">Other Operations</a>
146 <ol>
147 <li><a href="#i_icmp">'<tt>icmp</tt>' Instruction</a></li>
148 <li><a href="#i_fcmp">'<tt>fcmp</tt>' Instruction</a></li>
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]149 <li><a href="#i_vicmp">'<tt>vicmp</tt>' Instruction</a></li>
150 <li><a href="#i_vfcmp">'<tt>vfcmp</tt>' Instruction</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]151 <li><a href="#i_phi">'<tt>phi</tt>' Instruction</a></li>
152 <li><a href="#i_select">'<tt>select</tt>' Instruction</a></li>
153 <li><a href="#i_call">'<tt>call</tt>' Instruction</a></li>
154 <li><a href="#i_va_arg">'<tt>va_arg</tt>' Instruction</a></li>
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]155 <li><a href="#i_getresult">'<tt>getresult</tt>' Instruction</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]156 </ol>
157 </li>
158 </ol>
159 </li>
160 <li><a href="#intrinsics">Intrinsic Functions</a>
161 <ol>
162 <li><a href="#int_varargs">Variable Argument Handling Intrinsics</a>
163 <ol>
164 <li><a href="#int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a></li>
165 <li><a href="#int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a></li>
166 <li><a href="#int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a></li>
167 </ol>
168 </li>
169 <li><a href="#int_gc">Accurate Garbage Collection Intrinsics</a>
170 <ol>
171 <li><a href="#int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a></li>
172 <li><a href="#int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a></li>
173 <li><a href="#int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a></li>
174 </ol>
175 </li>
176 <li><a href="#int_codegen">Code Generator Intrinsics</a>
177 <ol>
178 <li><a href="#int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a></li>
179 <li><a href="#int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a></li>
180 <li><a href="#int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a></li>
181 <li><a href="#int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a></li>
182 <li><a href="#int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a></li>
183 <li><a href="#int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a></li>
184 <li><a href="#int_readcyclecounter"><tt>llvm.readcyclecounter</tt>' Intrinsic</a></li>
185 </ol>
186 </li>
187 <li><a href="#int_libc">Standard C Library Intrinsics</a>
188 <ol>
189 <li><a href="#int_memcpy">'<tt>llvm.memcpy.*</tt>' Intrinsic</a></li>
190 <li><a href="#int_memmove">'<tt>llvm.memmove.*</tt>' Intrinsic</a></li>
191 <li><a href="#int_memset">'<tt>llvm.memset.*</tt>' Intrinsic</a></li>
192 <li><a href="#int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a></li>
193 <li><a href="#int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a></li>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]194 <li><a href="#int_sin">'<tt>llvm.sin.*</tt>' Intrinsic</a></li>
195 <li><a href="#int_cos">'<tt>llvm.cos.*</tt>' Intrinsic</a></li>
196 <li><a href="#int_pow">'<tt>llvm.pow.*</tt>' Intrinsic</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]197 </ol>
198 </li>
199 <li><a href="#int_manip">Bit Manipulation Intrinsics</a>
200 <ol>
201 <li><a href="#int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a></li>
202 <li><a href="#int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic </a></li>
203 <li><a href="#int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic </a></li>
204 <li><a href="#int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic </a></li>
205 <li><a href="#int_part_select">'<tt>llvm.part.select.*</tt>' Intrinsic </a></li>
206 <li><a href="#int_part_set">'<tt>llvm.part.set.*</tt>' Intrinsic </a></li>
207 </ol>
208 </li>
209 <li><a href="#int_debugger">Debugger intrinsics</a></li>
210 <li><a href="#int_eh">Exception Handling intrinsics</a></li>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]211 <li><a href="#int_trampoline">Trampoline Intrinsic</a>
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]212 <ol>
213 <li><a href="#int_it">'<tt>llvm.init.trampoline</tt>' Intrinsic</a></li>
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]214 </ol>
215 </li>
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]216 <li><a href="#int_atomics">Atomic intrinsics</a>
217 <ol>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]218 <li><a href="#int_memory_barrier"><tt>llvm.memory_barrier</tt></a></li>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]219 <li><a href="#int_atomic_cmp_swap"><tt>llvm.atomic.cmp.swap</tt></a></li>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]220 <li><a href="#int_atomic_swap"><tt>llvm.atomic.swap</tt></a></li>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]221 <li><a href="#int_atomic_load_add"><tt>llvm.atomic.load.add</tt></a></li>
222 <li><a href="#int_atomic_load_sub"><tt>llvm.atomic.load.sub</tt></a></li>
223 <li><a href="#int_atomic_load_and"><tt>llvm.atomic.load.and</tt></a></li>
224 <li><a href="#int_atomic_load_nand"><tt>llvm.atomic.load.nand</tt></a></li>
225 <li><a href="#int_atomic_load_or"><tt>llvm.atomic.load.or</tt></a></li>
226 <li><a href="#int_atomic_load_xor"><tt>llvm.atomic.load.xor</tt></a></li>
227 <li><a href="#int_atomic_load_max"><tt>llvm.atomic.load.max</tt></a></li>
228 <li><a href="#int_atomic_load_min"><tt>llvm.atomic.load.min</tt></a></li>
229 <li><a href="#int_atomic_load_umax"><tt>llvm.atomic.load.umax</tt></a></li>
230 <li><a href="#int_atomic_load_umin"><tt>llvm.atomic.load.umin</tt></a></li>
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]231 </ol>
232 </li>
Reid Spencerb043f672007-07-20 19:59:11 +0000[diff] [blame]233 <li><a href="#int_general">General intrinsics</a>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]234 <ol>
Reid Spencerb043f672007-07-20 19:59:11 +0000[diff] [blame]235 <li><a href="#int_var_annotation">
Tanya Lattner51369f32007-09-22 00:01:26 +0000[diff] [blame]236 <tt>llvm.var.annotation</tt>' Intrinsic</a></li>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]237 <li><a href="#int_annotation">
Tanya Lattner51369f32007-09-22 00:01:26 +0000[diff] [blame]238 <tt>llvm.annotation.*</tt>' Intrinsic</a></li>
Anton Korobeynikove6e764f2008-01-15 22:31:34 +0000[diff] [blame]239 <li><a href="#int_trap">
240 <tt>llvm.trap</tt>' Intrinsic</a></li>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]241 </ol>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]242 </li>
243 </ol>
244 </li>
245</ol>
246
247<div class="doc_author">
248 <p>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a>
249 and <a href="mailto:vadve@cs.uiuc.edu">Vikram Adve</a></p>
250</div>
251
252<!-- *********************************************************************** -->
253<div class="doc_section"> <a name="abstract">Abstract </a></div>
254<!-- *********************************************************************** -->
255
256<div class="doc_text">
257<p>This document is a reference manual for the LLVM assembly language.
Bill Wendlinge7846a52008-08-05 22:29:16 +0000[diff] [blame]258LLVM is a Static Single Assignment (SSA) based representation that provides
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]259type safety, low-level operations, flexibility, and the capability of
260representing 'all' high-level languages cleanly. It is the common code
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]261representation used throughout all phases of the LLVM compilation
262strategy.</p>
263</div>
264
265<!-- *********************************************************************** -->
266<div class="doc_section"> <a name="introduction">Introduction</a> </div>
267<!-- *********************************************************************** -->
268
269<div class="doc_text">
270
271<p>The LLVM code representation is designed to be used in three
272different forms: as an in-memory compiler IR, as an on-disk bitcode
273representation (suitable for fast loading by a Just-In-Time compiler),
274and as a human readable assembly language representation. This allows
275LLVM to provide a powerful intermediate representation for efficient
276compiler transformations and analysis, while providing a natural means
277to debug and visualize the transformations. The three different forms
278of LLVM are all equivalent. This document describes the human readable
279representation and notation.</p>
280
281<p>The LLVM representation aims to be light-weight and low-level
282while being expressive, typed, and extensible at the same time. It
283aims to be a "universal IR" of sorts, by being at a low enough level
284that high-level ideas may be cleanly mapped to it (similar to how
285microprocessors are "universal IR's", allowing many source languages to
286be mapped to them). By providing type information, LLVM can be used as
287the target of optimizations: for example, through pointer analysis, it
288can be proven that a C automatic variable is never accessed outside of
289the current function... allowing it to be promoted to a simple SSA
290value instead of a memory location.</p>
291
292</div>
293
294<!-- _______________________________________________________________________ -->
295<div class="doc_subsubsection"> <a name="wellformed">Well-Formedness</a> </div>
296
297<div class="doc_text">
298
299<p>It is important to note that this document describes 'well formed'
300LLVM assembly language. There is a difference between what the parser
301accepts and what is considered 'well formed'. For example, the
302following instruction is syntactically okay, but not well formed:</p>
303
304<div class="doc_code">
305<pre>
306%x = <a href="#i_add">add</a> i32 1, %x
307</pre>
308</div>
309
310<p>...because the definition of <tt>%x</tt> does not dominate all of
311its uses. The LLVM infrastructure provides a verification pass that may
312be used to verify that an LLVM module is well formed. This pass is
313automatically run by the parser after parsing input assembly and by
314the optimizer before it outputs bitcode. The violations pointed out
315by the verifier pass indicate bugs in transformation passes or input to
316the parser.</p>
317</div>
318
Chris Lattnera83fdc02007-10-03 17:34:29 +0000[diff] [blame]319<!-- Describe the typesetting conventions here. -->
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]320
321<!-- *********************************************************************** -->
322<div class="doc_section"> <a name="identifiers">Identifiers</a> </div>
323<!-- *********************************************************************** -->
324
325<div class="doc_text">
326
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]327 <p>LLVM identifiers come in two basic types: global and local. Global
328 identifiers (functions, global variables) begin with the @ character. Local
329 identifiers (register names, types) begin with the % character. Additionally,
330 there are three different formats for identifiers, for different purposes:
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]331
332<ol>
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]333 <li>Named values are represented as a string of characters with their prefix.
334 For example, %foo, @DivisionByZero, %a.really.long.identifier. The actual
335 regular expression used is '<tt>[%@][a-zA-Z$._][a-zA-Z$._0-9]*</tt>'.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]336 Identifiers which require other characters in their names can be surrounded
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]337 with quotes. In this way, anything except a <tt>&quot;</tt> character can
338 be used in a named value.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]339
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]340 <li>Unnamed values are represented as an unsigned numeric value with their
341 prefix. For example, %12, @2, %44.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]342
343 <li>Constants, which are described in a <a href="#constants">section about
344 constants</a>, below.</li>
345</ol>
346
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]347<p>LLVM requires that values start with a prefix for two reasons: Compilers
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]348don't need to worry about name clashes with reserved words, and the set of
349reserved words may be expanded in the future without penalty. Additionally,
350unnamed identifiers allow a compiler to quickly come up with a temporary
351variable without having to avoid symbol table conflicts.</p>
352
353<p>Reserved words in LLVM are very similar to reserved words in other
354languages. There are keywords for different opcodes
355('<tt><a href="#i_add">add</a></tt>',
356 '<tt><a href="#i_bitcast">bitcast</a></tt>',
357 '<tt><a href="#i_ret">ret</a></tt>', etc...), for primitive type names ('<tt><a
358href="#t_void">void</a></tt>', '<tt><a href="#t_primitive">i32</a></tt>', etc...),
359and others. These reserved words cannot conflict with variable names, because
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]360none of them start with a prefix character ('%' or '@').</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]361
362<p>Here is an example of LLVM code to multiply the integer variable
363'<tt>%X</tt>' by 8:</p>
364
365<p>The easy way:</p>
366
367<div class="doc_code">
368<pre>
369%result = <a href="#i_mul">mul</a> i32 %X, 8
370</pre>
371</div>
372
373<p>After strength reduction:</p>
374
375<div class="doc_code">
376<pre>
377%result = <a href="#i_shl">shl</a> i32 %X, i8 3
378</pre>
379</div>
380
381<p>And the hard way:</p>
382
383<div class="doc_code">
384<pre>
385<a href="#i_add">add</a> i32 %X, %X <i>; yields {i32}:%0</i>
386<a href="#i_add">add</a> i32 %0, %0 <i>; yields {i32}:%1</i>
387%result = <a href="#i_add">add</a> i32 %1, %1
388</pre>
389</div>
390
391<p>This last way of multiplying <tt>%X</tt> by 8 illustrates several
392important lexical features of LLVM:</p>
393
394<ol>
395
396 <li>Comments are delimited with a '<tt>;</tt>' and go until the end of
397 line.</li>
398
399 <li>Unnamed temporaries are created when the result of a computation is not
400 assigned to a named value.</li>
401
402 <li>Unnamed temporaries are numbered sequentially</li>
403
404</ol>
405
406<p>...and it also shows a convention that we follow in this document. When
407demonstrating instructions, we will follow an instruction with a comment that
408defines the type and name of value produced. Comments are shown in italic
409text.</p>
410
411</div>
412
413<!-- *********************************************************************** -->
414<div class="doc_section"> <a name="highlevel">High Level Structure</a> </div>
415<!-- *********************************************************************** -->
416
417<!-- ======================================================================= -->
418<div class="doc_subsection"> <a name="modulestructure">Module Structure</a>
419</div>
420
421<div class="doc_text">
422
423<p>LLVM programs are composed of "Module"s, each of which is a
424translation unit of the input programs. Each module consists of
425functions, global variables, and symbol table entries. Modules may be
426combined together with the LLVM linker, which merges function (and
427global variable) definitions, resolves forward declarations, and merges
428symbol table entries. Here is an example of the "hello world" module:</p>
429
430<div class="doc_code">
431<pre><i>; Declare the string constant as a global constant...</i>
432<a href="#identifiers">@.LC0</a> = <a href="#linkage_internal">internal</a> <a
433 href="#globalvars">constant</a> <a href="#t_array">[13 x i8]</a> c"hello world\0A\00" <i>; [13 x i8]*</i>
434
435<i>; External declaration of the puts function</i>
436<a href="#functionstructure">declare</a> i32 @puts(i8 *) <i>; i32(i8 *)* </i>
437
438<i>; Definition of main function</i>
439define i32 @main() { <i>; i32()* </i>
440 <i>; Convert [13x i8 ]* to i8 *...</i>
441 %cast210 = <a
442 href="#i_getelementptr">getelementptr</a> [13 x i8 ]* @.LC0, i64 0, i64 0 <i>; i8 *</i>
443
444 <i>; Call puts function to write out the string to stdout...</i>
445 <a
446 href="#i_call">call</a> i32 @puts(i8 * %cast210) <i>; i32</i>
447 <a
448 href="#i_ret">ret</a> i32 0<br>}<br>
449</pre>
450</div>
451
452<p>This example is made up of a <a href="#globalvars">global variable</a>
453named "<tt>.LC0</tt>", an external declaration of the "<tt>puts</tt>"
454function, and a <a href="#functionstructure">function definition</a>
455for "<tt>main</tt>".</p>
456
457<p>In general, a module is made up of a list of global values,
458where both functions and global variables are global values. Global values are
459represented by a pointer to a memory location (in this case, a pointer to an
460array of char, and a pointer to a function), and have one of the following <a
461href="#linkage">linkage types</a>.</p>
462
463</div>
464
465<!-- ======================================================================= -->
466<div class="doc_subsection">
467 <a name="linkage">Linkage Types</a>
468</div>
469
470<div class="doc_text">
471
472<p>
473All Global Variables and Functions have one of the following types of linkage:
474</p>
475
476<dl>
477
Dale Johannesen96e7e092008-05-23 23:13:41 +0000[diff] [blame]478 <dt><tt><b><a name="linkage_internal">internal</a></b></tt>: </dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]479
480 <dd>Global values with internal linkage are only directly accessible by
481 objects in the current module. In particular, linking code into a module with
482 an internal global value may cause the internal to be renamed as necessary to
483 avoid collisions. Because the symbol is internal to the module, all
484 references can be updated. This corresponds to the notion of the
485 '<tt>static</tt>' keyword in C.
486 </dd>
487
488 <dt><tt><b><a name="linkage_linkonce">linkonce</a></b></tt>: </dt>
489
490 <dd>Globals with "<tt>linkonce</tt>" linkage are merged with other globals of
491 the same name when linkage occurs. This is typically used to implement
492 inline functions, templates, or other code which must be generated in each
493 translation unit that uses it. Unreferenced <tt>linkonce</tt> globals are
494 allowed to be discarded.
495 </dd>
496
Dale Johannesen96e7e092008-05-23 23:13:41 +0000[diff] [blame]497 <dt><tt><b><a name="linkage_common">common</a></b></tt>: </dt>
498
499 <dd>"<tt>common</tt>" linkage is exactly the same as <tt>linkonce</tt>
500 linkage, except that unreferenced <tt>common</tt> globals may not be
501 discarded. This is used for globals that may be emitted in multiple
502 translation units, but that are not guaranteed to be emitted into every
503 translation unit that uses them. One example of this is tentative
504 definitions in C, such as "<tt>int X;</tt>" at global scope.
505 </dd>
506
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]507 <dt><tt><b><a name="linkage_weak">weak</a></b></tt>: </dt>
508
Dale Johannesen96e7e092008-05-23 23:13:41 +0000[diff] [blame]509 <dd>"<tt>weak</tt>" linkage is the same as <tt>common</tt> linkage, except
510 that some targets may choose to emit different assembly sequences for them
511 for target-dependent reasons. This is used for globals that are declared
512 "weak" in C source code.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]513 </dd>
514
515 <dt><tt><b><a name="linkage_appending">appending</a></b></tt>: </dt>
516
517 <dd>"<tt>appending</tt>" linkage may only be applied to global variables of
518 pointer to array type. When two global variables with appending linkage are
519 linked together, the two global arrays are appended together. This is the
520 LLVM, typesafe, equivalent of having the system linker append together
521 "sections" with identical names when .o files are linked.
522 </dd>
523
524 <dt><tt><b><a name="linkage_externweak">extern_weak</a></b></tt>: </dt>
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]525 <dd>The semantics of this linkage follow the ELF object file model: the
526 symbol is weak until linked, if not linked, the symbol becomes null instead
527 of being an undefined reference.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]528 </dd>
529
530 <dt><tt><b><a name="linkage_external">externally visible</a></b></tt>:</dt>
531
532 <dd>If none of the above identifiers are used, the global is externally
533 visible, meaning that it participates in linkage and can be used to resolve
534 external symbol references.
535 </dd>
536</dl>
537
538 <p>
539 The next two types of linkage are targeted for Microsoft Windows platform
540 only. They are designed to support importing (exporting) symbols from (to)
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]541 DLLs (Dynamic Link Libraries).
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]542 </p>
543
544 <dl>
545 <dt><tt><b><a name="linkage_dllimport">dllimport</a></b></tt>: </dt>
546
547 <dd>"<tt>dllimport</tt>" linkage causes the compiler to reference a function
548 or variable via a global pointer to a pointer that is set up by the DLL
549 exporting the symbol. On Microsoft Windows targets, the pointer name is
550 formed by combining <code>_imp__</code> and the function or variable name.
551 </dd>
552
553 <dt><tt><b><a name="linkage_dllexport">dllexport</a></b></tt>: </dt>
554
555 <dd>"<tt>dllexport</tt>" linkage causes the compiler to provide a global
556 pointer to a pointer in a DLL, so that it can be referenced with the
557 <tt>dllimport</tt> attribute. On Microsoft Windows targets, the pointer
558 name is formed by combining <code>_imp__</code> and the function or variable
559 name.
560 </dd>
561
562</dl>
563
564<p><a name="linkage_external"></a>For example, since the "<tt>.LC0</tt>"
565variable is defined to be internal, if another module defined a "<tt>.LC0</tt>"
566variable and was linked with this one, one of the two would be renamed,
567preventing a collision. Since "<tt>main</tt>" and "<tt>puts</tt>" are
568external (i.e., lacking any linkage declarations), they are accessible
569outside of the current module.</p>
570<p>It is illegal for a function <i>declaration</i>
571to have any linkage type other than "externally visible", <tt>dllimport</tt>,
572or <tt>extern_weak</tt>.</p>
573<p>Aliases can have only <tt>external</tt>, <tt>internal</tt> and <tt>weak</tt>
574linkages.
575</div>
576
577<!-- ======================================================================= -->
578<div class="doc_subsection">
579 <a name="callingconv">Calling Conventions</a>
580</div>
581
582<div class="doc_text">
583
584<p>LLVM <a href="#functionstructure">functions</a>, <a href="#i_call">calls</a>
585and <a href="#i_invoke">invokes</a> can all have an optional calling convention
586specified for the call. The calling convention of any pair of dynamic
587caller/callee must match, or the behavior of the program is undefined. The
588following calling conventions are supported by LLVM, and more may be added in
589the future:</p>
590
591<dl>
592 <dt><b>"<tt>ccc</tt>" - The C calling convention</b>:</dt>
593
594 <dd>This calling convention (the default if no other calling convention is
595 specified) matches the target C calling conventions. This calling convention
596 supports varargs function calls and tolerates some mismatch in the declared
597 prototype and implemented declaration of the function (as does normal C).
598 </dd>
599
600 <dt><b>"<tt>fastcc</tt>" - The fast calling convention</b>:</dt>
601
602 <dd>This calling convention attempts to make calls as fast as possible
603 (e.g. by passing things in registers). This calling convention allows the
604 target to use whatever tricks it wants to produce fast code for the target,
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]605 without having to conform to an externally specified ABI (Application Binary
606 Interface). Implementations of this convention should allow arbitrary
Arnold Schwaighofer07444922008-05-14 09:17:12 +0000[diff] [blame]607 <a href="CodeGenerator.html#tailcallopt">tail call optimization</a> to be
608 supported. This calling convention does not support varargs and requires the
609 prototype of all callees to exactly match the prototype of the function
610 definition.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]611 </dd>
612
613 <dt><b>"<tt>coldcc</tt>" - The cold calling convention</b>:</dt>
614
615 <dd>This calling convention attempts to make code in the caller as efficient
616 as possible under the assumption that the call is not commonly executed. As
617 such, these calls often preserve all registers so that the call does not break
618 any live ranges in the caller side. This calling convention does not support
619 varargs and requires the prototype of all callees to exactly match the
620 prototype of the function definition.
621 </dd>
622
623 <dt><b>"<tt>cc &lt;<em>n</em>&gt;</tt>" - Numbered convention</b>:</dt>
624
625 <dd>Any calling convention may be specified by number, allowing
626 target-specific calling conventions to be used. Target specific calling
627 conventions start at 64.
628 </dd>
629</dl>
630
631<p>More calling conventions can be added/defined on an as-needed basis, to
632support pascal conventions or any other well-known target-independent
633convention.</p>
634
635</div>
636
637<!-- ======================================================================= -->
638<div class="doc_subsection">
639 <a name="visibility">Visibility Styles</a>
640</div>
641
642<div class="doc_text">
643
644<p>
645All Global Variables and Functions have one of the following visibility styles:
646</p>
647
648<dl>
649 <dt><b>"<tt>default</tt>" - Default style</b>:</dt>
650
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]651 <dd>On targets that use the ELF object file format, default visibility means
652 that the declaration is visible to other
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]653 modules and, in shared libraries, means that the declared entity may be
654 overridden. On Darwin, default visibility means that the declaration is
655 visible to other modules. Default visibility corresponds to "external
656 linkage" in the language.
657 </dd>
658
659 <dt><b>"<tt>hidden</tt>" - Hidden style</b>:</dt>
660
661 <dd>Two declarations of an object with hidden visibility refer to the same
662 object if they are in the same shared object. Usually, hidden visibility
663 indicates that the symbol will not be placed into the dynamic symbol table,
664 so no other module (executable or shared library) can reference it
665 directly.
666 </dd>
667
668 <dt><b>"<tt>protected</tt>" - Protected style</b>:</dt>
669
670 <dd>On ELF, protected visibility indicates that the symbol will be placed in
671 the dynamic symbol table, but that references within the defining module will
672 bind to the local symbol. That is, the symbol cannot be overridden by another
673 module.
674 </dd>
675</dl>
676
677</div>
678
679<!-- ======================================================================= -->
680<div class="doc_subsection">
681 <a name="globalvars">Global Variables</a>
682</div>
683
684<div class="doc_text">
685
686<p>Global variables define regions of memory allocated at compilation time
687instead of run-time. Global variables may optionally be initialized, may have
688an explicit section to be placed in, and may have an optional explicit alignment
689specified. A variable may be defined as "thread_local", which means that it
690will not be shared by threads (each thread will have a separated copy of the
691variable). A variable may be defined as a global "constant," which indicates
692that the contents of the variable will <b>never</b> be modified (enabling better
693optimization, allowing the global data to be placed in the read-only section of
694an executable, etc). Note that variables that need runtime initialization
695cannot be marked "constant" as there is a store to the variable.</p>
696
697<p>
698LLVM explicitly allows <em>declarations</em> of global variables to be marked
699constant, even if the final definition of the global is not. This capability
700can be used to enable slightly better optimization of the program, but requires
701the language definition to guarantee that optimizations based on the
702'constantness' are valid for the translation units that do not include the
703definition.
704</p>
705
706<p>As SSA values, global variables define pointer values that are in
707scope (i.e. they dominate) all basic blocks in the program. Global
708variables always define a pointer to their "content" type because they
709describe a region of memory, and all memory objects in LLVM are
710accessed through pointers.</p>
711
Christopher Lambdd0049d2007-12-11 09:31:00 +0000[diff] [blame]712<p>A global variable may be declared to reside in a target-specifc numbered
713address space. For targets that support them, address spaces may affect how
714optimizations are performed and/or what target instructions are used to access
Christopher Lamb20a39e92007-12-12 08:44:39 +0000[diff] [blame]715the variable. The default address space is zero. The address space qualifier
716must precede any other attributes.</p>
Christopher Lambdd0049d2007-12-11 09:31:00 +0000[diff] [blame]717
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]718<p>LLVM allows an explicit section to be specified for globals. If the target
719supports it, it will emit globals to the section specified.</p>
720
721<p>An explicit alignment may be specified for a global. If not present, or if
722the alignment is set to zero, the alignment of the global is set by the target
723to whatever it feels convenient. If an explicit alignment is specified, the
724global is forced to have at least that much alignment. All alignments must be
725a power of 2.</p>
726
Christopher Lambdd0049d2007-12-11 09:31:00 +0000[diff] [blame]727<p>For example, the following defines a global in a numbered address space with
728an initializer, section, and alignment:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]729
730<div class="doc_code">
731<pre>
Christopher Lambdd0049d2007-12-11 09:31:00 +0000[diff] [blame]732@G = constant float 1.0 addrspace(5), section "foo", align 4
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]733</pre>
734</div>
735
736</div>
737
738
739<!-- ======================================================================= -->
740<div class="doc_subsection">
741 <a name="functionstructure">Functions</a>
742</div>
743
744<div class="doc_text">
745
746<p>LLVM function definitions consist of the "<tt>define</tt>" keyord,
747an optional <a href="#linkage">linkage type</a>, an optional
748<a href="#visibility">visibility style</a>, an optional
749<a href="#callingconv">calling convention</a>, a return type, an optional
750<a href="#paramattrs">parameter attribute</a> for the return type, a function
751name, a (possibly empty) argument list (each with optional
752<a href="#paramattrs">parameter attributes</a>), an optional section, an
Devang Pateld468f1c2008-09-04 23:05:13 +0000[diff] [blame^]753optional alignment, an optional <a href="#gc">garbage collector name</a>,
754an optional <a href="#notes">function notes</a>, an
Gordon Henriksen13fe5e32007-12-10 03:18:06 +0000[diff] [blame]755opening curly brace, a list of basic blocks, and a closing curly brace.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]756
757LLVM function declarations consist of the "<tt>declare</tt>" keyword, an
758optional <a href="#linkage">linkage type</a>, an optional
759<a href="#visibility">visibility style</a>, an optional
760<a href="#callingconv">calling convention</a>, a return type, an optional
761<a href="#paramattrs">parameter attribute</a> for the return type, a function
Gordon Henriksen13fe5e32007-12-10 03:18:06 +0000[diff] [blame]762name, a possibly empty list of arguments, an optional alignment, and an optional
Gordon Henriksend606f5b2007-12-10 03:30:21 +0000[diff] [blame]763<a href="#gc">garbage collector name</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]764
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]765<p>A function definition contains a list of basic blocks, forming the CFG
766(Control Flow Graph) for
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]767the function. Each basic block may optionally start with a label (giving the
768basic block a symbol table entry), contains a list of instructions, and ends
769with a <a href="#terminators">terminator</a> instruction (such as a branch or
770function return).</p>
771
772<p>The first basic block in a function is special in two ways: it is immediately
773executed on entrance to the function, and it is not allowed to have predecessor
774basic blocks (i.e. there can not be any branches to the entry block of a
775function). Because the block can have no predecessors, it also cannot have any
776<a href="#i_phi">PHI nodes</a>.</p>
777
778<p>LLVM allows an explicit section to be specified for functions. If the target
779supports it, it will emit functions to the section specified.</p>
780
781<p>An explicit alignment may be specified for a function. If not present, or if
782the alignment is set to zero, the alignment of the function is set by the target
783to whatever it feels convenient. If an explicit alignment is specified, the
784function is forced to have at least that much alignment. All alignments must be
785a power of 2.</p>
786
787</div>
788
789
790<!-- ======================================================================= -->
791<div class="doc_subsection">
792 <a name="aliasstructure">Aliases</a>
793</div>
794<div class="doc_text">
795 <p>Aliases act as "second name" for the aliasee value (which can be either
Anton Korobeynikov96822812008-03-22 08:36:14 +0000[diff] [blame]796 function, global variable, another alias or bitcast of global value). Aliases
797 may have an optional <a href="#linkage">linkage type</a>, and an
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]798 optional <a href="#visibility">visibility style</a>.</p>
799
800 <h5>Syntax:</h5>
801
802<div class="doc_code">
803<pre>
804@&lt;Name&gt; = [Linkage] [Visibility] alias &lt;AliaseeTy&gt; @&lt;Aliasee&gt;
805</pre>
806</div>
807
808</div>
809
810
811
812<!-- ======================================================================= -->
813<div class="doc_subsection"><a name="paramattrs">Parameter Attributes</a></div>
814<div class="doc_text">
815 <p>The return type and each parameter of a function type may have a set of
816 <i>parameter attributes</i> associated with them. Parameter attributes are
817 used to communicate additional information about the result or parameters of
Duncan Sandsf5588dc2007-11-27 13:23:08 +0000[diff] [blame]818 a function. Parameter attributes are considered to be part of the function,
819 not of the function type, so functions with different parameter attributes
820 can have the same function type.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]821
822 <p>Parameter attributes are simple keywords that follow the type specified. If
823 multiple parameter attributes are needed, they are space separated. For
824 example:</p>
825
826<div class="doc_code">
827<pre>
Duncan Sandsf5588dc2007-11-27 13:23:08 +0000[diff] [blame]828declare i32 @printf(i8* noalias , ...) nounwind
829declare i32 @atoi(i8*) nounwind readonly
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]830</pre>
831</div>
832
Duncan Sandsf5588dc2007-11-27 13:23:08 +0000[diff] [blame]833 <p>Note that any attributes for the function result (<tt>nounwind</tt>,
834 <tt>readonly</tt>) come immediately after the argument list.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]835
836 <p>Currently, only the following parameter attributes are defined:</p>
837 <dl>
Reid Spencerf234bed2007-07-19 23:13:04 +0000[diff] [blame]838 <dt><tt>zeroext</tt></dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]839 <dd>This indicates that the parameter should be zero extended just before
840 a call to this function.</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]841
Reid Spencerf234bed2007-07-19 23:13:04 +0000[diff] [blame]842 <dt><tt>signext</tt></dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]843 <dd>This indicates that the parameter should be sign extended just before
844 a call to this function.</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]845
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]846 <dt><tt>inreg</tt></dt>
847 <dd>This indicates that the parameter should be placed in register (if
848 possible) during assembling function call. Support for this attribute is
849 target-specific</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]850
851 <dt><tt>byval</tt></dt>
Chris Lattner04c86182008-01-15 04:34:22 +0000[diff] [blame]852 <dd>This indicates that the pointer parameter should really be passed by
853 value to the function. The attribute implies that a hidden copy of the
854 pointee is made between the caller and the callee, so the callee is unable
Chris Lattner6a9f3c42008-08-05 18:21:08 +0000[diff] [blame]855 to modify the value in the callee. This attribute is only valid on LLVM
Chris Lattner04c86182008-01-15 04:34:22 +0000[diff] [blame]856 pointer arguments. It is generally used to pass structs and arrays by
857 value, but is also valid on scalars (even though this is silly).</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]858
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]859 <dt><tt>sret</tt></dt>
Duncan Sands616cc032008-02-18 04:19:38 +0000[diff] [blame]860 <dd>This indicates that the pointer parameter specifies the address of a
861 structure that is the return value of the function in the source program.
