Gitiles
Code Review Sign In
gerrit-public.fairphone.software / fp2-dev / platform / external / llvm / ee202543cf38b20fc1777f61cbebdb3f2addfaba / . / docs / LangRef.html
blob: 94ef37b97a7e79ef39daef4c30182d96225c32b3 [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>
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]156 <li><a href="#i_getresult">'<tt>getresult</tt>' Instruction</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]157 </ol>
158 </li>
159 </ol>
160 </li>
161 <li><a href="#intrinsics">Intrinsic Functions</a>
162 <ol>
163 <li><a href="#int_varargs">Variable Argument Handling Intrinsics</a>
164 <ol>
165 <li><a href="#int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a></li>
166 <li><a href="#int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a></li>
167 <li><a href="#int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a></li>
168 </ol>
169 </li>
170 <li><a href="#int_gc">Accurate Garbage Collection Intrinsics</a>
171 <ol>
172 <li><a href="#int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a></li>
173 <li><a href="#int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a></li>
174 <li><a href="#int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a></li>
175 </ol>
176 </li>
177 <li><a href="#int_codegen">Code Generator Intrinsics</a>
178 <ol>
179 <li><a href="#int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a></li>
180 <li><a href="#int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a></li>
181 <li><a href="#int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a></li>
182 <li><a href="#int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a></li>
183 <li><a href="#int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a></li>
184 <li><a href="#int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a></li>
185 <li><a href="#int_readcyclecounter"><tt>llvm.readcyclecounter</tt>' Intrinsic</a></li>
186 </ol>
187 </li>
188 <li><a href="#int_libc">Standard C Library Intrinsics</a>
189 <ol>
190 <li><a href="#int_memcpy">'<tt>llvm.memcpy.*</tt>' Intrinsic</a></li>
191 <li><a href="#int_memmove">'<tt>llvm.memmove.*</tt>' Intrinsic</a></li>
192 <li><a href="#int_memset">'<tt>llvm.memset.*</tt>' Intrinsic</a></li>
193 <li><a href="#int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a></li>
194 <li><a href="#int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a></li>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]195 <li><a href="#int_sin">'<tt>llvm.sin.*</tt>' Intrinsic</a></li>
196 <li><a href="#int_cos">'<tt>llvm.cos.*</tt>' Intrinsic</a></li>
197 <li><a href="#int_pow">'<tt>llvm.pow.*</tt>' Intrinsic</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]198 </ol>
199 </li>
200 <li><a href="#int_manip">Bit Manipulation Intrinsics</a>
201 <ol>
202 <li><a href="#int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a></li>
203 <li><a href="#int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic </a></li>
204 <li><a href="#int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic </a></li>
205 <li><a href="#int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic </a></li>
206 <li><a href="#int_part_select">'<tt>llvm.part.select.*</tt>' Intrinsic </a></li>
207 <li><a href="#int_part_set">'<tt>llvm.part.set.*</tt>' Intrinsic </a></li>
208 </ol>
209 </li>
210 <li><a href="#int_debugger">Debugger intrinsics</a></li>
211 <li><a href="#int_eh">Exception Handling intrinsics</a></li>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]212 <li><a href="#int_trampoline">Trampoline Intrinsic</a>
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]213 <ol>
214 <li><a href="#int_it">'<tt>llvm.init.trampoline</tt>' Intrinsic</a></li>
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]215 </ol>
216 </li>
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]217 <li><a href="#int_atomics">Atomic intrinsics</a>
218 <ol>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]219 <li><a href="#int_memory_barrier"><tt>llvm.memory_barrier</tt></a></li>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]220 <li><a href="#int_atomic_cmp_swap"><tt>llvm.atomic.cmp.swap</tt></a></li>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]221 <li><a href="#int_atomic_swap"><tt>llvm.atomic.swap</tt></a></li>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]222 <li><a href="#int_atomic_load_add"><tt>llvm.atomic.load.add</tt></a></li>
223 <li><a href="#int_atomic_load_sub"><tt>llvm.atomic.load.sub</tt></a></li>
224 <li><a href="#int_atomic_load_and"><tt>llvm.atomic.load.and</tt></a></li>
225 <li><a href="#int_atomic_load_nand"><tt>llvm.atomic.load.nand</tt></a></li>
226 <li><a href="#int_atomic_load_or"><tt>llvm.atomic.load.or</tt></a></li>
227 <li><a href="#int_atomic_load_xor"><tt>llvm.atomic.load.xor</tt></a></li>
228 <li><a href="#int_atomic_load_max"><tt>llvm.atomic.load.max</tt></a></li>
229 <li><a href="#int_atomic_load_min"><tt>llvm.atomic.load.min</tt></a></li>
230 <li><a href="#int_atomic_load_umax"><tt>llvm.atomic.load.umax</tt></a></li>
231 <li><a href="#int_atomic_load_umin"><tt>llvm.atomic.load.umin</tt></a></li>
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]232 </ol>
233 </li>
Reid Spencerb043f672007-07-20 19:59:11 +0000[diff] [blame]234 <li><a href="#int_general">General intrinsics</a>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]235 <ol>
Reid Spencerb043f672007-07-20 19:59:11 +0000[diff] [blame]236 <li><a href="#int_var_annotation">
Tanya Lattner51369f32007-09-22 00:01:26 +0000[diff] [blame]237 <tt>llvm.var.annotation</tt>' Intrinsic</a></li>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]238 <li><a href="#int_annotation">
Tanya Lattner51369f32007-09-22 00:01:26 +0000[diff] [blame]239 <tt>llvm.annotation.*</tt>' Intrinsic</a></li>
Anton Korobeynikove6e764f2008-01-15 22:31:34 +0000[diff] [blame]240 <li><a href="#int_trap">
241 <tt>llvm.trap</tt>' Intrinsic</a></li>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]242 </ol>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]243 </li>
244 </ol>
245 </li>
246</ol>
247
248<div class="doc_author">
249 <p>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a>
250 and <a href="mailto:vadve@cs.uiuc.edu">Vikram Adve</a></p>
251</div>
252
253<!-- *********************************************************************** -->
254<div class="doc_section"> <a name="abstract">Abstract </a></div>
255<!-- *********************************************************************** -->
256
257<div class="doc_text">
258<p>This document is a reference manual for the LLVM assembly language.
Bill Wendlinge7846a52008-08-05 22:29:16 +0000[diff] [blame]259LLVM is a Static Single Assignment (SSA) based representation that provides
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]260type safety, low-level operations, flexibility, and the capability of
261representing 'all' high-level languages cleanly. It is the common code
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]262representation used throughout all phases of the LLVM compilation
263strategy.</p>
264</div>
265
266<!-- *********************************************************************** -->
267<div class="doc_section"> <a name="introduction">Introduction</a> </div>
268<!-- *********************************************************************** -->
269
270<div class="doc_text">
271
272<p>The LLVM code representation is designed to be used in three
273different forms: as an in-memory compiler IR, as an on-disk bitcode
274representation (suitable for fast loading by a Just-In-Time compiler),
275and as a human readable assembly language representation. This allows
276LLVM to provide a powerful intermediate representation for efficient
277compiler transformations and analysis, while providing a natural means
278to debug and visualize the transformations. The three different forms
279of LLVM are all equivalent. This document describes the human readable
280representation and notation.</p>
281
282<p>The LLVM representation aims to be light-weight and low-level
283while being expressive, typed, and extensible at the same time. It
284aims to be a "universal IR" of sorts, by being at a low enough level
285that high-level ideas may be cleanly mapped to it (similar to how
286microprocessors are "universal IR's", allowing many source languages to
287be mapped to them). By providing type information, LLVM can be used as
288the target of optimizations: for example, through pointer analysis, it
289can be proven that a C automatic variable is never accessed outside of
290the current function... allowing it to be promoted to a simple SSA
291value instead of a memory location.</p>
292
293</div>
294
295<!-- _______________________________________________________________________ -->
296<div class="doc_subsubsection"> <a name="wellformed">Well-Formedness</a> </div>
297
298<div class="doc_text">
299
300<p>It is important to note that this document describes 'well formed'
301LLVM assembly language. There is a difference between what the parser
302accepts and what is considered 'well formed'. For example, the
303following instruction is syntactically okay, but not well formed:</p>
304
305<div class="doc_code">
306<pre>
307%x = <a href="#i_add">add</a> i32 1, %x
308</pre>
309</div>
310
311<p>...because the definition of <tt>%x</tt> does not dominate all of
312its uses. The LLVM infrastructure provides a verification pass that may
313be used to verify that an LLVM module is well formed. This pass is
314automatically run by the parser after parsing input assembly and by
315the optimizer before it outputs bitcode. The violations pointed out
316by the verifier pass indicate bugs in transformation passes or input to
317the parser.</p>
318</div>
319
Chris Lattnera83fdc02007-10-03 17:34:29 +0000[diff] [blame]320<!-- Describe the typesetting conventions here. -->
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]321
322<!-- *********************************************************************** -->
323<div class="doc_section"> <a name="identifiers">Identifiers</a> </div>
324<!-- *********************************************************************** -->
325
326<div class="doc_text">
327
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]328 <p>LLVM identifiers come in two basic types: global and local. Global
329 identifiers (functions, global variables) begin with the @ character. Local
330 identifiers (register names, types) begin with the % character. Additionally,
331 there are three different formats for identifiers, for different purposes:
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]332
333<ol>
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]334 <li>Named values are represented as a string of characters with their prefix.
335 For example, %foo, @DivisionByZero, %a.really.long.identifier. The actual
336 regular expression used is '<tt>[%@][a-zA-Z$._][a-zA-Z$._0-9]*</tt>'.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]337 Identifiers which require other characters in their names can be surrounded
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]338 with quotes. In this way, anything except a <tt>&quot;</tt> character can
339 be used in a named value.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]340
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]341 <li>Unnamed values are represented as an unsigned numeric value with their
342 prefix. For example, %12, @2, %44.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]343
344 <li>Constants, which are described in a <a href="#constants">section about
345 constants</a>, below.</li>
346</ol>
347
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]348<p>LLVM requires that values start with a prefix for two reasons: Compilers
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]349don't need to worry about name clashes with reserved words, and the set of
350reserved words may be expanded in the future without penalty. Additionally,
351unnamed identifiers allow a compiler to quickly come up with a temporary
352variable without having to avoid symbol table conflicts.</p>
353
354<p>Reserved words in LLVM are very similar to reserved words in other
355languages. There are keywords for different opcodes
356('<tt><a href="#i_add">add</a></tt>',
357 '<tt><a href="#i_bitcast">bitcast</a></tt>',
358 '<tt><a href="#i_ret">ret</a></tt>', etc...), for primitive type names ('<tt><a
359href="#t_void">void</a></tt>', '<tt><a href="#t_primitive">i32</a></tt>', etc...),
360and others. These reserved words cannot conflict with variable names, because
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]361none of them start with a prefix character ('%' or '@').</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]362
363<p>Here is an example of LLVM code to multiply the integer variable
364'<tt>%X</tt>' by 8:</p>
365
366<p>The easy way:</p>
367
368<div class="doc_code">
369<pre>
370%result = <a href="#i_mul">mul</a> i32 %X, 8
371</pre>
372</div>
373
374<p>After strength reduction:</p>
375
376<div class="doc_code">
377<pre>
378%result = <a href="#i_shl">shl</a> i32 %X, i8 3
379</pre>
380</div>
381
382<p>And the hard way:</p>
383
384<div class="doc_code">
385<pre>
386<a href="#i_add">add</a> i32 %X, %X <i>; yields {i32}:%0</i>
387<a href="#i_add">add</a> i32 %0, %0 <i>; yields {i32}:%1</i>
388%result = <a href="#i_add">add</a> i32 %1, %1
389</pre>
390</div>
391
392<p>This last way of multiplying <tt>%X</tt> by 8 illustrates several
393important lexical features of LLVM:</p>
394
395<ol>
396
397 <li>Comments are delimited with a '<tt>;</tt>' and go until the end of
398 line.</li>
399
400 <li>Unnamed temporaries are created when the result of a computation is not
401 assigned to a named value.</li>
402
403 <li>Unnamed temporaries are numbered sequentially</li>
404
405</ol>
406
407<p>...and it also shows a convention that we follow in this document. When
408demonstrating instructions, we will follow an instruction with a comment that
409defines the type and name of value produced. Comments are shown in italic
410text.</p>
411
412</div>
413
414<!-- *********************************************************************** -->
415<div class="doc_section"> <a name="highlevel">High Level Structure</a> </div>
416<!-- *********************************************************************** -->
417
418<!-- ======================================================================= -->
419<div class="doc_subsection"> <a name="modulestructure">Module Structure</a>
420</div>
421
422<div class="doc_text">
423
424<p>LLVM programs are composed of "Module"s, each of which is a
425translation unit of the input programs. Each module consists of
426functions, global variables, and symbol table entries. Modules may be
427combined together with the LLVM linker, which merges function (and
428global variable) definitions, resolves forward declarations, and merges
429symbol table entries. Here is an example of the "hello world" module:</p>
430
431<div class="doc_code">
432<pre><i>; Declare the string constant as a global constant...</i>
433<a href="#identifiers">@.LC0</a> = <a href="#linkage_internal">internal</a> <a
434 href="#globalvars">constant</a> <a href="#t_array">[13 x i8]</a> c"hello world\0A\00" <i>; [13 x i8]*</i>
435
436<i>; External declaration of the puts function</i>
437<a href="#functionstructure">declare</a> i32 @puts(i8 *) <i>; i32(i8 *)* </i>
438
439<i>; Definition of main function</i>
440define i32 @main() { <i>; i32()* </i>
441 <i>; Convert [13x i8 ]* to i8 *...</i>
442 %cast210 = <a
443 href="#i_getelementptr">getelementptr</a> [13 x i8 ]* @.LC0, i64 0, i64 0 <i>; i8 *</i>
444
445 <i>; Call puts function to write out the string to stdout...</i>
446 <a
447 href="#i_call">call</a> i32 @puts(i8 * %cast210) <i>; i32</i>
448 <a
449 href="#i_ret">ret</a> i32 0<br>}<br>
450</pre>
451</div>
452
453<p>This example is made up of a <a href="#globalvars">global variable</a>
454named "<tt>.LC0</tt>", an external declaration of the "<tt>puts</tt>"
455function, and a <a href="#functionstructure">function definition</a>
456for "<tt>main</tt>".</p>
457
458<p>In general, a module is made up of a list of global values,
459where both functions and global variables are global values. Global values are
460represented by a pointer to a memory location (in this case, a pointer to an
461array of char, and a pointer to a function), and have one of the following <a
462href="#linkage">linkage types</a>.</p>
463
464</div>
465
466<!-- ======================================================================= -->
467<div class="doc_subsection">
468 <a name="linkage">Linkage Types</a>
469</div>
470
471<div class="doc_text">
472
473<p>
474All Global Variables and Functions have one of the following types of linkage:
475</p>
476
477<dl>
478
Dale Johannesen96e7e092008-05-23 23:13:41 +0000[diff] [blame]479 <dt><tt><b><a name="linkage_internal">internal</a></b></tt>: </dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]480
481 <dd>Global values with internal linkage are only directly accessible by
482 objects in the current module. In particular, linking code into a module with
483 an internal global value may cause the internal to be renamed as necessary to
484 avoid collisions. Because the symbol is internal to the module, all
485 references can be updated. This corresponds to the notion of the
486 '<tt>static</tt>' keyword in C.
487 </dd>
488
489 <dt><tt><b><a name="linkage_linkonce">linkonce</a></b></tt>: </dt>
490
491 <dd>Globals with "<tt>linkonce</tt>" linkage are merged with other globals of
492 the same name when linkage occurs. This is typically used to implement
493 inline functions, templates, or other code which must be generated in each
494 translation unit that uses it. Unreferenced <tt>linkonce</tt> globals are
495 allowed to be discarded.
496 </dd>
497
Dale Johannesen96e7e092008-05-23 23:13:41 +0000[diff] [blame]498 <dt><tt><b><a name="linkage_common">common</a></b></tt>: </dt>
499
500 <dd>"<tt>common</tt>" linkage is exactly the same as <tt>linkonce</tt>
501 linkage, except that unreferenced <tt>common</tt> globals may not be
502 discarded. This is used for globals that may be emitted in multiple
503 translation units, but that are not guaranteed to be emitted into every
504 translation unit that uses them. One example of this is tentative
505 definitions in C, such as "<tt>int X;</tt>" at global scope.
506 </dd>
507
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]508 <dt><tt><b><a name="linkage_weak">weak</a></b></tt>: </dt>
509
Dale Johannesen96e7e092008-05-23 23:13:41 +0000[diff] [blame]510 <dd>"<tt>weak</tt>" linkage is the same as <tt>common</tt> linkage, except
511 that some targets may choose to emit different assembly sequences for them
512 for target-dependent reasons. This is used for globals that are declared
513 "weak" in C source code.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]514 </dd>
515
516 <dt><tt><b><a name="linkage_appending">appending</a></b></tt>: </dt>
517
518 <dd>"<tt>appending</tt>" linkage may only be applied to global variables of
519 pointer to array type. When two global variables with appending linkage are
520 linked together, the two global arrays are appended together. This is the
521 LLVM, typesafe, equivalent of having the system linker append together
522 "sections" with identical names when .o files are linked.
523 </dd>
524
525 <dt><tt><b><a name="linkage_externweak">extern_weak</a></b></tt>: </dt>
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]526 <dd>The semantics of this linkage follow the ELF object file model: the
527 symbol is weak until linked, if not linked, the symbol becomes null instead
528 of being an undefined reference.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]529 </dd>
530
531 <dt><tt><b><a name="linkage_external">externally visible</a></b></tt>:</dt>
532
533 <dd>If none of the above identifiers are used, the global is externally
534 visible, meaning that it participates in linkage and can be used to resolve
535 external symbol references.
536 </dd>
537</dl>
538
539 <p>
540 The next two types of linkage are targeted for Microsoft Windows platform
541 only. They are designed to support importing (exporting) symbols from (to)
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]542 DLLs (Dynamic Link Libraries).
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]543 </p>
544
545 <dl>
546 <dt><tt><b><a name="linkage_dllimport">dllimport</a></b></tt>: </dt>
547
548 <dd>"<tt>dllimport</tt>" linkage causes the compiler to reference a function
549 or variable via a global pointer to a pointer that is set up by the DLL
550 exporting the symbol. On Microsoft Windows targets, the pointer name is
551 formed by combining <code>_imp__</code> and the function or variable name.
552 </dd>
553
554 <dt><tt><b><a name="linkage_dllexport">dllexport</a></b></tt>: </dt>
555
556 <dd>"<tt>dllexport</tt>" linkage causes the compiler to provide a global
557 pointer to a pointer in a DLL, so that it can be referenced with the
558 <tt>dllimport</tt> attribute. On Microsoft Windows targets, the pointer
559 name is formed by combining <code>_imp__</code> and the function or variable
560 name.
