Gitiles
Code Review Sign In
gerrit-public.fairphone.software / fp2-dev / platform / external / llvm / dfaabbaf60826299aefc5d535c6f6bc6349c63a7 / . / docs / LangRef.html
blob: 221cc492ebbc881ae7b04194e419c0e1c3672954 [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>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]27 <li><a href="#aliasstructure">Aliases</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]28 <li><a href="#paramattrs">Parameter Attributes</a></li>
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]29 <li><a href="#fnattrs">Function Attributes</a></li>
Gordon Henriksen13fe5e32007-12-10 03:18:06 +0000[diff] [blame]30 <li><a href="#gc">Garbage Collector Names</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]31 <li><a href="#moduleasm">Module-Level Inline Assembly</a></li>
32 <li><a href="#datalayout">Data Layout</a></li>
33 </ol>
34 </li>
35 <li><a href="#typesystem">Type System</a>
36 <ol>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]37 <li><a href="#t_classifications">Type Classifications</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]38 <li><a href="#t_primitive">Primitive Types</a>
39 <ol>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]40 <li><a href="#t_floating">Floating Point Types</a></li>
41 <li><a href="#t_void">Void Type</a></li>
42 <li><a href="#t_label">Label Type</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]43 </ol>
44 </li>
45 <li><a href="#t_derived">Derived Types</a>
46 <ol>
Chris Lattner251ab812007-12-18 06:18:21 +0000[diff] [blame]47 <li><a href="#t_integer">Integer Type</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]48 <li><a href="#t_array">Array Type</a></li>
49 <li><a href="#t_function">Function Type</a></li>
50 <li><a href="#t_pointer">Pointer Type</a></li>
51 <li><a href="#t_struct">Structure Type</a></li>
52 <li><a href="#t_pstruct">Packed Structure Type</a></li>
53 <li><a href="#t_vector">Vector Type</a></li>
54 <li><a href="#t_opaque">Opaque Type</a></li>
55 </ol>
56 </li>
57 </ol>
58 </li>
59 <li><a href="#constants">Constants</a>
60 <ol>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]61 <li><a href="#simpleconstants">Simple Constants</a></li>
62 <li><a href="#aggregateconstants">Aggregate Constants</a></li>
63 <li><a href="#globalconstants">Global Variable and Function Addresses</a></li>
64 <li><a href="#undefvalues">Undefined Values</a></li>
65 <li><a href="#constantexprs">Constant Expressions</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]66 </ol>
67 </li>
68 <li><a href="#othervalues">Other Values</a>
69 <ol>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]70 <li><a href="#inlineasm">Inline Assembler Expressions</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]71 </ol>
72 </li>
73 <li><a href="#instref">Instruction Reference</a>
74 <ol>
75 <li><a href="#terminators">Terminator Instructions</a>
76 <ol>
77 <li><a href="#i_ret">'<tt>ret</tt>' Instruction</a></li>
78 <li><a href="#i_br">'<tt>br</tt>' Instruction</a></li>
79 <li><a href="#i_switch">'<tt>switch</tt>' Instruction</a></li>
80 <li><a href="#i_invoke">'<tt>invoke</tt>' Instruction</a></li>
81 <li><a href="#i_unwind">'<tt>unwind</tt>' Instruction</a></li>
82 <li><a href="#i_unreachable">'<tt>unreachable</tt>' Instruction</a></li>
83 </ol>
84 </li>
85 <li><a href="#binaryops">Binary Operations</a>
86 <ol>
87 <li><a href="#i_add">'<tt>add</tt>' Instruction</a></li>
88 <li><a href="#i_sub">'<tt>sub</tt>' Instruction</a></li>
89 <li><a href="#i_mul">'<tt>mul</tt>' Instruction</a></li>
90 <li><a href="#i_udiv">'<tt>udiv</tt>' Instruction</a></li>
91 <li><a href="#i_sdiv">'<tt>sdiv</tt>' Instruction</a></li>
92 <li><a href="#i_fdiv">'<tt>fdiv</tt>' Instruction</a></li>
93 <li><a href="#i_urem">'<tt>urem</tt>' Instruction</a></li>
94 <li><a href="#i_srem">'<tt>srem</tt>' Instruction</a></li>
95 <li><a href="#i_frem">'<tt>frem</tt>' Instruction</a></li>
96 </ol>
97 </li>
98 <li><a href="#bitwiseops">Bitwise Binary Operations</a>
99 <ol>
100 <li><a href="#i_shl">'<tt>shl</tt>' Instruction</a></li>
101 <li><a href="#i_lshr">'<tt>lshr</tt>' Instruction</a></li>
102 <li><a href="#i_ashr">'<tt>ashr</tt>' Instruction</a></li>
103 <li><a href="#i_and">'<tt>and</tt>' Instruction</a></li>
104 <li><a href="#i_or">'<tt>or</tt>' Instruction</a></li>
105 <li><a href="#i_xor">'<tt>xor</tt>' Instruction</a></li>
106 </ol>
107 </li>
108 <li><a href="#vectorops">Vector Operations</a>
109 <ol>
110 <li><a href="#i_extractelement">'<tt>extractelement</tt>' Instruction</a></li>
111 <li><a href="#i_insertelement">'<tt>insertelement</tt>' Instruction</a></li>
112 <li><a href="#i_shufflevector">'<tt>shufflevector</tt>' Instruction</a></li>
113 </ol>
114 </li>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]115 <li><a href="#aggregateops">Aggregate Operations</a>
116 <ol>
117 <li><a href="#i_extractvalue">'<tt>extractvalue</tt>' Instruction</a></li>
118 <li><a href="#i_insertvalue">'<tt>insertvalue</tt>' Instruction</a></li>
119 </ol>
120 </li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]121 <li><a href="#memoryops">Memory Access and Addressing Operations</a>
122 <ol>
123 <li><a href="#i_malloc">'<tt>malloc</tt>' Instruction</a></li>
124 <li><a href="#i_free">'<tt>free</tt>' Instruction</a></li>
125 <li><a href="#i_alloca">'<tt>alloca</tt>' Instruction</a></li>
126 <li><a href="#i_load">'<tt>load</tt>' Instruction</a></li>
127 <li><a href="#i_store">'<tt>store</tt>' Instruction</a></li>
128 <li><a href="#i_getelementptr">'<tt>getelementptr</tt>' Instruction</a></li>
129 </ol>
130 </li>
131 <li><a href="#convertops">Conversion Operations</a>
132 <ol>
133 <li><a href="#i_trunc">'<tt>trunc .. to</tt>' Instruction</a></li>
134 <li><a href="#i_zext">'<tt>zext .. to</tt>' Instruction</a></li>
135 <li><a href="#i_sext">'<tt>sext .. to</tt>' Instruction</a></li>
136 <li><a href="#i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a></li>
137 <li><a href="#i_fpext">'<tt>fpext .. to</tt>' Instruction</a></li>
138 <li><a href="#i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a></li>
139 <li><a href="#i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a></li>
140 <li><a href="#i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a></li>
141 <li><a href="#i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a></li>
142 <li><a href="#i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a></li>
143 <li><a href="#i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a></li>
144 <li><a href="#i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a></li>
145 </ol>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]146 </li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]147 <li><a href="#otherops">Other Operations</a>
148 <ol>
149 <li><a href="#i_icmp">'<tt>icmp</tt>' Instruction</a></li>
150 <li><a href="#i_fcmp">'<tt>fcmp</tt>' Instruction</a></li>
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]151 <li><a href="#i_vicmp">'<tt>vicmp</tt>' Instruction</a></li>
152 <li><a href="#i_vfcmp">'<tt>vfcmp</tt>' Instruction</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]153 <li><a href="#i_phi">'<tt>phi</tt>' Instruction</a></li>
154 <li><a href="#i_select">'<tt>select</tt>' Instruction</a></li>
155 <li><a href="#i_call">'<tt>call</tt>' Instruction</a></li>
156 <li><a href="#i_va_arg">'<tt>va_arg</tt>' Instruction</a></li>
157 </ol>
158 </li>
159 </ol>
160 </li>
161 <li><a href="#intrinsics">Intrinsic Functions</a>
162 <ol>
163 <li><a href="#int_varargs">Variable Argument Handling Intrinsics</a>
164 <ol>
165 <li><a href="#int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a></li>
166 <li><a href="#int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a></li>
167 <li><a href="#int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a></li>
168 </ol>
169 </li>
170 <li><a href="#int_gc">Accurate Garbage Collection Intrinsics</a>
171 <ol>
172 <li><a href="#int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a></li>
173 <li><a href="#int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a></li>
174 <li><a href="#int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a></li>
175 </ol>
176 </li>
177 <li><a href="#int_codegen">Code Generator Intrinsics</a>
178 <ol>
179 <li><a href="#int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a></li>
180 <li><a href="#int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a></li>
181 <li><a href="#int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a></li>
182 <li><a href="#int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a></li>
183 <li><a href="#int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a></li>
184 <li><a href="#int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a></li>
185 <li><a href="#int_readcyclecounter"><tt>llvm.readcyclecounter</tt>' Intrinsic</a></li>
186 </ol>
187 </li>
188 <li><a href="#int_libc">Standard C Library Intrinsics</a>
189 <ol>
190 <li><a href="#int_memcpy">'<tt>llvm.memcpy.*</tt>' Intrinsic</a></li>
191 <li><a href="#int_memmove">'<tt>llvm.memmove.*</tt>' Intrinsic</a></li>
192 <li><a href="#int_memset">'<tt>llvm.memset.*</tt>' Intrinsic</a></li>
193 <li><a href="#int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a></li>
194 <li><a href="#int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a></li>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]195 <li><a href="#int_sin">'<tt>llvm.sin.*</tt>' Intrinsic</a></li>
196 <li><a href="#int_cos">'<tt>llvm.cos.*</tt>' Intrinsic</a></li>
197 <li><a href="#int_pow">'<tt>llvm.pow.*</tt>' Intrinsic</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]198 </ol>
199 </li>
200 <li><a href="#int_manip">Bit Manipulation Intrinsics</a>
201 <ol>
202 <li><a href="#int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a></li>
203 <li><a href="#int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic </a></li>
204 <li><a href="#int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic </a></li>
205 <li><a href="#int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic </a></li>
206 <li><a href="#int_part_select">'<tt>llvm.part.select.*</tt>' Intrinsic </a></li>
207 <li><a href="#int_part_set">'<tt>llvm.part.set.*</tt>' Intrinsic </a></li>
208 </ol>
209 </li>
210 <li><a href="#int_debugger">Debugger intrinsics</a></li>
211 <li><a href="#int_eh">Exception Handling intrinsics</a></li>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]212 <li><a href="#int_trampoline">Trampoline Intrinsic</a>
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]213 <ol>
214 <li><a href="#int_it">'<tt>llvm.init.trampoline</tt>' Intrinsic</a></li>
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]215 </ol>
216 </li>
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]217 <li><a href="#int_atomics">Atomic intrinsics</a>
218 <ol>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]219 <li><a href="#int_memory_barrier"><tt>llvm.memory_barrier</tt></a></li>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]220 <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]221 <li><a href="#int_atomic_swap"><tt>llvm.atomic.swap</tt></a></li>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]222 <li><a href="#int_atomic_load_add"><tt>llvm.atomic.load.add</tt></a></li>
223 <li><a href="#int_atomic_load_sub"><tt>llvm.atomic.load.sub</tt></a></li>
224 <li><a href="#int_atomic_load_and"><tt>llvm.atomic.load.and</tt></a></li>
225 <li><a href="#int_atomic_load_nand"><tt>llvm.atomic.load.nand</tt></a></li>
226 <li><a href="#int_atomic_load_or"><tt>llvm.atomic.load.or</tt></a></li>
227 <li><a href="#int_atomic_load_xor"><tt>llvm.atomic.load.xor</tt></a></li>
228 <li><a href="#int_atomic_load_max"><tt>llvm.atomic.load.max</tt></a></li>
229 <li><a href="#int_atomic_load_min"><tt>llvm.atomic.load.min</tt></a></li>
230 <li><a href="#int_atomic_load_umax"><tt>llvm.atomic.load.umax</tt></a></li>
231 <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]232 </ol>
233 </li>
Reid Spencerb043f672007-07-20 19:59:11 +0000[diff] [blame]234 <li><a href="#int_general">General intrinsics</a>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]235 <ol>
Reid Spencerb043f672007-07-20 19:59:11 +0000[diff] [blame]236 <li><a href="#int_var_annotation">
Tanya Lattner51369f32007-09-22 00:01:26 +0000[diff] [blame]237 <tt>llvm.var.annotation</tt>' Intrinsic</a></li>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]238 <li><a href="#int_annotation">
Tanya Lattner51369f32007-09-22 00:01:26 +0000[diff] [blame]239 <tt>llvm.annotation.*</tt>' Intrinsic</a></li>
Anton Korobeynikove6e764f2008-01-15 22:31:34 +0000[diff] [blame]240 <li><a href="#int_trap">
241 <tt>llvm.trap</tt>' Intrinsic</a></li>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]242 </ol>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]243 </li>
244 </ol>
245 </li>
246</ol>
247
248<div class="doc_author">
249 <p>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a>
250 and <a href="mailto:vadve@cs.uiuc.edu">Vikram Adve</a></p>
251</div>
252
253<!-- *********************************************************************** -->
254<div class="doc_section"> <a name="abstract">Abstract </a></div>
255<!-- *********************************************************************** -->
256
257<div class="doc_text">
258<p>This document is a reference manual for the LLVM assembly language.
Bill Wendlinge7846a52008-08-05 22:29:16 +0000[diff] [blame]259LLVM is a Static Single Assignment (SSA) based representation that provides
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]260type safety, low-level operations, flexibility, and the capability of
261representing 'all' high-level languages cleanly. It is the common code
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]262representation used throughout all phases of the LLVM compilation
263strategy.</p>
264</div>
265
266<!-- *********************************************************************** -->
267<div class="doc_section"> <a name="introduction">Introduction</a> </div>
268<!-- *********************************************************************** -->
269
270<div class="doc_text">
271
272<p>The LLVM code representation is designed to be used in three
273different forms: as an in-memory compiler IR, as an on-disk bitcode
274representation (suitable for fast loading by a Just-In-Time compiler),
275and as a human readable assembly language representation. This allows
276LLVM to provide a powerful intermediate representation for efficient
277compiler transformations and analysis, while providing a natural means
278to debug and visualize the transformations. The three different forms
279of LLVM are all equivalent. This document describes the human readable
280representation and notation.</p>
281
282<p>The LLVM representation aims to be light-weight and low-level
283while being expressive, typed, and extensible at the same time. It
284aims to be a "universal IR" of sorts, by being at a low enough level
285that high-level ideas may be cleanly mapped to it (similar to how
286microprocessors are "universal IR's", allowing many source languages to
287be mapped to them). By providing type information, LLVM can be used as
288the target of optimizations: for example, through pointer analysis, it
289can be proven that a C automatic variable is never accessed outside of
290the current function... allowing it to be promoted to a simple SSA
291value instead of a memory location.</p>
292
293</div>
294
295<!-- _______________________________________________________________________ -->
296<div class="doc_subsubsection"> <a name="wellformed">Well-Formedness</a> </div>
297
298<div class="doc_text">
299
300<p>It is important to note that this document describes 'well formed'
301LLVM assembly language. There is a difference between what the parser
302accepts and what is considered 'well formed'. For example, the
303following instruction is syntactically okay, but not well formed:</p>
304
305<div class="doc_code">
306<pre>
307%x = <a href="#i_add">add</a> i32 1, %x
308</pre>
309</div>
310
311<p>...because the definition of <tt>%x</tt> does not dominate all of
312its uses. The LLVM infrastructure provides a verification pass that may
313be used to verify that an LLVM module is well formed. This pass is
314automatically run by the parser after parsing input assembly and by
315the optimizer before it outputs bitcode. The violations pointed out
316by the verifier pass indicate bugs in transformation passes or input to
317the parser.</p>
318</div>
319
Chris Lattnera83fdc02007-10-03 17:34:29 +0000[diff] [blame]320<!-- Describe the typesetting conventions here. -->
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]321
322<!-- *********************************************************************** -->
323<div class="doc_section"> <a name="identifiers">Identifiers</a> </div>
324<!-- *********************************************************************** -->
325
326<div class="doc_text">
327
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]328 <p>LLVM identifiers come in two basic types: global and local. Global
329 identifiers (functions, global variables) begin with the @ character. Local
330 identifiers (register names, types) begin with the % character. Additionally,
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]331 there are three different formats for identifiers, for different purposes:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]332
333<ol>
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]334 <li>Named values are represented as a string of characters with their prefix.
335 For example, %foo, @DivisionByZero, %a.really.long.identifier. The actual
336 regular expression used is '<tt>[%@][a-zA-Z$._][a-zA-Z$._0-9]*</tt>'.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]337 Identifiers which require other characters in their names can be surrounded
Daniel Dunbar73f9a772008-10-14 23:51:43 +0000[diff] [blame]338 with quotes. Special characters may be escaped using "\xx" where xx is the
339 ASCII code for the character in hexadecimal. In this way, any character can
340 be used in a name value, even quotes themselves.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]341
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]342 <li>Unnamed values are represented as an unsigned numeric value with their
343 prefix. For example, %12, @2, %44.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]344
345 <li>Constants, which are described in a <a href="#constants">section about
346 constants</a>, below.</li>
347</ol>
348
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]349<p>LLVM requires that values start with a prefix for two reasons: Compilers
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]350don't need to worry about name clashes with reserved words, and the set of
351reserved words may be expanded in the future without penalty. Additionally,
352unnamed identifiers allow a compiler to quickly come up with a temporary
353variable without having to avoid symbol table conflicts.</p>
354
355<p>Reserved words in LLVM are very similar to reserved words in other
356languages. There are keywords for different opcodes
357('<tt><a href="#i_add">add</a></tt>',
358 '<tt><a href="#i_bitcast">bitcast</a></tt>',
359 '<tt><a href="#i_ret">ret</a></tt>', etc...), for primitive type names ('<tt><a
360href="#t_void">void</a></tt>', '<tt><a href="#t_primitive">i32</a></tt>', etc...),
361and others. These reserved words cannot conflict with variable names, because
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]362none of them start with a prefix character ('%' or '@').</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]363
364<p>Here is an example of LLVM code to multiply the integer variable
365'<tt>%X</tt>' by 8:</p>
366
367<p>The easy way:</p>
368
369<div class="doc_code">
370<pre>
371%result = <a href="#i_mul">mul</a> i32 %X, 8
372</pre>
373</div>
374
375<p>After strength reduction:</p>
376
377<div class="doc_code">
378<pre>
379%result = <a href="#i_shl">shl</a> i32 %X, i8 3
380</pre>
381</div>
382
383<p>And the hard way:</p>
384
385<div class="doc_code">
386<pre>
387<a href="#i_add">add</a> i32 %X, %X <i>; yields {i32}:%0</i>
388<a href="#i_add">add</a> i32 %0, %0 <i>; yields {i32}:%1</i>
389%result = <a href="#i_add">add</a> i32 %1, %1
390</pre>
391</div>
392
393<p>This last way of multiplying <tt>%X</tt> by 8 illustrates several
394important lexical features of LLVM:</p>
395
396<ol>
397
398 <li>Comments are delimited with a '<tt>;</tt>' and go until the end of
399 line.</li>
400
401 <li>Unnamed temporaries are created when the result of a computation is not
402 assigned to a named value.</li>
403
404 <li>Unnamed temporaries are numbered sequentially</li>
405
406</ol>
407
408<p>...and it also shows a convention that we follow in this document. When
409demonstrating instructions, we will follow an instruction with a comment that
410defines the type and name of value produced. Comments are shown in italic
411text.</p>
412
413</div>
414
415<!-- *********************************************************************** -->
416<div class="doc_section"> <a name="highlevel">High Level Structure</a> </div>
417<!-- *********************************************************************** -->
418
419<!-- ======================================================================= -->
420<div class="doc_subsection"> <a name="modulestructure">Module Structure</a>
421</div>
422
423<div class="doc_text">
424
425<p>LLVM programs are composed of "Module"s, each of which is a
426translation unit of the input programs. Each module consists of
427functions, global variables, and symbol table entries. Modules may be
428combined together with the LLVM linker, which merges function (and
429global variable) definitions, resolves forward declarations, and merges
430symbol table entries. Here is an example of the "hello world" module:</p>
431
432<div class="doc_code">
433<pre><i>; Declare the string constant as a global constant...</i>
434<a href="#identifiers">@.LC0</a> = <a href="#linkage_internal">internal</a> <a
435 href="#globalvars">constant</a> <a href="#t_array">[13 x i8]</a> c"hello world\0A\00" <i>; [13 x i8]*</i>
436
437<i>; External declaration of the puts function</i>
438<a href="#functionstructure">declare</a> i32 @puts(i8 *) <i>; i32(i8 *)* </i>
439
440<i>; Definition of main function</i>
441define i32 @main() { <i>; i32()* </i>
442 <i>; Convert [13x i8 ]* to i8 *...</i>
443 %cast210 = <a
444 href="#i_getelementptr">getelementptr</a> [13 x i8 ]* @.LC0, i64 0, i64 0 <i>; i8 *</i>
445
446 <i>; Call puts function to write out the string to stdout...</i>
447 <a
448 href="#i_call">call</a> i32 @puts(i8 * %cast210) <i>; i32</i>
449 <a
450 href="#i_ret">ret</a> i32 0<br>}<br>
451</pre>
452</div>
453
454<p>This example is made up of a <a href="#globalvars">global variable</a>
455named "<tt>.LC0</tt>", an external declaration of the "<tt>puts</tt>"
456function, and a <a href="#functionstructure">function definition</a>
457for "<tt>main</tt>".</p>
458
459<p>In general, a module is made up of a list of global values,
460where both functions and global variables are global values. Global values are
461represented by a pointer to a memory location (in this case, a pointer to an
462array of char, and a pointer to a function), and have one of the following <a
463href="#linkage">linkage types</a>.</p>
464
465</div>
466
467<!-- ======================================================================= -->
468<div class="doc_subsection">
469 <a name="linkage">Linkage Types</a>
470</div>
471
472<div class="doc_text">
473
474<p>
475All Global Variables and Functions have one of the following types of linkage:
476</p>
477
478<dl>
479
Dale Johannesen96e7e092008-05-23 23:13:41 +0000[diff] [blame]480 <dt><tt><b><a name="linkage_internal">internal</a></b></tt>: </dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]481
482 <dd>Global values with internal linkage are only directly accessible by
483 objects in the current module. In particular, linking code into a module with
484 an internal global value may cause the internal to be renamed as necessary to
485 avoid collisions. Because the symbol is internal to the module, all
486 references can be updated. This corresponds to the notion of the
487 '<tt>static</tt>' keyword in C.
488 </dd>
489
490 <dt><tt><b><a name="linkage_linkonce">linkonce</a></b></tt>: </dt>
491
492 <dd>Globals with "<tt>linkonce</tt>" linkage are merged with other globals of
493 the same name when linkage occurs. This is typically used to implement
494 inline functions, templates, or other code which must be generated in each
495 translation unit that uses it. Unreferenced <tt>linkonce</tt> globals are
496 allowed to be discarded.
497 </dd>
498
Dale Johannesen96e7e092008-05-23 23:13:41 +0000[diff] [blame]499 <dt><tt><b><a name="linkage_common">common</a></b></tt>: </dt>
500
501 <dd>"<tt>common</tt>" linkage is exactly the same as <tt>linkonce</tt>
502 linkage, except that unreferenced <tt>common</tt> globals may not be
503 discarded. This is used for globals that may be emitted in multiple
504 translation units, but that are not guaranteed to be emitted into every
505 translation unit that uses them. One example of this is tentative
506 definitions in C, such as "<tt>int X;</tt>" at global scope.
507 </dd>
508
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]509 <dt><tt><b><a name="linkage_weak">weak</a></b></tt>: </dt>
510
Dale Johannesen96e7e092008-05-23 23:13:41 +0000[diff] [blame]511 <dd>"<tt>weak</tt>" linkage is the same as <tt>common</tt> linkage, except
512 that some targets may choose to emit different assembly sequences for them
513 for target-dependent reasons. This is used for globals that are declared
514 "weak" in C source code.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]515 </dd>
516
517 <dt><tt><b><a name="linkage_appending">appending</a></b></tt>: </dt>
518
519 <dd>"<tt>appending</tt>" linkage may only be applied to global variables of
520 pointer to array type. When two global variables with appending linkage are
521 linked together, the two global arrays are appended together. This is the
522 LLVM, typesafe, equivalent of having the system linker append together
523 "sections" with identical names when .o files are linked.
524 </dd>
525
526 <dt><tt><b><a name="linkage_externweak">extern_weak</a></b></tt>: </dt>
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]527 <dd>The semantics of this linkage follow the ELF object file model: the
528 symbol is weak until linked, if not linked, the symbol becomes null instead
529 of being an undefined reference.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]530 </dd>
531
532 <dt><tt><b><a name="linkage_external">externally visible</a></b></tt>:</dt>
533
534 <dd>If none of the above identifiers are used, the global is externally
535 visible, meaning that it participates in linkage and can be used to resolve
536 external symbol references.
537 </dd>
538</dl>
539
540 <p>
541 The next two types of linkage are targeted for Microsoft Windows platform
542 only. They are designed to support importing (exporting) symbols from (to)
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]543 DLLs (Dynamic Link Libraries).
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]544 </p>
545
546 <dl>
547 <dt><tt><b><a name="linkage_dllimport">dllimport</a></b></tt>: </dt>
548
549 <dd>"<tt>dllimport</tt>" linkage causes the compiler to reference a function
550 or variable via a global pointer to a pointer that is set up by the DLL
551 exporting the symbol. On Microsoft Windows targets, the pointer name is
552 formed by combining <code>_imp__</code> and the function or variable name.
553 </dd>
554
555 <dt><tt><b><a name="linkage_dllexport">dllexport</a></b></tt>: </dt>
556
557 <dd>"<tt>dllexport</tt>" linkage causes the compiler to provide a global
558 pointer to a pointer in a DLL, so that it can be referenced with the
559 <tt>dllimport</tt> attribute. On Microsoft Windows targets, the pointer
560 name is formed by combining <code>_imp__</code> and the function or variable
561 name.
562 </dd>
563
564</dl>
565
566<p><a name="linkage_external"></a>For example, since the "<tt>.LC0</tt>"
567variable is defined to be internal, if another module defined a "<tt>.LC0</tt>"
568variable and was linked with this one, one of the two would be renamed,
569preventing a collision. Since "<tt>main</tt>" and "<tt>puts</tt>" are
570external (i.e., lacking any linkage declarations), they are accessible
571outside of the current module.</p>
572<p>It is illegal for a function <i>declaration</i>
573to have any linkage type other than "externally visible", <tt>dllimport</tt>,
574or <tt>extern_weak</tt>.</p>
575<p>Aliases can have only <tt>external</tt>, <tt>internal</tt> and <tt>weak</tt>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]576linkages.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]577</div>
578
579<!-- ======================================================================= -->
580<div class="doc_subsection">
581 <a name="callingconv">Calling Conventions</a>
582</div>
583
584<div class="doc_text">
585
586<p>LLVM <a href="#functionstructure">functions</a>, <a href="#i_call">calls</a>
587and <a href="#i_invoke">invokes</a> can all have an optional calling convention
588specified for the call. The calling convention of any pair of dynamic
589caller/callee must match, or the behavior of the program is undefined. The
590following calling conventions are supported by LLVM, and more may be added in
591the future:</p>
592
593<dl>
594 <dt><b>"<tt>ccc</tt>" - The C calling convention</b>:</dt>
595
596 <dd>This calling convention (the default if no other calling convention is
597 specified) matches the target C calling conventions. This calling convention
598 supports varargs function calls and tolerates some mismatch in the declared
599 prototype and implemented declaration of the function (as does normal C).
600 </dd>
601
602 <dt><b>"<tt>fastcc</tt>" - The fast calling convention</b>:</dt>
603
604 <dd>This calling convention attempts to make calls as fast as possible
605 (e.g. by passing things in registers). This calling convention allows the
606 target to use whatever tricks it wants to produce fast code for the target,
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]607 without having to conform to an externally specified ABI (Application Binary
608 Interface). Implementations of this convention should allow arbitrary
Arnold Schwaighofer07444922008-05-14 09:17:12 +0000[diff] [blame]609 <a href="CodeGenerator.html#tailcallopt">tail call optimization</a> to be
610 supported. This calling convention does not support varargs and requires the
611 prototype of all callees to exactly match the prototype of the function
612 definition.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]613 </dd>
614
615 <dt><b>"<tt>coldcc</tt>" - The cold calling convention</b>:</dt>
616
617 <dd>This calling convention attempts to make code in the caller as efficient
618 as possible under the assumption that the call is not commonly executed. As
619 such, these calls often preserve all registers so that the call does not break
620 any live ranges in the caller side. This calling convention does not support
621 varargs and requires the prototype of all callees to exactly match the
622 prototype of the function definition.
623 </dd>
624
625 <dt><b>"<tt>cc &lt;<em>n</em>&gt;</tt>" - Numbered convention</b>:</dt>
626
627 <dd>Any calling convention may be specified by number, allowing
628 target-specific calling conventions to be used. Target specific calling
629 conventions start at 64.
630 </dd>
631</dl>
632
633<p>More calling conventions can be added/defined on an as-needed basis, to
634support pascal conventions or any other well-known target-independent
635convention.</p>
636
637</div>
638
639<!-- ======================================================================= -->
640<div class="doc_subsection">
641 <a name="visibility">Visibility Styles</a>
642</div>
643
644<div class="doc_text">
645
646<p>
647All Global Variables and Functions have one of the following visibility styles:
648</p>
649
650<dl>
651 <dt><b>"<tt>default</tt>" - Default style</b>:</dt>
652
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]653 <dd>On targets that use the ELF object file format, default visibility means
654 that the declaration is visible to other
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]655 modules and, in shared libraries, means that the declared entity may be
656 overridden. On Darwin, default visibility means that the declaration is
657 visible to other modules. Default visibility corresponds to "external
658 linkage" in the language.
659 </dd>
660
661 <dt><b>"<tt>hidden</tt>" - Hidden style</b>:</dt>
662
663 <dd>Two declarations of an object with hidden visibility refer to the same
664 object if they are in the same shared object. Usually, hidden visibility
665 indicates that the symbol will not be placed into the dynamic symbol table,
666 so no other module (executable or shared library) can reference it
667 directly.
668 </dd>
669
670 <dt><b>"<tt>protected</tt>" - Protected style</b>:</dt>
671
672 <dd>On ELF, protected visibility indicates that the symbol will be placed in
673 the dynamic symbol table, but that references within the defining module will
674 bind to the local symbol. That is, the symbol cannot be overridden by another
675 module.
676 </dd>
677</dl>
678
679</div>
680
681<!-- ======================================================================= -->
682<div class="doc_subsection">
683 <a name="globalvars">Global Variables</a>
684</div>
685
686<div class="doc_text">
687
688<p>Global variables define regions of memory allocated at compilation time
689instead of run-time. Global variables may optionally be initialized, may have
690an explicit section to be placed in, and may have an optional explicit alignment
691specified. A variable may be defined as "thread_local", which means that it
692will not be shared by threads (each thread will have a separated copy of the
693variable). A variable may be defined as a global "constant," which indicates
694that the contents of the variable will <b>never</b> be modified (enabling better
695optimization, allowing the global data to be placed in the read-only section of
696an executable, etc). Note that variables that need runtime initialization
697cannot be marked "constant" as there is a store to the variable.</p>
698
699<p>
700LLVM explicitly allows <em>declarations</em> of global variables to be marked
701constant, even if the final definition of the global is not. This capability
702can be used to enable slightly better optimization of the program, but requires
703the language definition to guarantee that optimizations based on the
704'constantness' are valid for the translation units that do not include the
705definition.
