Gitiles
Code Review Sign In
gerrit-public.fairphone.software / fp2-dev / platform / external / llvm / 9901e73caec653599f27d04075961afe4a7c29a1 / . / docs / LangRef.html
blob: 48b619a74ba8d7f044daa05fec53ecbc9cbd9c4c [file] [log] [blame]
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN"
2 "http://www.w3.org/TR/html4/strict.dtd">
3<html>
4<head>
5 <title>LLVM Assembly Language Reference Manual</title>
6 <meta http-equiv="Content-Type" content="text/html; charset=utf-8">
7 <meta name="author" content="Chris Lattner">
8 <meta name="description"
9 content="LLVM Assembly Language Reference Manual.">
10 <link rel="stylesheet" href="llvm.css" type="text/css">
11</head>
12
13<body>
14
15<div class="doc_title"> LLVM Language Reference Manual </div>
16<ol>
17 <li><a href="#abstract">Abstract</a></li>
18 <li><a href="#introduction">Introduction</a></li>
19 <li><a href="#identifiers">Identifiers</a></li>
20 <li><a href="#highlevel">High Level Structure</a>
21 <ol>
22 <li><a href="#modulestructure">Module Structure</a></li>
23 <li><a href="#linkage">Linkage Types</a></li>
24 <li><a href="#callingconv">Calling Conventions</a></li>
25 <li><a href="#globalvars">Global Variables</a></li>
26 <li><a href="#functionstructure">Functions</a></li>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]27 <li><a href="#aliasstructure">Aliases</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]28 <li><a href="#paramattrs">Parameter Attributes</a></li>
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]29 <li><a href="#fnattrs">Function Attributes</a></li>
Gordon Henriksen13fe5e32007-12-10 03:18:06 +0000[diff] [blame]30 <li><a href="#gc">Garbage Collector Names</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]31 <li><a href="#moduleasm">Module-Level Inline Assembly</a></li>
32 <li><a href="#datalayout">Data Layout</a></li>
33 </ol>
34 </li>
35 <li><a href="#typesystem">Type System</a>
36 <ol>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]37 <li><a href="#t_classifications">Type Classifications</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]38 <li><a href="#t_primitive">Primitive Types</a>
39 <ol>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]40 <li><a href="#t_floating">Floating Point Types</a></li>
41 <li><a href="#t_void">Void Type</a></li>
42 <li><a href="#t_label">Label Type</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]43 </ol>
44 </li>
45 <li><a href="#t_derived">Derived Types</a>
46 <ol>
Chris Lattner251ab812007-12-18 06:18:21 +0000[diff] [blame]47 <li><a href="#t_integer">Integer Type</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]48 <li><a href="#t_array">Array Type</a></li>
49 <li><a href="#t_function">Function Type</a></li>
50 <li><a href="#t_pointer">Pointer Type</a></li>
51 <li><a href="#t_struct">Structure Type</a></li>
52 <li><a href="#t_pstruct">Packed Structure Type</a></li>
53 <li><a href="#t_vector">Vector Type</a></li>
54 <li><a href="#t_opaque">Opaque Type</a></li>
55 </ol>
56 </li>
57 </ol>
58 </li>
59 <li><a href="#constants">Constants</a>
60 <ol>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]61 <li><a href="#simpleconstants">Simple Constants</a></li>
62 <li><a href="#aggregateconstants">Aggregate Constants</a></li>
63 <li><a href="#globalconstants">Global Variable and Function Addresses</a></li>
64 <li><a href="#undefvalues">Undefined Values</a></li>
65 <li><a href="#constantexprs">Constant Expressions</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]66 </ol>
67 </li>
68 <li><a href="#othervalues">Other Values</a>
69 <ol>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]70 <li><a href="#inlineasm">Inline Assembler Expressions</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]71 </ol>
72 </li>
73 <li><a href="#instref">Instruction Reference</a>
74 <ol>
75 <li><a href="#terminators">Terminator Instructions</a>
76 <ol>
77 <li><a href="#i_ret">'<tt>ret</tt>' Instruction</a></li>
78 <li><a href="#i_br">'<tt>br</tt>' Instruction</a></li>
79 <li><a href="#i_switch">'<tt>switch</tt>' Instruction</a></li>
80 <li><a href="#i_invoke">'<tt>invoke</tt>' Instruction</a></li>
81 <li><a href="#i_unwind">'<tt>unwind</tt>' Instruction</a></li>
82 <li><a href="#i_unreachable">'<tt>unreachable</tt>' Instruction</a></li>
83 </ol>
84 </li>
85 <li><a href="#binaryops">Binary Operations</a>
86 <ol>
87 <li><a href="#i_add">'<tt>add</tt>' Instruction</a></li>
88 <li><a href="#i_sub">'<tt>sub</tt>' Instruction</a></li>
89 <li><a href="#i_mul">'<tt>mul</tt>' Instruction</a></li>
90 <li><a href="#i_udiv">'<tt>udiv</tt>' Instruction</a></li>
91 <li><a href="#i_sdiv">'<tt>sdiv</tt>' Instruction</a></li>
92 <li><a href="#i_fdiv">'<tt>fdiv</tt>' Instruction</a></li>
93 <li><a href="#i_urem">'<tt>urem</tt>' Instruction</a></li>
94 <li><a href="#i_srem">'<tt>srem</tt>' Instruction</a></li>
95 <li><a href="#i_frem">'<tt>frem</tt>' Instruction</a></li>
96 </ol>
97 </li>
98 <li><a href="#bitwiseops">Bitwise Binary Operations</a>
99 <ol>
100 <li><a href="#i_shl">'<tt>shl</tt>' Instruction</a></li>
101 <li><a href="#i_lshr">'<tt>lshr</tt>' Instruction</a></li>
102 <li><a href="#i_ashr">'<tt>ashr</tt>' Instruction</a></li>
103 <li><a href="#i_and">'<tt>and</tt>' Instruction</a></li>
104 <li><a href="#i_or">'<tt>or</tt>' Instruction</a></li>
105 <li><a href="#i_xor">'<tt>xor</tt>' Instruction</a></li>
106 </ol>
107 </li>
108 <li><a href="#vectorops">Vector Operations</a>
109 <ol>
110 <li><a href="#i_extractelement">'<tt>extractelement</tt>' Instruction</a></li>
111 <li><a href="#i_insertelement">'<tt>insertelement</tt>' Instruction</a></li>
112 <li><a href="#i_shufflevector">'<tt>shufflevector</tt>' Instruction</a></li>
113 </ol>
114 </li>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]115 <li><a href="#aggregateops">Aggregate Operations</a>
116 <ol>
117 <li><a href="#i_extractvalue">'<tt>extractvalue</tt>' Instruction</a></li>
118 <li><a href="#i_insertvalue">'<tt>insertvalue</tt>' Instruction</a></li>
119 </ol>
120 </li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]121 <li><a href="#memoryops">Memory Access and Addressing Operations</a>
122 <ol>
123 <li><a href="#i_malloc">'<tt>malloc</tt>' Instruction</a></li>
124 <li><a href="#i_free">'<tt>free</tt>' Instruction</a></li>
125 <li><a href="#i_alloca">'<tt>alloca</tt>' Instruction</a></li>
126 <li><a href="#i_load">'<tt>load</tt>' Instruction</a></li>
127 <li><a href="#i_store">'<tt>store</tt>' Instruction</a></li>
128 <li><a href="#i_getelementptr">'<tt>getelementptr</tt>' Instruction</a></li>
129 </ol>
130 </li>
131 <li><a href="#convertops">Conversion Operations</a>
132 <ol>
133 <li><a href="#i_trunc">'<tt>trunc .. to</tt>' Instruction</a></li>
134 <li><a href="#i_zext">'<tt>zext .. to</tt>' Instruction</a></li>
135 <li><a href="#i_sext">'<tt>sext .. to</tt>' Instruction</a></li>
136 <li><a href="#i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a></li>
137 <li><a href="#i_fpext">'<tt>fpext .. to</tt>' Instruction</a></li>
138 <li><a href="#i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a></li>
139 <li><a href="#i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a></li>
140 <li><a href="#i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a></li>
141 <li><a href="#i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a></li>
142 <li><a href="#i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a></li>
143 <li><a href="#i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a></li>
144 <li><a href="#i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a></li>
145 </ol>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]146 </li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]147 <li><a href="#otherops">Other Operations</a>
148 <ol>
149 <li><a href="#i_icmp">'<tt>icmp</tt>' Instruction</a></li>
150 <li><a href="#i_fcmp">'<tt>fcmp</tt>' Instruction</a></li>
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]151 <li><a href="#i_vicmp">'<tt>vicmp</tt>' Instruction</a></li>
152 <li><a href="#i_vfcmp">'<tt>vfcmp</tt>' Instruction</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]153 <li><a href="#i_phi">'<tt>phi</tt>' Instruction</a></li>
154 <li><a href="#i_select">'<tt>select</tt>' Instruction</a></li>
155 <li><a href="#i_call">'<tt>call</tt>' Instruction</a></li>
156 <li><a href="#i_va_arg">'<tt>va_arg</tt>' Instruction</a></li>
157 </ol>
158 </li>
159 </ol>
160 </li>
161 <li><a href="#intrinsics">Intrinsic Functions</a>
162 <ol>
163 <li><a href="#int_varargs">Variable Argument Handling Intrinsics</a>
164 <ol>
165 <li><a href="#int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a></li>
166 <li><a href="#int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a></li>
167 <li><a href="#int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a></li>
168 </ol>
169 </li>
170 <li><a href="#int_gc">Accurate Garbage Collection Intrinsics</a>
171 <ol>
172 <li><a href="#int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a></li>
173 <li><a href="#int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a></li>
174 <li><a href="#int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a></li>
175 </ol>
176 </li>
177 <li><a href="#int_codegen">Code Generator Intrinsics</a>
178 <ol>
179 <li><a href="#int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a></li>
180 <li><a href="#int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a></li>
181 <li><a href="#int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a></li>
182 <li><a href="#int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a></li>
183 <li><a href="#int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a></li>
184 <li><a href="#int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a></li>
185 <li><a href="#int_readcyclecounter"><tt>llvm.readcyclecounter</tt>' Intrinsic</a></li>
186 </ol>
187 </li>
188 <li><a href="#int_libc">Standard C Library Intrinsics</a>
189 <ol>
190 <li><a href="#int_memcpy">'<tt>llvm.memcpy.*</tt>' Intrinsic</a></li>
191 <li><a href="#int_memmove">'<tt>llvm.memmove.*</tt>' Intrinsic</a></li>
192 <li><a href="#int_memset">'<tt>llvm.memset.*</tt>' Intrinsic</a></li>
193 <li><a href="#int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a></li>
194 <li><a href="#int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a></li>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]195 <li><a href="#int_sin">'<tt>llvm.sin.*</tt>' Intrinsic</a></li>
196 <li><a href="#int_cos">'<tt>llvm.cos.*</tt>' Intrinsic</a></li>
197 <li><a href="#int_pow">'<tt>llvm.pow.*</tt>' Intrinsic</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]198 </ol>
199 </li>
200 <li><a href="#int_manip">Bit Manipulation Intrinsics</a>
201 <ol>
202 <li><a href="#int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a></li>
203 <li><a href="#int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic </a></li>
204 <li><a href="#int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic </a></li>
205 <li><a href="#int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic </a></li>
206 <li><a href="#int_part_select">'<tt>llvm.part.select.*</tt>' Intrinsic </a></li>
207 <li><a href="#int_part_set">'<tt>llvm.part.set.*</tt>' Intrinsic </a></li>
208 </ol>
209 </li>
210 <li><a href="#int_debugger">Debugger intrinsics</a></li>
211 <li><a href="#int_eh">Exception Handling intrinsics</a></li>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]212 <li><a href="#int_trampoline">Trampoline Intrinsic</a>
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]213 <ol>
214 <li><a href="#int_it">'<tt>llvm.init.trampoline</tt>' Intrinsic</a></li>
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]215 </ol>
216 </li>
Bill Wendling9127adb2008-11-18 22:10:53 +0000[diff] [blame]217 <li><a href="#int_atomics">Atomic intrinsics</a>
218 <ol>
219 <li><a href="#int_memory_barrier"><tt>llvm.memory_barrier</tt></a></li>
220 <li><a href="#int_atomic_cmp_swap"><tt>llvm.atomic.cmp.swap</tt></a></li>
221 <li><a href="#int_atomic_swap"><tt>llvm.atomic.swap</tt></a></li>
222 <li><a href="#int_atomic_load_add"><tt>llvm.atomic.load.add</tt></a></li>
223 <li><a href="#int_atomic_load_sub"><tt>llvm.atomic.load.sub</tt></a></li>
224 <li><a href="#int_atomic_load_and"><tt>llvm.atomic.load.and</tt></a></li>
225 <li><a href="#int_atomic_load_nand"><tt>llvm.atomic.load.nand</tt></a></li>
226 <li><a href="#int_atomic_load_or"><tt>llvm.atomic.load.or</tt></a></li>
227 <li><a href="#int_atomic_load_xor"><tt>llvm.atomic.load.xor</tt></a></li>
228 <li><a href="#int_atomic_load_max"><tt>llvm.atomic.load.max</tt></a></li>
229 <li><a href="#int_atomic_load_min"><tt>llvm.atomic.load.min</tt></a></li>
230 <li><a href="#int_atomic_load_umax"><tt>llvm.atomic.load.umax</tt></a></li>
231 <li><a href="#int_atomic_load_umin"><tt>llvm.atomic.load.umin</tt></a></li>
232 </ol>
233 </li>
Reid Spencerb043f672007-07-20 19:59:11 +0000[diff] [blame]234 <li><a href="#int_general">General intrinsics</a>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]235 <ol>
Reid Spencerb043f672007-07-20 19:59:11 +0000[diff] [blame]236 <li><a href="#int_var_annotation">
Bill Wendlinge4164592008-11-19 05:56:17 +0000[diff] [blame]237 '<tt>llvm.var.annotation</tt>' Intrinsic</a></li>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]238 <li><a href="#int_annotation">
Bill Wendlinge4164592008-11-19 05:56:17 +0000[diff] [blame]239 '<tt>llvm.annotation.*</tt>' Intrinsic</a></li>
Anton Korobeynikove6e764f2008-01-15 22:31:34 +0000[diff] [blame]240 <li><a href="#int_trap">
Bill Wendlinge4164592008-11-19 05:56:17 +0000[diff] [blame]241 '<tt>llvm.trap</tt>' Intrinsic</a></li>
242 <li><a href="#int_stackprotector">
243 '<tt>llvm.stackprotector</tt>' Intrinsic</a></li>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]244 </ol>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]245 </li>
246 </ol>
247 </li>
248</ol>
249
250<div class="doc_author">
251 <p>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a>
252 and <a href="mailto:vadve@cs.uiuc.edu">Vikram Adve</a></p>
253</div>
254
255<!-- *********************************************************************** -->
256<div class="doc_section"> <a name="abstract">Abstract </a></div>
257<!-- *********************************************************************** -->
258
259<div class="doc_text">
260<p>This document is a reference manual for the LLVM assembly language.
Bill Wendlinge7846a52008-08-05 22:29:16 +0000[diff] [blame]261LLVM is a Static Single Assignment (SSA) based representation that provides
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]262type safety, low-level operations, flexibility, and the capability of
263representing 'all' high-level languages cleanly. It is the common code
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]264representation used throughout all phases of the LLVM compilation
265strategy.</p>
266</div>
267
268<!-- *********************************************************************** -->
269<div class="doc_section"> <a name="introduction">Introduction</a> </div>
270<!-- *********************************************************************** -->
271
272<div class="doc_text">
273
274<p>The LLVM code representation is designed to be used in three
275different forms: as an in-memory compiler IR, as an on-disk bitcode
276representation (suitable for fast loading by a Just-In-Time compiler),
277and as a human readable assembly language representation. This allows
278LLVM to provide a powerful intermediate representation for efficient
279compiler transformations and analysis, while providing a natural means
280to debug and visualize the transformations. The three different forms
281of LLVM are all equivalent. This document describes the human readable
282representation and notation.</p>
283
284<p>The LLVM representation aims to be light-weight and low-level
285while being expressive, typed, and extensible at the same time. It
286aims to be a "universal IR" of sorts, by being at a low enough level
287that high-level ideas may be cleanly mapped to it (similar to how
288microprocessors are "universal IR's", allowing many source languages to
289be mapped to them). By providing type information, LLVM can be used as
290the target of optimizations: for example, through pointer analysis, it
291can be proven that a C automatic variable is never accessed outside of
292the current function... allowing it to be promoted to a simple SSA
293value instead of a memory location.</p>
294
295</div>
296
297<!-- _______________________________________________________________________ -->
298<div class="doc_subsubsection"> <a name="wellformed">Well-Formedness</a> </div>
299
300<div class="doc_text">
301
302<p>It is important to note that this document describes 'well formed'
303LLVM assembly language. There is a difference between what the parser
304accepts and what is considered 'well formed'. For example, the
305following instruction is syntactically okay, but not well formed:</p>
306
307<div class="doc_code">
308<pre>
309%x = <a href="#i_add">add</a> i32 1, %x
310</pre>
311</div>
312
313<p>...because the definition of <tt>%x</tt> does not dominate all of
314its uses. The LLVM infrastructure provides a verification pass that may
315be used to verify that an LLVM module is well formed. This pass is
316automatically run by the parser after parsing input assembly and by
317the optimizer before it outputs bitcode. The violations pointed out
318by the verifier pass indicate bugs in transformation passes or input to
319the parser.</p>
320</div>
321
Chris Lattnera83fdc02007-10-03 17:34:29 +0000[diff] [blame]322<!-- Describe the typesetting conventions here. -->
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]323
324<!-- *********************************************************************** -->
325<div class="doc_section"> <a name="identifiers">Identifiers</a> </div>
326<!-- *********************************************************************** -->
327
328<div class="doc_text">
329
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]330 <p>LLVM identifiers come in two basic types: global and local. Global
331 identifiers (functions, global variables) begin with the @ character. Local
332 identifiers (register names, types) begin with the % character. Additionally,
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]333 there are three different formats for identifiers, for different purposes:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]334
335<ol>
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]336 <li>Named values are represented as a string of characters with their prefix.
337 For example, %foo, @DivisionByZero, %a.really.long.identifier. The actual
338 regular expression used is '<tt>[%@][a-zA-Z$._][a-zA-Z$._0-9]*</tt>'.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]339 Identifiers which require other characters in their names can be surrounded
Daniel Dunbar73f9a772008-10-14 23:51:43 +0000[diff] [blame]340 with quotes. Special characters may be escaped using "\xx" where xx is the
341 ASCII code for the character in hexadecimal. In this way, any character can
342 be used in a name value, even quotes themselves.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]343
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]344 <li>Unnamed values are represented as an unsigned numeric value with their
345 prefix. For example, %12, @2, %44.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]346
347 <li>Constants, which are described in a <a href="#constants">section about
348 constants</a>, below.</li>
349</ol>
350
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]351<p>LLVM requires that values start with a prefix for two reasons: Compilers
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]352don't need to worry about name clashes with reserved words, and the set of
353reserved words may be expanded in the future without penalty. Additionally,
354unnamed identifiers allow a compiler to quickly come up with a temporary
355variable without having to avoid symbol table conflicts.</p>
356
357<p>Reserved words in LLVM are very similar to reserved words in other
358languages. There are keywords for different opcodes
359('<tt><a href="#i_add">add</a></tt>',
360 '<tt><a href="#i_bitcast">bitcast</a></tt>',
361 '<tt><a href="#i_ret">ret</a></tt>', etc...), for primitive type names ('<tt><a
362href="#t_void">void</a></tt>', '<tt><a href="#t_primitive">i32</a></tt>', etc...),
363and others. These reserved words cannot conflict with variable names, because
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]364none of them start with a prefix character ('%' or '@').</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]365
366<p>Here is an example of LLVM code to multiply the integer variable
367'<tt>%X</tt>' by 8:</p>
368
369<p>The easy way:</p>
370
371<div class="doc_code">
372<pre>
373%result = <a href="#i_mul">mul</a> i32 %X, 8
374</pre>
375</div>
376
377<p>After strength reduction:</p>
378
379<div class="doc_code">
380<pre>
381%result = <a href="#i_shl">shl</a> i32 %X, i8 3
382</pre>
383</div>
384
385<p>And the hard way:</p>
386
387<div class="doc_code">
388<pre>
389<a href="#i_add">add</a> i32 %X, %X <i>; yields {i32}:%0</i>
390<a href="#i_add">add</a> i32 %0, %0 <i>; yields {i32}:%1</i>
391%result = <a href="#i_add">add</a> i32 %1, %1
392</pre>
393</div>
394
395<p>This last way of multiplying <tt>%X</tt> by 8 illustrates several
396important lexical features of LLVM:</p>
397
398<ol>
399
400 <li>Comments are delimited with a '<tt>;</tt>' and go until the end of
401 line.</li>
402
403 <li>Unnamed temporaries are created when the result of a computation is not
404 assigned to a named value.</li>
405
406 <li>Unnamed temporaries are numbered sequentially</li>
407
408</ol>
409
410<p>...and it also shows a convention that we follow in this document. When
411demonstrating instructions, we will follow an instruction with a comment that
412defines the type and name of value produced. Comments are shown in italic
413text.</p>
414
415</div>
416
417<!-- *********************************************************************** -->
418<div class="doc_section"> <a name="highlevel">High Level Structure</a> </div>
419<!-- *********************************************************************** -->
420
421<!-- ======================================================================= -->
422<div class="doc_subsection"> <a name="modulestructure">Module Structure</a>
423</div>
424
425<div class="doc_text">
426
427<p>LLVM programs are composed of "Module"s, each of which is a
428translation unit of the input programs. Each module consists of
429functions, global variables, and symbol table entries. Modules may be
430combined together with the LLVM linker, which merges function (and
431global variable) definitions, resolves forward declarations, and merges
432symbol table entries. Here is an example of the "hello world" module:</p>
433
434<div class="doc_code">
435<pre><i>; Declare the string constant as a global constant...</i>
436<a href="#identifiers">@.LC0</a> = <a href="#linkage_internal">internal</a> <a
437 href="#globalvars">constant</a> <a href="#t_array">[13 x i8]</a> c"hello world\0A\00" <i>; [13 x i8]*</i>
438
439<i>; External declaration of the puts function</i>
440<a href="#functionstructure">declare</a> i32 @puts(i8 *) <i>; i32(i8 *)* </i>
441
442<i>; Definition of main function</i>
443define i32 @main() { <i>; i32()* </i>
444 <i>; Convert [13x i8 ]* to i8 *...</i>
445 %cast210 = <a
446 href="#i_getelementptr">getelementptr</a> [13 x i8 ]* @.LC0, i64 0, i64 0 <i>; i8 *</i>
447
448 <i>; Call puts function to write out the string to stdout...</i>
449 <a
450 href="#i_call">call</a> i32 @puts(i8 * %cast210) <i>; i32</i>
451 <a
452 href="#i_ret">ret</a> i32 0<br>}<br>
453</pre>
454</div>
455
456<p>This example is made up of a <a href="#globalvars">global variable</a>
457named "<tt>.LC0</tt>", an external declaration of the "<tt>puts</tt>"
458function, and a <a href="#functionstructure">function definition</a>
459for "<tt>main</tt>".</p>
460
461<p>In general, a module is made up of a list of global values,
462where both functions and global variables are global values. Global values are
463represented by a pointer to a memory location (in this case, a pointer to an
464array of char, and a pointer to a function), and have one of the following <a
465href="#linkage">linkage types</a>.</p>
466
467</div>
468
469<!-- ======================================================================= -->
470<div class="doc_subsection">
471 <a name="linkage">Linkage Types</a>
472</div>
473
474<div class="doc_text">
475
476<p>
477All Global Variables and Functions have one of the following types of linkage:
478</p>
479
480<dl>
481
Dale Johannesen96e7e092008-05-23 23:13:41 +0000[diff] [blame]482 <dt><tt><b><a name="linkage_internal">internal</a></b></tt>: </dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]483
484 <dd>Global values with internal linkage are only directly accessible by
485 objects in the current module. In particular, linking code into a module with
486 an internal global value may cause the internal to be renamed as necessary to
487 avoid collisions. Because the symbol is internal to the module, all
488 references can be updated. This corresponds to the notion of the
489 '<tt>static</tt>' keyword in C.
490 </dd>
491
492 <dt><tt><b><a name="linkage_linkonce">linkonce</a></b></tt>: </dt>
493
494 <dd>Globals with "<tt>linkonce</tt>" linkage are merged with other globals of
495 the same name when linkage occurs. This is typically used to implement
496 inline functions, templates, or other code which must be generated in each
497 translation unit that uses it. Unreferenced <tt>linkonce</tt> globals are
498 allowed to be discarded.
499 </dd>
500
Dale Johannesen96e7e092008-05-23 23:13:41 +0000[diff] [blame]501 <dt><tt><b><a name="linkage_common">common</a></b></tt>: </dt>
502
503 <dd>"<tt>common</tt>" linkage is exactly the same as <tt>linkonce</tt>
504 linkage, except that unreferenced <tt>common</tt> globals may not be
505 discarded. This is used for globals that may be emitted in multiple
506 translation units, but that are not guaranteed to be emitted into every
507 translation unit that uses them. One example of this is tentative
508 definitions in C, such as "<tt>int X;</tt>" at global scope.
509 </dd>
510
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]511 <dt><tt><b><a name="linkage_weak">weak</a></b></tt>: </dt>
512
Dale Johannesen96e7e092008-05-23 23:13:41 +0000[diff] [blame]513 <dd>"<tt>weak</tt>" linkage is the same as <tt>common</tt> linkage, except
514 that some targets may choose to emit different assembly sequences for them
515 for target-dependent reasons. This is used for globals that are declared
516 "weak" in C source code.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]517 </dd>
518
519 <dt><tt><b><a name="linkage_appending">appending</a></b></tt>: </dt>
520
521 <dd>"<tt>appending</tt>" linkage may only be applied to global variables of
522 pointer to array type. When two global variables with appending linkage are
523 linked together, the two global arrays are appended together. This is the
524 LLVM, typesafe, equivalent of having the system linker append together
525 "sections" with identical names when .o files are linked.
526 </dd>
527
528 <dt><tt><b><a name="linkage_externweak">extern_weak</a></b></tt>: </dt>
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]529 <dd>The semantics of this linkage follow the ELF object file model: the
530 symbol is weak until linked, if not linked, the symbol becomes null instead
531 of being an undefined reference.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]532 </dd>
533
534 <dt><tt><b><a name="linkage_external">externally visible</a></b></tt>:</dt>
535
536 <dd>If none of the above identifiers are used, the global is externally
537 visible, meaning that it participates in linkage and can be used to resolve
538 external symbol references.
539 </dd>
540</dl>
541
542 <p>
543 The next two types of linkage are targeted for Microsoft Windows platform
544 only. They are designed to support importing (exporting) symbols from (to)
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]545 DLLs (Dynamic Link Libraries).
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]546 </p>
547
548 <dl>
549 <dt><tt><b><a name="linkage_dllimport">dllimport</a></b></tt>: </dt>
550
551 <dd>"<tt>dllimport</tt>" linkage causes the compiler to reference a function
552 or variable via a global pointer to a pointer that is set up by the DLL
553 exporting the symbol. On Microsoft Windows targets, the pointer name is
554 formed by combining <code>_imp__</code> and the function or variable name.
555 </dd>
556
557 <dt><tt><b><a name="linkage_dllexport">dllexport</a></b></tt>: </dt>
558
559 <dd>"<tt>dllexport</tt>" linkage causes the compiler to provide a global
560 pointer to a pointer in a DLL, so that it can be referenced with the
561 <tt>dllimport</tt> attribute. On Microsoft Windows targets, the pointer
562 name is formed by combining <code>_imp__</code> and the function or variable
563 name.
564 </dd>
565
566</dl>
567
Dan Gohman4dfac702008-11-24 17:18:39 +0000[diff] [blame]568<p>For example, since the "<tt>.LC0</tt>"
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]569variable is defined to be internal, if another module defined a "<tt>.LC0</tt>"
570variable and was linked with this one, one of the two would be renamed,
571preventing a collision. Since "<tt>main</tt>" and "<tt>puts</tt>" are
572external (i.e., lacking any linkage declarations), they are accessible
573outside of the current module.</p>
574<p>It is illegal for a function <i>declaration</i>
575to have any linkage type other than "externally visible", <tt>dllimport</tt>,
576or <tt>extern_weak</tt>.</p>
577<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]578linkages.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]579</div>
580
581<!-- ======================================================================= -->
582<div class="doc_subsection">
583 <a name="callingconv">Calling Conventions</a>
584</div>
585
586<div class="doc_text">
587
588<p>LLVM <a href="#functionstructure">functions</a>, <a href="#i_call">calls</a>
589and <a href="#i_invoke">invokes</a> can all have an optional calling convention
590specified for the call. The calling convention of any pair of dynamic
591caller/callee must match, or the behavior of the program is undefined. The
592following calling conventions are supported by LLVM, and more may be added in
593the future:</p>
594
595<dl>
596 <dt><b>"<tt>ccc</tt>" - The C calling convention</b>:</dt>
597
598 <dd>This calling convention (the default if no other calling convention is
599 specified) matches the target C calling conventions. This calling convention
600 supports varargs function calls and tolerates some mismatch in the declared
601 prototype and implemented declaration of the function (as does normal C).
602 </dd>
603
604 <dt><b>"<tt>fastcc</tt>" - The fast calling convention</b>:</dt>
605
606 <dd>This calling convention attempts to make calls as fast as possible
607 (e.g. by passing things in registers). This calling convention allows the
608 target to use whatever tricks it wants to produce fast code for the target,
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]609 without having to conform to an externally specified ABI (Application Binary
610 Interface). Implementations of this convention should allow arbitrary
Arnold Schwaighofer07444922008-05-14 09:17:12 +0000[diff] [blame]611 <a href="CodeGenerator.html#tailcallopt">tail call optimization</a> to be
612 supported. This calling convention does not support varargs and requires the
613 prototype of all callees to exactly match the prototype of the function
614 definition.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]615 </dd>
616
617 <dt><b>"<tt>coldcc</tt>" - The cold calling convention</b>:</dt>
618
619 <dd>This calling convention attempts to make code in the caller as efficient
620 as possible under the assumption that the call is not commonly executed. As
621 such, these calls often preserve all registers so that the call does not break
622 any live ranges in the caller side. This calling convention does not support
623 varargs and requires the prototype of all callees to exactly match the
624 prototype of the function definition.
625 </dd>
626
627 <dt><b>"<tt>cc &lt;<em>n</em>&gt;</tt>" - Numbered convention</b>:</dt>
628
629 <dd>Any calling convention may be specified by number, allowing
630 target-specific calling conventions to be used. Target specific calling
631 conventions start at 64.
632 </dd>
633</dl>
634
635<p>More calling conventions can be added/defined on an as-needed basis, to
636support pascal conventions or any other well-known target-independent
637convention.</p>
638
639</div>
640
641<!-- ======================================================================= -->
642<div class="doc_subsection">
643 <a name="visibility">Visibility Styles</a>
644</div>
645
646<div class="doc_text">
647
648<p>
649All Global Variables and Functions have one of the following visibility styles:
650</p>
651
652<dl>
653 <dt><b>"<tt>default</tt>" - Default style</b>:</dt>
654
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]655 <dd>On targets that use the ELF object file format, default visibility means
656 that the declaration is visible to other
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]657 modules and, in shared libraries, means that the declared entity may be
658 overridden. On Darwin, default visibility means that the declaration is
659 visible to other modules. Default visibility corresponds to "external
660 linkage" in the language.
661 </dd>
662
663 <dt><b>"<tt>hidden</tt>" - Hidden style</b>:</dt>
664
665 <dd>Two declarations of an object with hidden visibility refer to the same
666 object if they are in the same shared object. Usually, hidden visibility
667 indicates that the symbol will not be placed into the dynamic symbol table,
668 so no other module (executable or shared library) can reference it
669 directly.
670 </dd>
671
672 <dt><b>"<tt>protected</tt>" - Protected style</b>:</dt>
673
674 <dd>On ELF, protected visibility indicates that the symbol will be placed in
675 the dynamic symbol table, but that references within the defining module will
676 bind to the local symbol. That is, the symbol cannot be overridden by another
677 module.
678 </dd>
679</dl>
680
681</div>
682
683<!-- ======================================================================= -->
684<div class="doc_subsection">
685 <a name="globalvars">Global Variables</a>
686</div>
687
688<div class="doc_text">
689
690<p>Global variables define regions of memory allocated at compilation time
691instead of run-time. Global variables may optionally be initialized, may have
692an explicit section to be placed in, and may have an optional explicit alignment
693specified. A variable may be defined as "thread_local", which means that it
694will not be shared by threads (each thread will have a separated copy of the
695variable). A variable may be defined as a global "constant," which indicates
696that the contents of the variable will <b>never</b> be modified (enabling better
697optimization, allowing the global data to be placed in the read-only section of
698an executable, etc). Note that variables that need runtime initialization
699cannot be marked "constant" as there is a store to the variable.</p>
700
701<p>
702LLVM explicitly allows <em>declarations</em> of global variables to be marked
703constant, even if the final definition of the global is not. This capability
704can be used to enable slightly better optimization of the program, but requires
705the language definition to guarantee that optimizations based on the
706'constantness' are valid for the translation units that do not include the
707definition.
