Gitiles
Code Review Sign In
gerrit-public.fairphone.software / fp2-dev / platform / external / llvm / 5dd7f4d87faab237a6e3fc14fe5f8ff0e8b8b38a / . / docs / LangRef.html
blob: b6ea8e9b423dd44616d07470622f6317bd8feaf8 [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
568<p><a name="linkage_external"></a>For example, since the "<tt>.LC0</tt>"
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>
Owen Andersonc4fc4cd2008-02-18 04:09:01 +0000[diff] [blame]893 <dd>This indicates that the parameter does not alias any global or any other
894 parameter. The caller is responsible for ensuring that this is the case,
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]895 usually by placing the value in a stack allocation. This is not a valid
896 attribute for return values.</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]897
Duncan Sands4ee46812007-07-27 19:57:41 +0000[diff] [blame]898 <dt><tt>nest</tt></dt>
Duncan Sandsf1a7d4c2008-07-08 09:27:25 +0000[diff] [blame]899 <dd>This indicates that the pointer parameter can be excised using the
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]900 <a href="#int_trampoline">trampoline intrinsics</a>. This is not a valid
901 attribute for return values.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]902 </dl>
903
904</div>
905
906<!-- ======================================================================= -->
907<div class="doc_subsection">
Gordon Henriksen13fe5e32007-12-10 03:18:06 +0000[diff] [blame]908 <a name="gc">Garbage Collector Names</a>
909</div>
910
911<div class="doc_text">
912<p>Each function may specify a garbage collector name, which is simply a
913string.</p>
914
915<div class="doc_code"><pre
916>define void @f() gc "name" { ...</pre></div>
917
918<p>The compiler declares the supported values of <i>name</i>. Specifying a
919collector which will cause the compiler to alter its output in order to support
920the named garbage collection algorithm.</p>
921</div>
922
923<!-- ======================================================================= -->
924<div class="doc_subsection">
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]925 <a name="fnattrs">Function Attributes</a>
Devang Pateld468f1c2008-09-04 23:05:13 +0000[diff] [blame]926</div>
927
928<div class="doc_text">
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]929
930<p>Function attributes are set to communicate additional information about
931 a function. Function attributes are considered to be part of the function,
932 not of the function type, so functions with different parameter attributes
933 can have the same function type.</p>
934
935 <p>Function attributes are simple keywords that follow the type specified. If
936 multiple attributes are needed, they are space separated. For
937 example:</p>
Devang Pateld468f1c2008-09-04 23:05:13 +0000[diff] [blame]938
939<div class="doc_code">
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]940<pre>
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]941define void @f() noinline { ... }
942define void @f() alwaysinline { ... }
943define void @f() alwaysinline optsize { ... }
944define void @f() optsize
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]945</pre>
Devang Pateld468f1c2008-09-04 23:05:13 +0000[diff] [blame]946</div>
947
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]948<dl>
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]949<dt><tt>alwaysinline</tt></dt>
Chris Lattner82d70f92008-10-04 18:23:17 +0000[diff] [blame]950<dd>This attribute indicates that the inliner should attempt to inline this
951function into callers whenever possible, ignoring any active inlining size
952threshold for this caller.</dd>
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]953
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]954<dt><tt>noinline</tt></dt>
Chris Lattner82d70f92008-10-04 18:23:17 +0000[diff] [blame]955<dd>This attribute indicates that the inliner should never inline this function
Chris Lattner377f3ae2008-10-05 17:14:59 +0000[diff] [blame]956in any situation. This attribute may not be used together with the
Chris Lattner82d70f92008-10-04 18:23:17 +0000[diff] [blame]957<tt>alwaysinline</tt> attribute.</dd>
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]958
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]959<dt><tt>optsize</tt></dt>
Devang Patelc29099b2008-09-29 18:34:44 +0000[diff] [blame]960<dd>This attribute suggests that optimization passes and code generator passes
Chris Lattner82d70f92008-10-04 18:23:17 +0000[diff] [blame]961make choices that keep the code size of this function low, and otherwise do
962optimizations specifically to reduce code size.</dd>
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]963
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]964<dt><tt>noreturn</tt></dt>
Chris Lattner82d70f92008-10-04 18:23:17 +0000[diff] [blame]965<dd>This function attribute indicates that the function never returns normally.
966This produces undefined behavior at runtime if the function ever does
967dynamically return.</dd>
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]968
969<dt><tt>nounwind</tt></dt>
Chris Lattner82d70f92008-10-04 18:23:17 +0000[diff] [blame]970<dd>This function attribute indicates that the function never returns with an
971unwind or exceptional control flow. If the function does unwind, its runtime
972behavior is undefined.</dd>
973
974<dt><tt>readnone</tt></dt>
Duncan Sands7d94e5a2008-10-06 08:14:18 +0000[diff] [blame]975<dd>This attribute indicates that the function computes its result (or the
976exception it throws) based strictly on its arguments, without dereferencing any
977pointer arguments or otherwise accessing any mutable state (e.g. memory, control
978registers, etc) visible to caller functions. It does not write through any
979pointer arguments (including <tt><a href="#byval">byval</a></tt> arguments) and
980never changes any state visible to callers.</dd>
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]981
Duncan Sands7d94e5a2008-10-06 08:14:18 +0000[diff] [blame]982<dt><tt><a name="readonly">readonly</a></tt></dt>
983<dd>This attribute indicates that the function does not write through any
984pointer arguments (including <tt><a href="#byval">byval</a></tt> arguments)
985or otherwise modify any state (e.g. memory, control registers, etc) visible to
986caller functions. It may dereference pointer arguments and read state that may
987be set in the caller. A readonly function always returns the same value (or
988throws the same exception) when called with the same set of arguments and global
989state.</dd>
Bill Wendlingdfaabba2008-11-13 01:02:51 +0000[diff] [blame]990
991<dt><tt><a name="ssp">ssp</a></tt></dt>
992<dd>This attribute indicates that the function should emit a stack smashing
993protector. It is in the form of a "canary"&mdash;a random value placed on the
994stack before the local variables that's checked upon return from the function to
995see if it has been overwritten. A heuristic is used to determine if a function
996needs stack protectors or not.</dd>
997
998<dt><tt>ssp-req</tt></dt>
999<dd>This attribute indicates that the function should <em>always</em> emit a
1000stack smashing protector. This overrides the <tt><a href="#ssp">ssp</a></tt>
1001function attribute.</dd>
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]1002</dl>
1003
Devang Pateld468f1c2008-09-04 23:05:13 +0000[diff] [blame]1004</div>
1005
1006<!-- ======================================================================= -->
1007<div class="doc_subsection">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1008 <a name="moduleasm">Module-Level Inline Assembly</a>
1009</div>
1010
1011<div class="doc_text">
1012<p>
1013Modules may contain "module-level inline asm" blocks, which corresponds to the
1014GCC "file scope inline asm" blocks. These blocks are internally concatenated by
1015LLVM and treated as a single unit, but may be separated in the .ll file if
1016desired. The syntax is very simple:
1017</p>
1018
1019<div class="doc_code">
1020<pre>
1021module asm "inline asm code goes here"
1022module asm "more can go here"
1023</pre>
1024</div>
1025
1026<p>The strings can contain any character by escaping non-printable characters.
1027 The escape sequence used is simply "\xx" where "xx" is the two digit hex code
1028 for the number.
1029</p>
1030
1031<p>
1032 The inline asm code is simply printed to the machine code .s file when
1033 assembly code is generated.
1034</p>
1035</div>
1036
1037<!-- ======================================================================= -->
1038<div class="doc_subsection">
1039 <a name="datalayout">Data Layout</a>
1040</div>
1041
1042<div class="doc_text">
1043<p>A module may specify a target specific data layout string that specifies how
1044data is to be laid out in memory. The syntax for the data layout is simply:</p>
1045<pre> target datalayout = "<i>layout specification</i>"</pre>
1046<p>The <i>layout specification</i> consists of a list of specifications
1047separated by the minus sign character ('-'). Each specification starts with a
1048letter and may include other information after the letter to define some
1049aspect of the data layout. The specifications accepted are as follows: </p>
1050<dl>
1051 <dt><tt>E</tt></dt>
1052 <dd>Specifies that the target lays out data in big-endian form. That is, the
1053 bits with the most significance have the lowest address location.</dd>
1054 <dt><tt>e</tt></dt>
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]1055 <dd>Specifies that the target lays out data in little-endian form. That is,
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1056 the bits with the least significance have the lowest address location.</dd>
1057 <dt><tt>p:<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1058 <dd>This specifies the <i>size</i> of a pointer and its <i>abi</i> and
1059 <i>preferred</i> alignments. All sizes are in bits. Specifying the <i>pref</i>
1060 alignment is optional. If omitted, the preceding <tt>:</tt> should be omitted
1061 too.</dd>
1062 <dt><tt>i<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1063 <dd>This specifies the alignment for an integer type of a given bit
1064 <i>size</i>. The value of <i>size</i> must be in the range [1,2^23).</dd>
1065 <dt><tt>v<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1066 <dd>This specifies the alignment for a vector type of a given bit
1067 <i>size</i>.</dd>
1068 <dt><tt>f<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1069 <dd>This specifies the alignment for a floating point type of a given bit
1070 <i>size</i>. The value of <i>size</i> must be either 32 (float) or 64
1071 (double).</dd>
1072 <dt><tt>a<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1073 <dd>This specifies the alignment for an aggregate type of a given bit
1074 <i>size</i>.</dd>
1075</dl>
1076<p>When constructing the data layout for a given target, LLVM starts with a
1077default set of specifications which are then (possibly) overriden by the
1078specifications in the <tt>datalayout</tt> keyword. The default specifications
1079are given in this list:</p>
1080<ul>
1081 <li><tt>E</tt> - big endian</li>
1082 <li><tt>p:32:64:64</tt> - 32-bit pointers with 64-bit alignment</li>
1083 <li><tt>i1:8:8</tt> - i1 is 8-bit (byte) aligned</li>
1084 <li><tt>i8:8:8</tt> - i8 is 8-bit (byte) aligned</li>
1085 <li><tt>i16:16:16</tt> - i16 is 16-bit aligned</li>
1086 <li><tt>i32:32:32</tt> - i32 is 32-bit aligned</li>
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]1087 <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]1088 alignment of 64-bits</li>
1089 <li><tt>f32:32:32</tt> - float is 32-bit aligned</li>
1090 <li><tt>f64:64:64</tt> - double is 64-bit aligned</li>
1091 <li><tt>v64:64:64</tt> - 64-bit vector is 64-bit aligned</li>
1092 <li><tt>v128:128:128</tt> - 128-bit vector is 128-bit aligned</li>
1093 <li><tt>a0:0:1</tt> - aggregates are 8-bit aligned</li>
1094</ul>
Chris Lattner6a9f3c42008-08-05 18:21:08 +0000[diff] [blame]1095<p>When LLVM is determining the alignment for a given type, it uses the
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]1096following rules:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1097<ol>
1098 <li>If the type sought is an exact match for one of the specifications, that
1099 specification is used.</li>
1100 <li>If no match is found, and the type sought is an integer type, then the
1101 smallest integer type that is larger than the bitwidth of the sought type is
1102 used. If none of the specifications are larger than the bitwidth then the the
1103 largest integer type is used. For example, given the default specifications
1104 above, the i7 type will use the alignment of i8 (next largest) while both
1105 i65 and i256 will use the alignment of i64 (largest specified).</li>
1106 <li>If no match is found, and the type sought is a vector type, then the
1107 largest vector type that is smaller than the sought vector type will be used
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]1108 as a fall back. This happens because &lt;128 x double&gt; can be implemented
1109 in terms of 64 &lt;2 x double&gt;, for example.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1110</ol>
1111</div>
1112
1113<!-- *********************************************************************** -->
1114<div class="doc_section"> <a name="typesystem">Type System</a> </div>
1115<!-- *********************************************************************** -->
1116
1117<div class="doc_text">
1118
1119<p>The LLVM type system is one of the most important features of the
1120intermediate representation. Being typed enables a number of
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]1121optimizations to be performed on the intermediate representation directly,
1122without having to do
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1123extra analyses on the side before the transformation. A strong type
1124system makes it easier to read the generated code and enables novel
1125analyses and transformations that are not feasible to perform on normal
1126three address code representations.</p>
1127
1128</div>
1129
1130<!-- ======================================================================= -->
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1131<div class="doc_subsection"> <a name="t_classifications">Type
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1132Classifications</a> </div>
1133<div class="doc_text">
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1134<p>The types fall into a few useful
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1135classifications:</p>
1136
1137<table border="1" cellspacing="0" cellpadding="4">
1138 <tbody>
1139 <tr><th>Classification</th><th>Types</th></tr>
1140 <tr>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1141 <td><a href="#t_integer">integer</a></td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1142 <td><tt>i1, i2, i3, ... i8, ... i16, ... i32, ... i64, ... </tt></td>
1143 </tr>
1144 <tr>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1145 <td><a href="#t_floating">floating point</a></td>
1146 <td><tt>float, double, x86_fp80, fp128, ppc_fp128</tt></td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1147 </tr>
1148 <tr>
1149 <td><a name="t_firstclass">first class</a></td>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1150 <td><a href="#t_integer">integer</a>,
1151 <a href="#t_floating">floating point</a>,
1152 <a href="#t_pointer">pointer</a>,
Dan Gohmanf6237db2008-06-18 18:42:13 +0000[diff] [blame]1153 <a href="#t_vector">vector</a>,
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]1154 <a href="#t_struct">structure</a>,
1155 <a href="#t_array">array</a>,
Dan Gohmand13951e2008-05-23 22:50:26 +0000[diff] [blame]1156 <a href="#t_label">label</a>.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1157 </td>
1158 </tr>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1159 <tr>
1160 <td><a href="#t_primitive">primitive</a></td>
1161 <td><a href="#t_label">label</a>,
1162 <a href="#t_void">void</a>,
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1163 <a href="#t_floating">floating point</a>.</td>
1164 </tr>
1165 <tr>
1166 <td><a href="#t_derived">derived</a></td>
1167 <td><a href="#t_integer">integer</a>,
1168 <a href="#t_array">array</a>,
1169 <a href="#t_function">function</a>,
1170 <a href="#t_pointer">pointer</a>,
1171 <a href="#t_struct">structure</a>,
1172 <a href="#t_pstruct">packed structure</a>,
1173 <a href="#t_vector">vector</a>,
1174 <a href="#t_opaque">opaque</a>.
Dan Gohman032ba852008-10-14 16:32:04 +0000[diff] [blame]1175 </td>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1176 </tr>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1177 </tbody>
1178</table>
1179
1180<p>The <a href="#t_firstclass">first class</a> types are perhaps the
1181most important. Values of these types are the only ones which can be
1182produced by instructions, passed as arguments, or used as operands to
Dan Gohman4f29e422008-05-23 21:53:15 +0000[diff] [blame]1183instructions.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1184</div>
1185
1186<!-- ======================================================================= -->
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1187<div class="doc_subsection"> <a name="t_primitive">Primitive Types</a> </div>
Chris Lattner86437612008-01-04 04:34:14 +0000[diff] [blame]1188
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1189<div class="doc_text">
1190<p>The primitive types are the fundamental building blocks of the LLVM
1191system.</p>
1192
Chris Lattner86437612008-01-04 04:34:14 +0000[diff] [blame]1193</div>
1194
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1195<!-- _______________________________________________________________________ -->
1196<div class="doc_subsubsection"> <a name="t_floating">Floating Point Types</a> </div>
1197
1198<div class="doc_text">
1199 <table>
1200 <tbody>
1201 <tr><th>Type</th><th>Description</th></tr>
1202 <tr><td><tt>float</tt></td><td>32-bit floating point value</td></tr>
1203 <tr><td><tt>double</tt></td><td>64-bit floating point value</td></tr>
1204 <tr><td><tt>fp128</tt></td><td>128-bit floating point value (112-bit mantissa)</td></tr>
1205 <tr><td><tt>x86_fp80</tt></td><td>80-bit floating point value (X87)</td></tr>
1206 <tr><td><tt>ppc_fp128</tt></td><td>128-bit floating point value (two 64-bits)</td></tr>
1207 </tbody>
1208 </table>
1209</div>
1210
1211<!-- _______________________________________________________________________ -->
1212<div class="doc_subsubsection"> <a name="t_void">Void Type</a> </div>
1213
1214<div class="doc_text">
1215<h5>Overview:</h5>
1216<p>The void type does not represent any value and has no size.</p>
1217
1218<h5>Syntax:</h5>
1219
1220<pre>
1221 void
1222</pre>
1223</div>
1224
1225<!-- _______________________________________________________________________ -->
1226<div class="doc_subsubsection"> <a name="t_label">Label Type</a> </div>
1227
1228<div class="doc_text">
1229<h5>Overview:</h5>
1230<p>The label type represents code labels.</p>
1231
1232<h5>Syntax:</h5>
1233
1234<pre>
1235 label
1236</pre>
1237</div>
1238
1239
1240<!-- ======================================================================= -->
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1241<div class="doc_subsection"> <a name="t_derived">Derived Types</a> </div>
1242
1243<div class="doc_text">
1244
1245<p>The real power in LLVM comes from the derived types in the system.
1246This is what allows a programmer to represent arrays, functions,
1247pointers, and other useful types. Note that these derived types may be
1248recursive: For example, it is possible to have a two dimensional array.</p>
1249
1250</div>
1251
1252<!-- _______________________________________________________________________ -->
1253<div class="doc_subsubsection"> <a name="t_integer">Integer Type</a> </div>
1254
1255<div class="doc_text">
1256
1257<h5>Overview:</h5>
1258<p>The integer type is a very simple derived type that simply specifies an
1259arbitrary bit width for the integer type desired. Any bit width from 1 bit to
12602^23-1 (about 8 million) can be specified.</p>
1261
1262<h5>Syntax:</h5>
1263
1264<pre>
1265 iN
1266</pre>
1267
1268<p>The number of bits the integer will occupy is specified by the <tt>N</tt>
1269value.</p>
1270
1271<h5>Examples:</h5>
1272<table class="layout">
Chris Lattner251ab812007-12-18 06:18:21 +0000[diff] [blame]1273 <tbody>
1274 <tr>
1275 <td><tt>i1</tt></td>
1276 <td>a single-bit integer.</td>
1277 </tr><tr>
1278 <td><tt>i32</tt></td>
1279 <td>a 32-bit integer.</td>
1280 </tr><tr>
1281 <td><tt>i1942652</tt></td>
1282 <td>a really big integer of over 1 million bits.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1283 </tr>
Chris Lattner251ab812007-12-18 06:18:21 +0000[diff] [blame]1284 </tbody>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1285</table>
1286</div>
1287
1288<!-- _______________________________________________________________________ -->
1289<div class="doc_subsubsection"> <a name="t_array">Array Type</a> </div>
1290
1291<div class="doc_text">
1292
1293<h5>Overview:</h5>
1294
1295<p>The array type is a very simple derived type that arranges elements
1296sequentially in memory. The array type requires a size (number of
1297elements) and an underlying data type.</p>
1298
1299<h5>Syntax:</h5>
1300
1301<pre>
1302 [&lt;# elements&gt; x &lt;elementtype&gt;]
1303</pre>
1304
1305<p>The number of elements is a constant integer value; elementtype may
1306be any type with a size.</p>
1307
1308<h5>Examples:</h5>
1309<table class="layout">
1310 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1311 <td class="left"><tt>[40 x i32]</tt></td>
1312 <td class="left">Array of 40 32-bit integer values.</td>
1313 </tr>
1314 <tr class="layout">
1315 <td class="left"><tt>[41 x i32]</tt></td>
1316 <td class="left">Array of 41 32-bit integer values.</td>
1317 </tr>
1318 <tr class="layout">
1319 <td class="left"><tt>[4 x i8]</tt></td>
1320 <td class="left">Array of 4 8-bit integer values.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1321 </tr>
1322</table>
1323<p>Here are some examples of multidimensional arrays:</p>
1324<table class="layout">
1325 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1326 <td class="left"><tt>[3 x [4 x i32]]</tt></td>
1327 <td class="left">3x4 array of 32-bit integer values.</td>
1328 </tr>
1329 <tr class="layout">
1330 <td class="left"><tt>[12 x [10 x float]]</tt></td>
1331 <td class="left">12x10 array of single precision floating point values.</td>
1332 </tr>
1333 <tr class="layout">
1334 <td class="left"><tt>[2 x [3 x [4 x i16]]]</tt></td>
1335 <td class="left">2x3x4 array of 16-bit integer values.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1336 </tr>
1337</table>
1338
1339<p>Note that 'variable sized arrays' can be implemented in LLVM with a zero
1340length array. Normally, accesses past the end of an array are undefined in
1341LLVM (e.g. it is illegal to access the 5th element of a 3 element array).
1342As a special case, however, zero length arrays are recognized to be variable
1343length. This allows implementation of 'pascal style arrays' with the LLVM
1344type "{ i32, [0 x float]}", for example.</p>
1345
1346</div>
1347
1348<!-- _______________________________________________________________________ -->
1349<div class="doc_subsubsection"> <a name="t_function">Function Type</a> </div>
1350<div class="doc_text">
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1351
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1352<h5>Overview:</h5>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1353
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1354<p>The function type can be thought of as a function signature. It
Devang Pateld4ba41d2008-03-24 05:35:41 +0000[diff] [blame]1355consists of a return type and a list of formal parameter types. The
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1356return 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]1357If 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]1358class types, and the struct must have at least one element.</p>
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]1359
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1360<h5>Syntax:</h5>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1361
1362<pre>
1363 &lt;returntype list&gt; (&lt;parameter list&gt;)
1364</pre>
1365
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1366<p>...where '<tt>&lt;parameter list&gt;</tt>' is a comma-separated list of type
1367specifiers. Optionally, the parameter list may include a type <tt>...</tt>,
1368which indicates that the function takes a variable number of arguments.
1369Variable argument functions can access their arguments with the <a
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]1370 href="#int_varargs">variable argument handling intrinsic</a> functions.
1371'<tt>&lt;returntype list&gt;</tt>' is a comma-separated list of
1372<a href="#t_firstclass">first class</a> type specifiers.</p>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1373
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1374<h5>Examples:</h5>
1375<table class="layout">
1376 <tr class="layout">
1377 <td class="left"><tt>i32 (i32)</tt></td>
1378 <td class="left">function taking an <tt>i32</tt>, returning an <tt>i32</tt>
1379 </td>
1380 </tr><tr class="layout">
Reid Spencerf234bed2007-07-19 23:13:04 +0000[diff] [blame]1381 <td class="left"><tt>float&nbsp;(i16&nbsp;signext,&nbsp;i32&nbsp;*)&nbsp;*
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1382 </tt></td>
1383 <td class="left"><a href="#t_pointer">Pointer</a> to a function that takes
1384 an <tt>i16</tt> that should be sign extended and a
1385 <a href="#t_pointer">pointer</a> to <tt>i32</tt>, returning
1386 <tt>float</tt>.
1387 </td>
1388 </tr><tr class="layout">
1389 <td class="left"><tt>i32 (i8*, ...)</tt></td>
1390 <td class="left">A vararg function that takes at least one
1391 <a href="#t_pointer">pointer</a> to <tt>i8 </tt> (char in C),
1392 which returns an integer. This is the signature for <tt>printf</tt> in
1393 LLVM.
1394 </td>
Devang Pateld4ba41d2008-03-24 05:35:41 +0000[diff] [blame]1395 </tr><tr class="layout">
1396 <td class="left"><tt>{i32, i32} (i32)</tt></td>
Devang Patel6dddba22008-03-24 18:10:52 +0000[diff] [blame]1397 <td class="left">A function taking an <tt>i32></tt>, returning two
1398 <tt> i32 </tt> values as an aggregate of type <tt>{ i32, i32 }</tt>
Devang Pateld4ba41d2008-03-24 05:35:41 +0000[diff] [blame]1399 </td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1400 </tr>
1401</table>
1402
1403</div>
1404<!-- _______________________________________________________________________ -->
1405<div class="doc_subsubsection"> <a name="t_struct">Structure Type</a> </div>
1406<div class="doc_text">
1407<h5>Overview:</h5>
1408<p>The structure type is used to represent a collection of data members
1409together in memory. The packing of the field types is defined to match
1410the ABI of the underlying processor. The elements of a structure may
1411be any type that has a size.</p>
1412<p>Structures are accessed using '<tt><a href="#i_load">load</a></tt>
1413and '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a
1414field with the '<tt><a href="#i_getelementptr">getelementptr</a></tt>'
1415instruction.</p>
1416<h5>Syntax:</h5>
1417<pre> { &lt;type list&gt; }<br></pre>
1418<h5>Examples:</h5>
1419<table class="layout">
1420 <tr class="layout">
1421 <td class="left"><tt>{ i32, i32, i32 }</tt></td>
1422 <td class="left">A triple of three <tt>i32</tt> values</td>
1423 </tr><tr class="layout">
1424 <td class="left"><tt>{&nbsp;float,&nbsp;i32&nbsp;(i32)&nbsp;*&nbsp;}</tt></td>
1425 <td class="left">A pair, where the first element is a <tt>float</tt> and the
1426 second element is a <a href="#t_pointer">pointer</a> to a
1427 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
1428 an <tt>i32</tt>.</td>
1429 </tr>
1430</table>
1431</div>
1432
1433<!-- _______________________________________________________________________ -->
1434<div class="doc_subsubsection"> <a name="t_pstruct">Packed Structure Type</a>
1435</div>
1436<div class="doc_text">
1437<h5>Overview:</h5>
1438<p>The packed structure type is used to represent a collection of data members
1439together in memory. There is no padding between fields. Further, the alignment
1440of a packed structure is 1 byte. The elements of a packed structure may
1441be any type that has a size.</p>
1442<p>Structures are accessed using '<tt><a href="#i_load">load</a></tt>
1443and '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a
1444field with the '<tt><a href="#i_getelementptr">getelementptr</a></tt>'
1445instruction.</p>
1446<h5>Syntax:</h5>
1447<pre> &lt; { &lt;type list&gt; } &gt; <br></pre>
1448<h5>Examples:</h5>
1449<table class="layout">
1450 <tr class="layout">
1451 <td class="left"><tt>&lt; { i32, i32, i32 } &gt;</tt></td>
1452 <td class="left">A triple of three <tt>i32</tt> values</td>
1453 </tr><tr class="layout">
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]1454 <td class="left">
1455<tt>&lt;&nbsp;{&nbsp;float,&nbsp;i32&nbsp;(i32)*&nbsp;}&nbsp;&gt;</tt></td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1456 <td class="left">A pair, where the first element is a <tt>float</tt> and the
1457 second element is a <a href="#t_pointer">pointer</a> to a
1458 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
1459 an <tt>i32</tt>.</td>
1460 </tr>
1461</table>
1462</div>
1463
1464<!-- _______________________________________________________________________ -->
1465<div class="doc_subsubsection"> <a name="t_pointer">Pointer Type</a> </div>
1466<div class="doc_text">
1467<h5>Overview:</h5>
1468<p>As in many languages, the pointer type represents a pointer or
Christopher Lambdd0049d2007-12-11 09:31:00 +0000[diff] [blame]1469reference to another object, which must live in memory. Pointer types may have
1470an optional address space attribute defining the target-specific numbered
1471address space where the pointed-to object resides. The default address space is
1472zero.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1473<h5>Syntax:</h5>
1474<pre> &lt;type&gt; *<br></pre>
1475<h5>Examples:</h5>
1476<table class="layout">
1477 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1478 <td class="left"><tt>[4x i32]*</tt></td>
1479 <td class="left">A <a href="#t_pointer">pointer</a> to <a
1480 href="#t_array">array</a> of four <tt>i32</tt> values.</td>
1481 </tr>
1482 <tr class="layout">
1483 <td class="left"><tt>i32 (i32 *) *</tt></td>
1484 <td class="left"> A <a href="#t_pointer">pointer</a> to a <a
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1485 href="#t_function">function</a> that takes an <tt>i32*</tt>, returning an
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1486 <tt>i32</tt>.</td>
1487 </tr>
1488 <tr class="layout">
1489 <td class="left"><tt>i32 addrspace(5)*</tt></td>
1490 <td class="left">A <a href="#t_pointer">pointer</a> to an <tt>i32</tt> value
1491 that resides in address space #5.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1492 </tr>
1493</table>
1494</div>
1495
1496<!-- _______________________________________________________________________ -->
1497<div class="doc_subsubsection"> <a name="t_vector">Vector Type</a> </div>
1498<div class="doc_text">
1499
1500<h5>Overview:</h5>
1501
1502<p>A vector type is a simple derived type that represents a vector
1503of elements. Vector types are used when multiple primitive data
1504are operated in parallel using a single instruction (SIMD).