706</p>
707
708<p>As SSA values, global variables define pointer values that are in
709scope (i.e. they dominate) all basic blocks in the program. Global
710variables always define a pointer to their "content" type because they
711describe a region of memory, and all memory objects in LLVM are
712accessed through pointers.</p>
713
Christopher Lambdd0049d2007-12-11 09:31:00 +0000[diff] [blame]714<p>A global variable may be declared to reside in a target-specifc numbered
715address space. For targets that support them, address spaces may affect how
716optimizations are performed and/or what target instructions are used to access
Christopher Lamb20a39e92007-12-12 08:44:39 +0000[diff] [blame]717the variable. The default address space is zero. The address space qualifier
718must precede any other attributes.</p>
Christopher Lambdd0049d2007-12-11 09:31:00 +0000[diff] [blame]719
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]720<p>LLVM allows an explicit section to be specified for globals. If the target
721supports it, it will emit globals to the section specified.</p>
722
723<p>An explicit alignment may be specified for a global. If not present, or if
724the alignment is set to zero, the alignment of the global is set by the target
725to whatever it feels convenient. If an explicit alignment is specified, the
726global is forced to have at least that much alignment. All alignments must be
727a power of 2.</p>
728
Christopher Lambdd0049d2007-12-11 09:31:00 +0000[diff] [blame]729<p>For example, the following defines a global in a numbered address space with
730an initializer, section, and alignment:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]731
732<div class="doc_code">
733<pre>
Christopher Lambdd0049d2007-12-11 09:31:00 +0000[diff] [blame]734@G = constant float 1.0 addrspace(5), section "foo", align 4
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]735</pre>
736</div>
737
738</div>
739
740
741<!-- ======================================================================= -->
742<div class="doc_subsection">
743 <a name="functionstructure">Functions</a>
744</div>
745
746<div class="doc_text">
747
748<p>LLVM function definitions consist of the "<tt>define</tt>" keyord,
749an optional <a href="#linkage">linkage type</a>, an optional
750<a href="#visibility">visibility style</a>, an optional
751<a href="#callingconv">calling convention</a>, a return type, an optional
752<a href="#paramattrs">parameter attribute</a> for the return type, a function
753name, a (possibly empty) argument list (each with optional
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]754<a href="#paramattrs">parameter attributes</a>), optional
755<a href="#fnattrs">function attributes</a>, an optional section,
756an optional alignment, an optional <a href="#gc">garbage collector name</a>,
Chris Lattneree202542008-10-04 18:10:21 +0000[diff] [blame]757an opening curly brace, a list of basic blocks, and a closing curly brace.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]758
759LLVM function declarations consist of the "<tt>declare</tt>" keyword, an
760optional <a href="#linkage">linkage type</a>, an optional
761<a href="#visibility">visibility style</a>, an optional
762<a href="#callingconv">calling convention</a>, a return type, an optional
763<a href="#paramattrs">parameter attribute</a> for the return type, a function
Gordon Henriksen13fe5e32007-12-10 03:18:06 +0000[diff] [blame]764name, a possibly empty list of arguments, an optional alignment, and an optional
Gordon Henriksend606f5b2007-12-10 03:30:21 +0000[diff] [blame]765<a href="#gc">garbage collector name</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]766
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]767<p>A function definition contains a list of basic blocks, forming the CFG
768(Control Flow Graph) for
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]769the function. Each basic block may optionally start with a label (giving the
770basic block a symbol table entry), contains a list of instructions, and ends
771with a <a href="#terminators">terminator</a> instruction (such as a branch or
772function return).</p>
773
774<p>The first basic block in a function is special in two ways: it is immediately
775executed on entrance to the function, and it is not allowed to have predecessor
776basic blocks (i.e. there can not be any branches to the entry block of a
777function). Because the block can have no predecessors, it also cannot have any
778<a href="#i_phi">PHI nodes</a>.</p>
779
780<p>LLVM allows an explicit section to be specified for functions. If the target
781supports it, it will emit functions to the section specified.</p>
782
783<p>An explicit alignment may be specified for a function. If not present, or if
784the alignment is set to zero, the alignment of the function is set by the target
785to whatever it feels convenient. If an explicit alignment is specified, the
786function is forced to have at least that much alignment. All alignments must be
787a power of 2.</p>
788
Devang Pateld0bfcc72008-10-07 17:48:33 +0000[diff] [blame]789 <h5>Syntax:</h5>
790
791<div class="doc_code">
Chris Lattner1e5c5cd02008-10-13 16:55:18 +0000[diff] [blame]792<tt>
793define [<a href="#linkage">linkage</a>] [<a href="#visibility">visibility</a>]
794 [<a href="#callingconv">cconv</a>] [<a href="#paramattrs">ret attrs</a>]
795 &lt;ResultType&gt; @&lt;FunctionName&gt; ([argument list])
796 [<a href="#fnattrs">fn Attrs</a>] [section "name"] [align N]
797 [<a href="#gc">gc</a>] { ... }
798</tt>
Devang Pateld0bfcc72008-10-07 17:48:33 +0000[diff] [blame]799</div>
800
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]801</div>
802
803
804<!-- ======================================================================= -->
805<div class="doc_subsection">
806 <a name="aliasstructure">Aliases</a>
807</div>
808<div class="doc_text">
809 <p>Aliases act as "second name" for the aliasee value (which can be either
Anton Korobeynikov96822812008-03-22 08:36:14 +0000[diff] [blame]810 function, global variable, another alias or bitcast of global value). Aliases
811 may have an optional <a href="#linkage">linkage type</a>, and an
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]812 optional <a href="#visibility">visibility style</a>.</p>
813
814 <h5>Syntax:</h5>
815
816<div class="doc_code">
817<pre>
Duncan Sandsd7bfabf2008-09-12 20:48:21 +0000[diff] [blame]818@&lt;Name&gt; = alias [Linkage] [Visibility] &lt;AliaseeTy&gt; @&lt;Aliasee&gt;
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]819</pre>
820</div>
821
822</div>
823
824
825
826<!-- ======================================================================= -->
827<div class="doc_subsection"><a name="paramattrs">Parameter Attributes</a></div>
828<div class="doc_text">
829 <p>The return type and each parameter of a function type may have a set of
830 <i>parameter attributes</i> associated with them. Parameter attributes are
831 used to communicate additional information about the result or parameters of
Duncan Sandsf5588dc2007-11-27 13:23:08 +0000[diff] [blame]832 a function. Parameter attributes are considered to be part of the function,
833 not of the function type, so functions with different parameter attributes
834 can have the same function type.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]835
836 <p>Parameter attributes are simple keywords that follow the type specified. If
837 multiple parameter attributes are needed, they are space separated. For
838 example:</p>
839
840<div class="doc_code">
841<pre>
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]842declare i32 @printf(i8* noalias , ...)
Chris Lattnerf33b8452008-10-04 18:33:34 +0000[diff] [blame]843declare i32 @atoi(i8 zeroext)
844declare signext i8 @returns_signed_char()
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]845</pre>
846</div>
847
Duncan Sandsf5588dc2007-11-27 13:23:08 +0000[diff] [blame]848 <p>Note that any attributes for the function result (<tt>nounwind</tt>,
849 <tt>readonly</tt>) come immediately after the argument list.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]850
851 <p>Currently, only the following parameter attributes are defined:</p>
852 <dl>
Reid Spencerf234bed2007-07-19 23:13:04 +0000[diff] [blame]853 <dt><tt>zeroext</tt></dt>
Chris Lattnerf33b8452008-10-04 18:33:34 +0000[diff] [blame]854 <dd>This indicates to the code generator that the parameter or return value
855 should be zero-extended to a 32-bit value by the caller (for a parameter)
856 or the callee (for a return value).</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]857
Reid Spencerf234bed2007-07-19 23:13:04 +0000[diff] [blame]858 <dt><tt>signext</tt></dt>
Chris Lattnerf33b8452008-10-04 18:33:34 +0000[diff] [blame]859 <dd>This indicates to the code generator that the parameter or return value
860 should be sign-extended to a 32-bit value by the caller (for a parameter)
861 or the callee (for a return value).</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]862
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]863 <dt><tt>inreg</tt></dt>
Dale Johannesenc08a0e22008-09-25 20:47:45 +0000[diff] [blame]864 <dd>This indicates that this parameter or return value should be treated
865 in a special target-dependent fashion during while emitting code for a
866 function call or return (usually, by putting it in a register as opposed
Chris Lattnerf33b8452008-10-04 18:33:34 +0000[diff] [blame]867 to memory, though some targets use it to distinguish between two different
868 kinds of registers). Use of this attribute is target-specific.</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]869
Duncan Sands7d94e5a2008-10-06 08:14:18 +0000[diff] [blame]870 <dt><tt><a name="byval">byval</a></tt></dt>
Chris Lattner04c86182008-01-15 04:34:22 +0000[diff] [blame]871 <dd>This indicates that the pointer parameter should really be passed by
872 value to the function. The attribute implies that a hidden copy of the
873 pointee is made between the caller and the callee, so the callee is unable
Chris Lattner6a9f3c42008-08-05 18:21:08 +0000[diff] [blame]874 to modify the value in the callee. This attribute is only valid on LLVM
Chris Lattner04c86182008-01-15 04:34:22 +0000[diff] [blame]875 pointer arguments. It is generally used to pass structs and arrays by
Duncan Sands7d94e5a2008-10-06 08:14:18 +0000[diff] [blame]876 value, but is also valid on pointers to scalars. The copy is considered to
877 belong to the caller not the callee (for example,
878 <tt><a href="#readonly">readonly</a></tt> functions should not write to
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]879 <tt>byval</tt> parameters). This is not a valid attribute for return
880 values. </dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]881
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]882 <dt><tt>sret</tt></dt>
Duncan Sands616cc032008-02-18 04:19:38 +0000[diff] [blame]883 <dd>This indicates that the pointer parameter specifies the address of a
884 structure that is the return value of the function in the source program.
Chris Lattnerf33b8452008-10-04 18:33:34 +0000[diff] [blame]885 This pointer must be guaranteed by the caller to be valid: loads and stores
886 to the structure may be assumed by the callee to not to trap. This may only
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]887 be applied to the first parameter. This is not a valid attribute for
888 return values. </dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]889
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]890 <dt><tt>noalias</tt></dt>
Owen Andersonc4fc4cd2008-02-18 04:09:01 +0000[diff] [blame]891 <dd>This indicates that the parameter does not alias any global or any other
892 parameter. The caller is responsible for ensuring that this is the case,
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]893 usually by placing the value in a stack allocation. This is not a valid
894 attribute for return values.</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]895
Duncan Sands4ee46812007-07-27 19:57:41 +0000[diff] [blame]896 <dt><tt>nest</tt></dt>
Duncan Sandsf1a7d4c2008-07-08 09:27:25 +0000[diff] [blame]897 <dd>This indicates that the pointer parameter can be excised using the
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]898 <a href="#int_trampoline">trampoline intrinsics</a>. This is not a valid
899 attribute for return values.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]900 </dl>
901
902</div>
903
904<!-- ======================================================================= -->
905<div class="doc_subsection">
Gordon Henriksen13fe5e32007-12-10 03:18:06 +0000[diff] [blame]906 <a name="gc">Garbage Collector Names</a>
907</div>
908
909<div class="doc_text">
910<p>Each function may specify a garbage collector name, which is simply a
911string.</p>
912
913<div class="doc_code"><pre
914>define void @f() gc "name" { ...</pre></div>
915
916<p>The compiler declares the supported values of <i>name</i>. Specifying a
917collector which will cause the compiler to alter its output in order to support
918the named garbage collection algorithm.</p>
919</div>
920
921<!-- ======================================================================= -->
922<div class="doc_subsection">
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]923 <a name="fnattrs">Function Attributes</a>
Devang Pateld468f1c2008-09-04 23:05:13 +0000[diff] [blame]924</div>
925
926<div class="doc_text">
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]927
928<p>Function attributes are set to communicate additional information about
929 a function. Function attributes are considered to be part of the function,
930 not of the function type, so functions with different parameter attributes
931 can have the same function type.</p>
932
933 <p>Function attributes are simple keywords that follow the type specified. If
934 multiple attributes are needed, they are space separated. For
935 example:</p>
Devang Pateld468f1c2008-09-04 23:05:13 +0000[diff] [blame]936
937<div class="doc_code">
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]938<pre>
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]939define void @f() noinline { ... }
940define void @f() alwaysinline { ... }
941define void @f() alwaysinline optsize { ... }
942define void @f() optsize
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]943</pre>
Devang Pateld468f1c2008-09-04 23:05:13 +0000[diff] [blame]944</div>
945
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]946<dl>
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]947<dt><tt>alwaysinline</tt></dt>
Chris Lattner82d70f92008-10-04 18:23:17 +0000[diff] [blame]948<dd>This attribute indicates that the inliner should attempt to inline this
949function into callers whenever possible, ignoring any active inlining size
950threshold for this caller.</dd>
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]951
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]952<dt><tt>noinline</tt></dt>
Chris Lattner82d70f92008-10-04 18:23:17 +0000[diff] [blame]953<dd>This attribute indicates that the inliner should never inline this function
Chris Lattner377f3ae2008-10-05 17:14:59 +0000[diff] [blame]954in any situation. This attribute may not be used together with the
Chris Lattner82d70f92008-10-04 18:23:17 +0000[diff] [blame]955<tt>alwaysinline</tt> attribute.</dd>
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]956
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]957<dt><tt>optsize</tt></dt>
Devang Patelc29099b2008-09-29 18:34:44 +0000[diff] [blame]958<dd>This attribute suggests that optimization passes and code generator passes
Chris Lattner82d70f92008-10-04 18:23:17 +0000[diff] [blame]959make choices that keep the code size of this function low, and otherwise do
960optimizations specifically to reduce code size.</dd>
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]961
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]962<dt><tt>noreturn</tt></dt>
Chris Lattner82d70f92008-10-04 18:23:17 +0000[diff] [blame]963<dd>This function attribute indicates that the function never returns normally.
964This produces undefined behavior at runtime if the function ever does
965dynamically return.</dd>
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]966
967<dt><tt>nounwind</tt></dt>
Chris Lattner82d70f92008-10-04 18:23:17 +0000[diff] [blame]968<dd>This function attribute indicates that the function never returns with an
969unwind or exceptional control flow. If the function does unwind, its runtime
970behavior is undefined.</dd>
971
972<dt><tt>readnone</tt></dt>
Duncan Sands7d94e5a2008-10-06 08:14:18 +0000[diff] [blame]973<dd>This attribute indicates that the function computes its result (or the
974exception it throws) based strictly on its arguments, without dereferencing any
975pointer arguments or otherwise accessing any mutable state (e.g. memory, control
976registers, etc) visible to caller functions. It does not write through any
977pointer arguments (including <tt><a href="#byval">byval</a></tt> arguments) and
978never changes any state visible to callers.</dd>
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]979
Duncan Sands7d94e5a2008-10-06 08:14:18 +0000[diff] [blame]980<dt><tt><a name="readonly">readonly</a></tt></dt>
981<dd>This attribute indicates that the function does not write through any
982pointer arguments (including <tt><a href="#byval">byval</a></tt> arguments)
983or otherwise modify any state (e.g. memory, control registers, etc) visible to
984caller functions. It may dereference pointer arguments and read state that may
985be set in the caller. A readonly function always returns the same value (or
986throws the same exception) when called with the same set of arguments and global
987state.</dd>
Bill Wendlingdfaabba2008-11-13 01:02:51 +0000[diff] [blame^]988
989<dt><tt><a name="ssp">ssp</a></tt></dt>
990<dd>This attribute indicates that the function should emit a stack smashing
991protector. It is in the form of a "canary"&mdash;a random value placed on the
992stack before the local variables that's checked upon return from the function to
993see if it has been overwritten. A heuristic is used to determine if a function
994needs stack protectors or not.</dd>
995
996<dt><tt>ssp-req</tt></dt>
997<dd>This attribute indicates that the function should <em>always</em> emit a
998stack smashing protector. This overrides the <tt><a href="#ssp">ssp</a></tt>
999function attribute.</dd>
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]1000</dl>
1001
Devang Pateld468f1c2008-09-04 23:05:13 +0000[diff] [blame]1002</div>
1003
1004<!-- ======================================================================= -->
1005<div class="doc_subsection">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1006 <a name="moduleasm">Module-Level Inline Assembly</a>
1007</div>
1008
1009<div class="doc_text">
1010<p>
1011Modules may contain "module-level inline asm" blocks, which corresponds to the
1012GCC "file scope inline asm" blocks. These blocks are internally concatenated by
1013LLVM and treated as a single unit, but may be separated in the .ll file if
1014desired. The syntax is very simple:
1015</p>
1016
1017<div class="doc_code">
1018<pre>
1019module asm "inline asm code goes here"
1020module asm "more can go here"
1021</pre>
1022</div>
1023
1024<p>The strings can contain any character by escaping non-printable characters.
1025 The escape sequence used is simply "\xx" where "xx" is the two digit hex code
1026 for the number.
1027</p>
1028
1029<p>
1030 The inline asm code is simply printed to the machine code .s file when
1031 assembly code is generated.
1032</p>
1033</div>
1034
1035<!-- ======================================================================= -->
1036<div class="doc_subsection">
1037 <a name="datalayout">Data Layout</a>
1038</div>
1039
1040<div class="doc_text">
1041<p>A module may specify a target specific data layout string that specifies how
1042data is to be laid out in memory. The syntax for the data layout is simply:</p>
1043<pre> target datalayout = "<i>layout specification</i>"</pre>
1044<p>The <i>layout specification</i> consists of a list of specifications
1045separated by the minus sign character ('-'). Each specification starts with a
1046letter and may include other information after the letter to define some
1047aspect of the data layout. The specifications accepted are as follows: </p>
1048<dl>
1049 <dt><tt>E</tt></dt>
1050 <dd>Specifies that the target lays out data in big-endian form. That is, the
1051 bits with the most significance have the lowest address location.</dd>
1052 <dt><tt>e</tt></dt>
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]1053 <dd>Specifies that the target lays out data in little-endian form. That is,
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1054 the bits with the least significance have the lowest address location.</dd>
1055 <dt><tt>p:<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1056 <dd>This specifies the <i>size</i> of a pointer and its <i>abi</i> and
1057 <i>preferred</i> alignments. All sizes are in bits. Specifying the <i>pref</i>
1058 alignment is optional. If omitted, the preceding <tt>:</tt> should be omitted
1059 too.</dd>
1060 <dt><tt>i<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1061 <dd>This specifies the alignment for an integer type of a given bit
1062 <i>size</i>. The value of <i>size</i> must be in the range [1,2^23).</dd>
1063 <dt><tt>v<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1064 <dd>This specifies the alignment for a vector type of a given bit
1065 <i>size</i>.</dd>
1066 <dt><tt>f<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1067 <dd>This specifies the alignment for a floating point type of a given bit
1068 <i>size</i>. The value of <i>size</i> must be either 32 (float) or 64
1069 (double).</dd>
1070 <dt><tt>a<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1071 <dd>This specifies the alignment for an aggregate type of a given bit
1072 <i>size</i>.</dd>
1073</dl>
1074<p>When constructing the data layout for a given target, LLVM starts with a
1075default set of specifications which are then (possibly) overriden by the
1076specifications in the <tt>datalayout</tt> keyword. The default specifications
1077are given in this list:</p>
1078<ul>
1079 <li><tt>E</tt> - big endian</li>
1080 <li><tt>p:32:64:64</tt> - 32-bit pointers with 64-bit alignment</li>
1081 <li><tt>i1:8:8</tt> - i1 is 8-bit (byte) aligned</li>
1082 <li><tt>i8:8:8</tt> - i8 is 8-bit (byte) aligned</li>
1083 <li><tt>i16:16:16</tt> - i16 is 16-bit aligned</li>
1084 <li><tt>i32:32:32</tt> - i32 is 32-bit aligned</li>
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]1085 <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]1086 alignment of 64-bits</li>
1087 <li><tt>f32:32:32</tt> - float is 32-bit aligned</li>
1088 <li><tt>f64:64:64</tt> - double is 64-bit aligned</li>
1089 <li><tt>v64:64:64</tt> - 64-bit vector is 64-bit aligned</li>
1090 <li><tt>v128:128:128</tt> - 128-bit vector is 128-bit aligned</li>
1091 <li><tt>a0:0:1</tt> - aggregates are 8-bit aligned</li>
1092</ul>
Chris Lattner6a9f3c42008-08-05 18:21:08 +0000[diff] [blame]1093<p>When LLVM is determining the alignment for a given type, it uses the
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]1094following rules:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1095<ol>
1096 <li>If the type sought is an exact match for one of the specifications, that
1097 specification is used.</li>
1098 <li>If no match is found, and the type sought is an integer type, then the
1099 smallest integer type that is larger than the bitwidth of the sought type is
1100 used. If none of the specifications are larger than the bitwidth then the the
1101 largest integer type is used. For example, given the default specifications
1102 above, the i7 type will use the alignment of i8 (next largest) while both
1103 i65 and i256 will use the alignment of i64 (largest specified).</li>
1104 <li>If no match is found, and the type sought is a vector type, then the
1105 largest vector type that is smaller than the sought vector type will be used
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]1106 as a fall back. This happens because &lt;128 x double&gt; can be implemented
1107 in terms of 64 &lt;2 x double&gt;, for example.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1108</ol>
1109</div>
1110
1111<!-- *********************************************************************** -->
1112<div class="doc_section"> <a name="typesystem">Type System</a> </div>
1113<!-- *********************************************************************** -->
1114
1115<div class="doc_text">
1116
1117<p>The LLVM type system is one of the most important features of the
1118intermediate representation. Being typed enables a number of
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]1119optimizations to be performed on the intermediate representation directly,
1120without having to do
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1121extra analyses on the side before the transformation. A strong type
1122system makes it easier to read the generated code and enables novel
1123analyses and transformations that are not feasible to perform on normal
1124three address code representations.</p>
1125
1126</div>
1127
1128<!-- ======================================================================= -->
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1129<div class="doc_subsection"> <a name="t_classifications">Type
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1130Classifications</a> </div>
1131<div class="doc_text">
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1132<p>The types fall into a few useful
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1133classifications:</p>
1134
1135<table border="1" cellspacing="0" cellpadding="4">
1136 <tbody>
1137 <tr><th>Classification</th><th>Types</th></tr>
1138 <tr>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1139 <td><a href="#t_integer">integer</a></td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1140 <td><tt>i1, i2, i3, ... i8, ... i16, ... i32, ... i64, ... </tt></td>
1141 </tr>
1142 <tr>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1143 <td><a href="#t_floating">floating point</a></td>
1144 <td><tt>float, double, x86_fp80, fp128, ppc_fp128</tt></td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1145 </tr>
1146 <tr>
1147 <td><a name="t_firstclass">first class</a></td>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1148 <td><a href="#t_integer">integer</a>,
1149 <a href="#t_floating">floating point</a>,
1150 <a href="#t_pointer">pointer</a>,
Dan Gohmanf6237db2008-06-18 18:42:13 +0000[diff] [blame]1151 <a href="#t_vector">vector</a>,
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]1152 <a href="#t_struct">structure</a>,
1153 <a href="#t_array">array</a>,
Dan Gohmand13951e2008-05-23 22:50:26 +0000[diff] [blame]1154 <a href="#t_label">label</a>.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1155 </td>
1156 </tr>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1157 <tr>
1158 <td><a href="#t_primitive">primitive</a></td>
1159 <td><a href="#t_label">label</a>,
1160 <a href="#t_void">void</a>,
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1161 <a href="#t_floating">floating point</a>.</td>
1162 </tr>
1163 <tr>
1164 <td><a href="#t_derived">derived</a></td>
1165 <td><a href="#t_integer">integer</a>,
1166 <a href="#t_array">array</a>,
1167 <a href="#t_function">function</a>,
1168 <a href="#t_pointer">pointer</a>,
1169 <a href="#t_struct">structure</a>,
1170 <a href="#t_pstruct">packed structure</a>,
1171 <a href="#t_vector">vector</a>,
1172 <a href="#t_opaque">opaque</a>.
Dan Gohman032ba852008-10-14 16:32:04 +0000[diff] [blame]1173 </td>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1174 </tr>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1175 </tbody>
1176</table>
1177
1178<p>The <a href="#t_firstclass">first class</a> types are perhaps the
1179most important. Values of these types are the only ones which can be
1180produced by instructions, passed as arguments, or used as operands to
Dan Gohman4f29e422008-05-23 21:53:15 +0000[diff] [blame]1181instructions.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1182</div>
1183
1184<!-- ======================================================================= -->
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1185<div class="doc_subsection"> <a name="t_primitive">Primitive Types</a> </div>
Chris Lattner86437612008-01-04 04:34:14 +0000[diff] [blame]1186
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1187<div class="doc_text">
1188<p>The primitive types are the fundamental building blocks of the LLVM
1189system.</p>
1190
Chris Lattner86437612008-01-04 04:34:14 +0000[diff] [blame]1191</div>
1192
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1193<!-- _______________________________________________________________________ -->
1194<div class="doc_subsubsection"> <a name="t_floating">Floating Point Types</a> </div>
1195
1196<div class="doc_text">
1197 <table>
1198 <tbody>
1199 <tr><th>Type</th><th>Description</th></tr>
1200 <tr><td><tt>float</tt></td><td>32-bit floating point value</td></tr>
1201 <tr><td><tt>double</tt></td><td>64-bit floating point value</td></tr>
1202 <tr><td><tt>fp128</tt></td><td>128-bit floating point value (112-bit mantissa)</td></tr>
1203 <tr><td><tt>x86_fp80</tt></td><td>80-bit floating point value (X87)</td></tr>
1204 <tr><td><tt>ppc_fp128</tt></td><td>128-bit floating point value (two 64-bits)</td></tr>
1205 </tbody>
1206 </table>
1207</div>
1208
1209<!-- _______________________________________________________________________ -->
1210<div class="doc_subsubsection"> <a name="t_void">Void Type</a> </div>
1211
1212<div class="doc_text">
1213<h5>Overview:</h5>
1214<p>The void type does not represent any value and has no size.</p>
1215
1216<h5>Syntax:</h5>
1217
1218<pre>
1219 void
1220</pre>
1221</div>
1222
1223<!-- _______________________________________________________________________ -->
1224<div class="doc_subsubsection"> <a name="t_label">Label Type</a> </div>
1225
1226<div class="doc_text">
1227<h5>Overview:</h5>
1228<p>The label type represents code labels.</p>
1229
1230<h5>Syntax:</h5>
1231
1232<pre>
1233 label
1234</pre>
1235</div>
1236
1237
1238<!-- ======================================================================= -->
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1239<div class="doc_subsection"> <a name="t_derived">Derived Types</a> </div>
1240
1241<div class="doc_text">
1242
1243<p>The real power in LLVM comes from the derived types in the system.
1244This is what allows a programmer to represent arrays, functions,
1245pointers, and other useful types. Note that these derived types may be
1246recursive: For example, it is possible to have a two dimensional array.</p>
1247
1248</div>
1249
1250<!-- _______________________________________________________________________ -->
1251<div class="doc_subsubsection"> <a name="t_integer">Integer Type</a> </div>
1252
1253<div class="doc_text">
1254
1255<h5>Overview:</h5>
1256<p>The integer type is a very simple derived type that simply specifies an
1257arbitrary bit width for the integer type desired. Any bit width from 1 bit to
12582^23-1 (about 8 million) can be specified.</p>
1259
1260<h5>Syntax:</h5>
1261
1262<pre>
1263 iN
1264</pre>
1265
1266<p>The number of bits the integer will occupy is specified by the <tt>N</tt>
1267value.</p>
1268
1269<h5>Examples:</h5>
1270<table class="layout">
Chris Lattner251ab812007-12-18 06:18:21 +0000[diff] [blame]1271 <tbody>
1272 <tr>
1273 <td><tt>i1</tt></td>
1274 <td>a single-bit integer.</td>
1275 </tr><tr>
1276 <td><tt>i32</tt></td>
1277 <td>a 32-bit integer.</td>
1278 </tr><tr>
1279 <td><tt>i1942652</tt></td>
1280 <td>a really big integer of over 1 million bits.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1281 </tr>
Chris Lattner251ab812007-12-18 06:18:21 +0000[diff] [blame]1282 </tbody>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1283</table>
1284</div>
1285
1286<!-- _______________________________________________________________________ -->
1287<div class="doc_subsubsection"> <a name="t_array">Array Type</a> </div>
1288
1289<div class="doc_text">
1290
1291<h5>Overview:</h5>
1292
1293<p>The array type is a very simple derived type that arranges elements
1294sequentially in memory. The array type requires a size (number of
1295elements) and an underlying data type.</p>
1296
1297<h5>Syntax:</h5>
1298
1299<pre>
1300 [&lt;# elements&gt; x &lt;elementtype&gt;]
1301</pre>
1302
1303<p>The number of elements is a constant integer value; elementtype may
1304be any type with a size.</p>
1305
1306<h5>Examples:</h5>
1307<table class="layout">
1308 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1309 <td class="left"><tt>[40 x i32]</tt></td>
1310 <td class="left">Array of 40 32-bit integer values.</td>
1311 </tr>
1312 <tr class="layout">
1313 <td class="left"><tt>[41 x i32]</tt></td>
1314 <td class="left">Array of 41 32-bit integer values.</td>
1315 </tr>
1316 <tr class="layout">
1317 <td class="left"><tt>[4 x i8]</tt></td>
1318 <td class="left">Array of 4 8-bit integer values.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1319 </tr>
1320</table>
1321<p>Here are some examples of multidimensional arrays:</p>
1322<table class="layout">
1323 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1324 <td class="left"><tt>[3 x [4 x i32]]</tt></td>
1325 <td class="left">3x4 array of 32-bit integer values.</td>
1326 </tr>
1327 <tr class="layout">
1328 <td class="left"><tt>[12 x [10 x float]]</tt></td>
1329 <td class="left">12x10 array of single precision floating point values.</td>
1330 </tr>
1331 <tr class="layout">
1332 <td class="left"><tt>[2 x [3 x [4 x i16]]]</tt></td>
1333 <td class="left">2x3x4 array of 16-bit integer values.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1334 </tr>
1335</table>
1336
1337<p>Note that 'variable sized arrays' can be implemented in LLVM with a zero
1338length array. Normally, accesses past the end of an array are undefined in
1339LLVM (e.g. it is illegal to access the 5th element of a 3 element array).
1340As a special case, however, zero length arrays are recognized to be variable
1341length. This allows implementation of 'pascal style arrays' with the LLVM
1342type "{ i32, [0 x float]}", for example.</p>
1343
1344</div>
1345
1346<!-- _______________________________________________________________________ -->
1347<div class="doc_subsubsection"> <a name="t_function">Function Type</a> </div>
1348<div class="doc_text">
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1349
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1350<h5>Overview:</h5>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1351
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1352<p>The function type can be thought of as a function signature. It
Devang Pateld4ba41d2008-03-24 05:35:41 +0000[diff] [blame]1353consists of a return type and a list of formal parameter types. The
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1354return 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]1355If 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]1356class types, and the struct must have at least one element.</p>
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]1357
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1358<h5>Syntax:</h5>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1359
1360<pre>
1361 &lt;returntype list&gt; (&lt;parameter list&gt;)
1362</pre>
1363
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1364<p>...where '<tt>&lt;parameter list&gt;</tt>' is a comma-separated list of type
1365specifiers. Optionally, the parameter list may include a type <tt>...</tt>,
1366which indicates that the function takes a variable number of arguments.
1367Variable argument functions can access their arguments with the <a
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]1368 href="#int_varargs">variable argument handling intrinsic</a> functions.
1369'<tt>&lt;returntype list&gt;</tt>' is a comma-separated list of
1370<a href="#t_firstclass">first class</a> type specifiers.</p>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1371
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1372<h5>Examples:</h5>
1373<table class="layout">
1374 <tr class="layout">
1375 <td class="left"><tt>i32 (i32)</tt></td>
1376 <td class="left">function taking an <tt>i32</tt>, returning an <tt>i32</tt>
1377 </td>
1378 </tr><tr class="layout">
Reid Spencerf234bed2007-07-19 23:13:04 +0000[diff] [blame]1379 <td class="left"><tt>float&nbsp;(i16&nbsp;signext,&nbsp;i32&nbsp;*)&nbsp;*
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1380 </tt></td>
1381 <td class="left"><a href="#t_pointer">Pointer</a> to a function that takes
1382 an <tt>i16</tt> that should be sign extended and a
1383 <a href="#t_pointer">pointer</a> to <tt>i32</tt>, returning
1384 <tt>float</tt>.
1385 </td>
1386 </tr><tr class="layout">
1387 <td class="left"><tt>i32 (i8*, ...)</tt></td>
1388 <td class="left">A vararg function that takes at least one
1389 <a href="#t_pointer">pointer</a> to <tt>i8 </tt> (char in C),
1390 which returns an integer. This is the signature for <tt>printf</tt> in
1391 LLVM.
1392 </td>
Devang Pateld4ba41d2008-03-24 05:35:41 +0000[diff] [blame]1393 </tr><tr class="layout">
1394 <td class="left"><tt>{i32, i32} (i32)</tt></td>
Devang Patel6dddba22008-03-24 18:10:52 +0000[diff] [blame]1395 <td class="left">A function taking an <tt>i32></tt>, returning two
1396 <tt> i32 </tt> values as an aggregate of type <tt>{ i32, i32 }</tt>
Devang Pateld4ba41d2008-03-24 05:35:41 +0000[diff] [blame]1397 </td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1398 </tr>
1399</table>
1400
1401</div>
1402<!-- _______________________________________________________________________ -->
1403<div class="doc_subsubsection"> <a name="t_struct">Structure Type</a> </div>
1404<div class="doc_text">
1405<h5>Overview:</h5>
1406<p>The structure type is used to represent a collection of data members
1407together in memory. The packing of the field types is defined to match
1408the ABI of the underlying processor. The elements of a structure may
1409be any type that has a size.</p>
1410<p>Structures are accessed using '<tt><a href="#i_load">load</a></tt>
1411and '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a
1412field with the '<tt><a href="#i_getelementptr">getelementptr</a></tt>'
1413instruction.</p>
1414<h5>Syntax:</h5>
1415<pre> { &lt;type list&gt; }<br></pre>
1416<h5>Examples:</h5>
1417<table class="layout">
1418 <tr class="layout">
1419 <td class="left"><tt>{ i32, i32, i32 }</tt></td>
1420 <td class="left">A triple of three <tt>i32</tt> values</td>
1421 </tr><tr class="layout">
1422 <td class="left"><tt>{&nbsp;float,&nbsp;i32&nbsp;(i32)&nbsp;*&nbsp;}</tt></td>
1423 <td class="left">A pair, where the first element is a <tt>float</tt> and the
1424 second element is a <a href="#t_pointer">pointer</a> to a
1425 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
1426 an <tt>i32</tt>.</td>
1427 </tr>
1428</table>
1429</div>
1430
1431<!-- _______________________________________________________________________ -->
1432<div class="doc_subsubsection"> <a name="t_pstruct">Packed Structure Type</a>
1433</div>
1434<div class="doc_text">
1435<h5>Overview:</h5>
1436<p>The packed structure type is used to represent a collection of data members
1437together in memory. There is no padding between fields. Further, the alignment
1438of a packed structure is 1 byte. The elements of a packed structure may
1439be any type that has a size.</p>
1440<p>Structures are accessed using '<tt><a href="#i_load">load</a></tt>
1441and '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a
1442field with the '<tt><a href="#i_getelementptr">getelementptr</a></tt>'
1443instruction.</p>
1444<h5>Syntax:</h5>
1445<pre> &lt; { &lt;type list&gt; } &gt; <br></pre>
1446<h5>Examples:</h5>
1447<table class="layout">
1448 <tr class="layout">
1449 <td class="left"><tt>&lt; { i32, i32, i32 } &gt;</tt></td>
1450 <td class="left">A triple of three <tt>i32</tt> values</td>
1451 </tr><tr class="layout">
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]1452 <td class="left">
1453<tt>&lt;&nbsp;{&nbsp;float,&nbsp;i32&nbsp;(i32)*&nbsp;}&nbsp;&gt;</tt></td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1454 <td class="left">A pair, where the first element is a <tt>float</tt> and the
1455 second element is a <a href="#t_pointer">pointer</a> to a
1456 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
1457 an <tt>i32</tt>.</td>
1458 </tr>
1459</table>
1460</div>
1461
1462<!-- _______________________________________________________________________ -->
1463<div class="doc_subsubsection"> <a name="t_pointer">Pointer Type</a> </div>
1464<div class="doc_text">
1465<h5>Overview:</h5>
1466<p>As in many languages, the pointer type represents a pointer or
Christopher Lambdd0049d2007-12-11 09:31:00 +0000[diff] [blame]1467reference to another object, which must live in memory. Pointer types may have
1468an optional address space attribute defining the target-specific numbered
1469address space where the pointed-to object resides. The default address space is
1470zero.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1471<h5>Syntax:</h5>
1472<pre> &lt;type&gt; *<br></pre>
1473<h5>Examples:</h5>
1474<table class="layout">
1475 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1476 <td class="left"><tt>[4x i32]*</tt></td>
1477 <td class="left">A <a href="#t_pointer">pointer</a> to <a
1478 href="#t_array">array</a> of four <tt>i32</tt> values.</td>
1479 </tr>
1480 <tr class="layout">
1481 <td class="left"><tt>i32 (i32 *) *</tt></td>
1482 <td class="left"> A <a href="#t_pointer">pointer</a> to a <a
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1483 href="#t_function">function</a> that takes an <tt>i32*</tt>, returning an
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1484 <tt>i32</tt>.</td>
1485 </tr>
1486 <tr class="layout">
1487 <td class="left"><tt>i32 addrspace(5)*</tt></td>
1488 <td class="left">A <a href="#t_pointer">pointer</a> to an <tt>i32</tt> value
1489 that resides in address space #5.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1490 </tr>
1491</table>
1492</div>
1493
1494<!-- _______________________________________________________________________ -->
1495<div class="doc_subsubsection"> <a name="t_vector">Vector Type</a> </div>
1496<div class="doc_text">
1497
1498<h5>Overview:</h5>
1499
1500<p>A vector type is a simple derived type that represents a vector
1501of elements. Vector types are used when multiple primitive data
1502are operated in parallel using a single instruction (SIMD).
