Gitiles
Code Review Sign In
gerrit-public.fairphone.software / platform / external / llvm / e39c5bf00cfa04139ca6b40642c4d3568c3c929b / . / docs / LangRef.html
blob: afadd2169e82b3b3032b7f05bbeaffa586cb949d [file] [log] [blame]
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN"
2 "http://www.w3.org/TR/html4/strict.dtd">
3<html>
4<head>
5 <title>LLVM Assembly Language Reference Manual</title>
6 <meta http-equiv="Content-Type" content="text/html; charset=utf-8">
7 <meta name="author" content="Chris Lattner">
8 <meta name="description"
9 content="LLVM Assembly Language Reference Manual.">
10 <link rel="stylesheet" href="llvm.css" type="text/css">
11</head>
12
13<body>
14
15<div class="doc_title"> LLVM Language Reference Manual </div>
16<ol>
17 <li><a href="#abstract">Abstract</a></li>
18 <li><a href="#introduction">Introduction</a></li>
19 <li><a href="#identifiers">Identifiers</a></li>
20 <li><a href="#highlevel">High Level Structure</a>
21 <ol>
22 <li><a href="#modulestructure">Module Structure</a></li>
23 <li><a href="#linkage">Linkage Types</a></li>
24 <li><a href="#callingconv">Calling Conventions</a></li>
Chris Lattner5b6dc6e2009-01-11 20:53:49 +0000[diff] [blame]25 <li><a href="#namedtypes">Named Types</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]26 <li><a href="#globalvars">Global Variables</a></li>
27 <li><a href="#functionstructure">Functions</a></li>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]28 <li><a href="#aliasstructure">Aliases</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]29 <li><a href="#paramattrs">Parameter Attributes</a></li>
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]30 <li><a href="#fnattrs">Function Attributes</a></li>
Gordon Henriksen13fe5e32007-12-10 03:18:06 +0000[diff] [blame]31 <li><a href="#gc">Garbage Collector Names</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]32 <li><a href="#moduleasm">Module-Level Inline Assembly</a></li>
33 <li><a href="#datalayout">Data Layout</a></li>
34 </ol>
35 </li>
36 <li><a href="#typesystem">Type System</a>
37 <ol>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]38 <li><a href="#t_classifications">Type Classifications</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]39 <li><a href="#t_primitive">Primitive Types</a>
40 <ol>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]41 <li><a href="#t_floating">Floating Point Types</a></li>
42 <li><a href="#t_void">Void Type</a></li>
43 <li><a href="#t_label">Label Type</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]44 </ol>
45 </li>
46 <li><a href="#t_derived">Derived Types</a>
47 <ol>
Chris Lattner251ab812007-12-18 06:18:21 +0000[diff] [blame]48 <li><a href="#t_integer">Integer Type</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]49 <li><a href="#t_array">Array Type</a></li>
50 <li><a href="#t_function">Function Type</a></li>
51 <li><a href="#t_pointer">Pointer Type</a></li>
52 <li><a href="#t_struct">Structure Type</a></li>
53 <li><a href="#t_pstruct">Packed Structure Type</a></li>
54 <li><a href="#t_vector">Vector Type</a></li>
55 <li><a href="#t_opaque">Opaque Type</a></li>
56 </ol>
57 </li>
Chris Lattner515195a2009-02-02 07:32:36 +0000[diff] [blame]58 <li><a href="#t_uprefs">Type Up-references</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]59 </ol>
60 </li>
61 <li><a href="#constants">Constants</a>
62 <ol>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]63 <li><a href="#simpleconstants">Simple Constants</a></li>
64 <li><a href="#aggregateconstants">Aggregate Constants</a></li>
65 <li><a href="#globalconstants">Global Variable and Function Addresses</a></li>
66 <li><a href="#undefvalues">Undefined Values</a></li>
67 <li><a href="#constantexprs">Constant Expressions</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]68 </ol>
69 </li>
70 <li><a href="#othervalues">Other Values</a>
71 <ol>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]72 <li><a href="#inlineasm">Inline Assembler Expressions</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]73 </ol>
74 </li>
75 <li><a href="#instref">Instruction Reference</a>
76 <ol>
77 <li><a href="#terminators">Terminator Instructions</a>
78 <ol>
79 <li><a href="#i_ret">'<tt>ret</tt>' Instruction</a></li>
80 <li><a href="#i_br">'<tt>br</tt>' Instruction</a></li>
81 <li><a href="#i_switch">'<tt>switch</tt>' Instruction</a></li>
82 <li><a href="#i_invoke">'<tt>invoke</tt>' Instruction</a></li>
83 <li><a href="#i_unwind">'<tt>unwind</tt>' Instruction</a></li>
84 <li><a href="#i_unreachable">'<tt>unreachable</tt>' Instruction</a></li>
85 </ol>
86 </li>
87 <li><a href="#binaryops">Binary Operations</a>
88 <ol>
89 <li><a href="#i_add">'<tt>add</tt>' Instruction</a></li>
90 <li><a href="#i_sub">'<tt>sub</tt>' Instruction</a></li>
91 <li><a href="#i_mul">'<tt>mul</tt>' Instruction</a></li>
92 <li><a href="#i_udiv">'<tt>udiv</tt>' Instruction</a></li>
93 <li><a href="#i_sdiv">'<tt>sdiv</tt>' Instruction</a></li>
94 <li><a href="#i_fdiv">'<tt>fdiv</tt>' Instruction</a></li>
95 <li><a href="#i_urem">'<tt>urem</tt>' Instruction</a></li>
96 <li><a href="#i_srem">'<tt>srem</tt>' Instruction</a></li>
97 <li><a href="#i_frem">'<tt>frem</tt>' Instruction</a></li>
98 </ol>
99 </li>
100 <li><a href="#bitwiseops">Bitwise Binary Operations</a>
101 <ol>
102 <li><a href="#i_shl">'<tt>shl</tt>' Instruction</a></li>
103 <li><a href="#i_lshr">'<tt>lshr</tt>' Instruction</a></li>
104 <li><a href="#i_ashr">'<tt>ashr</tt>' Instruction</a></li>
105 <li><a href="#i_and">'<tt>and</tt>' Instruction</a></li>
106 <li><a href="#i_or">'<tt>or</tt>' Instruction</a></li>
107 <li><a href="#i_xor">'<tt>xor</tt>' Instruction</a></li>
108 </ol>
109 </li>
110 <li><a href="#vectorops">Vector Operations</a>
111 <ol>
112 <li><a href="#i_extractelement">'<tt>extractelement</tt>' Instruction</a></li>
113 <li><a href="#i_insertelement">'<tt>insertelement</tt>' Instruction</a></li>
114 <li><a href="#i_shufflevector">'<tt>shufflevector</tt>' Instruction</a></li>
115 </ol>
116 </li>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]117 <li><a href="#aggregateops">Aggregate Operations</a>
118 <ol>
119 <li><a href="#i_extractvalue">'<tt>extractvalue</tt>' Instruction</a></li>
120 <li><a href="#i_insertvalue">'<tt>insertvalue</tt>' Instruction</a></li>
121 </ol>
122 </li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]123 <li><a href="#memoryops">Memory Access and Addressing Operations</a>
124 <ol>
125 <li><a href="#i_malloc">'<tt>malloc</tt>' Instruction</a></li>
126 <li><a href="#i_free">'<tt>free</tt>' Instruction</a></li>
127 <li><a href="#i_alloca">'<tt>alloca</tt>' Instruction</a></li>
128 <li><a href="#i_load">'<tt>load</tt>' Instruction</a></li>
129 <li><a href="#i_store">'<tt>store</tt>' Instruction</a></li>
130 <li><a href="#i_getelementptr">'<tt>getelementptr</tt>' Instruction</a></li>
131 </ol>
132 </li>
133 <li><a href="#convertops">Conversion Operations</a>
134 <ol>
135 <li><a href="#i_trunc">'<tt>trunc .. to</tt>' Instruction</a></li>
136 <li><a href="#i_zext">'<tt>zext .. to</tt>' Instruction</a></li>
137 <li><a href="#i_sext">'<tt>sext .. to</tt>' Instruction</a></li>
138 <li><a href="#i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a></li>
139 <li><a href="#i_fpext">'<tt>fpext .. to</tt>' Instruction</a></li>
140 <li><a href="#i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a></li>
141 <li><a href="#i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a></li>
142 <li><a href="#i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a></li>
143 <li><a href="#i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a></li>
144 <li><a href="#i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a></li>
145 <li><a href="#i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a></li>
146 <li><a href="#i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a></li>
147 </ol>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]148 </li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]149 <li><a href="#otherops">Other Operations</a>
150 <ol>
151 <li><a href="#i_icmp">'<tt>icmp</tt>' Instruction</a></li>
152 <li><a href="#i_fcmp">'<tt>fcmp</tt>' Instruction</a></li>
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]153 <li><a href="#i_vicmp">'<tt>vicmp</tt>' Instruction</a></li>
154 <li><a href="#i_vfcmp">'<tt>vfcmp</tt>' Instruction</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]155 <li><a href="#i_phi">'<tt>phi</tt>' Instruction</a></li>
156 <li><a href="#i_select">'<tt>select</tt>' Instruction</a></li>
157 <li><a href="#i_call">'<tt>call</tt>' Instruction</a></li>
158 <li><a href="#i_va_arg">'<tt>va_arg</tt>' Instruction</a></li>
159 </ol>
160 </li>
161 </ol>
162 </li>
163 <li><a href="#intrinsics">Intrinsic Functions</a>
164 <ol>
165 <li><a href="#int_varargs">Variable Argument Handling Intrinsics</a>
166 <ol>
167 <li><a href="#int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a></li>
168 <li><a href="#int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a></li>
169 <li><a href="#int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a></li>
170 </ol>
171 </li>
172 <li><a href="#int_gc">Accurate Garbage Collection Intrinsics</a>
173 <ol>
174 <li><a href="#int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a></li>
175 <li><a href="#int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a></li>
176 <li><a href="#int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a></li>
177 </ol>
178 </li>
179 <li><a href="#int_codegen">Code Generator Intrinsics</a>
180 <ol>
181 <li><a href="#int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a></li>
182 <li><a href="#int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a></li>
183 <li><a href="#int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a></li>
184 <li><a href="#int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a></li>
185 <li><a href="#int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a></li>
186 <li><a href="#int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a></li>
187 <li><a href="#int_readcyclecounter"><tt>llvm.readcyclecounter</tt>' Intrinsic</a></li>
188 </ol>
189 </li>
190 <li><a href="#int_libc">Standard C Library Intrinsics</a>
191 <ol>
192 <li><a href="#int_memcpy">'<tt>llvm.memcpy.*</tt>' Intrinsic</a></li>
193 <li><a href="#int_memmove">'<tt>llvm.memmove.*</tt>' Intrinsic</a></li>
194 <li><a href="#int_memset">'<tt>llvm.memset.*</tt>' Intrinsic</a></li>
195 <li><a href="#int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a></li>
196 <li><a href="#int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a></li>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]197 <li><a href="#int_sin">'<tt>llvm.sin.*</tt>' Intrinsic</a></li>
198 <li><a href="#int_cos">'<tt>llvm.cos.*</tt>' Intrinsic</a></li>
199 <li><a href="#int_pow">'<tt>llvm.pow.*</tt>' Intrinsic</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]200 </ol>
201 </li>
202 <li><a href="#int_manip">Bit Manipulation Intrinsics</a>
203 <ol>
204 <li><a href="#int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a></li>
205 <li><a href="#int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic </a></li>
206 <li><a href="#int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic </a></li>
207 <li><a href="#int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic </a></li>
208 <li><a href="#int_part_select">'<tt>llvm.part.select.*</tt>' Intrinsic </a></li>
209 <li><a href="#int_part_set">'<tt>llvm.part.set.*</tt>' Intrinsic </a></li>
210 </ol>
211 </li>
212 <li><a href="#int_debugger">Debugger intrinsics</a></li>
213 <li><a href="#int_eh">Exception Handling intrinsics</a></li>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]214 <li><a href="#int_trampoline">Trampoline Intrinsic</a>
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]215 <ol>
216 <li><a href="#int_it">'<tt>llvm.init.trampoline</tt>' Intrinsic</a></li>
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]217 </ol>
218 </li>
Bill Wendling9127adb2008-11-18 22:10:53 +0000[diff] [blame]219 <li><a href="#int_atomics">Atomic intrinsics</a>
220 <ol>
221 <li><a href="#int_memory_barrier"><tt>llvm.memory_barrier</tt></a></li>
222 <li><a href="#int_atomic_cmp_swap"><tt>llvm.atomic.cmp.swap</tt></a></li>
223 <li><a href="#int_atomic_swap"><tt>llvm.atomic.swap</tt></a></li>
224 <li><a href="#int_atomic_load_add"><tt>llvm.atomic.load.add</tt></a></li>
225 <li><a href="#int_atomic_load_sub"><tt>llvm.atomic.load.sub</tt></a></li>
226 <li><a href="#int_atomic_load_and"><tt>llvm.atomic.load.and</tt></a></li>
227 <li><a href="#int_atomic_load_nand"><tt>llvm.atomic.load.nand</tt></a></li>
228 <li><a href="#int_atomic_load_or"><tt>llvm.atomic.load.or</tt></a></li>
229 <li><a href="#int_atomic_load_xor"><tt>llvm.atomic.load.xor</tt></a></li>
230 <li><a href="#int_atomic_load_max"><tt>llvm.atomic.load.max</tt></a></li>
231 <li><a href="#int_atomic_load_min"><tt>llvm.atomic.load.min</tt></a></li>
232 <li><a href="#int_atomic_load_umax"><tt>llvm.atomic.load.umax</tt></a></li>
233 <li><a href="#int_atomic_load_umin"><tt>llvm.atomic.load.umin</tt></a></li>
234 </ol>
235 </li>
Reid Spencerb043f672007-07-20 19:59:11 +0000[diff] [blame]236 <li><a href="#int_general">General intrinsics</a>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]237 <ol>
Reid Spencerb043f672007-07-20 19:59:11 +0000[diff] [blame]238 <li><a href="#int_var_annotation">
Bill Wendlinge4164592008-11-19 05:56:17 +0000[diff] [blame]239 '<tt>llvm.var.annotation</tt>' Intrinsic</a></li>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]240 <li><a href="#int_annotation">
Bill Wendlinge4164592008-11-19 05:56:17 +0000[diff] [blame]241 '<tt>llvm.annotation.*</tt>' Intrinsic</a></li>
Anton Korobeynikove6e764f2008-01-15 22:31:34 +0000[diff] [blame]242 <li><a href="#int_trap">
Bill Wendlinge4164592008-11-19 05:56:17 +0000[diff] [blame]243 '<tt>llvm.trap</tt>' Intrinsic</a></li>
244 <li><a href="#int_stackprotector">
245 '<tt>llvm.stackprotector</tt>' Intrinsic</a></li>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]246 </ol>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]247 </li>
248 </ol>
249 </li>
250</ol>
251
252<div class="doc_author">
253 <p>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a>
254 and <a href="mailto:vadve@cs.uiuc.edu">Vikram Adve</a></p>
255</div>
256
257<!-- *********************************************************************** -->
258<div class="doc_section"> <a name="abstract">Abstract </a></div>
259<!-- *********************************************************************** -->
260
261<div class="doc_text">
262<p>This document is a reference manual for the LLVM assembly language.
Bill Wendlinge7846a52008-08-05 22:29:16 +0000[diff] [blame]263LLVM is a Static Single Assignment (SSA) based representation that provides
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]264type safety, low-level operations, flexibility, and the capability of
265representing 'all' high-level languages cleanly. It is the common code
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]266representation used throughout all phases of the LLVM compilation
267strategy.</p>
268</div>
269
270<!-- *********************************************************************** -->
271<div class="doc_section"> <a name="introduction">Introduction</a> </div>
272<!-- *********************************************************************** -->
273
274<div class="doc_text">
275
276<p>The LLVM code representation is designed to be used in three
277different forms: as an in-memory compiler IR, as an on-disk bitcode
278representation (suitable for fast loading by a Just-In-Time compiler),
279and as a human readable assembly language representation. This allows
280LLVM to provide a powerful intermediate representation for efficient
281compiler transformations and analysis, while providing a natural means
282to debug and visualize the transformations. The three different forms
283of LLVM are all equivalent. This document describes the human readable
284representation and notation.</p>
285
286<p>The LLVM representation aims to be light-weight and low-level
287while being expressive, typed, and extensible at the same time. It
288aims to be a "universal IR" of sorts, by being at a low enough level
289that high-level ideas may be cleanly mapped to it (similar to how
290microprocessors are "universal IR's", allowing many source languages to
291be mapped to them). By providing type information, LLVM can be used as
292the target of optimizations: for example, through pointer analysis, it
293can be proven that a C automatic variable is never accessed outside of
294the current function... allowing it to be promoted to a simple SSA
295value instead of a memory location.</p>
296
297</div>
298
299<!-- _______________________________________________________________________ -->
300<div class="doc_subsubsection"> <a name="wellformed">Well-Formedness</a> </div>
301
302<div class="doc_text">
303
304<p>It is important to note that this document describes 'well formed'
305LLVM assembly language. There is a difference between what the parser
306accepts and what is considered 'well formed'. For example, the
307following instruction is syntactically okay, but not well formed:</p>
308
309<div class="doc_code">
310<pre>
311%x = <a href="#i_add">add</a> i32 1, %x
312</pre>
313</div>
314
315<p>...because the definition of <tt>%x</tt> does not dominate all of
316its uses. The LLVM infrastructure provides a verification pass that may
317be used to verify that an LLVM module is well formed. This pass is
318automatically run by the parser after parsing input assembly and by
319the optimizer before it outputs bitcode. The violations pointed out
320by the verifier pass indicate bugs in transformation passes or input to
321the parser.</p>
322</div>
323
Chris Lattnera83fdc02007-10-03 17:34:29 +0000[diff] [blame]324<!-- Describe the typesetting conventions here. -->
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]325
326<!-- *********************************************************************** -->
327<div class="doc_section"> <a name="identifiers">Identifiers</a> </div>
328<!-- *********************************************************************** -->
329
330<div class="doc_text">
331
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]332 <p>LLVM identifiers come in two basic types: global and local. Global
333 identifiers (functions, global variables) begin with the @ character. Local
334 identifiers (register names, types) begin with the % character. Additionally,
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]335 there are three different formats for identifiers, for different purposes:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]336
337<ol>
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]338 <li>Named values are represented as a string of characters with their prefix.
339 For example, %foo, @DivisionByZero, %a.really.long.identifier. The actual
340 regular expression used is '<tt>[%@][a-zA-Z$._][a-zA-Z$._0-9]*</tt>'.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]341 Identifiers which require other characters in their names can be surrounded
Daniel Dunbar73f9a772008-10-14 23:51:43 +0000[diff] [blame]342 with quotes. Special characters may be escaped using "\xx" where xx is the
343 ASCII code for the character in hexadecimal. In this way, any character can
344 be used in a name value, even quotes themselves.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]345
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]346 <li>Unnamed values are represented as an unsigned numeric value with their
347 prefix. For example, %12, @2, %44.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]348
349 <li>Constants, which are described in a <a href="#constants">section about
350 constants</a>, below.</li>
351</ol>
352
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]353<p>LLVM requires that values start with a prefix for two reasons: Compilers
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]354don't need to worry about name clashes with reserved words, and the set of
355reserved words may be expanded in the future without penalty. Additionally,
356unnamed identifiers allow a compiler to quickly come up with a temporary
357variable without having to avoid symbol table conflicts.</p>
358
359<p>Reserved words in LLVM are very similar to reserved words in other
360languages. There are keywords for different opcodes
361('<tt><a href="#i_add">add</a></tt>',
362 '<tt><a href="#i_bitcast">bitcast</a></tt>',
363 '<tt><a href="#i_ret">ret</a></tt>', etc...), for primitive type names ('<tt><a
364href="#t_void">void</a></tt>', '<tt><a href="#t_primitive">i32</a></tt>', etc...),
365and others. These reserved words cannot conflict with variable names, because
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]366none of them start with a prefix character ('%' or '@').</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]367
368<p>Here is an example of LLVM code to multiply the integer variable
369'<tt>%X</tt>' by 8:</p>
370
371<p>The easy way:</p>
372
373<div class="doc_code">
374<pre>
375%result = <a href="#i_mul">mul</a> i32 %X, 8
376</pre>
377</div>
378
379<p>After strength reduction:</p>
380
381<div class="doc_code">
382<pre>
383%result = <a href="#i_shl">shl</a> i32 %X, i8 3
384</pre>
385</div>
386
387<p>And the hard way:</p>
388
389<div class="doc_code">
390<pre>
391<a href="#i_add">add</a> i32 %X, %X <i>; yields {i32}:%0</i>
392<a href="#i_add">add</a> i32 %0, %0 <i>; yields {i32}:%1</i>
393%result = <a href="#i_add">add</a> i32 %1, %1
394</pre>
395</div>
396
397<p>This last way of multiplying <tt>%X</tt> by 8 illustrates several
398important lexical features of LLVM:</p>
399
400<ol>
401
402 <li>Comments are delimited with a '<tt>;</tt>' and go until the end of
403 line.</li>
404
405 <li>Unnamed temporaries are created when the result of a computation is not
406 assigned to a named value.</li>
407
408 <li>Unnamed temporaries are numbered sequentially</li>
409
410</ol>
411
412<p>...and it also shows a convention that we follow in this document. When
413demonstrating instructions, we will follow an instruction with a comment that
414defines the type and name of value produced. Comments are shown in italic
415text.</p>
416
417</div>
418
419<!-- *********************************************************************** -->
420<div class="doc_section"> <a name="highlevel">High Level Structure</a> </div>
421<!-- *********************************************************************** -->
422
423<!-- ======================================================================= -->
424<div class="doc_subsection"> <a name="modulestructure">Module Structure</a>
425</div>
426
427<div class="doc_text">
428
429<p>LLVM programs are composed of "Module"s, each of which is a
430translation unit of the input programs. Each module consists of
431functions, global variables, and symbol table entries. Modules may be
432combined together with the LLVM linker, which merges function (and
433global variable) definitions, resolves forward declarations, and merges
434symbol table entries. Here is an example of the "hello world" module:</p>
435
436<div class="doc_code">
437<pre><i>; Declare the string constant as a global constant...</i>
438<a href="#identifiers">@.LC0</a> = <a href="#linkage_internal">internal</a> <a
439 href="#globalvars">constant</a> <a href="#t_array">[13 x i8]</a> c"hello world\0A\00" <i>; [13 x i8]*</i>
440
441<i>; External declaration of the puts function</i>
442<a href="#functionstructure">declare</a> i32 @puts(i8 *) <i>; i32(i8 *)* </i>
443
444<i>; Definition of main function</i>
445define i32 @main() { <i>; i32()* </i>
Dan Gohman01852382009-01-04 23:44:43 +0000[diff] [blame]446 <i>; Convert [13 x i8]* to i8 *...</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]447 %cast210 = <a
Dan Gohman01852382009-01-04 23:44:43 +0000[diff] [blame]448 href="#i_getelementptr">getelementptr</a> [13 x i8]* @.LC0, i64 0, i64 0 <i>; i8 *</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]449
450 <i>; Call puts function to write out the string to stdout...</i>
451 <a
452 href="#i_call">call</a> i32 @puts(i8 * %cast210) <i>; i32</i>
453 <a
454 href="#i_ret">ret</a> i32 0<br>}<br>
455</pre>
456</div>
457
458<p>This example is made up of a <a href="#globalvars">global variable</a>
459named "<tt>.LC0</tt>", an external declaration of the "<tt>puts</tt>"
460function, and a <a href="#functionstructure">function definition</a>
461for "<tt>main</tt>".</p>
462
463<p>In general, a module is made up of a list of global values,
464where both functions and global variables are global values. Global values are
465represented by a pointer to a memory location (in this case, a pointer to an
466array of char, and a pointer to a function), and have one of the following <a
467href="#linkage">linkage types</a>.</p>
468
469</div>
470
471<!-- ======================================================================= -->
472<div class="doc_subsection">
473 <a name="linkage">Linkage Types</a>
474</div>
475
476<div class="doc_text">
477
478<p>
479All Global Variables and Functions have one of the following types of linkage:
480</p>
481
482<dl>
483
Rafael Espindolaa168fc92009-01-15 20:18:42 +0000[diff] [blame]484 <dt><tt><b><a name="linkage_private">private</a></b></tt>: </dt>
485
486 <dd>Global values with private linkage are only directly accessible by
487 objects in the current module. In particular, linking code into a module with
488 an private global value may cause the private to be renamed as necessary to
489 avoid collisions. Because the symbol is private to the module, all
490 references can be updated. This doesn't show up in any symbol table in the
491 object file.
492 </dd>
493
Dale Johannesen96e7e092008-05-23 23:13:41 +0000[diff] [blame]494 <dt><tt><b><a name="linkage_internal">internal</a></b></tt>: </dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]495
Duncan Sandsa75223a2009-01-16 09:29:46 +0000[diff] [blame]496 <dd> Similar to private, but the value shows as a local symbol (STB_LOCAL in
Rafael Espindolaa168fc92009-01-15 20:18:42 +0000[diff] [blame]497 the case of ELF) in the object file. This corresponds to the notion of the
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]498 '<tt>static</tt>' keyword in C.
499 </dd>
500
501 <dt><tt><b><a name="linkage_linkonce">linkonce</a></b></tt>: </dt>
502
503 <dd>Globals with "<tt>linkonce</tt>" linkage are merged with other globals of
504 the same name when linkage occurs. This is typically used to implement
505 inline functions, templates, or other code which must be generated in each
506 translation unit that uses it. Unreferenced <tt>linkonce</tt> globals are
507 allowed to be discarded.
508 </dd>
509
Dale Johannesen96e7e092008-05-23 23:13:41 +0000[diff] [blame]510 <dt><tt><b><a name="linkage_common">common</a></b></tt>: </dt>
511
512 <dd>"<tt>common</tt>" linkage is exactly the same as <tt>linkonce</tt>
513 linkage, except that unreferenced <tt>common</tt> globals may not be
514 discarded. This is used for globals that may be emitted in multiple
515 translation units, but that are not guaranteed to be emitted into every
516 translation unit that uses them. One example of this is tentative
517 definitions in C, such as "<tt>int X;</tt>" at global scope.
518 </dd>
519
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]520 <dt><tt><b><a name="linkage_weak">weak</a></b></tt>: </dt>
521
Dale Johannesen96e7e092008-05-23 23:13:41 +0000[diff] [blame]522 <dd>"<tt>weak</tt>" linkage is the same as <tt>common</tt> linkage, except
523 that some targets may choose to emit different assembly sequences for them
524 for target-dependent reasons. This is used for globals that are declared
525 "weak" in C source code.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]526 </dd>
527
528 <dt><tt><b><a name="linkage_appending">appending</a></b></tt>: </dt>
529
530 <dd>"<tt>appending</tt>" linkage may only be applied to global variables of
531 pointer to array type. When two global variables with appending linkage are
532 linked together, the two global arrays are appended together. This is the
533 LLVM, typesafe, equivalent of having the system linker append together
534 "sections" with identical names when .o files are linked.
535 </dd>
536
537 <dt><tt><b><a name="linkage_externweak">extern_weak</a></b></tt>: </dt>
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]538 <dd>The semantics of this linkage follow the ELF object file model: the
539 symbol is weak until linked, if not linked, the symbol becomes null instead
540 of being an undefined reference.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]541 </dd>
542
543 <dt><tt><b><a name="linkage_external">externally visible</a></b></tt>:</dt>
544
545 <dd>If none of the above identifiers are used, the global is externally
546 visible, meaning that it participates in linkage and can be used to resolve
547 external symbol references.
548 </dd>
549</dl>
550
551 <p>
552 The next two types of linkage are targeted for Microsoft Windows platform
553 only. They are designed to support importing (exporting) symbols from (to)
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]554 DLLs (Dynamic Link Libraries).
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]555 </p>
556
557 <dl>
558 <dt><tt><b><a name="linkage_dllimport">dllimport</a></b></tt>: </dt>
559
560 <dd>"<tt>dllimport</tt>" linkage causes the compiler to reference a function
561 or variable via a global pointer to a pointer that is set up by the DLL
562 exporting the symbol. On Microsoft Windows targets, the pointer name is
Dan Gohman5ec99832009-01-12 21:35:55 +0000[diff] [blame]563 formed by combining <code>__imp_</code> and the function or variable name.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]564 </dd>
565
566 <dt><tt><b><a name="linkage_dllexport">dllexport</a></b></tt>: </dt>
567
568 <dd>"<tt>dllexport</tt>" linkage causes the compiler to provide a global
569 pointer to a pointer in a DLL, so that it can be referenced with the
570 <tt>dllimport</tt> attribute. On Microsoft Windows targets, the pointer
Dan Gohman5ec99832009-01-12 21:35:55 +0000[diff] [blame]571 name is formed by combining <code>__imp_</code> and the function or variable
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]572 name.
573 </dd>
574
575</dl>
576
Dan Gohman4dfac702008-11-24 17:18:39 +0000[diff] [blame]577<p>For example, since the "<tt>.LC0</tt>"
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]578variable is defined to be internal, if another module defined a "<tt>.LC0</tt>"
579variable and was linked with this one, one of the two would be renamed,
580preventing a collision. Since "<tt>main</tt>" and "<tt>puts</tt>" are
581external (i.e., lacking any linkage declarations), they are accessible
582outside of the current module.</p>
583<p>It is illegal for a function <i>declaration</i>
584to have any linkage type other than "externally visible", <tt>dllimport</tt>,
585or <tt>extern_weak</tt>.</p>
586<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]587linkages.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]588</div>
589
590<!-- ======================================================================= -->
591<div class="doc_subsection">
592 <a name="callingconv">Calling Conventions</a>
593</div>
594
595<div class="doc_text">
596
597<p>LLVM <a href="#functionstructure">functions</a>, <a href="#i_call">calls</a>
598and <a href="#i_invoke">invokes</a> can all have an optional calling convention
599specified for the call. The calling convention of any pair of dynamic
600caller/callee must match, or the behavior of the program is undefined. The
601following calling conventions are supported by LLVM, and more may be added in
602the future:</p>
603
604<dl>
605 <dt><b>"<tt>ccc</tt>" - The C calling convention</b>:</dt>
606
607 <dd>This calling convention (the default if no other calling convention is
608 specified) matches the target C calling conventions. This calling convention
609 supports varargs function calls and tolerates some mismatch in the declared
610 prototype and implemented declaration of the function (as does normal C).
611 </dd>
612
613 <dt><b>"<tt>fastcc</tt>" - The fast calling convention</b>:</dt>
614
615 <dd>This calling convention attempts to make calls as fast as possible
616 (e.g. by passing things in registers). This calling convention allows the
617 target to use whatever tricks it wants to produce fast code for the target,
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]618 without having to conform to an externally specified ABI (Application Binary
619 Interface). Implementations of this convention should allow arbitrary
Arnold Schwaighofer07444922008-05-14 09:17:12 +0000[diff] [blame]620 <a href="CodeGenerator.html#tailcallopt">tail call optimization</a> to be
621 supported. This calling convention does not support varargs and requires the
622 prototype of all callees to exactly match the prototype of the function
623 definition.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]624 </dd>
625
626 <dt><b>"<tt>coldcc</tt>" - The cold calling convention</b>:</dt>
627
628 <dd>This calling convention attempts to make code in the caller as efficient
629 as possible under the assumption that the call is not commonly executed. As
630 such, these calls often preserve all registers so that the call does not break
631 any live ranges in the caller side. This calling convention does not support
632 varargs and requires the prototype of all callees to exactly match the
633 prototype of the function definition.
634 </dd>
635
636 <dt><b>"<tt>cc &lt;<em>n</em>&gt;</tt>" - Numbered convention</b>:</dt>
637
638 <dd>Any calling convention may be specified by number, allowing
639 target-specific calling conventions to be used. Target specific calling
640 conventions start at 64.
641 </dd>
642</dl>
643
644<p>More calling conventions can be added/defined on an as-needed basis, to
645support pascal conventions or any other well-known target-independent
646convention.</p>
647
648</div>
649
650<!-- ======================================================================= -->
651<div class="doc_subsection">
652 <a name="visibility">Visibility Styles</a>
653</div>
654
655<div class="doc_text">
656
657<p>
658All Global Variables and Functions have one of the following visibility styles:
659</p>
660
661<dl>
662 <dt><b>"<tt>default</tt>" - Default style</b>:</dt>
663
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]664 <dd>On targets that use the ELF object file format, default visibility means
665 that the declaration is visible to other
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]666 modules and, in shared libraries, means that the declared entity may be
667 overridden. On Darwin, default visibility means that the declaration is
668 visible to other modules. Default visibility corresponds to "external
669 linkage" in the language.
670 </dd>
671
672 <dt><b>"<tt>hidden</tt>" - Hidden style</b>:</dt>
673
674 <dd>Two declarations of an object with hidden visibility refer to the same
675 object if they are in the same shared object. Usually, hidden visibility
676 indicates that the symbol will not be placed into the dynamic symbol table,
677 so no other module (executable or shared library) can reference it
678 directly.
679 </dd>
680
681 <dt><b>"<tt>protected</tt>" - Protected style</b>:</dt>
682
683 <dd>On ELF, protected visibility indicates that the symbol will be placed in
684 the dynamic symbol table, but that references within the defining module will
685 bind to the local symbol. That is, the symbol cannot be overridden by another
686 module.
687 </dd>
688</dl>
689
690</div>
691
692<!-- ======================================================================= -->
693<div class="doc_subsection">
Chris Lattner5b6dc6e2009-01-11 20:53:49 +0000[diff] [blame]694 <a name="namedtypes">Named Types</a>
695</div>
696
697<div class="doc_text">
698
699<p>LLVM IR allows you to specify name aliases for certain types. This can make
700it easier to read the IR and make the IR more condensed (particularly when
701recursive types are involved). An example of a name specification is:
702</p>
703
704<div class="doc_code">
705<pre>
706%mytype = type { %mytype*, i32 }
707</pre>
708</div>
709
710<p>You may give a name to any <a href="#typesystem">type</a> except "<a
711href="t_void">void</a>". Type name aliases may be used anywhere a type is
712expected with the syntax "%mytype".</p>
713
714<p>Note that type names are aliases for the structural type that they indicate,
715and that you can therefore specify multiple names for the same type. This often
716leads to confusing behavior when dumping out a .ll file. Since LLVM IR uses
717structural typing, the name is not part of the type. When printing out LLVM IR,
718the printer will pick <em>one name</em> to render all types of a particular
719shape. This means that if you have code where two different source types end up
720having the same LLVM type, that the dumper will sometimes print the "wrong" or
721unexpected type. This is an important design point and isn't going to
722change.</p>
723
724</div>
725
Chris Lattner5b6dc6e2009-01-11 20:53:49 +0000[diff] [blame]726<!-- ======================================================================= -->
727<div class="doc_subsection">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]728 <a name="globalvars">Global Variables</a>
729</div>
730
731<div class="doc_text">
732
733<p>Global variables define regions of memory allocated at compilation time
734instead of run-time. Global variables may optionally be initialized, may have
735an explicit section to be placed in, and may have an optional explicit alignment
736specified. A variable may be defined as "thread_local", which means that it
737will not be shared by threads (each thread will have a separated copy of the
738variable). A variable may be defined as a global "constant," which indicates
739that the contents of the variable will <b>never</b> be modified (enabling better
740optimization, allowing the global data to be placed in the read-only section of
741an executable, etc). Note that variables that need runtime initialization
742cannot be marked "constant" as there is a store to the variable.</p>
743
744<p>
745LLVM explicitly allows <em>declarations</em> of global variables to be marked
746constant, even if the final definition of the global is not. This capability
747can be used to enable slightly better optimization of the program, but requires
748the language definition to guarantee that optimizations based on the
749'constantness' are valid for the translation units that do not include the
750definition.