561 </dd>
562
563</dl>
564
565<p><a name="linkage_external"></a>For example, since the "<tt>.LC0</tt>"
566variable is defined to be internal, if another module defined a "<tt>.LC0</tt>"
567variable and was linked with this one, one of the two would be renamed,
568preventing a collision. Since "<tt>main</tt>" and "<tt>puts</tt>" are
569external (i.e., lacking any linkage declarations), they are accessible
570outside of the current module.</p>
571<p>It is illegal for a function <i>declaration</i>
572to have any linkage type other than "externally visible", <tt>dllimport</tt>,
573or <tt>extern_weak</tt>.</p>
574<p>Aliases can have only <tt>external</tt>, <tt>internal</tt> and <tt>weak</tt>
575linkages.
576</div>
577
578<!-- ======================================================================= -->
579<div class="doc_subsection">
580 <a name="callingconv">Calling Conventions</a>
581</div>
582
583<div class="doc_text">
584
585<p>LLVM <a href="#functionstructure">functions</a>, <a href="#i_call">calls</a>
586and <a href="#i_invoke">invokes</a> can all have an optional calling convention
587specified for the call. The calling convention of any pair of dynamic
588caller/callee must match, or the behavior of the program is undefined. The
589following calling conventions are supported by LLVM, and more may be added in
590the future:</p>
591
592<dl>
593 <dt><b>"<tt>ccc</tt>" - The C calling convention</b>:</dt>
594
595 <dd>This calling convention (the default if no other calling convention is
596 specified) matches the target C calling conventions. This calling convention
597 supports varargs function calls and tolerates some mismatch in the declared
598 prototype and implemented declaration of the function (as does normal C).
599 </dd>
600
601 <dt><b>"<tt>fastcc</tt>" - The fast calling convention</b>:</dt>
602
603 <dd>This calling convention attempts to make calls as fast as possible
604 (e.g. by passing things in registers). This calling convention allows the
605 target to use whatever tricks it wants to produce fast code for the target,
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]606 without having to conform to an externally specified ABI (Application Binary
607 Interface). Implementations of this convention should allow arbitrary
Arnold Schwaighofer07444922008-05-14 09:17:12 +0000[diff] [blame]608 <a href="CodeGenerator.html#tailcallopt">tail call optimization</a> to be
609 supported. This calling convention does not support varargs and requires the
610 prototype of all callees to exactly match the prototype of the function
611 definition.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]612 </dd>
613
614 <dt><b>"<tt>coldcc</tt>" - The cold calling convention</b>:</dt>
615
616 <dd>This calling convention attempts to make code in the caller as efficient
617 as possible under the assumption that the call is not commonly executed. As
618 such, these calls often preserve all registers so that the call does not break
619 any live ranges in the caller side. This calling convention does not support
620 varargs and requires the prototype of all callees to exactly match the
621 prototype of the function definition.
622 </dd>
623
624 <dt><b>"<tt>cc &lt;<em>n</em>&gt;</tt>" - Numbered convention</b>:</dt>
625
626 <dd>Any calling convention may be specified by number, allowing
627 target-specific calling conventions to be used. Target specific calling
628 conventions start at 64.
629 </dd>
630</dl>
631
632<p>More calling conventions can be added/defined on an as-needed basis, to
633support pascal conventions or any other well-known target-independent
634convention.</p>
635
636</div>
637
638<!-- ======================================================================= -->
639<div class="doc_subsection">
640 <a name="visibility">Visibility Styles</a>
641</div>
642
643<div class="doc_text">
644
645<p>
646All Global Variables and Functions have one of the following visibility styles:
647</p>
648
649<dl>
650 <dt><b>"<tt>default</tt>" - Default style</b>:</dt>
651
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]652 <dd>On targets that use the ELF object file format, default visibility means
653 that the declaration is visible to other
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]654 modules and, in shared libraries, means that the declared entity may be
655 overridden. On Darwin, default visibility means that the declaration is
656 visible to other modules. Default visibility corresponds to "external
657 linkage" in the language.
658 </dd>
659
660 <dt><b>"<tt>hidden</tt>" - Hidden style</b>:</dt>
661
662 <dd>Two declarations of an object with hidden visibility refer to the same
663 object if they are in the same shared object. Usually, hidden visibility
664 indicates that the symbol will not be placed into the dynamic symbol table,
665 so no other module (executable or shared library) can reference it
666 directly.
667 </dd>
668
669 <dt><b>"<tt>protected</tt>" - Protected style</b>:</dt>
670
671 <dd>On ELF, protected visibility indicates that the symbol will be placed in
672 the dynamic symbol table, but that references within the defining module will
673 bind to the local symbol. That is, the symbol cannot be overridden by another
674 module.
675 </dd>
676</dl>
677
678</div>
679
680<!-- ======================================================================= -->
681<div class="doc_subsection">
682 <a name="globalvars">Global Variables</a>
683</div>
684
685<div class="doc_text">
686
687<p>Global variables define regions of memory allocated at compilation time
688instead of run-time. Global variables may optionally be initialized, may have
689an explicit section to be placed in, and may have an optional explicit alignment
690specified. A variable may be defined as "thread_local", which means that it
691will not be shared by threads (each thread will have a separated copy of the
692variable). A variable may be defined as a global "constant," which indicates
693that the contents of the variable will <b>never</b> be modified (enabling better
694optimization, allowing the global data to be placed in the read-only section of
695an executable, etc). Note that variables that need runtime initialization
696cannot be marked "constant" as there is a store to the variable.</p>
697
698<p>
699LLVM explicitly allows <em>declarations</em> of global variables to be marked
700constant, even if the final definition of the global is not. This capability
701can be used to enable slightly better optimization of the program, but requires
702the language definition to guarantee that optimizations based on the
703'constantness' are valid for the translation units that do not include the
704definition.
705</p>
706
707<p>As SSA values, global variables define pointer values that are in
708scope (i.e. they dominate) all basic blocks in the program. Global
709variables always define a pointer to their "content" type because they
710describe a region of memory, and all memory objects in LLVM are
711accessed through pointers.</p>
712
Christopher Lambdd0049d2007-12-11 09:31:00 +0000[diff] [blame]713<p>A global variable may be declared to reside in a target-specifc numbered
714address space. For targets that support them, address spaces may affect how
715optimizations are performed and/or what target instructions are used to access
Christopher Lamb20a39e92007-12-12 08:44:39 +0000[diff] [blame]716the variable. The default address space is zero. The address space qualifier
717must precede any other attributes.</p>
Christopher Lambdd0049d2007-12-11 09:31:00 +0000[diff] [blame]718
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]719<p>LLVM allows an explicit section to be specified for globals. If the target
720supports it, it will emit globals to the section specified.</p>
721
722<p>An explicit alignment may be specified for a global. If not present, or if
723the alignment is set to zero, the alignment of the global is set by the target
724to whatever it feels convenient. If an explicit alignment is specified, the
725global is forced to have at least that much alignment. All alignments must be
726a power of 2.</p>
727
Christopher Lambdd0049d2007-12-11 09:31:00 +0000[diff] [blame]728<p>For example, the following defines a global in a numbered address space with
729an initializer, section, and alignment:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]730
731<div class="doc_code">
732<pre>
Christopher Lambdd0049d2007-12-11 09:31:00 +0000[diff] [blame]733@G = constant float 1.0 addrspace(5), section "foo", align 4
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]734</pre>
735</div>
736
737</div>
738
739
740<!-- ======================================================================= -->
741<div class="doc_subsection">
742 <a name="functionstructure">Functions</a>
743</div>
744
745<div class="doc_text">
746
747<p>LLVM function definitions consist of the "<tt>define</tt>" keyord,
748an optional <a href="#linkage">linkage type</a>, an optional
749<a href="#visibility">visibility style</a>, an optional
750<a href="#callingconv">calling convention</a>, a return type, an optional
751<a href="#paramattrs">parameter attribute</a> for the return type, a function
752name, a (possibly empty) argument list (each with optional
753<a href="#paramattrs">parameter attributes</a>), an optional section, an
Devang Pateld468f1c2008-09-04 23:05:13 +0000[diff] [blame]754optional 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
787</div>
788
789
790<!-- ======================================================================= -->
791<div class="doc_subsection">
792 <a name="aliasstructure">Aliases</a>
793</div>
794<div class="doc_text">
795 <p>Aliases act as "second name" for the aliasee value (which can be either
Anton Korobeynikov96822812008-03-22 08:36:14 +0000[diff] [blame]796 function, global variable, another alias or bitcast of global value). Aliases
797 may have an optional <a href="#linkage">linkage type</a>, and an
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]798 optional <a href="#visibility">visibility style</a>.</p>
799
800 <h5>Syntax:</h5>
801
802<div class="doc_code">
803<pre>
Duncan Sandsd7bfabf2008-09-12 20:48:21 +0000[diff] [blame]804@&lt;Name&gt; = alias [Linkage] [Visibility] &lt;AliaseeTy&gt; @&lt;Aliasee&gt;
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]805</pre>
806</div>
807
808</div>
809
810
811
812<!-- ======================================================================= -->
813<div class="doc_subsection"><a name="paramattrs">Parameter Attributes</a></div>
814<div class="doc_text">
815 <p>The return type and each parameter of a function type may have a set of
816 <i>parameter attributes</i> associated with them. Parameter attributes are
817 used to communicate additional information about the result or parameters of
Duncan Sandsf5588dc2007-11-27 13:23:08 +0000[diff] [blame]818 a function. Parameter attributes are considered to be part of the function,
819 not of the function type, so functions with different parameter attributes
820 can have the same function type.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]821
822 <p>Parameter attributes are simple keywords that follow the type specified. If
823 multiple parameter attributes are needed, they are space separated. For
824 example:</p>
825
826<div class="doc_code">
827<pre>
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]828declare i32 @printf(i8* noalias , ...)
829declare i32 @atoi(i8 zeroext*)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]830</pre>
831</div>
832
Duncan Sandsf5588dc2007-11-27 13:23:08 +0000[diff] [blame]833 <p>Note that any attributes for the function result (<tt>nounwind</tt>,
834 <tt>readonly</tt>) come immediately after the argument list.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]835
836 <p>Currently, only the following parameter attributes are defined:</p>
837 <dl>
Reid Spencerf234bed2007-07-19 23:13:04 +0000[diff] [blame]838 <dt><tt>zeroext</tt></dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]839 <dd>This indicates that the parameter should be zero extended just before
840 a call to this function.</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]841
Reid Spencerf234bed2007-07-19 23:13:04 +0000[diff] [blame]842 <dt><tt>signext</tt></dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]843 <dd>This indicates that the parameter should be sign extended just before
844 a call to this function.</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]845
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]846 <dt><tt>inreg</tt></dt>
Dale Johannesenc08a0e22008-09-25 20:47:45 +0000[diff] [blame]847 <dd>This indicates that this parameter or return value should be treated
848 in a special target-dependent fashion during while emitting code for a
849 function call or return (usually, by putting it in a register as opposed
850 to memory; in some places it is used to distinguish between two different
851 kinds of registers). Use of this attribute is target-specific</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]852
853 <dt><tt>byval</tt></dt>
Chris Lattner04c86182008-01-15 04:34:22 +0000[diff] [blame]854 <dd>This indicates that the pointer parameter should really be passed by
855 value to the function. The attribute implies that a hidden copy of the
856 pointee is made between the caller and the callee, so the callee is unable
Chris Lattner6a9f3c42008-08-05 18:21:08 +0000[diff] [blame]857 to modify the value in the callee. This attribute is only valid on LLVM
Chris Lattner04c86182008-01-15 04:34:22 +0000[diff] [blame]858 pointer arguments. It is generally used to pass structs and arrays by
859 value, but is also valid on scalars (even though this is silly).</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>sret</tt></dt>
Duncan Sands616cc032008-02-18 04:19:38 +0000[diff] [blame]862 <dd>This indicates that the pointer parameter specifies the address of a
863 structure that is the return value of the function in the source program.
Duncan Sandsef0e9e42008-03-17 12:17:41 +0000[diff] [blame]864 Loads and stores to the structure are assumed not to trap.
Duncan Sands616cc032008-02-18 04:19:38 +0000[diff] [blame]865 May only be applied to the first parameter.</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]866
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]867 <dt><tt>noalias</tt></dt>
Owen Andersonc4fc4cd2008-02-18 04:09:01 +0000[diff] [blame]868 <dd>This indicates that the parameter does not alias any global or any other
869 parameter. The caller is responsible for ensuring that this is the case,
870 usually by placing the value in a stack allocation.</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]871
Duncan Sands4ee46812007-07-27 19:57:41 +0000[diff] [blame]872 <dt><tt>nest</tt></dt>
Duncan Sandsf1a7d4c2008-07-08 09:27:25 +0000[diff] [blame]873 <dd>This indicates that the pointer parameter can be excised using the
Duncan Sands4ee46812007-07-27 19:57:41 +0000[diff] [blame]874 <a href="#int_trampoline">trampoline intrinsics</a>.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]875 </dl>
876
877</div>
878
879<!-- ======================================================================= -->
880<div class="doc_subsection">
Gordon Henriksen13fe5e32007-12-10 03:18:06 +0000[diff] [blame]881 <a name="gc">Garbage Collector Names</a>
882</div>
883
884<div class="doc_text">
885<p>Each function may specify a garbage collector name, which is simply a
886string.</p>
887
888<div class="doc_code"><pre
889>define void @f() gc "name" { ...</pre></div>
890
891<p>The compiler declares the supported values of <i>name</i>. Specifying a
892collector which will cause the compiler to alter its output in order to support
893the named garbage collection algorithm.</p>
894</div>
895
896<!-- ======================================================================= -->
897<div class="doc_subsection">
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]898 <a name="fnattrs">Function Attributes</a>
Devang Pateld468f1c2008-09-04 23:05:13 +0000[diff] [blame]899</div>
900
901<div class="doc_text">
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]902
903<p>Function attributes are set to communicate additional information about
904 a function. Function attributes are considered to be part of the function,
905 not of the function type, so functions with different parameter attributes
906 can have the same function type.</p>
907
908 <p>Function attributes are simple keywords that follow the type specified. If
909 multiple attributes are needed, they are space separated. For
910 example:</p>
Devang Pateld468f1c2008-09-04 23:05:13 +0000[diff] [blame]911
912<div class="doc_code">
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]913<pre>
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]914define void @f() noinline { ... }
915define void @f() alwaysinline { ... }
916define void @f() alwaysinline optsize { ... }
917define void @f() optsize
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]918</pre>
Devang Pateld468f1c2008-09-04 23:05:13 +0000[diff] [blame]919</div>
920
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]921<dl>
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]922<dt><tt>alwaysinline</tt></dt>
923<dd>This attribute requests inliner to inline this function irrespective of
924inlining size threshold for this function.</dd>
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]925
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]926<dt><tt>noinline</tt></dt>
927<dd>This attributes requests inliner to never inline this function in any
928situation. This attribute may not be used together with <tt>alwaysinline</tt>
929 attribute.</dd>
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]930
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]931<dt><tt>optsize</tt></dt>
Devang Patelc29099b2008-09-29 18:34:44 +0000[diff] [blame]932<dd>This attribute suggests that optimization passes and code generator passes
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]933make choices that help reduce code size.</dd>
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]934
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]935<dt><tt>noreturn</tt></dt>
936<dd>This function attribute indicates that the function never returns. This
Devang Patelcbce03c2008-09-29 18:16:09 +0000[diff] [blame]937 tells LLVM that every call to this function should be treated as if
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]938 an <tt>unreachable</tt> instruction immediately followed the call.</dd>
939
940<dt><tt>nounwind</tt></dt>
941<dd>This function attribute indicates that no exceptions unwind out of the
942 function. Usually this is because the function makes no use of exceptions,
943 but it may also be that the function catches any exceptions thrown when
944 executing it.</dd>
945
946<dt><tt>readonly</tt></dt>
947<dd>This function attribute indicates that the function has no side-effects
948 except for producing a return value or throwing an exception. The value
949 returned must only depend on the function arguments and/or global variables.
950 It may use values obtained by dereferencing pointers.</dd>
951<dt><tt>readnone</tt></dt>
952<dd>A <tt>readnone</tt> function has the same restrictions as a <tt>readonly</tt>
953 function, but in addition it is not allowed to dereference any pointer arguments
954 or global variables.
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]955</dl>
956
Devang Pateld468f1c2008-09-04 23:05:13 +0000[diff] [blame]957</div>
958
959<!-- ======================================================================= -->
960<div class="doc_subsection">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]961 <a name="moduleasm">Module-Level Inline Assembly</a>
962</div>
963
964<div class="doc_text">
965<p>
966Modules may contain "module-level inline asm" blocks, which corresponds to the
967GCC "file scope inline asm" blocks. These blocks are internally concatenated by
968LLVM and treated as a single unit, but may be separated in the .ll file if
969desired. The syntax is very simple:
970</p>
971
972<div class="doc_code">
973<pre>
974module asm "inline asm code goes here"
975module asm "more can go here"
976</pre>
977</div>
978
979<p>The strings can contain any character by escaping non-printable characters.
980 The escape sequence used is simply "\xx" where "xx" is the two digit hex code
981 for the number.
982</p>
983
984<p>
985 The inline asm code is simply printed to the machine code .s file when
986 assembly code is generated.