708</p>
709
710<p>As SSA values, global variables define pointer values that are in
711scope (i.e. they dominate) all basic blocks in the program. Global
712variables always define a pointer to their "content" type because they
713describe a region of memory, and all memory objects in LLVM are
714accessed through pointers.</p>
715
Christopher Lambdd0049d2007-12-11 09:31:00 +0000[diff] [blame]716<p>A global variable may be declared to reside in a target-specifc numbered
717address space. For targets that support them, address spaces may affect how
718optimizations are performed and/or what target instructions are used to access
Christopher Lamb20a39e92007-12-12 08:44:39 +0000[diff] [blame]719the variable. The default address space is zero. The address space qualifier
720must precede any other attributes.</p>
Christopher Lambdd0049d2007-12-11 09:31:00 +0000[diff] [blame]721
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]722<p>LLVM allows an explicit section to be specified for globals. If the target
723supports it, it will emit globals to the section specified.</p>
724
725<p>An explicit alignment may be specified for a global. If not present, or if
726the alignment is set to zero, the alignment of the global is set by the target
727to whatever it feels convenient. If an explicit alignment is specified, the
728global is forced to have at least that much alignment. All alignments must be
729a power of 2.</p>
730
Christopher Lambdd0049d2007-12-11 09:31:00 +0000[diff] [blame]731<p>For example, the following defines a global in a numbered address space with
732an initializer, section, and alignment:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]733
734<div class="doc_code">
735<pre>
Christopher Lambdd0049d2007-12-11 09:31:00 +0000[diff] [blame]736@G = constant float 1.0 addrspace(5), section "foo", align 4
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]737</pre>
738</div>
739
740</div>
741
742
743<!-- ======================================================================= -->
744<div class="doc_subsection">
745 <a name="functionstructure">Functions</a>
746</div>
747
748<div class="doc_text">
749
750<p>LLVM function definitions consist of the "<tt>define</tt>" keyord,
751an optional <a href="#linkage">linkage type</a>, an optional
752<a href="#visibility">visibility style</a>, an optional
753<a href="#callingconv">calling convention</a>, a return type, an optional
754<a href="#paramattrs">parameter attribute</a> for the return type, a function
755name, a (possibly empty) argument list (each with optional
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]756<a href="#paramattrs">parameter attributes</a>), optional
757<a href="#fnattrs">function attributes</a>, an optional section,
758an optional alignment, an optional <a href="#gc">garbage collector name</a>,
Chris Lattneree202542008-10-04 18:10:21 +0000[diff] [blame]759an opening curly brace, a list of basic blocks, and a closing curly brace.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]760
761LLVM function declarations consist of the "<tt>declare</tt>" keyword, an
762optional <a href="#linkage">linkage type</a>, an optional
763<a href="#visibility">visibility style</a>, an optional
764<a href="#callingconv">calling convention</a>, a return type, an optional
765<a href="#paramattrs">parameter attribute</a> for the return type, a function
Gordon Henriksen13fe5e32007-12-10 03:18:06 +0000[diff] [blame]766name, a possibly empty list of arguments, an optional alignment, and an optional
Gordon Henriksend606f5b2007-12-10 03:30:21 +0000[diff] [blame]767<a href="#gc">garbage collector name</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]768
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]769<p>A function definition contains a list of basic blocks, forming the CFG
770(Control Flow Graph) for
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]771the function. Each basic block may optionally start with a label (giving the
772basic block a symbol table entry), contains a list of instructions, and ends
773with a <a href="#terminators">terminator</a> instruction (such as a branch or
774function return).</p>
775
776<p>The first basic block in a function is special in two ways: it is immediately
777executed on entrance to the function, and it is not allowed to have predecessor
778basic blocks (i.e. there can not be any branches to the entry block of a
779function). Because the block can have no predecessors, it also cannot have any
780<a href="#i_phi">PHI nodes</a>.</p>
781
782<p>LLVM allows an explicit section to be specified for functions. If the target
783supports it, it will emit functions to the section specified.</p>
784
785<p>An explicit alignment may be specified for a function. If not present, or if
786the alignment is set to zero, the alignment of the function is set by the target
787to whatever it feels convenient. If an explicit alignment is specified, the
788function is forced to have at least that much alignment. All alignments must be
789a power of 2.</p>
790
Devang Pateld0bfcc72008-10-07 17:48:33 +0000[diff] [blame]791 <h5>Syntax:</h5>
792
793<div class="doc_code">
Chris Lattner1e5c5cd02008-10-13 16:55:18 +0000[diff] [blame]794<tt>
795define [<a href="#linkage">linkage</a>] [<a href="#visibility">visibility</a>]
796 [<a href="#callingconv">cconv</a>] [<a href="#paramattrs">ret attrs</a>]
797 &lt;ResultType&gt; @&lt;FunctionName&gt; ([argument list])
798 [<a href="#fnattrs">fn Attrs</a>] [section "name"] [align N]
799 [<a href="#gc">gc</a>] { ... }
800</tt>
Devang Pateld0bfcc72008-10-07 17:48:33 +0000[diff] [blame]801</div>
802
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]803</div>
804
805
806<!-- ======================================================================= -->
807<div class="doc_subsection">
808 <a name="aliasstructure">Aliases</a>
809</div>
810<div class="doc_text">
811 <p>Aliases act as "second name" for the aliasee value (which can be either
Anton Korobeynikov96822812008-03-22 08:36:14 +0000[diff] [blame]812 function, global variable, another alias or bitcast of global value). Aliases
813 may have an optional <a href="#linkage">linkage type</a>, and an
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]814 optional <a href="#visibility">visibility style</a>.</p>
815
816 <h5>Syntax:</h5>
817
818<div class="doc_code">
819<pre>
Duncan Sandsd7bfabf2008-09-12 20:48:21 +0000[diff] [blame]820@&lt;Name&gt; = alias [Linkage] [Visibility] &lt;AliaseeTy&gt; @&lt;Aliasee&gt;
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]821</pre>
822</div>
823
824</div>
825
826
827
828<!-- ======================================================================= -->
829<div class="doc_subsection"><a name="paramattrs">Parameter Attributes</a></div>
830<div class="doc_text">
831 <p>The return type and each parameter of a function type may have a set of
832 <i>parameter attributes</i> associated with them. Parameter attributes are
833 used to communicate additional information about the result or parameters of
Duncan Sandsf5588dc2007-11-27 13:23:08 +0000[diff] [blame]834 a function. Parameter attributes are considered to be part of the function,
835 not of the function type, so functions with different parameter attributes
836 can have the same function type.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]837
838 <p>Parameter attributes are simple keywords that follow the type specified. If
839 multiple parameter attributes are needed, they are space separated. For
840 example:</p>
841
842<div class="doc_code">
843<pre>
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]844declare i32 @printf(i8* noalias , ...)
Chris Lattnerf33b8452008-10-04 18:33:34 +0000[diff] [blame]845declare i32 @atoi(i8 zeroext)
846declare signext i8 @returns_signed_char()
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]847</pre>
848</div>
849
Duncan Sandsf5588dc2007-11-27 13:23:08 +0000[diff] [blame]850 <p>Note that any attributes for the function result (<tt>nounwind</tt>,
851 <tt>readonly</tt>) come immediately after the argument list.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]852
853 <p>Currently, only the following parameter attributes are defined:</p>
854 <dl>
Reid Spencerf234bed2007-07-19 23:13:04 +0000[diff] [blame]855 <dt><tt>zeroext</tt></dt>
Chris Lattnerf33b8452008-10-04 18:33:34 +0000[diff] [blame]856 <dd>This indicates to the code generator that the parameter or return value
857 should be zero-extended to a 32-bit value by the caller (for a parameter)
858 or the callee (for a return value).</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]859
Reid Spencerf234bed2007-07-19 23:13:04 +0000[diff] [blame]860 <dt><tt>signext</tt></dt>
Chris Lattnerf33b8452008-10-04 18:33:34 +0000[diff] [blame]861 <dd>This indicates to the code generator that the parameter or return value
862 should be sign-extended to a 32-bit value by the caller (for a parameter)
863 or the callee (for a return value).</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]864
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]865 <dt><tt>inreg</tt></dt>
Dale Johannesenc08a0e22008-09-25 20:47:45 +0000[diff] [blame]866 <dd>This indicates that this parameter or return value should be treated
867 in a special target-dependent fashion during while emitting code for a
868 function call or return (usually, by putting it in a register as opposed
Chris Lattnerf33b8452008-10-04 18:33:34 +0000[diff] [blame]869 to memory, though some targets use it to distinguish between two different
870 kinds of registers). Use of this attribute is target-specific.</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]871
Duncan Sands7d94e5a2008-10-06 08:14:18 +0000[diff] [blame]872 <dt><tt><a name="byval">byval</a></tt></dt>
Chris Lattner04c86182008-01-15 04:34:22 +0000[diff] [blame]873 <dd>This indicates that the pointer parameter should really be passed by
874 value to the function. The attribute implies that a hidden copy of the
875 pointee is made between the caller and the callee, so the callee is unable
Chris Lattner6a9f3c42008-08-05 18:21:08 +0000[diff] [blame]876 to modify the value in the callee. This attribute is only valid on LLVM
Chris Lattner04c86182008-01-15 04:34:22 +0000[diff] [blame]877 pointer arguments. It is generally used to pass structs and arrays by
Duncan Sands7d94e5a2008-10-06 08:14:18 +0000[diff] [blame]878 value, but is also valid on pointers to scalars. The copy is considered to
879 belong to the caller not the callee (for example,
880 <tt><a href="#readonly">readonly</a></tt> functions should not write to
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]881 <tt>byval</tt> parameters). This is not a valid attribute for return
882 values. </dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]883
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]884 <dt><tt>sret</tt></dt>
Duncan Sands616cc032008-02-18 04:19:38 +0000[diff] [blame]885 <dd>This indicates that the pointer parameter specifies the address of a
886 structure that is the return value of the function in the source program.
Chris Lattnerf33b8452008-10-04 18:33:34 +0000[diff] [blame]887 This pointer must be guaranteed by the caller to be valid: loads and stores
888 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]889 be applied to the first parameter. This is not a valid attribute for
890 return values. </dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]891
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]892 <dt><tt>noalias</tt></dt>
Nick Lewyckyff384472008-11-24 03:41:24 +0000[diff] [blame]893 <dd>This indicates that the pointer does not alias any global or any other
894 parameter. The caller is responsible for ensuring that this is the
Nick Lewyckya604a942008-11-24 05:00:44 +0000[diff] [blame]895 case. On a function return value, <tt>noalias</tt> additionally indicates
896 that the pointer does not alias any other pointers visible to the
Edwin Török76433cc2008-11-24 08:02:24 +0000[diff] [blame]897 caller. Note that this applies only to pointers that can be used to actually
898 load/store a value: NULL, unique pointers from malloc(0), and freed pointers
899 are considered to not alias anything.</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]900
Duncan Sands4ee46812007-07-27 19:57:41 +0000[diff] [blame]901 <dt><tt>nest</tt></dt>
Duncan Sandsf1a7d4c2008-07-08 09:27:25 +0000[diff] [blame]902 <dd>This indicates that the pointer parameter can be excised using the
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]903 <a href="#int_trampoline">trampoline intrinsics</a>. This is not a valid
904 attribute for return values.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]905 </dl>
906
907</div>
908
909<!-- ======================================================================= -->
910<div class="doc_subsection">
Gordon Henriksen13fe5e32007-12-10 03:18:06 +0000[diff] [blame]911 <a name="gc">Garbage Collector Names</a>
912</div>
913
914<div class="doc_text">
915<p>Each function may specify a garbage collector name, which is simply a
916string.</p>
917
918<div class="doc_code"><pre
919>define void @f() gc "name" { ...</pre></div>
920
921<p>The compiler declares the supported values of <i>name</i>. Specifying a
922collector which will cause the compiler to alter its output in order to support
923the named garbage collection algorithm.</p>
924</div>
925
926<!-- ======================================================================= -->
927<div class="doc_subsection">
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]928 <a name="fnattrs">Function Attributes</a>
Devang Pateld468f1c2008-09-04 23:05:13 +0000[diff] [blame]929</div>
930
931<div class="doc_text">
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]932
933<p>Function attributes are set to communicate additional information about
934 a function. Function attributes are considered to be part of the function,
935 not of the function type, so functions with different parameter attributes
936 can have the same function type.</p>
937
938 <p>Function attributes are simple keywords that follow the type specified. If
939 multiple attributes are needed, they are space separated. For
940 example:</p>
Devang Pateld468f1c2008-09-04 23:05:13 +0000[diff] [blame]941
942<div class="doc_code">
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]943<pre>
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]944define void @f() noinline { ... }
945define void @f() alwaysinline { ... }
946define void @f() alwaysinline optsize { ... }
947define void @f() optsize
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]948</pre>
Devang Pateld468f1c2008-09-04 23:05:13 +0000[diff] [blame]949</div>
950
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]951<dl>
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]952<dt><tt>alwaysinline</tt></dt>
Chris Lattner82d70f92008-10-04 18:23:17 +0000[diff] [blame]953<dd>This attribute indicates that the inliner should attempt to inline this
954function into callers whenever possible, ignoring any active inlining size
955threshold for this caller.</dd>
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]956
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]957<dt><tt>noinline</tt></dt>
Chris Lattner82d70f92008-10-04 18:23:17 +0000[diff] [blame]958<dd>This attribute indicates that the inliner should never inline this function
Chris Lattner377f3ae2008-10-05 17:14:59 +0000[diff] [blame]959in any situation. This attribute may not be used together with the
Chris Lattner82d70f92008-10-04 18:23:17 +0000[diff] [blame]960<tt>alwaysinline</tt> attribute.</dd>
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]961
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]962<dt><tt>optsize</tt></dt>
Devang Patelc29099b2008-09-29 18:34:44 +0000[diff] [blame]963<dd>This attribute suggests that optimization passes and code generator passes
Chris Lattner82d70f92008-10-04 18:23:17 +0000[diff] [blame]964make choices that keep the code size of this function low, and otherwise do
965optimizations specifically to reduce code size.</dd>
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]966
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]967<dt><tt>noreturn</tt></dt>
Chris Lattner82d70f92008-10-04 18:23:17 +0000[diff] [blame]968<dd>This function attribute indicates that the function never returns normally.
969This produces undefined behavior at runtime if the function ever does
970dynamically return.</dd>
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]971
972<dt><tt>nounwind</tt></dt>
Chris Lattner82d70f92008-10-04 18:23:17 +0000[diff] [blame]973<dd>This function attribute indicates that the function never returns with an
974unwind or exceptional control flow. If the function does unwind, its runtime
975behavior is undefined.</dd>
976
977<dt><tt>readnone</tt></dt>
Duncan Sands7d94e5a2008-10-06 08:14:18 +0000[diff] [blame]978<dd>This attribute indicates that the function computes its result (or the
979exception it throws) based strictly on its arguments, without dereferencing any
980pointer arguments or otherwise accessing any mutable state (e.g. memory, control
981registers, etc) visible to caller functions. It does not write through any
982pointer arguments (including <tt><a href="#byval">byval</a></tt> arguments) and
983never changes any state visible to callers.</dd>
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]984
Duncan Sands7d94e5a2008-10-06 08:14:18 +0000[diff] [blame]985<dt><tt><a name="readonly">readonly</a></tt></dt>
986<dd>This attribute indicates that the function does not write through any
987pointer arguments (including <tt><a href="#byval">byval</a></tt> arguments)
988or otherwise modify any state (e.g. memory, control registers, etc) visible to
989caller functions. It may dereference pointer arguments and read state that may
990be set in the caller. A readonly function always returns the same value (or
991throws the same exception) when called with the same set of arguments and global
992state.</dd>
Bill Wendlingdfaabba2008-11-13 01:02:51 +0000[diff] [blame]993
994<dt><tt><a name="ssp">ssp</a></tt></dt>
Bill Wendling34f70d82008-11-26 19:19:05 +0000[diff] [blame]995<dd>This attribute indicates that the function should emit a stack smashing
Bill Wendlingdfaabba2008-11-13 01:02:51 +0000[diff] [blame]996protector. It is in the form of a "canary"&mdash;a random value placed on the
997stack before the local variables that's checked upon return from the function to
998see if it has been overwritten. A heuristic is used to determine if a function
Bill Wendling34f70d82008-11-26 19:19:05 +0000[diff] [blame]999needs stack protectors or not.
Bill Wendlingdfaabba2008-11-13 01:02:51 +0000[diff] [blame]1000
Bill Wendlingbe9ec3f2008-11-26 19:07:40 +0000[diff] [blame]1001<p>If a function that has an <tt>ssp</tt> attribute is inlined into a function
1002that doesn't have an <tt>ssp</tt> attribute, then the resulting function will
1003have an <tt>ssp</tt> attribute.</p></dd>
1004
1005<dt><tt>sspreq</tt></dt>
Bill Wendling34f70d82008-11-26 19:19:05 +0000[diff] [blame]1006<dd>This attribute indicates that the function should <em>always</em> emit a
Bill Wendlingdfaabba2008-11-13 01:02:51 +0000[diff] [blame]1007stack smashing protector. This overrides the <tt><a href="#ssp">ssp</a></tt>
Bill Wendling34f70d82008-11-26 19:19:05 +0000[diff] [blame]1008function attribute.
Bill Wendlingbe9ec3f2008-11-26 19:07:40 +0000[diff] [blame]1009
1010<p>If a function that has an <tt>sspreq</tt> attribute is inlined into a
1011function that doesn't have an <tt>sspreq</tt> attribute or which has
1012an <tt>ssp</tt> attribute, then the resulting function will have
1013an <tt>sspreq</tt> attribute.</p></dd>
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]1014</dl>
1015
Devang Pateld468f1c2008-09-04 23:05:13 +0000[diff] [blame]1016</div>
1017
1018<!-- ======================================================================= -->
1019<div class="doc_subsection">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1020 <a name="moduleasm">Module-Level Inline Assembly</a>
1021</div>
1022
1023<div class="doc_text">
1024<p>
1025Modules may contain "module-level inline asm" blocks, which corresponds to the
1026GCC "file scope inline asm" blocks. These blocks are internally concatenated by
1027LLVM and treated as a single unit, but may be separated in the .ll file if
1028desired. The syntax is very simple:
1029</p>
1030
1031<div class="doc_code">
1032<pre>
1033module asm "inline asm code goes here"
1034module asm "more can go here"
1035</pre>
1036</div>
1037
1038<p>The strings can contain any character by escaping non-printable characters.
1039 The escape sequence used is simply "\xx" where "xx" is the two digit hex code
1040 for the number.
1041</p>
1042
1043<p>
1044 The inline asm code is simply printed to the machine code .s file when
1045 assembly code is generated.
1046</p>
1047</div>
1048
1049<!-- ======================================================================= -->
1050<div class="doc_subsection">
1051 <a name="datalayout">Data Layout</a>
1052</div>
1053
1054<div class="doc_text">
1055<p>A module may specify a target specific data layout string that specifies how
1056data is to be laid out in memory. The syntax for the data layout is simply:</p>
1057<pre> target datalayout = "<i>layout specification</i>"</pre>
1058<p>The <i>layout specification</i> consists of a list of specifications
1059separated by the minus sign character ('-'). Each specification starts with a
1060letter and may include other information after the letter to define some
1061aspect of the data layout. The specifications accepted are as follows: </p>
1062<dl>
1063 <dt><tt>E</tt></dt>
1064 <dd>Specifies that the target lays out data in big-endian form. That is, the
1065 bits with the most significance have the lowest address location.</dd>
1066 <dt><tt>e</tt></dt>
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]1067 <dd>Specifies that the target lays out data in little-endian form. That is,
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1068 the bits with the least significance have the lowest address location.</dd>
1069 <dt><tt>p:<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1070 <dd>This specifies the <i>size</i> of a pointer and its <i>abi</i> and
1071 <i>preferred</i> alignments. All sizes are in bits. Specifying the <i>pref</i>
1072 alignment is optional. If omitted, the preceding <tt>:</tt> should be omitted
1073 too.</dd>
1074 <dt><tt>i<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1075 <dd>This specifies the alignment for an integer type of a given bit
1076 <i>size</i>. The value of <i>size</i> must be in the range [1,2^23).</dd>
1077 <dt><tt>v<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1078 <dd>This specifies the alignment for a vector type of a given bit
1079 <i>size</i>.</dd>
1080 <dt><tt>f<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1081 <dd>This specifies the alignment for a floating point type of a given bit
1082 <i>size</i>. The value of <i>size</i> must be either 32 (float) or 64
1083 (double).</dd>
1084 <dt><tt>a<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1085 <dd>This specifies the alignment for an aggregate type of a given bit
1086 <i>size</i>.</dd>
1087</dl>
1088<p>When constructing the data layout for a given target, LLVM starts with a
1089default set of specifications which are then (possibly) overriden by the
1090specifications in the <tt>datalayout</tt> keyword. The default specifications
1091are given in this list:</p>
1092<ul>
1093 <li><tt>E</tt> - big endian</li>
1094 <li><tt>p:32:64:64</tt> - 32-bit pointers with 64-bit alignment</li>
1095 <li><tt>i1:8:8</tt> - i1 is 8-bit (byte) aligned</li>
1096 <li><tt>i8:8:8</tt> - i8 is 8-bit (byte) aligned</li>
1097 <li><tt>i16:16:16</tt> - i16 is 16-bit aligned</li>
1098 <li><tt>i32:32:32</tt> - i32 is 32-bit aligned</li>
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]1099 <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]1100 alignment of 64-bits</li>
1101 <li><tt>f32:32:32</tt> - float is 32-bit aligned</li>
1102 <li><tt>f64:64:64</tt> - double is 64-bit aligned</li>
1103 <li><tt>v64:64:64</tt> - 64-bit vector is 64-bit aligned</li>
1104 <li><tt>v128:128:128</tt> - 128-bit vector is 128-bit aligned</li>
1105 <li><tt>a0:0:1</tt> - aggregates are 8-bit aligned</li>
1106</ul>
Chris Lattner6a9f3c42008-08-05 18:21:08 +0000[diff] [blame]1107<p>When LLVM is determining the alignment for a given type, it uses the
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]1108following rules:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1109<ol>
1110 <li>If the type sought is an exact match for one of the specifications, that
1111 specification is used.</li>
1112 <li>If no match is found, and the type sought is an integer type, then the
1113 smallest integer type that is larger than the bitwidth of the sought type is
1114 used. If none of the specifications are larger than the bitwidth then the the
1115 largest integer type is used. For example, given the default specifications
1116 above, the i7 type will use the alignment of i8 (next largest) while both
1117 i65 and i256 will use the alignment of i64 (largest specified).</li>
1118 <li>If no match is found, and the type sought is a vector type, then the
1119 largest vector type that is smaller than the sought vector type will be used
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]1120 as a fall back. This happens because &lt;128 x double&gt; can be implemented
1121 in terms of 64 &lt;2 x double&gt;, for example.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1122</ol>
1123</div>
1124
1125<!-- *********************************************************************** -->
1126<div class="doc_section"> <a name="typesystem">Type System</a> </div>
1127<!-- *********************************************************************** -->
1128
1129<div class="doc_text">
1130
1131<p>The LLVM type system is one of the most important features of the
1132intermediate representation. Being typed enables a number of
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]1133optimizations to be performed on the intermediate representation directly,
1134without having to do
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1135extra analyses on the side before the transformation. A strong type
1136system makes it easier to read the generated code and enables novel
1137analyses and transformations that are not feasible to perform on normal
1138three address code representations.</p>
1139
1140</div>
1141
1142<!-- ======================================================================= -->
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1143<div class="doc_subsection"> <a name="t_classifications">Type
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1144Classifications</a> </div>
1145<div class="doc_text">
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1146<p>The types fall into a few useful
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1147classifications:</p>
1148
1149<table border="1" cellspacing="0" cellpadding="4">
1150 <tbody>
1151 <tr><th>Classification</th><th>Types</th></tr>
1152 <tr>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1153 <td><a href="#t_integer">integer</a></td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1154 <td><tt>i1, i2, i3, ... i8, ... i16, ... i32, ... i64, ... </tt></td>
1155 </tr>
1156 <tr>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1157 <td><a href="#t_floating">floating point</a></td>
1158 <td><tt>float, double, x86_fp80, fp128, ppc_fp128</tt></td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1159 </tr>
1160 <tr>
1161 <td><a name="t_firstclass">first class</a></td>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1162 <td><a href="#t_integer">integer</a>,
1163 <a href="#t_floating">floating point</a>,
1164 <a href="#t_pointer">pointer</a>,
Dan Gohmanf6237db2008-06-18 18:42:13 +0000[diff] [blame]1165 <a href="#t_vector">vector</a>,
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]1166 <a href="#t_struct">structure</a>,
1167 <a href="#t_array">array</a>,
Dan Gohmand13951e2008-05-23 22:50:26 +0000[diff] [blame]1168 <a href="#t_label">label</a>.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1169 </td>
1170 </tr>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1171 <tr>
1172 <td><a href="#t_primitive">primitive</a></td>
1173 <td><a href="#t_label">label</a>,
1174 <a href="#t_void">void</a>,
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1175 <a href="#t_floating">floating point</a>.</td>
1176 </tr>
1177 <tr>
1178 <td><a href="#t_derived">derived</a></td>
1179 <td><a href="#t_integer">integer</a>,
1180 <a href="#t_array">array</a>,
1181 <a href="#t_function">function</a>,
1182 <a href="#t_pointer">pointer</a>,
1183 <a href="#t_struct">structure</a>,
1184 <a href="#t_pstruct">packed structure</a>,
1185 <a href="#t_vector">vector</a>,
1186 <a href="#t_opaque">opaque</a>.
Dan Gohman032ba852008-10-14 16:32:04 +0000[diff] [blame]1187 </td>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1188 </tr>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1189 </tbody>
1190</table>
1191
1192<p>The <a href="#t_firstclass">first class</a> types are perhaps the
1193most important. Values of these types are the only ones which can be
1194produced by instructions, passed as arguments, or used as operands to
Dan Gohman4f29e422008-05-23 21:53:15 +0000[diff] [blame]1195instructions.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1196</div>
1197
1198<!-- ======================================================================= -->
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1199<div class="doc_subsection"> <a name="t_primitive">Primitive Types</a> </div>
Chris Lattner86437612008-01-04 04:34:14 +0000[diff] [blame]1200
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1201<div class="doc_text">
1202<p>The primitive types are the fundamental building blocks of the LLVM
1203system.</p>
1204
Chris Lattner86437612008-01-04 04:34:14 +0000[diff] [blame]1205</div>
1206
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1207<!-- _______________________________________________________________________ -->
1208<div class="doc_subsubsection"> <a name="t_floating">Floating Point Types</a> </div>
1209
1210<div class="doc_text">
1211 <table>
1212 <tbody>
1213 <tr><th>Type</th><th>Description</th></tr>
1214 <tr><td><tt>float</tt></td><td>32-bit floating point value</td></tr>
1215 <tr><td><tt>double</tt></td><td>64-bit floating point value</td></tr>
1216 <tr><td><tt>fp128</tt></td><td>128-bit floating point value (112-bit mantissa)</td></tr>
1217 <tr><td><tt>x86_fp80</tt></td><td>80-bit floating point value (X87)</td></tr>
1218 <tr><td><tt>ppc_fp128</tt></td><td>128-bit floating point value (two 64-bits)</td></tr>
1219 </tbody>
1220 </table>
1221</div>
1222
1223<!-- _______________________________________________________________________ -->
1224<div class="doc_subsubsection"> <a name="t_void">Void Type</a> </div>
1225
1226<div class="doc_text">
1227<h5>Overview:</h5>
1228<p>The void type does not represent any value and has no size.</p>
1229
1230<h5>Syntax:</h5>
1231
1232<pre>
1233 void
1234</pre>
1235</div>
1236
1237<!-- _______________________________________________________________________ -->
1238<div class="doc_subsubsection"> <a name="t_label">Label Type</a> </div>
1239
1240<div class="doc_text">
1241<h5>Overview:</h5>
1242<p>The label type represents code labels.</p>
1243
1244<h5>Syntax:</h5>
1245
1246<pre>
1247 label
1248</pre>
1249</div>
1250
1251
1252<!-- ======================================================================= -->
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1253<div class="doc_subsection"> <a name="t_derived">Derived Types</a> </div>
1254
1255<div class="doc_text">
1256
1257<p>The real power in LLVM comes from the derived types in the system.
1258This is what allows a programmer to represent arrays, functions,
1259pointers, and other useful types. Note that these derived types may be
1260recursive: For example, it is possible to have a two dimensional array.</p>
1261
1262</div>
1263
1264<!-- _______________________________________________________________________ -->
1265<div class="doc_subsubsection"> <a name="t_integer">Integer Type</a> </div>
1266
1267<div class="doc_text">
1268
1269<h5>Overview:</h5>
1270<p>The integer type is a very simple derived type that simply specifies an
1271arbitrary bit width for the integer type desired. Any bit width from 1 bit to
12722^23-1 (about 8 million) can be specified.</p>
1273
1274<h5>Syntax:</h5>
1275
1276<pre>
1277 iN
1278</pre>
1279
1280<p>The number of bits the integer will occupy is specified by the <tt>N</tt>
1281value.</p>
1282
1283<h5>Examples:</h5>
1284<table class="layout">
Chris Lattner251ab812007-12-18 06:18:21 +0000[diff] [blame]1285 <tbody>
1286 <tr>
1287 <td><tt>i1</tt></td>
1288 <td>a single-bit integer.</td>
1289 </tr><tr>
1290 <td><tt>i32</tt></td>
1291 <td>a 32-bit integer.</td>
1292 </tr><tr>
1293 <td><tt>i1942652</tt></td>
1294 <td>a really big integer of over 1 million bits.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1295 </tr>
Chris Lattner251ab812007-12-18 06:18:21 +0000[diff] [blame]1296 </tbody>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1297</table>
1298</div>
1299
1300<!-- _______________________________________________________________________ -->
1301<div class="doc_subsubsection"> <a name="t_array">Array Type</a> </div>
1302
1303<div class="doc_text">
1304
1305<h5>Overview:</h5>
1306
1307<p>The array type is a very simple derived type that arranges elements
1308sequentially in memory. The array type requires a size (number of
1309elements) and an underlying data type.</p>
1310
1311<h5>Syntax:</h5>
1312
1313<pre>
1314 [&lt;# elements&gt; x &lt;elementtype&gt;]
1315</pre>
1316
1317<p>The number of elements is a constant integer value; elementtype may
1318be any type with a size.</p>
1319
1320<h5>Examples:</h5>
1321<table class="layout">
1322 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1323 <td class="left"><tt>[40 x i32]</tt></td>
1324 <td class="left">Array of 40 32-bit integer values.</td>
1325 </tr>
1326 <tr class="layout">
1327 <td class="left"><tt>[41 x i32]</tt></td>
1328 <td class="left">Array of 41 32-bit integer values.</td>
1329 </tr>
1330 <tr class="layout">
1331 <td class="left"><tt>[4 x i8]</tt></td>
1332 <td class="left">Array of 4 8-bit integer values.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1333 </tr>
1334</table>
1335<p>Here are some examples of multidimensional arrays:</p>
1336<table class="layout">
1337 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1338 <td class="left"><tt>[3 x [4 x i32]]</tt></td>
1339 <td class="left">3x4 array of 32-bit integer values.</td>
1340 </tr>
1341 <tr class="layout">
1342 <td class="left"><tt>[12 x [10 x float]]</tt></td>
1343 <td class="left">12x10 array of single precision floating point values.</td>
1344 </tr>
1345 <tr class="layout">
1346 <td class="left"><tt>[2 x [3 x [4 x i16]]]</tt></td>
1347 <td class="left">2x3x4 array of 16-bit integer values.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1348 </tr>
1349</table>
1350
1351<p>Note that 'variable sized arrays' can be implemented in LLVM with a zero
1352length array. Normally, accesses past the end of an array are undefined in
1353LLVM (e.g. it is illegal to access the 5th element of a 3 element array).
1354As a special case, however, zero length arrays are recognized to be variable
1355length. This allows implementation of 'pascal style arrays' with the LLVM
1356type "{ i32, [0 x float]}", for example.</p>
1357
1358</div>
1359
1360<!-- _______________________________________________________________________ -->
1361<div class="doc_subsubsection"> <a name="t_function">Function Type</a> </div>
1362<div class="doc_text">
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1363
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1364<h5>Overview:</h5>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1365
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1366<p>The function type can be thought of as a function signature. It
Devang Pateld4ba41d2008-03-24 05:35:41 +0000[diff] [blame]1367consists of a return type and a list of formal parameter types. The
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1368return 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]1369If 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]1370class types, and the struct must have at least one element.</p>
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]1371
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1372<h5>Syntax:</h5>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1373
1374<pre>
1375 &lt;returntype list&gt; (&lt;parameter list&gt;)
1376</pre>
1377
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1378<p>...where '<tt>&lt;parameter list&gt;</tt>' is a comma-separated list of type
1379specifiers. Optionally, the parameter list may include a type <tt>...</tt>,
1380which indicates that the function takes a variable number of arguments.
1381Variable argument functions can access their arguments with the <a
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]1382 href="#int_varargs">variable argument handling intrinsic</a> functions.
1383'<tt>&lt;returntype list&gt;</tt>' is a comma-separated list of
1384<a href="#t_firstclass">first class</a> type specifiers.</p>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1385
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1386<h5>Examples:</h5>
1387<table class="layout">
1388 <tr class="layout">
1389 <td class="left"><tt>i32 (i32)</tt></td>
1390 <td class="left">function taking an <tt>i32</tt>, returning an <tt>i32</tt>
1391 </td>
1392 </tr><tr class="layout">
Reid Spencerf234bed2007-07-19 23:13:04 +0000[diff] [blame]1393 <td class="left"><tt>float&nbsp;(i16&nbsp;signext,&nbsp;i32&nbsp;*)&nbsp;*
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1394 </tt></td>
1395 <td class="left"><a href="#t_pointer">Pointer</a> to a function that takes
1396 an <tt>i16</tt> that should be sign extended and a
1397 <a href="#t_pointer">pointer</a> to <tt>i32</tt>, returning
1398 <tt>float</tt>.