1505A vector type requires a size (number of
1506elements) and an underlying primitive data type. Vectors must have a power
1507of two length (1, 2, 4, 8, 16 ...). Vector types are
1508considered <a href="#t_firstclass">first class</a>.</p>
1509
1510<h5>Syntax:</h5>
1511
1512<pre>
1513 &lt; &lt;# elements&gt; x &lt;elementtype&gt; &gt;
1514</pre>
1515
1516<p>The number of elements is a constant integer value; elementtype may
1517be any integer or floating point type.</p>
1518
1519<h5>Examples:</h5>
1520
1521<table class="layout">
1522 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1523 <td class="left"><tt>&lt;4 x i32&gt;</tt></td>
1524 <td class="left">Vector of 4 32-bit integer values.</td>
1525 </tr>
1526 <tr class="layout">
1527 <td class="left"><tt>&lt;8 x float&gt;</tt></td>
1528 <td class="left">Vector of 8 32-bit floating-point values.</td>
1529 </tr>
1530 <tr class="layout">
1531 <td class="left"><tt>&lt;2 x i64&gt;</tt></td>
1532 <td class="left">Vector of 2 64-bit integer values.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1533 </tr>
1534</table>
1535</div>
1536
1537<!-- _______________________________________________________________________ -->
1538<div class="doc_subsubsection"> <a name="t_opaque">Opaque Type</a> </div>
1539<div class="doc_text">
1540
1541<h5>Overview:</h5>
1542
1543<p>Opaque types are used to represent unknown types in the system. This
Gordon Henriksenda0706e2007-10-14 00:34:53 +0000[diff] [blame]1544corresponds (for example) to the C notion of a forward declared structure type.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1545In LLVM, opaque types can eventually be resolved to any type (not just a
1546structure type).</p>
1547
1548<h5>Syntax:</h5>
1549
1550<pre>
1551 opaque
1552</pre>
1553
1554<h5>Examples:</h5>
1555
1556<table class="layout">
1557 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1558 <td class="left"><tt>opaque</tt></td>
1559 <td class="left">An opaque type.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1560 </tr>
1561</table>
1562</div>
1563
1564
1565<!-- *********************************************************************** -->
1566<div class="doc_section"> <a name="constants">Constants</a> </div>
1567<!-- *********************************************************************** -->
1568
1569<div class="doc_text">
1570
1571<p>LLVM has several different basic types of constants. This section describes
1572them all and their syntax.</p>
1573
1574</div>
1575
1576<!-- ======================================================================= -->
1577<div class="doc_subsection"><a name="simpleconstants">Simple Constants</a></div>
1578
1579<div class="doc_text">
1580
1581<dl>
1582 <dt><b>Boolean constants</b></dt>
1583
1584 <dd>The two strings '<tt>true</tt>' and '<tt>false</tt>' are both valid
1585 constants of the <tt><a href="#t_primitive">i1</a></tt> type.
1586 </dd>
1587
1588 <dt><b>Integer constants</b></dt>
1589
1590 <dd>Standard integers (such as '4') are constants of the <a
1591 href="#t_integer">integer</a> type. Negative numbers may be used with
1592 integer types.
1593 </dd>
1594
1595 <dt><b>Floating point constants</b></dt>
1596
1597 <dd>Floating point constants use standard decimal notation (e.g. 123.421),
1598 exponential notation (e.g. 1.23421e+2), or a more precise hexadecimal
Chris Lattnerdc15b1d2008-04-01 18:45:27 +0000[diff] [blame]1599 notation (see below). The assembler requires the exact decimal value of
1600 a floating-point constant. For example, the assembler accepts 1.25 but
1601 rejects 1.3 because 1.3 is a repeating decimal in binary. Floating point
1602 constants must have a <a href="#t_floating">floating point</a> type. </dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1603
1604 <dt><b>Null pointer constants</b></dt>
1605
1606 <dd>The identifier '<tt>null</tt>' is recognized as a null pointer constant
1607 and must be of <a href="#t_pointer">pointer type</a>.</dd>
1608
1609</dl>
1610
1611<p>The one non-intuitive notation for constants is the optional hexadecimal form
1612of floating point constants. For example, the form '<tt>double
16130x432ff973cafa8000</tt>' is equivalent to (but harder to read than) '<tt>double
16144.5e+15</tt>'. The only time hexadecimal floating point constants are required
1615(and the only time that they are generated by the disassembler) is when a
1616floating point constant must be emitted but it cannot be represented as a
1617decimal floating point number. For example, NaN's, infinities, and other
1618special values are represented in their IEEE hexadecimal format so that
1619assembly and disassembly do not cause any bits to change in the constants.</p>
1620
1621</div>
1622
1623<!-- ======================================================================= -->
1624<div class="doc_subsection"><a name="aggregateconstants">Aggregate Constants</a>
1625</div>
1626
1627<div class="doc_text">
1628<p>Aggregate constants arise from aggregation of simple constants
1629and smaller aggregate constants.</p>
1630
1631<dl>
1632 <dt><b>Structure constants</b></dt>
1633
1634 <dd>Structure constants are represented with notation similar to structure
1635 type definitions (a comma separated list of elements, surrounded by braces
Chris Lattner6c8de962007-12-25 20:34:52 +0000[diff] [blame]1636 (<tt>{}</tt>)). For example: "<tt>{ i32 4, float 17.0, i32* @G }</tt>",
1637 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]1638 must have <a href="#t_struct">structure type</a>, and the number and
1639 types of elements must match those specified by the type.
1640 </dd>
1641
1642 <dt><b>Array constants</b></dt>
1643
1644 <dd>Array constants are represented with notation similar to array type
1645 definitions (a comma separated list of elements, surrounded by square brackets
1646 (<tt>[]</tt>)). For example: "<tt>[ i32 42, i32 11, i32 74 ]</tt>". Array
1647 constants must have <a href="#t_array">array type</a>, and the number and
1648 types of elements must match those specified by the type.
1649 </dd>
1650
1651 <dt><b>Vector constants</b></dt>
1652
1653 <dd>Vector constants are represented with notation similar to vector type
1654 definitions (a comma separated list of elements, surrounded by
1655 less-than/greater-than's (<tt>&lt;&gt;</tt>)). For example: "<tt>&lt; i32 42,
1656 i32 11, i32 74, i32 100 &gt;</tt>". Vector constants must have <a
1657 href="#t_vector">vector type</a>, and the number and types of elements must
1658 match those specified by the type.
1659 </dd>
1660
1661 <dt><b>Zero initialization</b></dt>
1662
1663 <dd>The string '<tt>zeroinitializer</tt>' can be used to zero initialize a
1664 value to zero of <em>any</em> type, including scalar and aggregate types.
1665 This is often used to avoid having to print large zero initializers (e.g. for
1666 large arrays) and is always exactly equivalent to using explicit zero
1667 initializers.
1668 </dd>
1669</dl>
1670
1671</div>
1672
1673<!-- ======================================================================= -->
1674<div class="doc_subsection">
1675 <a name="globalconstants">Global Variable and Function Addresses</a>
1676</div>
1677
1678<div class="doc_text">
1679
1680<p>The addresses of <a href="#globalvars">global variables</a> and <a
1681href="#functionstructure">functions</a> are always implicitly valid (link-time)
1682constants. These constants are explicitly referenced when the <a
1683href="#identifiers">identifier for the global</a> is used and always have <a
1684href="#t_pointer">pointer</a> type. For example, the following is a legal LLVM
1685file:</p>
1686
1687<div class="doc_code">
1688<pre>
1689@X = global i32 17
1690@Y = global i32 42
1691@Z = global [2 x i32*] [ i32* @X, i32* @Y ]
1692</pre>
1693</div>
1694
1695</div>
1696
1697<!-- ======================================================================= -->
1698<div class="doc_subsection"><a name="undefvalues">Undefined Values</a></div>
1699<div class="doc_text">
1700 <p>The string '<tt>undef</tt>' is recognized as a type-less constant that has
1701 no specific value. Undefined values may be of any type and be used anywhere
1702 a constant is permitted.</p>
1703
1704 <p>Undefined values indicate to the compiler that the program is well defined
1705 no matter what value is used, giving the compiler more freedom to optimize.
1706 </p>
1707</div>
1708
1709<!-- ======================================================================= -->
1710<div class="doc_subsection"><a name="constantexprs">Constant Expressions</a>
1711</div>
1712
1713<div class="doc_text">
1714
1715<p>Constant expressions are used to allow expressions involving other constants
1716to be used as constants. Constant expressions may be of any <a
1717href="#t_firstclass">first class</a> type and may involve any LLVM operation
1718that does not have side effects (e.g. load and call are not supported). The
1719following is the syntax for constant expressions:</p>
1720
1721<dl>
1722 <dt><b><tt>trunc ( CST to TYPE )</tt></b></dt>
1723 <dd>Truncate a constant to another type. The bit size of CST must be larger
1724 than the bit size of TYPE. Both types must be integers.</dd>
1725
1726 <dt><b><tt>zext ( CST to TYPE )</tt></b></dt>
1727 <dd>Zero extend a constant to another type. The bit size of CST must be
1728 smaller or equal to the bit size of TYPE. Both types must be integers.</dd>
1729
1730 <dt><b><tt>sext ( CST to TYPE )</tt></b></dt>
1731 <dd>Sign extend a constant to another type. The bit size of CST must be
1732 smaller or equal to the bit size of TYPE. Both types must be integers.</dd>
1733
1734 <dt><b><tt>fptrunc ( CST to TYPE )</tt></b></dt>
1735 <dd>Truncate a floating point constant to another floating point type. The
1736 size of CST must be larger than the size of TYPE. Both types must be
1737 floating point.</dd>
1738
1739 <dt><b><tt>fpext ( CST to TYPE )</tt></b></dt>
1740 <dd>Floating point extend a constant to another type. The size of CST must be
1741 smaller or equal to the size of TYPE. Both types must be floating point.</dd>
1742
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]1743 <dt><b><tt>fptoui ( CST to TYPE )</tt></b></dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1744 <dd>Convert a floating point constant to the corresponding unsigned integer
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]1745 constant. TYPE must be a scalar or vector integer type. CST must be of scalar
1746 or vector floating point type. Both CST and TYPE must be scalars, or vectors
1747 of the same number of elements. If the value won't fit in the integer type,
1748 the results are undefined.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1749
1750 <dt><b><tt>fptosi ( CST to TYPE )</tt></b></dt>
1751 <dd>Convert a floating point constant to the corresponding signed integer
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]1752 constant. TYPE must be a scalar or vector integer type. CST must be of scalar
1753 or vector floating point type. Both CST and TYPE must be scalars, or vectors
1754 of the same number of elements. If the value won't fit in the integer type,
1755 the results are undefined.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1756
1757 <dt><b><tt>uitofp ( CST to TYPE )</tt></b></dt>
1758 <dd>Convert an unsigned integer constant to the corresponding floating point
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]1759 constant. TYPE must be a scalar or vector floating point type. CST must be of
1760 scalar or vector integer type. Both CST and TYPE must be scalars, or vectors
1761 of the same number of elements. If the value won't fit in the floating point
1762 type, the results are undefined.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1763
1764 <dt><b><tt>sitofp ( CST to TYPE )</tt></b></dt>
1765 <dd>Convert a signed integer constant to the corresponding floating point
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]1766 constant. TYPE must be a scalar or vector floating point type. CST must be of
1767 scalar or vector integer type. Both CST and TYPE must be scalars, or vectors
1768 of the same number of elements. If the value won't fit in the floating point
1769 type, the results are undefined.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1770
1771 <dt><b><tt>ptrtoint ( CST to TYPE )</tt></b></dt>
1772 <dd>Convert a pointer typed constant to the corresponding integer constant
1773 TYPE must be an integer type. CST must be of pointer type. The CST value is
1774 zero extended, truncated, or unchanged to make it fit in TYPE.</dd>
1775
1776 <dt><b><tt>inttoptr ( CST to TYPE )</tt></b></dt>
1777 <dd>Convert a integer constant to a pointer constant. TYPE must be a
1778 pointer type. CST must be of integer type. The CST value is zero extended,
1779 truncated, or unchanged to make it fit in a pointer size. This one is
1780 <i>really</i> dangerous!</dd>
1781
1782 <dt><b><tt>bitcast ( CST to TYPE )</tt></b></dt>
1783 <dd>Convert a constant, CST, to another TYPE. The size of CST and TYPE must be
1784 identical (same number of bits). The conversion is done as if the CST value
1785 was stored to memory and read back as TYPE. In other words, no bits change
1786 with this operator, just the type. This can be used for conversion of
1787 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]1788 pointers it is only valid to cast to another pointer type. It is not valid
1789 to bitcast to or from an aggregate type.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1790 </dd>
1791
1792 <dt><b><tt>getelementptr ( CSTPTR, IDX0, IDX1, ... )</tt></b></dt>
1793
1794 <dd>Perform the <a href="#i_getelementptr">getelementptr operation</a> on
1795 constants. As with the <a href="#i_getelementptr">getelementptr</a>
1796 instruction, the index list may have zero or more indexes, which are required
1797 to make sense for the type of "CSTPTR".</dd>
1798
1799 <dt><b><tt>select ( COND, VAL1, VAL2 )</tt></b></dt>
1800
1801 <dd>Perform the <a href="#i_select">select operation</a> on
1802 constants.</dd>
1803
1804 <dt><b><tt>icmp COND ( VAL1, VAL2 )</tt></b></dt>
1805 <dd>Performs the <a href="#i_icmp">icmp operation</a> on constants.</dd>
1806
1807 <dt><b><tt>fcmp COND ( VAL1, VAL2 )</tt></b></dt>
1808 <dd>Performs the <a href="#i_fcmp">fcmp operation</a> on constants.</dd>
1809
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]1810 <dt><b><tt>vicmp COND ( VAL1, VAL2 )</tt></b></dt>
1811 <dd>Performs the <a href="#i_vicmp">vicmp operation</a> on constants.</dd>
1812
1813 <dt><b><tt>vfcmp COND ( VAL1, VAL2 )</tt></b></dt>
1814 <dd>Performs the <a href="#i_vfcmp">vfcmp operation</a> on constants.</dd>
1815
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1816 <dt><b><tt>extractelement ( VAL, IDX )</tt></b></dt>
1817
1818 <dd>Perform the <a href="#i_extractelement">extractelement
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]1819 operation</a> on constants.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1820
1821 <dt><b><tt>insertelement ( VAL, ELT, IDX )</tt></b></dt>
1822
1823 <dd>Perform the <a href="#i_insertelement">insertelement
1824 operation</a> on constants.</dd>
1825
1826
1827 <dt><b><tt>shufflevector ( VEC1, VEC2, IDXMASK )</tt></b></dt>
1828
1829 <dd>Perform the <a href="#i_shufflevector">shufflevector
1830 operation</a> on constants.</dd>
1831
1832 <dt><b><tt>OPCODE ( LHS, RHS )</tt></b></dt>
1833
1834 <dd>Perform the specified operation of the LHS and RHS constants. OPCODE may
1835 be any of the <a href="#binaryops">binary</a> or <a href="#bitwiseops">bitwise
1836 binary</a> operations. The constraints on operands are the same as those for
1837 the corresponding instruction (e.g. no bitwise operations on floating point
1838 values are allowed).</dd>
1839</dl>
1840</div>
1841
1842<!-- *********************************************************************** -->
1843<div class="doc_section"> <a name="othervalues">Other Values</a> </div>
1844<!-- *********************************************************************** -->
1845
1846<!-- ======================================================================= -->
1847<div class="doc_subsection">
1848<a name="inlineasm">Inline Assembler Expressions</a>
1849</div>
1850
1851<div class="doc_text">
1852
1853<p>
1854LLVM supports inline assembler expressions (as opposed to <a href="#moduleasm">
1855Module-Level Inline Assembly</a>) through the use of a special value. This
1856value represents the inline assembler as a string (containing the instructions
1857to emit), a list of operand constraints (stored as a string), and a flag that
1858indicates whether or not the inline asm expression has side effects. An example
1859inline assembler expression is:
1860</p>
1861
1862<div class="doc_code">
1863<pre>
1864i32 (i32) asm "bswap $0", "=r,r"
1865</pre>
1866</div>
1867
1868<p>
1869Inline assembler expressions may <b>only</b> be used as the callee operand of
1870a <a href="#i_call"><tt>call</tt> instruction</a>. Thus, typically we have:
1871</p>
1872
1873<div class="doc_code">
1874<pre>
1875%X = call i32 asm "<a href="#int_bswap">bswap</a> $0", "=r,r"(i32 %Y)
1876</pre>
1877</div>
1878
1879<p>
1880Inline asms with side effects not visible in the constraint list must be marked
1881as having side effects. This is done through the use of the
1882'<tt>sideeffect</tt>' keyword, like so:
1883</p>
1884
1885<div class="doc_code">
1886<pre>
1887call void asm sideeffect "eieio", ""()
1888</pre>
1889</div>
1890
1891<p>TODO: The format of the asm and constraints string still need to be
1892documented here. Constraints on what can be done (e.g. duplication, moving, etc
Chris Lattner1cdd64e2008-10-04 18:36:02 +0000[diff] [blame]1893need to be documented). This is probably best done by reference to another
1894document that covers inline asm from a holistic perspective.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1895</p>
1896
1897</div>
1898
1899<!-- *********************************************************************** -->
1900<div class="doc_section"> <a name="instref">Instruction Reference</a> </div>
1901<!-- *********************************************************************** -->
1902
1903<div class="doc_text">
1904
1905<p>The LLVM instruction set consists of several different
1906classifications of instructions: <a href="#terminators">terminator
1907instructions</a>, <a href="#binaryops">binary instructions</a>,
1908<a href="#bitwiseops">bitwise binary instructions</a>, <a
1909 href="#memoryops">memory instructions</a>, and <a href="#otherops">other
1910instructions</a>.</p>
1911
1912</div>
1913
1914<!-- ======================================================================= -->
1915<div class="doc_subsection"> <a name="terminators">Terminator
1916Instructions</a> </div>
1917
1918<div class="doc_text">
1919
1920<p>As mentioned <a href="#functionstructure">previously</a>, every
1921basic block in a program ends with a "Terminator" instruction, which
1922indicates which block should be executed after the current block is
1923finished. These terminator instructions typically yield a '<tt>void</tt>'
1924value: they produce control flow, not values (the one exception being
1925the '<a href="#i_invoke"><tt>invoke</tt></a>' instruction).</p>
1926<p>There are six different terminator instructions: the '<a
1927 href="#i_ret"><tt>ret</tt></a>' instruction, the '<a href="#i_br"><tt>br</tt></a>'
1928instruction, the '<a href="#i_switch"><tt>switch</tt></a>' instruction,
1929the '<a href="#i_invoke"><tt>invoke</tt></a>' instruction, the '<a
1930 href="#i_unwind"><tt>unwind</tt></a>' instruction, and the '<a
1931 href="#i_unreachable"><tt>unreachable</tt></a>' instruction.</p>
1932
1933</div>
1934
1935<!-- _______________________________________________________________________ -->
1936<div class="doc_subsubsection"> <a name="i_ret">'<tt>ret</tt>'
1937Instruction</a> </div>
1938<div class="doc_text">
1939<h5>Syntax:</h5>
Dan Gohman3e700032008-10-04 19:00:07 +0000[diff] [blame]1940<pre>
1941 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]1942 ret void <i>; Return from void function</i>
1943</pre>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1944
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1945<h5>Overview:</h5>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1946
Dan Gohman3e700032008-10-04 19:00:07 +0000[diff] [blame]1947<p>The '<tt>ret</tt>' instruction is used to return control flow (and
1948optionally a value) from a function back to the caller.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1949<p>There are two forms of the '<tt>ret</tt>' instruction: one that
Dan Gohman3e700032008-10-04 19:00:07 +0000[diff] [blame]1950returns a value and then causes control flow, and one that just causes
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1951control flow to occur.</p>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1952
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1953<h5>Arguments:</h5>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1954
Dan Gohman3e700032008-10-04 19:00:07 +0000[diff] [blame]1955<p>The '<tt>ret</tt>' instruction optionally accepts a single argument,
1956the return value. The type of the return value must be a
1957'<a href="#t_firstclass">first class</a>' type.</p>
1958
1959<p>A function is not <a href="#wellformed">well formed</a> if
1960it it has a non-void return type and contains a '<tt>ret</tt>'
1961instruction with no return value or a return value with a type that
1962does not match its type, or if it has a void return type and contains
1963a '<tt>ret</tt>' instruction with a return value.</p>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1964
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1965<h5>Semantics:</h5>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1966
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1967<p>When the '<tt>ret</tt>' instruction is executed, control flow
1968returns back to the calling function's context. If the caller is a "<a
1969 href="#i_call"><tt>call</tt></a>" instruction, execution continues at
1970the instruction after the call. If the caller was an "<a
1971 href="#i_invoke"><tt>invoke</tt></a>" instruction, execution continues
1972at the beginning of the "normal" destination block. If the instruction
1973returns a value, that value shall set the call or invoke instruction's
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]1974return value.</p>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1975
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1976<h5>Example:</h5>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1977
1978<pre>
1979 ret i32 5 <i>; Return an integer value of 5</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1980 ret void <i>; Return from a void function</i>
Dan Gohman3e700032008-10-04 19:00:07 +0000[diff] [blame]1981 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]1982</pre>
1983</div>
1984<!-- _______________________________________________________________________ -->
1985<div class="doc_subsubsection"> <a name="i_br">'<tt>br</tt>' Instruction</a> </div>
1986<div class="doc_text">
1987<h5>Syntax:</h5>
1988<pre> br i1 &lt;cond&gt;, label &lt;iftrue&gt;, label &lt;iffalse&gt;<br> br label &lt;dest&gt; <i>; Unconditional branch</i>
1989</pre>
1990<h5>Overview:</h5>
1991<p>The '<tt>br</tt>' instruction is used to cause control flow to
1992transfer to a different basic block in the current function. There are
1993two forms of this instruction, corresponding to a conditional branch
1994and an unconditional branch.</p>
1995<h5>Arguments:</h5>
1996<p>The conditional branch form of the '<tt>br</tt>' instruction takes a
1997single '<tt>i1</tt>' value and two '<tt>label</tt>' values. The
1998unconditional form of the '<tt>br</tt>' instruction takes a single
1999'<tt>label</tt>' value as a target.</p>
2000<h5>Semantics:</h5>
2001<p>Upon execution of a conditional '<tt>br</tt>' instruction, the '<tt>i1</tt>'
2002argument is evaluated. If the value is <tt>true</tt>, control flows
2003to the '<tt>iftrue</tt>' <tt>label</tt> argument. If "cond" is <tt>false</tt>,
2004control flows to the '<tt>iffalse</tt>' <tt>label</tt> argument.</p>
2005<h5>Example:</h5>
2006<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
2007 href="#i_ret">ret</a> i32 1<br>IfUnequal:<br> <a href="#i_ret">ret</a> i32 0<br></pre>
2008</div>
2009<!-- _______________________________________________________________________ -->
2010<div class="doc_subsubsection">
2011 <a name="i_switch">'<tt>switch</tt>' Instruction</a>
2012</div>
2013
2014<div class="doc_text">
2015<h5>Syntax:</h5>
2016
2017<pre>
2018 switch &lt;intty&gt; &lt;value&gt;, label &lt;defaultdest&gt; [ &lt;intty&gt; &lt;val&gt;, label &lt;dest&gt; ... ]
2019</pre>
2020
2021<h5>Overview:</h5>
2022
2023<p>The '<tt>switch</tt>' instruction is used to transfer control flow to one of
2024several different places. It is a generalization of the '<tt>br</tt>'
2025instruction, allowing a branch to occur to one of many possible
2026destinations.</p>
2027
2028
2029<h5>Arguments:</h5>
2030
2031<p>The '<tt>switch</tt>' instruction uses three parameters: an integer
2032comparison value '<tt>value</tt>', a default '<tt>label</tt>' destination, and
2033an array of pairs of comparison value constants and '<tt>label</tt>'s. The
2034table is not allowed to contain duplicate constant entries.</p>
2035
2036<h5>Semantics:</h5>
2037
2038<p>The <tt>switch</tt> instruction specifies a table of values and
2039destinations. When the '<tt>switch</tt>' instruction is executed, this
2040table is searched for the given value. If the value is found, control flow is
2041transfered to the corresponding destination; otherwise, control flow is
2042transfered to the default destination.</p>
2043
2044<h5>Implementation:</h5>
2045
2046<p>Depending on properties of the target machine and the particular
2047<tt>switch</tt> instruction, this instruction may be code generated in different
2048ways. For example, it could be generated as a series of chained conditional
2049branches or with a lookup table.</p>
2050
2051<h5>Example:</h5>
2052
2053<pre>
2054 <i>; Emulate a conditional br instruction</i>
2055 %Val = <a href="#i_zext">zext</a> i1 %value to i32
2056 switch i32 %Val, label %truedest [i32 0, label %falsedest ]
2057
2058 <i>; Emulate an unconditional br instruction</i>
2059 switch i32 0, label %dest [ ]
2060
2061 <i>; Implement a jump table:</i>
2062 switch i32 %val, label %otherwise [ i32 0, label %onzero
2063 i32 1, label %onone
2064 i32 2, label %ontwo ]
2065</pre>
2066</div>
2067
2068<!-- _______________________________________________________________________ -->
2069<div class="doc_subsubsection">
2070 <a name="i_invoke">'<tt>invoke</tt>' Instruction</a>
2071</div>
2072
2073<div class="doc_text">
2074
2075<h5>Syntax:</h5>
2076
2077<pre>
Devang Pateld0bfcc72008-10-07 17:48:33 +0000[diff] [blame]2078 &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]2079 to label &lt;normal label&gt; unwind label &lt;exception label&gt;
2080</pre>
2081
2082<h5>Overview:</h5>
2083
2084<p>The '<tt>invoke</tt>' instruction causes control to transfer to a specified
2085function, with the possibility of control flow transfer to either the
2086'<tt>normal</tt>' label or the
2087'<tt>exception</tt>' label. If the callee function returns with the
2088"<tt><a href="#i_ret">ret</a></tt>" instruction, control flow will return to the
2089"normal" label. If the callee (or any indirect callees) returns with the "<a
2090href="#i_unwind"><tt>unwind</tt></a>" instruction, control is interrupted and
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]2091continued at the dynamically nearest "exception" label.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2092
2093<h5>Arguments:</h5>
2094
2095<p>This instruction requires several arguments:</p>
2096
2097<ol>
2098 <li>
2099 The optional "cconv" marker indicates which <a href="#callingconv">calling
2100 convention</a> the call should use. If none is specified, the call defaults
2101 to using C calling conventions.
2102 </li>
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]2103
2104 <li>The optional <a href="#paramattrs">Parameter Attributes</a> list for
2105 return values. Only '<tt>zeroext</tt>', '<tt>signext</tt>',
2106 and '<tt>inreg</tt>' attributes are valid here.</li>
2107
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2108 <li>'<tt>ptr to function ty</tt>': shall be the signature of the pointer to
2109 function value being invoked. In most cases, this is a direct function
2110 invocation, but indirect <tt>invoke</tt>s are just as possible, branching off
2111 an arbitrary pointer to function value.