Duncan Sandsef0e9e42008-03-17 12:17:41 +0000[diff] [blame]862 Loads and stores to the structure are assumed not to trap.
Duncan Sands616cc032008-02-18 04:19:38 +0000[diff] [blame]863 May only be applied to the first parameter.</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]864
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]865 <dt><tt>noalias</tt></dt>
Owen Andersonc4fc4cd2008-02-18 04:09:01 +0000[diff] [blame]866 <dd>This indicates that the parameter does not alias any global or any other
867 parameter. The caller is responsible for ensuring that this is the case,
868 usually by placing the value in a stack allocation.</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]869
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]870 <dt><tt>noreturn</tt></dt>
871 <dd>This function attribute indicates that the function never returns. This
872 indicates to LLVM that every call to this function should be treated as if
873 an <tt>unreachable</tt> instruction immediately followed the call.</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]874
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]875 <dt><tt>nounwind</tt></dt>
Duncan Sandsef0e9e42008-03-17 12:17:41 +0000[diff] [blame]876 <dd>This function attribute indicates that no exceptions unwind out of the
877 function. Usually this is because the function makes no use of exceptions,
878 but it may also be that the function catches any exceptions thrown when
879 executing it.</dd>
880
Duncan Sands4ee46812007-07-27 19:57:41 +0000[diff] [blame]881 <dt><tt>nest</tt></dt>
Duncan Sandsf1a7d4c2008-07-08 09:27:25 +0000[diff] [blame]882 <dd>This indicates that the pointer parameter can be excised using the
Duncan Sands4ee46812007-07-27 19:57:41 +0000[diff] [blame]883 <a href="#int_trampoline">trampoline intrinsics</a>.</dd>
Duncan Sands13e13f82007-11-22 20:23:04 +0000[diff] [blame]884 <dt><tt>readonly</tt></dt>
Duncan Sandsd69c0e62007-11-14 21:14:02 +0000[diff] [blame]885 <dd>This function attribute indicates that the function has no side-effects
Duncan Sands13e13f82007-11-22 20:23:04 +0000[diff] [blame]886 except for producing a return value or throwing an exception. The value
887 returned must only depend on the function arguments and/or global variables.
888 It may use values obtained by dereferencing pointers.</dd>
889 <dt><tt>readnone</tt></dt>
890 <dd>A <tt>readnone</tt> function has the same restrictions as a <tt>readonly</tt>
Duncan Sandsd69c0e62007-11-14 21:14:02 +0000[diff] [blame]891 function, but in addition it is not allowed to dereference any pointer arguments
892 or global variables.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]893 </dl>
894
895</div>
896
897<!-- ======================================================================= -->
898<div class="doc_subsection">
Gordon Henriksen13fe5e32007-12-10 03:18:06 +0000[diff] [blame]899 <a name="gc">Garbage Collector Names</a>
900</div>
901
902<div class="doc_text">
903<p>Each function may specify a garbage collector name, which is simply a
904string.</p>
905
906<div class="doc_code"><pre
907>define void @f() gc "name" { ...</pre></div>
908
909<p>The compiler declares the supported values of <i>name</i>. Specifying a
910collector which will cause the compiler to alter its output in order to support
911the named garbage collection algorithm.</p>
912</div>
913
914<!-- ======================================================================= -->
915<div class="doc_subsection">
Devang Pateld468f1c2008-09-04 23:05:13 +0000[diff] [blame^]916 <a name="notes">Function Notes</a>
917</div>
918
919<div class="doc_text">
920<p>Each function may specify function notes.</p>
921
922<div class="doc_code">
923<pre>define void @f() notes(inline=Always) { ... }</pre>
924<pre>define void @f() notes(inline=Always,opt-size) { ... }</pre>
925<pre>define void @f() notes(inline=Never,opt-size) { ... }</pre>
926<pre>define void @f() notes(opt-size) { ... }</pre>
927</div>
928
929<p>
930<li>inline=Always
931<p>
932This note requests inliner to inline this function irrespective of
933inlining size threshold for this function.
934</p></li>
935<li>inline=Never
936<p>
937This note requests inliner to never inline this function in any situation.
938This note may not be used together with inline=Always note.
939</p></li>
940<li>opt-size
941<p>
942This note suggests optimization passes and code generator passes to make
943choices that help reduce code size.
944</p></li>
945<p>
946The notes that are not documented here are considered invalid notes.
947</div>
948
949<!-- ======================================================================= -->
950<div class="doc_subsection">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]951 <a name="moduleasm">Module-Level Inline Assembly</a>
952</div>
953
954<div class="doc_text">
955<p>
956Modules may contain "module-level inline asm" blocks, which corresponds to the
957GCC "file scope inline asm" blocks. These blocks are internally concatenated by
958LLVM and treated as a single unit, but may be separated in the .ll file if
959desired. The syntax is very simple:
960</p>
961
962<div class="doc_code">
963<pre>
964module asm "inline asm code goes here"
965module asm "more can go here"
966</pre>
967</div>
968
969<p>The strings can contain any character by escaping non-printable characters.
970 The escape sequence used is simply "\xx" where "xx" is the two digit hex code
971 for the number.
972</p>
973
974<p>
975 The inline asm code is simply printed to the machine code .s file when
976 assembly code is generated.
977</p>
978</div>
979
980<!-- ======================================================================= -->
981<div class="doc_subsection">
982 <a name="datalayout">Data Layout</a>
983</div>
984
985<div class="doc_text">
986<p>A module may specify a target specific data layout string that specifies how
987data is to be laid out in memory. The syntax for the data layout is simply:</p>
988<pre> target datalayout = "<i>layout specification</i>"</pre>
989<p>The <i>layout specification</i> consists of a list of specifications
990separated by the minus sign character ('-'). Each specification starts with a
991letter and may include other information after the letter to define some
992aspect of the data layout. The specifications accepted are as follows: </p>
993<dl>
994 <dt><tt>E</tt></dt>
995 <dd>Specifies that the target lays out data in big-endian form. That is, the
996 bits with the most significance have the lowest address location.</dd>
997 <dt><tt>e</tt></dt>
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]998 <dd>Specifies that the target lays out data in little-endian form. That is,
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]999 the bits with the least significance have the lowest address location.</dd>
1000 <dt><tt>p:<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1001 <dd>This specifies the <i>size</i> of a pointer and its <i>abi</i> and
1002 <i>preferred</i> alignments. All sizes are in bits. Specifying the <i>pref</i>
1003 alignment is optional. If omitted, the preceding <tt>:</tt> should be omitted
1004 too.</dd>
1005 <dt><tt>i<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1006 <dd>This specifies the alignment for an integer type of a given bit
1007 <i>size</i>. The value of <i>size</i> must be in the range [1,2^23).</dd>
1008 <dt><tt>v<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1009 <dd>This specifies the alignment for a vector type of a given bit
1010 <i>size</i>.</dd>
1011 <dt><tt>f<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1012 <dd>This specifies the alignment for a floating point type of a given bit
1013 <i>size</i>. The value of <i>size</i> must be either 32 (float) or 64
1014 (double).</dd>
1015 <dt><tt>a<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1016 <dd>This specifies the alignment for an aggregate type of a given bit
1017 <i>size</i>.</dd>
1018</dl>
1019<p>When constructing the data layout for a given target, LLVM starts with a
1020default set of specifications which are then (possibly) overriden by the
1021specifications in the <tt>datalayout</tt> keyword. The default specifications
1022are given in this list:</p>
1023<ul>
1024 <li><tt>E</tt> - big endian</li>
1025 <li><tt>p:32:64:64</tt> - 32-bit pointers with 64-bit alignment</li>
1026 <li><tt>i1:8:8</tt> - i1 is 8-bit (byte) aligned</li>
1027 <li><tt>i8:8:8</tt> - i8 is 8-bit (byte) aligned</li>
1028 <li><tt>i16:16:16</tt> - i16 is 16-bit aligned</li>
1029 <li><tt>i32:32:32</tt> - i32 is 32-bit aligned</li>
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]1030 <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]1031 alignment of 64-bits</li>
1032 <li><tt>f32:32:32</tt> - float is 32-bit aligned</li>
1033 <li><tt>f64:64:64</tt> - double is 64-bit aligned</li>
1034 <li><tt>v64:64:64</tt> - 64-bit vector is 64-bit aligned</li>
1035 <li><tt>v128:128:128</tt> - 128-bit vector is 128-bit aligned</li>
1036 <li><tt>a0:0:1</tt> - aggregates are 8-bit aligned</li>
1037</ul>
Chris Lattner6a9f3c42008-08-05 18:21:08 +0000[diff] [blame]1038<p>When LLVM is determining the alignment for a given type, it uses the
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1039following rules:
1040<ol>
1041 <li>If the type sought is an exact match for one of the specifications, that
1042 specification is used.</li>
1043 <li>If no match is found, and the type sought is an integer type, then the
1044 smallest integer type that is larger than the bitwidth of the sought type is
1045 used. If none of the specifications are larger than the bitwidth then the the
1046 largest integer type is used. For example, given the default specifications
1047 above, the i7 type will use the alignment of i8 (next largest) while both
1048 i65 and i256 will use the alignment of i64 (largest specified).</li>
1049 <li>If no match is found, and the type sought is a vector type, then the
1050 largest vector type that is smaller than the sought vector type will be used
1051 as a fall back. This happens because <128 x double> can be implemented in
1052 terms of 64 <2 x double>, for example.</li>
1053</ol>
1054</div>
1055
1056<!-- *********************************************************************** -->
1057<div class="doc_section"> <a name="typesystem">Type System</a> </div>
1058<!-- *********************************************************************** -->
1059
1060<div class="doc_text">
1061
1062<p>The LLVM type system is one of the most important features of the
1063intermediate representation. Being typed enables a number of
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]1064optimizations to be performed on the intermediate representation directly,
1065without having to do
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1066extra analyses on the side before the transformation. A strong type
1067system makes it easier to read the generated code and enables novel
1068analyses and transformations that are not feasible to perform on normal
1069three address code representations.</p>
1070
1071</div>
1072
1073<!-- ======================================================================= -->
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1074<div class="doc_subsection"> <a name="t_classifications">Type
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1075Classifications</a> </div>
1076<div class="doc_text">
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1077<p>The types fall into a few useful
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1078classifications:</p>
1079
1080<table border="1" cellspacing="0" cellpadding="4">
1081 <tbody>
1082 <tr><th>Classification</th><th>Types</th></tr>
1083 <tr>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1084 <td><a href="#t_integer">integer</a></td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1085 <td><tt>i1, i2, i3, ... i8, ... i16, ... i32, ... i64, ... </tt></td>
1086 </tr>
1087 <tr>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1088 <td><a href="#t_floating">floating point</a></td>
1089 <td><tt>float, double, x86_fp80, fp128, ppc_fp128</tt></td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1090 </tr>
1091 <tr>
1092 <td><a name="t_firstclass">first class</a></td>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1093 <td><a href="#t_integer">integer</a>,
1094 <a href="#t_floating">floating point</a>,
1095 <a href="#t_pointer">pointer</a>,
Dan Gohmanf6237db2008-06-18 18:42:13 +0000[diff] [blame]1096 <a href="#t_vector">vector</a>,
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]1097 <a href="#t_struct">structure</a>,
1098 <a href="#t_array">array</a>,
Dan Gohmand13951e2008-05-23 22:50:26 +0000[diff] [blame]1099 <a href="#t_label">label</a>.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1100 </td>
1101 </tr>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1102 <tr>
1103 <td><a href="#t_primitive">primitive</a></td>
1104 <td><a href="#t_label">label</a>,
1105 <a href="#t_void">void</a>,
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1106 <a href="#t_floating">floating point</a>.</td>
1107 </tr>
1108 <tr>
1109 <td><a href="#t_derived">derived</a></td>
1110 <td><a href="#t_integer">integer</a>,
1111 <a href="#t_array">array</a>,
1112 <a href="#t_function">function</a>,
1113 <a href="#t_pointer">pointer</a>,
1114 <a href="#t_struct">structure</a>,
1115 <a href="#t_pstruct">packed structure</a>,
1116 <a href="#t_vector">vector</a>,
1117 <a href="#t_opaque">opaque</a>.
1118 </tr>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1119 </tbody>
1120</table>
1121
1122<p>The <a href="#t_firstclass">first class</a> types are perhaps the
1123most important. Values of these types are the only ones which can be
1124produced by instructions, passed as arguments, or used as operands to
Dan Gohman4f29e422008-05-23 21:53:15 +0000[diff] [blame]1125instructions.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1126</div>
1127
1128<!-- ======================================================================= -->
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1129<div class="doc_subsection"> <a name="t_primitive">Primitive Types</a> </div>
Chris Lattner86437612008-01-04 04:34:14 +0000[diff] [blame]1130
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1131<div class="doc_text">
1132<p>The primitive types are the fundamental building blocks of the LLVM
1133system.</p>
1134
Chris Lattner86437612008-01-04 04:34:14 +0000[diff] [blame]1135</div>
1136
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1137<!-- _______________________________________________________________________ -->
1138<div class="doc_subsubsection"> <a name="t_floating">Floating Point Types</a> </div>
1139
1140<div class="doc_text">
1141 <table>
1142 <tbody>
1143 <tr><th>Type</th><th>Description</th></tr>
1144 <tr><td><tt>float</tt></td><td>32-bit floating point value</td></tr>
1145 <tr><td><tt>double</tt></td><td>64-bit floating point value</td></tr>
1146 <tr><td><tt>fp128</tt></td><td>128-bit floating point value (112-bit mantissa)</td></tr>
1147 <tr><td><tt>x86_fp80</tt></td><td>80-bit floating point value (X87)</td></tr>
1148 <tr><td><tt>ppc_fp128</tt></td><td>128-bit floating point value (two 64-bits)</td></tr>
1149 </tbody>
1150 </table>
1151</div>
1152
1153<!-- _______________________________________________________________________ -->
1154<div class="doc_subsubsection"> <a name="t_void">Void Type</a> </div>
1155
1156<div class="doc_text">
1157<h5>Overview:</h5>
1158<p>The void type does not represent any value and has no size.</p>
1159
1160<h5>Syntax:</h5>
1161
1162<pre>
1163 void
1164</pre>
1165</div>
1166
1167<!-- _______________________________________________________________________ -->
1168<div class="doc_subsubsection"> <a name="t_label">Label Type</a> </div>
1169
1170<div class="doc_text">
1171<h5>Overview:</h5>
1172<p>The label type represents code labels.</p>
1173
1174<h5>Syntax:</h5>
1175
1176<pre>
1177 label
1178</pre>
1179</div>
1180
1181
1182<!-- ======================================================================= -->
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1183<div class="doc_subsection"> <a name="t_derived">Derived Types</a> </div>
1184
1185<div class="doc_text">
1186
1187<p>The real power in LLVM comes from the derived types in the system.
1188This is what allows a programmer to represent arrays, functions,
1189pointers, and other useful types. Note that these derived types may be
1190recursive: For example, it is possible to have a two dimensional array.</p>
1191
1192</div>
1193
1194<!-- _______________________________________________________________________ -->
1195<div class="doc_subsubsection"> <a name="t_integer">Integer Type</a> </div>
1196
1197<div class="doc_text">
1198
1199<h5>Overview:</h5>
1200<p>The integer type is a very simple derived type that simply specifies an
1201arbitrary bit width for the integer type desired. Any bit width from 1 bit to
12022^23-1 (about 8 million) can be specified.</p>
1203
1204<h5>Syntax:</h5>
1205
1206<pre>
1207 iN
1208</pre>
1209
1210<p>The number of bits the integer will occupy is specified by the <tt>N</tt>
1211value.</p>
1212
1213<h5>Examples:</h5>
1214<table class="layout">
Chris Lattner251ab812007-12-18 06:18:21 +0000[diff] [blame]1215 <tbody>
1216 <tr>
1217 <td><tt>i1</tt></td>
1218 <td>a single-bit integer.</td>
1219 </tr><tr>
1220 <td><tt>i32</tt></td>
1221 <td>a 32-bit integer.</td>
1222 </tr><tr>
1223 <td><tt>i1942652</tt></td>
1224 <td>a really big integer of over 1 million bits.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1225 </tr>
Chris Lattner251ab812007-12-18 06:18:21 +0000[diff] [blame]1226 </tbody>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1227</table>
1228</div>
1229
1230<!-- _______________________________________________________________________ -->
1231<div class="doc_subsubsection"> <a name="t_array">Array Type</a> </div>
1232
1233<div class="doc_text">
1234
1235<h5>Overview:</h5>
1236
1237<p>The array type is a very simple derived type that arranges elements
1238sequentially in memory. The array type requires a size (number of
1239elements) and an underlying data type.</p>
1240
1241<h5>Syntax:</h5>
1242
1243<pre>
1244 [&lt;# elements&gt; x &lt;elementtype&gt;]
1245</pre>
1246
1247<p>The number of elements is a constant integer value; elementtype may
1248be any type with a size.</p>
1249
1250<h5>Examples:</h5>
1251<table class="layout">
1252 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1253 <td class="left"><tt>[40 x i32]</tt></td>
1254 <td class="left">Array of 40 32-bit integer values.</td>
1255 </tr>
1256 <tr class="layout">
1257 <td class="left"><tt>[41 x i32]</tt></td>
1258 <td class="left">Array of 41 32-bit integer values.</td>
1259 </tr>
1260 <tr class="layout">
1261 <td class="left"><tt>[4 x i8]</tt></td>
1262 <td class="left">Array of 4 8-bit integer values.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1263 </tr>
1264</table>
1265<p>Here are some examples of multidimensional arrays:</p>
1266<table class="layout">
1267 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1268 <td class="left"><tt>[3 x [4 x i32]]</tt></td>
1269 <td class="left">3x4 array of 32-bit integer values.</td>
1270 </tr>
1271 <tr class="layout">
1272 <td class="left"><tt>[12 x [10 x float]]</tt></td>
1273 <td class="left">12x10 array of single precision floating point values.</td>
1274 </tr>
1275 <tr class="layout">
1276 <td class="left"><tt>[2 x [3 x [4 x i16]]]</tt></td>
1277 <td class="left">2x3x4 array of 16-bit integer values.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1278 </tr>
1279</table>
1280
1281<p>Note that 'variable sized arrays' can be implemented in LLVM with a zero
1282length array. Normally, accesses past the end of an array are undefined in
1283LLVM (e.g. it is illegal to access the 5th element of a 3 element array).
1284As a special case, however, zero length arrays are recognized to be variable
1285length. This allows implementation of 'pascal style arrays' with the LLVM
1286type "{ i32, [0 x float]}", for example.</p>
1287
1288</div>
1289
1290<!-- _______________________________________________________________________ -->
1291<div class="doc_subsubsection"> <a name="t_function">Function Type</a> </div>
1292<div class="doc_text">
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1293
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1294<h5>Overview:</h5>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1295
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1296<p>The function type can be thought of as a function signature. It
Devang Pateld4ba41d2008-03-24 05:35:41 +0000[diff] [blame]1297consists of a return type and a list of formal parameter types. The
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1298return 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]1299If 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]1300class types, and the struct must have at least one element.</p>
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]1301
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1302<h5>Syntax:</h5>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1303
1304<pre>
1305 &lt;returntype list&gt; (&lt;parameter list&gt;)
1306</pre>
1307
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1308<p>...where '<tt>&lt;parameter list&gt;</tt>' is a comma-separated list of type
1309specifiers. Optionally, the parameter list may include a type <tt>...</tt>,
1310which indicates that the function takes a variable number of arguments.
1311Variable argument functions can access their arguments with the <a
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]1312 href="#int_varargs">variable argument handling intrinsic</a> functions.
1313'<tt>&lt;returntype list&gt;</tt>' is a comma-separated list of
1314<a href="#t_firstclass">first class</a> type specifiers.</p>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1315
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1316<h5>Examples:</h5>
1317<table class="layout">
1318 <tr class="layout">
1319 <td class="left"><tt>i32 (i32)</tt></td>
1320 <td class="left">function taking an <tt>i32</tt>, returning an <tt>i32</tt>
1321 </td>
1322 </tr><tr class="layout">
Reid Spencerf234bed2007-07-19 23:13:04 +0000[diff] [blame]1323 <td class="left"><tt>float&nbsp;(i16&nbsp;signext,&nbsp;i32&nbsp;*)&nbsp;*
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1324 </tt></td>
1325 <td class="left"><a href="#t_pointer">Pointer</a> to a function that takes
1326 an <tt>i16</tt> that should be sign extended and a
1327 <a href="#t_pointer">pointer</a> to <tt>i32</tt>, returning
1328 <tt>float</tt>.
1329 </td>
1330 </tr><tr class="layout">
1331 <td class="left"><tt>i32 (i8*, ...)</tt></td>
1332 <td class="left">A vararg function that takes at least one
1333 <a href="#t_pointer">pointer</a> to <tt>i8 </tt> (char in C),
1334 which returns an integer. This is the signature for <tt>printf</tt> in
1335 LLVM.
1336 </td>
Devang Pateld4ba41d2008-03-24 05:35:41 +0000[diff] [blame]1337 </tr><tr class="layout">
1338 <td class="left"><tt>{i32, i32} (i32)</tt></td>
Devang Patel6dddba22008-03-24 18:10:52 +0000[diff] [blame]1339 <td class="left">A function taking an <tt>i32></tt>, returning two
1340 <tt> i32 </tt> values as an aggregate of type <tt>{ i32, i32 }</tt>
Devang Pateld4ba41d2008-03-24 05:35:41 +0000[diff] [blame]1341 </td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1342 </tr>
1343</table>
1344
1345</div>
1346<!-- _______________________________________________________________________ -->
1347<div class="doc_subsubsection"> <a name="t_struct">Structure Type</a> </div>
1348<div class="doc_text">
1349<h5>Overview:</h5>
1350<p>The structure type is used to represent a collection of data members
1351together in memory. The packing of the field types is defined to match
1352the ABI of the underlying processor. The elements of a structure may
1353be any type that has a size.</p>
1354<p>Structures are accessed using '<tt><a href="#i_load">load</a></tt>
1355and '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a
1356field with the '<tt><a href="#i_getelementptr">getelementptr</a></tt>'
1357instruction.</p>
1358<h5>Syntax:</h5>
1359<pre> { &lt;type list&gt; }<br></pre>
1360<h5>Examples:</h5>
1361<table class="layout">
1362 <tr class="layout">
1363 <td class="left"><tt>{ i32, i32, i32 }</tt></td>
1364 <td class="left">A triple of three <tt>i32</tt> values</td>
1365 </tr><tr class="layout">
1366 <td class="left"><tt>{&nbsp;float,&nbsp;i32&nbsp;(i32)&nbsp;*&nbsp;}</tt></td>
1367 <td class="left">A pair, where the first element is a <tt>float</tt> and the
1368 second element is a <a href="#t_pointer">pointer</a> to a
1369 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
1370 an <tt>i32</tt>.</td>
1371 </tr>
1372</table>
1373</div>
1374
1375<!-- _______________________________________________________________________ -->
1376<div class="doc_subsubsection"> <a name="t_pstruct">Packed Structure Type</a>
1377</div>
1378<div class="doc_text">
1379<h5>Overview:</h5>
1380<p>The packed structure type is used to represent a collection of data members
1381together in memory. There is no padding between fields. Further, the alignment
1382of a packed structure is 1 byte. The elements of a packed structure may
1383be any type that has a size.</p>
1384<p>Structures are accessed using '<tt><a href="#i_load">load</a></tt>
1385and '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a
1386field with the '<tt><a href="#i_getelementptr">getelementptr</a></tt>'
1387instruction.</p>
1388<h5>Syntax:</h5>
1389<pre> &lt; { &lt;type list&gt; } &gt; <br></pre>
1390<h5>Examples:</h5>
1391<table class="layout">
1392 <tr class="layout">
1393 <td class="left"><tt>&lt; { i32, i32, i32 } &gt;</tt></td>
1394 <td class="left">A triple of three <tt>i32</tt> values</td>
1395 </tr><tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1396 <td class="left"><tt>&lt; { float, i32 (i32)* } &gt;</tt></td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1397 <td class="left">A pair, where the first element is a <tt>float</tt> and the
1398 second element is a <a href="#t_pointer">pointer</a> to a
1399 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
1400 an <tt>i32</tt>.</td>
1401 </tr>
1402</table>
1403</div>
1404
1405<!-- _______________________________________________________________________ -->
1406<div class="doc_subsubsection"> <a name="t_pointer">Pointer Type</a> </div>
1407<div class="doc_text">
1408<h5>Overview:</h5>
1409<p>As in many languages, the pointer type represents a pointer or
Christopher Lambdd0049d2007-12-11 09:31:00 +0000[diff] [blame]1410reference to another object, which must live in memory. Pointer types may have
1411an optional address space attribute defining the target-specific numbered
1412address space where the pointed-to object resides. The default address space is
1413zero.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1414<h5>Syntax:</h5>
1415<pre> &lt;type&gt; *<br></pre>
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>[4x i32]*</tt></td>
1420 <td class="left">A <a href="#t_pointer">pointer</a> to <a
1421 href="#t_array">array</a> of four <tt>i32</tt> values.</td>
1422 </tr>
1423 <tr class="layout">
1424 <td class="left"><tt>i32 (i32 *) *</tt></td>
1425 <td class="left"> A <a href="#t_pointer">pointer</a> to a <a
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1426 href="#t_function">function</a> that takes an <tt>i32*</tt>, returning an
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1427 <tt>i32</tt>.</td>
1428 </tr>
1429 <tr class="layout">
1430 <td class="left"><tt>i32 addrspace(5)*</tt></td>
1431 <td class="left">A <a href="#t_pointer">pointer</a> to an <tt>i32</tt> value
1432 that resides in address space #5.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1433 </tr>
1434</table>
1435</div>
1436
1437<!-- _______________________________________________________________________ -->
1438<div class="doc_subsubsection"> <a name="t_vector">Vector Type</a> </div>
1439<div class="doc_text">
1440
1441<h5>Overview:</h5>
1442
1443<p>A vector type is a simple derived type that represents a vector
1444of elements. Vector types are used when multiple primitive data
1445are operated in parallel using a single instruction (SIMD).
1446A vector type requires a size (number of
1447elements) and an underlying primitive data type. Vectors must have a power
1448of two length (1, 2, 4, 8, 16 ...). Vector types are
1449considered <a href="#t_firstclass">first class</a>.</p>
1450
1451<h5>Syntax:</h5>
1452
1453<pre>
1454 &lt; &lt;# elements&gt; x &lt;elementtype&gt; &gt;
1455</pre>
1456
1457<p>The number of elements is a constant integer value; elementtype may
1458be any integer or floating point type.</p>
1459
1460<h5>Examples:</h5>
1461
1462<table class="layout">
1463 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1464 <td class="left"><tt>&lt;4 x i32&gt;</tt></td>
1465 <td class="left">Vector of 4 32-bit integer values.</td>
1466 </tr>
1467 <tr class="layout">
1468 <td class="left"><tt>&lt;8 x float&gt;</tt></td>
1469 <td class="left">Vector of 8 32-bit floating-point values.</td>
1470 </tr>
1471 <tr class="layout">
1472 <td class="left"><tt>&lt;2 x i64&gt;</tt></td>
1473 <td class="left">Vector of 2 64-bit integer values.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1474 </tr>
1475</table>
1476</div>
1477
1478<!-- _______________________________________________________________________ -->
1479<div class="doc_subsubsection"> <a name="t_opaque">Opaque Type</a> </div>
1480<div class="doc_text">
1481
1482<h5>Overview:</h5>
1483
1484<p>Opaque types are used to represent unknown types in the system. This
Gordon Henriksenda0706e2007-10-14 00:34:53 +0000[diff] [blame]1485corresponds (for example) to the C notion of a forward declared structure type.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1486In LLVM, opaque types can eventually be resolved to any type (not just a
1487structure type).</p>
1488
1489<h5>Syntax:</h5>
1490
1491<pre>
1492 opaque
1493</pre>
1494
1495<h5>Examples:</h5>
1496
1497<table class="layout">
1498 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1499 <td class="left"><tt>opaque</tt></td>
1500 <td class="left">An opaque type.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1501 </tr>
1502</table>
1503</div>
1504
1505
1506<!-- *********************************************************************** -->
1507<div class="doc_section"> <a name="constants">Constants</a> </div>
1508<!-- *********************************************************************** -->
1509
1510<div class="doc_text">
1511
1512<p>LLVM has several different basic types of constants. This section describes
1513them all and their syntax.</p>
1514
1515</div>
1516
1517<!-- ======================================================================= -->
1518<div class="doc_subsection"><a name="simpleconstants">Simple Constants</a></div>
1519
1520<div class="doc_text">
1521
1522<dl>
1523 <dt><b>Boolean constants</b></dt>
1524
1525 <dd>The two strings '<tt>true</tt>' and '<tt>false</tt>' are both valid
1526 constants of the <tt><a href="#t_primitive">i1</a></tt> type.
1527 </dd>
1528
1529 <dt><b>Integer constants</b></dt>
1530
1531 <dd>Standard integers (such as '4') are constants of the <a
1532 href="#t_integer">integer</a> type. Negative numbers may be used with
1533 integer types.
1534 </dd>
1535
1536 <dt><b>Floating point constants</b></dt>
1537
1538 <dd>Floating point constants use standard decimal notation (e.g. 123.421),
1539 exponential notation (e.g. 1.23421e+2), or a more precise hexadecimal
Chris Lattnerdc15b1d2008-04-01 18:45:27 +0000[diff] [blame]1540 notation (see below). The assembler requires the exact decimal value of
1541 a floating-point constant. For example, the assembler accepts 1.25 but
1542 rejects 1.3 because 1.3 is a repeating decimal in binary. Floating point
1543 constants must have a <a href="#t_floating">floating point</a> type. </dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1544
1545 <dt><b>Null pointer constants</b></dt>
1546
1547 <dd>The identifier '<tt>null</tt>' is recognized as a null pointer constant
1548 and must be of <a href="#t_pointer">pointer type</a>.</dd>
1549
1550</dl>
1551
1552<p>The one non-intuitive notation for constants is the optional hexadecimal form
1553of floating point constants. For example, the form '<tt>double
15540x432ff973cafa8000</tt>' is equivalent to (but harder to read than) '<tt>double
15554.5e+15</tt>'. The only time hexadecimal floating point constants are required
1556(and the only time that they are generated by the disassembler) is when a
1557floating point constant must be emitted but it cannot be represented as a
1558decimal floating point number. For example, NaN's, infinities, and other
1559special values are represented in their IEEE hexadecimal format so that
1560assembly and disassembly do not cause any bits to change in the constants.</p>
1561
1562</div>
1563
1564<!-- ======================================================================= -->
1565<div class="doc_subsection"><a name="aggregateconstants">Aggregate Constants</a>
1566</div>
1567
1568<div class="doc_text">
1569<p>Aggregate constants arise from aggregation of simple constants
1570and smaller aggregate constants.</p>
1571
1572<dl>
1573 <dt><b>Structure constants</b></dt>
1574
1575 <dd>Structure constants are represented with notation similar to structure
1576 type definitions (a comma separated list of elements, surrounded by braces
Chris Lattner6c8de962007-12-25 20:34:52 +0000[diff] [blame]1577 (<tt>{}</tt>)). For example: "<tt>{ i32 4, float 17.0, i32* @G }</tt>",
1578 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]1579 must have <a href="#t_struct">structure type</a>, and the number and
1580 types of elements must match those specified by the type.
1581 </dd>
1582
1583 <dt><b>Array constants</b></dt>
1584
1585 <dd>Array constants are represented with notation similar to array type
1586 definitions (a comma separated list of elements, surrounded by square brackets
1587 (<tt>[]</tt>)). For example: "<tt>[ i32 42, i32 11, i32 74 ]</tt>". Array
1588 constants must have <a href="#t_array">array type</a>, and the number and
1589 types of elements must match those specified by the type.
1590 </dd>
1591
1592 <dt><b>Vector constants</b></dt>
1593
1594 <dd>Vector constants are represented with notation similar to vector type
1595 definitions (a comma separated list of elements, surrounded by
1596 less-than/greater-than's (<tt>&lt;&gt;</tt>)). For example: "<tt>&lt; i32 42,
1597 i32 11, i32 74, i32 100 &gt;</tt>". Vector constants must have <a
1598 href="#t_vector">vector type</a>, and the number and types of elements must
1599 match those specified by the type.
1600 </dd>
1601
1602 <dt><b>Zero initialization</b></dt>
1603
1604 <dd>The string '<tt>zeroinitializer</tt>' can be used to zero initialize a
1605 value to zero of <em>any</em> type, including scalar and aggregate types.
1606 This is often used to avoid having to print large zero initializers (e.g. for
1607 large arrays) and is always exactly equivalent to using explicit zero
1608 initializers.
