Gitiles
Code Review Sign In
gerrit-public.fairphone.software / fp2-dev / platform / external / llvm / 032ba85af6400782c9169db9bade59e337f3026e / . / docs / LangRef.html
blob: e578a0d8402f6d1d03a00e4af23a081ba6bf5d4d [file] [log] [blame]
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN"
2 "http://www.w3.org/TR/html4/strict.dtd">
3<html>
4<head>
5 <title>LLVM Assembly Language Reference Manual</title>
6 <meta http-equiv="Content-Type" content="text/html; charset=utf-8">
7 <meta name="author" content="Chris Lattner">
8 <meta name="description"
9 content="LLVM Assembly Language Reference Manual.">
10 <link rel="stylesheet" href="llvm.css" type="text/css">
11</head>
12
13<body>
14
15<div class="doc_title"> LLVM Language Reference Manual </div>
16<ol>
17 <li><a href="#abstract">Abstract</a></li>
18 <li><a href="#introduction">Introduction</a></li>
19 <li><a href="#identifiers">Identifiers</a></li>
20 <li><a href="#highlevel">High Level Structure</a>
21 <ol>
22 <li><a href="#modulestructure">Module Structure</a></li>
23 <li><a href="#linkage">Linkage Types</a></li>
24 <li><a href="#callingconv">Calling Conventions</a></li>
25 <li><a href="#globalvars">Global Variables</a></li>
26 <li><a href="#functionstructure">Functions</a></li>
27 <li><a href="#aliasstructure">Aliases</a>
28 <li><a href="#paramattrs">Parameter Attributes</a></li>
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>
61 <li><a href="#simpleconstants">Simple Constants</a>
62 <li><a href="#aggregateconstants">Aggregate Constants</a>
63 <li><a href="#globalconstants">Global Variable and Function Addresses</a>
64 <li><a href="#undefvalues">Undefined Values</a>
65 <li><a href="#constantexprs">Constant Expressions</a>
66 </ol>
67 </li>
68 <li><a href="#othervalues">Other Values</a>
69 <ol>
70 <li><a href="#inlineasm">Inline Assembler Expressions</a>
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>
146 <li><a href="#otherops">Other Operations</a>
147 <ol>
148 <li><a href="#i_icmp">'<tt>icmp</tt>' Instruction</a></li>
149 <li><a href="#i_fcmp">'<tt>fcmp</tt>' Instruction</a></li>
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]150 <li><a href="#i_vicmp">'<tt>vicmp</tt>' Instruction</a></li>
151 <li><a href="#i_vfcmp">'<tt>vfcmp</tt>' Instruction</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]152 <li><a href="#i_phi">'<tt>phi</tt>' Instruction</a></li>
153 <li><a href="#i_select">'<tt>select</tt>' Instruction</a></li>
154 <li><a href="#i_call">'<tt>call</tt>' Instruction</a></li>
155 <li><a href="#i_va_arg">'<tt>va_arg</tt>' Instruction</a></li>
156 </ol>
157 </li>
158 </ol>
159 </li>
160 <li><a href="#intrinsics">Intrinsic Functions</a>
161 <ol>
162 <li><a href="#int_varargs">Variable Argument Handling Intrinsics</a>
163 <ol>
164 <li><a href="#int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a></li>
165 <li><a href="#int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a></li>
166 <li><a href="#int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a></li>
167 </ol>
168 </li>
169 <li><a href="#int_gc">Accurate Garbage Collection Intrinsics</a>
170 <ol>
171 <li><a href="#int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a></li>
172 <li><a href="#int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a></li>
173 <li><a href="#int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a></li>
174 </ol>
175 </li>
176 <li><a href="#int_codegen">Code Generator Intrinsics</a>
177 <ol>
178 <li><a href="#int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a></li>
179 <li><a href="#int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a></li>
180 <li><a href="#int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a></li>
181 <li><a href="#int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a></li>
182 <li><a href="#int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a></li>
183 <li><a href="#int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a></li>
184 <li><a href="#int_readcyclecounter"><tt>llvm.readcyclecounter</tt>' Intrinsic</a></li>
185 </ol>
186 </li>
187 <li><a href="#int_libc">Standard C Library Intrinsics</a>
188 <ol>
189 <li><a href="#int_memcpy">'<tt>llvm.memcpy.*</tt>' Intrinsic</a></li>
190 <li><a href="#int_memmove">'<tt>llvm.memmove.*</tt>' Intrinsic</a></li>
191 <li><a href="#int_memset">'<tt>llvm.memset.*</tt>' Intrinsic</a></li>
192 <li><a href="#int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a></li>
193 <li><a href="#int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a></li>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]194 <li><a href="#int_sin">'<tt>llvm.sin.*</tt>' Intrinsic</a></li>
195 <li><a href="#int_cos">'<tt>llvm.cos.*</tt>' Intrinsic</a></li>
196 <li><a href="#int_pow">'<tt>llvm.pow.*</tt>' Intrinsic</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]197 </ol>
198 </li>
199 <li><a href="#int_manip">Bit Manipulation Intrinsics</a>
200 <ol>
201 <li><a href="#int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a></li>
202 <li><a href="#int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic </a></li>
203 <li><a href="#int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic </a></li>
204 <li><a href="#int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic </a></li>
205 <li><a href="#int_part_select">'<tt>llvm.part.select.*</tt>' Intrinsic </a></li>
206 <li><a href="#int_part_set">'<tt>llvm.part.set.*</tt>' Intrinsic </a></li>
207 </ol>
208 </li>
209 <li><a href="#int_debugger">Debugger intrinsics</a></li>
210 <li><a href="#int_eh">Exception Handling intrinsics</a></li>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]211 <li><a href="#int_trampoline">Trampoline Intrinsic</a>
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]212 <ol>
213 <li><a href="#int_it">'<tt>llvm.init.trampoline</tt>' Intrinsic</a></li>
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]214 </ol>
215 </li>
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]216 <li><a href="#int_atomics">Atomic intrinsics</a>
217 <ol>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]218 <li><a href="#int_memory_barrier"><tt>llvm.memory_barrier</tt></a></li>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]219 <li><a href="#int_atomic_cmp_swap"><tt>llvm.atomic.cmp.swap</tt></a></li>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]220 <li><a href="#int_atomic_swap"><tt>llvm.atomic.swap</tt></a></li>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]221 <li><a href="#int_atomic_load_add"><tt>llvm.atomic.load.add</tt></a></li>
222 <li><a href="#int_atomic_load_sub"><tt>llvm.atomic.load.sub</tt></a></li>
223 <li><a href="#int_atomic_load_and"><tt>llvm.atomic.load.and</tt></a></li>
224 <li><a href="#int_atomic_load_nand"><tt>llvm.atomic.load.nand</tt></a></li>
225 <li><a href="#int_atomic_load_or"><tt>llvm.atomic.load.or</tt></a></li>
226 <li><a href="#int_atomic_load_xor"><tt>llvm.atomic.load.xor</tt></a></li>
227 <li><a href="#int_atomic_load_max"><tt>llvm.atomic.load.max</tt></a></li>
228 <li><a href="#int_atomic_load_min"><tt>llvm.atomic.load.min</tt></a></li>
229 <li><a href="#int_atomic_load_umax"><tt>llvm.atomic.load.umax</tt></a></li>
230 <li><a href="#int_atomic_load_umin"><tt>llvm.atomic.load.umin</tt></a></li>
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]231 </ol>
232 </li>
Reid Spencerb043f672007-07-20 19:59:11 +0000[diff] [blame]233 <li><a href="#int_general">General intrinsics</a>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]234 <ol>
Reid Spencerb043f672007-07-20 19:59:11 +0000[diff] [blame]235 <li><a href="#int_var_annotation">
Tanya Lattner51369f32007-09-22 00:01:26 +0000[diff] [blame]236 <tt>llvm.var.annotation</tt>' Intrinsic</a></li>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]237 <li><a href="#int_annotation">
Tanya Lattner51369f32007-09-22 00:01:26 +0000[diff] [blame]238 <tt>llvm.annotation.*</tt>' Intrinsic</a></li>
Anton Korobeynikove6e764f2008-01-15 22:31:34 +0000[diff] [blame]239 <li><a href="#int_trap">
240 <tt>llvm.trap</tt>' Intrinsic</a></li>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]241 </ol>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]242 </li>
243 </ol>
244 </li>
245</ol>
246
247<div class="doc_author">
248 <p>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a>
249 and <a href="mailto:vadve@cs.uiuc.edu">Vikram Adve</a></p>
250</div>
251
252<!-- *********************************************************************** -->
253<div class="doc_section"> <a name="abstract">Abstract </a></div>
254<!-- *********************************************************************** -->
255
256<div class="doc_text">
257<p>This document is a reference manual for the LLVM assembly language.
Bill Wendlinge7846a52008-08-05 22:29:16 +0000[diff] [blame]258LLVM is a Static Single Assignment (SSA) based representation that provides
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]259type safety, low-level operations, flexibility, and the capability of
260representing 'all' high-level languages cleanly. It is the common code
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]261representation used throughout all phases of the LLVM compilation
262strategy.</p>
263</div>
264
265<!-- *********************************************************************** -->
266<div class="doc_section"> <a name="introduction">Introduction</a> </div>
267<!-- *********************************************************************** -->
268
269<div class="doc_text">
270
271<p>The LLVM code representation is designed to be used in three
272different forms: as an in-memory compiler IR, as an on-disk bitcode
273representation (suitable for fast loading by a Just-In-Time compiler),
274and as a human readable assembly language representation. This allows
275LLVM to provide a powerful intermediate representation for efficient
276compiler transformations and analysis, while providing a natural means
277to debug and visualize the transformations. The three different forms
278of LLVM are all equivalent. This document describes the human readable
279representation and notation.</p>
280
281<p>The LLVM representation aims to be light-weight and low-level
282while being expressive, typed, and extensible at the same time. It
283aims to be a "universal IR" of sorts, by being at a low enough level
284that high-level ideas may be cleanly mapped to it (similar to how
285microprocessors are "universal IR's", allowing many source languages to
286be mapped to them). By providing type information, LLVM can be used as
287the target of optimizations: for example, through pointer analysis, it
288can be proven that a C automatic variable is never accessed outside of
289the current function... allowing it to be promoted to a simple SSA
290value instead of a memory location.</p>
291
292</div>
293
294<!-- _______________________________________________________________________ -->
295<div class="doc_subsubsection"> <a name="wellformed">Well-Formedness</a> </div>
296
297<div class="doc_text">
298
299<p>It is important to note that this document describes 'well formed'
300LLVM assembly language. There is a difference between what the parser
301accepts and what is considered 'well formed'. For example, the
302following instruction is syntactically okay, but not well formed:</p>
303
304<div class="doc_code">
305<pre>
306%x = <a href="#i_add">add</a> i32 1, %x
307</pre>
308</div>
309
310<p>...because the definition of <tt>%x</tt> does not dominate all of
311its uses. The LLVM infrastructure provides a verification pass that may
312be used to verify that an LLVM module is well formed. This pass is
313automatically run by the parser after parsing input assembly and by
314the optimizer before it outputs bitcode. The violations pointed out
315by the verifier pass indicate bugs in transformation passes or input to
316the parser.</p>
317</div>
318
Chris Lattnera83fdc02007-10-03 17:34:29 +0000[diff] [blame]319<!-- Describe the typesetting conventions here. -->
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]320
321<!-- *********************************************************************** -->
322<div class="doc_section"> <a name="identifiers">Identifiers</a> </div>
323<!-- *********************************************************************** -->
324
325<div class="doc_text">
326
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]327 <p>LLVM identifiers come in two basic types: global and local. Global
328 identifiers (functions, global variables) begin with the @ character. Local
329 identifiers (register names, types) begin with the % character. Additionally,
330 there are three different formats for identifiers, for different purposes:
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]331
332<ol>
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]333 <li>Named values are represented as a string of characters with their prefix.
334 For example, %foo, @DivisionByZero, %a.really.long.identifier. The actual
335 regular expression used is '<tt>[%@][a-zA-Z$._][a-zA-Z$._0-9]*</tt>'.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]336 Identifiers which require other characters in their names can be surrounded
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]337 with quotes. In this way, anything except a <tt>&quot;</tt> character can
338 be used in a named value.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]339
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]340 <li>Unnamed values are represented as an unsigned numeric value with their
341 prefix. For example, %12, @2, %44.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]342
343 <li>Constants, which are described in a <a href="#constants">section about
344 constants</a>, below.</li>
345</ol>
346
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]347<p>LLVM requires that values start with a prefix for two reasons: Compilers
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]348don't need to worry about name clashes with reserved words, and the set of
349reserved words may be expanded in the future without penalty. Additionally,
350unnamed identifiers allow a compiler to quickly come up with a temporary
351variable without having to avoid symbol table conflicts.</p>
352
353<p>Reserved words in LLVM are very similar to reserved words in other
354languages. There are keywords for different opcodes
355('<tt><a href="#i_add">add</a></tt>',
356 '<tt><a href="#i_bitcast">bitcast</a></tt>',
357 '<tt><a href="#i_ret">ret</a></tt>', etc...), for primitive type names ('<tt><a
358href="#t_void">void</a></tt>', '<tt><a href="#t_primitive">i32</a></tt>', etc...),
359and others. These reserved words cannot conflict with variable names, because
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]360none of them start with a prefix character ('%' or '@').</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]361
362<p>Here is an example of LLVM code to multiply the integer variable
363'<tt>%X</tt>' by 8:</p>
364
365<p>The easy way:</p>
366
367<div class="doc_code">
368<pre>
369%result = <a href="#i_mul">mul</a> i32 %X, 8
370</pre>
371</div>
372
373<p>After strength reduction:</p>
374
375<div class="doc_code">
376<pre>
377%result = <a href="#i_shl">shl</a> i32 %X, i8 3
378</pre>
379</div>
380
381<p>And the hard way:</p>
382
383<div class="doc_code">
384<pre>
385<a href="#i_add">add</a> i32 %X, %X <i>; yields {i32}:%0</i>
386<a href="#i_add">add</a> i32 %0, %0 <i>; yields {i32}:%1</i>
387%result = <a href="#i_add">add</a> i32 %1, %1
388</pre>
389</div>
390
391<p>This last way of multiplying <tt>%X</tt> by 8 illustrates several
392important lexical features of LLVM:</p>
393
394<ol>
395
396 <li>Comments are delimited with a '<tt>;</tt>' and go until the end of
397 line.</li>
398
399 <li>Unnamed temporaries are created when the result of a computation is not
400 assigned to a named value.</li>
401
402 <li>Unnamed temporaries are numbered sequentially</li>
403
404</ol>
405
406<p>...and it also shows a convention that we follow in this document. When
407demonstrating instructions, we will follow an instruction with a comment that
408defines the type and name of value produced. Comments are shown in italic
409text.</p>
410
411</div>
412
413<!-- *********************************************************************** -->
414<div class="doc_section"> <a name="highlevel">High Level Structure</a> </div>
415<!-- *********************************************************************** -->
416
417<!-- ======================================================================= -->
418<div class="doc_subsection"> <a name="modulestructure">Module Structure</a>
419</div>
420
421<div class="doc_text">
422
423<p>LLVM programs are composed of "Module"s, each of which is a
424translation unit of the input programs. Each module consists of
425functions, global variables, and symbol table entries. Modules may be
426combined together with the LLVM linker, which merges function (and
427global variable) definitions, resolves forward declarations, and merges
428symbol table entries. Here is an example of the "hello world" module:</p>
429
430<div class="doc_code">
431<pre><i>; Declare the string constant as a global constant...</i>
432<a href="#identifiers">@.LC0</a> = <a href="#linkage_internal">internal</a> <a
433 href="#globalvars">constant</a> <a href="#t_array">[13 x i8]</a> c"hello world\0A\00" <i>; [13 x i8]*</i>
434
435<i>; External declaration of the puts function</i>
436<a href="#functionstructure">declare</a> i32 @puts(i8 *) <i>; i32(i8 *)* </i>
437
438<i>; Definition of main function</i>
439define i32 @main() { <i>; i32()* </i>
440 <i>; Convert [13x i8 ]* to i8 *...</i>
441 %cast210 = <a
442 href="#i_getelementptr">getelementptr</a> [13 x i8 ]* @.LC0, i64 0, i64 0 <i>; i8 *</i>
443
444 <i>; Call puts function to write out the string to stdout...</i>
445 <a
446 href="#i_call">call</a> i32 @puts(i8 * %cast210) <i>; i32</i>
447 <a
448 href="#i_ret">ret</a> i32 0<br>}<br>
449</pre>
450</div>
451
452<p>This example is made up of a <a href="#globalvars">global variable</a>
453named "<tt>.LC0</tt>", an external declaration of the "<tt>puts</tt>"
454function, and a <a href="#functionstructure">function definition</a>
455for "<tt>main</tt>".</p>
456
457<p>In general, a module is made up of a list of global values,
458where both functions and global variables are global values. Global values are
459represented by a pointer to a memory location (in this case, a pointer to an
460array of char, and a pointer to a function), and have one of the following <a
461href="#linkage">linkage types</a>.</p>
462
463</div>
464
465<!-- ======================================================================= -->
466<div class="doc_subsection">
467 <a name="linkage">Linkage Types</a>
468</div>
469
470<div class="doc_text">
471
472<p>
473All Global Variables and Functions have one of the following types of linkage:
474</p>
475
476<dl>
477
Dale Johannesen96e7e092008-05-23 23:13:41 +0000[diff] [blame]478 <dt><tt><b><a name="linkage_internal">internal</a></b></tt>: </dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]479
480 <dd>Global values with internal linkage are only directly accessible by
481 objects in the current module. In particular, linking code into a module with
482 an internal global value may cause the internal to be renamed as necessary to
483 avoid collisions. Because the symbol is internal to the module, all
484 references can be updated. This corresponds to the notion of the
485 '<tt>static</tt>' keyword in C.
486 </dd>
487
488 <dt><tt><b><a name="linkage_linkonce">linkonce</a></b></tt>: </dt>
489
490 <dd>Globals with "<tt>linkonce</tt>" linkage are merged with other globals of
491 the same name when linkage occurs. This is typically used to implement
492 inline functions, templates, or other code which must be generated in each
493 translation unit that uses it. Unreferenced <tt>linkonce</tt> globals are
494 allowed to be discarded.
495 </dd>
496
Dale Johannesen96e7e092008-05-23 23:13:41 +0000[diff] [blame]497 <dt><tt><b><a name="linkage_common">common</a></b></tt>: </dt>
498
499 <dd>"<tt>common</tt>" linkage is exactly the same as <tt>linkonce</tt>
500 linkage, except that unreferenced <tt>common</tt> globals may not be
501 discarded. This is used for globals that may be emitted in multiple
502 translation units, but that are not guaranteed to be emitted into every
503 translation unit that uses them. One example of this is tentative
504 definitions in C, such as "<tt>int X;</tt>" at global scope.
505 </dd>
506
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]507 <dt><tt><b><a name="linkage_weak">weak</a></b></tt>: </dt>
508
Dale Johannesen96e7e092008-05-23 23:13:41 +0000[diff] [blame]509 <dd>"<tt>weak</tt>" linkage is the same as <tt>common</tt> linkage, except
510 that some targets may choose to emit different assembly sequences for them
511 for target-dependent reasons. This is used for globals that are declared
512 "weak" in C source code.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]513 </dd>
514
515 <dt><tt><b><a name="linkage_appending">appending</a></b></tt>: </dt>
516
517 <dd>"<tt>appending</tt>" linkage may only be applied to global variables of
518 pointer to array type. When two global variables with appending linkage are
519 linked together, the two global arrays are appended together. This is the
520 LLVM, typesafe, equivalent of having the system linker append together
521 "sections" with identical names when .o files are linked.
522 </dd>
523
524 <dt><tt><b><a name="linkage_externweak">extern_weak</a></b></tt>: </dt>
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]525 <dd>The semantics of this linkage follow the ELF object file model: the
526 symbol is weak until linked, if not linked, the symbol becomes null instead
527 of being an undefined reference.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]528 </dd>
529
530 <dt><tt><b><a name="linkage_external">externally visible</a></b></tt>:</dt>
531
532 <dd>If none of the above identifiers are used, the global is externally
533 visible, meaning that it participates in linkage and can be used to resolve
534 external symbol references.
535 </dd>
536</dl>
537
538 <p>
539 The next two types of linkage are targeted for Microsoft Windows platform
540 only. They are designed to support importing (exporting) symbols from (to)
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]541 DLLs (Dynamic Link Libraries).
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]542 </p>
543
544 <dl>
545 <dt><tt><b><a name="linkage_dllimport">dllimport</a></b></tt>: </dt>
546
547 <dd>"<tt>dllimport</tt>" linkage causes the compiler to reference a function
548 or variable via a global pointer to a pointer that is set up by the DLL
549 exporting the symbol. On Microsoft Windows targets, the pointer name is
550 formed by combining <code>_imp__</code> and the function or variable name.
551 </dd>
552
553 <dt><tt><b><a name="linkage_dllexport">dllexport</a></b></tt>: </dt>
554
555 <dd>"<tt>dllexport</tt>" linkage causes the compiler to provide a global
556 pointer to a pointer in a DLL, so that it can be referenced with the
557 <tt>dllimport</tt> attribute. On Microsoft Windows targets, the pointer
558 name is formed by combining <code>_imp__</code> and the function or variable
559 name.
560 </dd>
561
562</dl>
563
564<p><a name="linkage_external"></a>For example, since the "<tt>.LC0</tt>"
565variable is defined to be internal, if another module defined a "<tt>.LC0</tt>"
566variable and was linked with this one, one of the two would be renamed,
567preventing a collision. Since "<tt>main</tt>" and "<tt>puts</tt>" are
568external (i.e., lacking any linkage declarations), they are accessible
569outside of the current module.</p>
570<p>It is illegal for a function <i>declaration</i>
571to have any linkage type other than "externally visible", <tt>dllimport</tt>,
572or <tt>extern_weak</tt>.</p>
573<p>Aliases can have only <tt>external</tt>, <tt>internal</tt> and <tt>weak</tt>
574linkages.
575</div>
576
577<!-- ======================================================================= -->
578<div class="doc_subsection">
579 <a name="callingconv">Calling Conventions</a>
580</div>
581
582<div class="doc_text">
583
584<p>LLVM <a href="#functionstructure">functions</a>, <a href="#i_call">calls</a>
585and <a href="#i_invoke">invokes</a> can all have an optional calling convention
586specified for the call. The calling convention of any pair of dynamic
587caller/callee must match, or the behavior of the program is undefined. The
588following calling conventions are supported by LLVM, and more may be added in
589the future:</p>
590
591<dl>
592 <dt><b>"<tt>ccc</tt>" - The C calling convention</b>:</dt>
593
594 <dd>This calling convention (the default if no other calling convention is
595 specified) matches the target C calling conventions. This calling convention
596 supports varargs function calls and tolerates some mismatch in the declared
597 prototype and implemented declaration of the function (as does normal C).
598 </dd>
599
600 <dt><b>"<tt>fastcc</tt>" - The fast calling convention</b>:</dt>
601
602 <dd>This calling convention attempts to make calls as fast as possible
603 (e.g. by passing things in registers). This calling convention allows the
604 target to use whatever tricks it wants to produce fast code for the target,
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]605 without having to conform to an externally specified ABI (Application Binary
606 Interface). Implementations of this convention should allow arbitrary
Arnold Schwaighofer07444922008-05-14 09:17:12 +0000[diff] [blame]607 <a href="CodeGenerator.html#tailcallopt">tail call optimization</a> to be
608 supported. This calling convention does not support varargs and requires the
609 prototype of all callees to exactly match the prototype of the function
610 definition.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]611 </dd>
612
613 <dt><b>"<tt>coldcc</tt>" - The cold calling convention</b>:</dt>
614
615 <dd>This calling convention attempts to make code in the caller as efficient
616 as possible under the assumption that the call is not commonly executed. As
617 such, these calls often preserve all registers so that the call does not break
618 any live ranges in the caller side. This calling convention does not support
619 varargs and requires the prototype of all callees to exactly match the
620 prototype of the function definition.
621 </dd>
622
623 <dt><b>"<tt>cc &lt;<em>n</em>&gt;</tt>" - Numbered convention</b>:</dt>
624
625 <dd>Any calling convention may be specified by number, allowing
626 target-specific calling conventions to be used. Target specific calling
627 conventions start at 64.
628 </dd>
629</dl>
630
631<p>More calling conventions can be added/defined on an as-needed basis, to
632support pascal conventions or any other well-known target-independent
633convention.</p>
634
635</div>
636
637<!-- ======================================================================= -->
638<div class="doc_subsection">
639 <a name="visibility">Visibility Styles</a>
640</div>
641
642<div class="doc_text">
643
644<p>
645All Global Variables and Functions have one of the following visibility styles:
646</p>
647
648<dl>
649 <dt><b>"<tt>default</tt>" - Default style</b>:</dt>
650
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]651 <dd>On targets that use the ELF object file format, default visibility means
652 that the declaration is visible to other
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]653 modules and, in shared libraries, means that the declared entity may be
654 overridden. On Darwin, default visibility means that the declaration is
655 visible to other modules. Default visibility corresponds to "external
656 linkage" in the language.
657 </dd>
658
659 <dt><b>"<tt>hidden</tt>" - Hidden style</b>:</dt>
660
661 <dd>Two declarations of an object with hidden visibility refer to the same
662 object if they are in the same shared object. Usually, hidden visibility
663 indicates that the symbol will not be placed into the dynamic symbol table,
664 so no other module (executable or shared library) can reference it
665 directly.
666 </dd>
667
668 <dt><b>"<tt>protected</tt>" - Protected style</b>:</dt>
669
670 <dd>On ELF, protected visibility indicates that the symbol will be placed in
671 the dynamic symbol table, but that references within the defining module will
672 bind to the local symbol. That is, the symbol cannot be overridden by another
673 module.
674 </dd>
675</dl>
676
677</div>
678
679<!-- ======================================================================= -->
680<div class="doc_subsection">
681 <a name="globalvars">Global Variables</a>
682</div>
683
684<div class="doc_text">
685
686<p>Global variables define regions of memory allocated at compilation time
687instead of run-time. Global variables may optionally be initialized, may have
688an explicit section to be placed in, and may have an optional explicit alignment
689specified. A variable may be defined as "thread_local", which means that it
690will not be shared by threads (each thread will have a separated copy of the
691variable). A variable may be defined as a global "constant," which indicates
692that the contents of the variable will <b>never</b> be modified (enabling better
693optimization, allowing the global data to be placed in the read-only section of
694an executable, etc). Note that variables that need runtime initialization
695cannot be marked "constant" as there is a store to the variable.</p>
696
697<p>
698LLVM explicitly allows <em>declarations</em> of global variables to be marked
699constant, even if the final definition of the global is not. This capability
700can be used to enable slightly better optimization of the program, but requires
701the language definition to guarantee that optimizations based on the
702'constantness' are valid for the translation units that do not include the
703definition.
704</p>
705
706<p>As SSA values, global variables define pointer values that are in
707scope (i.e. they dominate) all basic blocks in the program. Global
708variables always define a pointer to their "content" type because they
709describe a region of memory, and all memory objects in LLVM are
710accessed through pointers.</p>
711
Christopher Lambdd0049d2007-12-11 09:31:00 +0000[diff] [blame]712<p>A global variable may be declared to reside in a target-specifc numbered
713address space. For targets that support them, address spaces may affect how
714optimizations are performed and/or what target instructions are used to access
Christopher Lamb20a39e92007-12-12 08:44:39 +0000[diff] [blame]715the variable. The default address space is zero. The address space qualifier
716must precede any other attributes.</p>
Christopher Lambdd0049d2007-12-11 09:31:00 +0000[diff] [blame]717
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]718<p>LLVM allows an explicit section to be specified for globals. If the target
719supports it, it will emit globals to the section specified.</p>
720
721<p>An explicit alignment may be specified for a global. If not present, or if
722the alignment is set to zero, the alignment of the global is set by the target
723to whatever it feels convenient. If an explicit alignment is specified, the
724global is forced to have at least that much alignment. All alignments must be
725a power of 2.</p>
726
Christopher Lambdd0049d2007-12-11 09:31:00 +0000[diff] [blame]727<p>For example, the following defines a global in a numbered address space with
728an initializer, section, and alignment:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]729
730<div class="doc_code">
731<pre>
Christopher Lambdd0049d2007-12-11 09:31:00 +0000[diff] [blame]732@G = constant float 1.0 addrspace(5), section "foo", align 4
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]733</pre>
734</div>
735
736</div>
737
738
739<!-- ======================================================================= -->
740<div class="doc_subsection">
741 <a name="functionstructure">Functions</a>
742</div>
743
744<div class="doc_text">
745
746<p>LLVM function definitions consist of the "<tt>define</tt>" keyord,
747an optional <a href="#linkage">linkage type</a>, an optional
748<a href="#visibility">visibility style</a>, an optional
749<a href="#callingconv">calling convention</a>, a return type, an optional
750<a href="#paramattrs">parameter attribute</a> for the return type, a function
751name, a (possibly empty) argument list (each with optional
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]752<a href="#paramattrs">parameter attributes</a>), optional
753<a href="#fnattrs">function attributes</a>, an optional section,
754an optional alignment, an optional <a href="#gc">garbage collector name</a>,
Chris Lattneree202542008-10-04 18:10:21 +0000[diff] [blame]755an opening curly brace, a list of basic blocks, and a closing curly brace.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]756
757LLVM function declarations consist of the "<tt>declare</tt>" keyword, an
758optional <a href="#linkage">linkage type</a>, an optional
759<a href="#visibility">visibility style</a>, an optional
760<a href="#callingconv">calling convention</a>, a return type, an optional
761<a href="#paramattrs">parameter attribute</a> for the return type, a function
Gordon Henriksen13fe5e32007-12-10 03:18:06 +0000[diff] [blame]762name, a possibly empty list of arguments, an optional alignment, and an optional
Gordon Henriksend606f5b2007-12-10 03:30:21 +0000[diff] [blame]763<a href="#gc">garbage collector name</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]764
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]765<p>A function definition contains a list of basic blocks, forming the CFG
766(Control Flow Graph) for
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]767the function. Each basic block may optionally start with a label (giving the
768basic block a symbol table entry), contains a list of instructions, and ends
769with a <a href="#terminators">terminator</a> instruction (such as a branch or
770function return).</p>
771
772<p>The first basic block in a function is special in two ways: it is immediately
773executed on entrance to the function, and it is not allowed to have predecessor
774basic blocks (i.e. there can not be any branches to the entry block of a
775function). Because the block can have no predecessors, it also cannot have any
776<a href="#i_phi">PHI nodes</a>.</p>
777
778<p>LLVM allows an explicit section to be specified for functions. If the target
779supports it, it will emit functions to the section specified.</p>
780
781<p>An explicit alignment may be specified for a function. If not present, or if
782the alignment is set to zero, the alignment of the function is set by the target
783to whatever it feels convenient. If an explicit alignment is specified, the
784function is forced to have at least that much alignment. All alignments must be
785a power of 2.</p>
786
Devang Pateld0bfcc72008-10-07 17:48:33 +0000[diff] [blame]787 <h5>Syntax:</h5>
788
789<div class="doc_code">
Chris Lattner1e5c5cd02008-10-13 16:55:18 +0000[diff] [blame]790<tt>
791define [<a href="#linkage">linkage</a>] [<a href="#visibility">visibility</a>]
792 [<a href="#callingconv">cconv</a>] [<a href="#paramattrs">ret attrs</a>]
793 &lt;ResultType&gt; @&lt;FunctionName&gt; ([argument list])
794 [<a href="#fnattrs">fn Attrs</a>] [section "name"] [align N]
795 [<a href="#gc">gc</a>] { ... }
796</tt>
Devang Pateld0bfcc72008-10-07 17:48:33 +0000[diff] [blame]797</div>
798
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]799</div>
800
801
802<!-- ======================================================================= -->
803<div class="doc_subsection">
804 <a name="aliasstructure">Aliases</a>
805</div>
806<div class="doc_text">
807 <p>Aliases act as "second name" for the aliasee value (which can be either
Anton Korobeynikov96822812008-03-22 08:36:14 +0000[diff] [blame]808 function, global variable, another alias or bitcast of global value). Aliases
809 may have an optional <a href="#linkage">linkage type</a>, and an
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]810 optional <a href="#visibility">visibility style</a>.</p>
811
812 <h5>Syntax:</h5>
813
814<div class="doc_code">
815<pre>
Duncan Sandsd7bfabf2008-09-12 20:48:21 +0000[diff] [blame]816@&lt;Name&gt; = alias [Linkage] [Visibility] &lt;AliaseeTy&gt; @&lt;Aliasee&gt;
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]817</pre>
818</div>
819
820</div>
821
822
823
824<!-- ======================================================================= -->
825<div class="doc_subsection"><a name="paramattrs">Parameter Attributes</a></div>
826<div class="doc_text">
827 <p>The return type and each parameter of a function type may have a set of
828 <i>parameter attributes</i> associated with them. Parameter attributes are
829 used to communicate additional information about the result or parameters of
Duncan Sandsf5588dc2007-11-27 13:23:08 +0000[diff] [blame]830 a function. Parameter attributes are considered to be part of the function,
831 not of the function type, so functions with different parameter attributes
832 can have the same function type.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]833
834 <p>Parameter attributes are simple keywords that follow the type specified. If
835 multiple parameter attributes are needed, they are space separated. For
836 example:</p>
837
838<div class="doc_code">
839<pre>
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]840declare i32 @printf(i8* noalias , ...)