751</p>
752
753<p>As SSA values, global variables define pointer values that are in
754scope (i.e. they dominate) all basic blocks in the program. Global
755variables always define a pointer to their "content" type because they
756describe a region of memory, and all memory objects in LLVM are
757accessed through pointers.</p>
758
Christopher Lambdd0049d2007-12-11 09:31:00 +0000[diff] [blame]759<p>A global variable may be declared to reside in a target-specifc numbered
760address space. For targets that support them, address spaces may affect how
761optimizations are performed and/or what target instructions are used to access
Christopher Lamb20a39e92007-12-12 08:44:39 +0000[diff] [blame]762the variable. The default address space is zero. The address space qualifier
763must precede any other attributes.</p>
Christopher Lambdd0049d2007-12-11 09:31:00 +0000[diff] [blame]764
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]765<p>LLVM allows an explicit section to be specified for globals. If the target
766supports it, it will emit globals to the section specified.</p>
767
768<p>An explicit alignment may be specified for a global. If not present, or if
769the alignment is set to zero, the alignment of the global is set by the target
770to whatever it feels convenient. If an explicit alignment is specified, the
771global is forced to have at least that much alignment. All alignments must be
772a power of 2.</p>
773
Christopher Lambdd0049d2007-12-11 09:31:00 +0000[diff] [blame]774<p>For example, the following defines a global in a numbered address space with
775an initializer, section, and alignment:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]776
777<div class="doc_code">
778<pre>
Dan Gohman21ef02c2009-01-11 00:40:00 +0000[diff] [blame]779@G = addrspace(5) constant float 1.0, section "foo", align 4
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]780</pre>
781</div>
782
783</div>
784
785
786<!-- ======================================================================= -->
787<div class="doc_subsection">
788 <a name="functionstructure">Functions</a>
789</div>
790
791<div class="doc_text">
792
793<p>LLVM function definitions consist of the "<tt>define</tt>" keyord,
794an optional <a href="#linkage">linkage type</a>, an optional
795<a href="#visibility">visibility style</a>, an optional
796<a href="#callingconv">calling convention</a>, a return type, an optional
797<a href="#paramattrs">parameter attribute</a> for the return type, a function
798name, a (possibly empty) argument list (each with optional
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]799<a href="#paramattrs">parameter attributes</a>), optional
800<a href="#fnattrs">function attributes</a>, an optional section,
801an optional alignment, an optional <a href="#gc">garbage collector name</a>,
Chris Lattneree202542008-10-04 18:10:21 +0000[diff] [blame]802an opening curly brace, a list of basic blocks, and a closing curly brace.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]803
804LLVM function declarations consist of the "<tt>declare</tt>" keyword, an
805optional <a href="#linkage">linkage type</a>, an optional
806<a href="#visibility">visibility style</a>, an optional
807<a href="#callingconv">calling convention</a>, a return type, an optional
808<a href="#paramattrs">parameter attribute</a> for the return type, a function
Gordon Henriksen13fe5e32007-12-10 03:18:06 +0000[diff] [blame]809name, a possibly empty list of arguments, an optional alignment, and an optional
Gordon Henriksend606f5b2007-12-10 03:30:21 +0000[diff] [blame]810<a href="#gc">garbage collector name</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]811
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]812<p>A function definition contains a list of basic blocks, forming the CFG
813(Control Flow Graph) for
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]814the function. Each basic block may optionally start with a label (giving the
815basic block a symbol table entry), contains a list of instructions, and ends
816with a <a href="#terminators">terminator</a> instruction (such as a branch or
817function return).</p>
818
819<p>The first basic block in a function is special in two ways: it is immediately
820executed on entrance to the function, and it is not allowed to have predecessor
821basic blocks (i.e. there can not be any branches to the entry block of a
822function). Because the block can have no predecessors, it also cannot have any
823<a href="#i_phi">PHI nodes</a>.</p>
824
825<p>LLVM allows an explicit section to be specified for functions. If the target
826supports it, it will emit functions to the section specified.</p>
827
828<p>An explicit alignment may be specified for a function. If not present, or if
829the alignment is set to zero, the alignment of the function is set by the target
830to whatever it feels convenient. If an explicit alignment is specified, the
831function is forced to have at least that much alignment. All alignments must be
832a power of 2.</p>
833
Devang Pateld0bfcc72008-10-07 17:48:33 +0000[diff] [blame]834 <h5>Syntax:</h5>
835
836<div class="doc_code">
Chris Lattner1e5c5cd02008-10-13 16:55:18 +0000[diff] [blame]837<tt>
838define [<a href="#linkage">linkage</a>] [<a href="#visibility">visibility</a>]
839 [<a href="#callingconv">cconv</a>] [<a href="#paramattrs">ret attrs</a>]
840 &lt;ResultType&gt; @&lt;FunctionName&gt; ([argument list])
841 [<a href="#fnattrs">fn Attrs</a>] [section "name"] [align N]
842 [<a href="#gc">gc</a>] { ... }
843</tt>
Devang Pateld0bfcc72008-10-07 17:48:33 +0000[diff] [blame]844</div>
845
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]846</div>
847
848
849<!-- ======================================================================= -->
850<div class="doc_subsection">
851 <a name="aliasstructure">Aliases</a>
852</div>
853<div class="doc_text">
854 <p>Aliases act as "second name" for the aliasee value (which can be either
Anton Korobeynikov96822812008-03-22 08:36:14 +0000[diff] [blame]855 function, global variable, another alias or bitcast of global value). Aliases
856 may have an optional <a href="#linkage">linkage type</a>, and an
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]857 optional <a href="#visibility">visibility style</a>.</p>
858
859 <h5>Syntax:</h5>
860
861<div class="doc_code">
862<pre>
Duncan Sandsd7bfabf2008-09-12 20:48:21 +0000[diff] [blame]863@&lt;Name&gt; = alias [Linkage] [Visibility] &lt;AliaseeTy&gt; @&lt;Aliasee&gt;
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]864</pre>
865</div>
866
867</div>
868
869
870
871<!-- ======================================================================= -->
872<div class="doc_subsection"><a name="paramattrs">Parameter Attributes</a></div>
873<div class="doc_text">
874 <p>The return type and each parameter of a function type may have a set of
875 <i>parameter attributes</i> associated with them. Parameter attributes are
876 used to communicate additional information about the result or parameters of
Duncan Sandsf5588dc2007-11-27 13:23:08 +0000[diff] [blame]877 a function. Parameter attributes are considered to be part of the function,
878 not of the function type, so functions with different parameter attributes
879 can have the same function type.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]880
881 <p>Parameter attributes are simple keywords that follow the type specified. If
882 multiple parameter attributes are needed, they are space separated. For
883 example:</p>
884
885<div class="doc_code">
886<pre>
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]887declare i32 @printf(i8* noalias , ...)
Chris Lattnerf33b8452008-10-04 18:33:34 +0000[diff] [blame]888declare i32 @atoi(i8 zeroext)
889declare signext i8 @returns_signed_char()
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]890</pre>
891</div>
892
Duncan Sandsf5588dc2007-11-27 13:23:08 +0000[diff] [blame]893 <p>Note that any attributes for the function result (<tt>nounwind</tt>,
894 <tt>readonly</tt>) come immediately after the argument list.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]895
896 <p>Currently, only the following parameter attributes are defined:</p>
897 <dl>
Reid Spencerf234bed2007-07-19 23:13:04 +0000[diff] [blame]898 <dt><tt>zeroext</tt></dt>
Chris Lattnerf33b8452008-10-04 18:33:34 +0000[diff] [blame]899 <dd>This indicates to the code generator that the parameter or return value
900 should be zero-extended to a 32-bit value by the caller (for a parameter)
901 or the callee (for a return value).</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]902
Reid Spencerf234bed2007-07-19 23:13:04 +0000[diff] [blame]903 <dt><tt>signext</tt></dt>
Chris Lattnerf33b8452008-10-04 18:33:34 +0000[diff] [blame]904 <dd>This indicates to the code generator that the parameter or return value
905 should be sign-extended to a 32-bit value by the caller (for a parameter)
906 or the callee (for a return value).</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]907
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]908 <dt><tt>inreg</tt></dt>
Dale Johannesenc08a0e22008-09-25 20:47:45 +0000[diff] [blame]909 <dd>This indicates that this parameter or return value should be treated
910 in a special target-dependent fashion during while emitting code for a
911 function call or return (usually, by putting it in a register as opposed
Chris Lattnerf33b8452008-10-04 18:33:34 +0000[diff] [blame]912 to memory, though some targets use it to distinguish between two different
913 kinds of registers). Use of this attribute is target-specific.</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]914
Duncan Sands7d94e5a2008-10-06 08:14:18 +0000[diff] [blame]915 <dt><tt><a name="byval">byval</a></tt></dt>
Chris Lattner04c86182008-01-15 04:34:22 +0000[diff] [blame]916 <dd>This indicates that the pointer parameter should really be passed by
917 value to the function. The attribute implies that a hidden copy of the
918 pointee is made between the caller and the callee, so the callee is unable
Chris Lattner6a9f3c42008-08-05 18:21:08 +0000[diff] [blame]919 to modify the value in the callee. This attribute is only valid on LLVM
Chris Lattner04c86182008-01-15 04:34:22 +0000[diff] [blame]920 pointer arguments. It is generally used to pass structs and arrays by
Duncan Sands7d94e5a2008-10-06 08:14:18 +0000[diff] [blame]921 value, but is also valid on pointers to scalars. The copy is considered to
922 belong to the caller not the callee (for example,
923 <tt><a href="#readonly">readonly</a></tt> functions should not write to
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]924 <tt>byval</tt> parameters). This is not a valid attribute for return
Chris Lattnere39c5bf2009-02-05 05:42:28 +0000[diff] [blame^]925 values. The byval attribute also supports specifying an alignment with the
926 align attribute. This has a target-specific effect on the code generator
927 that usually indicates a desired alignment for the synthesized stack
928 slot.</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]929
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]930 <dt><tt>sret</tt></dt>
Duncan Sands616cc032008-02-18 04:19:38 +0000[diff] [blame]931 <dd>This indicates that the pointer parameter specifies the address of a
932 structure that is the return value of the function in the source program.
Chris Lattnerf33b8452008-10-04 18:33:34 +0000[diff] [blame]933 This pointer must be guaranteed by the caller to be valid: loads and stores
934 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]935 be applied to the first parameter. This is not a valid attribute for
936 return values. </dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]937
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]938 <dt><tt>noalias</tt></dt>
Nick Lewyckyff384472008-11-24 03:41:24 +0000[diff] [blame]939 <dd>This indicates that the pointer does not alias any global or any other
940 parameter. The caller is responsible for ensuring that this is the
Nick Lewyckya604a942008-11-24 05:00:44 +0000[diff] [blame]941 case. On a function return value, <tt>noalias</tt> additionally indicates
942 that the pointer does not alias any other pointers visible to the
Nick Lewycky40fb6f22008-12-19 06:39:12 +0000[diff] [blame]943 caller. For further details, please see the discussion of the NoAlias
944 response in
945 <a href="http://llvm.org/docs/AliasAnalysis.html#MustMayNo">alias
946 analysis</a>.</dd>
947
948 <dt><tt>nocapture</tt></dt>
949 <dd>This indicates that the callee does not make any copies of the pointer
950 that outlive the callee itself. This is not a valid attribute for return
951 values.</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]952
Duncan Sands4ee46812007-07-27 19:57:41 +0000[diff] [blame]953 <dt><tt>nest</tt></dt>
Duncan Sandsf1a7d4c2008-07-08 09:27:25 +0000[diff] [blame]954 <dd>This indicates that the pointer parameter can be excised using the
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]955 <a href="#int_trampoline">trampoline intrinsics</a>. This is not a valid
956 attribute for return values.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]957 </dl>
958
959</div>
960
961<!-- ======================================================================= -->
962<div class="doc_subsection">
Gordon Henriksen13fe5e32007-12-10 03:18:06 +0000[diff] [blame]963 <a name="gc">Garbage Collector Names</a>
964</div>
965
966<div class="doc_text">
967<p>Each function may specify a garbage collector name, which is simply a
968string.</p>
969
970<div class="doc_code"><pre
971>define void @f() gc "name" { ...</pre></div>
972
973<p>The compiler declares the supported values of <i>name</i>. Specifying a
974collector which will cause the compiler to alter its output in order to support
975the named garbage collection algorithm.</p>
976</div>
977
978<!-- ======================================================================= -->
979<div class="doc_subsection">
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]980 <a name="fnattrs">Function Attributes</a>
Devang Pateld468f1c2008-09-04 23:05:13 +0000[diff] [blame]981</div>
982
983<div class="doc_text">
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]984
985<p>Function attributes are set to communicate additional information about
986 a function. Function attributes are considered to be part of the function,
987 not of the function type, so functions with different parameter attributes
988 can have the same function type.</p>
989
990 <p>Function attributes are simple keywords that follow the type specified. If
991 multiple attributes are needed, they are space separated. For
992 example:</p>
Devang Pateld468f1c2008-09-04 23:05:13 +0000[diff] [blame]993
994<div class="doc_code">
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]995<pre>
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]996define void @f() noinline { ... }
997define void @f() alwaysinline { ... }
998define void @f() alwaysinline optsize { ... }
999define void @f() optsize
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]1000</pre>
Devang Pateld468f1c2008-09-04 23:05:13 +0000[diff] [blame]1001</div>
1002
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]1003<dl>
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]1004<dt><tt>alwaysinline</tt></dt>
Chris Lattner82d70f92008-10-04 18:23:17 +0000[diff] [blame]1005<dd>This attribute indicates that the inliner should attempt to inline this
1006function into callers whenever possible, ignoring any active inlining size
1007threshold for this caller.</dd>
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]1008
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]1009<dt><tt>noinline</tt></dt>
Chris Lattner82d70f92008-10-04 18:23:17 +0000[diff] [blame]1010<dd>This attribute indicates that the inliner should never inline this function
Chris Lattner377f3ae2008-10-05 17:14:59 +0000[diff] [blame]1011in any situation. This attribute may not be used together with the
Chris Lattner82d70f92008-10-04 18:23:17 +0000[diff] [blame]1012<tt>alwaysinline</tt> attribute.</dd>
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]1013
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]1014<dt><tt>optsize</tt></dt>
Devang Patelc29099b2008-09-29 18:34:44 +0000[diff] [blame]1015<dd>This attribute suggests that optimization passes and code generator passes
Chris Lattner82d70f92008-10-04 18:23:17 +0000[diff] [blame]1016make choices that keep the code size of this function low, and otherwise do
1017optimizations specifically to reduce code size.</dd>
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]1018
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]1019<dt><tt>noreturn</tt></dt>
Chris Lattner82d70f92008-10-04 18:23:17 +0000[diff] [blame]1020<dd>This function attribute indicates that the function never returns normally.
1021This produces undefined behavior at runtime if the function ever does
1022dynamically return.</dd>
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]1023
1024<dt><tt>nounwind</tt></dt>
Chris Lattner82d70f92008-10-04 18:23:17 +0000[diff] [blame]1025<dd>This function attribute indicates that the function never returns with an
1026unwind or exceptional control flow. If the function does unwind, its runtime
1027behavior is undefined.</dd>
1028
1029<dt><tt>readnone</tt></dt>
Duncan Sands7d94e5a2008-10-06 08:14:18 +0000[diff] [blame]1030<dd>This attribute indicates that the function computes its result (or the
1031exception it throws) based strictly on its arguments, without dereferencing any
1032pointer arguments or otherwise accessing any mutable state (e.g. memory, control
1033registers, etc) visible to caller functions. It does not write through any
1034pointer arguments (including <tt><a href="#byval">byval</a></tt> arguments) and
1035never changes any state visible to callers.</dd>
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]1036
Duncan Sands7d94e5a2008-10-06 08:14:18 +0000[diff] [blame]1037<dt><tt><a name="readonly">readonly</a></tt></dt>
1038<dd>This attribute indicates that the function does not write through any
1039pointer arguments (including <tt><a href="#byval">byval</a></tt> arguments)
1040or otherwise modify any state (e.g. memory, control registers, etc) visible to
1041caller functions. It may dereference pointer arguments and read state that may
1042be set in the caller. A readonly function always returns the same value (or
1043throws the same exception) when called with the same set of arguments and global
1044state.</dd>
Bill Wendlingdfaabba2008-11-13 01:02:51 +0000[diff] [blame]1045
1046<dt><tt><a name="ssp">ssp</a></tt></dt>
Bill Wendling34f70d82008-11-26 19:19:05 +0000[diff] [blame]1047<dd>This attribute indicates that the function should emit a stack smashing
Bill Wendlingdfaabba2008-11-13 01:02:51 +0000[diff] [blame]1048protector. It is in the form of a "canary"&mdash;a random value placed on the
1049stack before the local variables that's checked upon return from the function to
1050see if it has been overwritten. A heuristic is used to determine if a function
Bill Wendling34f70d82008-11-26 19:19:05 +0000[diff] [blame]1051needs stack protectors or not.
Bill Wendlingdfaabba2008-11-13 01:02:51 +0000[diff] [blame]1052
Bill Wendlingbe9ec3f2008-11-26 19:07:40 +0000[diff] [blame]1053<p>If a function that has an <tt>ssp</tt> attribute is inlined into a function
1054that doesn't have an <tt>ssp</tt> attribute, then the resulting function will
1055have an <tt>ssp</tt> attribute.</p></dd>
1056
1057<dt><tt>sspreq</tt></dt>
Bill Wendling34f70d82008-11-26 19:19:05 +0000[diff] [blame]1058<dd>This attribute indicates that the function should <em>always</em> emit a
Bill Wendlingdfaabba2008-11-13 01:02:51 +0000[diff] [blame]1059stack smashing protector. This overrides the <tt><a href="#ssp">ssp</a></tt>
Bill Wendling34f70d82008-11-26 19:19:05 +0000[diff] [blame]1060function attribute.
Bill Wendlingbe9ec3f2008-11-26 19:07:40 +0000[diff] [blame]1061
1062<p>If a function that has an <tt>sspreq</tt> attribute is inlined into a
1063function that doesn't have an <tt>sspreq</tt> attribute or which has
1064an <tt>ssp</tt> attribute, then the resulting function will have
1065an <tt>sspreq</tt> attribute.</p></dd>
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]1066</dl>
1067
Devang Pateld468f1c2008-09-04 23:05:13 +0000[diff] [blame]1068</div>
1069
1070<!-- ======================================================================= -->
1071<div class="doc_subsection">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1072 <a name="moduleasm">Module-Level Inline Assembly</a>
1073</div>
1074
1075<div class="doc_text">
1076<p>
1077Modules may contain "module-level inline asm" blocks, which corresponds to the
1078GCC "file scope inline asm" blocks. These blocks are internally concatenated by
1079LLVM and treated as a single unit, but may be separated in the .ll file if
1080desired. The syntax is very simple:
1081</p>
1082
1083<div class="doc_code">
1084<pre>
1085module asm "inline asm code goes here"
1086module asm "more can go here"
1087</pre>
1088</div>
1089
1090<p>The strings can contain any character by escaping non-printable characters.
1091 The escape sequence used is simply "\xx" where "xx" is the two digit hex code
1092 for the number.
1093</p>
1094
1095<p>
1096 The inline asm code is simply printed to the machine code .s file when
1097 assembly code is generated.
1098</p>
1099</div>
1100
1101<!-- ======================================================================= -->
1102<div class="doc_subsection">
1103 <a name="datalayout">Data Layout</a>
1104</div>
1105
1106<div class="doc_text">
1107<p>A module may specify a target specific data layout string that specifies how
1108data is to be laid out in memory. The syntax for the data layout is simply:</p>
1109<pre> target datalayout = "<i>layout specification</i>"</pre>
1110<p>The <i>layout specification</i> consists of a list of specifications
1111separated by the minus sign character ('-'). Each specification starts with a
1112letter and may include other information after the letter to define some
1113aspect of the data layout. The specifications accepted are as follows: </p>
1114<dl>
1115 <dt><tt>E</tt></dt>
1116 <dd>Specifies that the target lays out data in big-endian form. That is, the
1117 bits with the most significance have the lowest address location.</dd>
1118 <dt><tt>e</tt></dt>
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]1119 <dd>Specifies that the target lays out data in little-endian form. That is,
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1120 the bits with the least significance have the lowest address location.</dd>
1121 <dt><tt>p:<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1122 <dd>This specifies the <i>size</i> of a pointer and its <i>abi</i> and
1123 <i>preferred</i> alignments. All sizes are in bits. Specifying the <i>pref</i>
1124 alignment is optional. If omitted, the preceding <tt>:</tt> should be omitted
1125 too.</dd>
1126 <dt><tt>i<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1127 <dd>This specifies the alignment for an integer type of a given bit
1128 <i>size</i>. The value of <i>size</i> must be in the range [1,2^23).</dd>
1129 <dt><tt>v<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1130 <dd>This specifies the alignment for a vector type of a given bit
1131 <i>size</i>.</dd>
1132 <dt><tt>f<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1133 <dd>This specifies the alignment for a floating point type of a given bit
1134 <i>size</i>. The value of <i>size</i> must be either 32 (float) or 64
1135 (double).</dd>
1136 <dt><tt>a<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1137 <dd>This specifies the alignment for an aggregate type of a given bit
1138 <i>size</i>.</dd>
1139</dl>
1140<p>When constructing the data layout for a given target, LLVM starts with a
1141default set of specifications which are then (possibly) overriden by the
1142specifications in the <tt>datalayout</tt> keyword. The default specifications
1143are given in this list:</p>
1144<ul>
1145 <li><tt>E</tt> - big endian</li>
1146 <li><tt>p:32:64:64</tt> - 32-bit pointers with 64-bit alignment</li>
1147 <li><tt>i1:8:8</tt> - i1 is 8-bit (byte) aligned</li>
1148 <li><tt>i8:8:8</tt> - i8 is 8-bit (byte) aligned</li>
1149 <li><tt>i16:16:16</tt> - i16 is 16-bit aligned</li>
1150 <li><tt>i32:32:32</tt> - i32 is 32-bit aligned</li>
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]1151 <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]1152 alignment of 64-bits</li>
1153 <li><tt>f32:32:32</tt> - float is 32-bit aligned</li>
1154 <li><tt>f64:64:64</tt> - double is 64-bit aligned</li>
1155 <li><tt>v64:64:64</tt> - 64-bit vector is 64-bit aligned</li>
1156 <li><tt>v128:128:128</tt> - 128-bit vector is 128-bit aligned</li>
1157 <li><tt>a0:0:1</tt> - aggregates are 8-bit aligned</li>
1158</ul>
Chris Lattner6a9f3c42008-08-05 18:21:08 +0000[diff] [blame]1159<p>When LLVM is determining the alignment for a given type, it uses the
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]1160following rules:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1161<ol>
1162 <li>If the type sought is an exact match for one of the specifications, that
1163 specification is used.</li>
1164 <li>If no match is found, and the type sought is an integer type, then the
1165 smallest integer type that is larger than the bitwidth of the sought type is
1166 used. If none of the specifications are larger than the bitwidth then the the
1167 largest integer type is used. For example, given the default specifications
1168 above, the i7 type will use the alignment of i8 (next largest) while both
1169 i65 and i256 will use the alignment of i64 (largest specified).</li>
1170 <li>If no match is found, and the type sought is a vector type, then the
1171 largest vector type that is smaller than the sought vector type will be used
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]1172 as a fall back. This happens because &lt;128 x double&gt; can be implemented
1173 in terms of 64 &lt;2 x double&gt;, for example.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1174</ol>
1175</div>
1176
1177<!-- *********************************************************************** -->
1178<div class="doc_section"> <a name="typesystem">Type System</a> </div>
1179<!-- *********************************************************************** -->
1180
1181<div class="doc_text">
1182
1183<p>The LLVM type system is one of the most important features of the
1184intermediate representation. Being typed enables a number of
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]1185optimizations to be performed on the intermediate representation directly,
1186without having to do
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1187extra analyses on the side before the transformation. A strong type
1188system makes it easier to read the generated code and enables novel
1189analyses and transformations that are not feasible to perform on normal
1190three address code representations.</p>
1191
1192</div>
1193
1194<!-- ======================================================================= -->
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1195<div class="doc_subsection"> <a name="t_classifications">Type
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1196Classifications</a> </div>
1197<div class="doc_text">
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1198<p>The types fall into a few useful
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1199classifications:</p>
1200
1201<table border="1" cellspacing="0" cellpadding="4">
1202 <tbody>
1203 <tr><th>Classification</th><th>Types</th></tr>
1204 <tr>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1205 <td><a href="#t_integer">integer</a></td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1206 <td><tt>i1, i2, i3, ... i8, ... i16, ... i32, ... i64, ... </tt></td>
1207 </tr>
1208 <tr>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1209 <td><a href="#t_floating">floating point</a></td>
1210 <td><tt>float, double, x86_fp80, fp128, ppc_fp128</tt></td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1211 </tr>
1212 <tr>
1213 <td><a name="t_firstclass">first class</a></td>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1214 <td><a href="#t_integer">integer</a>,
1215 <a href="#t_floating">floating point</a>,
1216 <a href="#t_pointer">pointer</a>,
Dan Gohmanf6237db2008-06-18 18:42:13 +0000[diff] [blame]1217 <a href="#t_vector">vector</a>,
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]1218 <a href="#t_struct">structure</a>,
1219 <a href="#t_array">array</a>,
Dan Gohmand13951e2008-05-23 22:50:26 +0000[diff] [blame]1220 <a href="#t_label">label</a>.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1221 </td>
1222 </tr>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1223 <tr>
1224 <td><a href="#t_primitive">primitive</a></td>
1225 <td><a href="#t_label">label</a>,
1226 <a href="#t_void">void</a>,
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1227 <a href="#t_floating">floating point</a>.</td>
1228 </tr>
1229 <tr>
1230 <td><a href="#t_derived">derived</a></td>
1231 <td><a href="#t_integer">integer</a>,
1232 <a href="#t_array">array</a>,
1233 <a href="#t_function">function</a>,
1234 <a href="#t_pointer">pointer</a>,
1235 <a href="#t_struct">structure</a>,
1236 <a href="#t_pstruct">packed structure</a>,
1237 <a href="#t_vector">vector</a>,
1238 <a href="#t_opaque">opaque</a>.
Dan Gohman032ba852008-10-14 16:32:04 +0000[diff] [blame]1239 </td>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1240 </tr>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1241 </tbody>
1242</table>
1243
1244<p>The <a href="#t_firstclass">first class</a> types are perhaps the
1245most important. Values of these types are the only ones which can be
1246produced by instructions, passed as arguments, or used as operands to
Dan Gohman4f29e422008-05-23 21:53:15 +0000[diff] [blame]1247instructions.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1248</div>
1249
1250<!-- ======================================================================= -->
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1251<div class="doc_subsection"> <a name="t_primitive">Primitive Types</a> </div>
Chris Lattner86437612008-01-04 04:34:14 +0000[diff] [blame]1252
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1253<div class="doc_text">
1254<p>The primitive types are the fundamental building blocks of the LLVM
1255system.</p>
1256
Chris Lattner86437612008-01-04 04:34:14 +0000[diff] [blame]1257</div>
1258
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1259<!-- _______________________________________________________________________ -->
1260<div class="doc_subsubsection"> <a name="t_floating">Floating Point Types</a> </div>
1261
1262<div class="doc_text">
1263 <table>
1264 <tbody>
1265 <tr><th>Type</th><th>Description</th></tr>
1266 <tr><td><tt>float</tt></td><td>32-bit floating point value</td></tr>
1267 <tr><td><tt>double</tt></td><td>64-bit floating point value</td></tr>
1268 <tr><td><tt>fp128</tt></td><td>128-bit floating point value (112-bit mantissa)</td></tr>
1269 <tr><td><tt>x86_fp80</tt></td><td>80-bit floating point value (X87)</td></tr>
1270 <tr><td><tt>ppc_fp128</tt></td><td>128-bit floating point value (two 64-bits)</td></tr>
1271 </tbody>
1272 </table>
1273</div>
1274
1275<!-- _______________________________________________________________________ -->
1276<div class="doc_subsubsection"> <a name="t_void">Void Type</a> </div>
1277
1278<div class="doc_text">
1279<h5>Overview:</h5>
1280<p>The void type does not represent any value and has no size.</p>
1281
1282<h5>Syntax:</h5>
1283
1284<pre>
1285 void
1286</pre>
1287</div>
1288
1289<!-- _______________________________________________________________________ -->
1290<div class="doc_subsubsection"> <a name="t_label">Label Type</a> </div>
1291
1292<div class="doc_text">
1293<h5>Overview:</h5>
1294<p>The label type represents code labels.</p>
1295
1296<h5>Syntax:</h5>
1297
1298<pre>
1299 label
1300</pre>
1301</div>
1302
1303
1304<!-- ======================================================================= -->
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1305<div class="doc_subsection"> <a name="t_derived">Derived Types</a> </div>
1306
1307<div class="doc_text">
1308
1309<p>The real power in LLVM comes from the derived types in the system.
1310This is what allows a programmer to represent arrays, functions,
1311pointers, and other useful types. Note that these derived types may be
1312recursive: For example, it is possible to have a two dimensional array.</p>
1313
1314</div>
1315
1316<!-- _______________________________________________________________________ -->
1317<div class="doc_subsubsection"> <a name="t_integer">Integer Type</a> </div>
1318
1319<div class="doc_text">
1320
1321<h5>Overview:</h5>
1322<p>The integer type is a very simple derived type that simply specifies an
1323arbitrary bit width for the integer type desired. Any bit width from 1 bit to
13242^23-1 (about 8 million) can be specified.</p>
1325
1326<h5>Syntax:</h5>
1327
1328<pre>
1329 iN
1330</pre>
1331
1332<p>The number of bits the integer will occupy is specified by the <tt>N</tt>
1333value.</p>
1334
1335<h5>Examples:</h5>
1336<table class="layout">
Chris Lattner251ab812007-12-18 06:18:21 +0000[diff] [blame]1337 <tbody>
1338 <tr>
1339 <td><tt>i1</tt></td>
1340 <td>a single-bit integer.</td>
1341 </tr><tr>
1342 <td><tt>i32</tt></td>
1343 <td>a 32-bit integer.</td>
1344 </tr><tr>
1345 <td><tt>i1942652</tt></td>
1346 <td>a really big integer of over 1 million bits.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1347 </tr>
Chris Lattner251ab812007-12-18 06:18:21 +0000[diff] [blame]1348 </tbody>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1349</table>
djge93155c2009-01-24 15:58:40 +0000[diff] [blame]1350
1351<p>Note that the code generator does not yet support large integer types
1352to be used as function return types. The specific limit on how large a
1353return type the code generator can currently handle is target-dependent;
1354currently it's often 64 bits for 32-bit targets and 128 bits for 64-bit
1355targets.</p>
1356
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1357</div>
1358
1359<!-- _______________________________________________________________________ -->
1360<div class="doc_subsubsection"> <a name="t_array">Array Type</a> </div>
1361
1362<div class="doc_text">
1363
1364<h5>Overview:</h5>
1365
1366<p>The array type is a very simple derived type that arranges elements
1367sequentially in memory. The array type requires a size (number of
1368elements) and an underlying data type.</p>
1369
1370<h5>Syntax:</h5>
1371
1372<pre>
1373 [&lt;# elements&gt; x &lt;elementtype&gt;]
1374</pre>
1375
1376<p>The number of elements is a constant integer value; elementtype may
1377be any type with a size.</p>
1378
1379<h5>Examples:</h5>
1380<table class="layout">
1381 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1382 <td class="left"><tt>[40 x i32]</tt></td>
1383 <td class="left">Array of 40 32-bit integer values.</td>
1384 </tr>
1385 <tr class="layout">
1386 <td class="left"><tt>[41 x i32]</tt></td>
1387 <td class="left">Array of 41 32-bit integer values.</td>
1388 </tr>
1389 <tr class="layout">
1390 <td class="left"><tt>[4 x i8]</tt></td>
1391 <td class="left">Array of 4 8-bit integer values.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1392 </tr>
1393</table>
1394<p>Here are some examples of multidimensional arrays:</p>
1395<table class="layout">
1396 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1397 <td class="left"><tt>[3 x [4 x i32]]</tt></td>
1398 <td class="left">3x4 array of 32-bit integer values.</td>
1399 </tr>
1400 <tr class="layout">
1401 <td class="left"><tt>[12 x [10 x float]]</tt></td>
1402 <td class="left">12x10 array of single precision floating point values.</td>
1403 </tr>
1404 <tr class="layout">
1405 <td class="left"><tt>[2 x [3 x [4 x i16]]]</tt></td>
1406 <td class="left">2x3x4 array of 16-bit integer values.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1407 </tr>
1408</table>
1409
1410<p>Note that 'variable sized arrays' can be implemented in LLVM with a zero
1411length array. Normally, accesses past the end of an array are undefined in
1412LLVM (e.g. it is illegal to access the 5th element of a 3 element array).
1413As a special case, however, zero length arrays are recognized to be variable
1414length. This allows implementation of 'pascal style arrays' with the LLVM
1415type "{ i32, [0 x float]}", for example.</p>
1416
djge93155c2009-01-24 15:58:40 +0000[diff] [blame]1417<p>Note that the code generator does not yet support large aggregate types
1418to be used as function return types. The specific limit on how large an
1419aggregate return type the code generator can currently handle is
1420target-dependent, and also dependent on the aggregate element types.</p>
1421
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1422</div>
1423
1424<!-- _______________________________________________________________________ -->
1425<div class="doc_subsubsection"> <a name="t_function">Function Type</a> </div>
1426<div class="doc_text">
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1427
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1428<h5>Overview:</h5>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1429
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1430<p>The function type can be thought of as a function signature. It
Devang Pateld4ba41d2008-03-24 05:35:41 +0000[diff] [blame]1431consists of a return type and a list of formal parameter types. The
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1432return 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]1433If 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]1434class types, and the struct must have at least one element.</p>
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]1435
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1436<h5>Syntax:</h5>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1437
1438<pre>
1439 &lt;returntype list&gt; (&lt;parameter list&gt;)
1440</pre>
1441
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1442<p>...where '<tt>&lt;parameter list&gt;</tt>' is a comma-separated list of type
1443specifiers. Optionally, the parameter list may include a type <tt>...</tt>,
1444which indicates that the function takes a variable number of arguments.
1445Variable argument functions can access their arguments with the <a
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]1446 href="#int_varargs">variable argument handling intrinsic</a> functions.
1447'<tt>&lt;returntype list&gt;</tt>' is a comma-separated list of
1448<a href="#t_firstclass">first class</a> type specifiers.</p>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1449
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1450<h5>Examples:</h5>
1451<table class="layout">
1452 <tr class="layout">
1453 <td class="left"><tt>i32 (i32)</tt></td>
1454 <td class="left">function taking an <tt>i32</tt>, returning an <tt>i32</tt>
1455 </td>
1456 </tr><tr class="layout">
Reid Spencerf234bed2007-07-19 23:13:04 +0000[diff] [blame]1457 <td class="left"><tt>float&nbsp;(i16&nbsp;signext,&nbsp;i32&nbsp;*)&nbsp;*
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1458 </tt></td>
1459 <td class="left"><a href="#t_pointer">Pointer</a> to a function that takes
1460 an <tt>i16</tt> that should be sign extended and a
1461 <a href="#t_pointer">pointer</a> to <tt>i32</tt>, returning
1462 <tt>float</tt>.