1609 </dd>
1610</dl>
1611
1612</div>
1613
1614<!-- ======================================================================= -->
1615<div class="doc_subsection">
1616 <a name="globalconstants">Global Variable and Function Addresses</a>
1617</div>
1618
1619<div class="doc_text">
1620
1621<p>The addresses of <a href="#globalvars">global variables</a> and <a
1622href="#functionstructure">functions</a> are always implicitly valid (link-time)
1623constants. These constants are explicitly referenced when the <a
1624href="#identifiers">identifier for the global</a> is used and always have <a
1625href="#t_pointer">pointer</a> type. For example, the following is a legal LLVM
1626file:</p>
1627
1628<div class="doc_code">
1629<pre>
1630@X = global i32 17
1631@Y = global i32 42
1632@Z = global [2 x i32*] [ i32* @X, i32* @Y ]
1633</pre>
1634</div>
1635
1636</div>
1637
1638<!-- ======================================================================= -->
1639<div class="doc_subsection"><a name="undefvalues">Undefined Values</a></div>
1640<div class="doc_text">
1641 <p>The string '<tt>undef</tt>' is recognized as a type-less constant that has
1642 no specific value. Undefined values may be of any type and be used anywhere
1643 a constant is permitted.</p>
1644
1645 <p>Undefined values indicate to the compiler that the program is well defined
1646 no matter what value is used, giving the compiler more freedom to optimize.
1647 </p>
1648</div>
1649
1650<!-- ======================================================================= -->
1651<div class="doc_subsection"><a name="constantexprs">Constant Expressions</a>
1652</div>
1653
1654<div class="doc_text">
1655
1656<p>Constant expressions are used to allow expressions involving other constants
1657to be used as constants. Constant expressions may be of any <a
1658href="#t_firstclass">first class</a> type and may involve any LLVM operation
1659that does not have side effects (e.g. load and call are not supported). The
1660following is the syntax for constant expressions:</p>
1661
1662<dl>
1663 <dt><b><tt>trunc ( CST to TYPE )</tt></b></dt>
1664 <dd>Truncate a constant to another type. The bit size of CST must be larger
1665 than the bit size of TYPE. Both types must be integers.</dd>
1666
1667 <dt><b><tt>zext ( CST to TYPE )</tt></b></dt>
1668 <dd>Zero extend a constant to another type. The bit size of CST must be
1669 smaller or equal to the bit size of TYPE. Both types must be integers.</dd>
1670
1671 <dt><b><tt>sext ( CST to TYPE )</tt></b></dt>
1672 <dd>Sign extend a constant to another type. The bit size of CST must be
1673 smaller or equal to the bit size of TYPE. Both types must be integers.</dd>
1674
1675 <dt><b><tt>fptrunc ( CST to TYPE )</tt></b></dt>
1676 <dd>Truncate a floating point constant to another floating point type. The
1677 size of CST must be larger than the size of TYPE. Both types must be
1678 floating point.</dd>
1679
1680 <dt><b><tt>fpext ( CST to TYPE )</tt></b></dt>
1681 <dd>Floating point extend a constant to another type. The size of CST must be
1682 smaller or equal to the size of TYPE. Both types must be floating point.</dd>
1683
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]1684 <dt><b><tt>fptoui ( CST to TYPE )</tt></b></dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1685 <dd>Convert a floating point constant to the corresponding unsigned integer
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]1686 constant. TYPE must be a scalar or vector integer type. CST must be of scalar
1687 or vector floating point type. Both CST and TYPE must be scalars, or vectors
1688 of the same number of elements. If the value won't fit in the integer type,
1689 the results are undefined.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1690
1691 <dt><b><tt>fptosi ( CST to TYPE )</tt></b></dt>
1692 <dd>Convert a floating point constant to the corresponding signed integer
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]1693 constant. TYPE must be a scalar or vector integer type. CST must be of scalar
1694 or vector floating point type. Both CST and TYPE must be scalars, or vectors
1695 of the same number of elements. If the value won't fit in the integer type,
1696 the results are undefined.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1697
1698 <dt><b><tt>uitofp ( CST to TYPE )</tt></b></dt>
1699 <dd>Convert an unsigned integer constant to the corresponding floating point
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]1700 constant. TYPE must be a scalar or vector floating point type. CST must be of
1701 scalar or vector integer type. Both CST and TYPE must be scalars, or vectors
1702 of the same number of elements. If the value won't fit in the floating point
1703 type, the results are undefined.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1704
1705 <dt><b><tt>sitofp ( CST to TYPE )</tt></b></dt>
1706 <dd>Convert a signed integer constant to the corresponding floating point
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]1707 constant. TYPE must be a scalar or vector floating point type. CST must be of
1708 scalar or vector integer type. Both CST and TYPE must be scalars, or vectors
1709 of the same number of elements. If the value won't fit in the floating point
1710 type, the results are undefined.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1711
1712 <dt><b><tt>ptrtoint ( CST to TYPE )</tt></b></dt>
1713 <dd>Convert a pointer typed constant to the corresponding integer constant
1714 TYPE must be an integer type. CST must be of pointer type. The CST value is
1715 zero extended, truncated, or unchanged to make it fit in TYPE.</dd>
1716
1717 <dt><b><tt>inttoptr ( CST to TYPE )</tt></b></dt>
1718 <dd>Convert a integer constant to a pointer constant. TYPE must be a
1719 pointer type. CST must be of integer type. The CST value is zero extended,
1720 truncated, or unchanged to make it fit in a pointer size. This one is
1721 <i>really</i> dangerous!</dd>
1722
1723 <dt><b><tt>bitcast ( CST to TYPE )</tt></b></dt>
1724 <dd>Convert a constant, CST, to another TYPE. The size of CST and TYPE must be
1725 identical (same number of bits). The conversion is done as if the CST value
1726 was stored to memory and read back as TYPE. In other words, no bits change
1727 with this operator, just the type. This can be used for conversion of
1728 vector types to any other type, as long as they have the same bit width. For
1729 pointers it is only valid to cast to another pointer type.
1730 </dd>
1731
1732 <dt><b><tt>getelementptr ( CSTPTR, IDX0, IDX1, ... )</tt></b></dt>
1733
1734 <dd>Perform the <a href="#i_getelementptr">getelementptr operation</a> on
1735 constants. As with the <a href="#i_getelementptr">getelementptr</a>
1736 instruction, the index list may have zero or more indexes, which are required
1737 to make sense for the type of "CSTPTR".</dd>
1738
1739 <dt><b><tt>select ( COND, VAL1, VAL2 )</tt></b></dt>
1740
1741 <dd>Perform the <a href="#i_select">select operation</a> on
1742 constants.</dd>
1743
1744 <dt><b><tt>icmp COND ( VAL1, VAL2 )</tt></b></dt>
1745 <dd>Performs the <a href="#i_icmp">icmp operation</a> on constants.</dd>
1746
1747 <dt><b><tt>fcmp COND ( VAL1, VAL2 )</tt></b></dt>
1748 <dd>Performs the <a href="#i_fcmp">fcmp operation</a> on constants.</dd>
1749
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]1750 <dt><b><tt>vicmp COND ( VAL1, VAL2 )</tt></b></dt>
1751 <dd>Performs the <a href="#i_vicmp">vicmp operation</a> on constants.</dd>
1752
1753 <dt><b><tt>vfcmp COND ( VAL1, VAL2 )</tt></b></dt>
1754 <dd>Performs the <a href="#i_vfcmp">vfcmp operation</a> on constants.</dd>
1755
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1756 <dt><b><tt>extractelement ( VAL, IDX )</tt></b></dt>
1757
1758 <dd>Perform the <a href="#i_extractelement">extractelement
1759 operation</a> on constants.
1760
1761 <dt><b><tt>insertelement ( VAL, ELT, IDX )</tt></b></dt>
1762
1763 <dd>Perform the <a href="#i_insertelement">insertelement
1764 operation</a> on constants.</dd>
1765
1766
1767 <dt><b><tt>shufflevector ( VEC1, VEC2, IDXMASK )</tt></b></dt>
1768
1769 <dd>Perform the <a href="#i_shufflevector">shufflevector
1770 operation</a> on constants.</dd>
1771
1772 <dt><b><tt>OPCODE ( LHS, RHS )</tt></b></dt>
1773
1774 <dd>Perform the specified operation of the LHS and RHS constants. OPCODE may
1775 be any of the <a href="#binaryops">binary</a> or <a href="#bitwiseops">bitwise
1776 binary</a> operations. The constraints on operands are the same as those for
1777 the corresponding instruction (e.g. no bitwise operations on floating point
1778 values are allowed).</dd>
1779</dl>
1780</div>
1781
1782<!-- *********************************************************************** -->
1783<div class="doc_section"> <a name="othervalues">Other Values</a> </div>
1784<!-- *********************************************************************** -->
1785
1786<!-- ======================================================================= -->
1787<div class="doc_subsection">
1788<a name="inlineasm">Inline Assembler Expressions</a>
1789</div>
1790
1791<div class="doc_text">
1792
1793<p>
1794LLVM supports inline assembler expressions (as opposed to <a href="#moduleasm">
1795Module-Level Inline Assembly</a>) through the use of a special value. This
1796value represents the inline assembler as a string (containing the instructions
1797to emit), a list of operand constraints (stored as a string), and a flag that
1798indicates whether or not the inline asm expression has side effects. An example
1799inline assembler expression is:
1800</p>
1801
1802<div class="doc_code">
1803<pre>
1804i32 (i32) asm "bswap $0", "=r,r"
1805</pre>
1806</div>
1807
1808<p>
1809Inline assembler expressions may <b>only</b> be used as the callee operand of
1810a <a href="#i_call"><tt>call</tt> instruction</a>. Thus, typically we have:
1811</p>
1812
1813<div class="doc_code">
1814<pre>
1815%X = call i32 asm "<a href="#int_bswap">bswap</a> $0", "=r,r"(i32 %Y)
1816</pre>
1817</div>
1818
1819<p>
1820Inline asms with side effects not visible in the constraint list must be marked
1821as having side effects. This is done through the use of the
1822'<tt>sideeffect</tt>' keyword, like so:
1823</p>
1824
1825<div class="doc_code">
1826<pre>
1827call void asm sideeffect "eieio", ""()
1828</pre>
1829</div>
1830
1831<p>TODO: The format of the asm and constraints string still need to be
1832documented here. Constraints on what can be done (e.g. duplication, moving, etc
1833need to be documented).
1834</p>
1835
1836</div>
1837
1838<!-- *********************************************************************** -->
1839<div class="doc_section"> <a name="instref">Instruction Reference</a> </div>
1840<!-- *********************************************************************** -->
1841
1842<div class="doc_text">
1843
1844<p>The LLVM instruction set consists of several different
1845classifications of instructions: <a href="#terminators">terminator
1846instructions</a>, <a href="#binaryops">binary instructions</a>,
1847<a href="#bitwiseops">bitwise binary instructions</a>, <a
1848 href="#memoryops">memory instructions</a>, and <a href="#otherops">other
1849instructions</a>.</p>
1850
1851</div>
1852
1853<!-- ======================================================================= -->
1854<div class="doc_subsection"> <a name="terminators">Terminator
1855Instructions</a> </div>
1856
1857<div class="doc_text">
1858
1859<p>As mentioned <a href="#functionstructure">previously</a>, every
1860basic block in a program ends with a "Terminator" instruction, which
1861indicates which block should be executed after the current block is
1862finished. These terminator instructions typically yield a '<tt>void</tt>'
1863value: they produce control flow, not values (the one exception being
1864the '<a href="#i_invoke"><tt>invoke</tt></a>' instruction).</p>
1865<p>There are six different terminator instructions: the '<a
1866 href="#i_ret"><tt>ret</tt></a>' instruction, the '<a href="#i_br"><tt>br</tt></a>'
1867instruction, the '<a href="#i_switch"><tt>switch</tt></a>' instruction,
1868the '<a href="#i_invoke"><tt>invoke</tt></a>' instruction, the '<a
1869 href="#i_unwind"><tt>unwind</tt></a>' instruction, and the '<a
1870 href="#i_unreachable"><tt>unreachable</tt></a>' instruction.</p>
1871
1872</div>
1873
1874<!-- _______________________________________________________________________ -->
1875<div class="doc_subsubsection"> <a name="i_ret">'<tt>ret</tt>'
1876Instruction</a> </div>
1877<div class="doc_text">
1878<h5>Syntax:</h5>
1879<pre> ret &lt;type&gt; &lt;value&gt; <i>; Return a value from a non-void function</i>
1880 ret void <i>; Return from void function</i>
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]1881 ret &lt;type&gt; &lt;value&gt;, &lt;type&gt; &lt;value&gt; <i>; Return two values from a non-void function </i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1882</pre>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1883
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1884<h5>Overview:</h5>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1885
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1886<p>The '<tt>ret</tt>' instruction is used to return control flow (and a
1887value) from a function back to the caller.</p>
1888<p>There are two forms of the '<tt>ret</tt>' instruction: one that
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1889returns value(s) and then causes control flow, and one that just causes
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1890control flow to occur.</p>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1891
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1892<h5>Arguments:</h5>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1893
1894<p>The '<tt>ret</tt>' instruction may return zero, one or multiple values.
1895The type of each return value must be a '<a href="#t_firstclass">first
1896class</a>' type. Note that a function is not <a href="#wellformed">well
1897formed</a> if there exists a '<tt>ret</tt>' instruction inside of the
1898function that returns values that do not match the return type of the
1899function.</p>
1900
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1901<h5>Semantics:</h5>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1902
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1903<p>When the '<tt>ret</tt>' instruction is executed, control flow
1904returns back to the calling function's context. If the caller is a "<a
1905 href="#i_call"><tt>call</tt></a>" instruction, execution continues at
1906the instruction after the call. If the caller was an "<a
1907 href="#i_invoke"><tt>invoke</tt></a>" instruction, execution continues
1908at the beginning of the "normal" destination block. If the instruction
1909returns a value, that value shall set the call or invoke instruction's
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]1910return value. If the instruction returns multiple values then these
Devang Patelec8a5b02008-03-11 05:51:59 +0000[diff] [blame]1911values can only be accessed through a '<a href="#i_getresult"><tt>getresult</tt>
1912</a>' instruction.</p>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1913
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1914<h5>Example:</h5>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1915
1916<pre>
1917 ret i32 5 <i>; Return an integer value of 5</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1918 ret void <i>; Return from a void function</i>
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]1919 ret i32 4, i8 2 <i>; Return two values 4 and 2 </i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1920</pre>
1921</div>
1922<!-- _______________________________________________________________________ -->
1923<div class="doc_subsubsection"> <a name="i_br">'<tt>br</tt>' Instruction</a> </div>
1924<div class="doc_text">
1925<h5>Syntax:</h5>
1926<pre> br i1 &lt;cond&gt;, label &lt;iftrue&gt;, label &lt;iffalse&gt;<br> br label &lt;dest&gt; <i>; Unconditional branch</i>
1927</pre>
1928<h5>Overview:</h5>
1929<p>The '<tt>br</tt>' instruction is used to cause control flow to
1930transfer to a different basic block in the current function. There are
1931two forms of this instruction, corresponding to a conditional branch
1932and an unconditional branch.</p>
1933<h5>Arguments:</h5>
1934<p>The conditional branch form of the '<tt>br</tt>' instruction takes a
1935single '<tt>i1</tt>' value and two '<tt>label</tt>' values. The
1936unconditional form of the '<tt>br</tt>' instruction takes a single
1937'<tt>label</tt>' value as a target.</p>
1938<h5>Semantics:</h5>
1939<p>Upon execution of a conditional '<tt>br</tt>' instruction, the '<tt>i1</tt>'
1940argument is evaluated. If the value is <tt>true</tt>, control flows
1941to the '<tt>iftrue</tt>' <tt>label</tt> argument. If "cond" is <tt>false</tt>,
1942control flows to the '<tt>iffalse</tt>' <tt>label</tt> argument.</p>
1943<h5>Example:</h5>
1944<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
1945 href="#i_ret">ret</a> i32 1<br>IfUnequal:<br> <a href="#i_ret">ret</a> i32 0<br></pre>
1946</div>
1947<!-- _______________________________________________________________________ -->
1948<div class="doc_subsubsection">
1949 <a name="i_switch">'<tt>switch</tt>' Instruction</a>
1950</div>
1951
1952<div class="doc_text">
1953<h5>Syntax:</h5>
1954
1955<pre>
1956 switch &lt;intty&gt; &lt;value&gt;, label &lt;defaultdest&gt; [ &lt;intty&gt; &lt;val&gt;, label &lt;dest&gt; ... ]
1957</pre>
1958
1959<h5>Overview:</h5>
1960
1961<p>The '<tt>switch</tt>' instruction is used to transfer control flow to one of
1962several different places. It is a generalization of the '<tt>br</tt>'
1963instruction, allowing a branch to occur to one of many possible
1964destinations.</p>
1965
1966
1967<h5>Arguments:</h5>
1968
1969<p>The '<tt>switch</tt>' instruction uses three parameters: an integer
1970comparison value '<tt>value</tt>', a default '<tt>label</tt>' destination, and
1971an array of pairs of comparison value constants and '<tt>label</tt>'s. The
1972table is not allowed to contain duplicate constant entries.</p>
1973
1974<h5>Semantics:</h5>
1975
1976<p>The <tt>switch</tt> instruction specifies a table of values and
1977destinations. When the '<tt>switch</tt>' instruction is executed, this
1978table is searched for the given value. If the value is found, control flow is
1979transfered to the corresponding destination; otherwise, control flow is
1980transfered to the default destination.</p>
1981
1982<h5>Implementation:</h5>
1983
1984<p>Depending on properties of the target machine and the particular
1985<tt>switch</tt> instruction, this instruction may be code generated in different
1986ways. For example, it could be generated as a series of chained conditional
1987branches or with a lookup table.</p>
1988
1989<h5>Example:</h5>
1990
1991<pre>
1992 <i>; Emulate a conditional br instruction</i>
1993 %Val = <a href="#i_zext">zext</a> i1 %value to i32
1994 switch i32 %Val, label %truedest [i32 0, label %falsedest ]
1995
1996 <i>; Emulate an unconditional br instruction</i>
1997 switch i32 0, label %dest [ ]
1998
1999 <i>; Implement a jump table:</i>
2000 switch i32 %val, label %otherwise [ i32 0, label %onzero
2001 i32 1, label %onone
2002 i32 2, label %ontwo ]
2003</pre>
2004</div>
2005
2006<!-- _______________________________________________________________________ -->
2007<div class="doc_subsubsection">
2008 <a name="i_invoke">'<tt>invoke</tt>' Instruction</a>
2009</div>
2010
2011<div class="doc_text">
2012
2013<h5>Syntax:</h5>
2014
2015<pre>
Nick Lewyckya1c11a12008-03-16 07:18:12 +0000[diff] [blame]2016 &lt;result&gt; = invoke [<a href="#callingconv">cconv</a>] &lt;ptr to function ty&gt; &lt;function ptr val&gt;(&lt;function args&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2017 to label &lt;normal label&gt; unwind label &lt;exception label&gt;
2018</pre>
2019
2020<h5>Overview:</h5>
2021
2022<p>The '<tt>invoke</tt>' instruction causes control to transfer to a specified
2023function, with the possibility of control flow transfer to either the
2024'<tt>normal</tt>' label or the
2025'<tt>exception</tt>' label. If the callee function returns with the
2026"<tt><a href="#i_ret">ret</a></tt>" instruction, control flow will return to the
2027"normal" label. If the callee (or any indirect callees) returns with the "<a
2028href="#i_unwind"><tt>unwind</tt></a>" instruction, control is interrupted and
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]2029continued at the dynamically nearest "exception" label. If the callee function
Devang Patelec8a5b02008-03-11 05:51:59 +0000[diff] [blame]2030returns multiple values then individual return values are only accessible through
2031a '<tt><a href="#i_getresult">getresult</a></tt>' instruction.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2032
2033<h5>Arguments:</h5>
2034
2035<p>This instruction requires several arguments:</p>
2036
2037<ol>
2038 <li>
2039 The optional "cconv" marker indicates which <a href="#callingconv">calling
2040 convention</a> the call should use. If none is specified, the call defaults
2041 to using C calling conventions.
2042 </li>
2043 <li>'<tt>ptr to function ty</tt>': shall be the signature of the pointer to
2044 function value being invoked. In most cases, this is a direct function
2045 invocation, but indirect <tt>invoke</tt>s are just as possible, branching off
2046 an arbitrary pointer to function value.
2047 </li>
2048
2049 <li>'<tt>function ptr val</tt>': An LLVM value containing a pointer to a
2050 function to be invoked. </li>
2051
2052 <li>'<tt>function args</tt>': argument list whose types match the function
2053 signature argument types. If the function signature indicates the function
2054 accepts a variable number of arguments, the extra arguments can be
2055 specified. </li>
2056
2057 <li>'<tt>normal label</tt>': the label reached when the called function
2058 executes a '<tt><a href="#i_ret">ret</a></tt>' instruction. </li>
2059
2060 <li>'<tt>exception label</tt>': the label reached when a callee returns with
2061 the <a href="#i_unwind"><tt>unwind</tt></a> instruction. </li>
2062
2063</ol>
2064
2065<h5>Semantics:</h5>
2066
2067<p>This instruction is designed to operate as a standard '<tt><a
2068href="#i_call">call</a></tt>' instruction in most regards. The primary
2069difference is that it establishes an association with a label, which is used by
2070the runtime library to unwind the stack.</p>
2071
2072<p>This instruction is used in languages with destructors to ensure that proper
2073cleanup is performed in the case of either a <tt>longjmp</tt> or a thrown
2074exception. Additionally, this is important for implementation of
2075'<tt>catch</tt>' clauses in high-level languages that support them.</p>
2076
2077<h5>Example:</h5>
2078<pre>
Nick Lewyckya1c11a12008-03-16 07:18:12 +0000[diff] [blame]2079 %retval = invoke i32 @Test(i32 15) to label %Continue
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2080 unwind label %TestCleanup <i>; {i32}:retval set</i>
Nick Lewyckya1c11a12008-03-16 07:18:12 +0000[diff] [blame]2081 %retval = invoke <a href="#callingconv">coldcc</a> i32 %Testfnptr(i32 15) to label %Continue
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2082 unwind label %TestCleanup <i>; {i32}:retval set</i>
2083</pre>
2084</div>
2085
2086
2087<!-- _______________________________________________________________________ -->
2088
2089<div class="doc_subsubsection"> <a name="i_unwind">'<tt>unwind</tt>'
2090Instruction</a> </div>
2091
2092<div class="doc_text">
2093
2094<h5>Syntax:</h5>
2095<pre>
2096 unwind
2097</pre>
2098
2099<h5>Overview:</h5>
2100
2101<p>The '<tt>unwind</tt>' instruction unwinds the stack, continuing control flow
2102at the first callee in the dynamic call stack which used an <a
2103href="#i_invoke"><tt>invoke</tt></a> instruction to perform the call. This is
2104primarily used to implement exception handling.</p>
2105
2106<h5>Semantics:</h5>
2107
Chris Lattner8b094fc2008-04-19 21:01:16 +0000[diff] [blame]2108<p>The '<tt>unwind</tt>' instruction causes execution of the current function to
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2109immediately halt. The dynamic call stack is then searched for the first <a
2110href="#i_invoke"><tt>invoke</tt></a> instruction on the call stack. Once found,
2111execution continues at the "exceptional" destination block specified by the
2112<tt>invoke</tt> instruction. If there is no <tt>invoke</tt> instruction in the
2113dynamic call chain, undefined behavior results.</p>
2114</div>
2115
2116<!-- _______________________________________________________________________ -->
2117
2118<div class="doc_subsubsection"> <a name="i_unreachable">'<tt>unreachable</tt>'
2119Instruction</a> </div>
2120
2121<div class="doc_text">
2122
2123<h5>Syntax:</h5>
2124<pre>
2125 unreachable
2126</pre>
2127
2128<h5>Overview:</h5>
2129
2130<p>The '<tt>unreachable</tt>' instruction has no defined semantics. This
2131instruction is used to inform the optimizer that a particular portion of the
2132code is not reachable. This can be used to indicate that the code after a
2133no-return function cannot be reached, and other facts.</p>
2134
2135<h5>Semantics:</h5>
2136
2137<p>The '<tt>unreachable</tt>' instruction has no defined semantics.</p>
2138</div>
2139
2140
2141
2142<!-- ======================================================================= -->
2143<div class="doc_subsection"> <a name="binaryops">Binary Operations</a> </div>
2144<div class="doc_text">
2145<p>Binary operators are used to do most of the computation in a
Chris Lattnerab596d92008-04-01 18:47:32 +0000[diff] [blame]2146program. They require two operands of the same type, execute an operation on them, and
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2147produce a single value. The operands might represent
2148multiple 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]2149The result value has the same type as its operands.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2150<p>There are several different binary operators:</p>
2151</div>
2152<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2153<div class="doc_subsubsection">
2154 <a name="i_add">'<tt>add</tt>' Instruction</a>
2155</div>
2156
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2157<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2158
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2159<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2160
2161<pre>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2162 &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]2163</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2164
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2165<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2166
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2167<p>The '<tt>add</tt>' instruction returns the sum of its two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2168
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2169<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2170
2171<p>The two arguments to the '<tt>add</tt>' instruction must be <a
2172 href="#t_integer">integer</a>, <a href="#t_floating">floating point</a>, or
2173 <a href="#t_vector">vector</a> values. Both arguments must have identical
2174 types.</p>
2175
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2176<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2177
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2178<p>The value produced is the integer or floating point sum of the two
2179operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2180
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2181<p>If an integer sum has unsigned overflow, the result returned is the
2182mathematical result modulo 2<sup>n</sup>, where n is the bit width of
2183the result.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2184
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2185<p>Because LLVM integers use a two's complement representation, this
2186instruction is appropriate for both signed and unsigned integers.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2187
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2188<h5>Example:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2189
2190<pre>
2191 &lt;result&gt; = add i32 4, %var <i>; yields {i32}:result = 4 + %var</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2192</pre>
2193</div>
2194<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2195<div class="doc_subsubsection">
2196 <a name="i_sub">'<tt>sub</tt>' Instruction</a>
2197</div>
2198
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2199<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2200
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2201<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2202
2203<pre>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2204 &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]2205</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2206
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2207<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2208
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2209<p>The '<tt>sub</tt>' instruction returns the difference of its two
2210operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2211
2212<p>Note that the '<tt>sub</tt>' instruction is used to represent the
2213'<tt>neg</tt>' instruction present in most other intermediate
2214representations.</p>
2215
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2216<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2217
2218<p>The two arguments to the '<tt>sub</tt>' instruction must be <a
2219 href="#t_integer">integer</a>, <a href="#t_floating">floating point</a>,
2220 or <a href="#t_vector">vector</a> values. Both arguments must have identical
2221 types.</p>
2222
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2223<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2224
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2225<p>The value produced is the integer or floating point difference of
2226the two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2227
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2228<p>If an integer difference has unsigned overflow, the result returned is the
2229mathematical result modulo 2<sup>n</sup>, where n is the bit width of
2230the result.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2231
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2232<p>Because LLVM integers use a two's complement representation, this
2233instruction is appropriate for both signed and unsigned integers.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2234
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2235<h5>Example:</h5>
2236<pre>
2237 &lt;result&gt; = sub i32 4, %var <i>; yields {i32}:result = 4 - %var</i>
2238 &lt;result&gt; = sub i32 0, %val <i>; yields {i32}:result = -%var</i>
2239</pre>
2240</div>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2241
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2242<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2243<div class="doc_subsubsection">
2244 <a name="i_mul">'<tt>mul</tt>' Instruction</a>
2245</div>
2246
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2247<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2248
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2249<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2250<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]2251</pre>
2252<h5>Overview:</h5>
2253<p>The '<tt>mul</tt>' instruction returns the product of its two
2254operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2255
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2256<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2257
2258<p>The two arguments to the '<tt>mul</tt>' instruction must be <a
2259href="#t_integer">integer</a>, <a href="#t_floating">floating point</a>,
2260or <a href="#t_vector">vector</a> values. Both arguments must have identical
2261types.</p>
2262
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2263<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2264
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2265<p>The value produced is the integer or floating point product of the
2266two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2267
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2268<p>If the result of an integer multiplication has unsigned overflow,
2269the result returned is the mathematical result modulo
22702<sup>n</sup>, where n is the bit width of the result.</p>
2271<p>Because LLVM integers use a two's complement representation, and the
2272result is the same width as the operands, this instruction returns the
2273correct result for both signed and unsigned integers. If a full product
2274(e.g. <tt>i32</tt>x<tt>i32</tt>-><tt>i64</tt>) is needed, the operands
2275should be sign-extended or zero-extended as appropriate to the
2276width of the full product.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2277<h5>Example:</h5>
2278<pre> &lt;result&gt; = mul i32 4, %var <i>; yields {i32}:result = 4 * %var</i>
2279</pre>
2280</div>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2281
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2282<!-- _______________________________________________________________________ -->
2283<div class="doc_subsubsection"> <a name="i_udiv">'<tt>udiv</tt>' Instruction
2284</a></div>
2285<div class="doc_text">
2286<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2287<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]2288</pre>
2289<h5>Overview:</h5>
2290<p>The '<tt>udiv</tt>' instruction returns the quotient of its two
2291operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2292
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2293<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2294
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2295<p>The two arguments to the '<tt>udiv</tt>' instruction must be
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2296<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2297values. Both arguments must have identical types.</p>
2298
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2299<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2300
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2301<p>The value produced is the unsigned integer quotient of the two operands.</p>
2302<p>Note that unsigned integer division and signed integer division are distinct
2303operations; for signed integer division, use '<tt>sdiv</tt>'.</p>
2304<p>Division by zero leads to undefined behavior.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2305<h5>Example:</h5>
2306<pre> &lt;result&gt; = udiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
2307</pre>
2308</div>
2309<!-- _______________________________________________________________________ -->
2310<div class="doc_subsubsection"> <a name="i_sdiv">'<tt>sdiv</tt>' Instruction
2311</a> </div>
2312<div class="doc_text">
2313<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2314<pre>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2315 &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]2316</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2317
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2318<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2319
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2320<p>The '<tt>sdiv</tt>' instruction returns the quotient of its two
2321operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2322
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2323<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2324
2325<p>The two arguments to the '<tt>sdiv</tt>' instruction must be
2326<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2327values. Both arguments must have identical types.</p>
2328
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2329<h5>Semantics:</h5>
Chris Lattnerdc15b1d2008-04-01 18:45:27 +0000[diff] [blame]2330<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]2331<p>Note that signed integer division and unsigned integer division are distinct
2332operations; for unsigned integer division, use '<tt>udiv</tt>'.</p>
2333<p>Division by zero leads to undefined behavior. Overflow also leads to
2334undefined behavior; this is a rare case, but can occur, for example,
2335by doing a 32-bit division of -2147483648 by -1.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2336<h5>Example:</h5>
2337<pre> &lt;result&gt; = sdiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
2338</pre>
2339</div>
2340<!-- _______________________________________________________________________ -->
2341<div class="doc_subsubsection"> <a name="i_fdiv">'<tt>fdiv</tt>'
2342Instruction</a> </div>
2343<div class="doc_text">
2344<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2345<pre>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2346 &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]2347</pre>
2348<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2349
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2350<p>The '<tt>fdiv</tt>' instruction returns the quotient of its two
2351operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2352
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2353<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2354
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2355<p>The two arguments to the '<tt>fdiv</tt>' instruction must be
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2356<a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a>
2357of floating point values. Both arguments must have identical types.</p>
2358
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2359<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2360
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2361<p>The value produced is the floating point quotient of the two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2362
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2363<h5>Example:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2364
2365<pre>
2366 &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]2367</pre>
2368</div>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2369
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2370<!-- _______________________________________________________________________ -->
2371<div class="doc_subsubsection"> <a name="i_urem">'<tt>urem</tt>' Instruction</a>
2372</div>
2373<div class="doc_text">
2374<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2375<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]2376</pre>
2377<h5>Overview:</h5>
2378<p>The '<tt>urem</tt>' instruction returns the remainder from the
2379unsigned division of its two arguments.</p>
2380<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2381<p>The two arguments to the '<tt>urem</tt>' instruction must be
2382<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2383values. Both arguments must have identical types.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2384<h5>Semantics:</h5>
2385<p>This instruction returns the unsigned integer <i>remainder</i> of a division.