987</p>
988</div>
989
990<!-- ======================================================================= -->
991<div class="doc_subsection">
992 <a name="datalayout">Data Layout</a>
993</div>
994
995<div class="doc_text">
996<p>A module may specify a target specific data layout string that specifies how
997data is to be laid out in memory. The syntax for the data layout is simply:</p>
998<pre> target datalayout = "<i>layout specification</i>"</pre>
999<p>The <i>layout specification</i> consists of a list of specifications
1000separated by the minus sign character ('-'). Each specification starts with a
1001letter and may include other information after the letter to define some
1002aspect of the data layout. The specifications accepted are as follows: </p>
1003<dl>
1004 <dt><tt>E</tt></dt>
1005 <dd>Specifies that the target lays out data in big-endian form. That is, the
1006 bits with the most significance have the lowest address location.</dd>
1007 <dt><tt>e</tt></dt>
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]1008 <dd>Specifies that the target lays out data in little-endian form. That is,
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1009 the bits with the least significance have the lowest address location.</dd>
1010 <dt><tt>p:<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1011 <dd>This specifies the <i>size</i> of a pointer and its <i>abi</i> and
1012 <i>preferred</i> alignments. All sizes are in bits. Specifying the <i>pref</i>
1013 alignment is optional. If omitted, the preceding <tt>:</tt> should be omitted
1014 too.</dd>
1015 <dt><tt>i<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1016 <dd>This specifies the alignment for an integer type of a given bit
1017 <i>size</i>. The value of <i>size</i> must be in the range [1,2^23).</dd>
1018 <dt><tt>v<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1019 <dd>This specifies the alignment for a vector type of a given bit
1020 <i>size</i>.</dd>
1021 <dt><tt>f<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1022 <dd>This specifies the alignment for a floating point type of a given bit
1023 <i>size</i>. The value of <i>size</i> must be either 32 (float) or 64
1024 (double).</dd>
1025 <dt><tt>a<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1026 <dd>This specifies the alignment for an aggregate type of a given bit
1027 <i>size</i>.</dd>
1028</dl>
1029<p>When constructing the data layout for a given target, LLVM starts with a
1030default set of specifications which are then (possibly) overriden by the
1031specifications in the <tt>datalayout</tt> keyword. The default specifications
1032are given in this list:</p>
1033<ul>
1034 <li><tt>E</tt> - big endian</li>
1035 <li><tt>p:32:64:64</tt> - 32-bit pointers with 64-bit alignment</li>
1036 <li><tt>i1:8:8</tt> - i1 is 8-bit (byte) aligned</li>
1037 <li><tt>i8:8:8</tt> - i8 is 8-bit (byte) aligned</li>
1038 <li><tt>i16:16:16</tt> - i16 is 16-bit aligned</li>
1039 <li><tt>i32:32:32</tt> - i32 is 32-bit aligned</li>
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]1040 <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]1041 alignment of 64-bits</li>
1042 <li><tt>f32:32:32</tt> - float is 32-bit aligned</li>
1043 <li><tt>f64:64:64</tt> - double is 64-bit aligned</li>
1044 <li><tt>v64:64:64</tt> - 64-bit vector is 64-bit aligned</li>
1045 <li><tt>v128:128:128</tt> - 128-bit vector is 128-bit aligned</li>
1046 <li><tt>a0:0:1</tt> - aggregates are 8-bit aligned</li>
1047</ul>
Chris Lattner6a9f3c42008-08-05 18:21:08 +0000[diff] [blame]1048<p>When LLVM is determining the alignment for a given type, it uses the
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1049following rules:
1050<ol>
1051 <li>If the type sought is an exact match for one of the specifications, that
1052 specification is used.</li>
1053 <li>If no match is found, and the type sought is an integer type, then the
1054 smallest integer type that is larger than the bitwidth of the sought type is
1055 used. If none of the specifications are larger than the bitwidth then the the
1056 largest integer type is used. For example, given the default specifications
1057 above, the i7 type will use the alignment of i8 (next largest) while both
1058 i65 and i256 will use the alignment of i64 (largest specified).</li>
1059 <li>If no match is found, and the type sought is a vector type, then the
1060 largest vector type that is smaller than the sought vector type will be used
1061 as a fall back. This happens because <128 x double> can be implemented in
1062 terms of 64 <2 x double>, for example.</li>
1063</ol>
1064</div>
1065
1066<!-- *********************************************************************** -->
1067<div class="doc_section"> <a name="typesystem">Type System</a> </div>
1068<!-- *********************************************************************** -->
1069
1070<div class="doc_text">
1071
1072<p>The LLVM type system is one of the most important features of the
1073intermediate representation. Being typed enables a number of
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]1074optimizations to be performed on the intermediate representation directly,
1075without having to do
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1076extra analyses on the side before the transformation. A strong type
1077system makes it easier to read the generated code and enables novel
1078analyses and transformations that are not feasible to perform on normal
1079three address code representations.</p>
1080
1081</div>
1082
1083<!-- ======================================================================= -->
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1084<div class="doc_subsection"> <a name="t_classifications">Type
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1085Classifications</a> </div>
1086<div class="doc_text">
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1087<p>The types fall into a few useful
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1088classifications:</p>
1089
1090<table border="1" cellspacing="0" cellpadding="4">
1091 <tbody>
1092 <tr><th>Classification</th><th>Types</th></tr>
1093 <tr>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1094 <td><a href="#t_integer">integer</a></td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1095 <td><tt>i1, i2, i3, ... i8, ... i16, ... i32, ... i64, ... </tt></td>
1096 </tr>
1097 <tr>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1098 <td><a href="#t_floating">floating point</a></td>
1099 <td><tt>float, double, x86_fp80, fp128, ppc_fp128</tt></td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1100 </tr>
1101 <tr>
1102 <td><a name="t_firstclass">first class</a></td>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1103 <td><a href="#t_integer">integer</a>,
1104 <a href="#t_floating">floating point</a>,
1105 <a href="#t_pointer">pointer</a>,
Dan Gohmanf6237db2008-06-18 18:42:13 +0000[diff] [blame]1106 <a href="#t_vector">vector</a>,
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]1107 <a href="#t_struct">structure</a>,
1108 <a href="#t_array">array</a>,
Dan Gohmand13951e2008-05-23 22:50:26 +0000[diff] [blame]1109 <a href="#t_label">label</a>.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1110 </td>
1111 </tr>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1112 <tr>
1113 <td><a href="#t_primitive">primitive</a></td>
1114 <td><a href="#t_label">label</a>,
1115 <a href="#t_void">void</a>,
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1116 <a href="#t_floating">floating point</a>.</td>
1117 </tr>
1118 <tr>
1119 <td><a href="#t_derived">derived</a></td>
1120 <td><a href="#t_integer">integer</a>,
1121 <a href="#t_array">array</a>,
1122 <a href="#t_function">function</a>,
1123 <a href="#t_pointer">pointer</a>,
1124 <a href="#t_struct">structure</a>,
1125 <a href="#t_pstruct">packed structure</a>,
1126 <a href="#t_vector">vector</a>,
1127 <a href="#t_opaque">opaque</a>.
1128 </tr>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1129 </tbody>
1130</table>
1131
1132<p>The <a href="#t_firstclass">first class</a> types are perhaps the
1133most important. Values of these types are the only ones which can be
1134produced by instructions, passed as arguments, or used as operands to
Dan Gohman4f29e422008-05-23 21:53:15 +0000[diff] [blame]1135instructions.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1136</div>
1137
1138<!-- ======================================================================= -->
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1139<div class="doc_subsection"> <a name="t_primitive">Primitive Types</a> </div>
Chris Lattner86437612008-01-04 04:34:14 +0000[diff] [blame]1140
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1141<div class="doc_text">
1142<p>The primitive types are the fundamental building blocks of the LLVM
1143system.</p>
1144
Chris Lattner86437612008-01-04 04:34:14 +0000[diff] [blame]1145</div>
1146
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1147<!-- _______________________________________________________________________ -->
1148<div class="doc_subsubsection"> <a name="t_floating">Floating Point Types</a> </div>
1149
1150<div class="doc_text">
1151 <table>
1152 <tbody>
1153 <tr><th>Type</th><th>Description</th></tr>
1154 <tr><td><tt>float</tt></td><td>32-bit floating point value</td></tr>
1155 <tr><td><tt>double</tt></td><td>64-bit floating point value</td></tr>
1156 <tr><td><tt>fp128</tt></td><td>128-bit floating point value (112-bit mantissa)</td></tr>
1157 <tr><td><tt>x86_fp80</tt></td><td>80-bit floating point value (X87)</td></tr>
1158 <tr><td><tt>ppc_fp128</tt></td><td>128-bit floating point value (two 64-bits)</td></tr>
1159 </tbody>
1160 </table>
1161</div>
1162
1163<!-- _______________________________________________________________________ -->
1164<div class="doc_subsubsection"> <a name="t_void">Void Type</a> </div>
1165
1166<div class="doc_text">
1167<h5>Overview:</h5>
1168<p>The void type does not represent any value and has no size.</p>
1169
1170<h5>Syntax:</h5>
1171
1172<pre>
1173 void
1174</pre>
1175</div>
1176
1177<!-- _______________________________________________________________________ -->
1178<div class="doc_subsubsection"> <a name="t_label">Label Type</a> </div>
1179
1180<div class="doc_text">
1181<h5>Overview:</h5>
1182<p>The label type represents code labels.</p>
1183
1184<h5>Syntax:</h5>
1185
1186<pre>
1187 label
1188</pre>
1189</div>
1190
1191
1192<!-- ======================================================================= -->
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1193<div class="doc_subsection"> <a name="t_derived">Derived Types</a> </div>
1194
1195<div class="doc_text">
1196
1197<p>The real power in LLVM comes from the derived types in the system.
1198This is what allows a programmer to represent arrays, functions,
1199pointers, and other useful types. Note that these derived types may be
1200recursive: For example, it is possible to have a two dimensional array.</p>
1201
1202</div>
1203
1204<!-- _______________________________________________________________________ -->
1205<div class="doc_subsubsection"> <a name="t_integer">Integer Type</a> </div>
1206
1207<div class="doc_text">
1208
1209<h5>Overview:</h5>
1210<p>The integer type is a very simple derived type that simply specifies an
1211arbitrary bit width for the integer type desired. Any bit width from 1 bit to
12122^23-1 (about 8 million) can be specified.</p>
1213
1214<h5>Syntax:</h5>
1215
1216<pre>
1217 iN
1218</pre>
1219
1220<p>The number of bits the integer will occupy is specified by the <tt>N</tt>
1221value.</p>
1222
1223<h5>Examples:</h5>
1224<table class="layout">
Chris Lattner251ab812007-12-18 06:18:21 +0000[diff] [blame]1225 <tbody>
1226 <tr>
1227 <td><tt>i1</tt></td>
1228 <td>a single-bit integer.</td>
1229 </tr><tr>
1230 <td><tt>i32</tt></td>
1231 <td>a 32-bit integer.</td>
1232 </tr><tr>
1233 <td><tt>i1942652</tt></td>
1234 <td>a really big integer of over 1 million bits.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1235 </tr>
Chris Lattner251ab812007-12-18 06:18:21 +0000[diff] [blame]1236 </tbody>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1237</table>
1238</div>
1239
1240<!-- _______________________________________________________________________ -->
1241<div class="doc_subsubsection"> <a name="t_array">Array Type</a> </div>
1242
1243<div class="doc_text">
1244
1245<h5>Overview:</h5>
1246
1247<p>The array type is a very simple derived type that arranges elements
1248sequentially in memory. The array type requires a size (number of
1249elements) and an underlying data type.</p>
1250
1251<h5>Syntax:</h5>
1252
1253<pre>
1254 [&lt;# elements&gt; x &lt;elementtype&gt;]
1255</pre>
1256
1257<p>The number of elements is a constant integer value; elementtype may
1258be any type with a size.</p>
1259
1260<h5>Examples:</h5>
1261<table class="layout">
1262 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1263 <td class="left"><tt>[40 x i32]</tt></td>
1264 <td class="left">Array of 40 32-bit integer values.</td>
1265 </tr>
1266 <tr class="layout">
1267 <td class="left"><tt>[41 x i32]</tt></td>
1268 <td class="left">Array of 41 32-bit integer values.</td>
1269 </tr>
1270 <tr class="layout">
1271 <td class="left"><tt>[4 x i8]</tt></td>
1272 <td class="left">Array of 4 8-bit integer values.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1273 </tr>
1274</table>
1275<p>Here are some examples of multidimensional arrays:</p>
1276<table class="layout">
1277 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1278 <td class="left"><tt>[3 x [4 x i32]]</tt></td>
1279 <td class="left">3x4 array of 32-bit integer values.</td>
1280 </tr>
1281 <tr class="layout">
1282 <td class="left"><tt>[12 x [10 x float]]</tt></td>
1283 <td class="left">12x10 array of single precision floating point values.</td>
1284 </tr>
1285 <tr class="layout">
1286 <td class="left"><tt>[2 x [3 x [4 x i16]]]</tt></td>
1287 <td class="left">2x3x4 array of 16-bit integer values.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1288 </tr>
1289</table>
1290
1291<p>Note that 'variable sized arrays' can be implemented in LLVM with a zero
1292length array. Normally, accesses past the end of an array are undefined in
1293LLVM (e.g. it is illegal to access the 5th element of a 3 element array).
1294As a special case, however, zero length arrays are recognized to be variable
1295length. This allows implementation of 'pascal style arrays' with the LLVM
1296type "{ i32, [0 x float]}", for example.</p>
1297
1298</div>
1299
1300<!-- _______________________________________________________________________ -->
1301<div class="doc_subsubsection"> <a name="t_function">Function Type</a> </div>
1302<div class="doc_text">
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1303
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1304<h5>Overview:</h5>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1305
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1306<p>The function type can be thought of as a function signature. It
Devang Pateld4ba41d2008-03-24 05:35:41 +0000[diff] [blame]1307consists of a return type and a list of formal parameter types. The
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1308return 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]1309If 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]1310class types, and the struct must have at least one element.</p>
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]1311
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1312<h5>Syntax:</h5>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1313
1314<pre>
1315 &lt;returntype list&gt; (&lt;parameter list&gt;)
1316</pre>
1317
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1318<p>...where '<tt>&lt;parameter list&gt;</tt>' is a comma-separated list of type
1319specifiers. Optionally, the parameter list may include a type <tt>...</tt>,
1320which indicates that the function takes a variable number of arguments.
1321Variable argument functions can access their arguments with the <a
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]1322 href="#int_varargs">variable argument handling intrinsic</a> functions.
1323'<tt>&lt;returntype list&gt;</tt>' is a comma-separated list of
1324<a href="#t_firstclass">first class</a> type specifiers.</p>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1325
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1326<h5>Examples:</h5>
1327<table class="layout">
1328 <tr class="layout">
1329 <td class="left"><tt>i32 (i32)</tt></td>
1330 <td class="left">function taking an <tt>i32</tt>, returning an <tt>i32</tt>
1331 </td>
1332 </tr><tr class="layout">
Reid Spencerf234bed2007-07-19 23:13:04 +0000[diff] [blame]1333 <td class="left"><tt>float&nbsp;(i16&nbsp;signext,&nbsp;i32&nbsp;*)&nbsp;*
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1334 </tt></td>
1335 <td class="left"><a href="#t_pointer">Pointer</a> to a function that takes
1336 an <tt>i16</tt> that should be sign extended and a
1337 <a href="#t_pointer">pointer</a> to <tt>i32</tt>, returning
1338 <tt>float</tt>.
1339 </td>
1340 </tr><tr class="layout">
1341 <td class="left"><tt>i32 (i8*, ...)</tt></td>
1342 <td class="left">A vararg function that takes at least one
1343 <a href="#t_pointer">pointer</a> to <tt>i8 </tt> (char in C),
1344 which returns an integer. This is the signature for <tt>printf</tt> in
1345 LLVM.
1346 </td>
Devang Pateld4ba41d2008-03-24 05:35:41 +0000[diff] [blame]1347 </tr><tr class="layout">
1348 <td class="left"><tt>{i32, i32} (i32)</tt></td>
Devang Patel6dddba22008-03-24 18:10:52 +0000[diff] [blame]1349 <td class="left">A function taking an <tt>i32></tt>, returning two
1350 <tt> i32 </tt> values as an aggregate of type <tt>{ i32, i32 }</tt>
Devang Pateld4ba41d2008-03-24 05:35:41 +0000[diff] [blame]1351 </td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1352 </tr>
1353</table>
1354
1355</div>
1356<!-- _______________________________________________________________________ -->
1357<div class="doc_subsubsection"> <a name="t_struct">Structure Type</a> </div>
1358<div class="doc_text">
1359<h5>Overview:</h5>
1360<p>The structure type is used to represent a collection of data members
1361together in memory. The packing of the field types is defined to match
1362the ABI of the underlying processor. The elements of a structure may
1363be any type that has a size.</p>
1364<p>Structures are accessed using '<tt><a href="#i_load">load</a></tt>
1365and '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a
1366field with the '<tt><a href="#i_getelementptr">getelementptr</a></tt>'
1367instruction.</p>
1368<h5>Syntax:</h5>
1369<pre> { &lt;type list&gt; }<br></pre>
1370<h5>Examples:</h5>
1371<table class="layout">
1372 <tr class="layout">
1373 <td class="left"><tt>{ i32, i32, i32 }</tt></td>
1374 <td class="left">A triple of three <tt>i32</tt> values</td>
1375 </tr><tr class="layout">
1376 <td class="left"><tt>{&nbsp;float,&nbsp;i32&nbsp;(i32)&nbsp;*&nbsp;}</tt></td>
1377 <td class="left">A pair, where the first element is a <tt>float</tt> and the
1378 second element is a <a href="#t_pointer">pointer</a> to a
1379 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
1380 an <tt>i32</tt>.</td>
1381 </tr>
1382</table>
1383</div>
1384
1385<!-- _______________________________________________________________________ -->
1386<div class="doc_subsubsection"> <a name="t_pstruct">Packed Structure Type</a>
1387</div>
1388<div class="doc_text">
1389<h5>Overview:</h5>
1390<p>The packed structure type is used to represent a collection of data members
1391together in memory. There is no padding between fields. Further, the alignment
1392of a packed structure is 1 byte. The elements of a packed structure may
1393be any type that has a size.</p>
1394<p>Structures are accessed using '<tt><a href="#i_load">load</a></tt>
1395and '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a
1396field with the '<tt><a href="#i_getelementptr">getelementptr</a></tt>'
1397instruction.</p>
1398<h5>Syntax:</h5>
1399<pre> &lt; { &lt;type list&gt; } &gt; <br></pre>
1400<h5>Examples:</h5>
1401<table class="layout">
1402 <tr class="layout">
1403 <td class="left"><tt>&lt; { i32, i32, i32 } &gt;</tt></td>
1404 <td class="left">A triple of three <tt>i32</tt> values</td>
1405 </tr><tr class="layout">
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]1406 <td class="left">
1407<tt>&lt;&nbsp;{&nbsp;float,&nbsp;i32&nbsp;(i32)*&nbsp;}&nbsp;&gt;</tt></td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1408 <td class="left">A pair, where the first element is a <tt>float</tt> and the
1409 second element is a <a href="#t_pointer">pointer</a> to a
1410 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
1411 an <tt>i32</tt>.</td>
1412 </tr>
1413</table>
1414</div>
1415
1416<!-- _______________________________________________________________________ -->
1417<div class="doc_subsubsection"> <a name="t_pointer">Pointer Type</a> </div>
1418<div class="doc_text">
1419<h5>Overview:</h5>
1420<p>As in many languages, the pointer type represents a pointer or
Christopher Lambdd0049d2007-12-11 09:31:00 +0000[diff] [blame]1421reference to another object, which must live in memory. Pointer types may have
1422an optional address space attribute defining the target-specific numbered
1423address space where the pointed-to object resides. The default address space is
1424zero.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1425<h5>Syntax:</h5>
1426<pre> &lt;type&gt; *<br></pre>
1427<h5>Examples:</h5>
1428<table class="layout">
1429 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1430 <td class="left"><tt>[4x i32]*</tt></td>
1431 <td class="left">A <a href="#t_pointer">pointer</a> to <a
1432 href="#t_array">array</a> of four <tt>i32</tt> values.</td>
1433 </tr>
1434 <tr class="layout">
1435 <td class="left"><tt>i32 (i32 *) *</tt></td>
1436 <td class="left"> A <a href="#t_pointer">pointer</a> to a <a
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1437 href="#t_function">function</a> that takes an <tt>i32*</tt>, returning an
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1438 <tt>i32</tt>.</td>
1439 </tr>
1440 <tr class="layout">
1441 <td class="left"><tt>i32 addrspace(5)*</tt></td>
1442 <td class="left">A <a href="#t_pointer">pointer</a> to an <tt>i32</tt> value
1443 that resides in address space #5.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1444 </tr>
1445</table>
1446</div>
1447
1448<!-- _______________________________________________________________________ -->
1449<div class="doc_subsubsection"> <a name="t_vector">Vector Type</a> </div>
1450<div class="doc_text">
1451
1452<h5>Overview:</h5>
1453
1454<p>A vector type is a simple derived type that represents a vector
1455of elements. Vector types are used when multiple primitive data
1456are operated in parallel using a single instruction (SIMD).