1463 </td>
1464 </tr><tr class="layout">
1465 <td class="left"><tt>i32 (i8*, ...)</tt></td>
1466 <td class="left">A vararg function that takes at least one
1467 <a href="#t_pointer">pointer</a> to <tt>i8 </tt> (char in C),
1468 which returns an integer. This is the signature for <tt>printf</tt> in
1469 LLVM.
1470 </td>
Devang Pateld4ba41d2008-03-24 05:35:41 +0000[diff] [blame]1471 </tr><tr class="layout">
1472 <td class="left"><tt>{i32, i32} (i32)</tt></td>
Misha Brukmanafc88b02008-11-27 06:41:20 +0000[diff] [blame]1473 <td class="left">A function taking an <tt>i32</tt>, returning two
1474 <tt>i32</tt> values as an aggregate of type <tt>{ i32, i32 }</tt>
Devang Pateld4ba41d2008-03-24 05:35:41 +0000[diff] [blame]1475 </td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1476 </tr>
1477</table>
1478
1479</div>
1480<!-- _______________________________________________________________________ -->
1481<div class="doc_subsubsection"> <a name="t_struct">Structure Type</a> </div>
1482<div class="doc_text">
1483<h5>Overview:</h5>
1484<p>The structure type is used to represent a collection of data members
1485together in memory. The packing of the field types is defined to match
1486the ABI of the underlying processor. The elements of a structure may
1487be any type that has a size.</p>
1488<p>Structures are accessed using '<tt><a href="#i_load">load</a></tt>
1489and '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a
1490field with the '<tt><a href="#i_getelementptr">getelementptr</a></tt>'
1491instruction.</p>
1492<h5>Syntax:</h5>
1493<pre> { &lt;type list&gt; }<br></pre>
1494<h5>Examples:</h5>
1495<table class="layout">
1496 <tr class="layout">
1497 <td class="left"><tt>{ i32, i32, i32 }</tt></td>
1498 <td class="left">A triple of three <tt>i32</tt> values</td>
1499 </tr><tr class="layout">
1500 <td class="left"><tt>{&nbsp;float,&nbsp;i32&nbsp;(i32)&nbsp;*&nbsp;}</tt></td>
1501 <td class="left">A pair, where the first element is a <tt>float</tt> and the
1502 second element is a <a href="#t_pointer">pointer</a> to a
1503 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
1504 an <tt>i32</tt>.</td>
1505 </tr>
1506</table>
djge93155c2009-01-24 15:58:40 +0000[diff] [blame]1507
1508<p>Note that the code generator does not yet support large aggregate types
1509to be used as function return types. The specific limit on how large an
1510aggregate return type the code generator can currently handle is
1511target-dependent, and also dependent on the aggregate element types.</p>
1512
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1513</div>
1514
1515<!-- _______________________________________________________________________ -->
1516<div class="doc_subsubsection"> <a name="t_pstruct">Packed Structure Type</a>
1517</div>
1518<div class="doc_text">
1519<h5>Overview:</h5>
1520<p>The packed structure type is used to represent a collection of data members
1521together in memory. There is no padding between fields. Further, the alignment
1522of a packed structure is 1 byte. The elements of a packed structure may
1523be any type that has a size.</p>
1524<p>Structures are accessed using '<tt><a href="#i_load">load</a></tt>
1525and '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a
1526field with the '<tt><a href="#i_getelementptr">getelementptr</a></tt>'
1527instruction.</p>
1528<h5>Syntax:</h5>
1529<pre> &lt; { &lt;type list&gt; } &gt; <br></pre>
1530<h5>Examples:</h5>
1531<table class="layout">
1532 <tr class="layout">
1533 <td class="left"><tt>&lt; { i32, i32, i32 } &gt;</tt></td>
1534 <td class="left">A triple of three <tt>i32</tt> values</td>
1535 </tr><tr class="layout">
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]1536 <td class="left">
1537<tt>&lt;&nbsp;{&nbsp;float,&nbsp;i32&nbsp;(i32)*&nbsp;}&nbsp;&gt;</tt></td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1538 <td class="left">A pair, where the first element is a <tt>float</tt> and the
1539 second element is a <a href="#t_pointer">pointer</a> to a
1540 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
1541 an <tt>i32</tt>.</td>
1542 </tr>
1543</table>
1544</div>
1545
1546<!-- _______________________________________________________________________ -->
1547<div class="doc_subsubsection"> <a name="t_pointer">Pointer Type</a> </div>
1548<div class="doc_text">
1549<h5>Overview:</h5>
1550<p>As in many languages, the pointer type represents a pointer or
Christopher Lambdd0049d2007-12-11 09:31:00 +0000[diff] [blame]1551reference to another object, which must live in memory. Pointer types may have
1552an optional address space attribute defining the target-specific numbered
1553address space where the pointed-to object resides. The default address space is
1554zero.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1555<h5>Syntax:</h5>
1556<pre> &lt;type&gt; *<br></pre>
1557<h5>Examples:</h5>
1558<table class="layout">
1559 <tr class="layout">
Dan Gohman01852382009-01-04 23:44:43 +0000[diff] [blame]1560 <td class="left"><tt>[4 x i32]*</tt></td>
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1561 <td class="left">A <a href="#t_pointer">pointer</a> to <a
1562 href="#t_array">array</a> of four <tt>i32</tt> values.</td>
1563 </tr>
1564 <tr class="layout">
1565 <td class="left"><tt>i32 (i32 *) *</tt></td>
1566 <td class="left"> A <a href="#t_pointer">pointer</a> to a <a
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1567 href="#t_function">function</a> that takes an <tt>i32*</tt>, returning an
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1568 <tt>i32</tt>.</td>
1569 </tr>
1570 <tr class="layout">
1571 <td class="left"><tt>i32 addrspace(5)*</tt></td>
1572 <td class="left">A <a href="#t_pointer">pointer</a> to an <tt>i32</tt> value
1573 that resides in address space #5.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1574 </tr>
1575</table>
1576</div>
1577
1578<!-- _______________________________________________________________________ -->
1579<div class="doc_subsubsection"> <a name="t_vector">Vector Type</a> </div>
1580<div class="doc_text">
1581
1582<h5>Overview:</h5>
1583
1584<p>A vector type is a simple derived type that represents a vector
1585of elements. Vector types are used when multiple primitive data
1586are operated in parallel using a single instruction (SIMD).
1587A vector type requires a size (number of
1588elements) and an underlying primitive data type. Vectors must have a power
1589of two length (1, 2, 4, 8, 16 ...). Vector types are
1590considered <a href="#t_firstclass">first class</a>.</p>
1591
1592<h5>Syntax:</h5>
1593
1594<pre>
1595 &lt; &lt;# elements&gt; x &lt;elementtype&gt; &gt;
1596</pre>
1597
1598<p>The number of elements is a constant integer value; elementtype may
1599be any integer or floating point type.</p>
1600
1601<h5>Examples:</h5>
1602
1603<table class="layout">
1604 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1605 <td class="left"><tt>&lt;4 x i32&gt;</tt></td>
1606 <td class="left">Vector of 4 32-bit integer values.</td>
1607 </tr>
1608 <tr class="layout">
1609 <td class="left"><tt>&lt;8 x float&gt;</tt></td>
1610 <td class="left">Vector of 8 32-bit floating-point values.</td>
1611 </tr>
1612 <tr class="layout">
1613 <td class="left"><tt>&lt;2 x i64&gt;</tt></td>
1614 <td class="left">Vector of 2 64-bit integer values.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1615 </tr>
1616</table>
djge93155c2009-01-24 15:58:40 +0000[diff] [blame]1617
1618<p>Note that the code generator does not yet support large vector types
1619to be used as function return types. The specific limit on how large a
1620vector return type codegen can currently handle is target-dependent;
1621currently it's often a few times longer than a hardware vector register.</p>
1622
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1623</div>
1624
1625<!-- _______________________________________________________________________ -->
1626<div class="doc_subsubsection"> <a name="t_opaque">Opaque Type</a> </div>
1627<div class="doc_text">
1628
1629<h5>Overview:</h5>
1630
1631<p>Opaque types are used to represent unknown types in the system. This
Gordon Henriksenda0706e2007-10-14 00:34:53 +0000[diff] [blame]1632corresponds (for example) to the C notion of a forward declared structure type.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1633In LLVM, opaque types can eventually be resolved to any type (not just a
1634structure type).</p>
1635
1636<h5>Syntax:</h5>
1637
1638<pre>
1639 opaque
1640</pre>
1641
1642<h5>Examples:</h5>
1643
1644<table class="layout">
1645 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1646 <td class="left"><tt>opaque</tt></td>
1647 <td class="left">An opaque type.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1648 </tr>
1649</table>
1650</div>
1651
Chris Lattner515195a2009-02-02 07:32:36 +0000[diff] [blame]1652<!-- ======================================================================= -->
1653<div class="doc_subsection">
1654 <a name="t_uprefs">Type Up-references</a>
1655</div>
1656
1657<div class="doc_text">
1658<h5>Overview:</h5>
1659<p>
1660An "up reference" allows you to refer to a lexically enclosing type without
1661requiring it to have a name. For instance, a structure declaration may contain a
1662pointer to any of the types it is lexically a member of. Example of up
1663references (with their equivalent as named type declarations) include:</p>
1664
1665<pre>
1666 { \2 * } %x = type { %t* }
1667 { \2 }* %y = type { %y }*
1668 \1* %z = type %z*
1669</pre>
1670
1671<p>
1672An up reference is needed by the asmprinter for printing out cyclic types when
1673there is no declared name for a type in the cycle. Because the asmprinter does
1674not want to print out an infinite type string, it needs a syntax to handle
1675recursive types that have no names (all names are optional in llvm IR).
1676</p>
1677
1678<h5>Syntax:</h5>
1679<pre>
1680 \&lt;level&gt;
1681</pre>
1682
1683<p>
1684The level is the count of the lexical type that is being referred to.
1685</p>
1686
1687<h5>Examples:</h5>
1688
1689<table class="layout">
1690 <tr class="layout">
1691 <td class="left"><tt>\1*</tt></td>
1692 <td class="left">Self-referential pointer.</td>
1693 </tr>
1694 <tr class="layout">
1695 <td class="left"><tt>{ { \3*, i8 }, i32 }</tt></td>
1696 <td class="left">Recursive structure where the upref refers to the out-most
1697 structure.</td>
1698 </tr>
1699</table>
1700</div>
1701
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1702
1703<!-- *********************************************************************** -->
1704<div class="doc_section"> <a name="constants">Constants</a> </div>
1705<!-- *********************************************************************** -->
1706
1707<div class="doc_text">
1708
1709<p>LLVM has several different basic types of constants. This section describes
1710them all and their syntax.</p>
1711
1712</div>
1713
1714<!-- ======================================================================= -->
1715<div class="doc_subsection"><a name="simpleconstants">Simple Constants</a></div>
1716
1717<div class="doc_text">
1718
1719<dl>
1720 <dt><b>Boolean constants</b></dt>
1721
1722 <dd>The two strings '<tt>true</tt>' and '<tt>false</tt>' are both valid
1723 constants of the <tt><a href="#t_primitive">i1</a></tt> type.
1724 </dd>
1725
1726 <dt><b>Integer constants</b></dt>
1727
1728 <dd>Standard integers (such as '4') are constants of the <a
1729 href="#t_integer">integer</a> type. Negative numbers may be used with
1730 integer types.
1731 </dd>
1732
1733 <dt><b>Floating point constants</b></dt>
1734
1735 <dd>Floating point constants use standard decimal notation (e.g. 123.421),
1736 exponential notation (e.g. 1.23421e+2), or a more precise hexadecimal
Chris Lattnerdc15b1d2008-04-01 18:45:27 +0000[diff] [blame]1737 notation (see below). The assembler requires the exact decimal value of
1738 a floating-point constant. For example, the assembler accepts 1.25 but
1739 rejects 1.3 because 1.3 is a repeating decimal in binary. Floating point
1740 constants must have a <a href="#t_floating">floating point</a> type. </dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1741
1742 <dt><b>Null pointer constants</b></dt>
1743
1744 <dd>The identifier '<tt>null</tt>' is recognized as a null pointer constant
1745 and must be of <a href="#t_pointer">pointer type</a>.</dd>
1746
1747</dl>
1748
1749<p>The one non-intuitive notation for constants is the optional hexadecimal form
1750of floating point constants. For example, the form '<tt>double
17510x432ff973cafa8000</tt>' is equivalent to (but harder to read than) '<tt>double
17524.5e+15</tt>'. The only time hexadecimal floating point constants are required
1753(and the only time that they are generated by the disassembler) is when a
1754floating point constant must be emitted but it cannot be represented as a
1755decimal floating point number. For example, NaN's, infinities, and other
1756special values are represented in their IEEE hexadecimal format so that
1757assembly and disassembly do not cause any bits to change in the constants.</p>
1758
1759</div>
1760
1761<!-- ======================================================================= -->
1762<div class="doc_subsection"><a name="aggregateconstants">Aggregate Constants</a>
1763</div>
1764
1765<div class="doc_text">
1766<p>Aggregate constants arise from aggregation of simple constants
1767and smaller aggregate constants.</p>
1768
1769<dl>
1770 <dt><b>Structure constants</b></dt>
1771
1772 <dd>Structure constants are represented with notation similar to structure
1773 type definitions (a comma separated list of elements, surrounded by braces
Chris Lattner6c8de962007-12-25 20:34:52 +0000[diff] [blame]1774 (<tt>{}</tt>)). For example: "<tt>{ i32 4, float 17.0, i32* @G }</tt>",
1775 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]1776 must have <a href="#t_struct">structure type</a>, and the number and
1777 types of elements must match those specified by the type.
1778 </dd>
1779
1780 <dt><b>Array constants</b></dt>
1781
1782 <dd>Array constants are represented with notation similar to array type
1783 definitions (a comma separated list of elements, surrounded by square brackets
1784 (<tt>[]</tt>)). For example: "<tt>[ i32 42, i32 11, i32 74 ]</tt>". Array
1785 constants must have <a href="#t_array">array type</a>, and the number and
1786 types of elements must match those specified by the type.
1787 </dd>
1788
1789 <dt><b>Vector constants</b></dt>
1790
1791 <dd>Vector constants are represented with notation similar to vector type
1792 definitions (a comma separated list of elements, surrounded by
1793 less-than/greater-than's (<tt>&lt;&gt;</tt>)). For example: "<tt>&lt; i32 42,
1794 i32 11, i32 74, i32 100 &gt;</tt>". Vector constants must have <a
1795 href="#t_vector">vector type</a>, and the number and types of elements must
1796 match those specified by the type.
1797 </dd>
1798
1799 <dt><b>Zero initialization</b></dt>
1800
1801 <dd>The string '<tt>zeroinitializer</tt>' can be used to zero initialize a
1802 value to zero of <em>any</em> type, including scalar and aggregate types.
1803 This is often used to avoid having to print large zero initializers (e.g. for
1804 large arrays) and is always exactly equivalent to using explicit zero
1805 initializers.
1806 </dd>
1807</dl>
1808
1809</div>
1810
1811<!-- ======================================================================= -->
1812<div class="doc_subsection">
1813 <a name="globalconstants">Global Variable and Function Addresses</a>
1814</div>
1815
1816<div class="doc_text">
1817
1818<p>The addresses of <a href="#globalvars">global variables</a> and <a
1819href="#functionstructure">functions</a> are always implicitly valid (link-time)
1820constants. These constants are explicitly referenced when the <a
1821href="#identifiers">identifier for the global</a> is used and always have <a
1822href="#t_pointer">pointer</a> type. For example, the following is a legal LLVM
1823file:</p>
1824
1825<div class="doc_code">
1826<pre>
1827@X = global i32 17
1828@Y = global i32 42
1829@Z = global [2 x i32*] [ i32* @X, i32* @Y ]
1830</pre>
1831</div>
1832
1833</div>
1834
1835<!-- ======================================================================= -->
1836<div class="doc_subsection"><a name="undefvalues">Undefined Values</a></div>
1837<div class="doc_text">
1838 <p>The string '<tt>undef</tt>' is recognized as a type-less constant that has
1839 no specific value. Undefined values may be of any type and be used anywhere
1840 a constant is permitted.</p>
1841
1842 <p>Undefined values indicate to the compiler that the program is well defined
1843 no matter what value is used, giving the compiler more freedom to optimize.
1844 </p>
1845</div>
1846
1847<!-- ======================================================================= -->
1848<div class="doc_subsection"><a name="constantexprs">Constant Expressions</a>
1849</div>
1850
1851<div class="doc_text">
1852
1853<p>Constant expressions are used to allow expressions involving other constants
1854to be used as constants. Constant expressions may be of any <a
1855href="#t_firstclass">first class</a> type and may involve any LLVM operation
1856that does not have side effects (e.g. load and call are not supported). The
1857following is the syntax for constant expressions:</p>
1858
1859<dl>
1860 <dt><b><tt>trunc ( CST to TYPE )</tt></b></dt>
1861 <dd>Truncate a constant to another type. The bit size of CST must be larger
1862 than the bit size of TYPE. Both types must be integers.</dd>
1863
1864 <dt><b><tt>zext ( CST to TYPE )</tt></b></dt>
1865 <dd>Zero extend a constant to another type. The bit size of CST must be
1866 smaller or equal to the bit size of TYPE. Both types must be integers.</dd>
1867
1868 <dt><b><tt>sext ( CST to TYPE )</tt></b></dt>
1869 <dd>Sign extend a constant to another type. The bit size of CST must be
1870 smaller or equal to the bit size of TYPE. Both types must be integers.</dd>
1871
1872 <dt><b><tt>fptrunc ( CST to TYPE )</tt></b></dt>
1873 <dd>Truncate a floating point constant to another floating point type. The
1874 size of CST must be larger than the size of TYPE. Both types must be
1875 floating point.</dd>
1876
1877 <dt><b><tt>fpext ( CST to TYPE )</tt></b></dt>
1878 <dd>Floating point extend a constant to another type. The size of CST must be
1879 smaller or equal to the size of TYPE. Both types must be floating point.</dd>
1880
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]1881 <dt><b><tt>fptoui ( CST to TYPE )</tt></b></dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1882 <dd>Convert a floating point constant to the corresponding unsigned integer
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]1883 constant. TYPE must be a scalar or vector integer type. CST must be of scalar
1884 or vector floating point type. Both CST and TYPE must be scalars, or vectors
1885 of the same number of elements. If the value won't fit in the integer type,
1886 the results are undefined.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1887
1888 <dt><b><tt>fptosi ( CST to TYPE )</tt></b></dt>
1889 <dd>Convert a floating point constant to the corresponding signed integer
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]1890 constant. TYPE must be a scalar or vector integer type. CST must be of scalar
1891 or vector floating point type. Both CST and TYPE must be scalars, or vectors
1892 of the same number of elements. If the value won't fit in the integer type,
1893 the results are undefined.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1894
1895 <dt><b><tt>uitofp ( CST to TYPE )</tt></b></dt>
1896 <dd>Convert an unsigned integer constant to the corresponding floating point
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]1897 constant. TYPE must be a scalar or vector floating point type. CST must be of
1898 scalar or vector integer type. Both CST and TYPE must be scalars, or vectors
1899 of the same number of elements. If the value won't fit in the floating point
1900 type, the results are undefined.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1901
1902 <dt><b><tt>sitofp ( CST to TYPE )</tt></b></dt>
1903 <dd>Convert a signed integer constant to the corresponding floating point
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]1904 constant. TYPE must be a scalar or vector floating point type. CST must be of
1905 scalar or vector integer type. Both CST and TYPE must be scalars, or vectors
1906 of the same number of elements. If the value won't fit in the floating point
1907 type, the results are undefined.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1908
1909 <dt><b><tt>ptrtoint ( CST to TYPE )</tt></b></dt>
1910 <dd>Convert a pointer typed constant to the corresponding integer constant
1911 TYPE must be an integer type. CST must be of pointer type. The CST value is
1912 zero extended, truncated, or unchanged to make it fit in TYPE.</dd>
1913
1914 <dt><b><tt>inttoptr ( CST to TYPE )</tt></b></dt>
1915 <dd>Convert a integer constant to a pointer constant. TYPE must be a
1916 pointer type. CST must be of integer type. The CST value is zero extended,
1917 truncated, or unchanged to make it fit in a pointer size. This one is
1918 <i>really</i> dangerous!</dd>
1919
1920 <dt><b><tt>bitcast ( CST to TYPE )</tt></b></dt>
1921 <dd>Convert a constant, CST, to another TYPE. The size of CST and TYPE must be
1922 identical (same number of bits). The conversion is done as if the CST value
1923 was stored to memory and read back as TYPE. In other words, no bits change
1924 with this operator, just the type. This can be used for conversion of
1925 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]1926 pointers it is only valid to cast to another pointer type. It is not valid
1927 to bitcast to or from an aggregate type.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1928 </dd>
1929
1930 <dt><b><tt>getelementptr ( CSTPTR, IDX0, IDX1, ... )</tt></b></dt>
1931
1932 <dd>Perform the <a href="#i_getelementptr">getelementptr operation</a> on
1933 constants. As with the <a href="#i_getelementptr">getelementptr</a>
1934 instruction, the index list may have zero or more indexes, which are required
1935 to make sense for the type of "CSTPTR".</dd>
1936
1937 <dt><b><tt>select ( COND, VAL1, VAL2 )</tt></b></dt>
1938
1939 <dd>Perform the <a href="#i_select">select operation</a> on
1940 constants.</dd>
1941
1942 <dt><b><tt>icmp COND ( VAL1, VAL2 )</tt></b></dt>
1943 <dd>Performs the <a href="#i_icmp">icmp operation</a> on constants.</dd>
1944
1945 <dt><b><tt>fcmp COND ( VAL1, VAL2 )</tt></b></dt>
1946 <dd>Performs the <a href="#i_fcmp">fcmp operation</a> on constants.</dd>
1947
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]1948 <dt><b><tt>vicmp COND ( VAL1, VAL2 )</tt></b></dt>
1949 <dd>Performs the <a href="#i_vicmp">vicmp operation</a> on constants.</dd>
1950
1951 <dt><b><tt>vfcmp COND ( VAL1, VAL2 )</tt></b></dt>
1952 <dd>Performs the <a href="#i_vfcmp">vfcmp operation</a> on constants.</dd>
1953
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1954 <dt><b><tt>extractelement ( VAL, IDX )</tt></b></dt>
1955
1956 <dd>Perform the <a href="#i_extractelement">extractelement
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]1957 operation</a> on constants.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1958
1959 <dt><b><tt>insertelement ( VAL, ELT, IDX )</tt></b></dt>
1960
1961 <dd>Perform the <a href="#i_insertelement">insertelement
1962 operation</a> on constants.</dd>
1963
1964
1965 <dt><b><tt>shufflevector ( VEC1, VEC2, IDXMASK )</tt></b></dt>
1966
1967 <dd>Perform the <a href="#i_shufflevector">shufflevector
1968 operation</a> on constants.</dd>
1969
1970 <dt><b><tt>OPCODE ( LHS, RHS )</tt></b></dt>
1971
1972 <dd>Perform the specified operation of the LHS and RHS constants. OPCODE may
1973 be any of the <a href="#binaryops">binary</a> or <a href="#bitwiseops">bitwise
1974 binary</a> operations. The constraints on operands are the same as those for
1975 the corresponding instruction (e.g. no bitwise operations on floating point
1976 values are allowed).</dd>
1977</dl>
1978</div>
1979
1980<!-- *********************************************************************** -->
1981<div class="doc_section"> <a name="othervalues">Other Values</a> </div>
1982<!-- *********************************************************************** -->
1983
1984<!-- ======================================================================= -->
1985<div class="doc_subsection">
1986<a name="inlineasm">Inline Assembler Expressions</a>
1987</div>
1988
1989<div class="doc_text">
1990
1991<p>
1992LLVM supports inline assembler expressions (as opposed to <a href="#moduleasm">
1993Module-Level Inline Assembly</a>) through the use of a special value. This
1994value represents the inline assembler as a string (containing the instructions
1995to emit), a list of operand constraints (stored as a string), and a flag that
1996indicates whether or not the inline asm expression has side effects. An example
1997inline assembler expression is:
1998</p>
1999
2000<div class="doc_code">
2001<pre>
2002i32 (i32) asm "bswap $0", "=r,r"
2003</pre>
2004</div>
2005
2006<p>
2007Inline assembler expressions may <b>only</b> be used as the callee operand of
2008a <a href="#i_call"><tt>call</tt> instruction</a>. Thus, typically we have:
2009</p>
2010
2011<div class="doc_code">
2012<pre>
2013%X = call i32 asm "<a href="#int_bswap">bswap</a> $0", "=r,r"(i32 %Y)
2014</pre>
2015</div>
2016
2017<p>
2018Inline asms with side effects not visible in the constraint list must be marked
2019as having side effects. This is done through the use of the
2020'<tt>sideeffect</tt>' keyword, like so:
2021</p>
2022
2023<div class="doc_code">
2024<pre>
2025call void asm sideeffect "eieio", ""()
2026</pre>
2027</div>
2028
2029<p>TODO: The format of the asm and constraints string still need to be
2030documented here. Constraints on what can be done (e.g. duplication, moving, etc
Chris Lattner1cdd64e2008-10-04 18:36:02 +0000[diff] [blame]2031need to be documented). This is probably best done by reference to another
2032document that covers inline asm from a holistic perspective.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2033</p>
2034
2035</div>
2036
2037<!-- *********************************************************************** -->
2038<div class="doc_section"> <a name="instref">Instruction Reference</a> </div>
2039<!-- *********************************************************************** -->
2040
2041<div class="doc_text">
2042
2043<p>The LLVM instruction set consists of several different
2044classifications of instructions: <a href="#terminators">terminator
2045instructions</a>, <a href="#binaryops">binary instructions</a>,
2046<a href="#bitwiseops">bitwise binary instructions</a>, <a
2047 href="#memoryops">memory instructions</a>, and <a href="#otherops">other
2048instructions</a>.</p>
2049
2050</div>
2051
2052<!-- ======================================================================= -->
2053<div class="doc_subsection"> <a name="terminators">Terminator
2054Instructions</a> </div>
2055
2056<div class="doc_text">
2057
2058<p>As mentioned <a href="#functionstructure">previously</a>, every
2059basic block in a program ends with a "Terminator" instruction, which
2060indicates which block should be executed after the current block is
2061finished. These terminator instructions typically yield a '<tt>void</tt>'
2062value: they produce control flow, not values (the one exception being
2063the '<a href="#i_invoke"><tt>invoke</tt></a>' instruction).</p>
2064<p>There are six different terminator instructions: the '<a
2065 href="#i_ret"><tt>ret</tt></a>' instruction, the '<a href="#i_br"><tt>br</tt></a>'
2066instruction, the '<a href="#i_switch"><tt>switch</tt></a>' instruction,
2067the '<a href="#i_invoke"><tt>invoke</tt></a>' instruction, the '<a
2068 href="#i_unwind"><tt>unwind</tt></a>' instruction, and the '<a
2069 href="#i_unreachable"><tt>unreachable</tt></a>' instruction.</p>
2070
2071</div>
2072
2073<!-- _______________________________________________________________________ -->
2074<div class="doc_subsubsection"> <a name="i_ret">'<tt>ret</tt>'
2075Instruction</a> </div>
2076<div class="doc_text">
2077<h5>Syntax:</h5>
Dan Gohman3e700032008-10-04 19:00:07 +0000[diff] [blame]2078<pre>
2079 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]2080 ret void <i>; Return from void function</i>
2081</pre>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]2082
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2083<h5>Overview:</h5>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]2084
Dan Gohman3e700032008-10-04 19:00:07 +0000[diff] [blame]2085<p>The '<tt>ret</tt>' instruction is used to return control flow (and
2086optionally a value) from a function back to the caller.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2087<p>There are two forms of the '<tt>ret</tt>' instruction: one that
Dan Gohman3e700032008-10-04 19:00:07 +0000[diff] [blame]2088returns a value and then causes control flow, and one that just causes
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2089control flow to occur.</p>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]2090
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2091<h5>Arguments:</h5>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]2092
Dan Gohman3e700032008-10-04 19:00:07 +0000[diff] [blame]2093<p>The '<tt>ret</tt>' instruction optionally accepts a single argument,
2094the return value. The type of the return value must be a
2095'<a href="#t_firstclass">first class</a>' type.</p>
2096
2097<p>A function is not <a href="#wellformed">well formed</a> if
2098it it has a non-void return type and contains a '<tt>ret</tt>'
2099instruction with no return value or a return value with a type that
2100does not match its type, or if it has a void return type and contains
2101a '<tt>ret</tt>' instruction with a return value.</p>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]2102
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2103<h5>Semantics:</h5>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]2104
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2105<p>When the '<tt>ret</tt>' instruction is executed, control flow
2106returns back to the calling function's context. If the caller is a "<a
2107 href="#i_call"><tt>call</tt></a>" instruction, execution continues at
2108the instruction after the call. If the caller was an "<a
2109 href="#i_invoke"><tt>invoke</tt></a>" instruction, execution continues
2110at the beginning of the "normal" destination block. If the instruction
2111returns a value, that value shall set the call or invoke instruction's
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]2112return value.</p>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]2113
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2114<h5>Example:</h5>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]2115
2116<pre>
2117 ret i32 5 <i>; Return an integer value of 5</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2118 ret void <i>; Return from a void function</i>
Dan Gohman3e700032008-10-04 19:00:07 +0000[diff] [blame]2119 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]2120</pre>
Dan Gohman60967192009-01-12 23:12:39 +0000[diff] [blame]2121
djge93155c2009-01-24 15:58:40 +0000[diff] [blame]2122<p>Note that the code generator does not yet fully support large
2123 return values. The specific sizes that are currently supported are
2124 dependent on the target. For integers, on 32-bit targets the limit
2125 is often 64 bits, and on 64-bit targets the limit is often 128 bits.
2126 For aggregate types, the current limits are dependent on the element
2127 types; for example targets are often limited to 2 total integer
2128 elements and 2 total floating-point elements.</p>
Dan Gohman60967192009-01-12 23:12:39 +0000[diff] [blame]2129
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2130</div>
2131<!-- _______________________________________________________________________ -->
2132<div class="doc_subsubsection"> <a name="i_br">'<tt>br</tt>' Instruction</a> </div>
2133<div class="doc_text">
2134<h5>Syntax:</h5>
2135<pre> br i1 &lt;cond&gt;, label &lt;iftrue&gt;, label &lt;iffalse&gt;<br> br label &lt;dest&gt; <i>; Unconditional branch</i>
2136</pre>
2137<h5>Overview:</h5>
2138<p>The '<tt>br</tt>' instruction is used to cause control flow to
2139transfer to a different basic block in the current function. There are
2140two forms of this instruction, corresponding to a conditional branch
2141and an unconditional branch.</p>
2142<h5>Arguments:</h5>
2143<p>The conditional branch form of the '<tt>br</tt>' instruction takes a
2144single '<tt>i1</tt>' value and two '<tt>label</tt>' values. The
2145unconditional form of the '<tt>br</tt>' instruction takes a single
2146'<tt>label</tt>' value as a target.</p>
2147<h5>Semantics:</h5>
2148<p>Upon execution of a conditional '<tt>br</tt>' instruction, the '<tt>i1</tt>'
2149argument is evaluated. If the value is <tt>true</tt>, control flows
2150to the '<tt>iftrue</tt>' <tt>label</tt> argument. If "cond" is <tt>false</tt>,
2151control flows to the '<tt>iffalse</tt>' <tt>label</tt> argument.</p>
2152<h5>Example:</h5>
2153<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
2154 href="#i_ret">ret</a> i32 1<br>IfUnequal:<br> <a href="#i_ret">ret</a> i32 0<br></pre>
2155</div>
2156<!-- _______________________________________________________________________ -->
2157<div class="doc_subsubsection">
2158 <a name="i_switch">'<tt>switch</tt>' Instruction</a>
2159</div>
2160
2161<div class="doc_text">
2162<h5>Syntax:</h5>
2163
2164<pre>
2165 switch &lt;intty&gt; &lt;value&gt;, label &lt;defaultdest&gt; [ &lt;intty&gt; &lt;val&gt;, label &lt;dest&gt; ... ]
2166</pre>
2167
2168<h5>Overview:</h5>
2169
2170<p>The '<tt>switch</tt>' instruction is used to transfer control flow to one of
2171several different places. It is a generalization of the '<tt>br</tt>'
2172instruction, allowing a branch to occur to one of many possible
2173destinations.</p>
2174
2175
2176<h5>Arguments:</h5>
2177
2178<p>The '<tt>switch</tt>' instruction uses three parameters: an integer
2179comparison value '<tt>value</tt>', a default '<tt>label</tt>' destination, and
2180an array of pairs of comparison value constants and '<tt>label</tt>'s. The
2181table is not allowed to contain duplicate constant entries.</p>
2182
2183<h5>Semantics:</h5>
2184
2185<p>The <tt>switch</tt> instruction specifies a table of values and
2186destinations. When the '<tt>switch</tt>' instruction is executed, this
2187table is searched for the given value. If the value is found, control flow is
2188transfered to the corresponding destination; otherwise, control flow is
2189transfered to the default destination.</p>
2190
2191<h5>Implementation:</h5>
2192
2193<p>Depending on properties of the target machine and the particular
2194<tt>switch</tt> instruction, this instruction may be code generated in different
2195ways. For example, it could be generated as a series of chained conditional
2196branches or with a lookup table.</p>
2197
2198<h5>Example:</h5>
2199
2200<pre>
2201 <i>; Emulate a conditional br instruction</i>
2202 %Val = <a href="#i_zext">zext</a> i1 %value to i32
Dan Gohman01852382009-01-04 23:44:43 +0000[diff] [blame]2203 switch i32 %Val, label %truedest [ i32 0, label %falsedest ]
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2204
2205 <i>; Emulate an unconditional br instruction</i>
2206 switch i32 0, label %dest [ ]
2207
2208 <i>; Implement a jump table:</i>
Dan Gohman01852382009-01-04 23:44:43 +0000[diff] [blame]2209 switch i32 %val, label %otherwise [ i32 0, label %onzero
2210 i32 1, label %onone
2211 i32 2, label %ontwo ]
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2212</pre>
2213</div>
2214
2215<!-- _______________________________________________________________________ -->
2216<div class="doc_subsubsection">
2217 <a name="i_invoke">'<tt>invoke</tt>' Instruction</a>
2218</div>
2219
2220<div class="doc_text">
2221
2222<h5>Syntax:</h5>
2223
2224<pre>
Devang Pateld0bfcc72008-10-07 17:48:33 +0000[diff] [blame]2225 &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]2226 to label &lt;normal label&gt; unwind label &lt;exception label&gt;
2227</pre>
2228
2229<h5>Overview:</h5>
2230
2231<p>The '<tt>invoke</tt>' instruction causes control to transfer to a specified
2232function, with the possibility of control flow transfer to either the
2233'<tt>normal</tt>' label or the
2234'<tt>exception</tt>' label. If the callee function returns with the
2235"<tt><a href="#i_ret">ret</a></tt>" instruction, control flow will return to the
2236"normal" label. If the callee (or any indirect callees) returns with the "<a
2237href="#i_unwind"><tt>unwind</tt></a>" instruction, control is interrupted and
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]2238continued at the dynamically nearest "exception" label.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2239
2240<h5>Arguments:</h5>
2241
2242<p>This instruction requires several arguments:</p>
2243
2244<ol>
2245 <li>
2246 The optional "cconv" marker indicates which <a href="#callingconv">calling
2247 convention</a> the call should use. If none is specified, the call defaults
2248 to using C calling conventions.
2249 </li>
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]2250
2251 <li>The optional <a href="#paramattrs">Parameter Attributes</a> list for
2252 return values. Only '<tt>zeroext</tt>', '<tt>signext</tt>',
2253 and '<tt>inreg</tt>' attributes are valid here.</li>
2254
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2255 <li>'<tt>ptr to function ty</tt>': shall be the signature of the pointer to
2256 function value being invoked. In most cases, this is a direct function
2257 invocation, but indirect <tt>invoke</tt>s are just as possible, branching off
2258 an arbitrary pointer to function value.