Chris Lattnerdc15b1d2008-04-01 18:45:27 +0000[diff] [blame]2386This instruction always performs an unsigned division to get the remainder.</p>
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2387<p>Note that unsigned integer remainder and signed integer remainder are
2388distinct operations; for signed integer remainder, use '<tt>srem</tt>'.</p>
2389<p>Taking the remainder of a division by zero leads to undefined behavior.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2390<h5>Example:</h5>
2391<pre> &lt;result&gt; = urem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
2392</pre>
2393
2394</div>
2395<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2396<div class="doc_subsubsection">
2397 <a name="i_srem">'<tt>srem</tt>' Instruction</a>
2398</div>
2399
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2400<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2401
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2402<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2403
2404<pre>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2405 &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]2406</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2407
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2408<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2409
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2410<p>The '<tt>srem</tt>' instruction returns the remainder from the
Dan Gohman3e3fd8c2007-11-05 23:35:22 +0000[diff] [blame]2411signed division of its two operands. This instruction can also take
2412<a href="#t_vector">vector</a> versions of the values in which case
2413the elements must be integers.</p>
Chris Lattner08497ce2008-01-04 04:33:49 +0000[diff] [blame]2414
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2415<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2416
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2417<p>The two arguments to the '<tt>srem</tt>' instruction must be
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2418<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2419values. Both arguments must have identical types.</p>
2420
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2421<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2422
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2423<p>This instruction returns the <i>remainder</i> of a division (where the result
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2424has the same sign as the dividend, <tt>op1</tt>), not the <i>modulo</i>
2425operator (where the result has the same sign as the divisor, <tt>op2</tt>) of
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2426a value. For more information about the difference, see <a
2427 href="http://mathforum.org/dr.math/problems/anne.4.28.99.html">The
2428Math Forum</a>. For a table of how this is implemented in various languages,
2429please see <a href="http://en.wikipedia.org/wiki/Modulo_operation">
2430Wikipedia: modulo operation</a>.</p>
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2431<p>Note that signed integer remainder and unsigned integer remainder are
2432distinct operations; for unsigned integer remainder, use '<tt>urem</tt>'.</p>
2433<p>Taking the remainder of a division by zero leads to undefined behavior.
2434Overflow also leads to undefined behavior; this is a rare case, but can occur,
2435for example, by taking the remainder of a 32-bit division of -2147483648 by -1.
2436(The remainder doesn't actually overflow, but this rule lets srem be
2437implemented using instructions that return both the result of the division
2438and the remainder.)</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2439<h5>Example:</h5>
2440<pre> &lt;result&gt; = srem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
2441</pre>
2442
2443</div>
2444<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2445<div class="doc_subsubsection">
2446 <a name="i_frem">'<tt>frem</tt>' Instruction</a> </div>
2447
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2448<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2449
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2450<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2451<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]2452</pre>
2453<h5>Overview:</h5>
2454<p>The '<tt>frem</tt>' instruction returns the remainder from the
2455division of its two operands.</p>
2456<h5>Arguments:</h5>
2457<p>The two arguments to the '<tt>frem</tt>' instruction must be
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2458<a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a>
2459of floating point values. Both arguments must have identical types.</p>
2460
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2461<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2462
Chris Lattnerdc15b1d2008-04-01 18:45:27 +0000[diff] [blame]2463<p>This instruction returns the <i>remainder</i> of a division.
2464The remainder has the same sign as the dividend.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2465
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2466<h5>Example:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2467
2468<pre>
2469 &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]2470</pre>
2471</div>
2472
2473<!-- ======================================================================= -->
2474<div class="doc_subsection"> <a name="bitwiseops">Bitwise Binary
2475Operations</a> </div>
2476<div class="doc_text">
2477<p>Bitwise binary operators are used to do various forms of
2478bit-twiddling in a program. They are generally very efficient
2479instructions and can commonly be strength reduced from other
Chris Lattnerdc15b1d2008-04-01 18:45:27 +0000[diff] [blame]2480instructions. They require two operands of the same type, execute an operation on them,
2481and 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]2482</div>
2483
2484<!-- _______________________________________________________________________ -->
2485<div class="doc_subsubsection"> <a name="i_shl">'<tt>shl</tt>'
2486Instruction</a> </div>
2487<div class="doc_text">
2488<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2489<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]2490</pre>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2491
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2492<h5>Overview:</h5>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2493
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2494<p>The '<tt>shl</tt>' instruction returns the first operand shifted to
2495the left a specified number of bits.</p>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2496
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2497<h5>Arguments:</h5>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2498
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2499<p>Both arguments to the '<tt>shl</tt>' instruction must be the same <a
Nate Begemanbb1ce942008-07-29 15:49:41 +0000[diff] [blame]2500 href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2501type. '<tt>op2</tt>' is treated as an unsigned value.</p>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2502
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2503<h5>Semantics:</h5>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2504
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2505<p>The value produced is <tt>op1</tt> * 2<sup><tt>op2</tt></sup> mod 2<sup>n</sup>,
2506where n is the width of the result. If <tt>op2</tt> is (statically or dynamically) negative or
2507equal to or larger than the number of bits in <tt>op1</tt>, the result is undefined.</p>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2508
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2509<h5>Example:</h5><pre>
2510 &lt;result&gt; = shl i32 4, %var <i>; yields {i32}: 4 &lt;&lt; %var</i>
2511 &lt;result&gt; = shl i32 4, 2 <i>; yields {i32}: 16</i>
2512 &lt;result&gt; = shl i32 1, 10 <i>; yields {i32}: 1024</i>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2513 &lt;result&gt; = shl i32 1, 32 <i>; undefined</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2514</pre>
2515</div>
2516<!-- _______________________________________________________________________ -->
2517<div class="doc_subsubsection"> <a name="i_lshr">'<tt>lshr</tt>'
2518Instruction</a> </div>
2519<div class="doc_text">
2520<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2521<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]2522</pre>
2523
2524<h5>Overview:</h5>
2525<p>The '<tt>lshr</tt>' instruction (logical shift right) returns the first
2526operand shifted to the right a specified number of bits with zero fill.</p>
2527
2528<h5>Arguments:</h5>
2529<p>Both arguments to the '<tt>lshr</tt>' instruction must be the same
Nate Begemanbb1ce942008-07-29 15:49:41 +0000[diff] [blame]2530<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]2531type. '<tt>op2</tt>' is treated as an unsigned value.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2532
2533<h5>Semantics:</h5>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2534
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2535<p>This instruction always performs a logical shift right operation. The most
2536significant bits of the result will be filled with zero bits after the
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2537shift. If <tt>op2</tt> is (statically or dynamically) equal to or larger than
2538the number of bits in <tt>op1</tt>, the result is undefined.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2539
2540<h5>Example:</h5>
2541<pre>
2542 &lt;result&gt; = lshr i32 4, 1 <i>; yields {i32}:result = 2</i>
2543 &lt;result&gt; = lshr i32 4, 2 <i>; yields {i32}:result = 1</i>
2544 &lt;result&gt; = lshr i8 4, 3 <i>; yields {i8}:result = 0</i>
2545 &lt;result&gt; = lshr i8 -2, 1 <i>; yields {i8}:result = 0x7FFFFFFF </i>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2546 &lt;result&gt; = lshr i32 1, 32 <i>; undefined</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2547</pre>
2548</div>
2549
2550<!-- _______________________________________________________________________ -->
2551<div class="doc_subsubsection"> <a name="i_ashr">'<tt>ashr</tt>'
2552Instruction</a> </div>
2553<div class="doc_text">
2554
2555<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2556<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]2557</pre>
2558
2559<h5>Overview:</h5>
2560<p>The '<tt>ashr</tt>' instruction (arithmetic shift right) returns the first
2561operand shifted to the right a specified number of bits with sign extension.</p>
2562
2563<h5>Arguments:</h5>
2564<p>Both arguments to the '<tt>ashr</tt>' instruction must be the same
Nate Begemanbb1ce942008-07-29 15:49:41 +0000[diff] [blame]2565<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]2566type. '<tt>op2</tt>' is treated as an unsigned value.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2567
2568<h5>Semantics:</h5>
2569<p>This instruction always performs an arithmetic shift right operation,
2570The most significant bits of the result will be filled with the sign bit
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2571of <tt>op1</tt>. If <tt>op2</tt> is (statically or dynamically) equal to or
2572larger than the number of bits in <tt>op1</tt>, the result is undefined.
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2573</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2574
2575<h5>Example:</h5>
2576<pre>
2577 &lt;result&gt; = ashr i32 4, 1 <i>; yields {i32}:result = 2</i>
2578 &lt;result&gt; = ashr i32 4, 2 <i>; yields {i32}:result = 1</i>
2579 &lt;result&gt; = ashr i8 4, 3 <i>; yields {i8}:result = 0</i>
2580 &lt;result&gt; = ashr i8 -2, 1 <i>; yields {i8}:result = -1</i>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2581 &lt;result&gt; = ashr i32 1, 32 <i>; undefined</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2582</pre>
2583</div>
2584
2585<!-- _______________________________________________________________________ -->
2586<div class="doc_subsubsection"> <a name="i_and">'<tt>and</tt>'
2587Instruction</a> </div>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2588
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2589<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2590
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2591<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2592
2593<pre>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2594 &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]2595</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2596
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2597<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2598
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2599<p>The '<tt>and</tt>' instruction returns the bitwise logical and of
2600its two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2601
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2602<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2603
2604<p>The two arguments to the '<tt>and</tt>' instruction must be
2605<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2606values. Both arguments must have identical types.</p>
2607
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2608<h5>Semantics:</h5>
2609<p>The truth table used for the '<tt>and</tt>' instruction is:</p>
2610<p> </p>
2611<div style="align: center">
2612<table border="1" cellspacing="0" cellpadding="4">
2613 <tbody>
2614 <tr>
2615 <td>In0</td>
2616 <td>In1</td>
2617 <td>Out</td>
2618 </tr>
2619 <tr>
2620 <td>0</td>
2621 <td>0</td>
2622 <td>0</td>
2623 </tr>
2624 <tr>
2625 <td>0</td>
2626 <td>1</td>
2627 <td>0</td>
2628 </tr>
2629 <tr>
2630 <td>1</td>
2631 <td>0</td>
2632 <td>0</td>
2633 </tr>
2634 <tr>
2635 <td>1</td>
2636 <td>1</td>
2637 <td>1</td>
2638 </tr>
2639 </tbody>
2640</table>
2641</div>
2642<h5>Example:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2643<pre>
2644 &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]2645 &lt;result&gt; = and i32 15, 40 <i>; yields {i32}:result = 8</i>
2646 &lt;result&gt; = and i32 4, 8 <i>; yields {i32}:result = 0</i>
2647</pre>
2648</div>
2649<!-- _______________________________________________________________________ -->
2650<div class="doc_subsubsection"> <a name="i_or">'<tt>or</tt>' Instruction</a> </div>
2651<div class="doc_text">
2652<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2653<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]2654</pre>
2655<h5>Overview:</h5>
2656<p>The '<tt>or</tt>' instruction returns the bitwise logical inclusive
2657or of its two operands.</p>
2658<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2659
2660<p>The two arguments to the '<tt>or</tt>' instruction must be
2661<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2662values. Both arguments must have identical types.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2663<h5>Semantics:</h5>
2664<p>The truth table used for the '<tt>or</tt>' instruction is:</p>
2665<p> </p>
2666<div style="align: center">
2667<table border="1" cellspacing="0" cellpadding="4">
2668 <tbody>
2669 <tr>
2670 <td>In0</td>
2671 <td>In1</td>
2672 <td>Out</td>
2673 </tr>
2674 <tr>
2675 <td>0</td>
2676 <td>0</td>
2677 <td>0</td>
2678 </tr>
2679 <tr>
2680 <td>0</td>
2681 <td>1</td>
2682 <td>1</td>
2683 </tr>
2684 <tr>
2685 <td>1</td>
2686 <td>0</td>
2687 <td>1</td>
2688 </tr>
2689 <tr>
2690 <td>1</td>
2691 <td>1</td>
2692 <td>1</td>
2693 </tr>
2694 </tbody>
2695</table>
2696</div>
2697<h5>Example:</h5>
2698<pre> &lt;result&gt; = or i32 4, %var <i>; yields {i32}:result = 4 | %var</i>
2699 &lt;result&gt; = or i32 15, 40 <i>; yields {i32}:result = 47</i>
2700 &lt;result&gt; = or i32 4, 8 <i>; yields {i32}:result = 12</i>
2701</pre>
2702</div>
2703<!-- _______________________________________________________________________ -->
2704<div class="doc_subsubsection"> <a name="i_xor">'<tt>xor</tt>'
2705Instruction</a> </div>
2706<div class="doc_text">
2707<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2708<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]2709</pre>
2710<h5>Overview:</h5>
2711<p>The '<tt>xor</tt>' instruction returns the bitwise logical exclusive
2712or of its two operands. The <tt>xor</tt> is used to implement the
2713"one's complement" operation, which is the "~" operator in C.</p>
2714<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2715<p>The two arguments to the '<tt>xor</tt>' instruction must be
2716<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2717values. Both arguments must have identical types.</p>
2718
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2719<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2720
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2721<p>The truth table used for the '<tt>xor</tt>' instruction is:</p>
2722<p> </p>
2723<div style="align: center">
2724<table border="1" cellspacing="0" cellpadding="4">
2725 <tbody>
2726 <tr>
2727 <td>In0</td>
2728 <td>In1</td>
2729 <td>Out</td>
2730 </tr>
2731 <tr>
2732 <td>0</td>
2733 <td>0</td>
2734 <td>0</td>
2735 </tr>
2736 <tr>
2737 <td>0</td>
2738 <td>1</td>
2739 <td>1</td>
2740 </tr>
2741 <tr>
2742 <td>1</td>
2743 <td>0</td>
2744 <td>1</td>
2745 </tr>
2746 <tr>
2747 <td>1</td>
2748 <td>1</td>
2749 <td>0</td>
2750 </tr>
2751 </tbody>
2752</table>
2753</div>
2754<p> </p>
2755<h5>Example:</h5>
2756<pre> &lt;result&gt; = xor i32 4, %var <i>; yields {i32}:result = 4 ^ %var</i>
2757 &lt;result&gt; = xor i32 15, 40 <i>; yields {i32}:result = 39</i>
2758 &lt;result&gt; = xor i32 4, 8 <i>; yields {i32}:result = 12</i>
2759 &lt;result&gt; = xor i32 %V, -1 <i>; yields {i32}:result = ~%V</i>
2760</pre>
2761</div>
2762
2763<!-- ======================================================================= -->
2764<div class="doc_subsection">
2765 <a name="vectorops">Vector Operations</a>
2766</div>
2767
2768<div class="doc_text">
2769
2770<p>LLVM supports several instructions to represent vector operations in a
2771target-independent manner. These instructions cover the element-access and
2772vector-specific operations needed to process vectors effectively. While LLVM
2773does directly support these vector operations, many sophisticated algorithms
2774will want to use target-specific intrinsics to take full advantage of a specific
2775target.</p>
2776
2777</div>
2778
2779<!-- _______________________________________________________________________ -->
2780<div class="doc_subsubsection">
2781 <a name="i_extractelement">'<tt>extractelement</tt>' Instruction</a>
2782</div>
2783
2784<div class="doc_text">
2785
2786<h5>Syntax:</h5>
2787
2788<pre>
2789 &lt;result&gt; = extractelement &lt;n x &lt;ty&gt;&gt; &lt;val&gt;, i32 &lt;idx&gt; <i>; yields &lt;ty&gt;</i>
2790</pre>
2791
2792<h5>Overview:</h5>
2793
2794<p>
2795The '<tt>extractelement</tt>' instruction extracts a single scalar
2796element from a vector at a specified index.
2797</p>
2798
2799
2800<h5>Arguments:</h5>
2801
2802<p>
2803The first operand of an '<tt>extractelement</tt>' instruction is a
2804value of <a href="#t_vector">vector</a> type. The second operand is
2805an index indicating the position from which to extract the element.
2806The index may be a variable.</p>
2807
2808<h5>Semantics:</h5>
2809
2810<p>
2811The result is a scalar of the same type as the element type of
2812<tt>val</tt>. Its value is the value at position <tt>idx</tt> of
2813<tt>val</tt>. If <tt>idx</tt> exceeds the length of <tt>val</tt>, the
2814results are undefined.
2815</p>
2816
2817<h5>Example:</h5>
2818
2819<pre>
2820 %result = extractelement &lt;4 x i32&gt; %vec, i32 0 <i>; yields i32</i>
2821</pre>
2822</div>
2823
2824
2825<!-- _______________________________________________________________________ -->
2826<div class="doc_subsubsection">
2827 <a name="i_insertelement">'<tt>insertelement</tt>' Instruction</a>
2828</div>
2829
2830<div class="doc_text">
2831
2832<h5>Syntax:</h5>
2833
2834<pre>
Dan Gohmanbcc3c502008-05-12 23:38:42 +0000[diff] [blame]2835 &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]2836</pre>
2837
2838<h5>Overview:</h5>
2839
2840<p>
2841The '<tt>insertelement</tt>' instruction inserts a scalar
2842element into a vector at a specified index.
2843</p>
2844
2845
2846<h5>Arguments:</h5>
2847
2848<p>
2849The first operand of an '<tt>insertelement</tt>' instruction is a
2850value of <a href="#t_vector">vector</a> type. The second operand is a
2851scalar value whose type must equal the element type of the first
2852operand. The third operand is an index indicating the position at
2853which to insert the value. The index may be a variable.</p>
2854
2855<h5>Semantics:</h5>
2856
2857<p>
2858The result is a vector of the same type as <tt>val</tt>. Its
2859element values are those of <tt>val</tt> except at position
2860<tt>idx</tt>, where it gets the value <tt>elt</tt>. If <tt>idx</tt>
2861exceeds the length of <tt>val</tt>, the results are undefined.
2862</p>
2863
2864<h5>Example:</h5>
2865
2866<pre>
2867 %result = insertelement &lt;4 x i32&gt; %vec, i32 1, i32 0 <i>; yields &lt;4 x i32&gt;</i>
2868</pre>
2869</div>
2870
2871<!-- _______________________________________________________________________ -->
2872<div class="doc_subsubsection">
2873 <a name="i_shufflevector">'<tt>shufflevector</tt>' Instruction</a>
2874</div>
2875
2876<div class="doc_text">
2877
2878<h5>Syntax:</h5>
2879
2880<pre>
2881 &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;n x i32&gt; &lt;mask&gt; <i>; yields &lt;n x &lt;ty&gt;&gt;</i>
2882</pre>
2883
2884<h5>Overview:</h5>
2885
2886<p>
2887The '<tt>shufflevector</tt>' instruction constructs a permutation of elements
2888from two input vectors, returning a vector of the same type.
2889</p>
2890
2891<h5>Arguments:</h5>
2892
2893<p>
2894The first two operands of a '<tt>shufflevector</tt>' instruction are vectors
2895with types that match each other and types that match the result of the
2896instruction. The third argument is a shuffle mask, which has the same number
2897of elements as the other vector type, but whose element type is always 'i32'.
2898</p>
2899
2900<p>
2901The shuffle mask operand is required to be a constant vector with either
2902constant integer or undef values.
2903</p>
2904
2905<h5>Semantics:</h5>
2906
2907<p>
2908The elements of the two input vectors are numbered from left to right across
2909both of the vectors. The shuffle mask operand specifies, for each element of
2910the result vector, which element of the two input registers the result element
2911gets. The element selector may be undef (meaning "don't care") and the second
2912operand may be undef if performing a shuffle from only one vector.
2913</p>
2914
2915<h5>Example:</h5>
2916
2917<pre>
2918 %result = shufflevector &lt;4 x i32&gt; %v1, &lt;4 x i32&gt; %v2,
2919 &lt;4 x i32&gt; &lt;i32 0, i32 4, i32 1, i32 5&gt; <i>; yields &lt;4 x i32&gt;</i>
2920 %result = shufflevector &lt;4 x i32&gt; %v1, &lt;4 x i32&gt; undef,
2921 &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.
2922</pre>
2923</div>
2924
2925
2926<!-- ======================================================================= -->
2927<div class="doc_subsection">
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]2928 <a name="aggregateops">Aggregate Operations</a>
2929</div>
2930
2931<div class="doc_text">
2932
2933<p>LLVM supports several instructions for working with aggregate values.
2934</p>
2935
2936</div>
2937
2938<!-- _______________________________________________________________________ -->
2939<div class="doc_subsubsection">
2940 <a name="i_extractvalue">'<tt>extractvalue</tt>' Instruction</a>
2941</div>
2942
2943<div class="doc_text">
2944
2945<h5>Syntax:</h5>
2946
2947<pre>
2948 &lt;result&gt; = extractvalue &lt;aggregate type&gt; &lt;val&gt;, &lt;idx&gt;{, &lt;idx&gt;}*
2949</pre>
2950
2951<h5>Overview:</h5>
2952
2953<p>
Dan Gohman6c6dea02008-05-13 18:16:06 +0000[diff] [blame]2954The '<tt>extractvalue</tt>' instruction extracts the value of a struct field
2955or array element from an aggregate value.
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]2956</p>
2957
2958
2959<h5>Arguments:</h5>
2960
2961<p>
2962The first operand of an '<tt>extractvalue</tt>' instruction is a
2963value of <a href="#t_struct">struct</a> or <a href="#t_array">array</a>
Dan Gohman6c6dea02008-05-13 18:16:06 +0000[diff] [blame]2964type. The operands are constant indices to specify which value to extract
Dan Gohmane5febe42008-05-31 00:58:22 +0000[diff] [blame]2965in a similar manner as indices in a
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]2966'<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.
2967</p>
2968
2969<h5>Semantics:</h5>
2970
2971<p>
2972The result is the value at the position in the aggregate specified by
2973the index operands.
2974</p>
2975
2976<h5>Example:</h5>
2977
2978<pre>
Dan Gohmane5febe42008-05-31 00:58:22 +0000[diff] [blame]2979 %result = extractvalue {i32, float} %agg, 0 <i>; yields i32</i>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]2980</pre>
2981</div>
2982
2983
2984<!-- _______________________________________________________________________ -->
2985<div class="doc_subsubsection">
2986 <a name="i_insertvalue">'<tt>insertvalue</tt>' Instruction</a>
2987</div>
2988
2989<div class="doc_text">
2990
2991<h5>Syntax:</h5>
2992
2993<pre>
Dan Gohmane5febe42008-05-31 00:58:22 +0000[diff] [blame]2994 &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]2995</pre>
2996
2997<h5>Overview:</h5>
2998
2999<p>
3000The '<tt>insertvalue</tt>' instruction inserts a value
Dan Gohman6c6dea02008-05-13 18:16:06 +0000[diff] [blame]3001into a struct field or array element in an aggregate.
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3002</p>
3003
3004
3005<h5>Arguments:</h5>
3006
3007<p>
3008The first operand of an '<tt>insertvalue</tt>' instruction is a
3009value of <a href="#t_struct">struct</a> or <a href="#t_array">array</a> type.
3010The second operand is a first-class value to insert.
Dan Gohman4f29e422008-05-23 21:53:15 +0000[diff] [blame]3011The following operands are constant indices
Dan Gohmane5febe42008-05-31 00:58:22 +0000[diff] [blame]3012indicating the position at which to insert the value in a similar manner as
Dan Gohman6c6dea02008-05-13 18:16:06 +0000[diff] [blame]3013indices in a
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3014'<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.
3015The value to insert must have the same type as the value identified
Dan Gohman6c6dea02008-05-13 18:16:06 +0000[diff] [blame]3016by the indices.
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3017
3018<h5>Semantics:</h5>
3019
3020<p>
3021The result is an aggregate of the same type as <tt>val</tt>. Its
3022value is that of <tt>val</tt> except that the value at the position
Dan Gohman6c6dea02008-05-13 18:16:06 +0000[diff] [blame]3023specified by the indices is that of <tt>elt</tt>.
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3024</p>
3025
3026<h5>Example:</h5>
3027
3028<pre>
Dan Gohmanb1aab4e2008-06-23 15:26:37 +0000[diff] [blame]3029 %result = insertvalue {i32, float} %agg, i32 1, 0 <i>; yields {i32, float}</i>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3030</pre>
3031</div>
3032
3033
3034<!-- ======================================================================= -->
3035<div class="doc_subsection">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3036 <a name="memoryops">Memory Access and Addressing Operations</a>
3037</div>
3038
3039<div class="doc_text">
3040
3041<p>A key design point of an SSA-based representation is how it
3042represents memory. In LLVM, no memory locations are in SSA form, which
3043makes things very simple. This section describes how to read, write,
3044allocate, and free memory in LLVM.</p>
3045
3046</div>
3047
3048<!-- _______________________________________________________________________ -->
3049<div class="doc_subsubsection">
3050 <a name="i_malloc">'<tt>malloc</tt>' Instruction</a>
3051</div>
3052
3053<div class="doc_text">
3054
3055<h5>Syntax:</h5>
3056
3057<pre>
3058 &lt;result&gt; = malloc &lt;type&gt;[, i32 &lt;NumElements&gt;][, align &lt;alignment&gt;] <i>; yields {type*}:result</i>
3059</pre>
3060
3061<h5>Overview:</h5>
3062
3063<p>The '<tt>malloc</tt>' instruction allocates memory from the system
Christopher Lambcfe00962007-12-17 01:00:21 +0000[diff] [blame]3064heap and returns a pointer to it. The object is always allocated in the generic
3065address space (address space zero).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3066
3067<h5>Arguments:</h5>
3068
3069<p>The '<tt>malloc</tt>' instruction allocates
3070<tt>sizeof(&lt;type&gt;)*NumElements</tt>
3071bytes of memory from the operating system and returns a pointer of the
3072appropriate type to the program. If "NumElements" is specified, it is the
Gabor Greif5082cf42008-02-09 22:24:34 +0000[diff] [blame]3073number of elements allocated, otherwise "NumElements" is defaulted to be one.
Chris Lattner10368b62008-04-02 00:38:26 +0000[diff] [blame]3074If a constant alignment is specified, the value result of the allocation is guaranteed to
Gabor Greif5082cf42008-02-09 22:24:34 +0000[diff] [blame]3075be aligned to at least that boundary. If not specified, or if zero, the target can
3076choose to align the allocation on any convenient boundary.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3077
3078<p>'<tt>type</tt>' must be a sized type.</p>
3079
3080<h5>Semantics:</h5>
3081
3082<p>Memory is allocated using the system "<tt>malloc</tt>" function, and
Chris Lattner8b094fc2008-04-19 21:01:16 +0000[diff] [blame]3083a pointer is returned. The result of a zero byte allocattion is undefined. The
3084result is null if there is insufficient memory available.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3085
3086<h5>Example:</h5>
3087
3088<pre>
3089 %array = malloc [4 x i8 ] <i>; yields {[%4 x i8]*}:array</i>
3090
3091 %size = <a href="#i_add">add</a> i32 2, 2 <i>; yields {i32}:size = i32 4</i>
3092 %array1 = malloc i8, i32 4 <i>; yields {i8*}:array1</i>
3093 %array2 = malloc [12 x i8], i32 %size <i>; yields {[12 x i8]*}:array2</i>
3094 %array3 = malloc i32, i32 4, align 1024 <i>; yields {i32*}:array3</i>
3095 %array4 = malloc i32, align 1024 <i>; yields {i32*}:array4</i>
3096</pre>
3097</div>
3098
3099<!-- _______________________________________________________________________ -->
3100<div class="doc_subsubsection">
3101 <a name="i_free">'<tt>free</tt>' Instruction</a>
3102</div>
3103
3104<div class="doc_text">
3105
3106<h5>Syntax:</h5>
3107
3108<pre>
3109 free &lt;type&gt; &lt;value&gt; <i>; yields {void}</i>
3110</pre>
3111
3112<h5>Overview:</h5>
3113
3114<p>The '<tt>free</tt>' instruction returns memory back to the unused
3115memory heap to be reallocated in the future.</p>
3116
3117<h5>Arguments:</h5>
3118
3119<p>'<tt>value</tt>' shall be a pointer value that points to a value
3120that was allocated with the '<tt><a href="#i_malloc">malloc</a></tt>'
3121instruction.</p>
3122
3123<h5>Semantics:</h5>
3124
3125<p>Access to the memory pointed to by the pointer is no longer defined
Chris Lattner329d6532008-04-19 22:41:32 +0000[diff] [blame]3126after this instruction executes. If the pointer is null, the operation
3127is a noop.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3128
3129<h5>Example:</h5>
3130
3131<pre>
3132 %array = <a href="#i_malloc">malloc</a> [4 x i8] <i>; yields {[4 x i8]*}:array</i>
3133 free [4 x i8]* %array
3134</pre>
3135</div>
3136
3137<!-- _______________________________________________________________________ -->
3138<div class="doc_subsubsection">
3139 <a name="i_alloca">'<tt>alloca</tt>' Instruction</a>
3140</div>
3141
3142<div class="doc_text">
3143
3144<h5>Syntax:</h5>
3145
3146<pre>
3147 &lt;result&gt; = alloca &lt;type&gt;[, i32 &lt;NumElements&gt;][, align &lt;alignment&gt;] <i>; yields {type*}:result</i>
3148</pre>
3149
3150<h5>Overview:</h5>
3151
3152<p>The '<tt>alloca</tt>' instruction allocates memory on the stack frame of the
3153currently executing function, to be automatically released when this function
Christopher Lambcfe00962007-12-17 01:00:21 +0000[diff] [blame]3154returns to its caller. The object is always allocated in the generic address
3155space (address space zero).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3156
3157<h5>Arguments:</h5>
3158
3159<p>The '<tt>alloca</tt>' instruction allocates <tt>sizeof(&lt;type&gt;)*NumElements</tt>
3160bytes of memory on the runtime stack, returning a pointer of the
Gabor Greif5082cf42008-02-09 22:24:34 +0000[diff] [blame]3161appropriate type to the program. If "NumElements" is specified, it is the
3162number of elements allocated, otherwise "NumElements" is defaulted to be one.
Chris Lattner10368b62008-04-02 00:38:26 +0000[diff] [blame]3163If a constant alignment is specified, the value result of the allocation is guaranteed
Gabor Greif5082cf42008-02-09 22:24:34 +0000[diff] [blame]3164to be aligned to at least that boundary. If not specified, or if zero, the target
3165can choose to align the allocation on any convenient boundary.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3166
3167<p>'<tt>type</tt>' may be any sized type.</p>
3168
3169<h5>Semantics:</h5>
3170
Chris Lattner8b094fc2008-04-19 21:01:16 +0000[diff] [blame]3171<p>Memory is allocated; a pointer is returned. The operation is undefiend if
3172there is insufficient stack space for the allocation. '<tt>alloca</tt>'d
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3173memory is automatically released when the function returns. The '<tt>alloca</tt>'
3174instruction is commonly used to represent automatic variables that must
3175have an address available. When the function returns (either with the <tt><a
3176 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]3177instructions), the memory is reclaimed. Allocating zero bytes
3178is legal, but the result is undefined.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3179
3180<h5>Example:</h5>
3181
3182<pre>
3183 %ptr = alloca i32 <i>; yields {i32*}:ptr</i>
3184 %ptr = alloca i32, i32 4 <i>; yields {i32*}:ptr</i>
3185 %ptr = alloca i32, i32 4, align 1024 <i>; yields {i32*}:ptr</i>
3186 %ptr = alloca i32, align 1024 <i>; yields {i32*}:ptr</i>
3187</pre>
3188</div>
3189
3190<!-- _______________________________________________________________________ -->
3191<div class="doc_subsubsection"> <a name="i_load">'<tt>load</tt>'
3192Instruction</a> </div>
3193<div class="doc_text">
3194<h5>Syntax:</h5>
3195<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>
3196<h5>Overview:</h5>
3197<p>The '<tt>load</tt>' instruction is used to read from memory.</p>
3198<h5>Arguments:</h5>
3199<p>The argument to the '<tt>load</tt>' instruction specifies the memory
3200address from which to load. The pointer must point to a <a
3201 href="#t_firstclass">first class</a> type. If the <tt>load</tt> is
3202marked as <tt>volatile</tt>, then the optimizer is not allowed to modify
3203the number or order of execution of this <tt>load</tt> with other
3204volatile <tt>load</tt> and <tt><a href="#i_store">store</a></tt>
3205instructions. </p>
Chris Lattner21003c82008-01-06 21:04:43 +0000[diff] [blame]3206<p>
Chris Lattner10368b62008-04-02 00:38:26 +0000[diff] [blame]3207The optional constant "align" argument specifies the alignment of the operation
Chris Lattner21003c82008-01-06 21:04:43 +0000[diff] [blame]3208(that is, the alignment of the memory address). A value of 0 or an
3209omitted "align" argument means that the operation has the preferential
3210alignment for the target. It is the responsibility of the code emitter
3211to ensure that the alignment information is correct. Overestimating
3212the alignment results in an undefined behavior. Underestimating the
3213alignment may produce less efficient code. An alignment of 1 is always
3214safe.