1399 </td>
1400 </tr><tr class="layout">
1401 <td class="left"><tt>i32 (i8*, ...)</tt></td>
1402 <td class="left">A vararg function that takes at least one
1403 <a href="#t_pointer">pointer</a> to <tt>i8 </tt> (char in C),
1404 which returns an integer. This is the signature for <tt>printf</tt> in
1405 LLVM.
1406 </td>
Devang Pateld4ba41d2008-03-24 05:35:41 +0000[diff] [blame]1407 </tr><tr class="layout">
1408 <td class="left"><tt>{i32, i32} (i32)</tt></td>
Misha Brukmanafc88b02008-11-27 06:41:20 +0000[diff] [blame]1409 <td class="left">A function taking an <tt>i32</tt>, returning two
1410 <tt>i32</tt> values as an aggregate of type <tt>{ i32, i32 }</tt>
Devang Pateld4ba41d2008-03-24 05:35:41 +0000[diff] [blame]1411 </td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1412 </tr>
1413</table>
1414
1415</div>
1416<!-- _______________________________________________________________________ -->
1417<div class="doc_subsubsection"> <a name="t_struct">Structure Type</a> </div>
1418<div class="doc_text">
1419<h5>Overview:</h5>
1420<p>The structure type is used to represent a collection of data members
1421together in memory. The packing of the field types is defined to match
1422the ABI of the underlying processor. The elements of a structure may
1423be any type that has a size.</p>
1424<p>Structures are accessed using '<tt><a href="#i_load">load</a></tt>
1425and '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a
1426field with the '<tt><a href="#i_getelementptr">getelementptr</a></tt>'
1427instruction.</p>
1428<h5>Syntax:</h5>
1429<pre> { &lt;type list&gt; }<br></pre>
1430<h5>Examples:</h5>
1431<table class="layout">
1432 <tr class="layout">
1433 <td class="left"><tt>{ i32, i32, i32 }</tt></td>
1434 <td class="left">A triple of three <tt>i32</tt> values</td>
1435 </tr><tr class="layout">
1436 <td class="left"><tt>{&nbsp;float,&nbsp;i32&nbsp;(i32)&nbsp;*&nbsp;}</tt></td>
1437 <td class="left">A pair, where the first element is a <tt>float</tt> and the
1438 second element is a <a href="#t_pointer">pointer</a> to a
1439 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
1440 an <tt>i32</tt>.</td>
1441 </tr>
1442</table>
1443</div>
1444
1445<!-- _______________________________________________________________________ -->
1446<div class="doc_subsubsection"> <a name="t_pstruct">Packed Structure Type</a>
1447</div>
1448<div class="doc_text">
1449<h5>Overview:</h5>
1450<p>The packed structure type is used to represent a collection of data members
1451together in memory. There is no padding between fields. Further, the alignment
1452of a packed structure is 1 byte. The elements of a packed structure may
1453be any type that has a size.</p>
1454<p>Structures are accessed using '<tt><a href="#i_load">load</a></tt>
1455and '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a
1456field with the '<tt><a href="#i_getelementptr">getelementptr</a></tt>'
1457instruction.</p>
1458<h5>Syntax:</h5>
1459<pre> &lt; { &lt;type list&gt; } &gt; <br></pre>
1460<h5>Examples:</h5>
1461<table class="layout">
1462 <tr class="layout">
1463 <td class="left"><tt>&lt; { i32, i32, i32 } &gt;</tt></td>
1464 <td class="left">A triple of three <tt>i32</tt> values</td>
1465 </tr><tr class="layout">
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]1466 <td class="left">
1467<tt>&lt;&nbsp;{&nbsp;float,&nbsp;i32&nbsp;(i32)*&nbsp;}&nbsp;&gt;</tt></td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1468 <td class="left">A pair, where the first element is a <tt>float</tt> and the
1469 second element is a <a href="#t_pointer">pointer</a> to a
1470 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
1471 an <tt>i32</tt>.</td>
1472 </tr>
1473</table>
1474</div>
1475
1476<!-- _______________________________________________________________________ -->
1477<div class="doc_subsubsection"> <a name="t_pointer">Pointer Type</a> </div>
1478<div class="doc_text">
1479<h5>Overview:</h5>
1480<p>As in many languages, the pointer type represents a pointer or
Christopher Lambdd0049d2007-12-11 09:31:00 +0000[diff] [blame]1481reference to another object, which must live in memory. Pointer types may have
1482an optional address space attribute defining the target-specific numbered
1483address space where the pointed-to object resides. The default address space is
1484zero.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1485<h5>Syntax:</h5>
1486<pre> &lt;type&gt; *<br></pre>
1487<h5>Examples:</h5>
1488<table class="layout">
1489 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1490 <td class="left"><tt>[4x i32]*</tt></td>
1491 <td class="left">A <a href="#t_pointer">pointer</a> to <a
1492 href="#t_array">array</a> of four <tt>i32</tt> values.</td>
1493 </tr>
1494 <tr class="layout">
1495 <td class="left"><tt>i32 (i32 *) *</tt></td>
1496 <td class="left"> A <a href="#t_pointer">pointer</a> to a <a
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1497 href="#t_function">function</a> that takes an <tt>i32*</tt>, returning an
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1498 <tt>i32</tt>.</td>
1499 </tr>
1500 <tr class="layout">
1501 <td class="left"><tt>i32 addrspace(5)*</tt></td>
1502 <td class="left">A <a href="#t_pointer">pointer</a> to an <tt>i32</tt> value
1503 that resides in address space #5.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1504 </tr>
1505</table>
1506</div>
1507
1508<!-- _______________________________________________________________________ -->
1509<div class="doc_subsubsection"> <a name="t_vector">Vector Type</a> </div>
1510<div class="doc_text">
1511
1512<h5>Overview:</h5>
1513
1514<p>A vector type is a simple derived type that represents a vector
1515of elements. Vector types are used when multiple primitive data
1516are operated in parallel using a single instruction (SIMD).
1517A vector type requires a size (number of
1518elements) and an underlying primitive data type. Vectors must have a power
1519of two length (1, 2, 4, 8, 16 ...). Vector types are
1520considered <a href="#t_firstclass">first class</a>.</p>
1521
1522<h5>Syntax:</h5>
1523
1524<pre>
1525 &lt; &lt;# elements&gt; x &lt;elementtype&gt; &gt;
1526</pre>
1527
1528<p>The number of elements is a constant integer value; elementtype may
1529be any integer or floating point type.</p>
1530
1531<h5>Examples:</h5>
1532
1533<table class="layout">
1534 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1535 <td class="left"><tt>&lt;4 x i32&gt;</tt></td>
1536 <td class="left">Vector of 4 32-bit integer values.</td>
1537 </tr>
1538 <tr class="layout">
1539 <td class="left"><tt>&lt;8 x float&gt;</tt></td>
1540 <td class="left">Vector of 8 32-bit floating-point values.</td>
1541 </tr>
1542 <tr class="layout">
1543 <td class="left"><tt>&lt;2 x i64&gt;</tt></td>
1544 <td class="left">Vector of 2 64-bit integer values.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1545 </tr>
1546</table>
1547</div>
1548
1549<!-- _______________________________________________________________________ -->
1550<div class="doc_subsubsection"> <a name="t_opaque">Opaque Type</a> </div>
1551<div class="doc_text">
1552
1553<h5>Overview:</h5>
1554
1555<p>Opaque types are used to represent unknown types in the system. This
Gordon Henriksenda0706e2007-10-14 00:34:53 +0000[diff] [blame]1556corresponds (for example) to the C notion of a forward declared structure type.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1557In LLVM, opaque types can eventually be resolved to any type (not just a
1558structure type).</p>
1559
1560<h5>Syntax:</h5>
1561
1562<pre>
1563 opaque
1564</pre>
1565
1566<h5>Examples:</h5>
1567
1568<table class="layout">
1569 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1570 <td class="left"><tt>opaque</tt></td>
1571 <td class="left">An opaque type.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1572 </tr>
1573</table>
1574</div>
1575
1576
1577<!-- *********************************************************************** -->
1578<div class="doc_section"> <a name="constants">Constants</a> </div>
1579<!-- *********************************************************************** -->
1580
1581<div class="doc_text">
1582
1583<p>LLVM has several different basic types of constants. This section describes
1584them all and their syntax.</p>
1585
1586</div>
1587
1588<!-- ======================================================================= -->
1589<div class="doc_subsection"><a name="simpleconstants">Simple Constants</a></div>
1590
1591<div class="doc_text">
1592
1593<dl>
1594 <dt><b>Boolean constants</b></dt>
1595
1596 <dd>The two strings '<tt>true</tt>' and '<tt>false</tt>' are both valid
1597 constants of the <tt><a href="#t_primitive">i1</a></tt> type.
1598 </dd>
1599
1600 <dt><b>Integer constants</b></dt>
1601
1602 <dd>Standard integers (such as '4') are constants of the <a
1603 href="#t_integer">integer</a> type. Negative numbers may be used with
1604 integer types.
1605 </dd>
1606
1607 <dt><b>Floating point constants</b></dt>
1608
1609 <dd>Floating point constants use standard decimal notation (e.g. 123.421),
1610 exponential notation (e.g. 1.23421e+2), or a more precise hexadecimal
Chris Lattnerdc15b1d2008-04-01 18:45:27 +0000[diff] [blame]1611 notation (see below). The assembler requires the exact decimal value of
1612 a floating-point constant. For example, the assembler accepts 1.25 but
1613 rejects 1.3 because 1.3 is a repeating decimal in binary. Floating point
1614 constants must have a <a href="#t_floating">floating point</a> type. </dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1615
1616 <dt><b>Null pointer constants</b></dt>
1617
1618 <dd>The identifier '<tt>null</tt>' is recognized as a null pointer constant
1619 and must be of <a href="#t_pointer">pointer type</a>.</dd>
1620
1621</dl>
1622
1623<p>The one non-intuitive notation for constants is the optional hexadecimal form
1624of floating point constants. For example, the form '<tt>double
16250x432ff973cafa8000</tt>' is equivalent to (but harder to read than) '<tt>double
16264.5e+15</tt>'. The only time hexadecimal floating point constants are required
1627(and the only time that they are generated by the disassembler) is when a
1628floating point constant must be emitted but it cannot be represented as a
1629decimal floating point number. For example, NaN's, infinities, and other
1630special values are represented in their IEEE hexadecimal format so that
1631assembly and disassembly do not cause any bits to change in the constants.</p>
1632
1633</div>
1634
1635<!-- ======================================================================= -->
1636<div class="doc_subsection"><a name="aggregateconstants">Aggregate Constants</a>
1637</div>
1638
1639<div class="doc_text">
1640<p>Aggregate constants arise from aggregation of simple constants
1641and smaller aggregate constants.</p>
1642
1643<dl>
1644 <dt><b>Structure constants</b></dt>
1645
1646 <dd>Structure constants are represented with notation similar to structure
1647 type definitions (a comma separated list of elements, surrounded by braces
Chris Lattner6c8de962007-12-25 20:34:52 +0000[diff] [blame]1648 (<tt>{}</tt>)). For example: "<tt>{ i32 4, float 17.0, i32* @G }</tt>",
1649 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]1650 must have <a href="#t_struct">structure type</a>, and the number and
1651 types of elements must match those specified by the type.
1652 </dd>
1653
1654 <dt><b>Array constants</b></dt>
1655
1656 <dd>Array constants are represented with notation similar to array type
1657 definitions (a comma separated list of elements, surrounded by square brackets
1658 (<tt>[]</tt>)). For example: "<tt>[ i32 42, i32 11, i32 74 ]</tt>". Array
1659 constants must have <a href="#t_array">array type</a>, and the number and
1660 types of elements must match those specified by the type.
1661 </dd>
1662
1663 <dt><b>Vector constants</b></dt>
1664
1665 <dd>Vector constants are represented with notation similar to vector type
1666 definitions (a comma separated list of elements, surrounded by
1667 less-than/greater-than's (<tt>&lt;&gt;</tt>)). For example: "<tt>&lt; i32 42,
1668 i32 11, i32 74, i32 100 &gt;</tt>". Vector constants must have <a
1669 href="#t_vector">vector type</a>, and the number and types of elements must
1670 match those specified by the type.
1671 </dd>
1672
1673 <dt><b>Zero initialization</b></dt>
1674
1675 <dd>The string '<tt>zeroinitializer</tt>' can be used to zero initialize a
1676 value to zero of <em>any</em> type, including scalar and aggregate types.
1677 This is often used to avoid having to print large zero initializers (e.g. for
1678 large arrays) and is always exactly equivalent to using explicit zero
1679 initializers.
1680 </dd>
1681</dl>
1682
1683</div>
1684
1685<!-- ======================================================================= -->
1686<div class="doc_subsection">
1687 <a name="globalconstants">Global Variable and Function Addresses</a>
1688</div>
1689
1690<div class="doc_text">
1691
1692<p>The addresses of <a href="#globalvars">global variables</a> and <a
1693href="#functionstructure">functions</a> are always implicitly valid (link-time)
1694constants. These constants are explicitly referenced when the <a
1695href="#identifiers">identifier for the global</a> is used and always have <a
1696href="#t_pointer">pointer</a> type. For example, the following is a legal LLVM
1697file:</p>
1698
1699<div class="doc_code">
1700<pre>
1701@X = global i32 17
1702@Y = global i32 42
1703@Z = global [2 x i32*] [ i32* @X, i32* @Y ]
1704</pre>
1705</div>
1706
1707</div>
1708
1709<!-- ======================================================================= -->
1710<div class="doc_subsection"><a name="undefvalues">Undefined Values</a></div>
1711<div class="doc_text">
1712 <p>The string '<tt>undef</tt>' is recognized as a type-less constant that has
1713 no specific value. Undefined values may be of any type and be used anywhere
1714 a constant is permitted.</p>
1715
1716 <p>Undefined values indicate to the compiler that the program is well defined
1717 no matter what value is used, giving the compiler more freedom to optimize.
1718 </p>
1719</div>
1720
1721<!-- ======================================================================= -->
1722<div class="doc_subsection"><a name="constantexprs">Constant Expressions</a>
1723</div>
1724
1725<div class="doc_text">
1726
1727<p>Constant expressions are used to allow expressions involving other constants
1728to be used as constants. Constant expressions may be of any <a
1729href="#t_firstclass">first class</a> type and may involve any LLVM operation
1730that does not have side effects (e.g. load and call are not supported). The
1731following is the syntax for constant expressions:</p>
1732
1733<dl>
1734 <dt><b><tt>trunc ( CST to TYPE )</tt></b></dt>
1735 <dd>Truncate a constant to another type. The bit size of CST must be larger
1736 than the bit size of TYPE. Both types must be integers.</dd>
1737
1738 <dt><b><tt>zext ( CST to TYPE )</tt></b></dt>
1739 <dd>Zero extend a constant to another type. The bit size of CST must be
1740 smaller or equal to the bit size of TYPE. Both types must be integers.</dd>
1741
1742 <dt><b><tt>sext ( CST to TYPE )</tt></b></dt>
1743 <dd>Sign extend a constant to another type. The bit size of CST must be
1744 smaller or equal to the bit size of TYPE. Both types must be integers.</dd>
1745
1746 <dt><b><tt>fptrunc ( CST to TYPE )</tt></b></dt>
1747 <dd>Truncate a floating point constant to another floating point type. The
1748 size of CST must be larger than the size of TYPE. Both types must be
1749 floating point.</dd>
1750
1751 <dt><b><tt>fpext ( CST to TYPE )</tt></b></dt>
1752 <dd>Floating point extend a constant to another type. The size of CST must be
1753 smaller or equal to the size of TYPE. Both types must be floating point.</dd>
1754
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]1755 <dt><b><tt>fptoui ( CST to TYPE )</tt></b></dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1756 <dd>Convert a floating point constant to the corresponding unsigned integer
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]1757 constant. TYPE must be a scalar or vector integer type. CST must be of scalar
1758 or vector floating point type. Both CST and TYPE must be scalars, or vectors
1759 of the same number of elements. If the value won't fit in the integer type,
1760 the results are undefined.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1761
1762 <dt><b><tt>fptosi ( CST to TYPE )</tt></b></dt>
1763 <dd>Convert a floating point constant to the corresponding signed integer
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]1764 constant. TYPE must be a scalar or vector integer type. CST must be of scalar
1765 or vector floating point type. Both CST and TYPE must be scalars, or vectors
1766 of the same number of elements. If the value won't fit in the integer type,
1767 the results are undefined.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1768
1769 <dt><b><tt>uitofp ( CST to TYPE )</tt></b></dt>
1770 <dd>Convert an unsigned integer constant to the corresponding floating point
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]1771 constant. TYPE must be a scalar or vector floating point type. CST must be of
1772 scalar or vector integer type. Both CST and TYPE must be scalars, or vectors
1773 of the same number of elements. If the value won't fit in the floating point
1774 type, the results are undefined.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1775
1776 <dt><b><tt>sitofp ( CST to TYPE )</tt></b></dt>
1777 <dd>Convert a signed integer constant to the corresponding floating point
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]1778 constant. TYPE must be a scalar or vector floating point type. CST must be of
1779 scalar or vector integer type. Both CST and TYPE must be scalars, or vectors
1780 of the same number of elements. If the value won't fit in the floating point
1781 type, the results are undefined.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1782
1783 <dt><b><tt>ptrtoint ( CST to TYPE )</tt></b></dt>
1784 <dd>Convert a pointer typed constant to the corresponding integer constant
1785 TYPE must be an integer type. CST must be of pointer type. The CST value is
1786 zero extended, truncated, or unchanged to make it fit in TYPE.</dd>
1787
1788 <dt><b><tt>inttoptr ( CST to TYPE )</tt></b></dt>
1789 <dd>Convert a integer constant to a pointer constant. TYPE must be a
1790 pointer type. CST must be of integer type. The CST value is zero extended,
1791 truncated, or unchanged to make it fit in a pointer size. This one is
1792 <i>really</i> dangerous!</dd>
1793
1794 <dt><b><tt>bitcast ( CST to TYPE )</tt></b></dt>
1795 <dd>Convert a constant, CST, to another TYPE. The size of CST and TYPE must be
1796 identical (same number of bits). The conversion is done as if the CST value
1797 was stored to memory and read back as TYPE. In other words, no bits change
1798 with this operator, just the type. This can be used for conversion of
1799 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]1800 pointers it is only valid to cast to another pointer type. It is not valid
1801 to bitcast to or from an aggregate type.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1802 </dd>
1803
1804 <dt><b><tt>getelementptr ( CSTPTR, IDX0, IDX1, ... )</tt></b></dt>
1805
1806 <dd>Perform the <a href="#i_getelementptr">getelementptr operation</a> on
1807 constants. As with the <a href="#i_getelementptr">getelementptr</a>
1808 instruction, the index list may have zero or more indexes, which are required
1809 to make sense for the type of "CSTPTR".</dd>
1810
1811 <dt><b><tt>select ( COND, VAL1, VAL2 )</tt></b></dt>
1812
1813 <dd>Perform the <a href="#i_select">select operation</a> on
1814 constants.</dd>
1815
1816 <dt><b><tt>icmp COND ( VAL1, VAL2 )</tt></b></dt>
1817 <dd>Performs the <a href="#i_icmp">icmp operation</a> on constants.</dd>
1818
1819 <dt><b><tt>fcmp COND ( VAL1, VAL2 )</tt></b></dt>
1820 <dd>Performs the <a href="#i_fcmp">fcmp operation</a> on constants.</dd>
1821
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]1822 <dt><b><tt>vicmp COND ( VAL1, VAL2 )</tt></b></dt>
1823 <dd>Performs the <a href="#i_vicmp">vicmp operation</a> on constants.</dd>
1824
1825 <dt><b><tt>vfcmp COND ( VAL1, VAL2 )</tt></b></dt>
1826 <dd>Performs the <a href="#i_vfcmp">vfcmp operation</a> on constants.</dd>
1827
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1828 <dt><b><tt>extractelement ( VAL, IDX )</tt></b></dt>
1829
1830 <dd>Perform the <a href="#i_extractelement">extractelement
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]1831 operation</a> on constants.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1832
1833 <dt><b><tt>insertelement ( VAL, ELT, IDX )</tt></b></dt>
1834
1835 <dd>Perform the <a href="#i_insertelement">insertelement
1836 operation</a> on constants.</dd>
1837
1838
1839 <dt><b><tt>shufflevector ( VEC1, VEC2, IDXMASK )</tt></b></dt>
1840
1841 <dd>Perform the <a href="#i_shufflevector">shufflevector
1842 operation</a> on constants.</dd>
1843
1844 <dt><b><tt>OPCODE ( LHS, RHS )</tt></b></dt>
1845
1846 <dd>Perform the specified operation of the LHS and RHS constants. OPCODE may
1847 be any of the <a href="#binaryops">binary</a> or <a href="#bitwiseops">bitwise
1848 binary</a> operations. The constraints on operands are the same as those for
1849 the corresponding instruction (e.g. no bitwise operations on floating point
1850 values are allowed).</dd>
1851</dl>
1852</div>
1853
1854<!-- *********************************************************************** -->
1855<div class="doc_section"> <a name="othervalues">Other Values</a> </div>
1856<!-- *********************************************************************** -->
1857
1858<!-- ======================================================================= -->
1859<div class="doc_subsection">
1860<a name="inlineasm">Inline Assembler Expressions</a>
1861</div>
1862
1863<div class="doc_text">
1864
1865<p>
1866LLVM supports inline assembler expressions (as opposed to <a href="#moduleasm">
1867Module-Level Inline Assembly</a>) through the use of a special value. This
1868value represents the inline assembler as a string (containing the instructions
1869to emit), a list of operand constraints (stored as a string), and a flag that
1870indicates whether or not the inline asm expression has side effects. An example
1871inline assembler expression is:
1872</p>
1873
1874<div class="doc_code">
1875<pre>
1876i32 (i32) asm "bswap $0", "=r,r"
1877</pre>
1878</div>
1879
1880<p>
1881Inline assembler expressions may <b>only</b> be used as the callee operand of
1882a <a href="#i_call"><tt>call</tt> instruction</a>. Thus, typically we have:
1883</p>
1884
1885<div class="doc_code">
1886<pre>
1887%X = call i32 asm "<a href="#int_bswap">bswap</a> $0", "=r,r"(i32 %Y)
1888</pre>
1889</div>
1890
1891<p>
1892Inline asms with side effects not visible in the constraint list must be marked
1893as having side effects. This is done through the use of the
1894'<tt>sideeffect</tt>' keyword, like so:
1895</p>
1896
1897<div class="doc_code">
1898<pre>
1899call void asm sideeffect "eieio", ""()
1900</pre>
1901</div>
1902
1903<p>TODO: The format of the asm and constraints string still need to be
1904documented here. Constraints on what can be done (e.g. duplication, moving, etc
Chris Lattner1cdd64e2008-10-04 18:36:02 +0000[diff] [blame]1905need to be documented). This is probably best done by reference to another
1906document that covers inline asm from a holistic perspective.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1907</p>
1908
1909</div>
1910
1911<!-- *********************************************************************** -->
1912<div class="doc_section"> <a name="instref">Instruction Reference</a> </div>
1913<!-- *********************************************************************** -->
1914
1915<div class="doc_text">
1916
1917<p>The LLVM instruction set consists of several different
1918classifications of instructions: <a href="#terminators">terminator
1919instructions</a>, <a href="#binaryops">binary instructions</a>,
1920<a href="#bitwiseops">bitwise binary instructions</a>, <a
1921 href="#memoryops">memory instructions</a>, and <a href="#otherops">other
1922instructions</a>.</p>
1923
1924</div>
1925
1926<!-- ======================================================================= -->
1927<div class="doc_subsection"> <a name="terminators">Terminator
1928Instructions</a> </div>
1929
1930<div class="doc_text">
1931
1932<p>As mentioned <a href="#functionstructure">previously</a>, every
1933basic block in a program ends with a "Terminator" instruction, which
1934indicates which block should be executed after the current block is
1935finished. These terminator instructions typically yield a '<tt>void</tt>'
1936value: they produce control flow, not values (the one exception being
1937the '<a href="#i_invoke"><tt>invoke</tt></a>' instruction).</p>
1938<p>There are six different terminator instructions: the '<a
1939 href="#i_ret"><tt>ret</tt></a>' instruction, the '<a href="#i_br"><tt>br</tt></a>'
1940instruction, the '<a href="#i_switch"><tt>switch</tt></a>' instruction,
1941the '<a href="#i_invoke"><tt>invoke</tt></a>' instruction, the '<a
1942 href="#i_unwind"><tt>unwind</tt></a>' instruction, and the '<a
1943 href="#i_unreachable"><tt>unreachable</tt></a>' instruction.</p>
1944
1945</div>
1946
1947<!-- _______________________________________________________________________ -->
1948<div class="doc_subsubsection"> <a name="i_ret">'<tt>ret</tt>'
1949Instruction</a> </div>
1950<div class="doc_text">
1951<h5>Syntax:</h5>
Dan Gohman3e700032008-10-04 19:00:07 +0000[diff] [blame]1952<pre>
1953 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]1954 ret void <i>; Return from void function</i>
1955</pre>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1956
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1957<h5>Overview:</h5>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1958
Dan Gohman3e700032008-10-04 19:00:07 +0000[diff] [blame]1959<p>The '<tt>ret</tt>' instruction is used to return control flow (and
1960optionally a value) from a function back to the caller.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1961<p>There are two forms of the '<tt>ret</tt>' instruction: one that
Dan Gohman3e700032008-10-04 19:00:07 +0000[diff] [blame]1962returns a value and then causes control flow, and one that just causes
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1963control flow to occur.</p>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1964
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1965<h5>Arguments:</h5>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1966
Dan Gohman3e700032008-10-04 19:00:07 +0000[diff] [blame]1967<p>The '<tt>ret</tt>' instruction optionally accepts a single argument,
1968the return value. The type of the return value must be a
1969'<a href="#t_firstclass">first class</a>' type.</p>
1970
1971<p>A function is not <a href="#wellformed">well formed</a> if
1972it it has a non-void return type and contains a '<tt>ret</tt>'
1973instruction with no return value or a return value with a type that
1974does not match its type, or if it has a void return type and contains
1975a '<tt>ret</tt>' instruction with a return value.</p>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1976
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1977<h5>Semantics:</h5>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1978
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1979<p>When the '<tt>ret</tt>' instruction is executed, control flow
1980returns back to the calling function's context. If the caller is a "<a
1981 href="#i_call"><tt>call</tt></a>" instruction, execution continues at
1982the instruction after the call. If the caller was an "<a
1983 href="#i_invoke"><tt>invoke</tt></a>" instruction, execution continues
1984at the beginning of the "normal" destination block. If the instruction
1985returns a value, that value shall set the call or invoke instruction's
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]1986return value.</p>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1987
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1988<h5>Example:</h5>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1989
1990<pre>
1991 ret i32 5 <i>; Return an integer value of 5</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1992 ret void <i>; Return from a void function</i>
Dan Gohman3e700032008-10-04 19:00:07 +0000[diff] [blame]1993 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]1994</pre>
1995</div>
1996<!-- _______________________________________________________________________ -->
1997<div class="doc_subsubsection"> <a name="i_br">'<tt>br</tt>' Instruction</a> </div>
1998<div class="doc_text">
1999<h5>Syntax:</h5>
2000<pre> br i1 &lt;cond&gt;, label &lt;iftrue&gt;, label &lt;iffalse&gt;<br> br label &lt;dest&gt; <i>; Unconditional branch</i>
2001</pre>
2002<h5>Overview:</h5>
2003<p>The '<tt>br</tt>' instruction is used to cause control flow to
2004transfer to a different basic block in the current function. There are
2005two forms of this instruction, corresponding to a conditional branch
2006and an unconditional branch.</p>
2007<h5>Arguments:</h5>
2008<p>The conditional branch form of the '<tt>br</tt>' instruction takes a
2009single '<tt>i1</tt>' value and two '<tt>label</tt>' values. The
2010unconditional form of the '<tt>br</tt>' instruction takes a single
2011'<tt>label</tt>' value as a target.</p>
2012<h5>Semantics:</h5>
2013<p>Upon execution of a conditional '<tt>br</tt>' instruction, the '<tt>i1</tt>'
2014argument is evaluated. If the value is <tt>true</tt>, control flows
2015to the '<tt>iftrue</tt>' <tt>label</tt> argument. If "cond" is <tt>false</tt>,
2016control flows to the '<tt>iffalse</tt>' <tt>label</tt> argument.</p>
2017<h5>Example:</h5>
2018<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
2019 href="#i_ret">ret</a> i32 1<br>IfUnequal:<br> <a href="#i_ret">ret</a> i32 0<br></pre>
2020</div>
2021<!-- _______________________________________________________________________ -->
2022<div class="doc_subsubsection">
2023 <a name="i_switch">'<tt>switch</tt>' Instruction</a>
2024</div>
2025
2026<div class="doc_text">
2027<h5>Syntax:</h5>
2028
2029<pre>
2030 switch &lt;intty&gt; &lt;value&gt;, label &lt;defaultdest&gt; [ &lt;intty&gt; &lt;val&gt;, label &lt;dest&gt; ... ]
2031</pre>
2032
2033<h5>Overview:</h5>
2034
2035<p>The '<tt>switch</tt>' instruction is used to transfer control flow to one of
2036several different places. It is a generalization of the '<tt>br</tt>'
2037instruction, allowing a branch to occur to one of many possible
2038destinations.</p>
2039
2040
2041<h5>Arguments:</h5>
2042
2043<p>The '<tt>switch</tt>' instruction uses three parameters: an integer
2044comparison value '<tt>value</tt>', a default '<tt>label</tt>' destination, and
2045an array of pairs of comparison value constants and '<tt>label</tt>'s. The
2046table is not allowed to contain duplicate constant entries.</p>
2047
2048<h5>Semantics:</h5>
2049
2050<p>The <tt>switch</tt> instruction specifies a table of values and
2051destinations. When the '<tt>switch</tt>' instruction is executed, this
2052table is searched for the given value. If the value is found, control flow is
2053transfered to the corresponding destination; otherwise, control flow is
2054transfered to the default destination.</p>
2055
2056<h5>Implementation:</h5>
2057
2058<p>Depending on properties of the target machine and the particular
2059<tt>switch</tt> instruction, this instruction may be code generated in different
2060ways. For example, it could be generated as a series of chained conditional
2061branches or with a lookup table.</p>
2062
2063<h5>Example:</h5>
2064
2065<pre>
2066 <i>; Emulate a conditional br instruction</i>
2067 %Val = <a href="#i_zext">zext</a> i1 %value to i32
2068 switch i32 %Val, label %truedest [i32 0, label %falsedest ]
2069
2070 <i>; Emulate an unconditional br instruction</i>
2071 switch i32 0, label %dest [ ]
2072
2073 <i>; Implement a jump table:</i>
2074 switch i32 %val, label %otherwise [ i32 0, label %onzero
2075 i32 1, label %onone
2076 i32 2, label %ontwo ]
2077</pre>
2078</div>
2079
2080<!-- _______________________________________________________________________ -->
2081<div class="doc_subsubsection">
2082 <a name="i_invoke">'<tt>invoke</tt>' Instruction</a>
2083</div>
2084
2085<div class="doc_text">
2086
2087<h5>Syntax:</h5>
2088
2089<pre>
Devang Pateld0bfcc72008-10-07 17:48:33 +0000[diff] [blame]2090 &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]2091 to label &lt;normal label&gt; unwind label &lt;exception label&gt;
2092</pre>
2093
2094<h5>Overview:</h5>
2095
2096<p>The '<tt>invoke</tt>' instruction causes control to transfer to a specified
2097function, with the possibility of control flow transfer to either the
2098'<tt>normal</tt>' label or the
2099'<tt>exception</tt>' label. If the callee function returns with the
2100"<tt><a href="#i_ret">ret</a></tt>" instruction, control flow will return to the
2101"normal" label. If the callee (or any indirect callees) returns with the "<a
2102href="#i_unwind"><tt>unwind</tt></a>" instruction, control is interrupted and
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]2103continued at the dynamically nearest "exception" label.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2104
2105<h5>Arguments:</h5>
2106
2107<p>This instruction requires several arguments:</p>
2108
2109<ol>
2110 <li>
2111 The optional "cconv" marker indicates which <a href="#callingconv">calling
2112 convention</a> the call should use. If none is specified, the call defaults
2113 to using C calling conventions.
2114 </li>
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]2115
2116 <li>The optional <a href="#paramattrs">Parameter Attributes</a> list for
2117 return values. Only '<tt>zeroext</tt>', '<tt>signext</tt>',
2118 and '<tt>inreg</tt>' attributes are valid here.</li>
2119
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2120 <li>'<tt>ptr to function ty</tt>': shall be the signature of the pointer to
2121 function value being invoked. In most cases, this is a direct function
2122 invocation, but indirect <tt>invoke</tt>s are just as possible, branching off
2123 an arbitrary pointer to function value.