1457A vector type requires a size (number of
1458elements) and an underlying primitive data type. Vectors must have a power
1459of two length (1, 2, 4, 8, 16 ...). Vector types are
1460considered <a href="#t_firstclass">first class</a>.</p>
1461
1462<h5>Syntax:</h5>
1463
1464<pre>
1465 &lt; &lt;# elements&gt; x &lt;elementtype&gt; &gt;
1466</pre>
1467
1468<p>The number of elements is a constant integer value; elementtype may
1469be any integer or floating point type.</p>
1470
1471<h5>Examples:</h5>
1472
1473<table class="layout">
1474 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1475 <td class="left"><tt>&lt;4 x i32&gt;</tt></td>
1476 <td class="left">Vector of 4 32-bit integer values.</td>
1477 </tr>
1478 <tr class="layout">
1479 <td class="left"><tt>&lt;8 x float&gt;</tt></td>
1480 <td class="left">Vector of 8 32-bit floating-point values.</td>
1481 </tr>
1482 <tr class="layout">
1483 <td class="left"><tt>&lt;2 x i64&gt;</tt></td>
1484 <td class="left">Vector of 2 64-bit integer values.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1485 </tr>
1486</table>
1487</div>
1488
1489<!-- _______________________________________________________________________ -->
1490<div class="doc_subsubsection"> <a name="t_opaque">Opaque Type</a> </div>
1491<div class="doc_text">
1492
1493<h5>Overview:</h5>
1494
1495<p>Opaque types are used to represent unknown types in the system. This
Gordon Henriksenda0706e2007-10-14 00:34:53 +0000[diff] [blame]1496corresponds (for example) to the C notion of a forward declared structure type.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1497In LLVM, opaque types can eventually be resolved to any type (not just a
1498structure type).</p>
1499
1500<h5>Syntax:</h5>
1501
1502<pre>
1503 opaque
1504</pre>
1505
1506<h5>Examples:</h5>
1507
1508<table class="layout">
1509 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1510 <td class="left"><tt>opaque</tt></td>
1511 <td class="left">An opaque type.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1512 </tr>
1513</table>
1514</div>
1515
1516
1517<!-- *********************************************************************** -->
1518<div class="doc_section"> <a name="constants">Constants</a> </div>
1519<!-- *********************************************************************** -->
1520
1521<div class="doc_text">
1522
1523<p>LLVM has several different basic types of constants. This section describes
1524them all and their syntax.</p>
1525
1526</div>
1527
1528<!-- ======================================================================= -->
1529<div class="doc_subsection"><a name="simpleconstants">Simple Constants</a></div>
1530
1531<div class="doc_text">
1532
1533<dl>
1534 <dt><b>Boolean constants</b></dt>
1535
1536 <dd>The two strings '<tt>true</tt>' and '<tt>false</tt>' are both valid
1537 constants of the <tt><a href="#t_primitive">i1</a></tt> type.
1538 </dd>
1539
1540 <dt><b>Integer constants</b></dt>
1541
1542 <dd>Standard integers (such as '4') are constants of the <a
1543 href="#t_integer">integer</a> type. Negative numbers may be used with
1544 integer types.
1545 </dd>
1546
1547 <dt><b>Floating point constants</b></dt>
1548
1549 <dd>Floating point constants use standard decimal notation (e.g. 123.421),
1550 exponential notation (e.g. 1.23421e+2), or a more precise hexadecimal
Chris Lattnerdc15b1d2008-04-01 18:45:27 +0000[diff] [blame]1551 notation (see below). The assembler requires the exact decimal value of
1552 a floating-point constant. For example, the assembler accepts 1.25 but
1553 rejects 1.3 because 1.3 is a repeating decimal in binary. Floating point
1554 constants must have a <a href="#t_floating">floating point</a> type. </dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1555
1556 <dt><b>Null pointer constants</b></dt>
1557
1558 <dd>The identifier '<tt>null</tt>' is recognized as a null pointer constant
1559 and must be of <a href="#t_pointer">pointer type</a>.</dd>
1560
1561</dl>
1562
1563<p>The one non-intuitive notation for constants is the optional hexadecimal form
1564of floating point constants. For example, the form '<tt>double
15650x432ff973cafa8000</tt>' is equivalent to (but harder to read than) '<tt>double
15664.5e+15</tt>'. The only time hexadecimal floating point constants are required
1567(and the only time that they are generated by the disassembler) is when a
1568floating point constant must be emitted but it cannot be represented as a
1569decimal floating point number. For example, NaN's, infinities, and other
1570special values are represented in their IEEE hexadecimal format so that
1571assembly and disassembly do not cause any bits to change in the constants.</p>
1572
1573</div>
1574
1575<!-- ======================================================================= -->
1576<div class="doc_subsection"><a name="aggregateconstants">Aggregate Constants</a>
1577</div>
1578
1579<div class="doc_text">
1580<p>Aggregate constants arise from aggregation of simple constants
1581and smaller aggregate constants.</p>
1582
1583<dl>
1584 <dt><b>Structure constants</b></dt>
1585
1586 <dd>Structure constants are represented with notation similar to structure
1587 type definitions (a comma separated list of elements, surrounded by braces
Chris Lattner6c8de962007-12-25 20:34:52 +0000[diff] [blame]1588 (<tt>{}</tt>)). For example: "<tt>{ i32 4, float 17.0, i32* @G }</tt>",
1589 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]1590 must have <a href="#t_struct">structure type</a>, and the number and
1591 types of elements must match those specified by the type.
1592 </dd>
1593
1594 <dt><b>Array constants</b></dt>
1595
1596 <dd>Array constants are represented with notation similar to array type
1597 definitions (a comma separated list of elements, surrounded by square brackets
1598 (<tt>[]</tt>)). For example: "<tt>[ i32 42, i32 11, i32 74 ]</tt>". Array
1599 constants must have <a href="#t_array">array type</a>, and the number and
1600 types of elements must match those specified by the type.
1601 </dd>
1602
1603 <dt><b>Vector constants</b></dt>
1604
1605 <dd>Vector constants are represented with notation similar to vector type
1606 definitions (a comma separated list of elements, surrounded by
1607 less-than/greater-than's (<tt>&lt;&gt;</tt>)). For example: "<tt>&lt; i32 42,
1608 i32 11, i32 74, i32 100 &gt;</tt>". Vector constants must have <a
1609 href="#t_vector">vector type</a>, and the number and types of elements must
1610 match those specified by the type.
1611 </dd>
1612
1613 <dt><b>Zero initialization</b></dt>
1614
1615 <dd>The string '<tt>zeroinitializer</tt>' can be used to zero initialize a
1616 value to zero of <em>any</em> type, including scalar and aggregate types.
1617 This is often used to avoid having to print large zero initializers (e.g. for
1618 large arrays) and is always exactly equivalent to using explicit zero
1619 initializers.
1620 </dd>
1621</dl>
1622
1623</div>
1624
1625<!-- ======================================================================= -->
1626<div class="doc_subsection">
1627 <a name="globalconstants">Global Variable and Function Addresses</a>
1628</div>
1629
1630<div class="doc_text">
1631
1632<p>The addresses of <a href="#globalvars">global variables</a> and <a
1633href="#functionstructure">functions</a> are always implicitly valid (link-time)
1634constants. These constants are explicitly referenced when the <a
1635href="#identifiers">identifier for the global</a> is used and always have <a
1636href="#t_pointer">pointer</a> type. For example, the following is a legal LLVM
1637file:</p>
1638
1639<div class="doc_code">
1640<pre>
1641@X = global i32 17
1642@Y = global i32 42
1643@Z = global [2 x i32*] [ i32* @X, i32* @Y ]
1644</pre>
1645</div>
1646
1647</div>
1648
1649<!-- ======================================================================= -->
1650<div class="doc_subsection"><a name="undefvalues">Undefined Values</a></div>
1651<div class="doc_text">
1652 <p>The string '<tt>undef</tt>' is recognized as a type-less constant that has
1653 no specific value. Undefined values may be of any type and be used anywhere
1654 a constant is permitted.</p>
1655
1656 <p>Undefined values indicate to the compiler that the program is well defined
1657 no matter what value is used, giving the compiler more freedom to optimize.
1658 </p>
1659</div>
1660
1661<!-- ======================================================================= -->
1662<div class="doc_subsection"><a name="constantexprs">Constant Expressions</a>
1663</div>
1664
1665<div class="doc_text">
1666
1667<p>Constant expressions are used to allow expressions involving other constants
1668to be used as constants. Constant expressions may be of any <a
1669href="#t_firstclass">first class</a> type and may involve any LLVM operation
1670that does not have side effects (e.g. load and call are not supported). The
1671following is the syntax for constant expressions:</p>
1672
1673<dl>
1674 <dt><b><tt>trunc ( CST to TYPE )</tt></b></dt>
1675 <dd>Truncate a constant to another type. The bit size of CST must be larger
1676 than the bit size of TYPE. Both types must be integers.</dd>
1677
1678 <dt><b><tt>zext ( CST to TYPE )</tt></b></dt>
1679 <dd>Zero extend a constant to another type. The bit size of CST must be
1680 smaller or equal to the bit size of TYPE. Both types must be integers.</dd>
1681
1682 <dt><b><tt>sext ( CST to TYPE )</tt></b></dt>
1683 <dd>Sign extend a constant to another type. The bit size of CST must be
1684 smaller or equal to the bit size of TYPE. Both types must be integers.</dd>
1685
1686 <dt><b><tt>fptrunc ( CST to TYPE )</tt></b></dt>
1687 <dd>Truncate a floating point constant to another floating point type. The
1688 size of CST must be larger than the size of TYPE. Both types must be
1689 floating point.</dd>
1690
1691 <dt><b><tt>fpext ( CST to TYPE )</tt></b></dt>
1692 <dd>Floating point extend a constant to another type. The size of CST must be
1693 smaller or equal to the size of TYPE. Both types must be floating point.</dd>
1694
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]1695 <dt><b><tt>fptoui ( CST to TYPE )</tt></b></dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1696 <dd>Convert a floating point constant to the corresponding unsigned integer
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]1697 constant. TYPE must be a scalar or vector integer type. CST must be of scalar
1698 or vector floating point type. Both CST and TYPE must be scalars, or vectors
1699 of the same number of elements. If the value won't fit in the integer type,
1700 the results are undefined.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1701
1702 <dt><b><tt>fptosi ( CST to TYPE )</tt></b></dt>
1703 <dd>Convert a floating point constant to the corresponding signed integer
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]1704 constant. TYPE must be a scalar or vector integer type. CST must be of scalar
1705 or vector floating point type. Both CST and TYPE must be scalars, or vectors
1706 of the same number of elements. If the value won't fit in the integer type,
1707 the results are undefined.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1708
1709 <dt><b><tt>uitofp ( CST to TYPE )</tt></b></dt>
1710 <dd>Convert an unsigned integer constant to the corresponding floating point
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]1711 constant. TYPE must be a scalar or vector floating point type. CST must be of
1712 scalar or vector integer type. Both CST and TYPE must be scalars, or vectors
1713 of the same number of elements. If the value won't fit in the floating point
1714 type, the results are undefined.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1715
1716 <dt><b><tt>sitofp ( CST to TYPE )</tt></b></dt>
1717 <dd>Convert a signed integer constant to the corresponding floating point
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]1718 constant. TYPE must be a scalar or vector floating point type. CST must be of
1719 scalar or vector integer type. Both CST and TYPE must be scalars, or vectors
1720 of the same number of elements. If the value won't fit in the floating point
1721 type, the results are undefined.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1722
1723 <dt><b><tt>ptrtoint ( CST to TYPE )</tt></b></dt>
1724 <dd>Convert a pointer typed constant to the corresponding integer constant
1725 TYPE must be an integer type. CST must be of pointer type. The CST value is
1726 zero extended, truncated, or unchanged to make it fit in TYPE.</dd>
1727
1728 <dt><b><tt>inttoptr ( CST to TYPE )</tt></b></dt>
1729 <dd>Convert a integer constant to a pointer constant. TYPE must be a
1730 pointer type. CST must be of integer type. The CST value is zero extended,
1731 truncated, or unchanged to make it fit in a pointer size. This one is
1732 <i>really</i> dangerous!</dd>
1733
1734 <dt><b><tt>bitcast ( CST to TYPE )</tt></b></dt>
1735 <dd>Convert a constant, CST, to another TYPE. The size of CST and TYPE must be
1736 identical (same number of bits). The conversion is done as if the CST value
1737 was stored to memory and read back as TYPE. In other words, no bits change
1738 with this operator, just the type. This can be used for conversion of
1739 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]1740 pointers it is only valid to cast to another pointer type. It is not valid
1741 to bitcast to or from an aggregate type.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1742 </dd>
1743
1744 <dt><b><tt>getelementptr ( CSTPTR, IDX0, IDX1, ... )</tt></b></dt>
1745
1746 <dd>Perform the <a href="#i_getelementptr">getelementptr operation</a> on
1747 constants. As with the <a href="#i_getelementptr">getelementptr</a>
1748 instruction, the index list may have zero or more indexes, which are required
1749 to make sense for the type of "CSTPTR".</dd>
1750
1751 <dt><b><tt>select ( COND, VAL1, VAL2 )</tt></b></dt>
1752
1753 <dd>Perform the <a href="#i_select">select operation</a> on
1754 constants.</dd>
1755
1756 <dt><b><tt>icmp COND ( VAL1, VAL2 )</tt></b></dt>
1757 <dd>Performs the <a href="#i_icmp">icmp operation</a> on constants.</dd>
1758
1759 <dt><b><tt>fcmp COND ( VAL1, VAL2 )</tt></b></dt>
1760 <dd>Performs the <a href="#i_fcmp">fcmp operation</a> on constants.</dd>
1761
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]1762 <dt><b><tt>vicmp COND ( VAL1, VAL2 )</tt></b></dt>
1763 <dd>Performs the <a href="#i_vicmp">vicmp operation</a> on constants.</dd>
1764
1765 <dt><b><tt>vfcmp COND ( VAL1, VAL2 )</tt></b></dt>
1766 <dd>Performs the <a href="#i_vfcmp">vfcmp operation</a> on constants.</dd>
1767
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1768 <dt><b><tt>extractelement ( VAL, IDX )</tt></b></dt>
1769
1770 <dd>Perform the <a href="#i_extractelement">extractelement
1771 operation</a> on constants.
1772
1773 <dt><b><tt>insertelement ( VAL, ELT, IDX )</tt></b></dt>
1774
1775 <dd>Perform the <a href="#i_insertelement">insertelement
1776 operation</a> on constants.</dd>
1777
1778
1779 <dt><b><tt>shufflevector ( VEC1, VEC2, IDXMASK )</tt></b></dt>
1780
1781 <dd>Perform the <a href="#i_shufflevector">shufflevector
1782 operation</a> on constants.</dd>
1783
1784 <dt><b><tt>OPCODE ( LHS, RHS )</tt></b></dt>
1785
1786 <dd>Perform the specified operation of the LHS and RHS constants. OPCODE may
1787 be any of the <a href="#binaryops">binary</a> or <a href="#bitwiseops">bitwise
1788 binary</a> operations. The constraints on operands are the same as those for
1789 the corresponding instruction (e.g. no bitwise operations on floating point
1790 values are allowed).</dd>
1791</dl>
1792</div>
1793
1794<!-- *********************************************************************** -->
1795<div class="doc_section"> <a name="othervalues">Other Values</a> </div>
1796<!-- *********************************************************************** -->
1797
1798<!-- ======================================================================= -->
1799<div class="doc_subsection">
1800<a name="inlineasm">Inline Assembler Expressions</a>
1801</div>
1802
1803<div class="doc_text">
1804
1805<p>
1806LLVM supports inline assembler expressions (as opposed to <a href="#moduleasm">
1807Module-Level Inline Assembly</a>) through the use of a special value. This
1808value represents the inline assembler as a string (containing the instructions
1809to emit), a list of operand constraints (stored as a string), and a flag that
1810indicates whether or not the inline asm expression has side effects. An example
1811inline assembler expression is:
1812</p>
1813
1814<div class="doc_code">
1815<pre>
1816i32 (i32) asm "bswap $0", "=r,r"
1817</pre>
1818</div>
1819
1820<p>
1821Inline assembler expressions may <b>only</b> be used as the callee operand of
1822a <a href="#i_call"><tt>call</tt> instruction</a>. Thus, typically we have:
1823</p>
1824
1825<div class="doc_code">
1826<pre>
1827%X = call i32 asm "<a href="#int_bswap">bswap</a> $0", "=r,r"(i32 %Y)
1828</pre>
1829</div>
1830
1831<p>
1832Inline asms with side effects not visible in the constraint list must be marked
1833as having side effects. This is done through the use of the
1834'<tt>sideeffect</tt>' keyword, like so:
1835</p>
1836
1837<div class="doc_code">
1838<pre>
1839call void asm sideeffect "eieio", ""()
1840</pre>
1841</div>
1842
1843<p>TODO: The format of the asm and constraints string still need to be
1844documented here. Constraints on what can be done (e.g. duplication, moving, etc
1845need to be documented).