2259 </li>
2260
2261 <li>'<tt>function ptr val</tt>': An LLVM value containing a pointer to a
2262 function to be invoked. </li>
2263
2264 <li>'<tt>function args</tt>': argument list whose types match the function
2265 signature argument types. If the function signature indicates the function
2266 accepts a variable number of arguments, the extra arguments can be
2267 specified. </li>
2268
2269 <li>'<tt>normal label</tt>': the label reached when the called function
2270 executes a '<tt><a href="#i_ret">ret</a></tt>' instruction. </li>
2271
2272 <li>'<tt>exception label</tt>': the label reached when a callee returns with
2273 the <a href="#i_unwind"><tt>unwind</tt></a> instruction. </li>
2274
Devang Pateld0bfcc72008-10-07 17:48:33 +0000[diff] [blame]2275 <li>The optional <a href="#fnattrs">function attributes</a> list. Only
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]2276 '<tt>noreturn</tt>', '<tt>nounwind</tt>', '<tt>readonly</tt>' and
2277 '<tt>readnone</tt>' attributes are valid here.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2278</ol>
2279
2280<h5>Semantics:</h5>
2281
2282<p>This instruction is designed to operate as a standard '<tt><a
2283href="#i_call">call</a></tt>' instruction in most regards. The primary
2284difference is that it establishes an association with a label, which is used by
2285the runtime library to unwind the stack.</p>
2286
2287<p>This instruction is used in languages with destructors to ensure that proper
2288cleanup is performed in the case of either a <tt>longjmp</tt> or a thrown
2289exception. Additionally, this is important for implementation of
2290'<tt>catch</tt>' clauses in high-level languages that support them.</p>
2291
2292<h5>Example:</h5>
2293<pre>
Nick Lewyckya1c11a12008-03-16 07:18:12 +0000[diff] [blame]2294 %retval = invoke i32 @Test(i32 15) to label %Continue
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2295 unwind label %TestCleanup <i>; {i32}:retval set</i>
Nick Lewyckya1c11a12008-03-16 07:18:12 +0000[diff] [blame]2296 %retval = invoke <a href="#callingconv">coldcc</a> i32 %Testfnptr(i32 15) to label %Continue
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2297 unwind label %TestCleanup <i>; {i32}:retval set</i>
2298</pre>
2299</div>
2300
2301
2302<!-- _______________________________________________________________________ -->
2303
2304<div class="doc_subsubsection"> <a name="i_unwind">'<tt>unwind</tt>'
2305Instruction</a> </div>
2306
2307<div class="doc_text">
2308
2309<h5>Syntax:</h5>
2310<pre>
2311 unwind
2312</pre>
2313
2314<h5>Overview:</h5>
2315
2316<p>The '<tt>unwind</tt>' instruction unwinds the stack, continuing control flow
2317at the first callee in the dynamic call stack which used an <a
2318href="#i_invoke"><tt>invoke</tt></a> instruction to perform the call. This is
2319primarily used to implement exception handling.</p>
2320
2321<h5>Semantics:</h5>
2322
Chris Lattner8b094fc2008-04-19 21:01:16 +0000[diff] [blame]2323<p>The '<tt>unwind</tt>' instruction causes execution of the current function to
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2324immediately halt. The dynamic call stack is then searched for the first <a
2325href="#i_invoke"><tt>invoke</tt></a> instruction on the call stack. Once found,
2326execution continues at the "exceptional" destination block specified by the
2327<tt>invoke</tt> instruction. If there is no <tt>invoke</tt> instruction in the
2328dynamic call chain, undefined behavior results.</p>
2329</div>
2330
2331<!-- _______________________________________________________________________ -->
2332
2333<div class="doc_subsubsection"> <a name="i_unreachable">'<tt>unreachable</tt>'
2334Instruction</a> </div>
2335
2336<div class="doc_text">
2337
2338<h5>Syntax:</h5>
2339<pre>
2340 unreachable
2341</pre>
2342
2343<h5>Overview:</h5>
2344
2345<p>The '<tt>unreachable</tt>' instruction has no defined semantics. This
2346instruction is used to inform the optimizer that a particular portion of the
2347code is not reachable. This can be used to indicate that the code after a
2348no-return function cannot be reached, and other facts.</p>
2349
2350<h5>Semantics:</h5>
2351
2352<p>The '<tt>unreachable</tt>' instruction has no defined semantics.</p>
2353</div>
2354
2355
2356
2357<!-- ======================================================================= -->
2358<div class="doc_subsection"> <a name="binaryops">Binary Operations</a> </div>
2359<div class="doc_text">
2360<p>Binary operators are used to do most of the computation in a
Chris Lattnerab596d92008-04-01 18:47:32 +0000[diff] [blame]2361program. They require two operands of the same type, execute an operation on them, and
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2362produce a single value. The operands might represent
2363multiple 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]2364The result value has the same type as its operands.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2365<p>There are several different binary operators:</p>
2366</div>
2367<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2368<div class="doc_subsubsection">
2369 <a name="i_add">'<tt>add</tt>' Instruction</a>
2370</div>
2371
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2372<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2373
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2374<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2375
2376<pre>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2377 &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]2378</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2379
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2380<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2381
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2382<p>The '<tt>add</tt>' instruction returns the sum of its two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2383
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2384<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2385
2386<p>The two arguments to the '<tt>add</tt>' instruction must be <a
2387 href="#t_integer">integer</a>, <a href="#t_floating">floating point</a>, or
2388 <a href="#t_vector">vector</a> values. Both arguments must have identical
2389 types.</p>
2390
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2391<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2392
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2393<p>The value produced is the integer or floating point sum of the two
2394operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2395
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2396<p>If an integer sum has unsigned overflow, the result returned is the
2397mathematical result modulo 2<sup>n</sup>, where n is the bit width of
2398the result.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2399
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2400<p>Because LLVM integers use a two's complement representation, this
2401instruction is appropriate for both signed and unsigned integers.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2402
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2403<h5>Example:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2404
2405<pre>
2406 &lt;result&gt; = add i32 4, %var <i>; yields {i32}:result = 4 + %var</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2407</pre>
2408</div>
2409<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2410<div class="doc_subsubsection">
2411 <a name="i_sub">'<tt>sub</tt>' Instruction</a>
2412</div>
2413
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2414<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2415
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2416<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2417
2418<pre>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2419 &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]2420</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2421
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2422<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2423
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2424<p>The '<tt>sub</tt>' instruction returns the difference of its two
2425operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2426
2427<p>Note that the '<tt>sub</tt>' instruction is used to represent the
2428'<tt>neg</tt>' instruction present in most other intermediate
2429representations.</p>
2430
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2431<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2432
2433<p>The two arguments to the '<tt>sub</tt>' instruction must be <a
2434 href="#t_integer">integer</a>, <a href="#t_floating">floating point</a>,
2435 or <a href="#t_vector">vector</a> values. Both arguments must have identical
2436 types.</p>
2437
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2438<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2439
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2440<p>The value produced is the integer or floating point difference of
2441the two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2442
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2443<p>If an integer difference has unsigned overflow, the result returned is the
2444mathematical result modulo 2<sup>n</sup>, where n is the bit width of
2445the result.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2446
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2447<p>Because LLVM integers use a two's complement representation, this
2448instruction is appropriate for both signed and unsigned integers.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2449
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2450<h5>Example:</h5>
2451<pre>
2452 &lt;result&gt; = sub i32 4, %var <i>; yields {i32}:result = 4 - %var</i>
2453 &lt;result&gt; = sub i32 0, %val <i>; yields {i32}:result = -%var</i>
2454</pre>
2455</div>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2456
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2457<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2458<div class="doc_subsubsection">
2459 <a name="i_mul">'<tt>mul</tt>' Instruction</a>
2460</div>
2461
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2462<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2463
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2464<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2465<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]2466</pre>
2467<h5>Overview:</h5>
2468<p>The '<tt>mul</tt>' instruction returns the product of its two
2469operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2470
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2471<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2472
2473<p>The two arguments to the '<tt>mul</tt>' instruction must be <a
2474href="#t_integer">integer</a>, <a href="#t_floating">floating point</a>,
2475or <a href="#t_vector">vector</a> values. Both arguments must have identical
2476types.</p>
2477
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2478<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2479
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2480<p>The value produced is the integer or floating point product of the
2481two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2482
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2483<p>If the result of an integer multiplication has unsigned overflow,
2484the result returned is the mathematical result modulo
24852<sup>n</sup>, where n is the bit width of the result.</p>
2486<p>Because LLVM integers use a two's complement representation, and the
2487result is the same width as the operands, this instruction returns the
2488correct result for both signed and unsigned integers. If a full product
2489(e.g. <tt>i32</tt>x<tt>i32</tt>-><tt>i64</tt>) is needed, the operands
2490should be sign-extended or zero-extended as appropriate to the
2491width of the full product.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2492<h5>Example:</h5>
2493<pre> &lt;result&gt; = mul i32 4, %var <i>; yields {i32}:result = 4 * %var</i>
2494</pre>
2495</div>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2496
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2497<!-- _______________________________________________________________________ -->
2498<div class="doc_subsubsection"> <a name="i_udiv">'<tt>udiv</tt>' Instruction
2499</a></div>
2500<div class="doc_text">
2501<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2502<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]2503</pre>
2504<h5>Overview:</h5>
2505<p>The '<tt>udiv</tt>' instruction returns the quotient of its two
2506operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2507
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2508<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2509
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2510<p>The two arguments to the '<tt>udiv</tt>' instruction must be
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2511<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2512values. Both arguments must have identical types.</p>
2513
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2514<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2515
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2516<p>The value produced is the unsigned integer quotient of the two operands.</p>
2517<p>Note that unsigned integer division and signed integer division are distinct
2518operations; for signed integer division, use '<tt>sdiv</tt>'.</p>
2519<p>Division by zero leads to undefined behavior.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2520<h5>Example:</h5>
2521<pre> &lt;result&gt; = udiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
2522</pre>
2523</div>
2524<!-- _______________________________________________________________________ -->
2525<div class="doc_subsubsection"> <a name="i_sdiv">'<tt>sdiv</tt>' Instruction
2526</a> </div>
2527<div class="doc_text">
2528<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2529<pre>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2530 &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]2531</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2532
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2533<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2534
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2535<p>The '<tt>sdiv</tt>' instruction returns the quotient of its two
2536operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2537
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2538<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2539
2540<p>The two arguments to the '<tt>sdiv</tt>' instruction must be
2541<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2542values. Both arguments must have identical types.</p>
2543
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2544<h5>Semantics:</h5>
Chris Lattnerdc15b1d2008-04-01 18:45:27 +0000[diff] [blame]2545<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]2546<p>Note that signed integer division and unsigned integer division are distinct
2547operations; for unsigned integer division, use '<tt>udiv</tt>'.</p>
2548<p>Division by zero leads to undefined behavior. Overflow also leads to
2549undefined behavior; this is a rare case, but can occur, for example,
2550by doing a 32-bit division of -2147483648 by -1.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2551<h5>Example:</h5>
2552<pre> &lt;result&gt; = sdiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
2553</pre>
2554</div>
2555<!-- _______________________________________________________________________ -->
2556<div class="doc_subsubsection"> <a name="i_fdiv">'<tt>fdiv</tt>'
2557Instruction</a> </div>
2558<div class="doc_text">
2559<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2560<pre>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2561 &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]2562</pre>
2563<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2564
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2565<p>The '<tt>fdiv</tt>' instruction returns the quotient of its two
2566operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2567
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2568<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2569
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2570<p>The two arguments to the '<tt>fdiv</tt>' instruction must be
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2571<a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a>
2572of floating point values. Both arguments must have identical types.</p>
2573
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2574<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2575
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2576<p>The value produced is the floating point quotient of the two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2577
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2578<h5>Example:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2579
2580<pre>
2581 &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]2582</pre>
2583</div>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2584
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2585<!-- _______________________________________________________________________ -->
2586<div class="doc_subsubsection"> <a name="i_urem">'<tt>urem</tt>' Instruction</a>
2587</div>
2588<div class="doc_text">
2589<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2590<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]2591</pre>
2592<h5>Overview:</h5>
2593<p>The '<tt>urem</tt>' instruction returns the remainder from the
2594unsigned division of its two arguments.</p>
2595<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2596<p>The two arguments to the '<tt>urem</tt>' instruction must be
2597<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2598values. Both arguments must have identical types.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2599<h5>Semantics:</h5>
2600<p>This instruction returns the unsigned integer <i>remainder</i> of a division.
Chris Lattnerdc15b1d2008-04-01 18:45:27 +0000[diff] [blame]2601This instruction always performs an unsigned division to get the remainder.</p>
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2602<p>Note that unsigned integer remainder and signed integer remainder are
2603distinct operations; for signed integer remainder, use '<tt>srem</tt>'.</p>
2604<p>Taking the remainder of a division by zero leads to undefined behavior.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2605<h5>Example:</h5>
2606<pre> &lt;result&gt; = urem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
2607</pre>
2608
2609</div>
2610<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2611<div class="doc_subsubsection">
2612 <a name="i_srem">'<tt>srem</tt>' Instruction</a>
2613</div>
2614
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2615<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2616
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2617<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2618
2619<pre>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2620 &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]2621</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2622
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2623<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2624
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2625<p>The '<tt>srem</tt>' instruction returns the remainder from the
Dan Gohman3e3fd8c2007-11-05 23:35:22 +0000[diff] [blame]2626signed division of its two operands. This instruction can also take
2627<a href="#t_vector">vector</a> versions of the values in which case
2628the elements must be integers.</p>
Chris Lattner08497ce2008-01-04 04:33:49 +0000[diff] [blame]2629
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2630<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2631
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2632<p>The two arguments to the '<tt>srem</tt>' instruction must be
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2633<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2634values. Both arguments must have identical types.</p>
2635
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2636<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2637
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2638<p>This instruction returns the <i>remainder</i> of a division (where the result
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2639has the same sign as the dividend, <tt>op1</tt>), not the <i>modulo</i>
2640operator (where the result has the same sign as the divisor, <tt>op2</tt>) of
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2641a value. For more information about the difference, see <a
2642 href="http://mathforum.org/dr.math/problems/anne.4.28.99.html">The
2643Math Forum</a>. For a table of how this is implemented in various languages,
2644please see <a href="http://en.wikipedia.org/wiki/Modulo_operation">
2645Wikipedia: modulo operation</a>.</p>
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2646<p>Note that signed integer remainder and unsigned integer remainder are
2647distinct operations; for unsigned integer remainder, use '<tt>urem</tt>'.</p>
2648<p>Taking the remainder of a division by zero leads to undefined behavior.
2649Overflow also leads to undefined behavior; this is a rare case, but can occur,
2650for example, by taking the remainder of a 32-bit division of -2147483648 by -1.
2651(The remainder doesn't actually overflow, but this rule lets srem be
2652implemented using instructions that return both the result of the division
2653and the remainder.)</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2654<h5>Example:</h5>
2655<pre> &lt;result&gt; = srem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
2656</pre>
2657
2658</div>
2659<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2660<div class="doc_subsubsection">
2661 <a name="i_frem">'<tt>frem</tt>' Instruction</a> </div>
2662
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2663<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2664
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2665<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2666<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]2667</pre>
2668<h5>Overview:</h5>
2669<p>The '<tt>frem</tt>' instruction returns the remainder from the
2670division of its two operands.</p>
2671<h5>Arguments:</h5>
2672<p>The two arguments to the '<tt>frem</tt>' instruction must be
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2673<a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a>
2674of floating point values. Both arguments must have identical types.</p>
2675
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2676<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2677
Chris Lattnerdc15b1d2008-04-01 18:45:27 +0000[diff] [blame]2678<p>This instruction returns the <i>remainder</i> of a division.
2679The remainder has the same sign as the dividend.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2680
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2681<h5>Example:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2682
2683<pre>
2684 &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]2685</pre>
2686</div>
2687
2688<!-- ======================================================================= -->
2689<div class="doc_subsection"> <a name="bitwiseops">Bitwise Binary
2690Operations</a> </div>
2691<div class="doc_text">
2692<p>Bitwise binary operators are used to do various forms of
2693bit-twiddling in a program. They are generally very efficient
2694instructions and can commonly be strength reduced from other
Chris Lattnerdc15b1d2008-04-01 18:45:27 +0000[diff] [blame]2695instructions. They require two operands of the same type, execute an operation on them,
2696and 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]2697</div>
2698
2699<!-- _______________________________________________________________________ -->
2700<div class="doc_subsubsection"> <a name="i_shl">'<tt>shl</tt>'
2701Instruction</a> </div>
2702<div class="doc_text">
2703<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2704<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]2705</pre>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2706
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2707<h5>Overview:</h5>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2708
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2709<p>The '<tt>shl</tt>' instruction returns the first operand shifted to
2710the left a specified number of bits.</p>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2711
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2712<h5>Arguments:</h5>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2713
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2714<p>Both arguments to the '<tt>shl</tt>' instruction must be the same <a
Nate Begemanbb1ce942008-07-29 15:49:41 +0000[diff] [blame]2715 href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2716type. '<tt>op2</tt>' is treated as an unsigned value.</p>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2717
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2718<h5>Semantics:</h5>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2719
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2720<p>The value produced is <tt>op1</tt> * 2<sup><tt>op2</tt></sup> mod 2<sup>n</sup>,
2721where n is the width of the result. If <tt>op2</tt> is (statically or dynamically) negative or
Mon P Wangb0f51822008-12-10 08:55:09 +0000[diff] [blame]2722equal to or larger than the number of bits in <tt>op1</tt>, the result is undefined.
2723If the arguments are vectors, each vector element of <tt>op1</tt> is shifted by the
2724corresponding shift amount in <tt>op2</tt>.</p>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2725
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2726<h5>Example:</h5><pre>
2727 &lt;result&gt; = shl i32 4, %var <i>; yields {i32}: 4 &lt;&lt; %var</i>
2728 &lt;result&gt; = shl i32 4, 2 <i>; yields {i32}: 16</i>
2729 &lt;result&gt; = shl i32 1, 10 <i>; yields {i32}: 1024</i>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2730 &lt;result&gt; = shl i32 1, 32 <i>; undefined</i>
Mon P Wang9901e732008-12-09 05:46:39 +0000[diff] [blame]2731 &lt;result&gt; = shl &lt;2 x i32&gt; &lt; i32 1, i32 1&gt;, &lt; i32 1, i32 2&gt; <i>; yields: result=&lt;2 x i32&gt; &lt; i32 2, i32 4&gt;</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2732</pre>
2733</div>
2734<!-- _______________________________________________________________________ -->
2735<div class="doc_subsubsection"> <a name="i_lshr">'<tt>lshr</tt>'
2736Instruction</a> </div>
2737<div class="doc_text">
2738<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2739<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]2740</pre>
2741
2742<h5>Overview:</h5>
2743<p>The '<tt>lshr</tt>' instruction (logical shift right) returns the first
2744operand shifted to the right a specified number of bits with zero fill.</p>
2745
2746<h5>Arguments:</h5>
2747<p>Both arguments to the '<tt>lshr</tt>' instruction must be the same
Nate Begemanbb1ce942008-07-29 15:49:41 +0000[diff] [blame]2748<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]2749type. '<tt>op2</tt>' is treated as an unsigned value.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2750
2751<h5>Semantics:</h5>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2752
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2753<p>This instruction always performs a logical shift right operation. The most
2754significant bits of the result will be filled with zero bits after the
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2755shift. If <tt>op2</tt> is (statically or dynamically) equal to or larger than
Mon P Wangb0f51822008-12-10 08:55:09 +0000[diff] [blame]2756the number of bits in <tt>op1</tt>, the result is undefined. If the arguments are
2757vectors, each vector element of <tt>op1</tt> is shifted by the corresponding shift
2758amount in <tt>op2</tt>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2759
2760<h5>Example:</h5>
2761<pre>
2762 &lt;result&gt; = lshr i32 4, 1 <i>; yields {i32}:result = 2</i>
2763 &lt;result&gt; = lshr i32 4, 2 <i>; yields {i32}:result = 1</i>
2764 &lt;result&gt; = lshr i8 4, 3 <i>; yields {i8}:result = 0</i>
2765 &lt;result&gt; = lshr i8 -2, 1 <i>; yields {i8}:result = 0x7FFFFFFF </i>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2766 &lt;result&gt; = lshr i32 1, 32 <i>; undefined</i>
Mon P Wang9901e732008-12-09 05:46:39 +0000[diff] [blame]2767 &lt;result&gt; = lshr &lt;2 x i32&gt; &lt; i32 -2, i32 4&gt;, &lt; i32 1, i32 2&gt; <i>; yields: result=&lt;2 x i32&gt; &lt; i32 0x7FFFFFFF, i32 1&gt;</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2768</pre>
2769</div>
2770
2771<!-- _______________________________________________________________________ -->
2772<div class="doc_subsubsection"> <a name="i_ashr">'<tt>ashr</tt>'
2773Instruction</a> </div>
2774<div class="doc_text">
2775
2776<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2777<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]2778</pre>
2779
2780<h5>Overview:</h5>
2781<p>The '<tt>ashr</tt>' instruction (arithmetic shift right) returns the first
2782operand shifted to the right a specified number of bits with sign extension.</p>
2783
2784<h5>Arguments:</h5>
2785<p>Both arguments to the '<tt>ashr</tt>' instruction must be the same
Nate Begemanbb1ce942008-07-29 15:49:41 +0000[diff] [blame]2786<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]2787type. '<tt>op2</tt>' is treated as an unsigned value.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2788
2789<h5>Semantics:</h5>
2790<p>This instruction always performs an arithmetic shift right operation,
2791The most significant bits of the result will be filled with the sign bit
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2792of <tt>op1</tt>. If <tt>op2</tt> is (statically or dynamically) equal to or
Mon P Wangb0f51822008-12-10 08:55:09 +0000[diff] [blame]2793larger than the number of bits in <tt>op1</tt>, the result is undefined. If the
2794arguments are vectors, each vector element of <tt>op1</tt> is shifted by the
2795corresponding shift amount in <tt>op2</tt>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2796
2797<h5>Example:</h5>
2798<pre>
2799 &lt;result&gt; = ashr i32 4, 1 <i>; yields {i32}:result = 2</i>
2800 &lt;result&gt; = ashr i32 4, 2 <i>; yields {i32}:result = 1</i>
2801 &lt;result&gt; = ashr i8 4, 3 <i>; yields {i8}:result = 0</i>
2802 &lt;result&gt; = ashr i8 -2, 1 <i>; yields {i8}:result = -1</i>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2803 &lt;result&gt; = ashr i32 1, 32 <i>; undefined</i>
Mon P Wang9901e732008-12-09 05:46:39 +0000[diff] [blame]2804 &lt;result&gt; = ashr &lt;2 x i32&gt; &lt; i32 -2, i32 4&gt;, &lt; i32 1, i32 3&gt; <i>; yields: result=&lt;2 x i32&gt; &lt; i32 -1, i32 0&gt;</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2805</pre>
2806</div>
2807
2808<!-- _______________________________________________________________________ -->
2809<div class="doc_subsubsection"> <a name="i_and">'<tt>and</tt>'
2810Instruction</a> </div>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2811
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2812<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2813
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2814<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2815
2816<pre>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2817 &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]2818</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2819
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2820<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2821
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2822<p>The '<tt>and</tt>' instruction returns the bitwise logical and of
2823its two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2824
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2825<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2826
2827<p>The two arguments to the '<tt>and</tt>' instruction must be
2828<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2829values. Both arguments must have identical types.</p>
2830
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2831<h5>Semantics:</h5>
2832<p>The truth table used for the '<tt>and</tt>' instruction is:</p>
2833<p> </p>
Bill Wendling1e19eac2008-09-07 10:29:20 +0000[diff] [blame]2834<div>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2835<table border="1" cellspacing="0" cellpadding="4">
2836 <tbody>
2837 <tr>
2838 <td>In0</td>
2839 <td>In1</td>
2840 <td>Out</td>
2841 </tr>
2842 <tr>
2843 <td>0</td>
2844 <td>0</td>
2845 <td>0</td>
2846 </tr>
2847 <tr>
2848 <td>0</td>
2849 <td>1</td>
2850 <td>0</td>
2851 </tr>
2852 <tr>
2853 <td>1</td>
2854 <td>0</td>
2855 <td>0</td>
2856 </tr>
2857 <tr>
2858 <td>1</td>
2859 <td>1</td>
2860 <td>1</td>
2861 </tr>
2862 </tbody>
2863</table>
2864</div>
2865<h5>Example:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2866<pre>
2867 &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]2868 &lt;result&gt; = and i32 15, 40 <i>; yields {i32}:result = 8</i>
2869 &lt;result&gt; = and i32 4, 8 <i>; yields {i32}:result = 0</i>
2870</pre>
2871</div>
2872<!-- _______________________________________________________________________ -->
2873<div class="doc_subsubsection"> <a name="i_or">'<tt>or</tt>' Instruction</a> </div>
2874<div class="doc_text">
2875<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2876<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]2877</pre>
2878<h5>Overview:</h5>
2879<p>The '<tt>or</tt>' instruction returns the bitwise logical inclusive
2880or of its two operands.</p>
2881<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2882
2883<p>The two arguments to the '<tt>or</tt>' instruction must be
2884<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2885values. Both arguments must have identical types.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2886<h5>Semantics:</h5>
2887<p>The truth table used for the '<tt>or</tt>' instruction is:</p>
2888<p> </p>
Bill Wendling1e19eac2008-09-07 10:29:20 +0000[diff] [blame]2889<div>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2890<table border="1" cellspacing="0" cellpadding="4">
2891 <tbody>
2892 <tr>
2893 <td>In0</td>
2894 <td>In1</td>
2895 <td>Out</td>
2896 </tr>
2897 <tr>
2898 <td>0</td>
2899 <td>0</td>
2900 <td>0</td>
2901 </tr>
2902 <tr>
2903 <td>0</td>
2904 <td>1</td>
2905 <td>1</td>
2906 </tr>
2907 <tr>
2908 <td>1</td>
2909 <td>0</td>
2910 <td>1</td>
2911 </tr>
2912 <tr>
2913 <td>1</td>
2914 <td>1</td>
2915 <td>1</td>
2916 </tr>
2917 </tbody>
2918</table>
2919</div>
2920<h5>Example:</h5>
2921<pre> &lt;result&gt; = or i32 4, %var <i>; yields {i32}:result = 4 | %var</i>
2922 &lt;result&gt; = or i32 15, 40 <i>; yields {i32}:result = 47</i>
2923 &lt;result&gt; = or i32 4, 8 <i>; yields {i32}:result = 12</i>
2924</pre>
2925</div>
2926<!-- _______________________________________________________________________ -->
2927<div class="doc_subsubsection"> <a name="i_xor">'<tt>xor</tt>'
2928Instruction</a> </div>
2929<div class="doc_text">
2930<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2931<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]2932</pre>
2933<h5>Overview:</h5>
2934<p>The '<tt>xor</tt>' instruction returns the bitwise logical exclusive
2935or of its two operands. The <tt>xor</tt> is used to implement the
2936"one's complement" operation, which is the "~" operator in C.</p>
2937<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2938<p>The two arguments to the '<tt>xor</tt>' instruction must be
2939<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2940values. Both arguments must have identical types.</p>
2941
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2942<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2943
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2944<p>The truth table used for the '<tt>xor</tt>' instruction is:</p>
2945<p> </p>
Bill Wendling1e19eac2008-09-07 10:29:20 +0000[diff] [blame]2946<div>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2947<table border="1" cellspacing="0" cellpadding="4">
2948 <tbody>
2949 <tr>
2950 <td>In0</td>
2951 <td>In1</td>
2952 <td>Out</td>
2953 </tr>
2954 <tr>
2955 <td>0</td>
2956 <td>0</td>
2957 <td>0</td>
2958 </tr>
2959 <tr>
2960 <td>0</td>
2961 <td>1</td>
2962 <td>1</td>
2963 </tr>
2964 <tr>
2965 <td>1</td>
2966 <td>0</td>
2967 <td>1</td>
2968 </tr>
2969 <tr>
2970 <td>1</td>
2971 <td>1</td>
2972 <td>0</td>
2973 </tr>
2974 </tbody>
2975</table>
2976</div>
2977<p> </p>
2978<h5>Example:</h5>
2979<pre> &lt;result&gt; = xor i32 4, %var <i>; yields {i32}:result = 4 ^ %var</i>
2980 &lt;result&gt; = xor i32 15, 40 <i>; yields {i32}:result = 39</i>
2981 &lt;result&gt; = xor i32 4, 8 <i>; yields {i32}:result = 12</i>
2982 &lt;result&gt; = xor i32 %V, -1 <i>; yields {i32}:result = ~%V</i>
2983</pre>
2984</div>
2985
2986<!-- ======================================================================= -->
2987<div class="doc_subsection">
2988 <a name="vectorops">Vector Operations</a>
2989</div>
2990
2991<div class="doc_text">
2992
2993<p>LLVM supports several instructions to represent vector operations in a
2994target-independent manner. These instructions cover the element-access and
2995vector-specific operations needed to process vectors effectively. While LLVM
2996does directly support these vector operations, many sophisticated algorithms
2997will want to use target-specific intrinsics to take full advantage of a specific
2998target.</p>
2999
3000</div>
3001
3002<!-- _______________________________________________________________________ -->
3003<div class="doc_subsubsection">
3004 <a name="i_extractelement">'<tt>extractelement</tt>' Instruction</a>
3005</div>
3006
3007<div class="doc_text">
3008
3009<h5>Syntax:</h5>
3010
3011<pre>
3012 &lt;result&gt; = extractelement &lt;n x &lt;ty&gt;&gt; &lt;val&gt;, i32 &lt;idx&gt; <i>; yields &lt;ty&gt;</i>
3013</pre>
3014
3015<h5>Overview:</h5>
3016
3017<p>
3018The '<tt>extractelement</tt>' instruction extracts a single scalar
3019element from a vector at a specified index.
3020</p>
3021
3022
3023<h5>Arguments:</h5>
3024
3025<p>
3026The first operand of an '<tt>extractelement</tt>' instruction is a
3027value of <a href="#t_vector">vector</a> type. The second operand is
3028an index indicating the position from which to extract the element.
3029The index may be a variable.</p>
3030
3031<h5>Semantics:</h5>
3032
3033<p>
3034The result is a scalar of the same type as the element type of
3035<tt>val</tt>. Its value is the value at position <tt>idx</tt> of
3036<tt>val</tt>. If <tt>idx</tt> exceeds the length of <tt>val</tt>, the
3037results are undefined.
3038</p>
3039
3040<h5>Example:</h5>
3041
3042<pre>
3043 %result = extractelement &lt;4 x i32&gt; %vec, i32 0 <i>; yields i32</i>
3044</pre>
3045</div>
3046
3047
3048<!-- _______________________________________________________________________ -->
3049<div class="doc_subsubsection">
3050 <a name="i_insertelement">'<tt>insertelement</tt>' Instruction</a>
3051</div>
3052
3053<div class="doc_text">
3054
3055<h5>Syntax:</h5>
3056
3057<pre>
Dan Gohmanbcc3c502008-05-12 23:38:42 +0000[diff] [blame]3058 &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]3059</pre>
3060
3061<h5>Overview:</h5>
3062
3063<p>
3064The '<tt>insertelement</tt>' instruction inserts a scalar
3065element into a vector at a specified index.
3066</p>
3067
3068
3069<h5>Arguments:</h5>
3070
3071<p>
3072The first operand of an '<tt>insertelement</tt>' instruction is a
3073value of <a href="#t_vector">vector</a> type. The second operand is a
3074scalar value whose type must equal the element type of the first
3075operand. The third operand is an index indicating the position at
3076which to insert the value. The index may be a variable.</p>
3077
3078<h5>Semantics:</h5>
3079
3080<p>
3081The result is a vector of the same type as <tt>val</tt>. Its
3082element values are those of <tt>val</tt> except at position
3083<tt>idx</tt>, where it gets the value <tt>elt</tt>. If <tt>idx</tt>
3084exceeds the length of <tt>val</tt>, the results are undefined.
3085</p>
3086
3087<h5>Example:</h5>
3088
3089<pre>
3090 %result = insertelement &lt;4 x i32&gt; %vec, i32 1, i32 0 <i>; yields &lt;4 x i32&gt;</i>
3091</pre>
3092</div>
3093
3094<!-- _______________________________________________________________________ -->
3095<div class="doc_subsubsection">
3096 <a name="i_shufflevector">'<tt>shufflevector</tt>' Instruction</a>
3097</div>
3098
3099<div class="doc_text">
3100
3101<h5>Syntax:</h5>
3102
3103<pre>
Mon P Wangbff5d9c2008-11-10 04:46:22 +0000[diff] [blame]3104 &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]3105</pre>
3106
3107<h5>Overview:</h5>
3108
3109<p>
3110The '<tt>shufflevector</tt>' instruction constructs a permutation of elements
Mon P Wangbff5d9c2008-11-10 04:46:22 +0000[diff] [blame]3111from two input vectors, returning a vector with the same element type as
3112the input and length that is the same as the shuffle mask.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3113</p>
3114
3115<h5>Arguments:</h5>
3116
3117<p>
Mon P Wangbff5d9c2008-11-10 04:46:22 +0000[diff] [blame]3118The first two operands of a '<tt>shufflevector</tt>' instruction are vectors
3119with types that match each other. The third argument is a shuffle mask whose
3120element type is always 'i32'. The result of the instruction is a vector whose
3121length is the same as the shuffle mask and whose element type is the same as
3122the element type of the first two operands.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3123</p>
3124
3125<p>
3126The shuffle mask operand is required to be a constant vector with either
3127constant integer or undef values.
3128</p>
3129
3130<h5>Semantics:</h5>
3131
3132<p>
3133The elements of the two input vectors are numbered from left to right across
3134both of the vectors. The shuffle mask operand specifies, for each element of
Mon P Wangbff5d9c2008-11-10 04:46:22 +0000[diff] [blame]3135the result vector, which element of the two input vectors the result element
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3136gets. The element selector may be undef (meaning "don't care") and the second
3137operand may be undef if performing a shuffle from only one vector.
3138</p>
3139
3140<h5>Example:</h5>
3141
3142<pre>
3143 %result = shufflevector &lt;4 x i32&gt; %v1, &lt;4 x i32&gt; %v2,
3144 &lt;4 x i32&gt; &lt;i32 0, i32 4, i32 1, i32 5&gt; <i>; yields &lt;4 x i32&gt;</i>
3145 %result = shufflevector &lt;4 x i32&gt; %v1, &lt;4 x i32&gt; undef,
3146 &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]3147 %result = shufflevector &lt;8 x i32&gt; %v1, &lt;8 x i32&gt; undef,
3148 &lt;4 x i32&gt; &lt;i32 0, i32 1, i32 2, i32 3&gt; <i>; yields &lt;4 x i32&gt;</i>
3149 %result = shufflevector &lt;4 x i32&gt; %v1, &lt;4 x i32&gt; %v2,
3150 &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]3151</pre>
3152</div>
3153
3154
3155<!-- ======================================================================= -->
3156<div class="doc_subsection">
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3157 <a name="aggregateops">Aggregate Operations</a>
3158</div>
3159
3160<div class="doc_text">
3161
3162<p>LLVM supports several instructions for working with aggregate values.
3163</p>
3164
3165</div>
3166
3167<!-- _______________________________________________________________________ -->
3168<div class="doc_subsubsection">
3169 <a name="i_extractvalue">'<tt>extractvalue</tt>' Instruction</a>
3170</div>
3171
3172<div class="doc_text">
3173
3174<h5>Syntax:</h5>
3175
3176<pre>
3177 &lt;result&gt; = extractvalue &lt;aggregate type&gt; &lt;val&gt;, &lt;idx&gt;{, &lt;idx&gt;}*
3178</pre>
3179
3180<h5>Overview:</h5>
3181
3182<p>
Dan Gohman6c6dea02008-05-13 18:16:06 +0000[diff] [blame]3183The '<tt>extractvalue</tt>' instruction extracts the value of a struct field
3184or array element from an aggregate value.
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3185</p>
3186
3187
3188<h5>Arguments:</h5>
3189
3190<p>
3191The first operand of an '<tt>extractvalue</tt>' instruction is a
3192value of <a href="#t_struct">struct</a> or <a href="#t_array">array</a>
Dan Gohman6c6dea02008-05-13 18:16:06 +0000[diff] [blame]3193type. The operands are constant indices to specify which value to extract
Dan Gohmane5febe42008-05-31 00:58:22 +0000[diff] [blame]3194in a similar manner as indices in a
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3195'<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.
3196</p>
3197
3198<h5>Semantics:</h5>
3199
3200<p>
3201The result is the value at the position in the aggregate specified by
3202the index operands.
3203</p>
3204
3205<h5>Example:</h5>
3206
3207<pre>
Dan Gohmane5febe42008-05-31 00:58:22 +0000[diff] [blame]3208 %result = extractvalue {i32, float} %agg, 0 <i>; yields i32</i>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3209</pre>
3210</div>
3211
3212
3213<!-- _______________________________________________________________________ -->
3214<div class="doc_subsubsection">
3215 <a name="i_insertvalue">'<tt>insertvalue</tt>' Instruction</a>
3216</div>
3217
3218<div class="doc_text">
3219
3220<h5>Syntax:</h5>
3221
3222<pre>
Dan Gohmane5febe42008-05-31 00:58:22 +0000[diff] [blame]3223 &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]3224</pre>
3225
3226<h5>Overview:</h5>
3227
3228<p>
3229The '<tt>insertvalue</tt>' instruction inserts a value
Dan Gohman6c6dea02008-05-13 18:16:06 +0000[diff] [blame]3230into a struct field or array element in an aggregate.