2124 </li>
2125
2126 <li>'<tt>function ptr val</tt>': An LLVM value containing a pointer to a
2127 function to be invoked. </li>
2128
2129 <li>'<tt>function args</tt>': argument list whose types match the function
2130 signature argument types. If the function signature indicates the function
2131 accepts a variable number of arguments, the extra arguments can be
2132 specified. </li>
2133
2134 <li>'<tt>normal label</tt>': the label reached when the called function
2135 executes a '<tt><a href="#i_ret">ret</a></tt>' instruction. </li>
2136
2137 <li>'<tt>exception label</tt>': the label reached when a callee returns with
2138 the <a href="#i_unwind"><tt>unwind</tt></a> instruction. </li>
2139
Devang Pateld0bfcc72008-10-07 17:48:33 +0000[diff] [blame]2140 <li>The optional <a href="#fnattrs">function attributes</a> list. Only
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]2141 '<tt>noreturn</tt>', '<tt>nounwind</tt>', '<tt>readonly</tt>' and
2142 '<tt>readnone</tt>' attributes are valid here.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2143</ol>
2144
2145<h5>Semantics:</h5>
2146
2147<p>This instruction is designed to operate as a standard '<tt><a
2148href="#i_call">call</a></tt>' instruction in most regards. The primary
2149difference is that it establishes an association with a label, which is used by
2150the runtime library to unwind the stack.</p>
2151
2152<p>This instruction is used in languages with destructors to ensure that proper
2153cleanup is performed in the case of either a <tt>longjmp</tt> or a thrown
2154exception. Additionally, this is important for implementation of
2155'<tt>catch</tt>' clauses in high-level languages that support them.</p>
2156
2157<h5>Example:</h5>
2158<pre>
Nick Lewyckya1c11a12008-03-16 07:18:12 +0000[diff] [blame]2159 %retval = invoke i32 @Test(i32 15) to label %Continue
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2160 unwind label %TestCleanup <i>; {i32}:retval set</i>
Nick Lewyckya1c11a12008-03-16 07:18:12 +0000[diff] [blame]2161 %retval = invoke <a href="#callingconv">coldcc</a> i32 %Testfnptr(i32 15) to label %Continue
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2162 unwind label %TestCleanup <i>; {i32}:retval set</i>
2163</pre>
2164</div>
2165
2166
2167<!-- _______________________________________________________________________ -->
2168
2169<div class="doc_subsubsection"> <a name="i_unwind">'<tt>unwind</tt>'
2170Instruction</a> </div>
2171
2172<div class="doc_text">
2173
2174<h5>Syntax:</h5>
2175<pre>
2176 unwind
2177</pre>
2178
2179<h5>Overview:</h5>
2180
2181<p>The '<tt>unwind</tt>' instruction unwinds the stack, continuing control flow
2182at the first callee in the dynamic call stack which used an <a
2183href="#i_invoke"><tt>invoke</tt></a> instruction to perform the call. This is
2184primarily used to implement exception handling.</p>
2185
2186<h5>Semantics:</h5>
2187
Chris Lattner8b094fc2008-04-19 21:01:16 +0000[diff] [blame]2188<p>The '<tt>unwind</tt>' instruction causes execution of the current function to
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2189immediately halt. The dynamic call stack is then searched for the first <a
2190href="#i_invoke"><tt>invoke</tt></a> instruction on the call stack. Once found,
2191execution continues at the "exceptional" destination block specified by the
2192<tt>invoke</tt> instruction. If there is no <tt>invoke</tt> instruction in the
2193dynamic call chain, undefined behavior results.</p>
2194</div>
2195
2196<!-- _______________________________________________________________________ -->
2197
2198<div class="doc_subsubsection"> <a name="i_unreachable">'<tt>unreachable</tt>'
2199Instruction</a> </div>
2200
2201<div class="doc_text">
2202
2203<h5>Syntax:</h5>
2204<pre>
2205 unreachable
2206</pre>
2207
2208<h5>Overview:</h5>
2209
2210<p>The '<tt>unreachable</tt>' instruction has no defined semantics. This
2211instruction is used to inform the optimizer that a particular portion of the
2212code is not reachable. This can be used to indicate that the code after a
2213no-return function cannot be reached, and other facts.</p>
2214
2215<h5>Semantics:</h5>
2216
2217<p>The '<tt>unreachable</tt>' instruction has no defined semantics.</p>
2218</div>
2219
2220
2221
2222<!-- ======================================================================= -->
2223<div class="doc_subsection"> <a name="binaryops">Binary Operations</a> </div>
2224<div class="doc_text">
2225<p>Binary operators are used to do most of the computation in a
Chris Lattnerab596d92008-04-01 18:47:32 +0000[diff] [blame]2226program. They require two operands of the same type, execute an operation on them, and
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2227produce a single value. The operands might represent
2228multiple 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]2229The result value has the same type as its operands.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2230<p>There are several different binary operators:</p>
2231</div>
2232<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2233<div class="doc_subsubsection">
2234 <a name="i_add">'<tt>add</tt>' Instruction</a>
2235</div>
2236
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2237<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2238
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2239<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2240
2241<pre>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2242 &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]2243</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2244
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2245<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2246
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2247<p>The '<tt>add</tt>' instruction returns the sum of its two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2248
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2249<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2250
2251<p>The two arguments to the '<tt>add</tt>' instruction must be <a
2252 href="#t_integer">integer</a>, <a href="#t_floating">floating point</a>, or
2253 <a href="#t_vector">vector</a> values. Both arguments must have identical
2254 types.</p>
2255
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2256<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2257
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2258<p>The value produced is the integer or floating point sum of the two
2259operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2260
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2261<p>If an integer sum has unsigned overflow, the result returned is the
2262mathematical result modulo 2<sup>n</sup>, where n is the bit width of
2263the result.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2264
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2265<p>Because LLVM integers use a two's complement representation, this
2266instruction is appropriate for both signed and unsigned integers.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2267
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2268<h5>Example:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2269
2270<pre>
2271 &lt;result&gt; = add i32 4, %var <i>; yields {i32}:result = 4 + %var</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2272</pre>
2273</div>
2274<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2275<div class="doc_subsubsection">
2276 <a name="i_sub">'<tt>sub</tt>' Instruction</a>
2277</div>
2278
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2279<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2280
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2281<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2282
2283<pre>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2284 &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]2285</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2286
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2287<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2288
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2289<p>The '<tt>sub</tt>' instruction returns the difference of its two
2290operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2291
2292<p>Note that the '<tt>sub</tt>' instruction is used to represent the
2293'<tt>neg</tt>' instruction present in most other intermediate
2294representations.</p>
2295
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2296<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2297
2298<p>The two arguments to the '<tt>sub</tt>' instruction must be <a
2299 href="#t_integer">integer</a>, <a href="#t_floating">floating point</a>,
2300 or <a href="#t_vector">vector</a> values. Both arguments must have identical
2301 types.</p>
2302
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2303<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2304
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2305<p>The value produced is the integer or floating point difference of
2306the two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2307
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2308<p>If an integer difference has unsigned overflow, the result returned is the
2309mathematical result modulo 2<sup>n</sup>, where n is the bit width of
2310the result.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2311
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2312<p>Because LLVM integers use a two's complement representation, this
2313instruction is appropriate for both signed and unsigned integers.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2314
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2315<h5>Example:</h5>
2316<pre>
2317 &lt;result&gt; = sub i32 4, %var <i>; yields {i32}:result = 4 - %var</i>
2318 &lt;result&gt; = sub i32 0, %val <i>; yields {i32}:result = -%var</i>
2319</pre>
2320</div>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2321
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2322<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2323<div class="doc_subsubsection">
2324 <a name="i_mul">'<tt>mul</tt>' Instruction</a>
2325</div>
2326
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2327<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2328
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2329<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2330<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]2331</pre>
2332<h5>Overview:</h5>
2333<p>The '<tt>mul</tt>' instruction returns the product of its two
2334operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2335
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2336<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2337
2338<p>The two arguments to the '<tt>mul</tt>' instruction must be <a
2339href="#t_integer">integer</a>, <a href="#t_floating">floating point</a>,
2340or <a href="#t_vector">vector</a> values. Both arguments must have identical
2341types.</p>
2342
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2343<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2344
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2345<p>The value produced is the integer or floating point product of the
2346two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2347
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2348<p>If the result of an integer multiplication has unsigned overflow,
2349the result returned is the mathematical result modulo
23502<sup>n</sup>, where n is the bit width of the result.</p>
2351<p>Because LLVM integers use a two's complement representation, and the
2352result is the same width as the operands, this instruction returns the
2353correct result for both signed and unsigned integers. If a full product
2354(e.g. <tt>i32</tt>x<tt>i32</tt>-><tt>i64</tt>) is needed, the operands
2355should be sign-extended or zero-extended as appropriate to the
2356width of the full product.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2357<h5>Example:</h5>
2358<pre> &lt;result&gt; = mul i32 4, %var <i>; yields {i32}:result = 4 * %var</i>
2359</pre>
2360</div>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2361
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2362<!-- _______________________________________________________________________ -->
2363<div class="doc_subsubsection"> <a name="i_udiv">'<tt>udiv</tt>' Instruction
2364</a></div>
2365<div class="doc_text">
2366<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2367<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]2368</pre>
2369<h5>Overview:</h5>
2370<p>The '<tt>udiv</tt>' instruction returns the quotient of its two
2371operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2372
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2373<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2374
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2375<p>The two arguments to the '<tt>udiv</tt>' instruction must be
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2376<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2377values. Both arguments must have identical types.</p>
2378
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2379<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2380
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2381<p>The value produced is the unsigned integer quotient of the two operands.</p>
2382<p>Note that unsigned integer division and signed integer division are distinct
2383operations; for signed integer division, use '<tt>sdiv</tt>'.</p>
2384<p>Division by zero leads to undefined behavior.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2385<h5>Example:</h5>
2386<pre> &lt;result&gt; = udiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
2387</pre>
2388</div>
2389<!-- _______________________________________________________________________ -->
2390<div class="doc_subsubsection"> <a name="i_sdiv">'<tt>sdiv</tt>' Instruction
2391</a> </div>
2392<div class="doc_text">
2393<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2394<pre>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2395 &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]2396</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2397
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2398<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2399
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2400<p>The '<tt>sdiv</tt>' instruction returns the quotient of its two
2401operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2402
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2403<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2404
2405<p>The two arguments to the '<tt>sdiv</tt>' instruction must be
2406<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2407values. Both arguments must have identical types.</p>
2408
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2409<h5>Semantics:</h5>
Chris Lattnerdc15b1d2008-04-01 18:45:27 +0000[diff] [blame]2410<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]2411<p>Note that signed integer division and unsigned integer division are distinct
2412operations; for unsigned integer division, use '<tt>udiv</tt>'.</p>
2413<p>Division by zero leads to undefined behavior. Overflow also leads to
2414undefined behavior; this is a rare case, but can occur, for example,
2415by doing a 32-bit division of -2147483648 by -1.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2416<h5>Example:</h5>
2417<pre> &lt;result&gt; = sdiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
2418</pre>
2419</div>
2420<!-- _______________________________________________________________________ -->
2421<div class="doc_subsubsection"> <a name="i_fdiv">'<tt>fdiv</tt>'
2422Instruction</a> </div>
2423<div class="doc_text">
2424<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2425<pre>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2426 &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]2427</pre>
2428<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2429
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2430<p>The '<tt>fdiv</tt>' instruction returns the quotient of its two
2431operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2432
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2433<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2434
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2435<p>The two arguments to the '<tt>fdiv</tt>' instruction must be
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2436<a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a>
2437of floating point values. Both arguments must have identical types.</p>
2438
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2439<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2440
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2441<p>The value produced is the floating point quotient of the two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2442
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2443<h5>Example:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2444
2445<pre>
2446 &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]2447</pre>
2448</div>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2449
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2450<!-- _______________________________________________________________________ -->
2451<div class="doc_subsubsection"> <a name="i_urem">'<tt>urem</tt>' Instruction</a>
2452</div>
2453<div class="doc_text">
2454<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2455<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]2456</pre>
2457<h5>Overview:</h5>
2458<p>The '<tt>urem</tt>' instruction returns the remainder from the
2459unsigned division of its two arguments.</p>
2460<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2461<p>The two arguments to the '<tt>urem</tt>' instruction must be
2462<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2463values. Both arguments must have identical types.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2464<h5>Semantics:</h5>
2465<p>This instruction returns the unsigned integer <i>remainder</i> of a division.
Chris Lattnerdc15b1d2008-04-01 18:45:27 +0000[diff] [blame]2466This instruction always performs an unsigned division to get the remainder.</p>
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2467<p>Note that unsigned integer remainder and signed integer remainder are
2468distinct operations; for signed integer remainder, use '<tt>srem</tt>'.</p>
2469<p>Taking the remainder of a division by zero leads to undefined behavior.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2470<h5>Example:</h5>
2471<pre> &lt;result&gt; = urem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
2472</pre>
2473
2474</div>
2475<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2476<div class="doc_subsubsection">
2477 <a name="i_srem">'<tt>srem</tt>' Instruction</a>
2478</div>
2479
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2480<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2481
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2482<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2483
2484<pre>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2485 &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]2486</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2487
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2488<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2489
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2490<p>The '<tt>srem</tt>' instruction returns the remainder from the
Dan Gohman3e3fd8c2007-11-05 23:35:22 +0000[diff] [blame]2491signed division of its two operands. This instruction can also take
2492<a href="#t_vector">vector</a> versions of the values in which case
2493the elements must be integers.</p>
Chris Lattner08497ce2008-01-04 04:33:49 +0000[diff] [blame]2494
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2495<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2496
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2497<p>The two arguments to the '<tt>srem</tt>' instruction must be
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2498<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2499values. Both arguments must have identical types.</p>
2500
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2501<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2502
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2503<p>This instruction returns the <i>remainder</i> of a division (where the result
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2504has the same sign as the dividend, <tt>op1</tt>), not the <i>modulo</i>
2505operator (where the result has the same sign as the divisor, <tt>op2</tt>) of
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2506a value. For more information about the difference, see <a
2507 href="http://mathforum.org/dr.math/problems/anne.4.28.99.html">The
2508Math Forum</a>. For a table of how this is implemented in various languages,
2509please see <a href="http://en.wikipedia.org/wiki/Modulo_operation">
2510Wikipedia: modulo operation</a>.</p>
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2511<p>Note that signed integer remainder and unsigned integer remainder are
2512distinct operations; for unsigned integer remainder, use '<tt>urem</tt>'.</p>
2513<p>Taking the remainder of a division by zero leads to undefined behavior.
2514Overflow also leads to undefined behavior; this is a rare case, but can occur,
2515for example, by taking the remainder of a 32-bit division of -2147483648 by -1.
2516(The remainder doesn't actually overflow, but this rule lets srem be
2517implemented using instructions that return both the result of the division
2518and the remainder.)</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2519<h5>Example:</h5>
2520<pre> &lt;result&gt; = srem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
2521</pre>
2522
2523</div>
2524<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2525<div class="doc_subsubsection">
2526 <a name="i_frem">'<tt>frem</tt>' Instruction</a> </div>
2527
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2528<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2529
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2530<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2531<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]2532</pre>
2533<h5>Overview:</h5>
2534<p>The '<tt>frem</tt>' instruction returns the remainder from the
2535division of its two operands.</p>
2536<h5>Arguments:</h5>
2537<p>The two arguments to the '<tt>frem</tt>' instruction must be
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2538<a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a>
2539of floating point values. Both arguments must have identical types.</p>
2540
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2541<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2542
Chris Lattnerdc15b1d2008-04-01 18:45:27 +0000[diff] [blame]2543<p>This instruction returns the <i>remainder</i> of a division.
2544The remainder has the same sign as the dividend.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2545
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2546<h5>Example:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2547
2548<pre>
2549 &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]2550</pre>
2551</div>
2552
2553<!-- ======================================================================= -->
2554<div class="doc_subsection"> <a name="bitwiseops">Bitwise Binary
2555Operations</a> </div>
2556<div class="doc_text">
2557<p>Bitwise binary operators are used to do various forms of
2558bit-twiddling in a program. They are generally very efficient
2559instructions and can commonly be strength reduced from other
Chris Lattnerdc15b1d2008-04-01 18:45:27 +0000[diff] [blame]2560instructions. They require two operands of the same type, execute an operation on them,
2561and 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]2562</div>
2563
2564<!-- _______________________________________________________________________ -->
2565<div class="doc_subsubsection"> <a name="i_shl">'<tt>shl</tt>'
2566Instruction</a> </div>
2567<div class="doc_text">
2568<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2569<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]2570</pre>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2571
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2572<h5>Overview:</h5>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2573
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2574<p>The '<tt>shl</tt>' instruction returns the first operand shifted to
2575the left a specified number of bits.</p>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2576
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2577<h5>Arguments:</h5>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2578
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2579<p>Both arguments to the '<tt>shl</tt>' instruction must be the same <a
Nate Begemanbb1ce942008-07-29 15:49:41 +0000[diff] [blame]2580 href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2581type. '<tt>op2</tt>' is treated as an unsigned value.</p>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2582
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2583<h5>Semantics:</h5>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2584
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2585<p>The value produced is <tt>op1</tt> * 2<sup><tt>op2</tt></sup> mod 2<sup>n</sup>,
2586where n is the width of the result. If <tt>op2</tt> is (statically or dynamically) negative or
2587equal to or larger than the number of bits in <tt>op1</tt>, the result is undefined.</p>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2588
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2589<h5>Example:</h5><pre>
2590 &lt;result&gt; = shl i32 4, %var <i>; yields {i32}: 4 &lt;&lt; %var</i>
2591 &lt;result&gt; = shl i32 4, 2 <i>; yields {i32}: 16</i>
2592 &lt;result&gt; = shl i32 1, 10 <i>; yields {i32}: 1024</i>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2593 &lt;result&gt; = shl i32 1, 32 <i>; undefined</i>
Mon P Wang9901e732008-12-09 05:46:39 +0000[diff] [blame^]2594 &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]2595</pre>
2596</div>
2597<!-- _______________________________________________________________________ -->
2598<div class="doc_subsubsection"> <a name="i_lshr">'<tt>lshr</tt>'
2599Instruction</a> </div>
2600<div class="doc_text">
2601<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2602<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]2603</pre>
2604
2605<h5>Overview:</h5>
2606<p>The '<tt>lshr</tt>' instruction (logical shift right) returns the first
2607operand shifted to the right a specified number of bits with zero fill.</p>
2608
2609<h5>Arguments:</h5>
2610<p>Both arguments to the '<tt>lshr</tt>' instruction must be the same
Nate Begemanbb1ce942008-07-29 15:49:41 +0000[diff] [blame]2611<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]2612type. '<tt>op2</tt>' is treated as an unsigned value.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2613
2614<h5>Semantics:</h5>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2615
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2616<p>This instruction always performs a logical shift right operation. The most
2617significant bits of the result will be filled with zero bits after the
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2618shift. If <tt>op2</tt> is (statically or dynamically) equal to or larger than
2619the number of bits in <tt>op1</tt>, the result is undefined.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2620
2621<h5>Example:</h5>
2622<pre>
2623 &lt;result&gt; = lshr i32 4, 1 <i>; yields {i32}:result = 2</i>
2624 &lt;result&gt; = lshr i32 4, 2 <i>; yields {i32}:result = 1</i>
2625 &lt;result&gt; = lshr i8 4, 3 <i>; yields {i8}:result = 0</i>
2626 &lt;result&gt; = lshr i8 -2, 1 <i>; yields {i8}:result = 0x7FFFFFFF </i>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2627 &lt;result&gt; = lshr i32 1, 32 <i>; undefined</i>
Mon P Wang9901e732008-12-09 05:46:39 +0000[diff] [blame^]2628 &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]2629</pre>
2630</div>
2631
2632<!-- _______________________________________________________________________ -->
2633<div class="doc_subsubsection"> <a name="i_ashr">'<tt>ashr</tt>'
2634Instruction</a> </div>
2635<div class="doc_text">
2636
2637<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2638<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]2639</pre>
2640
2641<h5>Overview:</h5>
2642<p>The '<tt>ashr</tt>' instruction (arithmetic shift right) returns the first
2643operand shifted to the right a specified number of bits with sign extension.</p>
2644
2645<h5>Arguments:</h5>
2646<p>Both arguments to the '<tt>ashr</tt>' instruction must be the same
Nate Begemanbb1ce942008-07-29 15:49:41 +0000[diff] [blame]2647<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]2648type. '<tt>op2</tt>' is treated as an unsigned value.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2649
2650<h5>Semantics:</h5>
2651<p>This instruction always performs an arithmetic shift right operation,
2652The most significant bits of the result will be filled with the sign bit
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2653of <tt>op1</tt>. If <tt>op2</tt> is (statically or dynamically) equal to or
2654larger than the number of bits in <tt>op1</tt>, the result is undefined.
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2655</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2656
2657<h5>Example:</h5>
2658<pre>
2659 &lt;result&gt; = ashr i32 4, 1 <i>; yields {i32}:result = 2</i>
2660 &lt;result&gt; = ashr i32 4, 2 <i>; yields {i32}:result = 1</i>
2661 &lt;result&gt; = ashr i8 4, 3 <i>; yields {i8}:result = 0</i>
2662 &lt;result&gt; = ashr i8 -2, 1 <i>; yields {i8}:result = -1</i>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2663 &lt;result&gt; = ashr i32 1, 32 <i>; undefined</i>
Mon P Wang9901e732008-12-09 05:46:39 +0000[diff] [blame^]2664 &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]2665</pre>
2666</div>
2667
2668<!-- _______________________________________________________________________ -->
2669<div class="doc_subsubsection"> <a name="i_and">'<tt>and</tt>'
2670Instruction</a> </div>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2671
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2672<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2673
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2674<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2675
2676<pre>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2677 &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]2678</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2679
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2680<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2681
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2682<p>The '<tt>and</tt>' instruction returns the bitwise logical and of
2683its two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2684
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2685<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2686
2687<p>The two arguments to the '<tt>and</tt>' instruction must be
2688<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2689values. Both arguments must have identical types.</p>
2690
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2691<h5>Semantics:</h5>
2692<p>The truth table used for the '<tt>and</tt>' instruction is:</p>
2693<p> </p>
Bill Wendling1e19eac2008-09-07 10:29:20 +0000[diff] [blame]2694<div>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2695<table border="1" cellspacing="0" cellpadding="4">
2696 <tbody>
2697 <tr>
2698 <td>In0</td>
2699 <td>In1</td>
2700 <td>Out</td>
2701 </tr>
2702 <tr>
2703 <td>0</td>
2704 <td>0</td>
2705 <td>0</td>
2706 </tr>
2707 <tr>
2708 <td>0</td>
2709 <td>1</td>
2710 <td>0</td>
2711 </tr>
2712 <tr>
2713 <td>1</td>
2714 <td>0</td>
2715 <td>0</td>
2716 </tr>
2717 <tr>
2718 <td>1</td>
2719 <td>1</td>
2720 <td>1</td>
2721 </tr>
2722 </tbody>
2723</table>
2724</div>
2725<h5>Example:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2726<pre>
2727 &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]2728 &lt;result&gt; = and i32 15, 40 <i>; yields {i32}:result = 8</i>
2729 &lt;result&gt; = and i32 4, 8 <i>; yields {i32}:result = 0</i>
2730</pre>
2731</div>
2732<!-- _______________________________________________________________________ -->
2733<div class="doc_subsubsection"> <a name="i_or">'<tt>or</tt>' Instruction</a> </div>
2734<div class="doc_text">
2735<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2736<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]2737</pre>
2738<h5>Overview:</h5>
2739<p>The '<tt>or</tt>' instruction returns the bitwise logical inclusive
2740or of its two operands.</p>
2741<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2742
2743<p>The two arguments to the '<tt>or</tt>' instruction must be
2744<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2745values. Both arguments must have identical types.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2746<h5>Semantics:</h5>
2747<p>The truth table used for the '<tt>or</tt>' instruction is:</p>
2748<p> </p>
Bill Wendling1e19eac2008-09-07 10:29:20 +0000[diff] [blame]2749<div>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2750<table border="1" cellspacing="0" cellpadding="4">
2751 <tbody>
2752 <tr>
2753 <td>In0</td>
2754 <td>In1</td>
2755 <td>Out</td>
2756 </tr>
2757 <tr>
2758 <td>0</td>
2759 <td>0</td>
2760 <td>0</td>
2761 </tr>
2762 <tr>
2763 <td>0</td>
2764 <td>1</td>
2765 <td>1</td>
2766 </tr>
2767 <tr>
2768 <td>1</td>
2769 <td>0</td>
2770 <td>1</td>
2771 </tr>
2772 <tr>
2773 <td>1</td>
2774 <td>1</td>
2775 <td>1</td>
2776 </tr>
2777 </tbody>
2778</table>
2779</div>
2780<h5>Example:</h5>
2781<pre> &lt;result&gt; = or i32 4, %var <i>; yields {i32}:result = 4 | %var</i>
2782 &lt;result&gt; = or i32 15, 40 <i>; yields {i32}:result = 47</i>
2783 &lt;result&gt; = or i32 4, 8 <i>; yields {i32}:result = 12</i>
2784</pre>
2785</div>
2786<!-- _______________________________________________________________________ -->
2787<div class="doc_subsubsection"> <a name="i_xor">'<tt>xor</tt>'
2788Instruction</a> </div>
2789<div class="doc_text">
2790<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2791<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]2792</pre>
2793<h5>Overview:</h5>
2794<p>The '<tt>xor</tt>' instruction returns the bitwise logical exclusive
2795or of its two operands. The <tt>xor</tt> is used to implement the
2796"one's complement" operation, which is the "~" operator in C.</p>
2797<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2798<p>The two arguments to the '<tt>xor</tt>' instruction must be
2799<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2800values. Both arguments must have identical types.</p>
2801
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2802<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2803
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2804<p>The truth table used for the '<tt>xor</tt>' instruction is:</p>
2805<p> </p>
Bill Wendling1e19eac2008-09-07 10:29:20 +0000[diff] [blame]2806<div>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2807<table border="1" cellspacing="0" cellpadding="4">
2808 <tbody>
2809 <tr>
2810 <td>In0</td>
2811 <td>In1</td>
2812 <td>Out</td>
2813 </tr>
2814 <tr>
2815 <td>0</td>
2816 <td>0</td>
2817 <td>0</td>
2818 </tr>
2819 <tr>
2820 <td>0</td>
2821 <td>1</td>
2822 <td>1</td>
2823 </tr>
2824 <tr>
2825 <td>1</td>
2826 <td>0</td>
2827 <td>1</td>
2828 </tr>
2829 <tr>
2830 <td>1</td>
2831 <td>1</td>
2832 <td>0</td>
2833 </tr>
2834 </tbody>
2835</table>
2836</div>
2837<p> </p>
2838<h5>Example:</h5>
2839<pre> &lt;result&gt; = xor i32 4, %var <i>; yields {i32}:result = 4 ^ %var</i>
2840 &lt;result&gt; = xor i32 15, 40 <i>; yields {i32}:result = 39</i>
2841 &lt;result&gt; = xor i32 4, 8 <i>; yields {i32}:result = 12</i>
2842 &lt;result&gt; = xor i32 %V, -1 <i>; yields {i32}:result = ~%V</i>
2843</pre>
2844</div>
2845
2846<!-- ======================================================================= -->
2847<div class="doc_subsection">
2848 <a name="vectorops">Vector Operations</a>
2849</div>
2850
2851<div class="doc_text">
2852
2853<p>LLVM supports several instructions to represent vector operations in a
2854target-independent manner. These instructions cover the element-access and
2855vector-specific operations needed to process vectors effectively. While LLVM
2856does directly support these vector operations, many sophisticated algorithms
2857will want to use target-specific intrinsics to take full advantage of a specific
2858target.</p>
2859
2860</div>
2861
2862<!-- _______________________________________________________________________ -->
2863<div class="doc_subsubsection">
2864 <a name="i_extractelement">'<tt>extractelement</tt>' Instruction</a>
2865</div>
2866
2867<div class="doc_text">
2868
2869<h5>Syntax:</h5>
2870
2871<pre>
2872 &lt;result&gt; = extractelement &lt;n x &lt;ty&gt;&gt; &lt;val&gt;, i32 &lt;idx&gt; <i>; yields &lt;ty&gt;</i>
2873</pre>
2874
2875<h5>Overview:</h5>
2876
2877<p>
2878The '<tt>extractelement</tt>' instruction extracts a single scalar
2879element from a vector at a specified index.
2880</p>
2881
2882
2883<h5>Arguments:</h5>
2884
2885<p>
2886The first operand of an '<tt>extractelement</tt>' instruction is a
2887value of <a href="#t_vector">vector</a> type. The second operand is
2888an index indicating the position from which to extract the element.
2889The index may be a variable.</p>
2890
2891<h5>Semantics:</h5>
2892
2893<p>
2894The result is a scalar of the same type as the element type of
2895<tt>val</tt>. Its value is the value at position <tt>idx</tt> of
2896<tt>val</tt>. If <tt>idx</tt> exceeds the length of <tt>val</tt>, the
2897results are undefined.
2898</p>
2899
2900<h5>Example:</h5>
2901
2902<pre>
2903 %result = extractelement &lt;4 x i32&gt; %vec, i32 0 <i>; yields i32</i>
2904</pre>
2905</div>
2906
2907
2908<!-- _______________________________________________________________________ -->
2909<div class="doc_subsubsection">
2910 <a name="i_insertelement">'<tt>insertelement</tt>' Instruction</a>
2911</div>
2912
2913<div class="doc_text">
2914
2915<h5>Syntax:</h5>
2916
2917<pre>
Dan Gohmanbcc3c502008-05-12 23:38:42 +0000[diff] [blame]2918 &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]2919</pre>
2920
2921<h5>Overview:</h5>
2922
2923<p>
2924The '<tt>insertelement</tt>' instruction inserts a scalar
2925element into a vector at a specified index.
2926</p>
2927
2928
2929<h5>Arguments:</h5>
2930
2931<p>
2932The first operand of an '<tt>insertelement</tt>' instruction is a
2933value of <a href="#t_vector">vector</a> type. The second operand is a
2934scalar value whose type must equal the element type of the first
2935operand. The third operand is an index indicating the position at
2936which to insert the value. The index may be a variable.</p>
2937
2938<h5>Semantics:</h5>
2939
2940<p>
2941The result is a vector of the same type as <tt>val</tt>. Its
2942element values are those of <tt>val</tt> except at position
2943<tt>idx</tt>, where it gets the value <tt>elt</tt>. If <tt>idx</tt>
2944exceeds the length of <tt>val</tt>, the results are undefined.
2945</p>
2946
2947<h5>Example:</h5>
2948
2949<pre>
2950 %result = insertelement &lt;4 x i32&gt; %vec, i32 1, i32 0 <i>; yields &lt;4 x i32&gt;</i>
2951</pre>
2952</div>
2953
2954<!-- _______________________________________________________________________ -->
2955<div class="doc_subsubsection">
2956 <a name="i_shufflevector">'<tt>shufflevector</tt>' Instruction</a>
2957</div>
2958
2959<div class="doc_text">
2960
2961<h5>Syntax:</h5>
2962
2963<pre>
Mon P Wangbff5d9c2008-11-10 04:46:22 +0000[diff] [blame]2964 &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]2965</pre>
2966
2967<h5>Overview:</h5>
2968
2969<p>
2970The '<tt>shufflevector</tt>' instruction constructs a permutation of elements
Mon P Wangbff5d9c2008-11-10 04:46:22 +0000[diff] [blame]2971from two input vectors, returning a vector with the same element type as
2972the input and length that is the same as the shuffle mask.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2973</p>
2974
2975<h5>Arguments:</h5>
2976
2977<p>
Mon P Wangbff5d9c2008-11-10 04:46:22 +0000[diff] [blame]2978The first two operands of a '<tt>shufflevector</tt>' instruction are vectors
2979with types that match each other. The third argument is a shuffle mask whose
2980element type is always 'i32'. The result of the instruction is a vector whose
2981length is the same as the shuffle mask and whose element type is the same as
2982the element type of the first two operands.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2983</p>
2984
2985<p>
2986The shuffle mask operand is required to be a constant vector with either
2987constant integer or undef values.
2988</p>
2989
2990<h5>Semantics:</h5>
2991
2992<p>
2993The elements of the two input vectors are numbered from left to right across
2994both of the vectors. The shuffle mask operand specifies, for each element of
Mon P Wangbff5d9c2008-11-10 04:46:22 +0000[diff] [blame]2995the result vector, which element of the two input vectors the result element
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2996gets. The element selector may be undef (meaning "don't care") and the second
2997operand may be undef if performing a shuffle from only one vector.
2998</p>
2999
3000<h5>Example:</h5>
3001
3002<pre>
3003 %result = shufflevector &lt;4 x i32&gt; %v1, &lt;4 x i32&gt; %v2,
3004 &lt;4 x i32&gt; &lt;i32 0, i32 4, i32 1, i32 5&gt; <i>; yields &lt;4 x i32&gt;</i>
3005 %result = shufflevector &lt;4 x i32&gt; %v1, &lt;4 x i32&gt; undef,
3006 &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]3007 %result = shufflevector &lt;8 x i32&gt; %v1, &lt;8 x i32&gt; undef,
3008 &lt;4 x i32&gt; &lt;i32 0, i32 1, i32 2, i32 3&gt; <i>; yields &lt;4 x i32&gt;</i>
3009 %result = shufflevector &lt;4 x i32&gt; %v1, &lt;4 x i32&gt; %v2,
3010 &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]3011</pre>
3012</div>
3013
3014
3015<!-- ======================================================================= -->
3016<div class="doc_subsection">
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3017 <a name="aggregateops">Aggregate Operations</a>
3018</div>
3019
3020<div class="doc_text">
3021
3022<p>LLVM supports several instructions for working with aggregate values.
3023</p>
3024
3025</div>
3026
3027<!-- _______________________________________________________________________ -->
3028<div class="doc_subsubsection">
3029 <a name="i_extractvalue">'<tt>extractvalue</tt>' Instruction</a>
3030</div>
3031
3032<div class="doc_text">
3033
3034<h5>Syntax:</h5>
3035
3036<pre>
3037 &lt;result&gt; = extractvalue &lt;aggregate type&gt; &lt;val&gt;, &lt;idx&gt;{, &lt;idx&gt;}*
3038</pre>
3039
3040<h5>Overview:</h5>
3041
3042<p>
Dan Gohman6c6dea02008-05-13 18:16:06 +0000[diff] [blame]3043The '<tt>extractvalue</tt>' instruction extracts the value of a struct field
3044or array element from an aggregate value.
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3045</p>
3046
3047
3048<h5>Arguments:</h5>
3049
3050<p>
3051The first operand of an '<tt>extractvalue</tt>' instruction is a
3052value of <a href="#t_struct">struct</a> or <a href="#t_array">array</a>
Dan Gohman6c6dea02008-05-13 18:16:06 +0000[diff] [blame]3053type. The operands are constant indices to specify which value to extract
Dan Gohmane5febe42008-05-31 00:58:22 +0000[diff] [blame]3054in a similar manner as indices in a
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3055'<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.
3056</p>
3057
3058<h5>Semantics:</h5>
3059
3060<p>
3061The result is the value at the position in the aggregate specified by
3062the index operands.
3063</p>
3064
3065<h5>Example:</h5>
3066
3067<pre>
Dan Gohmane5febe42008-05-31 00:58:22 +0000[diff] [blame]3068 %result = extractvalue {i32, float} %agg, 0 <i>; yields i32</i>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3069</pre>
3070</div>
3071
3072
3073<!-- _______________________________________________________________________ -->
3074<div class="doc_subsubsection">
3075 <a name="i_insertvalue">'<tt>insertvalue</tt>' Instruction</a>
3076</div>
3077
3078<div class="doc_text">
3079
3080<h5>Syntax:</h5>
3081
3082<pre>
Dan Gohmane5febe42008-05-31 00:58:22 +0000[diff] [blame]3083 &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]3084</pre>
3085
3086<h5>Overview:</h5>
3087
3088<p>
3089The '<tt>insertvalue</tt>' instruction inserts a value
Dan Gohman6c6dea02008-05-13 18:16:06 +0000[diff] [blame]3090into a struct field or array element in an aggregate.
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3091</p>
3092
3093
3094<h5>Arguments:</h5>
3095
3096<p>
3097The first operand of an '<tt>insertvalue</tt>' instruction is a
3098value of <a href="#t_struct">struct</a> or <a href="#t_array">array</a> type.
3099The second operand is a first-class value to insert.
Dan Gohman4f29e422008-05-23 21:53:15 +0000[diff] [blame]3100The following operands are constant indices
Dan Gohmane5febe42008-05-31 00:58:22 +0000[diff] [blame]3101indicating the position at which to insert the value in a similar manner as
Dan Gohman6c6dea02008-05-13 18:16:06 +0000[diff] [blame]3102indices in a
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3103'<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.
3104The value to insert must have the same type as the value identified
Dan Gohman6c6dea02008-05-13 18:16:06 +0000[diff] [blame]3105by the indices.
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]3106</p>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3107
3108<h5>Semantics:</h5>
3109
3110<p>
3111The result is an aggregate of the same type as <tt>val</tt>. Its
3112value is that of <tt>val</tt> except that the value at the position
Dan Gohman6c6dea02008-05-13 18:16:06 +0000[diff] [blame]3113specified by the indices is that of <tt>elt</tt>.
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3114</p>
3115
3116<h5>Example:</h5>
3117
3118<pre>
Dan Gohmanb1aab4e2008-06-23 15:26:37 +0000[diff] [blame]3119 %result = insertvalue {i32, float} %agg, i32 1, 0 <i>; yields {i32, float}</i>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3120</pre>
3121</div>
3122
3123
3124<!-- ======================================================================= -->
3125<div class="doc_subsection">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3126 <a name="memoryops">Memory Access and Addressing Operations</a>
3127</div>
3128
3129<div class="doc_text">
3130
3131<p>A key design point of an SSA-based representation is how it
3132represents memory. In LLVM, no memory locations are in SSA form, which
3133makes things very simple. This section describes how to read, write,
3134allocate, and free memory in LLVM.</p>
3135
3136</div>
3137
3138<!-- _______________________________________________________________________ -->
3139<div class="doc_subsubsection">
3140 <a name="i_malloc">'<tt>malloc</tt>' Instruction</a>
3141</div>
3142
3143<div class="doc_text">
3144
3145<h5>Syntax:</h5>
3146
3147<pre>
3148 &lt;result&gt; = malloc &lt;type&gt;[, i32 &lt;NumElements&gt;][, align &lt;alignment&gt;] <i>; yields {type*}:result</i>
3149</pre>
3150
3151<h5>Overview:</h5>
3152
3153<p>The '<tt>malloc</tt>' instruction allocates memory from the system
Christopher Lambcfe00962007-12-17 01:00:21 +0000[diff] [blame]3154heap and returns a pointer to it. The object is always allocated in the generic
3155address space (address space zero).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3156
3157<h5>Arguments:</h5>
3158
3159<p>The '<tt>malloc</tt>' instruction allocates
3160<tt>sizeof(&lt;type&gt;)*NumElements</tt>
3161bytes of memory from the operating system and returns a pointer of the
3162appropriate type to the program. If "NumElements" is specified, it is the
Gabor Greif5082cf42008-02-09 22:24:34 +0000[diff] [blame]3163number of elements allocated, otherwise "NumElements" is defaulted to be one.