1503A vector type requires a size (number of
1504elements) and an underlying primitive data type. Vectors must have a power
1505of two length (1, 2, 4, 8, 16 ...). Vector types are
1506considered <a href="#t_firstclass">first class</a>.</p>
1507
1508<h5>Syntax:</h5>
1509
1510<pre>
1511 &lt; &lt;# elements&gt; x &lt;elementtype&gt; &gt;
1512</pre>
1513
1514<p>The number of elements is a constant integer value; elementtype may
1515be any integer or floating point type.</p>
1516
1517<h5>Examples:</h5>
1518
1519<table class="layout">
1520 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1521 <td class="left"><tt>&lt;4 x i32&gt;</tt></td>
1522 <td class="left">Vector of 4 32-bit integer values.</td>
1523 </tr>
1524 <tr class="layout">
1525 <td class="left"><tt>&lt;8 x float&gt;</tt></td>
1526 <td class="left">Vector of 8 32-bit floating-point values.</td>
1527 </tr>
1528 <tr class="layout">
1529 <td class="left"><tt>&lt;2 x i64&gt;</tt></td>
1530 <td class="left">Vector of 2 64-bit integer values.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1531 </tr>
1532</table>
1533</div>
1534
1535<!-- _______________________________________________________________________ -->
1536<div class="doc_subsubsection"> <a name="t_opaque">Opaque Type</a> </div>
1537<div class="doc_text">
1538
1539<h5>Overview:</h5>
1540
1541<p>Opaque types are used to represent unknown types in the system. This
Gordon Henriksenda0706e2007-10-14 00:34:53 +0000[diff] [blame]1542corresponds (for example) to the C notion of a forward declared structure type.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1543In LLVM, opaque types can eventually be resolved to any type (not just a
1544structure type).</p>
1545
1546<h5>Syntax:</h5>
1547
1548<pre>
1549 opaque
1550</pre>
1551
1552<h5>Examples:</h5>
1553
1554<table class="layout">
1555 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1556 <td class="left"><tt>opaque</tt></td>
1557 <td class="left">An opaque type.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1558 </tr>
1559</table>
1560</div>
1561
1562
1563<!-- *********************************************************************** -->
1564<div class="doc_section"> <a name="constants">Constants</a> </div>
1565<!-- *********************************************************************** -->
1566
1567<div class="doc_text">
1568
1569<p>LLVM has several different basic types of constants. This section describes
1570them all and their syntax.</p>
1571
1572</div>
1573
1574<!-- ======================================================================= -->
1575<div class="doc_subsection"><a name="simpleconstants">Simple Constants</a></div>
1576
1577<div class="doc_text">
1578
1579<dl>
1580 <dt><b>Boolean constants</b></dt>
1581
1582 <dd>The two strings '<tt>true</tt>' and '<tt>false</tt>' are both valid
1583 constants of the <tt><a href="#t_primitive">i1</a></tt> type.
1584 </dd>
1585
1586 <dt><b>Integer constants</b></dt>
1587
1588 <dd>Standard integers (such as '4') are constants of the <a
1589 href="#t_integer">integer</a> type. Negative numbers may be used with
1590 integer types.
1591 </dd>
1592
1593 <dt><b>Floating point constants</b></dt>
1594
1595 <dd>Floating point constants use standard decimal notation (e.g. 123.421),
1596 exponential notation (e.g. 1.23421e+2), or a more precise hexadecimal
Chris Lattnerdc15b1d2008-04-01 18:45:27 +0000[diff] [blame]1597 notation (see below). The assembler requires the exact decimal value of
1598 a floating-point constant. For example, the assembler accepts 1.25 but
1599 rejects 1.3 because 1.3 is a repeating decimal in binary. Floating point
1600 constants must have a <a href="#t_floating">floating point</a> type. </dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1601
1602 <dt><b>Null pointer constants</b></dt>
1603
1604 <dd>The identifier '<tt>null</tt>' is recognized as a null pointer constant
1605 and must be of <a href="#t_pointer">pointer type</a>.</dd>
1606
1607</dl>
1608
1609<p>The one non-intuitive notation for constants is the optional hexadecimal form
1610of floating point constants. For example, the form '<tt>double
16110x432ff973cafa8000</tt>' is equivalent to (but harder to read than) '<tt>double
16124.5e+15</tt>'. The only time hexadecimal floating point constants are required
1613(and the only time that they are generated by the disassembler) is when a
1614floating point constant must be emitted but it cannot be represented as a
1615decimal floating point number. For example, NaN's, infinities, and other
1616special values are represented in their IEEE hexadecimal format so that
1617assembly and disassembly do not cause any bits to change in the constants.</p>
1618
1619</div>
1620
1621<!-- ======================================================================= -->
1622<div class="doc_subsection"><a name="aggregateconstants">Aggregate Constants</a>
1623</div>
1624
1625<div class="doc_text">
1626<p>Aggregate constants arise from aggregation of simple constants
1627and smaller aggregate constants.</p>
1628
1629<dl>
1630 <dt><b>Structure constants</b></dt>
1631
1632 <dd>Structure constants are represented with notation similar to structure
1633 type definitions (a comma separated list of elements, surrounded by braces
Chris Lattner6c8de962007-12-25 20:34:52 +0000[diff] [blame]1634 (<tt>{}</tt>)). For example: "<tt>{ i32 4, float 17.0, i32* @G }</tt>",
1635 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]1636 must have <a href="#t_struct">structure type</a>, and the number and
1637 types of elements must match those specified by the type.
1638 </dd>
1639
1640 <dt><b>Array constants</b></dt>
1641
1642 <dd>Array constants are represented with notation similar to array type
1643 definitions (a comma separated list of elements, surrounded by square brackets
1644 (<tt>[]</tt>)). For example: "<tt>[ i32 42, i32 11, i32 74 ]</tt>". Array
1645 constants must have <a href="#t_array">array type</a>, and the number and
1646 types of elements must match those specified by the type.
1647 </dd>
1648
1649 <dt><b>Vector constants</b></dt>
1650
1651 <dd>Vector constants are represented with notation similar to vector type
1652 definitions (a comma separated list of elements, surrounded by
1653 less-than/greater-than's (<tt>&lt;&gt;</tt>)). For example: "<tt>&lt; i32 42,
1654 i32 11, i32 74, i32 100 &gt;</tt>". Vector constants must have <a
1655 href="#t_vector">vector type</a>, and the number and types of elements must
1656 match those specified by the type.
1657 </dd>
1658
1659 <dt><b>Zero initialization</b></dt>
1660
1661 <dd>The string '<tt>zeroinitializer</tt>' can be used to zero initialize a
1662 value to zero of <em>any</em> type, including scalar and aggregate types.
1663 This is often used to avoid having to print large zero initializers (e.g. for
1664 large arrays) and is always exactly equivalent to using explicit zero
1665 initializers.
1666 </dd>
1667</dl>
1668
1669</div>
1670
1671<!-- ======================================================================= -->
1672<div class="doc_subsection">
1673 <a name="globalconstants">Global Variable and Function Addresses</a>
1674</div>
1675
1676<div class="doc_text">
1677
1678<p>The addresses of <a href="#globalvars">global variables</a> and <a
1679href="#functionstructure">functions</a> are always implicitly valid (link-time)
1680constants. These constants are explicitly referenced when the <a
1681href="#identifiers">identifier for the global</a> is used and always have <a
1682href="#t_pointer">pointer</a> type. For example, the following is a legal LLVM
1683file:</p>
1684
1685<div class="doc_code">
1686<pre>
1687@X = global i32 17
1688@Y = global i32 42
1689@Z = global [2 x i32*] [ i32* @X, i32* @Y ]
1690</pre>
1691</div>
1692
1693</div>
1694
1695<!-- ======================================================================= -->
1696<div class="doc_subsection"><a name="undefvalues">Undefined Values</a></div>
1697<div class="doc_text">
1698 <p>The string '<tt>undef</tt>' is recognized as a type-less constant that has
1699 no specific value. Undefined values may be of any type and be used anywhere
1700 a constant is permitted.</p>
1701
1702 <p>Undefined values indicate to the compiler that the program is well defined
1703 no matter what value is used, giving the compiler more freedom to optimize.
1704 </p>
1705</div>
1706
1707<!-- ======================================================================= -->
1708<div class="doc_subsection"><a name="constantexprs">Constant Expressions</a>
1709</div>
1710
1711<div class="doc_text">
1712
1713<p>Constant expressions are used to allow expressions involving other constants
1714to be used as constants. Constant expressions may be of any <a
1715href="#t_firstclass">first class</a> type and may involve any LLVM operation
1716that does not have side effects (e.g. load and call are not supported). The
1717following is the syntax for constant expressions:</p>
1718
1719<dl>
1720 <dt><b><tt>trunc ( CST to TYPE )</tt></b></dt>
1721 <dd>Truncate a constant to another type. The bit size of CST must be larger
1722 than the bit size of TYPE. Both types must be integers.</dd>
1723
1724 <dt><b><tt>zext ( CST to TYPE )</tt></b></dt>
1725 <dd>Zero extend a constant to another type. The bit size of CST must be
1726 smaller or equal to the bit size of TYPE. Both types must be integers.</dd>
1727
1728 <dt><b><tt>sext ( CST to TYPE )</tt></b></dt>
1729 <dd>Sign extend a constant to another type. The bit size of CST must be
1730 smaller or equal to the bit size of TYPE. Both types must be integers.</dd>
1731
1732 <dt><b><tt>fptrunc ( CST to TYPE )</tt></b></dt>
1733 <dd>Truncate a floating point constant to another floating point type. The
1734 size of CST must be larger than the size of TYPE. Both types must be
1735 floating point.</dd>
1736
1737 <dt><b><tt>fpext ( CST to TYPE )</tt></b></dt>
1738 <dd>Floating point extend a constant to another type. The size of CST must be
1739 smaller or equal to the size of TYPE. Both types must be floating point.</dd>
1740
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]1741 <dt><b><tt>fptoui ( CST to TYPE )</tt></b></dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1742 <dd>Convert a floating point constant to the corresponding unsigned integer
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]1743 constant. TYPE must be a scalar or vector integer type. CST must be of scalar
1744 or vector floating point type. Both CST and TYPE must be scalars, or vectors
1745 of the same number of elements. If the value won't fit in the integer type,
1746 the results are undefined.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1747
1748 <dt><b><tt>fptosi ( CST to TYPE )</tt></b></dt>
1749 <dd>Convert a floating point constant to the corresponding signed integer
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]1750 constant. TYPE must be a scalar or vector integer type. CST must be of scalar
1751 or vector floating point type. Both CST and TYPE must be scalars, or vectors
1752 of the same number of elements. If the value won't fit in the integer type,
1753 the results are undefined.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1754
1755 <dt><b><tt>uitofp ( CST to TYPE )</tt></b></dt>
1756 <dd>Convert an unsigned integer constant to the corresponding floating point
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]1757 constant. TYPE must be a scalar or vector floating point type. CST must be of
1758 scalar or vector integer type. Both CST and TYPE must be scalars, or vectors
1759 of the same number of elements. If the value won't fit in the floating point
1760 type, the results are undefined.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1761
1762 <dt><b><tt>sitofp ( CST to TYPE )</tt></b></dt>
1763 <dd>Convert a signed integer constant to the corresponding floating point
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]1764 constant. TYPE must be a scalar or vector floating point type. CST must be of
1765 scalar or vector integer type. Both CST and TYPE must be scalars, or vectors
1766 of the same number of elements. If the value won't fit in the floating point
1767 type, the results are undefined.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1768
1769 <dt><b><tt>ptrtoint ( CST to TYPE )</tt></b></dt>
1770 <dd>Convert a pointer typed constant to the corresponding integer constant
1771 TYPE must be an integer type. CST must be of pointer type. The CST value is
1772 zero extended, truncated, or unchanged to make it fit in TYPE.</dd>
1773
1774 <dt><b><tt>inttoptr ( CST to TYPE )</tt></b></dt>
1775 <dd>Convert a integer constant to a pointer constant. TYPE must be a
1776 pointer type. CST must be of integer type. The CST value is zero extended,
1777 truncated, or unchanged to make it fit in a pointer size. This one is
1778 <i>really</i> dangerous!</dd>
1779
1780 <dt><b><tt>bitcast ( CST to TYPE )</tt></b></dt>
1781 <dd>Convert a constant, CST, to another TYPE. The size of CST and TYPE must be
1782 identical (same number of bits). The conversion is done as if the CST value
1783 was stored to memory and read back as TYPE. In other words, no bits change
1784 with this operator, just the type. This can be used for conversion of
1785 vector types to any other type, as long as they have the same bit width. For
Dan Gohman7305fa02008-09-08 16:45:59 +0000[diff] [blame]1786 pointers it is only valid to cast to another pointer type. It is not valid
1787 to bitcast to or from an aggregate type.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1788 </dd>
1789
1790 <dt><b><tt>getelementptr ( CSTPTR, IDX0, IDX1, ... )</tt></b></dt>
1791
1792 <dd>Perform the <a href="#i_getelementptr">getelementptr operation</a> on
1793 constants. As with the <a href="#i_getelementptr">getelementptr</a>
1794 instruction, the index list may have zero or more indexes, which are required
1795 to make sense for the type of "CSTPTR".</dd>
1796
1797 <dt><b><tt>select ( COND, VAL1, VAL2 )</tt></b></dt>
1798
1799 <dd>Perform the <a href="#i_select">select operation</a> on
1800 constants.</dd>
1801
1802 <dt><b><tt>icmp COND ( VAL1, VAL2 )</tt></b></dt>
1803 <dd>Performs the <a href="#i_icmp">icmp operation</a> on constants.</dd>
1804
1805 <dt><b><tt>fcmp COND ( VAL1, VAL2 )</tt></b></dt>
1806 <dd>Performs the <a href="#i_fcmp">fcmp operation</a> on constants.</dd>
1807
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]1808 <dt><b><tt>vicmp COND ( VAL1, VAL2 )</tt></b></dt>
1809 <dd>Performs the <a href="#i_vicmp">vicmp operation</a> on constants.</dd>
1810
1811 <dt><b><tt>vfcmp COND ( VAL1, VAL2 )</tt></b></dt>
1812 <dd>Performs the <a href="#i_vfcmp">vfcmp operation</a> on constants.</dd>
1813
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1814 <dt><b><tt>extractelement ( VAL, IDX )</tt></b></dt>
1815
1816 <dd>Perform the <a href="#i_extractelement">extractelement
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]1817 operation</a> on constants.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1818
1819 <dt><b><tt>insertelement ( VAL, ELT, IDX )</tt></b></dt>
1820
1821 <dd>Perform the <a href="#i_insertelement">insertelement
1822 operation</a> on constants.</dd>
1823
1824
1825 <dt><b><tt>shufflevector ( VEC1, VEC2, IDXMASK )</tt></b></dt>
1826
1827 <dd>Perform the <a href="#i_shufflevector">shufflevector
1828 operation</a> on constants.</dd>
1829
1830 <dt><b><tt>OPCODE ( LHS, RHS )</tt></b></dt>
1831
1832 <dd>Perform the specified operation of the LHS and RHS constants. OPCODE may
1833 be any of the <a href="#binaryops">binary</a> or <a href="#bitwiseops">bitwise
1834 binary</a> operations. The constraints on operands are the same as those for
1835 the corresponding instruction (e.g. no bitwise operations on floating point
1836 values are allowed).</dd>
1837</dl>
1838</div>
1839
1840<!-- *********************************************************************** -->
1841<div class="doc_section"> <a name="othervalues">Other Values</a> </div>
1842<!-- *********************************************************************** -->
1843
1844<!-- ======================================================================= -->
1845<div class="doc_subsection">
1846<a name="inlineasm">Inline Assembler Expressions</a>
1847</div>
1848
1849<div class="doc_text">
1850
1851<p>
1852LLVM supports inline assembler expressions (as opposed to <a href="#moduleasm">
1853Module-Level Inline Assembly</a>) through the use of a special value. This
1854value represents the inline assembler as a string (containing the instructions
1855to emit), a list of operand constraints (stored as a string), and a flag that
1856indicates whether or not the inline asm expression has side effects. An example
1857inline assembler expression is:
1858</p>
1859
1860<div class="doc_code">
1861<pre>
1862i32 (i32) asm "bswap $0", "=r,r"
1863</pre>
1864</div>
1865
1866<p>
1867Inline assembler expressions may <b>only</b> be used as the callee operand of
1868a <a href="#i_call"><tt>call</tt> instruction</a>. Thus, typically we have:
1869</p>
1870
1871<div class="doc_code">
1872<pre>
1873%X = call i32 asm "<a href="#int_bswap">bswap</a> $0", "=r,r"(i32 %Y)
1874</pre>
1875</div>
1876
1877<p>
1878Inline asms with side effects not visible in the constraint list must be marked
1879as having side effects. This is done through the use of the
1880'<tt>sideeffect</tt>' keyword, like so:
1881</p>
1882
1883<div class="doc_code">
1884<pre>
1885call void asm sideeffect "eieio", ""()
1886</pre>
1887</div>
1888
1889<p>TODO: The format of the asm and constraints string still need to be
1890documented here. Constraints on what can be done (e.g. duplication, moving, etc
Chris Lattner1cdd64e2008-10-04 18:36:02 +0000[diff] [blame]1891need to be documented). This is probably best done by reference to another
1892document that covers inline asm from a holistic perspective.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1893</p>
1894
1895</div>
1896
1897<!-- *********************************************************************** -->
1898<div class="doc_section"> <a name="instref">Instruction Reference</a> </div>
1899<!-- *********************************************************************** -->
1900
1901<div class="doc_text">
1902
1903<p>The LLVM instruction set consists of several different
1904classifications of instructions: <a href="#terminators">terminator
1905instructions</a>, <a href="#binaryops">binary instructions</a>,
1906<a href="#bitwiseops">bitwise binary instructions</a>, <a
1907 href="#memoryops">memory instructions</a>, and <a href="#otherops">other
1908instructions</a>.</p>
1909
1910</div>
1911
1912<!-- ======================================================================= -->
1913<div class="doc_subsection"> <a name="terminators">Terminator
1914Instructions</a> </div>
1915
1916<div class="doc_text">
1917
1918<p>As mentioned <a href="#functionstructure">previously</a>, every
1919basic block in a program ends with a "Terminator" instruction, which
1920indicates which block should be executed after the current block is
1921finished. These terminator instructions typically yield a '<tt>void</tt>'
1922value: they produce control flow, not values (the one exception being
1923the '<a href="#i_invoke"><tt>invoke</tt></a>' instruction).</p>
1924<p>There are six different terminator instructions: the '<a
1925 href="#i_ret"><tt>ret</tt></a>' instruction, the '<a href="#i_br"><tt>br</tt></a>'
1926instruction, the '<a href="#i_switch"><tt>switch</tt></a>' instruction,
1927the '<a href="#i_invoke"><tt>invoke</tt></a>' instruction, the '<a
1928 href="#i_unwind"><tt>unwind</tt></a>' instruction, and the '<a
1929 href="#i_unreachable"><tt>unreachable</tt></a>' instruction.</p>
1930
1931</div>
1932
1933<!-- _______________________________________________________________________ -->
1934<div class="doc_subsubsection"> <a name="i_ret">'<tt>ret</tt>'
1935Instruction</a> </div>
1936<div class="doc_text">
1937<h5>Syntax:</h5>
Dan Gohman3e700032008-10-04 19:00:07 +0000[diff] [blame]1938<pre>
1939 ret &lt;type&gt; &lt;value&gt; <i>; Return a value from a non-void function</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1940 ret void <i>; Return from void function</i>
1941</pre>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1942
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1943<h5>Overview:</h5>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1944
Dan Gohman3e700032008-10-04 19:00:07 +0000[diff] [blame]1945<p>The '<tt>ret</tt>' instruction is used to return control flow (and
1946optionally a value) from a function back to the caller.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1947<p>There are two forms of the '<tt>ret</tt>' instruction: one that
Dan Gohman3e700032008-10-04 19:00:07 +0000[diff] [blame]1948returns a value and then causes control flow, and one that just causes
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1949control flow to occur.</p>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1950
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1951<h5>Arguments:</h5>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1952
Dan Gohman3e700032008-10-04 19:00:07 +0000[diff] [blame]1953<p>The '<tt>ret</tt>' instruction optionally accepts a single argument,
1954the return value. The type of the return value must be a
1955'<a href="#t_firstclass">first class</a>' type.</p>
1956
1957<p>A function is not <a href="#wellformed">well formed</a> if
1958it it has a non-void return type and contains a '<tt>ret</tt>'
1959instruction with no return value or a return value with a type that
1960does not match its type, or if it has a void return type and contains
1961a '<tt>ret</tt>' instruction with a return value.</p>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1962
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1963<h5>Semantics:</h5>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1964
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1965<p>When the '<tt>ret</tt>' instruction is executed, control flow
1966returns back to the calling function's context. If the caller is a "<a
1967 href="#i_call"><tt>call</tt></a>" instruction, execution continues at
1968the instruction after the call. If the caller was an "<a
1969 href="#i_invoke"><tt>invoke</tt></a>" instruction, execution continues
1970at the beginning of the "normal" destination block. If the instruction
1971returns a value, that value shall set the call or invoke instruction's
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]1972return value.</p>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1973
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1974<h5>Example:</h5>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1975
1976<pre>
1977 ret i32 5 <i>; Return an integer value of 5</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1978 ret void <i>; Return from a void function</i>
Dan Gohman3e700032008-10-04 19:00:07 +0000[diff] [blame]1979 ret { i32, i8 } { i32 4, i8 2 } <i>; Return an aggregate of values 4 and 2</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1980</pre>
1981</div>
1982<!-- _______________________________________________________________________ -->
1983<div class="doc_subsubsection"> <a name="i_br">'<tt>br</tt>' Instruction</a> </div>
1984<div class="doc_text">
1985<h5>Syntax:</h5>
1986<pre> br i1 &lt;cond&gt;, label &lt;iftrue&gt;, label &lt;iffalse&gt;<br> br label &lt;dest&gt; <i>; Unconditional branch</i>
1987</pre>
1988<h5>Overview:</h5>
1989<p>The '<tt>br</tt>' instruction is used to cause control flow to
1990transfer to a different basic block in the current function. There are
1991two forms of this instruction, corresponding to a conditional branch
1992and an unconditional branch.</p>
1993<h5>Arguments:</h5>
1994<p>The conditional branch form of the '<tt>br</tt>' instruction takes a
1995single '<tt>i1</tt>' value and two '<tt>label</tt>' values. The
1996unconditional form of the '<tt>br</tt>' instruction takes a single
1997'<tt>label</tt>' value as a target.</p>
1998<h5>Semantics:</h5>
1999<p>Upon execution of a conditional '<tt>br</tt>' instruction, the '<tt>i1</tt>'
2000argument is evaluated. If the value is <tt>true</tt>, control flows
2001to the '<tt>iftrue</tt>' <tt>label</tt> argument. If "cond" is <tt>false</tt>,
2002control flows to the '<tt>iffalse</tt>' <tt>label</tt> argument.</p>
2003<h5>Example:</h5>
2004<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
2005 href="#i_ret">ret</a> i32 1<br>IfUnequal:<br> <a href="#i_ret">ret</a> i32 0<br></pre>
2006</div>
2007<!-- _______________________________________________________________________ -->
2008<div class="doc_subsubsection">
2009 <a name="i_switch">'<tt>switch</tt>' Instruction</a>
2010</div>
2011
2012<div class="doc_text">
2013<h5>Syntax:</h5>
2014
2015<pre>
2016 switch &lt;intty&gt; &lt;value&gt;, label &lt;defaultdest&gt; [ &lt;intty&gt; &lt;val&gt;, label &lt;dest&gt; ... ]
2017</pre>
2018
2019<h5>Overview:</h5>
2020
2021<p>The '<tt>switch</tt>' instruction is used to transfer control flow to one of
2022several different places. It is a generalization of the '<tt>br</tt>'
2023instruction, allowing a branch to occur to one of many possible
2024destinations.</p>
2025
2026
2027<h5>Arguments:</h5>
2028
2029<p>The '<tt>switch</tt>' instruction uses three parameters: an integer
2030comparison value '<tt>value</tt>', a default '<tt>label</tt>' destination, and
2031an array of pairs of comparison value constants and '<tt>label</tt>'s. The
2032table is not allowed to contain duplicate constant entries.</p>
2033
2034<h5>Semantics:</h5>
2035
2036<p>The <tt>switch</tt> instruction specifies a table of values and
2037destinations. When the '<tt>switch</tt>' instruction is executed, this
2038table is searched for the given value. If the value is found, control flow is
2039transfered to the corresponding destination; otherwise, control flow is
2040transfered to the default destination.</p>
2041
2042<h5>Implementation:</h5>
2043
2044<p>Depending on properties of the target machine and the particular
2045<tt>switch</tt> instruction, this instruction may be code generated in different
2046ways. For example, it could be generated as a series of chained conditional
2047branches or with a lookup table.</p>
2048
2049<h5>Example:</h5>
2050
2051<pre>
2052 <i>; Emulate a conditional br instruction</i>
2053 %Val = <a href="#i_zext">zext</a> i1 %value to i32
2054 switch i32 %Val, label %truedest [i32 0, label %falsedest ]
2055
2056 <i>; Emulate an unconditional br instruction</i>
2057 switch i32 0, label %dest [ ]
2058
2059 <i>; Implement a jump table:</i>
2060 switch i32 %val, label %otherwise [ i32 0, label %onzero
2061 i32 1, label %onone
2062 i32 2, label %ontwo ]
2063</pre>
2064</div>
2065
2066<!-- _______________________________________________________________________ -->
2067<div class="doc_subsubsection">
2068 <a name="i_invoke">'<tt>invoke</tt>' Instruction</a>
2069</div>
2070
2071<div class="doc_text">
2072
2073<h5>Syntax:</h5>
2074
2075<pre>
Devang Pateld0bfcc72008-10-07 17:48:33 +0000[diff] [blame]2076 &lt;result&gt; = invoke [<a href="#callingconv">cconv</a>] [<a href="#paramattrs">ret attrs</a>] &lt;ptr to function ty&gt; &lt;function ptr val&gt;(&lt;function args&gt;) [<a href="#fnattrs">fn attrs</a>]
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2077 to label &lt;normal label&gt; unwind label &lt;exception label&gt;
2078</pre>
2079
2080<h5>Overview:</h5>
2081
2082<p>The '<tt>invoke</tt>' instruction causes control to transfer to a specified
2083function, with the possibility of control flow transfer to either the
2084'<tt>normal</tt>' label or the
2085'<tt>exception</tt>' label. If the callee function returns with the
2086"<tt><a href="#i_ret">ret</a></tt>" instruction, control flow will return to the
2087"normal" label. If the callee (or any indirect callees) returns with the "<a
2088href="#i_unwind"><tt>unwind</tt></a>" instruction, control is interrupted and
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]2089continued at the dynamically nearest "exception" label.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2090
2091<h5>Arguments:</h5>
2092
2093<p>This instruction requires several arguments:</p>
2094
2095<ol>
2096 <li>
2097 The optional "cconv" marker indicates which <a href="#callingconv">calling
2098 convention</a> the call should use. If none is specified, the call defaults
2099 to using C calling conventions.
2100 </li>
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]2101
2102 <li>The optional <a href="#paramattrs">Parameter Attributes</a> list for
2103 return values. Only '<tt>zeroext</tt>', '<tt>signext</tt>',
2104 and '<tt>inreg</tt>' attributes are valid here.</li>
2105
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2106 <li>'<tt>ptr to function ty</tt>': shall be the signature of the pointer to
2107 function value being invoked. In most cases, this is a direct function
2108 invocation, but indirect <tt>invoke</tt>s are just as possible, branching off
2109 an arbitrary pointer to function value.