2112 </li>
2113
2114 <li>'<tt>function ptr val</tt>': An LLVM value containing a pointer to a
2115 function to be invoked. </li>
2116
2117 <li>'<tt>function args</tt>': argument list whose types match the function
2118 signature argument types. If the function signature indicates the function
2119 accepts a variable number of arguments, the extra arguments can be
2120 specified. </li>
2121
2122 <li>'<tt>normal label</tt>': the label reached when the called function
2123 executes a '<tt><a href="#i_ret">ret</a></tt>' instruction. </li>
2124
2125 <li>'<tt>exception label</tt>': the label reached when a callee returns with
2126 the <a href="#i_unwind"><tt>unwind</tt></a> instruction. </li>
2127
Devang Pateld0bfcc72008-10-07 17:48:33 +0000[diff] [blame]2128 <li>The optional <a href="#fnattrs">function attributes</a> list. Only
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]2129 '<tt>noreturn</tt>', '<tt>nounwind</tt>', '<tt>readonly</tt>' and
2130 '<tt>readnone</tt>' attributes are valid here.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2131</ol>
2132
2133<h5>Semantics:</h5>
2134
2135<p>This instruction is designed to operate as a standard '<tt><a
2136href="#i_call">call</a></tt>' instruction in most regards. The primary
2137difference is that it establishes an association with a label, which is used by
2138the runtime library to unwind the stack.</p>
2139
2140<p>This instruction is used in languages with destructors to ensure that proper
2141cleanup is performed in the case of either a <tt>longjmp</tt> or a thrown
2142exception. Additionally, this is important for implementation of
2143'<tt>catch</tt>' clauses in high-level languages that support them.</p>
2144
2145<h5>Example:</h5>
2146<pre>
Nick Lewyckya1c11a12008-03-16 07:18:12 +0000[diff] [blame]2147 %retval = invoke i32 @Test(i32 15) to label %Continue
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2148 unwind label %TestCleanup <i>; {i32}:retval set</i>
Nick Lewyckya1c11a12008-03-16 07:18:12 +0000[diff] [blame]2149 %retval = invoke <a href="#callingconv">coldcc</a> i32 %Testfnptr(i32 15) to label %Continue
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2150 unwind label %TestCleanup <i>; {i32}:retval set</i>
2151</pre>
2152</div>
2153
2154
2155<!-- _______________________________________________________________________ -->
2156
2157<div class="doc_subsubsection"> <a name="i_unwind">'<tt>unwind</tt>'
2158Instruction</a> </div>
2159
2160<div class="doc_text">
2161
2162<h5>Syntax:</h5>
2163<pre>
2164 unwind
2165</pre>
2166
2167<h5>Overview:</h5>
2168
2169<p>The '<tt>unwind</tt>' instruction unwinds the stack, continuing control flow
2170at the first callee in the dynamic call stack which used an <a
2171href="#i_invoke"><tt>invoke</tt></a> instruction to perform the call. This is
2172primarily used to implement exception handling.</p>
2173
2174<h5>Semantics:</h5>
2175
Chris Lattner8b094fc2008-04-19 21:01:16 +0000[diff] [blame]2176<p>The '<tt>unwind</tt>' instruction causes execution of the current function to
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2177immediately halt. The dynamic call stack is then searched for the first <a
2178href="#i_invoke"><tt>invoke</tt></a> instruction on the call stack. Once found,
2179execution continues at the "exceptional" destination block specified by the
2180<tt>invoke</tt> instruction. If there is no <tt>invoke</tt> instruction in the
2181dynamic call chain, undefined behavior results.</p>
2182</div>
2183
2184<!-- _______________________________________________________________________ -->
2185
2186<div class="doc_subsubsection"> <a name="i_unreachable">'<tt>unreachable</tt>'
2187Instruction</a> </div>
2188
2189<div class="doc_text">
2190
2191<h5>Syntax:</h5>
2192<pre>
2193 unreachable
2194</pre>
2195
2196<h5>Overview:</h5>
2197
2198<p>The '<tt>unreachable</tt>' instruction has no defined semantics. This
2199instruction is used to inform the optimizer that a particular portion of the
2200code is not reachable. This can be used to indicate that the code after a
2201no-return function cannot be reached, and other facts.</p>
2202
2203<h5>Semantics:</h5>
2204
2205<p>The '<tt>unreachable</tt>' instruction has no defined semantics.</p>
2206</div>
2207
2208
2209
2210<!-- ======================================================================= -->
2211<div class="doc_subsection"> <a name="binaryops">Binary Operations</a> </div>
2212<div class="doc_text">
2213<p>Binary operators are used to do most of the computation in a
Chris Lattnerab596d92008-04-01 18:47:32 +0000[diff] [blame]2214program. They require two operands of the same type, execute an operation on them, and
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2215produce a single value. The operands might represent
2216multiple 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]2217The result value has the same type as its operands.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2218<p>There are several different binary operators:</p>
2219</div>
2220<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2221<div class="doc_subsubsection">
2222 <a name="i_add">'<tt>add</tt>' Instruction</a>
2223</div>
2224
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2225<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2226
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2227<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2228
2229<pre>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2230 &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]2231</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2232
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2233<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2234
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2235<p>The '<tt>add</tt>' instruction returns the sum of its two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2236
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2237<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2238
2239<p>The two arguments to the '<tt>add</tt>' instruction must be <a
2240 href="#t_integer">integer</a>, <a href="#t_floating">floating point</a>, or
2241 <a href="#t_vector">vector</a> values. Both arguments must have identical
2242 types.</p>
2243
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2244<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2245
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2246<p>The value produced is the integer or floating point sum of the two
2247operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2248
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2249<p>If an integer sum has unsigned overflow, the result returned is the
2250mathematical result modulo 2<sup>n</sup>, where n is the bit width of
2251the result.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2252
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2253<p>Because LLVM integers use a two's complement representation, this
2254instruction is appropriate for both signed and unsigned integers.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2255
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2256<h5>Example:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2257
2258<pre>
2259 &lt;result&gt; = add i32 4, %var <i>; yields {i32}:result = 4 + %var</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2260</pre>
2261</div>
2262<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2263<div class="doc_subsubsection">
2264 <a name="i_sub">'<tt>sub</tt>' Instruction</a>
2265</div>
2266
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2267<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2268
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2269<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2270
2271<pre>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2272 &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]2273</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2274
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2275<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2276
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2277<p>The '<tt>sub</tt>' instruction returns the difference of its two
2278operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2279
2280<p>Note that the '<tt>sub</tt>' instruction is used to represent the
2281'<tt>neg</tt>' instruction present in most other intermediate
2282representations.</p>
2283
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2284<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2285
2286<p>The two arguments to the '<tt>sub</tt>' instruction must be <a
2287 href="#t_integer">integer</a>, <a href="#t_floating">floating point</a>,
2288 or <a href="#t_vector">vector</a> values. Both arguments must have identical
2289 types.</p>
2290
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2291<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2292
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2293<p>The value produced is the integer or floating point difference of
2294the two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2295
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2296<p>If an integer difference has unsigned overflow, the result returned is the
2297mathematical result modulo 2<sup>n</sup>, where n is the bit width of
2298the result.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2299
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2300<p>Because LLVM integers use a two's complement representation, this
2301instruction is appropriate for both signed and unsigned integers.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2302
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2303<h5>Example:</h5>
2304<pre>
2305 &lt;result&gt; = sub i32 4, %var <i>; yields {i32}:result = 4 - %var</i>
2306 &lt;result&gt; = sub i32 0, %val <i>; yields {i32}:result = -%var</i>
2307</pre>
2308</div>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2309
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2310<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2311<div class="doc_subsubsection">
2312 <a name="i_mul">'<tt>mul</tt>' Instruction</a>
2313</div>
2314
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2315<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2316
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2317<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2318<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]2319</pre>
2320<h5>Overview:</h5>
2321<p>The '<tt>mul</tt>' instruction returns the product of its two
2322operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2323
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2324<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2325
2326<p>The two arguments to the '<tt>mul</tt>' instruction must be <a
2327href="#t_integer">integer</a>, <a href="#t_floating">floating point</a>,
2328or <a href="#t_vector">vector</a> values. Both arguments must have identical
2329types.</p>
2330
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2331<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2332
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2333<p>The value produced is the integer or floating point product of the
2334two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2335
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2336<p>If the result of an integer multiplication has unsigned overflow,
2337the result returned is the mathematical result modulo
23382<sup>n</sup>, where n is the bit width of the result.</p>
2339<p>Because LLVM integers use a two's complement representation, and the
2340result is the same width as the operands, this instruction returns the
2341correct result for both signed and unsigned integers. If a full product
2342(e.g. <tt>i32</tt>x<tt>i32</tt>-><tt>i64</tt>) is needed, the operands
2343should be sign-extended or zero-extended as appropriate to the
2344width of the full product.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2345<h5>Example:</h5>
2346<pre> &lt;result&gt; = mul i32 4, %var <i>; yields {i32}:result = 4 * %var</i>
2347</pre>
2348</div>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2349
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2350<!-- _______________________________________________________________________ -->
2351<div class="doc_subsubsection"> <a name="i_udiv">'<tt>udiv</tt>' Instruction
2352</a></div>
2353<div class="doc_text">
2354<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2355<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]2356</pre>
2357<h5>Overview:</h5>
2358<p>The '<tt>udiv</tt>' instruction returns the quotient of its two
2359operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2360
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2361<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2362
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2363<p>The two arguments to the '<tt>udiv</tt>' instruction must be
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2364<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2365values. Both arguments must have identical types.</p>
2366
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2367<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2368
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2369<p>The value produced is the unsigned integer quotient of the two operands.</p>
2370<p>Note that unsigned integer division and signed integer division are distinct
2371operations; for signed integer division, use '<tt>sdiv</tt>'.</p>
2372<p>Division by zero leads to undefined behavior.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2373<h5>Example:</h5>
2374<pre> &lt;result&gt; = udiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
2375</pre>
2376</div>
2377<!-- _______________________________________________________________________ -->
2378<div class="doc_subsubsection"> <a name="i_sdiv">'<tt>sdiv</tt>' Instruction
2379</a> </div>
2380<div class="doc_text">
2381<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2382<pre>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2383 &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]2384</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2385
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2386<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2387
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2388<p>The '<tt>sdiv</tt>' instruction returns the quotient of its two
2389operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2390
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2391<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2392
2393<p>The two arguments to the '<tt>sdiv</tt>' instruction must be
2394<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2395values. Both arguments must have identical types.</p>
2396
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2397<h5>Semantics:</h5>
Chris Lattnerdc15b1d2008-04-01 18:45:27 +0000[diff] [blame]2398<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]2399<p>Note that signed integer division and unsigned integer division are distinct
2400operations; for unsigned integer division, use '<tt>udiv</tt>'.</p>
2401<p>Division by zero leads to undefined behavior. Overflow also leads to
2402undefined behavior; this is a rare case, but can occur, for example,
2403by doing a 32-bit division of -2147483648 by -1.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2404<h5>Example:</h5>
2405<pre> &lt;result&gt; = sdiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
2406</pre>
2407</div>
2408<!-- _______________________________________________________________________ -->
2409<div class="doc_subsubsection"> <a name="i_fdiv">'<tt>fdiv</tt>'
2410Instruction</a> </div>
2411<div class="doc_text">
2412<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2413<pre>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2414 &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]2415</pre>
2416<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2417
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2418<p>The '<tt>fdiv</tt>' instruction returns the quotient of its two
2419operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2420
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2421<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2422
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2423<p>The two arguments to the '<tt>fdiv</tt>' instruction must be
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2424<a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a>
2425of floating point values. Both arguments must have identical types.</p>
2426
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2427<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2428
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2429<p>The value produced is the floating point quotient of the two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2430
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2431<h5>Example:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2432
2433<pre>
2434 &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]2435</pre>
2436</div>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2437
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2438<!-- _______________________________________________________________________ -->
2439<div class="doc_subsubsection"> <a name="i_urem">'<tt>urem</tt>' Instruction</a>
2440</div>
2441<div class="doc_text">
2442<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2443<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]2444</pre>
2445<h5>Overview:</h5>
2446<p>The '<tt>urem</tt>' instruction returns the remainder from the
2447unsigned division of its two arguments.</p>
2448<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2449<p>The two arguments to the '<tt>urem</tt>' instruction must be
2450<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2451values. Both arguments must have identical types.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2452<h5>Semantics:</h5>
2453<p>This instruction returns the unsigned integer <i>remainder</i> of a division.
Chris Lattnerdc15b1d2008-04-01 18:45:27 +0000[diff] [blame]2454This instruction always performs an unsigned division to get the remainder.</p>
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2455<p>Note that unsigned integer remainder and signed integer remainder are
2456distinct operations; for signed integer remainder, use '<tt>srem</tt>'.</p>
2457<p>Taking the remainder of a division by zero leads to undefined behavior.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2458<h5>Example:</h5>
2459<pre> &lt;result&gt; = urem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
2460</pre>
2461
2462</div>
2463<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2464<div class="doc_subsubsection">
2465 <a name="i_srem">'<tt>srem</tt>' Instruction</a>
2466</div>
2467
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2468<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2469
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2470<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2471
2472<pre>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2473 &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]2474</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2475
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2476<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2477
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2478<p>The '<tt>srem</tt>' instruction returns the remainder from the
Dan Gohman3e3fd8c2007-11-05 23:35:22 +0000[diff] [blame]2479signed division of its two operands. This instruction can also take
2480<a href="#t_vector">vector</a> versions of the values in which case
2481the elements must be integers.</p>
Chris Lattner08497ce2008-01-04 04:33:49 +0000[diff] [blame]2482
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2483<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2484
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2485<p>The two arguments to the '<tt>srem</tt>' instruction must be
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2486<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2487values. Both arguments must have identical types.</p>
2488
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2489<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2490
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2491<p>This instruction returns the <i>remainder</i> of a division (where the result
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2492has the same sign as the dividend, <tt>op1</tt>), not the <i>modulo</i>
2493operator (where the result has the same sign as the divisor, <tt>op2</tt>) of
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2494a value. For more information about the difference, see <a
2495 href="http://mathforum.org/dr.math/problems/anne.4.28.99.html">The
2496Math Forum</a>. For a table of how this is implemented in various languages,
2497please see <a href="http://en.wikipedia.org/wiki/Modulo_operation">
2498Wikipedia: modulo operation</a>.</p>
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2499<p>Note that signed integer remainder and unsigned integer remainder are
2500distinct operations; for unsigned integer remainder, use '<tt>urem</tt>'.</p>
2501<p>Taking the remainder of a division by zero leads to undefined behavior.
2502Overflow also leads to undefined behavior; this is a rare case, but can occur,
2503for example, by taking the remainder of a 32-bit division of -2147483648 by -1.
2504(The remainder doesn't actually overflow, but this rule lets srem be
2505implemented using instructions that return both the result of the division
2506and the remainder.)</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2507<h5>Example:</h5>
2508<pre> &lt;result&gt; = srem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
2509</pre>
2510
2511</div>
2512<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2513<div class="doc_subsubsection">
2514 <a name="i_frem">'<tt>frem</tt>' Instruction</a> </div>
2515
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2516<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2517
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2518<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2519<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]2520</pre>
2521<h5>Overview:</h5>
2522<p>The '<tt>frem</tt>' instruction returns the remainder from the
2523division of its two operands.</p>
2524<h5>Arguments:</h5>
2525<p>The two arguments to the '<tt>frem</tt>' instruction must be
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2526<a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a>
2527of floating point values. Both arguments must have identical types.</p>
2528
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2529<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2530
Chris Lattnerdc15b1d2008-04-01 18:45:27 +0000[diff] [blame]2531<p>This instruction returns the <i>remainder</i> of a division.
2532The remainder has the same sign as the dividend.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2533
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2534<h5>Example:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2535
2536<pre>
2537 &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]2538</pre>
2539</div>
2540
2541<!-- ======================================================================= -->
2542<div class="doc_subsection"> <a name="bitwiseops">Bitwise Binary
2543Operations</a> </div>
2544<div class="doc_text">
2545<p>Bitwise binary operators are used to do various forms of
2546bit-twiddling in a program. They are generally very efficient
2547instructions and can commonly be strength reduced from other
Chris Lattnerdc15b1d2008-04-01 18:45:27 +0000[diff] [blame]2548instructions. They require two operands of the same type, execute an operation on them,
2549and 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]2550</div>
2551
2552<!-- _______________________________________________________________________ -->
2553<div class="doc_subsubsection"> <a name="i_shl">'<tt>shl</tt>'
2554Instruction</a> </div>
2555<div class="doc_text">
2556<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2557<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]2558</pre>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2559
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2560<h5>Overview:</h5>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2561
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2562<p>The '<tt>shl</tt>' instruction returns the first operand shifted to
2563the left a specified number of bits.</p>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2564
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2565<h5>Arguments:</h5>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2566
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2567<p>Both arguments to the '<tt>shl</tt>' instruction must be the same <a
Nate Begemanbb1ce942008-07-29 15:49:41 +0000[diff] [blame]2568 href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2569type. '<tt>op2</tt>' is treated as an unsigned value.</p>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2570
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2571<h5>Semantics:</h5>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2572
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2573<p>The value produced is <tt>op1</tt> * 2<sup><tt>op2</tt></sup> mod 2<sup>n</sup>,
2574where n is the width of the result. If <tt>op2</tt> is (statically or dynamically) negative or
2575equal 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]2576
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2577<h5>Example:</h5><pre>
2578 &lt;result&gt; = shl i32 4, %var <i>; yields {i32}: 4 &lt;&lt; %var</i>
2579 &lt;result&gt; = shl i32 4, 2 <i>; yields {i32}: 16</i>
2580 &lt;result&gt; = shl i32 1, 10 <i>; yields {i32}: 1024</i>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2581 &lt;result&gt; = shl i32 1, 32 <i>; undefined</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2582</pre>
2583</div>
2584<!-- _______________________________________________________________________ -->
2585<div class="doc_subsubsection"> <a name="i_lshr">'<tt>lshr</tt>'
2586Instruction</a> </div>
2587<div class="doc_text">
2588<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2589<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]2590</pre>
2591
2592<h5>Overview:</h5>
2593<p>The '<tt>lshr</tt>' instruction (logical shift right) returns the first
2594operand shifted to the right a specified number of bits with zero fill.</p>
2595
2596<h5>Arguments:</h5>
2597<p>Both arguments to the '<tt>lshr</tt>' instruction must be the same
Nate Begemanbb1ce942008-07-29 15:49:41 +0000[diff] [blame]2598<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]2599type. '<tt>op2</tt>' is treated as an unsigned value.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2600
2601<h5>Semantics:</h5>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2602
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2603<p>This instruction always performs a logical shift right operation. The most
2604significant bits of the result will be filled with zero bits after the
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2605shift. If <tt>op2</tt> is (statically or dynamically) equal to or larger than
2606the number of bits in <tt>op1</tt>, the result is undefined.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2607
2608<h5>Example:</h5>
2609<pre>
2610 &lt;result&gt; = lshr i32 4, 1 <i>; yields {i32}:result = 2</i>
2611 &lt;result&gt; = lshr i32 4, 2 <i>; yields {i32}:result = 1</i>
2612 &lt;result&gt; = lshr i8 4, 3 <i>; yields {i8}:result = 0</i>
2613 &lt;result&gt; = lshr i8 -2, 1 <i>; yields {i8}:result = 0x7FFFFFFF </i>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2614 &lt;result&gt; = lshr i32 1, 32 <i>; undefined</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2615</pre>
2616</div>
2617
2618<!-- _______________________________________________________________________ -->
2619<div class="doc_subsubsection"> <a name="i_ashr">'<tt>ashr</tt>'
2620Instruction</a> </div>
2621<div class="doc_text">
2622
2623<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2624<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]2625</pre>
2626
2627<h5>Overview:</h5>
2628<p>The '<tt>ashr</tt>' instruction (arithmetic shift right) returns the first
2629operand shifted to the right a specified number of bits with sign extension.</p>
2630
2631<h5>Arguments:</h5>
2632<p>Both arguments to the '<tt>ashr</tt>' instruction must be the same
Nate Begemanbb1ce942008-07-29 15:49:41 +0000[diff] [blame]2633<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]2634type. '<tt>op2</tt>' is treated as an unsigned value.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2635
2636<h5>Semantics:</h5>
2637<p>This instruction always performs an arithmetic shift right operation,
2638The most significant bits of the result will be filled with the sign bit
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2639of <tt>op1</tt>. If <tt>op2</tt> is (statically or dynamically) equal to or
2640larger than the number of bits in <tt>op1</tt>, the result is undefined.
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2641</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2642
2643<h5>Example:</h5>
2644<pre>
2645 &lt;result&gt; = ashr i32 4, 1 <i>; yields {i32}:result = 2</i>
2646 &lt;result&gt; = ashr i32 4, 2 <i>; yields {i32}:result = 1</i>
2647 &lt;result&gt; = ashr i8 4, 3 <i>; yields {i8}:result = 0</i>
2648 &lt;result&gt; = ashr i8 -2, 1 <i>; yields {i8}:result = -1</i>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2649 &lt;result&gt; = ashr i32 1, 32 <i>; undefined</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2650</pre>
2651</div>
2652
2653<!-- _______________________________________________________________________ -->
2654<div class="doc_subsubsection"> <a name="i_and">'<tt>and</tt>'
2655Instruction</a> </div>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2656
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2657<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2658
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2659<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2660
2661<pre>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2662 &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]2663</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2664
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2665<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2666
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2667<p>The '<tt>and</tt>' instruction returns the bitwise logical and of
2668its two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2669
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2670<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2671
2672<p>The two arguments to the '<tt>and</tt>' instruction must be
2673<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2674values. Both arguments must have identical types.</p>
2675
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2676<h5>Semantics:</h5>
2677<p>The truth table used for the '<tt>and</tt>' instruction is:</p>
2678<p> </p>
Bill Wendling1e19eac2008-09-07 10:29:20 +0000[diff] [blame]2679<div>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2680<table border="1" cellspacing="0" cellpadding="4">
2681 <tbody>
2682 <tr>
2683 <td>In0</td>
2684 <td>In1</td>
2685 <td>Out</td>
2686 </tr>
2687 <tr>
2688 <td>0</td>
2689 <td>0</td>
2690 <td>0</td>
2691 </tr>
2692 <tr>
2693 <td>0</td>
2694 <td>1</td>
2695 <td>0</td>
2696 </tr>
2697 <tr>
2698 <td>1</td>
2699 <td>0</td>
2700 <td>0</td>
2701 </tr>
2702 <tr>
2703 <td>1</td>
2704 <td>1</td>
2705 <td>1</td>
2706 </tr>
2707 </tbody>
2708</table>
2709</div>
2710<h5>Example:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2711<pre>
2712 &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]2713 &lt;result&gt; = and i32 15, 40 <i>; yields {i32}:result = 8</i>
2714 &lt;result&gt; = and i32 4, 8 <i>; yields {i32}:result = 0</i>
2715</pre>
2716</div>
2717<!-- _______________________________________________________________________ -->
2718<div class="doc_subsubsection"> <a name="i_or">'<tt>or</tt>' Instruction</a> </div>
2719<div class="doc_text">
2720<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2721<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]2722</pre>
2723<h5>Overview:</h5>
2724<p>The '<tt>or</tt>' instruction returns the bitwise logical inclusive
2725or of its two operands.</p>
2726<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2727
2728<p>The two arguments to the '<tt>or</tt>' instruction must be
2729<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2730values. Both arguments must have identical types.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2731<h5>Semantics:</h5>
2732<p>The truth table used for the '<tt>or</tt>' instruction is:</p>
2733<p> </p>
Bill Wendling1e19eac2008-09-07 10:29:20 +0000[diff] [blame]2734<div>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2735<table border="1" cellspacing="0" cellpadding="4">
2736 <tbody>
2737 <tr>
2738 <td>In0</td>
2739 <td>In1</td>
2740 <td>Out</td>
2741 </tr>
2742 <tr>
2743 <td>0</td>
2744 <td>0</td>
2745 <td>0</td>
2746 </tr>
2747 <tr>
2748 <td>0</td>
2749 <td>1</td>
2750 <td>1</td>
2751 </tr>
2752 <tr>
2753 <td>1</td>
2754 <td>0</td>
2755 <td>1</td>
2756 </tr>
2757 <tr>
2758 <td>1</td>
2759 <td>1</td>
2760 <td>1</td>
2761 </tr>
2762 </tbody>
2763</table>
2764</div>
2765<h5>Example:</h5>
2766<pre> &lt;result&gt; = or i32 4, %var <i>; yields {i32}:result = 4 | %var</i>
2767 &lt;result&gt; = or i32 15, 40 <i>; yields {i32}:result = 47</i>
2768 &lt;result&gt; = or i32 4, 8 <i>; yields {i32}:result = 12</i>
2769</pre>
2770</div>
2771<!-- _______________________________________________________________________ -->
2772<div class="doc_subsubsection"> <a name="i_xor">'<tt>xor</tt>'
2773Instruction</a> </div>
2774<div class="doc_text">
2775<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2776<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]2777</pre>
2778<h5>Overview:</h5>
2779<p>The '<tt>xor</tt>' instruction returns the bitwise logical exclusive
2780or of its two operands. The <tt>xor</tt> is used to implement the
2781"one's complement" operation, which is the "~" operator in C.</p>
2782<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2783<p>The two arguments to the '<tt>xor</tt>' instruction must be
2784<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2785values. Both arguments must have identical types.</p>
2786
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2787<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2788
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2789<p>The truth table used for the '<tt>xor</tt>' instruction is:</p>
2790<p> </p>
Bill Wendling1e19eac2008-09-07 10:29:20 +0000[diff] [blame]2791<div>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2792<table border="1" cellspacing="0" cellpadding="4">
2793 <tbody>
2794 <tr>
2795 <td>In0</td>
2796 <td>In1</td>
2797 <td>Out</td>
2798 </tr>
2799 <tr>
2800 <td>0</td>
2801 <td>0</td>
2802 <td>0</td>
2803 </tr>
2804 <tr>
2805 <td>0</td>
2806 <td>1</td>
2807 <td>1</td>
2808 </tr>
2809 <tr>
2810 <td>1</td>
2811 <td>0</td>
2812 <td>1</td>
2813 </tr>
2814 <tr>
2815 <td>1</td>
2816 <td>1</td>
2817 <td>0</td>
2818 </tr>
2819 </tbody>
2820</table>
2821</div>
2822<p> </p>
2823<h5>Example:</h5>
2824<pre> &lt;result&gt; = xor i32 4, %var <i>; yields {i32}:result = 4 ^ %var</i>
2825 &lt;result&gt; = xor i32 15, 40 <i>; yields {i32}:result = 39</i>
2826 &lt;result&gt; = xor i32 4, 8 <i>; yields {i32}:result = 12</i>
2827 &lt;result&gt; = xor i32 %V, -1 <i>; yields {i32}:result = ~%V</i>
2828</pre>
2829</div>
2830
2831<!-- ======================================================================= -->
2832<div class="doc_subsection">
2833 <a name="vectorops">Vector Operations</a>
2834</div>
2835
2836<div class="doc_text">
2837
2838<p>LLVM supports several instructions to represent vector operations in a
2839target-independent manner. These instructions cover the element-access and
2840vector-specific operations needed to process vectors effectively. While LLVM
2841does directly support these vector operations, many sophisticated algorithms
2842will want to use target-specific intrinsics to take full advantage of a specific
2843target.</p>
2844
2845</div>
2846
2847<!-- _______________________________________________________________________ -->
2848<div class="doc_subsubsection">
2849 <a name="i_extractelement">'<tt>extractelement</tt>' Instruction</a>
2850</div>
2851
2852<div class="doc_text">
2853
2854<h5>Syntax:</h5>
2855
2856<pre>
2857 &lt;result&gt; = extractelement &lt;n x &lt;ty&gt;&gt; &lt;val&gt;, i32 &lt;idx&gt; <i>; yields &lt;ty&gt;</i>
2858</pre>
2859
2860<h5>Overview:</h5>
2861
2862<p>
2863The '<tt>extractelement</tt>' instruction extracts a single scalar
2864element from a vector at a specified index.
2865</p>
2866
2867
2868<h5>Arguments:</h5>
2869
2870<p>
2871The first operand of an '<tt>extractelement</tt>' instruction is a
2872value of <a href="#t_vector">vector</a> type. The second operand is
2873an index indicating the position from which to extract the element.
2874The index may be a variable.</p>
2875
2876<h5>Semantics:</h5>
2877
2878<p>
2879The result is a scalar of the same type as the element type of
2880<tt>val</tt>. Its value is the value at position <tt>idx</tt> of
2881<tt>val</tt>. If <tt>idx</tt> exceeds the length of <tt>val</tt>, the
2882results are undefined.
2883</p>
2884
2885<h5>Example:</h5>
2886
2887<pre>
2888 %result = extractelement &lt;4 x i32&gt; %vec, i32 0 <i>; yields i32</i>
2889</pre>
2890</div>
2891
2892
2893<!-- _______________________________________________________________________ -->
2894<div class="doc_subsubsection">
2895 <a name="i_insertelement">'<tt>insertelement</tt>' Instruction</a>
2896</div>
2897
2898<div class="doc_text">
2899
2900<h5>Syntax:</h5>
2901
2902<pre>
Dan Gohmanbcc3c502008-05-12 23:38:42 +0000[diff] [blame]2903 &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]2904</pre>
2905
2906<h5>Overview:</h5>
2907
2908<p>
2909The '<tt>insertelement</tt>' instruction inserts a scalar
2910element into a vector at a specified index.
2911</p>
2912
2913
2914<h5>Arguments:</h5>
2915
2916<p>
2917The first operand of an '<tt>insertelement</tt>' instruction is a
2918value of <a href="#t_vector">vector</a> type. The second operand is a
2919scalar value whose type must equal the element type of the first
2920operand. The third operand is an index indicating the position at
2921which to insert the value. The index may be a variable.</p>
2922
2923<h5>Semantics:</h5>
2924
2925<p>
2926The result is a vector of the same type as <tt>val</tt>. Its
2927element values are those of <tt>val</tt> except at position
2928<tt>idx</tt>, where it gets the value <tt>elt</tt>. If <tt>idx</tt>
2929exceeds the length of <tt>val</tt>, the results are undefined.
2930</p>
2931
2932<h5>Example:</h5>
2933
2934<pre>
2935 %result = insertelement &lt;4 x i32&gt; %vec, i32 1, i32 0 <i>; yields &lt;4 x i32&gt;</i>
2936</pre>
2937</div>
2938
2939<!-- _______________________________________________________________________ -->
2940<div class="doc_subsubsection">
2941 <a name="i_shufflevector">'<tt>shufflevector</tt>' Instruction</a>
2942</div>
2943
2944<div class="doc_text">
2945
2946<h5>Syntax:</h5>
2947
2948<pre>
Mon P Wangbff5d9c2008-11-10 04:46:22 +0000[diff] [blame]2949 &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]2950</pre>
2951
2952<h5>Overview:</h5>
2953
2954<p>
2955The '<tt>shufflevector</tt>' instruction constructs a permutation of elements
Mon P Wangbff5d9c2008-11-10 04:46:22 +0000[diff] [blame]2956from two input vectors, returning a vector with the same element type as
2957the input and length that is the same as the shuffle mask.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2958</p>
2959
2960<h5>Arguments:</h5>
2961
2962<p>
Mon P Wangbff5d9c2008-11-10 04:46:22 +0000[diff] [blame]2963The first two operands of a '<tt>shufflevector</tt>' instruction are vectors
2964with types that match each other. The third argument is a shuffle mask whose
2965element type is always 'i32'. The result of the instruction is a vector whose
2966length is the same as the shuffle mask and whose element type is the same as
2967the element type of the first two operands.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2968</p>
2969
2970<p>
2971The shuffle mask operand is required to be a constant vector with either
2972constant integer or undef values.
2973</p>
2974
2975<h5>Semantics:</h5>
2976
2977<p>
2978The elements of the two input vectors are numbered from left to right across
2979both of the vectors. The shuffle mask operand specifies, for each element of
Mon P Wangbff5d9c2008-11-10 04:46:22 +0000[diff] [blame]2980the result vector, which element of the two input vectors the result element
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2981gets. The element selector may be undef (meaning "don't care") and the second
2982operand may be undef if performing a shuffle from only one vector.