Chris Lattner10368b62008-04-02 00:38:26 +0000[diff] [blame]3164If a constant alignment is specified, the value result of the allocation is guaranteed to
Gabor Greif5082cf42008-02-09 22:24:34 +0000[diff] [blame]3165be aligned to at least that boundary. If not specified, or if zero, the target can
3166choose to align the allocation on any convenient boundary.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3167
3168<p>'<tt>type</tt>' must be a sized type.</p>
3169
3170<h5>Semantics:</h5>
3171
3172<p>Memory is allocated using the system "<tt>malloc</tt>" function, and
Nick Lewyckyff384472008-11-24 03:41:24 +0000[diff] [blame]3173a pointer is returned. The result of a zero byte allocation is undefined. The
Chris Lattner8b094fc2008-04-19 21:01:16 +0000[diff] [blame]3174result is null if there is insufficient memory available.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3175
3176<h5>Example:</h5>
3177
3178<pre>
3179 %array = malloc [4 x i8 ] <i>; yields {[%4 x i8]*}:array</i>
3180
3181 %size = <a href="#i_add">add</a> i32 2, 2 <i>; yields {i32}:size = i32 4</i>
3182 %array1 = malloc i8, i32 4 <i>; yields {i8*}:array1</i>
3183 %array2 = malloc [12 x i8], i32 %size <i>; yields {[12 x i8]*}:array2</i>
3184 %array3 = malloc i32, i32 4, align 1024 <i>; yields {i32*}:array3</i>
3185 %array4 = malloc i32, align 1024 <i>; yields {i32*}:array4</i>
3186</pre>
3187</div>
3188
3189<!-- _______________________________________________________________________ -->
3190<div class="doc_subsubsection">
3191 <a name="i_free">'<tt>free</tt>' Instruction</a>
3192</div>
3193
3194<div class="doc_text">
3195
3196<h5>Syntax:</h5>
3197
3198<pre>
3199 free &lt;type&gt; &lt;value&gt; <i>; yields {void}</i>
3200</pre>
3201
3202<h5>Overview:</h5>
3203
3204<p>The '<tt>free</tt>' instruction returns memory back to the unused
3205memory heap to be reallocated in the future.</p>
3206
3207<h5>Arguments:</h5>
3208
3209<p>'<tt>value</tt>' shall be a pointer value that points to a value
3210that was allocated with the '<tt><a href="#i_malloc">malloc</a></tt>'
3211instruction.</p>
3212
3213<h5>Semantics:</h5>
3214
3215<p>Access to the memory pointed to by the pointer is no longer defined
Chris Lattner329d6532008-04-19 22:41:32 +0000[diff] [blame]3216after this instruction executes. If the pointer is null, the operation
3217is a noop.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3218
3219<h5>Example:</h5>
3220
3221<pre>
3222 %array = <a href="#i_malloc">malloc</a> [4 x i8] <i>; yields {[4 x i8]*}:array</i>
3223 free [4 x i8]* %array
3224</pre>
3225</div>
3226
3227<!-- _______________________________________________________________________ -->
3228<div class="doc_subsubsection">
3229 <a name="i_alloca">'<tt>alloca</tt>' Instruction</a>
3230</div>
3231
3232<div class="doc_text">
3233
3234<h5>Syntax:</h5>
3235
3236<pre>
3237 &lt;result&gt; = alloca &lt;type&gt;[, i32 &lt;NumElements&gt;][, align &lt;alignment&gt;] <i>; yields {type*}:result</i>
3238</pre>
3239
3240<h5>Overview:</h5>
3241
3242<p>The '<tt>alloca</tt>' instruction allocates memory on the stack frame of the
3243currently executing function, to be automatically released when this function
Christopher Lambcfe00962007-12-17 01:00:21 +0000[diff] [blame]3244returns to its caller. The object is always allocated in the generic address
3245space (address space zero).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3246
3247<h5>Arguments:</h5>
3248
3249<p>The '<tt>alloca</tt>' instruction allocates <tt>sizeof(&lt;type&gt;)*NumElements</tt>
3250bytes of memory on the runtime stack, returning a pointer of the
Gabor Greif5082cf42008-02-09 22:24:34 +0000[diff] [blame]3251appropriate type to the program. If "NumElements" is specified, it is the
3252number of elements allocated, otherwise "NumElements" is defaulted to be one.
Chris Lattner10368b62008-04-02 00:38:26 +0000[diff] [blame]3253If a constant alignment is specified, the value result of the allocation is guaranteed
Gabor Greif5082cf42008-02-09 22:24:34 +0000[diff] [blame]3254to be aligned to at least that boundary. If not specified, or if zero, the target
3255can choose to align the allocation on any convenient boundary.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3256
3257<p>'<tt>type</tt>' may be any sized type.</p>
3258
3259<h5>Semantics:</h5>
3260
Chris Lattner8b094fc2008-04-19 21:01:16 +0000[diff] [blame]3261<p>Memory is allocated; a pointer is returned. The operation is undefiend if
3262there is insufficient stack space for the allocation. '<tt>alloca</tt>'d
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3263memory is automatically released when the function returns. The '<tt>alloca</tt>'
3264instruction is commonly used to represent automatic variables that must
3265have an address available. When the function returns (either with the <tt><a
3266 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]3267instructions), the memory is reclaimed. Allocating zero bytes
3268is legal, but the result is undefined.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3269
3270<h5>Example:</h5>
3271
3272<pre>
3273 %ptr = alloca i32 <i>; yields {i32*}:ptr</i>
3274 %ptr = alloca i32, i32 4 <i>; yields {i32*}:ptr</i>
3275 %ptr = alloca i32, i32 4, align 1024 <i>; yields {i32*}:ptr</i>
3276 %ptr = alloca i32, align 1024 <i>; yields {i32*}:ptr</i>
3277</pre>
3278</div>
3279
3280<!-- _______________________________________________________________________ -->
3281<div class="doc_subsubsection"> <a name="i_load">'<tt>load</tt>'
3282Instruction</a> </div>
3283<div class="doc_text">
3284<h5>Syntax:</h5>
3285<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>
3286<h5>Overview:</h5>
3287<p>The '<tt>load</tt>' instruction is used to read from memory.</p>
3288<h5>Arguments:</h5>
3289<p>The argument to the '<tt>load</tt>' instruction specifies the memory
3290address from which to load. The pointer must point to a <a
3291 href="#t_firstclass">first class</a> type. If the <tt>load</tt> is
3292marked as <tt>volatile</tt>, then the optimizer is not allowed to modify
3293the number or order of execution of this <tt>load</tt> with other
3294volatile <tt>load</tt> and <tt><a href="#i_store">store</a></tt>
3295instructions. </p>
Chris Lattner21003c82008-01-06 21:04:43 +0000[diff] [blame]3296<p>
Chris Lattner10368b62008-04-02 00:38:26 +0000[diff] [blame]3297The optional constant "align" argument specifies the alignment of the operation
Chris Lattner21003c82008-01-06 21:04:43 +0000[diff] [blame]3298(that is, the alignment of the memory address). A value of 0 or an
3299omitted "align" argument means that the operation has the preferential
3300alignment for the target. It is the responsibility of the code emitter
3301to ensure that the alignment information is correct. Overestimating
3302the alignment results in an undefined behavior. Underestimating the
3303alignment may produce less efficient code. An alignment of 1 is always
3304safe.
3305</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3306<h5>Semantics:</h5>
3307<p>The location of memory pointed to is loaded.</p>
3308<h5>Examples:</h5>
3309<pre> %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
3310 <a
3311 href="#i_store">store</a> i32 3, i32* %ptr <i>; yields {void}</i>
3312 %val = load i32* %ptr <i>; yields {i32}:val = i32 3</i>
3313</pre>
3314</div>
3315<!-- _______________________________________________________________________ -->
3316<div class="doc_subsubsection"> <a name="i_store">'<tt>store</tt>'
3317Instruction</a> </div>
3318<div class="doc_text">
3319<h5>Syntax:</h5>
3320<pre> store &lt;ty&gt; &lt;value&gt;, &lt;ty&gt;* &lt;pointer&gt;[, align &lt;alignment&gt;] <i>; yields {void}</i>
3321 volatile store &lt;ty&gt; &lt;value&gt;, &lt;ty&gt;* &lt;pointer&gt;[, align &lt;alignment&gt;] <i>; yields {void}</i>
3322</pre>
3323<h5>Overview:</h5>
3324<p>The '<tt>store</tt>' instruction is used to write to memory.</p>
3325<h5>Arguments:</h5>
3326<p>There are two arguments to the '<tt>store</tt>' instruction: a value
3327to 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]3328operand must be a pointer to the <a href="#t_firstclass">first class</a> type
3329of the '<tt>&lt;value&gt;</tt>'
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3330operand. If the <tt>store</tt> is marked as <tt>volatile</tt>, then the
3331optimizer is not allowed to modify the number or order of execution of
3332this <tt>store</tt> with other volatile <tt>load</tt> and <tt><a
3333 href="#i_store">store</a></tt> instructions.</p>
Chris Lattner21003c82008-01-06 21:04:43 +0000[diff] [blame]3334<p>
Chris Lattner10368b62008-04-02 00:38:26 +0000[diff] [blame]3335The optional constant "align" argument specifies the alignment of the operation
Chris Lattner21003c82008-01-06 21:04:43 +0000[diff] [blame]3336(that is, the alignment of the memory address). A value of 0 or an
3337omitted "align" argument means that the operation has the preferential
3338alignment for the target. It is the responsibility of the code emitter
3339to ensure that the alignment information is correct. Overestimating
3340the alignment results in an undefined behavior. Underestimating the
3341alignment may produce less efficient code. An alignment of 1 is always
3342safe.
3343</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3344<h5>Semantics:</h5>
3345<p>The contents of memory are updated to contain '<tt>&lt;value&gt;</tt>'
3346at the location specified by the '<tt>&lt;pointer&gt;</tt>' operand.</p>
3347<h5>Example:</h5>
3348<pre> %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
Bill Wendling63ffa142007-10-22 05:10:05 +0000[diff] [blame]3349 store i32 3, i32* %ptr <i>; yields {void}</i>
3350 %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]3351</pre>
3352</div>
3353
3354<!-- _______________________________________________________________________ -->
3355<div class="doc_subsubsection">
3356 <a name="i_getelementptr">'<tt>getelementptr</tt>' Instruction</a>
3357</div>
3358
3359<div class="doc_text">
3360<h5>Syntax:</h5>
3361<pre>
Matthijs Kooijman32a080e2008-10-13 13:44:15 +0000[diff] [blame]3362 &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]3363</pre>
3364
3365<h5>Overview:</h5>
3366
3367<p>
3368The '<tt>getelementptr</tt>' instruction is used to get the address of a
Matthijs Kooijman32a080e2008-10-13 13:44:15 +0000[diff] [blame]3369subelement of an aggregate data structure. It performs address calculation only
3370and does not access memory.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3371
3372<h5>Arguments:</h5>
3373
Matthijs Kooijman32a080e2008-10-13 13:44:15 +0000[diff] [blame]3374<p>The first argument is always a pointer, and forms the basis of the
3375calculation. The remaining arguments are indices, that indicate which of the
3376elements of the aggregate object are indexed. The interpretation of each index
3377is dependent on the type being indexed into. The first index always indexes the
3378pointer value given as the first argument, the second index indexes a value of
3379the type pointed to (not necessarily the value directly pointed to, since the
3380first index can be non-zero), etc. The first type indexed into must be a pointer
3381value, subsequent types can be arrays, vectors and structs. Note that subsequent
3382types being indexed into can never be pointers, since that would require loading
3383the pointer before continuing calculation.</p>
3384
3385<p>The type of each index argument depends on the type it is indexing into.
3386When indexing into a (packed) structure, only <tt>i32</tt> integer
3387<b>constants</b> are allowed. When indexing into an array, pointer or vector,
3388only integers of 32 or 64 bits are allowed (also non-constants). 32-bit values
3389will be sign extended to 64-bits if required.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3390
3391<p>For example, let's consider a C code fragment and how it gets
3392compiled to LLVM:</p>
3393
3394<div class="doc_code">
3395<pre>
3396struct RT {
3397 char A;
3398 int B[10][20];
3399 char C;
3400};
3401struct ST {
3402 int X;
3403 double Y;
3404 struct RT Z;
3405};
3406
3407int *foo(struct ST *s) {
3408 return &amp;s[1].Z.B[5][13];
3409}
3410</pre>
3411</div>
3412
3413<p>The LLVM code generated by the GCC frontend is:</p>
3414
3415<div class="doc_code">
3416<pre>
3417%RT = type { i8 , [10 x [20 x i32]], i8 }
3418%ST = type { i32, double, %RT }
3419
3420define i32* %foo(%ST* %s) {
3421entry:
3422 %reg = getelementptr %ST* %s, i32 1, i32 2, i32 1, i32 5, i32 13
3423 ret i32* %reg
3424}
3425</pre>
3426</div>
3427
3428<h5>Semantics:</h5>
3429
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3430<p>In the example above, the first index is indexing into the '<tt>%ST*</tt>'
3431type, which is a pointer, yielding a '<tt>%ST</tt>' = '<tt>{ i32, double, %RT
3432}</tt>' type, a structure. The second index indexes into the third element of
3433the structure, yielding a '<tt>%RT</tt>' = '<tt>{ i8 , [10 x [20 x i32]],
3434i8 }</tt>' type, another structure. The third index indexes into the second
3435element of the structure, yielding a '<tt>[10 x [20 x i32]]</tt>' type, an
3436array. The two dimensions of the array are subscripted into, yielding an
3437'<tt>i32</tt>' type. The '<tt>getelementptr</tt>' instruction returns a pointer
3438to this element, thus computing a value of '<tt>i32*</tt>' type.</p>
3439
3440<p>Note that it is perfectly legal to index partially through a
3441structure, returning a pointer to an inner element. Because of this,
3442the LLVM code for the given testcase is equivalent to:</p>
3443
3444<pre>
3445 define i32* %foo(%ST* %s) {
3446 %t1 = getelementptr %ST* %s, i32 1 <i>; yields %ST*:%t1</i>
3447 %t2 = getelementptr %ST* %t1, i32 0, i32 2 <i>; yields %RT*:%t2</i>
3448 %t3 = getelementptr %RT* %t2, i32 0, i32 1 <i>; yields [10 x [20 x i32]]*:%t3</i>
3449 %t4 = getelementptr [10 x [20 x i32]]* %t3, i32 0, i32 5 <i>; yields [20 x i32]*:%t4</i>
3450 %t5 = getelementptr [20 x i32]* %t4, i32 0, i32 13 <i>; yields i32*:%t5</i>
3451 ret i32* %t5
3452 }
3453</pre>
3454
3455<p>Note that it is undefined to access an array out of bounds: array and
3456pointer indexes must always be within the defined bounds of the array type.
Chris Lattnera7d94ba2008-04-24 05:59:56 +0000[diff] [blame]3457The one exception for this rule is zero length arrays. These arrays are
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3458defined to be accessible as variable length arrays, which requires access
3459beyond the zero'th element.</p>
3460
3461<p>The getelementptr instruction is often confusing. For some more insight
3462into how it works, see <a href="GetElementPtr.html">the getelementptr
3463FAQ</a>.</p>
3464
3465<h5>Example:</h5>
3466
3467<pre>
3468 <i>; yields [12 x i8]*:aptr</i>
Matthijs Kooijman32a080e2008-10-13 13:44:15 +0000[diff] [blame]3469 %aptr = getelementptr {i32, [12 x i8]}* %saptr, i64 0, i32 1
3470 <i>; yields i8*:vptr</i>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]3471 %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]3472 <i>; yields i8*:eptr</i>
3473 %eptr = getelementptr [12 x i8]* %aptr, i64 0, i32 1
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3474</pre>
3475</div>
3476
3477<!-- ======================================================================= -->
3478<div class="doc_subsection"> <a name="convertops">Conversion Operations</a>
3479</div>
3480<div class="doc_text">
3481<p>The instructions in this category are the conversion instructions (casting)
3482which all take a single operand and a type. They perform various bit conversions
3483on the operand.</p>
3484</div>
3485
3486<!-- _______________________________________________________________________ -->
3487<div class="doc_subsubsection">
3488 <a name="i_trunc">'<tt>trunc .. to</tt>' Instruction</a>
3489</div>
3490<div class="doc_text">
3491
3492<h5>Syntax:</h5>
3493<pre>
3494 &lt;result&gt; = trunc &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3495</pre>
3496
3497<h5>Overview:</h5>
3498<p>
3499The '<tt>trunc</tt>' instruction truncates its operand to the type <tt>ty2</tt>.
3500</p>
3501
3502<h5>Arguments:</h5>
3503<p>
3504The '<tt>trunc</tt>' instruction takes a <tt>value</tt> to trunc, which must
3505be an <a href="#t_integer">integer</a> type, and a type that specifies the size
3506and type of the result, which must be an <a href="#t_integer">integer</a>
3507type. The bit size of <tt>value</tt> must be larger than the bit size of
3508<tt>ty2</tt>. Equal sized types are not allowed.</p>
3509
3510<h5>Semantics:</h5>
3511<p>
3512The '<tt>trunc</tt>' instruction truncates the high order bits in <tt>value</tt>
3513and converts the remaining bits to <tt>ty2</tt>. Since the source size must be
3514larger than the destination size, <tt>trunc</tt> cannot be a <i>no-op cast</i>.
3515It will always truncate bits.</p>
3516
3517<h5>Example:</h5>
3518<pre>
3519 %X = trunc i32 257 to i8 <i>; yields i8:1</i>
3520 %Y = trunc i32 123 to i1 <i>; yields i1:true</i>
3521 %Y = trunc i32 122 to i1 <i>; yields i1:false</i>
3522</pre>
3523</div>
3524
3525<!-- _______________________________________________________________________ -->
3526<div class="doc_subsubsection">
3527 <a name="i_zext">'<tt>zext .. to</tt>' Instruction</a>
3528</div>
3529<div class="doc_text">
3530
3531<h5>Syntax:</h5>
3532<pre>
3533 &lt;result&gt; = zext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3534</pre>
3535
3536<h5>Overview:</h5>
3537<p>The '<tt>zext</tt>' instruction zero extends its operand to type
3538<tt>ty2</tt>.</p>
3539
3540
3541<h5>Arguments:</h5>
3542<p>The '<tt>zext</tt>' instruction takes a value to cast, which must be of
3543<a href="#t_integer">integer</a> type, and a type to cast it to, which must
3544also be of <a href="#t_integer">integer</a> type. The bit size of the
3545<tt>value</tt> must be smaller than the bit size of the destination type,
3546<tt>ty2</tt>.</p>
3547
3548<h5>Semantics:</h5>
3549<p>The <tt>zext</tt> fills the high order bits of the <tt>value</tt> with zero
3550bits until it reaches the size of the destination type, <tt>ty2</tt>.</p>
3551
3552<p>When zero extending from i1, the result will always be either 0 or 1.</p>
3553
3554<h5>Example:</h5>
3555<pre>
3556 %X = zext i32 257 to i64 <i>; yields i64:257</i>
3557 %Y = zext i1 true to i32 <i>; yields i32:1</i>
3558</pre>
3559</div>
3560
3561<!-- _______________________________________________________________________ -->
3562<div class="doc_subsubsection">
3563 <a name="i_sext">'<tt>sext .. to</tt>' Instruction</a>
3564</div>
3565<div class="doc_text">
3566
3567<h5>Syntax:</h5>
3568<pre>
3569 &lt;result&gt; = sext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3570</pre>
3571
3572<h5>Overview:</h5>
3573<p>The '<tt>sext</tt>' sign extends <tt>value</tt> to the type <tt>ty2</tt>.</p>
3574
3575<h5>Arguments:</h5>
3576<p>
3577The '<tt>sext</tt>' instruction takes a value to cast, which must be of
3578<a href="#t_integer">integer</a> type, and a type to cast it to, which must
3579also be of <a href="#t_integer">integer</a> type. The bit size of the
3580<tt>value</tt> must be smaller than the bit size of the destination type,
3581<tt>ty2</tt>.</p>
3582
3583<h5>Semantics:</h5>
3584<p>
3585The '<tt>sext</tt>' instruction performs a sign extension by copying the sign
3586bit (highest order bit) of the <tt>value</tt> until it reaches the bit size of
3587the type <tt>ty2</tt>.</p>
3588
3589<p>When sign extending from i1, the extension always results in -1 or 0.</p>
3590
3591<h5>Example:</h5>
3592<pre>
3593 %X = sext i8 -1 to i16 <i>; yields i16 :65535</i>
3594 %Y = sext i1 true to i32 <i>; yields i32:-1</i>
3595</pre>
3596</div>
3597
3598<!-- _______________________________________________________________________ -->
3599<div class="doc_subsubsection">
3600 <a name="i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a>
3601</div>
3602
3603<div class="doc_text">
3604
3605<h5>Syntax:</h5>
3606
3607<pre>
3608 &lt;result&gt; = fptrunc &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3609</pre>
3610
3611<h5>Overview:</h5>
3612<p>The '<tt>fptrunc</tt>' instruction truncates <tt>value</tt> to type
3613<tt>ty2</tt>.</p>
3614
3615
3616<h5>Arguments:</h5>
3617<p>The '<tt>fptrunc</tt>' instruction takes a <a href="#t_floating">floating
3618 point</a> value to cast and a <a href="#t_floating">floating point</a> type to
3619cast it to. The size of <tt>value</tt> must be larger than the size of
3620<tt>ty2</tt>. This implies that <tt>fptrunc</tt> cannot be used to make a
3621<i>no-op cast</i>.</p>
3622
3623<h5>Semantics:</h5>
3624<p> The '<tt>fptrunc</tt>' instruction truncates a <tt>value</tt> from a larger
3625<a href="#t_floating">floating point</a> type to a smaller
3626<a href="#t_floating">floating point</a> type. If the value cannot fit within
3627the destination type, <tt>ty2</tt>, then the results are undefined.</p>
3628
3629<h5>Example:</h5>
3630<pre>
3631 %X = fptrunc double 123.0 to float <i>; yields float:123.0</i>
3632 %Y = fptrunc double 1.0E+300 to float <i>; yields undefined</i>
3633</pre>
3634</div>
3635
3636<!-- _______________________________________________________________________ -->
3637<div class="doc_subsubsection">
3638 <a name="i_fpext">'<tt>fpext .. to</tt>' Instruction</a>
3639</div>
3640<div class="doc_text">
3641
3642<h5>Syntax:</h5>
3643<pre>
3644 &lt;result&gt; = fpext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3645</pre>
3646
3647<h5>Overview:</h5>
3648<p>The '<tt>fpext</tt>' extends a floating point <tt>value</tt> to a larger
3649floating point value.</p>
3650
3651<h5>Arguments:</h5>
3652<p>The '<tt>fpext</tt>' instruction takes a
3653<a href="#t_floating">floating point</a> <tt>value</tt> to cast,
3654and a <a href="#t_floating">floating point</a> type to cast it to. The source
3655type must be smaller than the destination type.</p>
3656
3657<h5>Semantics:</h5>
3658<p>The '<tt>fpext</tt>' instruction extends the <tt>value</tt> from a smaller
3659<a href="#t_floating">floating point</a> type to a larger
3660<a href="#t_floating">floating point</a> type. The <tt>fpext</tt> cannot be
3661used to make a <i>no-op cast</i> because it always changes bits. Use
3662<tt>bitcast</tt> to make a <i>no-op cast</i> for a floating point cast.</p>
3663
3664<h5>Example:</h5>
3665<pre>
3666 %X = fpext float 3.1415 to double <i>; yields double:3.1415</i>
3667 %Y = fpext float 1.0 to float <i>; yields float:1.0 (no-op)</i>
3668</pre>
3669</div>
3670
3671<!-- _______________________________________________________________________ -->
3672<div class="doc_subsubsection">
3673 <a name="i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a>
3674</div>
3675<div class="doc_text">
3676
3677<h5>Syntax:</h5>
3678<pre>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3679 &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]3680</pre>
3681
3682<h5>Overview:</h5>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3683<p>The '<tt>fptoui</tt>' converts a floating point <tt>value</tt> to its
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3684unsigned integer equivalent of type <tt>ty2</tt>.
3685</p>
3686
3687<h5>Arguments:</h5>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3688<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]3689scalar or vector <a href="#t_floating">floating point</a> value, and a type
3690to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
3691type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
3692vector integer type with the same number of elements as <tt>ty</tt></p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3693
3694<h5>Semantics:</h5>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3695<p> The '<tt>fptoui</tt>' instruction converts its
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3696<a href="#t_floating">floating point</a> operand into the nearest (rounding
3697towards zero) unsigned integer value. If the value cannot fit in <tt>ty2</tt>,
3698the results are undefined.</p>
3699
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3700<h5>Example:</h5>
3701<pre>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3702 %X = fptoui double 123.0 to i32 <i>; yields i32:123</i>
Chris Lattner681f1e82007-09-22 03:17:52 +0000[diff] [blame]3703 %Y = fptoui float 1.0E+300 to i1 <i>; yields undefined:1</i>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3704 %X = fptoui float 1.04E+17 to i8 <i>; yields undefined:1</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3705</pre>
3706</div>
3707
3708<!-- _______________________________________________________________________ -->
3709<div class="doc_subsubsection">
3710 <a name="i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a>
3711</div>
3712<div class="doc_text">
3713
3714<h5>Syntax:</h5>
3715<pre>
3716 &lt;result&gt; = fptosi &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3717</pre>
3718
3719<h5>Overview:</h5>
3720<p>The '<tt>fptosi</tt>' instruction converts
3721<a href="#t_floating">floating point</a> <tt>value</tt> to type <tt>ty2</tt>.
3722</p>
3723
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3724<h5>Arguments:</h5>
3725<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]3726scalar or vector <a href="#t_floating">floating point</a> value, and a type
3727to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
3728type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
3729vector integer type with the same number of elements as <tt>ty</tt></p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3730
3731<h5>Semantics:</h5>
3732<p>The '<tt>fptosi</tt>' instruction converts its
3733<a href="#t_floating">floating point</a> operand into the nearest (rounding
3734towards zero) signed integer value. If the value cannot fit in <tt>ty2</tt>,
3735the results are undefined.</p>
3736
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3737<h5>Example:</h5>
3738<pre>
3739 %X = fptosi double -123.0 to i32 <i>; yields i32:-123</i>
Chris Lattner681f1e82007-09-22 03:17:52 +0000[diff] [blame]3740 %Y = fptosi float 1.0E-247 to i1 <i>; yields undefined:1</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3741 %X = fptosi float 1.04E+17 to i8 <i>; yields undefined:1</i>
3742</pre>
3743</div>
3744
3745<!-- _______________________________________________________________________ -->
3746<div class="doc_subsubsection">
3747 <a name="i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a>
3748</div>
3749<div class="doc_text">
3750
3751<h5>Syntax:</h5>
3752<pre>
3753 &lt;result&gt; = uitofp &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3754</pre>
3755
3756<h5>Overview:</h5>
3757<p>The '<tt>uitofp</tt>' instruction regards <tt>value</tt> as an unsigned
3758integer and converts that value to the <tt>ty2</tt> type.</p>
3759
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3760<h5>Arguments:</h5>
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]3761<p>The '<tt>uitofp</tt>' instruction takes a value to cast, which must be a
3762scalar or vector <a href="#t_integer">integer</a> value, and a type to cast it
3763to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
3764type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
3765floating point type with the same number of elements as <tt>ty</tt></p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3766
3767<h5>Semantics:</h5>
3768<p>The '<tt>uitofp</tt>' instruction interprets its operand as an unsigned
3769integer quantity and converts it to the corresponding floating point value. If
3770the value cannot fit in the floating point value, the results are undefined.</p>
3771
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3772<h5>Example:</h5>
3773<pre>
3774 %X = uitofp i32 257 to float <i>; yields float:257.0</i>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]3775 %Y = uitofp i8 -1 to double <i>; yields double:255.0</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3776</pre>
3777</div>
3778
3779<!-- _______________________________________________________________________ -->
3780<div class="doc_subsubsection">
3781 <a name="i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a>
3782</div>
3783<div class="doc_text">
3784
3785<h5>Syntax:</h5>
3786<pre>
3787 &lt;result&gt; = sitofp &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3788</pre>
3789
3790<h5>Overview:</h5>
3791<p>The '<tt>sitofp</tt>' instruction regards <tt>value</tt> as a signed
3792integer and converts that value to the <tt>ty2</tt> type.</p>
3793
3794<h5>Arguments:</h5>
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]3795<p>The '<tt>sitofp</tt>' instruction takes a value to cast, which must be a
3796scalar or vector <a href="#t_integer">integer</a> value, and a type to cast it
3797to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
3798type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
3799floating point type with the same number of elements as <tt>ty</tt></p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3800
3801<h5>Semantics:</h5>
3802<p>The '<tt>sitofp</tt>' instruction interprets its operand as a signed
3803integer quantity and converts it to the corresponding floating point value. If
3804the value cannot fit in the floating point value, the results are undefined.</p>
3805
3806<h5>Example:</h5>
3807<pre>
3808 %X = sitofp i32 257 to float <i>; yields float:257.0</i>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]3809 %Y = sitofp i8 -1 to double <i>; yields double:-1.0</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3810</pre>
3811</div>
3812
3813<!-- _______________________________________________________________________ -->
3814<div class="doc_subsubsection">
3815 <a name="i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a>
3816</div>
3817<div class="doc_text">
3818
3819<h5>Syntax:</h5>
3820<pre>
3821 &lt;result&gt; = ptrtoint &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3822</pre>
3823
3824<h5>Overview:</h5>
3825<p>The '<tt>ptrtoint</tt>' instruction converts the pointer <tt>value</tt> to
3826the integer type <tt>ty2</tt>.</p>
3827
3828<h5>Arguments:</h5>
3829<p>The '<tt>ptrtoint</tt>' instruction takes a <tt>value</tt> to cast, which
3830must 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]3831<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]3832
3833<h5>Semantics:</h5>
3834<p>The '<tt>ptrtoint</tt>' instruction converts <tt>value</tt> to integer type
3835<tt>ty2</tt> by interpreting the pointer value as an integer and either
3836truncating or zero extending that value to the size of the integer type. If
3837<tt>value</tt> is smaller than <tt>ty2</tt> then a zero extension is done. If
3838<tt>value</tt> is larger than <tt>ty2</tt> then a truncation is done. If they
3839are the same size, then nothing is done (<i>no-op cast</i>) other than a type
3840change.</p>
3841
3842<h5>Example:</h5>
3843<pre>
3844 %X = ptrtoint i32* %X to i8 <i>; yields truncation on 32-bit architecture</i>
3845 %Y = ptrtoint i32* %x to i64 <i>; yields zero extension on 32-bit architecture</i>
3846</pre>
3847</div>
3848
3849<!-- _______________________________________________________________________ -->
3850<div class="doc_subsubsection">
3851 <a name="i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a>
3852</div>
3853<div class="doc_text">
3854
3855<h5>Syntax:</h5>
3856<pre>
3857 &lt;result&gt; = inttoptr &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3858</pre>
3859
3860<h5>Overview:</h5>
3861<p>The '<tt>inttoptr</tt>' instruction converts an integer <tt>value</tt> to
3862a pointer type, <tt>ty2</tt>.</p>
3863
3864<h5>Arguments:</h5>
3865<p>The '<tt>inttoptr</tt>' instruction takes an <a href="#t_integer">integer</a>
3866value to cast, and a type to cast it to, which must be a
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]3867<a href="#t_pointer">pointer</a> type.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3868
3869<h5>Semantics:</h5>
3870<p>The '<tt>inttoptr</tt>' instruction converts <tt>value</tt> to type
3871<tt>ty2</tt> by applying either a zero extension or a truncation depending on
3872the size of the integer <tt>value</tt>. If <tt>value</tt> is larger than the
3873size of a pointer then a truncation is done. If <tt>value</tt> is smaller than
3874the size of a pointer then a zero extension is done. If they are the same size,
3875nothing is done (<i>no-op cast</i>).</p>
3876
3877<h5>Example:</h5>
3878<pre>
3879 %X = inttoptr i32 255 to i32* <i>; yields zero extension on 64-bit architecture</i>
3880 %X = inttoptr i32 255 to i32* <i>; yields no-op on 32-bit architecture</i>
3881 %Y = inttoptr i64 0 to i32* <i>; yields truncation on 32-bit architecture</i>
3882</pre>
3883</div>
3884
3885<!-- _______________________________________________________________________ -->
3886<div class="doc_subsubsection">
3887 <a name="i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a>
3888</div>
3889<div class="doc_text">
3890
3891<h5>Syntax:</h5>
3892<pre>
3893 &lt;result&gt; = bitcast &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3894</pre>
3895
3896<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3897
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3898<p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
3899<tt>ty2</tt> without changing any bits.</p>
3900
3901<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3902
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3903<p>The '<tt>bitcast</tt>' instruction takes a value to cast, which must be
Dan Gohman7305fa02008-09-08 16:45:59 +0000[diff] [blame]3904a non-aggregate first class value, and a type to cast it to, which must also be
3905a non-aggregate <a href="#t_firstclass">first class</a> type. The bit sizes of
3906<tt>value</tt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3907and the destination type, <tt>ty2</tt>, must be identical. If the source
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3908type is a pointer, the destination type must also be a pointer. This
3909instruction supports bitwise conversion of vectors to integers and to vectors
3910of other types (as long as they have the same size).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3911
3912<h5>Semantics:</h5>
3913<p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
3914<tt>ty2</tt>. It is always a <i>no-op cast</i> because no bits change with
3915this conversion. The conversion is done as if the <tt>value</tt> had been
3916stored to memory and read back as type <tt>ty2</tt>. Pointer types may only be
3917converted to other pointer types with this instruction. To convert pointers to
3918other types, use the <a href="#i_inttoptr">inttoptr</a> or
3919<a href="#i_ptrtoint">ptrtoint</a> instructions first.</p>
3920
3921<h5>Example:</h5>
3922<pre>
3923 %X = bitcast i8 255 to i8 <i>; yields i8 :-1</i>
3924 %Y = bitcast i32* %x to sint* <i>; yields sint*:%x</i>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]3925 %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]3926</pre>
3927</div>
3928
3929<!-- ======================================================================= -->
3930<div class="doc_subsection"> <a name="otherops">Other Operations</a> </div>
3931<div class="doc_text">
3932<p>The instructions in this category are the "miscellaneous"
3933instructions, which defy better classification.</p>
3934</div>
3935
3936<!-- _______________________________________________________________________ -->
3937<div class="doc_subsubsection"><a name="i_icmp">'<tt>icmp</tt>' Instruction</a>
3938</div>
3939<div class="doc_text">
3940<h5>Syntax:</h5>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]3941<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]3942</pre>
3943<h5>Overview:</h5>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]3944<p>The '<tt>icmp</tt>' instruction returns a boolean value or
3945a vector of boolean values based on comparison
3946of its two integer, integer vector, or pointer operands.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3947<h5>Arguments:</h5>
3948<p>The '<tt>icmp</tt>' instruction takes three operands. The first operand is
3949the condition code indicating the kind of comparison to perform. It is not
3950a value, just a keyword. The possible condition code are:
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]3951</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3952<ol>
3953 <li><tt>eq</tt>: equal</li>
3954 <li><tt>ne</tt>: not equal </li>
3955 <li><tt>ugt</tt>: unsigned greater than</li>
3956 <li><tt>uge</tt>: unsigned greater or equal</li>
3957 <li><tt>ult</tt>: unsigned less than</li>
3958 <li><tt>ule</tt>: unsigned less or equal</li>
3959 <li><tt>sgt</tt>: signed greater than</li>
3960 <li><tt>sge</tt>: signed greater or equal</li>
3961 <li><tt>slt</tt>: signed less than</li>
3962 <li><tt>sle</tt>: signed less or equal</li>
3963</ol>
3964<p>The remaining two arguments must be <a href="#t_integer">integer</a> or
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]3965<a href="#t_pointer">pointer</a>
3966or integer <a href="#t_vector">vector</a> typed.