2110 </li>
2111
2112 <li>'<tt>function ptr val</tt>': An LLVM value containing a pointer to a
2113 function to be invoked. </li>
2114
2115 <li>'<tt>function args</tt>': argument list whose types match the function
2116 signature argument types. If the function signature indicates the function
2117 accepts a variable number of arguments, the extra arguments can be
2118 specified. </li>
2119
2120 <li>'<tt>normal label</tt>': the label reached when the called function
2121 executes a '<tt><a href="#i_ret">ret</a></tt>' instruction. </li>
2122
2123 <li>'<tt>exception label</tt>': the label reached when a callee returns with
2124 the <a href="#i_unwind"><tt>unwind</tt></a> instruction. </li>
2125
Devang Pateld0bfcc72008-10-07 17:48:33 +0000[diff] [blame]2126 <li>The optional <a href="#fnattrs">function attributes</a> list. Only
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]2127 '<tt>noreturn</tt>', '<tt>nounwind</tt>', '<tt>readonly</tt>' and
2128 '<tt>readnone</tt>' attributes are valid here.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2129</ol>
2130
2131<h5>Semantics:</h5>
2132
2133<p>This instruction is designed to operate as a standard '<tt><a
2134href="#i_call">call</a></tt>' instruction in most regards. The primary
2135difference is that it establishes an association with a label, which is used by
2136the runtime library to unwind the stack.</p>
2137
2138<p>This instruction is used in languages with destructors to ensure that proper
2139cleanup is performed in the case of either a <tt>longjmp</tt> or a thrown
2140exception. Additionally, this is important for implementation of
2141'<tt>catch</tt>' clauses in high-level languages that support them.</p>
2142
2143<h5>Example:</h5>
2144<pre>
Nick Lewyckya1c11a12008-03-16 07:18:12 +0000[diff] [blame]2145 %retval = invoke i32 @Test(i32 15) to label %Continue
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2146 unwind label %TestCleanup <i>; {i32}:retval set</i>
Nick Lewyckya1c11a12008-03-16 07:18:12 +0000[diff] [blame]2147 %retval = invoke <a href="#callingconv">coldcc</a> i32 %Testfnptr(i32 15) to label %Continue
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2148 unwind label %TestCleanup <i>; {i32}:retval set</i>
2149</pre>
2150</div>
2151
2152
2153<!-- _______________________________________________________________________ -->
2154
2155<div class="doc_subsubsection"> <a name="i_unwind">'<tt>unwind</tt>'
2156Instruction</a> </div>
2157
2158<div class="doc_text">
2159
2160<h5>Syntax:</h5>
2161<pre>
2162 unwind
2163</pre>
2164
2165<h5>Overview:</h5>
2166
2167<p>The '<tt>unwind</tt>' instruction unwinds the stack, continuing control flow
2168at the first callee in the dynamic call stack which used an <a
2169href="#i_invoke"><tt>invoke</tt></a> instruction to perform the call. This is
2170primarily used to implement exception handling.</p>
2171
2172<h5>Semantics:</h5>
2173
Chris Lattner8b094fc2008-04-19 21:01:16 +0000[diff] [blame]2174<p>The '<tt>unwind</tt>' instruction causes execution of the current function to
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2175immediately halt. The dynamic call stack is then searched for the first <a
2176href="#i_invoke"><tt>invoke</tt></a> instruction on the call stack. Once found,
2177execution continues at the "exceptional" destination block specified by the
2178<tt>invoke</tt> instruction. If there is no <tt>invoke</tt> instruction in the
2179dynamic call chain, undefined behavior results.</p>
2180</div>
2181
2182<!-- _______________________________________________________________________ -->
2183
2184<div class="doc_subsubsection"> <a name="i_unreachable">'<tt>unreachable</tt>'
2185Instruction</a> </div>
2186
2187<div class="doc_text">
2188
2189<h5>Syntax:</h5>
2190<pre>
2191 unreachable
2192</pre>
2193
2194<h5>Overview:</h5>
2195
2196<p>The '<tt>unreachable</tt>' instruction has no defined semantics. This
2197instruction is used to inform the optimizer that a particular portion of the
2198code is not reachable. This can be used to indicate that the code after a
2199no-return function cannot be reached, and other facts.</p>
2200
2201<h5>Semantics:</h5>
2202
2203<p>The '<tt>unreachable</tt>' instruction has no defined semantics.</p>
2204</div>
2205
2206
2207
2208<!-- ======================================================================= -->
2209<div class="doc_subsection"> <a name="binaryops">Binary Operations</a> </div>
2210<div class="doc_text">
2211<p>Binary operators are used to do most of the computation in a
Chris Lattnerab596d92008-04-01 18:47:32 +0000[diff] [blame]2212program. They require two operands of the same type, execute an operation on them, and
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2213produce a single value. The operands might represent
2214multiple 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]2215The result value has the same type as its operands.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2216<p>There are several different binary operators:</p>
2217</div>
2218<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2219<div class="doc_subsubsection">
2220 <a name="i_add">'<tt>add</tt>' Instruction</a>
2221</div>
2222
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2223<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2224
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2225<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2226
2227<pre>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2228 &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]2229</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2230
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2231<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2232
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2233<p>The '<tt>add</tt>' instruction returns the sum of its two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2234
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2235<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2236
2237<p>The two arguments to the '<tt>add</tt>' instruction must be <a
2238 href="#t_integer">integer</a>, <a href="#t_floating">floating point</a>, or
2239 <a href="#t_vector">vector</a> values. Both arguments must have identical
2240 types.</p>
2241
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2242<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2243
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2244<p>The value produced is the integer or floating point sum of the two
2245operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2246
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2247<p>If an integer sum has unsigned overflow, the result returned is the
2248mathematical result modulo 2<sup>n</sup>, where n is the bit width of
2249the result.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2250
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2251<p>Because LLVM integers use a two's complement representation, this
2252instruction is appropriate for both signed and unsigned integers.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2253
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2254<h5>Example:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2255
2256<pre>
2257 &lt;result&gt; = add i32 4, %var <i>; yields {i32}:result = 4 + %var</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2258</pre>
2259</div>
2260<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2261<div class="doc_subsubsection">
2262 <a name="i_sub">'<tt>sub</tt>' Instruction</a>
2263</div>
2264
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2265<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2266
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2267<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2268
2269<pre>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2270 &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]2271</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2272
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2273<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2274
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2275<p>The '<tt>sub</tt>' instruction returns the difference of its two
2276operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2277
2278<p>Note that the '<tt>sub</tt>' instruction is used to represent the
2279'<tt>neg</tt>' instruction present in most other intermediate
2280representations.</p>
2281
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2282<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2283
2284<p>The two arguments to the '<tt>sub</tt>' instruction must be <a
2285 href="#t_integer">integer</a>, <a href="#t_floating">floating point</a>,
2286 or <a href="#t_vector">vector</a> values. Both arguments must have identical
2287 types.</p>
2288
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2289<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2290
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2291<p>The value produced is the integer or floating point difference of
2292the two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2293
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2294<p>If an integer difference has unsigned overflow, the result returned is the
2295mathematical result modulo 2<sup>n</sup>, where n is the bit width of
2296the result.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2297
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2298<p>Because LLVM integers use a two's complement representation, this
2299instruction is appropriate for both signed and unsigned integers.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2300
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2301<h5>Example:</h5>
2302<pre>
2303 &lt;result&gt; = sub i32 4, %var <i>; yields {i32}:result = 4 - %var</i>
2304 &lt;result&gt; = sub i32 0, %val <i>; yields {i32}:result = -%var</i>
2305</pre>
2306</div>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2307
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2308<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2309<div class="doc_subsubsection">
2310 <a name="i_mul">'<tt>mul</tt>' Instruction</a>
2311</div>
2312
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2313<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2314
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2315<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2316<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]2317</pre>
2318<h5>Overview:</h5>
2319<p>The '<tt>mul</tt>' instruction returns the product of its two
2320operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2321
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2322<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2323
2324<p>The two arguments to the '<tt>mul</tt>' instruction must be <a
2325href="#t_integer">integer</a>, <a href="#t_floating">floating point</a>,
2326or <a href="#t_vector">vector</a> values. Both arguments must have identical
2327types.</p>
2328
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2329<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2330
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2331<p>The value produced is the integer or floating point product of the
2332two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2333
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2334<p>If the result of an integer multiplication has unsigned overflow,
2335the result returned is the mathematical result modulo
23362<sup>n</sup>, where n is the bit width of the result.</p>
2337<p>Because LLVM integers use a two's complement representation, and the
2338result is the same width as the operands, this instruction returns the
2339correct result for both signed and unsigned integers. If a full product
2340(e.g. <tt>i32</tt>x<tt>i32</tt>-><tt>i64</tt>) is needed, the operands
2341should be sign-extended or zero-extended as appropriate to the
2342width of the full product.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2343<h5>Example:</h5>
2344<pre> &lt;result&gt; = mul i32 4, %var <i>; yields {i32}:result = 4 * %var</i>
2345</pre>
2346</div>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2347
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2348<!-- _______________________________________________________________________ -->
2349<div class="doc_subsubsection"> <a name="i_udiv">'<tt>udiv</tt>' Instruction
2350</a></div>
2351<div class="doc_text">
2352<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2353<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]2354</pre>
2355<h5>Overview:</h5>
2356<p>The '<tt>udiv</tt>' instruction returns the quotient of its two
2357operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2358
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2359<h5>Arguments:</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 two arguments to the '<tt>udiv</tt>' instruction must be
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2362<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2363values. Both arguments must have identical types.</p>
2364
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2365<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2366
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2367<p>The value produced is the unsigned integer quotient of the two operands.</p>
2368<p>Note that unsigned integer division and signed integer division are distinct
2369operations; for signed integer division, use '<tt>sdiv</tt>'.</p>
2370<p>Division by zero leads to undefined behavior.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2371<h5>Example:</h5>
2372<pre> &lt;result&gt; = udiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
2373</pre>
2374</div>
2375<!-- _______________________________________________________________________ -->
2376<div class="doc_subsubsection"> <a name="i_sdiv">'<tt>sdiv</tt>' Instruction
2377</a> </div>
2378<div class="doc_text">
2379<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2380<pre>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2381 &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]2382</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2383
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2384<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2385
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2386<p>The '<tt>sdiv</tt>' instruction returns the quotient of its two
2387operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2388
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2389<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2390
2391<p>The two arguments to the '<tt>sdiv</tt>' instruction must be
2392<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2393values. Both arguments must have identical types.</p>
2394
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2395<h5>Semantics:</h5>
Chris Lattnerdc15b1d2008-04-01 18:45:27 +0000[diff] [blame]2396<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]2397<p>Note that signed integer division and unsigned integer division are distinct
2398operations; for unsigned integer division, use '<tt>udiv</tt>'.</p>
2399<p>Division by zero leads to undefined behavior. Overflow also leads to
2400undefined behavior; this is a rare case, but can occur, for example,
2401by doing a 32-bit division of -2147483648 by -1.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2402<h5>Example:</h5>
2403<pre> &lt;result&gt; = sdiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
2404</pre>
2405</div>
2406<!-- _______________________________________________________________________ -->
2407<div class="doc_subsubsection"> <a name="i_fdiv">'<tt>fdiv</tt>'
2408Instruction</a> </div>
2409<div class="doc_text">
2410<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2411<pre>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2412 &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]2413</pre>
2414<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2415
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2416<p>The '<tt>fdiv</tt>' instruction returns the quotient of its two
2417operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2418
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2419<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2420
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2421<p>The two arguments to the '<tt>fdiv</tt>' instruction must be
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2422<a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a>
2423of floating point values. Both arguments must have identical types.</p>
2424
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2425<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2426
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2427<p>The value produced is the floating point quotient of the two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2428
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2429<h5>Example:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2430
2431<pre>
2432 &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]2433</pre>
2434</div>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2435
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2436<!-- _______________________________________________________________________ -->
2437<div class="doc_subsubsection"> <a name="i_urem">'<tt>urem</tt>' Instruction</a>
2438</div>
2439<div class="doc_text">
2440<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2441<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]2442</pre>
2443<h5>Overview:</h5>
2444<p>The '<tt>urem</tt>' instruction returns the remainder from the
2445unsigned division of its two arguments.</p>
2446<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2447<p>The two arguments to the '<tt>urem</tt>' instruction must be
2448<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2449values. Both arguments must have identical types.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2450<h5>Semantics:</h5>
2451<p>This instruction returns the unsigned integer <i>remainder</i> of a division.
Chris Lattnerdc15b1d2008-04-01 18:45:27 +0000[diff] [blame]2452This instruction always performs an unsigned division to get the remainder.</p>
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2453<p>Note that unsigned integer remainder and signed integer remainder are
2454distinct operations; for signed integer remainder, use '<tt>srem</tt>'.</p>
2455<p>Taking the remainder of a division by zero leads to undefined behavior.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2456<h5>Example:</h5>
2457<pre> &lt;result&gt; = urem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
2458</pre>
2459
2460</div>
2461<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2462<div class="doc_subsubsection">
2463 <a name="i_srem">'<tt>srem</tt>' Instruction</a>
2464</div>
2465
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2466<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2467
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2468<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2469
2470<pre>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2471 &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]2472</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2473
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2474<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2475
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2476<p>The '<tt>srem</tt>' instruction returns the remainder from the
Dan Gohman3e3fd8c2007-11-05 23:35:22 +0000[diff] [blame]2477signed division of its two operands. This instruction can also take
2478<a href="#t_vector">vector</a> versions of the values in which case
2479the elements must be integers.</p>
Chris Lattner08497ce2008-01-04 04:33:49 +0000[diff] [blame]2480
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2481<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2482
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2483<p>The two arguments to the '<tt>srem</tt>' instruction must be
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2484<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2485values. Both arguments must have identical types.</p>
2486
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2487<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2488
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2489<p>This instruction returns the <i>remainder</i> of a division (where the result
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2490has the same sign as the dividend, <tt>op1</tt>), not the <i>modulo</i>
2491operator (where the result has the same sign as the divisor, <tt>op2</tt>) of
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2492a value. For more information about the difference, see <a
2493 href="http://mathforum.org/dr.math/problems/anne.4.28.99.html">The
2494Math Forum</a>. For a table of how this is implemented in various languages,
2495please see <a href="http://en.wikipedia.org/wiki/Modulo_operation">
2496Wikipedia: modulo operation</a>.</p>
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2497<p>Note that signed integer remainder and unsigned integer remainder are
2498distinct operations; for unsigned integer remainder, use '<tt>urem</tt>'.</p>
2499<p>Taking the remainder of a division by zero leads to undefined behavior.
2500Overflow also leads to undefined behavior; this is a rare case, but can occur,
2501for example, by taking the remainder of a 32-bit division of -2147483648 by -1.
2502(The remainder doesn't actually overflow, but this rule lets srem be
2503implemented using instructions that return both the result of the division
2504and the remainder.)</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2505<h5>Example:</h5>
2506<pre> &lt;result&gt; = srem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
2507</pre>
2508
2509</div>
2510<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2511<div class="doc_subsubsection">
2512 <a name="i_frem">'<tt>frem</tt>' Instruction</a> </div>
2513
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2514<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2515
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2516<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2517<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]2518</pre>
2519<h5>Overview:</h5>
2520<p>The '<tt>frem</tt>' instruction returns the remainder from the
2521division of its two operands.</p>
2522<h5>Arguments:</h5>
2523<p>The two arguments to the '<tt>frem</tt>' instruction must be
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2524<a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a>
2525of floating point values. Both arguments must have identical types.</p>
2526
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2527<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2528
Chris Lattnerdc15b1d2008-04-01 18:45:27 +0000[diff] [blame]2529<p>This instruction returns the <i>remainder</i> of a division.
2530The remainder has the same sign as the dividend.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2531
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2532<h5>Example:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2533
2534<pre>
2535 &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]2536</pre>
2537</div>
2538
2539<!-- ======================================================================= -->
2540<div class="doc_subsection"> <a name="bitwiseops">Bitwise Binary
2541Operations</a> </div>
2542<div class="doc_text">
2543<p>Bitwise binary operators are used to do various forms of
2544bit-twiddling in a program. They are generally very efficient
2545instructions and can commonly be strength reduced from other
Chris Lattnerdc15b1d2008-04-01 18:45:27 +0000[diff] [blame]2546instructions. They require two operands of the same type, execute an operation on them,
2547and 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]2548</div>
2549
2550<!-- _______________________________________________________________________ -->
2551<div class="doc_subsubsection"> <a name="i_shl">'<tt>shl</tt>'
2552Instruction</a> </div>
2553<div class="doc_text">
2554<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2555<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]2556</pre>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2557
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2558<h5>Overview:</h5>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2559
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2560<p>The '<tt>shl</tt>' instruction returns the first operand shifted to
2561the left a specified number of bits.</p>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2562
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2563<h5>Arguments:</h5>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2564
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2565<p>Both arguments to the '<tt>shl</tt>' instruction must be the same <a
Nate Begemanbb1ce942008-07-29 15:49:41 +0000[diff] [blame]2566 href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2567type. '<tt>op2</tt>' is treated as an unsigned value.</p>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2568
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2569<h5>Semantics:</h5>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2570
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2571<p>The value produced is <tt>op1</tt> * 2<sup><tt>op2</tt></sup> mod 2<sup>n</sup>,
2572where n is the width of the result. If <tt>op2</tt> is (statically or dynamically) negative or
2573equal 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]2574
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2575<h5>Example:</h5><pre>
2576 &lt;result&gt; = shl i32 4, %var <i>; yields {i32}: 4 &lt;&lt; %var</i>
2577 &lt;result&gt; = shl i32 4, 2 <i>; yields {i32}: 16</i>
2578 &lt;result&gt; = shl i32 1, 10 <i>; yields {i32}: 1024</i>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2579 &lt;result&gt; = shl i32 1, 32 <i>; undefined</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2580</pre>
2581</div>
2582<!-- _______________________________________________________________________ -->
2583<div class="doc_subsubsection"> <a name="i_lshr">'<tt>lshr</tt>'
2584Instruction</a> </div>
2585<div class="doc_text">
2586<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2587<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]2588</pre>
2589
2590<h5>Overview:</h5>
2591<p>The '<tt>lshr</tt>' instruction (logical shift right) returns the first
2592operand shifted to the right a specified number of bits with zero fill.</p>
2593
2594<h5>Arguments:</h5>
2595<p>Both arguments to the '<tt>lshr</tt>' instruction must be the same
Nate Begemanbb1ce942008-07-29 15:49:41 +0000[diff] [blame]2596<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]2597type. '<tt>op2</tt>' is treated as an unsigned value.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2598
2599<h5>Semantics:</h5>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2600
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2601<p>This instruction always performs a logical shift right operation. The most
2602significant bits of the result will be filled with zero bits after the
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2603shift. If <tt>op2</tt> is (statically or dynamically) equal to or larger than
2604the number of bits in <tt>op1</tt>, the result is undefined.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2605
2606<h5>Example:</h5>
2607<pre>
2608 &lt;result&gt; = lshr i32 4, 1 <i>; yields {i32}:result = 2</i>
2609 &lt;result&gt; = lshr i32 4, 2 <i>; yields {i32}:result = 1</i>
2610 &lt;result&gt; = lshr i8 4, 3 <i>; yields {i8}:result = 0</i>
2611 &lt;result&gt; = lshr i8 -2, 1 <i>; yields {i8}:result = 0x7FFFFFFF </i>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2612 &lt;result&gt; = lshr i32 1, 32 <i>; undefined</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2613</pre>
2614</div>
2615
2616<!-- _______________________________________________________________________ -->
2617<div class="doc_subsubsection"> <a name="i_ashr">'<tt>ashr</tt>'
2618Instruction</a> </div>
2619<div class="doc_text">
2620
2621<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2622<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]2623</pre>
2624
2625<h5>Overview:</h5>
2626<p>The '<tt>ashr</tt>' instruction (arithmetic shift right) returns the first
2627operand shifted to the right a specified number of bits with sign extension.</p>
2628
2629<h5>Arguments:</h5>
2630<p>Both arguments to the '<tt>ashr</tt>' instruction must be the same
Nate Begemanbb1ce942008-07-29 15:49:41 +0000[diff] [blame]2631<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]2632type. '<tt>op2</tt>' is treated as an unsigned value.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2633
2634<h5>Semantics:</h5>
2635<p>This instruction always performs an arithmetic shift right operation,
2636The most significant bits of the result will be filled with the sign bit
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2637of <tt>op1</tt>. If <tt>op2</tt> is (statically or dynamically) equal to or
2638larger than the number of bits in <tt>op1</tt>, the result is undefined.
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2639</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2640
2641<h5>Example:</h5>
2642<pre>
2643 &lt;result&gt; = ashr i32 4, 1 <i>; yields {i32}:result = 2</i>
2644 &lt;result&gt; = ashr i32 4, 2 <i>; yields {i32}:result = 1</i>
2645 &lt;result&gt; = ashr i8 4, 3 <i>; yields {i8}:result = 0</i>
2646 &lt;result&gt; = ashr i8 -2, 1 <i>; yields {i8}:result = -1</i>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2647 &lt;result&gt; = ashr i32 1, 32 <i>; undefined</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2648</pre>
2649</div>
2650
2651<!-- _______________________________________________________________________ -->
2652<div class="doc_subsubsection"> <a name="i_and">'<tt>and</tt>'
2653Instruction</a> </div>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2654
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2655<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2656
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2657<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2658
2659<pre>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2660 &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]2661</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2662
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2663<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2664
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2665<p>The '<tt>and</tt>' instruction returns the bitwise logical and of
2666its two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2667
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2668<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2669
2670<p>The two arguments to the '<tt>and</tt>' instruction must be
2671<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2672values. Both arguments must have identical types.</p>
2673
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2674<h5>Semantics:</h5>
2675<p>The truth table used for the '<tt>and</tt>' instruction is:</p>
2676<p> </p>
Bill Wendling1e19eac2008-09-07 10:29:20 +0000[diff] [blame]2677<div>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2678<table border="1" cellspacing="0" cellpadding="4">
2679 <tbody>
2680 <tr>
2681 <td>In0</td>
2682 <td>In1</td>
2683 <td>Out</td>
2684 </tr>
2685 <tr>
2686 <td>0</td>
2687 <td>0</td>
2688 <td>0</td>
2689 </tr>
2690 <tr>
2691 <td>0</td>
2692 <td>1</td>
2693 <td>0</td>
2694 </tr>
2695 <tr>
2696 <td>1</td>
2697 <td>0</td>
2698 <td>0</td>
2699 </tr>
2700 <tr>
2701 <td>1</td>
2702 <td>1</td>
2703 <td>1</td>
2704 </tr>
2705 </tbody>
2706</table>
2707</div>
2708<h5>Example:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2709<pre>
2710 &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]2711 &lt;result&gt; = and i32 15, 40 <i>; yields {i32}:result = 8</i>
2712 &lt;result&gt; = and i32 4, 8 <i>; yields {i32}:result = 0</i>
2713</pre>
2714</div>
2715<!-- _______________________________________________________________________ -->
2716<div class="doc_subsubsection"> <a name="i_or">'<tt>or</tt>' Instruction</a> </div>
2717<div class="doc_text">
2718<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2719<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]2720</pre>
2721<h5>Overview:</h5>
2722<p>The '<tt>or</tt>' instruction returns the bitwise logical inclusive
2723or of its two operands.</p>
2724<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2725
2726<p>The two arguments to the '<tt>or</tt>' instruction must be
2727<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2728values. Both arguments must have identical types.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2729<h5>Semantics:</h5>
2730<p>The truth table used for the '<tt>or</tt>' instruction is:</p>
2731<p> </p>
Bill Wendling1e19eac2008-09-07 10:29:20 +0000[diff] [blame]2732<div>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2733<table border="1" cellspacing="0" cellpadding="4">
2734 <tbody>
2735 <tr>
2736 <td>In0</td>
2737 <td>In1</td>
2738 <td>Out</td>
2739 </tr>
2740 <tr>
2741 <td>0</td>
2742 <td>0</td>
2743 <td>0</td>
2744 </tr>
2745 <tr>
2746 <td>0</td>
2747 <td>1</td>
2748 <td>1</td>
2749 </tr>
2750 <tr>
2751 <td>1</td>
2752 <td>0</td>
2753 <td>1</td>
2754 </tr>
2755 <tr>
2756 <td>1</td>
2757 <td>1</td>
2758 <td>1</td>
2759 </tr>
2760 </tbody>
2761</table>
2762</div>
2763<h5>Example:</h5>
2764<pre> &lt;result&gt; = or i32 4, %var <i>; yields {i32}:result = 4 | %var</i>
2765 &lt;result&gt; = or i32 15, 40 <i>; yields {i32}:result = 47</i>
2766 &lt;result&gt; = or i32 4, 8 <i>; yields {i32}:result = 12</i>
2767</pre>
2768</div>
2769<!-- _______________________________________________________________________ -->
2770<div class="doc_subsubsection"> <a name="i_xor">'<tt>xor</tt>'
2771Instruction</a> </div>
2772<div class="doc_text">
2773<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2774<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]2775</pre>
2776<h5>Overview:</h5>
2777<p>The '<tt>xor</tt>' instruction returns the bitwise logical exclusive
2778or of its two operands. The <tt>xor</tt> is used to implement the
2779"one's complement" operation, which is the "~" operator in C.</p>
2780<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2781<p>The two arguments to the '<tt>xor</tt>' instruction must be
2782<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2783values. Both arguments must have identical types.</p>
2784
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2785<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2786
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2787<p>The truth table used for the '<tt>xor</tt>' instruction is:</p>
2788<p> </p>
Bill Wendling1e19eac2008-09-07 10:29:20 +0000[diff] [blame]2789<div>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2790<table border="1" cellspacing="0" cellpadding="4">
2791 <tbody>
2792 <tr>
2793 <td>In0</td>
2794 <td>In1</td>
2795 <td>Out</td>
2796 </tr>
2797 <tr>
2798 <td>0</td>
2799 <td>0</td>
2800 <td>0</td>
2801 </tr>
2802 <tr>
2803 <td>0</td>
2804 <td>1</td>
2805 <td>1</td>
2806 </tr>
2807 <tr>
2808 <td>1</td>
2809 <td>0</td>
2810 <td>1</td>
2811 </tr>
2812 <tr>
2813 <td>1</td>
2814 <td>1</td>
2815 <td>0</td>
2816 </tr>
2817 </tbody>
2818</table>
2819</div>
2820<p> </p>
2821<h5>Example:</h5>
2822<pre> &lt;result&gt; = xor i32 4, %var <i>; yields {i32}:result = 4 ^ %var</i>
2823 &lt;result&gt; = xor i32 15, 40 <i>; yields {i32}:result = 39</i>
2824 &lt;result&gt; = xor i32 4, 8 <i>; yields {i32}:result = 12</i>
2825 &lt;result&gt; = xor i32 %V, -1 <i>; yields {i32}:result = ~%V</i>
2826</pre>
2827</div>
2828
2829<!-- ======================================================================= -->
2830<div class="doc_subsection">
2831 <a name="vectorops">Vector Operations</a>
2832</div>
2833
2834<div class="doc_text">
2835
2836<p>LLVM supports several instructions to represent vector operations in a
2837target-independent manner. These instructions cover the element-access and
2838vector-specific operations needed to process vectors effectively. While LLVM
2839does directly support these vector operations, many sophisticated algorithms
2840will want to use target-specific intrinsics to take full advantage of a specific
2841target.</p>
2842
2843</div>
2844
2845<!-- _______________________________________________________________________ -->
2846<div class="doc_subsubsection">
2847 <a name="i_extractelement">'<tt>extractelement</tt>' Instruction</a>
2848</div>
2849
2850<div class="doc_text">
2851
2852<h5>Syntax:</h5>
2853
2854<pre>
2855 &lt;result&gt; = extractelement &lt;n x &lt;ty&gt;&gt; &lt;val&gt;, i32 &lt;idx&gt; <i>; yields &lt;ty&gt;</i>
2856</pre>
2857
2858<h5>Overview:</h5>
2859
2860<p>
2861The '<tt>extractelement</tt>' instruction extracts a single scalar
2862element from a vector at a specified index.
2863</p>
2864
2865
2866<h5>Arguments:</h5>
2867
2868<p>
2869The first operand of an '<tt>extractelement</tt>' instruction is a
2870value of <a href="#t_vector">vector</a> type. The second operand is
2871an index indicating the position from which to extract the element.
2872The index may be a variable.</p>
2873
2874<h5>Semantics:</h5>
2875
2876<p>
2877The result is a scalar of the same type as the element type of
2878<tt>val</tt>. Its value is the value at position <tt>idx</tt> of
2879<tt>val</tt>. If <tt>idx</tt> exceeds the length of <tt>val</tt>, the
2880results are undefined.
2881</p>
2882
2883<h5>Example:</h5>
2884
2885<pre>
2886 %result = extractelement &lt;4 x i32&gt; %vec, i32 0 <i>; yields i32</i>
2887</pre>
2888</div>
2889
2890
2891<!-- _______________________________________________________________________ -->
2892<div class="doc_subsubsection">
2893 <a name="i_insertelement">'<tt>insertelement</tt>' Instruction</a>
2894</div>
2895
2896<div class="doc_text">
2897
2898<h5>Syntax:</h5>
2899
2900<pre>
Dan Gohmanbcc3c502008-05-12 23:38:42 +0000[diff] [blame]2901 &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]2902</pre>
2903
2904<h5>Overview:</h5>
2905
2906<p>
2907The '<tt>insertelement</tt>' instruction inserts a scalar
2908element into a vector at a specified index.
2909</p>
2910
2911
2912<h5>Arguments:</h5>
2913
2914<p>
2915The first operand of an '<tt>insertelement</tt>' instruction is a
2916value of <a href="#t_vector">vector</a> type. The second operand is a
2917scalar value whose type must equal the element type of the first
2918operand. The third operand is an index indicating the position at
2919which to insert the value. The index may be a variable.</p>
2920
2921<h5>Semantics:</h5>
2922
2923<p>
2924The result is a vector of the same type as <tt>val</tt>. Its
2925element values are those of <tt>val</tt> except at position
2926<tt>idx</tt>, where it gets the value <tt>elt</tt>. If <tt>idx</tt>
2927exceeds the length of <tt>val</tt>, the results are undefined.
2928</p>
2929
2930<h5>Example:</h5>
2931
2932<pre>
2933 %result = insertelement &lt;4 x i32&gt; %vec, i32 1, i32 0 <i>; yields &lt;4 x i32&gt;</i>
2934</pre>
2935</div>
2936
2937<!-- _______________________________________________________________________ -->
2938<div class="doc_subsubsection">
2939 <a name="i_shufflevector">'<tt>shufflevector</tt>' Instruction</a>
2940</div>
2941
2942<div class="doc_text">
2943
2944<h5>Syntax:</h5>
2945
2946<pre>
Mon P Wangbff5d9c2008-11-10 04:46:22 +0000[diff] [blame]2947 &lt;result&gt; = shufflevector &lt;n x &lt;ty&gt;&gt; &lt;v1&gt;, &lt;n x &lt;ty&gt;&gt; &lt;v2&gt;, &lt;m x i32&gt; &lt;mask&gt; <i>; yields &lt;m x &lt;ty&gt;&gt;</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2948</pre>
2949
2950<h5>Overview:</h5>
2951
2952<p>
2953The '<tt>shufflevector</tt>' instruction constructs a permutation of elements
Mon P Wangbff5d9c2008-11-10 04:46:22 +0000[diff] [blame]2954from two input vectors, returning a vector with the same element type as
2955the input and length that is the same as the shuffle mask.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2956</p>
2957
2958<h5>Arguments:</h5>
2959
2960<p>
Mon P Wangbff5d9c2008-11-10 04:46:22 +0000[diff] [blame]2961The first two operands of a '<tt>shufflevector</tt>' instruction are vectors
2962with types that match each other. The third argument is a shuffle mask whose
2963element type is always 'i32'. The result of the instruction is a vector whose
2964length is the same as the shuffle mask and whose element type is the same as
2965the element type of the first two operands.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2966</p>
2967
2968<p>
2969The shuffle mask operand is required to be a constant vector with either
2970constant integer or undef values.
2971</p>
2972
2973<h5>Semantics:</h5>
2974
2975<p>
2976The elements of the two input vectors are numbered from left to right across
2977both of the vectors. The shuffle mask operand specifies, for each element of
Mon P Wangbff5d9c2008-11-10 04:46:22 +0000[diff] [blame]2978the result vector, which element of the two input vectors the result element
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2979gets. The element selector may be undef (meaning "don't care") and the second
2980operand may be undef if performing a shuffle from only one vector.
2981</p>
2982
2983<h5>Example:</h5>
2984
2985<pre>
2986 %result = shufflevector &lt;4 x i32&gt; %v1, &lt;4 x i32&gt; %v2,
2987 &lt;4 x i32&gt; &lt;i32 0, i32 4, i32 1, i32 5&gt; <i>; yields &lt;4 x i32&gt;</i>
2988 %result = shufflevector &lt;4 x i32&gt; %v1, &lt;4 x i32&gt; undef,
2989 &lt;4 x i32&gt; &lt;i32 0, i32 1, i32 2, i32 3&gt; <i>; yields &lt;4 x i32&gt;</i> - Identity shuffle.
Mon P Wangbff5d9c2008-11-10 04:46:22 +0000[diff] [blame]2990 %result = shufflevector &lt;8 x i32&gt; %v1, &lt;8 x i32&gt; undef,
2991 &lt;4 x i32&gt; &lt;i32 0, i32 1, i32 2, i32 3&gt; <i>; yields &lt;4 x i32&gt;</i>
2992 %result = shufflevector &lt;4 x i32&gt; %v1, &lt;4 x i32&gt; %v2,
2993 &lt;8 x i32&gt; &lt;i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7 &gt; <i>; yields &lt;8 x i32&gt;</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2994</pre>
2995</div>
2996
2997
2998<!-- ======================================================================= -->
2999<div class="doc_subsection">
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3000 <a name="aggregateops">Aggregate Operations</a>
3001</div>
3002
3003<div class="doc_text">
3004
3005<p>LLVM supports several instructions for working with aggregate values.
3006</p>
3007
3008</div>
3009
3010<!-- _______________________________________________________________________ -->
3011<div class="doc_subsubsection">
3012 <a name="i_extractvalue">'<tt>extractvalue</tt>' Instruction</a>
3013</div>
3014
3015<div class="doc_text">
3016
3017<h5>Syntax:</h5>
3018
3019<pre>
3020 &lt;result&gt; = extractvalue &lt;aggregate type&gt; &lt;val&gt;, &lt;idx&gt;{, &lt;idx&gt;}*
3021</pre>
3022
3023<h5>Overview:</h5>
3024
3025<p>
Dan Gohman6c6dea02008-05-13 18:16:06 +0000[diff] [blame]3026The '<tt>extractvalue</tt>' instruction extracts the value of a struct field
3027or array element from an aggregate value.
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3028</p>
3029
3030
3031<h5>Arguments:</h5>
3032
3033<p>
3034The first operand of an '<tt>extractvalue</tt>' instruction is a
3035value of <a href="#t_struct">struct</a> or <a href="#t_array">array</a>
Dan Gohman6c6dea02008-05-13 18:16:06 +0000[diff] [blame]3036type. The operands are constant indices to specify which value to extract
Dan Gohmane5febe42008-05-31 00:58:22 +0000[diff] [blame]3037in a similar manner as indices in a
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3038'<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.
3039</p>
3040
3041<h5>Semantics:</h5>
3042
3043<p>
3044The result is the value at the position in the aggregate specified by
3045the index operands.
3046</p>
3047
3048<h5>Example:</h5>
3049
3050<pre>
Dan Gohmane5febe42008-05-31 00:58:22 +0000[diff] [blame]3051 %result = extractvalue {i32, float} %agg, 0 <i>; yields i32</i>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3052</pre>
3053</div>
3054
3055
3056<!-- _______________________________________________________________________ -->
3057<div class="doc_subsubsection">
3058 <a name="i_insertvalue">'<tt>insertvalue</tt>' Instruction</a>
3059</div>
3060
3061<div class="doc_text">
3062
3063<h5>Syntax:</h5>
3064
3065<pre>
Dan Gohmane5febe42008-05-31 00:58:22 +0000[diff] [blame]3066 &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]3067</pre>
3068
3069<h5>Overview:</h5>
3070
3071<p>
3072The '<tt>insertvalue</tt>' instruction inserts a value
Dan Gohman6c6dea02008-05-13 18:16:06 +0000[diff] [blame]3073into a struct field or array element in an aggregate.
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3074</p>
3075
3076
3077<h5>Arguments:</h5>
3078
3079<p>
3080The first operand of an '<tt>insertvalue</tt>' instruction is a
3081value of <a href="#t_struct">struct</a> or <a href="#t_array">array</a> type.
3082The second operand is a first-class value to insert.
Dan Gohman4f29e422008-05-23 21:53:15 +0000[diff] [blame]3083The following operands are constant indices
Dan Gohmane5febe42008-05-31 00:58:22 +0000[diff] [blame]3084indicating the position at which to insert the value in a similar manner as
Dan Gohman6c6dea02008-05-13 18:16:06 +0000[diff] [blame]3085indices in a
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3086'<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.
3087The value to insert must have the same type as the value identified
Dan Gohman6c6dea02008-05-13 18:16:06 +0000[diff] [blame]3088by the indices.
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]3089</p>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3090
3091<h5>Semantics:</h5>
3092
3093<p>
3094The result is an aggregate of the same type as <tt>val</tt>. Its
3095value is that of <tt>val</tt> except that the value at the position
Dan Gohman6c6dea02008-05-13 18:16:06 +0000[diff] [blame]3096specified by the indices is that of <tt>elt</tt>.
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3097</p>
3098
3099<h5>Example:</h5>
3100
3101<pre>
Dan Gohmanb1aab4e2008-06-23 15:26:37 +0000[diff] [blame]3102 %result = insertvalue {i32, float} %agg, i32 1, 0 <i>; yields {i32, float}</i>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3103</pre>
3104</div>
3105
3106
3107<!-- ======================================================================= -->
3108<div class="doc_subsection">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3109 <a name="memoryops">Memory Access and Addressing Operations</a>
3110</div>
3111
3112<div class="doc_text">
3113
3114<p>A key design point of an SSA-based representation is how it
3115represents memory. In LLVM, no memory locations are in SSA form, which
3116makes things very simple. This section describes how to read, write,
3117allocate, and free memory in LLVM.</p>
3118
3119</div>
3120
3121<!-- _______________________________________________________________________ -->
3122<div class="doc_subsubsection">
3123 <a name="i_malloc">'<tt>malloc</tt>' Instruction</a>
3124</div>
3125
3126<div class="doc_text">
3127
3128<h5>Syntax:</h5>
3129
3130<pre>
3131 &lt;result&gt; = malloc &lt;type&gt;[, i32 &lt;NumElements&gt;][, align &lt;alignment&gt;] <i>; yields {type*}:result</i>
3132</pre>
3133
3134<h5>Overview:</h5>
3135
3136<p>The '<tt>malloc</tt>' instruction allocates memory from the system
Christopher Lambcfe00962007-12-17 01:00:21 +0000[diff] [blame]3137heap and returns a pointer to it. The object is always allocated in the generic
3138address space (address space zero).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3139
3140<h5>Arguments:</h5>
3141
3142<p>The '<tt>malloc</tt>' instruction allocates
3143<tt>sizeof(&lt;type&gt;)*NumElements</tt>
3144bytes of memory from the operating system and returns a pointer of the
3145appropriate type to the program. If "NumElements" is specified, it is the
Gabor Greif5082cf42008-02-09 22:24:34 +0000[diff] [blame]3146number of elements allocated, otherwise "NumElements" is defaulted to be one.