2983</p>
2984
2985<h5>Example:</h5>
2986
2987<pre>
2988 %result = shufflevector &lt;4 x i32&gt; %v1, &lt;4 x i32&gt; %v2,
2989 &lt;4 x i32&gt; &lt;i32 0, i32 4, i32 1, i32 5&gt; <i>; yields &lt;4 x i32&gt;</i>
2990 %result = shufflevector &lt;4 x i32&gt; %v1, &lt;4 x i32&gt; undef,
2991 &lt;4 x i32&gt; &lt;i32 0, i32 1, i32 2, i32 3&gt; <i>; yields &lt;4 x i32&gt;</i> - Identity shuffle.
Mon P Wangbff5d9c2008-11-10 04:46:22 +0000[diff] [blame]2992 %result = shufflevector &lt;8 x i32&gt; %v1, &lt;8 x i32&gt; undef,
2993 &lt;4 x i32&gt; &lt;i32 0, i32 1, i32 2, i32 3&gt; <i>; yields &lt;4 x i32&gt;</i>
2994 %result = shufflevector &lt;4 x i32&gt; %v1, &lt;4 x i32&gt; %v2,
2995 &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]2996</pre>
2997</div>
2998
2999
3000<!-- ======================================================================= -->
3001<div class="doc_subsection">
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3002 <a name="aggregateops">Aggregate Operations</a>
3003</div>
3004
3005<div class="doc_text">
3006
3007<p>LLVM supports several instructions for working with aggregate values.
3008</p>
3009
3010</div>
3011
3012<!-- _______________________________________________________________________ -->
3013<div class="doc_subsubsection">
3014 <a name="i_extractvalue">'<tt>extractvalue</tt>' Instruction</a>
3015</div>
3016
3017<div class="doc_text">
3018
3019<h5>Syntax:</h5>
3020
3021<pre>
3022 &lt;result&gt; = extractvalue &lt;aggregate type&gt; &lt;val&gt;, &lt;idx&gt;{, &lt;idx&gt;}*
3023</pre>
3024
3025<h5>Overview:</h5>
3026
3027<p>
Dan Gohman6c6dea02008-05-13 18:16:06 +0000[diff] [blame]3028The '<tt>extractvalue</tt>' instruction extracts the value of a struct field
3029or array element from an aggregate value.
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3030</p>
3031
3032
3033<h5>Arguments:</h5>
3034
3035<p>
3036The first operand of an '<tt>extractvalue</tt>' instruction is a
3037value of <a href="#t_struct">struct</a> or <a href="#t_array">array</a>
Dan Gohman6c6dea02008-05-13 18:16:06 +0000[diff] [blame]3038type. The operands are constant indices to specify which value to extract
Dan Gohmane5febe42008-05-31 00:58:22 +0000[diff] [blame]3039in a similar manner as indices in a
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3040'<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.
3041</p>
3042
3043<h5>Semantics:</h5>
3044
3045<p>
3046The result is the value at the position in the aggregate specified by
3047the index operands.
3048</p>
3049
3050<h5>Example:</h5>
3051
3052<pre>
Dan Gohmane5febe42008-05-31 00:58:22 +0000[diff] [blame]3053 %result = extractvalue {i32, float} %agg, 0 <i>; yields i32</i>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3054</pre>
3055</div>
3056
3057
3058<!-- _______________________________________________________________________ -->
3059<div class="doc_subsubsection">
3060 <a name="i_insertvalue">'<tt>insertvalue</tt>' Instruction</a>
3061</div>
3062
3063<div class="doc_text">
3064
3065<h5>Syntax:</h5>
3066
3067<pre>
Dan Gohmane5febe42008-05-31 00:58:22 +0000[diff] [blame]3068 &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]3069</pre>
3070
3071<h5>Overview:</h5>
3072
3073<p>
3074The '<tt>insertvalue</tt>' instruction inserts a value
Dan Gohman6c6dea02008-05-13 18:16:06 +0000[diff] [blame]3075into a struct field or array element in an aggregate.
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3076</p>
3077
3078
3079<h5>Arguments:</h5>
3080
3081<p>
3082The first operand of an '<tt>insertvalue</tt>' instruction is a
3083value of <a href="#t_struct">struct</a> or <a href="#t_array">array</a> type.
3084The second operand is a first-class value to insert.
Dan Gohman4f29e422008-05-23 21:53:15 +0000[diff] [blame]3085The following operands are constant indices
Dan Gohmane5febe42008-05-31 00:58:22 +0000[diff] [blame]3086indicating the position at which to insert the value in a similar manner as
Dan Gohman6c6dea02008-05-13 18:16:06 +0000[diff] [blame]3087indices in a
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3088'<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.
3089The value to insert must have the same type as the value identified
Dan Gohman6c6dea02008-05-13 18:16:06 +0000[diff] [blame]3090by the indices.
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]3091</p>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3092
3093<h5>Semantics:</h5>
3094
3095<p>
3096The result is an aggregate of the same type as <tt>val</tt>. Its
3097value is that of <tt>val</tt> except that the value at the position
Dan Gohman6c6dea02008-05-13 18:16:06 +0000[diff] [blame]3098specified by the indices is that of <tt>elt</tt>.
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3099</p>
3100
3101<h5>Example:</h5>
3102
3103<pre>
Dan Gohmanb1aab4e2008-06-23 15:26:37 +0000[diff] [blame]3104 %result = insertvalue {i32, float} %agg, i32 1, 0 <i>; yields {i32, float}</i>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3105</pre>
3106</div>
3107
3108
3109<!-- ======================================================================= -->
3110<div class="doc_subsection">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3111 <a name="memoryops">Memory Access and Addressing Operations</a>
3112</div>
3113
3114<div class="doc_text">
3115
3116<p>A key design point of an SSA-based representation is how it
3117represents memory. In LLVM, no memory locations are in SSA form, which
3118makes things very simple. This section describes how to read, write,
3119allocate, and free memory in LLVM.</p>
3120
3121</div>
3122
3123<!-- _______________________________________________________________________ -->
3124<div class="doc_subsubsection">
3125 <a name="i_malloc">'<tt>malloc</tt>' Instruction</a>
3126</div>
3127
3128<div class="doc_text">
3129
3130<h5>Syntax:</h5>
3131
3132<pre>
3133 &lt;result&gt; = malloc &lt;type&gt;[, i32 &lt;NumElements&gt;][, align &lt;alignment&gt;] <i>; yields {type*}:result</i>
3134</pre>
3135
3136<h5>Overview:</h5>
3137
3138<p>The '<tt>malloc</tt>' instruction allocates memory from the system
Christopher Lambcfe00962007-12-17 01:00:21 +0000[diff] [blame]3139heap and returns a pointer to it. The object is always allocated in the generic
3140address space (address space zero).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3141
3142<h5>Arguments:</h5>
3143
3144<p>The '<tt>malloc</tt>' instruction allocates
3145<tt>sizeof(&lt;type&gt;)*NumElements</tt>
3146bytes of memory from the operating system and returns a pointer of the
3147appropriate type to the program. If "NumElements" is specified, it is the
Gabor Greif5082cf42008-02-09 22:24:34 +0000[diff] [blame]3148number of elements allocated, otherwise "NumElements" is defaulted to be one.
Chris Lattner10368b62008-04-02 00:38:26 +0000[diff] [blame]3149If a constant alignment is specified, the value result of the allocation is guaranteed to
Gabor Greif5082cf42008-02-09 22:24:34 +0000[diff] [blame]3150be aligned to at least that boundary. If not specified, or if zero, the target can
3151choose to align the allocation on any convenient boundary.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3152
3153<p>'<tt>type</tt>' must be a sized type.</p>
3154
3155<h5>Semantics:</h5>
3156
3157<p>Memory is allocated using the system "<tt>malloc</tt>" function, and
Chris Lattner8b094fc2008-04-19 21:01:16 +0000[diff] [blame]3158a pointer is returned. The result of a zero byte allocattion is undefined. The
3159result is null if there is insufficient memory available.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3160
3161<h5>Example:</h5>
3162
3163<pre>
3164 %array = malloc [4 x i8 ] <i>; yields {[%4 x i8]*}:array</i>
3165
3166 %size = <a href="#i_add">add</a> i32 2, 2 <i>; yields {i32}:size = i32 4</i>
3167 %array1 = malloc i8, i32 4 <i>; yields {i8*}:array1</i>
3168 %array2 = malloc [12 x i8], i32 %size <i>; yields {[12 x i8]*}:array2</i>
3169 %array3 = malloc i32, i32 4, align 1024 <i>; yields {i32*}:array3</i>
3170 %array4 = malloc i32, align 1024 <i>; yields {i32*}:array4</i>
3171</pre>
3172</div>
3173
3174<!-- _______________________________________________________________________ -->
3175<div class="doc_subsubsection">
3176 <a name="i_free">'<tt>free</tt>' Instruction</a>
3177</div>
3178
3179<div class="doc_text">
3180
3181<h5>Syntax:</h5>
3182
3183<pre>
3184 free &lt;type&gt; &lt;value&gt; <i>; yields {void}</i>
3185</pre>
3186
3187<h5>Overview:</h5>
3188
3189<p>The '<tt>free</tt>' instruction returns memory back to the unused
3190memory heap to be reallocated in the future.</p>
3191
3192<h5>Arguments:</h5>
3193
3194<p>'<tt>value</tt>' shall be a pointer value that points to a value
3195that was allocated with the '<tt><a href="#i_malloc">malloc</a></tt>'
3196instruction.</p>
3197
3198<h5>Semantics:</h5>
3199
3200<p>Access to the memory pointed to by the pointer is no longer defined
Chris Lattner329d6532008-04-19 22:41:32 +0000[diff] [blame]3201after this instruction executes. If the pointer is null, the operation
3202is a noop.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3203
3204<h5>Example:</h5>
3205
3206<pre>
3207 %array = <a href="#i_malloc">malloc</a> [4 x i8] <i>; yields {[4 x i8]*}:array</i>
3208 free [4 x i8]* %array
3209</pre>
3210</div>
3211
3212<!-- _______________________________________________________________________ -->
3213<div class="doc_subsubsection">
3214 <a name="i_alloca">'<tt>alloca</tt>' Instruction</a>
3215</div>
3216
3217<div class="doc_text">
3218
3219<h5>Syntax:</h5>
3220
3221<pre>
3222 &lt;result&gt; = alloca &lt;type&gt;[, i32 &lt;NumElements&gt;][, align &lt;alignment&gt;] <i>; yields {type*}:result</i>
3223</pre>
3224
3225<h5>Overview:</h5>
3226
3227<p>The '<tt>alloca</tt>' instruction allocates memory on the stack frame of the
3228currently executing function, to be automatically released when this function
Christopher Lambcfe00962007-12-17 01:00:21 +0000[diff] [blame]3229returns to its caller. The object is always allocated in the generic address
3230space (address space zero).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3231
3232<h5>Arguments:</h5>
3233
3234<p>The '<tt>alloca</tt>' instruction allocates <tt>sizeof(&lt;type&gt;)*NumElements</tt>
3235bytes of memory on the runtime stack, returning a pointer of the
Gabor Greif5082cf42008-02-09 22:24:34 +0000[diff] [blame]3236appropriate type to the program. If "NumElements" is specified, it is the
3237number of elements allocated, otherwise "NumElements" is defaulted to be one.
Chris Lattner10368b62008-04-02 00:38:26 +0000[diff] [blame]3238If a constant alignment is specified, the value result of the allocation is guaranteed
Gabor Greif5082cf42008-02-09 22:24:34 +0000[diff] [blame]3239to be aligned to at least that boundary. If not specified, or if zero, the target
3240can choose to align the allocation on any convenient boundary.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3241
3242<p>'<tt>type</tt>' may be any sized type.</p>
3243
3244<h5>Semantics:</h5>
3245
Chris Lattner8b094fc2008-04-19 21:01:16 +0000[diff] [blame]3246<p>Memory is allocated; a pointer is returned. The operation is undefiend if
3247there is insufficient stack space for the allocation. '<tt>alloca</tt>'d
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3248memory is automatically released when the function returns. The '<tt>alloca</tt>'
3249instruction is commonly used to represent automatic variables that must
3250have an address available. When the function returns (either with the <tt><a
3251 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]3252instructions), the memory is reclaimed. Allocating zero bytes
3253is legal, but the result is undefined.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3254
3255<h5>Example:</h5>
3256
3257<pre>
3258 %ptr = alloca i32 <i>; yields {i32*}:ptr</i>
3259 %ptr = alloca i32, i32 4 <i>; yields {i32*}:ptr</i>
3260 %ptr = alloca i32, i32 4, align 1024 <i>; yields {i32*}:ptr</i>
3261 %ptr = alloca i32, align 1024 <i>; yields {i32*}:ptr</i>
3262</pre>
3263</div>
3264
3265<!-- _______________________________________________________________________ -->
3266<div class="doc_subsubsection"> <a name="i_load">'<tt>load</tt>'
3267Instruction</a> </div>
3268<div class="doc_text">
3269<h5>Syntax:</h5>
3270<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>
3271<h5>Overview:</h5>
3272<p>The '<tt>load</tt>' instruction is used to read from memory.</p>
3273<h5>Arguments:</h5>
3274<p>The argument to the '<tt>load</tt>' instruction specifies the memory
3275address from which to load. The pointer must point to a <a
3276 href="#t_firstclass">first class</a> type. If the <tt>load</tt> is
3277marked as <tt>volatile</tt>, then the optimizer is not allowed to modify
3278the number or order of execution of this <tt>load</tt> with other
3279volatile <tt>load</tt> and <tt><a href="#i_store">store</a></tt>
3280instructions. </p>
Chris Lattner21003c82008-01-06 21:04:43 +0000[diff] [blame]3281<p>
Chris Lattner10368b62008-04-02 00:38:26 +0000[diff] [blame]3282The optional constant "align" argument specifies the alignment of the operation
Chris Lattner21003c82008-01-06 21:04:43 +0000[diff] [blame]3283(that is, the alignment of the memory address). A value of 0 or an
3284omitted "align" argument means that the operation has the preferential
3285alignment for the target. It is the responsibility of the code emitter
3286to ensure that the alignment information is correct. Overestimating
3287the alignment results in an undefined behavior. Underestimating the
3288alignment may produce less efficient code. An alignment of 1 is always
3289safe.
3290</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3291<h5>Semantics:</h5>
3292<p>The location of memory pointed to is loaded.</p>
3293<h5>Examples:</h5>
3294<pre> %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
3295 <a
3296 href="#i_store">store</a> i32 3, i32* %ptr <i>; yields {void}</i>
3297 %val = load i32* %ptr <i>; yields {i32}:val = i32 3</i>
3298</pre>
3299</div>
3300<!-- _______________________________________________________________________ -->
3301<div class="doc_subsubsection"> <a name="i_store">'<tt>store</tt>'
3302Instruction</a> </div>
3303<div class="doc_text">
3304<h5>Syntax:</h5>
3305<pre> store &lt;ty&gt; &lt;value&gt;, &lt;ty&gt;* &lt;pointer&gt;[, align &lt;alignment&gt;] <i>; yields {void}</i>
3306 volatile store &lt;ty&gt; &lt;value&gt;, &lt;ty&gt;* &lt;pointer&gt;[, align &lt;alignment&gt;] <i>; yields {void}</i>
3307</pre>
3308<h5>Overview:</h5>
3309<p>The '<tt>store</tt>' instruction is used to write to memory.</p>
3310<h5>Arguments:</h5>
3311<p>There are two arguments to the '<tt>store</tt>' instruction: a value
3312to 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]3313operand must be a pointer to the <a href="#t_firstclass">first class</a> type
3314of the '<tt>&lt;value&gt;</tt>'
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3315operand. If the <tt>store</tt> is marked as <tt>volatile</tt>, then the
3316optimizer is not allowed to modify the number or order of execution of
3317this <tt>store</tt> with other volatile <tt>load</tt> and <tt><a
3318 href="#i_store">store</a></tt> instructions.</p>
Chris Lattner21003c82008-01-06 21:04:43 +0000[diff] [blame]3319<p>
Chris Lattner10368b62008-04-02 00:38:26 +0000[diff] [blame]3320The optional constant "align" argument specifies the alignment of the operation
Chris Lattner21003c82008-01-06 21:04:43 +0000[diff] [blame]3321(that is, the alignment of the memory address). A value of 0 or an
3322omitted "align" argument means that the operation has the preferential
3323alignment for the target. It is the responsibility of the code emitter
3324to ensure that the alignment information is correct. Overestimating
3325the alignment results in an undefined behavior. Underestimating the
3326alignment may produce less efficient code. An alignment of 1 is always
3327safe.
3328</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3329<h5>Semantics:</h5>
3330<p>The contents of memory are updated to contain '<tt>&lt;value&gt;</tt>'
3331at the location specified by the '<tt>&lt;pointer&gt;</tt>' operand.</p>
3332<h5>Example:</h5>
3333<pre> %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
Bill Wendling63ffa142007-10-22 05:10:05 +0000[diff] [blame]3334 store i32 3, i32* %ptr <i>; yields {void}</i>
3335 %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]3336</pre>
3337</div>
3338
3339<!-- _______________________________________________________________________ -->
3340<div class="doc_subsubsection">
3341 <a name="i_getelementptr">'<tt>getelementptr</tt>' Instruction</a>
3342</div>
3343
3344<div class="doc_text">
3345<h5>Syntax:</h5>
3346<pre>
Matthijs Kooijman32a080e2008-10-13 13:44:15 +0000[diff] [blame]3347 &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]3348</pre>
3349
3350<h5>Overview:</h5>
3351
3352<p>
3353The '<tt>getelementptr</tt>' instruction is used to get the address of a
Matthijs Kooijman32a080e2008-10-13 13:44:15 +0000[diff] [blame]3354subelement of an aggregate data structure. It performs address calculation only
3355and does not access memory.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3356
3357<h5>Arguments:</h5>
3358
Matthijs Kooijman32a080e2008-10-13 13:44:15 +0000[diff] [blame]3359<p>The first argument is always a pointer, and forms the basis of the
3360calculation. The remaining arguments are indices, that indicate which of the
3361elements of the aggregate object are indexed. The interpretation of each index
3362is dependent on the type being indexed into. The first index always indexes the
3363pointer value given as the first argument, the second index indexes a value of
3364the type pointed to (not necessarily the value directly pointed to, since the
3365first index can be non-zero), etc. The first type indexed into must be a pointer
3366value, subsequent types can be arrays, vectors and structs. Note that subsequent
3367types being indexed into can never be pointers, since that would require loading
3368the pointer before continuing calculation.</p>
3369
3370<p>The type of each index argument depends on the type it is indexing into.
3371When indexing into a (packed) structure, only <tt>i32</tt> integer
3372<b>constants</b> are allowed. When indexing into an array, pointer or vector,
3373only integers of 32 or 64 bits are allowed (also non-constants). 32-bit values
3374will be sign extended to 64-bits if required.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3375
3376<p>For example, let's consider a C code fragment and how it gets
3377compiled to LLVM:</p>
3378
3379<div class="doc_code">
3380<pre>
3381struct RT {
3382 char A;
3383 int B[10][20];
3384 char C;
3385};
3386struct ST {
3387 int X;
3388 double Y;
3389 struct RT Z;
3390};
3391
3392int *foo(struct ST *s) {
3393 return &amp;s[1].Z.B[5][13];
3394}
3395</pre>
3396</div>
3397
3398<p>The LLVM code generated by the GCC frontend is:</p>
3399
3400<div class="doc_code">
3401<pre>
3402%RT = type { i8 , [10 x [20 x i32]], i8 }
3403%ST = type { i32, double, %RT }
3404
3405define i32* %foo(%ST* %s) {
3406entry:
3407 %reg = getelementptr %ST* %s, i32 1, i32 2, i32 1, i32 5, i32 13
3408 ret i32* %reg
3409}
3410</pre>
3411</div>
3412
3413<h5>Semantics:</h5>
3414
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3415<p>In the example above, the first index is indexing into the '<tt>%ST*</tt>'
3416type, which is a pointer, yielding a '<tt>%ST</tt>' = '<tt>{ i32, double, %RT
3417}</tt>' type, a structure. The second index indexes into the third element of
3418the structure, yielding a '<tt>%RT</tt>' = '<tt>{ i8 , [10 x [20 x i32]],
3419i8 }</tt>' type, another structure. The third index indexes into the second
3420element of the structure, yielding a '<tt>[10 x [20 x i32]]</tt>' type, an
3421array. The two dimensions of the array are subscripted into, yielding an
3422'<tt>i32</tt>' type. The '<tt>getelementptr</tt>' instruction returns a pointer
3423to this element, thus computing a value of '<tt>i32*</tt>' type.</p>
3424
3425<p>Note that it is perfectly legal to index partially through a
3426structure, returning a pointer to an inner element. Because of this,
3427the LLVM code for the given testcase is equivalent to:</p>
3428
3429<pre>
3430 define i32* %foo(%ST* %s) {
3431 %t1 = getelementptr %ST* %s, i32 1 <i>; yields %ST*:%t1</i>
3432 %t2 = getelementptr %ST* %t1, i32 0, i32 2 <i>; yields %RT*:%t2</i>
3433 %t3 = getelementptr %RT* %t2, i32 0, i32 1 <i>; yields [10 x [20 x i32]]*:%t3</i>
3434 %t4 = getelementptr [10 x [20 x i32]]* %t3, i32 0, i32 5 <i>; yields [20 x i32]*:%t4</i>
3435 %t5 = getelementptr [20 x i32]* %t4, i32 0, i32 13 <i>; yields i32*:%t5</i>
3436 ret i32* %t5
3437 }
3438</pre>
3439
3440<p>Note that it is undefined to access an array out of bounds: array and
3441pointer indexes must always be within the defined bounds of the array type.
Chris Lattnera7d94ba2008-04-24 05:59:56 +0000[diff] [blame]3442The one exception for this rule is zero length arrays. These arrays are
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3443defined to be accessible as variable length arrays, which requires access
3444beyond the zero'th element.</p>
3445
3446<p>The getelementptr instruction is often confusing. For some more insight
3447into how it works, see <a href="GetElementPtr.html">the getelementptr
3448FAQ</a>.</p>
3449
3450<h5>Example:</h5>
3451
3452<pre>
3453 <i>; yields [12 x i8]*:aptr</i>
Matthijs Kooijman32a080e2008-10-13 13:44:15 +0000[diff] [blame]3454 %aptr = getelementptr {i32, [12 x i8]}* %saptr, i64 0, i32 1
3455 <i>; yields i8*:vptr</i>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]3456 %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]3457 <i>; yields i8*:eptr</i>
3458 %eptr = getelementptr [12 x i8]* %aptr, i64 0, i32 1
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3459</pre>
3460</div>
3461
3462<!-- ======================================================================= -->
3463<div class="doc_subsection"> <a name="convertops">Conversion Operations</a>
3464</div>
3465<div class="doc_text">
3466<p>The instructions in this category are the conversion instructions (casting)
3467which all take a single operand and a type. They perform various bit conversions
3468on the operand.</p>
3469</div>
3470
3471<!-- _______________________________________________________________________ -->
3472<div class="doc_subsubsection">
3473 <a name="i_trunc">'<tt>trunc .. to</tt>' Instruction</a>
3474</div>
3475<div class="doc_text">
3476
3477<h5>Syntax:</h5>
3478<pre>
3479 &lt;result&gt; = trunc &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3480</pre>
3481
3482<h5>Overview:</h5>
3483<p>
3484The '<tt>trunc</tt>' instruction truncates its operand to the type <tt>ty2</tt>.
3485</p>
3486
3487<h5>Arguments:</h5>
3488<p>
3489The '<tt>trunc</tt>' instruction takes a <tt>value</tt> to trunc, which must
3490be an <a href="#t_integer">integer</a> type, and a type that specifies the size
3491and type of the result, which must be an <a href="#t_integer">integer</a>
3492type. The bit size of <tt>value</tt> must be larger than the bit size of
3493<tt>ty2</tt>. Equal sized types are not allowed.</p>
3494
3495<h5>Semantics:</h5>
3496<p>
3497The '<tt>trunc</tt>' instruction truncates the high order bits in <tt>value</tt>
3498and converts the remaining bits to <tt>ty2</tt>. Since the source size must be
3499larger than the destination size, <tt>trunc</tt> cannot be a <i>no-op cast</i>.
3500It will always truncate bits.</p>
3501
3502<h5>Example:</h5>
3503<pre>
3504 %X = trunc i32 257 to i8 <i>; yields i8:1</i>
3505 %Y = trunc i32 123 to i1 <i>; yields i1:true</i>
3506 %Y = trunc i32 122 to i1 <i>; yields i1:false</i>
3507</pre>
3508</div>
3509
3510<!-- _______________________________________________________________________ -->
3511<div class="doc_subsubsection">
3512 <a name="i_zext">'<tt>zext .. to</tt>' Instruction</a>
3513</div>
3514<div class="doc_text">
3515
3516<h5>Syntax:</h5>
3517<pre>
3518 &lt;result&gt; = zext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3519</pre>
3520
3521<h5>Overview:</h5>
3522<p>The '<tt>zext</tt>' instruction zero extends its operand to type
3523<tt>ty2</tt>.</p>
3524
3525
3526<h5>Arguments:</h5>
3527<p>The '<tt>zext</tt>' instruction takes a value to cast, which must be of
3528<a href="#t_integer">integer</a> type, and a type to cast it to, which must
3529also be of <a href="#t_integer">integer</a> type. The bit size of the
3530<tt>value</tt> must be smaller than the bit size of the destination type,
3531<tt>ty2</tt>.</p>
3532
3533<h5>Semantics:</h5>
3534<p>The <tt>zext</tt> fills the high order bits of the <tt>value</tt> with zero
3535bits until it reaches the size of the destination type, <tt>ty2</tt>.</p>
3536
3537<p>When zero extending from i1, the result will always be either 0 or 1.</p>
3538
3539<h5>Example:</h5>
3540<pre>
3541 %X = zext i32 257 to i64 <i>; yields i64:257</i>
3542 %Y = zext i1 true to i32 <i>; yields i32:1</i>
3543</pre>
3544</div>
3545
3546<!-- _______________________________________________________________________ -->
3547<div class="doc_subsubsection">
3548 <a name="i_sext">'<tt>sext .. to</tt>' Instruction</a>
3549</div>
3550<div class="doc_text">
3551
3552<h5>Syntax:</h5>
3553<pre>
3554 &lt;result&gt; = sext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3555</pre>
3556
3557<h5>Overview:</h5>
3558<p>The '<tt>sext</tt>' sign extends <tt>value</tt> to the type <tt>ty2</tt>.</p>
3559
3560<h5>Arguments:</h5>
3561<p>
3562The '<tt>sext</tt>' instruction takes a value to cast, which must be of
3563<a href="#t_integer">integer</a> type, and a type to cast it to, which must
3564also be of <a href="#t_integer">integer</a> type. The bit size of the
3565<tt>value</tt> must be smaller than the bit size of the destination type,
3566<tt>ty2</tt>.</p>
3567
3568<h5>Semantics:</h5>
3569<p>
3570The '<tt>sext</tt>' instruction performs a sign extension by copying the sign
3571bit (highest order bit) of the <tt>value</tt> until it reaches the bit size of
3572the type <tt>ty2</tt>.</p>
3573
3574<p>When sign extending from i1, the extension always results in -1 or 0.</p>
3575
3576<h5>Example:</h5>
3577<pre>
3578 %X = sext i8 -1 to i16 <i>; yields i16 :65535</i>
3579 %Y = sext i1 true to i32 <i>; yields i32:-1</i>
3580</pre>
3581</div>
3582
3583<!-- _______________________________________________________________________ -->
3584<div class="doc_subsubsection">
3585 <a name="i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a>
3586</div>
3587
3588<div class="doc_text">
3589
3590<h5>Syntax:</h5>
3591
3592<pre>
3593 &lt;result&gt; = fptrunc &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3594</pre>
3595
3596<h5>Overview:</h5>
3597<p>The '<tt>fptrunc</tt>' instruction truncates <tt>value</tt> to type
3598<tt>ty2</tt>.</p>
3599
3600
3601<h5>Arguments:</h5>
3602<p>The '<tt>fptrunc</tt>' instruction takes a <a href="#t_floating">floating
3603 point</a> value to cast and a <a href="#t_floating">floating point</a> type to
3604cast it to. The size of <tt>value</tt> must be larger than the size of
3605<tt>ty2</tt>. This implies that <tt>fptrunc</tt> cannot be used to make a
3606<i>no-op cast</i>.</p>
3607
3608<h5>Semantics:</h5>
3609<p> The '<tt>fptrunc</tt>' instruction truncates a <tt>value</tt> from a larger
3610<a href="#t_floating">floating point</a> type to a smaller
3611<a href="#t_floating">floating point</a> type. If the value cannot fit within
3612the destination type, <tt>ty2</tt>, then the results are undefined.</p>
3613
3614<h5>Example:</h5>
3615<pre>
3616 %X = fptrunc double 123.0 to float <i>; yields float:123.0</i>
3617 %Y = fptrunc double 1.0E+300 to float <i>; yields undefined</i>
3618</pre>
3619</div>
3620
3621<!-- _______________________________________________________________________ -->
3622<div class="doc_subsubsection">
3623 <a name="i_fpext">'<tt>fpext .. to</tt>' Instruction</a>
3624</div>
3625<div class="doc_text">
3626
3627<h5>Syntax:</h5>
3628<pre>
3629 &lt;result&gt; = fpext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3630</pre>
3631
3632<h5>Overview:</h5>
3633<p>The '<tt>fpext</tt>' extends a floating point <tt>value</tt> to a larger
3634floating point value.</p>
3635
3636<h5>Arguments:</h5>
3637<p>The '<tt>fpext</tt>' instruction takes a
3638<a href="#t_floating">floating point</a> <tt>value</tt> to cast,
3639and a <a href="#t_floating">floating point</a> type to cast it to. The source
3640type must be smaller than the destination type.</p>
3641
3642<h5>Semantics:</h5>
3643<p>The '<tt>fpext</tt>' instruction extends the <tt>value</tt> from a smaller
3644<a href="#t_floating">floating point</a> type to a larger
3645<a href="#t_floating">floating point</a> type. The <tt>fpext</tt> cannot be
3646used to make a <i>no-op cast</i> because it always changes bits. Use
3647<tt>bitcast</tt> to make a <i>no-op cast</i> for a floating point cast.</p>
3648
3649<h5>Example:</h5>
3650<pre>
3651 %X = fpext float 3.1415 to double <i>; yields double:3.1415</i>
3652 %Y = fpext float 1.0 to float <i>; yields float:1.0 (no-op)</i>
3653</pre>
3654</div>
3655
3656<!-- _______________________________________________________________________ -->
3657<div class="doc_subsubsection">
3658 <a name="i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a>
3659</div>
3660<div class="doc_text">
3661
3662<h5>Syntax:</h5>
3663<pre>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3664 &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]3665</pre>
3666
3667<h5>Overview:</h5>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3668<p>The '<tt>fptoui</tt>' converts a floating point <tt>value</tt> to its
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3669unsigned integer equivalent of type <tt>ty2</tt>.