Chris Lattnerf33b8452008-10-04 18:33:34 +0000[diff] [blame]841declare i32 @atoi(i8 zeroext)
842declare signext i8 @returns_signed_char()
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]843</pre>
844</div>
845
Duncan Sandsf5588dc2007-11-27 13:23:08 +0000[diff] [blame]846 <p>Note that any attributes for the function result (<tt>nounwind</tt>,
847 <tt>readonly</tt>) come immediately after the argument list.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]848
849 <p>Currently, only the following parameter attributes are defined:</p>
850 <dl>
Reid Spencerf234bed2007-07-19 23:13:04 +0000[diff] [blame]851 <dt><tt>zeroext</tt></dt>
Chris Lattnerf33b8452008-10-04 18:33:34 +0000[diff] [blame]852 <dd>This indicates to the code generator that the parameter or return value
853 should be zero-extended to a 32-bit value by the caller (for a parameter)
854 or the callee (for a return value).</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]855
Reid Spencerf234bed2007-07-19 23:13:04 +0000[diff] [blame]856 <dt><tt>signext</tt></dt>
Chris Lattnerf33b8452008-10-04 18:33:34 +0000[diff] [blame]857 <dd>This indicates to the code generator that the parameter or return value
858 should be sign-extended to a 32-bit value by the caller (for a parameter)
859 or the callee (for a return value).</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]860
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]861 <dt><tt>inreg</tt></dt>
Dale Johannesenc08a0e22008-09-25 20:47:45 +0000[diff] [blame]862 <dd>This indicates that this parameter or return value should be treated
863 in a special target-dependent fashion during while emitting code for a
864 function call or return (usually, by putting it in a register as opposed
Chris Lattnerf33b8452008-10-04 18:33:34 +0000[diff] [blame]865 to memory, though some targets use it to distinguish between two different
866 kinds of registers). Use of this attribute is target-specific.</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]867
Duncan Sands7d94e5a2008-10-06 08:14:18 +0000[diff] [blame]868 <dt><tt><a name="byval">byval</a></tt></dt>
Chris Lattner04c86182008-01-15 04:34:22 +0000[diff] [blame]869 <dd>This indicates that the pointer parameter should really be passed by
870 value to the function. The attribute implies that a hidden copy of the
871 pointee is made between the caller and the callee, so the callee is unable
Chris Lattner6a9f3c42008-08-05 18:21:08 +0000[diff] [blame]872 to modify the value in the callee. This attribute is only valid on LLVM
Chris Lattner04c86182008-01-15 04:34:22 +0000[diff] [blame]873 pointer arguments. It is generally used to pass structs and arrays by
Duncan Sands7d94e5a2008-10-06 08:14:18 +0000[diff] [blame]874 value, but is also valid on pointers to scalars. The copy is considered to
875 belong to the caller not the callee (for example,
876 <tt><a href="#readonly">readonly</a></tt> functions should not write to
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]877 <tt>byval</tt> parameters). This is not a valid attribute for return
878 values. </dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]879
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]880 <dt><tt>sret</tt></dt>
Duncan Sands616cc032008-02-18 04:19:38 +0000[diff] [blame]881 <dd>This indicates that the pointer parameter specifies the address of a
882 structure that is the return value of the function in the source program.
Chris Lattnerf33b8452008-10-04 18:33:34 +0000[diff] [blame]883 This pointer must be guaranteed by the caller to be valid: loads and stores
884 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]885 be applied to the first parameter. This is not a valid attribute for
886 return values. </dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]887
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]888 <dt><tt>noalias</tt></dt>
Owen Andersonc4fc4cd2008-02-18 04:09:01 +0000[diff] [blame]889 <dd>This indicates that the parameter does not alias any global or any other
890 parameter. The caller is responsible for ensuring that this is the case,
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]891 usually by placing the value in a stack allocation. This is not a valid
892 attribute for return values.</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]893
Duncan Sands4ee46812007-07-27 19:57:41 +0000[diff] [blame]894 <dt><tt>nest</tt></dt>
Duncan Sandsf1a7d4c2008-07-08 09:27:25 +0000[diff] [blame]895 <dd>This indicates that the pointer parameter can be excised using the
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]896 <a href="#int_trampoline">trampoline intrinsics</a>. This is not a valid
897 attribute for return values.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]898 </dl>
899
900</div>
901
902<!-- ======================================================================= -->
903<div class="doc_subsection">
Gordon Henriksen13fe5e32007-12-10 03:18:06 +0000[diff] [blame]904 <a name="gc">Garbage Collector Names</a>
905</div>
906
907<div class="doc_text">
908<p>Each function may specify a garbage collector name, which is simply a
909string.</p>
910
911<div class="doc_code"><pre
912>define void @f() gc "name" { ...</pre></div>
913
914<p>The compiler declares the supported values of <i>name</i>. Specifying a
915collector which will cause the compiler to alter its output in order to support
916the named garbage collection algorithm.</p>
917</div>
918
919<!-- ======================================================================= -->
920<div class="doc_subsection">
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]921 <a name="fnattrs">Function Attributes</a>
Devang Pateld468f1c2008-09-04 23:05:13 +0000[diff] [blame]922</div>
923
924<div class="doc_text">
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]925
926<p>Function attributes are set to communicate additional information about
927 a function. Function attributes are considered to be part of the function,
928 not of the function type, so functions with different parameter attributes
929 can have the same function type.</p>
930
931 <p>Function attributes are simple keywords that follow the type specified. If
932 multiple attributes are needed, they are space separated. For
933 example:</p>
Devang Pateld468f1c2008-09-04 23:05:13 +0000[diff] [blame]934
935<div class="doc_code">
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]936<pre>
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]937define void @f() noinline { ... }
938define void @f() alwaysinline { ... }
939define void @f() alwaysinline optsize { ... }
940define void @f() optsize
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]941</pre>
Devang Pateld468f1c2008-09-04 23:05:13 +0000[diff] [blame]942</div>
943
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]944<dl>
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]945<dt><tt>alwaysinline</tt></dt>
Chris Lattner82d70f92008-10-04 18:23:17 +0000[diff] [blame]946<dd>This attribute indicates that the inliner should attempt to inline this
947function into callers whenever possible, ignoring any active inlining size
948threshold for this caller.</dd>
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]949
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]950<dt><tt>noinline</tt></dt>
Chris Lattner82d70f92008-10-04 18:23:17 +0000[diff] [blame]951<dd>This attribute indicates that the inliner should never inline this function
Chris Lattner377f3ae2008-10-05 17:14:59 +0000[diff] [blame]952in any situation. This attribute may not be used together with the
Chris Lattner82d70f92008-10-04 18:23:17 +0000[diff] [blame]953<tt>alwaysinline</tt> attribute.</dd>
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]954
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]955<dt><tt>optsize</tt></dt>
Devang Patelc29099b2008-09-29 18:34:44 +0000[diff] [blame]956<dd>This attribute suggests that optimization passes and code generator passes
Chris Lattner82d70f92008-10-04 18:23:17 +0000[diff] [blame]957make choices that keep the code size of this function low, and otherwise do
958optimizations specifically to reduce code size.</dd>
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]959
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]960<dt><tt>noreturn</tt></dt>
Chris Lattner82d70f92008-10-04 18:23:17 +0000[diff] [blame]961<dd>This function attribute indicates that the function never returns normally.
962This produces undefined behavior at runtime if the function ever does
963dynamically return.</dd>
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]964
965<dt><tt>nounwind</tt></dt>
Chris Lattner82d70f92008-10-04 18:23:17 +0000[diff] [blame]966<dd>This function attribute indicates that the function never returns with an
967unwind or exceptional control flow. If the function does unwind, its runtime
968behavior is undefined.</dd>
969
970<dt><tt>readnone</tt></dt>
Duncan Sands7d94e5a2008-10-06 08:14:18 +0000[diff] [blame]971<dd>This attribute indicates that the function computes its result (or the
972exception it throws) based strictly on its arguments, without dereferencing any
973pointer arguments or otherwise accessing any mutable state (e.g. memory, control
974registers, etc) visible to caller functions. It does not write through any
975pointer arguments (including <tt><a href="#byval">byval</a></tt> arguments) and
976never changes any state visible to callers.</dd>
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]977
Duncan Sands7d94e5a2008-10-06 08:14:18 +0000[diff] [blame]978<dt><tt><a name="readonly">readonly</a></tt></dt>
979<dd>This attribute indicates that the function does not write through any
980pointer arguments (including <tt><a href="#byval">byval</a></tt> arguments)
981or otherwise modify any state (e.g. memory, control registers, etc) visible to
982caller functions. It may dereference pointer arguments and read state that may
983be set in the caller. A readonly function always returns the same value (or
984throws the same exception) when called with the same set of arguments and global
985state.</dd>
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]986</dl>
987
Devang Pateld468f1c2008-09-04 23:05:13 +0000[diff] [blame]988</div>
989
990<!-- ======================================================================= -->
991<div class="doc_subsection">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]992 <a name="moduleasm">Module-Level Inline Assembly</a>
993</div>
994
995<div class="doc_text">
996<p>
997Modules may contain "module-level inline asm" blocks, which corresponds to the
998GCC "file scope inline asm" blocks. These blocks are internally concatenated by
999LLVM and treated as a single unit, but may be separated in the .ll file if
1000desired. The syntax is very simple:
1001</p>
1002
1003<div class="doc_code">
1004<pre>
1005module asm "inline asm code goes here"
1006module asm "more can go here"
1007</pre>
1008</div>
1009
1010<p>The strings can contain any character by escaping non-printable characters.
1011 The escape sequence used is simply "\xx" where "xx" is the two digit hex code
1012 for the number.
1013</p>
1014
1015<p>
1016 The inline asm code is simply printed to the machine code .s file when
1017 assembly code is generated.
1018</p>
1019</div>
1020
1021<!-- ======================================================================= -->
1022<div class="doc_subsection">
1023 <a name="datalayout">Data Layout</a>
1024</div>
1025
1026<div class="doc_text">
1027<p>A module may specify a target specific data layout string that specifies how
1028data is to be laid out in memory. The syntax for the data layout is simply:</p>
1029<pre> target datalayout = "<i>layout specification</i>"</pre>
1030<p>The <i>layout specification</i> consists of a list of specifications
1031separated by the minus sign character ('-'). Each specification starts with a
1032letter and may include other information after the letter to define some
1033aspect of the data layout. The specifications accepted are as follows: </p>
1034<dl>
1035 <dt><tt>E</tt></dt>
1036 <dd>Specifies that the target lays out data in big-endian form. That is, the
1037 bits with the most significance have the lowest address location.</dd>
1038 <dt><tt>e</tt></dt>
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]1039 <dd>Specifies that the target lays out data in little-endian form. That is,
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1040 the bits with the least significance have the lowest address location.</dd>
1041 <dt><tt>p:<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1042 <dd>This specifies the <i>size</i> of a pointer and its <i>abi</i> and
1043 <i>preferred</i> alignments. All sizes are in bits. Specifying the <i>pref</i>
1044 alignment is optional. If omitted, the preceding <tt>:</tt> should be omitted
1045 too.</dd>
1046 <dt><tt>i<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1047 <dd>This specifies the alignment for an integer type of a given bit
1048 <i>size</i>. The value of <i>size</i> must be in the range [1,2^23).</dd>
1049 <dt><tt>v<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1050 <dd>This specifies the alignment for a vector type of a given bit
1051 <i>size</i>.</dd>
1052 <dt><tt>f<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1053 <dd>This specifies the alignment for a floating point type of a given bit
1054 <i>size</i>. The value of <i>size</i> must be either 32 (float) or 64
1055 (double).</dd>
1056 <dt><tt>a<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1057 <dd>This specifies the alignment for an aggregate type of a given bit
1058 <i>size</i>.</dd>
1059</dl>
1060<p>When constructing the data layout for a given target, LLVM starts with a
1061default set of specifications which are then (possibly) overriden by the
1062specifications in the <tt>datalayout</tt> keyword. The default specifications
1063are given in this list:</p>
1064<ul>
1065 <li><tt>E</tt> - big endian</li>
1066 <li><tt>p:32:64:64</tt> - 32-bit pointers with 64-bit alignment</li>
1067 <li><tt>i1:8:8</tt> - i1 is 8-bit (byte) aligned</li>
1068 <li><tt>i8:8:8</tt> - i8 is 8-bit (byte) aligned</li>
1069 <li><tt>i16:16:16</tt> - i16 is 16-bit aligned</li>
1070 <li><tt>i32:32:32</tt> - i32 is 32-bit aligned</li>
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]1071 <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]1072 alignment of 64-bits</li>
1073 <li><tt>f32:32:32</tt> - float is 32-bit aligned</li>
1074 <li><tt>f64:64:64</tt> - double is 64-bit aligned</li>
1075 <li><tt>v64:64:64</tt> - 64-bit vector is 64-bit aligned</li>
1076 <li><tt>v128:128:128</tt> - 128-bit vector is 128-bit aligned</li>
1077 <li><tt>a0:0:1</tt> - aggregates are 8-bit aligned</li>
1078</ul>
Chris Lattner6a9f3c42008-08-05 18:21:08 +0000[diff] [blame]1079<p>When LLVM is determining the alignment for a given type, it uses the
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1080following rules:
1081<ol>
1082 <li>If the type sought is an exact match for one of the specifications, that
1083 specification is used.</li>
1084 <li>If no match is found, and the type sought is an integer type, then the
1085 smallest integer type that is larger than the bitwidth of the sought type is
1086 used. If none of the specifications are larger than the bitwidth then the the
1087 largest integer type is used. For example, given the default specifications
1088 above, the i7 type will use the alignment of i8 (next largest) while both
1089 i65 and i256 will use the alignment of i64 (largest specified).</li>
1090 <li>If no match is found, and the type sought is a vector type, then the
1091 largest vector type that is smaller than the sought vector type will be used
1092 as a fall back. This happens because <128 x double> can be implemented in
1093 terms of 64 <2 x double>, for example.</li>
1094</ol>
1095</div>
1096
1097<!-- *********************************************************************** -->
1098<div class="doc_section"> <a name="typesystem">Type System</a> </div>
1099<!-- *********************************************************************** -->
1100
1101<div class="doc_text">
1102
1103<p>The LLVM type system is one of the most important features of the
1104intermediate representation. Being typed enables a number of
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]1105optimizations to be performed on the intermediate representation directly,
1106without having to do
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1107extra analyses on the side before the transformation. A strong type
1108system makes it easier to read the generated code and enables novel
1109analyses and transformations that are not feasible to perform on normal
1110three address code representations.</p>
1111
1112</div>
1113
1114<!-- ======================================================================= -->
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1115<div class="doc_subsection"> <a name="t_classifications">Type
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1116Classifications</a> </div>
1117<div class="doc_text">
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1118<p>The types fall into a few useful
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1119classifications:</p>
1120
1121<table border="1" cellspacing="0" cellpadding="4">
1122 <tbody>
1123 <tr><th>Classification</th><th>Types</th></tr>
1124 <tr>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1125 <td><a href="#t_integer">integer</a></td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1126 <td><tt>i1, i2, i3, ... i8, ... i16, ... i32, ... i64, ... </tt></td>
1127 </tr>
1128 <tr>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1129 <td><a href="#t_floating">floating point</a></td>
1130 <td><tt>float, double, x86_fp80, fp128, ppc_fp128</tt></td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1131 </tr>
1132 <tr>
1133 <td><a name="t_firstclass">first class</a></td>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1134 <td><a href="#t_integer">integer</a>,
1135 <a href="#t_floating">floating point</a>,
1136 <a href="#t_pointer">pointer</a>,
Dan Gohmanf6237db2008-06-18 18:42:13 +0000[diff] [blame]1137 <a href="#t_vector">vector</a>,
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]1138 <a href="#t_struct">structure</a>,
1139 <a href="#t_array">array</a>,
Dan Gohmand13951e2008-05-23 22:50:26 +0000[diff] [blame]1140 <a href="#t_label">label</a>.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1141 </td>
1142 </tr>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1143 <tr>
1144 <td><a href="#t_primitive">primitive</a></td>
1145 <td><a href="#t_label">label</a>,
1146 <a href="#t_void">void</a>,
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1147 <a href="#t_floating">floating point</a>.</td>
1148 </tr>
1149 <tr>
1150 <td><a href="#t_derived">derived</a></td>
1151 <td><a href="#t_integer">integer</a>,
1152 <a href="#t_array">array</a>,
1153 <a href="#t_function">function</a>,
1154 <a href="#t_pointer">pointer</a>,
1155 <a href="#t_struct">structure</a>,
1156 <a href="#t_pstruct">packed structure</a>,
1157 <a href="#t_vector">vector</a>,
1158 <a href="#t_opaque">opaque</a>.
Dan Gohman032ba852008-10-14 16:32:04 +0000[diff] [blame^]1159 </td>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1160 </tr>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1161 </tbody>
1162</table>
1163
1164<p>The <a href="#t_firstclass">first class</a> types are perhaps the
1165most important. Values of these types are the only ones which can be
1166produced by instructions, passed as arguments, or used as operands to
Dan Gohman4f29e422008-05-23 21:53:15 +0000[diff] [blame]1167instructions.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1168</div>
1169
1170<!-- ======================================================================= -->
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1171<div class="doc_subsection"> <a name="t_primitive">Primitive Types</a> </div>
Chris Lattner86437612008-01-04 04:34:14 +0000[diff] [blame]1172
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1173<div class="doc_text">
1174<p>The primitive types are the fundamental building blocks of the LLVM
1175system.</p>
1176
Chris Lattner86437612008-01-04 04:34:14 +0000[diff] [blame]1177</div>
1178
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1179<!-- _______________________________________________________________________ -->
1180<div class="doc_subsubsection"> <a name="t_floating">Floating Point Types</a> </div>
1181
1182<div class="doc_text">
1183 <table>
1184 <tbody>
1185 <tr><th>Type</th><th>Description</th></tr>
1186 <tr><td><tt>float</tt></td><td>32-bit floating point value</td></tr>
1187 <tr><td><tt>double</tt></td><td>64-bit floating point value</td></tr>
1188 <tr><td><tt>fp128</tt></td><td>128-bit floating point value (112-bit mantissa)</td></tr>
1189 <tr><td><tt>x86_fp80</tt></td><td>80-bit floating point value (X87)</td></tr>
1190 <tr><td><tt>ppc_fp128</tt></td><td>128-bit floating point value (two 64-bits)</td></tr>
1191 </tbody>
1192 </table>
1193</div>
1194
1195<!-- _______________________________________________________________________ -->
1196<div class="doc_subsubsection"> <a name="t_void">Void Type</a> </div>
1197
1198<div class="doc_text">
1199<h5>Overview:</h5>
1200<p>The void type does not represent any value and has no size.</p>
1201
1202<h5>Syntax:</h5>
1203
1204<pre>
1205 void
1206</pre>
1207</div>
1208
1209<!-- _______________________________________________________________________ -->
1210<div class="doc_subsubsection"> <a name="t_label">Label Type</a> </div>
1211
1212<div class="doc_text">
1213<h5>Overview:</h5>
1214<p>The label type represents code labels.</p>
1215
1216<h5>Syntax:</h5>
1217
1218<pre>
1219 label
1220</pre>
1221</div>
1222
1223
1224<!-- ======================================================================= -->
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1225<div class="doc_subsection"> <a name="t_derived">Derived Types</a> </div>
1226
1227<div class="doc_text">
1228
1229<p>The real power in LLVM comes from the derived types in the system.
1230This is what allows a programmer to represent arrays, functions,
1231pointers, and other useful types. Note that these derived types may be
1232recursive: For example, it is possible to have a two dimensional array.</p>
1233
1234</div>
1235
1236<!-- _______________________________________________________________________ -->
1237<div class="doc_subsubsection"> <a name="t_integer">Integer Type</a> </div>
1238
1239<div class="doc_text">
1240
1241<h5>Overview:</h5>
1242<p>The integer type is a very simple derived type that simply specifies an
1243arbitrary bit width for the integer type desired. Any bit width from 1 bit to
12442^23-1 (about 8 million) can be specified.</p>
1245
1246<h5>Syntax:</h5>
1247
1248<pre>
1249 iN
1250</pre>
1251
1252<p>The number of bits the integer will occupy is specified by the <tt>N</tt>
1253value.</p>
1254
1255<h5>Examples:</h5>
1256<table class="layout">
Chris Lattner251ab812007-12-18 06:18:21 +0000[diff] [blame]1257 <tbody>
1258 <tr>
1259 <td><tt>i1</tt></td>
1260 <td>a single-bit integer.</td>
1261 </tr><tr>
1262 <td><tt>i32</tt></td>
1263 <td>a 32-bit integer.</td>
1264 </tr><tr>
1265 <td><tt>i1942652</tt></td>
1266 <td>a really big integer of over 1 million bits.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1267 </tr>
Chris Lattner251ab812007-12-18 06:18:21 +0000[diff] [blame]1268 </tbody>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1269</table>
1270</div>
1271
1272<!-- _______________________________________________________________________ -->
1273<div class="doc_subsubsection"> <a name="t_array">Array Type</a> </div>
1274
1275<div class="doc_text">
1276
1277<h5>Overview:</h5>
1278
1279<p>The array type is a very simple derived type that arranges elements
1280sequentially in memory. The array type requires a size (number of
1281elements) and an underlying data type.</p>
1282
1283<h5>Syntax:</h5>
1284
1285<pre>
1286 [&lt;# elements&gt; x &lt;elementtype&gt;]
1287</pre>
1288
1289<p>The number of elements is a constant integer value; elementtype may
1290be any type with a size.</p>
1291
1292<h5>Examples:</h5>
1293<table class="layout">
1294 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1295 <td class="left"><tt>[40 x i32]</tt></td>
1296 <td class="left">Array of 40 32-bit integer values.</td>
1297 </tr>
1298 <tr class="layout">
1299 <td class="left"><tt>[41 x i32]</tt></td>
1300 <td class="left">Array of 41 32-bit integer values.</td>
1301 </tr>
1302 <tr class="layout">
1303 <td class="left"><tt>[4 x i8]</tt></td>
1304 <td class="left">Array of 4 8-bit integer values.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1305 </tr>
1306</table>
1307<p>Here are some examples of multidimensional arrays:</p>
1308<table class="layout">
1309 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1310 <td class="left"><tt>[3 x [4 x i32]]</tt></td>
1311 <td class="left">3x4 array of 32-bit integer values.</td>
1312 </tr>
1313 <tr class="layout">
1314 <td class="left"><tt>[12 x [10 x float]]</tt></td>
1315 <td class="left">12x10 array of single precision floating point values.</td>
1316 </tr>
1317 <tr class="layout">
1318 <td class="left"><tt>[2 x [3 x [4 x i16]]]</tt></td>
1319 <td class="left">2x3x4 array of 16-bit integer values.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1320 </tr>
1321</table>
1322
1323<p>Note that 'variable sized arrays' can be implemented in LLVM with a zero
1324length array. Normally, accesses past the end of an array are undefined in
1325LLVM (e.g. it is illegal to access the 5th element of a 3 element array).
1326As a special case, however, zero length arrays are recognized to be variable
1327length. This allows implementation of 'pascal style arrays' with the LLVM
1328type "{ i32, [0 x float]}", for example.</p>
1329
1330</div>
1331
1332<!-- _______________________________________________________________________ -->
1333<div class="doc_subsubsection"> <a name="t_function">Function Type</a> </div>
1334<div class="doc_text">
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1335
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1336<h5>Overview:</h5>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1337
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1338<p>The function type can be thought of as a function signature. It
Devang Pateld4ba41d2008-03-24 05:35:41 +0000[diff] [blame]1339consists of a return type and a list of formal parameter types. The
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1340return 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]1341If 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]1342class types, and the struct must have at least one element.</p>
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]1343
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1344<h5>Syntax:</h5>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1345
1346<pre>
1347 &lt;returntype list&gt; (&lt;parameter list&gt;)
1348</pre>
1349
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1350<p>...where '<tt>&lt;parameter list&gt;</tt>' is a comma-separated list of type
1351specifiers. Optionally, the parameter list may include a type <tt>...</tt>,
1352which indicates that the function takes a variable number of arguments.
1353Variable argument functions can access their arguments with the <a
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]1354 href="#int_varargs">variable argument handling intrinsic</a> functions.
1355'<tt>&lt;returntype list&gt;</tt>' is a comma-separated list of
1356<a href="#t_firstclass">first class</a> type specifiers.</p>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1357
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1358<h5>Examples:</h5>
1359<table class="layout">
1360 <tr class="layout">
1361 <td class="left"><tt>i32 (i32)</tt></td>
1362 <td class="left">function taking an <tt>i32</tt>, returning an <tt>i32</tt>
1363 </td>
1364 </tr><tr class="layout">
Reid Spencerf234bed2007-07-19 23:13:04 +0000[diff] [blame]1365 <td class="left"><tt>float&nbsp;(i16&nbsp;signext,&nbsp;i32&nbsp;*)&nbsp;*
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1366 </tt></td>
1367 <td class="left"><a href="#t_pointer">Pointer</a> to a function that takes
1368 an <tt>i16</tt> that should be sign extended and a
1369 <a href="#t_pointer">pointer</a> to <tt>i32</tt>, returning
1370 <tt>float</tt>.
1371 </td>
1372 </tr><tr class="layout">
1373 <td class="left"><tt>i32 (i8*, ...)</tt></td>
1374 <td class="left">A vararg function that takes at least one
1375 <a href="#t_pointer">pointer</a> to <tt>i8 </tt> (char in C),
1376 which returns an integer. This is the signature for <tt>printf</tt> in
1377 LLVM.
1378 </td>
Devang Pateld4ba41d2008-03-24 05:35:41 +0000[diff] [blame]1379 </tr><tr class="layout">
1380 <td class="left"><tt>{i32, i32} (i32)</tt></td>
Devang Patel6dddba22008-03-24 18:10:52 +0000[diff] [blame]1381 <td class="left">A function taking an <tt>i32></tt>, returning two
1382 <tt> i32 </tt> values as an aggregate of type <tt>{ i32, i32 }</tt>
Devang Pateld4ba41d2008-03-24 05:35:41 +0000[diff] [blame]1383 </td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1384 </tr>
1385</table>
1386
1387</div>
1388<!-- _______________________________________________________________________ -->
1389<div class="doc_subsubsection"> <a name="t_struct">Structure Type</a> </div>
1390<div class="doc_text">
1391<h5>Overview:</h5>
1392<p>The structure type is used to represent a collection of data members
1393together in memory. The packing of the field types is defined to match
1394the ABI of the underlying processor. The elements of a structure may
1395be any type that has a size.</p>
1396<p>Structures are accessed using '<tt><a href="#i_load">load</a></tt>
1397and '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a
1398field with the '<tt><a href="#i_getelementptr">getelementptr</a></tt>'
1399instruction.</p>
1400<h5>Syntax:</h5>
1401<pre> { &lt;type list&gt; }<br></pre>
1402<h5>Examples:</h5>
1403<table class="layout">
1404 <tr class="layout">
1405 <td class="left"><tt>{ i32, i32, i32 }</tt></td>
1406 <td class="left">A triple of three <tt>i32</tt> values</td>
1407 </tr><tr class="layout">
1408 <td class="left"><tt>{&nbsp;float,&nbsp;i32&nbsp;(i32)&nbsp;*&nbsp;}</tt></td>
1409 <td class="left">A pair, where the first element is a <tt>float</tt> and the
1410 second element is a <a href="#t_pointer">pointer</a> to a
1411 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
1412 an <tt>i32</tt>.</td>
1413 </tr>
1414</table>
1415</div>
1416
1417<!-- _______________________________________________________________________ -->
1418<div class="doc_subsubsection"> <a name="t_pstruct">Packed Structure Type</a>
1419</div>
1420<div class="doc_text">
1421<h5>Overview:</h5>
1422<p>The packed structure type is used to represent a collection of data members
1423together in memory. There is no padding between fields. Further, the alignment
1424of a packed structure is 1 byte. The elements of a packed structure may
1425be any type that has a size.</p>
1426<p>Structures are accessed using '<tt><a href="#i_load">load</a></tt>
1427and '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a
1428field with the '<tt><a href="#i_getelementptr">getelementptr</a></tt>'
1429instruction.</p>
1430<h5>Syntax:</h5>
1431<pre> &lt; { &lt;type list&gt; } &gt; <br></pre>
1432<h5>Examples:</h5>
1433<table class="layout">
1434 <tr class="layout">
1435 <td class="left"><tt>&lt; { i32, i32, i32 } &gt;</tt></td>
1436 <td class="left">A triple of three <tt>i32</tt> values</td>
1437 </tr><tr class="layout">
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]1438 <td class="left">
1439<tt>&lt;&nbsp;{&nbsp;float,&nbsp;i32&nbsp;(i32)*&nbsp;}&nbsp;&gt;</tt></td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1440 <td class="left">A pair, where the first element is a <tt>float</tt> and the
1441 second element is a <a href="#t_pointer">pointer</a> to a
1442 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
1443 an <tt>i32</tt>.</td>
1444 </tr>
1445</table>
1446</div>
1447
1448<!-- _______________________________________________________________________ -->
1449<div class="doc_subsubsection"> <a name="t_pointer">Pointer Type</a> </div>
1450<div class="doc_text">
1451<h5>Overview:</h5>
1452<p>As in many languages, the pointer type represents a pointer or
Christopher Lambdd0049d2007-12-11 09:31:00 +0000[diff] [blame]1453reference to another object, which must live in memory. Pointer types may have
1454an optional address space attribute defining the target-specific numbered
1455address space where the pointed-to object resides. The default address space is
1456zero.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1457<h5>Syntax:</h5>
1458<pre> &lt;type&gt; *<br></pre>
1459<h5>Examples:</h5>
1460<table class="layout">
1461 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1462 <td class="left"><tt>[4x i32]*</tt></td>
1463 <td class="left">A <a href="#t_pointer">pointer</a> to <a
1464 href="#t_array">array</a> of four <tt>i32</tt> values.</td>
1465 </tr>
1466 <tr class="layout">
1467 <td class="left"><tt>i32 (i32 *) *</tt></td>
1468 <td class="left"> A <a href="#t_pointer">pointer</a> to a <a
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1469 href="#t_function">function</a> that takes an <tt>i32*</tt>, returning an
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1470 <tt>i32</tt>.</td>
1471 </tr>
1472 <tr class="layout">
1473 <td class="left"><tt>i32 addrspace(5)*</tt></td>
1474 <td class="left">A <a href="#t_pointer">pointer</a> to an <tt>i32</tt> value
1475 that resides in address space #5.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1476 </tr>
1477</table>
1478</div>
1479
1480<!-- _______________________________________________________________________ -->
1481<div class="doc_subsubsection"> <a name="t_vector">Vector Type</a> </div>
1482<div class="doc_text">
1483
1484<h5>Overview:</h5>
1485
1486<p>A vector type is a simple derived type that represents a vector
1487of elements. Vector types are used when multiple primitive data
1488are operated in parallel using a single instruction (SIMD).
1489A vector type requires a size (number of
1490elements) and an underlying primitive data type. Vectors must have a power
1491of two length (1, 2, 4, 8, 16 ...). Vector types are
1492considered <a href="#t_firstclass">first class</a>.</p>
1493
1494<h5>Syntax:</h5>
1495
1496<pre>
1497 &lt; &lt;# elements&gt; x &lt;elementtype&gt; &gt;
1498</pre>
1499
1500<p>The number of elements is a constant integer value; elementtype may
1501be any integer or floating point type.</p>
1502
1503<h5>Examples:</h5>
1504
1505<table class="layout">
1506 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1507 <td class="left"><tt>&lt;4 x i32&gt;</tt></td>
1508 <td class="left">Vector of 4 32-bit integer values.</td>
1509 </tr>
1510 <tr class="layout">
1511 <td class="left"><tt>&lt;8 x float&gt;</tt></td>
1512 <td class="left">Vector of 8 32-bit floating-point values.</td>
1513 </tr>
1514 <tr class="layout">
1515 <td class="left"><tt>&lt;2 x i64&gt;</tt></td>
1516 <td class="left">Vector of 2 64-bit integer values.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1517 </tr>
1518</table>
1519</div>
1520
1521<!-- _______________________________________________________________________ -->
1522<div class="doc_subsubsection"> <a name="t_opaque">Opaque Type</a> </div>
1523<div class="doc_text">
1524
1525<h5>Overview:</h5>
1526
1527<p>Opaque types are used to represent unknown types in the system. This
Gordon Henriksenda0706e2007-10-14 00:34:53 +0000[diff] [blame]1528corresponds (for example) to the C notion of a forward declared structure type.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1529In LLVM, opaque types can eventually be resolved to any type (not just a
1530structure type).</p>
1531
1532<h5>Syntax:</h5>
1533
1534<pre>
1535 opaque
1536</pre>
1537
1538<h5>Examples:</h5>
1539
1540<table class="layout">
1541 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1542 <td class="left"><tt>opaque</tt></td>
1543 <td class="left">An opaque type.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1544 </tr>
1545</table>
1546</div>
1547
1548
1549<!-- *********************************************************************** -->
1550<div class="doc_section"> <a name="constants">Constants</a> </div>
1551<!-- *********************************************************************** -->
1552
1553<div class="doc_text">
1554
1555<p>LLVM has several different basic types of constants. This section describes
1556them all and their syntax.</p>
1557
1558</div>
1559
1560<!-- ======================================================================= -->
1561<div class="doc_subsection"><a name="simpleconstants">Simple Constants</a></div>
1562
1563<div class="doc_text">
1564
1565<dl>
1566 <dt><b>Boolean constants</b></dt>
1567
1568 <dd>The two strings '<tt>true</tt>' and '<tt>false</tt>' are both valid
1569 constants of the <tt><a href="#t_primitive">i1</a></tt> type.