Chris Lattner10368b62008-04-02 00:38:26 +0000[diff] [blame]3147If a constant alignment is specified, the value result of the allocation is guaranteed to
Gabor Greif5082cf42008-02-09 22:24:34 +0000[diff] [blame]3148be aligned to at least that boundary. If not specified, or if zero, the target can
3149choose to align the allocation on any convenient boundary.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3150
3151<p>'<tt>type</tt>' must be a sized type.</p>
3152
3153<h5>Semantics:</h5>
3154
3155<p>Memory is allocated using the system "<tt>malloc</tt>" function, and
Chris Lattner8b094fc2008-04-19 21:01:16 +0000[diff] [blame]3156a pointer is returned. The result of a zero byte allocattion is undefined. The
3157result is null if there is insufficient memory available.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3158
3159<h5>Example:</h5>
3160
3161<pre>
3162 %array = malloc [4 x i8 ] <i>; yields {[%4 x i8]*}:array</i>
3163
3164 %size = <a href="#i_add">add</a> i32 2, 2 <i>; yields {i32}:size = i32 4</i>
3165 %array1 = malloc i8, i32 4 <i>; yields {i8*}:array1</i>
3166 %array2 = malloc [12 x i8], i32 %size <i>; yields {[12 x i8]*}:array2</i>
3167 %array3 = malloc i32, i32 4, align 1024 <i>; yields {i32*}:array3</i>
3168 %array4 = malloc i32, align 1024 <i>; yields {i32*}:array4</i>
3169</pre>
3170</div>
3171
3172<!-- _______________________________________________________________________ -->
3173<div class="doc_subsubsection">
3174 <a name="i_free">'<tt>free</tt>' Instruction</a>
3175</div>
3176
3177<div class="doc_text">
3178
3179<h5>Syntax:</h5>
3180
3181<pre>
3182 free &lt;type&gt; &lt;value&gt; <i>; yields {void}</i>
3183</pre>
3184
3185<h5>Overview:</h5>
3186
3187<p>The '<tt>free</tt>' instruction returns memory back to the unused
3188memory heap to be reallocated in the future.</p>
3189
3190<h5>Arguments:</h5>
3191
3192<p>'<tt>value</tt>' shall be a pointer value that points to a value
3193that was allocated with the '<tt><a href="#i_malloc">malloc</a></tt>'
3194instruction.</p>
3195
3196<h5>Semantics:</h5>
3197
3198<p>Access to the memory pointed to by the pointer is no longer defined
Chris Lattner329d6532008-04-19 22:41:32 +0000[diff] [blame]3199after this instruction executes. If the pointer is null, the operation
3200is a noop.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3201
3202<h5>Example:</h5>
3203
3204<pre>
3205 %array = <a href="#i_malloc">malloc</a> [4 x i8] <i>; yields {[4 x i8]*}:array</i>
3206 free [4 x i8]* %array
3207</pre>
3208</div>
3209
3210<!-- _______________________________________________________________________ -->
3211<div class="doc_subsubsection">
3212 <a name="i_alloca">'<tt>alloca</tt>' Instruction</a>
3213</div>
3214
3215<div class="doc_text">
3216
3217<h5>Syntax:</h5>
3218
3219<pre>
3220 &lt;result&gt; = alloca &lt;type&gt;[, i32 &lt;NumElements&gt;][, align &lt;alignment&gt;] <i>; yields {type*}:result</i>
3221</pre>
3222
3223<h5>Overview:</h5>
3224
3225<p>The '<tt>alloca</tt>' instruction allocates memory on the stack frame of the
3226currently executing function, to be automatically released when this function
Christopher Lambcfe00962007-12-17 01:00:21 +0000[diff] [blame]3227returns to its caller. The object is always allocated in the generic address
3228space (address space zero).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3229
3230<h5>Arguments:</h5>
3231
3232<p>The '<tt>alloca</tt>' instruction allocates <tt>sizeof(&lt;type&gt;)*NumElements</tt>
3233bytes of memory on the runtime stack, returning a pointer of the
Gabor Greif5082cf42008-02-09 22:24:34 +0000[diff] [blame]3234appropriate type to the program. If "NumElements" is specified, it is the
3235number of elements allocated, otherwise "NumElements" is defaulted to be one.
Chris Lattner10368b62008-04-02 00:38:26 +0000[diff] [blame]3236If a constant alignment is specified, the value result of the allocation is guaranteed
Gabor Greif5082cf42008-02-09 22:24:34 +0000[diff] [blame]3237to be aligned to at least that boundary. If not specified, or if zero, the target
3238can choose to align the allocation on any convenient boundary.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3239
3240<p>'<tt>type</tt>' may be any sized type.</p>
3241
3242<h5>Semantics:</h5>
3243
Chris Lattner8b094fc2008-04-19 21:01:16 +0000[diff] [blame]3244<p>Memory is allocated; a pointer is returned. The operation is undefiend if
3245there is insufficient stack space for the allocation. '<tt>alloca</tt>'d
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3246memory is automatically released when the function returns. The '<tt>alloca</tt>'
3247instruction is commonly used to represent automatic variables that must
3248have an address available. When the function returns (either with the <tt><a
3249 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]3250instructions), the memory is reclaimed. Allocating zero bytes
3251is legal, but the result is undefined.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3252
3253<h5>Example:</h5>
3254
3255<pre>
3256 %ptr = alloca i32 <i>; yields {i32*}:ptr</i>
3257 %ptr = alloca i32, i32 4 <i>; yields {i32*}:ptr</i>
3258 %ptr = alloca i32, i32 4, align 1024 <i>; yields {i32*}:ptr</i>
3259 %ptr = alloca i32, align 1024 <i>; yields {i32*}:ptr</i>
3260</pre>
3261</div>
3262
3263<!-- _______________________________________________________________________ -->
3264<div class="doc_subsubsection"> <a name="i_load">'<tt>load</tt>'
3265Instruction</a> </div>
3266<div class="doc_text">
3267<h5>Syntax:</h5>
3268<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>
3269<h5>Overview:</h5>
3270<p>The '<tt>load</tt>' instruction is used to read from memory.</p>
3271<h5>Arguments:</h5>
3272<p>The argument to the '<tt>load</tt>' instruction specifies the memory
3273address from which to load. The pointer must point to a <a
3274 href="#t_firstclass">first class</a> type. If the <tt>load</tt> is
3275marked as <tt>volatile</tt>, then the optimizer is not allowed to modify
3276the number or order of execution of this <tt>load</tt> with other
3277volatile <tt>load</tt> and <tt><a href="#i_store">store</a></tt>
3278instructions. </p>
Chris Lattner21003c82008-01-06 21:04:43 +0000[diff] [blame]3279<p>
Chris Lattner10368b62008-04-02 00:38:26 +0000[diff] [blame]3280The optional constant "align" argument specifies the alignment of the operation
Chris Lattner21003c82008-01-06 21:04:43 +0000[diff] [blame]3281(that is, the alignment of the memory address). A value of 0 or an
3282omitted "align" argument means that the operation has the preferential
3283alignment for the target. It is the responsibility of the code emitter
3284to ensure that the alignment information is correct. Overestimating
3285the alignment results in an undefined behavior. Underestimating the
3286alignment may produce less efficient code. An alignment of 1 is always
3287safe.
3288</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3289<h5>Semantics:</h5>
3290<p>The location of memory pointed to is loaded.</p>
3291<h5>Examples:</h5>
3292<pre> %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
3293 <a
3294 href="#i_store">store</a> i32 3, i32* %ptr <i>; yields {void}</i>
3295 %val = load i32* %ptr <i>; yields {i32}:val = i32 3</i>
3296</pre>
3297</div>
3298<!-- _______________________________________________________________________ -->
3299<div class="doc_subsubsection"> <a name="i_store">'<tt>store</tt>'
3300Instruction</a> </div>
3301<div class="doc_text">
3302<h5>Syntax:</h5>
3303<pre> store &lt;ty&gt; &lt;value&gt;, &lt;ty&gt;* &lt;pointer&gt;[, align &lt;alignment&gt;] <i>; yields {void}</i>
3304 volatile store &lt;ty&gt; &lt;value&gt;, &lt;ty&gt;* &lt;pointer&gt;[, align &lt;alignment&gt;] <i>; yields {void}</i>
3305</pre>
3306<h5>Overview:</h5>
3307<p>The '<tt>store</tt>' instruction is used to write to memory.</p>
3308<h5>Arguments:</h5>
3309<p>There are two arguments to the '<tt>store</tt>' instruction: a value
3310to 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]3311operand must be a pointer to the <a href="#t_firstclass">first class</a> type
3312of the '<tt>&lt;value&gt;</tt>'
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3313operand. If the <tt>store</tt> is marked as <tt>volatile</tt>, then the
3314optimizer is not allowed to modify the number or order of execution of
3315this <tt>store</tt> with other volatile <tt>load</tt> and <tt><a
3316 href="#i_store">store</a></tt> instructions.</p>
Chris Lattner21003c82008-01-06 21:04:43 +0000[diff] [blame]3317<p>
Chris Lattner10368b62008-04-02 00:38:26 +0000[diff] [blame]3318The optional constant "align" argument specifies the alignment of the operation
Chris Lattner21003c82008-01-06 21:04:43 +0000[diff] [blame]3319(that is, the alignment of the memory address). A value of 0 or an
3320omitted "align" argument means that the operation has the preferential
3321alignment for the target. It is the responsibility of the code emitter
3322to ensure that the alignment information is correct. Overestimating
3323the alignment results in an undefined behavior. Underestimating the
3324alignment may produce less efficient code. An alignment of 1 is always
3325safe.
3326</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3327<h5>Semantics:</h5>
3328<p>The contents of memory are updated to contain '<tt>&lt;value&gt;</tt>'
3329at the location specified by the '<tt>&lt;pointer&gt;</tt>' operand.</p>
3330<h5>Example:</h5>
3331<pre> %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
Bill Wendling63ffa142007-10-22 05:10:05 +0000[diff] [blame]3332 store i32 3, i32* %ptr <i>; yields {void}</i>
3333 %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]3334</pre>
3335</div>
3336
3337<!-- _______________________________________________________________________ -->
3338<div class="doc_subsubsection">
3339 <a name="i_getelementptr">'<tt>getelementptr</tt>' Instruction</a>
3340</div>
3341
3342<div class="doc_text">
3343<h5>Syntax:</h5>
3344<pre>
Matthijs Kooijman32a080e2008-10-13 13:44:15 +0000[diff] [blame]3345 &lt;result&gt; = getelementptr &lt;pty&gt;* &lt;ptrval&gt;{, &lt;ty&gt; &lt;idx&gt;}*
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3346</pre>
3347
3348<h5>Overview:</h5>
3349
3350<p>
3351The '<tt>getelementptr</tt>' instruction is used to get the address of a
Matthijs Kooijman32a080e2008-10-13 13:44:15 +0000[diff] [blame]3352subelement of an aggregate data structure. It performs address calculation only
3353and does not access memory.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3354
3355<h5>Arguments:</h5>
3356
Matthijs Kooijman32a080e2008-10-13 13:44:15 +0000[diff] [blame]3357<p>The first argument is always a pointer, and forms the basis of the
3358calculation. The remaining arguments are indices, that indicate which of the
3359elements of the aggregate object are indexed. The interpretation of each index
3360is dependent on the type being indexed into. The first index always indexes the
3361pointer value given as the first argument, the second index indexes a value of
3362the type pointed to (not necessarily the value directly pointed to, since the
3363first index can be non-zero), etc. The first type indexed into must be a pointer
3364value, subsequent types can be arrays, vectors and structs. Note that subsequent
3365types being indexed into can never be pointers, since that would require loading
3366the pointer before continuing calculation.</p>
3367
3368<p>The type of each index argument depends on the type it is indexing into.
3369When indexing into a (packed) structure, only <tt>i32</tt> integer
3370<b>constants</b> are allowed. When indexing into an array, pointer or vector,
3371only integers of 32 or 64 bits are allowed (also non-constants). 32-bit values
3372will be sign extended to 64-bits if required.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3373
3374<p>For example, let's consider a C code fragment and how it gets
3375compiled to LLVM:</p>
3376
3377<div class="doc_code">
3378<pre>
3379struct RT {
3380 char A;
3381 int B[10][20];
3382 char C;
3383};
3384struct ST {
3385 int X;
3386 double Y;
3387 struct RT Z;
3388};
3389
3390int *foo(struct ST *s) {
3391 return &amp;s[1].Z.B[5][13];
3392}
3393</pre>
3394</div>
3395
3396<p>The LLVM code generated by the GCC frontend is:</p>
3397
3398<div class="doc_code">
3399<pre>
3400%RT = type { i8 , [10 x [20 x i32]], i8 }
3401%ST = type { i32, double, %RT }
3402
3403define i32* %foo(%ST* %s) {
3404entry:
3405 %reg = getelementptr %ST* %s, i32 1, i32 2, i32 1, i32 5, i32 13
3406 ret i32* %reg
3407}
3408</pre>
3409</div>
3410
3411<h5>Semantics:</h5>
3412
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3413<p>In the example above, the first index is indexing into the '<tt>%ST*</tt>'
3414type, which is a pointer, yielding a '<tt>%ST</tt>' = '<tt>{ i32, double, %RT
3415}</tt>' type, a structure. The second index indexes into the third element of
3416the structure, yielding a '<tt>%RT</tt>' = '<tt>{ i8 , [10 x [20 x i32]],
3417i8 }</tt>' type, another structure. The third index indexes into the second
3418element of the structure, yielding a '<tt>[10 x [20 x i32]]</tt>' type, an
3419array. The two dimensions of the array are subscripted into, yielding an
3420'<tt>i32</tt>' type. The '<tt>getelementptr</tt>' instruction returns a pointer
3421to this element, thus computing a value of '<tt>i32*</tt>' type.</p>
3422
3423<p>Note that it is perfectly legal to index partially through a
3424structure, returning a pointer to an inner element. Because of this,
3425the LLVM code for the given testcase is equivalent to:</p>
3426
3427<pre>
3428 define i32* %foo(%ST* %s) {
3429 %t1 = getelementptr %ST* %s, i32 1 <i>; yields %ST*:%t1</i>
3430 %t2 = getelementptr %ST* %t1, i32 0, i32 2 <i>; yields %RT*:%t2</i>
3431 %t3 = getelementptr %RT* %t2, i32 0, i32 1 <i>; yields [10 x [20 x i32]]*:%t3</i>
3432 %t4 = getelementptr [10 x [20 x i32]]* %t3, i32 0, i32 5 <i>; yields [20 x i32]*:%t4</i>
3433 %t5 = getelementptr [20 x i32]* %t4, i32 0, i32 13 <i>; yields i32*:%t5</i>
3434 ret i32* %t5
3435 }
3436</pre>
3437
3438<p>Note that it is undefined to access an array out of bounds: array and
3439pointer indexes must always be within the defined bounds of the array type.
Chris Lattnera7d94ba2008-04-24 05:59:56 +0000[diff] [blame]3440The one exception for this rule is zero length arrays. These arrays are
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3441defined to be accessible as variable length arrays, which requires access
3442beyond the zero'th element.</p>
3443
3444<p>The getelementptr instruction is often confusing. For some more insight
3445into how it works, see <a href="GetElementPtr.html">the getelementptr
3446FAQ</a>.</p>
3447
3448<h5>Example:</h5>
3449
3450<pre>
3451 <i>; yields [12 x i8]*:aptr</i>
Matthijs Kooijman32a080e2008-10-13 13:44:15 +0000[diff] [blame]3452 %aptr = getelementptr {i32, [12 x i8]}* %saptr, i64 0, i32 1
3453 <i>; yields i8*:vptr</i>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]3454 %vptr = getelementptr {i32, &lt;2 x i8&gt;}* %svptr, i64 0, i32 1, i32 1
Matthijs Kooijman32a080e2008-10-13 13:44:15 +0000[diff] [blame]3455 <i>; yields i8*:eptr</i>
3456 %eptr = getelementptr [12 x i8]* %aptr, i64 0, i32 1
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3457</pre>
3458</div>
3459
3460<!-- ======================================================================= -->
3461<div class="doc_subsection"> <a name="convertops">Conversion Operations</a>
3462</div>
3463<div class="doc_text">
3464<p>The instructions in this category are the conversion instructions (casting)
3465which all take a single operand and a type. They perform various bit conversions
3466on the operand.</p>
3467</div>
3468
3469<!-- _______________________________________________________________________ -->
3470<div class="doc_subsubsection">
3471 <a name="i_trunc">'<tt>trunc .. to</tt>' Instruction</a>
3472</div>
3473<div class="doc_text">
3474
3475<h5>Syntax:</h5>
3476<pre>
3477 &lt;result&gt; = trunc &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3478</pre>
3479
3480<h5>Overview:</h5>
3481<p>
3482The '<tt>trunc</tt>' instruction truncates its operand to the type <tt>ty2</tt>.
3483</p>
3484
3485<h5>Arguments:</h5>
3486<p>
3487The '<tt>trunc</tt>' instruction takes a <tt>value</tt> to trunc, which must
3488be an <a href="#t_integer">integer</a> type, and a type that specifies the size
3489and type of the result, which must be an <a href="#t_integer">integer</a>
3490type. The bit size of <tt>value</tt> must be larger than the bit size of
3491<tt>ty2</tt>. Equal sized types are not allowed.</p>
3492
3493<h5>Semantics:</h5>
3494<p>
3495The '<tt>trunc</tt>' instruction truncates the high order bits in <tt>value</tt>
3496and converts the remaining bits to <tt>ty2</tt>. Since the source size must be
3497larger than the destination size, <tt>trunc</tt> cannot be a <i>no-op cast</i>.
3498It will always truncate bits.</p>
3499
3500<h5>Example:</h5>
3501<pre>
3502 %X = trunc i32 257 to i8 <i>; yields i8:1</i>
3503 %Y = trunc i32 123 to i1 <i>; yields i1:true</i>
3504 %Y = trunc i32 122 to i1 <i>; yields i1:false</i>
3505</pre>
3506</div>
3507
3508<!-- _______________________________________________________________________ -->
3509<div class="doc_subsubsection">
3510 <a name="i_zext">'<tt>zext .. to</tt>' Instruction</a>
3511</div>
3512<div class="doc_text">
3513
3514<h5>Syntax:</h5>
3515<pre>
3516 &lt;result&gt; = zext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3517</pre>
3518
3519<h5>Overview:</h5>
3520<p>The '<tt>zext</tt>' instruction zero extends its operand to type
3521<tt>ty2</tt>.</p>
3522
3523
3524<h5>Arguments:</h5>
3525<p>The '<tt>zext</tt>' instruction takes a value to cast, which must be of
3526<a href="#t_integer">integer</a> type, and a type to cast it to, which must
3527also be of <a href="#t_integer">integer</a> type. The bit size of the
3528<tt>value</tt> must be smaller than the bit size of the destination type,
3529<tt>ty2</tt>.</p>
3530
3531<h5>Semantics:</h5>
3532<p>The <tt>zext</tt> fills the high order bits of the <tt>value</tt> with zero
3533bits until it reaches the size of the destination type, <tt>ty2</tt>.</p>
3534
3535<p>When zero extending from i1, the result will always be either 0 or 1.</p>
3536
3537<h5>Example:</h5>
3538<pre>
3539 %X = zext i32 257 to i64 <i>; yields i64:257</i>
3540 %Y = zext i1 true to i32 <i>; yields i32:1</i>
3541</pre>
3542</div>
3543
3544<!-- _______________________________________________________________________ -->
3545<div class="doc_subsubsection">
3546 <a name="i_sext">'<tt>sext .. to</tt>' Instruction</a>
3547</div>
3548<div class="doc_text">
3549
3550<h5>Syntax:</h5>
3551<pre>
3552 &lt;result&gt; = sext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3553</pre>
3554
3555<h5>Overview:</h5>
3556<p>The '<tt>sext</tt>' sign extends <tt>value</tt> to the type <tt>ty2</tt>.</p>
3557
3558<h5>Arguments:</h5>
3559<p>
3560The '<tt>sext</tt>' instruction takes a value to cast, which must be of
3561<a href="#t_integer">integer</a> type, and a type to cast it to, which must
3562also be of <a href="#t_integer">integer</a> type. The bit size of the
3563<tt>value</tt> must be smaller than the bit size of the destination type,
3564<tt>ty2</tt>.</p>
3565
3566<h5>Semantics:</h5>
3567<p>
3568The '<tt>sext</tt>' instruction performs a sign extension by copying the sign
3569bit (highest order bit) of the <tt>value</tt> until it reaches the bit size of
3570the type <tt>ty2</tt>.</p>
3571
3572<p>When sign extending from i1, the extension always results in -1 or 0.</p>
3573
3574<h5>Example:</h5>
3575<pre>
3576 %X = sext i8 -1 to i16 <i>; yields i16 :65535</i>
3577 %Y = sext i1 true to i32 <i>; yields i32:-1</i>
3578</pre>
3579</div>
3580
3581<!-- _______________________________________________________________________ -->
3582<div class="doc_subsubsection">
3583 <a name="i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a>
3584</div>
3585
3586<div class="doc_text">
3587
3588<h5>Syntax:</h5>
3589
3590<pre>
3591 &lt;result&gt; = fptrunc &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3592</pre>
3593
3594<h5>Overview:</h5>
3595<p>The '<tt>fptrunc</tt>' instruction truncates <tt>value</tt> to type
3596<tt>ty2</tt>.</p>
3597
3598
3599<h5>Arguments:</h5>
3600<p>The '<tt>fptrunc</tt>' instruction takes a <a href="#t_floating">floating
3601 point</a> value to cast and a <a href="#t_floating">floating point</a> type to
3602cast it to. The size of <tt>value</tt> must be larger than the size of
3603<tt>ty2</tt>. This implies that <tt>fptrunc</tt> cannot be used to make a
3604<i>no-op cast</i>.</p>
3605
3606<h5>Semantics:</h5>
3607<p> The '<tt>fptrunc</tt>' instruction truncates a <tt>value</tt> from a larger
3608<a href="#t_floating">floating point</a> type to a smaller
3609<a href="#t_floating">floating point</a> type. If the value cannot fit within
3610the destination type, <tt>ty2</tt>, then the results are undefined.</p>
3611
3612<h5>Example:</h5>
3613<pre>
3614 %X = fptrunc double 123.0 to float <i>; yields float:123.0</i>
3615 %Y = fptrunc double 1.0E+300 to float <i>; yields undefined</i>
3616</pre>
3617</div>
3618
3619<!-- _______________________________________________________________________ -->
3620<div class="doc_subsubsection">
3621 <a name="i_fpext">'<tt>fpext .. to</tt>' Instruction</a>
3622</div>
3623<div class="doc_text">
3624
3625<h5>Syntax:</h5>
3626<pre>
3627 &lt;result&gt; = fpext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3628</pre>
3629
3630<h5>Overview:</h5>
3631<p>The '<tt>fpext</tt>' extends a floating point <tt>value</tt> to a larger
3632floating point value.</p>
3633
3634<h5>Arguments:</h5>
3635<p>The '<tt>fpext</tt>' instruction takes a
3636<a href="#t_floating">floating point</a> <tt>value</tt> to cast,
3637and a <a href="#t_floating">floating point</a> type to cast it to. The source
3638type must be smaller than the destination type.</p>
3639
3640<h5>Semantics:</h5>
3641<p>The '<tt>fpext</tt>' instruction extends the <tt>value</tt> from a smaller
3642<a href="#t_floating">floating point</a> type to a larger
3643<a href="#t_floating">floating point</a> type. The <tt>fpext</tt> cannot be
3644used to make a <i>no-op cast</i> because it always changes bits. Use
3645<tt>bitcast</tt> to make a <i>no-op cast</i> for a floating point cast.</p>
3646
3647<h5>Example:</h5>
3648<pre>
3649 %X = fpext float 3.1415 to double <i>; yields double:3.1415</i>
3650 %Y = fpext float 1.0 to float <i>; yields float:1.0 (no-op)</i>
3651</pre>
3652</div>
3653
3654<!-- _______________________________________________________________________ -->
3655<div class="doc_subsubsection">
3656 <a name="i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a>
3657</div>
3658<div class="doc_text">
3659
3660<h5>Syntax:</h5>
3661<pre>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3662 &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]3663</pre>
3664
3665<h5>Overview:</h5>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3666<p>The '<tt>fptoui</tt>' converts a floating point <tt>value</tt> to its
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3667unsigned integer equivalent of type <tt>ty2</tt>.
3668</p>
3669
3670<h5>Arguments:</h5>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3671<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]3672scalar or vector <a href="#t_floating">floating point</a> value, and a type
3673to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
3674type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
3675vector integer type with the same number of elements as <tt>ty</tt></p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3676
3677<h5>Semantics:</h5>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3678<p> The '<tt>fptoui</tt>' instruction converts its
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3679<a href="#t_floating">floating point</a> operand into the nearest (rounding
3680towards zero) unsigned integer value. If the value cannot fit in <tt>ty2</tt>,
3681the results are undefined.</p>
3682
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3683<h5>Example:</h5>
3684<pre>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3685 %X = fptoui double 123.0 to i32 <i>; yields i32:123</i>
Chris Lattner681f1e82007-09-22 03:17:52 +0000[diff] [blame]3686 %Y = fptoui float 1.0E+300 to i1 <i>; yields undefined:1</i>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3687 %X = fptoui float 1.04E+17 to i8 <i>; yields undefined:1</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3688</pre>
3689</div>
3690
3691<!-- _______________________________________________________________________ -->
3692<div class="doc_subsubsection">
3693 <a name="i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a>
3694</div>
3695<div class="doc_text">
3696
3697<h5>Syntax:</h5>
3698<pre>
3699 &lt;result&gt; = fptosi &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3700</pre>
3701
3702<h5>Overview:</h5>
3703<p>The '<tt>fptosi</tt>' instruction converts
3704<a href="#t_floating">floating point</a> <tt>value</tt> to type <tt>ty2</tt>.
3705</p>
3706
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3707<h5>Arguments:</h5>
3708<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]3709scalar or vector <a href="#t_floating">floating point</a> value, and a type
3710to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
3711type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
3712vector integer type with the same number of elements as <tt>ty</tt></p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3713
3714<h5>Semantics:</h5>
3715<p>The '<tt>fptosi</tt>' instruction converts its
3716<a href="#t_floating">floating point</a> operand into the nearest (rounding
3717towards zero) signed integer value. If the value cannot fit in <tt>ty2</tt>,
3718the results are undefined.</p>
3719
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3720<h5>Example:</h5>
3721<pre>
3722 %X = fptosi double -123.0 to i32 <i>; yields i32:-123</i>
Chris Lattner681f1e82007-09-22 03:17:52 +0000[diff] [blame]3723 %Y = fptosi float 1.0E-247 to i1 <i>; yields undefined:1</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3724 %X = fptosi float 1.04E+17 to i8 <i>; yields undefined:1</i>
3725</pre>
3726</div>
3727
3728<!-- _______________________________________________________________________ -->
3729<div class="doc_subsubsection">
3730 <a name="i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a>
3731</div>
3732<div class="doc_text">
3733
3734<h5>Syntax:</h5>
3735<pre>
3736 &lt;result&gt; = uitofp &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3737</pre>
3738
3739<h5>Overview:</h5>
3740<p>The '<tt>uitofp</tt>' instruction regards <tt>value</tt> as an unsigned
3741integer and converts that value to the <tt>ty2</tt> type.</p>
3742
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3743<h5>Arguments:</h5>
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]3744<p>The '<tt>uitofp</tt>' instruction takes a value to cast, which must be a
3745scalar or vector <a href="#t_integer">integer</a> value, and a type to cast it
3746to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
3747type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
3748floating point type with the same number of elements as <tt>ty</tt></p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3749
3750<h5>Semantics:</h5>
3751<p>The '<tt>uitofp</tt>' instruction interprets its operand as an unsigned
3752integer quantity and converts it to the corresponding floating point value. If
3753the value cannot fit in the floating point value, the results are undefined.</p>
3754
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3755<h5>Example:</h5>
3756<pre>
3757 %X = uitofp i32 257 to float <i>; yields float:257.0</i>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]3758 %Y = uitofp i8 -1 to double <i>; yields double:255.0</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3759</pre>
3760</div>
3761
3762<!-- _______________________________________________________________________ -->
3763<div class="doc_subsubsection">
3764 <a name="i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a>
3765</div>
3766<div class="doc_text">
3767
3768<h5>Syntax:</h5>
3769<pre>
3770 &lt;result&gt; = sitofp &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3771</pre>
3772
3773<h5>Overview:</h5>
3774<p>The '<tt>sitofp</tt>' instruction regards <tt>value</tt> as a signed
3775integer and converts that value to the <tt>ty2</tt> type.</p>
3776
3777<h5>Arguments:</h5>
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]3778<p>The '<tt>sitofp</tt>' instruction takes a value to cast, which must be a
3779scalar or vector <a href="#t_integer">integer</a> value, and a type to cast it
3780to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
3781type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
3782floating point type with the same number of elements as <tt>ty</tt></p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3783
3784<h5>Semantics:</h5>
3785<p>The '<tt>sitofp</tt>' instruction interprets its operand as a signed
3786integer quantity and converts it to the corresponding floating point value. If
3787the value cannot fit in the floating point value, the results are undefined.</p>
3788
3789<h5>Example:</h5>
3790<pre>
3791 %X = sitofp i32 257 to float <i>; yields float:257.0</i>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]3792 %Y = sitofp i8 -1 to double <i>; yields double:-1.0</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3793</pre>
3794</div>
3795
3796<!-- _______________________________________________________________________ -->
3797<div class="doc_subsubsection">
3798 <a name="i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a>
3799</div>
3800<div class="doc_text">
3801
3802<h5>Syntax:</h5>
3803<pre>
3804 &lt;result&gt; = ptrtoint &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3805</pre>
3806
3807<h5>Overview:</h5>
3808<p>The '<tt>ptrtoint</tt>' instruction converts the pointer <tt>value</tt> to
3809the integer type <tt>ty2</tt>.</p>
3810
3811<h5>Arguments:</h5>
3812<p>The '<tt>ptrtoint</tt>' instruction takes a <tt>value</tt> to cast, which
3813must be a <a href="#t_pointer">pointer</a> value, and a type to cast it to
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]3814<tt>ty2</tt>, which must be an <a href="#t_integer">integer</a> type.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3815
3816<h5>Semantics:</h5>
3817<p>The '<tt>ptrtoint</tt>' instruction converts <tt>value</tt> to integer type
3818<tt>ty2</tt> by interpreting the pointer value as an integer and either
3819truncating or zero extending that value to the size of the integer type. If
3820<tt>value</tt> is smaller than <tt>ty2</tt> then a zero extension is done. If
3821<tt>value</tt> is larger than <tt>ty2</tt> then a truncation is done. If they
3822are the same size, then nothing is done (<i>no-op cast</i>) other than a type
3823change.</p>
3824
3825<h5>Example:</h5>
3826<pre>
3827 %X = ptrtoint i32* %X to i8 <i>; yields truncation on 32-bit architecture</i>
3828 %Y = ptrtoint i32* %x to i64 <i>; yields zero extension on 32-bit architecture</i>
3829</pre>
3830</div>
3831
3832<!-- _______________________________________________________________________ -->
3833<div class="doc_subsubsection">
3834 <a name="i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a>
3835</div>
3836<div class="doc_text">
3837
3838<h5>Syntax:</h5>
3839<pre>
3840 &lt;result&gt; = inttoptr &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3841</pre>
3842
3843<h5>Overview:</h5>
3844<p>The '<tt>inttoptr</tt>' instruction converts an integer <tt>value</tt> to
3845a pointer type, <tt>ty2</tt>.</p>
3846
3847<h5>Arguments:</h5>
3848<p>The '<tt>inttoptr</tt>' instruction takes an <a href="#t_integer">integer</a>
3849value to cast, and a type to cast it to, which must be a
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]3850<a href="#t_pointer">pointer</a> type.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3851
3852<h5>Semantics:</h5>
3853<p>The '<tt>inttoptr</tt>' instruction converts <tt>value</tt> to type
3854<tt>ty2</tt> by applying either a zero extension or a truncation depending on
3855the size of the integer <tt>value</tt>. If <tt>value</tt> is larger than the
3856size of a pointer then a truncation is done. If <tt>value</tt> is smaller than
3857the size of a pointer then a zero extension is done. If they are the same size,
3858nothing is done (<i>no-op cast</i>).</p>
3859
3860<h5>Example:</h5>
3861<pre>
3862 %X = inttoptr i32 255 to i32* <i>; yields zero extension on 64-bit architecture</i>
3863 %X = inttoptr i32 255 to i32* <i>; yields no-op on 32-bit architecture</i>
3864 %Y = inttoptr i64 0 to i32* <i>; yields truncation on 32-bit architecture</i>
3865</pre>
3866</div>
3867
3868<!-- _______________________________________________________________________ -->
3869<div class="doc_subsubsection">
3870 <a name="i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a>
3871</div>
3872<div class="doc_text">
3873
3874<h5>Syntax:</h5>
3875<pre>
3876 &lt;result&gt; = bitcast &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3877</pre>
3878
3879<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3880
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3881<p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
3882<tt>ty2</tt> without changing any bits.</p>
3883
3884<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3885
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3886<p>The '<tt>bitcast</tt>' instruction takes a value to cast, which must be
Dan Gohman7305fa02008-09-08 16:45:59 +0000[diff] [blame]3887a non-aggregate first class value, and a type to cast it to, which must also be
3888a non-aggregate <a href="#t_firstclass">first class</a> type. The bit sizes of
3889<tt>value</tt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3890and the destination type, <tt>ty2</tt>, must be identical. If the source
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3891type is a pointer, the destination type must also be a pointer. This
3892instruction supports bitwise conversion of vectors to integers and to vectors
3893of other types (as long as they have the same size).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3894
3895<h5>Semantics:</h5>
3896<p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
3897<tt>ty2</tt>. It is always a <i>no-op cast</i> because no bits change with
3898this conversion. The conversion is done as if the <tt>value</tt> had been
3899stored to memory and read back as type <tt>ty2</tt>. Pointer types may only be
3900converted to other pointer types with this instruction. To convert pointers to
3901other types, use the <a href="#i_inttoptr">inttoptr</a> or
3902<a href="#i_ptrtoint">ptrtoint</a> instructions first.</p>
3903
3904<h5>Example:</h5>
3905<pre>
3906 %X = bitcast i8 255 to i8 <i>; yields i8 :-1</i>
3907 %Y = bitcast i32* %x to sint* <i>; yields sint*:%x</i>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]3908 %Z = bitcast &lt;2 x int&gt; %V to i64; <i>; yields i64: %V</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3909</pre>
3910</div>
3911
3912<!-- ======================================================================= -->
3913<div class="doc_subsection"> <a name="otherops">Other Operations</a> </div>
3914<div class="doc_text">
3915<p>The instructions in this category are the "miscellaneous"
3916instructions, which defy better classification.</p>
3917</div>
3918
3919<!-- _______________________________________________________________________ -->
3920<div class="doc_subsubsection"><a name="i_icmp">'<tt>icmp</tt>' Instruction</a>
3921</div>
3922<div class="doc_text">
3923<h5>Syntax:</h5>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]3924<pre> &lt;result&gt; = icmp &lt;cond&gt; &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {i1} or {&lt;N x i1&gt;}:result</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3925</pre>
3926<h5>Overview:</h5>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]3927<p>The '<tt>icmp</tt>' instruction returns a boolean value or
3928a vector of boolean values based on comparison
3929of its two integer, integer vector, or pointer operands.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3930<h5>Arguments:</h5>
3931<p>The '<tt>icmp</tt>' instruction takes three operands. The first operand is
3932the condition code indicating the kind of comparison to perform. It is not
3933a value, just a keyword. The possible condition code are:
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]3934</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3935<ol>
3936 <li><tt>eq</tt>: equal</li>
3937 <li><tt>ne</tt>: not equal </li>
3938 <li><tt>ugt</tt>: unsigned greater than</li>
3939 <li><tt>uge</tt>: unsigned greater or equal</li>
3940 <li><tt>ult</tt>: unsigned less than</li>
3941 <li><tt>ule</tt>: unsigned less or equal</li>
3942 <li><tt>sgt</tt>: signed greater than</li>
3943 <li><tt>sge</tt>: signed greater or equal</li>
3944 <li><tt>slt</tt>: signed less than</li>
3945 <li><tt>sle</tt>: signed less or equal</li>
3946</ol>
3947<p>The remaining two arguments must be <a href="#t_integer">integer</a> or
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]3948<a href="#t_pointer">pointer</a>
3949or integer <a href="#t_vector">vector</a> typed.