3670</p>
3671
3672<h5>Arguments:</h5>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3673<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]3674scalar or vector <a href="#t_floating">floating point</a> value, and a type
3675to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
3676type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
3677vector integer type with the same number of elements as <tt>ty</tt></p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3678
3679<h5>Semantics:</h5>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3680<p> The '<tt>fptoui</tt>' instruction converts its
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3681<a href="#t_floating">floating point</a> operand into the nearest (rounding
3682towards zero) unsigned integer value. If the value cannot fit in <tt>ty2</tt>,
3683the results are undefined.</p>
3684
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3685<h5>Example:</h5>
3686<pre>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3687 %X = fptoui double 123.0 to i32 <i>; yields i32:123</i>
Chris Lattner681f1e82007-09-22 03:17:52 +0000[diff] [blame]3688 %Y = fptoui float 1.0E+300 to i1 <i>; yields undefined:1</i>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3689 %X = fptoui float 1.04E+17 to i8 <i>; yields undefined:1</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3690</pre>
3691</div>
3692
3693<!-- _______________________________________________________________________ -->
3694<div class="doc_subsubsection">
3695 <a name="i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a>
3696</div>
3697<div class="doc_text">
3698
3699<h5>Syntax:</h5>
3700<pre>
3701 &lt;result&gt; = fptosi &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3702</pre>
3703
3704<h5>Overview:</h5>
3705<p>The '<tt>fptosi</tt>' instruction converts
3706<a href="#t_floating">floating point</a> <tt>value</tt> to type <tt>ty2</tt>.
3707</p>
3708
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3709<h5>Arguments:</h5>
3710<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]3711scalar or vector <a href="#t_floating">floating point</a> value, and a type
3712to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
3713type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
3714vector integer type with the same number of elements as <tt>ty</tt></p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3715
3716<h5>Semantics:</h5>
3717<p>The '<tt>fptosi</tt>' instruction converts its
3718<a href="#t_floating">floating point</a> operand into the nearest (rounding
3719towards zero) signed integer value. If the value cannot fit in <tt>ty2</tt>,
3720the results are undefined.</p>
3721
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3722<h5>Example:</h5>
3723<pre>
3724 %X = fptosi double -123.0 to i32 <i>; yields i32:-123</i>
Chris Lattner681f1e82007-09-22 03:17:52 +0000[diff] [blame]3725 %Y = fptosi float 1.0E-247 to i1 <i>; yields undefined:1</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3726 %X = fptosi float 1.04E+17 to i8 <i>; yields undefined:1</i>
3727</pre>
3728</div>
3729
3730<!-- _______________________________________________________________________ -->
3731<div class="doc_subsubsection">
3732 <a name="i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a>
3733</div>
3734<div class="doc_text">
3735
3736<h5>Syntax:</h5>
3737<pre>
3738 &lt;result&gt; = uitofp &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3739</pre>
3740
3741<h5>Overview:</h5>
3742<p>The '<tt>uitofp</tt>' instruction regards <tt>value</tt> as an unsigned
3743integer and converts that value to the <tt>ty2</tt> type.</p>
3744
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3745<h5>Arguments:</h5>
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]3746<p>The '<tt>uitofp</tt>' instruction takes a value to cast, which must be a
3747scalar or vector <a href="#t_integer">integer</a> value, and a type to cast it
3748to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
3749type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
3750floating point type with the same number of elements as <tt>ty</tt></p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3751
3752<h5>Semantics:</h5>
3753<p>The '<tt>uitofp</tt>' instruction interprets its operand as an unsigned
3754integer quantity and converts it to the corresponding floating point value. If
3755the value cannot fit in the floating point value, the results are undefined.</p>
3756
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3757<h5>Example:</h5>
3758<pre>
3759 %X = uitofp i32 257 to float <i>; yields float:257.0</i>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]3760 %Y = uitofp i8 -1 to double <i>; yields double:255.0</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3761</pre>
3762</div>
3763
3764<!-- _______________________________________________________________________ -->
3765<div class="doc_subsubsection">
3766 <a name="i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a>
3767</div>
3768<div class="doc_text">
3769
3770<h5>Syntax:</h5>
3771<pre>
3772 &lt;result&gt; = sitofp &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3773</pre>
3774
3775<h5>Overview:</h5>
3776<p>The '<tt>sitofp</tt>' instruction regards <tt>value</tt> as a signed
3777integer and converts that value to the <tt>ty2</tt> type.</p>
3778
3779<h5>Arguments:</h5>
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]3780<p>The '<tt>sitofp</tt>' instruction takes a value to cast, which must be a
3781scalar or vector <a href="#t_integer">integer</a> value, and a type to cast it
3782to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
3783type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
3784floating point type with the same number of elements as <tt>ty</tt></p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3785
3786<h5>Semantics:</h5>
3787<p>The '<tt>sitofp</tt>' instruction interprets its operand as a signed
3788integer quantity and converts it to the corresponding floating point value. If
3789the value cannot fit in the floating point value, the results are undefined.</p>
3790
3791<h5>Example:</h5>
3792<pre>
3793 %X = sitofp i32 257 to float <i>; yields float:257.0</i>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]3794 %Y = sitofp i8 -1 to double <i>; yields double:-1.0</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3795</pre>
3796</div>
3797
3798<!-- _______________________________________________________________________ -->
3799<div class="doc_subsubsection">
3800 <a name="i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a>
3801</div>
3802<div class="doc_text">
3803
3804<h5>Syntax:</h5>
3805<pre>
3806 &lt;result&gt; = ptrtoint &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3807</pre>
3808
3809<h5>Overview:</h5>
3810<p>The '<tt>ptrtoint</tt>' instruction converts the pointer <tt>value</tt> to
3811the integer type <tt>ty2</tt>.</p>
3812
3813<h5>Arguments:</h5>
3814<p>The '<tt>ptrtoint</tt>' instruction takes a <tt>value</tt> to cast, which
3815must 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]3816<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]3817
3818<h5>Semantics:</h5>
3819<p>The '<tt>ptrtoint</tt>' instruction converts <tt>value</tt> to integer type
3820<tt>ty2</tt> by interpreting the pointer value as an integer and either
3821truncating or zero extending that value to the size of the integer type. If
3822<tt>value</tt> is smaller than <tt>ty2</tt> then a zero extension is done. If
3823<tt>value</tt> is larger than <tt>ty2</tt> then a truncation is done. If they
3824are the same size, then nothing is done (<i>no-op cast</i>) other than a type
3825change.</p>
3826
3827<h5>Example:</h5>
3828<pre>
3829 %X = ptrtoint i32* %X to i8 <i>; yields truncation on 32-bit architecture</i>
3830 %Y = ptrtoint i32* %x to i64 <i>; yields zero extension on 32-bit architecture</i>
3831</pre>
3832</div>
3833
3834<!-- _______________________________________________________________________ -->
3835<div class="doc_subsubsection">
3836 <a name="i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a>
3837</div>
3838<div class="doc_text">
3839
3840<h5>Syntax:</h5>
3841<pre>
3842 &lt;result&gt; = inttoptr &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3843</pre>
3844
3845<h5>Overview:</h5>
3846<p>The '<tt>inttoptr</tt>' instruction converts an integer <tt>value</tt> to
3847a pointer type, <tt>ty2</tt>.</p>
3848
3849<h5>Arguments:</h5>
3850<p>The '<tt>inttoptr</tt>' instruction takes an <a href="#t_integer">integer</a>
3851value to cast, and a type to cast it to, which must be a
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]3852<a href="#t_pointer">pointer</a> type.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3853
3854<h5>Semantics:</h5>
3855<p>The '<tt>inttoptr</tt>' instruction converts <tt>value</tt> to type
3856<tt>ty2</tt> by applying either a zero extension or a truncation depending on
3857the size of the integer <tt>value</tt>. If <tt>value</tt> is larger than the
3858size of a pointer then a truncation is done. If <tt>value</tt> is smaller than
3859the size of a pointer then a zero extension is done. If they are the same size,
3860nothing is done (<i>no-op cast</i>).</p>
3861
3862<h5>Example:</h5>
3863<pre>
3864 %X = inttoptr i32 255 to i32* <i>; yields zero extension on 64-bit architecture</i>
3865 %X = inttoptr i32 255 to i32* <i>; yields no-op on 32-bit architecture</i>
3866 %Y = inttoptr i64 0 to i32* <i>; yields truncation on 32-bit architecture</i>
3867</pre>
3868</div>
3869
3870<!-- _______________________________________________________________________ -->
3871<div class="doc_subsubsection">
3872 <a name="i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a>
3873</div>
3874<div class="doc_text">
3875
3876<h5>Syntax:</h5>
3877<pre>
3878 &lt;result&gt; = bitcast &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3879</pre>
3880
3881<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3882
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3883<p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
3884<tt>ty2</tt> without changing any bits.</p>
3885
3886<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3887
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3888<p>The '<tt>bitcast</tt>' instruction takes a value to cast, which must be
Dan Gohman7305fa02008-09-08 16:45:59 +0000[diff] [blame]3889a non-aggregate first class value, and a type to cast it to, which must also be
3890a non-aggregate <a href="#t_firstclass">first class</a> type. The bit sizes of
3891<tt>value</tt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3892and the destination type, <tt>ty2</tt>, must be identical. If the source
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3893type is a pointer, the destination type must also be a pointer. This
3894instruction supports bitwise conversion of vectors to integers and to vectors
3895of other types (as long as they have the same size).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3896
3897<h5>Semantics:</h5>
3898<p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
3899<tt>ty2</tt>. It is always a <i>no-op cast</i> because no bits change with
3900this conversion. The conversion is done as if the <tt>value</tt> had been
3901stored to memory and read back as type <tt>ty2</tt>. Pointer types may only be
3902converted to other pointer types with this instruction. To convert pointers to
3903other types, use the <a href="#i_inttoptr">inttoptr</a> or
3904<a href="#i_ptrtoint">ptrtoint</a> instructions first.</p>
3905
3906<h5>Example:</h5>
3907<pre>
3908 %X = bitcast i8 255 to i8 <i>; yields i8 :-1</i>
3909 %Y = bitcast i32* %x to sint* <i>; yields sint*:%x</i>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]3910 %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]3911</pre>
3912</div>
3913
3914<!-- ======================================================================= -->
3915<div class="doc_subsection"> <a name="otherops">Other Operations</a> </div>
3916<div class="doc_text">
3917<p>The instructions in this category are the "miscellaneous"
3918instructions, which defy better classification.</p>
3919</div>
3920
3921<!-- _______________________________________________________________________ -->
3922<div class="doc_subsubsection"><a name="i_icmp">'<tt>icmp</tt>' Instruction</a>
3923</div>
3924<div class="doc_text">
3925<h5>Syntax:</h5>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]3926<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]3927</pre>
3928<h5>Overview:</h5>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]3929<p>The '<tt>icmp</tt>' instruction returns a boolean value or
3930a vector of boolean values based on comparison
3931of its two integer, integer vector, or pointer operands.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3932<h5>Arguments:</h5>
3933<p>The '<tt>icmp</tt>' instruction takes three operands. The first operand is
3934the condition code indicating the kind of comparison to perform. It is not
3935a value, just a keyword. The possible condition code are:
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]3936</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3937<ol>
3938 <li><tt>eq</tt>: equal</li>
3939 <li><tt>ne</tt>: not equal </li>
3940 <li><tt>ugt</tt>: unsigned greater than</li>
3941 <li><tt>uge</tt>: unsigned greater or equal</li>
3942 <li><tt>ult</tt>: unsigned less than</li>
3943 <li><tt>ule</tt>: unsigned less or equal</li>
3944 <li><tt>sgt</tt>: signed greater than</li>
3945 <li><tt>sge</tt>: signed greater or equal</li>
3946 <li><tt>slt</tt>: signed less than</li>
3947 <li><tt>sle</tt>: signed less or equal</li>
3948</ol>
3949<p>The remaining two arguments must be <a href="#t_integer">integer</a> or
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]3950<a href="#t_pointer">pointer</a>
3951or integer <a href="#t_vector">vector</a> typed.
3952They must also be identical types.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3953<h5>Semantics:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3954<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]3955the condition code given as <tt>cond</tt>. The comparison performed always
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]3956yields 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]3957</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3958<ol>
3959 <li><tt>eq</tt>: yields <tt>true</tt> if the operands are equal,
3960 <tt>false</tt> otherwise. No sign interpretation is necessary or performed.
3961 </li>
3962 <li><tt>ne</tt>: yields <tt>true</tt> if the operands are unequal,
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]3963 <tt>false</tt> otherwise. No sign interpretation is necessary or performed.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3964 <li><tt>ugt</tt>: interprets the operands as unsigned values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3965 <tt>true</tt> if <tt>op1</tt> is greater than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3966 <li><tt>uge</tt>: interprets the operands as unsigned values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3967 <tt>true</tt> if <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3968 <li><tt>ult</tt>: interprets the operands as unsigned values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3969 <tt>true</tt> if <tt>op1</tt> is less than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3970 <li><tt>ule</tt>: interprets the operands as unsigned values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3971 <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]3972 <li><tt>sgt</tt>: interprets the operands as signed values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3973 <tt>true</tt> if <tt>op1</tt> is greater than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3974 <li><tt>sge</tt>: interprets the operands as signed values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3975 <tt>true</tt> if <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3976 <li><tt>slt</tt>: interprets the operands as signed values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3977 <tt>true</tt> if <tt>op1</tt> is less than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3978 <li><tt>sle</tt>: interprets the operands as signed values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3979 <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]3980</ol>
3981<p>If the operands are <a href="#t_pointer">pointer</a> typed, the pointer
3982values are compared as if they were integers.</p>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]3983<p>If the operands are integer vectors, then they are compared
3984element by element. The result is an <tt>i1</tt> vector with
3985the same number of elements as the values being compared.
3986Otherwise, the result is an <tt>i1</tt>.
3987</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3988
3989<h5>Example:</h5>
3990<pre> &lt;result&gt; = icmp eq i32 4, 5 <i>; yields: result=false</i>
3991 &lt;result&gt; = icmp ne float* %X, %X <i>; yields: result=false</i>
3992 &lt;result&gt; = icmp ult i16 4, 5 <i>; yields: result=true</i>
3993 &lt;result&gt; = icmp sgt i16 4, 5 <i>; yields: result=false</i>
3994 &lt;result&gt; = icmp ule i16 -4, 5 <i>; yields: result=false</i>
3995 &lt;result&gt; = icmp sge i16 4, 5 <i>; yields: result=false</i>
3996</pre>
3997</div>
3998
3999<!-- _______________________________________________________________________ -->
4000<div class="doc_subsubsection"><a name="i_fcmp">'<tt>fcmp</tt>' Instruction</a>
4001</div>
4002<div class="doc_text">
4003<h5>Syntax:</h5>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4004<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]4005</pre>
4006<h5>Overview:</h5>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4007<p>The '<tt>fcmp</tt>' instruction returns a boolean value
4008or vector of boolean values based on comparison
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4009of its operands.</p>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4010<p>
4011If the operands are floating point scalars, then the result
4012type is a boolean (<a href="#t_primitive"><tt>i1</tt></a>).
4013</p>
4014<p>If the operands are floating point vectors, then the result type
4015is a vector of boolean with the same number of elements as the
4016operands being compared.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4017<h5>Arguments:</h5>
4018<p>The '<tt>fcmp</tt>' instruction takes three operands. The first operand is
4019the condition code indicating the kind of comparison to perform. It is not
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4020a value, just a keyword. The possible condition code are:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4021<ol>
4022 <li><tt>false</tt>: no comparison, always returns false</li>
4023 <li><tt>oeq</tt>: ordered and equal</li>
4024 <li><tt>ogt</tt>: ordered and greater than </li>
4025 <li><tt>oge</tt>: ordered and greater than or equal</li>
4026 <li><tt>olt</tt>: ordered and less than </li>
4027 <li><tt>ole</tt>: ordered and less than or equal</li>
4028 <li><tt>one</tt>: ordered and not equal</li>
4029 <li><tt>ord</tt>: ordered (no nans)</li>
4030 <li><tt>ueq</tt>: unordered or equal</li>
4031 <li><tt>ugt</tt>: unordered or greater than </li>
4032 <li><tt>uge</tt>: unordered or greater than or equal</li>
4033 <li><tt>ult</tt>: unordered or less than </li>
4034 <li><tt>ule</tt>: unordered or less than or equal</li>
4035 <li><tt>une</tt>: unordered or not equal</li>
4036 <li><tt>uno</tt>: unordered (either nans)</li>
4037 <li><tt>true</tt>: no comparison, always returns true</li>
4038</ol>
4039<p><i>Ordered</i> means that neither operand is a QNAN while
4040<i>unordered</i> means that either operand may be a QNAN.</p>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4041<p>Each of <tt>val1</tt> and <tt>val2</tt> arguments must be
4042either a <a href="#t_floating">floating point</a> type
4043or a <a href="#t_vector">vector</a> of floating point type.
4044They must have identical types.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4045<h5>Semantics:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4046<p>The '<tt>fcmp</tt>' instruction compares <tt>op1</tt> and <tt>op2</tt>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4047according to the condition code given as <tt>cond</tt>.
4048If the operands are vectors, then the vectors are compared
4049element by element.
4050Each comparison performed
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4051always yields an <a href="#t_primitive">i1</a> result, as follows:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4052<ol>
4053 <li><tt>false</tt>: always yields <tt>false</tt>, regardless of operands.</li>
4054 <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]4055 <tt>op1</tt> is equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4056 <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]4057 <tt>op1</tt> is greather than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4058 <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]4059 <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4060 <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]4061 <tt>op1</tt> is less than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4062 <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]4063 <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4064 <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]4065 <tt>op1</tt> is not equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4066 <li><tt>ord</tt>: yields <tt>true</tt> if both operands are not a QNAN.</li>
4067 <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]4068 <tt>op1</tt> is equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4069 <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]4070 <tt>op1</tt> is greater than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4071 <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]4072 <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4073 <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]4074 <tt>op1</tt> is less than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4075 <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]4076 <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4077 <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]4078 <tt>op1</tt> is not equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4079 <li><tt>uno</tt>: yields <tt>true</tt> if either operand is a QNAN.</li>
4080 <li><tt>true</tt>: always yields <tt>true</tt>, regardless of operands.</li>
4081</ol>
4082
4083<h5>Example:</h5>
4084<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]4085 &lt;result&gt; = fcmp one float 4.0, 5.0 <i>; yields: result=true</i>
4086 &lt;result&gt; = fcmp olt float 4.0, 5.0 <i>; yields: result=true</i>
4087 &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]4088</pre>
4089</div>
4090
4091<!-- _______________________________________________________________________ -->
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]4092<div class="doc_subsubsection">
4093 <a name="i_vicmp">'<tt>vicmp</tt>' Instruction</a>
4094</div>
4095<div class="doc_text">
4096<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4097<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]4098</pre>
4099<h5>Overview:</h5>
4100<p>The '<tt>vicmp</tt>' instruction returns an integer vector value based on
4101element-wise comparison of its two integer vector operands.</p>
4102<h5>Arguments:</h5>
4103<p>The '<tt>vicmp</tt>' instruction takes three operands. The first operand is
4104the condition code indicating the kind of comparison to perform. It is not
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4105a value, just a keyword. The possible condition code are:</p>
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]4106<ol>
4107 <li><tt>eq</tt>: equal</li>
4108 <li><tt>ne</tt>: not equal </li>
4109 <li><tt>ugt</tt>: unsigned greater than</li>
4110 <li><tt>uge</tt>: unsigned greater or equal</li>
4111 <li><tt>ult</tt>: unsigned less than</li>
4112 <li><tt>ule</tt>: unsigned less or equal</li>
4113 <li><tt>sgt</tt>: signed greater than</li>
4114 <li><tt>sge</tt>: signed greater or equal</li>
4115 <li><tt>slt</tt>: signed less than</li>
4116 <li><tt>sle</tt>: signed less or equal</li>
4117</ol>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4118<p>The remaining two arguments must be <a href="#t_vector">vector</a> or
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]4119<a href="#t_integer">integer</a> typed. They must also be identical types.</p>
4120<h5>Semantics:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4121<p>The '<tt>vicmp</tt>' instruction compares <tt>op1</tt> and <tt>op2</tt>
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]4122according to the condition code given as <tt>cond</tt>. The comparison yields a
4123<a href="#t_vector">vector</a> of <a href="#t_integer">integer</a> result, of
4124identical type as the values being compared. The most significant bit in each
4125element is 1 if the element-wise comparison evaluates to true, and is 0
4126otherwise. All other bits of the result are undefined. The condition codes
4127are evaluated identically to the <a href="#i_icmp">'<tt>icmp</tt>'
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4128instruction</a>.</p>
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]4129
4130<h5>Example:</h5>
4131<pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4132 &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>
4133 &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]4134</pre>
4135</div>
4136
4137<!-- _______________________________________________________________________ -->
4138<div class="doc_subsubsection">
4139 <a name="i_vfcmp">'<tt>vfcmp</tt>' Instruction</a>
4140</div>
4141<div class="doc_text">
4142<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4143<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]4144<h5>Overview:</h5>
4145<p>The '<tt>vfcmp</tt>' instruction returns an integer vector value based on
4146element-wise comparison of its two floating point vector operands. The output
4147elements have the same width as the input elements.</p>
4148<h5>Arguments:</h5>
4149<p>The '<tt>vfcmp</tt>' instruction takes three operands. The first operand is
4150the condition code indicating the kind of comparison to perform. It is not
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4151a value, just a keyword. The possible condition code are:</p>
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]4152<ol>
4153 <li><tt>false</tt>: no comparison, always returns false</li>
4154 <li><tt>oeq</tt>: ordered and equal</li>
4155 <li><tt>ogt</tt>: ordered and greater than </li>
4156 <li><tt>oge</tt>: ordered and greater than or equal</li>
4157 <li><tt>olt</tt>: ordered and less than </li>
4158 <li><tt>ole</tt>: ordered and less than or equal</li>
4159 <li><tt>one</tt>: ordered and not equal</li>
4160 <li><tt>ord</tt>: ordered (no nans)</li>
4161 <li><tt>ueq</tt>: unordered or equal</li>
4162 <li><tt>ugt</tt>: unordered or greater than </li>
4163 <li><tt>uge</tt>: unordered or greater than or equal</li>
4164 <li><tt>ult</tt>: unordered or less than </li>
4165 <li><tt>ule</tt>: unordered or less than or equal</li>
4166 <li><tt>une</tt>: unordered or not equal</li>
4167 <li><tt>uno</tt>: unordered (either nans)</li>
4168 <li><tt>true</tt>: no comparison, always returns true</li>
4169</ol>
4170<p>The remaining two arguments must be <a href="#t_vector">vector</a> of
4171<a href="#t_floating">floating point</a> typed. They must also be identical
4172types.</p>
4173<h5>Semantics:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4174<p>The '<tt>vfcmp</tt>' instruction compares <tt>op1</tt> and <tt>op2</tt>
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]4175according to the condition code given as <tt>cond</tt>. The comparison yields a
4176<a href="#t_vector">vector</a> of <a href="#t_integer">integer</a> result, with
4177an identical number of elements as the values being compared, and each element
4178having identical with to the width of the floating point elements. The most
4179significant bit in each element is 1 if the element-wise comparison evaluates to
4180true, and is 0 otherwise. All other bits of the result are undefined. The
4181condition codes are evaluated identically to the
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4182<a href="#i_fcmp">'<tt>fcmp</tt>' instruction</a>.</p>
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]4183
4184<h5>Example:</h5>
4185<pre>
Chris Lattner1e5c5cd02008-10-13 16:55:18 +0000[diff] [blame]4186 <i>; yields: result=&lt;2 x i32&gt; &lt; i32 0, i32 -1 &gt;</i>
4187 &lt;result&gt; = vfcmp oeq &lt;2 x float&gt; &lt; float 4, float 0 &gt;, &lt; float 5, float 0 &gt;
4188
4189 <i>; yields: result=&lt;2 x i64&gt; &lt; i64 -1, i64 0 &gt;</i>
4190 &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]4191</pre>
4192</div>
4193
4194<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4195<div class="doc_subsubsection">
4196 <a name="i_phi">'<tt>phi</tt>' Instruction</a>
4197</div>
4198
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4199<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4200
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4201<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4202
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4203<pre> &lt;result&gt; = phi &lt;ty&gt; [ &lt;val0&gt;, &lt;label0&gt;], ...<br></pre>
4204<h5>Overview:</h5>
4205<p>The '<tt>phi</tt>' instruction is used to implement the &#966; node in
4206the SSA graph representing the function.</p>
4207<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4208
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4209<p>The type of the incoming values is specified with the first type
4210field. After this, the '<tt>phi</tt>' instruction takes a list of pairs
4211as arguments, with one pair for each predecessor basic block of the
4212current block. Only values of <a href="#t_firstclass">first class</a>
4213type may be used as the value arguments to the PHI node. Only labels
4214may be used as the label arguments.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4215
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4216<p>There must be no non-phi instructions between the start of a basic
4217block and the PHI instructions: i.e. PHI instructions must be first in
4218a basic block.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4219
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4220<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4221
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4222<p>At runtime, the '<tt>phi</tt>' instruction logically takes on the value
4223specified by the pair corresponding to the predecessor basic block that executed
4224just prior to the current block.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4225
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4226<h5>Example:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4227<pre>
4228Loop: ; Infinite loop that counts from 0 on up...
4229 %indvar = phi i32 [ 0, %LoopHeader ], [ %nextindvar, %Loop ]
4230 %nextindvar = add i32 %indvar, 1
4231 br label %Loop
4232</pre>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4233</div>
4234
4235<!-- _______________________________________________________________________ -->
4236<div class="doc_subsubsection">
4237 <a name="i_select">'<tt>select</tt>' Instruction</a>
4238</div>
4239
4240<div class="doc_text">
4241
4242<h5>Syntax:</h5>
4243
4244<pre>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4245 &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>
4246
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4247 <i>selty</i> is either i1 or {&lt;N x i1&gt;}
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4248</pre>
4249
4250<h5>Overview:</h5>
4251
4252<p>
4253The '<tt>select</tt>' instruction is used to choose one value based on a
4254condition, without branching.
4255</p>
4256
4257
4258<h5>Arguments:</h5>
4259
4260<p>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4261The '<tt>select</tt>' instruction requires an 'i1' value or
4262a vector of 'i1' values indicating the
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4263condition, and two values of the same <a href="#t_firstclass">first class</a>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4264type. If the val1/val2 are vectors and
4265the condition is a scalar, then entire vectors are selected, not
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4266individual elements.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4267</p>
4268
4269<h5>Semantics:</h5>
4270
4271<p>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4272If 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]4273value argument; otherwise, it returns the second value argument.
4274</p>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4275<p>
4276If the condition is a vector of i1, then the value arguments must
4277be vectors of the same size, and the selection is done element
4278by element.