1570 </dd>
1571
1572 <dt><b>Integer constants</b></dt>
1573
1574 <dd>Standard integers (such as '4') are constants of the <a
1575 href="#t_integer">integer</a> type. Negative numbers may be used with
1576 integer types.
1577 </dd>
1578
1579 <dt><b>Floating point constants</b></dt>
1580
1581 <dd>Floating point constants use standard decimal notation (e.g. 123.421),
1582 exponential notation (e.g. 1.23421e+2), or a more precise hexadecimal
Chris Lattnerdc15b1d2008-04-01 18:45:27 +0000[diff] [blame]1583 notation (see below). The assembler requires the exact decimal value of
1584 a floating-point constant. For example, the assembler accepts 1.25 but
1585 rejects 1.3 because 1.3 is a repeating decimal in binary. Floating point
1586 constants must have a <a href="#t_floating">floating point</a> type. </dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1587
1588 <dt><b>Null pointer constants</b></dt>
1589
1590 <dd>The identifier '<tt>null</tt>' is recognized as a null pointer constant
1591 and must be of <a href="#t_pointer">pointer type</a>.</dd>
1592
1593</dl>
1594
1595<p>The one non-intuitive notation for constants is the optional hexadecimal form
1596of floating point constants. For example, the form '<tt>double
15970x432ff973cafa8000</tt>' is equivalent to (but harder to read than) '<tt>double
15984.5e+15</tt>'. The only time hexadecimal floating point constants are required
1599(and the only time that they are generated by the disassembler) is when a
1600floating point constant must be emitted but it cannot be represented as a
1601decimal floating point number. For example, NaN's, infinities, and other
1602special values are represented in their IEEE hexadecimal format so that
1603assembly and disassembly do not cause any bits to change in the constants.</p>
1604
1605</div>
1606
1607<!-- ======================================================================= -->
1608<div class="doc_subsection"><a name="aggregateconstants">Aggregate Constants</a>
1609</div>
1610
1611<div class="doc_text">
1612<p>Aggregate constants arise from aggregation of simple constants
1613and smaller aggregate constants.</p>
1614
1615<dl>
1616 <dt><b>Structure constants</b></dt>
1617
1618 <dd>Structure constants are represented with notation similar to structure
1619 type definitions (a comma separated list of elements, surrounded by braces
Chris Lattner6c8de962007-12-25 20:34:52 +0000[diff] [blame]1620 (<tt>{}</tt>)). For example: "<tt>{ i32 4, float 17.0, i32* @G }</tt>",
1621 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]1622 must have <a href="#t_struct">structure type</a>, and the number and
1623 types of elements must match those specified by the type.
1624 </dd>
1625
1626 <dt><b>Array constants</b></dt>
1627
1628 <dd>Array constants are represented with notation similar to array type
1629 definitions (a comma separated list of elements, surrounded by square brackets
1630 (<tt>[]</tt>)). For example: "<tt>[ i32 42, i32 11, i32 74 ]</tt>". Array
1631 constants must have <a href="#t_array">array type</a>, and the number and
1632 types of elements must match those specified by the type.
1633 </dd>
1634
1635 <dt><b>Vector constants</b></dt>
1636
1637 <dd>Vector constants are represented with notation similar to vector type
1638 definitions (a comma separated list of elements, surrounded by
1639 less-than/greater-than's (<tt>&lt;&gt;</tt>)). For example: "<tt>&lt; i32 42,
1640 i32 11, i32 74, i32 100 &gt;</tt>". Vector constants must have <a
1641 href="#t_vector">vector type</a>, and the number and types of elements must
1642 match those specified by the type.
1643 </dd>
1644
1645 <dt><b>Zero initialization</b></dt>
1646
1647 <dd>The string '<tt>zeroinitializer</tt>' can be used to zero initialize a
1648 value to zero of <em>any</em> type, including scalar and aggregate types.
1649 This is often used to avoid having to print large zero initializers (e.g. for
1650 large arrays) and is always exactly equivalent to using explicit zero
1651 initializers.
1652 </dd>
1653</dl>
1654
1655</div>
1656
1657<!-- ======================================================================= -->
1658<div class="doc_subsection">
1659 <a name="globalconstants">Global Variable and Function Addresses</a>
1660</div>
1661
1662<div class="doc_text">
1663
1664<p>The addresses of <a href="#globalvars">global variables</a> and <a
1665href="#functionstructure">functions</a> are always implicitly valid (link-time)
1666constants. These constants are explicitly referenced when the <a
1667href="#identifiers">identifier for the global</a> is used and always have <a
1668href="#t_pointer">pointer</a> type. For example, the following is a legal LLVM
1669file:</p>
1670
1671<div class="doc_code">
1672<pre>
1673@X = global i32 17
1674@Y = global i32 42
1675@Z = global [2 x i32*] [ i32* @X, i32* @Y ]
1676</pre>
1677</div>
1678
1679</div>
1680
1681<!-- ======================================================================= -->
1682<div class="doc_subsection"><a name="undefvalues">Undefined Values</a></div>
1683<div class="doc_text">
1684 <p>The string '<tt>undef</tt>' is recognized as a type-less constant that has
1685 no specific value. Undefined values may be of any type and be used anywhere
1686 a constant is permitted.</p>
1687
1688 <p>Undefined values indicate to the compiler that the program is well defined
1689 no matter what value is used, giving the compiler more freedom to optimize.
1690 </p>
1691</div>
1692
1693<!-- ======================================================================= -->
1694<div class="doc_subsection"><a name="constantexprs">Constant Expressions</a>
1695</div>
1696
1697<div class="doc_text">
1698
1699<p>Constant expressions are used to allow expressions involving other constants
1700to be used as constants. Constant expressions may be of any <a
1701href="#t_firstclass">first class</a> type and may involve any LLVM operation
1702that does not have side effects (e.g. load and call are not supported). The
1703following is the syntax for constant expressions:</p>
1704
1705<dl>
1706 <dt><b><tt>trunc ( CST to TYPE )</tt></b></dt>
1707 <dd>Truncate a constant to another type. The bit size of CST must be larger
1708 than the bit size of TYPE. Both types must be integers.</dd>
1709
1710 <dt><b><tt>zext ( CST to TYPE )</tt></b></dt>
1711 <dd>Zero extend a constant to another type. The bit size of CST must be
1712 smaller or equal to the bit size of TYPE. Both types must be integers.</dd>
1713
1714 <dt><b><tt>sext ( CST to TYPE )</tt></b></dt>
1715 <dd>Sign extend a constant to another type. The bit size of CST must be
1716 smaller or equal to the bit size of TYPE. Both types must be integers.</dd>
1717
1718 <dt><b><tt>fptrunc ( CST to TYPE )</tt></b></dt>
1719 <dd>Truncate a floating point constant to another floating point type. The
1720 size of CST must be larger than the size of TYPE. Both types must be
1721 floating point.</dd>
1722
1723 <dt><b><tt>fpext ( CST to TYPE )</tt></b></dt>
1724 <dd>Floating point extend a constant to another type. The size of CST must be
1725 smaller or equal to the size of TYPE. Both types must be floating point.</dd>
1726
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]1727 <dt><b><tt>fptoui ( CST to TYPE )</tt></b></dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1728 <dd>Convert a floating point constant to the corresponding unsigned integer
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]1729 constant. TYPE must be a scalar or vector integer type. CST must be of scalar
1730 or vector floating point type. Both CST and TYPE must be scalars, or vectors
1731 of the same number of elements. If the value won't fit in the integer type,
1732 the results are undefined.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1733
1734 <dt><b><tt>fptosi ( CST to TYPE )</tt></b></dt>
1735 <dd>Convert a floating point constant to the corresponding signed integer
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]1736 constant. TYPE must be a scalar or vector integer type. CST must be of scalar
1737 or vector floating point type. Both CST and TYPE must be scalars, or vectors
1738 of the same number of elements. If the value won't fit in the integer type,
1739 the results are undefined.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1740
1741 <dt><b><tt>uitofp ( CST to TYPE )</tt></b></dt>
1742 <dd>Convert an unsigned integer constant to the corresponding floating point
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]1743 constant. TYPE must be a scalar or vector floating point type. CST must be of
1744 scalar or vector integer type. Both CST and TYPE must be scalars, or vectors
1745 of the same number of elements. If the value won't fit in the floating point
1746 type, the results are undefined.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1747
1748 <dt><b><tt>sitofp ( CST to TYPE )</tt></b></dt>
1749 <dd>Convert a signed integer constant to the corresponding floating point
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]1750 constant. TYPE must be a scalar or vector floating point type. CST must be of
1751 scalar or vector integer type. Both CST and TYPE must be scalars, or vectors
1752 of the same number of elements. If the value won't fit in the floating point
1753 type, the results are undefined.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1754
1755 <dt><b><tt>ptrtoint ( CST to TYPE )</tt></b></dt>
1756 <dd>Convert a pointer typed constant to the corresponding integer constant
1757 TYPE must be an integer type. CST must be of pointer type. The CST value is
1758 zero extended, truncated, or unchanged to make it fit in TYPE.</dd>
1759
1760 <dt><b><tt>inttoptr ( CST to TYPE )</tt></b></dt>
1761 <dd>Convert a integer constant to a pointer constant. TYPE must be a
1762 pointer type. CST must be of integer type. The CST value is zero extended,
1763 truncated, or unchanged to make it fit in a pointer size. This one is
1764 <i>really</i> dangerous!</dd>
1765
1766 <dt><b><tt>bitcast ( CST to TYPE )</tt></b></dt>
1767 <dd>Convert a constant, CST, to another TYPE. The size of CST and TYPE must be
1768 identical (same number of bits). The conversion is done as if the CST value
1769 was stored to memory and read back as TYPE. In other words, no bits change
1770 with this operator, just the type. This can be used for conversion of
1771 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]1772 pointers it is only valid to cast to another pointer type. It is not valid
1773 to bitcast to or from an aggregate type.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1774 </dd>
1775
1776 <dt><b><tt>getelementptr ( CSTPTR, IDX0, IDX1, ... )</tt></b></dt>
1777
1778 <dd>Perform the <a href="#i_getelementptr">getelementptr operation</a> on
1779 constants. As with the <a href="#i_getelementptr">getelementptr</a>
1780 instruction, the index list may have zero or more indexes, which are required
1781 to make sense for the type of "CSTPTR".</dd>
1782
1783 <dt><b><tt>select ( COND, VAL1, VAL2 )</tt></b></dt>
1784
1785 <dd>Perform the <a href="#i_select">select operation</a> on
1786 constants.</dd>
1787
1788 <dt><b><tt>icmp COND ( VAL1, VAL2 )</tt></b></dt>
1789 <dd>Performs the <a href="#i_icmp">icmp operation</a> on constants.</dd>
1790
1791 <dt><b><tt>fcmp COND ( VAL1, VAL2 )</tt></b></dt>
1792 <dd>Performs the <a href="#i_fcmp">fcmp operation</a> on constants.</dd>
1793
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]1794 <dt><b><tt>vicmp COND ( VAL1, VAL2 )</tt></b></dt>
1795 <dd>Performs the <a href="#i_vicmp">vicmp operation</a> on constants.</dd>
1796
1797 <dt><b><tt>vfcmp COND ( VAL1, VAL2 )</tt></b></dt>
1798 <dd>Performs the <a href="#i_vfcmp">vfcmp operation</a> on constants.</dd>
1799
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1800 <dt><b><tt>extractelement ( VAL, IDX )</tt></b></dt>
1801
1802 <dd>Perform the <a href="#i_extractelement">extractelement
1803 operation</a> on constants.
1804
1805 <dt><b><tt>insertelement ( VAL, ELT, IDX )</tt></b></dt>
1806
1807 <dd>Perform the <a href="#i_insertelement">insertelement
1808 operation</a> on constants.</dd>
1809
1810
1811 <dt><b><tt>shufflevector ( VEC1, VEC2, IDXMASK )</tt></b></dt>
1812
1813 <dd>Perform the <a href="#i_shufflevector">shufflevector
1814 operation</a> on constants.</dd>
1815
1816 <dt><b><tt>OPCODE ( LHS, RHS )</tt></b></dt>
1817
1818 <dd>Perform the specified operation of the LHS and RHS constants. OPCODE may
1819 be any of the <a href="#binaryops">binary</a> or <a href="#bitwiseops">bitwise
1820 binary</a> operations. The constraints on operands are the same as those for
1821 the corresponding instruction (e.g. no bitwise operations on floating point
1822 values are allowed).</dd>
1823</dl>
1824</div>
1825
1826<!-- *********************************************************************** -->
1827<div class="doc_section"> <a name="othervalues">Other Values</a> </div>
1828<!-- *********************************************************************** -->
1829
1830<!-- ======================================================================= -->
1831<div class="doc_subsection">
1832<a name="inlineasm">Inline Assembler Expressions</a>
1833</div>
1834
1835<div class="doc_text">
1836
1837<p>
1838LLVM supports inline assembler expressions (as opposed to <a href="#moduleasm">
1839Module-Level Inline Assembly</a>) through the use of a special value. This
1840value represents the inline assembler as a string (containing the instructions
1841to emit), a list of operand constraints (stored as a string), and a flag that
1842indicates whether or not the inline asm expression has side effects. An example
1843inline assembler expression is:
1844</p>
1845
1846<div class="doc_code">
1847<pre>
1848i32 (i32) asm "bswap $0", "=r,r"
1849</pre>
1850</div>
1851
1852<p>
1853Inline assembler expressions may <b>only</b> be used as the callee operand of
1854a <a href="#i_call"><tt>call</tt> instruction</a>. Thus, typically we have:
1855</p>
1856
1857<div class="doc_code">
1858<pre>
1859%X = call i32 asm "<a href="#int_bswap">bswap</a> $0", "=r,r"(i32 %Y)
1860</pre>
1861</div>
1862
1863<p>
1864Inline asms with side effects not visible in the constraint list must be marked
1865as having side effects. This is done through the use of the
1866'<tt>sideeffect</tt>' keyword, like so:
1867</p>
1868
1869<div class="doc_code">
1870<pre>
1871call void asm sideeffect "eieio", ""()
1872</pre>
1873</div>
1874
1875<p>TODO: The format of the asm and constraints string still need to be
1876documented here. Constraints on what can be done (e.g. duplication, moving, etc
Chris Lattner1cdd64e2008-10-04 18:36:02 +0000[diff] [blame]1877need to be documented). This is probably best done by reference to another
1878document that covers inline asm from a holistic perspective.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1879</p>
1880
1881</div>
1882
1883<!-- *********************************************************************** -->
1884<div class="doc_section"> <a name="instref">Instruction Reference</a> </div>
1885<!-- *********************************************************************** -->
1886
1887<div class="doc_text">
1888
1889<p>The LLVM instruction set consists of several different
1890classifications of instructions: <a href="#terminators">terminator
1891instructions</a>, <a href="#binaryops">binary instructions</a>,
1892<a href="#bitwiseops">bitwise binary instructions</a>, <a
1893 href="#memoryops">memory instructions</a>, and <a href="#otherops">other
1894instructions</a>.</p>
1895
1896</div>
1897
1898<!-- ======================================================================= -->
1899<div class="doc_subsection"> <a name="terminators">Terminator
1900Instructions</a> </div>
1901
1902<div class="doc_text">
1903
1904<p>As mentioned <a href="#functionstructure">previously</a>, every
1905basic block in a program ends with a "Terminator" instruction, which
1906indicates which block should be executed after the current block is
1907finished. These terminator instructions typically yield a '<tt>void</tt>'
1908value: they produce control flow, not values (the one exception being
1909the '<a href="#i_invoke"><tt>invoke</tt></a>' instruction).</p>
1910<p>There are six different terminator instructions: the '<a
1911 href="#i_ret"><tt>ret</tt></a>' instruction, the '<a href="#i_br"><tt>br</tt></a>'
1912instruction, the '<a href="#i_switch"><tt>switch</tt></a>' instruction,
1913the '<a href="#i_invoke"><tt>invoke</tt></a>' instruction, the '<a
1914 href="#i_unwind"><tt>unwind</tt></a>' instruction, and the '<a
1915 href="#i_unreachable"><tt>unreachable</tt></a>' instruction.</p>
1916
1917</div>
1918
1919<!-- _______________________________________________________________________ -->
1920<div class="doc_subsubsection"> <a name="i_ret">'<tt>ret</tt>'
1921Instruction</a> </div>
1922<div class="doc_text">
1923<h5>Syntax:</h5>
Dan Gohman3e700032008-10-04 19:00:07 +0000[diff] [blame]1924<pre>
1925 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]1926 ret void <i>; Return from void function</i>
1927</pre>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1928
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1929<h5>Overview:</h5>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1930
Dan Gohman3e700032008-10-04 19:00:07 +0000[diff] [blame]1931<p>The '<tt>ret</tt>' instruction is used to return control flow (and
1932optionally a value) from a function back to the caller.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1933<p>There are two forms of the '<tt>ret</tt>' instruction: one that
Dan Gohman3e700032008-10-04 19:00:07 +0000[diff] [blame]1934returns a value and then causes control flow, and one that just causes
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1935control flow to occur.</p>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1936
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1937<h5>Arguments:</h5>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1938
Dan Gohman3e700032008-10-04 19:00:07 +0000[diff] [blame]1939<p>The '<tt>ret</tt>' instruction optionally accepts a single argument,
1940the return value. The type of the return value must be a
1941'<a href="#t_firstclass">first class</a>' type.</p>
1942
1943<p>A function is not <a href="#wellformed">well formed</a> if
1944it it has a non-void return type and contains a '<tt>ret</tt>'
1945instruction with no return value or a return value with a type that
1946does not match its type, or if it has a void return type and contains
1947a '<tt>ret</tt>' instruction with a return value.</p>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1948
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1949<h5>Semantics:</h5>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1950
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1951<p>When the '<tt>ret</tt>' instruction is executed, control flow
1952returns back to the calling function's context. If the caller is a "<a
1953 href="#i_call"><tt>call</tt></a>" instruction, execution continues at
1954the instruction after the call. If the caller was an "<a
1955 href="#i_invoke"><tt>invoke</tt></a>" instruction, execution continues
1956at the beginning of the "normal" destination block. If the instruction
1957returns a value, that value shall set the call or invoke instruction's
Dan Gohman3e700032008-10-04 19:00:07 +0000[diff] [blame]1958return value.
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1959
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1960<h5>Example:</h5>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1961
1962<pre>
1963 ret i32 5 <i>; Return an integer value of 5</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1964 ret void <i>; Return from a void function</i>
Dan Gohman3e700032008-10-04 19:00:07 +0000[diff] [blame]1965 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]1966</pre>
1967</div>
1968<!-- _______________________________________________________________________ -->
1969<div class="doc_subsubsection"> <a name="i_br">'<tt>br</tt>' Instruction</a> </div>
1970<div class="doc_text">
1971<h5>Syntax:</h5>
1972<pre> br i1 &lt;cond&gt;, label &lt;iftrue&gt;, label &lt;iffalse&gt;<br> br label &lt;dest&gt; <i>; Unconditional branch</i>
1973</pre>
1974<h5>Overview:</h5>
1975<p>The '<tt>br</tt>' instruction is used to cause control flow to
1976transfer to a different basic block in the current function. There are
1977two forms of this instruction, corresponding to a conditional branch
1978and an unconditional branch.</p>
1979<h5>Arguments:</h5>
1980<p>The conditional branch form of the '<tt>br</tt>' instruction takes a
1981single '<tt>i1</tt>' value and two '<tt>label</tt>' values. The
1982unconditional form of the '<tt>br</tt>' instruction takes a single
1983'<tt>label</tt>' value as a target.</p>
1984<h5>Semantics:</h5>
1985<p>Upon execution of a conditional '<tt>br</tt>' instruction, the '<tt>i1</tt>'
1986argument is evaluated. If the value is <tt>true</tt>, control flows
1987to the '<tt>iftrue</tt>' <tt>label</tt> argument. If "cond" is <tt>false</tt>,
1988control flows to the '<tt>iffalse</tt>' <tt>label</tt> argument.</p>
1989<h5>Example:</h5>
1990<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
1991 href="#i_ret">ret</a> i32 1<br>IfUnequal:<br> <a href="#i_ret">ret</a> i32 0<br></pre>
1992</div>
1993<!-- _______________________________________________________________________ -->
1994<div class="doc_subsubsection">
1995 <a name="i_switch">'<tt>switch</tt>' Instruction</a>
1996</div>
1997
1998<div class="doc_text">
1999<h5>Syntax:</h5>
2000
2001<pre>
2002 switch &lt;intty&gt; &lt;value&gt;, label &lt;defaultdest&gt; [ &lt;intty&gt; &lt;val&gt;, label &lt;dest&gt; ... ]
2003</pre>
2004
2005<h5>Overview:</h5>
2006
2007<p>The '<tt>switch</tt>' instruction is used to transfer control flow to one of
2008several different places. It is a generalization of the '<tt>br</tt>'
2009instruction, allowing a branch to occur to one of many possible
2010destinations.</p>
2011
2012
2013<h5>Arguments:</h5>
2014
2015<p>The '<tt>switch</tt>' instruction uses three parameters: an integer
2016comparison value '<tt>value</tt>', a default '<tt>label</tt>' destination, and
2017an array of pairs of comparison value constants and '<tt>label</tt>'s. The
2018table is not allowed to contain duplicate constant entries.</p>
2019
2020<h5>Semantics:</h5>
2021
2022<p>The <tt>switch</tt> instruction specifies a table of values and
2023destinations. When the '<tt>switch</tt>' instruction is executed, this
2024table is searched for the given value. If the value is found, control flow is
2025transfered to the corresponding destination; otherwise, control flow is
2026transfered to the default destination.</p>
2027
2028<h5>Implementation:</h5>
2029
2030<p>Depending on properties of the target machine and the particular
2031<tt>switch</tt> instruction, this instruction may be code generated in different
2032ways. For example, it could be generated as a series of chained conditional
2033branches or with a lookup table.</p>
2034
2035<h5>Example:</h5>
2036
2037<pre>
2038 <i>; Emulate a conditional br instruction</i>
2039 %Val = <a href="#i_zext">zext</a> i1 %value to i32
2040 switch i32 %Val, label %truedest [i32 0, label %falsedest ]
2041
2042 <i>; Emulate an unconditional br instruction</i>
2043 switch i32 0, label %dest [ ]
2044
2045 <i>; Implement a jump table:</i>
2046 switch i32 %val, label %otherwise [ i32 0, label %onzero
2047 i32 1, label %onone
2048 i32 2, label %ontwo ]
2049</pre>
2050</div>
2051
2052<!-- _______________________________________________________________________ -->
2053<div class="doc_subsubsection">
2054 <a name="i_invoke">'<tt>invoke</tt>' Instruction</a>
2055</div>
2056
2057<div class="doc_text">
2058
2059<h5>Syntax:</h5>
2060
2061<pre>
Devang Pateld0bfcc72008-10-07 17:48:33 +0000[diff] [blame]2062 &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]2063 to label &lt;normal label&gt; unwind label &lt;exception label&gt;
2064</pre>
2065
2066<h5>Overview:</h5>
2067
2068<p>The '<tt>invoke</tt>' instruction causes control to transfer to a specified
2069function, with the possibility of control flow transfer to either the
2070'<tt>normal</tt>' label or the
2071'<tt>exception</tt>' label. If the callee function returns with the
2072"<tt><a href="#i_ret">ret</a></tt>" instruction, control flow will return to the
2073"normal" label. If the callee (or any indirect callees) returns with the "<a
2074href="#i_unwind"><tt>unwind</tt></a>" instruction, control is interrupted and
Dan Gohman3e700032008-10-04 19:00:07 +0000[diff] [blame]2075continued at the dynamically nearest "exception" label.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2076
2077<h5>Arguments:</h5>
2078
2079<p>This instruction requires several arguments:</p>
2080
2081<ol>
2082 <li>
2083 The optional "cconv" marker indicates which <a href="#callingconv">calling
2084 convention</a> the call should use. If none is specified, the call defaults
2085 to using C calling conventions.
2086 </li>
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]2087
2088 <li>The optional <a href="#paramattrs">Parameter Attributes</a> list for
2089 return values. Only '<tt>zeroext</tt>', '<tt>signext</tt>',
2090 and '<tt>inreg</tt>' attributes are valid here.</li>
2091
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2092 <li>'<tt>ptr to function ty</tt>': shall be the signature of the pointer to
2093 function value being invoked. In most cases, this is a direct function
2094 invocation, but indirect <tt>invoke</tt>s are just as possible, branching off
2095 an arbitrary pointer to function value.
2096 </li>
2097
2098 <li>'<tt>function ptr val</tt>': An LLVM value containing a pointer to a
2099 function to be invoked. </li>
2100
2101 <li>'<tt>function args</tt>': argument list whose types match the function
2102 signature argument types. If the function signature indicates the function
2103 accepts a variable number of arguments, the extra arguments can be
2104 specified. </li>
2105
2106 <li>'<tt>normal label</tt>': the label reached when the called function
2107 executes a '<tt><a href="#i_ret">ret</a></tt>' instruction. </li>
2108
2109 <li>'<tt>exception label</tt>': the label reached when a callee returns with
2110 the <a href="#i_unwind"><tt>unwind</tt></a> instruction. </li>
2111
Devang Pateld0bfcc72008-10-07 17:48:33 +0000[diff] [blame]2112 <li>The optional <a href="#fnattrs">function attributes</a> list. Only
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]2113 '<tt>noreturn</tt>', '<tt>nounwind</tt>', '<tt>readonly</tt>' and
2114 '<tt>readnone</tt>' attributes are valid here.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2115</ol>
2116
2117<h5>Semantics:</h5>
2118
2119<p>This instruction is designed to operate as a standard '<tt><a
2120href="#i_call">call</a></tt>' instruction in most regards. The primary
2121difference is that it establishes an association with a label, which is used by
2122the runtime library to unwind the stack.</p>
2123
2124<p>This instruction is used in languages with destructors to ensure that proper
2125cleanup is performed in the case of either a <tt>longjmp</tt> or a thrown
2126exception. Additionally, this is important for implementation of
2127'<tt>catch</tt>' clauses in high-level languages that support them.</p>
2128
2129<h5>Example:</h5>
2130<pre>
Nick Lewyckya1c11a12008-03-16 07:18:12 +0000[diff] [blame]2131 %retval = invoke i32 @Test(i32 15) to label %Continue
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2132 unwind label %TestCleanup <i>; {i32}:retval set</i>
Nick Lewyckya1c11a12008-03-16 07:18:12 +0000[diff] [blame]2133 %retval = invoke <a href="#callingconv">coldcc</a> i32 %Testfnptr(i32 15) to label %Continue
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2134 unwind label %TestCleanup <i>; {i32}:retval set</i>
2135</pre>
2136</div>
2137
2138
2139<!-- _______________________________________________________________________ -->
2140
2141<div class="doc_subsubsection"> <a name="i_unwind">'<tt>unwind</tt>'
2142Instruction</a> </div>
2143
2144<div class="doc_text">
2145
2146<h5>Syntax:</h5>
2147<pre>
2148 unwind
2149</pre>
2150
2151<h5>Overview:</h5>
2152
2153<p>The '<tt>unwind</tt>' instruction unwinds the stack, continuing control flow
2154at the first callee in the dynamic call stack which used an <a
2155href="#i_invoke"><tt>invoke</tt></a> instruction to perform the call. This is
2156primarily used to implement exception handling.</p>
2157
2158<h5>Semantics:</h5>
2159
Chris Lattner8b094fc2008-04-19 21:01:16 +0000[diff] [blame]2160<p>The '<tt>unwind</tt>' instruction causes execution of the current function to
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2161immediately halt. The dynamic call stack is then searched for the first <a
2162href="#i_invoke"><tt>invoke</tt></a> instruction on the call stack. Once found,
2163execution continues at the "exceptional" destination block specified by the
2164<tt>invoke</tt> instruction. If there is no <tt>invoke</tt> instruction in the
2165dynamic call chain, undefined behavior results.</p>
2166</div>
2167
2168<!-- _______________________________________________________________________ -->
2169
2170<div class="doc_subsubsection"> <a name="i_unreachable">'<tt>unreachable</tt>'
2171Instruction</a> </div>
2172
2173<div class="doc_text">
2174
2175<h5>Syntax:</h5>
2176<pre>
2177 unreachable
2178</pre>
2179
2180<h5>Overview:</h5>
2181
2182<p>The '<tt>unreachable</tt>' instruction has no defined semantics. This
2183instruction is used to inform the optimizer that a particular portion of the
2184code is not reachable. This can be used to indicate that the code after a
2185no-return function cannot be reached, and other facts.</p>
2186
2187<h5>Semantics:</h5>
2188
2189<p>The '<tt>unreachable</tt>' instruction has no defined semantics.</p>
2190</div>
2191
2192
2193
2194<!-- ======================================================================= -->
2195<div class="doc_subsection"> <a name="binaryops">Binary Operations</a> </div>
2196<div class="doc_text">
2197<p>Binary operators are used to do most of the computation in a
Chris Lattnerab596d92008-04-01 18:47:32 +0000[diff] [blame]2198program. They require two operands of the same type, execute an operation on them, and
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2199produce a single value. The operands might represent
2200multiple 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]2201The result value has the same type as its operands.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2202<p>There are several different binary operators:</p>
2203</div>
2204<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2205<div class="doc_subsubsection">
2206 <a name="i_add">'<tt>add</tt>' Instruction</a>
2207</div>
2208
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2209<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2210
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2211<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2212
2213<pre>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2214 &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]2215</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2216
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2217<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2218
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2219<p>The '<tt>add</tt>' instruction returns the sum of its two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2220
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2221<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2222
2223<p>The two arguments to the '<tt>add</tt>' instruction must be <a
2224 href="#t_integer">integer</a>, <a href="#t_floating">floating point</a>, or
2225 <a href="#t_vector">vector</a> values. Both arguments must have identical
2226 types.</p>
2227
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2228<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2229
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2230<p>The value produced is the integer or floating point sum of the two
2231operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2232
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2233<p>If an integer sum has unsigned overflow, the result returned is the
2234mathematical result modulo 2<sup>n</sup>, where n is the bit width of
2235the result.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2236
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2237<p>Because LLVM integers use a two's complement representation, this
2238instruction is appropriate for both signed and unsigned integers.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2239
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2240<h5>Example:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2241
2242<pre>
2243 &lt;result&gt; = add i32 4, %var <i>; yields {i32}:result = 4 + %var</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2244</pre>
2245</div>
2246<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2247<div class="doc_subsubsection">
2248 <a name="i_sub">'<tt>sub</tt>' Instruction</a>
2249</div>
2250
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2251<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2252
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2253<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2254
2255<pre>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2256 &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]2257</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2258
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2259<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2260
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2261<p>The '<tt>sub</tt>' instruction returns the difference of its two
2262operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2263
2264<p>Note that the '<tt>sub</tt>' instruction is used to represent the
2265'<tt>neg</tt>' instruction present in most other intermediate
2266representations.</p>
2267
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2268<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2269
2270<p>The two arguments to the '<tt>sub</tt>' instruction must be <a
2271 href="#t_integer">integer</a>, <a href="#t_floating">floating point</a>,
2272 or <a href="#t_vector">vector</a> values. Both arguments must have identical
2273 types.</p>
2274
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2275<h5>Semantics:</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 value produced is the integer or floating point difference of
2278the two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2279
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2280<p>If an integer difference has unsigned overflow, the result returned is the
2281mathematical result modulo 2<sup>n</sup>, where n is the bit width of
2282the result.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2283
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2284<p>Because LLVM integers use a two's complement representation, this
2285instruction is appropriate for both signed and unsigned integers.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2286
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2287<h5>Example:</h5>
2288<pre>
2289 &lt;result&gt; = sub i32 4, %var <i>; yields {i32}:result = 4 - %var</i>
2290 &lt;result&gt; = sub i32 0, %val <i>; yields {i32}:result = -%var</i>
2291</pre>
2292</div>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2293
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2294<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2295<div class="doc_subsubsection">
2296 <a name="i_mul">'<tt>mul</tt>' Instruction</a>
2297</div>
2298
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2299<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2300
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2301<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2302<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]2303</pre>
2304<h5>Overview:</h5>
2305<p>The '<tt>mul</tt>' instruction returns the product of its two
2306operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2307
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2308<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2309
2310<p>The two arguments to the '<tt>mul</tt>' instruction must be <a
2311href="#t_integer">integer</a>, <a href="#t_floating">floating point</a>,
2312or <a href="#t_vector">vector</a> values. Both arguments must have identical
2313types.</p>
2314
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2315<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2316
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2317<p>The value produced is the integer or floating point product of the
2318two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2319
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2320<p>If the result of an integer multiplication has unsigned overflow,
2321the result returned is the mathematical result modulo
23222<sup>n</sup>, where n is the bit width of the result.</p>
2323<p>Because LLVM integers use a two's complement representation, and the
2324result is the same width as the operands, this instruction returns the
2325correct result for both signed and unsigned integers. If a full product
2326(e.g. <tt>i32</tt>x<tt>i32</tt>-><tt>i64</tt>) is needed, the operands
2327should be sign-extended or zero-extended as appropriate to the
2328width of the full product.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2329<h5>Example:</h5>
2330<pre> &lt;result&gt; = mul i32 4, %var <i>; yields {i32}:result = 4 * %var</i>
2331</pre>
2332</div>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2333
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2334<!-- _______________________________________________________________________ -->
2335<div class="doc_subsubsection"> <a name="i_udiv">'<tt>udiv</tt>' Instruction
2336</a></div>
2337<div class="doc_text">
2338<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2339<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]2340</pre>
2341<h5>Overview:</h5>
2342<p>The '<tt>udiv</tt>' instruction returns the quotient of its two
2343operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2344
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2345<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2346
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2347<p>The two arguments to the '<tt>udiv</tt>' instruction must be
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2348<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2349values. Both arguments must have identical types.</p>
2350
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2351<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2352
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2353<p>The value produced is the unsigned integer quotient of the two operands.</p>
2354<p>Note that unsigned integer division and signed integer division are distinct
2355operations; for signed integer division, use '<tt>sdiv</tt>'.</p>
2356<p>Division by zero leads to undefined behavior.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2357<h5>Example:</h5>
2358<pre> &lt;result&gt; = udiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
2359</pre>
2360</div>
2361<!-- _______________________________________________________________________ -->
2362<div class="doc_subsubsection"> <a name="i_sdiv">'<tt>sdiv</tt>' Instruction
2363</a> </div>
2364<div class="doc_text">
2365<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2366<pre>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2367 &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]2368</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2369
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2370<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2371
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2372<p>The '<tt>sdiv</tt>' instruction returns the quotient of its two
2373operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2374
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2375<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2376
2377<p>The two arguments to the '<tt>sdiv</tt>' instruction must be
2378<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2379values. Both arguments must have identical types.</p>
2380
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2381<h5>Semantics:</h5>
Chris Lattnerdc15b1d2008-04-01 18:45:27 +0000[diff] [blame]2382<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]2383<p>Note that signed integer division and unsigned integer division are distinct
2384operations; for unsigned integer division, use '<tt>udiv</tt>'.</p>
2385<p>Division by zero leads to undefined behavior. Overflow also leads to
2386undefined behavior; this is a rare case, but can occur, for example,
2387by doing a 32-bit division of -2147483648 by -1.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2388<h5>Example:</h5>
2389<pre> &lt;result&gt; = sdiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
2390</pre>
2391</div>
2392<!-- _______________________________________________________________________ -->
2393<div class="doc_subsubsection"> <a name="i_fdiv">'<tt>fdiv</tt>'
2394Instruction</a> </div>
2395<div class="doc_text">
2396<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2397<pre>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2398 &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]2399</pre>
2400<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2401
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2402<p>The '<tt>fdiv</tt>' instruction returns the quotient of its two
2403operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2404
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2405<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2406
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2407<p>The two arguments to the '<tt>fdiv</tt>' instruction must be
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2408<a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a>
2409of floating point values. Both arguments must have identical types.</p>
2410
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2411<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2412
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2413<p>The value produced is the floating point quotient of the two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2414
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2415<h5>Example:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2416
2417<pre>
2418 &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]2419</pre>
2420</div>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2421
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2422<!-- _______________________________________________________________________ -->
2423<div class="doc_subsubsection"> <a name="i_urem">'<tt>urem</tt>' Instruction</a>
2424</div>
2425<div class="doc_text">
2426<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2427<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]2428</pre>
2429<h5>Overview:</h5>
2430<p>The '<tt>urem</tt>' instruction returns the remainder from the
2431unsigned division of its two arguments.</p>
2432<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2433<p>The two arguments to the '<tt>urem</tt>' instruction must be
2434<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2435values. Both arguments must have identical types.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2436<h5>Semantics:</h5>
2437<p>This instruction returns the unsigned integer <i>remainder</i> of a division.