3967They must also be identical types.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3968<h5>Semantics:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3969<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]3970the condition code given as <tt>cond</tt>. The comparison performed always
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]3971yields 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]3972</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3973<ol>
3974 <li><tt>eq</tt>: yields <tt>true</tt> if the operands are equal,
3975 <tt>false</tt> otherwise. No sign interpretation is necessary or performed.
3976 </li>
3977 <li><tt>ne</tt>: yields <tt>true</tt> if the operands are unequal,
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]3978 <tt>false</tt> otherwise. No sign interpretation is necessary or performed.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3979 <li><tt>ugt</tt>: interprets the operands as unsigned values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3980 <tt>true</tt> if <tt>op1</tt> is greater than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3981 <li><tt>uge</tt>: interprets the operands as unsigned values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3982 <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]3983 <li><tt>ult</tt>: interprets the operands as unsigned values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3984 <tt>true</tt> if <tt>op1</tt> is less than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3985 <li><tt>ule</tt>: interprets the operands as unsigned values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3986 <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]3987 <li><tt>sgt</tt>: interprets the operands as signed values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3988 <tt>true</tt> if <tt>op1</tt> is greater than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3989 <li><tt>sge</tt>: interprets the operands as signed values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3990 <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]3991 <li><tt>slt</tt>: interprets the operands as signed values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3992 <tt>true</tt> if <tt>op1</tt> is less than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3993 <li><tt>sle</tt>: interprets the operands as signed values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3994 <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]3995</ol>
3996<p>If the operands are <a href="#t_pointer">pointer</a> typed, the pointer
3997values are compared as if they were integers.</p>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]3998<p>If the operands are integer vectors, then they are compared
3999element by element. The result is an <tt>i1</tt> vector with
4000the same number of elements as the values being compared.
4001Otherwise, the result is an <tt>i1</tt>.
4002</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4003
4004<h5>Example:</h5>
4005<pre> &lt;result&gt; = icmp eq i32 4, 5 <i>; yields: result=false</i>
4006 &lt;result&gt; = icmp ne float* %X, %X <i>; yields: result=false</i>
4007 &lt;result&gt; = icmp ult i16 4, 5 <i>; yields: result=true</i>
4008 &lt;result&gt; = icmp sgt i16 4, 5 <i>; yields: result=false</i>
4009 &lt;result&gt; = icmp ule i16 -4, 5 <i>; yields: result=false</i>
4010 &lt;result&gt; = icmp sge i16 4, 5 <i>; yields: result=false</i>
4011</pre>
4012</div>
4013
4014<!-- _______________________________________________________________________ -->
4015<div class="doc_subsubsection"><a name="i_fcmp">'<tt>fcmp</tt>' Instruction</a>
4016</div>
4017<div class="doc_text">
4018<h5>Syntax:</h5>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4019<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]4020</pre>
4021<h5>Overview:</h5>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4022<p>The '<tt>fcmp</tt>' instruction returns a boolean value
4023or vector of boolean values based on comparison
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4024of its operands.</p>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4025<p>
4026If the operands are floating point scalars, then the result
4027type is a boolean (<a href="#t_primitive"><tt>i1</tt></a>).
4028</p>
4029<p>If the operands are floating point vectors, then the result type
4030is a vector of boolean with the same number of elements as the
4031operands being compared.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4032<h5>Arguments:</h5>
4033<p>The '<tt>fcmp</tt>' instruction takes three operands. The first operand is
4034the condition code indicating the kind of comparison to perform. It is not
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4035a value, just a keyword. The possible condition code are:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4036<ol>
4037 <li><tt>false</tt>: no comparison, always returns false</li>
4038 <li><tt>oeq</tt>: ordered and equal</li>
4039 <li><tt>ogt</tt>: ordered and greater than </li>
4040 <li><tt>oge</tt>: ordered and greater than or equal</li>
4041 <li><tt>olt</tt>: ordered and less than </li>
4042 <li><tt>ole</tt>: ordered and less than or equal</li>
4043 <li><tt>one</tt>: ordered and not equal</li>
4044 <li><tt>ord</tt>: ordered (no nans)</li>
4045 <li><tt>ueq</tt>: unordered or equal</li>
4046 <li><tt>ugt</tt>: unordered or greater than </li>
4047 <li><tt>uge</tt>: unordered or greater than or equal</li>
4048 <li><tt>ult</tt>: unordered or less than </li>
4049 <li><tt>ule</tt>: unordered or less than or equal</li>
4050 <li><tt>une</tt>: unordered or not equal</li>
4051 <li><tt>uno</tt>: unordered (either nans)</li>
4052 <li><tt>true</tt>: no comparison, always returns true</li>
4053</ol>
4054<p><i>Ordered</i> means that neither operand is a QNAN while
4055<i>unordered</i> means that either operand may be a QNAN.</p>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4056<p>Each of <tt>val1</tt> and <tt>val2</tt> arguments must be
4057either a <a href="#t_floating">floating point</a> type
4058or a <a href="#t_vector">vector</a> of floating point type.
4059They must have identical types.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4060<h5>Semantics:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4061<p>The '<tt>fcmp</tt>' instruction compares <tt>op1</tt> and <tt>op2</tt>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4062according to the condition code given as <tt>cond</tt>.
4063If the operands are vectors, then the vectors are compared
4064element by element.
4065Each comparison performed
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4066always yields an <a href="#t_primitive">i1</a> result, as follows:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4067<ol>
4068 <li><tt>false</tt>: always yields <tt>false</tt>, regardless of operands.</li>
4069 <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]4070 <tt>op1</tt> is equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4071 <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]4072 <tt>op1</tt> is greather than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4073 <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]4074 <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4075 <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]4076 <tt>op1</tt> is less than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4077 <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]4078 <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4079 <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]4080 <tt>op1</tt> is not equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4081 <li><tt>ord</tt>: yields <tt>true</tt> if both operands are not a QNAN.</li>
4082 <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]4083 <tt>op1</tt> is equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4084 <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]4085 <tt>op1</tt> is greater than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4086 <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]4087 <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4088 <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]4089 <tt>op1</tt> is less than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4090 <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]4091 <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4092 <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]4093 <tt>op1</tt> is not equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4094 <li><tt>uno</tt>: yields <tt>true</tt> if either operand is a QNAN.</li>
4095 <li><tt>true</tt>: always yields <tt>true</tt>, regardless of operands.</li>
4096</ol>
4097
4098<h5>Example:</h5>
4099<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]4100 &lt;result&gt; = fcmp one float 4.0, 5.0 <i>; yields: result=true</i>
4101 &lt;result&gt; = fcmp olt float 4.0, 5.0 <i>; yields: result=true</i>
4102 &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]4103</pre>
4104</div>
4105
4106<!-- _______________________________________________________________________ -->
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]4107<div class="doc_subsubsection">
4108 <a name="i_vicmp">'<tt>vicmp</tt>' Instruction</a>
4109</div>
4110<div class="doc_text">
4111<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4112<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]4113</pre>
4114<h5>Overview:</h5>
4115<p>The '<tt>vicmp</tt>' instruction returns an integer vector value based on
4116element-wise comparison of its two integer vector operands.</p>
4117<h5>Arguments:</h5>
4118<p>The '<tt>vicmp</tt>' instruction takes three operands. The first operand is
4119the condition code indicating the kind of comparison to perform. It is not
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4120a value, just a keyword. The possible condition code are:</p>
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]4121<ol>
4122 <li><tt>eq</tt>: equal</li>
4123 <li><tt>ne</tt>: not equal </li>
4124 <li><tt>ugt</tt>: unsigned greater than</li>
4125 <li><tt>uge</tt>: unsigned greater or equal</li>
4126 <li><tt>ult</tt>: unsigned less than</li>
4127 <li><tt>ule</tt>: unsigned less or equal</li>
4128 <li><tt>sgt</tt>: signed greater than</li>
4129 <li><tt>sge</tt>: signed greater or equal</li>
4130 <li><tt>slt</tt>: signed less than</li>
4131 <li><tt>sle</tt>: signed less or equal</li>
4132</ol>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4133<p>The remaining two arguments must be <a href="#t_vector">vector</a> or
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]4134<a href="#t_integer">integer</a> typed. They must also be identical types.</p>
4135<h5>Semantics:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4136<p>The '<tt>vicmp</tt>' instruction compares <tt>op1</tt> and <tt>op2</tt>
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]4137according to the condition code given as <tt>cond</tt>. The comparison yields a
4138<a href="#t_vector">vector</a> of <a href="#t_integer">integer</a> result, of
4139identical type as the values being compared. The most significant bit in each
4140element is 1 if the element-wise comparison evaluates to true, and is 0
4141otherwise. All other bits of the result are undefined. The condition codes
4142are evaluated identically to the <a href="#i_icmp">'<tt>icmp</tt>'
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4143instruction</a>.</p>
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]4144
4145<h5>Example:</h5>
4146<pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4147 &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>
4148 &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]4149</pre>
4150</div>
4151
4152<!-- _______________________________________________________________________ -->
4153<div class="doc_subsubsection">
4154 <a name="i_vfcmp">'<tt>vfcmp</tt>' Instruction</a>
4155</div>
4156<div class="doc_text">
4157<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4158<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]4159<h5>Overview:</h5>
4160<p>The '<tt>vfcmp</tt>' instruction returns an integer vector value based on
4161element-wise comparison of its two floating point vector operands. The output
4162elements have the same width as the input elements.</p>
4163<h5>Arguments:</h5>
4164<p>The '<tt>vfcmp</tt>' instruction takes three operands. The first operand is
4165the condition code indicating the kind of comparison to perform. It is not
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4166a value, just a keyword. The possible condition code are:</p>
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]4167<ol>
4168 <li><tt>false</tt>: no comparison, always returns false</li>
4169 <li><tt>oeq</tt>: ordered and equal</li>
4170 <li><tt>ogt</tt>: ordered and greater than </li>
4171 <li><tt>oge</tt>: ordered and greater than or equal</li>
4172 <li><tt>olt</tt>: ordered and less than </li>
4173 <li><tt>ole</tt>: ordered and less than or equal</li>
4174 <li><tt>one</tt>: ordered and not equal</li>
4175 <li><tt>ord</tt>: ordered (no nans)</li>
4176 <li><tt>ueq</tt>: unordered or equal</li>
4177 <li><tt>ugt</tt>: unordered or greater than </li>
4178 <li><tt>uge</tt>: unordered or greater than or equal</li>
4179 <li><tt>ult</tt>: unordered or less than </li>
4180 <li><tt>ule</tt>: unordered or less than or equal</li>
4181 <li><tt>une</tt>: unordered or not equal</li>
4182 <li><tt>uno</tt>: unordered (either nans)</li>
4183 <li><tt>true</tt>: no comparison, always returns true</li>
4184</ol>
4185<p>The remaining two arguments must be <a href="#t_vector">vector</a> of
4186<a href="#t_floating">floating point</a> typed. They must also be identical
4187types.</p>
4188<h5>Semantics:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4189<p>The '<tt>vfcmp</tt>' instruction compares <tt>op1</tt> and <tt>op2</tt>
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]4190according to the condition code given as <tt>cond</tt>. The comparison yields a
4191<a href="#t_vector">vector</a> of <a href="#t_integer">integer</a> result, with
4192an identical number of elements as the values being compared, and each element
4193having identical with to the width of the floating point elements. The most
4194significant bit in each element is 1 if the element-wise comparison evaluates to
4195true, and is 0 otherwise. All other bits of the result are undefined. The
4196condition codes are evaluated identically to the
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4197<a href="#i_fcmp">'<tt>fcmp</tt>' instruction</a>.</p>
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]4198
4199<h5>Example:</h5>
4200<pre>
Chris Lattner1e5c5cd02008-10-13 16:55:18 +0000[diff] [blame]4201 <i>; yields: result=&lt;2 x i32&gt; &lt; i32 0, i32 -1 &gt;</i>
4202 &lt;result&gt; = vfcmp oeq &lt;2 x float&gt; &lt; float 4, float 0 &gt;, &lt; float 5, float 0 &gt;
4203
4204 <i>; yields: result=&lt;2 x i64&gt; &lt; i64 -1, i64 0 &gt;</i>
4205 &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]4206</pre>
4207</div>
4208
4209<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4210<div class="doc_subsubsection">
4211 <a name="i_phi">'<tt>phi</tt>' Instruction</a>
4212</div>
4213
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4214<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4215
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4216<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4217
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4218<pre> &lt;result&gt; = phi &lt;ty&gt; [ &lt;val0&gt;, &lt;label0&gt;], ...<br></pre>
4219<h5>Overview:</h5>
4220<p>The '<tt>phi</tt>' instruction is used to implement the &#966; node in
4221the SSA graph representing the function.</p>
4222<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4223
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4224<p>The type of the incoming values is specified with the first type
4225field. After this, the '<tt>phi</tt>' instruction takes a list of pairs
4226as arguments, with one pair for each predecessor basic block of the
4227current block. Only values of <a href="#t_firstclass">first class</a>
4228type may be used as the value arguments to the PHI node. Only labels
4229may be used as the label arguments.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4230
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4231<p>There must be no non-phi instructions between the start of a basic
4232block and the PHI instructions: i.e. PHI instructions must be first in
4233a basic block.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4234
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4235<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4236
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4237<p>At runtime, the '<tt>phi</tt>' instruction logically takes on the value
4238specified by the pair corresponding to the predecessor basic block that executed
4239just prior to the current block.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4240
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4241<h5>Example:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4242<pre>
4243Loop: ; Infinite loop that counts from 0 on up...
4244 %indvar = phi i32 [ 0, %LoopHeader ], [ %nextindvar, %Loop ]
4245 %nextindvar = add i32 %indvar, 1
4246 br label %Loop
4247</pre>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4248</div>
4249
4250<!-- _______________________________________________________________________ -->
4251<div class="doc_subsubsection">
4252 <a name="i_select">'<tt>select</tt>' Instruction</a>
4253</div>
4254
4255<div class="doc_text">
4256
4257<h5>Syntax:</h5>
4258
4259<pre>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4260 &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>
4261
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4262 <i>selty</i> is either i1 or {&lt;N x i1&gt;}
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4263</pre>
4264
4265<h5>Overview:</h5>
4266
4267<p>
4268The '<tt>select</tt>' instruction is used to choose one value based on a
4269condition, without branching.
4270</p>
4271
4272
4273<h5>Arguments:</h5>
4274
4275<p>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4276The '<tt>select</tt>' instruction requires an 'i1' value or
4277a vector of 'i1' values indicating the
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4278condition, and two values of the same <a href="#t_firstclass">first class</a>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4279type. If the val1/val2 are vectors and
4280the condition is a scalar, then entire vectors are selected, not
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4281individual elements.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4282</p>
4283
4284<h5>Semantics:</h5>
4285
4286<p>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4287If 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]4288value argument; otherwise, it returns the second value argument.
4289</p>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4290<p>
4291If the condition is a vector of i1, then the value arguments must
4292be vectors of the same size, and the selection is done element
4293by element.
4294</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4295
4296<h5>Example:</h5>
4297
4298<pre>
4299 %X = select i1 true, i8 17, i8 42 <i>; yields i8:17</i>
4300</pre>
4301</div>
4302
4303
4304<!-- _______________________________________________________________________ -->
4305<div class="doc_subsubsection">
4306 <a name="i_call">'<tt>call</tt>' Instruction</a>
4307</div>
4308
4309<div class="doc_text">
4310
4311<h5>Syntax:</h5>
4312<pre>
Devang Pateld0bfcc72008-10-07 17:48:33 +0000[diff] [blame]4313 &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]4314</pre>
4315
4316<h5>Overview:</h5>
4317
4318<p>The '<tt>call</tt>' instruction represents a simple function call.</p>
4319
4320<h5>Arguments:</h5>
4321
4322<p>This instruction requires several arguments:</p>
4323
4324<ol>
4325 <li>
4326 <p>The optional "tail" marker indicates whether the callee function accesses
4327 any allocas or varargs in the caller. If the "tail" marker is present, the
4328 function call is eligible for tail call optimization. Note that calls may
4329 be marked "tail" even if they do not occur before a <a
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4330 href="#i_ret"><tt>ret</tt></a> instruction.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4331 </li>
4332 <li>
4333 <p>The optional "cconv" marker indicates which <a href="#callingconv">calling
4334 convention</a> the call should use. If none is specified, the call defaults
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4335 to using C calling conventions.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4336 </li>
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]4337
4338 <li>
4339 <p>The optional <a href="#paramattrs">Parameter Attributes</a> list for
4340 return values. Only '<tt>zeroext</tt>', '<tt>signext</tt>',
4341 and '<tt>inreg</tt>' attributes are valid here.</p>
4342 </li>
4343
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4344 <li>
Nick Lewycky93082fc2007-09-08 13:57:50 +0000[diff] [blame]4345 <p>'<tt>ty</tt>': the type of the call instruction itself which is also
4346 the type of the return value. Functions that return no value are marked
4347 <tt><a href="#t_void">void</a></tt>.</p>
4348 </li>
4349 <li>
4350 <p>'<tt>fnty</tt>': shall be the signature of the pointer to function
4351 value being invoked. The argument types must match the types implied by
4352 this signature. This type can be omitted if the function is not varargs
4353 and if the function type does not return a pointer to a function.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4354 </li>
4355 <li>
4356 <p>'<tt>fnptrval</tt>': An LLVM value containing a pointer to a function to
4357 be invoked. In most cases, this is a direct function invocation, but
4358 indirect <tt>call</tt>s are just as possible, calling an arbitrary pointer
4359 to function value.</p>
4360 </li>
4361 <li>
4362 <p>'<tt>function args</tt>': argument list whose types match the
4363 function signature argument types. All arguments must be of
4364 <a href="#t_firstclass">first class</a> type. If the function signature
4365 indicates the function accepts a variable number of arguments, the extra
4366 arguments can be specified.</p>
4367 </li>
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]4368 <li>
Devang Pateld0bfcc72008-10-07 17:48:33 +0000[diff] [blame]4369 <p>The optional <a href="#fnattrs">function attributes</a> list. Only
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]4370 '<tt>noreturn</tt>', '<tt>nounwind</tt>', '<tt>readonly</tt>' and
4371 '<tt>readnone</tt>' attributes are valid here.</p>
4372 </li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4373</ol>
4374
4375<h5>Semantics:</h5>
4376
4377<p>The '<tt>call</tt>' instruction is used to cause control flow to
4378transfer to a specified function, with its incoming arguments bound to
4379the specified values. Upon a '<tt><a href="#i_ret">ret</a></tt>'
4380instruction in the called function, control flow continues with the
4381instruction after the function call, and the return value of the
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4382function is bound to the result argument.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4383
4384<h5>Example:</h5>
4385
4386<pre>
Nick Lewycky93082fc2007-09-08 13:57:50 +0000[diff] [blame]4387 %retval = call i32 @test(i32 %argc)
Chris Lattner5e893ef2008-03-21 17:24:17 +0000[diff] [blame]4388 call i32 (i8 *, ...)* @printf(i8 * %msg, i32 12, i8 42) <i>; yields i32</i>
4389 %X = tail call i32 @foo() <i>; yields i32</i>
4390 %Y = tail call <a href="#callingconv">fastcc</a> i32 @foo() <i>; yields i32</i>
4391 call void %foo(i8 97 signext)
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]4392
4393 %struct.A = type { i32, i8 }
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]4394 %r = call %struct.A @foo() <i>; yields { 32, i8 }</i>
Dan Gohman3e700032008-10-04 19:00:07 +0000[diff] [blame]4395 %gr = extractvalue %struct.A %r, 0 <i>; yields i32</i>
4396 %gr1 = extractvalue %struct.A %r, 1 <i>; yields i8</i>
Chris Lattnerac454b32008-10-08 06:26:11 +0000[diff] [blame]4397 %Z = call void @foo() noreturn <i>; indicates that %foo never returns normally</i>
Matthijs Kooijman2c4e05a2008-10-07 10:03:45 +0000[diff] [blame]4398 %ZZ = call zeroext i32 @bar() <i>; Return value is %zero extended</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4399</pre>
4400
4401</div>
4402
4403<!-- _______________________________________________________________________ -->
4404<div class="doc_subsubsection">
4405 <a name="i_va_arg">'<tt>va_arg</tt>' Instruction</a>
4406</div>
4407
4408<div class="doc_text">
4409
4410<h5>Syntax:</h5>
4411
4412<pre>
4413 &lt;resultval&gt; = va_arg &lt;va_list*&gt; &lt;arglist&gt;, &lt;argty&gt;
4414</pre>
4415
4416<h5>Overview:</h5>
4417
4418<p>The '<tt>va_arg</tt>' instruction is used to access arguments passed through
4419the "variable argument" area of a function call. It is used to implement the
4420<tt>va_arg</tt> macro in C.</p>
4421
4422<h5>Arguments:</h5>
4423
4424<p>This instruction takes a <tt>va_list*</tt> value and the type of
4425the argument. It returns a value of the specified argument type and
4426increments the <tt>va_list</tt> to point to the next argument. The
4427actual type of <tt>va_list</tt> is target specific.</p>
4428
4429<h5>Semantics:</h5>
4430
4431<p>The '<tt>va_arg</tt>' instruction loads an argument of the specified
4432type from the specified <tt>va_list</tt> and causes the
4433<tt>va_list</tt> to point to the next argument. For more information,
4434see the variable argument handling <a href="#int_varargs">Intrinsic
4435Functions</a>.</p>
4436
4437<p>It is legal for this instruction to be called in a function which does not
4438take a variable number of arguments, for example, the <tt>vfprintf</tt>
4439function.</p>
4440
4441<p><tt>va_arg</tt> is an LLVM instruction instead of an <a
4442href="#intrinsics">intrinsic function</a> because it takes a type as an
4443argument.</p>
4444
4445<h5>Example:</h5>
4446
4447<p>See the <a href="#int_varargs">variable argument processing</a> section.</p>
4448
4449</div>
4450
4451<!-- *********************************************************************** -->
4452<div class="doc_section"> <a name="intrinsics">Intrinsic Functions</a> </div>
4453<!-- *********************************************************************** -->
4454
4455<div class="doc_text">
4456
4457<p>LLVM supports the notion of an "intrinsic function". These functions have
4458well known names and semantics and are required to follow certain restrictions.
4459Overall, these intrinsics represent an extension mechanism for the LLVM
4460language that does not require changing all of the transformations in LLVM when
4461adding to the language (or the bitcode reader/writer, the parser, etc...).</p>
4462
4463<p>Intrinsic function names must all start with an "<tt>llvm.</tt>" prefix. This
4464prefix is reserved in LLVM for intrinsic names; thus, function names may not
4465begin with this prefix. Intrinsic functions must always be external functions:
4466you cannot define the body of intrinsic functions. Intrinsic functions may
4467only be used in call or invoke instructions: it is illegal to take the address
4468of an intrinsic function. Additionally, because intrinsic functions are part
4469of the LLVM language, it is required if any are added that they be documented
4470here.</p>
4471
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]4472<p>Some intrinsic functions can be overloaded, i.e., the intrinsic represents
4473a family of functions that perform the same operation but on different data
4474types. Because LLVM can represent over 8 million different integer types,
4475overloading is used commonly to allow an intrinsic function to operate on any
4476integer type. One or more of the argument types or the result type can be
4477overloaded to accept any integer type. Argument types may also be defined as
4478exactly matching a previous argument's type or the result type. This allows an
4479intrinsic function which accepts multiple arguments, but needs all of them to
4480be of the same type, to only be overloaded with respect to a single argument or
4481the result.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4482
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]4483<p>Overloaded intrinsics will have the names of its overloaded argument types
4484encoded into its function name, each preceded by a period. Only those types
4485which are overloaded result in a name suffix. Arguments whose type is matched
4486against another type do not. For example, the <tt>llvm.ctpop</tt> function can
4487take an integer of any width and returns an integer of exactly the same integer
4488width. This leads to a family of functions such as
4489<tt>i8 @llvm.ctpop.i8(i8 %val)</tt> and <tt>i29 @llvm.ctpop.i29(i29 %val)</tt>.
4490Only one type, the return type, is overloaded, and only one type suffix is
4491required. Because the argument's type is matched against the return type, it
4492does not require its own name suffix.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4493
4494<p>To learn how to add an intrinsic function, please see the
4495<a href="ExtendingLLVM.html">Extending LLVM Guide</a>.
4496</p>
4497
4498</div>
4499
4500<!-- ======================================================================= -->
4501<div class="doc_subsection">
4502 <a name="int_varargs">Variable Argument Handling Intrinsics</a>
4503</div>
4504
4505<div class="doc_text">
4506
4507<p>Variable argument support is defined in LLVM with the <a
4508 href="#i_va_arg"><tt>va_arg</tt></a> instruction and these three
4509intrinsic functions. These functions are related to the similarly
4510named macros defined in the <tt>&lt;stdarg.h&gt;</tt> header file.</p>
4511
4512<p>All of these functions operate on arguments that use a
4513target-specific value type "<tt>va_list</tt>". The LLVM assembly
4514language reference manual does not define what this type is, so all
4515transformations should be prepared to handle these functions regardless of
4516the type used.</p>
4517
4518<p>This example shows how the <a href="#i_va_arg"><tt>va_arg</tt></a>
4519instruction and the variable argument handling intrinsic functions are
4520used.</p>
4521
4522<div class="doc_code">
4523<pre>
4524define i32 @test(i32 %X, ...) {
4525 ; Initialize variable argument processing
4526 %ap = alloca i8*
4527 %ap2 = bitcast i8** %ap to i8*
4528 call void @llvm.va_start(i8* %ap2)
4529
4530 ; Read a single integer argument
4531 %tmp = va_arg i8** %ap, i32
4532
4533 ; Demonstrate usage of llvm.va_copy and llvm.va_end
4534 %aq = alloca i8*
4535 %aq2 = bitcast i8** %aq to i8*
4536 call void @llvm.va_copy(i8* %aq2, i8* %ap2)
4537 call void @llvm.va_end(i8* %aq2)
4538
4539 ; Stop processing of arguments.
4540 call void @llvm.va_end(i8* %ap2)
4541 ret i32 %tmp
4542}
4543
4544declare void @llvm.va_start(i8*)
4545declare void @llvm.va_copy(i8*, i8*)
4546declare void @llvm.va_end(i8*)
4547</pre>
4548</div>
4549
4550</div>
4551
4552<!-- _______________________________________________________________________ -->
4553<div class="doc_subsubsection">
4554 <a name="int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a>
4555</div>
4556
4557
4558<div class="doc_text">
4559<h5>Syntax:</h5>
4560<pre> declare void %llvm.va_start(i8* &lt;arglist&gt;)<br></pre>
4561<h5>Overview:</h5>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4562<p>The '<tt>llvm.va_start</tt>' intrinsic initializes
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4563<tt>*&lt;arglist&gt;</tt> for subsequent use by <tt><a
4564href="#i_va_arg">va_arg</a></tt>.</p>
4565
4566<h5>Arguments:</h5>
4567
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4568<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]4569
4570<h5>Semantics:</h5>
4571
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4572<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]4573macro available in C. In a target-dependent way, it initializes the
4574<tt>va_list</tt> element to which the argument points, so that the next call to
4575<tt>va_arg</tt> will produce the first variable argument passed to the function.
4576Unlike the C <tt>va_start</tt> macro, this intrinsic does not need to know the
4577last argument of the function as the compiler can figure that out.</p>
4578
4579</div>
4580
4581<!-- _______________________________________________________________________ -->
4582<div class="doc_subsubsection">
4583 <a name="int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a>
4584</div>
4585
4586<div class="doc_text">
4587<h5>Syntax:</h5>
4588<pre> declare void @llvm.va_end(i8* &lt;arglist&gt;)<br></pre>
4589<h5>Overview:</h5>
4590
4591<p>The '<tt>llvm.va_end</tt>' intrinsic destroys <tt>*&lt;arglist&gt;</tt>,
4592which has been initialized previously with <tt><a href="#int_va_start">llvm.va_start</a></tt>
4593or <tt><a href="#i_va_copy">llvm.va_copy</a></tt>.</p>
4594
4595<h5>Arguments:</h5>
4596
4597<p>The argument is a pointer to a <tt>va_list</tt> to destroy.</p>
4598
4599<h5>Semantics:</h5>
4600
4601<p>The '<tt>llvm.va_end</tt>' intrinsic works just like the <tt>va_end</tt>
4602macro available in C. In a target-dependent way, it destroys the
4603<tt>va_list</tt> element to which the argument points. Calls to <a
4604href="#int_va_start"><tt>llvm.va_start</tt></a> and <a href="#int_va_copy">
4605<tt>llvm.va_copy</tt></a> must be matched exactly with calls to
4606<tt>llvm.va_end</tt>.</p>
4607
4608</div>
4609
4610<!-- _______________________________________________________________________ -->
4611<div class="doc_subsubsection">
4612 <a name="int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a>
4613</div>
4614
4615<div class="doc_text">
4616
4617<h5>Syntax:</h5>
4618
4619<pre>
4620 declare void @llvm.va_copy(i8* &lt;destarglist&gt;, i8* &lt;srcarglist&gt;)
4621</pre>
4622
4623<h5>Overview:</h5>
4624
4625<p>The '<tt>llvm.va_copy</tt>' intrinsic copies the current argument position
4626from the source argument list to the destination argument list.</p>
4627
4628<h5>Arguments:</h5>
4629
4630<p>The first argument is a pointer to a <tt>va_list</tt> element to initialize.
4631The second argument is a pointer to a <tt>va_list</tt> element to copy from.</p>
4632
4633
4634<h5>Semantics:</h5>
4635
4636<p>The '<tt>llvm.va_copy</tt>' intrinsic works just like the <tt>va_copy</tt>
4637macro available in C. In a target-dependent way, it copies the source
4638<tt>va_list</tt> element into the destination <tt>va_list</tt> element. This
4639intrinsic is necessary because the <tt><a href="#int_va_start">
4640llvm.va_start</a></tt> intrinsic may be arbitrarily complex and require, for
4641example, memory allocation.</p>
4642
4643</div>
4644
4645<!-- ======================================================================= -->
4646<div class="doc_subsection">
4647 <a name="int_gc">Accurate Garbage Collection Intrinsics</a>
4648</div>
4649
4650<div class="doc_text">
4651
4652<p>
4653LLVM support for <a href="GarbageCollection.html">Accurate Garbage
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]4654Collection</a> (GC) requires the implementation and generation of these
4655intrinsics.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4656These intrinsics allow identification of <a href="#int_gcroot">GC roots on the
4657stack</a>, as well as garbage collector implementations that require <a
4658href="#int_gcread">read</a> and <a href="#int_gcwrite">write</a> barriers.
4659Front-ends for type-safe garbage collected languages should generate these
4660intrinsics to make use of the LLVM garbage collectors. For more details, see <a
4661href="GarbageCollection.html">Accurate Garbage Collection with LLVM</a>.
4662</p>
Christopher Lambcfe00962007-12-17 01:00:21 +0000[diff] [blame]4663
4664<p>The garbage collection intrinsics only operate on objects in the generic
4665 address space (address space zero).</p>
4666
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4667</div>
4668
4669<!-- _______________________________________________________________________ -->
4670<div class="doc_subsubsection">
4671 <a name="int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a>
4672</div>
4673
4674<div class="doc_text">
4675
4676<h5>Syntax:</h5>
4677
4678<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4679 declare void @llvm.gcroot(i8** %ptrloc, i8* %metadata)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4680</pre>
4681
4682<h5>Overview:</h5>
4683
4684<p>The '<tt>llvm.gcroot</tt>' intrinsic declares the existence of a GC root to
4685the code generator, and allows some metadata to be associated with it.</p>
4686
4687<h5>Arguments:</h5>
4688
4689<p>The first argument specifies the address of a stack object that contains the
4690root pointer. The second pointer (which must be either a constant or a global
4691value address) contains the meta-data to be associated with the root.</p>
4692
4693<h5>Semantics:</h5>
4694
Chris Lattnera7d94ba2008-04-24 05:59:56 +0000[diff] [blame]4695<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]4696location. At compile-time, the code generator generates information to allow
Gordon Henriksen40393542007-12-25 02:31:26 +0000[diff] [blame]4697the runtime to find the pointer at GC safe points. The '<tt>llvm.gcroot</tt>'
4698intrinsic may only be used in a function which <a href="#gc">specifies a GC
4699algorithm</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4700
4701</div>
4702
4703
4704<!-- _______________________________________________________________________ -->
4705<div class="doc_subsubsection">
4706 <a name="int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a>
4707</div>
4708
4709<div class="doc_text">
4710
4711<h5>Syntax:</h5>
4712
4713<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4714 declare i8* @llvm.gcread(i8* %ObjPtr, i8** %Ptr)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4715</pre>
4716
4717<h5>Overview:</h5>
4718
4719<p>The '<tt>llvm.gcread</tt>' intrinsic identifies reads of references from heap
4720locations, allowing garbage collector implementations that require read
4721barriers.</p>
4722
4723<h5>Arguments:</h5>
4724
4725<p>The second argument is the address to read from, which should be an address
4726allocated from the garbage collector. The first object is a pointer to the
4727start of the referenced object, if needed by the language runtime (otherwise
4728null).</p>
4729
4730<h5>Semantics:</h5>
4731
4732<p>The '<tt>llvm.gcread</tt>' intrinsic has the same semantics as a load
4733instruction, but may be replaced with substantially more complex code by the
Gordon Henriksen40393542007-12-25 02:31:26 +0000[diff] [blame]4734garbage collector runtime, as needed. The '<tt>llvm.gcread</tt>' intrinsic
4735may only be used in a function which <a href="#gc">specifies a GC
4736algorithm</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4737
4738</div>
4739
4740
4741<!-- _______________________________________________________________________ -->
4742<div class="doc_subsubsection">
4743 <a name="int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a>
4744</div>
4745
4746<div class="doc_text">
4747
4748<h5>Syntax:</h5>
4749
4750<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4751 declare void @llvm.gcwrite(i8* %P1, i8* %Obj, i8** %P2)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4752</pre>
4753
4754<h5>Overview:</h5>
4755
4756<p>The '<tt>llvm.gcwrite</tt>' intrinsic identifies writes of references to heap
4757locations, allowing garbage collector implementations that require write
4758barriers (such as generational or reference counting collectors).</p>
4759
4760<h5>Arguments:</h5>
4761
4762<p>The first argument is the reference to store, the second is the start of the
4763object to store it to, and the third is the address of the field of Obj to
4764store to. If the runtime does not require a pointer to the object, Obj may be
4765null.</p>
4766
4767<h5>Semantics:</h5>
4768
4769<p>The '<tt>llvm.gcwrite</tt>' intrinsic has the same semantics as a store
4770instruction, but may be replaced with substantially more complex code by the
Gordon Henriksen40393542007-12-25 02:31:26 +0000[diff] [blame]4771garbage collector runtime, as needed. The '<tt>llvm.gcwrite</tt>' intrinsic
4772may only be used in a function which <a href="#gc">specifies a GC
4773algorithm</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4774
4775</div>
4776
4777
4778
4779<!-- ======================================================================= -->
4780<div class="doc_subsection">
4781 <a name="int_codegen">Code Generator Intrinsics</a>
4782</div>
4783
4784<div class="doc_text">
4785<p>
4786These intrinsics are provided by LLVM to expose special features that may only
4787be implemented with code generator support.