3950They must also be identical types.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3951<h5>Semantics:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3952<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]3953the condition code given as <tt>cond</tt>. The comparison performed always
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]3954yields either an <a href="#t_primitive"><tt>i1</tt></a> or vector of <tt>i1</tt> result, as follows:
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]3955</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3956<ol>
3957 <li><tt>eq</tt>: yields <tt>true</tt> if the operands are equal,
3958 <tt>false</tt> otherwise. No sign interpretation is necessary or performed.
3959 </li>
3960 <li><tt>ne</tt>: yields <tt>true</tt> if the operands are unequal,
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]3961 <tt>false</tt> otherwise. No sign interpretation is necessary or performed.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3962 <li><tt>ugt</tt>: interprets the operands as unsigned values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3963 <tt>true</tt> if <tt>op1</tt> is greater than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3964 <li><tt>uge</tt>: interprets the operands as unsigned values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3965 <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]3966 <li><tt>ult</tt>: interprets the operands as unsigned values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3967 <tt>true</tt> if <tt>op1</tt> is less than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3968 <li><tt>ule</tt>: interprets the operands as unsigned values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3969 <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]3970 <li><tt>sgt</tt>: interprets the operands as signed values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3971 <tt>true</tt> if <tt>op1</tt> is greater than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3972 <li><tt>sge</tt>: interprets the operands as signed values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3973 <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]3974 <li><tt>slt</tt>: interprets the operands as signed values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3975 <tt>true</tt> if <tt>op1</tt> is less than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3976 <li><tt>sle</tt>: interprets the operands as signed values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3977 <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]3978</ol>
3979<p>If the operands are <a href="#t_pointer">pointer</a> typed, the pointer
3980values are compared as if they were integers.</p>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]3981<p>If the operands are integer vectors, then they are compared
3982element by element. The result is an <tt>i1</tt> vector with
3983the same number of elements as the values being compared.
3984Otherwise, the result is an <tt>i1</tt>.
3985</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3986
3987<h5>Example:</h5>
3988<pre> &lt;result&gt; = icmp eq i32 4, 5 <i>; yields: result=false</i>
3989 &lt;result&gt; = icmp ne float* %X, %X <i>; yields: result=false</i>
3990 &lt;result&gt; = icmp ult i16 4, 5 <i>; yields: result=true</i>
3991 &lt;result&gt; = icmp sgt i16 4, 5 <i>; yields: result=false</i>
3992 &lt;result&gt; = icmp ule i16 -4, 5 <i>; yields: result=false</i>
3993 &lt;result&gt; = icmp sge i16 4, 5 <i>; yields: result=false</i>
3994</pre>
3995</div>
3996
3997<!-- _______________________________________________________________________ -->
3998<div class="doc_subsubsection"><a name="i_fcmp">'<tt>fcmp</tt>' Instruction</a>
3999</div>
4000<div class="doc_text">
4001<h5>Syntax:</h5>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4002<pre> &lt;result&gt; = fcmp &lt;cond&gt; &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {i1} or {&lt;N x i1&gt;}:result</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4003</pre>
4004<h5>Overview:</h5>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4005<p>The '<tt>fcmp</tt>' instruction returns a boolean value
4006or vector of boolean values based on comparison
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4007of its operands.</p>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4008<p>
4009If the operands are floating point scalars, then the result
4010type is a boolean (<a href="#t_primitive"><tt>i1</tt></a>).
4011</p>
4012<p>If the operands are floating point vectors, then the result type
4013is a vector of boolean with the same number of elements as the
4014operands being compared.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4015<h5>Arguments:</h5>
4016<p>The '<tt>fcmp</tt>' instruction takes three operands. The first operand is
4017the condition code indicating the kind of comparison to perform. It is not
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4018a value, just a keyword. The possible condition code are:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4019<ol>
4020 <li><tt>false</tt>: no comparison, always returns false</li>
4021 <li><tt>oeq</tt>: ordered and equal</li>
4022 <li><tt>ogt</tt>: ordered and greater than </li>
4023 <li><tt>oge</tt>: ordered and greater than or equal</li>
4024 <li><tt>olt</tt>: ordered and less than </li>
4025 <li><tt>ole</tt>: ordered and less than or equal</li>
4026 <li><tt>one</tt>: ordered and not equal</li>
4027 <li><tt>ord</tt>: ordered (no nans)</li>
4028 <li><tt>ueq</tt>: unordered or equal</li>
4029 <li><tt>ugt</tt>: unordered or greater than </li>
4030 <li><tt>uge</tt>: unordered or greater than or equal</li>
4031 <li><tt>ult</tt>: unordered or less than </li>
4032 <li><tt>ule</tt>: unordered or less than or equal</li>
4033 <li><tt>une</tt>: unordered or not equal</li>
4034 <li><tt>uno</tt>: unordered (either nans)</li>
4035 <li><tt>true</tt>: no comparison, always returns true</li>
4036</ol>
4037<p><i>Ordered</i> means that neither operand is a QNAN while
4038<i>unordered</i> means that either operand may be a QNAN.</p>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4039<p>Each of <tt>val1</tt> and <tt>val2</tt> arguments must be
4040either a <a href="#t_floating">floating point</a> type
4041or a <a href="#t_vector">vector</a> of floating point type.
4042They must have identical types.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4043<h5>Semantics:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4044<p>The '<tt>fcmp</tt>' instruction compares <tt>op1</tt> and <tt>op2</tt>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4045according to the condition code given as <tt>cond</tt>.
4046If the operands are vectors, then the vectors are compared
4047element by element.
4048Each comparison performed
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4049always yields an <a href="#t_primitive">i1</a> result, as follows:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4050<ol>
4051 <li><tt>false</tt>: always yields <tt>false</tt>, regardless of operands.</li>
4052 <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]4053 <tt>op1</tt> is equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4054 <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]4055 <tt>op1</tt> is greather than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4056 <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]4057 <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4058 <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]4059 <tt>op1</tt> is less than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4060 <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]4061 <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4062 <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]4063 <tt>op1</tt> is not equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4064 <li><tt>ord</tt>: yields <tt>true</tt> if both operands are not a QNAN.</li>
4065 <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]4066 <tt>op1</tt> is equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4067 <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]4068 <tt>op1</tt> is greater than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4069 <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]4070 <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4071 <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]4072 <tt>op1</tt> is less than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4073 <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]4074 <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4075 <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]4076 <tt>op1</tt> is not equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4077 <li><tt>uno</tt>: yields <tt>true</tt> if either operand is a QNAN.</li>
4078 <li><tt>true</tt>: always yields <tt>true</tt>, regardless of operands.</li>
4079</ol>
4080
4081<h5>Example:</h5>
4082<pre> &lt;result&gt; = fcmp oeq float 4.0, 5.0 <i>; yields: result=false</i>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4083 &lt;result&gt; = fcmp one float 4.0, 5.0 <i>; yields: result=true</i>
4084 &lt;result&gt; = fcmp olt float 4.0, 5.0 <i>; yields: result=true</i>
4085 &lt;result&gt; = fcmp ueq double 1.0, 2.0 <i>; yields: result=false</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4086</pre>
4087</div>
4088
4089<!-- _______________________________________________________________________ -->
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]4090<div class="doc_subsubsection">
4091 <a name="i_vicmp">'<tt>vicmp</tt>' Instruction</a>
4092</div>
4093<div class="doc_text">
4094<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4095<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]4096</pre>
4097<h5>Overview:</h5>
4098<p>The '<tt>vicmp</tt>' instruction returns an integer vector value based on
4099element-wise comparison of its two integer vector operands.</p>
4100<h5>Arguments:</h5>
4101<p>The '<tt>vicmp</tt>' instruction takes three operands. The first operand is
4102the condition code indicating the kind of comparison to perform. It is not
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4103a value, just a keyword. The possible condition code are:</p>
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]4104<ol>
4105 <li><tt>eq</tt>: equal</li>
4106 <li><tt>ne</tt>: not equal </li>
4107 <li><tt>ugt</tt>: unsigned greater than</li>
4108 <li><tt>uge</tt>: unsigned greater or equal</li>
4109 <li><tt>ult</tt>: unsigned less than</li>
4110 <li><tt>ule</tt>: unsigned less or equal</li>
4111 <li><tt>sgt</tt>: signed greater than</li>
4112 <li><tt>sge</tt>: signed greater or equal</li>
4113 <li><tt>slt</tt>: signed less than</li>
4114 <li><tt>sle</tt>: signed less or equal</li>
4115</ol>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4116<p>The remaining two arguments must be <a href="#t_vector">vector</a> or
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]4117<a href="#t_integer">integer</a> typed. They must also be identical types.</p>
4118<h5>Semantics:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4119<p>The '<tt>vicmp</tt>' instruction compares <tt>op1</tt> and <tt>op2</tt>
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]4120according to the condition code given as <tt>cond</tt>. The comparison yields a
4121<a href="#t_vector">vector</a> of <a href="#t_integer">integer</a> result, of
4122identical type as the values being compared. The most significant bit in each
4123element is 1 if the element-wise comparison evaluates to true, and is 0
4124otherwise. All other bits of the result are undefined. The condition codes
4125are evaluated identically to the <a href="#i_icmp">'<tt>icmp</tt>'
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4126instruction</a>.</p>
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]4127
4128<h5>Example:</h5>
4129<pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4130 &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>
4131 &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]4132</pre>
4133</div>
4134
4135<!-- _______________________________________________________________________ -->
4136<div class="doc_subsubsection">
4137 <a name="i_vfcmp">'<tt>vfcmp</tt>' Instruction</a>
4138</div>
4139<div class="doc_text">
4140<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4141<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]4142<h5>Overview:</h5>
4143<p>The '<tt>vfcmp</tt>' instruction returns an integer vector value based on
4144element-wise comparison of its two floating point vector operands. The output
4145elements have the same width as the input elements.</p>
4146<h5>Arguments:</h5>
4147<p>The '<tt>vfcmp</tt>' instruction takes three operands. The first operand is
4148the condition code indicating the kind of comparison to perform. It is not
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4149a value, just a keyword. The possible condition code are:</p>
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]4150<ol>
4151 <li><tt>false</tt>: no comparison, always returns false</li>
4152 <li><tt>oeq</tt>: ordered and equal</li>
4153 <li><tt>ogt</tt>: ordered and greater than </li>
4154 <li><tt>oge</tt>: ordered and greater than or equal</li>
4155 <li><tt>olt</tt>: ordered and less than </li>
4156 <li><tt>ole</tt>: ordered and less than or equal</li>
4157 <li><tt>one</tt>: ordered and not equal</li>
4158 <li><tt>ord</tt>: ordered (no nans)</li>
4159 <li><tt>ueq</tt>: unordered or equal</li>
4160 <li><tt>ugt</tt>: unordered or greater than </li>
4161 <li><tt>uge</tt>: unordered or greater than or equal</li>
4162 <li><tt>ult</tt>: unordered or less than </li>
4163 <li><tt>ule</tt>: unordered or less than or equal</li>
4164 <li><tt>une</tt>: unordered or not equal</li>
4165 <li><tt>uno</tt>: unordered (either nans)</li>
4166 <li><tt>true</tt>: no comparison, always returns true</li>
4167</ol>
4168<p>The remaining two arguments must be <a href="#t_vector">vector</a> of
4169<a href="#t_floating">floating point</a> typed. They must also be identical
4170types.</p>
4171<h5>Semantics:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4172<p>The '<tt>vfcmp</tt>' instruction compares <tt>op1</tt> and <tt>op2</tt>
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]4173according to the condition code given as <tt>cond</tt>. The comparison yields a
4174<a href="#t_vector">vector</a> of <a href="#t_integer">integer</a> result, with
4175an identical number of elements as the values being compared, and each element
4176having identical with to the width of the floating point elements. The most
4177significant bit in each element is 1 if the element-wise comparison evaluates to
4178true, and is 0 otherwise. All other bits of the result are undefined. The
4179condition codes are evaluated identically to the
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4180<a href="#i_fcmp">'<tt>fcmp</tt>' instruction</a>.</p>
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]4181
4182<h5>Example:</h5>
4183<pre>
Chris Lattner1e5c5cd02008-10-13 16:55:18 +0000[diff] [blame]4184 <i>; yields: result=&lt;2 x i32&gt; &lt; i32 0, i32 -1 &gt;</i>
4185 &lt;result&gt; = vfcmp oeq &lt;2 x float&gt; &lt; float 4, float 0 &gt;, &lt; float 5, float 0 &gt;
4186
4187 <i>; yields: result=&lt;2 x i64&gt; &lt; i64 -1, i64 0 &gt;</i>
4188 &lt;result&gt; = vfcmp ult &lt;2 x double&gt; &lt; double 1, double 2 &gt;, &lt; double 2, double 2&gt;
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]4189</pre>
4190</div>
4191
4192<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4193<div class="doc_subsubsection">
4194 <a name="i_phi">'<tt>phi</tt>' Instruction</a>
4195</div>
4196
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4197<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4198
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4199<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4200
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4201<pre> &lt;result&gt; = phi &lt;ty&gt; [ &lt;val0&gt;, &lt;label0&gt;], ...<br></pre>
4202<h5>Overview:</h5>
4203<p>The '<tt>phi</tt>' instruction is used to implement the &#966; node in
4204the SSA graph representing the function.</p>
4205<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4206
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4207<p>The type of the incoming values is specified with the first type
4208field. After this, the '<tt>phi</tt>' instruction takes a list of pairs
4209as arguments, with one pair for each predecessor basic block of the
4210current block. Only values of <a href="#t_firstclass">first class</a>
4211type may be used as the value arguments to the PHI node. Only labels
4212may be used as the label arguments.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4213
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4214<p>There must be no non-phi instructions between the start of a basic
4215block and the PHI instructions: i.e. PHI instructions must be first in
4216a basic block.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4217
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4218<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4219
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4220<p>At runtime, the '<tt>phi</tt>' instruction logically takes on the value
4221specified by the pair corresponding to the predecessor basic block that executed
4222just prior to the current block.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4223
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4224<h5>Example:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4225<pre>
4226Loop: ; Infinite loop that counts from 0 on up...
4227 %indvar = phi i32 [ 0, %LoopHeader ], [ %nextindvar, %Loop ]
4228 %nextindvar = add i32 %indvar, 1
4229 br label %Loop
4230</pre>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4231</div>
4232
4233<!-- _______________________________________________________________________ -->
4234<div class="doc_subsubsection">
4235 <a name="i_select">'<tt>select</tt>' Instruction</a>
4236</div>
4237
4238<div class="doc_text">
4239
4240<h5>Syntax:</h5>
4241
4242<pre>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4243 &lt;result&gt; = select <i>selty</i> &lt;cond&gt;, &lt;ty&gt; &lt;val1&gt;, &lt;ty&gt; &lt;val2&gt; <i>; yields ty</i>
4244
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4245 <i>selty</i> is either i1 or {&lt;N x i1&gt;}
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4246</pre>
4247
4248<h5>Overview:</h5>
4249
4250<p>
4251The '<tt>select</tt>' instruction is used to choose one value based on a
4252condition, without branching.
4253</p>
4254
4255
4256<h5>Arguments:</h5>
4257
4258<p>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4259The '<tt>select</tt>' instruction requires an 'i1' value or
4260a vector of 'i1' values indicating the
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4261condition, and two values of the same <a href="#t_firstclass">first class</a>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4262type. If the val1/val2 are vectors and
4263the condition is a scalar, then entire vectors are selected, not
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4264individual elements.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4265</p>
4266
4267<h5>Semantics:</h5>
4268
4269<p>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4270If the condition is an i1 and it evaluates to 1, the instruction returns the first
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4271value argument; otherwise, it returns the second value argument.
4272</p>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4273<p>
4274If the condition is a vector of i1, then the value arguments must
4275be vectors of the same size, and the selection is done element
4276by element.
4277</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4278
4279<h5>Example:</h5>
4280
4281<pre>
4282 %X = select i1 true, i8 17, i8 42 <i>; yields i8:17</i>
4283</pre>
4284</div>
4285
4286
4287<!-- _______________________________________________________________________ -->
4288<div class="doc_subsubsection">
4289 <a name="i_call">'<tt>call</tt>' Instruction</a>
4290</div>
4291
4292<div class="doc_text">
4293
4294<h5>Syntax:</h5>
4295<pre>
Devang Pateld0bfcc72008-10-07 17:48:33 +0000[diff] [blame]4296 &lt;result&gt; = [tail] call [<a href="#callingconv">cconv</a>] [<a href="#paramattrs">ret attrs</a>] &lt;ty&gt; [&lt;fnty&gt;*] &lt;fnptrval&gt;(&lt;function args&gt;) [<a href="#fnattrs">fn attrs</a>]
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4297</pre>
4298
4299<h5>Overview:</h5>
4300
4301<p>The '<tt>call</tt>' instruction represents a simple function call.</p>
4302
4303<h5>Arguments:</h5>
4304
4305<p>This instruction requires several arguments:</p>
4306
4307<ol>
4308 <li>
4309 <p>The optional "tail" marker indicates whether the callee function accesses
4310 any allocas or varargs in the caller. If the "tail" marker is present, the
4311 function call is eligible for tail call optimization. Note that calls may
4312 be marked "tail" even if they do not occur before a <a
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4313 href="#i_ret"><tt>ret</tt></a> instruction.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4314 </li>
4315 <li>
4316 <p>The optional "cconv" marker indicates which <a href="#callingconv">calling
4317 convention</a> the call should use. If none is specified, the call defaults
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4318 to using C calling conventions.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4319 </li>
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]4320
4321 <li>
4322 <p>The optional <a href="#paramattrs">Parameter Attributes</a> list for
4323 return values. Only '<tt>zeroext</tt>', '<tt>signext</tt>',
4324 and '<tt>inreg</tt>' attributes are valid here.</p>
4325 </li>
4326
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4327 <li>
Nick Lewycky93082fc2007-09-08 13:57:50 +0000[diff] [blame]4328 <p>'<tt>ty</tt>': the type of the call instruction itself which is also
4329 the type of the return value. Functions that return no value are marked
4330 <tt><a href="#t_void">void</a></tt>.</p>
4331 </li>
4332 <li>
4333 <p>'<tt>fnty</tt>': shall be the signature of the pointer to function
4334 value being invoked. The argument types must match the types implied by
4335 this signature. This type can be omitted if the function is not varargs
4336 and if the function type does not return a pointer to a function.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4337 </li>
4338 <li>
4339 <p>'<tt>fnptrval</tt>': An LLVM value containing a pointer to a function to
4340 be invoked. In most cases, this is a direct function invocation, but
4341 indirect <tt>call</tt>s are just as possible, calling an arbitrary pointer
4342 to function value.</p>
4343 </li>
4344 <li>
4345 <p>'<tt>function args</tt>': argument list whose types match the
4346 function signature argument types. All arguments must be of
4347 <a href="#t_firstclass">first class</a> type. If the function signature
4348 indicates the function accepts a variable number of arguments, the extra
4349 arguments can be specified.</p>
4350 </li>
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]4351 <li>
Devang Pateld0bfcc72008-10-07 17:48:33 +0000[diff] [blame]4352 <p>The optional <a href="#fnattrs">function attributes</a> list. Only
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]4353 '<tt>noreturn</tt>', '<tt>nounwind</tt>', '<tt>readonly</tt>' and
4354 '<tt>readnone</tt>' attributes are valid here.</p>
4355 </li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4356</ol>
4357
4358<h5>Semantics:</h5>
4359
4360<p>The '<tt>call</tt>' instruction is used to cause control flow to
4361transfer to a specified function, with its incoming arguments bound to
4362the specified values. Upon a '<tt><a href="#i_ret">ret</a></tt>'
4363instruction in the called function, control flow continues with the
4364instruction after the function call, and the return value of the
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4365function is bound to the result argument.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4366
4367<h5>Example:</h5>
4368
4369<pre>
Nick Lewycky93082fc2007-09-08 13:57:50 +0000[diff] [blame]4370 %retval = call i32 @test(i32 %argc)
Chris Lattner5e893ef2008-03-21 17:24:17 +0000[diff] [blame]4371 call i32 (i8 *, ...)* @printf(i8 * %msg, i32 12, i8 42) <i>; yields i32</i>
4372 %X = tail call i32 @foo() <i>; yields i32</i>
4373 %Y = tail call <a href="#callingconv">fastcc</a> i32 @foo() <i>; yields i32</i>
4374 call void %foo(i8 97 signext)
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]4375
4376 %struct.A = type { i32, i8 }
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]4377 %r = call %struct.A @foo() <i>; yields { 32, i8 }</i>
Dan Gohman3e700032008-10-04 19:00:07 +0000[diff] [blame]4378 %gr = extractvalue %struct.A %r, 0 <i>; yields i32</i>
4379 %gr1 = extractvalue %struct.A %r, 1 <i>; yields i8</i>
Chris Lattnerac454b32008-10-08 06:26:11 +0000[diff] [blame]4380 %Z = call void @foo() noreturn <i>; indicates that %foo never returns normally</i>
Matthijs Kooijman2c4e05a2008-10-07 10:03:45 +0000[diff] [blame]4381 %ZZ = call zeroext i32 @bar() <i>; Return value is %zero extended</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4382</pre>
4383
4384</div>
4385
4386<!-- _______________________________________________________________________ -->
4387<div class="doc_subsubsection">
4388 <a name="i_va_arg">'<tt>va_arg</tt>' Instruction</a>
4389</div>
4390
4391<div class="doc_text">
4392
4393<h5>Syntax:</h5>
4394
4395<pre>
4396 &lt;resultval&gt; = va_arg &lt;va_list*&gt; &lt;arglist&gt;, &lt;argty&gt;
4397</pre>
4398
4399<h5>Overview:</h5>
4400
4401<p>The '<tt>va_arg</tt>' instruction is used to access arguments passed through
4402the "variable argument" area of a function call. It is used to implement the
4403<tt>va_arg</tt> macro in C.</p>
4404
4405<h5>Arguments:</h5>
4406
4407<p>This instruction takes a <tt>va_list*</tt> value and the type of
4408the argument. It returns a value of the specified argument type and
4409increments the <tt>va_list</tt> to point to the next argument. The
4410actual type of <tt>va_list</tt> is target specific.</p>
4411
4412<h5>Semantics:</h5>
4413
4414<p>The '<tt>va_arg</tt>' instruction loads an argument of the specified
4415type from the specified <tt>va_list</tt> and causes the
4416<tt>va_list</tt> to point to the next argument. For more information,
4417see the variable argument handling <a href="#int_varargs">Intrinsic
4418Functions</a>.</p>
4419
4420<p>It is legal for this instruction to be called in a function which does not
4421take a variable number of arguments, for example, the <tt>vfprintf</tt>
4422function.</p>
4423
4424<p><tt>va_arg</tt> is an LLVM instruction instead of an <a
4425href="#intrinsics">intrinsic function</a> because it takes a type as an
4426argument.</p>
4427
4428<h5>Example:</h5>
4429
4430<p>See the <a href="#int_varargs">variable argument processing</a> section.</p>
4431
4432</div>
4433
4434<!-- *********************************************************************** -->
4435<div class="doc_section"> <a name="intrinsics">Intrinsic Functions</a> </div>
4436<!-- *********************************************************************** -->
4437
4438<div class="doc_text">
4439
4440<p>LLVM supports the notion of an "intrinsic function". These functions have
4441well known names and semantics and are required to follow certain restrictions.
4442Overall, these intrinsics represent an extension mechanism for the LLVM
4443language that does not require changing all of the transformations in LLVM when
4444adding to the language (or the bitcode reader/writer, the parser, etc...).</p>
4445
4446<p>Intrinsic function names must all start with an "<tt>llvm.</tt>" prefix. This
4447prefix is reserved in LLVM for intrinsic names; thus, function names may not
4448begin with this prefix. Intrinsic functions must always be external functions:
4449you cannot define the body of intrinsic functions. Intrinsic functions may
4450only be used in call or invoke instructions: it is illegal to take the address
4451of an intrinsic function. Additionally, because intrinsic functions are part
4452of the LLVM language, it is required if any are added that they be documented
4453here.</p>
4454
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]4455<p>Some intrinsic functions can be overloaded, i.e., the intrinsic represents
4456a family of functions that perform the same operation but on different data
4457types. Because LLVM can represent over 8 million different integer types,
4458overloading is used commonly to allow an intrinsic function to operate on any
4459integer type. One or more of the argument types or the result type can be
4460overloaded to accept any integer type. Argument types may also be defined as
4461exactly matching a previous argument's type or the result type. This allows an
4462intrinsic function which accepts multiple arguments, but needs all of them to
4463be of the same type, to only be overloaded with respect to a single argument or
4464the result.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4465
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]4466<p>Overloaded intrinsics will have the names of its overloaded argument types
4467encoded into its function name, each preceded by a period. Only those types
4468which are overloaded result in a name suffix. Arguments whose type is matched
4469against another type do not. For example, the <tt>llvm.ctpop</tt> function can
4470take an integer of any width and returns an integer of exactly the same integer
4471width. This leads to a family of functions such as
4472<tt>i8 @llvm.ctpop.i8(i8 %val)</tt> and <tt>i29 @llvm.ctpop.i29(i29 %val)</tt>.
4473Only one type, the return type, is overloaded, and only one type suffix is
4474required. Because the argument's type is matched against the return type, it
4475does not require its own name suffix.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4476
4477<p>To learn how to add an intrinsic function, please see the
4478<a href="ExtendingLLVM.html">Extending LLVM Guide</a>.
4479</p>
4480
4481</div>
4482
4483<!-- ======================================================================= -->
4484<div class="doc_subsection">
4485 <a name="int_varargs">Variable Argument Handling Intrinsics</a>
4486</div>
4487
4488<div class="doc_text">
4489
4490<p>Variable argument support is defined in LLVM with the <a
4491 href="#i_va_arg"><tt>va_arg</tt></a> instruction and these three
4492intrinsic functions. These functions are related to the similarly
4493named macros defined in the <tt>&lt;stdarg.h&gt;</tt> header file.</p>
4494
4495<p>All of these functions operate on arguments that use a
4496target-specific value type "<tt>va_list</tt>". The LLVM assembly
4497language reference manual does not define what this type is, so all
4498transformations should be prepared to handle these functions regardless of
4499the type used.</p>
4500
4501<p>This example shows how the <a href="#i_va_arg"><tt>va_arg</tt></a>
4502instruction and the variable argument handling intrinsic functions are
4503used.</p>
4504
4505<div class="doc_code">
4506<pre>
4507define i32 @test(i32 %X, ...) {
4508 ; Initialize variable argument processing
4509 %ap = alloca i8*
4510 %ap2 = bitcast i8** %ap to i8*
4511 call void @llvm.va_start(i8* %ap2)
4512
4513 ; Read a single integer argument
4514 %tmp = va_arg i8** %ap, i32
4515
4516 ; Demonstrate usage of llvm.va_copy and llvm.va_end
4517 %aq = alloca i8*
4518 %aq2 = bitcast i8** %aq to i8*
4519 call void @llvm.va_copy(i8* %aq2, i8* %ap2)
4520 call void @llvm.va_end(i8* %aq2)
4521
4522 ; Stop processing of arguments.
4523 call void @llvm.va_end(i8* %ap2)
4524 ret i32 %tmp
4525}
4526
4527declare void @llvm.va_start(i8*)
4528declare void @llvm.va_copy(i8*, i8*)
4529declare void @llvm.va_end(i8*)
4530</pre>
4531</div>
4532
4533</div>
4534
4535<!-- _______________________________________________________________________ -->
4536<div class="doc_subsubsection">
4537 <a name="int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a>
4538</div>
4539
4540
4541<div class="doc_text">
4542<h5>Syntax:</h5>
4543<pre> declare void %llvm.va_start(i8* &lt;arglist&gt;)<br></pre>
4544<h5>Overview:</h5>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4545<p>The '<tt>llvm.va_start</tt>' intrinsic initializes
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4546<tt>*&lt;arglist&gt;</tt> for subsequent use by <tt><a
4547href="#i_va_arg">va_arg</a></tt>.</p>
4548
4549<h5>Arguments:</h5>
4550
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4551<p>The argument is a pointer to a <tt>va_list</tt> element to initialize.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4552
4553<h5>Semantics:</h5>
4554
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4555<p>The '<tt>llvm.va_start</tt>' intrinsic works just like the <tt>va_start</tt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4556macro available in C. In a target-dependent way, it initializes the
4557<tt>va_list</tt> element to which the argument points, so that the next call to
4558<tt>va_arg</tt> will produce the first variable argument passed to the function.
4559Unlike the C <tt>va_start</tt> macro, this intrinsic does not need to know the
4560last argument of the function as the compiler can figure that out.</p>
4561
4562</div>
4563
4564<!-- _______________________________________________________________________ -->
4565<div class="doc_subsubsection">
4566 <a name="int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a>
4567</div>
4568
4569<div class="doc_text">
4570<h5>Syntax:</h5>
4571<pre> declare void @llvm.va_end(i8* &lt;arglist&gt;)<br></pre>
4572<h5>Overview:</h5>
4573
4574<p>The '<tt>llvm.va_end</tt>' intrinsic destroys <tt>*&lt;arglist&gt;</tt>,
4575which has been initialized previously with <tt><a href="#int_va_start">llvm.va_start</a></tt>
4576or <tt><a href="#i_va_copy">llvm.va_copy</a></tt>.</p>
4577
4578<h5>Arguments:</h5>
4579
4580<p>The argument is a pointer to a <tt>va_list</tt> to destroy.</p>
4581
4582<h5>Semantics:</h5>
4583
4584<p>The '<tt>llvm.va_end</tt>' intrinsic works just like the <tt>va_end</tt>
4585macro available in C. In a target-dependent way, it destroys the
4586<tt>va_list</tt> element to which the argument points. Calls to <a
4587href="#int_va_start"><tt>llvm.va_start</tt></a> and <a href="#int_va_copy">
4588<tt>llvm.va_copy</tt></a> must be matched exactly with calls to
4589<tt>llvm.va_end</tt>.</p>
4590
4591</div>
4592
4593<!-- _______________________________________________________________________ -->
4594<div class="doc_subsubsection">
4595 <a name="int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a>
4596</div>
4597
4598<div class="doc_text">
4599
4600<h5>Syntax:</h5>
4601
4602<pre>
4603 declare void @llvm.va_copy(i8* &lt;destarglist&gt;, i8* &lt;srcarglist&gt;)
4604</pre>
4605
4606<h5>Overview:</h5>
4607
4608<p>The '<tt>llvm.va_copy</tt>' intrinsic copies the current argument position
4609from the source argument list to the destination argument list.</p>
4610
4611<h5>Arguments:</h5>
4612
4613<p>The first argument is a pointer to a <tt>va_list</tt> element to initialize.
4614The second argument is a pointer to a <tt>va_list</tt> element to copy from.</p>
4615
4616
4617<h5>Semantics:</h5>
4618
4619<p>The '<tt>llvm.va_copy</tt>' intrinsic works just like the <tt>va_copy</tt>
4620macro available in C. In a target-dependent way, it copies the source
4621<tt>va_list</tt> element into the destination <tt>va_list</tt> element. This
4622intrinsic is necessary because the <tt><a href="#int_va_start">
4623llvm.va_start</a></tt> intrinsic may be arbitrarily complex and require, for
4624example, memory allocation.</p>
4625
4626</div>
4627
4628<!-- ======================================================================= -->
4629<div class="doc_subsection">
4630 <a name="int_gc">Accurate Garbage Collection Intrinsics</a>
4631</div>
4632
4633<div class="doc_text">
4634
4635<p>
4636LLVM support for <a href="GarbageCollection.html">Accurate Garbage
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]4637Collection</a> (GC) requires the implementation and generation of these
4638intrinsics.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4639These intrinsics allow identification of <a href="#int_gcroot">GC roots on the
4640stack</a>, as well as garbage collector implementations that require <a
4641href="#int_gcread">read</a> and <a href="#int_gcwrite">write</a> barriers.
4642Front-ends for type-safe garbage collected languages should generate these
4643intrinsics to make use of the LLVM garbage collectors. For more details, see <a
4644href="GarbageCollection.html">Accurate Garbage Collection with LLVM</a>.