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3231</p>
3232
3233
3234<h5>Arguments:</h5>
3235
3236<p>
3237The first operand of an '<tt>insertvalue</tt>' instruction is a
3238value of <a href="#t_struct">struct</a> or <a href="#t_array">array</a> type.
3239The second operand is a first-class value to insert.
Dan Gohman4f29e422008-05-23 21:53:15 +0000[diff] [blame]3240The following operands are constant indices
Dan Gohmane5febe42008-05-31 00:58:22 +0000[diff] [blame]3241indicating the position at which to insert the value in a similar manner as
Dan Gohman6c6dea02008-05-13 18:16:06 +0000[diff] [blame]3242indices in a
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3243'<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.
3244The value to insert must have the same type as the value identified
Dan Gohman6c6dea02008-05-13 18:16:06 +0000[diff] [blame]3245by the indices.
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]3246</p>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3247
3248<h5>Semantics:</h5>
3249
3250<p>
3251The result is an aggregate of the same type as <tt>val</tt>. Its
3252value is that of <tt>val</tt> except that the value at the position
Dan Gohman6c6dea02008-05-13 18:16:06 +0000[diff] [blame]3253specified by the indices is that of <tt>elt</tt>.
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3254</p>
3255
3256<h5>Example:</h5>
3257
3258<pre>
Dan Gohmanb1aab4e2008-06-23 15:26:37 +0000[diff] [blame]3259 %result = insertvalue {i32, float} %agg, i32 1, 0 <i>; yields {i32, float}</i>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3260</pre>
3261</div>
3262
3263
3264<!-- ======================================================================= -->
3265<div class="doc_subsection">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3266 <a name="memoryops">Memory Access and Addressing Operations</a>
3267</div>
3268
3269<div class="doc_text">
3270
3271<p>A key design point of an SSA-based representation is how it
3272represents memory. In LLVM, no memory locations are in SSA form, which
3273makes things very simple. This section describes how to read, write,
3274allocate, and free memory in LLVM.</p>
3275
3276</div>
3277
3278<!-- _______________________________________________________________________ -->
3279<div class="doc_subsubsection">
3280 <a name="i_malloc">'<tt>malloc</tt>' Instruction</a>
3281</div>
3282
3283<div class="doc_text">
3284
3285<h5>Syntax:</h5>
3286
3287<pre>
3288 &lt;result&gt; = malloc &lt;type&gt;[, i32 &lt;NumElements&gt;][, align &lt;alignment&gt;] <i>; yields {type*}:result</i>
3289</pre>
3290
3291<h5>Overview:</h5>
3292
3293<p>The '<tt>malloc</tt>' instruction allocates memory from the system
Christopher Lambcfe00962007-12-17 01:00:21 +0000[diff] [blame]3294heap and returns a pointer to it. The object is always allocated in the generic
3295address space (address space zero).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3296
3297<h5>Arguments:</h5>
3298
3299<p>The '<tt>malloc</tt>' instruction allocates
3300<tt>sizeof(&lt;type&gt;)*NumElements</tt>
3301bytes of memory from the operating system and returns a pointer of the
3302appropriate type to the program. If "NumElements" is specified, it is the
Gabor Greif5082cf42008-02-09 22:24:34 +0000[diff] [blame]3303number of elements allocated, otherwise "NumElements" is defaulted to be one.
Chris Lattner10368b62008-04-02 00:38:26 +0000[diff] [blame]3304If a constant alignment is specified, the value result of the allocation is guaranteed to
Gabor Greif5082cf42008-02-09 22:24:34 +0000[diff] [blame]3305be aligned to at least that boundary. If not specified, or if zero, the target can
3306choose to align the allocation on any convenient boundary.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3307
3308<p>'<tt>type</tt>' must be a sized type.</p>
3309
3310<h5>Semantics:</h5>
3311
3312<p>Memory is allocated using the system "<tt>malloc</tt>" function, and
Nick Lewyckyff384472008-11-24 03:41:24 +0000[diff] [blame]3313a pointer is returned. The result of a zero byte allocation is undefined. The
Chris Lattner8b094fc2008-04-19 21:01:16 +0000[diff] [blame]3314result is null if there is insufficient memory available.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3315
3316<h5>Example:</h5>
3317
3318<pre>
Dan Gohmanf54f50a2009-01-04 23:49:44 +0000[diff] [blame]3319 %array = malloc [4 x i8] <i>; yields {[%4 x i8]*}:array</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3320
3321 %size = <a href="#i_add">add</a> i32 2, 2 <i>; yields {i32}:size = i32 4</i>
3322 %array1 = malloc i8, i32 4 <i>; yields {i8*}:array1</i>
3323 %array2 = malloc [12 x i8], i32 %size <i>; yields {[12 x i8]*}:array2</i>
3324 %array3 = malloc i32, i32 4, align 1024 <i>; yields {i32*}:array3</i>
3325 %array4 = malloc i32, align 1024 <i>; yields {i32*}:array4</i>
3326</pre>
Dan Gohman60967192009-01-12 23:12:39 +0000[diff] [blame]3327
3328<p>Note that the code generator does not yet respect the
3329 alignment value.</p>
3330
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3331</div>
3332
3333<!-- _______________________________________________________________________ -->
3334<div class="doc_subsubsection">
3335 <a name="i_free">'<tt>free</tt>' Instruction</a>
3336</div>
3337
3338<div class="doc_text">
3339
3340<h5>Syntax:</h5>
3341
3342<pre>
Dan Gohmanf54f50a2009-01-04 23:49:44 +0000[diff] [blame]3343 free &lt;type&gt; &lt;value&gt; <i>; yields {void}</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3344</pre>
3345
3346<h5>Overview:</h5>
3347
3348<p>The '<tt>free</tt>' instruction returns memory back to the unused
3349memory heap to be reallocated in the future.</p>
3350
3351<h5>Arguments:</h5>
3352
3353<p>'<tt>value</tt>' shall be a pointer value that points to a value
3354that was allocated with the '<tt><a href="#i_malloc">malloc</a></tt>'
3355instruction.</p>
3356
3357<h5>Semantics:</h5>
3358
3359<p>Access to the memory pointed to by the pointer is no longer defined
Chris Lattner329d6532008-04-19 22:41:32 +0000[diff] [blame]3360after this instruction executes. If the pointer is null, the operation
3361is a noop.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3362
3363<h5>Example:</h5>
3364
3365<pre>
Dan Gohmanf54f50a2009-01-04 23:49:44 +0000[diff] [blame]3366 %array = <a href="#i_malloc">malloc</a> [4 x i8] <i>; yields {[4 x i8]*}:array</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3367 free [4 x i8]* %array
3368</pre>
3369</div>
3370
3371<!-- _______________________________________________________________________ -->
3372<div class="doc_subsubsection">
3373 <a name="i_alloca">'<tt>alloca</tt>' Instruction</a>
3374</div>
3375
3376<div class="doc_text">
3377
3378<h5>Syntax:</h5>
3379
3380<pre>
3381 &lt;result&gt; = alloca &lt;type&gt;[, i32 &lt;NumElements&gt;][, align &lt;alignment&gt;] <i>; yields {type*}:result</i>
3382</pre>
3383
3384<h5>Overview:</h5>
3385
3386<p>The '<tt>alloca</tt>' instruction allocates memory on the stack frame of the
3387currently executing function, to be automatically released when this function
Christopher Lambcfe00962007-12-17 01:00:21 +0000[diff] [blame]3388returns to its caller. The object is always allocated in the generic address
3389space (address space zero).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3390
3391<h5>Arguments:</h5>
3392
3393<p>The '<tt>alloca</tt>' instruction allocates <tt>sizeof(&lt;type&gt;)*NumElements</tt>
3394bytes of memory on the runtime stack, returning a pointer of the
Gabor Greif5082cf42008-02-09 22:24:34 +0000[diff] [blame]3395appropriate type to the program. If "NumElements" is specified, it is the
3396number of elements allocated, otherwise "NumElements" is defaulted to be one.
Chris Lattner10368b62008-04-02 00:38:26 +0000[diff] [blame]3397If a constant alignment is specified, the value result of the allocation is guaranteed
Gabor Greif5082cf42008-02-09 22:24:34 +0000[diff] [blame]3398to be aligned to at least that boundary. If not specified, or if zero, the target
3399can choose to align the allocation on any convenient boundary.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3400
3401<p>'<tt>type</tt>' may be any sized type.</p>
3402
3403<h5>Semantics:</h5>
3404
Chris Lattner8b094fc2008-04-19 21:01:16 +0000[diff] [blame]3405<p>Memory is allocated; a pointer is returned. The operation is undefiend if
3406there is insufficient stack space for the allocation. '<tt>alloca</tt>'d
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3407memory is automatically released when the function returns. The '<tt>alloca</tt>'
3408instruction is commonly used to represent automatic variables that must
3409have an address available. When the function returns (either with the <tt><a
3410 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]3411instructions), the memory is reclaimed. Allocating zero bytes
3412is legal, but the result is undefined.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3413
3414<h5>Example:</h5>
3415
3416<pre>
Dan Gohmanf54f50a2009-01-04 23:49:44 +0000[diff] [blame]3417 %ptr = alloca i32 <i>; yields {i32*}:ptr</i>
3418 %ptr = alloca i32, i32 4 <i>; yields {i32*}:ptr</i>
3419 %ptr = alloca i32, i32 4, align 1024 <i>; yields {i32*}:ptr</i>
3420 %ptr = alloca i32, align 1024 <i>; yields {i32*}:ptr</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3421</pre>
3422</div>
3423
3424<!-- _______________________________________________________________________ -->
3425<div class="doc_subsubsection"> <a name="i_load">'<tt>load</tt>'
3426Instruction</a> </div>
3427<div class="doc_text">
3428<h5>Syntax:</h5>
3429<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>
3430<h5>Overview:</h5>
3431<p>The '<tt>load</tt>' instruction is used to read from memory.</p>
3432<h5>Arguments:</h5>
3433<p>The argument to the '<tt>load</tt>' instruction specifies the memory
3434address from which to load. The pointer must point to a <a
3435 href="#t_firstclass">first class</a> type. If the <tt>load</tt> is
3436marked as <tt>volatile</tt>, then the optimizer is not allowed to modify
3437the number or order of execution of this <tt>load</tt> with other
3438volatile <tt>load</tt> and <tt><a href="#i_store">store</a></tt>
3439instructions. </p>
Chris Lattner21003c82008-01-06 21:04:43 +0000[diff] [blame]3440<p>
Chris Lattner10368b62008-04-02 00:38:26 +0000[diff] [blame]3441The optional constant "align" argument specifies the alignment of the operation
Chris Lattner21003c82008-01-06 21:04:43 +0000[diff] [blame]3442(that is, the alignment of the memory address). A value of 0 or an
3443omitted "align" argument means that the operation has the preferential
3444alignment for the target. It is the responsibility of the code emitter
3445to ensure that the alignment information is correct. Overestimating
3446the alignment results in an undefined behavior. Underestimating the
3447alignment may produce less efficient code. An alignment of 1 is always
3448safe.
3449</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3450<h5>Semantics:</h5>
3451<p>The location of memory pointed to is loaded.</p>
3452<h5>Examples:</h5>
3453<pre> %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
3454 <a
3455 href="#i_store">store</a> i32 3, i32* %ptr <i>; yields {void}</i>
3456 %val = load i32* %ptr <i>; yields {i32}:val = i32 3</i>
3457</pre>
3458</div>
3459<!-- _______________________________________________________________________ -->
3460<div class="doc_subsubsection"> <a name="i_store">'<tt>store</tt>'
3461Instruction</a> </div>
3462<div class="doc_text">
3463<h5>Syntax:</h5>
3464<pre> store &lt;ty&gt; &lt;value&gt;, &lt;ty&gt;* &lt;pointer&gt;[, align &lt;alignment&gt;] <i>; yields {void}</i>
3465 volatile store &lt;ty&gt; &lt;value&gt;, &lt;ty&gt;* &lt;pointer&gt;[, align &lt;alignment&gt;] <i>; yields {void}</i>
3466</pre>
3467<h5>Overview:</h5>
3468<p>The '<tt>store</tt>' instruction is used to write to memory.</p>
3469<h5>Arguments:</h5>
3470<p>There are two arguments to the '<tt>store</tt>' instruction: a value
3471to 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]3472operand must be a pointer to the <a href="#t_firstclass">first class</a> type
3473of the '<tt>&lt;value&gt;</tt>'
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3474operand. If the <tt>store</tt> is marked as <tt>volatile</tt>, then the
3475optimizer is not allowed to modify the number or order of execution of
3476this <tt>store</tt> with other volatile <tt>load</tt> and <tt><a
3477 href="#i_store">store</a></tt> instructions.</p>
Chris Lattner21003c82008-01-06 21:04:43 +0000[diff] [blame]3478<p>
Chris Lattner10368b62008-04-02 00:38:26 +0000[diff] [blame]3479The optional constant "align" argument specifies the alignment of the operation
Chris Lattner21003c82008-01-06 21:04:43 +0000[diff] [blame]3480(that is, the alignment of the memory address). A value of 0 or an
3481omitted "align" argument means that the operation has the preferential
3482alignment for the target. It is the responsibility of the code emitter
3483to ensure that the alignment information is correct. Overestimating
3484the alignment results in an undefined behavior. Underestimating the
3485alignment may produce less efficient code. An alignment of 1 is always
3486safe.
3487</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3488<h5>Semantics:</h5>
3489<p>The contents of memory are updated to contain '<tt>&lt;value&gt;</tt>'
3490at the location specified by the '<tt>&lt;pointer&gt;</tt>' operand.</p>
3491<h5>Example:</h5>
3492<pre> %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
Bill Wendling63ffa142007-10-22 05:10:05 +0000[diff] [blame]3493 store i32 3, i32* %ptr <i>; yields {void}</i>
3494 %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]3495</pre>
3496</div>
3497
3498<!-- _______________________________________________________________________ -->
3499<div class="doc_subsubsection">
3500 <a name="i_getelementptr">'<tt>getelementptr</tt>' Instruction</a>
3501</div>
3502
3503<div class="doc_text">
3504<h5>Syntax:</h5>
3505<pre>
Matthijs Kooijman32a080e2008-10-13 13:44:15 +0000[diff] [blame]3506 &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]3507</pre>
3508
3509<h5>Overview:</h5>
3510
3511<p>
3512The '<tt>getelementptr</tt>' instruction is used to get the address of a
Matthijs Kooijman32a080e2008-10-13 13:44:15 +0000[diff] [blame]3513subelement of an aggregate data structure. It performs address calculation only
3514and does not access memory.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3515
3516<h5>Arguments:</h5>
3517
Matthijs Kooijman32a080e2008-10-13 13:44:15 +0000[diff] [blame]3518<p>The first argument is always a pointer, and forms the basis of the
3519calculation. The remaining arguments are indices, that indicate which of the
3520elements of the aggregate object are indexed. The interpretation of each index
3521is dependent on the type being indexed into. The first index always indexes the
3522pointer value given as the first argument, the second index indexes a value of
3523the type pointed to (not necessarily the value directly pointed to, since the
3524first index can be non-zero), etc. The first type indexed into must be a pointer
3525value, subsequent types can be arrays, vectors and structs. Note that subsequent
3526types being indexed into can never be pointers, since that would require loading
3527the pointer before continuing calculation.</p>
3528
3529<p>The type of each index argument depends on the type it is indexing into.
3530When indexing into a (packed) structure, only <tt>i32</tt> integer
3531<b>constants</b> are allowed. When indexing into an array, pointer or vector,
3532only integers of 32 or 64 bits are allowed (also non-constants). 32-bit values
3533will be sign extended to 64-bits if required.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3534
3535<p>For example, let's consider a C code fragment and how it gets
3536compiled to LLVM:</p>
3537
3538<div class="doc_code">
3539<pre>
3540struct RT {
3541 char A;
3542 int B[10][20];
3543 char C;
3544};
3545struct ST {
3546 int X;
3547 double Y;
3548 struct RT Z;
3549};
3550
3551int *foo(struct ST *s) {
3552 return &amp;s[1].Z.B[5][13];
3553}
3554</pre>
3555</div>
3556
3557<p>The LLVM code generated by the GCC frontend is:</p>
3558
3559<div class="doc_code">
3560<pre>
Chris Lattner5b6dc6e2009-01-11 20:53:49 +0000[diff] [blame]3561%RT = <a href="#namedtypes">type</a> { i8 , [10 x [20 x i32]], i8 }
3562%ST = <a href="#namedtypes">type</a> { i32, double, %RT }
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3563
3564define i32* %foo(%ST* %s) {
3565entry:
3566 %reg = getelementptr %ST* %s, i32 1, i32 2, i32 1, i32 5, i32 13
3567 ret i32* %reg
3568}
3569</pre>
3570</div>
3571
3572<h5>Semantics:</h5>
3573
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3574<p>In the example above, the first index is indexing into the '<tt>%ST*</tt>'
3575type, which is a pointer, yielding a '<tt>%ST</tt>' = '<tt>{ i32, double, %RT
3576}</tt>' type, a structure. The second index indexes into the third element of
3577the structure, yielding a '<tt>%RT</tt>' = '<tt>{ i8 , [10 x [20 x i32]],
3578i8 }</tt>' type, another structure. The third index indexes into the second
3579element of the structure, yielding a '<tt>[10 x [20 x i32]]</tt>' type, an
3580array. The two dimensions of the array are subscripted into, yielding an
3581'<tt>i32</tt>' type. The '<tt>getelementptr</tt>' instruction returns a pointer
3582to this element, thus computing a value of '<tt>i32*</tt>' type.</p>
3583
3584<p>Note that it is perfectly legal to index partially through a
3585structure, returning a pointer to an inner element. Because of this,
3586the LLVM code for the given testcase is equivalent to:</p>
3587
3588<pre>
3589 define i32* %foo(%ST* %s) {
3590 %t1 = getelementptr %ST* %s, i32 1 <i>; yields %ST*:%t1</i>
3591 %t2 = getelementptr %ST* %t1, i32 0, i32 2 <i>; yields %RT*:%t2</i>
3592 %t3 = getelementptr %RT* %t2, i32 0, i32 1 <i>; yields [10 x [20 x i32]]*:%t3</i>
3593 %t4 = getelementptr [10 x [20 x i32]]* %t3, i32 0, i32 5 <i>; yields [20 x i32]*:%t4</i>
3594 %t5 = getelementptr [20 x i32]* %t4, i32 0, i32 13 <i>; yields i32*:%t5</i>
3595 ret i32* %t5
3596 }
3597</pre>
3598
3599<p>Note that it is undefined to access an array out of bounds: array and
3600pointer indexes must always be within the defined bounds of the array type.
Chris Lattnera7d94ba2008-04-24 05:59:56 +0000[diff] [blame]3601The one exception for this rule is zero length arrays. These arrays are
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3602defined to be accessible as variable length arrays, which requires access
3603beyond the zero'th element.</p>
3604
3605<p>The getelementptr instruction is often confusing. For some more insight
3606into how it works, see <a href="GetElementPtr.html">the getelementptr
3607FAQ</a>.</p>
3608
3609<h5>Example:</h5>
3610
3611<pre>
3612 <i>; yields [12 x i8]*:aptr</i>
Matthijs Kooijman32a080e2008-10-13 13:44:15 +0000[diff] [blame]3613 %aptr = getelementptr {i32, [12 x i8]}* %saptr, i64 0, i32 1
3614 <i>; yields i8*:vptr</i>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]3615 %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]3616 <i>; yields i8*:eptr</i>
3617 %eptr = getelementptr [12 x i8]* %aptr, i64 0, i32 1
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3618</pre>
3619</div>
3620
3621<!-- ======================================================================= -->
3622<div class="doc_subsection"> <a name="convertops">Conversion Operations</a>
3623</div>
3624<div class="doc_text">
3625<p>The instructions in this category are the conversion instructions (casting)
3626which all take a single operand and a type. They perform various bit conversions
3627on the operand.</p>
3628</div>
3629
3630<!-- _______________________________________________________________________ -->
3631<div class="doc_subsubsection">
3632 <a name="i_trunc">'<tt>trunc .. to</tt>' Instruction</a>
3633</div>
3634<div class="doc_text">
3635
3636<h5>Syntax:</h5>
3637<pre>
3638 &lt;result&gt; = trunc &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3639</pre>
3640
3641<h5>Overview:</h5>
3642<p>
3643The '<tt>trunc</tt>' instruction truncates its operand to the type <tt>ty2</tt>.
3644</p>
3645
3646<h5>Arguments:</h5>
3647<p>
3648The '<tt>trunc</tt>' instruction takes a <tt>value</tt> to trunc, which must
3649be an <a href="#t_integer">integer</a> type, and a type that specifies the size
3650and type of the result, which must be an <a href="#t_integer">integer</a>
3651type. The bit size of <tt>value</tt> must be larger than the bit size of
3652<tt>ty2</tt>. Equal sized types are not allowed.</p>
3653
3654<h5>Semantics:</h5>
3655<p>
3656The '<tt>trunc</tt>' instruction truncates the high order bits in <tt>value</tt>
3657and converts the remaining bits to <tt>ty2</tt>. Since the source size must be
3658larger than the destination size, <tt>trunc</tt> cannot be a <i>no-op cast</i>.
3659It will always truncate bits.</p>
3660
3661<h5>Example:</h5>
3662<pre>
3663 %X = trunc i32 257 to i8 <i>; yields i8:1</i>
3664 %Y = trunc i32 123 to i1 <i>; yields i1:true</i>
3665 %Y = trunc i32 122 to i1 <i>; yields i1:false</i>
3666</pre>
3667</div>
3668
3669<!-- _______________________________________________________________________ -->
3670<div class="doc_subsubsection">
3671 <a name="i_zext">'<tt>zext .. to</tt>' Instruction</a>
3672</div>
3673<div class="doc_text">
3674
3675<h5>Syntax:</h5>
3676<pre>
3677 &lt;result&gt; = zext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3678</pre>
3679
3680<h5>Overview:</h5>
3681<p>The '<tt>zext</tt>' instruction zero extends its operand to type
3682<tt>ty2</tt>.</p>
3683
3684
3685<h5>Arguments:</h5>
3686<p>The '<tt>zext</tt>' instruction takes a value to cast, which must be of
3687<a href="#t_integer">integer</a> type, and a type to cast it to, which must
3688also be of <a href="#t_integer">integer</a> type. The bit size of the
3689<tt>value</tt> must be smaller than the bit size of the destination type,
3690<tt>ty2</tt>.</p>
3691
3692<h5>Semantics:</h5>
3693<p>The <tt>zext</tt> fills the high order bits of the <tt>value</tt> with zero
3694bits until it reaches the size of the destination type, <tt>ty2</tt>.</p>
3695
3696<p>When zero extending from i1, the result will always be either 0 or 1.</p>
3697
3698<h5>Example:</h5>
3699<pre>
3700 %X = zext i32 257 to i64 <i>; yields i64:257</i>
3701 %Y = zext i1 true to i32 <i>; yields i32:1</i>
3702</pre>
3703</div>
3704
3705<!-- _______________________________________________________________________ -->
3706<div class="doc_subsubsection">
3707 <a name="i_sext">'<tt>sext .. to</tt>' Instruction</a>
3708</div>
3709<div class="doc_text">
3710
3711<h5>Syntax:</h5>
3712<pre>
3713 &lt;result&gt; = sext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3714</pre>
3715
3716<h5>Overview:</h5>
3717<p>The '<tt>sext</tt>' sign extends <tt>value</tt> to the type <tt>ty2</tt>.</p>
3718
3719<h5>Arguments:</h5>
3720<p>
3721The '<tt>sext</tt>' instruction takes a value to cast, which must be of
3722<a href="#t_integer">integer</a> type, and a type to cast it to, which must
3723also be of <a href="#t_integer">integer</a> type. The bit size of the
3724<tt>value</tt> must be smaller than the bit size of the destination type,
3725<tt>ty2</tt>.</p>
3726
3727<h5>Semantics:</h5>
3728<p>
3729The '<tt>sext</tt>' instruction performs a sign extension by copying the sign
3730bit (highest order bit) of the <tt>value</tt> until it reaches the bit size of
3731the type <tt>ty2</tt>.</p>
3732
3733<p>When sign extending from i1, the extension always results in -1 or 0.</p>
3734
3735<h5>Example:</h5>
3736<pre>
3737 %X = sext i8 -1 to i16 <i>; yields i16 :65535</i>
3738 %Y = sext i1 true to i32 <i>; yields i32:-1</i>
3739</pre>
3740</div>
3741
3742<!-- _______________________________________________________________________ -->
3743<div class="doc_subsubsection">
3744 <a name="i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a>
3745</div>
3746
3747<div class="doc_text">
3748
3749<h5>Syntax:</h5>
3750
3751<pre>
3752 &lt;result&gt; = fptrunc &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3753</pre>
3754
3755<h5>Overview:</h5>
3756<p>The '<tt>fptrunc</tt>' instruction truncates <tt>value</tt> to type
3757<tt>ty2</tt>.</p>
3758
3759
3760<h5>Arguments:</h5>
3761<p>The '<tt>fptrunc</tt>' instruction takes a <a href="#t_floating">floating
3762 point</a> value to cast and a <a href="#t_floating">floating point</a> type to
3763cast it to. The size of <tt>value</tt> must be larger than the size of
3764<tt>ty2</tt>. This implies that <tt>fptrunc</tt> cannot be used to make a
3765<i>no-op cast</i>.</p>
3766
3767<h5>Semantics:</h5>
3768<p> The '<tt>fptrunc</tt>' instruction truncates a <tt>value</tt> from a larger
3769<a href="#t_floating">floating point</a> type to a smaller
3770<a href="#t_floating">floating point</a> type. If the value cannot fit within
3771the destination type, <tt>ty2</tt>, then the results are undefined.</p>
3772
3773<h5>Example:</h5>
3774<pre>
3775 %X = fptrunc double 123.0 to float <i>; yields float:123.0</i>
3776 %Y = fptrunc double 1.0E+300 to float <i>; yields undefined</i>
3777</pre>
3778</div>
3779
3780<!-- _______________________________________________________________________ -->
3781<div class="doc_subsubsection">
3782 <a name="i_fpext">'<tt>fpext .. to</tt>' Instruction</a>
3783</div>
3784<div class="doc_text">
3785
3786<h5>Syntax:</h5>
3787<pre>
3788 &lt;result&gt; = fpext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3789</pre>
3790
3791<h5>Overview:</h5>
3792<p>The '<tt>fpext</tt>' extends a floating point <tt>value</tt> to a larger
3793floating point value.</p>
3794
3795<h5>Arguments:</h5>
3796<p>The '<tt>fpext</tt>' instruction takes a
3797<a href="#t_floating">floating point</a> <tt>value</tt> to cast,
3798and a <a href="#t_floating">floating point</a> type to cast it to. The source
3799type must be smaller than the destination type.</p>
3800
3801<h5>Semantics:</h5>
3802<p>The '<tt>fpext</tt>' instruction extends the <tt>value</tt> from a smaller
3803<a href="#t_floating">floating point</a> type to a larger
3804<a href="#t_floating">floating point</a> type. The <tt>fpext</tt> cannot be
3805used to make a <i>no-op cast</i> because it always changes bits. Use
3806<tt>bitcast</tt> to make a <i>no-op cast</i> for a floating point cast.</p>
3807
3808<h5>Example:</h5>
3809<pre>
3810 %X = fpext float 3.1415 to double <i>; yields double:3.1415</i>
3811 %Y = fpext float 1.0 to float <i>; yields float:1.0 (no-op)</i>
3812</pre>
3813</div>
3814
3815<!-- _______________________________________________________________________ -->
3816<div class="doc_subsubsection">
3817 <a name="i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a>
3818</div>
3819<div class="doc_text">
3820
3821<h5>Syntax:</h5>
3822<pre>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3823 &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]3824</pre>
3825
3826<h5>Overview:</h5>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3827<p>The '<tt>fptoui</tt>' converts a floating point <tt>value</tt> to its
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3828unsigned integer equivalent of type <tt>ty2</tt>.
3829</p>
3830
3831<h5>Arguments:</h5>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3832<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]3833scalar or vector <a href="#t_floating">floating point</a> value, and a type
3834to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
3835type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
3836vector integer type with the same number of elements as <tt>ty</tt></p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3837
3838<h5>Semantics:</h5>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3839<p> The '<tt>fptoui</tt>' instruction converts its
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3840<a href="#t_floating">floating point</a> operand into the nearest (rounding
3841towards zero) unsigned integer value. If the value cannot fit in <tt>ty2</tt>,
3842the results are undefined.</p>
3843
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3844<h5>Example:</h5>
3845<pre>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3846 %X = fptoui double 123.0 to i32 <i>; yields i32:123</i>
Chris Lattner681f1e82007-09-22 03:17:52 +0000[diff] [blame]3847 %Y = fptoui float 1.0E+300 to i1 <i>; yields undefined:1</i>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3848 %X = fptoui float 1.04E+17 to i8 <i>; yields undefined:1</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3849</pre>
3850</div>
3851
3852<!-- _______________________________________________________________________ -->
3853<div class="doc_subsubsection">
3854 <a name="i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a>
3855</div>
3856<div class="doc_text">
3857
3858<h5>Syntax:</h5>
3859<pre>
3860 &lt;result&gt; = fptosi &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3861</pre>
3862
3863<h5>Overview:</h5>
3864<p>The '<tt>fptosi</tt>' instruction converts
3865<a href="#t_floating">floating point</a> <tt>value</tt> to type <tt>ty2</tt>.
3866</p>
3867
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3868<h5>Arguments:</h5>
3869<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]3870scalar or vector <a href="#t_floating">floating point</a> value, and a type
3871to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
3872type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
3873vector integer type with the same number of elements as <tt>ty</tt></p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3874
3875<h5>Semantics:</h5>
3876<p>The '<tt>fptosi</tt>' instruction converts its
3877<a href="#t_floating">floating point</a> operand into the nearest (rounding
3878towards zero) signed integer value. If the value cannot fit in <tt>ty2</tt>,
3879the results are undefined.</p>
3880
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3881<h5>Example:</h5>
3882<pre>
3883 %X = fptosi double -123.0 to i32 <i>; yields i32:-123</i>
Chris Lattner681f1e82007-09-22 03:17:52 +0000[diff] [blame]3884 %Y = fptosi float 1.0E-247 to i1 <i>; yields undefined:1</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3885 %X = fptosi float 1.04E+17 to i8 <i>; yields undefined:1</i>
3886</pre>
3887</div>
3888
3889<!-- _______________________________________________________________________ -->
3890<div class="doc_subsubsection">
3891 <a name="i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a>
3892</div>
3893<div class="doc_text">
3894
3895<h5>Syntax:</h5>
3896<pre>
3897 &lt;result&gt; = uitofp &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3898</pre>
3899
3900<h5>Overview:</h5>
3901<p>The '<tt>uitofp</tt>' instruction regards <tt>value</tt> as an unsigned
3902integer and converts that value to the <tt>ty2</tt> type.</p>
3903
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3904<h5>Arguments:</h5>
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]3905<p>The '<tt>uitofp</tt>' instruction takes a value to cast, which must be a
3906scalar or vector <a href="#t_integer">integer</a> value, and a type to cast it
3907to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
3908type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
3909floating point type with the same number of elements as <tt>ty</tt></p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3910
3911<h5>Semantics:</h5>
3912<p>The '<tt>uitofp</tt>' instruction interprets its operand as an unsigned
3913integer quantity and converts it to the corresponding floating point value. If
3914the value cannot fit in the floating point value, the results are undefined.</p>
3915
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3916<h5>Example:</h5>
3917<pre>
3918 %X = uitofp i32 257 to float <i>; yields float:257.0</i>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]3919 %Y = uitofp i8 -1 to double <i>; yields double:255.0</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3920</pre>
3921</div>
3922
3923<!-- _______________________________________________________________________ -->
3924<div class="doc_subsubsection">
3925 <a name="i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a>
3926</div>
3927<div class="doc_text">
3928
3929<h5>Syntax:</h5>
3930<pre>
3931 &lt;result&gt; = sitofp &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3932</pre>
3933
3934<h5>Overview:</h5>
3935<p>The '<tt>sitofp</tt>' instruction regards <tt>value</tt> as a signed
3936integer and converts that value to the <tt>ty2</tt> type.</p>
3937
3938<h5>Arguments:</h5>
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]3939<p>The '<tt>sitofp</tt>' instruction takes a value to cast, which must be a
3940scalar or vector <a href="#t_integer">integer</a> value, and a type to cast it
3941to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
3942type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
3943floating point type with the same number of elements as <tt>ty</tt></p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3944
3945<h5>Semantics:</h5>
3946<p>The '<tt>sitofp</tt>' instruction interprets its operand as a signed
3947integer quantity and converts it to the corresponding floating point value. If
3948the value cannot fit in the floating point value, the results are undefined.</p>
3949
3950<h5>Example:</h5>
3951<pre>
3952 %X = sitofp i32 257 to float <i>; yields float:257.0</i>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]3953 %Y = sitofp i8 -1 to double <i>; yields double:-1.0</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3954</pre>
3955</div>
3956
3957<!-- _______________________________________________________________________ -->
3958<div class="doc_subsubsection">
3959 <a name="i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a>
3960</div>
3961<div class="doc_text">
3962
3963<h5>Syntax:</h5>
3964<pre>
3965 &lt;result&gt; = ptrtoint &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3966</pre>
3967
3968<h5>Overview:</h5>
3969<p>The '<tt>ptrtoint</tt>' instruction converts the pointer <tt>value</tt> to
3970the integer type <tt>ty2</tt>.</p>
3971
3972<h5>Arguments:</h5>
3973<p>The '<tt>ptrtoint</tt>' instruction takes a <tt>value</tt> to cast, which
3974must 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]3975<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]3976
3977<h5>Semantics:</h5>
3978<p>The '<tt>ptrtoint</tt>' instruction converts <tt>value</tt> to integer type
3979<tt>ty2</tt> by interpreting the pointer value as an integer and either
3980truncating or zero extending that value to the size of the integer type. If
3981<tt>value</tt> is smaller than <tt>ty2</tt> then a zero extension is done. If
3982<tt>value</tt> is larger than <tt>ty2</tt> then a truncation is done. If they
3983are the same size, then nothing is done (<i>no-op cast</i>) other than a type
3984change.</p>
3985
3986<h5>Example:</h5>
3987<pre>
3988 %X = ptrtoint i32* %X to i8 <i>; yields truncation on 32-bit architecture</i>
3989 %Y = ptrtoint i32* %x to i64 <i>; yields zero extension on 32-bit architecture</i>
3990</pre>
3991</div>
3992
3993<!-- _______________________________________________________________________ -->
3994<div class="doc_subsubsection">
3995 <a name="i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a>
3996</div>
3997<div class="doc_text">
3998
3999<h5>Syntax:</h5>
4000<pre>
4001 &lt;result&gt; = inttoptr &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
4002</pre>
4003
4004<h5>Overview:</h5>
4005<p>The '<tt>inttoptr</tt>' instruction converts an integer <tt>value</tt> to
4006a pointer type, <tt>ty2</tt>.</p>
4007
4008<h5>Arguments:</h5>
4009<p>The '<tt>inttoptr</tt>' instruction takes an <a href="#t_integer">integer</a>
4010value to cast, and a type to cast it to, which must be a
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4011<a href="#t_pointer">pointer</a> type.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4012
4013<h5>Semantics:</h5>
4014<p>The '<tt>inttoptr</tt>' instruction converts <tt>value</tt> to type
4015<tt>ty2</tt> by applying either a zero extension or a truncation depending on
4016the size of the integer <tt>value</tt>. If <tt>value</tt> is larger than the
4017size of a pointer then a truncation is done. If <tt>value</tt> is smaller than
4018the size of a pointer then a zero extension is done. If they are the same size,
4019nothing is done (<i>no-op cast</i>).</p>
4020
4021<h5>Example:</h5>
4022<pre>
4023 %X = inttoptr i32 255 to i32* <i>; yields zero extension on 64-bit architecture</i>
4024 %X = inttoptr i32 255 to i32* <i>; yields no-op on 32-bit architecture</i>
4025 %Y = inttoptr i64 0 to i32* <i>; yields truncation on 32-bit architecture</i>
4026</pre>
4027</div>
4028
4029<!-- _______________________________________________________________________ -->
4030<div class="doc_subsubsection">
4031 <a name="i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a>
4032</div>
4033<div class="doc_text">
4034
4035<h5>Syntax:</h5>
4036<pre>
4037 &lt;result&gt; = bitcast &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
4038</pre>
4039
4040<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4041
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4042<p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
4043<tt>ty2</tt> without changing any bits.</p>
4044
4045<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4046
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4047<p>The '<tt>bitcast</tt>' instruction takes a value to cast, which must be
Dan Gohman7305fa02008-09-08 16:45:59 +0000[diff] [blame]4048a non-aggregate first class value, and a type to cast it to, which must also be
4049a non-aggregate <a href="#t_firstclass">first class</a> type. The bit sizes of
4050<tt>value</tt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4051and the destination type, <tt>ty2</tt>, must be identical. If the source
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4052type is a pointer, the destination type must also be a pointer. This
4053instruction supports bitwise conversion of vectors to integers and to vectors
4054of other types (as long as they have the same size).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4055
4056<h5>Semantics:</h5>
4057<p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
4058<tt>ty2</tt>. It is always a <i>no-op cast</i> because no bits change with
4059this conversion. The conversion is done as if the <tt>value</tt> had been
4060stored to memory and read back as type <tt>ty2</tt>. Pointer types may only be
4061converted to other pointer types with this instruction. To convert pointers to
4062other types, use the <a href="#i_inttoptr">inttoptr</a> or
4063<a href="#i_ptrtoint">ptrtoint</a> instructions first.</p>
4064
4065<h5>Example:</h5>
4066<pre>
4067 %X = bitcast i8 255 to i8 <i>; yields i8 :-1</i>
4068 %Y = bitcast i32* %x to sint* <i>; yields sint*:%x</i>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4069 %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]4070</pre>
4071</div>
4072
4073<!-- ======================================================================= -->
4074<div class="doc_subsection"> <a name="otherops">Other Operations</a> </div>
4075<div class="doc_text">
4076<p>The instructions in this category are the "miscellaneous"
4077instructions, which defy better classification.</p>
4078</div>
4079
4080<!-- _______________________________________________________________________ -->
4081<div class="doc_subsubsection"><a name="i_icmp">'<tt>icmp</tt>' Instruction</a>
4082</div>
4083<div class="doc_text">
4084<h5>Syntax:</h5>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4085<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]4086</pre>
4087<h5>Overview:</h5>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4088<p>The '<tt>icmp</tt>' instruction returns a boolean value or
4089a vector of boolean values based on comparison
4090of its two integer, integer vector, or pointer operands.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4091<h5>Arguments:</h5>
4092<p>The '<tt>icmp</tt>' instruction takes three operands. The first operand is
4093the condition code indicating the kind of comparison to perform. It is not
4094a value, just a keyword. The possible condition code are:
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4095</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4096<ol>
4097 <li><tt>eq</tt>: equal</li>
4098 <li><tt>ne</tt>: not equal </li>
4099 <li><tt>ugt</tt>: unsigned greater than</li>
4100 <li><tt>uge</tt>: unsigned greater or equal</li>
4101 <li><tt>ult</tt>: unsigned less than</li>
4102 <li><tt>ule</tt>: unsigned less or equal</li>
4103 <li><tt>sgt</tt>: signed greater than</li>
4104 <li><tt>sge</tt>: signed greater or equal</li>
4105 <li><tt>slt</tt>: signed less than</li>
4106 <li><tt>sle</tt>: signed less or equal</li>
4107</ol>
4108<p>The remaining two arguments must be <a href="#t_integer">integer</a> or
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4109<a href="#t_pointer">pointer</a>
4110or integer <a href="#t_vector">vector</a> typed.