4279</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4280
4281<h5>Example:</h5>
4282
4283<pre>
4284 %X = select i1 true, i8 17, i8 42 <i>; yields i8:17</i>
4285</pre>
4286</div>
4287
4288
4289<!-- _______________________________________________________________________ -->
4290<div class="doc_subsubsection">
4291 <a name="i_call">'<tt>call</tt>' Instruction</a>
4292</div>
4293
4294<div class="doc_text">
4295
4296<h5>Syntax:</h5>
4297<pre>
Devang Pateld0bfcc72008-10-07 17:48:33 +0000[diff] [blame]4298 &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]4299</pre>
4300
4301<h5>Overview:</h5>
4302
4303<p>The '<tt>call</tt>' instruction represents a simple function call.</p>
4304
4305<h5>Arguments:</h5>
4306
4307<p>This instruction requires several arguments:</p>
4308
4309<ol>
4310 <li>
4311 <p>The optional "tail" marker indicates whether the callee function accesses
4312 any allocas or varargs in the caller. If the "tail" marker is present, the
4313 function call is eligible for tail call optimization. Note that calls may
4314 be marked "tail" even if they do not occur before a <a
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4315 href="#i_ret"><tt>ret</tt></a> instruction.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4316 </li>
4317 <li>
4318 <p>The optional "cconv" marker indicates which <a href="#callingconv">calling
4319 convention</a> the call should use. If none is specified, the call defaults
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4320 to using C calling conventions.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4321 </li>
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]4322
4323 <li>
4324 <p>The optional <a href="#paramattrs">Parameter Attributes</a> list for
4325 return values. Only '<tt>zeroext</tt>', '<tt>signext</tt>',
4326 and '<tt>inreg</tt>' attributes are valid here.</p>
4327 </li>
4328
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4329 <li>
Nick Lewycky93082fc2007-09-08 13:57:50 +0000[diff] [blame]4330 <p>'<tt>ty</tt>': the type of the call instruction itself which is also
4331 the type of the return value. Functions that return no value are marked
4332 <tt><a href="#t_void">void</a></tt>.</p>
4333 </li>
4334 <li>
4335 <p>'<tt>fnty</tt>': shall be the signature of the pointer to function
4336 value being invoked. The argument types must match the types implied by
4337 this signature. This type can be omitted if the function is not varargs
4338 and if the function type does not return a pointer to a function.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4339 </li>
4340 <li>
4341 <p>'<tt>fnptrval</tt>': An LLVM value containing a pointer to a function to
4342 be invoked. In most cases, this is a direct function invocation, but
4343 indirect <tt>call</tt>s are just as possible, calling an arbitrary pointer
4344 to function value.</p>
4345 </li>
4346 <li>
4347 <p>'<tt>function args</tt>': argument list whose types match the
4348 function signature argument types. All arguments must be of
4349 <a href="#t_firstclass">first class</a> type. If the function signature
4350 indicates the function accepts a variable number of arguments, the extra
4351 arguments can be specified.</p>
4352 </li>
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]4353 <li>
Devang Pateld0bfcc72008-10-07 17:48:33 +0000[diff] [blame]4354 <p>The optional <a href="#fnattrs">function attributes</a> list. Only
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]4355 '<tt>noreturn</tt>', '<tt>nounwind</tt>', '<tt>readonly</tt>' and
4356 '<tt>readnone</tt>' attributes are valid here.</p>
4357 </li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4358</ol>
4359
4360<h5>Semantics:</h5>
4361
4362<p>The '<tt>call</tt>' instruction is used to cause control flow to
4363transfer to a specified function, with its incoming arguments bound to
4364the specified values. Upon a '<tt><a href="#i_ret">ret</a></tt>'
4365instruction in the called function, control flow continues with the
4366instruction after the function call, and the return value of the
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4367function is bound to the result argument.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4368
4369<h5>Example:</h5>
4370
4371<pre>
Nick Lewycky93082fc2007-09-08 13:57:50 +0000[diff] [blame]4372 %retval = call i32 @test(i32 %argc)
Chris Lattner5e893ef2008-03-21 17:24:17 +0000[diff] [blame]4373 call i32 (i8 *, ...)* @printf(i8 * %msg, i32 12, i8 42) <i>; yields i32</i>
4374 %X = tail call i32 @foo() <i>; yields i32</i>
4375 %Y = tail call <a href="#callingconv">fastcc</a> i32 @foo() <i>; yields i32</i>
4376 call void %foo(i8 97 signext)
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]4377
4378 %struct.A = type { i32, i8 }
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]4379 %r = call %struct.A @foo() <i>; yields { 32, i8 }</i>
Dan Gohman3e700032008-10-04 19:00:07 +0000[diff] [blame]4380 %gr = extractvalue %struct.A %r, 0 <i>; yields i32</i>
4381 %gr1 = extractvalue %struct.A %r, 1 <i>; yields i8</i>
Chris Lattnerac454b32008-10-08 06:26:11 +0000[diff] [blame]4382 %Z = call void @foo() noreturn <i>; indicates that %foo never returns normally</i>
Matthijs Kooijman2c4e05a2008-10-07 10:03:45 +0000[diff] [blame]4383 %ZZ = call zeroext i32 @bar() <i>; Return value is %zero extended</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4384</pre>
4385
4386</div>
4387
4388<!-- _______________________________________________________________________ -->
4389<div class="doc_subsubsection">
4390 <a name="i_va_arg">'<tt>va_arg</tt>' Instruction</a>
4391</div>
4392
4393<div class="doc_text">
4394
4395<h5>Syntax:</h5>
4396
4397<pre>
4398 &lt;resultval&gt; = va_arg &lt;va_list*&gt; &lt;arglist&gt;, &lt;argty&gt;
4399</pre>
4400
4401<h5>Overview:</h5>
4402
4403<p>The '<tt>va_arg</tt>' instruction is used to access arguments passed through
4404the "variable argument" area of a function call. It is used to implement the
4405<tt>va_arg</tt> macro in C.</p>
4406
4407<h5>Arguments:</h5>
4408
4409<p>This instruction takes a <tt>va_list*</tt> value and the type of
4410the argument. It returns a value of the specified argument type and
4411increments the <tt>va_list</tt> to point to the next argument. The
4412actual type of <tt>va_list</tt> is target specific.</p>
4413
4414<h5>Semantics:</h5>
4415
4416<p>The '<tt>va_arg</tt>' instruction loads an argument of the specified
4417type from the specified <tt>va_list</tt> and causes the
4418<tt>va_list</tt> to point to the next argument. For more information,
4419see the variable argument handling <a href="#int_varargs">Intrinsic
4420Functions</a>.</p>
4421
4422<p>It is legal for this instruction to be called in a function which does not
4423take a variable number of arguments, for example, the <tt>vfprintf</tt>
4424function.</p>
4425
4426<p><tt>va_arg</tt> is an LLVM instruction instead of an <a
4427href="#intrinsics">intrinsic function</a> because it takes a type as an
4428argument.</p>
4429
4430<h5>Example:</h5>
4431
4432<p>See the <a href="#int_varargs">variable argument processing</a> section.</p>
4433
4434</div>
4435
4436<!-- *********************************************************************** -->
4437<div class="doc_section"> <a name="intrinsics">Intrinsic Functions</a> </div>
4438<!-- *********************************************************************** -->
4439
4440<div class="doc_text">
4441
4442<p>LLVM supports the notion of an "intrinsic function". These functions have
4443well known names and semantics and are required to follow certain restrictions.
4444Overall, these intrinsics represent an extension mechanism for the LLVM
4445language that does not require changing all of the transformations in LLVM when
4446adding to the language (or the bitcode reader/writer, the parser, etc...).</p>
4447
4448<p>Intrinsic function names must all start with an "<tt>llvm.</tt>" prefix. This
4449prefix is reserved in LLVM for intrinsic names; thus, function names may not
4450begin with this prefix. Intrinsic functions must always be external functions:
4451you cannot define the body of intrinsic functions. Intrinsic functions may
4452only be used in call or invoke instructions: it is illegal to take the address
4453of an intrinsic function. Additionally, because intrinsic functions are part
4454of the LLVM language, it is required if any are added that they be documented
4455here.</p>
4456
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]4457<p>Some intrinsic functions can be overloaded, i.e., the intrinsic represents
4458a family of functions that perform the same operation but on different data
4459types. Because LLVM can represent over 8 million different integer types,
4460overloading is used commonly to allow an intrinsic function to operate on any
4461integer type. One or more of the argument types or the result type can be
4462overloaded to accept any integer type. Argument types may also be defined as
4463exactly matching a previous argument's type or the result type. This allows an
4464intrinsic function which accepts multiple arguments, but needs all of them to
4465be of the same type, to only be overloaded with respect to a single argument or
4466the result.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4467
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]4468<p>Overloaded intrinsics will have the names of its overloaded argument types
4469encoded into its function name, each preceded by a period. Only those types
4470which are overloaded result in a name suffix. Arguments whose type is matched
4471against another type do not. For example, the <tt>llvm.ctpop</tt> function can
4472take an integer of any width and returns an integer of exactly the same integer
4473width. This leads to a family of functions such as
4474<tt>i8 @llvm.ctpop.i8(i8 %val)</tt> and <tt>i29 @llvm.ctpop.i29(i29 %val)</tt>.
4475Only one type, the return type, is overloaded, and only one type suffix is
4476required. Because the argument's type is matched against the return type, it
4477does not require its own name suffix.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4478
4479<p>To learn how to add an intrinsic function, please see the
4480<a href="ExtendingLLVM.html">Extending LLVM Guide</a>.
4481</p>
4482
4483</div>
4484
4485<!-- ======================================================================= -->
4486<div class="doc_subsection">
4487 <a name="int_varargs">Variable Argument Handling Intrinsics</a>
4488</div>
4489
4490<div class="doc_text">
4491
4492<p>Variable argument support is defined in LLVM with the <a
4493 href="#i_va_arg"><tt>va_arg</tt></a> instruction and these three
4494intrinsic functions. These functions are related to the similarly
4495named macros defined in the <tt>&lt;stdarg.h&gt;</tt> header file.</p>
4496
4497<p>All of these functions operate on arguments that use a
4498target-specific value type "<tt>va_list</tt>". The LLVM assembly
4499language reference manual does not define what this type is, so all
4500transformations should be prepared to handle these functions regardless of
4501the type used.</p>
4502
4503<p>This example shows how the <a href="#i_va_arg"><tt>va_arg</tt></a>
4504instruction and the variable argument handling intrinsic functions are
4505used.</p>
4506
4507<div class="doc_code">
4508<pre>
4509define i32 @test(i32 %X, ...) {
4510 ; Initialize variable argument processing
4511 %ap = alloca i8*
4512 %ap2 = bitcast i8** %ap to i8*
4513 call void @llvm.va_start(i8* %ap2)
4514
4515 ; Read a single integer argument
4516 %tmp = va_arg i8** %ap, i32
4517
4518 ; Demonstrate usage of llvm.va_copy and llvm.va_end
4519 %aq = alloca i8*
4520 %aq2 = bitcast i8** %aq to i8*
4521 call void @llvm.va_copy(i8* %aq2, i8* %ap2)
4522 call void @llvm.va_end(i8* %aq2)
4523
4524 ; Stop processing of arguments.
4525 call void @llvm.va_end(i8* %ap2)
4526 ret i32 %tmp
4527}
4528
4529declare void @llvm.va_start(i8*)
4530declare void @llvm.va_copy(i8*, i8*)
4531declare void @llvm.va_end(i8*)
4532</pre>
4533</div>
4534
4535</div>
4536
4537<!-- _______________________________________________________________________ -->
4538<div class="doc_subsubsection">
4539 <a name="int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a>
4540</div>
4541
4542
4543<div class="doc_text">
4544<h5>Syntax:</h5>
4545<pre> declare void %llvm.va_start(i8* &lt;arglist&gt;)<br></pre>
4546<h5>Overview:</h5>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4547<p>The '<tt>llvm.va_start</tt>' intrinsic initializes
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4548<tt>*&lt;arglist&gt;</tt> for subsequent use by <tt><a
4549href="#i_va_arg">va_arg</a></tt>.</p>
4550
4551<h5>Arguments:</h5>
4552
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4553<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]4554
4555<h5>Semantics:</h5>
4556
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4557<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]4558macro available in C. In a target-dependent way, it initializes the
4559<tt>va_list</tt> element to which the argument points, so that the next call to
4560<tt>va_arg</tt> will produce the first variable argument passed to the function.
4561Unlike the C <tt>va_start</tt> macro, this intrinsic does not need to know the
4562last argument of the function as the compiler can figure that out.</p>
4563
4564</div>
4565
4566<!-- _______________________________________________________________________ -->
4567<div class="doc_subsubsection">
4568 <a name="int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a>
4569</div>
4570
4571<div class="doc_text">
4572<h5>Syntax:</h5>
4573<pre> declare void @llvm.va_end(i8* &lt;arglist&gt;)<br></pre>
4574<h5>Overview:</h5>
4575
4576<p>The '<tt>llvm.va_end</tt>' intrinsic destroys <tt>*&lt;arglist&gt;</tt>,
4577which has been initialized previously with <tt><a href="#int_va_start">llvm.va_start</a></tt>
4578or <tt><a href="#i_va_copy">llvm.va_copy</a></tt>.</p>
4579
4580<h5>Arguments:</h5>
4581
4582<p>The argument is a pointer to a <tt>va_list</tt> to destroy.</p>
4583
4584<h5>Semantics:</h5>
4585
4586<p>The '<tt>llvm.va_end</tt>' intrinsic works just like the <tt>va_end</tt>
4587macro available in C. In a target-dependent way, it destroys the
4588<tt>va_list</tt> element to which the argument points. Calls to <a
4589href="#int_va_start"><tt>llvm.va_start</tt></a> and <a href="#int_va_copy">
4590<tt>llvm.va_copy</tt></a> must be matched exactly with calls to
4591<tt>llvm.va_end</tt>.</p>
4592
4593</div>
4594
4595<!-- _______________________________________________________________________ -->
4596<div class="doc_subsubsection">
4597 <a name="int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a>
4598</div>
4599
4600<div class="doc_text">
4601
4602<h5>Syntax:</h5>
4603
4604<pre>
4605 declare void @llvm.va_copy(i8* &lt;destarglist&gt;, i8* &lt;srcarglist&gt;)
4606</pre>
4607
4608<h5>Overview:</h5>
4609
4610<p>The '<tt>llvm.va_copy</tt>' intrinsic copies the current argument position
4611from the source argument list to the destination argument list.</p>
4612
4613<h5>Arguments:</h5>
4614
4615<p>The first argument is a pointer to a <tt>va_list</tt> element to initialize.
4616The second argument is a pointer to a <tt>va_list</tt> element to copy from.</p>
4617
4618
4619<h5>Semantics:</h5>
4620
4621<p>The '<tt>llvm.va_copy</tt>' intrinsic works just like the <tt>va_copy</tt>
4622macro available in C. In a target-dependent way, it copies the source
4623<tt>va_list</tt> element into the destination <tt>va_list</tt> element. This
4624intrinsic is necessary because the <tt><a href="#int_va_start">
4625llvm.va_start</a></tt> intrinsic may be arbitrarily complex and require, for
4626example, memory allocation.</p>
4627
4628</div>
4629
4630<!-- ======================================================================= -->
4631<div class="doc_subsection">
4632 <a name="int_gc">Accurate Garbage Collection Intrinsics</a>
4633</div>
4634
4635<div class="doc_text">
4636
4637<p>
4638LLVM support for <a href="GarbageCollection.html">Accurate Garbage
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]4639Collection</a> (GC) requires the implementation and generation of these
4640intrinsics.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4641These intrinsics allow identification of <a href="#int_gcroot">GC roots on the
4642stack</a>, as well as garbage collector implementations that require <a
4643href="#int_gcread">read</a> and <a href="#int_gcwrite">write</a> barriers.
4644Front-ends for type-safe garbage collected languages should generate these
4645intrinsics to make use of the LLVM garbage collectors. For more details, see <a
4646href="GarbageCollection.html">Accurate Garbage Collection with LLVM</a>.
4647</p>
Christopher Lambcfe00962007-12-17 01:00:21 +0000[diff] [blame]4648
4649<p>The garbage collection intrinsics only operate on objects in the generic
4650 address space (address space zero).</p>
4651
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4652</div>
4653
4654<!-- _______________________________________________________________________ -->
4655<div class="doc_subsubsection">
4656 <a name="int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a>
4657</div>
4658
4659<div class="doc_text">
4660
4661<h5>Syntax:</h5>
4662
4663<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4664 declare void @llvm.gcroot(i8** %ptrloc, i8* %metadata)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4665</pre>
4666
4667<h5>Overview:</h5>
4668
4669<p>The '<tt>llvm.gcroot</tt>' intrinsic declares the existence of a GC root to
4670the code generator, and allows some metadata to be associated with it.</p>
4671
4672<h5>Arguments:</h5>
4673
4674<p>The first argument specifies the address of a stack object that contains the
4675root pointer. The second pointer (which must be either a constant or a global
4676value address) contains the meta-data to be associated with the root.</p>
4677
4678<h5>Semantics:</h5>
4679
Chris Lattnera7d94ba2008-04-24 05:59:56 +0000[diff] [blame]4680<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]4681location. At compile-time, the code generator generates information to allow
Gordon Henriksen40393542007-12-25 02:31:26 +0000[diff] [blame]4682the runtime to find the pointer at GC safe points. The '<tt>llvm.gcroot</tt>'
4683intrinsic may only be used in a function which <a href="#gc">specifies a GC
4684algorithm</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4685
4686</div>
4687
4688
4689<!-- _______________________________________________________________________ -->
4690<div class="doc_subsubsection">
4691 <a name="int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a>
4692</div>
4693
4694<div class="doc_text">
4695
4696<h5>Syntax:</h5>
4697
4698<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4699 declare i8* @llvm.gcread(i8* %ObjPtr, i8** %Ptr)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4700</pre>
4701
4702<h5>Overview:</h5>
4703
4704<p>The '<tt>llvm.gcread</tt>' intrinsic identifies reads of references from heap
4705locations, allowing garbage collector implementations that require read
4706barriers.</p>
4707
4708<h5>Arguments:</h5>
4709
4710<p>The second argument is the address to read from, which should be an address
4711allocated from the garbage collector. The first object is a pointer to the
4712start of the referenced object, if needed by the language runtime (otherwise
4713null).</p>
4714
4715<h5>Semantics:</h5>
4716
4717<p>The '<tt>llvm.gcread</tt>' intrinsic has the same semantics as a load
4718instruction, but may be replaced with substantially more complex code by the
Gordon Henriksen40393542007-12-25 02:31:26 +0000[diff] [blame]4719garbage collector runtime, as needed. The '<tt>llvm.gcread</tt>' intrinsic
4720may only be used in a function which <a href="#gc">specifies a GC
4721algorithm</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4722
4723</div>
4724
4725
4726<!-- _______________________________________________________________________ -->
4727<div class="doc_subsubsection">
4728 <a name="int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a>
4729</div>
4730
4731<div class="doc_text">
4732
4733<h5>Syntax:</h5>
4734
4735<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4736 declare void @llvm.gcwrite(i8* %P1, i8* %Obj, i8** %P2)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4737</pre>
4738
4739<h5>Overview:</h5>
4740
4741<p>The '<tt>llvm.gcwrite</tt>' intrinsic identifies writes of references to heap
4742locations, allowing garbage collector implementations that require write
4743barriers (such as generational or reference counting collectors).</p>
4744
4745<h5>Arguments:</h5>
4746
4747<p>The first argument is the reference to store, the second is the start of the
4748object to store it to, and the third is the address of the field of Obj to
4749store to. If the runtime does not require a pointer to the object, Obj may be
4750null.</p>
4751
4752<h5>Semantics:</h5>
4753
4754<p>The '<tt>llvm.gcwrite</tt>' intrinsic has the same semantics as a store
4755instruction, but may be replaced with substantially more complex code by the
Gordon Henriksen40393542007-12-25 02:31:26 +0000[diff] [blame]4756garbage collector runtime, as needed. The '<tt>llvm.gcwrite</tt>' intrinsic
4757may only be used in a function which <a href="#gc">specifies a GC
4758algorithm</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4759
4760</div>
4761
4762
4763
4764<!-- ======================================================================= -->
4765<div class="doc_subsection">
4766 <a name="int_codegen">Code Generator Intrinsics</a>
4767</div>
4768
4769<div class="doc_text">
4770<p>
4771These intrinsics are provided by LLVM to expose special features that may only
4772be implemented with code generator support.
4773</p>
4774
4775</div>
4776
4777<!-- _______________________________________________________________________ -->
4778<div class="doc_subsubsection">
4779 <a name="int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a>
4780</div>
4781
4782<div class="doc_text">
4783
4784<h5>Syntax:</h5>
4785<pre>
4786 declare i8 *@llvm.returnaddress(i32 &lt;level&gt;)
4787</pre>
4788
4789<h5>Overview:</h5>
4790
4791<p>
4792The '<tt>llvm.returnaddress</tt>' intrinsic attempts to compute a
4793target-specific value indicating the return address of the current function
4794or one of its callers.
4795</p>
4796
4797<h5>Arguments:</h5>
4798
4799<p>
4800The argument to this intrinsic indicates which function to return the address
4801for. Zero indicates the calling function, one indicates its caller, etc. The
4802argument is <b>required</b> to be a constant integer value.
4803</p>
4804
4805<h5>Semantics:</h5>
4806
4807<p>
4808The '<tt>llvm.returnaddress</tt>' intrinsic either returns a pointer indicating
4809the return address of the specified call frame, or zero if it cannot be
4810identified. The value returned by this intrinsic is likely to be incorrect or 0
4811for arguments other than zero, so it should only be used for debugging purposes.
4812</p>
4813
4814<p>
4815Note that calling this intrinsic does not prevent function inlining or other
4816aggressive transformations, so the value returned may not be that of the obvious
4817source-language caller.
4818</p>
4819</div>
4820
4821
4822<!-- _______________________________________________________________________ -->
4823<div class="doc_subsubsection">
4824 <a name="int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a>
4825</div>
4826
4827<div class="doc_text">
4828
4829<h5>Syntax:</h5>
4830<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4831 declare i8 *@llvm.frameaddress(i32 &lt;level&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4832</pre>
4833
4834<h5>Overview:</h5>
4835
4836<p>
4837The '<tt>llvm.frameaddress</tt>' intrinsic attempts to return the
4838target-specific frame pointer value for the specified stack frame.
4839</p>
4840
4841<h5>Arguments:</h5>
4842
4843<p>
4844The argument to this intrinsic indicates which function to return the frame
4845pointer for. Zero indicates the calling function, one indicates its caller,
4846etc. The argument is <b>required</b> to be a constant integer value.
4847</p>
4848
4849<h5>Semantics:</h5>
4850
4851<p>
4852The '<tt>llvm.frameaddress</tt>' intrinsic either returns a pointer indicating
4853the frame address of the specified call frame, or zero if it cannot be
4854identified. The value returned by this intrinsic is likely to be incorrect or 0
4855for arguments other than zero, so it should only be used for debugging purposes.
4856</p>
4857
4858<p>
4859Note that calling this intrinsic does not prevent function inlining or other
4860aggressive transformations, so the value returned may not be that of the obvious
4861source-language caller.
4862</p>
4863</div>
4864
4865<!-- _______________________________________________________________________ -->
4866<div class="doc_subsubsection">
4867 <a name="int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a>
4868</div>
4869
4870<div class="doc_text">
4871
4872<h5>Syntax:</h5>
4873<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4874 declare i8 *@llvm.stacksave()
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4875</pre>
4876
4877<h5>Overview:</h5>
4878
4879<p>
4880The '<tt>llvm.stacksave</tt>' intrinsic is used to remember the current state of
4881the function stack, for use with <a href="#int_stackrestore">
4882<tt>llvm.stackrestore</tt></a>. This is useful for implementing language
4883features like scoped automatic variable sized arrays in C99.
4884</p>
4885
4886<h5>Semantics:</h5>
4887
4888<p>
4889This intrinsic returns a opaque pointer value that can be passed to <a
4890href="#int_stackrestore"><tt>llvm.stackrestore</tt></a>. When an
4891<tt>llvm.stackrestore</tt> intrinsic is executed with a value saved from
4892<tt>llvm.stacksave</tt>, it effectively restores the state of the stack to the
4893state it was in when the <tt>llvm.stacksave</tt> intrinsic executed. In
4894practice, this pops any <a href="#i_alloca">alloca</a> blocks from the stack
4895that were allocated after the <tt>llvm.stacksave</tt> was executed.
4896</p>
4897
4898</div>
4899
4900<!-- _______________________________________________________________________ -->
4901<div class="doc_subsubsection">
4902 <a name="int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a>
4903</div>
4904
4905<div class="doc_text">
4906
4907<h5>Syntax:</h5>
4908<pre>
4909 declare void @llvm.stackrestore(i8 * %ptr)
4910</pre>
4911
4912<h5>Overview:</h5>
4913
4914<p>
4915The '<tt>llvm.stackrestore</tt>' intrinsic is used to restore the state of
4916the function stack to the state it was in when the corresponding <a
4917href="#int_stacksave"><tt>llvm.stacksave</tt></a> intrinsic executed. This is
4918useful for implementing language features like scoped automatic variable sized
4919arrays in C99.
4920</p>
4921
4922<h5>Semantics:</h5>
4923
4924<p>
4925See the description for <a href="#int_stacksave"><tt>llvm.stacksave</tt></a>.
4926</p>
4927
4928</div>
4929
4930
4931<!-- _______________________________________________________________________ -->
4932<div class="doc_subsubsection">
4933 <a name="int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a>
4934</div>
4935
4936<div class="doc_text">
4937
4938<h5>Syntax:</h5>
4939<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4940 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]4941</pre>
4942
4943<h5>Overview:</h5>
4944
4945
4946<p>
4947The '<tt>llvm.prefetch</tt>' intrinsic is a hint to the code generator to insert
4948a prefetch instruction if supported; otherwise, it is a noop. Prefetches have
4949no
4950effect on the behavior of the program but can change its performance
4951characteristics.
4952</p>
4953
4954<h5>Arguments:</h5>
4955
4956<p>
4957<tt>address</tt> is the address to be prefetched, <tt>rw</tt> is the specifier
4958determining if the fetch should be for a read (0) or write (1), and
4959<tt>locality</tt> is a temporal locality specifier ranging from (0) - no
4960locality, to (3) - extremely local keep in cache. The <tt>rw</tt> and
4961<tt>locality</tt> arguments must be constant integers.
4962</p>
4963
4964<h5>Semantics:</h5>
4965
4966<p>
4967This intrinsic does not modify the behavior of the program. In particular,
4968prefetches cannot trap and do not produce a value. On targets that support this
4969intrinsic, the prefetch can provide hints to the processor cache for better
4970performance.
4971</p>
4972
4973</div>
4974
4975<!-- _______________________________________________________________________ -->
4976<div class="doc_subsubsection">
4977 <a name="int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a>
4978</div>
4979
4980<div class="doc_text">
4981
4982<h5>Syntax:</h5>
4983<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4984 declare void @llvm.pcmarker(i32 &lt;id&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4985</pre>
4986
4987<h5>Overview:</h5>
4988
4989
4990<p>
4991The '<tt>llvm.pcmarker</tt>' intrinsic is a method to export a Program Counter
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]4992(PC) in a region of
4993code to simulators and other tools. The method is target specific, but it is
4994expected that the marker will use exported symbols to transmit the PC of the
4995marker.
4996The marker makes no guarantees that it will remain with any specific instruction
4997after optimizations. It is possible that the presence of a marker will inhibit
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4998optimizations. The intended use is to be inserted after optimizations to allow
4999correlations of simulation runs.
5000</p>
5001
5002<h5>Arguments:</h5>
5003
5004<p>
5005<tt>id</tt> is a numerical id identifying the marker.
5006</p>
5007
5008<h5>Semantics:</h5>
5009
5010<p>
5011This intrinsic does not modify the behavior of the program. Backends that do not
5012support this intrinisic may ignore it.
5013</p>
5014
5015</div>
5016
5017<!-- _______________________________________________________________________ -->
5018<div class="doc_subsubsection">
5019 <a name="int_readcyclecounter">'<tt>llvm.readcyclecounter</tt>' Intrinsic</a>
5020</div>
5021
5022<div class="doc_text">
5023
5024<h5>Syntax:</h5>
5025<pre>
5026 declare i64 @llvm.readcyclecounter( )
5027</pre>
5028
5029<h5>Overview:</h5>
5030
5031
5032<p>
5033The '<tt>llvm.readcyclecounter</tt>' intrinsic provides access to the cycle
5034counter register (or similar low latency, high accuracy clocks) on those targets
5035that support it. On X86, it should map to RDTSC. On Alpha, it should map to RPCC.
5036As the backing counters overflow quickly (on the order of 9 seconds on alpha), this
5037should only be used for small timings.
5038</p>
5039
5040<h5>Semantics:</h5>
5041
5042<p>
5043When directly supported, reading the cycle counter should not modify any memory.
5044Implementations are allowed to either return a application specific value or a
5045system wide value. On backends without support, this is lowered to a constant 0.
5046</p>
5047
5048</div>
5049
5050<!-- ======================================================================= -->
5051<div class="doc_subsection">
5052 <a name="int_libc">Standard C Library Intrinsics</a>
5053</div>
5054
5055<div class="doc_text">
5056<p>
5057LLVM provides intrinsics for a few important standard C library functions.
5058These intrinsics allow source-language front-ends to pass information about the
5059alignment of the pointer arguments to the code generator, providing opportunity
5060for more efficient code generation.
5061</p>
5062
5063</div>
5064
5065<!-- _______________________________________________________________________ -->
5066<div class="doc_subsubsection">
5067 <a name="int_memcpy">'<tt>llvm.memcpy</tt>' Intrinsic</a>
5068</div>
5069
5070<div class="doc_text">
5071
5072<h5>Syntax:</h5>
Chris Lattner82c2e432008-11-21 16:42:48 +0000[diff] [blame]5073<p>This is an overloaded intrinsic. You can use llvm.memcpy on any integer bit
5074width. Not all targets support all bit widths however.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5075<pre>
Chris Lattner82c2e432008-11-21 16:42:48 +0000[diff] [blame]5076 declare void @llvm.memcpy.i8(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5077 i8 &lt;len&gt;, i32 &lt;align&gt;)
5078 declare void @llvm.memcpy.i16(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5079 i16 &lt;len&gt;, i32 &lt;align&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5080 declare void @llvm.memcpy.i32(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5081 i32 &lt;len&gt;, i32 &lt;align&gt;)
5082 declare void @llvm.memcpy.i64(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5083 i64 &lt;len&gt;, i32 &lt;align&gt;)
5084</pre>
5085
5086<h5>Overview:</h5>
5087
5088<p>
5089The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the source
5090location to the destination location.