4788</p>
4789
4790</div>
4791
4792<!-- _______________________________________________________________________ -->
4793<div class="doc_subsubsection">
4794 <a name="int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a>
4795</div>
4796
4797<div class="doc_text">
4798
4799<h5>Syntax:</h5>
4800<pre>
4801 declare i8 *@llvm.returnaddress(i32 &lt;level&gt;)
4802</pre>
4803
4804<h5>Overview:</h5>
4805
4806<p>
4807The '<tt>llvm.returnaddress</tt>' intrinsic attempts to compute a
4808target-specific value indicating the return address of the current function
4809or one of its callers.
4810</p>
4811
4812<h5>Arguments:</h5>
4813
4814<p>
4815The argument to this intrinsic indicates which function to return the address
4816for. Zero indicates the calling function, one indicates its caller, etc. The
4817argument is <b>required</b> to be a constant integer value.
4818</p>
4819
4820<h5>Semantics:</h5>
4821
4822<p>
4823The '<tt>llvm.returnaddress</tt>' intrinsic either returns a pointer indicating
4824the return address of the specified call frame, or zero if it cannot be
4825identified. The value returned by this intrinsic is likely to be incorrect or 0
4826for arguments other than zero, so it should only be used for debugging purposes.
4827</p>
4828
4829<p>
4830Note that calling this intrinsic does not prevent function inlining or other
4831aggressive transformations, so the value returned may not be that of the obvious
4832source-language caller.
4833</p>
4834</div>
4835
4836
4837<!-- _______________________________________________________________________ -->
4838<div class="doc_subsubsection">
4839 <a name="int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a>
4840</div>
4841
4842<div class="doc_text">
4843
4844<h5>Syntax:</h5>
4845<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4846 declare i8 *@llvm.frameaddress(i32 &lt;level&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4847</pre>
4848
4849<h5>Overview:</h5>
4850
4851<p>
4852The '<tt>llvm.frameaddress</tt>' intrinsic attempts to return the
4853target-specific frame pointer value for the specified stack frame.
4854</p>
4855
4856<h5>Arguments:</h5>
4857
4858<p>
4859The argument to this intrinsic indicates which function to return the frame
4860pointer for. Zero indicates the calling function, one indicates its caller,
4861etc. The argument is <b>required</b> to be a constant integer value.
4862</p>
4863
4864<h5>Semantics:</h5>
4865
4866<p>
4867The '<tt>llvm.frameaddress</tt>' intrinsic either returns a pointer indicating
4868the frame address of the specified call frame, or zero if it cannot be
4869identified. The value returned by this intrinsic is likely to be incorrect or 0
4870for arguments other than zero, so it should only be used for debugging purposes.
4871</p>
4872
4873<p>
4874Note that calling this intrinsic does not prevent function inlining or other
4875aggressive transformations, so the value returned may not be that of the obvious
4876source-language caller.
4877</p>
4878</div>
4879
4880<!-- _______________________________________________________________________ -->
4881<div class="doc_subsubsection">
4882 <a name="int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a>
4883</div>
4884
4885<div class="doc_text">
4886
4887<h5>Syntax:</h5>
4888<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4889 declare i8 *@llvm.stacksave()
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4890</pre>
4891
4892<h5>Overview:</h5>
4893
4894<p>
4895The '<tt>llvm.stacksave</tt>' intrinsic is used to remember the current state of
4896the function stack, for use with <a href="#int_stackrestore">
4897<tt>llvm.stackrestore</tt></a>. This is useful for implementing language
4898features like scoped automatic variable sized arrays in C99.
4899</p>
4900
4901<h5>Semantics:</h5>
4902
4903<p>
4904This intrinsic returns a opaque pointer value that can be passed to <a
4905href="#int_stackrestore"><tt>llvm.stackrestore</tt></a>. When an
4906<tt>llvm.stackrestore</tt> intrinsic is executed with a value saved from
4907<tt>llvm.stacksave</tt>, it effectively restores the state of the stack to the
4908state it was in when the <tt>llvm.stacksave</tt> intrinsic executed. In
4909practice, this pops any <a href="#i_alloca">alloca</a> blocks from the stack
4910that were allocated after the <tt>llvm.stacksave</tt> was executed.
4911</p>
4912
4913</div>
4914
4915<!-- _______________________________________________________________________ -->
4916<div class="doc_subsubsection">
4917 <a name="int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a>
4918</div>
4919
4920<div class="doc_text">
4921
4922<h5>Syntax:</h5>
4923<pre>
4924 declare void @llvm.stackrestore(i8 * %ptr)
4925</pre>
4926
4927<h5>Overview:</h5>
4928
4929<p>
4930The '<tt>llvm.stackrestore</tt>' intrinsic is used to restore the state of
4931the function stack to the state it was in when the corresponding <a
4932href="#int_stacksave"><tt>llvm.stacksave</tt></a> intrinsic executed. This is
4933useful for implementing language features like scoped automatic variable sized
4934arrays in C99.
4935</p>
4936
4937<h5>Semantics:</h5>
4938
4939<p>
4940See the description for <a href="#int_stacksave"><tt>llvm.stacksave</tt></a>.
4941</p>
4942
4943</div>
4944
4945
4946<!-- _______________________________________________________________________ -->
4947<div class="doc_subsubsection">
4948 <a name="int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a>
4949</div>
4950
4951<div class="doc_text">
4952
4953<h5>Syntax:</h5>
4954<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4955 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]4956</pre>
4957
4958<h5>Overview:</h5>
4959
4960
4961<p>
4962The '<tt>llvm.prefetch</tt>' intrinsic is a hint to the code generator to insert
4963a prefetch instruction if supported; otherwise, it is a noop. Prefetches have
4964no
4965effect on the behavior of the program but can change its performance
4966characteristics.
4967</p>
4968
4969<h5>Arguments:</h5>
4970
4971<p>
4972<tt>address</tt> is the address to be prefetched, <tt>rw</tt> is the specifier
4973determining if the fetch should be for a read (0) or write (1), and
4974<tt>locality</tt> is a temporal locality specifier ranging from (0) - no
4975locality, to (3) - extremely local keep in cache. The <tt>rw</tt> and
4976<tt>locality</tt> arguments must be constant integers.
4977</p>
4978
4979<h5>Semantics:</h5>
4980
4981<p>
4982This intrinsic does not modify the behavior of the program. In particular,
4983prefetches cannot trap and do not produce a value. On targets that support this
4984intrinsic, the prefetch can provide hints to the processor cache for better
4985performance.
4986</p>
4987
4988</div>
4989
4990<!-- _______________________________________________________________________ -->
4991<div class="doc_subsubsection">
4992 <a name="int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a>
4993</div>
4994
4995<div class="doc_text">
4996
4997<h5>Syntax:</h5>
4998<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4999 declare void @llvm.pcmarker(i32 &lt;id&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5000</pre>
5001
5002<h5>Overview:</h5>
5003
5004
5005<p>
5006The '<tt>llvm.pcmarker</tt>' intrinsic is a method to export a Program Counter
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]5007(PC) in a region of
5008code to simulators and other tools. The method is target specific, but it is
5009expected that the marker will use exported symbols to transmit the PC of the
5010marker.
5011The marker makes no guarantees that it will remain with any specific instruction
5012after optimizations. It is possible that the presence of a marker will inhibit
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5013optimizations. The intended use is to be inserted after optimizations to allow
5014correlations of simulation runs.
5015</p>
5016
5017<h5>Arguments:</h5>
5018
5019<p>
5020<tt>id</tt> is a numerical id identifying the marker.
5021</p>
5022
5023<h5>Semantics:</h5>
5024
5025<p>
5026This intrinsic does not modify the behavior of the program. Backends that do not
5027support this intrinisic may ignore it.
5028</p>
5029
5030</div>
5031
5032<!-- _______________________________________________________________________ -->
5033<div class="doc_subsubsection">
5034 <a name="int_readcyclecounter">'<tt>llvm.readcyclecounter</tt>' Intrinsic</a>
5035</div>
5036
5037<div class="doc_text">
5038
5039<h5>Syntax:</h5>
5040<pre>
5041 declare i64 @llvm.readcyclecounter( )
5042</pre>
5043
5044<h5>Overview:</h5>
5045
5046
5047<p>
5048The '<tt>llvm.readcyclecounter</tt>' intrinsic provides access to the cycle
5049counter register (or similar low latency, high accuracy clocks) on those targets
5050that support it. On X86, it should map to RDTSC. On Alpha, it should map to RPCC.
5051As the backing counters overflow quickly (on the order of 9 seconds on alpha), this
5052should only be used for small timings.
5053</p>
5054
5055<h5>Semantics:</h5>
5056
5057<p>
5058When directly supported, reading the cycle counter should not modify any memory.
5059Implementations are allowed to either return a application specific value or a
5060system wide value. On backends without support, this is lowered to a constant 0.
5061</p>
5062
5063</div>
5064
5065<!-- ======================================================================= -->
5066<div class="doc_subsection">
5067 <a name="int_libc">Standard C Library Intrinsics</a>
5068</div>
5069
5070<div class="doc_text">
5071<p>
5072LLVM provides intrinsics for a few important standard C library functions.
5073These intrinsics allow source-language front-ends to pass information about the
5074alignment of the pointer arguments to the code generator, providing opportunity
5075for more efficient code generation.
5076</p>
5077
5078</div>
5079
5080<!-- _______________________________________________________________________ -->
5081<div class="doc_subsubsection">
5082 <a name="int_memcpy">'<tt>llvm.memcpy</tt>' Intrinsic</a>
5083</div>
5084
5085<div class="doc_text">
5086
5087<h5>Syntax:</h5>
Chris Lattner82c2e432008-11-21 16:42:48 +0000[diff] [blame]5088<p>This is an overloaded intrinsic. You can use llvm.memcpy on any integer bit
5089width. Not all targets support all bit widths however.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5090<pre>
Chris Lattner82c2e432008-11-21 16:42:48 +0000[diff] [blame]5091 declare void @llvm.memcpy.i8(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5092 i8 &lt;len&gt;, i32 &lt;align&gt;)
5093 declare void @llvm.memcpy.i16(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5094 i16 &lt;len&gt;, i32 &lt;align&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5095 declare void @llvm.memcpy.i32(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5096 i32 &lt;len&gt;, i32 &lt;align&gt;)
5097 declare void @llvm.memcpy.i64(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5098 i64 &lt;len&gt;, i32 &lt;align&gt;)
5099</pre>
5100
5101<h5>Overview:</h5>
5102
5103<p>
5104The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the source
5105location to the destination location.
5106</p>
5107
5108<p>
5109Note that, unlike the standard libc function, the <tt>llvm.memcpy.*</tt>
5110intrinsics do not return a value, and takes an extra alignment argument.
5111</p>
5112
5113<h5>Arguments:</h5>
5114
5115<p>
5116The first argument is a pointer to the destination, the second is a pointer to
5117the source. The third argument is an integer argument
5118specifying the number of bytes to copy, and the fourth argument is the alignment
5119of the source and destination locations.
5120</p>
5121
5122<p>
5123If the call to this intrinisic has an alignment value that is not 0 or 1, then
5124the caller guarantees that both the source and destination pointers are aligned
5125to that boundary.
5126</p>
5127
5128<h5>Semantics:</h5>
5129
5130<p>
5131The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the source
5132location to the destination location, which are not allowed to overlap. It
5133copies "len" bytes of memory over. If the argument is known to be aligned to
5134some boundary, this can be specified as the fourth argument, otherwise it should
5135be set to 0 or 1.
5136</p>
5137</div>
5138
5139
5140<!-- _______________________________________________________________________ -->
5141<div class="doc_subsubsection">
5142 <a name="int_memmove">'<tt>llvm.memmove</tt>' Intrinsic</a>
5143</div>
5144
5145<div class="doc_text">
5146
5147<h5>Syntax:</h5>
Chris Lattner82c2e432008-11-21 16:42:48 +0000[diff] [blame]5148<p>This is an overloaded intrinsic. You can use llvm.memmove on any integer bit
5149width. Not all targets support all bit widths however.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5150<pre>
Chris Lattner82c2e432008-11-21 16:42:48 +0000[diff] [blame]5151 declare void @llvm.memmove.i8(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5152 i8 &lt;len&gt;, i32 &lt;align&gt;)
5153 declare void @llvm.memmove.i16(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5154 i16 &lt;len&gt;, i32 &lt;align&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5155 declare void @llvm.memmove.i32(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5156 i32 &lt;len&gt;, i32 &lt;align&gt;)
5157 declare void @llvm.memmove.i64(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5158 i64 &lt;len&gt;, i32 &lt;align&gt;)
5159</pre>
5160
5161<h5>Overview:</h5>
5162
5163<p>
5164The '<tt>llvm.memmove.*</tt>' intrinsics move a block of memory from the source
5165location to the destination location. It is similar to the
Chris Lattnerdba16ea2008-01-06 19:51:52 +0000[diff] [blame]5166'<tt>llvm.memcpy</tt>' intrinsic but allows the two memory locations to overlap.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5167</p>
5168
5169<p>
5170Note that, unlike the standard libc function, the <tt>llvm.memmove.*</tt>
5171intrinsics do not return a value, and takes an extra alignment argument.
5172</p>
5173
5174<h5>Arguments:</h5>
5175
5176<p>
5177The first argument is a pointer to the destination, the second is a pointer to
5178the source. The third argument is an integer argument
5179specifying the number of bytes to copy, and the fourth argument is the alignment
5180of the source and destination locations.
5181</p>
5182
5183<p>
5184If the call to this intrinisic has an alignment value that is not 0 or 1, then
5185the caller guarantees that the source and destination pointers are aligned to
5186that boundary.
5187</p>
5188
5189<h5>Semantics:</h5>
5190
5191<p>
5192The '<tt>llvm.memmove.*</tt>' intrinsics copy a block of memory from the source
5193location to the destination location, which may overlap. It
5194copies "len" bytes of memory over. If the argument is known to be aligned to
5195some boundary, this can be specified as the fourth argument, otherwise it should
5196be set to 0 or 1.
5197</p>
5198</div>
5199
5200
5201<!-- _______________________________________________________________________ -->
5202<div class="doc_subsubsection">
5203 <a name="int_memset">'<tt>llvm.memset.*</tt>' Intrinsics</a>
5204</div>
5205
5206<div class="doc_text">
5207
5208<h5>Syntax:</h5>
Chris Lattner82c2e432008-11-21 16:42:48 +0000[diff] [blame]5209<p>This is an overloaded intrinsic. You can use llvm.memset on any integer bit
5210width. Not all targets support all bit widths however.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5211<pre>
Chris Lattner82c2e432008-11-21 16:42:48 +0000[diff] [blame]5212 declare void @llvm.memset.i8(i8 * &lt;dest&gt;, i8 &lt;val&gt;,
5213 i8 &lt;len&gt;, i32 &lt;align&gt;)
5214 declare void @llvm.memset.i16(i8 * &lt;dest&gt;, i8 &lt;val&gt;,
5215 i16 &lt;len&gt;, i32 &lt;align&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5216 declare void @llvm.memset.i32(i8 * &lt;dest&gt;, i8 &lt;val&gt;,
5217 i32 &lt;len&gt;, i32 &lt;align&gt;)
5218 declare void @llvm.memset.i64(i8 * &lt;dest&gt;, i8 &lt;val&gt;,
5219 i64 &lt;len&gt;, i32 &lt;align&gt;)
5220</pre>
5221
5222<h5>Overview:</h5>
5223
5224<p>
5225The '<tt>llvm.memset.*</tt>' intrinsics fill a block of memory with a particular
5226byte value.
5227</p>
5228
5229<p>
5230Note that, unlike the standard libc function, the <tt>llvm.memset</tt> intrinsic
5231does not return a value, and takes an extra alignment argument.
5232</p>
5233
5234<h5>Arguments:</h5>
5235
5236<p>
5237The first argument is a pointer to the destination to fill, the second is the
5238byte value to fill it with, the third argument is an integer
5239argument specifying the number of bytes to fill, and the fourth argument is the
5240known alignment of destination location.
5241</p>
5242
5243<p>
5244If the call to this intrinisic has an alignment value that is not 0 or 1, then
5245the caller guarantees that the destination pointer is aligned to that boundary.
5246</p>
5247
5248<h5>Semantics:</h5>
5249
5250<p>
5251The '<tt>llvm.memset.*</tt>' intrinsics fill "len" bytes of memory starting at
5252the
5253destination location. If the argument is known to be aligned to some boundary,
5254this can be specified as the fourth argument, otherwise it should be set to 0 or
52551.
5256</p>
5257</div>
5258
5259
5260<!-- _______________________________________________________________________ -->
5261<div class="doc_subsubsection">
5262 <a name="int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a>
5263</div>
5264
5265<div class="doc_text">
5266
5267<h5>Syntax:</h5>
Dale Johannesenf9adbb62007-10-02 17:47:38 +0000[diff] [blame]5268<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]5269floating point or vector of floating point type. Not all targets support all
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5270types however.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5271<pre>
Dale Johannesenf9adbb62007-10-02 17:47:38 +0000[diff] [blame]5272 declare float @llvm.sqrt.f32(float %Val)
5273 declare double @llvm.sqrt.f64(double %Val)
5274 declare x86_fp80 @llvm.sqrt.f80(x86_fp80 %Val)
5275 declare fp128 @llvm.sqrt.f128(fp128 %Val)
5276 declare ppc_fp128 @llvm.sqrt.ppcf128(ppc_fp128 %Val)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5277</pre>
5278
5279<h5>Overview:</h5>
5280
5281<p>
5282The '<tt>llvm.sqrt</tt>' intrinsics return the sqrt of the specified operand,
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5283returning the same value as the libm '<tt>sqrt</tt>' functions would. Unlike
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5284<tt>sqrt</tt> in libm, however, <tt>llvm.sqrt</tt> has undefined behavior for
Chris Lattnerf914a002008-01-29 07:00:44 +0000[diff] [blame]5285negative numbers other than -0.0 (which allows for better optimization, because
5286there is no need to worry about errno being set). <tt>llvm.sqrt(-0.0)</tt> is
5287defined to return -0.0 like IEEE sqrt.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5288</p>
5289
5290<h5>Arguments:</h5>
5291
5292<p>
5293The argument and return value are floating point numbers of the same type.
5294</p>
5295
5296<h5>Semantics:</h5>
5297
5298<p>
5299This function returns the sqrt of the specified operand if it is a nonnegative
5300floating point number.
5301</p>
5302</div>
5303
5304<!-- _______________________________________________________________________ -->
5305<div class="doc_subsubsection">
5306 <a name="int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a>
5307</div>
5308
5309<div class="doc_text">
5310
5311<h5>Syntax:</h5>
Dale Johannesenf9adbb62007-10-02 17:47:38 +0000[diff] [blame]5312<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]5313floating point or vector of floating point type. Not all targets support all
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5314types however.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5315<pre>
Dale Johannesenf9adbb62007-10-02 17:47:38 +0000[diff] [blame]5316 declare float @llvm.powi.f32(float %Val, i32 %power)
5317 declare double @llvm.powi.f64(double %Val, i32 %power)
5318 declare x86_fp80 @llvm.powi.f80(x86_fp80 %Val, i32 %power)
5319 declare fp128 @llvm.powi.f128(fp128 %Val, i32 %power)
5320 declare ppc_fp128 @llvm.powi.ppcf128(ppc_fp128 %Val, i32 %power)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5321</pre>
5322
5323<h5>Overview:</h5>
5324
5325<p>
5326The '<tt>llvm.powi.*</tt>' intrinsics return the first operand raised to the
5327specified (positive or negative) power. The order of evaluation of
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5328multiplications is not defined. When a vector of floating point type is
5329used, the second argument remains a scalar integer value.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5330</p>
5331
5332<h5>Arguments:</h5>
5333
5334<p>
5335The second argument is an integer power, and the first is a value to raise to
5336that power.
5337</p>
5338
5339<h5>Semantics:</h5>
5340
5341<p>
5342This function returns the first value raised to the second power with an
5343unspecified sequence of rounding operations.</p>
5344</div>
5345
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5346<!-- _______________________________________________________________________ -->
5347<div class="doc_subsubsection">
5348 <a name="int_sin">'<tt>llvm.sin.*</tt>' Intrinsic</a>
5349</div>
5350
5351<div class="doc_text">
5352
5353<h5>Syntax:</h5>
5354<p>This is an overloaded intrinsic. You can use <tt>llvm.sin</tt> on any
5355floating point or vector of floating point type. Not all targets support all
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5356types however.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5357<pre>
5358 declare float @llvm.sin.f32(float %Val)
5359 declare double @llvm.sin.f64(double %Val)
5360 declare x86_fp80 @llvm.sin.f80(x86_fp80 %Val)
5361 declare fp128 @llvm.sin.f128(fp128 %Val)
5362 declare ppc_fp128 @llvm.sin.ppcf128(ppc_fp128 %Val)
5363</pre>
5364
5365<h5>Overview:</h5>
5366
5367<p>
5368The '<tt>llvm.sin.*</tt>' intrinsics return the sine of the operand.
5369</p>
5370
5371<h5>Arguments:</h5>
5372
5373<p>
5374The argument and return value are floating point numbers of the same type.
5375</p>
5376
5377<h5>Semantics:</h5>
5378
5379<p>
5380This function returns the sine of the specified operand, returning the
5381same values as the libm <tt>sin</tt> functions would, and handles error
Dan Gohmaneaba92e2007-10-17 18:05:13 +0000[diff] [blame]5382conditions in the same way.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5383</div>
5384
5385<!-- _______________________________________________________________________ -->
5386<div class="doc_subsubsection">
5387 <a name="int_cos">'<tt>llvm.cos.*</tt>' Intrinsic</a>
5388</div>
5389
5390<div class="doc_text">
5391
5392<h5>Syntax:</h5>
5393<p>This is an overloaded intrinsic. You can use <tt>llvm.cos</tt> on any
5394floating point or vector of floating point type. Not all targets support all
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5395types however.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5396<pre>
5397 declare float @llvm.cos.f32(float %Val)
5398 declare double @llvm.cos.f64(double %Val)
5399 declare x86_fp80 @llvm.cos.f80(x86_fp80 %Val)
5400 declare fp128 @llvm.cos.f128(fp128 %Val)
5401 declare ppc_fp128 @llvm.cos.ppcf128(ppc_fp128 %Val)
5402</pre>
5403
5404<h5>Overview:</h5>
5405
5406<p>
5407The '<tt>llvm.cos.*</tt>' intrinsics return the cosine of the operand.
5408</p>
5409
5410<h5>Arguments:</h5>
5411
5412<p>
5413The argument and return value are floating point numbers of the same type.
5414</p>
5415
5416<h5>Semantics:</h5>
5417
5418<p>
5419This function returns the cosine of the specified operand, returning the
5420same values as the libm <tt>cos</tt> functions would, and handles error
Dan Gohmaneaba92e2007-10-17 18:05:13 +0000[diff] [blame]5421conditions in the same way.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5422</div>
5423
5424<!-- _______________________________________________________________________ -->
5425<div class="doc_subsubsection">
5426 <a name="int_pow">'<tt>llvm.pow.*</tt>' Intrinsic</a>
5427</div>
5428
5429<div class="doc_text">
5430
5431<h5>Syntax:</h5>
5432<p>This is an overloaded intrinsic. You can use <tt>llvm.pow</tt> on any
5433floating point or vector of floating point type. Not all targets support all
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5434types however.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5435<pre>
5436 declare float @llvm.pow.f32(float %Val, float %Power)
5437 declare double @llvm.pow.f64(double %Val, double %Power)
5438 declare x86_fp80 @llvm.pow.f80(x86_fp80 %Val, x86_fp80 %Power)
5439 declare fp128 @llvm.pow.f128(fp128 %Val, fp128 %Power)
5440 declare ppc_fp128 @llvm.pow.ppcf128(ppc_fp128 %Val, ppc_fp128 Power)
5441</pre>
5442
5443<h5>Overview:</h5>
5444
5445<p>
5446The '<tt>llvm.pow.*</tt>' intrinsics return the first operand raised to the
5447specified (positive or negative) power.
5448</p>
5449
5450<h5>Arguments:</h5>
5451
5452<p>
5453The second argument is a floating point power, and the first is a value to
5454raise to that power.
5455</p>
5456
5457<h5>Semantics:</h5>
5458
5459<p>
5460This function returns the first value raised to the second power,
5461returning the
5462same values as the libm <tt>pow</tt> functions would, and handles error
Dan Gohmaneaba92e2007-10-17 18:05:13 +0000[diff] [blame]5463conditions in the same way.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5464</div>
5465
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5466
5467<!-- ======================================================================= -->
5468<div class="doc_subsection">
5469 <a name="int_manip">Bit Manipulation Intrinsics</a>
5470</div>
5471
5472<div class="doc_text">
5473<p>
5474LLVM provides intrinsics for a few important bit manipulation operations.
5475These allow efficient code generation for some algorithms.
5476</p>
5477
5478</div>
5479
5480<!-- _______________________________________________________________________ -->
5481<div class="doc_subsubsection">
5482 <a name="int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a>
5483</div>
5484
5485<div class="doc_text">
5486
5487<h5>Syntax:</h5>
5488<p>This is an overloaded intrinsic function. You can use bswap on any integer
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5489type that is an even number of bytes (i.e. BitWidth % 16 == 0).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5490<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5491 declare i16 @llvm.bswap.i16(i16 &lt;id&gt;)
5492 declare i32 @llvm.bswap.i32(i32 &lt;id&gt;)
5493 declare i64 @llvm.bswap.i64(i64 &lt;id&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5494</pre>
5495
5496<h5>Overview:</h5>
5497
5498<p>
5499The '<tt>llvm.bswap</tt>' family of intrinsics is used to byte swap integer
5500values with an even number of bytes (positive multiple of 16 bits). These are
5501useful for performing operations on data that is not in the target's native
5502byte order.
5503</p>
5504
5505<h5>Semantics:</h5>
5506
5507<p>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5508The <tt>llvm.bswap.i16</tt> intrinsic returns an i16 value that has the high
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5509and low byte of the input i16 swapped. Similarly, the <tt>llvm.bswap.i32</tt>
5510intrinsic returns an i32 value that has the four bytes of the input i32
5511swapped, 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]5512i32 will have its bytes in 3, 2, 1, 0 order. The <tt>llvm.bswap.i48</tt>,
5513<tt>llvm.bswap.i64</tt> and other intrinsics extend this concept to
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5514additional even-byte lengths (6 bytes, 8 bytes and more, respectively).
5515</p>
5516
5517</div>
5518
5519<!-- _______________________________________________________________________ -->
5520<div class="doc_subsubsection">
5521 <a name="int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic</a>
5522</div>
5523
5524<div class="doc_text">
5525
5526<h5>Syntax:</h5>
5527<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]5528width. Not all targets support all bit widths however.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5529<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5530 declare i8 @llvm.ctpop.i8 (i8 &lt;src&gt;)
5531 declare i16 @llvm.ctpop.i16(i16 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5532 declare i32 @llvm.ctpop.i32(i32 &lt;src&gt;)
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5533 declare i64 @llvm.ctpop.i64(i64 &lt;src&gt;)
5534 declare i256 @llvm.ctpop.i256(i256 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5535</pre>
5536
5537<h5>Overview:</h5>
5538
5539<p>
5540The '<tt>llvm.ctpop</tt>' family of intrinsics counts the number of bits set in a
5541value.
5542</p>
5543
5544<h5>Arguments:</h5>
5545
5546<p>
5547The only argument is the value to be counted. The argument may be of any
5548integer type. The return type must match the argument type.
5549</p>
5550
5551<h5>Semantics:</h5>
5552
5553<p>
5554The '<tt>llvm.ctpop</tt>' intrinsic counts the 1's in a variable.
5555</p>
5556</div>
5557
5558<!-- _______________________________________________________________________ -->
5559<div class="doc_subsubsection">
5560 <a name="int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic</a>
5561</div>
5562
5563<div class="doc_text">
5564
5565<h5>Syntax:</h5>
5566<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]5567integer bit width. Not all targets support all bit widths however.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5568<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5569 declare i8 @llvm.ctlz.i8 (i8 &lt;src&gt;)
5570 declare i16 @llvm.ctlz.i16(i16 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5571 declare i32 @llvm.ctlz.i32(i32 &lt;src&gt;)
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5572 declare i64 @llvm.ctlz.i64(i64 &lt;src&gt;)
5573 declare i256 @llvm.ctlz.i256(i256 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5574</pre>
5575
5576<h5>Overview:</h5>
5577
5578<p>
5579The '<tt>llvm.ctlz</tt>' family of intrinsic functions counts the number of
5580leading zeros in a variable.
5581</p>
5582
5583<h5>Arguments:</h5>
5584
5585<p>
5586The only argument is the value to be counted. The argument may be of any
5587integer type. The return type must match the argument type.
5588</p>
5589
5590<h5>Semantics:</h5>
5591
5592<p>
5593The '<tt>llvm.ctlz</tt>' intrinsic counts the leading (most significant) zeros
5594in a variable. If the src == 0 then the result is the size in bits of the type
5595of src. For example, <tt>llvm.ctlz(i32 2) = 30</tt>.
5596</p>
5597</div>
5598
5599
5600
5601<!-- _______________________________________________________________________ -->
5602<div class="doc_subsubsection">
5603 <a name="int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic</a>
5604</div>
5605
5606<div class="doc_text">
5607
5608<h5>Syntax:</h5>
5609<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]5610integer bit width. Not all targets support all bit widths however.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5611<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5612 declare i8 @llvm.cttz.i8 (i8 &lt;src&gt;)
5613 declare i16 @llvm.cttz.i16(i16 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5614 declare i32 @llvm.cttz.i32(i32 &lt;src&gt;)
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5615 declare i64 @llvm.cttz.i64(i64 &lt;src&gt;)
5616 declare i256 @llvm.cttz.i256(i256 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5617</pre>
5618
5619<h5>Overview:</h5>
5620
5621<p>
5622The '<tt>llvm.cttz</tt>' family of intrinsic functions counts the number of
5623trailing zeros.
5624</p>
5625
5626<h5>Arguments:</h5>
5627
5628<p>
5629The only argument is the value to be counted. The argument may be of any
5630integer type. The return type must match the argument type.
5631</p>
5632
5633<h5>Semantics:</h5>
5634
5635<p>
5636The '<tt>llvm.cttz</tt>' intrinsic counts the trailing (least significant) zeros
5637in a variable. If the src == 0 then the result is the size in bits of the type
5638of src. For example, <tt>llvm.cttz(2) = 1</tt>.
5639</p>
5640</div>
5641
5642<!-- _______________________________________________________________________ -->
5643<div class="doc_subsubsection">
5644 <a name="int_part_select">'<tt>llvm.part.select.*</tt>' Intrinsic</a>
5645</div>
5646
5647<div class="doc_text">
5648
5649<h5>Syntax:</h5>
5650<p>This is an overloaded intrinsic. You can use <tt>llvm.part.select</tt>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5651on any integer bit width.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5652<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5653 declare i17 @llvm.part.select.i17 (i17 %val, i32 %loBit, i32 %hiBit)
5654 declare i29 @llvm.part.select.i29 (i29 %val, i32 %loBit, i32 %hiBit)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5655</pre>
5656
5657<h5>Overview:</h5>
5658<p>The '<tt>llvm.part.select</tt>' family of intrinsic functions selects a
5659range of bits from an integer value and returns them in the same bit width as
5660the original value.</p>
5661
5662<h5>Arguments:</h5>
5663<p>The first argument, <tt>%val</tt> and the result may be integer types of
5664any bit width but they must have the same bit width. The second and third
5665arguments must be <tt>i32</tt> type since they specify only a bit index.</p>
5666
5667<h5>Semantics:</h5>
5668<p>The operation of the '<tt>llvm.part.select</tt>' intrinsic has two modes
5669of operation: forwards and reverse. If <tt>%loBit</tt> is greater than
5670<tt>%hiBits</tt> then the intrinsic operates in reverse mode. Otherwise it
5671operates in forward mode.</p>
5672<p>In forward mode, this intrinsic is the equivalent of shifting <tt>%val</tt>
5673right by <tt>%loBit</tt> bits and then ANDing it with a mask with
5674only the <tt>%hiBit - %loBit</tt> bits set, as follows:</p>
5675<ol>
5676 <li>The <tt>%val</tt> is shifted right (LSHR) by the number of bits specified
5677 by <tt>%loBits</tt>. This normalizes the value to the low order bits.</li>
5678 <li>The <tt>%loBits</tt> value is subtracted from the <tt>%hiBits</tt> value
5679 to determine the number of bits to retain.</li>
5680 <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]5681 <li>The mask is ANDed with <tt>%val</tt> to produce the result.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5682</ol>
5683<p>In reverse mode, a similar computation is made except that the bits are
5684returned in the reverse order. So, for example, if <tt>X</tt> has the value
5685<tt>i16 0x0ACF (101011001111)</tt> and we apply
5686<tt>part.select(i16 X, 8, 3)</tt> to it, we get back the value
5687<tt>i16 0x0026 (000000100110)</tt>.</p>
5688</div>
5689
5690<div class="doc_subsubsection">
5691 <a name="int_part_set">'<tt>llvm.part.set.*</tt>' Intrinsic</a>
5692</div>
5693
5694<div class="doc_text">
5695
5696<h5>Syntax:</h5>
5697<p>This is an overloaded intrinsic. You can use <tt>llvm.part.set</tt>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5698on any integer bit width.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5699<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5700 declare i17 @llvm.part.set.i17.i9 (i17 %val, i9 %repl, i32 %lo, i32 %hi)
5701 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]5702</pre>
5703
5704<h5>Overview:</h5>
5705<p>The '<tt>llvm.part.set</tt>' family of intrinsic functions replaces a range
5706of bits in an integer value with another integer value. It returns the integer
5707with the replaced bits.</p>
5708
5709<h5>Arguments:</h5>
5710<p>The first argument, <tt>%val</tt> and the result may be integer types of
5711any bit width but they must have the same bit width. <tt>%val</tt> is the value
5712whose bits will be replaced. The second argument, <tt>%repl</tt> may be an
5713integer of any bit width. The third and fourth arguments must be <tt>i32</tt>
5714type since they specify only a bit index.</p>
5715
5716<h5>Semantics:</h5>
5717<p>The operation of the '<tt>llvm.part.set</tt>' intrinsic has two modes
5718of operation: forwards and reverse. If <tt>%lo</tt> is greater than
5719<tt>%hi</tt> then the intrinsic operates in reverse mode. Otherwise it
5720operates in forward mode.</p>
5721<p>For both modes, the <tt>%repl</tt> value is prepared for use by either
5722truncating it down to the size of the replacement area or zero extending it
5723up to that size.</p>
5724<p>In forward mode, the bits between <tt>%lo</tt> and <tt>%hi</tt> (inclusive)
5725are replaced with corresponding bits from <tt>%repl</tt>. That is the 0th bit
5726in <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]5727to the <tt>%hi</tt>th bit.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5728<p>In reverse mode, a similar computation is made except that the bits are
5729reversed. That is, the <tt>0</tt>th bit in <tt>%repl</tt> replaces the
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5730<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]5731<h5>Examples:</h5>
5732<pre>
5733 llvm.part.set(0xFFFF, 0, 4, 7) -&gt; 0xFF0F
5734 llvm.part.set(0xFFFF, 0, 7, 4) -&gt; 0xFF0F
5735 llvm.part.set(0xFFFF, 1, 7, 4) -&gt; 0xFF8F
5736 llvm.part.set(0xFFFF, F, 8, 3) -&gt; 0xFFE7
5737 llvm.part.set(0xFFFF, 0, 3, 8) -&gt; 0xFE07
5738</pre>
5739</div>
5740
5741<!-- ======================================================================= -->
5742<div class="doc_subsection">
5743 <a name="int_debugger">Debugger Intrinsics</a>
5744</div>
5745
5746<div class="doc_text">
5747<p>
5748The LLVM debugger intrinsics (which all start with <tt>llvm.dbg.</tt> prefix),
5749are described in the <a
5750href="SourceLevelDebugging.html#format_common_intrinsics">LLVM Source Level
5751Debugging</a> document.