1846</p>
1847
1848</div>
1849
1850<!-- *********************************************************************** -->
1851<div class="doc_section"> <a name="instref">Instruction Reference</a> </div>
1852<!-- *********************************************************************** -->
1853
1854<div class="doc_text">
1855
1856<p>The LLVM instruction set consists of several different
1857classifications of instructions: <a href="#terminators">terminator
1858instructions</a>, <a href="#binaryops">binary instructions</a>,
1859<a href="#bitwiseops">bitwise binary instructions</a>, <a
1860 href="#memoryops">memory instructions</a>, and <a href="#otherops">other
1861instructions</a>.</p>
1862
1863</div>
1864
1865<!-- ======================================================================= -->
1866<div class="doc_subsection"> <a name="terminators">Terminator
1867Instructions</a> </div>
1868
1869<div class="doc_text">
1870
1871<p>As mentioned <a href="#functionstructure">previously</a>, every
1872basic block in a program ends with a "Terminator" instruction, which
1873indicates which block should be executed after the current block is
1874finished. These terminator instructions typically yield a '<tt>void</tt>'
1875value: they produce control flow, not values (the one exception being
1876the '<a href="#i_invoke"><tt>invoke</tt></a>' instruction).</p>
1877<p>There are six different terminator instructions: the '<a
1878 href="#i_ret"><tt>ret</tt></a>' instruction, the '<a href="#i_br"><tt>br</tt></a>'
1879instruction, the '<a href="#i_switch"><tt>switch</tt></a>' instruction,
1880the '<a href="#i_invoke"><tt>invoke</tt></a>' instruction, the '<a
1881 href="#i_unwind"><tt>unwind</tt></a>' instruction, and the '<a
1882 href="#i_unreachable"><tt>unreachable</tt></a>' instruction.</p>
1883
1884</div>
1885
1886<!-- _______________________________________________________________________ -->
1887<div class="doc_subsubsection"> <a name="i_ret">'<tt>ret</tt>'
1888Instruction</a> </div>
1889<div class="doc_text">
1890<h5>Syntax:</h5>
1891<pre> ret &lt;type&gt; &lt;value&gt; <i>; Return a value from a non-void function</i>
1892 ret void <i>; Return from void function</i>
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]1893 ret &lt;type&gt; &lt;value&gt;, &lt;type&gt; &lt;value&gt; <i>; Return two values from a non-void function </i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1894</pre>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1895
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1896<h5>Overview:</h5>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1897
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1898<p>The '<tt>ret</tt>' instruction is used to return control flow (and a
1899value) from a function back to the caller.</p>
1900<p>There are two forms of the '<tt>ret</tt>' instruction: one that
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1901returns value(s) and then causes control flow, and one that just causes
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1902control flow to occur.</p>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1903
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1904<h5>Arguments:</h5>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1905
1906<p>The '<tt>ret</tt>' instruction may return zero, one or multiple values.
1907The type of each return value must be a '<a href="#t_firstclass">first
1908class</a>' type. Note that a function is not <a href="#wellformed">well
1909formed</a> if there exists a '<tt>ret</tt>' instruction inside of the
1910function that returns values that do not match the return type of the
1911function.</p>
1912
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1913<h5>Semantics:</h5>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1914
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1915<p>When the '<tt>ret</tt>' instruction is executed, control flow
1916returns back to the calling function's context. If the caller is a "<a
1917 href="#i_call"><tt>call</tt></a>" instruction, execution continues at
1918the instruction after the call. If the caller was an "<a
1919 href="#i_invoke"><tt>invoke</tt></a>" instruction, execution continues
1920at the beginning of the "normal" destination block. If the instruction
1921returns a value, that value shall set the call or invoke instruction's
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]1922return value. If the instruction returns multiple values then these
Devang Patelec8a5b02008-03-11 05:51:59 +0000[diff] [blame]1923values can only be accessed through a '<a href="#i_getresult"><tt>getresult</tt>
1924</a>' instruction.</p>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1925
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1926<h5>Example:</h5>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1927
1928<pre>
1929 ret i32 5 <i>; Return an integer value of 5</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1930 ret void <i>; Return from a void function</i>
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]1931 ret i32 4, i8 2 <i>; Return two values 4 and 2 </i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1932</pre>
1933</div>
1934<!-- _______________________________________________________________________ -->
1935<div class="doc_subsubsection"> <a name="i_br">'<tt>br</tt>' Instruction</a> </div>
1936<div class="doc_text">
1937<h5>Syntax:</h5>
1938<pre> br i1 &lt;cond&gt;, label &lt;iftrue&gt;, label &lt;iffalse&gt;<br> br label &lt;dest&gt; <i>; Unconditional branch</i>
1939</pre>
1940<h5>Overview:</h5>
1941<p>The '<tt>br</tt>' instruction is used to cause control flow to
1942transfer to a different basic block in the current function. There are
1943two forms of this instruction, corresponding to a conditional branch
1944and an unconditional branch.</p>
1945<h5>Arguments:</h5>
1946<p>The conditional branch form of the '<tt>br</tt>' instruction takes a
1947single '<tt>i1</tt>' value and two '<tt>label</tt>' values. The
1948unconditional form of the '<tt>br</tt>' instruction takes a single
1949'<tt>label</tt>' value as a target.</p>
1950<h5>Semantics:</h5>
1951<p>Upon execution of a conditional '<tt>br</tt>' instruction, the '<tt>i1</tt>'
1952argument is evaluated. If the value is <tt>true</tt>, control flows
1953to the '<tt>iftrue</tt>' <tt>label</tt> argument. If "cond" is <tt>false</tt>,
1954control flows to the '<tt>iffalse</tt>' <tt>label</tt> argument.</p>
1955<h5>Example:</h5>
1956<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
1957 href="#i_ret">ret</a> i32 1<br>IfUnequal:<br> <a href="#i_ret">ret</a> i32 0<br></pre>
1958</div>
1959<!-- _______________________________________________________________________ -->
1960<div class="doc_subsubsection">
1961 <a name="i_switch">'<tt>switch</tt>' Instruction</a>
1962</div>
1963
1964<div class="doc_text">
1965<h5>Syntax:</h5>
1966
1967<pre>
1968 switch &lt;intty&gt; &lt;value&gt;, label &lt;defaultdest&gt; [ &lt;intty&gt; &lt;val&gt;, label &lt;dest&gt; ... ]
1969</pre>
1970
1971<h5>Overview:</h5>
1972
1973<p>The '<tt>switch</tt>' instruction is used to transfer control flow to one of
1974several different places. It is a generalization of the '<tt>br</tt>'
1975instruction, allowing a branch to occur to one of many possible
1976destinations.</p>
1977
1978
1979<h5>Arguments:</h5>
1980
1981<p>The '<tt>switch</tt>' instruction uses three parameters: an integer
1982comparison value '<tt>value</tt>', a default '<tt>label</tt>' destination, and
1983an array of pairs of comparison value constants and '<tt>label</tt>'s. The
1984table is not allowed to contain duplicate constant entries.</p>
1985
1986<h5>Semantics:</h5>
1987
1988<p>The <tt>switch</tt> instruction specifies a table of values and
1989destinations. When the '<tt>switch</tt>' instruction is executed, this
1990table is searched for the given value. If the value is found, control flow is
1991transfered to the corresponding destination; otherwise, control flow is
1992transfered to the default destination.</p>
1993
1994<h5>Implementation:</h5>
1995
1996<p>Depending on properties of the target machine and the particular
1997<tt>switch</tt> instruction, this instruction may be code generated in different
1998ways. For example, it could be generated as a series of chained conditional
1999branches or with a lookup table.</p>
2000
2001<h5>Example:</h5>
2002
2003<pre>
2004 <i>; Emulate a conditional br instruction</i>
2005 %Val = <a href="#i_zext">zext</a> i1 %value to i32
2006 switch i32 %Val, label %truedest [i32 0, label %falsedest ]
2007
2008 <i>; Emulate an unconditional br instruction</i>
2009 switch i32 0, label %dest [ ]
2010
2011 <i>; Implement a jump table:</i>
2012 switch i32 %val, label %otherwise [ i32 0, label %onzero
2013 i32 1, label %onone
2014 i32 2, label %ontwo ]
2015</pre>
2016</div>
2017
2018<!-- _______________________________________________________________________ -->
2019<div class="doc_subsubsection">
2020 <a name="i_invoke">'<tt>invoke</tt>' Instruction</a>
2021</div>
2022
2023<div class="doc_text">
2024
2025<h5>Syntax:</h5>
2026
2027<pre>
Nick Lewyckya1c11a12008-03-16 07:18:12 +0000[diff] [blame]2028 &lt;result&gt; = invoke [<a href="#callingconv">cconv</a>] &lt;ptr to function ty&gt; &lt;function ptr val&gt;(&lt;function args&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2029 to label &lt;normal label&gt; unwind label &lt;exception label&gt;
2030</pre>
2031
2032<h5>Overview:</h5>
2033
2034<p>The '<tt>invoke</tt>' instruction causes control to transfer to a specified
2035function, with the possibility of control flow transfer to either the
2036'<tt>normal</tt>' label or the
2037'<tt>exception</tt>' label. If the callee function returns with the
2038"<tt><a href="#i_ret">ret</a></tt>" instruction, control flow will return to the
2039"normal" label. If the callee (or any indirect callees) returns with the "<a
2040href="#i_unwind"><tt>unwind</tt></a>" instruction, control is interrupted and
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]2041continued at the dynamically nearest "exception" label. If the callee function
Devang Patelec8a5b02008-03-11 05:51:59 +0000[diff] [blame]2042returns multiple values then individual return values are only accessible through
2043a '<tt><a href="#i_getresult">getresult</a></tt>' instruction.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2044
2045<h5>Arguments:</h5>
2046
2047<p>This instruction requires several arguments:</p>
2048
2049<ol>
2050 <li>
2051 The optional "cconv" marker indicates which <a href="#callingconv">calling
2052 convention</a> the call should use. If none is specified, the call defaults
2053 to using C calling conventions.
2054 </li>
2055 <li>'<tt>ptr to function ty</tt>': shall be the signature of the pointer to
2056 function value being invoked. In most cases, this is a direct function
2057 invocation, but indirect <tt>invoke</tt>s are just as possible, branching off
2058 an arbitrary pointer to function value.
2059 </li>
2060
2061 <li>'<tt>function ptr val</tt>': An LLVM value containing a pointer to a
2062 function to be invoked. </li>
2063
2064 <li>'<tt>function args</tt>': argument list whose types match the function
2065 signature argument types. If the function signature indicates the function
2066 accepts a variable number of arguments, the extra arguments can be
2067 specified. </li>
2068
2069 <li>'<tt>normal label</tt>': the label reached when the called function
2070 executes a '<tt><a href="#i_ret">ret</a></tt>' instruction. </li>
2071
2072 <li>'<tt>exception label</tt>': the label reached when a callee returns with
2073 the <a href="#i_unwind"><tt>unwind</tt></a> instruction. </li>
2074
2075</ol>
2076
2077<h5>Semantics:</h5>
2078
2079<p>This instruction is designed to operate as a standard '<tt><a
2080href="#i_call">call</a></tt>' instruction in most regards. The primary
2081difference is that it establishes an association with a label, which is used by
2082the runtime library to unwind the stack.</p>
2083
2084<p>This instruction is used in languages with destructors to ensure that proper
2085cleanup is performed in the case of either a <tt>longjmp</tt> or a thrown
2086exception. Additionally, this is important for implementation of
2087'<tt>catch</tt>' clauses in high-level languages that support them.</p>
2088
2089<h5>Example:</h5>
2090<pre>
Nick Lewyckya1c11a12008-03-16 07:18:12 +0000[diff] [blame]2091 %retval = invoke i32 @Test(i32 15) to label %Continue
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2092 unwind label %TestCleanup <i>; {i32}:retval set</i>
Nick Lewyckya1c11a12008-03-16 07:18:12 +0000[diff] [blame]2093 %retval = invoke <a href="#callingconv">coldcc</a> i32 %Testfnptr(i32 15) to label %Continue
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2094 unwind label %TestCleanup <i>; {i32}:retval set</i>
2095</pre>
2096</div>
2097
2098
2099<!-- _______________________________________________________________________ -->
2100
2101<div class="doc_subsubsection"> <a name="i_unwind">'<tt>unwind</tt>'
2102Instruction</a> </div>
2103
2104<div class="doc_text">
2105
2106<h5>Syntax:</h5>
2107<pre>
2108 unwind
2109</pre>
2110
2111<h5>Overview:</h5>
2112
2113<p>The '<tt>unwind</tt>' instruction unwinds the stack, continuing control flow
2114at the first callee in the dynamic call stack which used an <a
2115href="#i_invoke"><tt>invoke</tt></a> instruction to perform the call. This is
2116primarily used to implement exception handling.</p>
2117
2118<h5>Semantics:</h5>
2119
Chris Lattner8b094fc2008-04-19 21:01:16 +0000[diff] [blame]2120<p>The '<tt>unwind</tt>' instruction causes execution of the current function to
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2121immediately halt. The dynamic call stack is then searched for the first <a
2122href="#i_invoke"><tt>invoke</tt></a> instruction on the call stack. Once found,
2123execution continues at the "exceptional" destination block specified by the
2124<tt>invoke</tt> instruction. If there is no <tt>invoke</tt> instruction in the
2125dynamic call chain, undefined behavior results.</p>
2126</div>
2127
2128<!-- _______________________________________________________________________ -->
2129
2130<div class="doc_subsubsection"> <a name="i_unreachable">'<tt>unreachable</tt>'
2131Instruction</a> </div>
2132
2133<div class="doc_text">
2134
2135<h5>Syntax:</h5>
2136<pre>
2137 unreachable
2138</pre>
2139
2140<h5>Overview:</h5>
2141
2142<p>The '<tt>unreachable</tt>' instruction has no defined semantics. This
2143instruction is used to inform the optimizer that a particular portion of the
2144code is not reachable. This can be used to indicate that the code after a
2145no-return function cannot be reached, and other facts.</p>
2146
2147<h5>Semantics:</h5>
2148
2149<p>The '<tt>unreachable</tt>' instruction has no defined semantics.</p>
2150</div>
2151
2152
2153
2154<!-- ======================================================================= -->
2155<div class="doc_subsection"> <a name="binaryops">Binary Operations</a> </div>
2156<div class="doc_text">
2157<p>Binary operators are used to do most of the computation in a
Chris Lattnerab596d92008-04-01 18:47:32 +0000[diff] [blame]2158program. They require two operands of the same type, execute an operation on them, and
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2159produce a single value. The operands might represent
2160multiple 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]2161The result value has the same type as its operands.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2162<p>There are several different binary operators:</p>
2163</div>
2164<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2165<div class="doc_subsubsection">
2166 <a name="i_add">'<tt>add</tt>' Instruction</a>
2167</div>
2168
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2169<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2170
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2171<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2172
2173<pre>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2174 &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]2175</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2176
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2177<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2178
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2179<p>The '<tt>add</tt>' instruction returns the sum of its two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2180
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2181<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2182
2183<p>The two arguments to the '<tt>add</tt>' instruction must be <a
2184 href="#t_integer">integer</a>, <a href="#t_floating">floating point</a>, or
2185 <a href="#t_vector">vector</a> values. Both arguments must have identical
2186 types.</p>
2187
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2188<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2189
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2190<p>The value produced is the integer or floating point sum of the two
2191operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2192
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2193<p>If an integer sum has unsigned overflow, the result returned is the
2194mathematical result modulo 2<sup>n</sup>, where n is the bit width of
2195the result.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2196
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2197<p>Because LLVM integers use a two's complement representation, this
2198instruction is appropriate for both signed and unsigned integers.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2199
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2200<h5>Example:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2201
2202<pre>
2203 &lt;result&gt; = add i32 4, %var <i>; yields {i32}:result = 4 + %var</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2204</pre>
2205</div>
2206<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2207<div class="doc_subsubsection">
2208 <a name="i_sub">'<tt>sub</tt>' Instruction</a>
2209</div>
2210
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2211<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2212
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2213<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2214
2215<pre>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2216 &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]2217</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2218
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2219<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2220
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2221<p>The '<tt>sub</tt>' instruction returns the difference of its two
2222operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2223
2224<p>Note that the '<tt>sub</tt>' instruction is used to represent the
2225'<tt>neg</tt>' instruction present in most other intermediate
2226representations.</p>
2227
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2228<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2229
2230<p>The two arguments to the '<tt>sub</tt>' instruction must be <a
2231 href="#t_integer">integer</a>, <a href="#t_floating">floating point</a>,
2232 or <a href="#t_vector">vector</a> values. Both arguments must have identical
2233 types.</p>
2234
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2235<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2236
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2237<p>The value produced is the integer or floating point difference of
2238the two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2239
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2240<p>If an integer difference has unsigned overflow, the result returned is the
2241mathematical result modulo 2<sup>n</sup>, where n is the bit width of
2242the result.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2243
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2244<p>Because LLVM integers use a two's complement representation, this
2245instruction is appropriate for both signed and unsigned integers.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2246
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2247<h5>Example:</h5>
2248<pre>
2249 &lt;result&gt; = sub i32 4, %var <i>; yields {i32}:result = 4 - %var</i>
2250 &lt;result&gt; = sub i32 0, %val <i>; yields {i32}:result = -%var</i>
2251</pre>
2252</div>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2253
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2254<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2255<div class="doc_subsubsection">
2256 <a name="i_mul">'<tt>mul</tt>' Instruction</a>
2257</div>
2258
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2259<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2260
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2261<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2262<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]2263</pre>
2264<h5>Overview:</h5>
2265<p>The '<tt>mul</tt>' instruction returns the product of its two
2266operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]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>mul</tt>' instruction must be <a
2271href="#t_integer">integer</a>, <a href="#t_floating">floating point</a>,
2272or <a href="#t_vector">vector</a> values. Both arguments must have identical
2273types.</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 product of the
2278two 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 the result of an integer multiplication has unsigned overflow,
2281the result returned is the mathematical result modulo
22822<sup>n</sup>, where n is the bit width of the result.</p>
2283<p>Because LLVM integers use a two's complement representation, and the
2284result is the same width as the operands, this instruction returns the
2285correct result for both signed and unsigned integers. If a full product
2286(e.g. <tt>i32</tt>x<tt>i32</tt>-><tt>i64</tt>) is needed, the operands
2287should be sign-extended or zero-extended as appropriate to the
2288width of the full product.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2289<h5>Example:</h5>
2290<pre> &lt;result&gt; = mul i32 4, %var <i>; yields {i32}:result = 4 * %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<!-- _______________________________________________________________________ -->
2295<div class="doc_subsubsection"> <a name="i_udiv">'<tt>udiv</tt>' Instruction
2296</a></div>
2297<div class="doc_text">
2298<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2299<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]2300</pre>
2301<h5>Overview:</h5>
2302<p>The '<tt>udiv</tt>' instruction returns the quotient of its two
2303operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2304
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2305<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2306
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2307<p>The two arguments to the '<tt>udiv</tt>' instruction must be
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2308<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2309values. Both arguments must have identical types.</p>
2310
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2311<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2312
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2313<p>The value produced is the unsigned integer quotient of the two operands.</p>
2314<p>Note that unsigned integer division and signed integer division are distinct
2315operations; for signed integer division, use '<tt>sdiv</tt>'.</p>
2316<p>Division by zero leads to undefined behavior.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2317<h5>Example:</h5>
2318<pre> &lt;result&gt; = udiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
2319</pre>
2320</div>
2321<!-- _______________________________________________________________________ -->
2322<div class="doc_subsubsection"> <a name="i_sdiv">'<tt>sdiv</tt>' Instruction
2323</a> </div>
2324<div class="doc_text">
2325<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2326<pre>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2327 &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]2328</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2329
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2330<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2331
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2332<p>The '<tt>sdiv</tt>' instruction returns the quotient of its two
2333operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2334
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2335<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2336
2337<p>The two arguments to the '<tt>sdiv</tt>' instruction must be
2338<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2339values. Both arguments must have identical types.</p>
2340
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2341<h5>Semantics:</h5>
Chris Lattnerdc15b1d2008-04-01 18:45:27 +0000[diff] [blame]2342<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]2343<p>Note that signed integer division and unsigned integer division are distinct
2344operations; for unsigned integer division, use '<tt>udiv</tt>'.</p>
2345<p>Division by zero leads to undefined behavior. Overflow also leads to
2346undefined behavior; this is a rare case, but can occur, for example,
2347by doing a 32-bit division of -2147483648 by -1.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2348<h5>Example:</h5>
2349<pre> &lt;result&gt; = sdiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
2350</pre>
2351</div>
2352<!-- _______________________________________________________________________ -->
2353<div class="doc_subsubsection"> <a name="i_fdiv">'<tt>fdiv</tt>'
2354Instruction</a> </div>
2355<div class="doc_text">
2356<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2357<pre>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2358 &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]2359</pre>
2360<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2361
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2362<p>The '<tt>fdiv</tt>' instruction returns the quotient of its two
2363operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2364
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2365<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2366
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2367<p>The two arguments to the '<tt>fdiv</tt>' instruction must be
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2368<a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a>
2369of floating point values. Both arguments must have identical types.</p>
2370
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2371<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2372
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2373<p>The value produced is the floating point quotient of the two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2374
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2375<h5>Example:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2376
2377<pre>
2378 &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]2379</pre>
2380</div>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2381
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2382<!-- _______________________________________________________________________ -->
2383<div class="doc_subsubsection"> <a name="i_urem">'<tt>urem</tt>' Instruction</a>
2384</div>
2385<div class="doc_text">
2386<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2387<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]2388</pre>
2389<h5>Overview:</h5>
2390<p>The '<tt>urem</tt>' instruction returns the remainder from the
2391unsigned division of its two arguments.</p>
2392<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2393<p>The two arguments to the '<tt>urem</tt>' instruction must be
2394<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2395values. Both arguments must have identical types.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2396<h5>Semantics:</h5>
2397<p>This instruction returns the unsigned integer <i>remainder</i> of a division.