Chris Lattnerdc15b1d2008-04-01 18:45:27 +0000[diff] [blame]2438This instruction always performs an unsigned division to get the remainder.</p>
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2439<p>Note that unsigned integer remainder and signed integer remainder are
2440distinct operations; for signed integer remainder, use '<tt>srem</tt>'.</p>
2441<p>Taking the remainder of a division by zero leads to undefined behavior.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2442<h5>Example:</h5>
2443<pre> &lt;result&gt; = urem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
2444</pre>
2445
2446</div>
2447<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2448<div class="doc_subsubsection">
2449 <a name="i_srem">'<tt>srem</tt>' Instruction</a>
2450</div>
2451
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2452<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2453
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2454<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2455
2456<pre>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2457 &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]2458</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2459
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2460<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2461
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2462<p>The '<tt>srem</tt>' instruction returns the remainder from the
Dan Gohman3e3fd8c2007-11-05 23:35:22 +0000[diff] [blame]2463signed division of its two operands. This instruction can also take
2464<a href="#t_vector">vector</a> versions of the values in which case
2465the elements must be integers.</p>
Chris Lattner08497ce2008-01-04 04:33:49 +0000[diff] [blame]2466
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2467<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2468
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2469<p>The two arguments to the '<tt>srem</tt>' instruction must be
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2470<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2471values. Both arguments must have identical types.</p>
2472
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2473<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2474
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2475<p>This instruction returns the <i>remainder</i> of a division (where the result
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2476has the same sign as the dividend, <tt>op1</tt>), not the <i>modulo</i>
2477operator (where the result has the same sign as the divisor, <tt>op2</tt>) of
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2478a value. For more information about the difference, see <a
2479 href="http://mathforum.org/dr.math/problems/anne.4.28.99.html">The
2480Math Forum</a>. For a table of how this is implemented in various languages,
2481please see <a href="http://en.wikipedia.org/wiki/Modulo_operation">
2482Wikipedia: modulo operation</a>.</p>
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2483<p>Note that signed integer remainder and unsigned integer remainder are
2484distinct operations; for unsigned integer remainder, use '<tt>urem</tt>'.</p>
2485<p>Taking the remainder of a division by zero leads to undefined behavior.
2486Overflow also leads to undefined behavior; this is a rare case, but can occur,
2487for example, by taking the remainder of a 32-bit division of -2147483648 by -1.
2488(The remainder doesn't actually overflow, but this rule lets srem be
2489implemented using instructions that return both the result of the division
2490and the remainder.)</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2491<h5>Example:</h5>
2492<pre> &lt;result&gt; = srem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
2493</pre>
2494
2495</div>
2496<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2497<div class="doc_subsubsection">
2498 <a name="i_frem">'<tt>frem</tt>' Instruction</a> </div>
2499
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2500<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2501
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2502<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2503<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]2504</pre>
2505<h5>Overview:</h5>
2506<p>The '<tt>frem</tt>' instruction returns the remainder from the
2507division of its two operands.</p>
2508<h5>Arguments:</h5>
2509<p>The two arguments to the '<tt>frem</tt>' instruction must be
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2510<a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a>
2511of floating point values. Both arguments must have identical types.</p>
2512
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2513<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2514
Chris Lattnerdc15b1d2008-04-01 18:45:27 +0000[diff] [blame]2515<p>This instruction returns the <i>remainder</i> of a division.
2516The remainder has the same sign as the dividend.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2517
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2518<h5>Example:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2519
2520<pre>
2521 &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]2522</pre>
2523</div>
2524
2525<!-- ======================================================================= -->
2526<div class="doc_subsection"> <a name="bitwiseops">Bitwise Binary
2527Operations</a> </div>
2528<div class="doc_text">
2529<p>Bitwise binary operators are used to do various forms of
2530bit-twiddling in a program. They are generally very efficient
2531instructions and can commonly be strength reduced from other
Chris Lattnerdc15b1d2008-04-01 18:45:27 +0000[diff] [blame]2532instructions. They require two operands of the same type, execute an operation on them,
2533and 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]2534</div>
2535
2536<!-- _______________________________________________________________________ -->
2537<div class="doc_subsubsection"> <a name="i_shl">'<tt>shl</tt>'
2538Instruction</a> </div>
2539<div class="doc_text">
2540<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2541<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]2542</pre>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2543
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2544<h5>Overview:</h5>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2545
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2546<p>The '<tt>shl</tt>' instruction returns the first operand shifted to
2547the left a specified number of bits.</p>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2548
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2549<h5>Arguments:</h5>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2550
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2551<p>Both arguments to the '<tt>shl</tt>' instruction must be the same <a
Nate Begemanbb1ce942008-07-29 15:49:41 +0000[diff] [blame]2552 href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2553type. '<tt>op2</tt>' is treated as an unsigned value.</p>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2554
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2555<h5>Semantics:</h5>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2556
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2557<p>The value produced is <tt>op1</tt> * 2<sup><tt>op2</tt></sup> mod 2<sup>n</sup>,
2558where n is the width of the result. If <tt>op2</tt> is (statically or dynamically) negative or
2559equal 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]2560
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2561<h5>Example:</h5><pre>
2562 &lt;result&gt; = shl i32 4, %var <i>; yields {i32}: 4 &lt;&lt; %var</i>
2563 &lt;result&gt; = shl i32 4, 2 <i>; yields {i32}: 16</i>
2564 &lt;result&gt; = shl i32 1, 10 <i>; yields {i32}: 1024</i>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2565 &lt;result&gt; = shl i32 1, 32 <i>; undefined</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2566</pre>
2567</div>
2568<!-- _______________________________________________________________________ -->
2569<div class="doc_subsubsection"> <a name="i_lshr">'<tt>lshr</tt>'
2570Instruction</a> </div>
2571<div class="doc_text">
2572<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2573<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]2574</pre>
2575
2576<h5>Overview:</h5>
2577<p>The '<tt>lshr</tt>' instruction (logical shift right) returns the first
2578operand shifted to the right a specified number of bits with zero fill.</p>
2579
2580<h5>Arguments:</h5>
2581<p>Both arguments to the '<tt>lshr</tt>' instruction must be the same
Nate Begemanbb1ce942008-07-29 15:49:41 +0000[diff] [blame]2582<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]2583type. '<tt>op2</tt>' is treated as an unsigned value.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2584
2585<h5>Semantics:</h5>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2586
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2587<p>This instruction always performs a logical shift right operation. The most
2588significant bits of the result will be filled with zero bits after the
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2589shift. If <tt>op2</tt> is (statically or dynamically) equal to or larger than
2590the number of bits in <tt>op1</tt>, the result is undefined.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2591
2592<h5>Example:</h5>
2593<pre>
2594 &lt;result&gt; = lshr i32 4, 1 <i>; yields {i32}:result = 2</i>
2595 &lt;result&gt; = lshr i32 4, 2 <i>; yields {i32}:result = 1</i>
2596 &lt;result&gt; = lshr i8 4, 3 <i>; yields {i8}:result = 0</i>
2597 &lt;result&gt; = lshr i8 -2, 1 <i>; yields {i8}:result = 0x7FFFFFFF </i>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2598 &lt;result&gt; = lshr i32 1, 32 <i>; undefined</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2599</pre>
2600</div>
2601
2602<!-- _______________________________________________________________________ -->
2603<div class="doc_subsubsection"> <a name="i_ashr">'<tt>ashr</tt>'
2604Instruction</a> </div>
2605<div class="doc_text">
2606
2607<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2608<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]2609</pre>
2610
2611<h5>Overview:</h5>
2612<p>The '<tt>ashr</tt>' instruction (arithmetic shift right) returns the first
2613operand shifted to the right a specified number of bits with sign extension.</p>
2614
2615<h5>Arguments:</h5>
2616<p>Both arguments to the '<tt>ashr</tt>' instruction must be the same
Nate Begemanbb1ce942008-07-29 15:49:41 +0000[diff] [blame]2617<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]2618type. '<tt>op2</tt>' is treated as an unsigned value.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2619
2620<h5>Semantics:</h5>
2621<p>This instruction always performs an arithmetic shift right operation,
2622The most significant bits of the result will be filled with the sign bit
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2623of <tt>op1</tt>. If <tt>op2</tt> is (statically or dynamically) equal to or
2624larger than the number of bits in <tt>op1</tt>, the result is undefined.
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2625</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2626
2627<h5>Example:</h5>
2628<pre>
2629 &lt;result&gt; = ashr i32 4, 1 <i>; yields {i32}:result = 2</i>
2630 &lt;result&gt; = ashr i32 4, 2 <i>; yields {i32}:result = 1</i>
2631 &lt;result&gt; = ashr i8 4, 3 <i>; yields {i8}:result = 0</i>
2632 &lt;result&gt; = ashr i8 -2, 1 <i>; yields {i8}:result = -1</i>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2633 &lt;result&gt; = ashr i32 1, 32 <i>; undefined</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2634</pre>
2635</div>
2636
2637<!-- _______________________________________________________________________ -->
2638<div class="doc_subsubsection"> <a name="i_and">'<tt>and</tt>'
2639Instruction</a> </div>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2640
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2641<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2642
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2643<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2644
2645<pre>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2646 &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]2647</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2648
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2649<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2650
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2651<p>The '<tt>and</tt>' instruction returns the bitwise logical and of
2652its two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2653
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2654<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2655
2656<p>The two arguments to the '<tt>and</tt>' instruction must be
2657<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2658values. Both arguments must have identical types.</p>
2659
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2660<h5>Semantics:</h5>
2661<p>The truth table used for the '<tt>and</tt>' instruction is:</p>
2662<p> </p>
Bill Wendling1e19eac2008-09-07 10:29:20 +0000[diff] [blame]2663<div>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2664<table border="1" cellspacing="0" cellpadding="4">
2665 <tbody>
2666 <tr>
2667 <td>In0</td>
2668 <td>In1</td>
2669 <td>Out</td>
2670 </tr>
2671 <tr>
2672 <td>0</td>
2673 <td>0</td>
2674 <td>0</td>
2675 </tr>
2676 <tr>
2677 <td>0</td>
2678 <td>1</td>
2679 <td>0</td>
2680 </tr>
2681 <tr>
2682 <td>1</td>
2683 <td>0</td>
2684 <td>0</td>
2685 </tr>
2686 <tr>
2687 <td>1</td>
2688 <td>1</td>
2689 <td>1</td>
2690 </tr>
2691 </tbody>
2692</table>
2693</div>
2694<h5>Example:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2695<pre>
2696 &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]2697 &lt;result&gt; = and i32 15, 40 <i>; yields {i32}:result = 8</i>
2698 &lt;result&gt; = and i32 4, 8 <i>; yields {i32}:result = 0</i>
2699</pre>
2700</div>
2701<!-- _______________________________________________________________________ -->
2702<div class="doc_subsubsection"> <a name="i_or">'<tt>or</tt>' Instruction</a> </div>
2703<div class="doc_text">
2704<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2705<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]2706</pre>
2707<h5>Overview:</h5>
2708<p>The '<tt>or</tt>' instruction returns the bitwise logical inclusive
2709or of its two operands.</p>
2710<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2711
2712<p>The two arguments to the '<tt>or</tt>' instruction must be
2713<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2714values. Both arguments must have identical types.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2715<h5>Semantics:</h5>
2716<p>The truth table used for the '<tt>or</tt>' instruction is:</p>
2717<p> </p>
Bill Wendling1e19eac2008-09-07 10:29:20 +0000[diff] [blame]2718<div>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2719<table border="1" cellspacing="0" cellpadding="4">
2720 <tbody>
2721 <tr>
2722 <td>In0</td>
2723 <td>In1</td>
2724 <td>Out</td>
2725 </tr>
2726 <tr>
2727 <td>0</td>
2728 <td>0</td>
2729 <td>0</td>
2730 </tr>
2731 <tr>
2732 <td>0</td>
2733 <td>1</td>
2734 <td>1</td>
2735 </tr>
2736 <tr>
2737 <td>1</td>
2738 <td>0</td>
2739 <td>1</td>
2740 </tr>
2741 <tr>
2742 <td>1</td>
2743 <td>1</td>
2744 <td>1</td>
2745 </tr>
2746 </tbody>
2747</table>
2748</div>
2749<h5>Example:</h5>
2750<pre> &lt;result&gt; = or i32 4, %var <i>; yields {i32}:result = 4 | %var</i>
2751 &lt;result&gt; = or i32 15, 40 <i>; yields {i32}:result = 47</i>
2752 &lt;result&gt; = or i32 4, 8 <i>; yields {i32}:result = 12</i>
2753</pre>
2754</div>
2755<!-- _______________________________________________________________________ -->
2756<div class="doc_subsubsection"> <a name="i_xor">'<tt>xor</tt>'
2757Instruction</a> </div>
2758<div class="doc_text">
2759<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2760<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]2761</pre>
2762<h5>Overview:</h5>
2763<p>The '<tt>xor</tt>' instruction returns the bitwise logical exclusive
2764or of its two operands. The <tt>xor</tt> is used to implement the
2765"one's complement" operation, which is the "~" operator in C.</p>
2766<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2767<p>The two arguments to the '<tt>xor</tt>' instruction must be
2768<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2769values. Both arguments must have identical types.</p>
2770
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2771<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2772
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2773<p>The truth table used for the '<tt>xor</tt>' instruction is:</p>
2774<p> </p>
Bill Wendling1e19eac2008-09-07 10:29:20 +0000[diff] [blame]2775<div>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2776<table border="1" cellspacing="0" cellpadding="4">
2777 <tbody>
2778 <tr>
2779 <td>In0</td>
2780 <td>In1</td>
2781 <td>Out</td>
2782 </tr>
2783 <tr>
2784 <td>0</td>
2785 <td>0</td>
2786 <td>0</td>
2787 </tr>
2788 <tr>
2789 <td>0</td>
2790 <td>1</td>
2791 <td>1</td>
2792 </tr>
2793 <tr>
2794 <td>1</td>
2795 <td>0</td>
2796 <td>1</td>
2797 </tr>
2798 <tr>
2799 <td>1</td>
2800 <td>1</td>
2801 <td>0</td>
2802 </tr>
2803 </tbody>
2804</table>
2805</div>
2806<p> </p>
2807<h5>Example:</h5>
2808<pre> &lt;result&gt; = xor i32 4, %var <i>; yields {i32}:result = 4 ^ %var</i>
2809 &lt;result&gt; = xor i32 15, 40 <i>; yields {i32}:result = 39</i>
2810 &lt;result&gt; = xor i32 4, 8 <i>; yields {i32}:result = 12</i>
2811 &lt;result&gt; = xor i32 %V, -1 <i>; yields {i32}:result = ~%V</i>
2812</pre>
2813</div>
2814
2815<!-- ======================================================================= -->
2816<div class="doc_subsection">
2817 <a name="vectorops">Vector Operations</a>
2818</div>
2819
2820<div class="doc_text">
2821
2822<p>LLVM supports several instructions to represent vector operations in a
2823target-independent manner. These instructions cover the element-access and
2824vector-specific operations needed to process vectors effectively. While LLVM
2825does directly support these vector operations, many sophisticated algorithms
2826will want to use target-specific intrinsics to take full advantage of a specific
2827target.</p>
2828
2829</div>
2830
2831<!-- _______________________________________________________________________ -->
2832<div class="doc_subsubsection">
2833 <a name="i_extractelement">'<tt>extractelement</tt>' Instruction</a>
2834</div>
2835
2836<div class="doc_text">
2837
2838<h5>Syntax:</h5>
2839
2840<pre>
2841 &lt;result&gt; = extractelement &lt;n x &lt;ty&gt;&gt; &lt;val&gt;, i32 &lt;idx&gt; <i>; yields &lt;ty&gt;</i>
2842</pre>
2843
2844<h5>Overview:</h5>
2845
2846<p>
2847The '<tt>extractelement</tt>' instruction extracts a single scalar
2848element from a vector at a specified index.
2849</p>
2850
2851
2852<h5>Arguments:</h5>
2853
2854<p>
2855The first operand of an '<tt>extractelement</tt>' instruction is a
2856value of <a href="#t_vector">vector</a> type. The second operand is
2857an index indicating the position from which to extract the element.
2858The index may be a variable.</p>
2859
2860<h5>Semantics:</h5>
2861
2862<p>
2863The result is a scalar of the same type as the element type of
2864<tt>val</tt>. Its value is the value at position <tt>idx</tt> of
2865<tt>val</tt>. If <tt>idx</tt> exceeds the length of <tt>val</tt>, the
2866results are undefined.
2867</p>
2868
2869<h5>Example:</h5>
2870
2871<pre>
2872 %result = extractelement &lt;4 x i32&gt; %vec, i32 0 <i>; yields i32</i>
2873</pre>
2874</div>
2875
2876
2877<!-- _______________________________________________________________________ -->
2878<div class="doc_subsubsection">
2879 <a name="i_insertelement">'<tt>insertelement</tt>' Instruction</a>
2880</div>
2881
2882<div class="doc_text">
2883
2884<h5>Syntax:</h5>
2885
2886<pre>
Dan Gohmanbcc3c502008-05-12 23:38:42 +0000[diff] [blame]2887 &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]2888</pre>
2889
2890<h5>Overview:</h5>
2891
2892<p>
2893The '<tt>insertelement</tt>' instruction inserts a scalar
2894element into a vector at a specified index.
2895</p>
2896
2897
2898<h5>Arguments:</h5>
2899
2900<p>
2901The first operand of an '<tt>insertelement</tt>' instruction is a
2902value of <a href="#t_vector">vector</a> type. The second operand is a
2903scalar value whose type must equal the element type of the first
2904operand. The third operand is an index indicating the position at
2905which to insert the value. The index may be a variable.</p>
2906
2907<h5>Semantics:</h5>
2908
2909<p>
2910The result is a vector of the same type as <tt>val</tt>. Its
2911element values are those of <tt>val</tt> except at position
2912<tt>idx</tt>, where it gets the value <tt>elt</tt>. If <tt>idx</tt>
2913exceeds the length of <tt>val</tt>, the results are undefined.
2914</p>
2915
2916<h5>Example:</h5>
2917
2918<pre>
2919 %result = insertelement &lt;4 x i32&gt; %vec, i32 1, i32 0 <i>; yields &lt;4 x i32&gt;</i>
2920</pre>
2921</div>
2922
2923<!-- _______________________________________________________________________ -->
2924<div class="doc_subsubsection">
2925 <a name="i_shufflevector">'<tt>shufflevector</tt>' Instruction</a>
2926</div>
2927
2928<div class="doc_text">
2929
2930<h5>Syntax:</h5>
2931
2932<pre>
2933 &lt;result&gt; = shufflevector &lt;n x &lt;ty&gt;&gt; &lt;v1&gt;, &lt;n x &lt;ty&gt;&gt; &lt;v2&gt;, &lt;n x i32&gt; &lt;mask&gt; <i>; yields &lt;n x &lt;ty&gt;&gt;</i>
2934</pre>
2935
2936<h5>Overview:</h5>
2937
2938<p>
2939The '<tt>shufflevector</tt>' instruction constructs a permutation of elements
2940from two input vectors, returning a vector of the same type.
2941</p>
2942
2943<h5>Arguments:</h5>
2944
2945<p>
2946The first two operands of a '<tt>shufflevector</tt>' instruction are vectors
2947with types that match each other and types that match the result of the
2948instruction. The third argument is a shuffle mask, which has the same number
2949of elements as the other vector type, but whose element type is always 'i32'.
2950</p>
2951
2952<p>
2953The shuffle mask operand is required to be a constant vector with either
2954constant integer or undef values.
2955</p>
2956
2957<h5>Semantics:</h5>
2958
2959<p>
2960The elements of the two input vectors are numbered from left to right across
2961both of the vectors. The shuffle mask operand specifies, for each element of
2962the result vector, which element of the two input registers the result element
2963gets. The element selector may be undef (meaning "don't care") and the second
2964operand may be undef if performing a shuffle from only one vector.
2965</p>
2966
2967<h5>Example:</h5>
2968
2969<pre>
2970 %result = shufflevector &lt;4 x i32&gt; %v1, &lt;4 x i32&gt; %v2,
2971 &lt;4 x i32&gt; &lt;i32 0, i32 4, i32 1, i32 5&gt; <i>; yields &lt;4 x i32&gt;</i>
2972 %result = shufflevector &lt;4 x i32&gt; %v1, &lt;4 x i32&gt; undef,
2973 &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.
2974</pre>
2975</div>
2976
2977
2978<!-- ======================================================================= -->
2979<div class="doc_subsection">
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]2980 <a name="aggregateops">Aggregate Operations</a>
2981</div>
2982
2983<div class="doc_text">
2984
2985<p>LLVM supports several instructions for working with aggregate values.
2986</p>
2987
2988</div>
2989
2990<!-- _______________________________________________________________________ -->
2991<div class="doc_subsubsection">
2992 <a name="i_extractvalue">'<tt>extractvalue</tt>' Instruction</a>
2993</div>
2994
2995<div class="doc_text">
2996
2997<h5>Syntax:</h5>
2998
2999<pre>
3000 &lt;result&gt; = extractvalue &lt;aggregate type&gt; &lt;val&gt;, &lt;idx&gt;{, &lt;idx&gt;}*
3001</pre>
3002
3003<h5>Overview:</h5>
3004
3005<p>
Dan Gohman6c6dea02008-05-13 18:16:06 +0000[diff] [blame]3006The '<tt>extractvalue</tt>' instruction extracts the value of a struct field
3007or array element from an aggregate value.
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3008</p>
3009
3010
3011<h5>Arguments:</h5>
3012
3013<p>
3014The first operand of an '<tt>extractvalue</tt>' instruction is a
3015value of <a href="#t_struct">struct</a> or <a href="#t_array">array</a>
Dan Gohman6c6dea02008-05-13 18:16:06 +0000[diff] [blame]3016type. The operands are constant indices to specify which value to extract
Dan Gohmane5febe42008-05-31 00:58:22 +0000[diff] [blame]3017in a similar manner as indices in a
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3018'<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.
3019</p>
3020
3021<h5>Semantics:</h5>
3022
3023<p>
3024The result is the value at the position in the aggregate specified by
3025the index operands.
3026</p>
3027
3028<h5>Example:</h5>
3029
3030<pre>
Dan Gohmane5febe42008-05-31 00:58:22 +0000[diff] [blame]3031 %result = extractvalue {i32, float} %agg, 0 <i>; yields i32</i>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3032</pre>
3033</div>
3034
3035
3036<!-- _______________________________________________________________________ -->
3037<div class="doc_subsubsection">
3038 <a name="i_insertvalue">'<tt>insertvalue</tt>' Instruction</a>
3039</div>
3040
3041<div class="doc_text">
3042
3043<h5>Syntax:</h5>
3044
3045<pre>
Dan Gohmane5febe42008-05-31 00:58:22 +0000[diff] [blame]3046 &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]3047</pre>
3048
3049<h5>Overview:</h5>
3050
3051<p>
3052The '<tt>insertvalue</tt>' instruction inserts a value
Dan Gohman6c6dea02008-05-13 18:16:06 +0000[diff] [blame]3053into a struct field or array element in an aggregate.
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3054</p>
3055
3056
3057<h5>Arguments:</h5>
3058
3059<p>
3060The first operand of an '<tt>insertvalue</tt>' instruction is a
3061value of <a href="#t_struct">struct</a> or <a href="#t_array">array</a> type.
3062The second operand is a first-class value to insert.
Dan Gohman4f29e422008-05-23 21:53:15 +0000[diff] [blame]3063The following operands are constant indices
Dan Gohmane5febe42008-05-31 00:58:22 +0000[diff] [blame]3064indicating the position at which to insert the value in a similar manner as
Dan Gohman6c6dea02008-05-13 18:16:06 +0000[diff] [blame]3065indices in a
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3066'<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.
3067The value to insert must have the same type as the value identified
Dan Gohman6c6dea02008-05-13 18:16:06 +0000[diff] [blame]3068by the indices.
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3069
3070<h5>Semantics:</h5>
3071
3072<p>
3073The result is an aggregate of the same type as <tt>val</tt>. Its
3074value is that of <tt>val</tt> except that the value at the position
Dan Gohman6c6dea02008-05-13 18:16:06 +0000[diff] [blame]3075specified by the indices is that of <tt>elt</tt>.
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3076</p>
3077
3078<h5>Example:</h5>
3079
3080<pre>
Dan Gohmanb1aab4e2008-06-23 15:26:37 +0000[diff] [blame]3081 %result = insertvalue {i32, float} %agg, i32 1, 0 <i>; yields {i32, float}</i>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3082</pre>
3083</div>
3084
3085
3086<!-- ======================================================================= -->
3087<div class="doc_subsection">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3088 <a name="memoryops">Memory Access and Addressing Operations</a>
3089</div>
3090
3091<div class="doc_text">
3092
3093<p>A key design point of an SSA-based representation is how it
3094represents memory. In LLVM, no memory locations are in SSA form, which
3095makes things very simple. This section describes how to read, write,
3096allocate, and free memory in LLVM.</p>
3097
3098</div>
3099
3100<!-- _______________________________________________________________________ -->
3101<div class="doc_subsubsection">
3102 <a name="i_malloc">'<tt>malloc</tt>' Instruction</a>
3103</div>
3104
3105<div class="doc_text">
3106
3107<h5>Syntax:</h5>
3108
3109<pre>
3110 &lt;result&gt; = malloc &lt;type&gt;[, i32 &lt;NumElements&gt;][, align &lt;alignment&gt;] <i>; yields {type*}:result</i>
3111</pre>
3112
3113<h5>Overview:</h5>
3114
3115<p>The '<tt>malloc</tt>' instruction allocates memory from the system
Christopher Lambcfe00962007-12-17 01:00:21 +0000[diff] [blame]3116heap and returns a pointer to it. The object is always allocated in the generic
3117address space (address space zero).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3118
3119<h5>Arguments:</h5>
3120
3121<p>The '<tt>malloc</tt>' instruction allocates
3122<tt>sizeof(&lt;type&gt;)*NumElements</tt>
3123bytes of memory from the operating system and returns a pointer of the
3124appropriate type to the program. If "NumElements" is specified, it is the
Gabor Greif5082cf42008-02-09 22:24:34 +0000[diff] [blame]3125number of elements allocated, otherwise "NumElements" is defaulted to be one.
Chris Lattner10368b62008-04-02 00:38:26 +0000[diff] [blame]3126If a constant alignment is specified, the value result of the allocation is guaranteed to
Gabor Greif5082cf42008-02-09 22:24:34 +0000[diff] [blame]3127be aligned to at least that boundary. If not specified, or if zero, the target can
3128choose to align the allocation on any convenient boundary.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3129
3130<p>'<tt>type</tt>' must be a sized type.</p>
3131
3132<h5>Semantics:</h5>
3133
3134<p>Memory is allocated using the system "<tt>malloc</tt>" function, and
Chris Lattner8b094fc2008-04-19 21:01:16 +0000[diff] [blame]3135a pointer is returned. The result of a zero byte allocattion is undefined. The
3136result is null if there is insufficient memory available.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3137
3138<h5>Example:</h5>
3139
3140<pre>
3141 %array = malloc [4 x i8 ] <i>; yields {[%4 x i8]*}:array</i>
3142
3143 %size = <a href="#i_add">add</a> i32 2, 2 <i>; yields {i32}:size = i32 4</i>
3144 %array1 = malloc i8, i32 4 <i>; yields {i8*}:array1</i>
3145 %array2 = malloc [12 x i8], i32 %size <i>; yields {[12 x i8]*}:array2</i>
3146 %array3 = malloc i32, i32 4, align 1024 <i>; yields {i32*}:array3</i>
3147 %array4 = malloc i32, align 1024 <i>; yields {i32*}:array4</i>
3148</pre>
3149</div>
3150
3151<!-- _______________________________________________________________________ -->
3152<div class="doc_subsubsection">
3153 <a name="i_free">'<tt>free</tt>' Instruction</a>
3154</div>
3155
3156<div class="doc_text">
3157
3158<h5>Syntax:</h5>
3159
3160<pre>
3161 free &lt;type&gt; &lt;value&gt; <i>; yields {void}</i>
3162</pre>
3163
3164<h5>Overview:</h5>
3165
3166<p>The '<tt>free</tt>' instruction returns memory back to the unused
3167memory heap to be reallocated in the future.</p>
3168
3169<h5>Arguments:</h5>
3170
3171<p>'<tt>value</tt>' shall be a pointer value that points to a value
3172that was allocated with the '<tt><a href="#i_malloc">malloc</a></tt>'
3173instruction.</p>
3174
3175<h5>Semantics:</h5>
3176
3177<p>Access to the memory pointed to by the pointer is no longer defined
Chris Lattner329d6532008-04-19 22:41:32 +0000[diff] [blame]3178after this instruction executes. If the pointer is null, the operation
3179is a noop.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3180
3181<h5>Example:</h5>
3182
3183<pre>
3184 %array = <a href="#i_malloc">malloc</a> [4 x i8] <i>; yields {[4 x i8]*}:array</i>
3185 free [4 x i8]* %array
3186</pre>
3187</div>
3188
3189<!-- _______________________________________________________________________ -->
3190<div class="doc_subsubsection">
3191 <a name="i_alloca">'<tt>alloca</tt>' Instruction</a>
3192</div>
3193
3194<div class="doc_text">
3195
3196<h5>Syntax:</h5>
3197
3198<pre>
3199 &lt;result&gt; = alloca &lt;type&gt;[, i32 &lt;NumElements&gt;][, align &lt;alignment&gt;] <i>; yields {type*}:result</i>
3200</pre>
3201
3202<h5>Overview:</h5>
3203
3204<p>The '<tt>alloca</tt>' instruction allocates memory on the stack frame of the
3205currently executing function, to be automatically released when this function
Christopher Lambcfe00962007-12-17 01:00:21 +0000[diff] [blame]3206returns to its caller. The object is always allocated in the generic address
3207space (address space zero).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3208
3209<h5>Arguments:</h5>
3210
3211<p>The '<tt>alloca</tt>' instruction allocates <tt>sizeof(&lt;type&gt;)*NumElements</tt>
3212bytes of memory on the runtime stack, returning a pointer of the
Gabor Greif5082cf42008-02-09 22:24:34 +0000[diff] [blame]3213appropriate type to the program. If "NumElements" is specified, it is the
3214number of elements allocated, otherwise "NumElements" is defaulted to be one.