4645</p>
Christopher Lambcfe00962007-12-17 01:00:21 +0000[diff] [blame]4646
4647<p>The garbage collection intrinsics only operate on objects in the generic
4648 address space (address space zero).</p>
4649
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4650</div>
4651
4652<!-- _______________________________________________________________________ -->
4653<div class="doc_subsubsection">
4654 <a name="int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a>
4655</div>
4656
4657<div class="doc_text">
4658
4659<h5>Syntax:</h5>
4660
4661<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4662 declare void @llvm.gcroot(i8** %ptrloc, i8* %metadata)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4663</pre>
4664
4665<h5>Overview:</h5>
4666
4667<p>The '<tt>llvm.gcroot</tt>' intrinsic declares the existence of a GC root to
4668the code generator, and allows some metadata to be associated with it.</p>
4669
4670<h5>Arguments:</h5>
4671
4672<p>The first argument specifies the address of a stack object that contains the
4673root pointer. The second pointer (which must be either a constant or a global
4674value address) contains the meta-data to be associated with the root.</p>
4675
4676<h5>Semantics:</h5>
4677
Chris Lattnera7d94ba2008-04-24 05:59:56 +0000[diff] [blame]4678<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]4679location. At compile-time, the code generator generates information to allow
Gordon Henriksen40393542007-12-25 02:31:26 +0000[diff] [blame]4680the runtime to find the pointer at GC safe points. The '<tt>llvm.gcroot</tt>'
4681intrinsic may only be used in a function which <a href="#gc">specifies a GC
4682algorithm</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4683
4684</div>
4685
4686
4687<!-- _______________________________________________________________________ -->
4688<div class="doc_subsubsection">
4689 <a name="int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a>
4690</div>
4691
4692<div class="doc_text">
4693
4694<h5>Syntax:</h5>
4695
4696<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4697 declare i8* @llvm.gcread(i8* %ObjPtr, i8** %Ptr)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4698</pre>
4699
4700<h5>Overview:</h5>
4701
4702<p>The '<tt>llvm.gcread</tt>' intrinsic identifies reads of references from heap
4703locations, allowing garbage collector implementations that require read
4704barriers.</p>
4705
4706<h5>Arguments:</h5>
4707
4708<p>The second argument is the address to read from, which should be an address
4709allocated from the garbage collector. The first object is a pointer to the
4710start of the referenced object, if needed by the language runtime (otherwise
4711null).</p>
4712
4713<h5>Semantics:</h5>
4714
4715<p>The '<tt>llvm.gcread</tt>' intrinsic has the same semantics as a load
4716instruction, but may be replaced with substantially more complex code by the
Gordon Henriksen40393542007-12-25 02:31:26 +0000[diff] [blame]4717garbage collector runtime, as needed. The '<tt>llvm.gcread</tt>' intrinsic
4718may only be used in a function which <a href="#gc">specifies a GC
4719algorithm</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4720
4721</div>
4722
4723
4724<!-- _______________________________________________________________________ -->
4725<div class="doc_subsubsection">
4726 <a name="int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a>
4727</div>
4728
4729<div class="doc_text">
4730
4731<h5>Syntax:</h5>
4732
4733<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4734 declare void @llvm.gcwrite(i8* %P1, i8* %Obj, i8** %P2)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4735</pre>
4736
4737<h5>Overview:</h5>
4738
4739<p>The '<tt>llvm.gcwrite</tt>' intrinsic identifies writes of references to heap
4740locations, allowing garbage collector implementations that require write
4741barriers (such as generational or reference counting collectors).</p>
4742
4743<h5>Arguments:</h5>
4744
4745<p>The first argument is the reference to store, the second is the start of the
4746object to store it to, and the third is the address of the field of Obj to
4747store to. If the runtime does not require a pointer to the object, Obj may be
4748null.</p>
4749
4750<h5>Semantics:</h5>
4751
4752<p>The '<tt>llvm.gcwrite</tt>' intrinsic has the same semantics as a store
4753instruction, but may be replaced with substantially more complex code by the
Gordon Henriksen40393542007-12-25 02:31:26 +0000[diff] [blame]4754garbage collector runtime, as needed. The '<tt>llvm.gcwrite</tt>' intrinsic
4755may only be used in a function which <a href="#gc">specifies a GC
4756algorithm</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4757
4758</div>
4759
4760
4761
4762<!-- ======================================================================= -->
4763<div class="doc_subsection">
4764 <a name="int_codegen">Code Generator Intrinsics</a>
4765</div>
4766
4767<div class="doc_text">
4768<p>
4769These intrinsics are provided by LLVM to expose special features that may only
4770be implemented with code generator support.
4771</p>
4772
4773</div>
4774
4775<!-- _______________________________________________________________________ -->
4776<div class="doc_subsubsection">
4777 <a name="int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a>
4778</div>
4779
4780<div class="doc_text">
4781
4782<h5>Syntax:</h5>
4783<pre>
4784 declare i8 *@llvm.returnaddress(i32 &lt;level&gt;)
4785</pre>
4786
4787<h5>Overview:</h5>
4788
4789<p>
4790The '<tt>llvm.returnaddress</tt>' intrinsic attempts to compute a
4791target-specific value indicating the return address of the current function
4792or one of its callers.
4793</p>
4794
4795<h5>Arguments:</h5>
4796
4797<p>
4798The argument to this intrinsic indicates which function to return the address
4799for. Zero indicates the calling function, one indicates its caller, etc. The
4800argument is <b>required</b> to be a constant integer value.
4801</p>
4802
4803<h5>Semantics:</h5>
4804
4805<p>
4806The '<tt>llvm.returnaddress</tt>' intrinsic either returns a pointer indicating
4807the return address of the specified call frame, or zero if it cannot be
4808identified. The value returned by this intrinsic is likely to be incorrect or 0
4809for arguments other than zero, so it should only be used for debugging purposes.
4810</p>
4811
4812<p>
4813Note that calling this intrinsic does not prevent function inlining or other
4814aggressive transformations, so the value returned may not be that of the obvious
4815source-language caller.
4816</p>
4817</div>
4818
4819
4820<!-- _______________________________________________________________________ -->
4821<div class="doc_subsubsection">
4822 <a name="int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a>
4823</div>
4824
4825<div class="doc_text">
4826
4827<h5>Syntax:</h5>
4828<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4829 declare i8 *@llvm.frameaddress(i32 &lt;level&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4830</pre>
4831
4832<h5>Overview:</h5>
4833
4834<p>
4835The '<tt>llvm.frameaddress</tt>' intrinsic attempts to return the
4836target-specific frame pointer value for the specified stack frame.
4837</p>
4838
4839<h5>Arguments:</h5>
4840
4841<p>
4842The argument to this intrinsic indicates which function to return the frame
4843pointer for. Zero indicates the calling function, one indicates its caller,
4844etc. The argument is <b>required</b> to be a constant integer value.
4845</p>
4846
4847<h5>Semantics:</h5>
4848
4849<p>
4850The '<tt>llvm.frameaddress</tt>' intrinsic either returns a pointer indicating
4851the frame address of the specified call frame, or zero if it cannot be
4852identified. The value returned by this intrinsic is likely to be incorrect or 0
4853for arguments other than zero, so it should only be used for debugging purposes.
4854</p>
4855
4856<p>
4857Note that calling this intrinsic does not prevent function inlining or other
4858aggressive transformations, so the value returned may not be that of the obvious
4859source-language caller.
4860</p>
4861</div>
4862
4863<!-- _______________________________________________________________________ -->
4864<div class="doc_subsubsection">
4865 <a name="int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a>
4866</div>
4867
4868<div class="doc_text">
4869
4870<h5>Syntax:</h5>
4871<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4872 declare i8 *@llvm.stacksave()
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4873</pre>
4874
4875<h5>Overview:</h5>
4876
4877<p>
4878The '<tt>llvm.stacksave</tt>' intrinsic is used to remember the current state of
4879the function stack, for use with <a href="#int_stackrestore">
4880<tt>llvm.stackrestore</tt></a>. This is useful for implementing language
4881features like scoped automatic variable sized arrays in C99.
4882</p>
4883
4884<h5>Semantics:</h5>
4885
4886<p>
4887This intrinsic returns a opaque pointer value that can be passed to <a
4888href="#int_stackrestore"><tt>llvm.stackrestore</tt></a>. When an
4889<tt>llvm.stackrestore</tt> intrinsic is executed with a value saved from
4890<tt>llvm.stacksave</tt>, it effectively restores the state of the stack to the
4891state it was in when the <tt>llvm.stacksave</tt> intrinsic executed. In
4892practice, this pops any <a href="#i_alloca">alloca</a> blocks from the stack
4893that were allocated after the <tt>llvm.stacksave</tt> was executed.
4894</p>
4895
4896</div>
4897
4898<!-- _______________________________________________________________________ -->
4899<div class="doc_subsubsection">
4900 <a name="int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a>
4901</div>
4902
4903<div class="doc_text">
4904
4905<h5>Syntax:</h5>
4906<pre>
4907 declare void @llvm.stackrestore(i8 * %ptr)
4908</pre>
4909
4910<h5>Overview:</h5>
4911
4912<p>
4913The '<tt>llvm.stackrestore</tt>' intrinsic is used to restore the state of
4914the function stack to the state it was in when the corresponding <a
4915href="#int_stacksave"><tt>llvm.stacksave</tt></a> intrinsic executed. This is
4916useful for implementing language features like scoped automatic variable sized
4917arrays in C99.
4918</p>
4919
4920<h5>Semantics:</h5>
4921
4922<p>
4923See the description for <a href="#int_stacksave"><tt>llvm.stacksave</tt></a>.
4924</p>
4925
4926</div>
4927
4928
4929<!-- _______________________________________________________________________ -->
4930<div class="doc_subsubsection">
4931 <a name="int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a>
4932</div>
4933
4934<div class="doc_text">
4935
4936<h5>Syntax:</h5>
4937<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4938 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]4939</pre>
4940
4941<h5>Overview:</h5>
4942
4943
4944<p>
4945The '<tt>llvm.prefetch</tt>' intrinsic is a hint to the code generator to insert
4946a prefetch instruction if supported; otherwise, it is a noop. Prefetches have
4947no
4948effect on the behavior of the program but can change its performance
4949characteristics.
4950</p>
4951
4952<h5>Arguments:</h5>
4953
4954<p>
4955<tt>address</tt> is the address to be prefetched, <tt>rw</tt> is the specifier
4956determining if the fetch should be for a read (0) or write (1), and
4957<tt>locality</tt> is a temporal locality specifier ranging from (0) - no
4958locality, to (3) - extremely local keep in cache. The <tt>rw</tt> and
4959<tt>locality</tt> arguments must be constant integers.
4960</p>
4961
4962<h5>Semantics:</h5>
4963
4964<p>
4965This intrinsic does not modify the behavior of the program. In particular,
4966prefetches cannot trap and do not produce a value. On targets that support this
4967intrinsic, the prefetch can provide hints to the processor cache for better
4968performance.
4969</p>
4970
4971</div>
4972
4973<!-- _______________________________________________________________________ -->
4974<div class="doc_subsubsection">
4975 <a name="int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a>
4976</div>
4977
4978<div class="doc_text">
4979
4980<h5>Syntax:</h5>
4981<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4982 declare void @llvm.pcmarker(i32 &lt;id&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4983</pre>
4984
4985<h5>Overview:</h5>
4986
4987
4988<p>
4989The '<tt>llvm.pcmarker</tt>' intrinsic is a method to export a Program Counter
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]4990(PC) in a region of
4991code to simulators and other tools. The method is target specific, but it is
4992expected that the marker will use exported symbols to transmit the PC of the
4993marker.
4994The marker makes no guarantees that it will remain with any specific instruction
4995after optimizations. It is possible that the presence of a marker will inhibit
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4996optimizations. The intended use is to be inserted after optimizations to allow
4997correlations of simulation runs.
4998</p>
4999
5000<h5>Arguments:</h5>
5001
5002<p>
5003<tt>id</tt> is a numerical id identifying the marker.
5004</p>
5005
5006<h5>Semantics:</h5>
5007
5008<p>
5009This intrinsic does not modify the behavior of the program. Backends that do not
5010support this intrinisic may ignore it.
5011</p>
5012
5013</div>
5014
5015<!-- _______________________________________________________________________ -->
5016<div class="doc_subsubsection">
5017 <a name="int_readcyclecounter">'<tt>llvm.readcyclecounter</tt>' Intrinsic</a>
5018</div>
5019
5020<div class="doc_text">
5021
5022<h5>Syntax:</h5>
5023<pre>
5024 declare i64 @llvm.readcyclecounter( )
5025</pre>
5026
5027<h5>Overview:</h5>
5028
5029
5030<p>
5031The '<tt>llvm.readcyclecounter</tt>' intrinsic provides access to the cycle
5032counter register (or similar low latency, high accuracy clocks) on those targets
5033that support it. On X86, it should map to RDTSC. On Alpha, it should map to RPCC.
5034As the backing counters overflow quickly (on the order of 9 seconds on alpha), this
5035should only be used for small timings.
5036</p>
5037
5038<h5>Semantics:</h5>
5039
5040<p>
5041When directly supported, reading the cycle counter should not modify any memory.
5042Implementations are allowed to either return a application specific value or a
5043system wide value. On backends without support, this is lowered to a constant 0.
5044</p>
5045
5046</div>
5047
5048<!-- ======================================================================= -->
5049<div class="doc_subsection">
5050 <a name="int_libc">Standard C Library Intrinsics</a>
5051</div>
5052
5053<div class="doc_text">
5054<p>
5055LLVM provides intrinsics for a few important standard C library functions.
5056These intrinsics allow source-language front-ends to pass information about the
5057alignment of the pointer arguments to the code generator, providing opportunity
5058for more efficient code generation.
5059</p>
5060
5061</div>
5062
5063<!-- _______________________________________________________________________ -->
5064<div class="doc_subsubsection">
5065 <a name="int_memcpy">'<tt>llvm.memcpy</tt>' Intrinsic</a>
5066</div>
5067
5068<div class="doc_text">
5069
5070<h5>Syntax:</h5>
5071<pre>
5072 declare void @llvm.memcpy.i32(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5073 i32 &lt;len&gt;, i32 &lt;align&gt;)
5074 declare void @llvm.memcpy.i64(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5075 i64 &lt;len&gt;, i32 &lt;align&gt;)
5076</pre>
5077
5078<h5>Overview:</h5>
5079
5080<p>
5081The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the source
5082location to the destination location.
5083</p>
5084
5085<p>
5086Note that, unlike the standard libc function, the <tt>llvm.memcpy.*</tt>
5087intrinsics do not return a value, and takes an extra alignment argument.
5088</p>
5089
5090<h5>Arguments:</h5>
5091
5092<p>
5093The first argument is a pointer to the destination, the second is a pointer to
5094the source. The third argument is an integer argument
5095specifying the number of bytes to copy, and the fourth argument is the alignment
5096of the source and destination locations.
5097</p>
5098
5099<p>
5100If the call to this intrinisic has an alignment value that is not 0 or 1, then
5101the caller guarantees that both the source and destination pointers are aligned
5102to that boundary.
5103</p>
5104
5105<h5>Semantics:</h5>
5106
5107<p>
5108The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the source
5109location to the destination location, which are not allowed to overlap. It
5110copies "len" bytes of memory over. If the argument is known to be aligned to
5111some boundary, this can be specified as the fourth argument, otherwise it should
5112be set to 0 or 1.
5113</p>
5114</div>
5115
5116
5117<!-- _______________________________________________________________________ -->
5118<div class="doc_subsubsection">
5119 <a name="int_memmove">'<tt>llvm.memmove</tt>' Intrinsic</a>
5120</div>
5121
5122<div class="doc_text">
5123
5124<h5>Syntax:</h5>
5125<pre>
5126 declare void @llvm.memmove.i32(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5127 i32 &lt;len&gt;, i32 &lt;align&gt;)
5128 declare void @llvm.memmove.i64(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5129 i64 &lt;len&gt;, i32 &lt;align&gt;)
5130</pre>
5131
5132<h5>Overview:</h5>
5133
5134<p>
5135The '<tt>llvm.memmove.*</tt>' intrinsics move a block of memory from the source
5136location to the destination location. It is similar to the
Chris Lattnerdba16ea2008-01-06 19:51:52 +0000[diff] [blame]5137'<tt>llvm.memcpy</tt>' intrinsic but allows the two memory locations to overlap.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5138</p>
5139
5140<p>
5141Note that, unlike the standard libc function, the <tt>llvm.memmove.*</tt>
5142intrinsics do not return a value, and takes an extra alignment argument.
5143</p>
5144
5145<h5>Arguments:</h5>
5146
5147<p>
5148The first argument is a pointer to the destination, the second is a pointer to
5149the source. The third argument is an integer argument
5150specifying the number of bytes to copy, and the fourth argument is the alignment
5151of the source and destination locations.
5152</p>
5153
5154<p>
5155If the call to this intrinisic has an alignment value that is not 0 or 1, then
5156the caller guarantees that the source and destination pointers are aligned to
5157that boundary.
5158</p>
5159
5160<h5>Semantics:</h5>
5161
5162<p>
5163The '<tt>llvm.memmove.*</tt>' intrinsics copy a block of memory from the source
5164location to the destination location, which may overlap. It
5165copies "len" bytes of memory over. If the argument is known to be aligned to
5166some boundary, this can be specified as the fourth argument, otherwise it should
5167be set to 0 or 1.
5168</p>
5169</div>
5170
5171
5172<!-- _______________________________________________________________________ -->
5173<div class="doc_subsubsection">
5174 <a name="int_memset">'<tt>llvm.memset.*</tt>' Intrinsics</a>
5175</div>
5176
5177<div class="doc_text">
5178
5179<h5>Syntax:</h5>
5180<pre>
5181 declare void @llvm.memset.i32(i8 * &lt;dest&gt;, i8 &lt;val&gt;,
5182 i32 &lt;len&gt;, i32 &lt;align&gt;)
5183 declare void @llvm.memset.i64(i8 * &lt;dest&gt;, i8 &lt;val&gt;,
5184 i64 &lt;len&gt;, i32 &lt;align&gt;)
5185</pre>
5186
5187<h5>Overview:</h5>
5188
5189<p>
5190The '<tt>llvm.memset.*</tt>' intrinsics fill a block of memory with a particular
5191byte value.
5192</p>
5193
5194<p>
5195Note that, unlike the standard libc function, the <tt>llvm.memset</tt> intrinsic
5196does not return a value, and takes an extra alignment argument.
5197</p>
5198
5199<h5>Arguments:</h5>
5200
5201<p>
5202The first argument is a pointer to the destination to fill, the second is the
5203byte value to fill it with, the third argument is an integer
5204argument specifying the number of bytes to fill, and the fourth argument is the
5205known alignment of destination location.
5206</p>
5207
5208<p>
5209If the call to this intrinisic has an alignment value that is not 0 or 1, then
5210the caller guarantees that the destination pointer is aligned to that boundary.
5211</p>
5212
5213<h5>Semantics:</h5>
5214
5215<p>
5216The '<tt>llvm.memset.*</tt>' intrinsics fill "len" bytes of memory starting at
5217the
5218destination location. If the argument is known to be aligned to some boundary,
5219this can be specified as the fourth argument, otherwise it should be set to 0 or
52201.
5221</p>
5222</div>
5223
5224
5225<!-- _______________________________________________________________________ -->
5226<div class="doc_subsubsection">
5227 <a name="int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a>
5228</div>
5229
5230<div class="doc_text">
5231
5232<h5>Syntax:</h5>
Dale Johannesenf9adbb62007-10-02 17:47:38 +0000[diff] [blame]5233<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]5234floating point or vector of floating point type. Not all targets support all
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5235types however.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5236<pre>
Dale Johannesenf9adbb62007-10-02 17:47:38 +0000[diff] [blame]5237 declare float @llvm.sqrt.f32(float %Val)
5238 declare double @llvm.sqrt.f64(double %Val)
5239 declare x86_fp80 @llvm.sqrt.f80(x86_fp80 %Val)
5240 declare fp128 @llvm.sqrt.f128(fp128 %Val)
5241 declare ppc_fp128 @llvm.sqrt.ppcf128(ppc_fp128 %Val)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5242</pre>
5243
5244<h5>Overview:</h5>
5245
5246<p>
5247The '<tt>llvm.sqrt</tt>' intrinsics return the sqrt of the specified operand,
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5248returning the same value as the libm '<tt>sqrt</tt>' functions would. Unlike
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5249<tt>sqrt</tt> in libm, however, <tt>llvm.sqrt</tt> has undefined behavior for
Chris Lattnerf914a002008-01-29 07:00:44 +0000[diff] [blame]5250negative numbers other than -0.0 (which allows for better optimization, because
5251there is no need to worry about errno being set). <tt>llvm.sqrt(-0.0)</tt> is
5252defined to return -0.0 like IEEE sqrt.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5253</p>
5254
5255<h5>Arguments:</h5>
5256
5257<p>
5258The argument and return value are floating point numbers of the same type.
5259</p>
5260
5261<h5>Semantics:</h5>
5262
5263<p>
5264This function returns the sqrt of the specified operand if it is a nonnegative
5265floating point number.
5266</p>
5267</div>
5268
5269<!-- _______________________________________________________________________ -->
5270<div class="doc_subsubsection">
5271 <a name="int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a>
5272</div>
5273
5274<div class="doc_text">
5275
5276<h5>Syntax:</h5>
Dale Johannesenf9adbb62007-10-02 17:47:38 +0000[diff] [blame]5277<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]5278floating point or vector of floating point type. Not all targets support all
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5279types however.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5280<pre>
Dale Johannesenf9adbb62007-10-02 17:47:38 +0000[diff] [blame]5281 declare float @llvm.powi.f32(float %Val, i32 %power)
5282 declare double @llvm.powi.f64(double %Val, i32 %power)
5283 declare x86_fp80 @llvm.powi.f80(x86_fp80 %Val, i32 %power)
5284 declare fp128 @llvm.powi.f128(fp128 %Val, i32 %power)
5285 declare ppc_fp128 @llvm.powi.ppcf128(ppc_fp128 %Val, i32 %power)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5286</pre>
5287
5288<h5>Overview:</h5>
5289
5290<p>
5291The '<tt>llvm.powi.*</tt>' intrinsics return the first operand raised to the
5292specified (positive or negative) power. The order of evaluation of
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5293multiplications is not defined. When a vector of floating point type is
5294used, the second argument remains a scalar integer value.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5295</p>
5296
5297<h5>Arguments:</h5>
5298
5299<p>
5300The second argument is an integer power, and the first is a value to raise to
5301that power.
5302</p>
5303
5304<h5>Semantics:</h5>
5305
5306<p>
5307This function returns the first value raised to the second power with an
5308unspecified sequence of rounding operations.</p>
5309</div>
5310
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5311<!-- _______________________________________________________________________ -->
5312<div class="doc_subsubsection">
5313 <a name="int_sin">'<tt>llvm.sin.*</tt>' Intrinsic</a>
5314</div>
5315
5316<div class="doc_text">
5317
5318<h5>Syntax:</h5>
5319<p>This is an overloaded intrinsic. You can use <tt>llvm.sin</tt> on any
5320floating point or vector of floating point type. Not all targets support all
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5321types however.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5322<pre>
5323 declare float @llvm.sin.f32(float %Val)
5324 declare double @llvm.sin.f64(double %Val)
5325 declare x86_fp80 @llvm.sin.f80(x86_fp80 %Val)
5326 declare fp128 @llvm.sin.f128(fp128 %Val)
5327 declare ppc_fp128 @llvm.sin.ppcf128(ppc_fp128 %Val)
5328</pre>
5329
5330<h5>Overview:</h5>
5331
5332<p>
5333The '<tt>llvm.sin.*</tt>' intrinsics return the sine of the operand.
5334</p>
5335
5336<h5>Arguments:</h5>
5337
5338<p>
5339The argument and return value are floating point numbers of the same type.
5340</p>
5341
5342<h5>Semantics:</h5>
5343
5344<p>
5345This function returns the sine of the specified operand, returning the
5346same values as the libm <tt>sin</tt> functions would, and handles error
Dan Gohmaneaba92e2007-10-17 18:05:13 +0000[diff] [blame]5347conditions in the same way.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5348</div>
5349
5350<!-- _______________________________________________________________________ -->
5351<div class="doc_subsubsection">
5352 <a name="int_cos">'<tt>llvm.cos.*</tt>' Intrinsic</a>
5353</div>
5354
5355<div class="doc_text">
5356
5357<h5>Syntax:</h5>
5358<p>This is an overloaded intrinsic. You can use <tt>llvm.cos</tt> on any
5359floating point or vector of floating point type. Not all targets support all
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5360types however.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5361<pre>
5362 declare float @llvm.cos.f32(float %Val)
5363 declare double @llvm.cos.f64(double %Val)
5364 declare x86_fp80 @llvm.cos.f80(x86_fp80 %Val)
5365 declare fp128 @llvm.cos.f128(fp128 %Val)
5366 declare ppc_fp128 @llvm.cos.ppcf128(ppc_fp128 %Val)
5367</pre>
5368
5369<h5>Overview:</h5>
5370
5371<p>
5372The '<tt>llvm.cos.*</tt>' intrinsics return the cosine of the operand.
5373</p>
5374
5375<h5>Arguments:</h5>
5376
5377<p>
5378The argument and return value are floating point numbers of the same type.
5379</p>
5380
5381<h5>Semantics:</h5>
5382
5383<p>
5384This function returns the cosine of the specified operand, returning the
5385same values as the libm <tt>cos</tt> functions would, and handles error
Dan Gohmaneaba92e2007-10-17 18:05:13 +0000[diff] [blame]5386conditions in the same way.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5387</div>
5388
5389<!-- _______________________________________________________________________ -->
5390<div class="doc_subsubsection">
5391 <a name="int_pow">'<tt>llvm.pow.*</tt>' Intrinsic</a>
5392</div>
5393
5394<div class="doc_text">
5395
5396<h5>Syntax:</h5>
5397<p>This is an overloaded intrinsic. You can use <tt>llvm.pow</tt> on any
5398floating point or vector of floating point type. Not all targets support all
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5399types however.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5400<pre>
5401 declare float @llvm.pow.f32(float %Val, float %Power)
5402 declare double @llvm.pow.f64(double %Val, double %Power)
5403 declare x86_fp80 @llvm.pow.f80(x86_fp80 %Val, x86_fp80 %Power)
5404 declare fp128 @llvm.pow.f128(fp128 %Val, fp128 %Power)
5405 declare ppc_fp128 @llvm.pow.ppcf128(ppc_fp128 %Val, ppc_fp128 Power)
5406</pre>
5407
5408<h5>Overview:</h5>
5409
5410<p>
5411The '<tt>llvm.pow.*</tt>' intrinsics return the first operand raised to the
5412specified (positive or negative) power.
5413</p>
5414
5415<h5>Arguments:</h5>
5416
5417<p>
5418The second argument is a floating point power, and the first is a value to
5419raise to that power.
5420</p>
5421
5422<h5>Semantics:</h5>
5423
5424<p>
5425This function returns the first value raised to the second power,
5426returning the
5427same values as the libm <tt>pow</tt> functions would, and handles error
Dan Gohmaneaba92e2007-10-17 18:05:13 +0000[diff] [blame]5428conditions in the same way.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5429</div>
5430
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5431
5432<!-- ======================================================================= -->
5433<div class="doc_subsection">
5434 <a name="int_manip">Bit Manipulation Intrinsics</a>
5435</div>
5436
5437<div class="doc_text">
5438<p>
5439LLVM provides intrinsics for a few important bit manipulation operations.
5440These allow efficient code generation for some algorithms.
5441</p>
5442
5443</div>
5444
5445<!-- _______________________________________________________________________ -->
5446<div class="doc_subsubsection">
5447 <a name="int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a>
5448</div>
5449
5450<div class="doc_text">
5451
5452<h5>Syntax:</h5>
5453<p>This is an overloaded intrinsic function. You can use bswap on any integer
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5454type that is an even number of bytes (i.e. BitWidth % 16 == 0).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5455<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5456 declare i16 @llvm.bswap.i16(i16 &lt;id&gt;)
5457 declare i32 @llvm.bswap.i32(i32 &lt;id&gt;)
5458 declare i64 @llvm.bswap.i64(i64 &lt;id&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5459</pre>
5460
5461<h5>Overview:</h5>
5462
5463<p>
5464The '<tt>llvm.bswap</tt>' family of intrinsics is used to byte swap integer
5465values with an even number of bytes (positive multiple of 16 bits). These are
5466useful for performing operations on data that is not in the target's native
5467byte order.
5468</p>
5469
5470<h5>Semantics:</h5>
5471
5472<p>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5473The <tt>llvm.bswap.i16</tt> intrinsic returns an i16 value that has the high
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5474and low byte of the input i16 swapped. Similarly, the <tt>llvm.bswap.i32</tt>
5475intrinsic returns an i32 value that has the four bytes of the input i32
5476swapped, 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]5477i32 will have its bytes in 3, 2, 1, 0 order. The <tt>llvm.bswap.i48</tt>,
5478<tt>llvm.bswap.i64</tt> and other intrinsics extend this concept to
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5479additional even-byte lengths (6 bytes, 8 bytes and more, respectively).
5480</p>
5481
5482</div>
5483
5484<!-- _______________________________________________________________________ -->
5485<div class="doc_subsubsection">
5486 <a name="int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic</a>
5487</div>
5488
5489<div class="doc_text">
5490
5491<h5>Syntax:</h5>
5492<p>This is an overloaded intrinsic. You can use llvm.ctpop on any integer bit
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5493width. Not all targets support all bit widths however.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5494<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5495 declare i8 @llvm.ctpop.i8 (i8 &lt;src&gt;)
5496 declare i16 @llvm.ctpop.i16(i16 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5497 declare i32 @llvm.ctpop.i32(i32 &lt;src&gt;)
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5498 declare i64 @llvm.ctpop.i64(i64 &lt;src&gt;)
5499 declare i256 @llvm.ctpop.i256(i256 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5500</pre>
5501
5502<h5>Overview:</h5>
5503
5504<p>
5505The '<tt>llvm.ctpop</tt>' family of intrinsics counts the number of bits set in a
5506value.
5507</p>
5508
5509<h5>Arguments:</h5>
5510
5511<p>
5512The only argument is the value to be counted. The argument may be of any
5513integer type. The return type must match the argument type.
5514</p>
5515
5516<h5>Semantics:</h5>
5517
5518<p>
5519The '<tt>llvm.ctpop</tt>' intrinsic counts the 1's in a variable.
5520</p>
5521</div>
5522
5523<!-- _______________________________________________________________________ -->
5524<div class="doc_subsubsection">
5525 <a name="int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic</a>
5526</div>
5527
5528<div class="doc_text">
5529
5530<h5>Syntax:</h5>
5531<p>This is an overloaded intrinsic. You can use <tt>llvm.ctlz</tt> on any
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5532integer bit width. Not all targets support all bit widths however.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5533<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5534 declare i8 @llvm.ctlz.i8 (i8 &lt;src&gt;)
5535 declare i16 @llvm.ctlz.i16(i16 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5536 declare i32 @llvm.ctlz.i32(i32 &lt;src&gt;)
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5537 declare i64 @llvm.ctlz.i64(i64 &lt;src&gt;)
5538 declare i256 @llvm.ctlz.i256(i256 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5539</pre>
5540
5541<h5>Overview:</h5>
5542
5543<p>
5544The '<tt>llvm.ctlz</tt>' family of intrinsic functions counts the number of
5545leading zeros in a variable.
5546</p>
5547
5548<h5>Arguments:</h5>
5549
5550<p>
5551The only argument is the value to be counted. The argument may be of any
5552integer type. The return type must match the argument type.
5553</p>
5554
5555<h5>Semantics:</h5>
5556
5557<p>
5558The '<tt>llvm.ctlz</tt>' intrinsic counts the leading (most significant) zeros
5559in a variable. If the src == 0 then the result is the size in bits of the type
5560of src. For example, <tt>llvm.ctlz(i32 2) = 30</tt>.
5561</p>
5562</div>
5563
5564
5565
5566<!-- _______________________________________________________________________ -->
5567<div class="doc_subsubsection">
5568 <a name="int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic</a>
5569</div>
5570
5571<div class="doc_text">
5572
5573<h5>Syntax:</h5>
5574<p>This is an overloaded intrinsic. You can use <tt>llvm.cttz</tt> on any
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5575integer bit width. Not all targets support all bit widths however.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5576<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5577 declare i8 @llvm.cttz.i8 (i8 &lt;src&gt;)
5578 declare i16 @llvm.cttz.i16(i16 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5579 declare i32 @llvm.cttz.i32(i32 &lt;src&gt;)
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5580 declare i64 @llvm.cttz.i64(i64 &lt;src&gt;)
5581 declare i256 @llvm.cttz.i256(i256 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5582</pre>
5583
5584<h5>Overview:</h5>
5585
5586<p>
5587The '<tt>llvm.cttz</tt>' family of intrinsic functions counts the number of
5588trailing zeros.
5589</p>
5590
5591<h5>Arguments:</h5>
5592
5593<p>
5594The only argument is the value to be counted. The argument may be of any
5595integer type. The return type must match the argument type.
5596</p>
5597
5598<h5>Semantics:</h5>
5599
5600<p>
5601The '<tt>llvm.cttz</tt>' intrinsic counts the trailing (least significant) zeros
5602in a variable. If the src == 0 then the result is the size in bits of the type
5603of src. For example, <tt>llvm.cttz(2) = 1</tt>.