Chris Lattnerdc15b1d2008-04-01 18:45:27 +0000[diff] [blame]2398This instruction always performs an unsigned division to get the remainder.</p>
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2399<p>Note that unsigned integer remainder and signed integer remainder are
2400distinct operations; for signed integer remainder, use '<tt>srem</tt>'.</p>
2401<p>Taking the remainder of a division by zero leads to undefined behavior.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2402<h5>Example:</h5>
2403<pre> &lt;result&gt; = urem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
2404</pre>
2405
2406</div>
2407<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2408<div class="doc_subsubsection">
2409 <a name="i_srem">'<tt>srem</tt>' Instruction</a>
2410</div>
2411
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2412<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2413
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2414<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2415
2416<pre>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2417 &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]2418</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2419
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2420<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2421
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2422<p>The '<tt>srem</tt>' instruction returns the remainder from the
Dan Gohman3e3fd8c2007-11-05 23:35:22 +0000[diff] [blame]2423signed division of its two operands. This instruction can also take
2424<a href="#t_vector">vector</a> versions of the values in which case
2425the elements must be integers.</p>
Chris Lattner08497ce2008-01-04 04:33:49 +0000[diff] [blame]2426
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2427<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2428
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2429<p>The two arguments to the '<tt>srem</tt>' instruction must be
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2430<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2431values. Both arguments must have identical types.</p>
2432
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2433<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2434
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2435<p>This instruction returns the <i>remainder</i> of a division (where the result
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2436has the same sign as the dividend, <tt>op1</tt>), not the <i>modulo</i>
2437operator (where the result has the same sign as the divisor, <tt>op2</tt>) of
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2438a value. For more information about the difference, see <a
2439 href="http://mathforum.org/dr.math/problems/anne.4.28.99.html">The
2440Math Forum</a>. For a table of how this is implemented in various languages,
2441please see <a href="http://en.wikipedia.org/wiki/Modulo_operation">
2442Wikipedia: modulo operation</a>.</p>
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2443<p>Note that signed integer remainder and unsigned integer remainder are
2444distinct operations; for unsigned integer remainder, use '<tt>urem</tt>'.</p>
2445<p>Taking the remainder of a division by zero leads to undefined behavior.
2446Overflow also leads to undefined behavior; this is a rare case, but can occur,
2447for example, by taking the remainder of a 32-bit division of -2147483648 by -1.
2448(The remainder doesn't actually overflow, but this rule lets srem be
2449implemented using instructions that return both the result of the division
2450and the remainder.)</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2451<h5>Example:</h5>
2452<pre> &lt;result&gt; = srem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
2453</pre>
2454
2455</div>
2456<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2457<div class="doc_subsubsection">
2458 <a name="i_frem">'<tt>frem</tt>' Instruction</a> </div>
2459
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2460<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2461
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2462<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2463<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]2464</pre>
2465<h5>Overview:</h5>
2466<p>The '<tt>frem</tt>' instruction returns the remainder from the
2467division of its two operands.</p>
2468<h5>Arguments:</h5>
2469<p>The two arguments to the '<tt>frem</tt>' instruction must be
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2470<a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a>
2471of floating point values. 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
Chris Lattnerdc15b1d2008-04-01 18:45:27 +0000[diff] [blame]2475<p>This instruction returns the <i>remainder</i> of a division.
2476The remainder has the same sign as the dividend.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2477
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2478<h5>Example:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2479
2480<pre>
2481 &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]2482</pre>
2483</div>
2484
2485<!-- ======================================================================= -->
2486<div class="doc_subsection"> <a name="bitwiseops">Bitwise Binary
2487Operations</a> </div>
2488<div class="doc_text">
2489<p>Bitwise binary operators are used to do various forms of
2490bit-twiddling in a program. They are generally very efficient
2491instructions and can commonly be strength reduced from other
Chris Lattnerdc15b1d2008-04-01 18:45:27 +0000[diff] [blame]2492instructions. They require two operands of the same type, execute an operation on them,
2493and 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]2494</div>
2495
2496<!-- _______________________________________________________________________ -->
2497<div class="doc_subsubsection"> <a name="i_shl">'<tt>shl</tt>'
2498Instruction</a> </div>
2499<div class="doc_text">
2500<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2501<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]2502</pre>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2503
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2504<h5>Overview:</h5>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2505
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2506<p>The '<tt>shl</tt>' instruction returns the first operand shifted to
2507the left a specified number of bits.</p>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2508
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2509<h5>Arguments:</h5>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2510
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2511<p>Both arguments to the '<tt>shl</tt>' instruction must be the same <a
Nate Begemanbb1ce942008-07-29 15:49:41 +0000[diff] [blame]2512 href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2513type. '<tt>op2</tt>' is treated as an unsigned value.</p>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2514
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2515<h5>Semantics:</h5>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2516
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2517<p>The value produced is <tt>op1</tt> * 2<sup><tt>op2</tt></sup> mod 2<sup>n</sup>,
2518where n is the width of the result. If <tt>op2</tt> is (statically or dynamically) negative or
2519equal 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]2520
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2521<h5>Example:</h5><pre>
2522 &lt;result&gt; = shl i32 4, %var <i>; yields {i32}: 4 &lt;&lt; %var</i>
2523 &lt;result&gt; = shl i32 4, 2 <i>; yields {i32}: 16</i>
2524 &lt;result&gt; = shl i32 1, 10 <i>; yields {i32}: 1024</i>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2525 &lt;result&gt; = shl i32 1, 32 <i>; undefined</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2526</pre>
2527</div>
2528<!-- _______________________________________________________________________ -->
2529<div class="doc_subsubsection"> <a name="i_lshr">'<tt>lshr</tt>'
2530Instruction</a> </div>
2531<div class="doc_text">
2532<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2533<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]2534</pre>
2535
2536<h5>Overview:</h5>
2537<p>The '<tt>lshr</tt>' instruction (logical shift right) returns the first
2538operand shifted to the right a specified number of bits with zero fill.</p>
2539
2540<h5>Arguments:</h5>
2541<p>Both arguments to the '<tt>lshr</tt>' instruction must be the same
Nate Begemanbb1ce942008-07-29 15:49:41 +0000[diff] [blame]2542<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]2543type. '<tt>op2</tt>' is treated as an unsigned value.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2544
2545<h5>Semantics:</h5>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2546
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2547<p>This instruction always performs a logical shift right operation. The most
2548significant bits of the result will be filled with zero bits after the
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2549shift. If <tt>op2</tt> is (statically or dynamically) equal to or larger than
2550the number of bits in <tt>op1</tt>, the result is undefined.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2551
2552<h5>Example:</h5>
2553<pre>
2554 &lt;result&gt; = lshr i32 4, 1 <i>; yields {i32}:result = 2</i>
2555 &lt;result&gt; = lshr i32 4, 2 <i>; yields {i32}:result = 1</i>
2556 &lt;result&gt; = lshr i8 4, 3 <i>; yields {i8}:result = 0</i>
2557 &lt;result&gt; = lshr i8 -2, 1 <i>; yields {i8}:result = 0x7FFFFFFF </i>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2558 &lt;result&gt; = lshr i32 1, 32 <i>; undefined</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2559</pre>
2560</div>
2561
2562<!-- _______________________________________________________________________ -->
2563<div class="doc_subsubsection"> <a name="i_ashr">'<tt>ashr</tt>'
2564Instruction</a> </div>
2565<div class="doc_text">
2566
2567<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2568<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]2569</pre>
2570
2571<h5>Overview:</h5>
2572<p>The '<tt>ashr</tt>' instruction (arithmetic shift right) returns the first
2573operand shifted to the right a specified number of bits with sign extension.</p>
2574
2575<h5>Arguments:</h5>
2576<p>Both arguments to the '<tt>ashr</tt>' instruction must be the same
Nate Begemanbb1ce942008-07-29 15:49:41 +0000[diff] [blame]2577<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]2578type. '<tt>op2</tt>' is treated as an unsigned value.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2579
2580<h5>Semantics:</h5>
2581<p>This instruction always performs an arithmetic shift right operation,
2582The most significant bits of the result will be filled with the sign bit
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2583of <tt>op1</tt>. If <tt>op2</tt> is (statically or dynamically) equal to or
2584larger than the number of bits in <tt>op1</tt>, the result is undefined.
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2585</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2586
2587<h5>Example:</h5>
2588<pre>
2589 &lt;result&gt; = ashr i32 4, 1 <i>; yields {i32}:result = 2</i>
2590 &lt;result&gt; = ashr i32 4, 2 <i>; yields {i32}:result = 1</i>
2591 &lt;result&gt; = ashr i8 4, 3 <i>; yields {i8}:result = 0</i>
2592 &lt;result&gt; = ashr i8 -2, 1 <i>; yields {i8}:result = -1</i>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2593 &lt;result&gt; = ashr i32 1, 32 <i>; undefined</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2594</pre>
2595</div>
2596
2597<!-- _______________________________________________________________________ -->
2598<div class="doc_subsubsection"> <a name="i_and">'<tt>and</tt>'
2599Instruction</a> </div>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2600
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2601<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2602
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2603<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2604
2605<pre>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2606 &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]2607</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2608
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2609<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2610
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2611<p>The '<tt>and</tt>' instruction returns the bitwise logical and of
2612its two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2613
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2614<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2615
2616<p>The two arguments to the '<tt>and</tt>' instruction must be
2617<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2618values. Both arguments must have identical types.</p>
2619
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2620<h5>Semantics:</h5>
2621<p>The truth table used for the '<tt>and</tt>' instruction is:</p>
2622<p> </p>
Bill Wendling1e19eac2008-09-07 10:29:20 +0000[diff] [blame]2623<div>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2624<table border="1" cellspacing="0" cellpadding="4">
2625 <tbody>
2626 <tr>
2627 <td>In0</td>
2628 <td>In1</td>
2629 <td>Out</td>
2630 </tr>
2631 <tr>
2632 <td>0</td>
2633 <td>0</td>
2634 <td>0</td>
2635 </tr>
2636 <tr>
2637 <td>0</td>
2638 <td>1</td>
2639 <td>0</td>
2640 </tr>
2641 <tr>
2642 <td>1</td>
2643 <td>0</td>
2644 <td>0</td>
2645 </tr>
2646 <tr>
2647 <td>1</td>
2648 <td>1</td>
2649 <td>1</td>
2650 </tr>
2651 </tbody>
2652</table>
2653</div>
2654<h5>Example:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2655<pre>
2656 &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]2657 &lt;result&gt; = and i32 15, 40 <i>; yields {i32}:result = 8</i>
2658 &lt;result&gt; = and i32 4, 8 <i>; yields {i32}:result = 0</i>
2659</pre>
2660</div>
2661<!-- _______________________________________________________________________ -->
2662<div class="doc_subsubsection"> <a name="i_or">'<tt>or</tt>' Instruction</a> </div>
2663<div class="doc_text">
2664<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2665<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]2666</pre>
2667<h5>Overview:</h5>
2668<p>The '<tt>or</tt>' instruction returns the bitwise logical inclusive
2669or of its two operands.</p>
2670<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2671
2672<p>The two arguments to the '<tt>or</tt>' instruction must be
2673<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2674values. Both arguments must have identical types.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2675<h5>Semantics:</h5>
2676<p>The truth table used for the '<tt>or</tt>' instruction is:</p>
2677<p> </p>
Bill Wendling1e19eac2008-09-07 10:29:20 +0000[diff] [blame]2678<div>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2679<table border="1" cellspacing="0" cellpadding="4">
2680 <tbody>
2681 <tr>
2682 <td>In0</td>
2683 <td>In1</td>
2684 <td>Out</td>
2685 </tr>
2686 <tr>
2687 <td>0</td>
2688 <td>0</td>
2689 <td>0</td>
2690 </tr>
2691 <tr>
2692 <td>0</td>
2693 <td>1</td>
2694 <td>1</td>
2695 </tr>
2696 <tr>
2697 <td>1</td>
2698 <td>0</td>
2699 <td>1</td>
2700 </tr>
2701 <tr>
2702 <td>1</td>
2703 <td>1</td>
2704 <td>1</td>
2705 </tr>
2706 </tbody>
2707</table>
2708</div>
2709<h5>Example:</h5>
2710<pre> &lt;result&gt; = or i32 4, %var <i>; yields {i32}:result = 4 | %var</i>
2711 &lt;result&gt; = or i32 15, 40 <i>; yields {i32}:result = 47</i>
2712 &lt;result&gt; = or i32 4, 8 <i>; yields {i32}:result = 12</i>
2713</pre>
2714</div>
2715<!-- _______________________________________________________________________ -->
2716<div class="doc_subsubsection"> <a name="i_xor">'<tt>xor</tt>'
2717Instruction</a> </div>
2718<div class="doc_text">
2719<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2720<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]2721</pre>
2722<h5>Overview:</h5>
2723<p>The '<tt>xor</tt>' instruction returns the bitwise logical exclusive
2724or of its two operands. The <tt>xor</tt> is used to implement the
2725"one's complement" operation, which is the "~" operator in C.</p>
2726<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2727<p>The two arguments to the '<tt>xor</tt>' instruction must be
2728<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2729values. Both arguments must have identical types.</p>
2730
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2731<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2732
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2733<p>The truth table used for the '<tt>xor</tt>' instruction is:</p>
2734<p> </p>
Bill Wendling1e19eac2008-09-07 10:29:20 +0000[diff] [blame]2735<div>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2736<table border="1" cellspacing="0" cellpadding="4">
2737 <tbody>
2738 <tr>
2739 <td>In0</td>
2740 <td>In1</td>
2741 <td>Out</td>
2742 </tr>
2743 <tr>
2744 <td>0</td>
2745 <td>0</td>
2746 <td>0</td>
2747 </tr>
2748 <tr>
2749 <td>0</td>
2750 <td>1</td>
2751 <td>1</td>
2752 </tr>
2753 <tr>
2754 <td>1</td>
2755 <td>0</td>
2756 <td>1</td>
2757 </tr>
2758 <tr>
2759 <td>1</td>
2760 <td>1</td>
2761 <td>0</td>
2762 </tr>
2763 </tbody>
2764</table>
2765</div>
2766<p> </p>
2767<h5>Example:</h5>
2768<pre> &lt;result&gt; = xor i32 4, %var <i>; yields {i32}:result = 4 ^ %var</i>
2769 &lt;result&gt; = xor i32 15, 40 <i>; yields {i32}:result = 39</i>
2770 &lt;result&gt; = xor i32 4, 8 <i>; yields {i32}:result = 12</i>
2771 &lt;result&gt; = xor i32 %V, -1 <i>; yields {i32}:result = ~%V</i>
2772</pre>
2773</div>
2774
2775<!-- ======================================================================= -->
2776<div class="doc_subsection">
2777 <a name="vectorops">Vector Operations</a>
2778</div>
2779
2780<div class="doc_text">
2781
2782<p>LLVM supports several instructions to represent vector operations in a
2783target-independent manner. These instructions cover the element-access and
2784vector-specific operations needed to process vectors effectively. While LLVM
2785does directly support these vector operations, many sophisticated algorithms
2786will want to use target-specific intrinsics to take full advantage of a specific
2787target.</p>
2788
2789</div>
2790
2791<!-- _______________________________________________________________________ -->
2792<div class="doc_subsubsection">
2793 <a name="i_extractelement">'<tt>extractelement</tt>' Instruction</a>
2794</div>
2795
2796<div class="doc_text">
2797
2798<h5>Syntax:</h5>
2799
2800<pre>
2801 &lt;result&gt; = extractelement &lt;n x &lt;ty&gt;&gt; &lt;val&gt;, i32 &lt;idx&gt; <i>; yields &lt;ty&gt;</i>
2802</pre>
2803
2804<h5>Overview:</h5>
2805
2806<p>
2807The '<tt>extractelement</tt>' instruction extracts a single scalar
2808element from a vector at a specified index.
2809</p>
2810
2811
2812<h5>Arguments:</h5>
2813
2814<p>
2815The first operand of an '<tt>extractelement</tt>' instruction is a
2816value of <a href="#t_vector">vector</a> type. The second operand is
2817an index indicating the position from which to extract the element.
2818The index may be a variable.</p>
2819
2820<h5>Semantics:</h5>
2821
2822<p>
2823The result is a scalar of the same type as the element type of
2824<tt>val</tt>. Its value is the value at position <tt>idx</tt> of
2825<tt>val</tt>. If <tt>idx</tt> exceeds the length of <tt>val</tt>, the
2826results are undefined.
2827</p>
2828
2829<h5>Example:</h5>
2830
2831<pre>
2832 %result = extractelement &lt;4 x i32&gt; %vec, i32 0 <i>; yields i32</i>
2833</pre>
2834</div>
2835
2836
2837<!-- _______________________________________________________________________ -->
2838<div class="doc_subsubsection">
2839 <a name="i_insertelement">'<tt>insertelement</tt>' Instruction</a>
2840</div>
2841
2842<div class="doc_text">
2843
2844<h5>Syntax:</h5>
2845
2846<pre>
Dan Gohmanbcc3c502008-05-12 23:38:42 +0000[diff] [blame]2847 &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]2848</pre>
2849
2850<h5>Overview:</h5>
2851
2852<p>
2853The '<tt>insertelement</tt>' instruction inserts a scalar
2854element into a vector at a specified index.