3215</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3216<h5>Semantics:</h5>
3217<p>The location of memory pointed to is loaded.</p>
3218<h5>Examples:</h5>
3219<pre> %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
3220 <a
3221 href="#i_store">store</a> i32 3, i32* %ptr <i>; yields {void}</i>
3222 %val = load i32* %ptr <i>; yields {i32}:val = i32 3</i>
3223</pre>
3224</div>
3225<!-- _______________________________________________________________________ -->
3226<div class="doc_subsubsection"> <a name="i_store">'<tt>store</tt>'
3227Instruction</a> </div>
3228<div class="doc_text">
3229<h5>Syntax:</h5>
3230<pre> store &lt;ty&gt; &lt;value&gt;, &lt;ty&gt;* &lt;pointer&gt;[, align &lt;alignment&gt;] <i>; yields {void}</i>
3231 volatile store &lt;ty&gt; &lt;value&gt;, &lt;ty&gt;* &lt;pointer&gt;[, align &lt;alignment&gt;] <i>; yields {void}</i>
3232</pre>
3233<h5>Overview:</h5>
3234<p>The '<tt>store</tt>' instruction is used to write to memory.</p>
3235<h5>Arguments:</h5>
3236<p>There are two arguments to the '<tt>store</tt>' instruction: a value
3237to 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]3238operand must be a pointer to the <a href="#t_firstclass">first class</a> type
3239of the '<tt>&lt;value&gt;</tt>'
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3240operand. If the <tt>store</tt> is marked as <tt>volatile</tt>, then the
3241optimizer is not allowed to modify the number or order of execution of
3242this <tt>store</tt> with other volatile <tt>load</tt> and <tt><a
3243 href="#i_store">store</a></tt> instructions.</p>
Chris Lattner21003c82008-01-06 21:04:43 +0000[diff] [blame]3244<p>
Chris Lattner10368b62008-04-02 00:38:26 +0000[diff] [blame]3245The optional constant "align" argument specifies the alignment of the operation
Chris Lattner21003c82008-01-06 21:04:43 +0000[diff] [blame]3246(that is, the alignment of the memory address). A value of 0 or an
3247omitted "align" argument means that the operation has the preferential
3248alignment for the target. It is the responsibility of the code emitter
3249to ensure that the alignment information is correct. Overestimating
3250the alignment results in an undefined behavior. Underestimating the
3251alignment may produce less efficient code. An alignment of 1 is always
3252safe.
3253</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3254<h5>Semantics:</h5>
3255<p>The contents of memory are updated to contain '<tt>&lt;value&gt;</tt>'
3256at the location specified by the '<tt>&lt;pointer&gt;</tt>' operand.</p>
3257<h5>Example:</h5>
3258<pre> %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
Bill Wendling63ffa142007-10-22 05:10:05 +0000[diff] [blame]3259 store i32 3, i32* %ptr <i>; yields {void}</i>
3260 %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]3261</pre>
3262</div>
3263
3264<!-- _______________________________________________________________________ -->
3265<div class="doc_subsubsection">
3266 <a name="i_getelementptr">'<tt>getelementptr</tt>' Instruction</a>
3267</div>
3268
3269<div class="doc_text">
3270<h5>Syntax:</h5>
3271<pre>
3272 &lt;result&gt; = getelementptr &lt;ty&gt;* &lt;ptrval&gt;{, &lt;ty&gt; &lt;idx&gt;}*
3273</pre>
3274
3275<h5>Overview:</h5>
3276
3277<p>
3278The '<tt>getelementptr</tt>' instruction is used to get the address of a
3279subelement of an aggregate data structure.</p>
3280
3281<h5>Arguments:</h5>
3282
3283<p>This instruction takes a list of integer operands that indicate what
3284elements of the aggregate object to index to. The actual types of the arguments
3285provided depend on the type of the first pointer argument. The
3286'<tt>getelementptr</tt>' instruction is used to index down through the type
3287levels of a structure or to a specific index in an array. When indexing into a
3288structure, only <tt>i32</tt> integer constants are allowed. When indexing
Chris Lattnera7d94ba2008-04-24 05:59:56 +0000[diff] [blame]3289into an array or pointer, only integers of 32 or 64 bits are allowed; 32-bit
3290values will be sign extended to 64-bits if required.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3291
3292<p>For example, let's consider a C code fragment and how it gets
3293compiled to LLVM:</p>
3294
3295<div class="doc_code">
3296<pre>
3297struct RT {
3298 char A;
3299 int B[10][20];
3300 char C;
3301};
3302struct ST {
3303 int X;
3304 double Y;
3305 struct RT Z;
3306};
3307
3308int *foo(struct ST *s) {
3309 return &amp;s[1].Z.B[5][13];
3310}
3311</pre>
3312</div>
3313
3314<p>The LLVM code generated by the GCC frontend is:</p>
3315
3316<div class="doc_code">
3317<pre>
3318%RT = type { i8 , [10 x [20 x i32]], i8 }
3319%ST = type { i32, double, %RT }
3320
3321define i32* %foo(%ST* %s) {
3322entry:
3323 %reg = getelementptr %ST* %s, i32 1, i32 2, i32 1, i32 5, i32 13
3324 ret i32* %reg
3325}
3326</pre>
3327</div>
3328
3329<h5>Semantics:</h5>
3330
3331<p>The index types specified for the '<tt>getelementptr</tt>' instruction depend
3332on the pointer type that is being indexed into. <a href="#t_pointer">Pointer</a>
3333and <a href="#t_array">array</a> types can use a 32-bit or 64-bit
3334<a href="#t_integer">integer</a> type but the value will always be sign extended
Chris Lattner10368b62008-04-02 00:38:26 +0000[diff] [blame]3335to 64-bits. <a href="#t_struct">Structure</a> and <a href="#t_pstruct">packed
3336structure</a> types require <tt>i32</tt> <b>constants</b>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3337
3338<p>In the example above, the first index is indexing into the '<tt>%ST*</tt>'
3339type, which is a pointer, yielding a '<tt>%ST</tt>' = '<tt>{ i32, double, %RT
3340}</tt>' type, a structure. The second index indexes into the third element of
3341the structure, yielding a '<tt>%RT</tt>' = '<tt>{ i8 , [10 x [20 x i32]],
3342i8 }</tt>' type, another structure. The third index indexes into the second
3343element of the structure, yielding a '<tt>[10 x [20 x i32]]</tt>' type, an
3344array. The two dimensions of the array are subscripted into, yielding an
3345'<tt>i32</tt>' type. The '<tt>getelementptr</tt>' instruction returns a pointer
3346to this element, thus computing a value of '<tt>i32*</tt>' type.</p>
3347
3348<p>Note that it is perfectly legal to index partially through a
3349structure, returning a pointer to an inner element. Because of this,
3350the LLVM code for the given testcase is equivalent to:</p>
3351
3352<pre>
3353 define i32* %foo(%ST* %s) {
3354 %t1 = getelementptr %ST* %s, i32 1 <i>; yields %ST*:%t1</i>
3355 %t2 = getelementptr %ST* %t1, i32 0, i32 2 <i>; yields %RT*:%t2</i>
3356 %t3 = getelementptr %RT* %t2, i32 0, i32 1 <i>; yields [10 x [20 x i32]]*:%t3</i>
3357 %t4 = getelementptr [10 x [20 x i32]]* %t3, i32 0, i32 5 <i>; yields [20 x i32]*:%t4</i>
3358 %t5 = getelementptr [20 x i32]* %t4, i32 0, i32 13 <i>; yields i32*:%t5</i>
3359 ret i32* %t5
3360 }
3361</pre>
3362
3363<p>Note that it is undefined to access an array out of bounds: array and
3364pointer indexes must always be within the defined bounds of the array type.
Chris Lattnera7d94ba2008-04-24 05:59:56 +0000[diff] [blame]3365The one exception for this rule is zero length arrays. These arrays are
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3366defined to be accessible as variable length arrays, which requires access
3367beyond the zero'th element.</p>
3368
3369<p>The getelementptr instruction is often confusing. For some more insight
3370into how it works, see <a href="GetElementPtr.html">the getelementptr
3371FAQ</a>.</p>
3372
3373<h5>Example:</h5>
3374
3375<pre>
3376 <i>; yields [12 x i8]*:aptr</i>
3377 %aptr = getelementptr {i32, [12 x i8]}* %sptr, i64 0, i32 1
3378</pre>
3379</div>
3380
3381<!-- ======================================================================= -->
3382<div class="doc_subsection"> <a name="convertops">Conversion Operations</a>
3383</div>
3384<div class="doc_text">
3385<p>The instructions in this category are the conversion instructions (casting)
3386which all take a single operand and a type. They perform various bit conversions
3387on the operand.</p>
3388</div>
3389
3390<!-- _______________________________________________________________________ -->
3391<div class="doc_subsubsection">
3392 <a name="i_trunc">'<tt>trunc .. to</tt>' Instruction</a>
3393</div>
3394<div class="doc_text">
3395
3396<h5>Syntax:</h5>
3397<pre>
3398 &lt;result&gt; = trunc &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3399</pre>
3400
3401<h5>Overview:</h5>
3402<p>
3403The '<tt>trunc</tt>' instruction truncates its operand to the type <tt>ty2</tt>.
3404</p>
3405
3406<h5>Arguments:</h5>
3407<p>
3408The '<tt>trunc</tt>' instruction takes a <tt>value</tt> to trunc, which must
3409be an <a href="#t_integer">integer</a> type, and a type that specifies the size
3410and type of the result, which must be an <a href="#t_integer">integer</a>
3411type. The bit size of <tt>value</tt> must be larger than the bit size of
3412<tt>ty2</tt>. Equal sized types are not allowed.</p>
3413
3414<h5>Semantics:</h5>
3415<p>
3416The '<tt>trunc</tt>' instruction truncates the high order bits in <tt>value</tt>
3417and converts the remaining bits to <tt>ty2</tt>. Since the source size must be
3418larger than the destination size, <tt>trunc</tt> cannot be a <i>no-op cast</i>.
3419It will always truncate bits.</p>
3420
3421<h5>Example:</h5>
3422<pre>
3423 %X = trunc i32 257 to i8 <i>; yields i8:1</i>
3424 %Y = trunc i32 123 to i1 <i>; yields i1:true</i>
3425 %Y = trunc i32 122 to i1 <i>; yields i1:false</i>
3426</pre>
3427</div>
3428
3429<!-- _______________________________________________________________________ -->
3430<div class="doc_subsubsection">
3431 <a name="i_zext">'<tt>zext .. to</tt>' Instruction</a>
3432</div>
3433<div class="doc_text">
3434
3435<h5>Syntax:</h5>
3436<pre>
3437 &lt;result&gt; = zext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3438</pre>
3439
3440<h5>Overview:</h5>
3441<p>The '<tt>zext</tt>' instruction zero extends its operand to type
3442<tt>ty2</tt>.</p>
3443
3444
3445<h5>Arguments:</h5>
3446<p>The '<tt>zext</tt>' instruction takes a value to cast, which must be of
3447<a href="#t_integer">integer</a> type, and a type to cast it to, which must
3448also be of <a href="#t_integer">integer</a> type. The bit size of the
3449<tt>value</tt> must be smaller than the bit size of the destination type,
3450<tt>ty2</tt>.</p>
3451
3452<h5>Semantics:</h5>
3453<p>The <tt>zext</tt> fills the high order bits of the <tt>value</tt> with zero
3454bits until it reaches the size of the destination type, <tt>ty2</tt>.</p>
3455
3456<p>When zero extending from i1, the result will always be either 0 or 1.</p>
3457
3458<h5>Example:</h5>
3459<pre>
3460 %X = zext i32 257 to i64 <i>; yields i64:257</i>
3461 %Y = zext i1 true to i32 <i>; yields i32:1</i>
3462</pre>
3463</div>
3464
3465<!-- _______________________________________________________________________ -->
3466<div class="doc_subsubsection">
3467 <a name="i_sext">'<tt>sext .. to</tt>' Instruction</a>
3468</div>
3469<div class="doc_text">
3470
3471<h5>Syntax:</h5>
3472<pre>
3473 &lt;result&gt; = sext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3474</pre>
3475
3476<h5>Overview:</h5>
3477<p>The '<tt>sext</tt>' sign extends <tt>value</tt> to the type <tt>ty2</tt>.</p>
3478
3479<h5>Arguments:</h5>
3480<p>
3481The '<tt>sext</tt>' instruction takes a value to cast, which must be of
3482<a href="#t_integer">integer</a> type, and a type to cast it to, which must
3483also be of <a href="#t_integer">integer</a> type. The bit size of the
3484<tt>value</tt> must be smaller than the bit size of the destination type,
3485<tt>ty2</tt>.</p>
3486
3487<h5>Semantics:</h5>
3488<p>
3489The '<tt>sext</tt>' instruction performs a sign extension by copying the sign
3490bit (highest order bit) of the <tt>value</tt> until it reaches the bit size of
3491the type <tt>ty2</tt>.</p>
3492
3493<p>When sign extending from i1, the extension always results in -1 or 0.</p>
3494
3495<h5>Example:</h5>
3496<pre>
3497 %X = sext i8 -1 to i16 <i>; yields i16 :65535</i>
3498 %Y = sext i1 true to i32 <i>; yields i32:-1</i>
3499</pre>
3500</div>
3501
3502<!-- _______________________________________________________________________ -->
3503<div class="doc_subsubsection">
3504 <a name="i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a>
3505</div>
3506
3507<div class="doc_text">
3508
3509<h5>Syntax:</h5>
3510
3511<pre>
3512 &lt;result&gt; = fptrunc &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3513</pre>
3514
3515<h5>Overview:</h5>
3516<p>The '<tt>fptrunc</tt>' instruction truncates <tt>value</tt> to type
3517<tt>ty2</tt>.</p>
3518
3519
3520<h5>Arguments:</h5>
3521<p>The '<tt>fptrunc</tt>' instruction takes a <a href="#t_floating">floating
3522 point</a> value to cast and a <a href="#t_floating">floating point</a> type to
3523cast it to. The size of <tt>value</tt> must be larger than the size of
3524<tt>ty2</tt>. This implies that <tt>fptrunc</tt> cannot be used to make a
3525<i>no-op cast</i>.</p>
3526
3527<h5>Semantics:</h5>
3528<p> The '<tt>fptrunc</tt>' instruction truncates a <tt>value</tt> from a larger
3529<a href="#t_floating">floating point</a> type to a smaller
3530<a href="#t_floating">floating point</a> type. If the value cannot fit within
3531the destination type, <tt>ty2</tt>, then the results are undefined.</p>
3532
3533<h5>Example:</h5>
3534<pre>
3535 %X = fptrunc double 123.0 to float <i>; yields float:123.0</i>
3536 %Y = fptrunc double 1.0E+300 to float <i>; yields undefined</i>
3537</pre>
3538</div>
3539
3540<!-- _______________________________________________________________________ -->
3541<div class="doc_subsubsection">
3542 <a name="i_fpext">'<tt>fpext .. to</tt>' Instruction</a>
3543</div>
3544<div class="doc_text">
3545
3546<h5>Syntax:</h5>
3547<pre>
3548 &lt;result&gt; = fpext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3549</pre>
3550
3551<h5>Overview:</h5>
3552<p>The '<tt>fpext</tt>' extends a floating point <tt>value</tt> to a larger
3553floating point value.</p>
3554
3555<h5>Arguments:</h5>
3556<p>The '<tt>fpext</tt>' instruction takes a
3557<a href="#t_floating">floating point</a> <tt>value</tt> to cast,
3558and a <a href="#t_floating">floating point</a> type to cast it to. The source
3559type must be smaller than the destination type.</p>
3560
3561<h5>Semantics:</h5>
3562<p>The '<tt>fpext</tt>' instruction extends the <tt>value</tt> from a smaller
3563<a href="#t_floating">floating point</a> type to a larger
3564<a href="#t_floating">floating point</a> type. The <tt>fpext</tt> cannot be
3565used to make a <i>no-op cast</i> because it always changes bits. Use
3566<tt>bitcast</tt> to make a <i>no-op cast</i> for a floating point cast.</p>
3567
3568<h5>Example:</h5>
3569<pre>
3570 %X = fpext float 3.1415 to double <i>; yields double:3.1415</i>
3571 %Y = fpext float 1.0 to float <i>; yields float:1.0 (no-op)</i>
3572</pre>
3573</div>
3574
3575<!-- _______________________________________________________________________ -->
3576<div class="doc_subsubsection">
3577 <a name="i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a>
3578</div>
3579<div class="doc_text">
3580
3581<h5>Syntax:</h5>
3582<pre>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3583 &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]3584</pre>
3585
3586<h5>Overview:</h5>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3587<p>The '<tt>fptoui</tt>' converts a floating point <tt>value</tt> to its
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3588unsigned integer equivalent of type <tt>ty2</tt>.
3589</p>
3590
3591<h5>Arguments:</h5>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3592<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]3593scalar or vector <a href="#t_floating">floating point</a> value, and a type
3594to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
3595type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
3596vector integer type with the same number of elements as <tt>ty</tt></p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3597
3598<h5>Semantics:</h5>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3599<p> The '<tt>fptoui</tt>' instruction converts its
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3600<a href="#t_floating">floating point</a> operand into the nearest (rounding
3601towards zero) unsigned integer value. If the value cannot fit in <tt>ty2</tt>,
3602the results are undefined.</p>
3603
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3604<h5>Example:</h5>
3605<pre>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3606 %X = fptoui double 123.0 to i32 <i>; yields i32:123</i>
Chris Lattner681f1e82007-09-22 03:17:52 +0000[diff] [blame]3607 %Y = fptoui float 1.0E+300 to i1 <i>; yields undefined:1</i>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3608 %X = fptoui float 1.04E+17 to i8 <i>; yields undefined:1</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3609</pre>
3610</div>
3611
3612<!-- _______________________________________________________________________ -->
3613<div class="doc_subsubsection">
3614 <a name="i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a>
3615</div>
3616<div class="doc_text">
3617
3618<h5>Syntax:</h5>
3619<pre>
3620 &lt;result&gt; = fptosi &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3621</pre>
3622
3623<h5>Overview:</h5>
3624<p>The '<tt>fptosi</tt>' instruction converts
3625<a href="#t_floating">floating point</a> <tt>value</tt> to type <tt>ty2</tt>.
3626</p>
3627
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3628<h5>Arguments:</h5>
3629<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]3630scalar or vector <a href="#t_floating">floating point</a> value, and a type
3631to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
3632type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
3633vector integer type with the same number of elements as <tt>ty</tt></p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3634
3635<h5>Semantics:</h5>
3636<p>The '<tt>fptosi</tt>' instruction converts its
3637<a href="#t_floating">floating point</a> operand into the nearest (rounding
3638towards zero) signed integer value. If the value cannot fit in <tt>ty2</tt>,
3639the results are undefined.</p>
3640
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3641<h5>Example:</h5>
3642<pre>
3643 %X = fptosi double -123.0 to i32 <i>; yields i32:-123</i>
Chris Lattner681f1e82007-09-22 03:17:52 +0000[diff] [blame]3644 %Y = fptosi float 1.0E-247 to i1 <i>; yields undefined:1</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3645 %X = fptosi float 1.04E+17 to i8 <i>; yields undefined:1</i>
3646</pre>
3647</div>
3648
3649<!-- _______________________________________________________________________ -->
3650<div class="doc_subsubsection">
3651 <a name="i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a>
3652</div>
3653<div class="doc_text">
3654
3655<h5>Syntax:</h5>
3656<pre>
3657 &lt;result&gt; = uitofp &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3658</pre>
3659
3660<h5>Overview:</h5>
3661<p>The '<tt>uitofp</tt>' instruction regards <tt>value</tt> as an unsigned
3662integer and converts that value to the <tt>ty2</tt> type.</p>
3663
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3664<h5>Arguments:</h5>
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]3665<p>The '<tt>uitofp</tt>' instruction takes a value to cast, which must be a
3666scalar or vector <a href="#t_integer">integer</a> value, and a type to cast it
3667to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
3668type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
3669floating point type with the same number of elements as <tt>ty</tt></p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3670
3671<h5>Semantics:</h5>
3672<p>The '<tt>uitofp</tt>' instruction interprets its operand as an unsigned
3673integer quantity and converts it to the corresponding floating point value. If
3674the value cannot fit in the floating point value, the results are undefined.</p>
3675
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3676<h5>Example:</h5>
3677<pre>
3678 %X = uitofp i32 257 to float <i>; yields float:257.0</i>
3679 %Y = uitofp i8 -1 to double <i>; yields double:255.0</i>
3680</pre>
3681</div>
3682
3683<!-- _______________________________________________________________________ -->
3684<div class="doc_subsubsection">
3685 <a name="i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a>
3686</div>
3687<div class="doc_text">
3688
3689<h5>Syntax:</h5>
3690<pre>
3691 &lt;result&gt; = sitofp &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3692</pre>
3693
3694<h5>Overview:</h5>
3695<p>The '<tt>sitofp</tt>' instruction regards <tt>value</tt> as a signed
3696integer and converts that value to the <tt>ty2</tt> type.</p>
3697
3698<h5>Arguments:</h5>
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]3699<p>The '<tt>sitofp</tt>' instruction takes a value to cast, which must be a
3700scalar or vector <a href="#t_integer">integer</a> value, and a type to cast it
3701to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
3702type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
3703floating point type with the same number of elements as <tt>ty</tt></p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3704
3705<h5>Semantics:</h5>
3706<p>The '<tt>sitofp</tt>' instruction interprets its operand as a signed
3707integer quantity and converts it to the corresponding floating point value. If
3708the value cannot fit in the floating point value, the results are undefined.</p>
3709
3710<h5>Example:</h5>
3711<pre>
3712 %X = sitofp i32 257 to float <i>; yields float:257.0</i>
3713 %Y = sitofp i8 -1 to double <i>; yields double:-1.0</i>
3714</pre>
3715</div>
3716
3717<!-- _______________________________________________________________________ -->
3718<div class="doc_subsubsection">
3719 <a name="i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a>
3720</div>
3721<div class="doc_text">
3722
3723<h5>Syntax:</h5>
3724<pre>
3725 &lt;result&gt; = ptrtoint &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3726</pre>
3727
3728<h5>Overview:</h5>
3729<p>The '<tt>ptrtoint</tt>' instruction converts the pointer <tt>value</tt> to
3730the integer type <tt>ty2</tt>.</p>
3731
3732<h5>Arguments:</h5>
3733<p>The '<tt>ptrtoint</tt>' instruction takes a <tt>value</tt> to cast, which
3734must be a <a href="#t_pointer">pointer</a> value, and a type to cast it to
3735<tt>ty2</tt>, which must be an <a href="#t_integer">integer</a> type.
3736
3737<h5>Semantics:</h5>
3738<p>The '<tt>ptrtoint</tt>' instruction converts <tt>value</tt> to integer type
3739<tt>ty2</tt> by interpreting the pointer value as an integer and either
3740truncating or zero extending that value to the size of the integer type. If
3741<tt>value</tt> is smaller than <tt>ty2</tt> then a zero extension is done. If
3742<tt>value</tt> is larger than <tt>ty2</tt> then a truncation is done. If they
3743are the same size, then nothing is done (<i>no-op cast</i>) other than a type
3744change.</p>
3745
3746<h5>Example:</h5>
3747<pre>
3748 %X = ptrtoint i32* %X to i8 <i>; yields truncation on 32-bit architecture</i>
3749 %Y = ptrtoint i32* %x to i64 <i>; yields zero extension on 32-bit architecture</i>
3750</pre>
3751</div>
3752
3753<!-- _______________________________________________________________________ -->
3754<div class="doc_subsubsection">
3755 <a name="i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a>
3756</div>
3757<div class="doc_text">
3758
3759<h5>Syntax:</h5>
3760<pre>
3761 &lt;result&gt; = inttoptr &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3762</pre>
3763
3764<h5>Overview:</h5>
3765<p>The '<tt>inttoptr</tt>' instruction converts an integer <tt>value</tt> to
3766a pointer type, <tt>ty2</tt>.</p>
3767
3768<h5>Arguments:</h5>
3769<p>The '<tt>inttoptr</tt>' instruction takes an <a href="#t_integer">integer</a>
3770value to cast, and a type to cast it to, which must be a
3771<a href="#t_pointer">pointer</a> type.
3772
3773<h5>Semantics:</h5>
3774<p>The '<tt>inttoptr</tt>' instruction converts <tt>value</tt> to type
3775<tt>ty2</tt> by applying either a zero extension or a truncation depending on
3776the size of the integer <tt>value</tt>. If <tt>value</tt> is larger than the
3777size of a pointer then a truncation is done. If <tt>value</tt> is smaller than
3778the size of a pointer then a zero extension is done. If they are the same size,
3779nothing is done (<i>no-op cast</i>).</p>
3780
3781<h5>Example:</h5>
3782<pre>
3783 %X = inttoptr i32 255 to i32* <i>; yields zero extension on 64-bit architecture</i>
3784 %X = inttoptr i32 255 to i32* <i>; yields no-op on 32-bit architecture</i>
3785 %Y = inttoptr i64 0 to i32* <i>; yields truncation on 32-bit architecture</i>
3786</pre>
3787</div>
3788
3789<!-- _______________________________________________________________________ -->
3790<div class="doc_subsubsection">
3791 <a name="i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a>
3792</div>
3793<div class="doc_text">
3794
3795<h5>Syntax:</h5>
3796<pre>
3797 &lt;result&gt; = bitcast &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3798</pre>
3799
3800<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3801
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3802<p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
3803<tt>ty2</tt> without changing any bits.</p>
3804
3805<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3806
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3807<p>The '<tt>bitcast</tt>' instruction takes a value to cast, which must be
3808a first class value, and a type to cast it to, which must also be a <a
3809 href="#t_firstclass">first class</a> type. The bit sizes of <tt>value</tt>
3810and the destination type, <tt>ty2</tt>, must be identical. If the source
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3811type is a pointer, the destination type must also be a pointer. This
3812instruction supports bitwise conversion of vectors to integers and to vectors
3813of other types (as long as they have the same size).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3814
3815<h5>Semantics:</h5>
3816<p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
3817<tt>ty2</tt>. It is always a <i>no-op cast</i> because no bits change with
3818this conversion. The conversion is done as if the <tt>value</tt> had been
3819stored to memory and read back as type <tt>ty2</tt>. Pointer types may only be
3820converted to other pointer types with this instruction. To convert pointers to
3821other types, use the <a href="#i_inttoptr">inttoptr</a> or
3822<a href="#i_ptrtoint">ptrtoint</a> instructions first.</p>
3823
3824<h5>Example:</h5>
3825<pre>
3826 %X = bitcast i8 255 to i8 <i>; yields i8 :-1</i>
3827 %Y = bitcast i32* %x to sint* <i>; yields sint*:%x</i>
3828 %Z = bitcast <2xint> %V to i64; <i>; yields i64: %V</i>
3829</pre>
3830</div>
3831
3832<!-- ======================================================================= -->
3833<div class="doc_subsection"> <a name="otherops">Other Operations</a> </div>
3834<div class="doc_text">
3835<p>The instructions in this category are the "miscellaneous"
3836instructions, which defy better classification.</p>
3837</div>
3838
3839<!-- _______________________________________________________________________ -->
3840<div class="doc_subsubsection"><a name="i_icmp">'<tt>icmp</tt>' Instruction</a>
3841</div>
3842<div class="doc_text">
3843<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3844<pre> &lt;result&gt; = icmp &lt;cond&gt; &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {i1}:result</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3845</pre>
3846<h5>Overview:</h5>
3847<p>The '<tt>icmp</tt>' instruction returns a boolean value based on comparison
Chris Lattner10368b62008-04-02 00:38:26 +0000[diff] [blame]3848of its two integer or pointer operands.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3849<h5>Arguments:</h5>
3850<p>The '<tt>icmp</tt>' instruction takes three operands. The first operand is
3851the condition code indicating the kind of comparison to perform. It is not
3852a value, just a keyword. The possible condition code are:
3853<ol>
3854 <li><tt>eq</tt>: equal</li>
3855 <li><tt>ne</tt>: not equal </li>
3856 <li><tt>ugt</tt>: unsigned greater than</li>
3857 <li><tt>uge</tt>: unsigned greater or equal</li>
3858 <li><tt>ult</tt>: unsigned less than</li>
3859 <li><tt>ule</tt>: unsigned less or equal</li>
3860 <li><tt>sgt</tt>: signed greater than</li>
3861 <li><tt>sge</tt>: signed greater or equal</li>
3862 <li><tt>slt</tt>: signed less than</li>
3863 <li><tt>sle</tt>: signed less or equal</li>
3864</ol>
3865<p>The remaining two arguments must be <a href="#t_integer">integer</a> or
3866<a href="#t_pointer">pointer</a> typed. They must also be identical types.</p>
3867<h5>Semantics:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3868<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]3869the condition code given as <tt>cond</tt>. The comparison performed always
3870yields a <a href="#t_primitive">i1</a> result, as follows:
3871<ol>
3872 <li><tt>eq</tt>: yields <tt>true</tt> if the operands are equal,
3873 <tt>false</tt> otherwise. No sign interpretation is necessary or performed.
3874 </li>
3875 <li><tt>ne</tt>: yields <tt>true</tt> if the operands are unequal,
3876 <tt>false</tt> otherwise. No sign interpretation is necessary or performed.