5604</p>
5605</div>
5606
5607<!-- _______________________________________________________________________ -->
5608<div class="doc_subsubsection">
5609 <a name="int_part_select">'<tt>llvm.part.select.*</tt>' Intrinsic</a>
5610</div>
5611
5612<div class="doc_text">
5613
5614<h5>Syntax:</h5>
5615<p>This is an overloaded intrinsic. You can use <tt>llvm.part.select</tt>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5616on any integer bit width.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5617<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5618 declare i17 @llvm.part.select.i17 (i17 %val, i32 %loBit, i32 %hiBit)
5619 declare i29 @llvm.part.select.i29 (i29 %val, i32 %loBit, i32 %hiBit)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5620</pre>
5621
5622<h5>Overview:</h5>
5623<p>The '<tt>llvm.part.select</tt>' family of intrinsic functions selects a
5624range of bits from an integer value and returns them in the same bit width as
5625the original value.</p>
5626
5627<h5>Arguments:</h5>
5628<p>The first argument, <tt>%val</tt> and the result may be integer types of
5629any bit width but they must have the same bit width. The second and third
5630arguments must be <tt>i32</tt> type since they specify only a bit index.</p>
5631
5632<h5>Semantics:</h5>
5633<p>The operation of the '<tt>llvm.part.select</tt>' intrinsic has two modes
5634of operation: forwards and reverse. If <tt>%loBit</tt> is greater than
5635<tt>%hiBits</tt> then the intrinsic operates in reverse mode. Otherwise it
5636operates in forward mode.</p>
5637<p>In forward mode, this intrinsic is the equivalent of shifting <tt>%val</tt>
5638right by <tt>%loBit</tt> bits and then ANDing it with a mask with
5639only the <tt>%hiBit - %loBit</tt> bits set, as follows:</p>
5640<ol>
5641 <li>The <tt>%val</tt> is shifted right (LSHR) by the number of bits specified
5642 by <tt>%loBits</tt>. This normalizes the value to the low order bits.</li>
5643 <li>The <tt>%loBits</tt> value is subtracted from the <tt>%hiBits</tt> value
5644 to determine the number of bits to retain.</li>
5645 <li>A mask of the retained bits is created by shifting a -1 value.</li>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5646 <li>The mask is ANDed with <tt>%val</tt> to produce the result.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5647</ol>
5648<p>In reverse mode, a similar computation is made except that the bits are
5649returned in the reverse order. So, for example, if <tt>X</tt> has the value
5650<tt>i16 0x0ACF (101011001111)</tt> and we apply
5651<tt>part.select(i16 X, 8, 3)</tt> to it, we get back the value
5652<tt>i16 0x0026 (000000100110)</tt>.</p>
5653</div>
5654
5655<div class="doc_subsubsection">
5656 <a name="int_part_set">'<tt>llvm.part.set.*</tt>' Intrinsic</a>
5657</div>
5658
5659<div class="doc_text">
5660
5661<h5>Syntax:</h5>
5662<p>This is an overloaded intrinsic. You can use <tt>llvm.part.set</tt>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5663on any integer bit width.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5664<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5665 declare i17 @llvm.part.set.i17.i9 (i17 %val, i9 %repl, i32 %lo, i32 %hi)
5666 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]5667</pre>
5668
5669<h5>Overview:</h5>
5670<p>The '<tt>llvm.part.set</tt>' family of intrinsic functions replaces a range
5671of bits in an integer value with another integer value. It returns the integer
5672with the replaced bits.</p>
5673
5674<h5>Arguments:</h5>
5675<p>The first argument, <tt>%val</tt> and the result may be integer types of
5676any bit width but they must have the same bit width. <tt>%val</tt> is the value
5677whose bits will be replaced. The second argument, <tt>%repl</tt> may be an
5678integer of any bit width. The third and fourth arguments must be <tt>i32</tt>
5679type since they specify only a bit index.</p>
5680
5681<h5>Semantics:</h5>
5682<p>The operation of the '<tt>llvm.part.set</tt>' intrinsic has two modes
5683of operation: forwards and reverse. If <tt>%lo</tt> is greater than
5684<tt>%hi</tt> then the intrinsic operates in reverse mode. Otherwise it
5685operates in forward mode.</p>
5686<p>For both modes, the <tt>%repl</tt> value is prepared for use by either
5687truncating it down to the size of the replacement area or zero extending it
5688up to that size.</p>
5689<p>In forward mode, the bits between <tt>%lo</tt> and <tt>%hi</tt> (inclusive)
5690are replaced with corresponding bits from <tt>%repl</tt>. That is the 0th bit
5691in <tt>%repl</tt> replaces the <tt>%lo</tt>th bit in <tt>%val</tt> and etc. up
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5692to the <tt>%hi</tt>th bit.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5693<p>In reverse mode, a similar computation is made except that the bits are
5694reversed. That is, the <tt>0</tt>th bit in <tt>%repl</tt> replaces the
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5695<tt>%hi</tt> bit in <tt>%val</tt> and etc. down to the <tt>%lo</tt>th bit.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5696<h5>Examples:</h5>
5697<pre>
5698 llvm.part.set(0xFFFF, 0, 4, 7) -&gt; 0xFF0F
5699 llvm.part.set(0xFFFF, 0, 7, 4) -&gt; 0xFF0F
5700 llvm.part.set(0xFFFF, 1, 7, 4) -&gt; 0xFF8F
5701 llvm.part.set(0xFFFF, F, 8, 3) -&gt; 0xFFE7
5702 llvm.part.set(0xFFFF, 0, 3, 8) -&gt; 0xFE07
5703</pre>
5704</div>
5705
5706<!-- ======================================================================= -->
5707<div class="doc_subsection">
5708 <a name="int_debugger">Debugger Intrinsics</a>
5709</div>
5710
5711<div class="doc_text">
5712<p>
5713The LLVM debugger intrinsics (which all start with <tt>llvm.dbg.</tt> prefix),
5714are described in the <a
5715href="SourceLevelDebugging.html#format_common_intrinsics">LLVM Source Level
5716Debugging</a> document.
5717</p>
5718</div>
5719
5720
5721<!-- ======================================================================= -->
5722<div class="doc_subsection">
5723 <a name="int_eh">Exception Handling Intrinsics</a>
5724</div>
5725
5726<div class="doc_text">
5727<p> The LLVM exception handling intrinsics (which all start with
5728<tt>llvm.eh.</tt> prefix), are described in the <a
5729href="ExceptionHandling.html#format_common_intrinsics">LLVM Exception
5730Handling</a> document. </p>
5731</div>
5732
5733<!-- ======================================================================= -->
5734<div class="doc_subsection">
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5735 <a name="int_trampoline">Trampoline Intrinsic</a>
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5736</div>
5737
5738<div class="doc_text">
5739<p>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5740 This intrinsic makes it possible to excise one parameter, marked with
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5741 the <tt>nest</tt> attribute, from a function. The result is a callable
5742 function pointer lacking the nest parameter - the caller does not need
5743 to provide a value for it. Instead, the value to use is stored in
5744 advance in a "trampoline", a block of memory usually allocated
5745 on the stack, which also contains code to splice the nest value into the
5746 argument list. This is used to implement the GCC nested function address
5747 extension.
5748</p>
5749<p>
5750 For example, if the function is
5751 <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]5752 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]5753<pre>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5754 %tramp = alloca [10 x i8], align 4 ; size and alignment only correct for X86
5755 %tramp1 = getelementptr [10 x i8]* %tramp, i32 0, i32 0
5756 %p = call i8* @llvm.init.trampoline( i8* %tramp1, i8* bitcast (i32 (i8* nest , i32, i32)* @f to i8*), i8* %nval )
5757 %fp = bitcast i8* %p to i32 (i32, i32)*
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5758</pre>
Bill Wendlinge262b4c2007-09-22 09:23:55 +0000[diff] [blame]5759 <p>The call <tt>%val = call i32 %fp( i32 %x, i32 %y )</tt> is then equivalent
5760 to <tt>%val = call i32 %f( i8* %nval, i32 %x, i32 %y )</tt>.</p>
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5761</div>
5762
5763<!-- _______________________________________________________________________ -->
5764<div class="doc_subsubsection">
5765 <a name="int_it">'<tt>llvm.init.trampoline</tt>' Intrinsic</a>
5766</div>
5767<div class="doc_text">
5768<h5>Syntax:</h5>
5769<pre>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5770declare 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]5771</pre>
5772<h5>Overview:</h5>
5773<p>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5774 This fills the memory pointed to by <tt>tramp</tt> with code
5775 and returns a function pointer suitable for executing it.
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5776</p>
5777<h5>Arguments:</h5>
5778<p>
5779 The <tt>llvm.init.trampoline</tt> intrinsic takes three arguments, all
5780 pointers. The <tt>tramp</tt> argument must point to a sufficiently large
5781 and sufficiently aligned block of memory; this memory is written to by the
Duncan Sands35012212007-08-22 23:39:54 +0000[diff] [blame]5782 intrinsic. Note that the size and the alignment are target-specific - LLVM
5783 currently provides no portable way of determining them, so a front-end that
5784 generates this intrinsic needs to have some target-specific knowledge.
5785 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]5786</p>
5787<h5>Semantics:</h5>
5788<p>
5789 The block of memory pointed to by <tt>tramp</tt> is filled with target
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5790 dependent code, turning it into a function. A pointer to this function is
5791 returned, but needs to be bitcast to an
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5792 <a href="#int_trampoline">appropriate function pointer type</a>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5793 before being called. The new function's signature is the same as that of
5794 <tt>func</tt> with any arguments marked with the <tt>nest</tt> attribute
5795 removed. At most one such <tt>nest</tt> argument is allowed, and it must be
5796 of pointer type. Calling the new function is equivalent to calling
5797 <tt>func</tt> with the same argument list, but with <tt>nval</tt> used for the
5798 missing <tt>nest</tt> argument. If, after calling
5799 <tt>llvm.init.trampoline</tt>, the memory pointed to by <tt>tramp</tt> is
5800 modified, then the effect of any later call to the returned function pointer is
5801 undefined.
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5802</p>
5803</div>
5804
5805<!-- ======================================================================= -->
5806<div class="doc_subsection">
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]5807 <a name="int_atomics">Atomic Operations and Synchronization Intrinsics</a>
5808</div>
5809
5810<div class="doc_text">
5811<p>
5812 These intrinsic functions expand the "universal IR" of LLVM to represent
5813 hardware constructs for atomic operations and memory synchronization. This
5814 provides an interface to the hardware, not an interface to the programmer. It
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]5815 is aimed at a low enough level to allow any programming models or APIs
5816 (Application Programming Interfaces) which
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]5817 need atomic behaviors to map cleanly onto it. It is also modeled primarily on
5818 hardware behavior. Just as hardware provides a "universal IR" for source
5819 languages, it also provides a starting point for developing a "universal"
5820 atomic operation and synchronization IR.
5821</p>
5822<p>
5823 These do <em>not</em> form an API such as high-level threading libraries,
5824 software transaction memory systems, atomic primitives, and intrinsic
5825 functions as found in BSD, GNU libc, atomic_ops, APR, and other system and
5826 application libraries. The hardware interface provided by LLVM should allow
5827 a clean implementation of all of these APIs and parallel programming models.
5828 No one model or paradigm should be selected above others unless the hardware
5829 itself ubiquitously does so.
5830
5831</p>
5832</div>
5833
5834<!-- _______________________________________________________________________ -->
5835<div class="doc_subsubsection">
5836 <a name="int_memory_barrier">'<tt>llvm.memory.barrier</tt>' Intrinsic</a>
5837</div>
5838<div class="doc_text">
5839<h5>Syntax:</h5>
5840<pre>
5841declare void @llvm.memory.barrier( i1 &lt;ll&gt;, i1 &lt;ls&gt;, i1 &lt;sl&gt;, i1 &lt;ss&gt;,
5842i1 &lt;device&gt; )
5843
5844</pre>
5845<h5>Overview:</h5>
5846<p>
5847 The <tt>llvm.memory.barrier</tt> intrinsic guarantees ordering between
5848 specific pairs of memory access types.
5849</p>
5850<h5>Arguments:</h5>
5851<p>
5852 The <tt>llvm.memory.barrier</tt> intrinsic requires five boolean arguments.
5853 The first four arguments enables a specific barrier as listed below. The fith
5854 argument specifies that the barrier applies to io or device or uncached memory.
5855
5856</p>
5857 <ul>
5858 <li><tt>ll</tt>: load-load barrier</li>
5859 <li><tt>ls</tt>: load-store barrier</li>
5860 <li><tt>sl</tt>: store-load barrier</li>
5861 <li><tt>ss</tt>: store-store barrier</li>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5862 <li><tt>device</tt>: barrier applies to device and uncached memory also.</li>
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]5863 </ul>
5864<h5>Semantics:</h5>
5865<p>
5866 This intrinsic causes the system to enforce some ordering constraints upon
5867 the loads and stores of the program. This barrier does not indicate
5868 <em>when</em> any events will occur, it only enforces an <em>order</em> in
5869 which they occur. For any of the specified pairs of load and store operations
5870 (f.ex. load-load, or store-load), all of the first operations preceding the
5871 barrier will complete before any of the second operations succeeding the
5872 barrier begin. Specifically the semantics for each pairing is as follows:
5873</p>
5874 <ul>
5875 <li><tt>ll</tt>: All loads before the barrier must complete before any load
5876 after the barrier begins.</li>
5877
5878 <li><tt>ls</tt>: All loads before the barrier must complete before any
5879 store after the barrier begins.</li>
5880 <li><tt>ss</tt>: All stores before the barrier must complete before any
5881 store after the barrier begins.</li>
5882 <li><tt>sl</tt>: All stores before the barrier must complete before any
5883 load after the barrier begins.</li>
5884 </ul>
5885<p>
5886 These semantics are applied with a logical "and" behavior when more than one
5887 is enabled in a single memory barrier intrinsic.
5888</p>
5889<p>
5890 Backends may implement stronger barriers than those requested when they do not
5891 support as fine grained a barrier as requested. Some architectures do not
5892 need all types of barriers and on such architectures, these become noops.
5893</p>
5894<h5>Example:</h5>
5895<pre>
5896%ptr = malloc i32
5897 store i32 4, %ptr
5898
5899%result1 = load i32* %ptr <i>; yields {i32}:result1 = 4</i>
5900 call void @llvm.memory.barrier( i1 false, i1 true, i1 false, i1 false )
5901 <i>; guarantee the above finishes</i>
5902 store i32 8, %ptr <i>; before this begins</i>
5903</pre>
5904</div>
5905
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]5906<!-- _______________________________________________________________________ -->
5907<div class="doc_subsubsection">
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]5908 <a name="int_atomic_cmp_swap">'<tt>llvm.atomic.cmp.swap.*</tt>' Intrinsic</a>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]5909</div>
5910<div class="doc_text">
5911<h5>Syntax:</h5>
5912<p>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]5913 This is an overloaded intrinsic. You can use <tt>llvm.atomic.cmp.swap</tt> on
5914 any integer bit width and for different address spaces. Not all targets
5915 support all bit widths however.</p>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]5916
5917<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]5918declare i8 @llvm.atomic.cmp.swap.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;cmp&gt;, i8 &lt;val&gt; )
5919declare i16 @llvm.atomic.cmp.swap.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;cmp&gt;, i16 &lt;val&gt; )
5920declare i32 @llvm.atomic.cmp.swap.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;cmp&gt;, i32 &lt;val&gt; )
5921declare 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]5922
5923</pre>
5924<h5>Overview:</h5>
5925<p>
5926 This loads a value in memory and compares it to a given value. If they are
5927 equal, it stores a new value into the memory.
5928</p>
5929<h5>Arguments:</h5>
5930<p>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]5931 The <tt>llvm.atomic.cmp.swap</tt> intrinsic takes three arguments. The result as
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]5932 well as both <tt>cmp</tt> and <tt>val</tt> must be integer values with the
5933 same bit width. The <tt>ptr</tt> argument must be a pointer to a value of
5934 this integer type. While any bit width integer may be used, targets may only
5935 lower representations they support in hardware.
5936
5937</p>
5938<h5>Semantics:</h5>
5939<p>
5940 This entire intrinsic must be executed atomically. It first loads the value
5941 in memory pointed to by <tt>ptr</tt> and compares it with the value
5942 <tt>cmp</tt>. If they are equal, <tt>val</tt> is stored into the memory. The
5943 loaded value is yielded in all cases. This provides the equivalent of an
5944 atomic compare-and-swap operation within the SSA framework.
5945</p>
5946<h5>Examples:</h5>
5947
5948<pre>
5949%ptr = malloc i32
5950 store i32 4, %ptr
5951
5952%val1 = add i32 4, 4
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]5953%result1 = call i32 @llvm.atomic.cmp.swap.i32.p0i32( i32* %ptr, i32 4, %val1 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]5954 <i>; yields {i32}:result1 = 4</i>
5955%stored1 = icmp eq i32 %result1, 4 <i>; yields {i1}:stored1 = true</i>
5956%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 8</i>
5957
5958%val2 = add i32 1, 1
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]5959%result2 = call i32 @llvm.atomic.cmp.swap.i32.p0i32( i32* %ptr, i32 5, %val2 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]5960 <i>; yields {i32}:result2 = 8</i>
5961%stored2 = icmp eq i32 %result2, 5 <i>; yields {i1}:stored2 = false</i>
5962
5963%memval2 = load i32* %ptr <i>; yields {i32}:memval2 = 8</i>
5964</pre>
5965</div>
5966
5967<!-- _______________________________________________________________________ -->
5968<div class="doc_subsubsection">
5969 <a name="int_atomic_swap">'<tt>llvm.atomic.swap.*</tt>' Intrinsic</a>
5970</div>
5971<div class="doc_text">
5972<h5>Syntax:</h5>
5973
5974<p>
5975 This is an overloaded intrinsic. You can use <tt>llvm.atomic.swap</tt> on any
5976 integer bit width. Not all targets support all bit widths however.</p>
5977<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]5978declare i8 @llvm.atomic.swap.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;val&gt; )
5979declare i16 @llvm.atomic.swap.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;val&gt; )
5980declare i32 @llvm.atomic.swap.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;val&gt; )
5981declare i64 @llvm.atomic.swap.i64.p0i64( i64* &lt;ptr&gt;, i64 &lt;val&gt; )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]5982
5983</pre>
5984<h5>Overview:</h5>
5985<p>
5986 This intrinsic loads the value stored in memory at <tt>ptr</tt> and yields
5987 the value from memory. It then stores the value in <tt>val</tt> in the memory
5988 at <tt>ptr</tt>.
5989</p>
5990<h5>Arguments:</h5>
5991
5992<p>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]5993 The <tt>llvm.atomic.swap</tt> intrinsic takes two arguments. Both the
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]5994 <tt>val</tt> argument and the result must be integers of the same bit width.
5995 The first argument, <tt>ptr</tt>, must be a pointer to a value of this
5996 integer type. The targets may only lower integer representations they
5997 support.
5998</p>
5999<h5>Semantics:</h5>
6000<p>
6001 This intrinsic loads the value pointed to by <tt>ptr</tt>, yields it, and
6002 stores <tt>val</tt> back into <tt>ptr</tt> atomically. This provides the
6003 equivalent of an atomic swap operation within the SSA framework.
6004
6005</p>
6006<h5>Examples:</h5>
6007<pre>
6008%ptr = malloc i32
6009 store i32 4, %ptr
6010
6011%val1 = add i32 4, 4
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6012%result1 = call i32 @llvm.atomic.swap.i32.p0i32( i32* %ptr, i32 %val1 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6013 <i>; yields {i32}:result1 = 4</i>
6014%stored1 = icmp eq i32 %result1, 4 <i>; yields {i1}:stored1 = true</i>
6015%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 8</i>
6016
6017%val2 = add i32 1, 1
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6018%result2 = call i32 @llvm.atomic.swap.i32.p0i32( i32* %ptr, i32 %val2 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6019 <i>; yields {i32}:result2 = 8</i>
6020
6021%stored2 = icmp eq i32 %result2, 8 <i>; yields {i1}:stored2 = true</i>
6022%memval2 = load i32* %ptr <i>; yields {i32}:memval2 = 2</i>
6023</pre>
6024</div>
6025
6026<!-- _______________________________________________________________________ -->
6027<div class="doc_subsubsection">
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6028 <a name="int_atomic_load_add">'<tt>llvm.atomic.load.add.*</tt>' Intrinsic</a>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6029
6030</div>
6031<div class="doc_text">
6032<h5>Syntax:</h5>
6033<p>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6034 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]6035 integer bit width. Not all targets support all bit widths however.</p>
6036<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6037declare i8 @llvm.atomic.load.add.i8..p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6038declare i16 @llvm.atomic.load.add.i16..p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6039declare i32 @llvm.atomic.load.add.i32..p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6040declare 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]6041
6042</pre>
6043<h5>Overview:</h5>
6044<p>
6045 This intrinsic adds <tt>delta</tt> to the value stored in memory at
6046 <tt>ptr</tt>. It yields the original value at <tt>ptr</tt>.
6047</p>
6048<h5>Arguments:</h5>
6049<p>
6050
6051 The intrinsic takes two arguments, the first a pointer to an integer value
6052 and the second an integer value. The result is also an integer value. These
6053 integer types can have any bit width, but they must all have the same bit
6054 width. The targets may only lower integer representations they support.
6055</p>
6056<h5>Semantics:</h5>
6057<p>
6058 This intrinsic does a series of operations atomically. It first loads the
6059 value stored at <tt>ptr</tt>. It then adds <tt>delta</tt>, stores the result
6060 to <tt>ptr</tt>. It yields the original value stored at <tt>ptr</tt>.
6061</p>
6062
6063<h5>Examples:</h5>
6064<pre>
6065%ptr = malloc i32
6066 store i32 4, %ptr
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6067%result1 = call i32 @llvm.atomic.load.add.i32.p0i32( i32* %ptr, i32 4 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6068 <i>; yields {i32}:result1 = 4</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6069%result2 = call i32 @llvm.atomic.load.add.i32.p0i32( i32* %ptr, i32 2 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6070 <i>; yields {i32}:result2 = 8</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6071%result3 = call i32 @llvm.atomic.load.add.i32.p0i32( i32* %ptr, i32 5 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6072 <i>; yields {i32}:result3 = 10</i>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6073%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 15</i>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6074</pre>
6075</div>
6076
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6077<!-- _______________________________________________________________________ -->
6078<div class="doc_subsubsection">
6079 <a name="int_atomic_load_sub">'<tt>llvm.atomic.load.sub.*</tt>' Intrinsic</a>
6080
6081</div>
6082<div class="doc_text">
6083<h5>Syntax:</h5>
6084<p>
6085 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]6086 any integer bit width and for different address spaces. Not all targets
6087 support all bit widths however.</p>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6088<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6089declare i8 @llvm.atomic.load.sub.i8.p0i32( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6090declare i16 @llvm.atomic.load.sub.i16.p0i32( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6091declare i32 @llvm.atomic.load.sub.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6092declare 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]6093
6094</pre>
6095<h5>Overview:</h5>
6096<p>
6097 This intrinsic subtracts <tt>delta</tt> to the value stored in memory at
6098 <tt>ptr</tt>. It yields the original value at <tt>ptr</tt>.
6099</p>
6100<h5>Arguments:</h5>
6101<p>
6102
6103 The intrinsic takes two arguments, the first a pointer to an integer value
6104 and the second an integer value. The result is also an integer value. These
6105 integer types can have any bit width, but they must all have the same bit
6106 width. The targets may only lower integer representations they support.
6107</p>
6108<h5>Semantics:</h5>
6109<p>
6110 This intrinsic does a series of operations atomically. It first loads the
6111 value stored at <tt>ptr</tt>. It then subtracts <tt>delta</tt>, stores the
6112 result to <tt>ptr</tt>. It yields the original value stored at <tt>ptr</tt>.
6113</p>
6114
6115<h5>Examples:</h5>
6116<pre>
6117%ptr = malloc i32
6118 store i32 8, %ptr
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6119%result1 = call i32 @llvm.atomic.load.sub.i32.p0i32( i32* %ptr, i32 4 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6120 <i>; yields {i32}:result1 = 8</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6121%result2 = call i32 @llvm.atomic.load.sub.i32.p0i32( i32* %ptr, i32 2 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6122 <i>; yields {i32}:result2 = 4</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6123%result3 = call i32 @llvm.atomic.load.sub.i32.p0i32( i32* %ptr, i32 5 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6124 <i>; yields {i32}:result3 = 2</i>
6125%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = -3</i>
6126</pre>
6127</div>
6128
6129<!-- _______________________________________________________________________ -->
6130<div class="doc_subsubsection">
6131 <a name="int_atomic_load_and">'<tt>llvm.atomic.load.and.*</tt>' Intrinsic</a><br>
6132 <a name="int_atomic_load_nand">'<tt>llvm.atomic.load.nand.*</tt>' Intrinsic</a><br>
6133 <a name="int_atomic_load_or">'<tt>llvm.atomic.load.or.*</tt>' Intrinsic</a><br>
6134 <a name="int_atomic_load_xor">'<tt>llvm.atomic.load.xor.*</tt>' Intrinsic</a><br>
6135
6136</div>
6137<div class="doc_text">
6138<h5>Syntax:</h5>
6139<p>
6140 These are overloaded intrinsics. You can use <tt>llvm.atomic.load_and</tt>,
6141 <tt>llvm.atomic.load_nand</tt>, <tt>llvm.atomic.load_or</tt>, and
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6142 <tt>llvm.atomic.load_xor</tt> on any integer bit width and for different
6143 address spaces. Not all targets support all bit widths however.</p>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6144<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6145declare i8 @llvm.atomic.load.and.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6146declare i16 @llvm.atomic.load.and.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6147declare i32 @llvm.atomic.load.and.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6148declare 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]6149
6150</pre>
6151
6152<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6153declare i8 @llvm.atomic.load.or.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6154declare i16 @llvm.atomic.load.or.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6155declare i32 @llvm.atomic.load.or.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6156declare 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]6157
6158</pre>
6159
6160<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6161declare i8 @llvm.atomic.load.nand.i8.p0i32( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6162declare i16 @llvm.atomic.load.nand.i16.p0i32( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6163declare i32 @llvm.atomic.load.nand.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6164declare 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]6165
6166</pre>
6167
6168<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6169declare i8 @llvm.atomic.load.xor.i8.p0i32( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6170declare i16 @llvm.atomic.load.xor.i16.p0i32( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6171declare i32 @llvm.atomic.load.xor.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6172declare 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]6173
6174</pre>
6175<h5>Overview:</h5>
6176<p>
6177 These intrinsics bitwise the operation (and, nand, or, xor) <tt>delta</tt> to
6178 the value stored in memory at <tt>ptr</tt>. It yields the original value
6179 at <tt>ptr</tt>.
6180</p>
6181<h5>Arguments:</h5>
6182<p>
6183
6184 These intrinsics take two arguments, the first a pointer to an integer value
6185 and the second an integer value. The result is also an integer value. These
6186 integer types can have any bit width, but they must all have the same bit
6187 width. The targets may only lower integer representations they support.
6188</p>
6189<h5>Semantics:</h5>
6190<p>
6191 These intrinsics does a series of operations atomically. They first load the
6192 value stored at <tt>ptr</tt>. They then do the bitwise operation
6193 <tt>delta</tt>, store the result to <tt>ptr</tt>. They yield the original
6194 value stored at <tt>ptr</tt>.
6195</p>
6196
6197<h5>Examples:</h5>
6198<pre>
6199%ptr = malloc i32
6200 store i32 0x0F0F, %ptr
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6201%result0 = call i32 @llvm.atomic.load.nand.i32.p0i32( i32* %ptr, i32 0xFF )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6202 <i>; yields {i32}:result0 = 0x0F0F</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6203%result1 = call i32 @llvm.atomic.load.and.i32.p0i32( i32* %ptr, i32 0xFF )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6204 <i>; yields {i32}:result1 = 0xFFFFFFF0</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6205%result2 = call i32 @llvm.atomic.load.or.i32.p0i32( i32* %ptr, i32 0F )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6206 <i>; yields {i32}:result2 = 0xF0</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6207%result3 = call i32 @llvm.atomic.load.xor.i32.p0i32( i32* %ptr, i32 0F )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6208 <i>; yields {i32}:result3 = FF</i>
6209%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = F0</i>
6210</pre>
6211</div>
6212
6213
6214<!-- _______________________________________________________________________ -->
6215<div class="doc_subsubsection">
6216 <a name="int_atomic_load_max">'<tt>llvm.atomic.load.max.*</tt>' Intrinsic</a><br>
6217 <a name="int_atomic_load_min">'<tt>llvm.atomic.load.min.*</tt>' Intrinsic</a><br>
6218 <a name="int_atomic_load_umax">'<tt>llvm.atomic.load.umax.*</tt>' Intrinsic</a><br>
6219 <a name="int_atomic_load_umin">'<tt>llvm.atomic.load.umin.*</tt>' Intrinsic</a><br>
6220
6221</div>
6222<div class="doc_text">
6223<h5>Syntax:</h5>
6224<p>
6225 These are overloaded intrinsics. You can use <tt>llvm.atomic.load_max</tt>,
6226 <tt>llvm.atomic.load_min</tt>, <tt>llvm.atomic.load_umax</tt>, and
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6227 <tt>llvm.atomic.load_umin</tt> on any integer bit width and for different
6228 address spaces. Not all targets
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6229 support all bit widths however.</p>
6230<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6231declare i8 @llvm.atomic.load.max.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6232declare i16 @llvm.atomic.load.max.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6233declare i32 @llvm.atomic.load.max.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6234declare 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]6235
6236</pre>
6237
6238<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6239declare i8 @llvm.atomic.load.min.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6240declare i16 @llvm.atomic.load.min.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6241declare i32 @llvm.atomic.load.min.i32..p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6242declare 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]6243
6244</pre>
6245
6246<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6247declare i8 @llvm.atomic.load.umax.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6248declare i16 @llvm.atomic.load.umax.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6249declare i32 @llvm.atomic.load.umax.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6250declare 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]6251
6252</pre>
6253
6254<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6255declare i8 @llvm.atomic.load.umin.i8..p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6256declare i16 @llvm.atomic.load.umin.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6257declare i32 @llvm.atomic.load.umin.i32..p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6258declare 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]6259
6260</pre>
6261<h5>Overview:</h5>
6262<p>
6263 These intrinsics takes the signed or unsigned minimum or maximum of
6264 <tt>delta</tt> and the value stored in memory at <tt>ptr</tt>. It yields the
6265 original value at <tt>ptr</tt>.
6266</p>
6267<h5>Arguments:</h5>
6268<p>
6269
6270 These intrinsics take two arguments, the first a pointer to an integer value
6271 and the second an integer value. The result is also an integer value. These
6272 integer types can have any bit width, but they must all have the same bit
6273 width. The targets may only lower integer representations they support.
6274</p>
6275<h5>Semantics:</h5>
6276<p>
6277 These intrinsics does a series of operations atomically. They first load the
6278 value stored at <tt>ptr</tt>. They then do the signed or unsigned min or max
6279 <tt>delta</tt> and the value, store the result to <tt>ptr</tt>. They yield
6280 the original value stored at <tt>ptr</tt>.
6281</p>
6282
6283<h5>Examples:</h5>
6284<pre>
6285%ptr = malloc i32
6286 store i32 7, %ptr
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6287%result0 = call i32 @llvm.atomic.load.min.i32.p0i32( i32* %ptr, i32 -2 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6288 <i>; yields {i32}:result0 = 7</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6289%result1 = call i32 @llvm.atomic.load.max.i32.p0i32( i32* %ptr, i32 8 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6290 <i>; yields {i32}:result1 = -2</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6291%result2 = call i32 @llvm.atomic.load.umin.i32.p0i32( i32* %ptr, i32 10 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6292 <i>; yields {i32}:result2 = 8</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6293%result3 = call i32 @llvm.atomic.load.umax.i32.p0i32( i32* %ptr, i32 30 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6294 <i>; yields {i32}:result3 = 8</i>
6295%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 30</i>
6296</pre>
6297</div>
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]6298
6299<!-- ======================================================================= -->
6300<div class="doc_subsection">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6301 <a name="int_general">General Intrinsics</a>
6302</div>
6303
6304<div class="doc_text">
6305<p> This class of intrinsics is designed to be generic and has
6306no specific purpose. </p>
6307</div>
6308
6309<!-- _______________________________________________________________________ -->
6310<div class="doc_subsubsection">
6311 <a name="int_var_annotation">'<tt>llvm.var.annotation</tt>' Intrinsic</a>
6312</div>
6313
6314<div class="doc_text">
6315
6316<h5>Syntax:</h5>
6317<pre>
6318 declare void @llvm.var.annotation(i8* &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
6319</pre>
6320
6321<h5>Overview:</h5>
6322
6323<p>
6324The '<tt>llvm.var.annotation</tt>' intrinsic
6325</p>
6326
6327<h5>Arguments:</h5>
6328
6329<p>
6330The first argument is a pointer to a value, the second is a pointer to a
6331global string, the third is a pointer to a global string which is the source
6332file name, and the last argument is the line number.
6333</p>
6334
6335<h5>Semantics:</h5>
6336
6337<p>
Anton Korobeynikove6e764f2008-01-15 22:31:34 +0000[diff] [blame]6338This intrinsic allows annotation of local variables with arbitrary strings.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6339This can be useful for special purpose optimizations that want to look for these
Anton Korobeynikove6e764f2008-01-15 22:31:34 +0000[diff] [blame]6340annotations. These have no other defined use, they are ignored by code
6341generation and optimization.
6342</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6343</div>
6344
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]6345<!-- _______________________________________________________________________ -->
6346<div class="doc_subsubsection">
Tanya Lattnerc9869b12007-09-21 23:57:59 +0000[diff] [blame]6347 <a name="int_annotation">'<tt>llvm.annotation.*</tt>' Intrinsic</a>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]6348</div>
6349
6350<div class="doc_text">
6351
6352<h5>Syntax:</h5>
Tanya Lattnere545be72007-09-21 23:56:27 +0000[diff] [blame]6353<p>This is an overloaded intrinsic. You can use '<tt>llvm.annotation</tt>' on
6354any integer bit width.
6355</p>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]6356<pre>
Tanya Lattner09161fe2007-09-22 00:03:01 +0000[diff] [blame]6357 declare i8 @llvm.annotation.i8(i8 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
6358 declare i16 @llvm.annotation.i16(i16 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
6359 declare i32 @llvm.annotation.i32(i32 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
6360 declare i64 @llvm.annotation.i64(i64 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
6361 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]6362</pre>
6363
6364<h5>Overview:</h5>
Tanya Lattnere545be72007-09-21 23:56:27 +0000[diff] [blame]6365
6366<p>
6367The '<tt>llvm.annotation</tt>' intrinsic.
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]6368</p>
6369
6370<h5>Arguments:</h5>
6371
6372<p>
6373The first argument is an integer value (result of some expression),
6374the second is a pointer to a global string, the third is a pointer to a global
6375string which is the source file name, and the last argument is the line number.
Tanya Lattnere545be72007-09-21 23:56:27 +0000[diff] [blame]6376It returns the value of the first argument.
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]6377</p>
6378
6379<h5>Semantics:</h5>
6380
6381<p>
6382This intrinsic allows annotations to be put on arbitrary expressions
6383with arbitrary strings. This can be useful for special purpose optimizations
6384that want to look for these annotations. These have no other defined use, they
6385are ignored by code generation and optimization.
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]6386</p>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]6387</div>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6388
Anton Korobeynikove6e764f2008-01-15 22:31:34 +0000[diff] [blame]6389<!-- _______________________________________________________________________ -->
6390<div class="doc_subsubsection">
6391 <a name="int_trap">'<tt>llvm.trap</tt>' Intrinsic</a>
6392</div>
6393
6394<div class="doc_text">
6395
6396<h5>Syntax:</h5>
6397<pre>
6398 declare void @llvm.trap()
6399</pre>
6400
6401<h5>Overview:</h5>
6402
6403<p>
6404The '<tt>llvm.trap</tt>' intrinsic
6405</p>
6406
6407<h5>Arguments:</h5>
6408
6409<p>
6410None
6411</p>
6412
6413<h5>Semantics:</h5>
6414
6415<p>
6416This intrinsics is lowered to the target dependent trap instruction. If the
6417target does not have a trap instruction, this intrinsic will be lowered to the
6418call of the abort() function.
6419</p>
6420</div>
6421
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6422<!-- *********************************************************************** -->
6423<hr>
6424<address>
6425 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
6426 src="http://jigsaw.w3.org/css-validator/images/vcss" alt="Valid CSS!"></a>
6427 <a href="http://validator.w3.org/check/referer"><img
Chris Lattner08497ce2008-01-04 04:33:49 +0000[diff] [blame]6428 src="http://www.w3.org/Icons/valid-html401" alt="Valid HTML 4.01!"></a>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6429
6430 <a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
6431 <a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br>
6432 Last modified: $Date$
6433</address>
Chris Lattner08497ce2008-01-04 04:33:49 +0000[diff] [blame]6434
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6435</body>
6436</html>