5091</p>
5092
5093<p>
5094Note that, unlike the standard libc function, the <tt>llvm.memcpy.*</tt>
5095intrinsics do not return a value, and takes an extra alignment argument.
5096</p>
5097
5098<h5>Arguments:</h5>
5099
5100<p>
5101The first argument is a pointer to the destination, the second is a pointer to
5102the source. The third argument is an integer argument
5103specifying the number of bytes to copy, and the fourth argument is the alignment
5104of the source and destination locations.
5105</p>
5106
5107<p>
5108If the call to this intrinisic has an alignment value that is not 0 or 1, then
5109the caller guarantees that both the source and destination pointers are aligned
5110to that boundary.
5111</p>
5112
5113<h5>Semantics:</h5>
5114
5115<p>
5116The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the source
5117location to the destination location, which are not allowed to overlap. It
5118copies "len" bytes of memory over. If the argument is known to be aligned to
5119some boundary, this can be specified as the fourth argument, otherwise it should
5120be set to 0 or 1.
5121</p>
5122</div>
5123
5124
5125<!-- _______________________________________________________________________ -->
5126<div class="doc_subsubsection">
5127 <a name="int_memmove">'<tt>llvm.memmove</tt>' Intrinsic</a>
5128</div>
5129
5130<div class="doc_text">
5131
5132<h5>Syntax:</h5>
Chris Lattner82c2e432008-11-21 16:42:48 +0000[diff] [blame]5133<p>This is an overloaded intrinsic. You can use llvm.memmove on any integer bit
5134width. Not all targets support all bit widths however.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5135<pre>
Chris Lattner82c2e432008-11-21 16:42:48 +0000[diff] [blame]5136 declare void @llvm.memmove.i8(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5137 i8 &lt;len&gt;, i32 &lt;align&gt;)
5138 declare void @llvm.memmove.i16(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5139 i16 &lt;len&gt;, i32 &lt;align&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5140 declare void @llvm.memmove.i32(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5141 i32 &lt;len&gt;, i32 &lt;align&gt;)
5142 declare void @llvm.memmove.i64(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5143 i64 &lt;len&gt;, i32 &lt;align&gt;)
5144</pre>
5145
5146<h5>Overview:</h5>
5147
5148<p>
5149The '<tt>llvm.memmove.*</tt>' intrinsics move a block of memory from the source
5150location to the destination location. It is similar to the
Chris Lattnerdba16ea2008-01-06 19:51:52 +0000[diff] [blame]5151'<tt>llvm.memcpy</tt>' intrinsic but allows the two memory locations to overlap.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5152</p>
5153
5154<p>
5155Note that, unlike the standard libc function, the <tt>llvm.memmove.*</tt>
5156intrinsics do not return a value, and takes an extra alignment argument.
5157</p>
5158
5159<h5>Arguments:</h5>
5160
5161<p>
5162The first argument is a pointer to the destination, the second is a pointer to
5163the source. The third argument is an integer argument
5164specifying the number of bytes to copy, and the fourth argument is the alignment
5165of the source and destination locations.
5166</p>
5167
5168<p>
5169If the call to this intrinisic has an alignment value that is not 0 or 1, then
5170the caller guarantees that the source and destination pointers are aligned to
5171that boundary.
5172</p>
5173
5174<h5>Semantics:</h5>
5175
5176<p>
5177The '<tt>llvm.memmove.*</tt>' intrinsics copy a block of memory from the source
5178location to the destination location, which may overlap. It
5179copies "len" bytes of memory over. If the argument is known to be aligned to
5180some boundary, this can be specified as the fourth argument, otherwise it should
5181be set to 0 or 1.
5182</p>
5183</div>
5184
5185
5186<!-- _______________________________________________________________________ -->
5187<div class="doc_subsubsection">
5188 <a name="int_memset">'<tt>llvm.memset.*</tt>' Intrinsics</a>
5189</div>
5190
5191<div class="doc_text">
5192
5193<h5>Syntax:</h5>
Chris Lattner82c2e432008-11-21 16:42:48 +0000[diff] [blame]5194<p>This is an overloaded intrinsic. You can use llvm.memset on any integer bit
5195width. Not all targets support all bit widths however.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5196<pre>
Chris Lattner82c2e432008-11-21 16:42:48 +0000[diff] [blame]5197 declare void @llvm.memset.i8(i8 * &lt;dest&gt;, i8 &lt;val&gt;,
5198 i8 &lt;len&gt;, i32 &lt;align&gt;)
5199 declare void @llvm.memset.i16(i8 * &lt;dest&gt;, i8 &lt;val&gt;,
5200 i16 &lt;len&gt;, i32 &lt;align&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5201 declare void @llvm.memset.i32(i8 * &lt;dest&gt;, i8 &lt;val&gt;,
5202 i32 &lt;len&gt;, i32 &lt;align&gt;)
5203 declare void @llvm.memset.i64(i8 * &lt;dest&gt;, i8 &lt;val&gt;,
5204 i64 &lt;len&gt;, i32 &lt;align&gt;)
5205</pre>
5206
5207<h5>Overview:</h5>
5208
5209<p>
5210The '<tt>llvm.memset.*</tt>' intrinsics fill a block of memory with a particular
5211byte value.
5212</p>
5213
5214<p>
5215Note that, unlike the standard libc function, the <tt>llvm.memset</tt> intrinsic
5216does not return a value, and takes an extra alignment argument.
5217</p>
5218
5219<h5>Arguments:</h5>
5220
5221<p>
5222The first argument is a pointer to the destination to fill, the second is the
5223byte value to fill it with, the third argument is an integer
5224argument specifying the number of bytes to fill, and the fourth argument is the
5225known alignment of destination location.
5226</p>
5227
5228<p>
5229If the call to this intrinisic has an alignment value that is not 0 or 1, then
5230the caller guarantees that the destination pointer is aligned to that boundary.
5231</p>
5232
5233<h5>Semantics:</h5>
5234
5235<p>
5236The '<tt>llvm.memset.*</tt>' intrinsics fill "len" bytes of memory starting at
5237the
5238destination location. If the argument is known to be aligned to some boundary,
5239this can be specified as the fourth argument, otherwise it should be set to 0 or
52401.
5241</p>
5242</div>
5243
5244
5245<!-- _______________________________________________________________________ -->
5246<div class="doc_subsubsection">
5247 <a name="int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a>
5248</div>
5249
5250<div class="doc_text">
5251
5252<h5>Syntax:</h5>
Dale Johannesenf9adbb62007-10-02 17:47:38 +0000[diff] [blame]5253<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]5254floating point or vector of floating point type. Not all targets support all
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5255types however.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5256<pre>
Dale Johannesenf9adbb62007-10-02 17:47:38 +0000[diff] [blame]5257 declare float @llvm.sqrt.f32(float %Val)
5258 declare double @llvm.sqrt.f64(double %Val)
5259 declare x86_fp80 @llvm.sqrt.f80(x86_fp80 %Val)
5260 declare fp128 @llvm.sqrt.f128(fp128 %Val)
5261 declare ppc_fp128 @llvm.sqrt.ppcf128(ppc_fp128 %Val)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5262</pre>
5263
5264<h5>Overview:</h5>
5265
5266<p>
5267The '<tt>llvm.sqrt</tt>' intrinsics return the sqrt of the specified operand,
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5268returning the same value as the libm '<tt>sqrt</tt>' functions would. Unlike
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5269<tt>sqrt</tt> in libm, however, <tt>llvm.sqrt</tt> has undefined behavior for
Chris Lattnerf914a002008-01-29 07:00:44 +0000[diff] [blame]5270negative numbers other than -0.0 (which allows for better optimization, because
5271there is no need to worry about errno being set). <tt>llvm.sqrt(-0.0)</tt> is
5272defined to return -0.0 like IEEE sqrt.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5273</p>
5274
5275<h5>Arguments:</h5>
5276
5277<p>
5278The argument and return value are floating point numbers of the same type.
5279</p>
5280
5281<h5>Semantics:</h5>
5282
5283<p>
5284This function returns the sqrt of the specified operand if it is a nonnegative
5285floating point number.
5286</p>
5287</div>
5288
5289<!-- _______________________________________________________________________ -->
5290<div class="doc_subsubsection">
5291 <a name="int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a>
5292</div>
5293
5294<div class="doc_text">
5295
5296<h5>Syntax:</h5>
Dale Johannesenf9adbb62007-10-02 17:47:38 +0000[diff] [blame]5297<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]5298floating point or vector of floating point type. Not all targets support all
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5299types however.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5300<pre>
Dale Johannesenf9adbb62007-10-02 17:47:38 +0000[diff] [blame]5301 declare float @llvm.powi.f32(float %Val, i32 %power)
5302 declare double @llvm.powi.f64(double %Val, i32 %power)
5303 declare x86_fp80 @llvm.powi.f80(x86_fp80 %Val, i32 %power)
5304 declare fp128 @llvm.powi.f128(fp128 %Val, i32 %power)
5305 declare ppc_fp128 @llvm.powi.ppcf128(ppc_fp128 %Val, i32 %power)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5306</pre>
5307
5308<h5>Overview:</h5>
5309
5310<p>
5311The '<tt>llvm.powi.*</tt>' intrinsics return the first operand raised to the
5312specified (positive or negative) power. The order of evaluation of
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5313multiplications is not defined. When a vector of floating point type is
5314used, the second argument remains a scalar integer value.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5315</p>
5316
5317<h5>Arguments:</h5>
5318
5319<p>
5320The second argument is an integer power, and the first is a value to raise to
5321that power.
5322</p>
5323
5324<h5>Semantics:</h5>
5325
5326<p>
5327This function returns the first value raised to the second power with an
5328unspecified sequence of rounding operations.</p>
5329</div>
5330
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5331<!-- _______________________________________________________________________ -->
5332<div class="doc_subsubsection">
5333 <a name="int_sin">'<tt>llvm.sin.*</tt>' Intrinsic</a>
5334</div>
5335
5336<div class="doc_text">
5337
5338<h5>Syntax:</h5>
5339<p>This is an overloaded intrinsic. You can use <tt>llvm.sin</tt> on any
5340floating point or vector of floating point type. Not all targets support all
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5341types however.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5342<pre>
5343 declare float @llvm.sin.f32(float %Val)
5344 declare double @llvm.sin.f64(double %Val)
5345 declare x86_fp80 @llvm.sin.f80(x86_fp80 %Val)
5346 declare fp128 @llvm.sin.f128(fp128 %Val)
5347 declare ppc_fp128 @llvm.sin.ppcf128(ppc_fp128 %Val)
5348</pre>
5349
5350<h5>Overview:</h5>
5351
5352<p>
5353The '<tt>llvm.sin.*</tt>' intrinsics return the sine of the operand.
5354</p>
5355
5356<h5>Arguments:</h5>
5357
5358<p>
5359The argument and return value are floating point numbers of the same type.
5360</p>
5361
5362<h5>Semantics:</h5>
5363
5364<p>
5365This function returns the sine of the specified operand, returning the
5366same values as the libm <tt>sin</tt> functions would, and handles error
Dan Gohmaneaba92e2007-10-17 18:05:13 +0000[diff] [blame]5367conditions in the same way.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5368</div>
5369
5370<!-- _______________________________________________________________________ -->
5371<div class="doc_subsubsection">
5372 <a name="int_cos">'<tt>llvm.cos.*</tt>' Intrinsic</a>
5373</div>
5374
5375<div class="doc_text">
5376
5377<h5>Syntax:</h5>
5378<p>This is an overloaded intrinsic. You can use <tt>llvm.cos</tt> on any
5379floating point or vector of floating point type. Not all targets support all
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5380types however.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5381<pre>
5382 declare float @llvm.cos.f32(float %Val)
5383 declare double @llvm.cos.f64(double %Val)
5384 declare x86_fp80 @llvm.cos.f80(x86_fp80 %Val)
5385 declare fp128 @llvm.cos.f128(fp128 %Val)
5386 declare ppc_fp128 @llvm.cos.ppcf128(ppc_fp128 %Val)
5387</pre>
5388
5389<h5>Overview:</h5>
5390
5391<p>
5392The '<tt>llvm.cos.*</tt>' intrinsics return the cosine of the operand.
5393</p>
5394
5395<h5>Arguments:</h5>
5396
5397<p>
5398The argument and return value are floating point numbers of the same type.
5399</p>
5400
5401<h5>Semantics:</h5>
5402
5403<p>
5404This function returns the cosine of the specified operand, returning the
5405same values as the libm <tt>cos</tt> functions would, and handles error
Dan Gohmaneaba92e2007-10-17 18:05:13 +0000[diff] [blame]5406conditions in the same way.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5407</div>
5408
5409<!-- _______________________________________________________________________ -->
5410<div class="doc_subsubsection">
5411 <a name="int_pow">'<tt>llvm.pow.*</tt>' Intrinsic</a>
5412</div>
5413
5414<div class="doc_text">
5415
5416<h5>Syntax:</h5>
5417<p>This is an overloaded intrinsic. You can use <tt>llvm.pow</tt> on any
5418floating point or vector of floating point type. Not all targets support all
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5419types however.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5420<pre>
5421 declare float @llvm.pow.f32(float %Val, float %Power)
5422 declare double @llvm.pow.f64(double %Val, double %Power)
5423 declare x86_fp80 @llvm.pow.f80(x86_fp80 %Val, x86_fp80 %Power)
5424 declare fp128 @llvm.pow.f128(fp128 %Val, fp128 %Power)
5425 declare ppc_fp128 @llvm.pow.ppcf128(ppc_fp128 %Val, ppc_fp128 Power)
5426</pre>
5427
5428<h5>Overview:</h5>
5429
5430<p>
5431The '<tt>llvm.pow.*</tt>' intrinsics return the first operand raised to the
5432specified (positive or negative) power.
5433</p>
5434
5435<h5>Arguments:</h5>
5436
5437<p>
5438The second argument is a floating point power, and the first is a value to
5439raise to that power.
5440</p>
5441
5442<h5>Semantics:</h5>
5443
5444<p>
5445This function returns the first value raised to the second power,
5446returning the
5447same values as the libm <tt>pow</tt> functions would, and handles error
Dan Gohmaneaba92e2007-10-17 18:05:13 +0000[diff] [blame]5448conditions in the same way.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5449</div>
5450
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5451
5452<!-- ======================================================================= -->
5453<div class="doc_subsection">
5454 <a name="int_manip">Bit Manipulation Intrinsics</a>
5455</div>
5456
5457<div class="doc_text">
5458<p>
5459LLVM provides intrinsics for a few important bit manipulation operations.
5460These allow efficient code generation for some algorithms.
5461</p>
5462
5463</div>
5464
5465<!-- _______________________________________________________________________ -->
5466<div class="doc_subsubsection">
5467 <a name="int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a>
5468</div>
5469
5470<div class="doc_text">
5471
5472<h5>Syntax:</h5>
5473<p>This is an overloaded intrinsic function. You can use bswap on any integer
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5474type that is an even number of bytes (i.e. BitWidth % 16 == 0).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5475<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5476 declare i16 @llvm.bswap.i16(i16 &lt;id&gt;)
5477 declare i32 @llvm.bswap.i32(i32 &lt;id&gt;)
5478 declare i64 @llvm.bswap.i64(i64 &lt;id&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5479</pre>
5480
5481<h5>Overview:</h5>
5482
5483<p>
5484The '<tt>llvm.bswap</tt>' family of intrinsics is used to byte swap integer
5485values with an even number of bytes (positive multiple of 16 bits). These are
5486useful for performing operations on data that is not in the target's native
5487byte order.
5488</p>
5489
5490<h5>Semantics:</h5>
5491
5492<p>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5493The <tt>llvm.bswap.i16</tt> intrinsic returns an i16 value that has the high
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5494and low byte of the input i16 swapped. Similarly, the <tt>llvm.bswap.i32</tt>
5495intrinsic returns an i32 value that has the four bytes of the input i32
5496swapped, 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]5497i32 will have its bytes in 3, 2, 1, 0 order. The <tt>llvm.bswap.i48</tt>,
5498<tt>llvm.bswap.i64</tt> and other intrinsics extend this concept to
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5499additional even-byte lengths (6 bytes, 8 bytes and more, respectively).
5500</p>
5501
5502</div>
5503
5504<!-- _______________________________________________________________________ -->
5505<div class="doc_subsubsection">
5506 <a name="int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic</a>
5507</div>
5508
5509<div class="doc_text">
5510
5511<h5>Syntax:</h5>
5512<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]5513width. Not all targets support all bit widths however.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5514<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5515 declare i8 @llvm.ctpop.i8 (i8 &lt;src&gt;)
5516 declare i16 @llvm.ctpop.i16(i16 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5517 declare i32 @llvm.ctpop.i32(i32 &lt;src&gt;)
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5518 declare i64 @llvm.ctpop.i64(i64 &lt;src&gt;)
5519 declare i256 @llvm.ctpop.i256(i256 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5520</pre>
5521
5522<h5>Overview:</h5>
5523
5524<p>
5525The '<tt>llvm.ctpop</tt>' family of intrinsics counts the number of bits set in a
5526value.
5527</p>
5528
5529<h5>Arguments:</h5>
5530
5531<p>
5532The only argument is the value to be counted. The argument may be of any
5533integer type. The return type must match the argument type.
5534</p>
5535
5536<h5>Semantics:</h5>
5537
5538<p>
5539The '<tt>llvm.ctpop</tt>' intrinsic counts the 1's in a variable.
5540</p>
5541</div>
5542
5543<!-- _______________________________________________________________________ -->
5544<div class="doc_subsubsection">
5545 <a name="int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic</a>
5546</div>
5547
5548<div class="doc_text">
5549
5550<h5>Syntax:</h5>
5551<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]5552integer bit width. Not all targets support all bit widths however.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5553<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5554 declare i8 @llvm.ctlz.i8 (i8 &lt;src&gt;)
5555 declare i16 @llvm.ctlz.i16(i16 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5556 declare i32 @llvm.ctlz.i32(i32 &lt;src&gt;)
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5557 declare i64 @llvm.ctlz.i64(i64 &lt;src&gt;)
5558 declare i256 @llvm.ctlz.i256(i256 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5559</pre>
5560
5561<h5>Overview:</h5>
5562
5563<p>
5564The '<tt>llvm.ctlz</tt>' family of intrinsic functions counts the number of
5565leading zeros in a variable.
5566</p>
5567
5568<h5>Arguments:</h5>
5569
5570<p>
5571The only argument is the value to be counted. The argument may be of any
5572integer type. The return type must match the argument type.
5573</p>
5574
5575<h5>Semantics:</h5>
5576
5577<p>
5578The '<tt>llvm.ctlz</tt>' intrinsic counts the leading (most significant) zeros
5579in a variable. If the src == 0 then the result is the size in bits of the type
5580of src. For example, <tt>llvm.ctlz(i32 2) = 30</tt>.
5581</p>
5582</div>
5583
5584
5585
5586<!-- _______________________________________________________________________ -->
5587<div class="doc_subsubsection">
5588 <a name="int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic</a>
5589</div>
5590
5591<div class="doc_text">
5592
5593<h5>Syntax:</h5>
5594<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]5595integer bit width. Not all targets support all bit widths however.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5596<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5597 declare i8 @llvm.cttz.i8 (i8 &lt;src&gt;)
5598 declare i16 @llvm.cttz.i16(i16 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5599 declare i32 @llvm.cttz.i32(i32 &lt;src&gt;)
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5600 declare i64 @llvm.cttz.i64(i64 &lt;src&gt;)
5601 declare i256 @llvm.cttz.i256(i256 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5602</pre>
5603
5604<h5>Overview:</h5>
5605
5606<p>
5607The '<tt>llvm.cttz</tt>' family of intrinsic functions counts the number of
5608trailing zeros.
5609</p>
5610
5611<h5>Arguments:</h5>
5612
5613<p>
5614The only argument is the value to be counted. The argument may be of any
5615integer type. The return type must match the argument type.
5616</p>
5617
5618<h5>Semantics:</h5>
5619
5620<p>
5621The '<tt>llvm.cttz</tt>' intrinsic counts the trailing (least significant) zeros
5622in a variable. If the src == 0 then the result is the size in bits of the type
5623of src. For example, <tt>llvm.cttz(2) = 1</tt>.
5624</p>
5625</div>
5626
5627<!-- _______________________________________________________________________ -->
5628<div class="doc_subsubsection">
5629 <a name="int_part_select">'<tt>llvm.part.select.*</tt>' Intrinsic</a>
5630</div>
5631
5632<div class="doc_text">
5633
5634<h5>Syntax:</h5>
5635<p>This is an overloaded intrinsic. You can use <tt>llvm.part.select</tt>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5636on any integer bit width.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5637<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5638 declare i17 @llvm.part.select.i17 (i17 %val, i32 %loBit, i32 %hiBit)
5639 declare i29 @llvm.part.select.i29 (i29 %val, i32 %loBit, i32 %hiBit)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5640</pre>
5641
5642<h5>Overview:</h5>
5643<p>The '<tt>llvm.part.select</tt>' family of intrinsic functions selects a
5644range of bits from an integer value and returns them in the same bit width as
5645the original value.</p>
5646
5647<h5>Arguments:</h5>
5648<p>The first argument, <tt>%val</tt> and the result may be integer types of
5649any bit width but they must have the same bit width. The second and third
5650arguments must be <tt>i32</tt> type since they specify only a bit index.</p>
5651
5652<h5>Semantics:</h5>
5653<p>The operation of the '<tt>llvm.part.select</tt>' intrinsic has two modes
5654of operation: forwards and reverse. If <tt>%loBit</tt> is greater than
5655<tt>%hiBits</tt> then the intrinsic operates in reverse mode. Otherwise it
5656operates in forward mode.</p>
5657<p>In forward mode, this intrinsic is the equivalent of shifting <tt>%val</tt>
5658right by <tt>%loBit</tt> bits and then ANDing it with a mask with
5659only the <tt>%hiBit - %loBit</tt> bits set, as follows:</p>
5660<ol>
5661 <li>The <tt>%val</tt> is shifted right (LSHR) by the number of bits specified
5662 by <tt>%loBits</tt>. This normalizes the value to the low order bits.</li>
5663 <li>The <tt>%loBits</tt> value is subtracted from the <tt>%hiBits</tt> value
5664 to determine the number of bits to retain.</li>
5665 <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]5666 <li>The mask is ANDed with <tt>%val</tt> to produce the result.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5667</ol>
5668<p>In reverse mode, a similar computation is made except that the bits are
5669returned in the reverse order. So, for example, if <tt>X</tt> has the value
5670<tt>i16 0x0ACF (101011001111)</tt> and we apply
5671<tt>part.select(i16 X, 8, 3)</tt> to it, we get back the value
5672<tt>i16 0x0026 (000000100110)</tt>.</p>
5673</div>
5674
5675<div class="doc_subsubsection">
5676 <a name="int_part_set">'<tt>llvm.part.set.*</tt>' Intrinsic</a>
5677</div>
5678
5679<div class="doc_text">
5680
5681<h5>Syntax:</h5>
5682<p>This is an overloaded intrinsic. You can use <tt>llvm.part.set</tt>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5683on any integer bit width.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5684<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5685 declare i17 @llvm.part.set.i17.i9 (i17 %val, i9 %repl, i32 %lo, i32 %hi)
5686 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]5687</pre>
5688
5689<h5>Overview:</h5>
5690<p>The '<tt>llvm.part.set</tt>' family of intrinsic functions replaces a range
5691of bits in an integer value with another integer value. It returns the integer
5692with the replaced bits.</p>
5693
5694<h5>Arguments:</h5>
5695<p>The first argument, <tt>%val</tt> and the result may be integer types of
5696any bit width but they must have the same bit width. <tt>%val</tt> is the value
5697whose bits will be replaced. The second argument, <tt>%repl</tt> may be an
5698integer of any bit width. The third and fourth arguments must be <tt>i32</tt>
5699type since they specify only a bit index.</p>
5700
5701<h5>Semantics:</h5>
5702<p>The operation of the '<tt>llvm.part.set</tt>' intrinsic has two modes
5703of operation: forwards and reverse. If <tt>%lo</tt> is greater than
5704<tt>%hi</tt> then the intrinsic operates in reverse mode. Otherwise it
5705operates in forward mode.</p>
5706<p>For both modes, the <tt>%repl</tt> value is prepared for use by either
5707truncating it down to the size of the replacement area or zero extending it
5708up to that size.</p>
5709<p>In forward mode, the bits between <tt>%lo</tt> and <tt>%hi</tt> (inclusive)
5710are replaced with corresponding bits from <tt>%repl</tt>. That is the 0th bit
5711in <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]5712to the <tt>%hi</tt>th bit.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5713<p>In reverse mode, a similar computation is made except that the bits are
5714reversed. That is, the <tt>0</tt>th bit in <tt>%repl</tt> replaces the
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5715<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]5716<h5>Examples:</h5>
5717<pre>
5718 llvm.part.set(0xFFFF, 0, 4, 7) -&gt; 0xFF0F
5719 llvm.part.set(0xFFFF, 0, 7, 4) -&gt; 0xFF0F
5720 llvm.part.set(0xFFFF, 1, 7, 4) -&gt; 0xFF8F
5721 llvm.part.set(0xFFFF, F, 8, 3) -&gt; 0xFFE7
5722 llvm.part.set(0xFFFF, 0, 3, 8) -&gt; 0xFE07
5723</pre>
5724</div>
5725
5726<!-- ======================================================================= -->
5727<div class="doc_subsection">
5728 <a name="int_debugger">Debugger Intrinsics</a>
5729</div>
5730
5731<div class="doc_text">
5732<p>
5733The LLVM debugger intrinsics (which all start with <tt>llvm.dbg.</tt> prefix),
5734are described in the <a
5735href="SourceLevelDebugging.html#format_common_intrinsics">LLVM Source Level
5736Debugging</a> document.
5737</p>
5738</div>
5739
5740
5741<!-- ======================================================================= -->
5742<div class="doc_subsection">
5743 <a name="int_eh">Exception Handling Intrinsics</a>
5744</div>
5745
5746<div class="doc_text">
5747<p> The LLVM exception handling intrinsics (which all start with
5748<tt>llvm.eh.</tt> prefix), are described in the <a
5749href="ExceptionHandling.html#format_common_intrinsics">LLVM Exception
5750Handling</a> document. </p>
5751</div>
5752
5753<!-- ======================================================================= -->
5754<div class="doc_subsection">
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5755 <a name="int_trampoline">Trampoline Intrinsic</a>
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5756</div>
5757
5758<div class="doc_text">
5759<p>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5760 This intrinsic makes it possible to excise one parameter, marked with
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5761 the <tt>nest</tt> attribute, from a function. The result is a callable
5762 function pointer lacking the nest parameter - the caller does not need
5763 to provide a value for it. Instead, the value to use is stored in
5764 advance in a "trampoline", a block of memory usually allocated
5765 on the stack, which also contains code to splice the nest value into the
5766 argument list. This is used to implement the GCC nested function address
5767 extension.
5768</p>
5769<p>
5770 For example, if the function is
5771 <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]5772 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]5773<pre>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5774 %tramp = alloca [10 x i8], align 4 ; size and alignment only correct for X86
5775 %tramp1 = getelementptr [10 x i8]* %tramp, i32 0, i32 0
5776 %p = call i8* @llvm.init.trampoline( i8* %tramp1, i8* bitcast (i32 (i8* nest , i32, i32)* @f to i8*), i8* %nval )
5777 %fp = bitcast i8* %p to i32 (i32, i32)*
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5778</pre>
Bill Wendlinge262b4c2007-09-22 09:23:55 +0000[diff] [blame]5779 <p>The call <tt>%val = call i32 %fp( i32 %x, i32 %y )</tt> is then equivalent
5780 to <tt>%val = call i32 %f( i8* %nval, i32 %x, i32 %y )</tt>.</p>
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5781</div>
5782
5783<!-- _______________________________________________________________________ -->
5784<div class="doc_subsubsection">
5785 <a name="int_it">'<tt>llvm.init.trampoline</tt>' Intrinsic</a>
5786</div>
5787<div class="doc_text">
5788<h5>Syntax:</h5>
5789<pre>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5790declare 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]5791</pre>
5792<h5>Overview:</h5>
5793<p>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5794 This fills the memory pointed to by <tt>tramp</tt> with code
5795 and returns a function pointer suitable for executing it.
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5796</p>
5797<h5>Arguments:</h5>
5798<p>
5799 The <tt>llvm.init.trampoline</tt> intrinsic takes three arguments, all
5800 pointers. The <tt>tramp</tt> argument must point to a sufficiently large
5801 and sufficiently aligned block of memory; this memory is written to by the
Duncan Sands35012212007-08-22 23:39:54 +0000[diff] [blame]5802 intrinsic. Note that the size and the alignment are target-specific - LLVM
5803 currently provides no portable way of determining them, so a front-end that
5804 generates this intrinsic needs to have some target-specific knowledge.
5805 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]5806</p>
5807<h5>Semantics:</h5>
5808<p>
5809 The block of memory pointed to by <tt>tramp</tt> is filled with target
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5810 dependent code, turning it into a function. A pointer to this function is
5811 returned, but needs to be bitcast to an
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5812 <a href="#int_trampoline">appropriate function pointer type</a>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5813 before being called. The new function's signature is the same as that of
5814 <tt>func</tt> with any arguments marked with the <tt>nest</tt> attribute
5815 removed. At most one such <tt>nest</tt> argument is allowed, and it must be
5816 of pointer type. Calling the new function is equivalent to calling
5817 <tt>func</tt> with the same argument list, but with <tt>nval</tt> used for the
5818 missing <tt>nest</tt> argument. If, after calling
5819 <tt>llvm.init.trampoline</tt>, the memory pointed to by <tt>tramp</tt> is
5820 modified, then the effect of any later call to the returned function pointer is
5821 undefined.