Chris Lattner10368b62008-04-02 00:38:26 +0000[diff] [blame]3215If a constant alignment is specified, the value result of the allocation is guaranteed
Gabor Greif5082cf42008-02-09 22:24:34 +0000[diff] [blame]3216to be aligned to at least that boundary. If not specified, or if zero, the target
3217can choose to align the allocation on any convenient boundary.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3218
3219<p>'<tt>type</tt>' may be any sized type.</p>
3220
3221<h5>Semantics:</h5>
3222
Chris Lattner8b094fc2008-04-19 21:01:16 +0000[diff] [blame]3223<p>Memory is allocated; a pointer is returned. The operation is undefiend if
3224there is insufficient stack space for the allocation. '<tt>alloca</tt>'d
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3225memory is automatically released when the function returns. The '<tt>alloca</tt>'
3226instruction is commonly used to represent automatic variables that must
3227have an address available. When the function returns (either with the <tt><a
3228 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]3229instructions), the memory is reclaimed. Allocating zero bytes
3230is legal, but the result is undefined.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3231
3232<h5>Example:</h5>
3233
3234<pre>
3235 %ptr = alloca i32 <i>; yields {i32*}:ptr</i>
3236 %ptr = alloca i32, i32 4 <i>; yields {i32*}:ptr</i>
3237 %ptr = alloca i32, i32 4, align 1024 <i>; yields {i32*}:ptr</i>
3238 %ptr = alloca i32, align 1024 <i>; yields {i32*}:ptr</i>
3239</pre>
3240</div>
3241
3242<!-- _______________________________________________________________________ -->
3243<div class="doc_subsubsection"> <a name="i_load">'<tt>load</tt>'
3244Instruction</a> </div>
3245<div class="doc_text">
3246<h5>Syntax:</h5>
3247<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>
3248<h5>Overview:</h5>
3249<p>The '<tt>load</tt>' instruction is used to read from memory.</p>
3250<h5>Arguments:</h5>
3251<p>The argument to the '<tt>load</tt>' instruction specifies the memory
3252address from which to load. The pointer must point to a <a
3253 href="#t_firstclass">first class</a> type. If the <tt>load</tt> is
3254marked as <tt>volatile</tt>, then the optimizer is not allowed to modify
3255the number or order of execution of this <tt>load</tt> with other
3256volatile <tt>load</tt> and <tt><a href="#i_store">store</a></tt>
3257instructions. </p>
Chris Lattner21003c82008-01-06 21:04:43 +0000[diff] [blame]3258<p>
Chris Lattner10368b62008-04-02 00:38:26 +0000[diff] [blame]3259The optional constant "align" argument specifies the alignment of the operation
Chris Lattner21003c82008-01-06 21:04:43 +0000[diff] [blame]3260(that is, the alignment of the memory address). A value of 0 or an
3261omitted "align" argument means that the operation has the preferential
3262alignment for the target. It is the responsibility of the code emitter
3263to ensure that the alignment information is correct. Overestimating
3264the alignment results in an undefined behavior. Underestimating the
3265alignment may produce less efficient code. An alignment of 1 is always
3266safe.
3267</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3268<h5>Semantics:</h5>
3269<p>The location of memory pointed to is loaded.</p>
3270<h5>Examples:</h5>
3271<pre> %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
3272 <a
3273 href="#i_store">store</a> i32 3, i32* %ptr <i>; yields {void}</i>
3274 %val = load i32* %ptr <i>; yields {i32}:val = i32 3</i>
3275</pre>
3276</div>
3277<!-- _______________________________________________________________________ -->
3278<div class="doc_subsubsection"> <a name="i_store">'<tt>store</tt>'
3279Instruction</a> </div>
3280<div class="doc_text">
3281<h5>Syntax:</h5>
3282<pre> store &lt;ty&gt; &lt;value&gt;, &lt;ty&gt;* &lt;pointer&gt;[, align &lt;alignment&gt;] <i>; yields {void}</i>
3283 volatile store &lt;ty&gt; &lt;value&gt;, &lt;ty&gt;* &lt;pointer&gt;[, align &lt;alignment&gt;] <i>; yields {void}</i>
3284</pre>
3285<h5>Overview:</h5>
3286<p>The '<tt>store</tt>' instruction is used to write to memory.</p>
3287<h5>Arguments:</h5>
3288<p>There are two arguments to the '<tt>store</tt>' instruction: a value
3289to 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]3290operand must be a pointer to the <a href="#t_firstclass">first class</a> type
3291of the '<tt>&lt;value&gt;</tt>'
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3292operand. If the <tt>store</tt> is marked as <tt>volatile</tt>, then the
3293optimizer is not allowed to modify the number or order of execution of
3294this <tt>store</tt> with other volatile <tt>load</tt> and <tt><a
3295 href="#i_store">store</a></tt> instructions.</p>
Chris Lattner21003c82008-01-06 21:04:43 +0000[diff] [blame]3296<p>
Chris Lattner10368b62008-04-02 00:38:26 +0000[diff] [blame]3297The optional constant "align" argument specifies the alignment of the operation
Chris Lattner21003c82008-01-06 21:04:43 +0000[diff] [blame]3298(that is, the alignment of the memory address). A value of 0 or an
3299omitted "align" argument means that the operation has the preferential
3300alignment for the target. It is the responsibility of the code emitter
3301to ensure that the alignment information is correct. Overestimating
3302the alignment results in an undefined behavior. Underestimating the
3303alignment may produce less efficient code. An alignment of 1 is always
3304safe.
3305</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3306<h5>Semantics:</h5>
3307<p>The contents of memory are updated to contain '<tt>&lt;value&gt;</tt>'
3308at the location specified by the '<tt>&lt;pointer&gt;</tt>' operand.</p>
3309<h5>Example:</h5>
3310<pre> %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
Bill Wendling63ffa142007-10-22 05:10:05 +0000[diff] [blame]3311 store i32 3, i32* %ptr <i>; yields {void}</i>
3312 %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]3313</pre>
3314</div>
3315
3316<!-- _______________________________________________________________________ -->
3317<div class="doc_subsubsection">
3318 <a name="i_getelementptr">'<tt>getelementptr</tt>' Instruction</a>
3319</div>
3320
3321<div class="doc_text">
3322<h5>Syntax:</h5>
3323<pre>
Matthijs Kooijman32a080e2008-10-13 13:44:15 +0000[diff] [blame]3324 &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]3325</pre>
3326
3327<h5>Overview:</h5>
3328
3329<p>
3330The '<tt>getelementptr</tt>' instruction is used to get the address of a
Matthijs Kooijman32a080e2008-10-13 13:44:15 +0000[diff] [blame]3331subelement of an aggregate data structure. It performs address calculation only
3332and does not access memory.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3333
3334<h5>Arguments:</h5>
3335
Matthijs Kooijman32a080e2008-10-13 13:44:15 +0000[diff] [blame]3336<p>The first argument is always a pointer, and forms the basis of the
3337calculation. The remaining arguments are indices, that indicate which of the
3338elements of the aggregate object are indexed. The interpretation of each index
3339is dependent on the type being indexed into. The first index always indexes the
3340pointer value given as the first argument, the second index indexes a value of
3341the type pointed to (not necessarily the value directly pointed to, since the
3342first index can be non-zero), etc. The first type indexed into must be a pointer
3343value, subsequent types can be arrays, vectors and structs. Note that subsequent
3344types being indexed into can never be pointers, since that would require loading
3345the pointer before continuing calculation.</p>
3346
3347<p>The type of each index argument depends on the type it is indexing into.
3348When indexing into a (packed) structure, only <tt>i32</tt> integer
3349<b>constants</b> are allowed. When indexing into an array, pointer or vector,
3350only integers of 32 or 64 bits are allowed (also non-constants). 32-bit values
3351will be sign extended to 64-bits if required.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3352
3353<p>For example, let's consider a C code fragment and how it gets
3354compiled to LLVM:</p>
3355
3356<div class="doc_code">
3357<pre>
3358struct RT {
3359 char A;
3360 int B[10][20];
3361 char C;
3362};
3363struct ST {
3364 int X;
3365 double Y;
3366 struct RT Z;
3367};
3368
3369int *foo(struct ST *s) {
3370 return &amp;s[1].Z.B[5][13];
3371}
3372</pre>
3373</div>
3374
3375<p>The LLVM code generated by the GCC frontend is:</p>
3376
3377<div class="doc_code">
3378<pre>
3379%RT = type { i8 , [10 x [20 x i32]], i8 }
3380%ST = type { i32, double, %RT }
3381
3382define i32* %foo(%ST* %s) {
3383entry:
3384 %reg = getelementptr %ST* %s, i32 1, i32 2, i32 1, i32 5, i32 13
3385 ret i32* %reg
3386}
3387</pre>
3388</div>
3389
3390<h5>Semantics:</h5>
3391
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3392<p>In the example above, the first index is indexing into the '<tt>%ST*</tt>'
3393type, which is a pointer, yielding a '<tt>%ST</tt>' = '<tt>{ i32, double, %RT
3394}</tt>' type, a structure. The second index indexes into the third element of
3395the structure, yielding a '<tt>%RT</tt>' = '<tt>{ i8 , [10 x [20 x i32]],
3396i8 }</tt>' type, another structure. The third index indexes into the second
3397element of the structure, yielding a '<tt>[10 x [20 x i32]]</tt>' type, an
3398array. The two dimensions of the array are subscripted into, yielding an
3399'<tt>i32</tt>' type. The '<tt>getelementptr</tt>' instruction returns a pointer
3400to this element, thus computing a value of '<tt>i32*</tt>' type.</p>
3401
3402<p>Note that it is perfectly legal to index partially through a
3403structure, returning a pointer to an inner element. Because of this,
3404the LLVM code for the given testcase is equivalent to:</p>
3405
3406<pre>
3407 define i32* %foo(%ST* %s) {
3408 %t1 = getelementptr %ST* %s, i32 1 <i>; yields %ST*:%t1</i>
3409 %t2 = getelementptr %ST* %t1, i32 0, i32 2 <i>; yields %RT*:%t2</i>
3410 %t3 = getelementptr %RT* %t2, i32 0, i32 1 <i>; yields [10 x [20 x i32]]*:%t3</i>
3411 %t4 = getelementptr [10 x [20 x i32]]* %t3, i32 0, i32 5 <i>; yields [20 x i32]*:%t4</i>
3412 %t5 = getelementptr [20 x i32]* %t4, i32 0, i32 13 <i>; yields i32*:%t5</i>
3413 ret i32* %t5
3414 }
3415</pre>
3416
3417<p>Note that it is undefined to access an array out of bounds: array and
3418pointer indexes must always be within the defined bounds of the array type.
Chris Lattnera7d94ba2008-04-24 05:59:56 +0000[diff] [blame]3419The one exception for this rule is zero length arrays. These arrays are
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3420defined to be accessible as variable length arrays, which requires access
3421beyond the zero'th element.</p>
3422
3423<p>The getelementptr instruction is often confusing. For some more insight
3424into how it works, see <a href="GetElementPtr.html">the getelementptr
3425FAQ</a>.</p>
3426
3427<h5>Example:</h5>
3428
3429<pre>
3430 <i>; yields [12 x i8]*:aptr</i>
Matthijs Kooijman32a080e2008-10-13 13:44:15 +0000[diff] [blame]3431 %aptr = getelementptr {i32, [12 x i8]}* %saptr, i64 0, i32 1
3432 <i>; yields i8*:vptr</i>
3433 %vptr = getelementptr {i32, <2 x i8>}* %svptr, i64 0, i32 1, i32 1
3434 <i>; yields i8*:eptr</i>
3435 %eptr = getelementptr [12 x i8]* %aptr, i64 0, i32 1
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3436</pre>
3437</div>
3438
3439<!-- ======================================================================= -->
3440<div class="doc_subsection"> <a name="convertops">Conversion Operations</a>
3441</div>
3442<div class="doc_text">
3443<p>The instructions in this category are the conversion instructions (casting)
3444which all take a single operand and a type. They perform various bit conversions
3445on the operand.</p>
3446</div>
3447
3448<!-- _______________________________________________________________________ -->
3449<div class="doc_subsubsection">
3450 <a name="i_trunc">'<tt>trunc .. to</tt>' Instruction</a>
3451</div>
3452<div class="doc_text">
3453
3454<h5>Syntax:</h5>
3455<pre>
3456 &lt;result&gt; = trunc &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3457</pre>
3458
3459<h5>Overview:</h5>
3460<p>
3461The '<tt>trunc</tt>' instruction truncates its operand to the type <tt>ty2</tt>.
3462</p>
3463
3464<h5>Arguments:</h5>
3465<p>
3466The '<tt>trunc</tt>' instruction takes a <tt>value</tt> to trunc, which must
3467be an <a href="#t_integer">integer</a> type, and a type that specifies the size
3468and type of the result, which must be an <a href="#t_integer">integer</a>
3469type. The bit size of <tt>value</tt> must be larger than the bit size of
3470<tt>ty2</tt>. Equal sized types are not allowed.</p>
3471
3472<h5>Semantics:</h5>
3473<p>
3474The '<tt>trunc</tt>' instruction truncates the high order bits in <tt>value</tt>
3475and converts the remaining bits to <tt>ty2</tt>. Since the source size must be
3476larger than the destination size, <tt>trunc</tt> cannot be a <i>no-op cast</i>.
3477It will always truncate bits.</p>
3478
3479<h5>Example:</h5>
3480<pre>
3481 %X = trunc i32 257 to i8 <i>; yields i8:1</i>
3482 %Y = trunc i32 123 to i1 <i>; yields i1:true</i>
3483 %Y = trunc i32 122 to i1 <i>; yields i1:false</i>
3484</pre>
3485</div>
3486
3487<!-- _______________________________________________________________________ -->
3488<div class="doc_subsubsection">
3489 <a name="i_zext">'<tt>zext .. to</tt>' Instruction</a>
3490</div>
3491<div class="doc_text">
3492
3493<h5>Syntax:</h5>
3494<pre>
3495 &lt;result&gt; = zext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3496</pre>
3497
3498<h5>Overview:</h5>
3499<p>The '<tt>zext</tt>' instruction zero extends its operand to type
3500<tt>ty2</tt>.</p>
3501
3502
3503<h5>Arguments:</h5>
3504<p>The '<tt>zext</tt>' instruction takes a value to cast, which must be of
3505<a href="#t_integer">integer</a> type, and a type to cast it to, which must
3506also be of <a href="#t_integer">integer</a> type. The bit size of the
3507<tt>value</tt> must be smaller than the bit size of the destination type,
3508<tt>ty2</tt>.</p>
3509
3510<h5>Semantics:</h5>
3511<p>The <tt>zext</tt> fills the high order bits of the <tt>value</tt> with zero
3512bits until it reaches the size of the destination type, <tt>ty2</tt>.</p>
3513
3514<p>When zero extending from i1, the result will always be either 0 or 1.</p>
3515
3516<h5>Example:</h5>
3517<pre>
3518 %X = zext i32 257 to i64 <i>; yields i64:257</i>
3519 %Y = zext i1 true to i32 <i>; yields i32:1</i>
3520</pre>
3521</div>
3522
3523<!-- _______________________________________________________________________ -->
3524<div class="doc_subsubsection">
3525 <a name="i_sext">'<tt>sext .. to</tt>' Instruction</a>
3526</div>
3527<div class="doc_text">
3528
3529<h5>Syntax:</h5>
3530<pre>
3531 &lt;result&gt; = sext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3532</pre>
3533
3534<h5>Overview:</h5>
3535<p>The '<tt>sext</tt>' sign extends <tt>value</tt> to the type <tt>ty2</tt>.</p>
3536
3537<h5>Arguments:</h5>
3538<p>
3539The '<tt>sext</tt>' instruction takes a value to cast, which must be of
3540<a href="#t_integer">integer</a> type, and a type to cast it to, which must
3541also be of <a href="#t_integer">integer</a> type. The bit size of the
3542<tt>value</tt> must be smaller than the bit size of the destination type,
3543<tt>ty2</tt>.</p>
3544
3545<h5>Semantics:</h5>
3546<p>
3547The '<tt>sext</tt>' instruction performs a sign extension by copying the sign
3548bit (highest order bit) of the <tt>value</tt> until it reaches the bit size of
3549the type <tt>ty2</tt>.</p>
3550
3551<p>When sign extending from i1, the extension always results in -1 or 0.</p>
3552
3553<h5>Example:</h5>
3554<pre>
3555 %X = sext i8 -1 to i16 <i>; yields i16 :65535</i>
3556 %Y = sext i1 true to i32 <i>; yields i32:-1</i>
3557</pre>
3558</div>
3559
3560<!-- _______________________________________________________________________ -->
3561<div class="doc_subsubsection">
3562 <a name="i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a>
3563</div>
3564
3565<div class="doc_text">
3566
3567<h5>Syntax:</h5>
3568
3569<pre>
3570 &lt;result&gt; = fptrunc &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3571</pre>
3572
3573<h5>Overview:</h5>
3574<p>The '<tt>fptrunc</tt>' instruction truncates <tt>value</tt> to type
3575<tt>ty2</tt>.</p>
3576
3577
3578<h5>Arguments:</h5>
3579<p>The '<tt>fptrunc</tt>' instruction takes a <a href="#t_floating">floating
3580 point</a> value to cast and a <a href="#t_floating">floating point</a> type to
3581cast it to. The size of <tt>value</tt> must be larger than the size of
3582<tt>ty2</tt>. This implies that <tt>fptrunc</tt> cannot be used to make a
3583<i>no-op cast</i>.</p>
3584
3585<h5>Semantics:</h5>
3586<p> The '<tt>fptrunc</tt>' instruction truncates a <tt>value</tt> from a larger
3587<a href="#t_floating">floating point</a> type to a smaller
3588<a href="#t_floating">floating point</a> type. If the value cannot fit within
3589the destination type, <tt>ty2</tt>, then the results are undefined.</p>
3590
3591<h5>Example:</h5>
3592<pre>
3593 %X = fptrunc double 123.0 to float <i>; yields float:123.0</i>
3594 %Y = fptrunc double 1.0E+300 to float <i>; yields undefined</i>
3595</pre>
3596</div>
3597
3598<!-- _______________________________________________________________________ -->
3599<div class="doc_subsubsection">
3600 <a name="i_fpext">'<tt>fpext .. to</tt>' Instruction</a>
3601</div>
3602<div class="doc_text">
3603
3604<h5>Syntax:</h5>
3605<pre>
3606 &lt;result&gt; = fpext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3607</pre>
3608
3609<h5>Overview:</h5>
3610<p>The '<tt>fpext</tt>' extends a floating point <tt>value</tt> to a larger
3611floating point value.</p>
3612
3613<h5>Arguments:</h5>
3614<p>The '<tt>fpext</tt>' instruction takes a
3615<a href="#t_floating">floating point</a> <tt>value</tt> to cast,
3616and a <a href="#t_floating">floating point</a> type to cast it to. The source
3617type must be smaller than the destination type.</p>
3618
3619<h5>Semantics:</h5>
3620<p>The '<tt>fpext</tt>' instruction extends the <tt>value</tt> from a smaller
3621<a href="#t_floating">floating point</a> type to a larger
3622<a href="#t_floating">floating point</a> type. The <tt>fpext</tt> cannot be
3623used to make a <i>no-op cast</i> because it always changes bits. Use
3624<tt>bitcast</tt> to make a <i>no-op cast</i> for a floating point cast.</p>
3625
3626<h5>Example:</h5>
3627<pre>
3628 %X = fpext float 3.1415 to double <i>; yields double:3.1415</i>
3629 %Y = fpext float 1.0 to float <i>; yields float:1.0 (no-op)</i>
3630</pre>
3631</div>
3632
3633<!-- _______________________________________________________________________ -->
3634<div class="doc_subsubsection">
3635 <a name="i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a>
3636</div>
3637<div class="doc_text">
3638
3639<h5>Syntax:</h5>
3640<pre>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3641 &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]3642</pre>
3643
3644<h5>Overview:</h5>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3645<p>The '<tt>fptoui</tt>' converts a floating point <tt>value</tt> to its
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3646unsigned integer equivalent of type <tt>ty2</tt>.
3647</p>
3648
3649<h5>Arguments:</h5>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3650<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]3651scalar or vector <a href="#t_floating">floating point</a> value, and a type
3652to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
3653type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
3654vector integer type with the same number of elements as <tt>ty</tt></p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3655
3656<h5>Semantics:</h5>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3657<p> The '<tt>fptoui</tt>' instruction converts its
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3658<a href="#t_floating">floating point</a> operand into the nearest (rounding
3659towards zero) unsigned integer value. If the value cannot fit in <tt>ty2</tt>,
3660the results are undefined.</p>
3661
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3662<h5>Example:</h5>
3663<pre>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3664 %X = fptoui double 123.0 to i32 <i>; yields i32:123</i>
Chris Lattner681f1e82007-09-22 03:17:52 +0000[diff] [blame]3665 %Y = fptoui float 1.0E+300 to i1 <i>; yields undefined:1</i>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3666 %X = fptoui float 1.04E+17 to i8 <i>; yields undefined:1</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3667</pre>
3668</div>
3669
3670<!-- _______________________________________________________________________ -->
3671<div class="doc_subsubsection">
3672 <a name="i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a>
3673</div>
3674<div class="doc_text">
3675
3676<h5>Syntax:</h5>
3677<pre>
3678 &lt;result&gt; = fptosi &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3679</pre>
3680
3681<h5>Overview:</h5>
3682<p>The '<tt>fptosi</tt>' instruction converts
3683<a href="#t_floating">floating point</a> <tt>value</tt> to type <tt>ty2</tt>.
3684</p>
3685
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3686<h5>Arguments:</h5>
3687<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]3688scalar or vector <a href="#t_floating">floating point</a> value, and a type
3689to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
3690type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
3691vector integer type with the same number of elements as <tt>ty</tt></p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3692
3693<h5>Semantics:</h5>
3694<p>The '<tt>fptosi</tt>' instruction converts its
3695<a href="#t_floating">floating point</a> operand into the nearest (rounding
3696towards zero) signed integer value. If the value cannot fit in <tt>ty2</tt>,
3697the results are undefined.</p>
3698
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3699<h5>Example:</h5>
3700<pre>
3701 %X = fptosi double -123.0 to i32 <i>; yields i32:-123</i>
Chris Lattner681f1e82007-09-22 03:17:52 +0000[diff] [blame]3702 %Y = fptosi float 1.0E-247 to i1 <i>; yields undefined:1</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3703 %X = fptosi float 1.04E+17 to i8 <i>; yields undefined:1</i>
3704</pre>
3705</div>
3706
3707<!-- _______________________________________________________________________ -->
3708<div class="doc_subsubsection">
3709 <a name="i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a>
3710</div>
3711<div class="doc_text">
3712
3713<h5>Syntax:</h5>
3714<pre>
3715 &lt;result&gt; = uitofp &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3716</pre>
3717
3718<h5>Overview:</h5>
3719<p>The '<tt>uitofp</tt>' instruction regards <tt>value</tt> as an unsigned
3720integer and converts that value to the <tt>ty2</tt> type.</p>
3721
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3722<h5>Arguments:</h5>
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]3723<p>The '<tt>uitofp</tt>' instruction takes a value to cast, which must be a
3724scalar or vector <a href="#t_integer">integer</a> value, and a type to cast it
3725to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
3726type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
3727floating point type with the same number of elements as <tt>ty</tt></p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3728
3729<h5>Semantics:</h5>
3730<p>The '<tt>uitofp</tt>' instruction interprets its operand as an unsigned
3731integer quantity and converts it to the corresponding floating point value. If
3732the value cannot fit in the floating point value, the results are undefined.</p>
3733
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3734<h5>Example:</h5>
3735<pre>
3736 %X = uitofp i32 257 to float <i>; yields float:257.0</i>
3737 %Y = uitofp i8 -1 to double <i>; yields double:255.0</i>
3738</pre>
3739</div>
3740
3741<!-- _______________________________________________________________________ -->
3742<div class="doc_subsubsection">
3743 <a name="i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a>
3744</div>
3745<div class="doc_text">
3746
3747<h5>Syntax:</h5>
3748<pre>
3749 &lt;result&gt; = sitofp &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3750</pre>
3751
3752<h5>Overview:</h5>
3753<p>The '<tt>sitofp</tt>' instruction regards <tt>value</tt> as a signed
3754integer and converts that value to the <tt>ty2</tt> type.</p>
3755
3756<h5>Arguments:</h5>
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]3757<p>The '<tt>sitofp</tt>' instruction takes a value to cast, which must be a
3758scalar or vector <a href="#t_integer">integer</a> value, and a type to cast it
3759to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
3760type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
3761floating point type with the same number of elements as <tt>ty</tt></p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3762
3763<h5>Semantics:</h5>
3764<p>The '<tt>sitofp</tt>' instruction interprets its operand as a signed
3765integer quantity and converts it to the corresponding floating point value. If
3766the value cannot fit in the floating point value, the results are undefined.</p>
3767
3768<h5>Example:</h5>
3769<pre>
3770 %X = sitofp i32 257 to float <i>; yields float:257.0</i>
3771 %Y = sitofp i8 -1 to double <i>; yields double:-1.0</i>
3772</pre>
3773</div>
3774
3775<!-- _______________________________________________________________________ -->
3776<div class="doc_subsubsection">
3777 <a name="i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a>
3778</div>
3779<div class="doc_text">
3780
3781<h5>Syntax:</h5>
3782<pre>
3783 &lt;result&gt; = ptrtoint &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3784</pre>
3785
3786<h5>Overview:</h5>
3787<p>The '<tt>ptrtoint</tt>' instruction converts the pointer <tt>value</tt> to
3788the integer type <tt>ty2</tt>.</p>
3789
3790<h5>Arguments:</h5>
3791<p>The '<tt>ptrtoint</tt>' instruction takes a <tt>value</tt> to cast, which
3792must be a <a href="#t_pointer">pointer</a> value, and a type to cast it to
3793<tt>ty2</tt>, which must be an <a href="#t_integer">integer</a> type.
3794
3795<h5>Semantics:</h5>
3796<p>The '<tt>ptrtoint</tt>' instruction converts <tt>value</tt> to integer type
3797<tt>ty2</tt> by interpreting the pointer value as an integer and either
3798truncating or zero extending that value to the size of the integer type. If
3799<tt>value</tt> is smaller than <tt>ty2</tt> then a zero extension is done. If
3800<tt>value</tt> is larger than <tt>ty2</tt> then a truncation is done. If they
3801are the same size, then nothing is done (<i>no-op cast</i>) other than a type
3802change.</p>
3803
3804<h5>Example:</h5>
3805<pre>
3806 %X = ptrtoint i32* %X to i8 <i>; yields truncation on 32-bit architecture</i>
3807 %Y = ptrtoint i32* %x to i64 <i>; yields zero extension on 32-bit architecture</i>
3808</pre>
3809</div>
3810
3811<!-- _______________________________________________________________________ -->
3812<div class="doc_subsubsection">
3813 <a name="i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a>
3814</div>
3815<div class="doc_text">
3816
3817<h5>Syntax:</h5>
3818<pre>
3819 &lt;result&gt; = inttoptr &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3820</pre>
3821
3822<h5>Overview:</h5>
3823<p>The '<tt>inttoptr</tt>' instruction converts an integer <tt>value</tt> to
3824a pointer type, <tt>ty2</tt>.</p>
3825
3826<h5>Arguments:</h5>
3827<p>The '<tt>inttoptr</tt>' instruction takes an <a href="#t_integer">integer</a>
3828value to cast, and a type to cast it to, which must be a
3829<a href="#t_pointer">pointer</a> type.
3830
3831<h5>Semantics:</h5>
3832<p>The '<tt>inttoptr</tt>' instruction converts <tt>value</tt> to type
3833<tt>ty2</tt> by applying either a zero extension or a truncation depending on
3834the size of the integer <tt>value</tt>. If <tt>value</tt> is larger than the
3835size of a pointer then a truncation is done. If <tt>value</tt> is smaller than
3836the size of a pointer then a zero extension is done. If they are the same size,
3837nothing is done (<i>no-op cast</i>).</p>
3838
3839<h5>Example:</h5>
3840<pre>
3841 %X = inttoptr i32 255 to i32* <i>; yields zero extension on 64-bit architecture</i>
3842 %X = inttoptr i32 255 to i32* <i>; yields no-op on 32-bit architecture</i>
3843 %Y = inttoptr i64 0 to i32* <i>; yields truncation on 32-bit architecture</i>
3844</pre>
3845</div>
3846
3847<!-- _______________________________________________________________________ -->
3848<div class="doc_subsubsection">
3849 <a name="i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a>
3850</div>
3851<div class="doc_text">
3852
3853<h5>Syntax:</h5>
3854<pre>
3855 &lt;result&gt; = bitcast &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3856</pre>
3857
3858<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3859
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3860<p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
3861<tt>ty2</tt> without changing any bits.</p>
3862
3863<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3864
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3865<p>The '<tt>bitcast</tt>' instruction takes a value to cast, which must be
Dan Gohman7305fa02008-09-08 16:45:59 +0000[diff] [blame]3866a non-aggregate first class value, and a type to cast it to, which must also be
3867a non-aggregate <a href="#t_firstclass">first class</a> type. The bit sizes of
3868<tt>value</tt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3869and the destination type, <tt>ty2</tt>, must be identical. If the source
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3870type is a pointer, the destination type must also be a pointer. This
3871instruction supports bitwise conversion of vectors to integers and to vectors
3872of other types (as long as they have the same size).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3873
3874<h5>Semantics:</h5>
3875<p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
3876<tt>ty2</tt>. It is always a <i>no-op cast</i> because no bits change with
3877this conversion. The conversion is done as if the <tt>value</tt> had been
3878stored to memory and read back as type <tt>ty2</tt>. Pointer types may only be
3879converted to other pointer types with this instruction. To convert pointers to
3880other types, use the <a href="#i_inttoptr">inttoptr</a> or
3881<a href="#i_ptrtoint">ptrtoint</a> instructions first.</p>
3882
3883<h5>Example:</h5>
3884<pre>
3885 %X = bitcast i8 255 to i8 <i>; yields i8 :-1</i>
3886 %Y = bitcast i32* %x to sint* <i>; yields sint*:%x</i>
3887 %Z = bitcast <2xint> %V to i64; <i>; yields i64: %V</i>
3888</pre>
3889</div>
3890
3891<!-- ======================================================================= -->
3892<div class="doc_subsection"> <a name="otherops">Other Operations</a> </div>
3893<div class="doc_text">
3894<p>The instructions in this category are the "miscellaneous"
3895instructions, which defy better classification.</p>
3896</div>
3897
3898<!-- _______________________________________________________________________ -->
3899<div class="doc_subsubsection"><a name="i_icmp">'<tt>icmp</tt>' Instruction</a>
3900</div>
3901<div class="doc_text">
3902<h5>Syntax:</h5>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]3903<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]3904</pre>
3905<h5>Overview:</h5>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]3906<p>The '<tt>icmp</tt>' instruction returns a boolean value or
3907a vector of boolean values based on comparison
3908of its two integer, integer vector, or pointer operands.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3909<h5>Arguments:</h5>
3910<p>The '<tt>icmp</tt>' instruction takes three operands. The first operand is
3911the condition code indicating the kind of comparison to perform. It is not
3912a value, just a keyword. The possible condition code are:
3913<ol>
3914 <li><tt>eq</tt>: equal</li>
3915 <li><tt>ne</tt>: not equal </li>
3916 <li><tt>ugt</tt>: unsigned greater than</li>
3917 <li><tt>uge</tt>: unsigned greater or equal</li>
3918 <li><tt>ult</tt>: unsigned less than</li>
3919 <li><tt>ule</tt>: unsigned less or equal</li>
3920 <li><tt>sgt</tt>: signed greater than</li>
3921 <li><tt>sge</tt>: signed greater or equal</li>
3922 <li><tt>slt</tt>: signed less than</li>
3923 <li><tt>sle</tt>: signed less or equal</li>
3924</ol>
3925<p>The remaining two arguments must be <a href="#t_integer">integer</a> or
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]3926<a href="#t_pointer">pointer</a>
3927or integer <a href="#t_vector">vector</a> typed.
3928They must also be identical types.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3929<h5>Semantics:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3930<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]3931the condition code given as <tt>cond</tt>. The comparison performed always
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]3932yields either an <a href="#t_primitive"><tt>i1</tt></a> or vector of <tt>i1</tt> result, as follows:
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3933<ol>
3934 <li><tt>eq</tt>: yields <tt>true</tt> if the operands are equal,
3935 <tt>false</tt> otherwise. No sign interpretation is necessary or performed.
3936 </li>
3937 <li><tt>ne</tt>: yields <tt>true</tt> if the operands are unequal,
3938 <tt>false</tt> otherwise. No sign interpretation is necessary or performed.