4111They must also be identical types.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4112<h5>Semantics:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4113<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]4114the condition code given as <tt>cond</tt>. The comparison performed always
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4115yields 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]4116</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4117<ol>
4118 <li><tt>eq</tt>: yields <tt>true</tt> if the operands are equal,
4119 <tt>false</tt> otherwise. No sign interpretation is necessary or performed.
4120 </li>
4121 <li><tt>ne</tt>: yields <tt>true</tt> if the operands are unequal,
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4122 <tt>false</tt> otherwise. No sign interpretation is necessary or performed.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4123 <li><tt>ugt</tt>: interprets the operands as unsigned values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4124 <tt>true</tt> if <tt>op1</tt> is greater than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4125 <li><tt>uge</tt>: interprets the operands as unsigned values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4126 <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]4127 <li><tt>ult</tt>: interprets the operands as unsigned values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4128 <tt>true</tt> if <tt>op1</tt> is less than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4129 <li><tt>ule</tt>: interprets the operands as unsigned values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4130 <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]4131 <li><tt>sgt</tt>: interprets the operands as signed values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4132 <tt>true</tt> if <tt>op1</tt> is greater than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4133 <li><tt>sge</tt>: interprets the operands as signed values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4134 <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]4135 <li><tt>slt</tt>: interprets the operands as signed values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4136 <tt>true</tt> if <tt>op1</tt> is less than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4137 <li><tt>sle</tt>: interprets the operands as signed values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4138 <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]4139</ol>
4140<p>If the operands are <a href="#t_pointer">pointer</a> typed, the pointer
4141values are compared as if they were integers.</p>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4142<p>If the operands are integer vectors, then they are compared
4143element by element. The result is an <tt>i1</tt> vector with
4144the same number of elements as the values being compared.
4145Otherwise, the result is an <tt>i1</tt>.
4146</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4147
4148<h5>Example:</h5>
4149<pre> &lt;result&gt; = icmp eq i32 4, 5 <i>; yields: result=false</i>
4150 &lt;result&gt; = icmp ne float* %X, %X <i>; yields: result=false</i>
4151 &lt;result&gt; = icmp ult i16 4, 5 <i>; yields: result=true</i>
4152 &lt;result&gt; = icmp sgt i16 4, 5 <i>; yields: result=false</i>
4153 &lt;result&gt; = icmp ule i16 -4, 5 <i>; yields: result=false</i>
4154 &lt;result&gt; = icmp sge i16 4, 5 <i>; yields: result=false</i>
4155</pre>
Dan Gohmana53eb382009-01-22 01:39:38 +0000[diff] [blame]4156
4157<p>Note that the code generator does not yet support vector types with
4158 the <tt>icmp</tt> instruction.</p>
4159
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4160</div>
4161
4162<!-- _______________________________________________________________________ -->
4163<div class="doc_subsubsection"><a name="i_fcmp">'<tt>fcmp</tt>' Instruction</a>
4164</div>
4165<div class="doc_text">
4166<h5>Syntax:</h5>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4167<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]4168</pre>
4169<h5>Overview:</h5>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4170<p>The '<tt>fcmp</tt>' instruction returns a boolean value
4171or vector of boolean values based on comparison
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4172of its operands.</p>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4173<p>
4174If the operands are floating point scalars, then the result
4175type is a boolean (<a href="#t_primitive"><tt>i1</tt></a>).
4176</p>
4177<p>If the operands are floating point vectors, then the result type
4178is a vector of boolean with the same number of elements as the
4179operands being compared.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4180<h5>Arguments:</h5>
4181<p>The '<tt>fcmp</tt>' instruction takes three operands. The first operand is
4182the condition code indicating the kind of comparison to perform. It is not
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4183a value, just a keyword. The possible condition code are:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4184<ol>
4185 <li><tt>false</tt>: no comparison, always returns false</li>
4186 <li><tt>oeq</tt>: ordered and equal</li>
4187 <li><tt>ogt</tt>: ordered and greater than </li>
4188 <li><tt>oge</tt>: ordered and greater than or equal</li>
4189 <li><tt>olt</tt>: ordered and less than </li>
4190 <li><tt>ole</tt>: ordered and less than or equal</li>
4191 <li><tt>one</tt>: ordered and not equal</li>
4192 <li><tt>ord</tt>: ordered (no nans)</li>
4193 <li><tt>ueq</tt>: unordered or equal</li>
4194 <li><tt>ugt</tt>: unordered or greater than </li>
4195 <li><tt>uge</tt>: unordered or greater than or equal</li>
4196 <li><tt>ult</tt>: unordered or less than </li>
4197 <li><tt>ule</tt>: unordered or less than or equal</li>
4198 <li><tt>une</tt>: unordered or not equal</li>
4199 <li><tt>uno</tt>: unordered (either nans)</li>
4200 <li><tt>true</tt>: no comparison, always returns true</li>
4201</ol>
4202<p><i>Ordered</i> means that neither operand is a QNAN while
4203<i>unordered</i> means that either operand may be a QNAN.</p>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4204<p>Each of <tt>val1</tt> and <tt>val2</tt> arguments must be
4205either a <a href="#t_floating">floating point</a> type
4206or a <a href="#t_vector">vector</a> of floating point type.
4207They must have identical types.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4208<h5>Semantics:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4209<p>The '<tt>fcmp</tt>' instruction compares <tt>op1</tt> and <tt>op2</tt>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4210according to the condition code given as <tt>cond</tt>.
4211If the operands are vectors, then the vectors are compared
4212element by element.
4213Each comparison performed
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4214always yields an <a href="#t_primitive">i1</a> result, as follows:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4215<ol>
4216 <li><tt>false</tt>: always yields <tt>false</tt>, regardless of operands.</li>
4217 <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]4218 <tt>op1</tt> is equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4219 <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]4220 <tt>op1</tt> is greather than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4221 <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]4222 <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4223 <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]4224 <tt>op1</tt> is less than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4225 <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]4226 <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4227 <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]4228 <tt>op1</tt> is not equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4229 <li><tt>ord</tt>: yields <tt>true</tt> if both operands are not a QNAN.</li>
4230 <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]4231 <tt>op1</tt> is equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4232 <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]4233 <tt>op1</tt> is greater than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4234 <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]4235 <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4236 <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]4237 <tt>op1</tt> is less than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4238 <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]4239 <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4240 <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]4241 <tt>op1</tt> is not equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4242 <li><tt>uno</tt>: yields <tt>true</tt> if either operand is a QNAN.</li>
4243 <li><tt>true</tt>: always yields <tt>true</tt>, regardless of operands.</li>
4244</ol>
4245
4246<h5>Example:</h5>
4247<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]4248 &lt;result&gt; = fcmp one float 4.0, 5.0 <i>; yields: result=true</i>
4249 &lt;result&gt; = fcmp olt float 4.0, 5.0 <i>; yields: result=true</i>
4250 &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]4251</pre>
Dan Gohmana53eb382009-01-22 01:39:38 +0000[diff] [blame]4252
4253<p>Note that the code generator does not yet support vector types with
4254 the <tt>fcmp</tt> instruction.</p>
4255
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4256</div>
4257
4258<!-- _______________________________________________________________________ -->
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]4259<div class="doc_subsubsection">
4260 <a name="i_vicmp">'<tt>vicmp</tt>' Instruction</a>
4261</div>
4262<div class="doc_text">
4263<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4264<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]4265</pre>
4266<h5>Overview:</h5>
4267<p>The '<tt>vicmp</tt>' instruction returns an integer vector value based on
4268element-wise comparison of its two integer vector operands.</p>
4269<h5>Arguments:</h5>
4270<p>The '<tt>vicmp</tt>' instruction takes three operands. The first operand is
4271the condition code indicating the kind of comparison to perform. It is not
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4272a value, just a keyword. The possible condition code are:</p>
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]4273<ol>
4274 <li><tt>eq</tt>: equal</li>
4275 <li><tt>ne</tt>: not equal </li>
4276 <li><tt>ugt</tt>: unsigned greater than</li>
4277 <li><tt>uge</tt>: unsigned greater or equal</li>
4278 <li><tt>ult</tt>: unsigned less than</li>
4279 <li><tt>ule</tt>: unsigned less or equal</li>
4280 <li><tt>sgt</tt>: signed greater than</li>
4281 <li><tt>sge</tt>: signed greater or equal</li>
4282 <li><tt>slt</tt>: signed less than</li>
4283 <li><tt>sle</tt>: signed less or equal</li>
4284</ol>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4285<p>The remaining two arguments must be <a href="#t_vector">vector</a> or
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]4286<a href="#t_integer">integer</a> typed. They must also be identical types.</p>
4287<h5>Semantics:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4288<p>The '<tt>vicmp</tt>' instruction compares <tt>op1</tt> and <tt>op2</tt>
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]4289according to the condition code given as <tt>cond</tt>. The comparison yields a
4290<a href="#t_vector">vector</a> of <a href="#t_integer">integer</a> result, of
4291identical type as the values being compared. The most significant bit in each
4292element is 1 if the element-wise comparison evaluates to true, and is 0
4293otherwise. All other bits of the result are undefined. The condition codes
4294are evaluated identically to the <a href="#i_icmp">'<tt>icmp</tt>'
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4295instruction</a>.</p>
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]4296
4297<h5>Example:</h5>
4298<pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4299 &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>
4300 &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]4301</pre>
4302</div>
4303
4304<!-- _______________________________________________________________________ -->
4305<div class="doc_subsubsection">
4306 <a name="i_vfcmp">'<tt>vfcmp</tt>' Instruction</a>
4307</div>
4308<div class="doc_text">
4309<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4310<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]4311<h5>Overview:</h5>
4312<p>The '<tt>vfcmp</tt>' instruction returns an integer vector value based on
4313element-wise comparison of its two floating point vector operands. The output
4314elements have the same width as the input elements.</p>
4315<h5>Arguments:</h5>
4316<p>The '<tt>vfcmp</tt>' instruction takes three operands. The first operand is
4317the condition code indicating the kind of comparison to perform. It is not
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4318a value, just a keyword. The possible condition code are:</p>
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]4319<ol>
4320 <li><tt>false</tt>: no comparison, always returns false</li>
4321 <li><tt>oeq</tt>: ordered and equal</li>
4322 <li><tt>ogt</tt>: ordered and greater than </li>
4323 <li><tt>oge</tt>: ordered and greater than or equal</li>
4324 <li><tt>olt</tt>: ordered and less than </li>
4325 <li><tt>ole</tt>: ordered and less than or equal</li>
4326 <li><tt>one</tt>: ordered and not equal</li>
4327 <li><tt>ord</tt>: ordered (no nans)</li>
4328 <li><tt>ueq</tt>: unordered or equal</li>
4329 <li><tt>ugt</tt>: unordered or greater than </li>
4330 <li><tt>uge</tt>: unordered or greater than or equal</li>
4331 <li><tt>ult</tt>: unordered or less than </li>
4332 <li><tt>ule</tt>: unordered or less than or equal</li>
4333 <li><tt>une</tt>: unordered or not equal</li>
4334 <li><tt>uno</tt>: unordered (either nans)</li>
4335 <li><tt>true</tt>: no comparison, always returns true</li>
4336</ol>
4337<p>The remaining two arguments must be <a href="#t_vector">vector</a> of
4338<a href="#t_floating">floating point</a> typed. They must also be identical
4339types.</p>
4340<h5>Semantics:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4341<p>The '<tt>vfcmp</tt>' instruction compares <tt>op1</tt> and <tt>op2</tt>
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]4342according to the condition code given as <tt>cond</tt>. The comparison yields a
4343<a href="#t_vector">vector</a> of <a href="#t_integer">integer</a> result, with
4344an identical number of elements as the values being compared, and each element
4345having identical with to the width of the floating point elements. The most
4346significant bit in each element is 1 if the element-wise comparison evaluates to
4347true, and is 0 otherwise. All other bits of the result are undefined. The
4348condition codes are evaluated identically to the
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4349<a href="#i_fcmp">'<tt>fcmp</tt>' instruction</a>.</p>
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]4350
4351<h5>Example:</h5>
4352<pre>
Chris Lattner1e5c5cd02008-10-13 16:55:18 +0000[diff] [blame]4353 <i>; yields: result=&lt;2 x i32&gt; &lt; i32 0, i32 -1 &gt;</i>
4354 &lt;result&gt; = vfcmp oeq &lt;2 x float&gt; &lt; float 4, float 0 &gt;, &lt; float 5, float 0 &gt;
4355
4356 <i>; yields: result=&lt;2 x i64&gt; &lt; i64 -1, i64 0 &gt;</i>
4357 &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]4358</pre>
4359</div>
4360
4361<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4362<div class="doc_subsubsection">
4363 <a name="i_phi">'<tt>phi</tt>' Instruction</a>
4364</div>
4365
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4366<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4367
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4368<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4369
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4370<pre> &lt;result&gt; = phi &lt;ty&gt; [ &lt;val0&gt;, &lt;label0&gt;], ...<br></pre>
4371<h5>Overview:</h5>
4372<p>The '<tt>phi</tt>' instruction is used to implement the &#966; node in
4373the SSA graph representing the function.</p>
4374<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4375
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4376<p>The type of the incoming values is specified with the first type
4377field. After this, the '<tt>phi</tt>' instruction takes a list of pairs
4378as arguments, with one pair for each predecessor basic block of the
4379current block. Only values of <a href="#t_firstclass">first class</a>
4380type may be used as the value arguments to the PHI node. Only labels
4381may be used as the label arguments.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4382
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4383<p>There must be no non-phi instructions between the start of a basic
4384block and the PHI instructions: i.e. PHI instructions must be first in
4385a basic block.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4386
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4387<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4388
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4389<p>At runtime, the '<tt>phi</tt>' instruction logically takes on the value
4390specified by the pair corresponding to the predecessor basic block that executed
4391just prior to the current block.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4392
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4393<h5>Example:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4394<pre>
4395Loop: ; Infinite loop that counts from 0 on up...
4396 %indvar = phi i32 [ 0, %LoopHeader ], [ %nextindvar, %Loop ]
4397 %nextindvar = add i32 %indvar, 1
4398 br label %Loop
4399</pre>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4400</div>
4401
4402<!-- _______________________________________________________________________ -->
4403<div class="doc_subsubsection">
4404 <a name="i_select">'<tt>select</tt>' Instruction</a>
4405</div>
4406
4407<div class="doc_text">
4408
4409<h5>Syntax:</h5>
4410
4411<pre>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4412 &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>
4413
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4414 <i>selty</i> is either i1 or {&lt;N x i1&gt;}
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4415</pre>
4416
4417<h5>Overview:</h5>
4418
4419<p>
4420The '<tt>select</tt>' instruction is used to choose one value based on a
4421condition, without branching.
4422</p>
4423
4424
4425<h5>Arguments:</h5>
4426
4427<p>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4428The '<tt>select</tt>' instruction requires an 'i1' value or
4429a vector of 'i1' values indicating the
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4430condition, and two values of the same <a href="#t_firstclass">first class</a>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4431type. If the val1/val2 are vectors and
4432the condition is a scalar, then entire vectors are selected, not
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4433individual elements.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4434</p>
4435
4436<h5>Semantics:</h5>
4437
4438<p>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4439If 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]4440value argument; otherwise, it returns the second value argument.
4441</p>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4442<p>
4443If the condition is a vector of i1, then the value arguments must
4444be vectors of the same size, and the selection is done element
4445by element.
4446</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4447
4448<h5>Example:</h5>
4449
4450<pre>
4451 %X = select i1 true, i8 17, i8 42 <i>; yields i8:17</i>
4452</pre>
Dan Gohmana53eb382009-01-22 01:39:38 +0000[diff] [blame]4453
4454<p>Note that the code generator does not yet support conditions
4455 with vector type.</p>
4456
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4457</div>
4458
4459
4460<!-- _______________________________________________________________________ -->
4461<div class="doc_subsubsection">
4462 <a name="i_call">'<tt>call</tt>' Instruction</a>
4463</div>
4464
4465<div class="doc_text">
4466
4467<h5>Syntax:</h5>
4468<pre>
Devang Pateld0bfcc72008-10-07 17:48:33 +0000[diff] [blame]4469 &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]4470</pre>
4471
4472<h5>Overview:</h5>
4473
4474<p>The '<tt>call</tt>' instruction represents a simple function call.</p>
4475
4476<h5>Arguments:</h5>
4477
4478<p>This instruction requires several arguments:</p>
4479
4480<ol>
4481 <li>
4482 <p>The optional "tail" marker indicates whether the callee function accesses
4483 any allocas or varargs in the caller. If the "tail" marker is present, the
4484 function call is eligible for tail call optimization. Note that calls may
4485 be marked "tail" even if they do not occur before a <a
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4486 href="#i_ret"><tt>ret</tt></a> instruction.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4487 </li>
4488 <li>
4489 <p>The optional "cconv" marker indicates which <a href="#callingconv">calling
4490 convention</a> the call should use. If none is specified, the call defaults
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4491 to using C calling conventions.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4492 </li>
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]4493
4494 <li>
4495 <p>The optional <a href="#paramattrs">Parameter Attributes</a> list for
4496 return values. Only '<tt>zeroext</tt>', '<tt>signext</tt>',
4497 and '<tt>inreg</tt>' attributes are valid here.</p>
4498 </li>
4499
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4500 <li>
Nick Lewycky93082fc2007-09-08 13:57:50 +0000[diff] [blame]4501 <p>'<tt>ty</tt>': the type of the call instruction itself which is also
4502 the type of the return value. Functions that return no value are marked
4503 <tt><a href="#t_void">void</a></tt>.</p>
4504 </li>
4505 <li>
4506 <p>'<tt>fnty</tt>': shall be the signature of the pointer to function
4507 value being invoked. The argument types must match the types implied by
4508 this signature. This type can be omitted if the function is not varargs
4509 and if the function type does not return a pointer to a function.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4510 </li>
4511 <li>
4512 <p>'<tt>fnptrval</tt>': An LLVM value containing a pointer to a function to
4513 be invoked. In most cases, this is a direct function invocation, but
4514 indirect <tt>call</tt>s are just as possible, calling an arbitrary pointer
4515 to function value.</p>
4516 </li>
4517 <li>
4518 <p>'<tt>function args</tt>': argument list whose types match the
4519 function signature argument types. All arguments must be of
4520 <a href="#t_firstclass">first class</a> type. If the function signature
4521 indicates the function accepts a variable number of arguments, the extra
4522 arguments can be specified.</p>
4523 </li>
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]4524 <li>
Devang Pateld0bfcc72008-10-07 17:48:33 +0000[diff] [blame]4525 <p>The optional <a href="#fnattrs">function attributes</a> list. Only
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]4526 '<tt>noreturn</tt>', '<tt>nounwind</tt>', '<tt>readonly</tt>' and
4527 '<tt>readnone</tt>' attributes are valid here.</p>
4528 </li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4529</ol>
4530
4531<h5>Semantics:</h5>
4532
4533<p>The '<tt>call</tt>' instruction is used to cause control flow to
4534transfer to a specified function, with its incoming arguments bound to
4535the specified values. Upon a '<tt><a href="#i_ret">ret</a></tt>'
4536instruction in the called function, control flow continues with the
4537instruction after the function call, and the return value of the
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4538function is bound to the result argument.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4539
4540<h5>Example:</h5>
4541
4542<pre>
Nick Lewycky93082fc2007-09-08 13:57:50 +0000[diff] [blame]4543 %retval = call i32 @test(i32 %argc)
Chris Lattner5e893ef2008-03-21 17:24:17 +0000[diff] [blame]4544 call i32 (i8 *, ...)* @printf(i8 * %msg, i32 12, i8 42) <i>; yields i32</i>
4545 %X = tail call i32 @foo() <i>; yields i32</i>
4546 %Y = tail call <a href="#callingconv">fastcc</a> i32 @foo() <i>; yields i32</i>
4547 call void %foo(i8 97 signext)
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]4548
4549 %struct.A = type { i32, i8 }
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]4550 %r = call %struct.A @foo() <i>; yields { 32, i8 }</i>
Dan Gohman3e700032008-10-04 19:00:07 +0000[diff] [blame]4551 %gr = extractvalue %struct.A %r, 0 <i>; yields i32</i>
4552 %gr1 = extractvalue %struct.A %r, 1 <i>; yields i8</i>
Chris Lattnerac454b32008-10-08 06:26:11 +0000[diff] [blame]4553 %Z = call void @foo() noreturn <i>; indicates that %foo never returns normally</i>
Matthijs Kooijman2c4e05a2008-10-07 10:03:45 +0000[diff] [blame]4554 %ZZ = call zeroext i32 @bar() <i>; Return value is %zero extended</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4555</pre>
4556
4557</div>
4558
4559<!-- _______________________________________________________________________ -->
4560<div class="doc_subsubsection">
4561 <a name="i_va_arg">'<tt>va_arg</tt>' Instruction</a>
4562</div>
4563
4564<div class="doc_text">
4565
4566<h5>Syntax:</h5>
4567
4568<pre>
4569 &lt;resultval&gt; = va_arg &lt;va_list*&gt; &lt;arglist&gt;, &lt;argty&gt;
4570</pre>
4571
4572<h5>Overview:</h5>
4573
4574<p>The '<tt>va_arg</tt>' instruction is used to access arguments passed through
4575the "variable argument" area of a function call. It is used to implement the
4576<tt>va_arg</tt> macro in C.</p>
4577
4578<h5>Arguments:</h5>
4579
4580<p>This instruction takes a <tt>va_list*</tt> value and the type of
4581the argument. It returns a value of the specified argument type and
4582increments the <tt>va_list</tt> to point to the next argument. The
4583actual type of <tt>va_list</tt> is target specific.</p>
4584
4585<h5>Semantics:</h5>
4586
4587<p>The '<tt>va_arg</tt>' instruction loads an argument of the specified
4588type from the specified <tt>va_list</tt> and causes the
4589<tt>va_list</tt> to point to the next argument. For more information,
4590see the variable argument handling <a href="#int_varargs">Intrinsic
4591Functions</a>.</p>
4592
4593<p>It is legal for this instruction to be called in a function which does not
4594take a variable number of arguments, for example, the <tt>vfprintf</tt>
4595function.</p>
4596
4597<p><tt>va_arg</tt> is an LLVM instruction instead of an <a
4598href="#intrinsics">intrinsic function</a> because it takes a type as an
4599argument.</p>
4600
4601<h5>Example:</h5>
4602
4603<p>See the <a href="#int_varargs">variable argument processing</a> section.</p>
4604
Dan Gohman60967192009-01-12 23:12:39 +0000[diff] [blame]4605<p>Note that the code generator does not yet fully support va_arg
4606 on many targets. Also, it does not currently support va_arg with
4607 aggregate types on any target.</p>
4608
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4609</div>
4610
4611<!-- *********************************************************************** -->
4612<div class="doc_section"> <a name="intrinsics">Intrinsic Functions</a> </div>
4613<!-- *********************************************************************** -->
4614
4615<div class="doc_text">
4616
4617<p>LLVM supports the notion of an "intrinsic function". These functions have
4618well known names and semantics and are required to follow certain restrictions.
4619Overall, these intrinsics represent an extension mechanism for the LLVM
4620language that does not require changing all of the transformations in LLVM when
4621adding to the language (or the bitcode reader/writer, the parser, etc...).</p>
4622
4623<p>Intrinsic function names must all start with an "<tt>llvm.</tt>" prefix. This
4624prefix is reserved in LLVM for intrinsic names; thus, function names may not
4625begin with this prefix. Intrinsic functions must always be external functions:
4626you cannot define the body of intrinsic functions. Intrinsic functions may
4627only be used in call or invoke instructions: it is illegal to take the address
4628of an intrinsic function. Additionally, because intrinsic functions are part
4629of the LLVM language, it is required if any are added that they be documented
4630here.</p>
4631
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]4632<p>Some intrinsic functions can be overloaded, i.e., the intrinsic represents
4633a family of functions that perform the same operation but on different data
4634types. Because LLVM can represent over 8 million different integer types,
4635overloading is used commonly to allow an intrinsic function to operate on any
4636integer type. One or more of the argument types or the result type can be
4637overloaded to accept any integer type. Argument types may also be defined as
4638exactly matching a previous argument's type or the result type. This allows an
4639intrinsic function which accepts multiple arguments, but needs all of them to
4640be of the same type, to only be overloaded with respect to a single argument or
4641the result.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4642
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]4643<p>Overloaded intrinsics will have the names of its overloaded argument types
4644encoded into its function name, each preceded by a period. Only those types
4645which are overloaded result in a name suffix. Arguments whose type is matched
4646against another type do not. For example, the <tt>llvm.ctpop</tt> function can
4647take an integer of any width and returns an integer of exactly the same integer
4648width. This leads to a family of functions such as
4649<tt>i8 @llvm.ctpop.i8(i8 %val)</tt> and <tt>i29 @llvm.ctpop.i29(i29 %val)</tt>.
4650Only one type, the return type, is overloaded, and only one type suffix is
4651required. Because the argument's type is matched against the return type, it
4652does not require its own name suffix.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4653
4654<p>To learn how to add an intrinsic function, please see the
4655<a href="ExtendingLLVM.html">Extending LLVM Guide</a>.
4656</p>
4657
4658</div>
4659
4660<!-- ======================================================================= -->
4661<div class="doc_subsection">
4662 <a name="int_varargs">Variable Argument Handling Intrinsics</a>
4663</div>
4664
4665<div class="doc_text">
4666
4667<p>Variable argument support is defined in LLVM with the <a
4668 href="#i_va_arg"><tt>va_arg</tt></a> instruction and these three
4669intrinsic functions. These functions are related to the similarly
4670named macros defined in the <tt>&lt;stdarg.h&gt;</tt> header file.</p>
4671
4672<p>All of these functions operate on arguments that use a
4673target-specific value type "<tt>va_list</tt>". The LLVM assembly
4674language reference manual does not define what this type is, so all
4675transformations should be prepared to handle these functions regardless of
4676the type used.</p>
4677
4678<p>This example shows how the <a href="#i_va_arg"><tt>va_arg</tt></a>
4679instruction and the variable argument handling intrinsic functions are
4680used.</p>
4681
4682<div class="doc_code">
4683<pre>
4684define i32 @test(i32 %X, ...) {
4685 ; Initialize variable argument processing
4686 %ap = alloca i8*
4687 %ap2 = bitcast i8** %ap to i8*
4688 call void @llvm.va_start(i8* %ap2)
4689
4690 ; Read a single integer argument
4691 %tmp = va_arg i8** %ap, i32
4692
4693 ; Demonstrate usage of llvm.va_copy and llvm.va_end
4694 %aq = alloca i8*
4695 %aq2 = bitcast i8** %aq to i8*
4696 call void @llvm.va_copy(i8* %aq2, i8* %ap2)
4697 call void @llvm.va_end(i8* %aq2)
4698
4699 ; Stop processing of arguments.
4700 call void @llvm.va_end(i8* %ap2)
4701 ret i32 %tmp
4702}
4703
4704declare void @llvm.va_start(i8*)
4705declare void @llvm.va_copy(i8*, i8*)
4706declare void @llvm.va_end(i8*)
4707</pre>
4708</div>
4709
4710</div>
4711
4712<!-- _______________________________________________________________________ -->
4713<div class="doc_subsubsection">
4714 <a name="int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a>
4715</div>
4716
4717
4718<div class="doc_text">
4719<h5>Syntax:</h5>
4720<pre> declare void %llvm.va_start(i8* &lt;arglist&gt;)<br></pre>
4721<h5>Overview:</h5>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4722<p>The '<tt>llvm.va_start</tt>' intrinsic initializes
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4723<tt>*&lt;arglist&gt;</tt> for subsequent use by <tt><a
4724href="#i_va_arg">va_arg</a></tt>.</p>
4725
4726<h5>Arguments:</h5>
4727
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4728<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]4729
4730<h5>Semantics:</h5>
4731
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4732<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]4733macro available in C. In a target-dependent way, it initializes the
4734<tt>va_list</tt> element to which the argument points, so that the next call to
4735<tt>va_arg</tt> will produce the first variable argument passed to the function.
4736Unlike the C <tt>va_start</tt> macro, this intrinsic does not need to know the
4737last argument of the function as the compiler can figure that out.</p>
4738
4739</div>
4740
4741<!-- _______________________________________________________________________ -->
4742<div class="doc_subsubsection">
4743 <a name="int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a>
4744</div>
4745
4746<div class="doc_text">
4747<h5>Syntax:</h5>
4748<pre> declare void @llvm.va_end(i8* &lt;arglist&gt;)<br></pre>
4749<h5>Overview:</h5>
4750
4751<p>The '<tt>llvm.va_end</tt>' intrinsic destroys <tt>*&lt;arglist&gt;</tt>,
4752which has been initialized previously with <tt><a href="#int_va_start">llvm.va_start</a></tt>
4753or <tt><a href="#i_va_copy">llvm.va_copy</a></tt>.</p>
4754
4755<h5>Arguments:</h5>
4756
4757<p>The argument is a pointer to a <tt>va_list</tt> to destroy.</p>
4758
4759<h5>Semantics:</h5>
4760
4761<p>The '<tt>llvm.va_end</tt>' intrinsic works just like the <tt>va_end</tt>
4762macro available in C. In a target-dependent way, it destroys the
4763<tt>va_list</tt> element to which the argument points. Calls to <a
4764href="#int_va_start"><tt>llvm.va_start</tt></a> and <a href="#int_va_copy">
4765<tt>llvm.va_copy</tt></a> must be matched exactly with calls to
4766<tt>llvm.va_end</tt>.</p>
4767
4768</div>
4769
4770<!-- _______________________________________________________________________ -->
4771<div class="doc_subsubsection">
4772 <a name="int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a>
4773</div>
4774
4775<div class="doc_text">
4776
4777<h5>Syntax:</h5>
4778
4779<pre>
4780 declare void @llvm.va_copy(i8* &lt;destarglist&gt;, i8* &lt;srcarglist&gt;)
4781</pre>
4782
4783<h5>Overview:</h5>
4784
4785<p>The '<tt>llvm.va_copy</tt>' intrinsic copies the current argument position
4786from the source argument list to the destination argument list.</p>
4787
4788<h5>Arguments:</h5>
4789
4790<p>The first argument is a pointer to a <tt>va_list</tt> element to initialize.
4791The second argument is a pointer to a <tt>va_list</tt> element to copy from.</p>
4792
4793
4794<h5>Semantics:</h5>
4795
4796<p>The '<tt>llvm.va_copy</tt>' intrinsic works just like the <tt>va_copy</tt>
4797macro available in C. In a target-dependent way, it copies the source
4798<tt>va_list</tt> element into the destination <tt>va_list</tt> element. This
4799intrinsic is necessary because the <tt><a href="#int_va_start">
4800llvm.va_start</a></tt> intrinsic may be arbitrarily complex and require, for
4801example, memory allocation.</p>
4802
4803</div>
4804
4805<!-- ======================================================================= -->
4806<div class="doc_subsection">
4807 <a name="int_gc">Accurate Garbage Collection Intrinsics</a>
4808</div>
4809
4810<div class="doc_text">
4811
4812<p>
4813LLVM support for <a href="GarbageCollection.html">Accurate Garbage
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]4814Collection</a> (GC) requires the implementation and generation of these
4815intrinsics.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4816These intrinsics allow identification of <a href="#int_gcroot">GC roots on the
4817stack</a>, as well as garbage collector implementations that require <a
4818href="#int_gcread">read</a> and <a href="#int_gcwrite">write</a> barriers.
4819Front-ends for type-safe garbage collected languages should generate these
4820intrinsics to make use of the LLVM garbage collectors. For more details, see <a
4821href="GarbageCollection.html">Accurate Garbage Collection with LLVM</a>.