2855</p>
2856
2857
2858<h5>Arguments:</h5>
2859
2860<p>
2861The first operand of an '<tt>insertelement</tt>' instruction is a
2862value of <a href="#t_vector">vector</a> type. The second operand is a
2863scalar value whose type must equal the element type of the first
2864operand. The third operand is an index indicating the position at
2865which to insert the value. The index may be a variable.</p>
2866
2867<h5>Semantics:</h5>
2868
2869<p>
2870The result is a vector of the same type as <tt>val</tt>. Its
2871element values are those of <tt>val</tt> except at position
2872<tt>idx</tt>, where it gets the value <tt>elt</tt>. If <tt>idx</tt>
2873exceeds the length of <tt>val</tt>, the results are undefined.
2874</p>
2875
2876<h5>Example:</h5>
2877
2878<pre>
2879 %result = insertelement &lt;4 x i32&gt; %vec, i32 1, i32 0 <i>; yields &lt;4 x i32&gt;</i>
2880</pre>
2881</div>
2882
2883<!-- _______________________________________________________________________ -->
2884<div class="doc_subsubsection">
2885 <a name="i_shufflevector">'<tt>shufflevector</tt>' Instruction</a>
2886</div>
2887
2888<div class="doc_text">
2889
2890<h5>Syntax:</h5>
2891
2892<pre>
2893 &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>
2894</pre>
2895
2896<h5>Overview:</h5>
2897
2898<p>
2899The '<tt>shufflevector</tt>' instruction constructs a permutation of elements
2900from two input vectors, returning a vector of the same type.
2901</p>
2902
2903<h5>Arguments:</h5>
2904
2905<p>
2906The first two operands of a '<tt>shufflevector</tt>' instruction are vectors
2907with types that match each other and types that match the result of the
2908instruction. The third argument is a shuffle mask, which has the same number
2909of elements as the other vector type, but whose element type is always 'i32'.
2910</p>
2911
2912<p>
2913The shuffle mask operand is required to be a constant vector with either
2914constant integer or undef values.
2915</p>
2916
2917<h5>Semantics:</h5>
2918
2919<p>
2920The elements of the two input vectors are numbered from left to right across
2921both of the vectors. The shuffle mask operand specifies, for each element of
2922the result vector, which element of the two input registers the result element
2923gets. The element selector may be undef (meaning "don't care") and the second
2924operand may be undef if performing a shuffle from only one vector.
2925</p>
2926
2927<h5>Example:</h5>
2928
2929<pre>
2930 %result = shufflevector &lt;4 x i32&gt; %v1, &lt;4 x i32&gt; %v2,
2931 &lt;4 x i32&gt; &lt;i32 0, i32 4, i32 1, i32 5&gt; <i>; yields &lt;4 x i32&gt;</i>
2932 %result = shufflevector &lt;4 x i32&gt; %v1, &lt;4 x i32&gt; undef,
2933 &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.
2934</pre>
2935</div>
2936
2937
2938<!-- ======================================================================= -->
2939<div class="doc_subsection">
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]2940 <a name="aggregateops">Aggregate Operations</a>
2941</div>
2942
2943<div class="doc_text">
2944
2945<p>LLVM supports several instructions for working with aggregate values.
2946</p>
2947
2948</div>
2949
2950<!-- _______________________________________________________________________ -->
2951<div class="doc_subsubsection">
2952 <a name="i_extractvalue">'<tt>extractvalue</tt>' Instruction</a>
2953</div>
2954
2955<div class="doc_text">
2956
2957<h5>Syntax:</h5>
2958
2959<pre>
2960 &lt;result&gt; = extractvalue &lt;aggregate type&gt; &lt;val&gt;, &lt;idx&gt;{, &lt;idx&gt;}*
2961</pre>
2962
2963<h5>Overview:</h5>
2964
2965<p>
Dan Gohman6c6dea02008-05-13 18:16:06 +0000[diff] [blame]2966The '<tt>extractvalue</tt>' instruction extracts the value of a struct field
2967or array element from an aggregate value.
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]2968</p>
2969
2970
2971<h5>Arguments:</h5>
2972
2973<p>
2974The first operand of an '<tt>extractvalue</tt>' instruction is a
2975value of <a href="#t_struct">struct</a> or <a href="#t_array">array</a>
Dan Gohman6c6dea02008-05-13 18:16:06 +0000[diff] [blame]2976type. The operands are constant indices to specify which value to extract
Dan Gohmane5febe42008-05-31 00:58:22 +0000[diff] [blame]2977in a similar manner as indices in a
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]2978'<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.
2979</p>
2980
2981<h5>Semantics:</h5>
2982
2983<p>
2984The result is the value at the position in the aggregate specified by
2985the index operands.
2986</p>
2987
2988<h5>Example:</h5>
2989
2990<pre>
Dan Gohmane5febe42008-05-31 00:58:22 +0000[diff] [blame]2991 %result = extractvalue {i32, float} %agg, 0 <i>; yields i32</i>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]2992</pre>
2993</div>
2994
2995
2996<!-- _______________________________________________________________________ -->
2997<div class="doc_subsubsection">
2998 <a name="i_insertvalue">'<tt>insertvalue</tt>' Instruction</a>
2999</div>
3000
3001<div class="doc_text">
3002
3003<h5>Syntax:</h5>
3004
3005<pre>
Dan Gohmane5febe42008-05-31 00:58:22 +0000[diff] [blame]3006 &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]3007</pre>
3008
3009<h5>Overview:</h5>
3010
3011<p>
3012The '<tt>insertvalue</tt>' instruction inserts a value
Dan Gohman6c6dea02008-05-13 18:16:06 +0000[diff] [blame]3013into a struct field or array element in an aggregate.
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3014</p>
3015
3016
3017<h5>Arguments:</h5>
3018
3019<p>
3020The first operand of an '<tt>insertvalue</tt>' instruction is a
3021value of <a href="#t_struct">struct</a> or <a href="#t_array">array</a> type.
3022The second operand is a first-class value to insert.
Dan Gohman4f29e422008-05-23 21:53:15 +0000[diff] [blame]3023The following operands are constant indices
Dan Gohmane5febe42008-05-31 00:58:22 +0000[diff] [blame]3024indicating the position at which to insert the value in a similar manner as
Dan Gohman6c6dea02008-05-13 18:16:06 +0000[diff] [blame]3025indices in a
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3026'<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.
3027The value to insert must have the same type as the value identified
Dan Gohman6c6dea02008-05-13 18:16:06 +0000[diff] [blame]3028by the indices.
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3029
3030<h5>Semantics:</h5>
3031
3032<p>
3033The result is an aggregate of the same type as <tt>val</tt>. Its
3034value is that of <tt>val</tt> except that the value at the position
Dan Gohman6c6dea02008-05-13 18:16:06 +0000[diff] [blame]3035specified by the indices is that of <tt>elt</tt>.
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3036</p>
3037
3038<h5>Example:</h5>
3039
3040<pre>
Dan Gohmanb1aab4e2008-06-23 15:26:37 +0000[diff] [blame]3041 %result = insertvalue {i32, float} %agg, i32 1, 0 <i>; yields {i32, float}</i>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3042</pre>
3043</div>
3044
3045
3046<!-- ======================================================================= -->
3047<div class="doc_subsection">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3048 <a name="memoryops">Memory Access and Addressing Operations</a>
3049</div>
3050
3051<div class="doc_text">
3052
3053<p>A key design point of an SSA-based representation is how it
3054represents memory. In LLVM, no memory locations are in SSA form, which
3055makes things very simple. This section describes how to read, write,
3056allocate, and free memory in LLVM.</p>
3057
3058</div>
3059
3060<!-- _______________________________________________________________________ -->
3061<div class="doc_subsubsection">
3062 <a name="i_malloc">'<tt>malloc</tt>' Instruction</a>
3063</div>
3064
3065<div class="doc_text">
3066
3067<h5>Syntax:</h5>
3068
3069<pre>
3070 &lt;result&gt; = malloc &lt;type&gt;[, i32 &lt;NumElements&gt;][, align &lt;alignment&gt;] <i>; yields {type*}:result</i>
3071</pre>
3072
3073<h5>Overview:</h5>
3074
3075<p>The '<tt>malloc</tt>' instruction allocates memory from the system
Christopher Lambcfe00962007-12-17 01:00:21 +0000[diff] [blame]3076heap and returns a pointer to it. The object is always allocated in the generic
3077address space (address space zero).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3078
3079<h5>Arguments:</h5>
3080
3081<p>The '<tt>malloc</tt>' instruction allocates
3082<tt>sizeof(&lt;type&gt;)*NumElements</tt>
3083bytes of memory from the operating system and returns a pointer of the
3084appropriate type to the program. If "NumElements" is specified, it is the
Gabor Greif5082cf42008-02-09 22:24:34 +0000[diff] [blame]3085number of elements allocated, otherwise "NumElements" is defaulted to be one.
Chris Lattner10368b62008-04-02 00:38:26 +0000[diff] [blame]3086If a constant alignment is specified, the value result of the allocation is guaranteed to
Gabor Greif5082cf42008-02-09 22:24:34 +0000[diff] [blame]3087be aligned to at least that boundary. If not specified, or if zero, the target can
3088choose to align the allocation on any convenient boundary.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3089
3090<p>'<tt>type</tt>' must be a sized type.</p>
3091
3092<h5>Semantics:</h5>
3093
3094<p>Memory is allocated using the system "<tt>malloc</tt>" function, and
Chris Lattner8b094fc2008-04-19 21:01:16 +0000[diff] [blame]3095a pointer is returned. The result of a zero byte allocattion is undefined. The
3096result is null if there is insufficient memory available.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3097
3098<h5>Example:</h5>
3099
3100<pre>
3101 %array = malloc [4 x i8 ] <i>; yields {[%4 x i8]*}:array</i>
3102
3103 %size = <a href="#i_add">add</a> i32 2, 2 <i>; yields {i32}:size = i32 4</i>
3104 %array1 = malloc i8, i32 4 <i>; yields {i8*}:array1</i>
3105 %array2 = malloc [12 x i8], i32 %size <i>; yields {[12 x i8]*}:array2</i>
3106 %array3 = malloc i32, i32 4, align 1024 <i>; yields {i32*}:array3</i>
3107 %array4 = malloc i32, align 1024 <i>; yields {i32*}:array4</i>
3108</pre>
3109</div>
3110
3111<!-- _______________________________________________________________________ -->
3112<div class="doc_subsubsection">
3113 <a name="i_free">'<tt>free</tt>' Instruction</a>
3114</div>
3115
3116<div class="doc_text">
3117
3118<h5>Syntax:</h5>
3119
3120<pre>
3121 free &lt;type&gt; &lt;value&gt; <i>; yields {void}</i>
3122</pre>
3123
3124<h5>Overview:</h5>
3125
3126<p>The '<tt>free</tt>' instruction returns memory back to the unused
3127memory heap to be reallocated in the future.</p>
3128
3129<h5>Arguments:</h5>
3130
3131<p>'<tt>value</tt>' shall be a pointer value that points to a value
3132that was allocated with the '<tt><a href="#i_malloc">malloc</a></tt>'
3133instruction.</p>
3134
3135<h5>Semantics:</h5>
3136
3137<p>Access to the memory pointed to by the pointer is no longer defined
Chris Lattner329d6532008-04-19 22:41:32 +0000[diff] [blame]3138after this instruction executes. If the pointer is null, the operation
3139is a noop.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3140
3141<h5>Example:</h5>
3142
3143<pre>
3144 %array = <a href="#i_malloc">malloc</a> [4 x i8] <i>; yields {[4 x i8]*}:array</i>
3145 free [4 x i8]* %array
3146</pre>
3147</div>
3148
3149<!-- _______________________________________________________________________ -->
3150<div class="doc_subsubsection">
3151 <a name="i_alloca">'<tt>alloca</tt>' Instruction</a>
3152</div>
3153
3154<div class="doc_text">
3155
3156<h5>Syntax:</h5>
3157
3158<pre>
3159 &lt;result&gt; = alloca &lt;type&gt;[, i32 &lt;NumElements&gt;][, align &lt;alignment&gt;] <i>; yields {type*}:result</i>
3160</pre>
3161
3162<h5>Overview:</h5>
3163
3164<p>The '<tt>alloca</tt>' instruction allocates memory on the stack frame of the
3165currently executing function, to be automatically released when this function
Christopher Lambcfe00962007-12-17 01:00:21 +0000[diff] [blame]3166returns to its caller. The object is always allocated in the generic address
3167space (address space zero).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3168
3169<h5>Arguments:</h5>
3170
3171<p>The '<tt>alloca</tt>' instruction allocates <tt>sizeof(&lt;type&gt;)*NumElements</tt>
3172bytes of memory on the runtime stack, returning a pointer of the
Gabor Greif5082cf42008-02-09 22:24:34 +0000[diff] [blame]3173appropriate type to the program. If "NumElements" is specified, it is the
3174number of elements allocated, otherwise "NumElements" is defaulted to be one.
Chris Lattner10368b62008-04-02 00:38:26 +0000[diff] [blame]3175If a constant alignment is specified, the value result of the allocation is guaranteed
Gabor Greif5082cf42008-02-09 22:24:34 +0000[diff] [blame]3176to be aligned to at least that boundary. If not specified, or if zero, the target
3177can choose to align the allocation on any convenient boundary.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3178
3179<p>'<tt>type</tt>' may be any sized type.</p>
3180
3181<h5>Semantics:</h5>
3182
Chris Lattner8b094fc2008-04-19 21:01:16 +0000[diff] [blame]3183<p>Memory is allocated; a pointer is returned. The operation is undefiend if
3184there is insufficient stack space for the allocation. '<tt>alloca</tt>'d
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3185memory is automatically released when the function returns. The '<tt>alloca</tt>'
3186instruction is commonly used to represent automatic variables that must
3187have an address available. When the function returns (either with the <tt><a
3188 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]3189instructions), the memory is reclaimed. Allocating zero bytes
3190is legal, but the result is undefined.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3191
3192<h5>Example:</h5>
3193
3194<pre>
3195 %ptr = alloca i32 <i>; yields {i32*}:ptr</i>
3196 %ptr = alloca i32, i32 4 <i>; yields {i32*}:ptr</i>
3197 %ptr = alloca i32, i32 4, align 1024 <i>; yields {i32*}:ptr</i>
3198 %ptr = alloca i32, align 1024 <i>; yields {i32*}:ptr</i>
3199</pre>
3200</div>
3201
3202<!-- _______________________________________________________________________ -->
3203<div class="doc_subsubsection"> <a name="i_load">'<tt>load</tt>'
3204Instruction</a> </div>
3205<div class="doc_text">
3206<h5>Syntax:</h5>
3207<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>
3208<h5>Overview:</h5>
3209<p>The '<tt>load</tt>' instruction is used to read from memory.</p>
3210<h5>Arguments:</h5>
3211<p>The argument to the '<tt>load</tt>' instruction specifies the memory
3212address from which to load. The pointer must point to a <a
3213 href="#t_firstclass">first class</a> type. If the <tt>load</tt> is
3214marked as <tt>volatile</tt>, then the optimizer is not allowed to modify
3215the number or order of execution of this <tt>load</tt> with other
3216volatile <tt>load</tt> and <tt><a href="#i_store">store</a></tt>
3217instructions. </p>
Chris Lattner21003c82008-01-06 21:04:43 +0000[diff] [blame]3218<p>
Chris Lattner10368b62008-04-02 00:38:26 +0000[diff] [blame]3219The optional constant "align" argument specifies the alignment of the operation
Chris Lattner21003c82008-01-06 21:04:43 +0000[diff] [blame]3220(that is, the alignment of the memory address). A value of 0 or an
3221omitted "align" argument means that the operation has the preferential
3222alignment for the target. It is the responsibility of the code emitter
3223to ensure that the alignment information is correct. Overestimating
3224the alignment results in an undefined behavior. Underestimating the
3225alignment may produce less efficient code. An alignment of 1 is always
3226safe.
3227</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3228<h5>Semantics:</h5>
3229<p>The location of memory pointed to is loaded.</p>
3230<h5>Examples:</h5>
3231<pre> %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
3232 <a
3233 href="#i_store">store</a> i32 3, i32* %ptr <i>; yields {void}</i>
3234 %val = load i32* %ptr <i>; yields {i32}:val = i32 3</i>
3235</pre>
3236</div>
3237<!-- _______________________________________________________________________ -->
3238<div class="doc_subsubsection"> <a name="i_store">'<tt>store</tt>'
3239Instruction</a> </div>
3240<div class="doc_text">
3241<h5>Syntax:</h5>
3242<pre> store &lt;ty&gt; &lt;value&gt;, &lt;ty&gt;* &lt;pointer&gt;[, align &lt;alignment&gt;] <i>; yields {void}</i>
3243 volatile store &lt;ty&gt; &lt;value&gt;, &lt;ty&gt;* &lt;pointer&gt;[, align &lt;alignment&gt;] <i>; yields {void}</i>
3244</pre>
3245<h5>Overview:</h5>
3246<p>The '<tt>store</tt>' instruction is used to write to memory.</p>
3247<h5>Arguments:</h5>
3248<p>There are two arguments to the '<tt>store</tt>' instruction: a value
3249to 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]3250operand must be a pointer to the <a href="#t_firstclass">first class</a> type
3251of the '<tt>&lt;value&gt;</tt>'
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3252operand. If the <tt>store</tt> is marked as <tt>volatile</tt>, then the
3253optimizer is not allowed to modify the number or order of execution of
3254this <tt>store</tt> with other volatile <tt>load</tt> and <tt><a
3255 href="#i_store">store</a></tt> instructions.</p>
Chris Lattner21003c82008-01-06 21:04:43 +0000[diff] [blame]3256<p>
Chris Lattner10368b62008-04-02 00:38:26 +0000[diff] [blame]3257The optional constant "align" argument specifies the alignment of the operation
Chris Lattner21003c82008-01-06 21:04:43 +0000[diff] [blame]3258(that is, the alignment of the memory address). A value of 0 or an
3259omitted "align" argument means that the operation has the preferential
3260alignment for the target. It is the responsibility of the code emitter
3261to ensure that the alignment information is correct. Overestimating
3262the alignment results in an undefined behavior. Underestimating the
3263alignment may produce less efficient code. An alignment of 1 is always
3264safe.