3939 <li><tt>ugt</tt>: interprets the operands as unsigned values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3940 <tt>true</tt> if <tt>op1</tt> is greater than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3941 <li><tt>uge</tt>: interprets the operands as unsigned values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3942 <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]3943 <li><tt>ult</tt>: interprets the operands as unsigned values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3944 <tt>true</tt> if <tt>op1</tt> is less than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3945 <li><tt>ule</tt>: interprets the operands as unsigned values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3946 <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]3947 <li><tt>sgt</tt>: interprets the operands as signed values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3948 <tt>true</tt> if <tt>op1</tt> is greater than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3949 <li><tt>sge</tt>: interprets the operands as signed values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3950 <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]3951 <li><tt>slt</tt>: interprets the operands as signed values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3952 <tt>true</tt> if <tt>op1</tt> is less than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3953 <li><tt>sle</tt>: interprets the operands as signed values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3954 <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]3955</ol>
3956<p>If the operands are <a href="#t_pointer">pointer</a> typed, the pointer
3957values are compared as if they were integers.</p>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]3958<p>If the operands are integer vectors, then they are compared
3959element by element. The result is an <tt>i1</tt> vector with
3960the same number of elements as the values being compared.
3961Otherwise, the result is an <tt>i1</tt>.
3962</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3963
3964<h5>Example:</h5>
3965<pre> &lt;result&gt; = icmp eq i32 4, 5 <i>; yields: result=false</i>
3966 &lt;result&gt; = icmp ne float* %X, %X <i>; yields: result=false</i>
3967 &lt;result&gt; = icmp ult i16 4, 5 <i>; yields: result=true</i>
3968 &lt;result&gt; = icmp sgt i16 4, 5 <i>; yields: result=false</i>
3969 &lt;result&gt; = icmp ule i16 -4, 5 <i>; yields: result=false</i>
3970 &lt;result&gt; = icmp sge i16 4, 5 <i>; yields: result=false</i>
3971</pre>
3972</div>
3973
3974<!-- _______________________________________________________________________ -->
3975<div class="doc_subsubsection"><a name="i_fcmp">'<tt>fcmp</tt>' Instruction</a>
3976</div>
3977<div class="doc_text">
3978<h5>Syntax:</h5>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]3979<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]3980</pre>
3981<h5>Overview:</h5>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]3982<p>The '<tt>fcmp</tt>' instruction returns a boolean value
3983or vector of boolean values based on comparison
3984of its operands.
3985<p>
3986If the operands are floating point scalars, then the result
3987type is a boolean (<a href="#t_primitive"><tt>i1</tt></a>).
3988</p>
3989<p>If the operands are floating point vectors, then the result type
3990is a vector of boolean with the same number of elements as the
3991operands being compared.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3992<h5>Arguments:</h5>
3993<p>The '<tt>fcmp</tt>' instruction takes three operands. The first operand is
3994the condition code indicating the kind of comparison to perform. It is not
3995a value, just a keyword. The possible condition code are:
3996<ol>
3997 <li><tt>false</tt>: no comparison, always returns false</li>
3998 <li><tt>oeq</tt>: ordered and equal</li>
3999 <li><tt>ogt</tt>: ordered and greater than </li>
4000 <li><tt>oge</tt>: ordered and greater than or equal</li>
4001 <li><tt>olt</tt>: ordered and less than </li>
4002 <li><tt>ole</tt>: ordered and less than or equal</li>
4003 <li><tt>one</tt>: ordered and not equal</li>
4004 <li><tt>ord</tt>: ordered (no nans)</li>
4005 <li><tt>ueq</tt>: unordered or equal</li>
4006 <li><tt>ugt</tt>: unordered or greater than </li>
4007 <li><tt>uge</tt>: unordered or greater than or equal</li>
4008 <li><tt>ult</tt>: unordered or less than </li>
4009 <li><tt>ule</tt>: unordered or less than or equal</li>
4010 <li><tt>une</tt>: unordered or not equal</li>
4011 <li><tt>uno</tt>: unordered (either nans)</li>
4012 <li><tt>true</tt>: no comparison, always returns true</li>
4013</ol>
4014<p><i>Ordered</i> means that neither operand is a QNAN while
4015<i>unordered</i> means that either operand may be a QNAN.</p>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4016<p>Each of <tt>val1</tt> and <tt>val2</tt> arguments must be
4017either a <a href="#t_floating">floating point</a> type
4018or a <a href="#t_vector">vector</a> of floating point type.
4019They must have identical types.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4020<h5>Semantics:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4021<p>The '<tt>fcmp</tt>' instruction compares <tt>op1</tt> and <tt>op2</tt>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4022according to the condition code given as <tt>cond</tt>.
4023If the operands are vectors, then the vectors are compared
4024element by element.
4025Each comparison performed
4026always yields an <a href="#t_primitive">i1</a> result, as follows:
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4027<ol>
4028 <li><tt>false</tt>: always yields <tt>false</tt>, regardless of operands.</li>
4029 <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]4030 <tt>op1</tt> is equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4031 <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]4032 <tt>op1</tt> is greather than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4033 <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]4034 <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4035 <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]4036 <tt>op1</tt> is less than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4037 <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]4038 <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4039 <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]4040 <tt>op1</tt> is not equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4041 <li><tt>ord</tt>: yields <tt>true</tt> if both operands are not a QNAN.</li>
4042 <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]4043 <tt>op1</tt> is equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4044 <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]4045 <tt>op1</tt> is greater than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4046 <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]4047 <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4048 <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]4049 <tt>op1</tt> is less than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4050 <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]4051 <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4052 <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]4053 <tt>op1</tt> is not equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4054 <li><tt>uno</tt>: yields <tt>true</tt> if either operand is a QNAN.</li>
4055 <li><tt>true</tt>: always yields <tt>true</tt>, regardless of operands.</li>
4056</ol>
4057
4058<h5>Example:</h5>
4059<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]4060 &lt;result&gt; = fcmp one float 4.0, 5.0 <i>; yields: result=true</i>
4061 &lt;result&gt; = fcmp olt float 4.0, 5.0 <i>; yields: result=true</i>
4062 &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]4063</pre>
4064</div>
4065
4066<!-- _______________________________________________________________________ -->
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]4067<div class="doc_subsubsection">
4068 <a name="i_vicmp">'<tt>vicmp</tt>' Instruction</a>
4069</div>
4070<div class="doc_text">
4071<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4072<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]4073</pre>
4074<h5>Overview:</h5>
4075<p>The '<tt>vicmp</tt>' instruction returns an integer vector value based on
4076element-wise comparison of its two integer vector operands.</p>
4077<h5>Arguments:</h5>
4078<p>The '<tt>vicmp</tt>' instruction takes three operands. The first operand is
4079the condition code indicating the kind of comparison to perform. It is not
4080a value, just a keyword. The possible condition code are:
4081<ol>
4082 <li><tt>eq</tt>: equal</li>
4083 <li><tt>ne</tt>: not equal </li>
4084 <li><tt>ugt</tt>: unsigned greater than</li>
4085 <li><tt>uge</tt>: unsigned greater or equal</li>
4086 <li><tt>ult</tt>: unsigned less than</li>
4087 <li><tt>ule</tt>: unsigned less or equal</li>
4088 <li><tt>sgt</tt>: signed greater than</li>
4089 <li><tt>sge</tt>: signed greater or equal</li>
4090 <li><tt>slt</tt>: signed less than</li>
4091 <li><tt>sle</tt>: signed less or equal</li>
4092</ol>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4093<p>The remaining two arguments must be <a href="#t_vector">vector</a> or
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]4094<a href="#t_integer">integer</a> typed. They must also be identical types.</p>
4095<h5>Semantics:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4096<p>The '<tt>vicmp</tt>' instruction compares <tt>op1</tt> and <tt>op2</tt>
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]4097according to the condition code given as <tt>cond</tt>. The comparison yields a
4098<a href="#t_vector">vector</a> of <a href="#t_integer">integer</a> result, of
4099identical type as the values being compared. The most significant bit in each
4100element is 1 if the element-wise comparison evaluates to true, and is 0
4101otherwise. All other bits of the result are undefined. The condition codes
4102are evaluated identically to the <a href="#i_icmp">'<tt>icmp</tt>'
4103instruction</a>.
4104
4105<h5>Example:</h5>
4106<pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4107 &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>
4108 &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]4109</pre>
4110</div>
4111
4112<!-- _______________________________________________________________________ -->
4113<div class="doc_subsubsection">
4114 <a name="i_vfcmp">'<tt>vfcmp</tt>' Instruction</a>
4115</div>
4116<div class="doc_text">
4117<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4118<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]4119<h5>Overview:</h5>
4120<p>The '<tt>vfcmp</tt>' instruction returns an integer vector value based on
4121element-wise comparison of its two floating point vector operands. The output
4122elements have the same width as the input elements.</p>
4123<h5>Arguments:</h5>
4124<p>The '<tt>vfcmp</tt>' instruction takes three operands. The first operand is
4125the condition code indicating the kind of comparison to perform. It is not
4126a value, just a keyword. The possible condition code are:
4127<ol>
4128 <li><tt>false</tt>: no comparison, always returns false</li>
4129 <li><tt>oeq</tt>: ordered and equal</li>
4130 <li><tt>ogt</tt>: ordered and greater than </li>
4131 <li><tt>oge</tt>: ordered and greater than or equal</li>
4132 <li><tt>olt</tt>: ordered and less than </li>
4133 <li><tt>ole</tt>: ordered and less than or equal</li>
4134 <li><tt>one</tt>: ordered and not equal</li>
4135 <li><tt>ord</tt>: ordered (no nans)</li>
4136 <li><tt>ueq</tt>: unordered or equal</li>
4137 <li><tt>ugt</tt>: unordered or greater than </li>
4138 <li><tt>uge</tt>: unordered or greater than or equal</li>
4139 <li><tt>ult</tt>: unordered or less than </li>
4140 <li><tt>ule</tt>: unordered or less than or equal</li>
4141 <li><tt>une</tt>: unordered or not equal</li>
4142 <li><tt>uno</tt>: unordered (either nans)</li>
4143 <li><tt>true</tt>: no comparison, always returns true</li>
4144</ol>
4145<p>The remaining two arguments must be <a href="#t_vector">vector</a> of
4146<a href="#t_floating">floating point</a> typed. They must also be identical
4147types.</p>
4148<h5>Semantics:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4149<p>The '<tt>vfcmp</tt>' instruction compares <tt>op1</tt> and <tt>op2</tt>
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]4150according to the condition code given as <tt>cond</tt>. The comparison yields a
4151<a href="#t_vector">vector</a> of <a href="#t_integer">integer</a> result, with
4152an identical number of elements as the values being compared, and each element
4153having identical with to the width of the floating point elements. The most
4154significant bit in each element is 1 if the element-wise comparison evaluates to
4155true, and is 0 otherwise. All other bits of the result are undefined. The
4156condition codes are evaluated identically to the
4157<a href="#i_fcmp">'<tt>fcmp</tt>' instruction</a>.
4158
4159<h5>Example:</h5>
4160<pre>
Chris Lattner1e5c5cd02008-10-13 16:55:18 +0000[diff] [blame]4161 <i>; yields: result=&lt;2 x i32&gt; &lt; i32 0, i32 -1 &gt;</i>
4162 &lt;result&gt; = vfcmp oeq &lt;2 x float&gt; &lt; float 4, float 0 &gt;, &lt; float 5, float 0 &gt;
4163
4164 <i>; yields: result=&lt;2 x i64&gt; &lt; i64 -1, i64 0 &gt;</i>
4165 &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]4166</pre>
4167</div>
4168
4169<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4170<div class="doc_subsubsection">
4171 <a name="i_phi">'<tt>phi</tt>' Instruction</a>
4172</div>
4173
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4174<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4175
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4176<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4177
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4178<pre> &lt;result&gt; = phi &lt;ty&gt; [ &lt;val0&gt;, &lt;label0&gt;], ...<br></pre>
4179<h5>Overview:</h5>
4180<p>The '<tt>phi</tt>' instruction is used to implement the &#966; node in
4181the SSA graph representing the function.</p>
4182<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4183
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4184<p>The type of the incoming values is specified with the first type
4185field. After this, the '<tt>phi</tt>' instruction takes a list of pairs
4186as arguments, with one pair for each predecessor basic block of the
4187current block. Only values of <a href="#t_firstclass">first class</a>
4188type may be used as the value arguments to the PHI node. Only labels
4189may be used as the label arguments.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4190
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4191<p>There must be no non-phi instructions between the start of a basic
4192block and the PHI instructions: i.e. PHI instructions must be first in
4193a basic block.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4194
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4195<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4196
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4197<p>At runtime, the '<tt>phi</tt>' instruction logically takes on the value
4198specified by the pair corresponding to the predecessor basic block that executed
4199just prior to the current block.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4200
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4201<h5>Example:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4202<pre>
4203Loop: ; Infinite loop that counts from 0 on up...
4204 %indvar = phi i32 [ 0, %LoopHeader ], [ %nextindvar, %Loop ]
4205 %nextindvar = add i32 %indvar, 1
4206 br label %Loop
4207</pre>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4208</div>
4209
4210<!-- _______________________________________________________________________ -->
4211<div class="doc_subsubsection">
4212 <a name="i_select">'<tt>select</tt>' Instruction</a>
4213</div>
4214
4215<div class="doc_text">
4216
4217<h5>Syntax:</h5>
4218
4219<pre>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4220 &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>
4221
4222 <i>selty</i> is either i1 or {&lt;N x i1&gt}
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4223</pre>
4224
4225<h5>Overview:</h5>
4226
4227<p>
4228The '<tt>select</tt>' instruction is used to choose one value based on a
4229condition, without branching.
4230</p>
4231
4232
4233<h5>Arguments:</h5>
4234
4235<p>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4236The '<tt>select</tt>' instruction requires an 'i1' value or
4237a vector of 'i1' values indicating the
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4238condition, and two values of the same <a href="#t_firstclass">first class</a>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4239type. If the val1/val2 are vectors and
4240the condition is a scalar, then entire vectors are selected, not
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4241individual elements.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4242</p>
4243
4244<h5>Semantics:</h5>
4245
4246<p>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4247If 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]4248value argument; otherwise, it returns the second value argument.
4249</p>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4250<p>
4251If the condition is a vector of i1, then the value arguments must
4252be vectors of the same size, and the selection is done element
4253by element.
4254</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4255
4256<h5>Example:</h5>
4257
4258<pre>
4259 %X = select i1 true, i8 17, i8 42 <i>; yields i8:17</i>
4260</pre>
4261</div>
4262
4263
4264<!-- _______________________________________________________________________ -->
4265<div class="doc_subsubsection">
4266 <a name="i_call">'<tt>call</tt>' Instruction</a>
4267</div>
4268
4269<div class="doc_text">
4270
4271<h5>Syntax:</h5>
4272<pre>
Devang Pateld0bfcc72008-10-07 17:48:33 +0000[diff] [blame]4273 &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]4274</pre>
4275
4276<h5>Overview:</h5>
4277
4278<p>The '<tt>call</tt>' instruction represents a simple function call.</p>
4279
4280<h5>Arguments:</h5>
4281
4282<p>This instruction requires several arguments:</p>
4283
4284<ol>
4285 <li>
4286 <p>The optional "tail" marker indicates whether the callee function accesses
4287 any allocas or varargs in the caller. If the "tail" marker is present, the
4288 function call is eligible for tail call optimization. Note that calls may
4289 be marked "tail" even if they do not occur before a <a
4290 href="#i_ret"><tt>ret</tt></a> instruction.
4291 </li>
4292 <li>
4293 <p>The optional "cconv" marker indicates which <a href="#callingconv">calling
4294 convention</a> the call should use. If none is specified, the call defaults
4295 to using C calling conventions.
4296 </li>
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]4297
4298 <li>
4299 <p>The optional <a href="#paramattrs">Parameter Attributes</a> list for
4300 return values. Only '<tt>zeroext</tt>', '<tt>signext</tt>',
4301 and '<tt>inreg</tt>' attributes are valid here.</p>
4302 </li>
4303
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4304 <li>
Nick Lewycky93082fc2007-09-08 13:57:50 +0000[diff] [blame]4305 <p>'<tt>ty</tt>': the type of the call instruction itself which is also
4306 the type of the return value. Functions that return no value are marked
4307 <tt><a href="#t_void">void</a></tt>.</p>
4308 </li>
4309 <li>
4310 <p>'<tt>fnty</tt>': shall be the signature of the pointer to function
4311 value being invoked. The argument types must match the types implied by
4312 this signature. This type can be omitted if the function is not varargs
4313 and if the function type does not return a pointer to a function.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4314 </li>
4315 <li>
4316 <p>'<tt>fnptrval</tt>': An LLVM value containing a pointer to a function to
4317 be invoked. In most cases, this is a direct function invocation, but
4318 indirect <tt>call</tt>s are just as possible, calling an arbitrary pointer
4319 to function value.</p>
4320 </li>
4321 <li>
4322 <p>'<tt>function args</tt>': argument list whose types match the
4323 function signature argument types. All arguments must be of
4324 <a href="#t_firstclass">first class</a> type. If the function signature
4325 indicates the function accepts a variable number of arguments, the extra
4326 arguments can be specified.</p>
4327 </li>
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]4328 <li>
Devang Pateld0bfcc72008-10-07 17:48:33 +0000[diff] [blame]4329 <p>The optional <a href="#fnattrs">function attributes</a> list. Only
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]4330 '<tt>noreturn</tt>', '<tt>nounwind</tt>', '<tt>readonly</tt>' and
4331 '<tt>readnone</tt>' attributes are valid here.</p>
4332 </li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4333</ol>
4334
4335<h5>Semantics:</h5>
4336
4337<p>The '<tt>call</tt>' instruction is used to cause control flow to
4338transfer to a specified function, with its incoming arguments bound to
4339the specified values. Upon a '<tt><a href="#i_ret">ret</a></tt>'
4340instruction in the called function, control flow continues with the
4341instruction after the function call, and the return value of the
Dan Gohman3e700032008-10-04 19:00:07 +0000[diff] [blame]4342function is bound to the result argument.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4343
4344<h5>Example:</h5>
4345
4346<pre>
Nick Lewycky93082fc2007-09-08 13:57:50 +0000[diff] [blame]4347 %retval = call i32 @test(i32 %argc)
Chris Lattner5e893ef2008-03-21 17:24:17 +0000[diff] [blame]4348 call i32 (i8 *, ...)* @printf(i8 * %msg, i32 12, i8 42) <i>; yields i32</i>
4349 %X = tail call i32 @foo() <i>; yields i32</i>
4350 %Y = tail call <a href="#callingconv">fastcc</a> i32 @foo() <i>; yields i32</i>
4351 call void %foo(i8 97 signext)
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]4352
4353 %struct.A = type { i32, i8 }
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]4354 %r = call %struct.A @foo() <i>; yields { 32, i8 }</i>
Dan Gohman3e700032008-10-04 19:00:07 +0000[diff] [blame]4355 %gr = extractvalue %struct.A %r, 0 <i>; yields i32</i>
4356 %gr1 = extractvalue %struct.A %r, 1 <i>; yields i8</i>
Chris Lattnerac454b32008-10-08 06:26:11 +0000[diff] [blame]4357 %Z = call void @foo() noreturn <i>; indicates that %foo never returns normally</i>
Matthijs Kooijman2c4e05a2008-10-07 10:03:45 +0000[diff] [blame]4358 %ZZ = call zeroext i32 @bar() <i>; Return value is %zero extended</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4359</pre>
4360
4361</div>
4362
4363<!-- _______________________________________________________________________ -->
4364<div class="doc_subsubsection">
4365 <a name="i_va_arg">'<tt>va_arg</tt>' Instruction</a>
4366</div>
4367
4368<div class="doc_text">
4369
4370<h5>Syntax:</h5>
4371
4372<pre>
4373 &lt;resultval&gt; = va_arg &lt;va_list*&gt; &lt;arglist&gt;, &lt;argty&gt;
4374</pre>
4375
4376<h5>Overview:</h5>
4377
4378<p>The '<tt>va_arg</tt>' instruction is used to access arguments passed through
4379the "variable argument" area of a function call. It is used to implement the
4380<tt>va_arg</tt> macro in C.</p>
4381
4382<h5>Arguments:</h5>
4383
4384<p>This instruction takes a <tt>va_list*</tt> value and the type of
4385the argument. It returns a value of the specified argument type and
4386increments the <tt>va_list</tt> to point to the next argument. The
4387actual type of <tt>va_list</tt> is target specific.</p>
4388
4389<h5>Semantics:</h5>
4390
4391<p>The '<tt>va_arg</tt>' instruction loads an argument of the specified
4392type from the specified <tt>va_list</tt> and causes the
4393<tt>va_list</tt> to point to the next argument. For more information,
4394see the variable argument handling <a href="#int_varargs">Intrinsic
4395Functions</a>.</p>
4396
4397<p>It is legal for this instruction to be called in a function which does not
4398take a variable number of arguments, for example, the <tt>vfprintf</tt>
4399function.</p>
4400
4401<p><tt>va_arg</tt> is an LLVM instruction instead of an <a
4402href="#intrinsics">intrinsic function</a> because it takes a type as an
4403argument.</p>
4404
4405<h5>Example:</h5>
4406
4407<p>See the <a href="#int_varargs">variable argument processing</a> section.</p>
4408
4409</div>
4410
4411<!-- *********************************************************************** -->
4412<div class="doc_section"> <a name="intrinsics">Intrinsic Functions</a> </div>
4413<!-- *********************************************************************** -->
4414
4415<div class="doc_text">
4416
4417<p>LLVM supports the notion of an "intrinsic function". These functions have
4418well known names and semantics and are required to follow certain restrictions.
4419Overall, these intrinsics represent an extension mechanism for the LLVM
4420language that does not require changing all of the transformations in LLVM when
4421adding to the language (or the bitcode reader/writer, the parser, etc...).</p>
4422
4423<p>Intrinsic function names must all start with an "<tt>llvm.</tt>" prefix. This
4424prefix is reserved in LLVM for intrinsic names; thus, function names may not
4425begin with this prefix. Intrinsic functions must always be external functions:
4426you cannot define the body of intrinsic functions. Intrinsic functions may
4427only be used in call or invoke instructions: it is illegal to take the address
4428of an intrinsic function. Additionally, because intrinsic functions are part
4429of the LLVM language, it is required if any are added that they be documented
4430here.</p>
4431
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]4432<p>Some intrinsic functions can be overloaded, i.e., the intrinsic represents
4433a family of functions that perform the same operation but on different data
4434types. Because LLVM can represent over 8 million different integer types,
4435overloading is used commonly to allow an intrinsic function to operate on any
4436integer type. One or more of the argument types or the result type can be
4437overloaded to accept any integer type. Argument types may also be defined as
4438exactly matching a previous argument's type or the result type. This allows an
4439intrinsic function which accepts multiple arguments, but needs all of them to
4440be of the same type, to only be overloaded with respect to a single argument or
4441the result.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4442
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]4443<p>Overloaded intrinsics will have the names of its overloaded argument types
4444encoded into its function name, each preceded by a period. Only those types
4445which are overloaded result in a name suffix. Arguments whose type is matched
4446against another type do not. For example, the <tt>llvm.ctpop</tt> function can
4447take an integer of any width and returns an integer of exactly the same integer
4448width. This leads to a family of functions such as
4449<tt>i8 @llvm.ctpop.i8(i8 %val)</tt> and <tt>i29 @llvm.ctpop.i29(i29 %val)</tt>.
4450Only one type, the return type, is overloaded, and only one type suffix is
4451required. Because the argument's type is matched against the return type, it
4452does not require its own name suffix.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4453
4454<p>To learn how to add an intrinsic function, please see the
4455<a href="ExtendingLLVM.html">Extending LLVM Guide</a>.
4456</p>
4457
4458</div>
4459
4460<!-- ======================================================================= -->
4461<div class="doc_subsection">
4462 <a name="int_varargs">Variable Argument Handling Intrinsics</a>
4463</div>
4464
4465<div class="doc_text">
4466
4467<p>Variable argument support is defined in LLVM with the <a
4468 href="#i_va_arg"><tt>va_arg</tt></a> instruction and these three
4469intrinsic functions. These functions are related to the similarly
4470named macros defined in the <tt>&lt;stdarg.h&gt;</tt> header file.</p>
4471
4472<p>All of these functions operate on arguments that use a
4473target-specific value type "<tt>va_list</tt>". The LLVM assembly
4474language reference manual does not define what this type is, so all
4475transformations should be prepared to handle these functions regardless of
4476the type used.</p>
4477
4478<p>This example shows how the <a href="#i_va_arg"><tt>va_arg</tt></a>
4479instruction and the variable argument handling intrinsic functions are
4480used.</p>
4481
4482<div class="doc_code">
4483<pre>
4484define i32 @test(i32 %X, ...) {
4485 ; Initialize variable argument processing
4486 %ap = alloca i8*
4487 %ap2 = bitcast i8** %ap to i8*
4488 call void @llvm.va_start(i8* %ap2)
4489
4490 ; Read a single integer argument
4491 %tmp = va_arg i8** %ap, i32
4492
4493 ; Demonstrate usage of llvm.va_copy and llvm.va_end
4494 %aq = alloca i8*
4495 %aq2 = bitcast i8** %aq to i8*
4496 call void @llvm.va_copy(i8* %aq2, i8* %ap2)
4497 call void @llvm.va_end(i8* %aq2)
4498
4499 ; Stop processing of arguments.
4500 call void @llvm.va_end(i8* %ap2)
4501 ret i32 %tmp
4502}
4503
4504declare void @llvm.va_start(i8*)
4505declare void @llvm.va_copy(i8*, i8*)
4506declare void @llvm.va_end(i8*)
4507</pre>
4508</div>
4509
4510</div>
4511
4512<!-- _______________________________________________________________________ -->
4513<div class="doc_subsubsection">
4514 <a name="int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a>
4515</div>
4516
4517
4518<div class="doc_text">
4519<h5>Syntax:</h5>
4520<pre> declare void %llvm.va_start(i8* &lt;arglist&gt;)<br></pre>
4521<h5>Overview:</h5>
4522<P>The '<tt>llvm.va_start</tt>' intrinsic initializes
4523<tt>*&lt;arglist&gt;</tt> for subsequent use by <tt><a
4524href="#i_va_arg">va_arg</a></tt>.</p>
4525
4526<h5>Arguments:</h5>
4527
4528<P>The argument is a pointer to a <tt>va_list</tt> element to initialize.</p>
4529
4530<h5>Semantics:</h5>
4531
4532<P>The '<tt>llvm.va_start</tt>' intrinsic works just like the <tt>va_start</tt>
4533macro available in C. In a target-dependent way, it initializes the
4534<tt>va_list</tt> element to which the argument points, so that the next call to
4535<tt>va_arg</tt> will produce the first variable argument passed to the function.
4536Unlike the C <tt>va_start</tt> macro, this intrinsic does not need to know the
4537last argument of the function as the compiler can figure that out.</p>
4538
4539</div>
4540
4541<!-- _______________________________________________________________________ -->
4542<div class="doc_subsubsection">
4543 <a name="int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a>
4544</div>
4545
4546<div class="doc_text">
4547<h5>Syntax:</h5>
4548<pre> declare void @llvm.va_end(i8* &lt;arglist&gt;)<br></pre>
4549<h5>Overview:</h5>
4550
4551<p>The '<tt>llvm.va_end</tt>' intrinsic destroys <tt>*&lt;arglist&gt;</tt>,
4552which has been initialized previously with <tt><a href="#int_va_start">llvm.va_start</a></tt>
4553or <tt><a href="#i_va_copy">llvm.va_copy</a></tt>.</p>
4554
4555<h5>Arguments:</h5>
4556
4557<p>The argument is a pointer to a <tt>va_list</tt> to destroy.</p>
4558
4559<h5>Semantics:</h5>
4560
4561<p>The '<tt>llvm.va_end</tt>' intrinsic works just like the <tt>va_end</tt>
4562macro available in C. In a target-dependent way, it destroys the
4563<tt>va_list</tt> element to which the argument points. Calls to <a
4564href="#int_va_start"><tt>llvm.va_start</tt></a> and <a href="#int_va_copy">
4565<tt>llvm.va_copy</tt></a> must be matched exactly with calls to
4566<tt>llvm.va_end</tt>.</p>
4567
4568</div>
4569
4570<!-- _______________________________________________________________________ -->
4571<div class="doc_subsubsection">
4572 <a name="int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a>
4573</div>
4574
4575<div class="doc_text">
4576
4577<h5>Syntax:</h5>
4578
4579<pre>
4580 declare void @llvm.va_copy(i8* &lt;destarglist&gt;, i8* &lt;srcarglist&gt;)
4581</pre>
4582
4583<h5>Overview:</h5>
4584
4585<p>The '<tt>llvm.va_copy</tt>' intrinsic copies the current argument position
4586from the source argument list to the destination argument list.</p>
4587
4588<h5>Arguments:</h5>
4589
4590<p>The first argument is a pointer to a <tt>va_list</tt> element to initialize.
4591The second argument is a pointer to a <tt>va_list</tt> element to copy from.</p>
4592
4593
4594<h5>Semantics:</h5>
4595
4596<p>The '<tt>llvm.va_copy</tt>' intrinsic works just like the <tt>va_copy</tt>
4597macro available in C. In a target-dependent way, it copies the source
4598<tt>va_list</tt> element into the destination <tt>va_list</tt> element. This
4599intrinsic is necessary because the <tt><a href="#int_va_start">
4600llvm.va_start</a></tt> intrinsic may be arbitrarily complex and require, for
4601example, memory allocation.</p>
4602
4603</div>
4604
4605<!-- ======================================================================= -->
4606<div class="doc_subsection">
4607 <a name="int_gc">Accurate Garbage Collection Intrinsics</a>
4608</div>
4609
4610<div class="doc_text">
4611
4612<p>
4613LLVM support for <a href="GarbageCollection.html">Accurate Garbage
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]4614Collection</a> (GC) requires the implementation and generation of these
4615intrinsics.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4616These intrinsics allow identification of <a href="#int_gcroot">GC roots on the
4617stack</a>, as well as garbage collector implementations that require <a
4618href="#int_gcread">read</a> and <a href="#int_gcwrite">write</a> barriers.
4619Front-ends for type-safe garbage collected languages should generate these
4620intrinsics to make use of the LLVM garbage collectors. For more details, see <a
4621href="GarbageCollection.html">Accurate Garbage Collection with LLVM</a>.
4622</p>
Christopher Lambcfe00962007-12-17 01:00:21 +0000[diff] [blame]4623
4624<p>The garbage collection intrinsics only operate on objects in the generic
4625 address space (address space zero).</p>
4626
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4627</div>
4628
4629<!-- _______________________________________________________________________ -->
4630<div class="doc_subsubsection">
4631 <a name="int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a>
4632</div>
4633
4634<div class="doc_text">
4635
4636<h5>Syntax:</h5>
4637
4638<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4639 declare void @llvm.gcroot(i8** %ptrloc, i8* %metadata)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4640</pre>
4641
4642<h5>Overview:</h5>
4643
4644<p>The '<tt>llvm.gcroot</tt>' intrinsic declares the existence of a GC root to
4645the code generator, and allows some metadata to be associated with it.</p>
4646
4647<h5>Arguments:</h5>
4648
4649<p>The first argument specifies the address of a stack object that contains the
4650root pointer. The second pointer (which must be either a constant or a global
4651value address) contains the meta-data to be associated with the root.</p>
4652
4653<h5>Semantics:</h5>
4654
Chris Lattnera7d94ba2008-04-24 05:59:56 +0000[diff] [blame]4655<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]4656location. At compile-time, the code generator generates information to allow
Gordon Henriksen40393542007-12-25 02:31:26 +0000[diff] [blame]4657the runtime to find the pointer at GC safe points. The '<tt>llvm.gcroot</tt>'
4658intrinsic may only be used in a function which <a href="#gc">specifies a GC
4659algorithm</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4660
4661</div>
4662
4663
4664<!-- _______________________________________________________________________ -->
4665<div class="doc_subsubsection">
4666 <a name="int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a>
4667</div>
4668
4669<div class="doc_text">
4670
4671<h5>Syntax:</h5>
4672
4673<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4674 declare i8* @llvm.gcread(i8* %ObjPtr, i8** %Ptr)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4675</pre>
4676
4677<h5>Overview:</h5>
4678
4679<p>The '<tt>llvm.gcread</tt>' intrinsic identifies reads of references from heap
4680locations, allowing garbage collector implementations that require read
4681barriers.</p>
4682
4683<h5>Arguments:</h5>
4684
4685<p>The second argument is the address to read from, which should be an address
4686allocated from the garbage collector. The first object is a pointer to the
4687start of the referenced object, if needed by the language runtime (otherwise
4688null).</p>
4689
4690<h5>Semantics:</h5>
4691
4692<p>The '<tt>llvm.gcread</tt>' intrinsic has the same semantics as a load
4693instruction, but may be replaced with substantially more complex code by the
Gordon Henriksen40393542007-12-25 02:31:26 +0000[diff] [blame]4694garbage collector runtime, as needed. The '<tt>llvm.gcread</tt>' intrinsic
4695may only be used in a function which <a href="#gc">specifies a GC
4696algorithm</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4697
4698</div>
4699
4700
4701<!-- _______________________________________________________________________ -->
4702<div class="doc_subsubsection">
4703 <a name="int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a>
4704</div>
4705
4706<div class="doc_text">
4707
4708<h5>Syntax:</h5>
4709
4710<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4711 declare void @llvm.gcwrite(i8* %P1, i8* %Obj, i8** %P2)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4712</pre>
4713
4714<h5>Overview:</h5>
4715
4716<p>The '<tt>llvm.gcwrite</tt>' intrinsic identifies writes of references to heap
4717locations, allowing garbage collector implementations that require write
4718barriers (such as generational or reference counting collectors).</p>
4719
4720<h5>Arguments:</h5>
4721
4722<p>The first argument is the reference to store, the second is the start of the
4723object to store it to, and the third is the address of the field of Obj to
4724store to. If the runtime does not require a pointer to the object, Obj may be
4725null.</p>
4726
4727<h5>Semantics:</h5>
4728
4729<p>The '<tt>llvm.gcwrite</tt>' intrinsic has the same semantics as a store
4730instruction, but may be replaced with substantially more complex code by the
Gordon Henriksen40393542007-12-25 02:31:26 +0000[diff] [blame]4731garbage collector runtime, as needed. The '<tt>llvm.gcwrite</tt>' intrinsic
4732may only be used in a function which <a href="#gc">specifies a GC
4733algorithm</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4734
4735</div>
4736
4737
4738
4739<!-- ======================================================================= -->
4740<div class="doc_subsection">
4741 <a name="int_codegen">Code Generator Intrinsics</a>
4742</div>
4743
4744<div class="doc_text">
4745<p>
4746These intrinsics are provided by LLVM to expose special features that may only
4747be implemented with code generator support.