4822</p>
Christopher Lambcfe00962007-12-17 01:00:21 +0000[diff] [blame]4823
4824<p>The garbage collection intrinsics only operate on objects in the generic
4825 address space (address space zero).</p>
4826
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4827</div>
4828
4829<!-- _______________________________________________________________________ -->
4830<div class="doc_subsubsection">
4831 <a name="int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a>
4832</div>
4833
4834<div class="doc_text">
4835
4836<h5>Syntax:</h5>
4837
4838<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4839 declare void @llvm.gcroot(i8** %ptrloc, i8* %metadata)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4840</pre>
4841
4842<h5>Overview:</h5>
4843
4844<p>The '<tt>llvm.gcroot</tt>' intrinsic declares the existence of a GC root to
4845the code generator, and allows some metadata to be associated with it.</p>
4846
4847<h5>Arguments:</h5>
4848
4849<p>The first argument specifies the address of a stack object that contains the
4850root pointer. The second pointer (which must be either a constant or a global
4851value address) contains the meta-data to be associated with the root.</p>
4852
4853<h5>Semantics:</h5>
4854
Chris Lattnera7d94ba2008-04-24 05:59:56 +0000[diff] [blame]4855<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]4856location. At compile-time, the code generator generates information to allow
Gordon Henriksen40393542007-12-25 02:31:26 +0000[diff] [blame]4857the runtime to find the pointer at GC safe points. The '<tt>llvm.gcroot</tt>'
4858intrinsic may only be used in a function which <a href="#gc">specifies a GC
4859algorithm</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4860
4861</div>
4862
4863
4864<!-- _______________________________________________________________________ -->
4865<div class="doc_subsubsection">
4866 <a name="int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a>
4867</div>
4868
4869<div class="doc_text">
4870
4871<h5>Syntax:</h5>
4872
4873<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4874 declare i8* @llvm.gcread(i8* %ObjPtr, i8** %Ptr)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4875</pre>
4876
4877<h5>Overview:</h5>
4878
4879<p>The '<tt>llvm.gcread</tt>' intrinsic identifies reads of references from heap
4880locations, allowing garbage collector implementations that require read
4881barriers.</p>
4882
4883<h5>Arguments:</h5>
4884
4885<p>The second argument is the address to read from, which should be an address
4886allocated from the garbage collector. The first object is a pointer to the
4887start of the referenced object, if needed by the language runtime (otherwise
4888null).</p>
4889
4890<h5>Semantics:</h5>
4891
4892<p>The '<tt>llvm.gcread</tt>' intrinsic has the same semantics as a load
4893instruction, but may be replaced with substantially more complex code by the
Gordon Henriksen40393542007-12-25 02:31:26 +0000[diff] [blame]4894garbage collector runtime, as needed. The '<tt>llvm.gcread</tt>' intrinsic
4895may only be used in a function which <a href="#gc">specifies a GC
4896algorithm</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4897
4898</div>
4899
4900
4901<!-- _______________________________________________________________________ -->
4902<div class="doc_subsubsection">
4903 <a name="int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a>
4904</div>
4905
4906<div class="doc_text">
4907
4908<h5>Syntax:</h5>
4909
4910<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4911 declare void @llvm.gcwrite(i8* %P1, i8* %Obj, i8** %P2)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4912</pre>
4913
4914<h5>Overview:</h5>
4915
4916<p>The '<tt>llvm.gcwrite</tt>' intrinsic identifies writes of references to heap
4917locations, allowing garbage collector implementations that require write
4918barriers (such as generational or reference counting collectors).</p>
4919
4920<h5>Arguments:</h5>
4921
4922<p>The first argument is the reference to store, the second is the start of the
4923object to store it to, and the third is the address of the field of Obj to
4924store to. If the runtime does not require a pointer to the object, Obj may be
4925null.</p>
4926
4927<h5>Semantics:</h5>
4928
4929<p>The '<tt>llvm.gcwrite</tt>' intrinsic has the same semantics as a store
4930instruction, but may be replaced with substantially more complex code by the
Gordon Henriksen40393542007-12-25 02:31:26 +0000[diff] [blame]4931garbage collector runtime, as needed. The '<tt>llvm.gcwrite</tt>' intrinsic
4932may only be used in a function which <a href="#gc">specifies a GC
4933algorithm</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4934
4935</div>
4936
4937
4938
4939<!-- ======================================================================= -->
4940<div class="doc_subsection">
4941 <a name="int_codegen">Code Generator Intrinsics</a>
4942</div>
4943
4944<div class="doc_text">
4945<p>
4946These intrinsics are provided by LLVM to expose special features that may only
4947be implemented with code generator support.
4948</p>
4949
4950</div>
4951
4952<!-- _______________________________________________________________________ -->
4953<div class="doc_subsubsection">
4954 <a name="int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a>
4955</div>
4956
4957<div class="doc_text">
4958
4959<h5>Syntax:</h5>
4960<pre>
4961 declare i8 *@llvm.returnaddress(i32 &lt;level&gt;)
4962</pre>
4963
4964<h5>Overview:</h5>
4965
4966<p>
4967The '<tt>llvm.returnaddress</tt>' intrinsic attempts to compute a
4968target-specific value indicating the return address of the current function
4969or one of its callers.
4970</p>
4971
4972<h5>Arguments:</h5>
4973
4974<p>
4975The argument to this intrinsic indicates which function to return the address
4976for. Zero indicates the calling function, one indicates its caller, etc. The
4977argument is <b>required</b> to be a constant integer value.
4978</p>
4979
4980<h5>Semantics:</h5>
4981
4982<p>
4983The '<tt>llvm.returnaddress</tt>' intrinsic either returns a pointer indicating
4984the return address of the specified call frame, or zero if it cannot be
4985identified. The value returned by this intrinsic is likely to be incorrect or 0
4986for arguments other than zero, so it should only be used for debugging purposes.
4987</p>
4988
4989<p>
4990Note that calling this intrinsic does not prevent function inlining or other
4991aggressive transformations, so the value returned may not be that of the obvious
4992source-language caller.
4993</p>
4994</div>
4995
4996
4997<!-- _______________________________________________________________________ -->
4998<div class="doc_subsubsection">
4999 <a name="int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a>
5000</div>
5001
5002<div class="doc_text">
5003
5004<h5>Syntax:</h5>
5005<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]5006 declare i8 *@llvm.frameaddress(i32 &lt;level&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5007</pre>
5008
5009<h5>Overview:</h5>
5010
5011<p>
5012The '<tt>llvm.frameaddress</tt>' intrinsic attempts to return the
5013target-specific frame pointer value for the specified stack frame.
5014</p>
5015
5016<h5>Arguments:</h5>
5017
5018<p>
5019The argument to this intrinsic indicates which function to return the frame
5020pointer for. Zero indicates the calling function, one indicates its caller,
5021etc. The argument is <b>required</b> to be a constant integer value.
5022</p>
5023
5024<h5>Semantics:</h5>
5025
5026<p>
5027The '<tt>llvm.frameaddress</tt>' intrinsic either returns a pointer indicating
5028the frame address of the specified call frame, or zero if it cannot be
5029identified. The value returned by this intrinsic is likely to be incorrect or 0
5030for arguments other than zero, so it should only be used for debugging purposes.
5031</p>
5032
5033<p>
5034Note that calling this intrinsic does not prevent function inlining or other
5035aggressive transformations, so the value returned may not be that of the obvious
5036source-language caller.
5037</p>
5038</div>
5039
5040<!-- _______________________________________________________________________ -->
5041<div class="doc_subsubsection">
5042 <a name="int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a>
5043</div>
5044
5045<div class="doc_text">
5046
5047<h5>Syntax:</h5>
5048<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]5049 declare i8 *@llvm.stacksave()
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5050</pre>
5051
5052<h5>Overview:</h5>
5053
5054<p>
5055The '<tt>llvm.stacksave</tt>' intrinsic is used to remember the current state of
5056the function stack, for use with <a href="#int_stackrestore">
5057<tt>llvm.stackrestore</tt></a>. This is useful for implementing language
5058features like scoped automatic variable sized arrays in C99.
5059</p>
5060
5061<h5>Semantics:</h5>
5062
5063<p>
5064This intrinsic returns a opaque pointer value that can be passed to <a
5065href="#int_stackrestore"><tt>llvm.stackrestore</tt></a>. When an
5066<tt>llvm.stackrestore</tt> intrinsic is executed with a value saved from
5067<tt>llvm.stacksave</tt>, it effectively restores the state of the stack to the
5068state it was in when the <tt>llvm.stacksave</tt> intrinsic executed. In
5069practice, this pops any <a href="#i_alloca">alloca</a> blocks from the stack
5070that were allocated after the <tt>llvm.stacksave</tt> was executed.
5071</p>
5072
5073</div>
5074
5075<!-- _______________________________________________________________________ -->
5076<div class="doc_subsubsection">
5077 <a name="int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a>
5078</div>
5079
5080<div class="doc_text">
5081
5082<h5>Syntax:</h5>
5083<pre>
5084 declare void @llvm.stackrestore(i8 * %ptr)
5085</pre>
5086
5087<h5>Overview:</h5>
5088
5089<p>
5090The '<tt>llvm.stackrestore</tt>' intrinsic is used to restore the state of
5091the function stack to the state it was in when the corresponding <a
5092href="#int_stacksave"><tt>llvm.stacksave</tt></a> intrinsic executed. This is
5093useful for implementing language features like scoped automatic variable sized
5094arrays in C99.
5095</p>
5096
5097<h5>Semantics:</h5>
5098
5099<p>
5100See the description for <a href="#int_stacksave"><tt>llvm.stacksave</tt></a>.
5101</p>
5102
5103</div>
5104
5105
5106<!-- _______________________________________________________________________ -->
5107<div class="doc_subsubsection">
5108 <a name="int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a>
5109</div>
5110
5111<div class="doc_text">
5112
5113<h5>Syntax:</h5>
5114<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]5115 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]5116</pre>
5117
5118<h5>Overview:</h5>
5119
5120
5121<p>
5122The '<tt>llvm.prefetch</tt>' intrinsic is a hint to the code generator to insert
5123a prefetch instruction if supported; otherwise, it is a noop. Prefetches have
5124no
5125effect on the behavior of the program but can change its performance
5126characteristics.
5127</p>
5128
5129<h5>Arguments:</h5>
5130
5131<p>
5132<tt>address</tt> is the address to be prefetched, <tt>rw</tt> is the specifier
5133determining if the fetch should be for a read (0) or write (1), and
5134<tt>locality</tt> is a temporal locality specifier ranging from (0) - no
5135locality, to (3) - extremely local keep in cache. The <tt>rw</tt> and
5136<tt>locality</tt> arguments must be constant integers.
5137</p>
5138
5139<h5>Semantics:</h5>
5140
5141<p>
5142This intrinsic does not modify the behavior of the program. In particular,
5143prefetches cannot trap and do not produce a value. On targets that support this
5144intrinsic, the prefetch can provide hints to the processor cache for better
5145performance.
5146</p>
5147
5148</div>
5149
5150<!-- _______________________________________________________________________ -->
5151<div class="doc_subsubsection">
5152 <a name="int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a>
5153</div>
5154
5155<div class="doc_text">
5156
5157<h5>Syntax:</h5>
5158<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]5159 declare void @llvm.pcmarker(i32 &lt;id&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5160</pre>
5161
5162<h5>Overview:</h5>
5163
5164
5165<p>
5166The '<tt>llvm.pcmarker</tt>' intrinsic is a method to export a Program Counter
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]5167(PC) in a region of
5168code to simulators and other tools. The method is target specific, but it is
5169expected that the marker will use exported symbols to transmit the PC of the
5170marker.
5171The marker makes no guarantees that it will remain with any specific instruction
5172after optimizations. It is possible that the presence of a marker will inhibit
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5173optimizations. The intended use is to be inserted after optimizations to allow
5174correlations of simulation runs.
5175</p>
5176
5177<h5>Arguments:</h5>
5178
5179<p>
5180<tt>id</tt> is a numerical id identifying the marker.
5181</p>
5182
5183<h5>Semantics:</h5>
5184
5185<p>
5186This intrinsic does not modify the behavior of the program. Backends that do not
5187support this intrinisic may ignore it.
5188</p>
5189
5190</div>
5191
5192<!-- _______________________________________________________________________ -->
5193<div class="doc_subsubsection">
5194 <a name="int_readcyclecounter">'<tt>llvm.readcyclecounter</tt>' Intrinsic</a>
5195</div>
5196
5197<div class="doc_text">
5198
5199<h5>Syntax:</h5>
5200<pre>
5201 declare i64 @llvm.readcyclecounter( )
5202</pre>
5203
5204<h5>Overview:</h5>
5205
5206
5207<p>
5208The '<tt>llvm.readcyclecounter</tt>' intrinsic provides access to the cycle
5209counter register (or similar low latency, high accuracy clocks) on those targets
5210that support it. On X86, it should map to RDTSC. On Alpha, it should map to RPCC.
5211As the backing counters overflow quickly (on the order of 9 seconds on alpha), this
5212should only be used for small timings.
5213</p>
5214
5215<h5>Semantics:</h5>
5216
5217<p>
5218When directly supported, reading the cycle counter should not modify any memory.
5219Implementations are allowed to either return a application specific value or a
5220system wide value. On backends without support, this is lowered to a constant 0.
5221</p>
5222
5223</div>
5224
5225<!-- ======================================================================= -->
5226<div class="doc_subsection">
5227 <a name="int_libc">Standard C Library Intrinsics</a>
5228</div>
5229
5230<div class="doc_text">
5231<p>
5232LLVM provides intrinsics for a few important standard C library functions.
5233These intrinsics allow source-language front-ends to pass information about the
5234alignment of the pointer arguments to the code generator, providing opportunity
5235for more efficient code generation.
5236</p>
5237
5238</div>
5239
5240<!-- _______________________________________________________________________ -->
5241<div class="doc_subsubsection">
5242 <a name="int_memcpy">'<tt>llvm.memcpy</tt>' Intrinsic</a>
5243</div>
5244
5245<div class="doc_text">
5246
5247<h5>Syntax:</h5>
Chris Lattner82c2e432008-11-21 16:42:48 +0000[diff] [blame]5248<p>This is an overloaded intrinsic. You can use llvm.memcpy on any integer bit
5249width. Not all targets support all bit widths however.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5250<pre>
Chris Lattner82c2e432008-11-21 16:42:48 +0000[diff] [blame]5251 declare void @llvm.memcpy.i8(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5252 i8 &lt;len&gt;, i32 &lt;align&gt;)
5253 declare void @llvm.memcpy.i16(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5254 i16 &lt;len&gt;, i32 &lt;align&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5255 declare void @llvm.memcpy.i32(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5256 i32 &lt;len&gt;, i32 &lt;align&gt;)
5257 declare void @llvm.memcpy.i64(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5258 i64 &lt;len&gt;, i32 &lt;align&gt;)
5259</pre>
5260
5261<h5>Overview:</h5>
5262
5263<p>
5264The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the source
5265location to the destination location.
5266</p>
5267
5268<p>
5269Note that, unlike the standard libc function, the <tt>llvm.memcpy.*</tt>
5270intrinsics do not return a value, and takes an extra alignment argument.
5271</p>
5272
5273<h5>Arguments:</h5>
5274
5275<p>
5276The first argument is a pointer to the destination, the second is a pointer to
5277the source. The third argument is an integer argument
5278specifying the number of bytes to copy, and the fourth argument is the alignment
5279of the source and destination locations.
5280</p>
5281
5282<p>
5283If the call to this intrinisic has an alignment value that is not 0 or 1, then
5284the caller guarantees that both the source and destination pointers are aligned
5285to that boundary.
5286</p>
5287
5288<h5>Semantics:</h5>
5289
5290<p>
5291The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the source
5292location to the destination location, which are not allowed to overlap. It
5293copies "len" bytes of memory over. If the argument is known to be aligned to
5294some boundary, this can be specified as the fourth argument, otherwise it should
5295be set to 0 or 1.
5296</p>
5297</div>
5298
5299
5300<!-- _______________________________________________________________________ -->
5301<div class="doc_subsubsection">
5302 <a name="int_memmove">'<tt>llvm.memmove</tt>' Intrinsic</a>
5303</div>
5304
5305<div class="doc_text">
5306
5307<h5>Syntax:</h5>
Chris Lattner82c2e432008-11-21 16:42:48 +0000[diff] [blame]5308<p>This is an overloaded intrinsic. You can use llvm.memmove on any integer bit
5309width. Not all targets support all bit widths however.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5310<pre>
Chris Lattner82c2e432008-11-21 16:42:48 +0000[diff] [blame]5311 declare void @llvm.memmove.i8(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5312 i8 &lt;len&gt;, i32 &lt;align&gt;)
5313 declare void @llvm.memmove.i16(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5314 i16 &lt;len&gt;, i32 &lt;align&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5315 declare void @llvm.memmove.i32(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5316 i32 &lt;len&gt;, i32 &lt;align&gt;)
5317 declare void @llvm.memmove.i64(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5318 i64 &lt;len&gt;, i32 &lt;align&gt;)
5319</pre>
5320
5321<h5>Overview:</h5>
5322
5323<p>
5324The '<tt>llvm.memmove.*</tt>' intrinsics move a block of memory from the source
5325location to the destination location. It is similar to the
Chris Lattnerdba16ea2008-01-06 19:51:52 +0000[diff] [blame]5326'<tt>llvm.memcpy</tt>' intrinsic but allows the two memory locations to overlap.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5327</p>
5328
5329<p>
5330Note that, unlike the standard libc function, the <tt>llvm.memmove.*</tt>
5331intrinsics do not return a value, and takes an extra alignment argument.
5332</p>
5333
5334<h5>Arguments:</h5>
5335
5336<p>
5337The first argument is a pointer to the destination, the second is a pointer to
5338the source. The third argument is an integer argument
5339specifying the number of bytes to copy, and the fourth argument is the alignment
5340of the source and destination locations.
5341</p>
5342
5343<p>
5344If the call to this intrinisic has an alignment value that is not 0 or 1, then
5345the caller guarantees that the source and destination pointers are aligned to
5346that boundary.
5347</p>
5348
5349<h5>Semantics:</h5>
5350
5351<p>
5352The '<tt>llvm.memmove.*</tt>' intrinsics copy a block of memory from the source
5353location to the destination location, which may overlap. It
5354copies "len" bytes of memory over. If the argument is known to be aligned to
5355some boundary, this can be specified as the fourth argument, otherwise it should
5356be set to 0 or 1.
5357</p>
5358</div>
5359
5360
5361<!-- _______________________________________________________________________ -->
5362<div class="doc_subsubsection">
5363 <a name="int_memset">'<tt>llvm.memset.*</tt>' Intrinsics</a>
5364</div>
5365
5366<div class="doc_text">
5367
5368<h5>Syntax:</h5>
Chris Lattner82c2e432008-11-21 16:42:48 +0000[diff] [blame]5369<p>This is an overloaded intrinsic. You can use llvm.memset on any integer bit
5370width. Not all targets support all bit widths however.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5371<pre>
Chris Lattner82c2e432008-11-21 16:42:48 +0000[diff] [blame]5372 declare void @llvm.memset.i8(i8 * &lt;dest&gt;, i8 &lt;val&gt;,
5373 i8 &lt;len&gt;, i32 &lt;align&gt;)
5374 declare void @llvm.memset.i16(i8 * &lt;dest&gt;, i8 &lt;val&gt;,
5375 i16 &lt;len&gt;, i32 &lt;align&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5376 declare void @llvm.memset.i32(i8 * &lt;dest&gt;, i8 &lt;val&gt;,
5377 i32 &lt;len&gt;, i32 &lt;align&gt;)
5378 declare void @llvm.memset.i64(i8 * &lt;dest&gt;, i8 &lt;val&gt;,
5379 i64 &lt;len&gt;, i32 &lt;align&gt;)
5380</pre>
5381
5382<h5>Overview:</h5>
5383
5384<p>
5385The '<tt>llvm.memset.*</tt>' intrinsics fill a block of memory with a particular
5386byte value.
5387</p>
5388
5389<p>
5390Note that, unlike the standard libc function, the <tt>llvm.memset</tt> intrinsic
5391does not return a value, and takes an extra alignment argument.
5392</p>
5393
5394<h5>Arguments:</h5>
5395
5396<p>
5397The first argument is a pointer to the destination to fill, the second is the
5398byte value to fill it with, the third argument is an integer
5399argument specifying the number of bytes to fill, and the fourth argument is the
5400known alignment of destination location.
5401</p>
5402
5403<p>
5404If the call to this intrinisic has an alignment value that is not 0 or 1, then
5405the caller guarantees that the destination pointer is aligned to that boundary.
5406</p>
5407
5408<h5>Semantics:</h5>
5409
5410<p>
5411The '<tt>llvm.memset.*</tt>' intrinsics fill "len" bytes of memory starting at
5412the
5413destination location. If the argument is known to be aligned to some boundary,
5414this can be specified as the fourth argument, otherwise it should be set to 0 or
54151.
5416</p>
5417</div>
5418
5419
5420<!-- _______________________________________________________________________ -->
5421<div class="doc_subsubsection">
5422 <a name="int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a>
5423</div>
5424
5425<div class="doc_text">
5426
5427<h5>Syntax:</h5>
Dale Johannesenf9adbb62007-10-02 17:47:38 +0000[diff] [blame]5428<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]5429floating point or vector of floating point type. Not all targets support all
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5430types however.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5431<pre>
Dale Johannesenf9adbb62007-10-02 17:47:38 +0000[diff] [blame]5432 declare float @llvm.sqrt.f32(float %Val)
5433 declare double @llvm.sqrt.f64(double %Val)
5434 declare x86_fp80 @llvm.sqrt.f80(x86_fp80 %Val)
5435 declare fp128 @llvm.sqrt.f128(fp128 %Val)
5436 declare ppc_fp128 @llvm.sqrt.ppcf128(ppc_fp128 %Val)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5437</pre>
5438
5439<h5>Overview:</h5>
5440
5441<p>
5442The '<tt>llvm.sqrt</tt>' intrinsics return the sqrt of the specified operand,
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5443returning the same value as the libm '<tt>sqrt</tt>' functions would. Unlike
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5444<tt>sqrt</tt> in libm, however, <tt>llvm.sqrt</tt> has undefined behavior for
Chris Lattnerf914a002008-01-29 07:00:44 +0000[diff] [blame]5445negative numbers other than -0.0 (which allows for better optimization, because
5446there is no need to worry about errno being set). <tt>llvm.sqrt(-0.0)</tt> is
5447defined to return -0.0 like IEEE sqrt.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5448</p>
5449
5450<h5>Arguments:</h5>
5451
5452<p>
5453The argument and return value are floating point numbers of the same type.
5454</p>
5455
5456<h5>Semantics:</h5>
5457
5458<p>
5459This function returns the sqrt of the specified operand if it is a nonnegative
5460floating point number.
5461</p>
5462</div>
5463
5464<!-- _______________________________________________________________________ -->
5465<div class="doc_subsubsection">
5466 <a name="int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a>
5467</div>
5468
5469<div class="doc_text">
5470
5471<h5>Syntax:</h5>
Dale Johannesenf9adbb62007-10-02 17:47:38 +0000[diff] [blame]5472<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]5473floating point or vector of floating point type. Not all targets support all
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5474types however.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5475<pre>
Dale Johannesenf9adbb62007-10-02 17:47:38 +0000[diff] [blame]5476 declare float @llvm.powi.f32(float %Val, i32 %power)
5477 declare double @llvm.powi.f64(double %Val, i32 %power)
5478 declare x86_fp80 @llvm.powi.f80(x86_fp80 %Val, i32 %power)
5479 declare fp128 @llvm.powi.f128(fp128 %Val, i32 %power)
5480 declare ppc_fp128 @llvm.powi.ppcf128(ppc_fp128 %Val, i32 %power)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5481</pre>
5482
5483<h5>Overview:</h5>
5484
5485<p>
5486The '<tt>llvm.powi.*</tt>' intrinsics return the first operand raised to the
5487specified (positive or negative) power. The order of evaluation of
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5488multiplications is not defined. When a vector of floating point type is
5489used, the second argument remains a scalar integer value.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5490</p>
5491
5492<h5>Arguments:</h5>
5493
5494<p>
5495The second argument is an integer power, and the first is a value to raise to
5496that power.
5497</p>
5498
5499<h5>Semantics:</h5>
5500
5501<p>
5502This function returns the first value raised to the second power with an
5503unspecified sequence of rounding operations.</p>
5504</div>
5505
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5506<!-- _______________________________________________________________________ -->
5507<div class="doc_subsubsection">
5508 <a name="int_sin">'<tt>llvm.sin.*</tt>' Intrinsic</a>
5509</div>
5510
5511<div class="doc_text">
5512
5513<h5>Syntax:</h5>
5514<p>This is an overloaded intrinsic. You can use <tt>llvm.sin</tt> on any
5515floating point or vector of floating point type. Not all targets support all
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5516types however.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5517<pre>
5518 declare float @llvm.sin.f32(float %Val)
5519 declare double @llvm.sin.f64(double %Val)
5520 declare x86_fp80 @llvm.sin.f80(x86_fp80 %Val)
5521 declare fp128 @llvm.sin.f128(fp128 %Val)
5522 declare ppc_fp128 @llvm.sin.ppcf128(ppc_fp128 %Val)
5523</pre>
5524
5525<h5>Overview:</h5>
5526
5527<p>
5528The '<tt>llvm.sin.*</tt>' intrinsics return the sine of the operand.
5529</p>
5530
5531<h5>Arguments:</h5>
5532
5533<p>
5534The argument and return value are floating point numbers of the same type.
5535</p>
5536
5537<h5>Semantics:</h5>
5538
5539<p>
5540This function returns the sine of the specified operand, returning the
5541same values as the libm <tt>sin</tt> functions would, and handles error
Dan Gohmaneaba92e2007-10-17 18:05:13 +0000[diff] [blame]5542conditions in the same way.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5543</div>
5544
5545<!-- _______________________________________________________________________ -->
5546<div class="doc_subsubsection">
5547 <a name="int_cos">'<tt>llvm.cos.*</tt>' Intrinsic</a>
5548</div>
5549
5550<div class="doc_text">
5551
5552<h5>Syntax:</h5>
5553<p>This is an overloaded intrinsic. You can use <tt>llvm.cos</tt> on any
5554floating point or vector of floating point type. Not all targets support all
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5555types however.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5556<pre>
5557 declare float @llvm.cos.f32(float %Val)
5558 declare double @llvm.cos.f64(double %Val)
5559 declare x86_fp80 @llvm.cos.f80(x86_fp80 %Val)
5560 declare fp128 @llvm.cos.f128(fp128 %Val)
5561 declare ppc_fp128 @llvm.cos.ppcf128(ppc_fp128 %Val)
5562</pre>
5563
5564<h5>Overview:</h5>
5565
5566<p>
5567The '<tt>llvm.cos.*</tt>' intrinsics return the cosine of the operand.
5568</p>
5569
5570<h5>Arguments:</h5>
5571
5572<p>
5573The argument and return value are floating point numbers of the same type.
5574</p>
5575
5576<h5>Semantics:</h5>
5577
5578<p>
5579This function returns the cosine of the specified operand, returning the
5580same values as the libm <tt>cos</tt> functions would, and handles error
Dan Gohmaneaba92e2007-10-17 18:05:13 +0000[diff] [blame]5581conditions in the same way.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5582</div>
5583
5584<!-- _______________________________________________________________________ -->
5585<div class="doc_subsubsection">
5586 <a name="int_pow">'<tt>llvm.pow.*</tt>' Intrinsic</a>
5587</div>
5588
5589<div class="doc_text">
5590
5591<h5>Syntax:</h5>
5592<p>This is an overloaded intrinsic. You can use <tt>llvm.pow</tt> on any
5593floating point or vector of floating point type. Not all targets support all
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5594types however.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5595<pre>
5596 declare float @llvm.pow.f32(float %Val, float %Power)
5597 declare double @llvm.pow.f64(double %Val, double %Power)
5598 declare x86_fp80 @llvm.pow.f80(x86_fp80 %Val, x86_fp80 %Power)
5599 declare fp128 @llvm.pow.f128(fp128 %Val, fp128 %Power)
5600 declare ppc_fp128 @llvm.pow.ppcf128(ppc_fp128 %Val, ppc_fp128 Power)
5601</pre>
5602
5603<h5>Overview:</h5>
5604
5605<p>
5606The '<tt>llvm.pow.*</tt>' intrinsics return the first operand raised to the
5607specified (positive or negative) power.
5608</p>
5609
5610<h5>Arguments:</h5>
5611
5612<p>
5613The second argument is a floating point power, and the first is a value to
5614raise to that power.
5615</p>
5616
5617<h5>Semantics:</h5>
5618
5619<p>
5620This function returns the first value raised to the second power,
5621returning the
5622same values as the libm <tt>pow</tt> functions would, and handles error
Dan Gohmaneaba92e2007-10-17 18:05:13 +0000[diff] [blame]5623conditions in the same way.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5624</div>
5625
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5626
5627<!-- ======================================================================= -->
5628<div class="doc_subsection">
5629 <a name="int_manip">Bit Manipulation Intrinsics</a>
5630</div>
5631
5632<div class="doc_text">
5633<p>
5634LLVM provides intrinsics for a few important bit manipulation operations.
5635These allow efficient code generation for some algorithms.
5636</p>
5637
5638</div>
5639
5640<!-- _______________________________________________________________________ -->
5641<div class="doc_subsubsection">
5642 <a name="int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a>
5643</div>
5644
5645<div class="doc_text">
5646
5647<h5>Syntax:</h5>
5648<p>This is an overloaded intrinsic function. You can use bswap on any integer
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5649type that is an even number of bytes (i.e. BitWidth % 16 == 0).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5650<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5651 declare i16 @llvm.bswap.i16(i16 &lt;id&gt;)
5652 declare i32 @llvm.bswap.i32(i32 &lt;id&gt;)
5653 declare i64 @llvm.bswap.i64(i64 &lt;id&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5654</pre>
5655
5656<h5>Overview:</h5>
5657
5658<p>
5659The '<tt>llvm.bswap</tt>' family of intrinsics is used to byte swap integer
5660values with an even number of bytes (positive multiple of 16 bits). These are
5661useful for performing operations on data that is not in the target's native
5662byte order.
5663</p>
5664
5665<h5>Semantics:</h5>
5666
5667<p>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5668The <tt>llvm.bswap.i16</tt> intrinsic returns an i16 value that has the high
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5669and low byte of the input i16 swapped. Similarly, the <tt>llvm.bswap.i32</tt>
5670intrinsic returns an i32 value that has the four bytes of the input i32
5671swapped, 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]5672i32 will have its bytes in 3, 2, 1, 0 order. The <tt>llvm.bswap.i48</tt>,
5673<tt>llvm.bswap.i64</tt> and other intrinsics extend this concept to
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5674additional even-byte lengths (6 bytes, 8 bytes and more, respectively).
5675</p>
5676
5677</div>
5678
5679<!-- _______________________________________________________________________ -->
5680<div class="doc_subsubsection">
5681 <a name="int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic</a>
5682</div>
5683
5684<div class="doc_text">
5685
5686<h5>Syntax:</h5>
5687<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]5688width. Not all targets support all bit widths however.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5689<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5690 declare i8 @llvm.ctpop.i8 (i8 &lt;src&gt;)
5691 declare i16 @llvm.ctpop.i16(i16 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5692 declare i32 @llvm.ctpop.i32(i32 &lt;src&gt;)
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5693 declare i64 @llvm.ctpop.i64(i64 &lt;src&gt;)
5694 declare i256 @llvm.ctpop.i256(i256 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5695</pre>
5696
5697<h5>Overview:</h5>
5698
5699<p>
5700The '<tt>llvm.ctpop</tt>' family of intrinsics counts the number of bits set in a
5701value.
5702</p>
5703
5704<h5>Arguments:</h5>
5705
5706<p>
5707The only argument is the value to be counted. The argument may be of any
5708integer type. The return type must match the argument type.
5709</p>
5710
5711<h5>Semantics:</h5>
5712
5713<p>
5714The '<tt>llvm.ctpop</tt>' intrinsic counts the 1's in a variable.
5715</p>
5716</div>
5717
5718<!-- _______________________________________________________________________ -->
5719<div class="doc_subsubsection">
5720 <a name="int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic</a>
5721</div>
5722
5723<div class="doc_text">
5724
5725<h5>Syntax:</h5>
5726<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]5727integer bit width. Not all targets support all bit widths however.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5728<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5729 declare i8 @llvm.ctlz.i8 (i8 &lt;src&gt;)
5730 declare i16 @llvm.ctlz.i16(i16 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5731 declare i32 @llvm.ctlz.i32(i32 &lt;src&gt;)
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5732 declare i64 @llvm.ctlz.i64(i64 &lt;src&gt;)
5733 declare i256 @llvm.ctlz.i256(i256 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5734</pre>
5735
5736<h5>Overview:</h5>
5737
5738<p>
5739The '<tt>llvm.ctlz</tt>' family of intrinsic functions counts the number of
5740leading zeros in a variable.
5741</p>
5742
5743<h5>Arguments:</h5>
5744
5745<p>
5746The only argument is the value to be counted. The argument may be of any
5747integer type. The return type must match the argument type.
5748</p>
5749
5750<h5>Semantics:</h5>
5751
5752<p>
5753The '<tt>llvm.ctlz</tt>' intrinsic counts the leading (most significant) zeros
5754in a variable. If the src == 0 then the result is the size in bits of the type
5755of src. For example, <tt>llvm.ctlz(i32 2) = 30</tt>.
5756</p>
5757</div>
5758
5759
5760
5761<!-- _______________________________________________________________________ -->
5762<div class="doc_subsubsection">
5763 <a name="int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic</a>
5764</div>
5765
5766<div class="doc_text">
5767
5768<h5>Syntax:</h5>
5769<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]5770integer bit width. Not all targets support all bit widths however.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5771<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5772 declare i8 @llvm.cttz.i8 (i8 &lt;src&gt;)
5773 declare i16 @llvm.cttz.i16(i16 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5774 declare i32 @llvm.cttz.i32(i32 &lt;src&gt;)
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5775 declare i64 @llvm.cttz.i64(i64 &lt;src&gt;)
5776 declare i256 @llvm.cttz.i256(i256 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5777</pre>
5778
5779<h5>Overview:</h5>
5780
5781<p>
5782The '<tt>llvm.cttz</tt>' family of intrinsic functions counts the number of
5783trailing zeros.
5784</p>
5785
5786<h5>Arguments:</h5>
5787
5788<p>
5789The only argument is the value to be counted. The argument may be of any
5790integer type. The return type must match the argument type.
5791</p>
5792
5793<h5>Semantics:</h5>
5794
5795<p>
5796The '<tt>llvm.cttz</tt>' intrinsic counts the trailing (least significant) zeros
5797in a variable. If the src == 0 then the result is the size in bits of the type
5798of src. For example, <tt>llvm.cttz(2) = 1</tt>.