3877 <li><tt>ugt</tt>: interprets the operands as unsigned values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3878 <tt>true</tt> if <tt>op1</tt> is greater than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3879 <li><tt>uge</tt>: interprets the operands as unsigned values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3880 <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]3881 <li><tt>ult</tt>: interprets the operands as unsigned values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3882 <tt>true</tt> if <tt>op1</tt> is less than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3883 <li><tt>ule</tt>: interprets the operands as unsigned values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3884 <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]3885 <li><tt>sgt</tt>: interprets the operands as signed values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3886 <tt>true</tt> if <tt>op1</tt> is greater than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3887 <li><tt>sge</tt>: interprets the operands as signed values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3888 <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]3889 <li><tt>slt</tt>: interprets the operands as signed values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3890 <tt>true</tt> if <tt>op1</tt> is less than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3891 <li><tt>sle</tt>: interprets the operands as signed values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3892 <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]3893</ol>
3894<p>If the operands are <a href="#t_pointer">pointer</a> typed, the pointer
3895values are compared as if they were integers.</p>
3896
3897<h5>Example:</h5>
3898<pre> &lt;result&gt; = icmp eq i32 4, 5 <i>; yields: result=false</i>
3899 &lt;result&gt; = icmp ne float* %X, %X <i>; yields: result=false</i>
3900 &lt;result&gt; = icmp ult i16 4, 5 <i>; yields: result=true</i>
3901 &lt;result&gt; = icmp sgt i16 4, 5 <i>; yields: result=false</i>
3902 &lt;result&gt; = icmp ule i16 -4, 5 <i>; yields: result=false</i>
3903 &lt;result&gt; = icmp sge i16 4, 5 <i>; yields: result=false</i>
3904</pre>
3905</div>
3906
3907<!-- _______________________________________________________________________ -->
3908<div class="doc_subsubsection"><a name="i_fcmp">'<tt>fcmp</tt>' Instruction</a>
3909</div>
3910<div class="doc_text">
3911<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3912<pre> &lt;result&gt; = fcmp &lt;cond&gt; &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {i1}:result</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3913</pre>
3914<h5>Overview:</h5>
3915<p>The '<tt>fcmp</tt>' instruction returns a boolean value based on comparison
3916of its floating point operands.</p>
3917<h5>Arguments:</h5>
3918<p>The '<tt>fcmp</tt>' instruction takes three operands. The first operand is
3919the condition code indicating the kind of comparison to perform. It is not
3920a value, just a keyword. The possible condition code are:
3921<ol>
3922 <li><tt>false</tt>: no comparison, always returns false</li>
3923 <li><tt>oeq</tt>: ordered and equal</li>
3924 <li><tt>ogt</tt>: ordered and greater than </li>
3925 <li><tt>oge</tt>: ordered and greater than or equal</li>
3926 <li><tt>olt</tt>: ordered and less than </li>
3927 <li><tt>ole</tt>: ordered and less than or equal</li>
3928 <li><tt>one</tt>: ordered and not equal</li>
3929 <li><tt>ord</tt>: ordered (no nans)</li>
3930 <li><tt>ueq</tt>: unordered or equal</li>
3931 <li><tt>ugt</tt>: unordered or greater than </li>
3932 <li><tt>uge</tt>: unordered or greater than or equal</li>
3933 <li><tt>ult</tt>: unordered or less than </li>
3934 <li><tt>ule</tt>: unordered or less than or equal</li>
3935 <li><tt>une</tt>: unordered or not equal</li>
3936 <li><tt>uno</tt>: unordered (either nans)</li>
3937 <li><tt>true</tt>: no comparison, always returns true</li>
3938</ol>
3939<p><i>Ordered</i> means that neither operand is a QNAN while
3940<i>unordered</i> means that either operand may be a QNAN.</p>
3941<p>The <tt>val1</tt> and <tt>val2</tt> arguments must be
3942<a href="#t_floating">floating point</a> typed. They must have identical
3943types.</p>
3944<h5>Semantics:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3945<p>The '<tt>fcmp</tt>' instruction compares <tt>op1</tt> and <tt>op2</tt>
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]3946according to the condition code given as <tt>cond</tt>. The comparison performed
3947always yields a <a href="#t_primitive">i1</a> result, as follows:
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3948<ol>
3949 <li><tt>false</tt>: always yields <tt>false</tt>, regardless of operands.</li>
3950 <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]3951 <tt>op1</tt> is equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3952 <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]3953 <tt>op1</tt> is greather than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3954 <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]3955 <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3956 <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]3957 <tt>op1</tt> is less than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3958 <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]3959 <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3960 <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]3961 <tt>op1</tt> is not equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3962 <li><tt>ord</tt>: yields <tt>true</tt> if both operands are not a QNAN.</li>
3963 <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]3964 <tt>op1</tt> is equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3965 <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]3966 <tt>op1</tt> is greater than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3967 <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]3968 <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3969 <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]3970 <tt>op1</tt> is less than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3971 <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]3972 <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3973 <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]3974 <tt>op1</tt> is not equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3975 <li><tt>uno</tt>: yields <tt>true</tt> if either operand is a QNAN.</li>
3976 <li><tt>true</tt>: always yields <tt>true</tt>, regardless of operands.</li>
3977</ol>
3978
3979<h5>Example:</h5>
3980<pre> &lt;result&gt; = fcmp oeq float 4.0, 5.0 <i>; yields: result=false</i>
3981 &lt;result&gt; = icmp one float 4.0, 5.0 <i>; yields: result=true</i>
3982 &lt;result&gt; = icmp olt float 4.0, 5.0 <i>; yields: result=true</i>
3983 &lt;result&gt; = icmp ueq double 1.0, 2.0 <i>; yields: result=false</i>
3984</pre>
3985</div>
3986
3987<!-- _______________________________________________________________________ -->
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]3988<div class="doc_subsubsection">
3989 <a name="i_vicmp">'<tt>vicmp</tt>' Instruction</a>
3990</div>
3991<div class="doc_text">
3992<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3993<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]3994</pre>
3995<h5>Overview:</h5>
3996<p>The '<tt>vicmp</tt>' instruction returns an integer vector value based on
3997element-wise comparison of its two integer vector operands.</p>
3998<h5>Arguments:</h5>
3999<p>The '<tt>vicmp</tt>' instruction takes three operands. The first operand is
4000the condition code indicating the kind of comparison to perform. It is not
4001a value, just a keyword. The possible condition code are:
4002<ol>
4003 <li><tt>eq</tt>: equal</li>
4004 <li><tt>ne</tt>: not equal </li>
4005 <li><tt>ugt</tt>: unsigned greater than</li>
4006 <li><tt>uge</tt>: unsigned greater or equal</li>
4007 <li><tt>ult</tt>: unsigned less than</li>
4008 <li><tt>ule</tt>: unsigned less or equal</li>
4009 <li><tt>sgt</tt>: signed greater than</li>
4010 <li><tt>sge</tt>: signed greater or equal</li>
4011 <li><tt>slt</tt>: signed less than</li>
4012 <li><tt>sle</tt>: signed less or equal</li>
4013</ol>
4014<p>The remaining two arguments must be <a href="#t_vector">vector</a> of
4015<a href="#t_integer">integer</a> typed. They must also be identical types.</p>
4016<h5>Semantics:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4017<p>The '<tt>vicmp</tt>' instruction compares <tt>op1</tt> and <tt>op2</tt>
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]4018according to the condition code given as <tt>cond</tt>. The comparison yields a
4019<a href="#t_vector">vector</a> of <a href="#t_integer">integer</a> result, of
4020identical type as the values being compared. The most significant bit in each
4021element is 1 if the element-wise comparison evaluates to true, and is 0
4022otherwise. All other bits of the result are undefined. The condition codes
4023are evaluated identically to the <a href="#i_icmp">'<tt>icmp</tt>'
4024instruction</a>.
4025
4026<h5>Example:</h5>
4027<pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4028 &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>
4029 &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]4030</pre>
4031</div>
4032
4033<!-- _______________________________________________________________________ -->
4034<div class="doc_subsubsection">
4035 <a name="i_vfcmp">'<tt>vfcmp</tt>' Instruction</a>
4036</div>
4037<div class="doc_text">
4038<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4039<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]4040<h5>Overview:</h5>
4041<p>The '<tt>vfcmp</tt>' instruction returns an integer vector value based on
4042element-wise comparison of its two floating point vector operands. The output
4043elements have the same width as the input elements.</p>
4044<h5>Arguments:</h5>
4045<p>The '<tt>vfcmp</tt>' instruction takes three operands. The first operand is
4046the condition code indicating the kind of comparison to perform. It is not
4047a value, just a keyword. The possible condition code are:
4048<ol>
4049 <li><tt>false</tt>: no comparison, always returns false</li>
4050 <li><tt>oeq</tt>: ordered and equal</li>
4051 <li><tt>ogt</tt>: ordered and greater than </li>
4052 <li><tt>oge</tt>: ordered and greater than or equal</li>
4053 <li><tt>olt</tt>: ordered and less than </li>
4054 <li><tt>ole</tt>: ordered and less than or equal</li>
4055 <li><tt>one</tt>: ordered and not equal</li>
4056 <li><tt>ord</tt>: ordered (no nans)</li>
4057 <li><tt>ueq</tt>: unordered or equal</li>
4058 <li><tt>ugt</tt>: unordered or greater than </li>
4059 <li><tt>uge</tt>: unordered or greater than or equal</li>
4060 <li><tt>ult</tt>: unordered or less than </li>
4061 <li><tt>ule</tt>: unordered or less than or equal</li>
4062 <li><tt>une</tt>: unordered or not equal</li>
4063 <li><tt>uno</tt>: unordered (either nans)</li>
4064 <li><tt>true</tt>: no comparison, always returns true</li>
4065</ol>
4066<p>The remaining two arguments must be <a href="#t_vector">vector</a> of
4067<a href="#t_floating">floating point</a> typed. They must also be identical
4068types.</p>
4069<h5>Semantics:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4070<p>The '<tt>vfcmp</tt>' instruction compares <tt>op1</tt> and <tt>op2</tt>
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]4071according to the condition code given as <tt>cond</tt>. The comparison yields a
4072<a href="#t_vector">vector</a> of <a href="#t_integer">integer</a> result, with
4073an identical number of elements as the values being compared, and each element
4074having identical with to the width of the floating point elements. The most
4075significant bit in each element is 1 if the element-wise comparison evaluates to
4076true, and is 0 otherwise. All other bits of the result are undefined. The
4077condition codes are evaluated identically to the
4078<a href="#i_fcmp">'<tt>fcmp</tt>' instruction</a>.
4079
4080<h5>Example:</h5>
4081<pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4082 &lt;result&gt; = vfcmp oeq &lt;2 x float&gt; &lt; float 4, float 0 &gt;, &lt; float 5, float 0 &gt; <i>; yields: result=&lt;2 x i32&gt; &lt; i32 0, i32 -1 &gt;</i>
4083 &lt;result&gt; = vfcmp ult &lt;2 x double&gt; &lt; double 1, double 2 &gt;, &lt; double 2, double 2&gt; <i>; yields: result=&lt;2 x i64&gt; &lt; i64 -1, i64 0 &gt;</i>
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]4084</pre>
4085</div>
4086
4087<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4088<div class="doc_subsubsection">
4089 <a name="i_phi">'<tt>phi</tt>' Instruction</a>
4090</div>
4091
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4092<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4093
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4094<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4095
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4096<pre> &lt;result&gt; = phi &lt;ty&gt; [ &lt;val0&gt;, &lt;label0&gt;], ...<br></pre>
4097<h5>Overview:</h5>
4098<p>The '<tt>phi</tt>' instruction is used to implement the &#966; node in
4099the SSA graph representing the function.</p>
4100<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4101
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4102<p>The type of the incoming values is specified with the first type
4103field. After this, the '<tt>phi</tt>' instruction takes a list of pairs
4104as arguments, with one pair for each predecessor basic block of the
4105current block. Only values of <a href="#t_firstclass">first class</a>
4106type may be used as the value arguments to the PHI node. Only labels
4107may be used as the label arguments.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4108
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4109<p>There must be no non-phi instructions between the start of a basic
4110block and the PHI instructions: i.e. PHI instructions must be first in
4111a basic block.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4112
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4113<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4114
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4115<p>At runtime, the '<tt>phi</tt>' instruction logically takes on the value
4116specified by the pair corresponding to the predecessor basic block that executed
4117just prior to the current block.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4118
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4119<h5>Example:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4120<pre>
4121Loop: ; Infinite loop that counts from 0 on up...
4122 %indvar = phi i32 [ 0, %LoopHeader ], [ %nextindvar, %Loop ]
4123 %nextindvar = add i32 %indvar, 1
4124 br label %Loop
4125</pre>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4126</div>
4127
4128<!-- _______________________________________________________________________ -->
4129<div class="doc_subsubsection">
4130 <a name="i_select">'<tt>select</tt>' Instruction</a>
4131</div>
4132
4133<div class="doc_text">
4134
4135<h5>Syntax:</h5>
4136
4137<pre>
4138 &lt;result&gt; = select i1 &lt;cond&gt;, &lt;ty&gt; &lt;val1&gt;, &lt;ty&gt; &lt;val2&gt; <i>; yields ty</i>
4139</pre>
4140
4141<h5>Overview:</h5>
4142
4143<p>
4144The '<tt>select</tt>' instruction is used to choose one value based on a
4145condition, without branching.
4146</p>
4147
4148
4149<h5>Arguments:</h5>
4150
4151<p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4152The '<tt>select</tt>' instruction requires an 'i1' value indicating the
4153condition, and two values of the same <a href="#t_firstclass">first class</a>
4154type. If the val1/val2 are vectors, the entire vectors are selected, not
4155individual elements.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4156</p>
4157
4158<h5>Semantics:</h5>
4159
4160<p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4161If the i1 condition evaluates is 1, the instruction returns the first
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4162value argument; otherwise, it returns the second value argument.
4163</p>
4164
4165<h5>Example:</h5>
4166
4167<pre>
4168 %X = select i1 true, i8 17, i8 42 <i>; yields i8:17</i>
4169</pre>
4170</div>
4171
4172
4173<!-- _______________________________________________________________________ -->
4174<div class="doc_subsubsection">
4175 <a name="i_call">'<tt>call</tt>' Instruction</a>
4176</div>
4177
4178<div class="doc_text">
4179
4180<h5>Syntax:</h5>
4181<pre>
Nick Lewycky93082fc2007-09-08 13:57:50 +0000[diff] [blame]4182 &lt;result&gt; = [tail] call [<a href="#callingconv">cconv</a>] &lt;ty&gt; [&lt;fnty&gt;*] &lt;fnptrval&gt;(&lt;param list&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4183</pre>
4184
4185<h5>Overview:</h5>
4186
4187<p>The '<tt>call</tt>' instruction represents a simple function call.</p>
4188
4189<h5>Arguments:</h5>
4190
4191<p>This instruction requires several arguments:</p>
4192
4193<ol>
4194 <li>
4195 <p>The optional "tail" marker indicates whether the callee function accesses
4196 any allocas or varargs in the caller. If the "tail" marker is present, the
4197 function call is eligible for tail call optimization. Note that calls may
4198 be marked "tail" even if they do not occur before a <a
4199 href="#i_ret"><tt>ret</tt></a> instruction.
4200 </li>
4201 <li>
4202 <p>The optional "cconv" marker indicates which <a href="#callingconv">calling
4203 convention</a> the call should use. If none is specified, the call defaults
4204 to using C calling conventions.
4205 </li>
4206 <li>
Nick Lewycky93082fc2007-09-08 13:57:50 +0000[diff] [blame]4207 <p>'<tt>ty</tt>': the type of the call instruction itself which is also
4208 the type of the return value. Functions that return no value are marked
4209 <tt><a href="#t_void">void</a></tt>.</p>
4210 </li>
4211 <li>
4212 <p>'<tt>fnty</tt>': shall be the signature of the pointer to function
4213 value being invoked. The argument types must match the types implied by
4214 this signature. This type can be omitted if the function is not varargs
4215 and if the function type does not return a pointer to a function.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4216 </li>
4217 <li>
4218 <p>'<tt>fnptrval</tt>': An LLVM value containing a pointer to a function to
4219 be invoked. In most cases, this is a direct function invocation, but
4220 indirect <tt>call</tt>s are just as possible, calling an arbitrary pointer
4221 to function value.</p>
4222 </li>
4223 <li>
4224 <p>'<tt>function args</tt>': argument list whose types match the
4225 function signature argument types. All arguments must be of
4226 <a href="#t_firstclass">first class</a> type. If the function signature
4227 indicates the function accepts a variable number of arguments, the extra
4228 arguments can be specified.</p>
4229 </li>
4230</ol>
4231
4232<h5>Semantics:</h5>
4233
4234<p>The '<tt>call</tt>' instruction is used to cause control flow to
4235transfer to a specified function, with its incoming arguments bound to
4236the specified values. Upon a '<tt><a href="#i_ret">ret</a></tt>'
4237instruction in the called function, control flow continues with the
4238instruction after the function call, and the return value of the
Chris Lattner5e893ef2008-03-21 17:24:17 +0000[diff] [blame]4239function is bound to the result argument. If the callee returns multiple
4240values then the return values of the function are only accessible through
4241the '<tt><a href="#i_getresult">getresult</a></tt>' instruction.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4242
4243<h5>Example:</h5>
4244
4245<pre>
Nick Lewycky93082fc2007-09-08 13:57:50 +0000[diff] [blame]4246 %retval = call i32 @test(i32 %argc)
Chris Lattner5e893ef2008-03-21 17:24:17 +0000[diff] [blame]4247 call i32 (i8 *, ...)* @printf(i8 * %msg, i32 12, i8 42) <i>; yields i32</i>
4248 %X = tail call i32 @foo() <i>; yields i32</i>
4249 %Y = tail call <a href="#callingconv">fastcc</a> i32 @foo() <i>; yields i32</i>
4250 call void %foo(i8 97 signext)
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]4251
4252 %struct.A = type { i32, i8 }
Chris Lattner5e893ef2008-03-21 17:24:17 +0000[diff] [blame]4253 %r = call %struct.A @foo() <i>; yields { 32, i8 }</i>
4254 %gr = getresult %struct.A %r, 0 <i>; yields i32</i>
4255 %gr1 = getresult %struct.A %r, 1 <i>; yields i8</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4256</pre>
4257
4258</div>
4259
4260<!-- _______________________________________________________________________ -->
4261<div class="doc_subsubsection">
4262 <a name="i_va_arg">'<tt>va_arg</tt>' Instruction</a>
4263</div>
4264
4265<div class="doc_text">
4266
4267<h5>Syntax:</h5>
4268
4269<pre>
4270 &lt;resultval&gt; = va_arg &lt;va_list*&gt; &lt;arglist&gt;, &lt;argty&gt;
4271</pre>
4272
4273<h5>Overview:</h5>
4274
4275<p>The '<tt>va_arg</tt>' instruction is used to access arguments passed through
4276the "variable argument" area of a function call. It is used to implement the
4277<tt>va_arg</tt> macro in C.</p>
4278
4279<h5>Arguments:</h5>
4280
4281<p>This instruction takes a <tt>va_list*</tt> value and the type of
4282the argument. It returns a value of the specified argument type and
4283increments the <tt>va_list</tt> to point to the next argument. The
4284actual type of <tt>va_list</tt> is target specific.</p>
4285
4286<h5>Semantics:</h5>
4287
4288<p>The '<tt>va_arg</tt>' instruction loads an argument of the specified
4289type from the specified <tt>va_list</tt> and causes the
4290<tt>va_list</tt> to point to the next argument. For more information,
4291see the variable argument handling <a href="#int_varargs">Intrinsic
4292Functions</a>.</p>
4293
4294<p>It is legal for this instruction to be called in a function which does not
4295take a variable number of arguments, for example, the <tt>vfprintf</tt>
4296function.</p>
4297
4298<p><tt>va_arg</tt> is an LLVM instruction instead of an <a
4299href="#intrinsics">intrinsic function</a> because it takes a type as an
4300argument.</p>
4301
4302<h5>Example:</h5>
4303
4304<p>See the <a href="#int_varargs">variable argument processing</a> section.</p>
4305
4306</div>
4307
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]4308<!-- _______________________________________________________________________ -->
4309<div class="doc_subsubsection">
4310 <a name="i_getresult">'<tt>getresult</tt>' Instruction</a>
4311</div>
4312
4313<div class="doc_text">
4314
4315<h5>Syntax:</h5>
4316<pre>
Chris Lattneree9da3f2008-03-21 17:20:51 +0000[diff] [blame]4317 &lt;resultval&gt; = getresult &lt;type&gt; &lt;retval&gt;, &lt;index&gt;
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]4318</pre>
Chris Lattneree9da3f2008-03-21 17:20:51 +0000[diff] [blame]4319
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]4320<h5>Overview:</h5>
4321
4322<p> The '<tt>getresult</tt>' instruction is used to extract individual values
Chris Lattneree9da3f2008-03-21 17:20:51 +0000[diff] [blame]4323from a '<tt><a href="#i_call">call</a></tt>'
4324or '<tt><a href="#i_invoke">invoke</a></tt>' instruction that returns multiple
4325results.</p>
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]4326
4327<h5>Arguments:</h5>
4328
Chris Lattneree9da3f2008-03-21 17:20:51 +0000[diff] [blame]4329<p>The '<tt>getresult</tt>' instruction takes a call or invoke value as its
Chris Lattnerd8dd3522008-04-23 04:06:52 +0000[diff] [blame]4330first argument, or an undef value. The value must have <a
4331href="#t_struct">structure type</a>. The second argument is a constant
4332unsigned index value which must be in range for the number of values returned
4333by the call.</p>
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]4334
4335<h5>Semantics:</h5>
4336
Chris Lattneree9da3f2008-03-21 17:20:51 +0000[diff] [blame]4337<p>The '<tt>getresult</tt>' instruction extracts the element identified by
4338'<tt>index</tt>' from the aggregate value.</p>
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]4339
4340<h5>Example:</h5>
4341
4342<pre>
4343 %struct.A = type { i32, i8 }
4344
4345 %r = call %struct.A @foo()
Chris Lattneree9da3f2008-03-21 17:20:51 +0000[diff] [blame]4346 %gr = getresult %struct.A %r, 0 <i>; yields i32:%gr</i>
4347 %gr1 = getresult %struct.A %r, 1 <i>; yields i8:%gr1</i>
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]4348 add i32 %gr, 42
4349 add i8 %gr1, 41
4350</pre>
4351
4352</div>
4353
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4354<!-- *********************************************************************** -->
4355<div class="doc_section"> <a name="intrinsics">Intrinsic Functions</a> </div>
4356<!-- *********************************************************************** -->
4357
4358<div class="doc_text">
4359
4360<p>LLVM supports the notion of an "intrinsic function". These functions have
4361well known names and semantics and are required to follow certain restrictions.
4362Overall, these intrinsics represent an extension mechanism for the LLVM
4363language that does not require changing all of the transformations in LLVM when
4364adding to the language (or the bitcode reader/writer, the parser, etc...).</p>
4365
4366<p>Intrinsic function names must all start with an "<tt>llvm.</tt>" prefix. This
4367prefix is reserved in LLVM for intrinsic names; thus, function names may not
4368begin with this prefix. Intrinsic functions must always be external functions:
4369you cannot define the body of intrinsic functions. Intrinsic functions may
4370only be used in call or invoke instructions: it is illegal to take the address
4371of an intrinsic function. Additionally, because intrinsic functions are part
4372of the LLVM language, it is required if any are added that they be documented
4373here.</p>
4374
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]4375<p>Some intrinsic functions can be overloaded, i.e., the intrinsic represents
4376a family of functions that perform the same operation but on different data
4377types. Because LLVM can represent over 8 million different integer types,
4378overloading is used commonly to allow an intrinsic function to operate on any
4379integer type. One or more of the argument types or the result type can be
4380overloaded to accept any integer type. Argument types may also be defined as
4381exactly matching a previous argument's type or the result type. This allows an
4382intrinsic function which accepts multiple arguments, but needs all of them to
4383be of the same type, to only be overloaded with respect to a single argument or
4384the result.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4385
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]4386<p>Overloaded intrinsics will have the names of its overloaded argument types
4387encoded into its function name, each preceded by a period. Only those types
4388which are overloaded result in a name suffix. Arguments whose type is matched
4389against another type do not. For example, the <tt>llvm.ctpop</tt> function can
4390take an integer of any width and returns an integer of exactly the same integer
4391width. This leads to a family of functions such as
4392<tt>i8 @llvm.ctpop.i8(i8 %val)</tt> and <tt>i29 @llvm.ctpop.i29(i29 %val)</tt>.
4393Only one type, the return type, is overloaded, and only one type suffix is
4394required. Because the argument's type is matched against the return type, it
4395does not require its own name suffix.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4396
4397<p>To learn how to add an intrinsic function, please see the
4398<a href="ExtendingLLVM.html">Extending LLVM Guide</a>.
4399</p>
4400
4401</div>
4402
4403<!-- ======================================================================= -->
4404<div class="doc_subsection">
4405 <a name="int_varargs">Variable Argument Handling Intrinsics</a>
4406</div>
4407
4408<div class="doc_text">
4409
4410<p>Variable argument support is defined in LLVM with the <a
4411 href="#i_va_arg"><tt>va_arg</tt></a> instruction and these three
4412intrinsic functions. These functions are related to the similarly
4413named macros defined in the <tt>&lt;stdarg.h&gt;</tt> header file.</p>
4414
4415<p>All of these functions operate on arguments that use a
4416target-specific value type "<tt>va_list</tt>". The LLVM assembly
4417language reference manual does not define what this type is, so all
4418transformations should be prepared to handle these functions regardless of
4419the type used.</p>
4420
4421<p>This example shows how the <a href="#i_va_arg"><tt>va_arg</tt></a>
4422instruction and the variable argument handling intrinsic functions are
4423used.</p>
4424
4425<div class="doc_code">
4426<pre>
4427define i32 @test(i32 %X, ...) {
4428 ; Initialize variable argument processing
4429 %ap = alloca i8*
4430 %ap2 = bitcast i8** %ap to i8*
4431 call void @llvm.va_start(i8* %ap2)
4432
4433 ; Read a single integer argument
4434 %tmp = va_arg i8** %ap, i32
4435
4436 ; Demonstrate usage of llvm.va_copy and llvm.va_end
4437 %aq = alloca i8*
4438 %aq2 = bitcast i8** %aq to i8*
4439 call void @llvm.va_copy(i8* %aq2, i8* %ap2)
4440 call void @llvm.va_end(i8* %aq2)
4441
4442 ; Stop processing of arguments.
4443 call void @llvm.va_end(i8* %ap2)
4444 ret i32 %tmp
4445}
4446
4447declare void @llvm.va_start(i8*)
4448declare void @llvm.va_copy(i8*, i8*)
4449declare void @llvm.va_end(i8*)
4450</pre>
4451</div>
4452
4453</div>
4454
4455<!-- _______________________________________________________________________ -->
4456<div class="doc_subsubsection">
4457 <a name="int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a>
4458</div>
4459
4460
4461<div class="doc_text">
4462<h5>Syntax:</h5>
4463<pre> declare void %llvm.va_start(i8* &lt;arglist&gt;)<br></pre>
4464<h5>Overview:</h5>
4465<P>The '<tt>llvm.va_start</tt>' intrinsic initializes
4466<tt>*&lt;arglist&gt;</tt> for subsequent use by <tt><a
4467href="#i_va_arg">va_arg</a></tt>.</p>
4468
4469<h5>Arguments:</h5>
4470
4471<P>The argument is a pointer to a <tt>va_list</tt> element to initialize.</p>
4472
4473<h5>Semantics:</h5>
4474
4475<P>The '<tt>llvm.va_start</tt>' intrinsic works just like the <tt>va_start</tt>
4476macro available in C. In a target-dependent way, it initializes the
4477<tt>va_list</tt> element to which the argument points, so that the next call to
4478<tt>va_arg</tt> will produce the first variable argument passed to the function.
4479Unlike the C <tt>va_start</tt> macro, this intrinsic does not need to know the
4480last argument of the function as the compiler can figure that out.</p>
4481
4482</div>
4483
4484<!-- _______________________________________________________________________ -->
4485<div class="doc_subsubsection">
4486 <a name="int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a>
4487</div>
4488
4489<div class="doc_text">
4490<h5>Syntax:</h5>
4491<pre> declare void @llvm.va_end(i8* &lt;arglist&gt;)<br></pre>
4492<h5>Overview:</h5>
4493
4494<p>The '<tt>llvm.va_end</tt>' intrinsic destroys <tt>*&lt;arglist&gt;</tt>,
4495which has been initialized previously with <tt><a href="#int_va_start">llvm.va_start</a></tt>
4496or <tt><a href="#i_va_copy">llvm.va_copy</a></tt>.</p>
4497
4498<h5>Arguments:</h5>
4499
4500<p>The argument is a pointer to a <tt>va_list</tt> to destroy.</p>
4501
4502<h5>Semantics:</h5>
4503
4504<p>The '<tt>llvm.va_end</tt>' intrinsic works just like the <tt>va_end</tt>
4505macro available in C. In a target-dependent way, it destroys the
4506<tt>va_list</tt> element to which the argument points. Calls to <a
4507href="#int_va_start"><tt>llvm.va_start</tt></a> and <a href="#int_va_copy">
4508<tt>llvm.va_copy</tt></a> must be matched exactly with calls to
4509<tt>llvm.va_end</tt>.</p>
4510
4511</div>
4512
4513<!-- _______________________________________________________________________ -->
4514<div class="doc_subsubsection">
4515 <a name="int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a>
4516</div>
4517
4518<div class="doc_text">
4519
4520<h5>Syntax:</h5>
4521
4522<pre>
4523 declare void @llvm.va_copy(i8* &lt;destarglist&gt;, i8* &lt;srcarglist&gt;)
4524</pre>
4525
4526<h5>Overview:</h5>
4527
4528<p>The '<tt>llvm.va_copy</tt>' intrinsic copies the current argument position
4529from the source argument list to the destination argument list.</p>
4530
4531<h5>Arguments:</h5>
4532
4533<p>The first argument is a pointer to a <tt>va_list</tt> element to initialize.
4534The second argument is a pointer to a <tt>va_list</tt> element to copy from.</p>
4535
4536
4537<h5>Semantics:</h5>
4538
4539<p>The '<tt>llvm.va_copy</tt>' intrinsic works just like the <tt>va_copy</tt>
4540macro available in C. In a target-dependent way, it copies the source
4541<tt>va_list</tt> element into the destination <tt>va_list</tt> element. This
4542intrinsic is necessary because the <tt><a href="#int_va_start">
4543llvm.va_start</a></tt> intrinsic may be arbitrarily complex and require, for
4544example, memory allocation.</p>
4545
4546</div>
4547
4548<!-- ======================================================================= -->
4549<div class="doc_subsection">
4550 <a name="int_gc">Accurate Garbage Collection Intrinsics</a>
4551</div>
4552
4553<div class="doc_text">
4554
4555<p>
4556LLVM support for <a href="GarbageCollection.html">Accurate Garbage
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]4557Collection</a> (GC) requires the implementation and generation of these
4558intrinsics.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4559These intrinsics allow identification of <a href="#int_gcroot">GC roots on the
4560stack</a>, as well as garbage collector implementations that require <a
4561href="#int_gcread">read</a> and <a href="#int_gcwrite">write</a> barriers.
4562Front-ends for type-safe garbage collected languages should generate these
4563intrinsics to make use of the LLVM garbage collectors. For more details, see <a
4564href="GarbageCollection.html">Accurate Garbage Collection with LLVM</a>.
4565</p>
Christopher Lambcfe00962007-12-17 01:00:21 +0000[diff] [blame]4566
4567<p>The garbage collection intrinsics only operate on objects in the generic
4568 address space (address space zero).</p>
4569
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4570</div>
4571
4572<!-- _______________________________________________________________________ -->
4573<div class="doc_subsubsection">
4574 <a name="int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a>
4575</div>
4576
4577<div class="doc_text">
4578
4579<h5>Syntax:</h5>
4580
4581<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4582 declare void @llvm.gcroot(i8** %ptrloc, i8* %metadata)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4583</pre>
4584
4585<h5>Overview:</h5>
4586
4587<p>The '<tt>llvm.gcroot</tt>' intrinsic declares the existence of a GC root to
4588the code generator, and allows some metadata to be associated with it.</p>
4589
4590<h5>Arguments:</h5>
4591
4592<p>The first argument specifies the address of a stack object that contains the
4593root pointer. The second pointer (which must be either a constant or a global
4594value address) contains the meta-data to be associated with the root.</p>
4595
4596<h5>Semantics:</h5>
4597
Chris Lattnera7d94ba2008-04-24 05:59:56 +0000[diff] [blame]4598<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]4599location. At compile-time, the code generator generates information to allow
Gordon Henriksen40393542007-12-25 02:31:26 +0000[diff] [blame]4600the runtime to find the pointer at GC safe points. The '<tt>llvm.gcroot</tt>'
4601intrinsic may only be used in a function which <a href="#gc">specifies a GC
4602algorithm</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4603
4604</div>
4605
4606
4607<!-- _______________________________________________________________________ -->
4608<div class="doc_subsubsection">
4609 <a name="int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a>
4610</div>
4611
4612<div class="doc_text">
4613
4614<h5>Syntax:</h5>
4615
4616<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4617 declare i8* @llvm.gcread(i8* %ObjPtr, i8** %Ptr)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4618</pre>
4619
4620<h5>Overview:</h5>
4621
4622<p>The '<tt>llvm.gcread</tt>' intrinsic identifies reads of references from heap
4623locations, allowing garbage collector implementations that require read
4624barriers.</p>
4625
4626<h5>Arguments:</h5>
4627
4628<p>The second argument is the address to read from, which should be an address
4629allocated from the garbage collector. The first object is a pointer to the
4630start of the referenced object, if needed by the language runtime (otherwise
4631null).</p>
4632
4633<h5>Semantics:</h5>
4634
4635<p>The '<tt>llvm.gcread</tt>' intrinsic has the same semantics as a load
4636instruction, but may be replaced with substantially more complex code by the
Gordon Henriksen40393542007-12-25 02:31:26 +0000[diff] [blame]4637garbage collector runtime, as needed. The '<tt>llvm.gcread</tt>' intrinsic
4638may only be used in a function which <a href="#gc">specifies a GC
4639algorithm</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4640
4641</div>
4642
4643
4644<!-- _______________________________________________________________________ -->
4645<div class="doc_subsubsection">
4646 <a name="int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a>
4647</div>
4648
4649<div class="doc_text">
4650
4651<h5>Syntax:</h5>
4652
4653<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4654 declare void @llvm.gcwrite(i8* %P1, i8* %Obj, i8** %P2)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4655</pre>
4656
4657<h5>Overview:</h5>
4658
4659<p>The '<tt>llvm.gcwrite</tt>' intrinsic identifies writes of references to heap
4660locations, allowing garbage collector implementations that require write
4661barriers (such as generational or reference counting collectors).</p>
4662
4663<h5>Arguments:</h5>
4664
4665<p>The first argument is the reference to store, the second is the start of the
4666object to store it to, and the third is the address of the field of Obj to
4667store to. If the runtime does not require a pointer to the object, Obj may be
4668null.</p>
4669
4670<h5>Semantics:</h5>
4671
4672<p>The '<tt>llvm.gcwrite</tt>' intrinsic has the same semantics as a store
4673instruction, but may be replaced with substantially more complex code by the
Gordon Henriksen40393542007-12-25 02:31:26 +0000[diff] [blame]4674garbage collector runtime, as needed. The '<tt>llvm.gcwrite</tt>' intrinsic
4675may only be used in a function which <a href="#gc">specifies a GC
4676algorithm</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4677
4678</div>
4679
4680
4681
4682<!-- ======================================================================= -->
4683<div class="doc_subsection">
4684 <a name="int_codegen">Code Generator Intrinsics</a>
4685</div>
4686
4687<div class="doc_text">
4688<p>
4689These intrinsics are provided by LLVM to expose special features that may only
4690be implemented with code generator support.
4691</p>
4692
4693</div>
4694
4695<!-- _______________________________________________________________________ -->
4696<div class="doc_subsubsection">
4697 <a name="int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a>
4698</div>
4699
4700<div class="doc_text">
4701
4702<h5>Syntax:</h5>
4703<pre>
4704 declare i8 *@llvm.returnaddress(i32 &lt;level&gt;)
4705</pre>
4706
4707<h5>Overview:</h5>
4708
4709<p>
4710The '<tt>llvm.returnaddress</tt>' intrinsic attempts to compute a
4711target-specific value indicating the return address of the current function
4712or one of its callers.