4748</p>
4749
4750</div>
4751
4752<!-- _______________________________________________________________________ -->
4753<div class="doc_subsubsection">
4754 <a name="int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a>
4755</div>
4756
4757<div class="doc_text">
4758
4759<h5>Syntax:</h5>
4760<pre>
4761 declare i8 *@llvm.returnaddress(i32 &lt;level&gt;)
4762</pre>
4763
4764<h5>Overview:</h5>
4765
4766<p>
4767The '<tt>llvm.returnaddress</tt>' intrinsic attempts to compute a
4768target-specific value indicating the return address of the current function
4769or one of its callers.
4770</p>
4771
4772<h5>Arguments:</h5>
4773
4774<p>
4775The argument to this intrinsic indicates which function to return the address
4776for. Zero indicates the calling function, one indicates its caller, etc. The
4777argument is <b>required</b> to be a constant integer value.
4778</p>
4779
4780<h5>Semantics:</h5>
4781
4782<p>
4783The '<tt>llvm.returnaddress</tt>' intrinsic either returns a pointer indicating
4784the return address of the specified call frame, or zero if it cannot be
4785identified. The value returned by this intrinsic is likely to be incorrect or 0
4786for arguments other than zero, so it should only be used for debugging purposes.
4787</p>
4788
4789<p>
4790Note that calling this intrinsic does not prevent function inlining or other
4791aggressive transformations, so the value returned may not be that of the obvious
4792source-language caller.
4793</p>
4794</div>
4795
4796
4797<!-- _______________________________________________________________________ -->
4798<div class="doc_subsubsection">
4799 <a name="int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a>
4800</div>
4801
4802<div class="doc_text">
4803
4804<h5>Syntax:</h5>
4805<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4806 declare i8 *@llvm.frameaddress(i32 &lt;level&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4807</pre>
4808
4809<h5>Overview:</h5>
4810
4811<p>
4812The '<tt>llvm.frameaddress</tt>' intrinsic attempts to return the
4813target-specific frame pointer value for the specified stack frame.
4814</p>
4815
4816<h5>Arguments:</h5>
4817
4818<p>
4819The argument to this intrinsic indicates which function to return the frame
4820pointer for. Zero indicates the calling function, one indicates its caller,
4821etc. The argument is <b>required</b> to be a constant integer value.
4822</p>
4823
4824<h5>Semantics:</h5>
4825
4826<p>
4827The '<tt>llvm.frameaddress</tt>' intrinsic either returns a pointer indicating
4828the frame address of the specified call frame, or zero if it cannot be
4829identified. The value returned by this intrinsic is likely to be incorrect or 0
4830for arguments other than zero, so it should only be used for debugging purposes.
4831</p>
4832
4833<p>
4834Note that calling this intrinsic does not prevent function inlining or other
4835aggressive transformations, so the value returned may not be that of the obvious
4836source-language caller.
4837</p>
4838</div>
4839
4840<!-- _______________________________________________________________________ -->
4841<div class="doc_subsubsection">
4842 <a name="int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a>
4843</div>
4844
4845<div class="doc_text">
4846
4847<h5>Syntax:</h5>
4848<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4849 declare i8 *@llvm.stacksave()
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4850</pre>
4851
4852<h5>Overview:</h5>
4853
4854<p>
4855The '<tt>llvm.stacksave</tt>' intrinsic is used to remember the current state of
4856the function stack, for use with <a href="#int_stackrestore">
4857<tt>llvm.stackrestore</tt></a>. This is useful for implementing language
4858features like scoped automatic variable sized arrays in C99.
4859</p>
4860
4861<h5>Semantics:</h5>
4862
4863<p>
4864This intrinsic returns a opaque pointer value that can be passed to <a
4865href="#int_stackrestore"><tt>llvm.stackrestore</tt></a>. When an
4866<tt>llvm.stackrestore</tt> intrinsic is executed with a value saved from
4867<tt>llvm.stacksave</tt>, it effectively restores the state of the stack to the
4868state it was in when the <tt>llvm.stacksave</tt> intrinsic executed. In
4869practice, this pops any <a href="#i_alloca">alloca</a> blocks from the stack
4870that were allocated after the <tt>llvm.stacksave</tt> was executed.
4871</p>
4872
4873</div>
4874
4875<!-- _______________________________________________________________________ -->
4876<div class="doc_subsubsection">
4877 <a name="int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a>
4878</div>
4879
4880<div class="doc_text">
4881
4882<h5>Syntax:</h5>
4883<pre>
4884 declare void @llvm.stackrestore(i8 * %ptr)
4885</pre>
4886
4887<h5>Overview:</h5>
4888
4889<p>
4890The '<tt>llvm.stackrestore</tt>' intrinsic is used to restore the state of
4891the function stack to the state it was in when the corresponding <a
4892href="#int_stacksave"><tt>llvm.stacksave</tt></a> intrinsic executed. This is
4893useful for implementing language features like scoped automatic variable sized
4894arrays in C99.
4895</p>
4896
4897<h5>Semantics:</h5>
4898
4899<p>
4900See the description for <a href="#int_stacksave"><tt>llvm.stacksave</tt></a>.
4901</p>
4902
4903</div>
4904
4905
4906<!-- _______________________________________________________________________ -->
4907<div class="doc_subsubsection">
4908 <a name="int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a>
4909</div>
4910
4911<div class="doc_text">
4912
4913<h5>Syntax:</h5>
4914<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4915 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]4916</pre>
4917
4918<h5>Overview:</h5>
4919
4920
4921<p>
4922The '<tt>llvm.prefetch</tt>' intrinsic is a hint to the code generator to insert
4923a prefetch instruction if supported; otherwise, it is a noop. Prefetches have
4924no
4925effect on the behavior of the program but can change its performance
4926characteristics.
4927</p>
4928
4929<h5>Arguments:</h5>
4930
4931<p>
4932<tt>address</tt> is the address to be prefetched, <tt>rw</tt> is the specifier
4933determining if the fetch should be for a read (0) or write (1), and
4934<tt>locality</tt> is a temporal locality specifier ranging from (0) - no
4935locality, to (3) - extremely local keep in cache. The <tt>rw</tt> and
4936<tt>locality</tt> arguments must be constant integers.
4937</p>
4938
4939<h5>Semantics:</h5>
4940
4941<p>
4942This intrinsic does not modify the behavior of the program. In particular,
4943prefetches cannot trap and do not produce a value. On targets that support this
4944intrinsic, the prefetch can provide hints to the processor cache for better
4945performance.
4946</p>
4947
4948</div>
4949
4950<!-- _______________________________________________________________________ -->
4951<div class="doc_subsubsection">
4952 <a name="int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a>
4953</div>
4954
4955<div class="doc_text">
4956
4957<h5>Syntax:</h5>
4958<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4959 declare void @llvm.pcmarker(i32 &lt;id&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4960</pre>
4961
4962<h5>Overview:</h5>
4963
4964
4965<p>
4966The '<tt>llvm.pcmarker</tt>' intrinsic is a method to export a Program Counter
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]4967(PC) in a region of
4968code to simulators and other tools. The method is target specific, but it is
4969expected that the marker will use exported symbols to transmit the PC of the
4970marker.
4971The marker makes no guarantees that it will remain with any specific instruction
4972after optimizations. It is possible that the presence of a marker will inhibit
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4973optimizations. The intended use is to be inserted after optimizations to allow
4974correlations of simulation runs.
4975</p>
4976
4977<h5>Arguments:</h5>
4978
4979<p>
4980<tt>id</tt> is a numerical id identifying the marker.
4981</p>
4982
4983<h5>Semantics:</h5>
4984
4985<p>
4986This intrinsic does not modify the behavior of the program. Backends that do not
4987support this intrinisic may ignore it.
4988</p>
4989
4990</div>
4991
4992<!-- _______________________________________________________________________ -->
4993<div class="doc_subsubsection">
4994 <a name="int_readcyclecounter">'<tt>llvm.readcyclecounter</tt>' Intrinsic</a>
4995</div>
4996
4997<div class="doc_text">
4998
4999<h5>Syntax:</h5>
5000<pre>
5001 declare i64 @llvm.readcyclecounter( )
5002</pre>
5003
5004<h5>Overview:</h5>
5005
5006
5007<p>
5008The '<tt>llvm.readcyclecounter</tt>' intrinsic provides access to the cycle
5009counter register (or similar low latency, high accuracy clocks) on those targets
5010that support it. On X86, it should map to RDTSC. On Alpha, it should map to RPCC.
5011As the backing counters overflow quickly (on the order of 9 seconds on alpha), this
5012should only be used for small timings.
5013</p>
5014
5015<h5>Semantics:</h5>
5016
5017<p>
5018When directly supported, reading the cycle counter should not modify any memory.
5019Implementations are allowed to either return a application specific value or a
5020system wide value. On backends without support, this is lowered to a constant 0.
5021</p>
5022
5023</div>
5024
5025<!-- ======================================================================= -->
5026<div class="doc_subsection">
5027 <a name="int_libc">Standard C Library Intrinsics</a>
5028</div>
5029
5030<div class="doc_text">
5031<p>
5032LLVM provides intrinsics for a few important standard C library functions.
5033These intrinsics allow source-language front-ends to pass information about the
5034alignment of the pointer arguments to the code generator, providing opportunity
5035for more efficient code generation.
5036</p>
5037
5038</div>
5039
5040<!-- _______________________________________________________________________ -->
5041<div class="doc_subsubsection">
5042 <a name="int_memcpy">'<tt>llvm.memcpy</tt>' Intrinsic</a>
5043</div>
5044
5045<div class="doc_text">
5046
5047<h5>Syntax:</h5>
5048<pre>
5049 declare void @llvm.memcpy.i32(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5050 i32 &lt;len&gt;, i32 &lt;align&gt;)
5051 declare void @llvm.memcpy.i64(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5052 i64 &lt;len&gt;, i32 &lt;align&gt;)
5053</pre>
5054
5055<h5>Overview:</h5>
5056
5057<p>
5058The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the source
5059location to the destination location.
5060</p>
5061
5062<p>
5063Note that, unlike the standard libc function, the <tt>llvm.memcpy.*</tt>
5064intrinsics do not return a value, and takes an extra alignment argument.
5065</p>
5066
5067<h5>Arguments:</h5>
5068
5069<p>
5070The first argument is a pointer to the destination, the second is a pointer to
5071the source. The third argument is an integer argument
5072specifying the number of bytes to copy, and the fourth argument is the alignment
5073of the source and destination locations.
5074</p>
5075
5076<p>
5077If the call to this intrinisic has an alignment value that is not 0 or 1, then
5078the caller guarantees that both the source and destination pointers are aligned
5079to that boundary.
5080</p>
5081
5082<h5>Semantics:</h5>
5083
5084<p>
5085The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the source
5086location to the destination location, which are not allowed to overlap. It
5087copies "len" bytes of memory over. If the argument is known to be aligned to
5088some boundary, this can be specified as the fourth argument, otherwise it should
5089be set to 0 or 1.
5090</p>
5091</div>
5092
5093
5094<!-- _______________________________________________________________________ -->
5095<div class="doc_subsubsection">
5096 <a name="int_memmove">'<tt>llvm.memmove</tt>' Intrinsic</a>
5097</div>
5098
5099<div class="doc_text">
5100
5101<h5>Syntax:</h5>
5102<pre>
5103 declare void @llvm.memmove.i32(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5104 i32 &lt;len&gt;, i32 &lt;align&gt;)
5105 declare void @llvm.memmove.i64(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5106 i64 &lt;len&gt;, i32 &lt;align&gt;)
5107</pre>
5108
5109<h5>Overview:</h5>
5110
5111<p>
5112The '<tt>llvm.memmove.*</tt>' intrinsics move a block of memory from the source
5113location to the destination location. It is similar to the
Chris Lattnerdba16ea2008-01-06 19:51:52 +0000[diff] [blame]5114'<tt>llvm.memcpy</tt>' intrinsic but allows the two memory locations to overlap.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5115</p>
5116
5117<p>
5118Note that, unlike the standard libc function, the <tt>llvm.memmove.*</tt>
5119intrinsics do not return a value, and takes an extra alignment argument.
5120</p>
5121
5122<h5>Arguments:</h5>
5123
5124<p>
5125The first argument is a pointer to the destination, the second is a pointer to
5126the source. The third argument is an integer argument
5127specifying the number of bytes to copy, and the fourth argument is the alignment
5128of the source and destination locations.
5129</p>
5130
5131<p>
5132If the call to this intrinisic has an alignment value that is not 0 or 1, then
5133the caller guarantees that the source and destination pointers are aligned to
5134that boundary.
5135</p>
5136
5137<h5>Semantics:</h5>
5138
5139<p>
5140The '<tt>llvm.memmove.*</tt>' intrinsics copy a block of memory from the source
5141location to the destination location, which may overlap. It
5142copies "len" bytes of memory over. If the argument is known to be aligned to
5143some boundary, this can be specified as the fourth argument, otherwise it should
5144be set to 0 or 1.
5145</p>
5146</div>
5147
5148
5149<!-- _______________________________________________________________________ -->
5150<div class="doc_subsubsection">
5151 <a name="int_memset">'<tt>llvm.memset.*</tt>' Intrinsics</a>
5152</div>
5153
5154<div class="doc_text">
5155
5156<h5>Syntax:</h5>
5157<pre>
5158 declare void @llvm.memset.i32(i8 * &lt;dest&gt;, i8 &lt;val&gt;,
5159 i32 &lt;len&gt;, i32 &lt;align&gt;)
5160 declare void @llvm.memset.i64(i8 * &lt;dest&gt;, i8 &lt;val&gt;,
5161 i64 &lt;len&gt;, i32 &lt;align&gt;)
5162</pre>
5163
5164<h5>Overview:</h5>
5165
5166<p>
5167The '<tt>llvm.memset.*</tt>' intrinsics fill a block of memory with a particular
5168byte value.
5169</p>
5170
5171<p>
5172Note that, unlike the standard libc function, the <tt>llvm.memset</tt> intrinsic
5173does not return a value, and takes an extra alignment argument.
5174</p>
5175
5176<h5>Arguments:</h5>
5177
5178<p>
5179The first argument is a pointer to the destination to fill, the second is the
5180byte value to fill it with, the third argument is an integer
5181argument specifying the number of bytes to fill, and the fourth argument is the
5182known alignment of destination location.
5183</p>
5184
5185<p>
5186If the call to this intrinisic has an alignment value that is not 0 or 1, then
5187the caller guarantees that the destination pointer is aligned to that boundary.
5188</p>
5189
5190<h5>Semantics:</h5>
5191
5192<p>
5193The '<tt>llvm.memset.*</tt>' intrinsics fill "len" bytes of memory starting at
5194the
5195destination location. If the argument is known to be aligned to some boundary,
5196this can be specified as the fourth argument, otherwise it should be set to 0 or
51971.
5198</p>
5199</div>
5200
5201
5202<!-- _______________________________________________________________________ -->
5203<div class="doc_subsubsection">
5204 <a name="int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a>
5205</div>
5206
5207<div class="doc_text">
5208
5209<h5>Syntax:</h5>
Dale Johannesenf9adbb62007-10-02 17:47:38 +0000[diff] [blame]5210<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]5211floating point or vector of floating point type. Not all targets support all
5212types however.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5213<pre>
Dale Johannesenf9adbb62007-10-02 17:47:38 +0000[diff] [blame]5214 declare float @llvm.sqrt.f32(float %Val)
5215 declare double @llvm.sqrt.f64(double %Val)
5216 declare x86_fp80 @llvm.sqrt.f80(x86_fp80 %Val)
5217 declare fp128 @llvm.sqrt.f128(fp128 %Val)
5218 declare ppc_fp128 @llvm.sqrt.ppcf128(ppc_fp128 %Val)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5219</pre>
5220
5221<h5>Overview:</h5>
5222
5223<p>
5224The '<tt>llvm.sqrt</tt>' intrinsics return the sqrt of the specified operand,
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5225returning the same value as the libm '<tt>sqrt</tt>' functions would. Unlike
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5226<tt>sqrt</tt> in libm, however, <tt>llvm.sqrt</tt> has undefined behavior for
Chris Lattnerf914a002008-01-29 07:00:44 +0000[diff] [blame]5227negative numbers other than -0.0 (which allows for better optimization, because
5228there is no need to worry about errno being set). <tt>llvm.sqrt(-0.0)</tt> is
5229defined to return -0.0 like IEEE sqrt.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5230</p>
5231
5232<h5>Arguments:</h5>
5233
5234<p>
5235The argument and return value are floating point numbers of the same type.
5236</p>
5237
5238<h5>Semantics:</h5>
5239
5240<p>
5241This function returns the sqrt of the specified operand if it is a nonnegative
5242floating point number.
5243</p>
5244</div>
5245
5246<!-- _______________________________________________________________________ -->
5247<div class="doc_subsubsection">
5248 <a name="int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a>
5249</div>
5250
5251<div class="doc_text">
5252
5253<h5>Syntax:</h5>
Dale Johannesenf9adbb62007-10-02 17:47:38 +0000[diff] [blame]5254<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]5255floating point or vector of floating point type. Not all targets support all
5256types however.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5257<pre>
Dale Johannesenf9adbb62007-10-02 17:47:38 +0000[diff] [blame]5258 declare float @llvm.powi.f32(float %Val, i32 %power)
5259 declare double @llvm.powi.f64(double %Val, i32 %power)
5260 declare x86_fp80 @llvm.powi.f80(x86_fp80 %Val, i32 %power)
5261 declare fp128 @llvm.powi.f128(fp128 %Val, i32 %power)
5262 declare ppc_fp128 @llvm.powi.ppcf128(ppc_fp128 %Val, i32 %power)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5263</pre>
5264
5265<h5>Overview:</h5>
5266
5267<p>
5268The '<tt>llvm.powi.*</tt>' intrinsics return the first operand raised to the
5269specified (positive or negative) power. The order of evaluation of
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5270multiplications is not defined. When a vector of floating point type is
5271used, the second argument remains a scalar integer value.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5272</p>
5273
5274<h5>Arguments:</h5>
5275
5276<p>
5277The second argument is an integer power, and the first is a value to raise to
5278that power.
5279</p>
5280
5281<h5>Semantics:</h5>
5282
5283<p>
5284This function returns the first value raised to the second power with an
5285unspecified sequence of rounding operations.</p>
5286</div>
5287
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5288<!-- _______________________________________________________________________ -->
5289<div class="doc_subsubsection">
5290 <a name="int_sin">'<tt>llvm.sin.*</tt>' Intrinsic</a>
5291</div>
5292
5293<div class="doc_text">
5294
5295<h5>Syntax:</h5>
5296<p>This is an overloaded intrinsic. You can use <tt>llvm.sin</tt> on any
5297floating point or vector of floating point type. Not all targets support all
5298types however.
5299<pre>
5300 declare float @llvm.sin.f32(float %Val)
5301 declare double @llvm.sin.f64(double %Val)
5302 declare x86_fp80 @llvm.sin.f80(x86_fp80 %Val)
5303 declare fp128 @llvm.sin.f128(fp128 %Val)
5304 declare ppc_fp128 @llvm.sin.ppcf128(ppc_fp128 %Val)
5305</pre>
5306
5307<h5>Overview:</h5>
5308
5309<p>
5310The '<tt>llvm.sin.*</tt>' intrinsics return the sine of the operand.
5311</p>
5312
5313<h5>Arguments:</h5>
5314
5315<p>
5316The argument and return value are floating point numbers of the same type.
5317</p>
5318
5319<h5>Semantics:</h5>
5320
5321<p>
5322This function returns the sine of the specified operand, returning the
5323same values as the libm <tt>sin</tt> functions would, and handles error
Dan Gohmaneaba92e2007-10-17 18:05:13 +0000[diff] [blame]5324conditions in the same way.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5325</div>
5326
5327<!-- _______________________________________________________________________ -->
5328<div class="doc_subsubsection">
5329 <a name="int_cos">'<tt>llvm.cos.*</tt>' Intrinsic</a>
5330</div>
5331
5332<div class="doc_text">
5333
5334<h5>Syntax:</h5>
5335<p>This is an overloaded intrinsic. You can use <tt>llvm.cos</tt> on any
5336floating point or vector of floating point type. Not all targets support all
5337types however.
5338<pre>
5339 declare float @llvm.cos.f32(float %Val)
5340 declare double @llvm.cos.f64(double %Val)
5341 declare x86_fp80 @llvm.cos.f80(x86_fp80 %Val)
5342 declare fp128 @llvm.cos.f128(fp128 %Val)
5343 declare ppc_fp128 @llvm.cos.ppcf128(ppc_fp128 %Val)
5344</pre>
5345
5346<h5>Overview:</h5>
5347
5348<p>
5349The '<tt>llvm.cos.*</tt>' intrinsics return the cosine of the operand.
5350</p>
5351
5352<h5>Arguments:</h5>
5353
5354<p>
5355The argument and return value are floating point numbers of the same type.
5356</p>
5357
5358<h5>Semantics:</h5>
5359
5360<p>
5361This function returns the cosine of the specified operand, returning the
5362same values as the libm <tt>cos</tt> functions would, and handles error
Dan Gohmaneaba92e2007-10-17 18:05:13 +0000[diff] [blame]5363conditions in the same way.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5364</div>
5365
5366<!-- _______________________________________________________________________ -->
5367<div class="doc_subsubsection">
5368 <a name="int_pow">'<tt>llvm.pow.*</tt>' Intrinsic</a>
5369</div>
5370
5371<div class="doc_text">
5372
5373<h5>Syntax:</h5>
5374<p>This is an overloaded intrinsic. You can use <tt>llvm.pow</tt> on any
5375floating point or vector of floating point type. Not all targets support all
5376types however.
5377<pre>
5378 declare float @llvm.pow.f32(float %Val, float %Power)
5379 declare double @llvm.pow.f64(double %Val, double %Power)
5380 declare x86_fp80 @llvm.pow.f80(x86_fp80 %Val, x86_fp80 %Power)
5381 declare fp128 @llvm.pow.f128(fp128 %Val, fp128 %Power)
5382 declare ppc_fp128 @llvm.pow.ppcf128(ppc_fp128 %Val, ppc_fp128 Power)
5383</pre>
5384
5385<h5>Overview:</h5>
5386
5387<p>
5388The '<tt>llvm.pow.*</tt>' intrinsics return the first operand raised to the
5389specified (positive or negative) power.
5390</p>
5391
5392<h5>Arguments:</h5>
5393
5394<p>
5395The second argument is a floating point power, and the first is a value to
5396raise to that power.
5397</p>
5398
5399<h5>Semantics:</h5>
5400
5401<p>
5402This function returns the first value raised to the second power,
5403returning the
5404same values as the libm <tt>pow</tt> functions would, and handles error
Dan Gohmaneaba92e2007-10-17 18:05:13 +0000[diff] [blame]5405conditions in the same way.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5406</div>
5407
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5408
5409<!-- ======================================================================= -->
5410<div class="doc_subsection">
5411 <a name="int_manip">Bit Manipulation Intrinsics</a>
5412</div>
5413
5414<div class="doc_text">
5415<p>
5416LLVM provides intrinsics for a few important bit manipulation operations.
5417These allow efficient code generation for some algorithms.
5418</p>
5419
5420</div>
5421
5422<!-- _______________________________________________________________________ -->
5423<div class="doc_subsubsection">
5424 <a name="int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a>
5425</div>
5426
5427<div class="doc_text">
5428
5429<h5>Syntax:</h5>
5430<p>This is an overloaded intrinsic function. You can use bswap on any integer
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5431type that is an even number of bytes (i.e. BitWidth % 16 == 0).
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5432<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5433 declare i16 @llvm.bswap.i16(i16 &lt;id&gt;)
5434 declare i32 @llvm.bswap.i32(i32 &lt;id&gt;)
5435 declare i64 @llvm.bswap.i64(i64 &lt;id&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5436</pre>
5437
5438<h5>Overview:</h5>
5439
5440<p>
5441The '<tt>llvm.bswap</tt>' family of intrinsics is used to byte swap integer
5442values with an even number of bytes (positive multiple of 16 bits). These are
5443useful for performing operations on data that is not in the target's native
5444byte order.
5445</p>
5446
5447<h5>Semantics:</h5>
5448
5449<p>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5450The <tt>llvm.bswap.i16</tt> intrinsic returns an i16 value that has the high
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5451and low byte of the input i16 swapped. Similarly, the <tt>llvm.bswap.i32</tt>
5452intrinsic returns an i32 value that has the four bytes of the input i32
5453swapped, 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]5454i32 will have its bytes in 3, 2, 1, 0 order. The <tt>llvm.bswap.i48</tt>,
5455<tt>llvm.bswap.i64</tt> and other intrinsics extend this concept to
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5456additional even-byte lengths (6 bytes, 8 bytes and more, respectively).
5457</p>
5458
5459</div>
5460
5461<!-- _______________________________________________________________________ -->
5462<div class="doc_subsubsection">
5463 <a name="int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic</a>
5464</div>
5465
5466<div class="doc_text">
5467
5468<h5>Syntax:</h5>
5469<p>This is an overloaded intrinsic. You can use llvm.ctpop on any integer bit
5470width. Not all targets support all bit widths however.
5471<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5472 declare i8 @llvm.ctpop.i8 (i8 &lt;src&gt;)
5473 declare i16 @llvm.ctpop.i16(i16 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5474 declare i32 @llvm.ctpop.i32(i32 &lt;src&gt;)
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5475 declare i64 @llvm.ctpop.i64(i64 &lt;src&gt;)
5476 declare i256 @llvm.ctpop.i256(i256 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5477</pre>
5478
5479<h5>Overview:</h5>
5480
5481<p>
5482The '<tt>llvm.ctpop</tt>' family of intrinsics counts the number of bits set in a
5483value.
5484</p>
5485
5486<h5>Arguments:</h5>
5487
5488<p>
5489The only argument is the value to be counted. The argument may be of any
5490integer type. The return type must match the argument type.
5491</p>
5492
5493<h5>Semantics:</h5>
5494
5495<p>
5496The '<tt>llvm.ctpop</tt>' intrinsic counts the 1's in a variable.
5497</p>
5498</div>
5499
5500<!-- _______________________________________________________________________ -->
5501<div class="doc_subsubsection">
5502 <a name="int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic</a>
5503</div>
5504
5505<div class="doc_text">
5506
5507<h5>Syntax:</h5>
5508<p>This is an overloaded intrinsic. You can use <tt>llvm.ctlz</tt> on any
5509integer bit width. Not all targets support all bit widths however.
5510<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5511 declare i8 @llvm.ctlz.i8 (i8 &lt;src&gt;)
5512 declare i16 @llvm.ctlz.i16(i16 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5513 declare i32 @llvm.ctlz.i32(i32 &lt;src&gt;)
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5514 declare i64 @llvm.ctlz.i64(i64 &lt;src&gt;)
5515 declare i256 @llvm.ctlz.i256(i256 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5516</pre>
5517
5518<h5>Overview:</h5>
5519
5520<p>
5521The '<tt>llvm.ctlz</tt>' family of intrinsic functions counts the number of
5522leading zeros in a variable.
5523</p>
5524
5525<h5>Arguments:</h5>
5526
5527<p>
5528The only argument is the value to be counted. The argument may be of any
5529integer type. The return type must match the argument type.
5530</p>
5531
5532<h5>Semantics:</h5>
5533
5534<p>
5535The '<tt>llvm.ctlz</tt>' intrinsic counts the leading (most significant) zeros
5536in a variable. If the src == 0 then the result is the size in bits of the type
5537of src. For example, <tt>llvm.ctlz(i32 2) = 30</tt>.
5538</p>
5539</div>
5540
5541
5542
5543<!-- _______________________________________________________________________ -->
5544<div class="doc_subsubsection">
5545 <a name="int_cttz">'<tt>llvm.cttz.*</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.cttz</tt> on any
5552integer bit width. Not all targets support all bit widths however.
5553<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5554 declare i8 @llvm.cttz.i8 (i8 &lt;src&gt;)
5555 declare i16 @llvm.cttz.i16(i16 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5556 declare i32 @llvm.cttz.i32(i32 &lt;src&gt;)
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5557 declare i64 @llvm.cttz.i64(i64 &lt;src&gt;)
5558 declare i256 @llvm.cttz.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.cttz</tt>' family of intrinsic functions counts the number of
5565trailing zeros.
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.cttz</tt>' intrinsic counts the trailing (least 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.cttz(2) = 1</tt>.
5581</p>
5582</div>
5583
5584<!-- _______________________________________________________________________ -->
5585<div class="doc_subsubsection">
5586 <a name="int_part_select">'<tt>llvm.part.select.*</tt>' Intrinsic</a>
5587</div>
5588
5589<div class="doc_text">
5590
5591<h5>Syntax:</h5>
5592<p>This is an overloaded intrinsic. You can use <tt>llvm.part.select</tt>
5593on any integer bit width.
5594<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5595 declare i17 @llvm.part.select.i17 (i17 %val, i32 %loBit, i32 %hiBit)
5596 declare i29 @llvm.part.select.i29 (i29 %val, i32 %loBit, i32 %hiBit)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5597</pre>
5598
5599<h5>Overview:</h5>
5600<p>The '<tt>llvm.part.select</tt>' family of intrinsic functions selects a
5601range of bits from an integer value and returns them in the same bit width as
5602the original value.</p>
5603
5604<h5>Arguments:</h5>
5605<p>The first argument, <tt>%val</tt> and the result may be integer types of
5606any bit width but they must have the same bit width. The second and third
5607arguments must be <tt>i32</tt> type since they specify only a bit index.</p>
5608
5609<h5>Semantics:</h5>
5610<p>The operation of the '<tt>llvm.part.select</tt>' intrinsic has two modes
5611of operation: forwards and reverse. If <tt>%loBit</tt> is greater than
5612<tt>%hiBits</tt> then the intrinsic operates in reverse mode. Otherwise it
5613operates in forward mode.</p>
5614<p>In forward mode, this intrinsic is the equivalent of shifting <tt>%val</tt>
5615right by <tt>%loBit</tt> bits and then ANDing it with a mask with
5616only the <tt>%hiBit - %loBit</tt> bits set, as follows:</p>
5617<ol>
5618 <li>The <tt>%val</tt> is shifted right (LSHR) by the number of bits specified
5619 by <tt>%loBits</tt>. This normalizes the value to the low order bits.</li>
5620 <li>The <tt>%loBits</tt> value is subtracted from the <tt>%hiBits</tt> value
5621 to determine the number of bits to retain.</li>
5622 <li>A mask of the retained bits is created by shifting a -1 value.</li>
5623 <li>The mask is ANDed with <tt>%val</tt> to produce the result.
5624</ol>
5625<p>In reverse mode, a similar computation is made except that the bits are
5626returned in the reverse order. So, for example, if <tt>X</tt> has the value
5627<tt>i16 0x0ACF (101011001111)</tt> and we apply
5628<tt>part.select(i16 X, 8, 3)</tt> to it, we get back the value
5629<tt>i16 0x0026 (000000100110)</tt>.</p>
5630</div>
5631
5632<div class="doc_subsubsection">
5633 <a name="int_part_set">'<tt>llvm.part.set.*</tt>' Intrinsic</a>
5634</div>
5635
5636<div class="doc_text">
5637
5638<h5>Syntax:</h5>
5639<p>This is an overloaded intrinsic. You can use <tt>llvm.part.set</tt>
5640on any integer bit width.