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5822</p>
5823</div>
5824
5825<!-- ======================================================================= -->
5826<div class="doc_subsection">
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]5827 <a name="int_atomics">Atomic Operations and Synchronization Intrinsics</a>
5828</div>
5829
5830<div class="doc_text">
5831<p>
5832 These intrinsic functions expand the "universal IR" of LLVM to represent
5833 hardware constructs for atomic operations and memory synchronization. This
5834 provides an interface to the hardware, not an interface to the programmer. It
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]5835 is aimed at a low enough level to allow any programming models or APIs
5836 (Application Programming Interfaces) which
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]5837 need atomic behaviors to map cleanly onto it. It is also modeled primarily on
5838 hardware behavior. Just as hardware provides a "universal IR" for source
5839 languages, it also provides a starting point for developing a "universal"
5840 atomic operation and synchronization IR.
5841</p>
5842<p>
5843 These do <em>not</em> form an API such as high-level threading libraries,
5844 software transaction memory systems, atomic primitives, and intrinsic
5845 functions as found in BSD, GNU libc, atomic_ops, APR, and other system and
5846 application libraries. The hardware interface provided by LLVM should allow
5847 a clean implementation of all of these APIs and parallel programming models.
5848 No one model or paradigm should be selected above others unless the hardware
5849 itself ubiquitously does so.
5850
5851</p>
5852</div>
5853
5854<!-- _______________________________________________________________________ -->
5855<div class="doc_subsubsection">
5856 <a name="int_memory_barrier">'<tt>llvm.memory.barrier</tt>' Intrinsic</a>
5857</div>
5858<div class="doc_text">
5859<h5>Syntax:</h5>
5860<pre>
5861declare void @llvm.memory.barrier( i1 &lt;ll&gt;, i1 &lt;ls&gt;, i1 &lt;sl&gt;, i1 &lt;ss&gt;,
5862i1 &lt;device&gt; )
5863
5864</pre>
5865<h5>Overview:</h5>
5866<p>
5867 The <tt>llvm.memory.barrier</tt> intrinsic guarantees ordering between
5868 specific pairs of memory access types.
5869</p>
5870<h5>Arguments:</h5>
5871<p>
5872 The <tt>llvm.memory.barrier</tt> intrinsic requires five boolean arguments.
5873 The first four arguments enables a specific barrier as listed below. The fith
5874 argument specifies that the barrier applies to io or device or uncached memory.
5875
5876</p>
5877 <ul>
5878 <li><tt>ll</tt>: load-load barrier</li>
5879 <li><tt>ls</tt>: load-store barrier</li>
5880 <li><tt>sl</tt>: store-load barrier</li>
5881 <li><tt>ss</tt>: store-store barrier</li>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5882 <li><tt>device</tt>: barrier applies to device and uncached memory also.</li>
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]5883 </ul>
5884<h5>Semantics:</h5>
5885<p>
5886 This intrinsic causes the system to enforce some ordering constraints upon
5887 the loads and stores of the program. This barrier does not indicate
5888 <em>when</em> any events will occur, it only enforces an <em>order</em> in
5889 which they occur. For any of the specified pairs of load and store operations
5890 (f.ex. load-load, or store-load), all of the first operations preceding the
5891 barrier will complete before any of the second operations succeeding the
5892 barrier begin. Specifically the semantics for each pairing is as follows:
5893</p>
5894 <ul>
5895 <li><tt>ll</tt>: All loads before the barrier must complete before any load
5896 after the barrier begins.</li>
5897
5898 <li><tt>ls</tt>: All loads before the barrier must complete before any
5899 store after the barrier begins.</li>
5900 <li><tt>ss</tt>: All stores before the barrier must complete before any
5901 store after the barrier begins.</li>
5902 <li><tt>sl</tt>: All stores before the barrier must complete before any
5903 load after the barrier begins.</li>
5904 </ul>
5905<p>
5906 These semantics are applied with a logical "and" behavior when more than one
5907 is enabled in a single memory barrier intrinsic.
5908</p>
5909<p>
5910 Backends may implement stronger barriers than those requested when they do not
5911 support as fine grained a barrier as requested. Some architectures do not
5912 need all types of barriers and on such architectures, these become noops.
5913</p>
5914<h5>Example:</h5>
5915<pre>
5916%ptr = malloc i32
5917 store i32 4, %ptr
5918
5919%result1 = load i32* %ptr <i>; yields {i32}:result1 = 4</i>
5920 call void @llvm.memory.barrier( i1 false, i1 true, i1 false, i1 false )
5921 <i>; guarantee the above finishes</i>
5922 store i32 8, %ptr <i>; before this begins</i>
5923</pre>
5924</div>
5925
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]5926<!-- _______________________________________________________________________ -->
5927<div class="doc_subsubsection">
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]5928 <a name="int_atomic_cmp_swap">'<tt>llvm.atomic.cmp.swap.*</tt>' Intrinsic</a>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]5929</div>
5930<div class="doc_text">
5931<h5>Syntax:</h5>
5932<p>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]5933 This is an overloaded intrinsic. You can use <tt>llvm.atomic.cmp.swap</tt> on
5934 any integer bit width and for different address spaces. Not all targets
5935 support all bit widths however.</p>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]5936
5937<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]5938declare i8 @llvm.atomic.cmp.swap.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;cmp&gt;, i8 &lt;val&gt; )
5939declare i16 @llvm.atomic.cmp.swap.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;cmp&gt;, i16 &lt;val&gt; )
5940declare i32 @llvm.atomic.cmp.swap.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;cmp&gt;, i32 &lt;val&gt; )
5941declare 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]5942
5943</pre>
5944<h5>Overview:</h5>
5945<p>
5946 This loads a value in memory and compares it to a given value. If they are
5947 equal, it stores a new value into the memory.
5948</p>
5949<h5>Arguments:</h5>
5950<p>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]5951 The <tt>llvm.atomic.cmp.swap</tt> intrinsic takes three arguments. The result as
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]5952 well as both <tt>cmp</tt> and <tt>val</tt> must be integer values with the
5953 same bit width. The <tt>ptr</tt> argument must be a pointer to a value of
5954 this integer type. While any bit width integer may be used, targets may only
5955 lower representations they support in hardware.
5956
5957</p>
5958<h5>Semantics:</h5>
5959<p>
5960 This entire intrinsic must be executed atomically. It first loads the value
5961 in memory pointed to by <tt>ptr</tt> and compares it with the value
5962 <tt>cmp</tt>. If they are equal, <tt>val</tt> is stored into the memory. The
5963 loaded value is yielded in all cases. This provides the equivalent of an
5964 atomic compare-and-swap operation within the SSA framework.
5965</p>
5966<h5>Examples:</h5>
5967
5968<pre>
5969%ptr = malloc i32
5970 store i32 4, %ptr
5971
5972%val1 = add i32 4, 4
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]5973%result1 = call i32 @llvm.atomic.cmp.swap.i32.p0i32( i32* %ptr, i32 4, %val1 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]5974 <i>; yields {i32}:result1 = 4</i>
5975%stored1 = icmp eq i32 %result1, 4 <i>; yields {i1}:stored1 = true</i>
5976%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 8</i>
5977
5978%val2 = add i32 1, 1
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]5979%result2 = call i32 @llvm.atomic.cmp.swap.i32.p0i32( i32* %ptr, i32 5, %val2 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]5980 <i>; yields {i32}:result2 = 8</i>
5981%stored2 = icmp eq i32 %result2, 5 <i>; yields {i1}:stored2 = false</i>
5982
5983%memval2 = load i32* %ptr <i>; yields {i32}:memval2 = 8</i>
5984</pre>
5985</div>
5986
5987<!-- _______________________________________________________________________ -->
5988<div class="doc_subsubsection">
5989 <a name="int_atomic_swap">'<tt>llvm.atomic.swap.*</tt>' Intrinsic</a>
5990</div>
5991<div class="doc_text">
5992<h5>Syntax:</h5>
5993
5994<p>
5995 This is an overloaded intrinsic. You can use <tt>llvm.atomic.swap</tt> on any
5996 integer bit width. Not all targets support all bit widths however.</p>
5997<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]5998declare i8 @llvm.atomic.swap.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;val&gt; )
5999declare i16 @llvm.atomic.swap.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;val&gt; )
6000declare i32 @llvm.atomic.swap.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;val&gt; )
6001declare i64 @llvm.atomic.swap.i64.p0i64( i64* &lt;ptr&gt;, i64 &lt;val&gt; )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6002
6003</pre>
6004<h5>Overview:</h5>
6005<p>
6006 This intrinsic loads the value stored in memory at <tt>ptr</tt> and yields
6007 the value from memory. It then stores the value in <tt>val</tt> in the memory
6008 at <tt>ptr</tt>.
6009</p>
6010<h5>Arguments:</h5>
6011
6012<p>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6013 The <tt>llvm.atomic.swap</tt> intrinsic takes two arguments. Both the
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6014 <tt>val</tt> argument and the result must be integers of the same bit width.
6015 The first argument, <tt>ptr</tt>, must be a pointer to a value of this
6016 integer type. The targets may only lower integer representations they
6017 support.
6018</p>
6019<h5>Semantics:</h5>
6020<p>
6021 This intrinsic loads the value pointed to by <tt>ptr</tt>, yields it, and
6022 stores <tt>val</tt> back into <tt>ptr</tt> atomically. This provides the
6023 equivalent of an atomic swap operation within the SSA framework.
6024
6025</p>
6026<h5>Examples:</h5>
6027<pre>
6028%ptr = malloc i32
6029 store i32 4, %ptr
6030
6031%val1 = add i32 4, 4
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6032%result1 = call i32 @llvm.atomic.swap.i32.p0i32( i32* %ptr, i32 %val1 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6033 <i>; yields {i32}:result1 = 4</i>
6034%stored1 = icmp eq i32 %result1, 4 <i>; yields {i1}:stored1 = true</i>
6035%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 8</i>
6036
6037%val2 = add i32 1, 1
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6038%result2 = call i32 @llvm.atomic.swap.i32.p0i32( i32* %ptr, i32 %val2 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6039 <i>; yields {i32}:result2 = 8</i>
6040
6041%stored2 = icmp eq i32 %result2, 8 <i>; yields {i1}:stored2 = true</i>
6042%memval2 = load i32* %ptr <i>; yields {i32}:memval2 = 2</i>
6043</pre>
6044</div>
6045
6046<!-- _______________________________________________________________________ -->
6047<div class="doc_subsubsection">
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6048 <a name="int_atomic_load_add">'<tt>llvm.atomic.load.add.*</tt>' Intrinsic</a>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6049
6050</div>
6051<div class="doc_text">
6052<h5>Syntax:</h5>
6053<p>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6054 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]6055 integer bit width. Not all targets support all bit widths however.</p>
6056<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6057declare i8 @llvm.atomic.load.add.i8..p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6058declare i16 @llvm.atomic.load.add.i16..p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6059declare i32 @llvm.atomic.load.add.i32..p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6060declare 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]6061
6062</pre>
6063<h5>Overview:</h5>
6064<p>
6065 This intrinsic adds <tt>delta</tt> to the value stored in memory at
6066 <tt>ptr</tt>. It yields the original value at <tt>ptr</tt>.
6067</p>
6068<h5>Arguments:</h5>
6069<p>
6070
6071 The intrinsic takes two arguments, the first a pointer to an integer value
6072 and the second an integer value. The result is also an integer value. These
6073 integer types can have any bit width, but they must all have the same bit
6074 width. The targets may only lower integer representations they support.
6075</p>
6076<h5>Semantics:</h5>
6077<p>
6078 This intrinsic does a series of operations atomically. It first loads the
6079 value stored at <tt>ptr</tt>. It then adds <tt>delta</tt>, stores the result
6080 to <tt>ptr</tt>. It yields the original value stored at <tt>ptr</tt>.
6081</p>
6082
6083<h5>Examples:</h5>
6084<pre>
6085%ptr = malloc i32
6086 store i32 4, %ptr
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6087%result1 = call i32 @llvm.atomic.load.add.i32.p0i32( i32* %ptr, i32 4 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6088 <i>; yields {i32}:result1 = 4</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6089%result2 = call i32 @llvm.atomic.load.add.i32.p0i32( i32* %ptr, i32 2 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6090 <i>; yields {i32}:result2 = 8</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6091%result3 = call i32 @llvm.atomic.load.add.i32.p0i32( i32* %ptr, i32 5 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6092 <i>; yields {i32}:result3 = 10</i>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6093%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 15</i>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6094</pre>
6095</div>
6096
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6097<!-- _______________________________________________________________________ -->
6098<div class="doc_subsubsection">
6099 <a name="int_atomic_load_sub">'<tt>llvm.atomic.load.sub.*</tt>' Intrinsic</a>
6100
6101</div>
6102<div class="doc_text">
6103<h5>Syntax:</h5>
6104<p>
6105 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]6106 any integer bit width and for different address spaces. Not all targets
6107 support all bit widths however.</p>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6108<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6109declare i8 @llvm.atomic.load.sub.i8.p0i32( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6110declare i16 @llvm.atomic.load.sub.i16.p0i32( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6111declare i32 @llvm.atomic.load.sub.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6112declare 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]6113
6114</pre>
6115<h5>Overview:</h5>
6116<p>
6117 This intrinsic subtracts <tt>delta</tt> to the value stored in memory at
6118 <tt>ptr</tt>. It yields the original value at <tt>ptr</tt>.
6119</p>
6120<h5>Arguments:</h5>
6121<p>
6122
6123 The intrinsic takes two arguments, the first a pointer to an integer value
6124 and the second an integer value. The result is also an integer value. These
6125 integer types can have any bit width, but they must all have the same bit
6126 width. The targets may only lower integer representations they support.
6127</p>
6128<h5>Semantics:</h5>
6129<p>
6130 This intrinsic does a series of operations atomically. It first loads the
6131 value stored at <tt>ptr</tt>. It then subtracts <tt>delta</tt>, stores the
6132 result to <tt>ptr</tt>. It yields the original value stored at <tt>ptr</tt>.
6133</p>
6134
6135<h5>Examples:</h5>
6136<pre>
6137%ptr = malloc i32
6138 store i32 8, %ptr
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6139%result1 = call i32 @llvm.atomic.load.sub.i32.p0i32( i32* %ptr, i32 4 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6140 <i>; yields {i32}:result1 = 8</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6141%result2 = call i32 @llvm.atomic.load.sub.i32.p0i32( i32* %ptr, i32 2 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6142 <i>; yields {i32}:result2 = 4</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6143%result3 = call i32 @llvm.atomic.load.sub.i32.p0i32( i32* %ptr, i32 5 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6144 <i>; yields {i32}:result3 = 2</i>
6145%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = -3</i>
6146</pre>
6147</div>
6148
6149<!-- _______________________________________________________________________ -->
6150<div class="doc_subsubsection">
6151 <a name="int_atomic_load_and">'<tt>llvm.atomic.load.and.*</tt>' Intrinsic</a><br>
6152 <a name="int_atomic_load_nand">'<tt>llvm.atomic.load.nand.*</tt>' Intrinsic</a><br>
6153 <a name="int_atomic_load_or">'<tt>llvm.atomic.load.or.*</tt>' Intrinsic</a><br>
6154 <a name="int_atomic_load_xor">'<tt>llvm.atomic.load.xor.*</tt>' Intrinsic</a><br>
6155
6156</div>
6157<div class="doc_text">
6158<h5>Syntax:</h5>
6159<p>
6160 These are overloaded intrinsics. You can use <tt>llvm.atomic.load_and</tt>,
6161 <tt>llvm.atomic.load_nand</tt>, <tt>llvm.atomic.load_or</tt>, and
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6162 <tt>llvm.atomic.load_xor</tt> on any integer bit width and for different
6163 address spaces. Not all targets support all bit widths however.</p>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6164<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6165declare i8 @llvm.atomic.load.and.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6166declare i16 @llvm.atomic.load.and.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6167declare i32 @llvm.atomic.load.and.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6168declare 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]6169
6170</pre>
6171
6172<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6173declare i8 @llvm.atomic.load.or.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6174declare i16 @llvm.atomic.load.or.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6175declare i32 @llvm.atomic.load.or.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6176declare 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]6177
6178</pre>
6179
6180<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6181declare i8 @llvm.atomic.load.nand.i8.p0i32( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6182declare i16 @llvm.atomic.load.nand.i16.p0i32( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6183declare i32 @llvm.atomic.load.nand.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6184declare 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]6185
6186</pre>
6187
6188<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6189declare i8 @llvm.atomic.load.xor.i8.p0i32( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6190declare i16 @llvm.atomic.load.xor.i16.p0i32( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6191declare i32 @llvm.atomic.load.xor.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6192declare 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]6193
6194</pre>
6195<h5>Overview:</h5>
6196<p>
6197 These intrinsics bitwise the operation (and, nand, or, xor) <tt>delta</tt> to
6198 the value stored in memory at <tt>ptr</tt>. It yields the original value
6199 at <tt>ptr</tt>.
6200</p>
6201<h5>Arguments:</h5>
6202<p>
6203
6204 These intrinsics take two arguments, the first a pointer to an integer value
6205 and the second an integer value. The result is also an integer value. These
6206 integer types can have any bit width, but they must all have the same bit
6207 width. The targets may only lower integer representations they support.
6208</p>
6209<h5>Semantics:</h5>
6210<p>
6211 These intrinsics does a series of operations atomically. They first load the
6212 value stored at <tt>ptr</tt>. They then do the bitwise operation
6213 <tt>delta</tt>, store the result to <tt>ptr</tt>. They yield the original
6214 value stored at <tt>ptr</tt>.
6215</p>
6216
6217<h5>Examples:</h5>
6218<pre>
6219%ptr = malloc i32
6220 store i32 0x0F0F, %ptr
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6221%result0 = call i32 @llvm.atomic.load.nand.i32.p0i32( i32* %ptr, i32 0xFF )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6222 <i>; yields {i32}:result0 = 0x0F0F</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6223%result1 = call i32 @llvm.atomic.load.and.i32.p0i32( i32* %ptr, i32 0xFF )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6224 <i>; yields {i32}:result1 = 0xFFFFFFF0</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6225%result2 = call i32 @llvm.atomic.load.or.i32.p0i32( i32* %ptr, i32 0F )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6226 <i>; yields {i32}:result2 = 0xF0</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6227%result3 = call i32 @llvm.atomic.load.xor.i32.p0i32( i32* %ptr, i32 0F )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6228 <i>; yields {i32}:result3 = FF</i>
6229%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = F0</i>
6230</pre>
6231</div>
6232
6233
6234<!-- _______________________________________________________________________ -->
6235<div class="doc_subsubsection">
6236 <a name="int_atomic_load_max">'<tt>llvm.atomic.load.max.*</tt>' Intrinsic</a><br>
6237 <a name="int_atomic_load_min">'<tt>llvm.atomic.load.min.*</tt>' Intrinsic</a><br>
6238 <a name="int_atomic_load_umax">'<tt>llvm.atomic.load.umax.*</tt>' Intrinsic</a><br>
6239 <a name="int_atomic_load_umin">'<tt>llvm.atomic.load.umin.*</tt>' Intrinsic</a><br>
6240
6241</div>
6242<div class="doc_text">
6243<h5>Syntax:</h5>
6244<p>
6245 These are overloaded intrinsics. You can use <tt>llvm.atomic.load_max</tt>,
6246 <tt>llvm.atomic.load_min</tt>, <tt>llvm.atomic.load_umax</tt>, and
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6247 <tt>llvm.atomic.load_umin</tt> on any integer bit width and for different
6248 address spaces. Not all targets
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6249 support all bit widths however.</p>
6250<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6251declare i8 @llvm.atomic.load.max.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6252declare i16 @llvm.atomic.load.max.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6253declare i32 @llvm.atomic.load.max.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6254declare 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]6255
6256</pre>
6257
6258<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6259declare i8 @llvm.atomic.load.min.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6260declare i16 @llvm.atomic.load.min.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6261declare i32 @llvm.atomic.load.min.i32..p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6262declare 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]6263
6264</pre>
6265
6266<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6267declare i8 @llvm.atomic.load.umax.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6268declare i16 @llvm.atomic.load.umax.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6269declare i32 @llvm.atomic.load.umax.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6270declare 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]6271
6272</pre>
6273
6274<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6275declare i8 @llvm.atomic.load.umin.i8..p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6276declare i16 @llvm.atomic.load.umin.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6277declare i32 @llvm.atomic.load.umin.i32..p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6278declare 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]6279
6280</pre>
6281<h5>Overview:</h5>
6282<p>
6283 These intrinsics takes the signed or unsigned minimum or maximum of
6284 <tt>delta</tt> and the value stored in memory at <tt>ptr</tt>. It yields the
6285 original value at <tt>ptr</tt>.
6286</p>
6287<h5>Arguments:</h5>
6288<p>
6289
6290 These intrinsics take two arguments, the first a pointer to an integer value
6291 and the second an integer value. The result is also an integer value. These
6292 integer types can have any bit width, but they must all have the same bit
6293 width. The targets may only lower integer representations they support.
6294</p>
6295<h5>Semantics:</h5>
6296<p>
6297 These intrinsics does a series of operations atomically. They first load the
6298 value stored at <tt>ptr</tt>. They then do the signed or unsigned min or max
6299 <tt>delta</tt> and the value, store the result to <tt>ptr</tt>. They yield
6300 the original value stored at <tt>ptr</tt>.
6301</p>
6302
6303<h5>Examples:</h5>
6304<pre>
6305%ptr = malloc i32
6306 store i32 7, %ptr
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6307%result0 = call i32 @llvm.atomic.load.min.i32.p0i32( i32* %ptr, i32 -2 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6308 <i>; yields {i32}:result0 = 7</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6309%result1 = call i32 @llvm.atomic.load.max.i32.p0i32( i32* %ptr, i32 8 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6310 <i>; yields {i32}:result1 = -2</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6311%result2 = call i32 @llvm.atomic.load.umin.i32.p0i32( i32* %ptr, i32 10 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6312 <i>; yields {i32}:result2 = 8</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6313%result3 = call i32 @llvm.atomic.load.umax.i32.p0i32( i32* %ptr, i32 30 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6314 <i>; yields {i32}:result3 = 8</i>
6315%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 30</i>
6316</pre>
6317</div>
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]6318
6319<!-- ======================================================================= -->
6320<div class="doc_subsection">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6321 <a name="int_general">General Intrinsics</a>
6322</div>
6323
6324<div class="doc_text">
6325<p> This class of intrinsics is designed to be generic and has
6326no specific purpose. </p>
6327</div>
6328
6329<!-- _______________________________________________________________________ -->
6330<div class="doc_subsubsection">
6331 <a name="int_var_annotation">'<tt>llvm.var.annotation</tt>' Intrinsic</a>
6332</div>
6333
6334<div class="doc_text">
6335
6336<h5>Syntax:</h5>
6337<pre>
6338 declare void @llvm.var.annotation(i8* &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
6339</pre>
6340
6341<h5>Overview:</h5>
6342
6343<p>
6344The '<tt>llvm.var.annotation</tt>' intrinsic
6345</p>
6346
6347<h5>Arguments:</h5>
6348
6349<p>
6350The first argument is a pointer to a value, the second is a pointer to a
6351global string, the third is a pointer to a global string which is the source
6352file name, and the last argument is the line number.
6353</p>
6354
6355<h5>Semantics:</h5>
6356
6357<p>
Anton Korobeynikove6e764f2008-01-15 22:31:34 +0000[diff] [blame]6358This intrinsic allows annotation of local variables with arbitrary strings.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6359This can be useful for special purpose optimizations that want to look for these
Anton Korobeynikove6e764f2008-01-15 22:31:34 +0000[diff] [blame]6360annotations. These have no other defined use, they are ignored by code
6361generation and optimization.
6362</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6363</div>
6364
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]6365<!-- _______________________________________________________________________ -->
6366<div class="doc_subsubsection">
Tanya Lattnerc9869b12007-09-21 23:57:59 +0000[diff] [blame]6367 <a name="int_annotation">'<tt>llvm.annotation.*</tt>' Intrinsic</a>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]6368</div>
6369
6370<div class="doc_text">
6371
6372<h5>Syntax:</h5>
Tanya Lattnere545be72007-09-21 23:56:27 +0000[diff] [blame]6373<p>This is an overloaded intrinsic. You can use '<tt>llvm.annotation</tt>' on
6374any integer bit width.
6375</p>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]6376<pre>
Tanya Lattner09161fe2007-09-22 00:03:01 +0000[diff] [blame]6377 declare i8 @llvm.annotation.i8(i8 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
6378 declare i16 @llvm.annotation.i16(i16 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
6379 declare i32 @llvm.annotation.i32(i32 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
6380 declare i64 @llvm.annotation.i64(i64 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
6381 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]6382</pre>
6383
6384<h5>Overview:</h5>
Tanya Lattnere545be72007-09-21 23:56:27 +0000[diff] [blame]6385
6386<p>
6387The '<tt>llvm.annotation</tt>' intrinsic.
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]6388</p>
6389
6390<h5>Arguments:</h5>
6391
6392<p>
6393The first argument is an integer value (result of some expression),
6394the second is a pointer to a global string, the third is a pointer to a global
6395string which is the source file name, and the last argument is the line number.
Tanya Lattnere545be72007-09-21 23:56:27 +0000[diff] [blame]6396It returns the value of the first argument.
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]6397</p>
6398
6399<h5>Semantics:</h5>
6400
6401<p>
6402This intrinsic allows annotations to be put on arbitrary expressions
6403with arbitrary strings. This can be useful for special purpose optimizations
6404that want to look for these annotations. These have no other defined use, they
6405are ignored by code generation and optimization.
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]6406</p>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]6407</div>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6408
Anton Korobeynikove6e764f2008-01-15 22:31:34 +0000[diff] [blame]6409<!-- _______________________________________________________________________ -->
6410<div class="doc_subsubsection">
6411 <a name="int_trap">'<tt>llvm.trap</tt>' Intrinsic</a>
6412</div>
6413
6414<div class="doc_text">
6415
6416<h5>Syntax:</h5>
6417<pre>
6418 declare void @llvm.trap()
6419</pre>
6420
6421<h5>Overview:</h5>
6422
6423<p>
6424The '<tt>llvm.trap</tt>' intrinsic
6425</p>
6426
6427<h5>Arguments:</h5>
6428
6429<p>
6430None
6431</p>
6432
6433<h5>Semantics:</h5>
6434
6435<p>
6436This intrinsics is lowered to the target dependent trap instruction. If the
6437target does not have a trap instruction, this intrinsic will be lowered to the
6438call of the abort() function.
6439</p>
6440</div>
6441
Bill Wendlinge4164592008-11-19 05:56:17 +0000[diff] [blame]6442<!-- _______________________________________________________________________ -->
6443<div class="doc_subsubsection">
Misha Brukman5dd7f4d2008-11-22 23:55:29 +0000[diff] [blame^]6444 <a name="int_stackprotector">'<tt>llvm.stackprotector</tt>' Intrinsic</a>
Bill Wendlinge4164592008-11-19 05:56:17 +0000[diff] [blame]6445</div>
6446<div class="doc_text">
6447<h5>Syntax:</h5>
6448<pre>
6449declare void @llvm.stackprotector( i8* &lt;guard&gt;, i8** &lt;slot&gt; )
6450
6451</pre>
6452<h5>Overview:</h5>
6453<p>
6454 The <tt>llvm.stackprotector</tt> intrinsic takes the <tt>guard</tt> and stores
6455 it onto the stack at <tt>slot</tt>. The stack slot is adjusted to ensure that
6456 it is placed on the stack before local variables.
6457</p>
6458<h5>Arguments:</h5>
6459<p>
6460 The <tt>llvm.stackprotector</tt> intrinsic requires two pointer arguments. The
6461 first argument is the value loaded from the stack guard
6462 <tt>@__stack_chk_guard</tt>. The second variable is an <tt>alloca</tt> that
6463 has enough space to hold the value of the guard.
6464</p>
6465<h5>Semantics:</h5>
6466<p>
6467 This intrinsic causes the prologue/epilogue inserter to force the position of
6468 the <tt>AllocaInst</tt> stack slot to be before local variables on the
6469 stack. This is to ensure that if a local variable on the stack is overwritten,
6470 it will destroy the value of the guard. When the function exits, the guard on
6471 the stack is checked against the original guard. If they're different, then
6472 the program aborts by calling the <tt>__stack_chk_fail()</tt> function.
6473</p>
6474</div>
6475
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6476<!-- *********************************************************************** -->
6477<hr>
6478<address>
6479 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
6480 src="http://jigsaw.w3.org/css-validator/images/vcss" alt="Valid CSS!"></a>
6481 <a href="http://validator.w3.org/check/referer"><img
Chris Lattner08497ce2008-01-04 04:33:49 +0000[diff] [blame]6482 src="http://www.w3.org/Icons/valid-html401" alt="Valid HTML 4.01!"></a>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6483
6484 <a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
6485 <a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br>
6486 Last modified: $Date$
6487</address>
Chris Lattner08497ce2008-01-04 04:33:49 +0000[diff] [blame]6488
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6489</body>
6490</html>