4713</p>
4714
4715<h5>Arguments:</h5>
4716
4717<p>
4718The argument to this intrinsic indicates which function to return the address
4719for. Zero indicates the calling function, one indicates its caller, etc. The
4720argument is <b>required</b> to be a constant integer value.
4721</p>
4722
4723<h5>Semantics:</h5>
4724
4725<p>
4726The '<tt>llvm.returnaddress</tt>' intrinsic either returns a pointer indicating
4727the return address of the specified call frame, or zero if it cannot be
4728identified. The value returned by this intrinsic is likely to be incorrect or 0
4729for arguments other than zero, so it should only be used for debugging purposes.
4730</p>
4731
4732<p>
4733Note that calling this intrinsic does not prevent function inlining or other
4734aggressive transformations, so the value returned may not be that of the obvious
4735source-language caller.
4736</p>
4737</div>
4738
4739
4740<!-- _______________________________________________________________________ -->
4741<div class="doc_subsubsection">
4742 <a name="int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a>
4743</div>
4744
4745<div class="doc_text">
4746
4747<h5>Syntax:</h5>
4748<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4749 declare i8 *@llvm.frameaddress(i32 &lt;level&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4750</pre>
4751
4752<h5>Overview:</h5>
4753
4754<p>
4755The '<tt>llvm.frameaddress</tt>' intrinsic attempts to return the
4756target-specific frame pointer value for the specified stack frame.
4757</p>
4758
4759<h5>Arguments:</h5>
4760
4761<p>
4762The argument to this intrinsic indicates which function to return the frame
4763pointer for. Zero indicates the calling function, one indicates its caller,
4764etc. The argument is <b>required</b> to be a constant integer value.
4765</p>
4766
4767<h5>Semantics:</h5>
4768
4769<p>
4770The '<tt>llvm.frameaddress</tt>' intrinsic either returns a pointer indicating
4771the frame address of the specified call frame, or zero if it cannot be
4772identified. The value returned by this intrinsic is likely to be incorrect or 0
4773for arguments other than zero, so it should only be used for debugging purposes.
4774</p>
4775
4776<p>
4777Note that calling this intrinsic does not prevent function inlining or other
4778aggressive transformations, so the value returned may not be that of the obvious
4779source-language caller.
4780</p>
4781</div>
4782
4783<!-- _______________________________________________________________________ -->
4784<div class="doc_subsubsection">
4785 <a name="int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a>
4786</div>
4787
4788<div class="doc_text">
4789
4790<h5>Syntax:</h5>
4791<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4792 declare i8 *@llvm.stacksave()
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4793</pre>
4794
4795<h5>Overview:</h5>
4796
4797<p>
4798The '<tt>llvm.stacksave</tt>' intrinsic is used to remember the current state of
4799the function stack, for use with <a href="#int_stackrestore">
4800<tt>llvm.stackrestore</tt></a>. This is useful for implementing language
4801features like scoped automatic variable sized arrays in C99.
4802</p>
4803
4804<h5>Semantics:</h5>
4805
4806<p>
4807This intrinsic returns a opaque pointer value that can be passed to <a
4808href="#int_stackrestore"><tt>llvm.stackrestore</tt></a>. When an
4809<tt>llvm.stackrestore</tt> intrinsic is executed with a value saved from
4810<tt>llvm.stacksave</tt>, it effectively restores the state of the stack to the
4811state it was in when the <tt>llvm.stacksave</tt> intrinsic executed. In
4812practice, this pops any <a href="#i_alloca">alloca</a> blocks from the stack
4813that were allocated after the <tt>llvm.stacksave</tt> was executed.
4814</p>
4815
4816</div>
4817
4818<!-- _______________________________________________________________________ -->
4819<div class="doc_subsubsection">
4820 <a name="int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a>
4821</div>
4822
4823<div class="doc_text">
4824
4825<h5>Syntax:</h5>
4826<pre>
4827 declare void @llvm.stackrestore(i8 * %ptr)
4828</pre>
4829
4830<h5>Overview:</h5>
4831
4832<p>
4833The '<tt>llvm.stackrestore</tt>' intrinsic is used to restore the state of
4834the function stack to the state it was in when the corresponding <a
4835href="#int_stacksave"><tt>llvm.stacksave</tt></a> intrinsic executed. This is
4836useful for implementing language features like scoped automatic variable sized
4837arrays in C99.
4838</p>
4839
4840<h5>Semantics:</h5>
4841
4842<p>
4843See the description for <a href="#int_stacksave"><tt>llvm.stacksave</tt></a>.
4844</p>
4845
4846</div>
4847
4848
4849<!-- _______________________________________________________________________ -->
4850<div class="doc_subsubsection">
4851 <a name="int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a>
4852</div>
4853
4854<div class="doc_text">
4855
4856<h5>Syntax:</h5>
4857<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4858 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]4859</pre>
4860
4861<h5>Overview:</h5>
4862
4863
4864<p>
4865The '<tt>llvm.prefetch</tt>' intrinsic is a hint to the code generator to insert
4866a prefetch instruction if supported; otherwise, it is a noop. Prefetches have
4867no
4868effect on the behavior of the program but can change its performance
4869characteristics.
4870</p>
4871
4872<h5>Arguments:</h5>
4873
4874<p>
4875<tt>address</tt> is the address to be prefetched, <tt>rw</tt> is the specifier
4876determining if the fetch should be for a read (0) or write (1), and
4877<tt>locality</tt> is a temporal locality specifier ranging from (0) - no
4878locality, to (3) - extremely local keep in cache. The <tt>rw</tt> and
4879<tt>locality</tt> arguments must be constant integers.
4880</p>
4881
4882<h5>Semantics:</h5>
4883
4884<p>
4885This intrinsic does not modify the behavior of the program. In particular,
4886prefetches cannot trap and do not produce a value. On targets that support this
4887intrinsic, the prefetch can provide hints to the processor cache for better
4888performance.
4889</p>
4890
4891</div>
4892
4893<!-- _______________________________________________________________________ -->
4894<div class="doc_subsubsection">
4895 <a name="int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a>
4896</div>
4897
4898<div class="doc_text">
4899
4900<h5>Syntax:</h5>
4901<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4902 declare void @llvm.pcmarker(i32 &lt;id&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4903</pre>
4904
4905<h5>Overview:</h5>
4906
4907
4908<p>
4909The '<tt>llvm.pcmarker</tt>' intrinsic is a method to export a Program Counter
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]4910(PC) in a region of
4911code to simulators and other tools. The method is target specific, but it is
4912expected that the marker will use exported symbols to transmit the PC of the
4913marker.
4914The marker makes no guarantees that it will remain with any specific instruction
4915after optimizations. It is possible that the presence of a marker will inhibit
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4916optimizations. The intended use is to be inserted after optimizations to allow
4917correlations of simulation runs.
4918</p>
4919
4920<h5>Arguments:</h5>
4921
4922<p>
4923<tt>id</tt> is a numerical id identifying the marker.
4924</p>
4925
4926<h5>Semantics:</h5>
4927
4928<p>
4929This intrinsic does not modify the behavior of the program. Backends that do not
4930support this intrinisic may ignore it.
4931</p>
4932
4933</div>
4934
4935<!-- _______________________________________________________________________ -->
4936<div class="doc_subsubsection">
4937 <a name="int_readcyclecounter">'<tt>llvm.readcyclecounter</tt>' Intrinsic</a>
4938</div>
4939
4940<div class="doc_text">
4941
4942<h5>Syntax:</h5>
4943<pre>
4944 declare i64 @llvm.readcyclecounter( )
4945</pre>
4946
4947<h5>Overview:</h5>
4948
4949
4950<p>
4951The '<tt>llvm.readcyclecounter</tt>' intrinsic provides access to the cycle
4952counter register (or similar low latency, high accuracy clocks) on those targets
4953that support it. On X86, it should map to RDTSC. On Alpha, it should map to RPCC.
4954As the backing counters overflow quickly (on the order of 9 seconds on alpha), this
4955should only be used for small timings.
4956</p>
4957
4958<h5>Semantics:</h5>
4959
4960<p>
4961When directly supported, reading the cycle counter should not modify any memory.
4962Implementations are allowed to either return a application specific value or a
4963system wide value. On backends without support, this is lowered to a constant 0.
4964</p>
4965
4966</div>
4967
4968<!-- ======================================================================= -->
4969<div class="doc_subsection">
4970 <a name="int_libc">Standard C Library Intrinsics</a>
4971</div>
4972
4973<div class="doc_text">
4974<p>
4975LLVM provides intrinsics for a few important standard C library functions.
4976These intrinsics allow source-language front-ends to pass information about the
4977alignment of the pointer arguments to the code generator, providing opportunity
4978for more efficient code generation.
4979</p>
4980
4981</div>
4982
4983<!-- _______________________________________________________________________ -->
4984<div class="doc_subsubsection">
4985 <a name="int_memcpy">'<tt>llvm.memcpy</tt>' Intrinsic</a>
4986</div>
4987
4988<div class="doc_text">
4989
4990<h5>Syntax:</h5>
4991<pre>
4992 declare void @llvm.memcpy.i32(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
4993 i32 &lt;len&gt;, i32 &lt;align&gt;)
4994 declare void @llvm.memcpy.i64(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
4995 i64 &lt;len&gt;, i32 &lt;align&gt;)
4996</pre>
4997
4998<h5>Overview:</h5>
4999
5000<p>
5001The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the source
5002location to the destination location.
5003</p>
5004
5005<p>
5006Note that, unlike the standard libc function, the <tt>llvm.memcpy.*</tt>
5007intrinsics do not return a value, and takes an extra alignment argument.
5008</p>
5009
5010<h5>Arguments:</h5>
5011
5012<p>
5013The first argument is a pointer to the destination, the second is a pointer to
5014the source. The third argument is an integer argument
5015specifying the number of bytes to copy, and the fourth argument is the alignment
5016of the source and destination locations.
5017</p>
5018
5019<p>
5020If the call to this intrinisic has an alignment value that is not 0 or 1, then
5021the caller guarantees that both the source and destination pointers are aligned
5022to that boundary.
5023</p>
5024
5025<h5>Semantics:</h5>
5026
5027<p>
5028The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the source
5029location to the destination location, which are not allowed to overlap. It
5030copies "len" bytes of memory over. If the argument is known to be aligned to
5031some boundary, this can be specified as the fourth argument, otherwise it should
5032be set to 0 or 1.
5033</p>
5034</div>
5035
5036
5037<!-- _______________________________________________________________________ -->
5038<div class="doc_subsubsection">
5039 <a name="int_memmove">'<tt>llvm.memmove</tt>' Intrinsic</a>
5040</div>
5041
5042<div class="doc_text">
5043
5044<h5>Syntax:</h5>
5045<pre>
5046 declare void @llvm.memmove.i32(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5047 i32 &lt;len&gt;, i32 &lt;align&gt;)
5048 declare void @llvm.memmove.i64(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5049 i64 &lt;len&gt;, i32 &lt;align&gt;)
5050</pre>
5051
5052<h5>Overview:</h5>
5053
5054<p>
5055The '<tt>llvm.memmove.*</tt>' intrinsics move a block of memory from the source
5056location to the destination location. It is similar to the
Chris Lattnerdba16ea2008-01-06 19:51:52 +0000[diff] [blame]5057'<tt>llvm.memcpy</tt>' intrinsic but allows the two memory locations to overlap.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5058</p>
5059
5060<p>
5061Note that, unlike the standard libc function, the <tt>llvm.memmove.*</tt>
5062intrinsics do not return a value, and takes an extra alignment argument.
5063</p>
5064
5065<h5>Arguments:</h5>
5066
5067<p>
5068The first argument is a pointer to the destination, the second is a pointer to
5069the source. The third argument is an integer argument
5070specifying the number of bytes to copy, and the fourth argument is the alignment
5071of the source and destination locations.
5072</p>
5073
5074<p>
5075If the call to this intrinisic has an alignment value that is not 0 or 1, then
5076the caller guarantees that the source and destination pointers are aligned to
5077that boundary.
5078</p>
5079
5080<h5>Semantics:</h5>
5081
5082<p>
5083The '<tt>llvm.memmove.*</tt>' intrinsics copy a block of memory from the source
5084location to the destination location, which may overlap. It
5085copies "len" bytes of memory over. If the argument is known to be aligned to
5086some boundary, this can be specified as the fourth argument, otherwise it should
5087be set to 0 or 1.
5088</p>
5089</div>
5090
5091
5092<!-- _______________________________________________________________________ -->
5093<div class="doc_subsubsection">
5094 <a name="int_memset">'<tt>llvm.memset.*</tt>' Intrinsics</a>
5095</div>
5096
5097<div class="doc_text">
5098
5099<h5>Syntax:</h5>
5100<pre>
5101 declare void @llvm.memset.i32(i8 * &lt;dest&gt;, i8 &lt;val&gt;,
5102 i32 &lt;len&gt;, i32 &lt;align&gt;)
5103 declare void @llvm.memset.i64(i8 * &lt;dest&gt;, i8 &lt;val&gt;,
5104 i64 &lt;len&gt;, i32 &lt;align&gt;)
5105</pre>
5106
5107<h5>Overview:</h5>
5108
5109<p>
5110The '<tt>llvm.memset.*</tt>' intrinsics fill a block of memory with a particular
5111byte value.
5112</p>
5113
5114<p>
5115Note that, unlike the standard libc function, the <tt>llvm.memset</tt> intrinsic
5116does not return a value, and takes an extra alignment argument.
5117</p>
5118
5119<h5>Arguments:</h5>
5120
5121<p>
5122The first argument is a pointer to the destination to fill, the second is the
5123byte value to fill it with, the third argument is an integer
5124argument specifying the number of bytes to fill, and the fourth argument is the
5125known alignment of destination location.
5126</p>
5127
5128<p>
5129If the call to this intrinisic has an alignment value that is not 0 or 1, then
5130the caller guarantees that the destination pointer is aligned to that boundary.
5131</p>
5132
5133<h5>Semantics:</h5>
5134
5135<p>
5136The '<tt>llvm.memset.*</tt>' intrinsics fill "len" bytes of memory starting at
5137the
5138destination location. If the argument is known to be aligned to some boundary,
5139this can be specified as the fourth argument, otherwise it should be set to 0 or
51401.
5141</p>
5142</div>
5143
5144
5145<!-- _______________________________________________________________________ -->
5146<div class="doc_subsubsection">
5147 <a name="int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a>
5148</div>
5149
5150<div class="doc_text">
5151
5152<h5>Syntax:</h5>
Dale Johannesenf9adbb62007-10-02 17:47:38 +0000[diff] [blame]5153<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]5154floating point or vector of floating point type. Not all targets support all
5155types however.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5156<pre>
Dale Johannesenf9adbb62007-10-02 17:47:38 +0000[diff] [blame]5157 declare float @llvm.sqrt.f32(float %Val)
5158 declare double @llvm.sqrt.f64(double %Val)
5159 declare x86_fp80 @llvm.sqrt.f80(x86_fp80 %Val)
5160 declare fp128 @llvm.sqrt.f128(fp128 %Val)
5161 declare ppc_fp128 @llvm.sqrt.ppcf128(ppc_fp128 %Val)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5162</pre>
5163
5164<h5>Overview:</h5>
5165
5166<p>
5167The '<tt>llvm.sqrt</tt>' intrinsics return the sqrt of the specified operand,
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5168returning the same value as the libm '<tt>sqrt</tt>' functions would. Unlike
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5169<tt>sqrt</tt> in libm, however, <tt>llvm.sqrt</tt> has undefined behavior for
Chris Lattnerf914a002008-01-29 07:00:44 +0000[diff] [blame]5170negative numbers other than -0.0 (which allows for better optimization, because
5171there is no need to worry about errno being set). <tt>llvm.sqrt(-0.0)</tt> is
5172defined to return -0.0 like IEEE sqrt.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5173</p>
5174
5175<h5>Arguments:</h5>
5176
5177<p>
5178The argument and return value are floating point numbers of the same type.
5179</p>
5180
5181<h5>Semantics:</h5>
5182
5183<p>
5184This function returns the sqrt of the specified operand if it is a nonnegative
5185floating point number.
5186</p>
5187</div>
5188
5189<!-- _______________________________________________________________________ -->
5190<div class="doc_subsubsection">
5191 <a name="int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a>
5192</div>
5193
5194<div class="doc_text">
5195
5196<h5>Syntax:</h5>
Dale Johannesenf9adbb62007-10-02 17:47:38 +0000[diff] [blame]5197<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]5198floating point or vector of floating point type. Not all targets support all
5199types however.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5200<pre>
Dale Johannesenf9adbb62007-10-02 17:47:38 +0000[diff] [blame]5201 declare float @llvm.powi.f32(float %Val, i32 %power)
5202 declare double @llvm.powi.f64(double %Val, i32 %power)
5203 declare x86_fp80 @llvm.powi.f80(x86_fp80 %Val, i32 %power)
5204 declare fp128 @llvm.powi.f128(fp128 %Val, i32 %power)
5205 declare ppc_fp128 @llvm.powi.ppcf128(ppc_fp128 %Val, i32 %power)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5206</pre>
5207
5208<h5>Overview:</h5>
5209
5210<p>
5211The '<tt>llvm.powi.*</tt>' intrinsics return the first operand raised to the
5212specified (positive or negative) power. The order of evaluation of
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5213multiplications is not defined. When a vector of floating point type is
5214used, the second argument remains a scalar integer value.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5215</p>
5216
5217<h5>Arguments:</h5>
5218
5219<p>
5220The second argument is an integer power, and the first is a value to raise to
5221that power.
5222</p>
5223
5224<h5>Semantics:</h5>
5225
5226<p>
5227This function returns the first value raised to the second power with an
5228unspecified sequence of rounding operations.</p>
5229</div>
5230
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5231<!-- _______________________________________________________________________ -->
5232<div class="doc_subsubsection">
5233 <a name="int_sin">'<tt>llvm.sin.*</tt>' Intrinsic</a>
5234</div>
5235
5236<div class="doc_text">
5237
5238<h5>Syntax:</h5>
5239<p>This is an overloaded intrinsic. You can use <tt>llvm.sin</tt> on any
5240floating point or vector of floating point type. Not all targets support all
5241types however.
5242<pre>
5243 declare float @llvm.sin.f32(float %Val)
5244 declare double @llvm.sin.f64(double %Val)
5245 declare x86_fp80 @llvm.sin.f80(x86_fp80 %Val)
5246 declare fp128 @llvm.sin.f128(fp128 %Val)
5247 declare ppc_fp128 @llvm.sin.ppcf128(ppc_fp128 %Val)
5248</pre>
5249
5250<h5>Overview:</h5>
5251
5252<p>
5253The '<tt>llvm.sin.*</tt>' intrinsics return the sine of the operand.
5254</p>
5255
5256<h5>Arguments:</h5>
5257
5258<p>
5259The argument and return value are floating point numbers of the same type.
5260</p>
5261
5262<h5>Semantics:</h5>
5263
5264<p>
5265This function returns the sine of the specified operand, returning the
5266same values as the libm <tt>sin</tt> functions would, and handles error
Dan Gohmaneaba92e2007-10-17 18:05:13 +0000[diff] [blame]5267conditions in the same way.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5268</div>
5269
5270<!-- _______________________________________________________________________ -->
5271<div class="doc_subsubsection">
5272 <a name="int_cos">'<tt>llvm.cos.*</tt>' Intrinsic</a>
5273</div>
5274
5275<div class="doc_text">
5276
5277<h5>Syntax:</h5>
5278<p>This is an overloaded intrinsic. You can use <tt>llvm.cos</tt> on any
5279floating point or vector of floating point type. Not all targets support all
5280types however.
5281<pre>
5282 declare float @llvm.cos.f32(float %Val)
5283 declare double @llvm.cos.f64(double %Val)
5284 declare x86_fp80 @llvm.cos.f80(x86_fp80 %Val)
5285 declare fp128 @llvm.cos.f128(fp128 %Val)
5286 declare ppc_fp128 @llvm.cos.ppcf128(ppc_fp128 %Val)
5287</pre>
5288
5289<h5>Overview:</h5>
5290
5291<p>
5292The '<tt>llvm.cos.*</tt>' intrinsics return the cosine of the operand.
5293</p>
5294
5295<h5>Arguments:</h5>
5296
5297<p>
5298The argument and return value are floating point numbers of the same type.
5299</p>
5300
5301<h5>Semantics:</h5>
5302
5303<p>
5304This function returns the cosine of the specified operand, returning the
5305same values as the libm <tt>cos</tt> functions would, and handles error
Dan Gohmaneaba92e2007-10-17 18:05:13 +0000[diff] [blame]5306conditions in the same way.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5307</div>
5308
5309<!-- _______________________________________________________________________ -->
5310<div class="doc_subsubsection">
5311 <a name="int_pow">'<tt>llvm.pow.*</tt>' Intrinsic</a>
5312</div>
5313
5314<div class="doc_text">
5315
5316<h5>Syntax:</h5>
5317<p>This is an overloaded intrinsic. You can use <tt>llvm.pow</tt> on any
5318floating point or vector of floating point type. Not all targets support all
5319types however.
5320<pre>
5321 declare float @llvm.pow.f32(float %Val, float %Power)
5322 declare double @llvm.pow.f64(double %Val, double %Power)
5323 declare x86_fp80 @llvm.pow.f80(x86_fp80 %Val, x86_fp80 %Power)
5324 declare fp128 @llvm.pow.f128(fp128 %Val, fp128 %Power)
5325 declare ppc_fp128 @llvm.pow.ppcf128(ppc_fp128 %Val, ppc_fp128 Power)
5326</pre>
5327
5328<h5>Overview:</h5>
5329
5330<p>
5331The '<tt>llvm.pow.*</tt>' intrinsics return the first operand raised to the
5332specified (positive or negative) power.
5333</p>
5334
5335<h5>Arguments:</h5>
5336
5337<p>
5338The second argument is a floating point power, and the first is a value to
5339raise to that power.
5340</p>
5341
5342<h5>Semantics:</h5>
5343
5344<p>
5345This function returns the first value raised to the second power,
5346returning the
5347same values as the libm <tt>pow</tt> functions would, and handles error
Dan Gohmaneaba92e2007-10-17 18:05:13 +0000[diff] [blame]5348conditions in the same way.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5349</div>
5350
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5351
5352<!-- ======================================================================= -->
5353<div class="doc_subsection">
5354 <a name="int_manip">Bit Manipulation Intrinsics</a>
5355</div>
5356
5357<div class="doc_text">
5358<p>
5359LLVM provides intrinsics for a few important bit manipulation operations.
5360These allow efficient code generation for some algorithms.
5361</p>
5362
5363</div>
5364
5365<!-- _______________________________________________________________________ -->
5366<div class="doc_subsubsection">
5367 <a name="int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a>
5368</div>
5369
5370<div class="doc_text">
5371
5372<h5>Syntax:</h5>
5373<p>This is an overloaded intrinsic function. You can use bswap on any integer
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5374type that is an even number of bytes (i.e. BitWidth % 16 == 0).
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5375<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5376 declare i16 @llvm.bswap.i16(i16 &lt;id&gt;)
5377 declare i32 @llvm.bswap.i32(i32 &lt;id&gt;)
5378 declare i64 @llvm.bswap.i64(i64 &lt;id&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5379</pre>
5380
5381<h5>Overview:</h5>
5382
5383<p>
5384The '<tt>llvm.bswap</tt>' family of intrinsics is used to byte swap integer
5385values with an even number of bytes (positive multiple of 16 bits). These are
5386useful for performing operations on data that is not in the target's native
5387byte order.
5388</p>
5389
5390<h5>Semantics:</h5>
5391
5392<p>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5393The <tt>llvm.bswap.i16</tt> intrinsic returns an i16 value that has the high
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5394and low byte of the input i16 swapped. Similarly, the <tt>llvm.bswap.i32</tt>
5395intrinsic returns an i32 value that has the four bytes of the input i32
5396swapped, 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]5397i32 will have its bytes in 3, 2, 1, 0 order. The <tt>llvm.bswap.i48</tt>,
5398<tt>llvm.bswap.i64</tt> and other intrinsics extend this concept to
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5399additional even-byte lengths (6 bytes, 8 bytes and more, respectively).
5400</p>
5401
5402</div>
5403
5404<!-- _______________________________________________________________________ -->
5405<div class="doc_subsubsection">
5406 <a name="int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic</a>
5407</div>
5408
5409<div class="doc_text">
5410
5411<h5>Syntax:</h5>
5412<p>This is an overloaded intrinsic. You can use llvm.ctpop on any integer bit
5413width. Not all targets support all bit widths however.
5414<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5415 declare i8 @llvm.ctpop.i8 (i8 &lt;src&gt;)
5416 declare i16 @llvm.ctpop.i16(i16 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5417 declare i32 @llvm.ctpop.i32(i32 &lt;src&gt;)
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5418 declare i64 @llvm.ctpop.i64(i64 &lt;src&gt;)
5419 declare i256 @llvm.ctpop.i256(i256 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5420</pre>
5421
5422<h5>Overview:</h5>
5423
5424<p>
5425The '<tt>llvm.ctpop</tt>' family of intrinsics counts the number of bits set in a
5426value.
5427</p>
5428
5429<h5>Arguments:</h5>
5430
5431<p>
5432The only argument is the value to be counted. The argument may be of any
5433integer type. The return type must match the argument type.
5434</p>
5435
5436<h5>Semantics:</h5>
5437
5438<p>
5439The '<tt>llvm.ctpop</tt>' intrinsic counts the 1's in a variable.
5440</p>
5441</div>
5442
5443<!-- _______________________________________________________________________ -->
5444<div class="doc_subsubsection">
5445 <a name="int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic</a>
5446</div>
5447
5448<div class="doc_text">
5449
5450<h5>Syntax:</h5>
5451<p>This is an overloaded intrinsic. You can use <tt>llvm.ctlz</tt> on any
5452integer bit width. Not all targets support all bit widths however.
5453<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5454 declare i8 @llvm.ctlz.i8 (i8 &lt;src&gt;)
5455 declare i16 @llvm.ctlz.i16(i16 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5456 declare i32 @llvm.ctlz.i32(i32 &lt;src&gt;)
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5457 declare i64 @llvm.ctlz.i64(i64 &lt;src&gt;)
5458 declare i256 @llvm.ctlz.i256(i256 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5459</pre>
5460
5461<h5>Overview:</h5>
5462
5463<p>
5464The '<tt>llvm.ctlz</tt>' family of intrinsic functions counts the number of
5465leading zeros in a variable.
5466</p>
5467
5468<h5>Arguments:</h5>
5469
5470<p>
5471The only argument is the value to be counted. The argument may be of any
5472integer type. The return type must match the argument type.
5473</p>
5474
5475<h5>Semantics:</h5>
5476
5477<p>
5478The '<tt>llvm.ctlz</tt>' intrinsic counts the leading (most significant) zeros
5479in a variable. If the src == 0 then the result is the size in bits of the type
5480of src. For example, <tt>llvm.ctlz(i32 2) = 30</tt>.
5481</p>
5482</div>
5483
5484
5485
5486<!-- _______________________________________________________________________ -->
5487<div class="doc_subsubsection">
5488 <a name="int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic</a>
5489</div>
5490
5491<div class="doc_text">
5492
5493<h5>Syntax:</h5>
5494<p>This is an overloaded intrinsic. You can use <tt>llvm.cttz</tt> on any
5495integer bit width. Not all targets support all bit widths however.
5496<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5497 declare i8 @llvm.cttz.i8 (i8 &lt;src&gt;)
5498 declare i16 @llvm.cttz.i16(i16 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5499 declare i32 @llvm.cttz.i32(i32 &lt;src&gt;)
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5500 declare i64 @llvm.cttz.i64(i64 &lt;src&gt;)
5501 declare i256 @llvm.cttz.i256(i256 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5502</pre>
5503
5504<h5>Overview:</h5>
5505
5506<p>
5507The '<tt>llvm.cttz</tt>' family of intrinsic functions counts the number of
5508trailing zeros.
5509</p>
5510
5511<h5>Arguments:</h5>
5512
5513<p>
5514The only argument is the value to be counted. The argument may be of any
5515integer type. The return type must match the argument type.
5516</p>
5517
5518<h5>Semantics:</h5>
5519
5520<p>
5521The '<tt>llvm.cttz</tt>' intrinsic counts the trailing (least significant) zeros
5522in a variable. If the src == 0 then the result is the size in bits of the type
5523of src. For example, <tt>llvm.cttz(2) = 1</tt>.
5524</p>
5525</div>
5526
5527<!-- _______________________________________________________________________ -->
5528<div class="doc_subsubsection">
5529 <a name="int_part_select">'<tt>llvm.part.select.*</tt>' Intrinsic</a>
5530</div>
5531
5532<div class="doc_text">
5533
5534<h5>Syntax:</h5>
5535<p>This is an overloaded intrinsic. You can use <tt>llvm.part.select</tt>
5536on any integer bit width.
5537<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5538 declare i17 @llvm.part.select.i17 (i17 %val, i32 %loBit, i32 %hiBit)
5539 declare i29 @llvm.part.select.i29 (i29 %val, i32 %loBit, i32 %hiBit)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5540</pre>
5541
5542<h5>Overview:</h5>
5543<p>The '<tt>llvm.part.select</tt>' family of intrinsic functions selects a
5544range of bits from an integer value and returns them in the same bit width as
5545the original value.</p>
5546
5547<h5>Arguments:</h5>
5548<p>The first argument, <tt>%val</tt> and the result may be integer types of
5549any bit width but they must have the same bit width. The second and third
5550arguments must be <tt>i32</tt> type since they specify only a bit index.</p>
5551
5552<h5>Semantics:</h5>
5553<p>The operation of the '<tt>llvm.part.select</tt>' intrinsic has two modes
5554of operation: forwards and reverse. If <tt>%loBit</tt> is greater than
5555<tt>%hiBits</tt> then the intrinsic operates in reverse mode. Otherwise it
5556operates in forward mode.</p>
5557<p>In forward mode, this intrinsic is the equivalent of shifting <tt>%val</tt>
5558right by <tt>%loBit</tt> bits and then ANDing it with a mask with
5559only the <tt>%hiBit - %loBit</tt> bits set, as follows:</p>
5560<ol>
5561 <li>The <tt>%val</tt> is shifted right (LSHR) by the number of bits specified
5562 by <tt>%loBits</tt>. This normalizes the value to the low order bits.</li>
5563 <li>The <tt>%loBits</tt> value is subtracted from the <tt>%hiBits</tt> value
5564 to determine the number of bits to retain.</li>
5565 <li>A mask of the retained bits is created by shifting a -1 value.</li>
5566 <li>The mask is ANDed with <tt>%val</tt> to produce the result.
5567</ol>
5568<p>In reverse mode, a similar computation is made except that the bits are
5569returned in the reverse order. So, for example, if <tt>X</tt> has the value
5570<tt>i16 0x0ACF (101011001111)</tt> and we apply
5571<tt>part.select(i16 X, 8, 3)</tt> to it, we get back the value
5572<tt>i16 0x0026 (000000100110)</tt>.</p>
5573</div>
5574
5575<div class="doc_subsubsection">
5576 <a name="int_part_set">'<tt>llvm.part.set.*</tt>' Intrinsic</a>
5577</div>
5578
5579<div class="doc_text">
5580
5581<h5>Syntax:</h5>
5582<p>This is an overloaded intrinsic. You can use <tt>llvm.part.set</tt>
5583on any integer bit width.
5584<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5585 declare i17 @llvm.part.set.i17.i9 (i17 %val, i9 %repl, i32 %lo, i32 %hi)
5586 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]5587</pre>
5588
5589<h5>Overview:</h5>
5590<p>The '<tt>llvm.part.set</tt>' family of intrinsic functions replaces a range
5591of bits in an integer value with another integer value. It returns the integer
5592with the replaced bits.</p>
5593
5594<h5>Arguments:</h5>
5595<p>The first argument, <tt>%val</tt> and the result may be integer types of
5596any bit width but they must have the same bit width. <tt>%val</tt> is the value
5597whose bits will be replaced. The second argument, <tt>%repl</tt> may be an
5598integer of any bit width. The third and fourth arguments must be <tt>i32</tt>
5599type since they specify only a bit index.</p>
5600
5601<h5>Semantics:</h5>
5602<p>The operation of the '<tt>llvm.part.set</tt>' intrinsic has two modes
5603of operation: forwards and reverse. If <tt>%lo</tt> is greater than
5604<tt>%hi</tt> then the intrinsic operates in reverse mode. Otherwise it
5605operates in forward mode.</p>
5606<p>For both modes, the <tt>%repl</tt> value is prepared for use by either
5607truncating it down to the size of the replacement area or zero extending it
5608up to that size.</p>
5609<p>In forward mode, the bits between <tt>%lo</tt> and <tt>%hi</tt> (inclusive)
5610are replaced with corresponding bits from <tt>%repl</tt>. That is the 0th bit
5611in <tt>%repl</tt> replaces the <tt>%lo</tt>th bit in <tt>%val</tt> and etc. up
5612to the <tt>%hi</tt>th bit.