5641<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5642 declare i17 @llvm.part.set.i17.i9 (i17 %val, i9 %repl, i32 %lo, i32 %hi)
5643 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]5644</pre>
5645
5646<h5>Overview:</h5>
5647<p>The '<tt>llvm.part.set</tt>' family of intrinsic functions replaces a range
5648of bits in an integer value with another integer value. It returns the integer
5649with the replaced bits.</p>
5650
5651<h5>Arguments:</h5>
5652<p>The first argument, <tt>%val</tt> and the result may be integer types of
5653any bit width but they must have the same bit width. <tt>%val</tt> is the value
5654whose bits will be replaced. The second argument, <tt>%repl</tt> may be an
5655integer of any bit width. The third and fourth arguments must be <tt>i32</tt>
5656type since they specify only a bit index.</p>
5657
5658<h5>Semantics:</h5>
5659<p>The operation of the '<tt>llvm.part.set</tt>' intrinsic has two modes
5660of operation: forwards and reverse. If <tt>%lo</tt> is greater than
5661<tt>%hi</tt> then the intrinsic operates in reverse mode. Otherwise it
5662operates in forward mode.</p>
5663<p>For both modes, the <tt>%repl</tt> value is prepared for use by either
5664truncating it down to the size of the replacement area or zero extending it
5665up to that size.</p>
5666<p>In forward mode, the bits between <tt>%lo</tt> and <tt>%hi</tt> (inclusive)
5667are replaced with corresponding bits from <tt>%repl</tt>. That is the 0th bit
5668in <tt>%repl</tt> replaces the <tt>%lo</tt>th bit in <tt>%val</tt> and etc. up
5669to the <tt>%hi</tt>th bit.
5670<p>In reverse mode, a similar computation is made except that the bits are
5671reversed. That is, the <tt>0</tt>th bit in <tt>%repl</tt> replaces the
5672<tt>%hi</tt> bit in <tt>%val</tt> and etc. down to the <tt>%lo</tt>th bit.
5673<h5>Examples:</h5>
5674<pre>
5675 llvm.part.set(0xFFFF, 0, 4, 7) -&gt; 0xFF0F
5676 llvm.part.set(0xFFFF, 0, 7, 4) -&gt; 0xFF0F
5677 llvm.part.set(0xFFFF, 1, 7, 4) -&gt; 0xFF8F
5678 llvm.part.set(0xFFFF, F, 8, 3) -&gt; 0xFFE7
5679 llvm.part.set(0xFFFF, 0, 3, 8) -&gt; 0xFE07
5680</pre>
5681</div>
5682
5683<!-- ======================================================================= -->
5684<div class="doc_subsection">
5685 <a name="int_debugger">Debugger Intrinsics</a>
5686</div>
5687
5688<div class="doc_text">
5689<p>
5690The LLVM debugger intrinsics (which all start with <tt>llvm.dbg.</tt> prefix),
5691are described in the <a
5692href="SourceLevelDebugging.html#format_common_intrinsics">LLVM Source Level
5693Debugging</a> document.
5694</p>
5695</div>
5696
5697
5698<!-- ======================================================================= -->
5699<div class="doc_subsection">
5700 <a name="int_eh">Exception Handling Intrinsics</a>
5701</div>
5702
5703<div class="doc_text">
5704<p> The LLVM exception handling intrinsics (which all start with
5705<tt>llvm.eh.</tt> prefix), are described in the <a
5706href="ExceptionHandling.html#format_common_intrinsics">LLVM Exception
5707Handling</a> document. </p>
5708</div>
5709
5710<!-- ======================================================================= -->
5711<div class="doc_subsection">
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5712 <a name="int_trampoline">Trampoline Intrinsic</a>
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5713</div>
5714
5715<div class="doc_text">
5716<p>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5717 This intrinsic makes it possible to excise one parameter, marked with
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5718 the <tt>nest</tt> attribute, from a function. The result is a callable
5719 function pointer lacking the nest parameter - the caller does not need
5720 to provide a value for it. Instead, the value to use is stored in
5721 advance in a "trampoline", a block of memory usually allocated
5722 on the stack, which also contains code to splice the nest value into the
5723 argument list. This is used to implement the GCC nested function address
5724 extension.
5725</p>
5726<p>
5727 For example, if the function is
5728 <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]5729 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]5730<pre>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5731 %tramp = alloca [10 x i8], align 4 ; size and alignment only correct for X86
5732 %tramp1 = getelementptr [10 x i8]* %tramp, i32 0, i32 0
5733 %p = call i8* @llvm.init.trampoline( i8* %tramp1, i8* bitcast (i32 (i8* nest , i32, i32)* @f to i8*), i8* %nval )
5734 %fp = bitcast i8* %p to i32 (i32, i32)*
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5735</pre>
Bill Wendlinge262b4c2007-09-22 09:23:55 +0000[diff] [blame]5736 <p>The call <tt>%val = call i32 %fp( i32 %x, i32 %y )</tt> is then equivalent
5737 to <tt>%val = call i32 %f( i8* %nval, i32 %x, i32 %y )</tt>.</p>
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5738</div>
5739
5740<!-- _______________________________________________________________________ -->
5741<div class="doc_subsubsection">
5742 <a name="int_it">'<tt>llvm.init.trampoline</tt>' Intrinsic</a>
5743</div>
5744<div class="doc_text">
5745<h5>Syntax:</h5>
5746<pre>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5747declare 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]5748</pre>
5749<h5>Overview:</h5>
5750<p>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5751 This fills the memory pointed to by <tt>tramp</tt> with code
5752 and returns a function pointer suitable for executing it.
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5753</p>
5754<h5>Arguments:</h5>
5755<p>
5756 The <tt>llvm.init.trampoline</tt> intrinsic takes three arguments, all
5757 pointers. The <tt>tramp</tt> argument must point to a sufficiently large
5758 and sufficiently aligned block of memory; this memory is written to by the
Duncan Sands35012212007-08-22 23:39:54 +0000[diff] [blame]5759 intrinsic. Note that the size and the alignment are target-specific - LLVM
5760 currently provides no portable way of determining them, so a front-end that
5761 generates this intrinsic needs to have some target-specific knowledge.
5762 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]5763</p>
5764<h5>Semantics:</h5>
5765<p>
5766 The block of memory pointed to by <tt>tramp</tt> is filled with target
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5767 dependent code, turning it into a function. A pointer to this function is
5768 returned, but needs to be bitcast to an
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5769 <a href="#int_trampoline">appropriate function pointer type</a>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5770 before being called. The new function's signature is the same as that of
5771 <tt>func</tt> with any arguments marked with the <tt>nest</tt> attribute
5772 removed. At most one such <tt>nest</tt> argument is allowed, and it must be
5773 of pointer type. Calling the new function is equivalent to calling
5774 <tt>func</tt> with the same argument list, but with <tt>nval</tt> used for the
5775 missing <tt>nest</tt> argument. If, after calling
5776 <tt>llvm.init.trampoline</tt>, the memory pointed to by <tt>tramp</tt> is
5777 modified, then the effect of any later call to the returned function pointer is
5778 undefined.
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5779</p>
5780</div>
5781
5782<!-- ======================================================================= -->
5783<div class="doc_subsection">
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]5784 <a name="int_atomics">Atomic Operations and Synchronization Intrinsics</a>
5785</div>
5786
5787<div class="doc_text">
5788<p>
5789 These intrinsic functions expand the "universal IR" of LLVM to represent
5790 hardware constructs for atomic operations and memory synchronization. This
5791 provides an interface to the hardware, not an interface to the programmer. It
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]5792 is aimed at a low enough level to allow any programming models or APIs
5793 (Application Programming Interfaces) which
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]5794 need atomic behaviors to map cleanly onto it. It is also modeled primarily on
5795 hardware behavior. Just as hardware provides a "universal IR" for source
5796 languages, it also provides a starting point for developing a "universal"
5797 atomic operation and synchronization IR.
5798</p>
5799<p>
5800 These do <em>not</em> form an API such as high-level threading libraries,
5801 software transaction memory systems, atomic primitives, and intrinsic
5802 functions as found in BSD, GNU libc, atomic_ops, APR, and other system and
5803 application libraries. The hardware interface provided by LLVM should allow
5804 a clean implementation of all of these APIs and parallel programming models.
5805 No one model or paradigm should be selected above others unless the hardware
5806 itself ubiquitously does so.
5807
5808</p>
5809</div>
5810
5811<!-- _______________________________________________________________________ -->
5812<div class="doc_subsubsection">
5813 <a name="int_memory_barrier">'<tt>llvm.memory.barrier</tt>' Intrinsic</a>
5814</div>
5815<div class="doc_text">
5816<h5>Syntax:</h5>
5817<pre>
5818declare void @llvm.memory.barrier( i1 &lt;ll&gt;, i1 &lt;ls&gt;, i1 &lt;sl&gt;, i1 &lt;ss&gt;,
5819i1 &lt;device&gt; )
5820
5821</pre>
5822<h5>Overview:</h5>
5823<p>
5824 The <tt>llvm.memory.barrier</tt> intrinsic guarantees ordering between
5825 specific pairs of memory access types.
5826</p>
5827<h5>Arguments:</h5>
5828<p>
5829 The <tt>llvm.memory.barrier</tt> intrinsic requires five boolean arguments.
5830 The first four arguments enables a specific barrier as listed below. The fith
5831 argument specifies that the barrier applies to io or device or uncached memory.
5832
5833</p>
5834 <ul>
5835 <li><tt>ll</tt>: load-load barrier</li>
5836 <li><tt>ls</tt>: load-store barrier</li>
5837 <li><tt>sl</tt>: store-load barrier</li>
5838 <li><tt>ss</tt>: store-store barrier</li>
5839 <li><tt>device</tt>: barrier applies to device and uncached memory also.
5840 </ul>
5841<h5>Semantics:</h5>
5842<p>
5843 This intrinsic causes the system to enforce some ordering constraints upon
5844 the loads and stores of the program. This barrier does not indicate
5845 <em>when</em> any events will occur, it only enforces an <em>order</em> in
5846 which they occur. For any of the specified pairs of load and store operations
5847 (f.ex. load-load, or store-load), all of the first operations preceding the
5848 barrier will complete before any of the second operations succeeding the
5849 barrier begin. Specifically the semantics for each pairing is as follows:
5850</p>
5851 <ul>
5852 <li><tt>ll</tt>: All loads before the barrier must complete before any load
5853 after the barrier begins.</li>
5854
5855 <li><tt>ls</tt>: All loads before the barrier must complete before any
5856 store after the barrier begins.</li>
5857 <li><tt>ss</tt>: All stores before the barrier must complete before any
5858 store after the barrier begins.</li>
5859 <li><tt>sl</tt>: All stores before the barrier must complete before any
5860 load after the barrier begins.</li>
5861 </ul>
5862<p>
5863 These semantics are applied with a logical "and" behavior when more than one
5864 is enabled in a single memory barrier intrinsic.
5865</p>
5866<p>
5867 Backends may implement stronger barriers than those requested when they do not
5868 support as fine grained a barrier as requested. Some architectures do not
5869 need all types of barriers and on such architectures, these become noops.
5870</p>
5871<h5>Example:</h5>
5872<pre>
5873%ptr = malloc i32
5874 store i32 4, %ptr
5875
5876%result1 = load i32* %ptr <i>; yields {i32}:result1 = 4</i>
5877 call void @llvm.memory.barrier( i1 false, i1 true, i1 false, i1 false )
5878 <i>; guarantee the above finishes</i>
5879 store i32 8, %ptr <i>; before this begins</i>
5880</pre>
5881</div>
5882
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]5883<!-- _______________________________________________________________________ -->
5884<div class="doc_subsubsection">
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]5885 <a name="int_atomic_cmp_swap">'<tt>llvm.atomic.cmp.swap.*</tt>' Intrinsic</a>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]5886</div>
5887<div class="doc_text">
5888<h5>Syntax:</h5>
5889<p>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]5890 This is an overloaded intrinsic. You can use <tt>llvm.atomic.cmp.swap</tt> on
5891 any integer bit width and for different address spaces. Not all targets
5892 support all bit widths however.</p>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]5893
5894<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]5895declare i8 @llvm.atomic.cmp.swap.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;cmp&gt;, i8 &lt;val&gt; )
5896declare i16 @llvm.atomic.cmp.swap.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;cmp&gt;, i16 &lt;val&gt; )
5897declare i32 @llvm.atomic.cmp.swap.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;cmp&gt;, i32 &lt;val&gt; )
5898declare 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]5899
5900</pre>
5901<h5>Overview:</h5>
5902<p>
5903 This loads a value in memory and compares it to a given value. If they are
5904 equal, it stores a new value into the memory.
5905</p>
5906<h5>Arguments:</h5>
5907<p>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]5908 The <tt>llvm.atomic.cmp.swap</tt> intrinsic takes three arguments. The result as
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]5909 well as both <tt>cmp</tt> and <tt>val</tt> must be integer values with the
5910 same bit width. The <tt>ptr</tt> argument must be a pointer to a value of
5911 this integer type. While any bit width integer may be used, targets may only
5912 lower representations they support in hardware.
5913
5914</p>
5915<h5>Semantics:</h5>
5916<p>
5917 This entire intrinsic must be executed atomically. It first loads the value
5918 in memory pointed to by <tt>ptr</tt> and compares it with the value
5919 <tt>cmp</tt>. If they are equal, <tt>val</tt> is stored into the memory. The
5920 loaded value is yielded in all cases. This provides the equivalent of an
5921 atomic compare-and-swap operation within the SSA framework.
5922</p>
5923<h5>Examples:</h5>
5924
5925<pre>
5926%ptr = malloc i32
5927 store i32 4, %ptr
5928
5929%val1 = add i32 4, 4
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]5930%result1 = call i32 @llvm.atomic.cmp.swap.i32.p0i32( i32* %ptr, i32 4, %val1 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]5931 <i>; yields {i32}:result1 = 4</i>
5932%stored1 = icmp eq i32 %result1, 4 <i>; yields {i1}:stored1 = true</i>
5933%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 8</i>
5934
5935%val2 = add i32 1, 1
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]5936%result2 = call i32 @llvm.atomic.cmp.swap.i32.p0i32( i32* %ptr, i32 5, %val2 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]5937 <i>; yields {i32}:result2 = 8</i>
5938%stored2 = icmp eq i32 %result2, 5 <i>; yields {i1}:stored2 = false</i>
5939
5940%memval2 = load i32* %ptr <i>; yields {i32}:memval2 = 8</i>
5941</pre>
5942</div>
5943
5944<!-- _______________________________________________________________________ -->
5945<div class="doc_subsubsection">
5946 <a name="int_atomic_swap">'<tt>llvm.atomic.swap.*</tt>' Intrinsic</a>
5947</div>
5948<div class="doc_text">
5949<h5>Syntax:</h5>
5950
5951<p>
5952 This is an overloaded intrinsic. You can use <tt>llvm.atomic.swap</tt> on any
5953 integer bit width. Not all targets support all bit widths however.</p>
5954<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]5955declare i8 @llvm.atomic.swap.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;val&gt; )
5956declare i16 @llvm.atomic.swap.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;val&gt; )
5957declare i32 @llvm.atomic.swap.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;val&gt; )
5958declare i64 @llvm.atomic.swap.i64.p0i64( i64* &lt;ptr&gt;, i64 &lt;val&gt; )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]5959
5960</pre>
5961<h5>Overview:</h5>
5962<p>
5963 This intrinsic loads the value stored in memory at <tt>ptr</tt> and yields
5964 the value from memory. It then stores the value in <tt>val</tt> in the memory
5965 at <tt>ptr</tt>.
5966</p>
5967<h5>Arguments:</h5>
5968
5969<p>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]5970 The <tt>llvm.atomic.swap</tt> intrinsic takes two arguments. Both the
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]5971 <tt>val</tt> argument and the result must be integers of the same bit width.
5972 The first argument, <tt>ptr</tt>, must be a pointer to a value of this
5973 integer type. The targets may only lower integer representations they
5974 support.
5975</p>
5976<h5>Semantics:</h5>
5977<p>
5978 This intrinsic loads the value pointed to by <tt>ptr</tt>, yields it, and
5979 stores <tt>val</tt> back into <tt>ptr</tt> atomically. This provides the
5980 equivalent of an atomic swap operation within the SSA framework.
5981
5982</p>
5983<h5>Examples:</h5>
5984<pre>
5985%ptr = malloc i32
5986 store i32 4, %ptr
5987
5988%val1 = add i32 4, 4
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]5989%result1 = call i32 @llvm.atomic.swap.i32.p0i32( i32* %ptr, i32 %val1 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]5990 <i>; yields {i32}:result1 = 4</i>
5991%stored1 = icmp eq i32 %result1, 4 <i>; yields {i1}:stored1 = true</i>
5992%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 8</i>
5993
5994%val2 = add i32 1, 1
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]5995%result2 = call i32 @llvm.atomic.swap.i32.p0i32( i32* %ptr, i32 %val2 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]5996 <i>; yields {i32}:result2 = 8</i>
5997
5998%stored2 = icmp eq i32 %result2, 8 <i>; yields {i1}:stored2 = true</i>
5999%memval2 = load i32* %ptr <i>; yields {i32}:memval2 = 2</i>
6000</pre>
6001</div>
6002
6003<!-- _______________________________________________________________________ -->
6004<div class="doc_subsubsection">
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6005 <a name="int_atomic_load_add">'<tt>llvm.atomic.load.add.*</tt>' Intrinsic</a>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6006
6007</div>
6008<div class="doc_text">
6009<h5>Syntax:</h5>
6010<p>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6011 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]6012 integer bit width. Not all targets support all bit widths however.</p>
6013<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6014declare i8 @llvm.atomic.load.add.i8..p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6015declare i16 @llvm.atomic.load.add.i16..p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6016declare i32 @llvm.atomic.load.add.i32..p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6017declare 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]6018
6019</pre>
6020<h5>Overview:</h5>
6021<p>
6022 This intrinsic adds <tt>delta</tt> to the value stored in memory at
6023 <tt>ptr</tt>. It yields the original value at <tt>ptr</tt>.
6024</p>
6025<h5>Arguments:</h5>
6026<p>
6027
6028 The intrinsic takes two arguments, the first a pointer to an integer value
6029 and the second an integer value. The result is also an integer value. These
6030 integer types can have any bit width, but they must all have the same bit
6031 width. The targets may only lower integer representations they support.
6032</p>
6033<h5>Semantics:</h5>
6034<p>
6035 This intrinsic does a series of operations atomically. It first loads the
6036 value stored at <tt>ptr</tt>. It then adds <tt>delta</tt>, stores the result
6037 to <tt>ptr</tt>. It yields the original value stored at <tt>ptr</tt>.
6038</p>
6039
6040<h5>Examples:</h5>
6041<pre>
6042%ptr = malloc i32
6043 store i32 4, %ptr
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6044%result1 = call i32 @llvm.atomic.load.add.i32.p0i32( i32* %ptr, i32 4 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6045 <i>; yields {i32}:result1 = 4</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6046%result2 = call i32 @llvm.atomic.load.add.i32.p0i32( i32* %ptr, i32 2 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6047 <i>; yields {i32}:result2 = 8</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6048%result3 = call i32 @llvm.atomic.load.add.i32.p0i32( i32* %ptr, i32 5 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6049 <i>; yields {i32}:result3 = 10</i>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6050%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 15</i>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6051</pre>
6052</div>
6053
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6054<!-- _______________________________________________________________________ -->
6055<div class="doc_subsubsection">
6056 <a name="int_atomic_load_sub">'<tt>llvm.atomic.load.sub.*</tt>' Intrinsic</a>
6057
6058</div>
6059<div class="doc_text">
6060<h5>Syntax:</h5>
6061<p>
6062 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]6063 any integer bit width and for different address spaces. Not all targets
6064 support all bit widths however.</p>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6065<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6066declare i8 @llvm.atomic.load.sub.i8.p0i32( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6067declare i16 @llvm.atomic.load.sub.i16.p0i32( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6068declare i32 @llvm.atomic.load.sub.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6069declare 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]6070
6071</pre>
6072<h5>Overview:</h5>
6073<p>
6074 This intrinsic subtracts <tt>delta</tt> to the value stored in memory at
6075 <tt>ptr</tt>. It yields the original value at <tt>ptr</tt>.
6076</p>
6077<h5>Arguments:</h5>
6078<p>
6079
6080 The intrinsic takes two arguments, the first a pointer to an integer value
6081 and the second an integer value. The result is also an integer value. These
6082 integer types can have any bit width, but they must all have the same bit
6083 width. The targets may only lower integer representations they support.
6084</p>
6085<h5>Semantics:</h5>
6086<p>
6087 This intrinsic does a series of operations atomically. It first loads the
6088 value stored at <tt>ptr</tt>. It then subtracts <tt>delta</tt>, stores the
6089 result to <tt>ptr</tt>. It yields the original value stored at <tt>ptr</tt>.
6090</p>
6091
6092<h5>Examples:</h5>
6093<pre>
6094%ptr = malloc i32
6095 store i32 8, %ptr
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6096%result1 = call i32 @llvm.atomic.load.sub.i32.p0i32( i32* %ptr, i32 4 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6097 <i>; yields {i32}:result1 = 8</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6098%result2 = call i32 @llvm.atomic.load.sub.i32.p0i32( i32* %ptr, i32 2 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6099 <i>; yields {i32}:result2 = 4</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6100%result3 = call i32 @llvm.atomic.load.sub.i32.p0i32( i32* %ptr, i32 5 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6101 <i>; yields {i32}:result3 = 2</i>
6102%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = -3</i>
6103</pre>
6104</div>
6105
6106<!-- _______________________________________________________________________ -->
6107<div class="doc_subsubsection">
6108 <a name="int_atomic_load_and">'<tt>llvm.atomic.load.and.*</tt>' Intrinsic</a><br>
6109 <a name="int_atomic_load_nand">'<tt>llvm.atomic.load.nand.*</tt>' Intrinsic</a><br>
6110 <a name="int_atomic_load_or">'<tt>llvm.atomic.load.or.*</tt>' Intrinsic</a><br>
6111 <a name="int_atomic_load_xor">'<tt>llvm.atomic.load.xor.*</tt>' Intrinsic</a><br>
6112
6113</div>
6114<div class="doc_text">
6115<h5>Syntax:</h5>
6116<p>
6117 These are overloaded intrinsics. You can use <tt>llvm.atomic.load_and</tt>,
6118 <tt>llvm.atomic.load_nand</tt>, <tt>llvm.atomic.load_or</tt>, and
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6119 <tt>llvm.atomic.load_xor</tt> on any integer bit width and for different
6120 address spaces. Not all targets support all bit widths however.</p>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6121<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6122declare i8 @llvm.atomic.load.and.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6123declare i16 @llvm.atomic.load.and.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6124declare i32 @llvm.atomic.load.and.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6125declare 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]6126
6127</pre>
6128
6129<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6130declare i8 @llvm.atomic.load.or.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6131declare i16 @llvm.atomic.load.or.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6132declare i32 @llvm.atomic.load.or.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6133declare 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]6134
6135</pre>
6136
6137<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6138declare i8 @llvm.atomic.load.nand.i8.p0i32( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6139declare i16 @llvm.atomic.load.nand.i16.p0i32( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6140declare i32 @llvm.atomic.load.nand.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6141declare 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]6142
6143</pre>
6144
6145<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6146declare i8 @llvm.atomic.load.xor.i8.p0i32( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6147declare i16 @llvm.atomic.load.xor.i16.p0i32( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6148declare i32 @llvm.atomic.load.xor.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6149declare 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]6150
6151</pre>
6152<h5>Overview:</h5>
6153<p>
6154 These intrinsics bitwise the operation (and, nand, or, xor) <tt>delta</tt> to
6155 the value stored in memory at <tt>ptr</tt>. It yields the original value
6156 at <tt>ptr</tt>.
6157</p>
6158<h5>Arguments:</h5>
6159<p>
6160
6161 These intrinsics take two arguments, the first a pointer to an integer value
6162 and the second an integer value. The result is also an integer value. These
6163 integer types can have any bit width, but they must all have the same bit
6164 width. The targets may only lower integer representations they support.
6165</p>
6166<h5>Semantics:</h5>
6167<p>
6168 These intrinsics does a series of operations atomically. They first load the
6169 value stored at <tt>ptr</tt>. They then do the bitwise operation
6170 <tt>delta</tt>, store the result to <tt>ptr</tt>. They yield the original
6171 value stored at <tt>ptr</tt>.
6172</p>
6173
6174<h5>Examples:</h5>
6175<pre>
6176%ptr = malloc i32
6177 store i32 0x0F0F, %ptr
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6178%result0 = call i32 @llvm.atomic.load.nand.i32.p0i32( i32* %ptr, i32 0xFF )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6179 <i>; yields {i32}:result0 = 0x0F0F</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6180%result1 = call i32 @llvm.atomic.load.and.i32.p0i32( i32* %ptr, i32 0xFF )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6181 <i>; yields {i32}:result1 = 0xFFFFFFF0</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6182%result2 = call i32 @llvm.atomic.load.or.i32.p0i32( i32* %ptr, i32 0F )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6183 <i>; yields {i32}:result2 = 0xF0</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6184%result3 = call i32 @llvm.atomic.load.xor.i32.p0i32( i32* %ptr, i32 0F )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6185 <i>; yields {i32}:result3 = FF</i>
6186%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = F0</i>
6187</pre>
6188</div>
6189
6190
6191<!-- _______________________________________________________________________ -->
6192<div class="doc_subsubsection">
6193 <a name="int_atomic_load_max">'<tt>llvm.atomic.load.max.*</tt>' Intrinsic</a><br>
6194 <a name="int_atomic_load_min">'<tt>llvm.atomic.load.min.*</tt>' Intrinsic</a><br>
6195 <a name="int_atomic_load_umax">'<tt>llvm.atomic.load.umax.*</tt>' Intrinsic</a><br>
6196 <a name="int_atomic_load_umin">'<tt>llvm.atomic.load.umin.*</tt>' Intrinsic</a><br>
6197
6198</div>
6199<div class="doc_text">
6200<h5>Syntax:</h5>
6201<p>
6202 These are overloaded intrinsics. You can use <tt>llvm.atomic.load_max</tt>,
6203 <tt>llvm.atomic.load_min</tt>, <tt>llvm.atomic.load_umax</tt>, and
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6204 <tt>llvm.atomic.load_umin</tt> on any integer bit width and for different
6205 address spaces. Not all targets
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6206 support all bit widths however.</p>
6207<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6208declare i8 @llvm.atomic.load.max.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6209declare i16 @llvm.atomic.load.max.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6210declare i32 @llvm.atomic.load.max.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6211declare 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]6212
6213</pre>
6214
6215<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6216declare i8 @llvm.atomic.load.min.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6217declare i16 @llvm.atomic.load.min.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6218declare i32 @llvm.atomic.load.min.i32..p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6219declare 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]6220
6221</pre>
6222
6223<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6224declare i8 @llvm.atomic.load.umax.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6225declare i16 @llvm.atomic.load.umax.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6226declare i32 @llvm.atomic.load.umax.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6227declare 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]6228
6229</pre>
6230
6231<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6232declare i8 @llvm.atomic.load.umin.i8..p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6233declare i16 @llvm.atomic.load.umin.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6234declare i32 @llvm.atomic.load.umin.i32..p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6235declare 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]6236
6237</pre>
6238<h5>Overview:</h5>
6239<p>
6240 These intrinsics takes the signed or unsigned minimum or maximum of
6241 <tt>delta</tt> and the value stored in memory at <tt>ptr</tt>. It yields the
6242 original value at <tt>ptr</tt>.
6243</p>
6244<h5>Arguments:</h5>
6245<p>
6246
6247 These intrinsics take two arguments, the first a pointer to an integer value
6248 and the second an integer value. The result is also an integer value. These
6249 integer types can have any bit width, but they must all have the same bit
6250 width. The targets may only lower integer representations they support.
6251</p>
6252<h5>Semantics:</h5>
6253<p>
6254 These intrinsics does a series of operations atomically. They first load the
6255 value stored at <tt>ptr</tt>. They then do the signed or unsigned min or max
6256 <tt>delta</tt> and the value, store the result to <tt>ptr</tt>. They yield
6257 the original value stored at <tt>ptr</tt>.
6258</p>
6259
6260<h5>Examples:</h5>
6261<pre>
6262%ptr = malloc i32
6263 store i32 7, %ptr
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6264%result0 = call i32 @llvm.atomic.load.min.i32.p0i32( i32* %ptr, i32 -2 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6265 <i>; yields {i32}:result0 = 7</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6266%result1 = call i32 @llvm.atomic.load.max.i32.p0i32( i32* %ptr, i32 8 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6267 <i>; yields {i32}:result1 = -2</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6268%result2 = call i32 @llvm.atomic.load.umin.i32.p0i32( i32* %ptr, i32 10 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6269 <i>; yields {i32}:result2 = 8</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6270%result3 = call i32 @llvm.atomic.load.umax.i32.p0i32( i32* %ptr, i32 30 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6271 <i>; yields {i32}:result3 = 8</i>
6272%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 30</i>
6273</pre>
6274</div>
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]6275
6276<!-- ======================================================================= -->
6277<div class="doc_subsection">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6278 <a name="int_general">General Intrinsics</a>
6279</div>
6280
6281<div class="doc_text">
6282<p> This class of intrinsics is designed to be generic and has
6283no specific purpose. </p>
6284</div>
6285
6286<!-- _______________________________________________________________________ -->
6287<div class="doc_subsubsection">
6288 <a name="int_var_annotation">'<tt>llvm.var.annotation</tt>' Intrinsic</a>
6289</div>
6290
6291<div class="doc_text">
6292
6293<h5>Syntax:</h5>
6294<pre>
6295 declare void @llvm.var.annotation(i8* &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
6296</pre>
6297
6298<h5>Overview:</h5>
6299
6300<p>
6301The '<tt>llvm.var.annotation</tt>' intrinsic
6302</p>
6303
6304<h5>Arguments:</h5>
6305
6306<p>
6307The first argument is a pointer to a value, the second is a pointer to a
6308global string, the third is a pointer to a global string which is the source
6309file name, and the last argument is the line number.
6310</p>
6311
6312<h5>Semantics:</h5>
6313
6314<p>
Anton Korobeynikove6e764f2008-01-15 22:31:34 +0000[diff] [blame]6315This intrinsic allows annotation of local variables with arbitrary strings.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6316This can be useful for special purpose optimizations that want to look for these
Anton Korobeynikove6e764f2008-01-15 22:31:34 +0000[diff] [blame]6317annotations. These have no other defined use, they are ignored by code
6318generation and optimization.
6319</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6320</div>
6321
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]6322<!-- _______________________________________________________________________ -->
6323<div class="doc_subsubsection">
Tanya Lattnerc9869b12007-09-21 23:57:59 +0000[diff] [blame]6324 <a name="int_annotation">'<tt>llvm.annotation.*</tt>' Intrinsic</a>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]6325</div>
6326
6327<div class="doc_text">
6328
6329<h5>Syntax:</h5>
Tanya Lattnere545be72007-09-21 23:56:27 +0000[diff] [blame]6330<p>This is an overloaded intrinsic. You can use '<tt>llvm.annotation</tt>' on
6331any integer bit width.
6332</p>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]6333<pre>
Tanya Lattner09161fe2007-09-22 00:03:01 +0000[diff] [blame]6334 declare i8 @llvm.annotation.i8(i8 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
6335 declare i16 @llvm.annotation.i16(i16 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
6336 declare i32 @llvm.annotation.i32(i32 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
6337 declare i64 @llvm.annotation.i64(i64 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
6338 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]6339</pre>
6340
6341<h5>Overview:</h5>
Tanya Lattnere545be72007-09-21 23:56:27 +0000[diff] [blame]6342
6343<p>
6344The '<tt>llvm.annotation</tt>' intrinsic.
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]6345</p>
6346
6347<h5>Arguments:</h5>
6348
6349<p>
6350The first argument is an integer value (result of some expression),
6351the second is a pointer to a global string, the third is a pointer to a global
6352string which is the source file name, and the last argument is the line number.
Tanya Lattnere545be72007-09-21 23:56:27 +0000[diff] [blame]6353It returns the value of the first argument.
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]6354</p>
6355
6356<h5>Semantics:</h5>
6357
6358<p>
6359This intrinsic allows annotations to be put on arbitrary expressions
6360with arbitrary strings. This can be useful for special purpose optimizations
6361that want to look for these annotations. These have no other defined use, they
6362are ignored by code generation and optimization.
6363</div>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6364
Anton Korobeynikove6e764f2008-01-15 22:31:34 +0000[diff] [blame]6365<!-- _______________________________________________________________________ -->
6366<div class="doc_subsubsection">
6367 <a name="int_trap">'<tt>llvm.trap</tt>' Intrinsic</a>
6368</div>
6369
6370<div class="doc_text">
6371
6372<h5>Syntax:</h5>
6373<pre>
6374 declare void @llvm.trap()
6375</pre>
6376
6377<h5>Overview:</h5>
6378
6379<p>
6380The '<tt>llvm.trap</tt>' intrinsic
6381</p>
6382
6383<h5>Arguments:</h5>
6384
6385<p>
6386None
6387</p>
6388
6389<h5>Semantics:</h5>
6390
6391<p>
6392This intrinsics is lowered to the target dependent trap instruction. If the
6393target does not have a trap instruction, this intrinsic will be lowered to the
6394call of the abort() function.
6395</p>
6396</div>
6397
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6398<!-- *********************************************************************** -->
6399<hr>
6400<address>
6401 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
6402 src="http://jigsaw.w3.org/css-validator/images/vcss" alt="Valid CSS!"></a>
6403 <a href="http://validator.w3.org/check/referer"><img
Chris Lattner08497ce2008-01-04 04:33:49 +0000[diff] [blame]6404 src="http://www.w3.org/Icons/valid-html401" alt="Valid HTML 4.01!"></a>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6405
6406 <a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
6407 <a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br>
6408 Last modified: $Date$
6409</address>
Chris Lattner08497ce2008-01-04 04:33:49 +0000[diff] [blame]6410
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6411</body>
6412</html>