5752</p>
5753</div>
5754
5755
5756<!-- ======================================================================= -->
5757<div class="doc_subsection">
5758 <a name="int_eh">Exception Handling Intrinsics</a>
5759</div>
5760
5761<div class="doc_text">
5762<p> The LLVM exception handling intrinsics (which all start with
5763<tt>llvm.eh.</tt> prefix), are described in the <a
5764href="ExceptionHandling.html#format_common_intrinsics">LLVM Exception
5765Handling</a> document. </p>
5766</div>
5767
5768<!-- ======================================================================= -->
5769<div class="doc_subsection">
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5770 <a name="int_trampoline">Trampoline Intrinsic</a>
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5771</div>
5772
5773<div class="doc_text">
5774<p>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5775 This intrinsic makes it possible to excise one parameter, marked with
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5776 the <tt>nest</tt> attribute, from a function. The result is a callable
5777 function pointer lacking the nest parameter - the caller does not need
5778 to provide a value for it. Instead, the value to use is stored in
5779 advance in a "trampoline", a block of memory usually allocated
5780 on the stack, which also contains code to splice the nest value into the
5781 argument list. This is used to implement the GCC nested function address
5782 extension.
5783</p>
5784<p>
5785 For example, if the function is
5786 <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]5787 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]5788<pre>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5789 %tramp = alloca [10 x i8], align 4 ; size and alignment only correct for X86
5790 %tramp1 = getelementptr [10 x i8]* %tramp, i32 0, i32 0
5791 %p = call i8* @llvm.init.trampoline( i8* %tramp1, i8* bitcast (i32 (i8* nest , i32, i32)* @f to i8*), i8* %nval )
5792 %fp = bitcast i8* %p to i32 (i32, i32)*
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5793</pre>
Bill Wendlinge262b4c2007-09-22 09:23:55 +0000[diff] [blame]5794 <p>The call <tt>%val = call i32 %fp( i32 %x, i32 %y )</tt> is then equivalent
5795 to <tt>%val = call i32 %f( i8* %nval, i32 %x, i32 %y )</tt>.</p>
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5796</div>
5797
5798<!-- _______________________________________________________________________ -->
5799<div class="doc_subsubsection">
5800 <a name="int_it">'<tt>llvm.init.trampoline</tt>' Intrinsic</a>
5801</div>
5802<div class="doc_text">
5803<h5>Syntax:</h5>
5804<pre>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5805declare 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]5806</pre>
5807<h5>Overview:</h5>
5808<p>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5809 This fills the memory pointed to by <tt>tramp</tt> with code
5810 and returns a function pointer suitable for executing it.
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5811</p>
5812<h5>Arguments:</h5>
5813<p>
5814 The <tt>llvm.init.trampoline</tt> intrinsic takes three arguments, all
5815 pointers. The <tt>tramp</tt> argument must point to a sufficiently large
5816 and sufficiently aligned block of memory; this memory is written to by the
Duncan Sands35012212007-08-22 23:39:54 +0000[diff] [blame]5817 intrinsic. Note that the size and the alignment are target-specific - LLVM
5818 currently provides no portable way of determining them, so a front-end that
5819 generates this intrinsic needs to have some target-specific knowledge.
5820 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]5821</p>
5822<h5>Semantics:</h5>
5823<p>
5824 The block of memory pointed to by <tt>tramp</tt> is filled with target
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5825 dependent code, turning it into a function. A pointer to this function is
5826 returned, but needs to be bitcast to an
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5827 <a href="#int_trampoline">appropriate function pointer type</a>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5828 before being called. The new function's signature is the same as that of
5829 <tt>func</tt> with any arguments marked with the <tt>nest</tt> attribute
5830 removed. At most one such <tt>nest</tt> argument is allowed, and it must be
5831 of pointer type. Calling the new function is equivalent to calling
5832 <tt>func</tt> with the same argument list, but with <tt>nval</tt> used for the
5833 missing <tt>nest</tt> argument. If, after calling
5834 <tt>llvm.init.trampoline</tt>, the memory pointed to by <tt>tramp</tt> is
5835 modified, then the effect of any later call to the returned function pointer is
5836 undefined.
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5837</p>
5838</div>
5839
5840<!-- ======================================================================= -->
5841<div class="doc_subsection">
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]5842 <a name="int_atomics">Atomic Operations and Synchronization Intrinsics</a>
5843</div>
5844
5845<div class="doc_text">
5846<p>
5847 These intrinsic functions expand the "universal IR" of LLVM to represent
5848 hardware constructs for atomic operations and memory synchronization. This
5849 provides an interface to the hardware, not an interface to the programmer. It
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]5850 is aimed at a low enough level to allow any programming models or APIs
5851 (Application Programming Interfaces) which
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]5852 need atomic behaviors to map cleanly onto it. It is also modeled primarily on
5853 hardware behavior. Just as hardware provides a "universal IR" for source
5854 languages, it also provides a starting point for developing a "universal"
5855 atomic operation and synchronization IR.
5856</p>
5857<p>
5858 These do <em>not</em> form an API such as high-level threading libraries,
5859 software transaction memory systems, atomic primitives, and intrinsic
5860 functions as found in BSD, GNU libc, atomic_ops, APR, and other system and
5861 application libraries. The hardware interface provided by LLVM should allow
5862 a clean implementation of all of these APIs and parallel programming models.
5863 No one model or paradigm should be selected above others unless the hardware
5864 itself ubiquitously does so.
5865
5866</p>
5867</div>
5868
5869<!-- _______________________________________________________________________ -->
5870<div class="doc_subsubsection">
5871 <a name="int_memory_barrier">'<tt>llvm.memory.barrier</tt>' Intrinsic</a>
5872</div>
5873<div class="doc_text">
5874<h5>Syntax:</h5>
5875<pre>
5876declare void @llvm.memory.barrier( i1 &lt;ll&gt;, i1 &lt;ls&gt;, i1 &lt;sl&gt;, i1 &lt;ss&gt;,
5877i1 &lt;device&gt; )
5878
5879</pre>
5880<h5>Overview:</h5>
5881<p>
5882 The <tt>llvm.memory.barrier</tt> intrinsic guarantees ordering between
5883 specific pairs of memory access types.
5884</p>
5885<h5>Arguments:</h5>
5886<p>
5887 The <tt>llvm.memory.barrier</tt> intrinsic requires five boolean arguments.
5888 The first four arguments enables a specific barrier as listed below. The fith
5889 argument specifies that the barrier applies to io or device or uncached memory.
5890
5891</p>
5892 <ul>
5893 <li><tt>ll</tt>: load-load barrier</li>
5894 <li><tt>ls</tt>: load-store barrier</li>
5895 <li><tt>sl</tt>: store-load barrier</li>
5896 <li><tt>ss</tt>: store-store barrier</li>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5897 <li><tt>device</tt>: barrier applies to device and uncached memory also.</li>
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]5898 </ul>
5899<h5>Semantics:</h5>
5900<p>
5901 This intrinsic causes the system to enforce some ordering constraints upon
5902 the loads and stores of the program. This barrier does not indicate
5903 <em>when</em> any events will occur, it only enforces an <em>order</em> in
5904 which they occur. For any of the specified pairs of load and store operations
5905 (f.ex. load-load, or store-load), all of the first operations preceding the
5906 barrier will complete before any of the second operations succeeding the
5907 barrier begin. Specifically the semantics for each pairing is as follows:
5908</p>
5909 <ul>
5910 <li><tt>ll</tt>: All loads before the barrier must complete before any load
5911 after the barrier begins.</li>
5912
5913 <li><tt>ls</tt>: All loads before the barrier must complete before any
5914 store after the barrier begins.</li>
5915 <li><tt>ss</tt>: All stores before the barrier must complete before any
5916 store after the barrier begins.</li>
5917 <li><tt>sl</tt>: All stores before the barrier must complete before any
5918 load after the barrier begins.</li>
5919 </ul>
5920<p>
5921 These semantics are applied with a logical "and" behavior when more than one
5922 is enabled in a single memory barrier intrinsic.
5923</p>
5924<p>
5925 Backends may implement stronger barriers than those requested when they do not
5926 support as fine grained a barrier as requested. Some architectures do not
5927 need all types of barriers and on such architectures, these become noops.
5928</p>
5929<h5>Example:</h5>
5930<pre>
5931%ptr = malloc i32
5932 store i32 4, %ptr
5933
5934%result1 = load i32* %ptr <i>; yields {i32}:result1 = 4</i>
5935 call void @llvm.memory.barrier( i1 false, i1 true, i1 false, i1 false )
5936 <i>; guarantee the above finishes</i>
5937 store i32 8, %ptr <i>; before this begins</i>
5938</pre>
5939</div>
5940
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]5941<!-- _______________________________________________________________________ -->
5942<div class="doc_subsubsection">
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]5943 <a name="int_atomic_cmp_swap">'<tt>llvm.atomic.cmp.swap.*</tt>' Intrinsic</a>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]5944</div>
5945<div class="doc_text">
5946<h5>Syntax:</h5>
5947<p>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]5948 This is an overloaded intrinsic. You can use <tt>llvm.atomic.cmp.swap</tt> on
5949 any integer bit width and for different address spaces. Not all targets
5950 support all bit widths however.</p>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]5951
5952<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]5953declare i8 @llvm.atomic.cmp.swap.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;cmp&gt;, i8 &lt;val&gt; )
5954declare i16 @llvm.atomic.cmp.swap.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;cmp&gt;, i16 &lt;val&gt; )
5955declare i32 @llvm.atomic.cmp.swap.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;cmp&gt;, i32 &lt;val&gt; )
5956declare 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]5957
5958</pre>
5959<h5>Overview:</h5>
5960<p>
5961 This loads a value in memory and compares it to a given value. If they are
5962 equal, it stores a new value into the memory.
5963</p>
5964<h5>Arguments:</h5>
5965<p>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]5966 The <tt>llvm.atomic.cmp.swap</tt> intrinsic takes three arguments. The result as
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]5967 well as both <tt>cmp</tt> and <tt>val</tt> must be integer values with the
5968 same bit width. The <tt>ptr</tt> argument must be a pointer to a value of
5969 this integer type. While any bit width integer may be used, targets may only
5970 lower representations they support in hardware.
5971
5972</p>
5973<h5>Semantics:</h5>
5974<p>
5975 This entire intrinsic must be executed atomically. It first loads the value
5976 in memory pointed to by <tt>ptr</tt> and compares it with the value
5977 <tt>cmp</tt>. If they are equal, <tt>val</tt> is stored into the memory. The
5978 loaded value is yielded in all cases. This provides the equivalent of an
5979 atomic compare-and-swap operation within the SSA framework.
5980</p>
5981<h5>Examples:</h5>
5982
5983<pre>
5984%ptr = malloc i32
5985 store i32 4, %ptr
5986
5987%val1 = add i32 4, 4
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]5988%result1 = call i32 @llvm.atomic.cmp.swap.i32.p0i32( i32* %ptr, i32 4, %val1 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]5989 <i>; yields {i32}:result1 = 4</i>
5990%stored1 = icmp eq i32 %result1, 4 <i>; yields {i1}:stored1 = true</i>
5991%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 8</i>
5992
5993%val2 = add i32 1, 1
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]5994%result2 = call i32 @llvm.atomic.cmp.swap.i32.p0i32( i32* %ptr, i32 5, %val2 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]5995 <i>; yields {i32}:result2 = 8</i>
5996%stored2 = icmp eq i32 %result2, 5 <i>; yields {i1}:stored2 = false</i>
5997
5998%memval2 = load i32* %ptr <i>; yields {i32}:memval2 = 8</i>
5999</pre>
6000</div>
6001
6002<!-- _______________________________________________________________________ -->
6003<div class="doc_subsubsection">
6004 <a name="int_atomic_swap">'<tt>llvm.atomic.swap.*</tt>' Intrinsic</a>
6005</div>
6006<div class="doc_text">
6007<h5>Syntax:</h5>
6008
6009<p>
6010 This is an overloaded intrinsic. You can use <tt>llvm.atomic.swap</tt> on any
6011 integer bit width. Not all targets support all bit widths however.</p>
6012<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6013declare i8 @llvm.atomic.swap.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;val&gt; )
6014declare i16 @llvm.atomic.swap.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;val&gt; )
6015declare i32 @llvm.atomic.swap.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;val&gt; )
6016declare i64 @llvm.atomic.swap.i64.p0i64( i64* &lt;ptr&gt;, i64 &lt;val&gt; )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6017
6018</pre>
6019<h5>Overview:</h5>
6020<p>
6021 This intrinsic loads the value stored in memory at <tt>ptr</tt> and yields
6022 the value from memory. It then stores the value in <tt>val</tt> in the memory
6023 at <tt>ptr</tt>.
6024</p>
6025<h5>Arguments:</h5>
6026
6027<p>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6028 The <tt>llvm.atomic.swap</tt> intrinsic takes two arguments. Both the
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6029 <tt>val</tt> argument and the result must be integers of the same bit width.
6030 The first argument, <tt>ptr</tt>, must be a pointer to a value of this
6031 integer type. The targets may only lower integer representations they
6032 support.
6033</p>
6034<h5>Semantics:</h5>
6035<p>
6036 This intrinsic loads the value pointed to by <tt>ptr</tt>, yields it, and
6037 stores <tt>val</tt> back into <tt>ptr</tt> atomically. This provides the
6038 equivalent of an atomic swap operation within the SSA framework.
6039
6040</p>
6041<h5>Examples:</h5>
6042<pre>
6043%ptr = malloc i32
6044 store i32 4, %ptr
6045
6046%val1 = add i32 4, 4
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6047%result1 = call i32 @llvm.atomic.swap.i32.p0i32( i32* %ptr, i32 %val1 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6048 <i>; yields {i32}:result1 = 4</i>
6049%stored1 = icmp eq i32 %result1, 4 <i>; yields {i1}:stored1 = true</i>
6050%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 8</i>
6051
6052%val2 = add i32 1, 1
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6053%result2 = call i32 @llvm.atomic.swap.i32.p0i32( i32* %ptr, i32 %val2 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6054 <i>; yields {i32}:result2 = 8</i>
6055
6056%stored2 = icmp eq i32 %result2, 8 <i>; yields {i1}:stored2 = true</i>
6057%memval2 = load i32* %ptr <i>; yields {i32}:memval2 = 2</i>
6058</pre>
6059</div>
6060
6061<!-- _______________________________________________________________________ -->
6062<div class="doc_subsubsection">
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6063 <a name="int_atomic_load_add">'<tt>llvm.atomic.load.add.*</tt>' Intrinsic</a>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6064
6065</div>
6066<div class="doc_text">
6067<h5>Syntax:</h5>
6068<p>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6069 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]6070 integer bit width. Not all targets support all bit widths however.</p>
6071<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6072declare i8 @llvm.atomic.load.add.i8..p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6073declare i16 @llvm.atomic.load.add.i16..p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6074declare i32 @llvm.atomic.load.add.i32..p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6075declare 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]6076
6077</pre>
6078<h5>Overview:</h5>
6079<p>
6080 This intrinsic adds <tt>delta</tt> to the value stored in memory at
6081 <tt>ptr</tt>. It yields the original value at <tt>ptr</tt>.
6082</p>
6083<h5>Arguments:</h5>
6084<p>
6085
6086 The intrinsic takes two arguments, the first a pointer to an integer value
6087 and the second an integer value. The result is also an integer value. These
6088 integer types can have any bit width, but they must all have the same bit
6089 width. The targets may only lower integer representations they support.
6090</p>
6091<h5>Semantics:</h5>
6092<p>
6093 This intrinsic does a series of operations atomically. It first loads the
6094 value stored at <tt>ptr</tt>. It then adds <tt>delta</tt>, stores the result
6095 to <tt>ptr</tt>. It yields the original value stored at <tt>ptr</tt>.
6096</p>
6097
6098<h5>Examples:</h5>
6099<pre>
6100%ptr = malloc i32
6101 store i32 4, %ptr
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6102%result1 = call i32 @llvm.atomic.load.add.i32.p0i32( i32* %ptr, i32 4 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6103 <i>; yields {i32}:result1 = 4</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6104%result2 = call i32 @llvm.atomic.load.add.i32.p0i32( i32* %ptr, i32 2 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6105 <i>; yields {i32}:result2 = 8</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6106%result3 = call i32 @llvm.atomic.load.add.i32.p0i32( i32* %ptr, i32 5 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6107 <i>; yields {i32}:result3 = 10</i>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6108%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 15</i>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6109</pre>
6110</div>
6111
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6112<!-- _______________________________________________________________________ -->
6113<div class="doc_subsubsection">
6114 <a name="int_atomic_load_sub">'<tt>llvm.atomic.load.sub.*</tt>' Intrinsic</a>
6115
6116</div>
6117<div class="doc_text">
6118<h5>Syntax:</h5>
6119<p>
6120 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]6121 any integer bit width and for different address spaces. Not all targets
6122 support all bit widths however.</p>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6123<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6124declare i8 @llvm.atomic.load.sub.i8.p0i32( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6125declare i16 @llvm.atomic.load.sub.i16.p0i32( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6126declare i32 @llvm.atomic.load.sub.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6127declare 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]6128
6129</pre>
6130<h5>Overview:</h5>
6131<p>
6132 This intrinsic subtracts <tt>delta</tt> to the value stored in memory at
6133 <tt>ptr</tt>. It yields the original value at <tt>ptr</tt>.
6134</p>
6135<h5>Arguments:</h5>
6136<p>
6137
6138 The intrinsic takes two arguments, the first a pointer to an integer value
6139 and the second an integer value. The result is also an integer value. These
6140 integer types can have any bit width, but they must all have the same bit
6141 width. The targets may only lower integer representations they support.
6142</p>
6143<h5>Semantics:</h5>
6144<p>
6145 This intrinsic does a series of operations atomically. It first loads the
6146 value stored at <tt>ptr</tt>. It then subtracts <tt>delta</tt>, stores the
6147 result to <tt>ptr</tt>. It yields the original value stored at <tt>ptr</tt>.
6148</p>
6149
6150<h5>Examples:</h5>
6151<pre>
6152%ptr = malloc i32
6153 store i32 8, %ptr
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6154%result1 = call i32 @llvm.atomic.load.sub.i32.p0i32( i32* %ptr, i32 4 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6155 <i>; yields {i32}:result1 = 8</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6156%result2 = call i32 @llvm.atomic.load.sub.i32.p0i32( i32* %ptr, i32 2 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6157 <i>; yields {i32}:result2 = 4</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6158%result3 = call i32 @llvm.atomic.load.sub.i32.p0i32( i32* %ptr, i32 5 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6159 <i>; yields {i32}:result3 = 2</i>
6160%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = -3</i>
6161</pre>
6162</div>
6163
6164<!-- _______________________________________________________________________ -->
6165<div class="doc_subsubsection">
6166 <a name="int_atomic_load_and">'<tt>llvm.atomic.load.and.*</tt>' Intrinsic</a><br>
6167 <a name="int_atomic_load_nand">'<tt>llvm.atomic.load.nand.*</tt>' Intrinsic</a><br>
6168 <a name="int_atomic_load_or">'<tt>llvm.atomic.load.or.*</tt>' Intrinsic</a><br>
6169 <a name="int_atomic_load_xor">'<tt>llvm.atomic.load.xor.*</tt>' Intrinsic</a><br>
6170
6171</div>
6172<div class="doc_text">
6173<h5>Syntax:</h5>
6174<p>
6175 These are overloaded intrinsics. You can use <tt>llvm.atomic.load_and</tt>,
6176 <tt>llvm.atomic.load_nand</tt>, <tt>llvm.atomic.load_or</tt>, and
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6177 <tt>llvm.atomic.load_xor</tt> on any integer bit width and for different
6178 address spaces. Not all targets support all bit widths however.</p>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6179<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6180declare i8 @llvm.atomic.load.and.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6181declare i16 @llvm.atomic.load.and.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6182declare i32 @llvm.atomic.load.and.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6183declare 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]6184
6185</pre>
6186
6187<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6188declare i8 @llvm.atomic.load.or.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6189declare i16 @llvm.atomic.load.or.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6190declare i32 @llvm.atomic.load.or.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6191declare 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]6192
6193</pre>
6194
6195<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6196declare i8 @llvm.atomic.load.nand.i8.p0i32( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6197declare i16 @llvm.atomic.load.nand.i16.p0i32( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6198declare i32 @llvm.atomic.load.nand.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6199declare 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]6200
6201</pre>
6202
6203<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6204declare i8 @llvm.atomic.load.xor.i8.p0i32( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6205declare i16 @llvm.atomic.load.xor.i16.p0i32( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6206declare i32 @llvm.atomic.load.xor.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6207declare 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]6208
6209</pre>
6210<h5>Overview:</h5>
6211<p>
6212 These intrinsics bitwise the operation (and, nand, or, xor) <tt>delta</tt> to
6213 the value stored in memory at <tt>ptr</tt>. It yields the original value
6214 at <tt>ptr</tt>.
6215</p>
6216<h5>Arguments:</h5>
6217<p>
6218
6219 These intrinsics take two arguments, the first a pointer to an integer value
6220 and the second an integer value. The result is also an integer value. These
6221 integer types can have any bit width, but they must all have the same bit
6222 width. The targets may only lower integer representations they support.
6223</p>
6224<h5>Semantics:</h5>
6225<p>
6226 These intrinsics does a series of operations atomically. They first load the
6227 value stored at <tt>ptr</tt>. They then do the bitwise operation
6228 <tt>delta</tt>, store the result to <tt>ptr</tt>. They yield the original
6229 value stored at <tt>ptr</tt>.
6230</p>
6231
6232<h5>Examples:</h5>
6233<pre>
6234%ptr = malloc i32
6235 store i32 0x0F0F, %ptr
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6236%result0 = call i32 @llvm.atomic.load.nand.i32.p0i32( i32* %ptr, i32 0xFF )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6237 <i>; yields {i32}:result0 = 0x0F0F</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6238%result1 = call i32 @llvm.atomic.load.and.i32.p0i32( i32* %ptr, i32 0xFF )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6239 <i>; yields {i32}:result1 = 0xFFFFFFF0</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6240%result2 = call i32 @llvm.atomic.load.or.i32.p0i32( i32* %ptr, i32 0F )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6241 <i>; yields {i32}:result2 = 0xF0</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6242%result3 = call i32 @llvm.atomic.load.xor.i32.p0i32( i32* %ptr, i32 0F )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6243 <i>; yields {i32}:result3 = FF</i>
6244%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = F0</i>
6245</pre>
6246</div>
6247
6248
6249<!-- _______________________________________________________________________ -->
6250<div class="doc_subsubsection">
6251 <a name="int_atomic_load_max">'<tt>llvm.atomic.load.max.*</tt>' Intrinsic</a><br>
6252 <a name="int_atomic_load_min">'<tt>llvm.atomic.load.min.*</tt>' Intrinsic</a><br>
6253 <a name="int_atomic_load_umax">'<tt>llvm.atomic.load.umax.*</tt>' Intrinsic</a><br>
6254 <a name="int_atomic_load_umin">'<tt>llvm.atomic.load.umin.*</tt>' Intrinsic</a><br>
6255
6256</div>
6257<div class="doc_text">
6258<h5>Syntax:</h5>
6259<p>
6260 These are overloaded intrinsics. You can use <tt>llvm.atomic.load_max</tt>,
6261 <tt>llvm.atomic.load_min</tt>, <tt>llvm.atomic.load_umax</tt>, and
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6262 <tt>llvm.atomic.load_umin</tt> on any integer bit width and for different
6263 address spaces. Not all targets
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6264 support all bit widths however.</p>
6265<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6266declare i8 @llvm.atomic.load.max.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6267declare i16 @llvm.atomic.load.max.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6268declare i32 @llvm.atomic.load.max.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6269declare 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]6270
6271</pre>
6272
6273<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6274declare i8 @llvm.atomic.load.min.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6275declare i16 @llvm.atomic.load.min.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6276declare i32 @llvm.atomic.load.min.i32..p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6277declare 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]6278
6279</pre>
6280
6281<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6282declare i8 @llvm.atomic.load.umax.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6283declare i16 @llvm.atomic.load.umax.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6284declare i32 @llvm.atomic.load.umax.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6285declare 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]6286
6287</pre>
6288
6289<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6290declare i8 @llvm.atomic.load.umin.i8..p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6291declare i16 @llvm.atomic.load.umin.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6292declare i32 @llvm.atomic.load.umin.i32..p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6293declare 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]6294
6295</pre>
6296<h5>Overview:</h5>
6297<p>
6298 These intrinsics takes the signed or unsigned minimum or maximum of
6299 <tt>delta</tt> and the value stored in memory at <tt>ptr</tt>. It yields the
6300 original value at <tt>ptr</tt>.
6301</p>
6302<h5>Arguments:</h5>
6303<p>
6304
6305 These intrinsics take two arguments, the first a pointer to an integer value
6306 and the second an integer value. The result is also an integer value. These
6307 integer types can have any bit width, but they must all have the same bit
6308 width. The targets may only lower integer representations they support.
6309</p>
6310<h5>Semantics:</h5>
6311<p>
6312 These intrinsics does a series of operations atomically. They first load the
6313 value stored at <tt>ptr</tt>. They then do the signed or unsigned min or max
6314 <tt>delta</tt> and the value, store the result to <tt>ptr</tt>. They yield
6315 the original value stored at <tt>ptr</tt>.
6316</p>
6317
6318<h5>Examples:</h5>
6319<pre>
6320%ptr = malloc i32
6321 store i32 7, %ptr
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6322%result0 = call i32 @llvm.atomic.load.min.i32.p0i32( i32* %ptr, i32 -2 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6323 <i>; yields {i32}:result0 = 7</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6324%result1 = call i32 @llvm.atomic.load.max.i32.p0i32( i32* %ptr, i32 8 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6325 <i>; yields {i32}:result1 = -2</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6326%result2 = call i32 @llvm.atomic.load.umin.i32.p0i32( i32* %ptr, i32 10 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6327 <i>; yields {i32}:result2 = 8</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6328%result3 = call i32 @llvm.atomic.load.umax.i32.p0i32( i32* %ptr, i32 30 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6329 <i>; yields {i32}:result3 = 8</i>
6330%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 30</i>
6331</pre>
6332</div>
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]6333
6334<!-- ======================================================================= -->
6335<div class="doc_subsection">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6336 <a name="int_general">General Intrinsics</a>
6337</div>
6338
6339<div class="doc_text">
6340<p> This class of intrinsics is designed to be generic and has
6341no specific purpose. </p>
6342</div>
6343
6344<!-- _______________________________________________________________________ -->
6345<div class="doc_subsubsection">
6346 <a name="int_var_annotation">'<tt>llvm.var.annotation</tt>' Intrinsic</a>
6347</div>
6348
6349<div class="doc_text">
6350
6351<h5>Syntax:</h5>
6352<pre>
6353 declare void @llvm.var.annotation(i8* &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
6354</pre>
6355
6356<h5>Overview:</h5>
6357
6358<p>
6359The '<tt>llvm.var.annotation</tt>' intrinsic
6360</p>
6361
6362<h5>Arguments:</h5>
6363
6364<p>
6365The first argument is a pointer to a value, the second is a pointer to a
6366global string, the third is a pointer to a global string which is the source
6367file name, and the last argument is the line number.
6368</p>
6369
6370<h5>Semantics:</h5>
6371
6372<p>
Anton Korobeynikove6e764f2008-01-15 22:31:34 +0000[diff] [blame]6373This intrinsic allows annotation of local variables with arbitrary strings.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6374This can be useful for special purpose optimizations that want to look for these
Anton Korobeynikove6e764f2008-01-15 22:31:34 +0000[diff] [blame]6375annotations. These have no other defined use, they are ignored by code
6376generation and optimization.
6377</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6378</div>
6379
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]6380<!-- _______________________________________________________________________ -->
6381<div class="doc_subsubsection">
Tanya Lattnerc9869b12007-09-21 23:57:59 +0000[diff] [blame]6382 <a name="int_annotation">'<tt>llvm.annotation.*</tt>' Intrinsic</a>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]6383</div>
6384
6385<div class="doc_text">
6386
6387<h5>Syntax:</h5>
Tanya Lattnere545be72007-09-21 23:56:27 +0000[diff] [blame]6388<p>This is an overloaded intrinsic. You can use '<tt>llvm.annotation</tt>' on
6389any integer bit width.
6390</p>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]6391<pre>
Tanya Lattner09161fe2007-09-22 00:03:01 +0000[diff] [blame]6392 declare i8 @llvm.annotation.i8(i8 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
6393 declare i16 @llvm.annotation.i16(i16 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
6394 declare i32 @llvm.annotation.i32(i32 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
6395 declare i64 @llvm.annotation.i64(i64 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
6396 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]6397</pre>
6398
6399<h5>Overview:</h5>
Tanya Lattnere545be72007-09-21 23:56:27 +0000[diff] [blame]6400
6401<p>
6402The '<tt>llvm.annotation</tt>' intrinsic.
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]6403</p>
6404
6405<h5>Arguments:</h5>
6406
6407<p>
6408The first argument is an integer value (result of some expression),
6409the second is a pointer to a global string, the third is a pointer to a global
6410string which is the source file name, and the last argument is the line number.
Tanya Lattnere545be72007-09-21 23:56:27 +0000[diff] [blame]6411It returns the value of the first argument.
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]6412</p>
6413
6414<h5>Semantics:</h5>
6415
6416<p>
6417This intrinsic allows annotations to be put on arbitrary expressions
6418with arbitrary strings. This can be useful for special purpose optimizations
6419that want to look for these annotations. These have no other defined use, they
6420are ignored by code generation and optimization.
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]6421</p>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]6422</div>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6423
Anton Korobeynikove6e764f2008-01-15 22:31:34 +0000[diff] [blame]6424<!-- _______________________________________________________________________ -->
6425<div class="doc_subsubsection">
6426 <a name="int_trap">'<tt>llvm.trap</tt>' Intrinsic</a>
6427</div>
6428
6429<div class="doc_text">
6430
6431<h5>Syntax:</h5>
6432<pre>
6433 declare void @llvm.trap()
6434</pre>
6435
6436<h5>Overview:</h5>
6437
6438<p>
6439The '<tt>llvm.trap</tt>' intrinsic
6440</p>
6441
6442<h5>Arguments:</h5>
6443
6444<p>
6445None
6446</p>
6447
6448<h5>Semantics:</h5>
6449
6450<p>
6451This intrinsics is lowered to the target dependent trap instruction. If the
6452target does not have a trap instruction, this intrinsic will be lowered to the
6453call of the abort() function.
6454</p>
6455</div>
6456
Bill Wendlinge4164592008-11-19 05:56:17 +0000[diff] [blame]6457<!-- _______________________________________________________________________ -->
6458<div class="doc_subsubsection">
Misha Brukman5dd7f4d2008-11-22 23:55:29 +0000[diff] [blame]6459 <a name="int_stackprotector">'<tt>llvm.stackprotector</tt>' Intrinsic</a>
Bill Wendlinge4164592008-11-19 05:56:17 +0000[diff] [blame]6460</div>
6461<div class="doc_text">
6462<h5>Syntax:</h5>
6463<pre>
6464declare void @llvm.stackprotector( i8* &lt;guard&gt;, i8** &lt;slot&gt; )
6465
6466</pre>
6467<h5>Overview:</h5>
6468<p>
6469 The <tt>llvm.stackprotector</tt> intrinsic takes the <tt>guard</tt> and stores
6470 it onto the stack at <tt>slot</tt>. The stack slot is adjusted to ensure that
6471 it is placed on the stack before local variables.
6472</p>
6473<h5>Arguments:</h5>
6474<p>
6475 The <tt>llvm.stackprotector</tt> intrinsic requires two pointer arguments. The
6476 first argument is the value loaded from the stack guard
6477 <tt>@__stack_chk_guard</tt>. The second variable is an <tt>alloca</tt> that
6478 has enough space to hold the value of the guard.
6479</p>
6480<h5>Semantics:</h5>
6481<p>
6482 This intrinsic causes the prologue/epilogue inserter to force the position of
6483 the <tt>AllocaInst</tt> stack slot to be before local variables on the
6484 stack. This is to ensure that if a local variable on the stack is overwritten,
6485 it will destroy the value of the guard. When the function exits, the guard on
6486 the stack is checked against the original guard. If they're different, then
6487 the program aborts by calling the <tt>__stack_chk_fail()</tt> function.
6488</p>
6489</div>
6490
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6491<!-- *********************************************************************** -->
6492<hr>
6493<address>
6494 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
6495 src="http://jigsaw.w3.org/css-validator/images/vcss" alt="Valid CSS!"></a>
6496 <a href="http://validator.w3.org/check/referer"><img
Chris Lattner08497ce2008-01-04 04:33:49 +0000[diff] [blame]6497 src="http://www.w3.org/Icons/valid-html401" alt="Valid HTML 4.01!"></a>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6498
6499 <a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
6500 <a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br>
6501 Last modified: $Date$
6502</address>
Chris Lattner08497ce2008-01-04 04:33:49 +0000[diff] [blame]6503
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6504</body>
6505</html>