5799</p>
5800</div>
5801
5802<!-- _______________________________________________________________________ -->
5803<div class="doc_subsubsection">
5804 <a name="int_part_select">'<tt>llvm.part.select.*</tt>' Intrinsic</a>
5805</div>
5806
5807<div class="doc_text">
5808
5809<h5>Syntax:</h5>
5810<p>This is an overloaded intrinsic. You can use <tt>llvm.part.select</tt>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5811on any integer bit width.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5812<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5813 declare i17 @llvm.part.select.i17 (i17 %val, i32 %loBit, i32 %hiBit)
5814 declare i29 @llvm.part.select.i29 (i29 %val, i32 %loBit, i32 %hiBit)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5815</pre>
5816
5817<h5>Overview:</h5>
5818<p>The '<tt>llvm.part.select</tt>' family of intrinsic functions selects a
5819range of bits from an integer value and returns them in the same bit width as
5820the original value.</p>
5821
5822<h5>Arguments:</h5>
5823<p>The first argument, <tt>%val</tt> and the result may be integer types of
5824any bit width but they must have the same bit width. The second and third
5825arguments must be <tt>i32</tt> type since they specify only a bit index.</p>
5826
5827<h5>Semantics:</h5>
5828<p>The operation of the '<tt>llvm.part.select</tt>' intrinsic has two modes
5829of operation: forwards and reverse. If <tt>%loBit</tt> is greater than
5830<tt>%hiBits</tt> then the intrinsic operates in reverse mode. Otherwise it
5831operates in forward mode.</p>
5832<p>In forward mode, this intrinsic is the equivalent of shifting <tt>%val</tt>
5833right by <tt>%loBit</tt> bits and then ANDing it with a mask with
5834only the <tt>%hiBit - %loBit</tt> bits set, as follows:</p>
5835<ol>
5836 <li>The <tt>%val</tt> is shifted right (LSHR) by the number of bits specified
5837 by <tt>%loBits</tt>. This normalizes the value to the low order bits.</li>
5838 <li>The <tt>%loBits</tt> value is subtracted from the <tt>%hiBits</tt> value
5839 to determine the number of bits to retain.</li>
5840 <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]5841 <li>The mask is ANDed with <tt>%val</tt> to produce the result.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5842</ol>
5843<p>In reverse mode, a similar computation is made except that the bits are
5844returned in the reverse order. So, for example, if <tt>X</tt> has the value
5845<tt>i16 0x0ACF (101011001111)</tt> and we apply
5846<tt>part.select(i16 X, 8, 3)</tt> to it, we get back the value
5847<tt>i16 0x0026 (000000100110)</tt>.</p>
5848</div>
5849
5850<div class="doc_subsubsection">
5851 <a name="int_part_set">'<tt>llvm.part.set.*</tt>' Intrinsic</a>
5852</div>
5853
5854<div class="doc_text">
5855
5856<h5>Syntax:</h5>
5857<p>This is an overloaded intrinsic. You can use <tt>llvm.part.set</tt>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5858on any integer bit width.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5859<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5860 declare i17 @llvm.part.set.i17.i9 (i17 %val, i9 %repl, i32 %lo, i32 %hi)
5861 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]5862</pre>
5863
5864<h5>Overview:</h5>
5865<p>The '<tt>llvm.part.set</tt>' family of intrinsic functions replaces a range
5866of bits in an integer value with another integer value. It returns the integer
5867with the replaced bits.</p>
5868
5869<h5>Arguments:</h5>
5870<p>The first argument, <tt>%val</tt> and the result may be integer types of
5871any bit width but they must have the same bit width. <tt>%val</tt> is the value
5872whose bits will be replaced. The second argument, <tt>%repl</tt> may be an
5873integer of any bit width. The third and fourth arguments must be <tt>i32</tt>
5874type since they specify only a bit index.</p>
5875
5876<h5>Semantics:</h5>
5877<p>The operation of the '<tt>llvm.part.set</tt>' intrinsic has two modes
5878of operation: forwards and reverse. If <tt>%lo</tt> is greater than
5879<tt>%hi</tt> then the intrinsic operates in reverse mode. Otherwise it
5880operates in forward mode.</p>
5881<p>For both modes, the <tt>%repl</tt> value is prepared for use by either
5882truncating it down to the size of the replacement area or zero extending it
5883up to that size.</p>
5884<p>In forward mode, the bits between <tt>%lo</tt> and <tt>%hi</tt> (inclusive)
5885are replaced with corresponding bits from <tt>%repl</tt>. That is the 0th bit
5886in <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]5887to the <tt>%hi</tt>th bit.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5888<p>In reverse mode, a similar computation is made except that the bits are
5889reversed. That is, the <tt>0</tt>th bit in <tt>%repl</tt> replaces the
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5890<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]5891<h5>Examples:</h5>
5892<pre>
5893 llvm.part.set(0xFFFF, 0, 4, 7) -&gt; 0xFF0F
5894 llvm.part.set(0xFFFF, 0, 7, 4) -&gt; 0xFF0F
5895 llvm.part.set(0xFFFF, 1, 7, 4) -&gt; 0xFF8F
5896 llvm.part.set(0xFFFF, F, 8, 3) -&gt; 0xFFE7
5897 llvm.part.set(0xFFFF, 0, 3, 8) -&gt; 0xFE07
5898</pre>
5899</div>
5900
5901<!-- ======================================================================= -->
5902<div class="doc_subsection">
5903 <a name="int_debugger">Debugger Intrinsics</a>
5904</div>
5905
5906<div class="doc_text">
5907<p>
5908The LLVM debugger intrinsics (which all start with <tt>llvm.dbg.</tt> prefix),
5909are described in the <a
5910href="SourceLevelDebugging.html#format_common_intrinsics">LLVM Source Level
5911Debugging</a> document.
5912</p>
5913</div>
5914
5915
5916<!-- ======================================================================= -->
5917<div class="doc_subsection">
5918 <a name="int_eh">Exception Handling Intrinsics</a>
5919</div>
5920
5921<div class="doc_text">
5922<p> The LLVM exception handling intrinsics (which all start with
5923<tt>llvm.eh.</tt> prefix), are described in the <a
5924href="ExceptionHandling.html#format_common_intrinsics">LLVM Exception
5925Handling</a> document. </p>
5926</div>
5927
5928<!-- ======================================================================= -->
5929<div class="doc_subsection">
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5930 <a name="int_trampoline">Trampoline Intrinsic</a>
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5931</div>
5932
5933<div class="doc_text">
5934<p>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5935 This intrinsic makes it possible to excise one parameter, marked with
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5936 the <tt>nest</tt> attribute, from a function. The result is a callable
5937 function pointer lacking the nest parameter - the caller does not need
5938 to provide a value for it. Instead, the value to use is stored in
5939 advance in a "trampoline", a block of memory usually allocated
5940 on the stack, which also contains code to splice the nest value into the
5941 argument list. This is used to implement the GCC nested function address
5942 extension.
5943</p>
5944<p>
5945 For example, if the function is
5946 <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]5947 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]5948<pre>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5949 %tramp = alloca [10 x i8], align 4 ; size and alignment only correct for X86
5950 %tramp1 = getelementptr [10 x i8]* %tramp, i32 0, i32 0
5951 %p = call i8* @llvm.init.trampoline( i8* %tramp1, i8* bitcast (i32 (i8* nest , i32, i32)* @f to i8*), i8* %nval )
5952 %fp = bitcast i8* %p to i32 (i32, i32)*
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5953</pre>
Bill Wendlinge262b4c2007-09-22 09:23:55 +0000[diff] [blame]5954 <p>The call <tt>%val = call i32 %fp( i32 %x, i32 %y )</tt> is then equivalent
5955 to <tt>%val = call i32 %f( i8* %nval, i32 %x, i32 %y )</tt>.</p>
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5956</div>
5957
5958<!-- _______________________________________________________________________ -->
5959<div class="doc_subsubsection">
5960 <a name="int_it">'<tt>llvm.init.trampoline</tt>' Intrinsic</a>
5961</div>
5962<div class="doc_text">
5963<h5>Syntax:</h5>
5964<pre>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5965declare 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]5966</pre>
5967<h5>Overview:</h5>
5968<p>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5969 This fills the memory pointed to by <tt>tramp</tt> with code
5970 and returns a function pointer suitable for executing it.
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5971</p>
5972<h5>Arguments:</h5>
5973<p>
5974 The <tt>llvm.init.trampoline</tt> intrinsic takes three arguments, all
5975 pointers. The <tt>tramp</tt> argument must point to a sufficiently large
5976 and sufficiently aligned block of memory; this memory is written to by the
Duncan Sands35012212007-08-22 23:39:54 +0000[diff] [blame]5977 intrinsic. Note that the size and the alignment are target-specific - LLVM
5978 currently provides no portable way of determining them, so a front-end that
5979 generates this intrinsic needs to have some target-specific knowledge.
5980 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]5981</p>
5982<h5>Semantics:</h5>
5983<p>
5984 The block of memory pointed to by <tt>tramp</tt> is filled with target
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5985 dependent code, turning it into a function. A pointer to this function is
5986 returned, but needs to be bitcast to an
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5987 <a href="#int_trampoline">appropriate function pointer type</a>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5988 before being called. The new function's signature is the same as that of
5989 <tt>func</tt> with any arguments marked with the <tt>nest</tt> attribute
5990 removed. At most one such <tt>nest</tt> argument is allowed, and it must be
5991 of pointer type. Calling the new function is equivalent to calling
5992 <tt>func</tt> with the same argument list, but with <tt>nval</tt> used for the
5993 missing <tt>nest</tt> argument. If, after calling
5994 <tt>llvm.init.trampoline</tt>, the memory pointed to by <tt>tramp</tt> is
5995 modified, then the effect of any later call to the returned function pointer is
5996 undefined.
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5997</p>
5998</div>
5999
6000<!-- ======================================================================= -->
6001<div class="doc_subsection">
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]6002 <a name="int_atomics">Atomic Operations and Synchronization Intrinsics</a>
6003</div>
6004
6005<div class="doc_text">
6006<p>
6007 These intrinsic functions expand the "universal IR" of LLVM to represent
6008 hardware constructs for atomic operations and memory synchronization. This
6009 provides an interface to the hardware, not an interface to the programmer. It
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]6010 is aimed at a low enough level to allow any programming models or APIs
6011 (Application Programming Interfaces) which
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]6012 need atomic behaviors to map cleanly onto it. It is also modeled primarily on
6013 hardware behavior. Just as hardware provides a "universal IR" for source
6014 languages, it also provides a starting point for developing a "universal"
6015 atomic operation and synchronization IR.
6016</p>
6017<p>
6018 These do <em>not</em> form an API such as high-level threading libraries,
6019 software transaction memory systems, atomic primitives, and intrinsic
6020 functions as found in BSD, GNU libc, atomic_ops, APR, and other system and
6021 application libraries. The hardware interface provided by LLVM should allow
6022 a clean implementation of all of these APIs and parallel programming models.
6023 No one model or paradigm should be selected above others unless the hardware
6024 itself ubiquitously does so.
6025
6026</p>
6027</div>
6028
6029<!-- _______________________________________________________________________ -->
6030<div class="doc_subsubsection">
6031 <a name="int_memory_barrier">'<tt>llvm.memory.barrier</tt>' Intrinsic</a>
6032</div>
6033<div class="doc_text">
6034<h5>Syntax:</h5>
6035<pre>
6036declare void @llvm.memory.barrier( i1 &lt;ll&gt;, i1 &lt;ls&gt;, i1 &lt;sl&gt;, i1 &lt;ss&gt;,
6037i1 &lt;device&gt; )
6038
6039</pre>
6040<h5>Overview:</h5>
6041<p>
6042 The <tt>llvm.memory.barrier</tt> intrinsic guarantees ordering between
6043 specific pairs of memory access types.
6044</p>
6045<h5>Arguments:</h5>
6046<p>
6047 The <tt>llvm.memory.barrier</tt> intrinsic requires five boolean arguments.
6048 The first four arguments enables a specific barrier as listed below. The fith
6049 argument specifies that the barrier applies to io or device or uncached memory.
6050
6051</p>
6052 <ul>
6053 <li><tt>ll</tt>: load-load barrier</li>
6054 <li><tt>ls</tt>: load-store barrier</li>
6055 <li><tt>sl</tt>: store-load barrier</li>
6056 <li><tt>ss</tt>: store-store barrier</li>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]6057 <li><tt>device</tt>: barrier applies to device and uncached memory also.</li>
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]6058 </ul>
6059<h5>Semantics:</h5>
6060<p>
6061 This intrinsic causes the system to enforce some ordering constraints upon
6062 the loads and stores of the program. This barrier does not indicate
6063 <em>when</em> any events will occur, it only enforces an <em>order</em> in
6064 which they occur. For any of the specified pairs of load and store operations
6065 (f.ex. load-load, or store-load), all of the first operations preceding the
6066 barrier will complete before any of the second operations succeeding the
6067 barrier begin. Specifically the semantics for each pairing is as follows:
6068</p>
6069 <ul>
6070 <li><tt>ll</tt>: All loads before the barrier must complete before any load
6071 after the barrier begins.</li>
6072
6073 <li><tt>ls</tt>: All loads before the barrier must complete before any
6074 store after the barrier begins.</li>
6075 <li><tt>ss</tt>: All stores before the barrier must complete before any
6076 store after the barrier begins.</li>
6077 <li><tt>sl</tt>: All stores before the barrier must complete before any
6078 load after the barrier begins.</li>
6079 </ul>
6080<p>
6081 These semantics are applied with a logical "and" behavior when more than one
6082 is enabled in a single memory barrier intrinsic.
6083</p>
6084<p>
6085 Backends may implement stronger barriers than those requested when they do not
6086 support as fine grained a barrier as requested. Some architectures do not
6087 need all types of barriers and on such architectures, these become noops.
6088</p>
6089<h5>Example:</h5>
6090<pre>
6091%ptr = malloc i32
6092 store i32 4, %ptr
6093
6094%result1 = load i32* %ptr <i>; yields {i32}:result1 = 4</i>
6095 call void @llvm.memory.barrier( i1 false, i1 true, i1 false, i1 false )
6096 <i>; guarantee the above finishes</i>
6097 store i32 8, %ptr <i>; before this begins</i>
6098</pre>
6099</div>
6100
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6101<!-- _______________________________________________________________________ -->
6102<div class="doc_subsubsection">
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6103 <a name="int_atomic_cmp_swap">'<tt>llvm.atomic.cmp.swap.*</tt>' Intrinsic</a>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6104</div>
6105<div class="doc_text">
6106<h5>Syntax:</h5>
6107<p>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6108 This is an overloaded intrinsic. You can use <tt>llvm.atomic.cmp.swap</tt> on
6109 any integer bit width and for different address spaces. Not all targets
6110 support all bit widths however.</p>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6111
6112<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6113declare i8 @llvm.atomic.cmp.swap.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;cmp&gt;, i8 &lt;val&gt; )
6114declare i16 @llvm.atomic.cmp.swap.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;cmp&gt;, i16 &lt;val&gt; )
6115declare i32 @llvm.atomic.cmp.swap.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;cmp&gt;, i32 &lt;val&gt; )
6116declare 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]6117
6118</pre>
6119<h5>Overview:</h5>
6120<p>
6121 This loads a value in memory and compares it to a given value. If they are
6122 equal, it stores a new value into the memory.
6123</p>
6124<h5>Arguments:</h5>
6125<p>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6126 The <tt>llvm.atomic.cmp.swap</tt> intrinsic takes three arguments. The result as
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6127 well as both <tt>cmp</tt> and <tt>val</tt> must be integer values with the
6128 same bit width. The <tt>ptr</tt> argument must be a pointer to a value of
6129 this integer type. While any bit width integer may be used, targets may only
6130 lower representations they support in hardware.
6131
6132</p>
6133<h5>Semantics:</h5>
6134<p>
6135 This entire intrinsic must be executed atomically. It first loads the value
6136 in memory pointed to by <tt>ptr</tt> and compares it with the value
6137 <tt>cmp</tt>. If they are equal, <tt>val</tt> is stored into the memory. The
6138 loaded value is yielded in all cases. This provides the equivalent of an
6139 atomic compare-and-swap operation within the SSA framework.
6140</p>
6141<h5>Examples:</h5>
6142
6143<pre>
6144%ptr = malloc i32
6145 store i32 4, %ptr
6146
6147%val1 = add i32 4, 4
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6148%result1 = call i32 @llvm.atomic.cmp.swap.i32.p0i32( i32* %ptr, i32 4, %val1 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6149 <i>; yields {i32}:result1 = 4</i>
6150%stored1 = icmp eq i32 %result1, 4 <i>; yields {i1}:stored1 = true</i>
6151%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 8</i>
6152
6153%val2 = add i32 1, 1
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6154%result2 = call i32 @llvm.atomic.cmp.swap.i32.p0i32( i32* %ptr, i32 5, %val2 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6155 <i>; yields {i32}:result2 = 8</i>
6156%stored2 = icmp eq i32 %result2, 5 <i>; yields {i1}:stored2 = false</i>
6157
6158%memval2 = load i32* %ptr <i>; yields {i32}:memval2 = 8</i>
6159</pre>
6160</div>
6161
6162<!-- _______________________________________________________________________ -->
6163<div class="doc_subsubsection">
6164 <a name="int_atomic_swap">'<tt>llvm.atomic.swap.*</tt>' Intrinsic</a>
6165</div>
6166<div class="doc_text">
6167<h5>Syntax:</h5>
6168
6169<p>
6170 This is an overloaded intrinsic. You can use <tt>llvm.atomic.swap</tt> on any
6171 integer bit width. Not all targets support all bit widths however.</p>
6172<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6173declare i8 @llvm.atomic.swap.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;val&gt; )
6174declare i16 @llvm.atomic.swap.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;val&gt; )
6175declare i32 @llvm.atomic.swap.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;val&gt; )
6176declare i64 @llvm.atomic.swap.i64.p0i64( i64* &lt;ptr&gt;, i64 &lt;val&gt; )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6177
6178</pre>
6179<h5>Overview:</h5>
6180<p>
6181 This intrinsic loads the value stored in memory at <tt>ptr</tt> and yields
6182 the value from memory. It then stores the value in <tt>val</tt> in the memory
6183 at <tt>ptr</tt>.
6184</p>
6185<h5>Arguments:</h5>
6186
6187<p>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6188 The <tt>llvm.atomic.swap</tt> intrinsic takes two arguments. Both the
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6189 <tt>val</tt> argument and the result must be integers of the same bit width.
6190 The first argument, <tt>ptr</tt>, must be a pointer to a value of this
6191 integer type. The targets may only lower integer representations they
6192 support.
6193</p>
6194<h5>Semantics:</h5>
6195<p>
6196 This intrinsic loads the value pointed to by <tt>ptr</tt>, yields it, and
6197 stores <tt>val</tt> back into <tt>ptr</tt> atomically. This provides the
6198 equivalent of an atomic swap operation within the SSA framework.
6199
6200</p>
6201<h5>Examples:</h5>
6202<pre>
6203%ptr = malloc i32
6204 store i32 4, %ptr
6205
6206%val1 = add i32 4, 4
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6207%result1 = call i32 @llvm.atomic.swap.i32.p0i32( i32* %ptr, i32 %val1 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6208 <i>; yields {i32}:result1 = 4</i>
6209%stored1 = icmp eq i32 %result1, 4 <i>; yields {i1}:stored1 = true</i>
6210%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 8</i>
6211
6212%val2 = add i32 1, 1
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6213%result2 = call i32 @llvm.atomic.swap.i32.p0i32( i32* %ptr, i32 %val2 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6214 <i>; yields {i32}:result2 = 8</i>
6215
6216%stored2 = icmp eq i32 %result2, 8 <i>; yields {i1}:stored2 = true</i>
6217%memval2 = load i32* %ptr <i>; yields {i32}:memval2 = 2</i>
6218</pre>
6219</div>
6220
6221<!-- _______________________________________________________________________ -->
6222<div class="doc_subsubsection">
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6223 <a name="int_atomic_load_add">'<tt>llvm.atomic.load.add.*</tt>' Intrinsic</a>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6224
6225</div>
6226<div class="doc_text">
6227<h5>Syntax:</h5>
6228<p>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6229 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]6230 integer bit width. Not all targets support all bit widths however.</p>
6231<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6232declare i8 @llvm.atomic.load.add.i8..p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6233declare i16 @llvm.atomic.load.add.i16..p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6234declare i32 @llvm.atomic.load.add.i32..p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6235declare 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]6236
6237</pre>
6238<h5>Overview:</h5>
6239<p>
6240 This intrinsic adds <tt>delta</tt> to the value stored in memory at
6241 <tt>ptr</tt>. It yields the original value at <tt>ptr</tt>.
6242</p>
6243<h5>Arguments:</h5>
6244<p>
6245
6246 The intrinsic takes two arguments, the first a pointer to an integer value
6247 and the second an integer value. The result is also an integer value. These
6248 integer types can have any bit width, but they must all have the same bit
6249 width. The targets may only lower integer representations they support.
6250</p>
6251<h5>Semantics:</h5>
6252<p>
6253 This intrinsic does a series of operations atomically. It first loads the
6254 value stored at <tt>ptr</tt>. It then adds <tt>delta</tt>, stores the result
6255 to <tt>ptr</tt>. It yields the original value stored at <tt>ptr</tt>.
6256</p>
6257
6258<h5>Examples:</h5>
6259<pre>
6260%ptr = malloc i32
6261 store i32 4, %ptr
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6262%result1 = call i32 @llvm.atomic.load.add.i32.p0i32( i32* %ptr, i32 4 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6263 <i>; yields {i32}:result1 = 4</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6264%result2 = call i32 @llvm.atomic.load.add.i32.p0i32( i32* %ptr, i32 2 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6265 <i>; yields {i32}:result2 = 8</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6266%result3 = call i32 @llvm.atomic.load.add.i32.p0i32( i32* %ptr, i32 5 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6267 <i>; yields {i32}:result3 = 10</i>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6268%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 15</i>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6269</pre>
6270</div>
6271
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6272<!-- _______________________________________________________________________ -->
6273<div class="doc_subsubsection">
6274 <a name="int_atomic_load_sub">'<tt>llvm.atomic.load.sub.*</tt>' Intrinsic</a>
6275
6276</div>
6277<div class="doc_text">
6278<h5>Syntax:</h5>
6279<p>
6280 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]6281 any integer bit width and for different address spaces. Not all targets
6282 support all bit widths however.</p>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6283<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6284declare i8 @llvm.atomic.load.sub.i8.p0i32( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6285declare i16 @llvm.atomic.load.sub.i16.p0i32( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6286declare i32 @llvm.atomic.load.sub.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6287declare 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]6288
6289</pre>
6290<h5>Overview:</h5>
6291<p>
6292 This intrinsic subtracts <tt>delta</tt> to the value stored in memory at
6293 <tt>ptr</tt>. It yields the original value at <tt>ptr</tt>.
6294</p>
6295<h5>Arguments:</h5>
6296<p>
6297
6298 The intrinsic takes two arguments, the first a pointer to an integer value
6299 and the second an integer value. The result is also an integer value. These
6300 integer types can have any bit width, but they must all have the same bit
6301 width. The targets may only lower integer representations they support.
6302</p>
6303<h5>Semantics:</h5>
6304<p>
6305 This intrinsic does a series of operations atomically. It first loads the
6306 value stored at <tt>ptr</tt>. It then subtracts <tt>delta</tt>, stores the
6307 result to <tt>ptr</tt>. It yields the original value stored at <tt>ptr</tt>.
6308</p>
6309
6310<h5>Examples:</h5>
6311<pre>
6312%ptr = malloc i32
6313 store i32 8, %ptr
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6314%result1 = call i32 @llvm.atomic.load.sub.i32.p0i32( i32* %ptr, i32 4 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6315 <i>; yields {i32}:result1 = 8</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6316%result2 = call i32 @llvm.atomic.load.sub.i32.p0i32( i32* %ptr, i32 2 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6317 <i>; yields {i32}:result2 = 4</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6318%result3 = call i32 @llvm.atomic.load.sub.i32.p0i32( i32* %ptr, i32 5 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6319 <i>; yields {i32}:result3 = 2</i>
6320%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = -3</i>
6321</pre>
6322</div>
6323
6324<!-- _______________________________________________________________________ -->
6325<div class="doc_subsubsection">
6326 <a name="int_atomic_load_and">'<tt>llvm.atomic.load.and.*</tt>' Intrinsic</a><br>
6327 <a name="int_atomic_load_nand">'<tt>llvm.atomic.load.nand.*</tt>' Intrinsic</a><br>
6328 <a name="int_atomic_load_or">'<tt>llvm.atomic.load.or.*</tt>' Intrinsic</a><br>
6329 <a name="int_atomic_load_xor">'<tt>llvm.atomic.load.xor.*</tt>' Intrinsic</a><br>
6330
6331</div>
6332<div class="doc_text">
6333<h5>Syntax:</h5>
6334<p>
6335 These are overloaded intrinsics. You can use <tt>llvm.atomic.load_and</tt>,
6336 <tt>llvm.atomic.load_nand</tt>, <tt>llvm.atomic.load_or</tt>, and
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6337 <tt>llvm.atomic.load_xor</tt> on any integer bit width and for different
6338 address spaces. Not all targets support all bit widths however.</p>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6339<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6340declare i8 @llvm.atomic.load.and.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6341declare i16 @llvm.atomic.load.and.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6342declare i32 @llvm.atomic.load.and.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6343declare 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]6344
6345</pre>
6346
6347<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6348declare i8 @llvm.atomic.load.or.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6349declare i16 @llvm.atomic.load.or.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6350declare i32 @llvm.atomic.load.or.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6351declare 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]6352
6353</pre>
6354
6355<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6356declare i8 @llvm.atomic.load.nand.i8.p0i32( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6357declare i16 @llvm.atomic.load.nand.i16.p0i32( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6358declare i32 @llvm.atomic.load.nand.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6359declare 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]6360
6361</pre>
6362
6363<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6364declare i8 @llvm.atomic.load.xor.i8.p0i32( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6365declare i16 @llvm.atomic.load.xor.i16.p0i32( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6366declare i32 @llvm.atomic.load.xor.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6367declare 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]6368
6369</pre>
6370<h5>Overview:</h5>
6371<p>
6372 These intrinsics bitwise the operation (and, nand, or, xor) <tt>delta</tt> to
6373 the value stored in memory at <tt>ptr</tt>. It yields the original value
6374 at <tt>ptr</tt>.
6375</p>
6376<h5>Arguments:</h5>
6377<p>
6378
6379 These intrinsics take two arguments, the first a pointer to an integer value
6380 and the second an integer value. The result is also an integer value. These
6381 integer types can have any bit width, but they must all have the same bit
6382 width. The targets may only lower integer representations they support.
6383</p>
6384<h5>Semantics:</h5>
6385<p>
6386 These intrinsics does a series of operations atomically. They first load the
6387 value stored at <tt>ptr</tt>. They then do the bitwise operation
6388 <tt>delta</tt>, store the result to <tt>ptr</tt>. They yield the original
6389 value stored at <tt>ptr</tt>.
6390</p>
6391
6392<h5>Examples:</h5>
6393<pre>
6394%ptr = malloc i32
6395 store i32 0x0F0F, %ptr
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6396%result0 = call i32 @llvm.atomic.load.nand.i32.p0i32( i32* %ptr, i32 0xFF )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6397 <i>; yields {i32}:result0 = 0x0F0F</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6398%result1 = call i32 @llvm.atomic.load.and.i32.p0i32( i32* %ptr, i32 0xFF )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6399 <i>; yields {i32}:result1 = 0xFFFFFFF0</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6400%result2 = call i32 @llvm.atomic.load.or.i32.p0i32( i32* %ptr, i32 0F )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6401 <i>; yields {i32}:result2 = 0xF0</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6402%result3 = call i32 @llvm.atomic.load.xor.i32.p0i32( i32* %ptr, i32 0F )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6403 <i>; yields {i32}:result3 = FF</i>
6404%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = F0</i>
6405</pre>
6406</div>
6407
6408
6409<!-- _______________________________________________________________________ -->
6410<div class="doc_subsubsection">
6411 <a name="int_atomic_load_max">'<tt>llvm.atomic.load.max.*</tt>' Intrinsic</a><br>
6412 <a name="int_atomic_load_min">'<tt>llvm.atomic.load.min.*</tt>' Intrinsic</a><br>
6413 <a name="int_atomic_load_umax">'<tt>llvm.atomic.load.umax.*</tt>' Intrinsic</a><br>
6414 <a name="int_atomic_load_umin">'<tt>llvm.atomic.load.umin.*</tt>' Intrinsic</a><br>
6415
6416</div>
6417<div class="doc_text">
6418<h5>Syntax:</h5>
6419<p>
6420 These are overloaded intrinsics. You can use <tt>llvm.atomic.load_max</tt>,
6421 <tt>llvm.atomic.load_min</tt>, <tt>llvm.atomic.load_umax</tt>, and
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6422 <tt>llvm.atomic.load_umin</tt> on any integer bit width and for different
6423 address spaces. Not all targets
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6424 support all bit widths however.</p>
6425<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6426declare i8 @llvm.atomic.load.max.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6427declare i16 @llvm.atomic.load.max.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6428declare i32 @llvm.atomic.load.max.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6429declare 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]6430
6431</pre>
6432
6433<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6434declare i8 @llvm.atomic.load.min.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6435declare i16 @llvm.atomic.load.min.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6436declare i32 @llvm.atomic.load.min.i32..p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6437declare 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]6438
6439</pre>
6440
6441<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6442declare i8 @llvm.atomic.load.umax.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6443declare i16 @llvm.atomic.load.umax.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6444declare i32 @llvm.atomic.load.umax.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6445declare 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]6446
6447</pre>
6448
6449<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6450declare i8 @llvm.atomic.load.umin.i8..p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6451declare i16 @llvm.atomic.load.umin.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6452declare i32 @llvm.atomic.load.umin.i32..p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6453declare 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]6454
6455</pre>
6456<h5>Overview:</h5>
6457<p>
6458 These intrinsics takes the signed or unsigned minimum or maximum of
6459 <tt>delta</tt> and the value stored in memory at <tt>ptr</tt>. It yields the
6460 original value at <tt>ptr</tt>.
6461</p>
6462<h5>Arguments:</h5>
6463<p>
6464
6465 These intrinsics take two arguments, the first a pointer to an integer value
6466 and the second an integer value. The result is also an integer value. These
6467 integer types can have any bit width, but they must all have the same bit
6468 width. The targets may only lower integer representations they support.
6469</p>
6470<h5>Semantics:</h5>
6471<p>
6472 These intrinsics does a series of operations atomically. They first load the
6473 value stored at <tt>ptr</tt>. They then do the signed or unsigned min or max
6474 <tt>delta</tt> and the value, store the result to <tt>ptr</tt>. They yield
6475 the original value stored at <tt>ptr</tt>.
6476</p>
6477
6478<h5>Examples:</h5>
6479<pre>
6480%ptr = malloc i32
6481 store i32 7, %ptr
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6482%result0 = call i32 @llvm.atomic.load.min.i32.p0i32( i32* %ptr, i32 -2 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6483 <i>; yields {i32}:result0 = 7</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6484%result1 = call i32 @llvm.atomic.load.max.i32.p0i32( i32* %ptr, i32 8 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6485 <i>; yields {i32}:result1 = -2</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6486%result2 = call i32 @llvm.atomic.load.umin.i32.p0i32( i32* %ptr, i32 10 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6487 <i>; yields {i32}:result2 = 8</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6488%result3 = call i32 @llvm.atomic.load.umax.i32.p0i32( i32* %ptr, i32 30 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6489 <i>; yields {i32}:result3 = 8</i>
6490%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 30</i>
6491</pre>
6492</div>
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]6493
6494<!-- ======================================================================= -->
6495<div class="doc_subsection">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6496 <a name="int_general">General Intrinsics</a>
6497</div>
6498
6499<div class="doc_text">
6500<p> This class of intrinsics is designed to be generic and has
6501no specific purpose. </p>
6502</div>
6503
6504<!-- _______________________________________________________________________ -->
6505<div class="doc_subsubsection">
6506 <a name="int_var_annotation">'<tt>llvm.var.annotation</tt>' Intrinsic</a>
6507</div>
6508
6509<div class="doc_text">
6510
6511<h5>Syntax:</h5>
6512<pre>
6513 declare void @llvm.var.annotation(i8* &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
6514</pre>
6515
6516<h5>Overview:</h5>
6517
6518<p>
6519The '<tt>llvm.var.annotation</tt>' intrinsic
6520</p>
6521
6522<h5>Arguments:</h5>
6523
6524<p>
6525The first argument is a pointer to a value, the second is a pointer to a
6526global string, the third is a pointer to a global string which is the source
6527file name, and the last argument is the line number.
6528</p>
6529
6530<h5>Semantics:</h5>
6531
6532<p>
Anton Korobeynikove6e764f2008-01-15 22:31:34 +0000[diff] [blame]6533This intrinsic allows annotation of local variables with arbitrary strings.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6534This can be useful for special purpose optimizations that want to look for these
Anton Korobeynikove6e764f2008-01-15 22:31:34 +0000[diff] [blame]6535annotations. These have no other defined use, they are ignored by code
6536generation and optimization.
6537</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6538</div>
6539
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]6540<!-- _______________________________________________________________________ -->
6541<div class="doc_subsubsection">
Tanya Lattnerc9869b12007-09-21 23:57:59 +0000[diff] [blame]6542 <a name="int_annotation">'<tt>llvm.annotation.*</tt>' Intrinsic</a>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]6543</div>
6544
6545<div class="doc_text">
6546
6547<h5>Syntax:</h5>
Tanya Lattnere545be72007-09-21 23:56:27 +0000[diff] [blame]6548<p>This is an overloaded intrinsic. You can use '<tt>llvm.annotation</tt>' on
6549any integer bit width.
6550</p>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]6551<pre>
Tanya Lattner09161fe2007-09-22 00:03:01 +0000[diff] [blame]6552 declare i8 @llvm.annotation.i8(i8 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
6553 declare i16 @llvm.annotation.i16(i16 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
6554 declare i32 @llvm.annotation.i32(i32 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
6555 declare i64 @llvm.annotation.i64(i64 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
6556 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]6557</pre>
6558
6559<h5>Overview:</h5>
Tanya Lattnere545be72007-09-21 23:56:27 +0000[diff] [blame]6560
6561<p>
6562The '<tt>llvm.annotation</tt>' intrinsic.
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]6563</p>
6564
6565<h5>Arguments:</h5>
6566
6567<p>
6568The first argument is an integer value (result of some expression),
6569the second is a pointer to a global string, the third is a pointer to a global
6570string which is the source file name, and the last argument is the line number.
Tanya Lattnere545be72007-09-21 23:56:27 +0000[diff] [blame]6571It returns the value of the first argument.
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]6572</p>
6573
6574<h5>Semantics:</h5>
6575
6576<p>
6577This intrinsic allows annotations to be put on arbitrary expressions
6578with arbitrary strings. This can be useful for special purpose optimizations
6579that want to look for these annotations. These have no other defined use, they
6580are ignored by code generation and optimization.
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]6581</p>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]6582</div>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6583
Anton Korobeynikove6e764f2008-01-15 22:31:34 +0000[diff] [blame]6584<!-- _______________________________________________________________________ -->
6585<div class="doc_subsubsection">
6586 <a name="int_trap">'<tt>llvm.trap</tt>' Intrinsic</a>
6587</div>
6588
6589<div class="doc_text">
6590
6591<h5>Syntax:</h5>
6592<pre>
6593 declare void @llvm.trap()
6594</pre>
6595
6596<h5>Overview:</h5>
6597
6598<p>
6599The '<tt>llvm.trap</tt>' intrinsic
6600</p>
6601
6602<h5>Arguments:</h5>
6603
6604<p>
6605None
6606</p>
6607
6608<h5>Semantics:</h5>
6609
6610<p>
6611This intrinsics is lowered to the target dependent trap instruction. If the
6612target does not have a trap instruction, this intrinsic will be lowered to the
6613call of the abort() function.
6614</p>
6615</div>
6616
Bill Wendlinge4164592008-11-19 05:56:17 +0000[diff] [blame]6617<!-- _______________________________________________________________________ -->
6618<div class="doc_subsubsection">
Misha Brukman5dd7f4d2008-11-22 23:55:29 +0000[diff] [blame]6619 <a name="int_stackprotector">'<tt>llvm.stackprotector</tt>' Intrinsic</a>
Bill Wendlinge4164592008-11-19 05:56:17 +0000[diff] [blame]6620</div>
6621<div class="doc_text">
6622<h5>Syntax:</h5>
6623<pre>
6624declare void @llvm.stackprotector( i8* &lt;guard&gt;, i8** &lt;slot&gt; )
6625
6626</pre>
6627<h5>Overview:</h5>
6628<p>
6629 The <tt>llvm.stackprotector</tt> intrinsic takes the <tt>guard</tt> and stores
6630 it onto the stack at <tt>slot</tt>. The stack slot is adjusted to ensure that
6631 it is placed on the stack before local variables.
6632</p>
6633<h5>Arguments:</h5>
6634<p>
6635 The <tt>llvm.stackprotector</tt> intrinsic requires two pointer arguments. The
6636 first argument is the value loaded from the stack guard
6637 <tt>@__stack_chk_guard</tt>. The second variable is an <tt>alloca</tt> that
6638 has enough space to hold the value of the guard.
6639</p>
6640<h5>Semantics:</h5>
6641<p>
6642 This intrinsic causes the prologue/epilogue inserter to force the position of
6643 the <tt>AllocaInst</tt> stack slot to be before local variables on the
6644 stack. This is to ensure that if a local variable on the stack is overwritten,
6645 it will destroy the value of the guard. When the function exits, the guard on
6646 the stack is checked against the original guard. If they're different, then
6647 the program aborts by calling the <tt>__stack_chk_fail()</tt> function.
6648</p>
6649</div>
6650
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6651<!-- *********************************************************************** -->
6652<hr>
6653<address>
6654 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
Misha Brukman947321d2008-12-11 17:34:48 +0000[diff] [blame]6655 src="http://jigsaw.w3.org/css-validator/images/vcss-blue" alt="Valid CSS"></a>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6656 <a href="http://validator.w3.org/check/referer"><img
Misha Brukman947321d2008-12-11 17:34:48 +0000[diff] [blame]6657 src="http://www.w3.org/Icons/valid-html401-blue" alt="Valid HTML 4.01"></a>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6658
6659 <a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
6660 <a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br>
6661 Last modified: $Date$
6662</address>
Chris Lattner08497ce2008-01-04 04:33:49 +0000[diff] [blame]6663
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6664</body>
6665</html>