5613<p>In reverse mode, a similar computation is made except that the bits are
5614reversed. That is, the <tt>0</tt>th bit in <tt>%repl</tt> replaces the
5615<tt>%hi</tt> bit in <tt>%val</tt> and etc. down to the <tt>%lo</tt>th bit.
5616<h5>Examples:</h5>
5617<pre>
5618 llvm.part.set(0xFFFF, 0, 4, 7) -&gt; 0xFF0F
5619 llvm.part.set(0xFFFF, 0, 7, 4) -&gt; 0xFF0F
5620 llvm.part.set(0xFFFF, 1, 7, 4) -&gt; 0xFF8F
5621 llvm.part.set(0xFFFF, F, 8, 3) -&gt; 0xFFE7
5622 llvm.part.set(0xFFFF, 0, 3, 8) -&gt; 0xFE07
5623</pre>
5624</div>
5625
5626<!-- ======================================================================= -->
5627<div class="doc_subsection">
5628 <a name="int_debugger">Debugger Intrinsics</a>
5629</div>
5630
5631<div class="doc_text">
5632<p>
5633The LLVM debugger intrinsics (which all start with <tt>llvm.dbg.</tt> prefix),
5634are described in the <a
5635href="SourceLevelDebugging.html#format_common_intrinsics">LLVM Source Level
5636Debugging</a> document.
5637</p>
5638</div>
5639
5640
5641<!-- ======================================================================= -->
5642<div class="doc_subsection">
5643 <a name="int_eh">Exception Handling Intrinsics</a>
5644</div>
5645
5646<div class="doc_text">
5647<p> The LLVM exception handling intrinsics (which all start with
5648<tt>llvm.eh.</tt> prefix), are described in the <a
5649href="ExceptionHandling.html#format_common_intrinsics">LLVM Exception
5650Handling</a> document. </p>
5651</div>
5652
5653<!-- ======================================================================= -->
5654<div class="doc_subsection">
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5655 <a name="int_trampoline">Trampoline Intrinsic</a>
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5656</div>
5657
5658<div class="doc_text">
5659<p>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5660 This intrinsic makes it possible to excise one parameter, marked with
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5661 the <tt>nest</tt> attribute, from a function. The result is a callable
5662 function pointer lacking the nest parameter - the caller does not need
5663 to provide a value for it. Instead, the value to use is stored in
5664 advance in a "trampoline", a block of memory usually allocated
5665 on the stack, which also contains code to splice the nest value into the
5666 argument list. This is used to implement the GCC nested function address
5667 extension.
5668</p>
5669<p>
5670 For example, if the function is
5671 <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]5672 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]5673<pre>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5674 %tramp = alloca [10 x i8], align 4 ; size and alignment only correct for X86
5675 %tramp1 = getelementptr [10 x i8]* %tramp, i32 0, i32 0
5676 %p = call i8* @llvm.init.trampoline( i8* %tramp1, i8* bitcast (i32 (i8* nest , i32, i32)* @f to i8*), i8* %nval )
5677 %fp = bitcast i8* %p to i32 (i32, i32)*
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5678</pre>
Bill Wendlinge262b4c2007-09-22 09:23:55 +0000[diff] [blame]5679 <p>The call <tt>%val = call i32 %fp( i32 %x, i32 %y )</tt> is then equivalent
5680 to <tt>%val = call i32 %f( i8* %nval, i32 %x, i32 %y )</tt>.</p>
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5681</div>
5682
5683<!-- _______________________________________________________________________ -->
5684<div class="doc_subsubsection">
5685 <a name="int_it">'<tt>llvm.init.trampoline</tt>' Intrinsic</a>
5686</div>
5687<div class="doc_text">
5688<h5>Syntax:</h5>
5689<pre>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5690declare 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]5691</pre>
5692<h5>Overview:</h5>
5693<p>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5694 This fills the memory pointed to by <tt>tramp</tt> with code
5695 and returns a function pointer suitable for executing it.
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5696</p>
5697<h5>Arguments:</h5>
5698<p>
5699 The <tt>llvm.init.trampoline</tt> intrinsic takes three arguments, all
5700 pointers. The <tt>tramp</tt> argument must point to a sufficiently large
5701 and sufficiently aligned block of memory; this memory is written to by the
Duncan Sands35012212007-08-22 23:39:54 +0000[diff] [blame]5702 intrinsic. Note that the size and the alignment are target-specific - LLVM
5703 currently provides no portable way of determining them, so a front-end that
5704 generates this intrinsic needs to have some target-specific knowledge.
5705 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]5706</p>
5707<h5>Semantics:</h5>
5708<p>
5709 The block of memory pointed to by <tt>tramp</tt> is filled with target
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5710 dependent code, turning it into a function. A pointer to this function is
5711 returned, but needs to be bitcast to an
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5712 <a href="#int_trampoline">appropriate function pointer type</a>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5713 before being called. The new function's signature is the same as that of
5714 <tt>func</tt> with any arguments marked with the <tt>nest</tt> attribute
5715 removed. At most one such <tt>nest</tt> argument is allowed, and it must be
5716 of pointer type. Calling the new function is equivalent to calling
5717 <tt>func</tt> with the same argument list, but with <tt>nval</tt> used for the
5718 missing <tt>nest</tt> argument. If, after calling
5719 <tt>llvm.init.trampoline</tt>, the memory pointed to by <tt>tramp</tt> is
5720 modified, then the effect of any later call to the returned function pointer is
5721 undefined.
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5722</p>
5723</div>
5724
5725<!-- ======================================================================= -->
5726<div class="doc_subsection">
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]5727 <a name="int_atomics">Atomic Operations and Synchronization Intrinsics</a>
5728</div>
5729
5730<div class="doc_text">
5731<p>
5732 These intrinsic functions expand the "universal IR" of LLVM to represent
5733 hardware constructs for atomic operations and memory synchronization. This
5734 provides an interface to the hardware, not an interface to the programmer. It
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]5735 is aimed at a low enough level to allow any programming models or APIs
5736 (Application Programming Interfaces) which
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]5737 need atomic behaviors to map cleanly onto it. It is also modeled primarily on
5738 hardware behavior. Just as hardware provides a "universal IR" for source
5739 languages, it also provides a starting point for developing a "universal"
5740 atomic operation and synchronization IR.
5741</p>
5742<p>
5743 These do <em>not</em> form an API such as high-level threading libraries,
5744 software transaction memory systems, atomic primitives, and intrinsic
5745 functions as found in BSD, GNU libc, atomic_ops, APR, and other system and
5746 application libraries. The hardware interface provided by LLVM should allow
5747 a clean implementation of all of these APIs and parallel programming models.
5748 No one model or paradigm should be selected above others unless the hardware
5749 itself ubiquitously does so.
5750
5751</p>
5752</div>
5753
5754<!-- _______________________________________________________________________ -->
5755<div class="doc_subsubsection">
5756 <a name="int_memory_barrier">'<tt>llvm.memory.barrier</tt>' Intrinsic</a>
5757</div>
5758<div class="doc_text">
5759<h5>Syntax:</h5>
5760<pre>
5761declare void @llvm.memory.barrier( i1 &lt;ll&gt;, i1 &lt;ls&gt;, i1 &lt;sl&gt;, i1 &lt;ss&gt;,
5762i1 &lt;device&gt; )
5763
5764</pre>
5765<h5>Overview:</h5>
5766<p>
5767 The <tt>llvm.memory.barrier</tt> intrinsic guarantees ordering between
5768 specific pairs of memory access types.
5769</p>
5770<h5>Arguments:</h5>
5771<p>
5772 The <tt>llvm.memory.barrier</tt> intrinsic requires five boolean arguments.
5773 The first four arguments enables a specific barrier as listed below. The fith
5774 argument specifies that the barrier applies to io or device or uncached memory.
5775
5776</p>
5777 <ul>
5778 <li><tt>ll</tt>: load-load barrier</li>
5779 <li><tt>ls</tt>: load-store barrier</li>
5780 <li><tt>sl</tt>: store-load barrier</li>
5781 <li><tt>ss</tt>: store-store barrier</li>
5782 <li><tt>device</tt>: barrier applies to device and uncached memory also.
5783 </ul>
5784<h5>Semantics:</h5>
5785<p>
5786 This intrinsic causes the system to enforce some ordering constraints upon
5787 the loads and stores of the program. This barrier does not indicate
5788 <em>when</em> any events will occur, it only enforces an <em>order</em> in
5789 which they occur. For any of the specified pairs of load and store operations
5790 (f.ex. load-load, or store-load), all of the first operations preceding the
5791 barrier will complete before any of the second operations succeeding the
5792 barrier begin. Specifically the semantics for each pairing is as follows:
5793</p>
5794 <ul>
5795 <li><tt>ll</tt>: All loads before the barrier must complete before any load
5796 after the barrier begins.</li>
5797
5798 <li><tt>ls</tt>: All loads before the barrier must complete before any
5799 store after the barrier begins.</li>
5800 <li><tt>ss</tt>: All stores before the barrier must complete before any
5801 store after the barrier begins.</li>
5802 <li><tt>sl</tt>: All stores before the barrier must complete before any
5803 load after the barrier begins.</li>
5804 </ul>
5805<p>
5806 These semantics are applied with a logical "and" behavior when more than one
5807 is enabled in a single memory barrier intrinsic.
5808</p>
5809<p>
5810 Backends may implement stronger barriers than those requested when they do not
5811 support as fine grained a barrier as requested. Some architectures do not
5812 need all types of barriers and on such architectures, these become noops.
5813</p>
5814<h5>Example:</h5>
5815<pre>
5816%ptr = malloc i32
5817 store i32 4, %ptr
5818
5819%result1 = load i32* %ptr <i>; yields {i32}:result1 = 4</i>
5820 call void @llvm.memory.barrier( i1 false, i1 true, i1 false, i1 false )
5821 <i>; guarantee the above finishes</i>
5822 store i32 8, %ptr <i>; before this begins</i>
5823</pre>
5824</div>
5825
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]5826<!-- _______________________________________________________________________ -->
5827<div class="doc_subsubsection">
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]5828 <a name="int_atomic_cmp_swap">'<tt>llvm.atomic.cmp.swap.*</tt>' Intrinsic</a>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]5829</div>
5830<div class="doc_text">
5831<h5>Syntax:</h5>
5832<p>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]5833 This is an overloaded intrinsic. You can use <tt>llvm.atomic.cmp.swap</tt> on
5834 any integer bit width and for different address spaces. Not all targets
5835 support all bit widths however.</p>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]5836
5837<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]5838declare i8 @llvm.atomic.cmp.swap.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;cmp&gt;, i8 &lt;val&gt; )
5839declare i16 @llvm.atomic.cmp.swap.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;cmp&gt;, i16 &lt;val&gt; )
5840declare i32 @llvm.atomic.cmp.swap.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;cmp&gt;, i32 &lt;val&gt; )
5841declare 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]5842
5843</pre>
5844<h5>Overview:</h5>
5845<p>
5846 This loads a value in memory and compares it to a given value. If they are
5847 equal, it stores a new value into the memory.
5848</p>
5849<h5>Arguments:</h5>
5850<p>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]5851 The <tt>llvm.atomic.cmp.swap</tt> intrinsic takes three arguments. The result as
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]5852 well as both <tt>cmp</tt> and <tt>val</tt> must be integer values with the
5853 same bit width. The <tt>ptr</tt> argument must be a pointer to a value of
5854 this integer type. While any bit width integer may be used, targets may only
5855 lower representations they support in hardware.
5856
5857</p>
5858<h5>Semantics:</h5>
5859<p>
5860 This entire intrinsic must be executed atomically. It first loads the value
5861 in memory pointed to by <tt>ptr</tt> and compares it with the value
5862 <tt>cmp</tt>. If they are equal, <tt>val</tt> is stored into the memory. The
5863 loaded value is yielded in all cases. This provides the equivalent of an
5864 atomic compare-and-swap operation within the SSA framework.
5865</p>
5866<h5>Examples:</h5>
5867
5868<pre>
5869%ptr = malloc i32
5870 store i32 4, %ptr
5871
5872%val1 = add i32 4, 4
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]5873%result1 = call i32 @llvm.atomic.cmp.swap.i32.p0i32( i32* %ptr, i32 4, %val1 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]5874 <i>; yields {i32}:result1 = 4</i>
5875%stored1 = icmp eq i32 %result1, 4 <i>; yields {i1}:stored1 = true</i>
5876%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 8</i>
5877
5878%val2 = add i32 1, 1
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]5879%result2 = call i32 @llvm.atomic.cmp.swap.i32.p0i32( i32* %ptr, i32 5, %val2 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]5880 <i>; yields {i32}:result2 = 8</i>
5881%stored2 = icmp eq i32 %result2, 5 <i>; yields {i1}:stored2 = false</i>
5882
5883%memval2 = load i32* %ptr <i>; yields {i32}:memval2 = 8</i>
5884</pre>
5885</div>
5886
5887<!-- _______________________________________________________________________ -->
5888<div class="doc_subsubsection">
5889 <a name="int_atomic_swap">'<tt>llvm.atomic.swap.*</tt>' Intrinsic</a>
5890</div>
5891<div class="doc_text">
5892<h5>Syntax:</h5>
5893
5894<p>
5895 This is an overloaded intrinsic. You can use <tt>llvm.atomic.swap</tt> on any
5896 integer bit width. Not all targets support all bit widths however.</p>
5897<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]5898declare i8 @llvm.atomic.swap.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;val&gt; )
5899declare i16 @llvm.atomic.swap.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;val&gt; )
5900declare i32 @llvm.atomic.swap.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;val&gt; )
5901declare i64 @llvm.atomic.swap.i64.p0i64( i64* &lt;ptr&gt;, i64 &lt;val&gt; )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]5902
5903</pre>
5904<h5>Overview:</h5>
5905<p>
5906 This intrinsic loads the value stored in memory at <tt>ptr</tt> and yields
5907 the value from memory. It then stores the value in <tt>val</tt> in the memory
5908 at <tt>ptr</tt>.
5909</p>
5910<h5>Arguments:</h5>
5911
5912<p>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]5913 The <tt>llvm.atomic.swap</tt> intrinsic takes two arguments. Both the
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]5914 <tt>val</tt> argument and the result must be integers of the same bit width.
5915 The first argument, <tt>ptr</tt>, must be a pointer to a value of this
5916 integer type. The targets may only lower integer representations they
5917 support.
5918</p>
5919<h5>Semantics:</h5>
5920<p>
5921 This intrinsic loads the value pointed to by <tt>ptr</tt>, yields it, and
5922 stores <tt>val</tt> back into <tt>ptr</tt> atomically. This provides the
5923 equivalent of an atomic swap operation within the SSA framework.
5924
5925</p>
5926<h5>Examples:</h5>
5927<pre>
5928%ptr = malloc i32
5929 store i32 4, %ptr
5930
5931%val1 = add i32 4, 4
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]5932%result1 = call i32 @llvm.atomic.swap.i32.p0i32( i32* %ptr, i32 %val1 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]5933 <i>; yields {i32}:result1 = 4</i>
5934%stored1 = icmp eq i32 %result1, 4 <i>; yields {i1}:stored1 = true</i>
5935%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 8</i>
5936
5937%val2 = add i32 1, 1
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]5938%result2 = call i32 @llvm.atomic.swap.i32.p0i32( i32* %ptr, i32 %val2 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]5939 <i>; yields {i32}:result2 = 8</i>
5940
5941%stored2 = icmp eq i32 %result2, 8 <i>; yields {i1}:stored2 = true</i>
5942%memval2 = load i32* %ptr <i>; yields {i32}:memval2 = 2</i>
5943</pre>
5944</div>
5945
5946<!-- _______________________________________________________________________ -->
5947<div class="doc_subsubsection">
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]5948 <a name="int_atomic_load_add">'<tt>llvm.atomic.load.add.*</tt>' Intrinsic</a>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]5949
5950</div>
5951<div class="doc_text">
5952<h5>Syntax:</h5>
5953<p>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]5954 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]5955 integer bit width. Not all targets support all bit widths however.</p>
5956<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]5957declare i8 @llvm.atomic.load.add.i8..p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
5958declare i16 @llvm.atomic.load.add.i16..p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
5959declare i32 @llvm.atomic.load.add.i32..p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
5960declare 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]5961
5962</pre>
5963<h5>Overview:</h5>
5964<p>
5965 This intrinsic adds <tt>delta</tt> to the value stored in memory at
5966 <tt>ptr</tt>. It yields the original value at <tt>ptr</tt>.
5967</p>
5968<h5>Arguments:</h5>
5969<p>
5970
5971 The intrinsic takes two arguments, the first a pointer to an integer value
5972 and the second an integer value. The result is also an integer value. These
5973 integer types can have any bit width, but they must all have the same bit
5974 width. The targets may only lower integer representations they support.
5975</p>
5976<h5>Semantics:</h5>
5977<p>
5978 This intrinsic does a series of operations atomically. It first loads the
5979 value stored at <tt>ptr</tt>. It then adds <tt>delta</tt>, stores the result
5980 to <tt>ptr</tt>. It yields the original value stored at <tt>ptr</tt>.
5981</p>
5982
5983<h5>Examples:</h5>
5984<pre>
5985%ptr = malloc i32
5986 store i32 4, %ptr
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]5987%result1 = call i32 @llvm.atomic.load.add.i32.p0i32( i32* %ptr, i32 4 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]5988 <i>; yields {i32}:result1 = 4</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]5989%result2 = call i32 @llvm.atomic.load.add.i32.p0i32( i32* %ptr, i32 2 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]5990 <i>; yields {i32}:result2 = 8</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]5991%result3 = call i32 @llvm.atomic.load.add.i32.p0i32( i32* %ptr, i32 5 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]5992 <i>; yields {i32}:result3 = 10</i>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]5993%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 15</i>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]5994</pre>
5995</div>
5996
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]5997<!-- _______________________________________________________________________ -->
5998<div class="doc_subsubsection">
5999 <a name="int_atomic_load_sub">'<tt>llvm.atomic.load.sub.*</tt>' Intrinsic</a>
6000
6001</div>
6002<div class="doc_text">
6003<h5>Syntax:</h5>
6004<p>
6005 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]6006 any integer bit width and for different address spaces. Not all targets
6007 support all bit widths however.</p>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6008<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6009declare i8 @llvm.atomic.load.sub.i8.p0i32( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6010declare i16 @llvm.atomic.load.sub.i16.p0i32( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6011declare i32 @llvm.atomic.load.sub.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6012declare 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]6013
6014</pre>
6015<h5>Overview:</h5>
6016<p>
6017 This intrinsic subtracts <tt>delta</tt> to the value stored in memory at
6018 <tt>ptr</tt>. It yields the original value at <tt>ptr</tt>.
6019</p>
6020<h5>Arguments:</h5>
6021<p>
6022
6023 The intrinsic takes two arguments, the first a pointer to an integer value
6024 and the second an integer value. The result is also an integer value. These
6025 integer types can have any bit width, but they must all have the same bit
6026 width. The targets may only lower integer representations they support.
6027</p>
6028<h5>Semantics:</h5>
6029<p>
6030 This intrinsic does a series of operations atomically. It first loads the
6031 value stored at <tt>ptr</tt>. It then subtracts <tt>delta</tt>, stores the
6032 result to <tt>ptr</tt>. It yields the original value stored at <tt>ptr</tt>.
6033</p>
6034
6035<h5>Examples:</h5>
6036<pre>
6037%ptr = malloc i32
6038 store i32 8, %ptr
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6039%result1 = call i32 @llvm.atomic.load.sub.i32.p0i32( i32* %ptr, i32 4 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6040 <i>; yields {i32}:result1 = 8</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6041%result2 = call i32 @llvm.atomic.load.sub.i32.p0i32( i32* %ptr, i32 2 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6042 <i>; yields {i32}:result2 = 4</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6043%result3 = call i32 @llvm.atomic.load.sub.i32.p0i32( i32* %ptr, i32 5 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6044 <i>; yields {i32}:result3 = 2</i>
6045%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = -3</i>
6046</pre>
6047</div>
6048
6049<!-- _______________________________________________________________________ -->
6050<div class="doc_subsubsection">
6051 <a name="int_atomic_load_and">'<tt>llvm.atomic.load.and.*</tt>' Intrinsic</a><br>
6052 <a name="int_atomic_load_nand">'<tt>llvm.atomic.load.nand.*</tt>' Intrinsic</a><br>
6053 <a name="int_atomic_load_or">'<tt>llvm.atomic.load.or.*</tt>' Intrinsic</a><br>
6054 <a name="int_atomic_load_xor">'<tt>llvm.atomic.load.xor.*</tt>' Intrinsic</a><br>
6055
6056</div>
6057<div class="doc_text">
6058<h5>Syntax:</h5>
6059<p>
6060 These are overloaded intrinsics. You can use <tt>llvm.atomic.load_and</tt>,
6061 <tt>llvm.atomic.load_nand</tt>, <tt>llvm.atomic.load_or</tt>, and
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6062 <tt>llvm.atomic.load_xor</tt> on any integer bit width and for different
6063 address spaces. Not all targets support all bit widths however.</p>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6064<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6065declare i8 @llvm.atomic.load.and.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6066declare i16 @llvm.atomic.load.and.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6067declare i32 @llvm.atomic.load.and.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6068declare 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]6069
6070</pre>
6071
6072<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6073declare i8 @llvm.atomic.load.or.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6074declare i16 @llvm.atomic.load.or.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6075declare i32 @llvm.atomic.load.or.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6076declare 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]6077
6078</pre>
6079
6080<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6081declare i8 @llvm.atomic.load.nand.i8.p0i32( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6082declare i16 @llvm.atomic.load.nand.i16.p0i32( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6083declare i32 @llvm.atomic.load.nand.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6084declare 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]6085
6086</pre>
6087
6088<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6089declare i8 @llvm.atomic.load.xor.i8.p0i32( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6090declare i16 @llvm.atomic.load.xor.i16.p0i32( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6091declare i32 @llvm.atomic.load.xor.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6092declare 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]6093
6094</pre>
6095<h5>Overview:</h5>
6096<p>
6097 These intrinsics bitwise the operation (and, nand, or, xor) <tt>delta</tt> to
6098 the value stored in memory at <tt>ptr</tt>. It yields the original value
6099 at <tt>ptr</tt>.
6100</p>
6101<h5>Arguments:</h5>
6102<p>
6103
6104 These intrinsics take two arguments, the first a pointer to an integer value
6105 and the second an integer value. The result is also an integer value. These
6106 integer types can have any bit width, but they must all have the same bit
6107 width. The targets may only lower integer representations they support.
6108</p>
6109<h5>Semantics:</h5>
6110<p>
6111 These intrinsics does a series of operations atomically. They first load the
6112 value stored at <tt>ptr</tt>. They then do the bitwise operation
6113 <tt>delta</tt>, store the result to <tt>ptr</tt>. They yield the original
6114 value stored at <tt>ptr</tt>.
6115</p>
6116
6117<h5>Examples:</h5>
6118<pre>
6119%ptr = malloc i32
6120 store i32 0x0F0F, %ptr
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6121%result0 = call i32 @llvm.atomic.load.nand.i32.p0i32( i32* %ptr, i32 0xFF )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6122 <i>; yields {i32}:result0 = 0x0F0F</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6123%result1 = call i32 @llvm.atomic.load.and.i32.p0i32( i32* %ptr, i32 0xFF )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6124 <i>; yields {i32}:result1 = 0xFFFFFFF0</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6125%result2 = call i32 @llvm.atomic.load.or.i32.p0i32( i32* %ptr, i32 0F )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6126 <i>; yields {i32}:result2 = 0xF0</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6127%result3 = call i32 @llvm.atomic.load.xor.i32.p0i32( i32* %ptr, i32 0F )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6128 <i>; yields {i32}:result3 = FF</i>
6129%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = F0</i>
6130</pre>
6131</div>
6132
6133
6134<!-- _______________________________________________________________________ -->
6135<div class="doc_subsubsection">
6136 <a name="int_atomic_load_max">'<tt>llvm.atomic.load.max.*</tt>' Intrinsic</a><br>
6137 <a name="int_atomic_load_min">'<tt>llvm.atomic.load.min.*</tt>' Intrinsic</a><br>
6138 <a name="int_atomic_load_umax">'<tt>llvm.atomic.load.umax.*</tt>' Intrinsic</a><br>
6139 <a name="int_atomic_load_umin">'<tt>llvm.atomic.load.umin.*</tt>' Intrinsic</a><br>
6140
6141</div>
6142<div class="doc_text">
6143<h5>Syntax:</h5>
6144<p>
6145 These are overloaded intrinsics. You can use <tt>llvm.atomic.load_max</tt>,
6146 <tt>llvm.atomic.load_min</tt>, <tt>llvm.atomic.load_umax</tt>, and
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6147 <tt>llvm.atomic.load_umin</tt> on any integer bit width and for different
6148 address spaces. Not all targets
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6149 support all bit widths however.</p>
6150<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6151declare i8 @llvm.atomic.load.max.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6152declare i16 @llvm.atomic.load.max.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6153declare i32 @llvm.atomic.load.max.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6154declare 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]6155
6156</pre>
6157
6158<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6159declare i8 @llvm.atomic.load.min.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6160declare i16 @llvm.atomic.load.min.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6161declare i32 @llvm.atomic.load.min.i32..p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6162declare 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]6163
6164</pre>
6165
6166<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6167declare i8 @llvm.atomic.load.umax.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6168declare i16 @llvm.atomic.load.umax.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6169declare i32 @llvm.atomic.load.umax.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6170declare 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]6171
6172</pre>
6173
6174<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6175declare i8 @llvm.atomic.load.umin.i8..p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6176declare i16 @llvm.atomic.load.umin.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6177declare i32 @llvm.atomic.load.umin.i32..p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6178declare 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]6179
6180</pre>
6181<h5>Overview:</h5>
6182<p>
6183 These intrinsics takes the signed or unsigned minimum or maximum of
6184 <tt>delta</tt> and the value stored in memory at <tt>ptr</tt>. It yields the
6185 original value at <tt>ptr</tt>.
6186</p>
6187<h5>Arguments:</h5>
6188<p>
6189
6190 These intrinsics take two arguments, the first a pointer to an integer value
6191 and the second an integer value. The result is also an integer value. These
6192 integer types can have any bit width, but they must all have the same bit
6193 width. The targets may only lower integer representations they support.
6194</p>
6195<h5>Semantics:</h5>
6196<p>
6197 These intrinsics does a series of operations atomically. They first load the
6198 value stored at <tt>ptr</tt>. They then do the signed or unsigned min or max
6199 <tt>delta</tt> and the value, store the result to <tt>ptr</tt>. They yield
6200 the original value stored at <tt>ptr</tt>.
6201</p>
6202
6203<h5>Examples:</h5>
6204<pre>
6205%ptr = malloc i32
6206 store i32 7, %ptr
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6207%result0 = call i32 @llvm.atomic.load.min.i32.p0i32( i32* %ptr, i32 -2 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6208 <i>; yields {i32}:result0 = 7</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6209%result1 = call i32 @llvm.atomic.load.max.i32.p0i32( i32* %ptr, i32 8 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6210 <i>; yields {i32}:result1 = -2</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6211%result2 = call i32 @llvm.atomic.load.umin.i32.p0i32( i32* %ptr, i32 10 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6212 <i>; yields {i32}:result2 = 8</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6213%result3 = call i32 @llvm.atomic.load.umax.i32.p0i32( i32* %ptr, i32 30 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6214 <i>; yields {i32}:result3 = 8</i>
6215%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 30</i>
6216</pre>
6217</div>
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]6218
6219<!-- ======================================================================= -->
6220<div class="doc_subsection">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6221 <a name="int_general">General Intrinsics</a>
6222</div>
6223
6224<div class="doc_text">
6225<p> This class of intrinsics is designed to be generic and has
6226no specific purpose. </p>
6227</div>
6228
6229<!-- _______________________________________________________________________ -->
6230<div class="doc_subsubsection">
6231 <a name="int_var_annotation">'<tt>llvm.var.annotation</tt>' Intrinsic</a>
6232</div>
6233
6234<div class="doc_text">
6235
6236<h5>Syntax:</h5>
6237<pre>
6238 declare void @llvm.var.annotation(i8* &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
6239</pre>
6240
6241<h5>Overview:</h5>
6242
6243<p>
6244The '<tt>llvm.var.annotation</tt>' intrinsic
6245</p>
6246
6247<h5>Arguments:</h5>
6248
6249<p>
6250The first argument is a pointer to a value, the second is a pointer to a
6251global string, the third is a pointer to a global string which is the source
6252file name, and the last argument is the line number.
6253</p>
6254
6255<h5>Semantics:</h5>
6256
6257<p>
Anton Korobeynikove6e764f2008-01-15 22:31:34 +0000[diff] [blame]6258This intrinsic allows annotation of local variables with arbitrary strings.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6259This can be useful for special purpose optimizations that want to look for these
Anton Korobeynikove6e764f2008-01-15 22:31:34 +0000[diff] [blame]6260annotations. These have no other defined use, they are ignored by code
6261generation and optimization.
6262</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6263</div>
6264
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]6265<!-- _______________________________________________________________________ -->
6266<div class="doc_subsubsection">
Tanya Lattnerc9869b12007-09-21 23:57:59 +0000[diff] [blame]6267 <a name="int_annotation">'<tt>llvm.annotation.*</tt>' Intrinsic</a>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]6268</div>
6269
6270<div class="doc_text">
6271
6272<h5>Syntax:</h5>
Tanya Lattnere545be72007-09-21 23:56:27 +0000[diff] [blame]6273<p>This is an overloaded intrinsic. You can use '<tt>llvm.annotation</tt>' on
6274any integer bit width.
6275</p>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]6276<pre>
Tanya Lattner09161fe2007-09-22 00:03:01 +0000[diff] [blame]6277 declare i8 @llvm.annotation.i8(i8 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
6278 declare i16 @llvm.annotation.i16(i16 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
6279 declare i32 @llvm.annotation.i32(i32 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
6280 declare i64 @llvm.annotation.i64(i64 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
6281 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]6282</pre>
6283
6284<h5>Overview:</h5>
Tanya Lattnere545be72007-09-21 23:56:27 +0000[diff] [blame]6285
6286<p>
6287The '<tt>llvm.annotation</tt>' intrinsic.
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]6288</p>
6289
6290<h5>Arguments:</h5>
6291
6292<p>
6293The first argument is an integer value (result of some expression),
6294the second is a pointer to a global string, the third is a pointer to a global
6295string which is the source file name, and the last argument is the line number.
Tanya Lattnere545be72007-09-21 23:56:27 +0000[diff] [blame]6296It returns the value of the first argument.
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]6297</p>
6298
6299<h5>Semantics:</h5>
6300
6301<p>
6302This intrinsic allows annotations to be put on arbitrary expressions
6303with arbitrary strings. This can be useful for special purpose optimizations
6304that want to look for these annotations. These have no other defined use, they
6305are ignored by code generation and optimization.
6306</div>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6307
Anton Korobeynikove6e764f2008-01-15 22:31:34 +0000[diff] [blame]6308<!-- _______________________________________________________________________ -->
6309<div class="doc_subsubsection">
6310 <a name="int_trap">'<tt>llvm.trap</tt>' Intrinsic</a>
6311</div>
6312
6313<div class="doc_text">
6314
6315<h5>Syntax:</h5>
6316<pre>
6317 declare void @llvm.trap()
6318</pre>
6319
6320<h5>Overview:</h5>
6321
6322<p>
6323The '<tt>llvm.trap</tt>' intrinsic
6324</p>
6325
6326<h5>Arguments:</h5>
6327
6328<p>
6329None
6330</p>
6331
6332<h5>Semantics:</h5>
6333
6334<p>
6335This intrinsics is lowered to the target dependent trap instruction. If the
6336target does not have a trap instruction, this intrinsic will be lowered to the
6337call of the abort() function.
6338</p>
6339</div>
6340
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6341<!-- *********************************************************************** -->
6342<hr>
6343<address>
6344 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
6345 src="http://jigsaw.w3.org/css-validator/images/vcss" alt="Valid CSS!"></a>
6346 <a href="http://validator.w3.org/check/referer"><img
Chris Lattner08497ce2008-01-04 04:33:49 +0000[diff] [blame]6347 src="http://www.w3.org/Icons/valid-html401" alt="Valid HTML 4.01!"></a>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6348
6349 <a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
6350 <a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br>
6351 Last modified: $Date$
6352</address>
Chris Lattner08497ce2008-01-04 04:33:49 +0000[diff] [blame]6353
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6354</body>
6355</html>