3265</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3266<h5>Semantics:</h5>
3267<p>The contents of memory are updated to contain '<tt>&lt;value&gt;</tt>'
3268at the location specified by the '<tt>&lt;pointer&gt;</tt>' operand.</p>
3269<h5>Example:</h5>
3270<pre> %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
Bill Wendling63ffa142007-10-22 05:10:05 +0000[diff] [blame]3271 store i32 3, i32* %ptr <i>; yields {void}</i>
3272 %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]3273</pre>
3274</div>
3275
3276<!-- _______________________________________________________________________ -->
3277<div class="doc_subsubsection">
3278 <a name="i_getelementptr">'<tt>getelementptr</tt>' Instruction</a>
3279</div>
3280
3281<div class="doc_text">
3282<h5>Syntax:</h5>
3283<pre>
3284 &lt;result&gt; = getelementptr &lt;ty&gt;* &lt;ptrval&gt;{, &lt;ty&gt; &lt;idx&gt;}*
3285</pre>
3286
3287<h5>Overview:</h5>
3288
3289<p>
3290The '<tt>getelementptr</tt>' instruction is used to get the address of a
3291subelement of an aggregate data structure.</p>
3292
3293<h5>Arguments:</h5>
3294
3295<p>This instruction takes a list of integer operands that indicate what
3296elements of the aggregate object to index to. The actual types of the arguments
3297provided depend on the type of the first pointer argument. The
3298'<tt>getelementptr</tt>' instruction is used to index down through the type
3299levels of a structure or to a specific index in an array. When indexing into a
3300structure, only <tt>i32</tt> integer constants are allowed. When indexing
Chris Lattnera7d94ba2008-04-24 05:59:56 +0000[diff] [blame]3301into an array or pointer, only integers of 32 or 64 bits are allowed; 32-bit
3302values will be sign extended to 64-bits if required.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3303
3304<p>For example, let's consider a C code fragment and how it gets
3305compiled to LLVM:</p>
3306
3307<div class="doc_code">
3308<pre>
3309struct RT {
3310 char A;
3311 int B[10][20];
3312 char C;
3313};
3314struct ST {
3315 int X;
3316 double Y;
3317 struct RT Z;
3318};
3319
3320int *foo(struct ST *s) {
3321 return &amp;s[1].Z.B[5][13];
3322}
3323</pre>
3324</div>
3325
3326<p>The LLVM code generated by the GCC frontend is:</p>
3327
3328<div class="doc_code">
3329<pre>
3330%RT = type { i8 , [10 x [20 x i32]], i8 }
3331%ST = type { i32, double, %RT }
3332
3333define i32* %foo(%ST* %s) {
3334entry:
3335 %reg = getelementptr %ST* %s, i32 1, i32 2, i32 1, i32 5, i32 13
3336 ret i32* %reg
3337}
3338</pre>
3339</div>
3340
3341<h5>Semantics:</h5>
3342
3343<p>The index types specified for the '<tt>getelementptr</tt>' instruction depend
3344on the pointer type that is being indexed into. <a href="#t_pointer">Pointer</a>
3345and <a href="#t_array">array</a> types can use a 32-bit or 64-bit
3346<a href="#t_integer">integer</a> type but the value will always be sign extended
Chris Lattner10368b62008-04-02 00:38:26 +0000[diff] [blame]3347to 64-bits. <a href="#t_struct">Structure</a> and <a href="#t_pstruct">packed
3348structure</a> types require <tt>i32</tt> <b>constants</b>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3349
3350<p>In the example above, the first index is indexing into the '<tt>%ST*</tt>'
3351type, which is a pointer, yielding a '<tt>%ST</tt>' = '<tt>{ i32, double, %RT
3352}</tt>' type, a structure. The second index indexes into the third element of
3353the structure, yielding a '<tt>%RT</tt>' = '<tt>{ i8 , [10 x [20 x i32]],
3354i8 }</tt>' type, another structure. The third index indexes into the second
3355element of the structure, yielding a '<tt>[10 x [20 x i32]]</tt>' type, an
3356array. The two dimensions of the array are subscripted into, yielding an
3357'<tt>i32</tt>' type. The '<tt>getelementptr</tt>' instruction returns a pointer
3358to this element, thus computing a value of '<tt>i32*</tt>' type.</p>
3359
3360<p>Note that it is perfectly legal to index partially through a
3361structure, returning a pointer to an inner element. Because of this,
3362the LLVM code for the given testcase is equivalent to:</p>
3363
3364<pre>
3365 define i32* %foo(%ST* %s) {
3366 %t1 = getelementptr %ST* %s, i32 1 <i>; yields %ST*:%t1</i>
3367 %t2 = getelementptr %ST* %t1, i32 0, i32 2 <i>; yields %RT*:%t2</i>
3368 %t3 = getelementptr %RT* %t2, i32 0, i32 1 <i>; yields [10 x [20 x i32]]*:%t3</i>
3369 %t4 = getelementptr [10 x [20 x i32]]* %t3, i32 0, i32 5 <i>; yields [20 x i32]*:%t4</i>
3370 %t5 = getelementptr [20 x i32]* %t4, i32 0, i32 13 <i>; yields i32*:%t5</i>
3371 ret i32* %t5
3372 }
3373</pre>
3374
3375<p>Note that it is undefined to access an array out of bounds: array and
3376pointer indexes must always be within the defined bounds of the array type.
Chris Lattnera7d94ba2008-04-24 05:59:56 +0000[diff] [blame]3377The one exception for this rule is zero length arrays. These arrays are
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3378defined to be accessible as variable length arrays, which requires access
3379beyond the zero'th element.</p>
3380
3381<p>The getelementptr instruction is often confusing. For some more insight
3382into how it works, see <a href="GetElementPtr.html">the getelementptr
3383FAQ</a>.</p>
3384
3385<h5>Example:</h5>
3386
3387<pre>
3388 <i>; yields [12 x i8]*:aptr</i>
3389 %aptr = getelementptr {i32, [12 x i8]}* %sptr, i64 0, i32 1
3390</pre>
3391</div>
3392
3393<!-- ======================================================================= -->
3394<div class="doc_subsection"> <a name="convertops">Conversion Operations</a>
3395</div>
3396<div class="doc_text">
3397<p>The instructions in this category are the conversion instructions (casting)
3398which all take a single operand and a type. They perform various bit conversions
3399on the operand.</p>
3400</div>
3401
3402<!-- _______________________________________________________________________ -->
3403<div class="doc_subsubsection">
3404 <a name="i_trunc">'<tt>trunc .. to</tt>' Instruction</a>
3405</div>
3406<div class="doc_text">
3407
3408<h5>Syntax:</h5>
3409<pre>
3410 &lt;result&gt; = trunc &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3411</pre>
3412
3413<h5>Overview:</h5>
3414<p>
3415The '<tt>trunc</tt>' instruction truncates its operand to the type <tt>ty2</tt>.
3416</p>
3417
3418<h5>Arguments:</h5>
3419<p>
3420The '<tt>trunc</tt>' instruction takes a <tt>value</tt> to trunc, which must
3421be an <a href="#t_integer">integer</a> type, and a type that specifies the size
3422and type of the result, which must be an <a href="#t_integer">integer</a>
3423type. The bit size of <tt>value</tt> must be larger than the bit size of
3424<tt>ty2</tt>. Equal sized types are not allowed.</p>
3425
3426<h5>Semantics:</h5>
3427<p>
3428The '<tt>trunc</tt>' instruction truncates the high order bits in <tt>value</tt>
3429and converts the remaining bits to <tt>ty2</tt>. Since the source size must be
3430larger than the destination size, <tt>trunc</tt> cannot be a <i>no-op cast</i>.
3431It will always truncate bits.</p>
3432
3433<h5>Example:</h5>
3434<pre>
3435 %X = trunc i32 257 to i8 <i>; yields i8:1</i>
3436 %Y = trunc i32 123 to i1 <i>; yields i1:true</i>
3437 %Y = trunc i32 122 to i1 <i>; yields i1:false</i>
3438</pre>
3439</div>
3440
3441<!-- _______________________________________________________________________ -->
3442<div class="doc_subsubsection">
3443 <a name="i_zext">'<tt>zext .. to</tt>' Instruction</a>
3444</div>
3445<div class="doc_text">
3446
3447<h5>Syntax:</h5>
3448<pre>
3449 &lt;result&gt; = zext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3450</pre>
3451
3452<h5>Overview:</h5>
3453<p>The '<tt>zext</tt>' instruction zero extends its operand to type
3454<tt>ty2</tt>.</p>
3455
3456
3457<h5>Arguments:</h5>
3458<p>The '<tt>zext</tt>' instruction takes a value to cast, which must be of
3459<a href="#t_integer">integer</a> type, and a type to cast it to, which must
3460also be of <a href="#t_integer">integer</a> type. The bit size of the
3461<tt>value</tt> must be smaller than the bit size of the destination type,
3462<tt>ty2</tt>.</p>
3463
3464<h5>Semantics:</h5>
3465<p>The <tt>zext</tt> fills the high order bits of the <tt>value</tt> with zero
3466bits until it reaches the size of the destination type, <tt>ty2</tt>.</p>
3467
3468<p>When zero extending from i1, the result will always be either 0 or 1.</p>
3469
3470<h5>Example:</h5>
3471<pre>
3472 %X = zext i32 257 to i64 <i>; yields i64:257</i>
3473 %Y = zext i1 true to i32 <i>; yields i32:1</i>
3474</pre>
3475</div>
3476
3477<!-- _______________________________________________________________________ -->
3478<div class="doc_subsubsection">
3479 <a name="i_sext">'<tt>sext .. to</tt>' Instruction</a>
3480</div>
3481<div class="doc_text">
3482
3483<h5>Syntax:</h5>
3484<pre>
3485 &lt;result&gt; = sext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3486</pre>
3487
3488<h5>Overview:</h5>
3489<p>The '<tt>sext</tt>' sign extends <tt>value</tt> to the type <tt>ty2</tt>.</p>
3490
3491<h5>Arguments:</h5>
3492<p>
3493The '<tt>sext</tt>' instruction takes a value to cast, which must be of
3494<a href="#t_integer">integer</a> type, and a type to cast it to, which must
3495also be of <a href="#t_integer">integer</a> type. The bit size of the
3496<tt>value</tt> must be smaller than the bit size of the destination type,
3497<tt>ty2</tt>.</p>
3498
3499<h5>Semantics:</h5>
3500<p>
3501The '<tt>sext</tt>' instruction performs a sign extension by copying the sign
3502bit (highest order bit) of the <tt>value</tt> until it reaches the bit size of
3503the type <tt>ty2</tt>.</p>
3504
3505<p>When sign extending from i1, the extension always results in -1 or 0.</p>
3506
3507<h5>Example:</h5>
3508<pre>
3509 %X = sext i8 -1 to i16 <i>; yields i16 :65535</i>
3510 %Y = sext i1 true to i32 <i>; yields i32:-1</i>
3511</pre>
3512</div>
3513
3514<!-- _______________________________________________________________________ -->
3515<div class="doc_subsubsection">
3516 <a name="i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a>
3517</div>
3518
3519<div class="doc_text">
3520
3521<h5>Syntax:</h5>
3522
3523<pre>
3524 &lt;result&gt; = fptrunc &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3525</pre>
3526
3527<h5>Overview:</h5>
3528<p>The '<tt>fptrunc</tt>' instruction truncates <tt>value</tt> to type
3529<tt>ty2</tt>.</p>
3530
3531
3532<h5>Arguments:</h5>
3533<p>The '<tt>fptrunc</tt>' instruction takes a <a href="#t_floating">floating
3534 point</a> value to cast and a <a href="#t_floating">floating point</a> type to
3535cast it to. The size of <tt>value</tt> must be larger than the size of
3536<tt>ty2</tt>. This implies that <tt>fptrunc</tt> cannot be used to make a
3537<i>no-op cast</i>.</p>
3538
3539<h5>Semantics:</h5>
3540<p> The '<tt>fptrunc</tt>' instruction truncates a <tt>value</tt> from a larger
3541<a href="#t_floating">floating point</a> type to a smaller
3542<a href="#t_floating">floating point</a> type. If the value cannot fit within
3543the destination type, <tt>ty2</tt>, then the results are undefined.</p>
3544
3545<h5>Example:</h5>
3546<pre>
3547 %X = fptrunc double 123.0 to float <i>; yields float:123.0</i>
3548 %Y = fptrunc double 1.0E+300 to float <i>; yields undefined</i>
3549</pre>
3550</div>
3551
3552<!-- _______________________________________________________________________ -->
3553<div class="doc_subsubsection">
3554 <a name="i_fpext">'<tt>fpext .. to</tt>' Instruction</a>
3555</div>
3556<div class="doc_text">
3557
3558<h5>Syntax:</h5>
3559<pre>
3560 &lt;result&gt; = fpext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3561</pre>
3562
3563<h5>Overview:</h5>
3564<p>The '<tt>fpext</tt>' extends a floating point <tt>value</tt> to a larger
3565floating point value.</p>
3566
3567<h5>Arguments:</h5>
3568<p>The '<tt>fpext</tt>' instruction takes a
3569<a href="#t_floating">floating point</a> <tt>value</tt> to cast,
3570and a <a href="#t_floating">floating point</a> type to cast it to. The source
3571type must be smaller than the destination type.</p>
3572
3573<h5>Semantics:</h5>
3574<p>The '<tt>fpext</tt>' instruction extends the <tt>value</tt> from a smaller
3575<a href="#t_floating">floating point</a> type to a larger
3576<a href="#t_floating">floating point</a> type. The <tt>fpext</tt> cannot be
3577used to make a <i>no-op cast</i> because it always changes bits. Use
3578<tt>bitcast</tt> to make a <i>no-op cast</i> for a floating point cast.</p>
3579
3580<h5>Example:</h5>
3581<pre>
3582 %X = fpext float 3.1415 to double <i>; yields double:3.1415</i>
3583 %Y = fpext float 1.0 to float <i>; yields float:1.0 (no-op)</i>
3584</pre>
3585</div>
3586
3587<!-- _______________________________________________________________________ -->
3588<div class="doc_subsubsection">
3589 <a name="i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a>
3590</div>
3591<div class="doc_text">
3592
3593<h5>Syntax:</h5>
3594<pre>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3595 &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]3596</pre>
3597
3598<h5>Overview:</h5>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3599<p>The '<tt>fptoui</tt>' converts a floating point <tt>value</tt> to its
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3600unsigned integer equivalent of type <tt>ty2</tt>.
3601</p>
3602
3603<h5>Arguments:</h5>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3604<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]3605scalar or vector <a href="#t_floating">floating point</a> value, and a type
3606to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
3607type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
3608vector integer type with the same number of elements as <tt>ty</tt></p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3609
3610<h5>Semantics:</h5>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3611<p> The '<tt>fptoui</tt>' instruction converts its
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3612<a href="#t_floating">floating point</a> operand into the nearest (rounding
3613towards zero) unsigned integer value. If the value cannot fit in <tt>ty2</tt>,
3614the results are undefined.</p>
3615
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3616<h5>Example:</h5>
3617<pre>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3618 %X = fptoui double 123.0 to i32 <i>; yields i32:123</i>
Chris Lattner681f1e82007-09-22 03:17:52 +0000[diff] [blame]3619 %Y = fptoui float 1.0E+300 to i1 <i>; yields undefined:1</i>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3620 %X = fptoui float 1.04E+17 to i8 <i>; yields undefined:1</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3621</pre>
3622</div>
3623
3624<!-- _______________________________________________________________________ -->
3625<div class="doc_subsubsection">
3626 <a name="i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a>
3627</div>
3628<div class="doc_text">
3629
3630<h5>Syntax:</h5>
3631<pre>
3632 &lt;result&gt; = fptosi &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3633</pre>
3634
3635<h5>Overview:</h5>
3636<p>The '<tt>fptosi</tt>' instruction converts
3637<a href="#t_floating">floating point</a> <tt>value</tt> to type <tt>ty2</tt>.
3638</p>
3639
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3640<h5>Arguments:</h5>
3641<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]3642scalar or vector <a href="#t_floating">floating point</a> value, and a type
3643to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
3644type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
3645vector integer type with the same number of elements as <tt>ty</tt></p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3646
3647<h5>Semantics:</h5>
3648<p>The '<tt>fptosi</tt>' instruction converts its
3649<a href="#t_floating">floating point</a> operand into the nearest (rounding
3650towards zero) signed integer value. If the value cannot fit in <tt>ty2</tt>,
3651the results are undefined.</p>
3652
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3653<h5>Example:</h5>
3654<pre>
3655 %X = fptosi double -123.0 to i32 <i>; yields i32:-123</i>
Chris Lattner681f1e82007-09-22 03:17:52 +0000[diff] [blame]3656 %Y = fptosi float 1.0E-247 to i1 <i>; yields undefined:1</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3657 %X = fptosi float 1.04E+17 to i8 <i>; yields undefined:1</i>
3658</pre>
3659</div>
3660
3661<!-- _______________________________________________________________________ -->
3662<div class="doc_subsubsection">
3663 <a name="i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a>
3664</div>
3665<div class="doc_text">
3666
3667<h5>Syntax:</h5>
3668<pre>
3669 &lt;result&gt; = uitofp &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3670</pre>
3671
3672<h5>Overview:</h5>
3673<p>The '<tt>uitofp</tt>' instruction regards <tt>value</tt> as an unsigned
3674integer and converts that value to the <tt>ty2</tt> type.</p>
3675
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3676<h5>Arguments:</h5>
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]3677<p>The '<tt>uitofp</tt>' instruction takes a value to cast, which must be a
3678scalar or vector <a href="#t_integer">integer</a> value, and a type to cast it
3679to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
3680type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
3681floating point type with the same number of elements as <tt>ty</tt></p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3682
3683<h5>Semantics:</h5>
3684<p>The '<tt>uitofp</tt>' instruction interprets its operand as an unsigned
3685integer quantity and converts it to the corresponding floating point value. If
3686the value cannot fit in the floating point value, the results are undefined.</p>
3687
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3688<h5>Example:</h5>
3689<pre>
3690 %X = uitofp i32 257 to float <i>; yields float:257.0</i>
3691 %Y = uitofp i8 -1 to double <i>; yields double:255.0</i>
3692</pre>
3693</div>
3694
3695<!-- _______________________________________________________________________ -->
3696<div class="doc_subsubsection">
3697 <a name="i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a>
3698</div>
3699<div class="doc_text">
3700
3701<h5>Syntax:</h5>
3702<pre>
3703 &lt;result&gt; = sitofp &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3704</pre>
3705
3706<h5>Overview:</h5>
3707<p>The '<tt>sitofp</tt>' instruction regards <tt>value</tt> as a signed
3708integer and converts that value to the <tt>ty2</tt> type.</p>
3709
3710<h5>Arguments:</h5>
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]3711<p>The '<tt>sitofp</tt>' instruction takes a value to cast, which must be a
3712scalar or vector <a href="#t_integer">integer</a> value, and a type to cast it
3713to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
3714type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
3715floating point type with the same number of elements as <tt>ty</tt></p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3716
3717<h5>Semantics:</h5>
3718<p>The '<tt>sitofp</tt>' instruction interprets its operand as a signed
3719integer quantity and converts it to the corresponding floating point value. If
3720the value cannot fit in the floating point value, the results are undefined.</p>
3721
3722<h5>Example:</h5>
3723<pre>
3724 %X = sitofp i32 257 to float <i>; yields float:257.0</i>
3725 %Y = sitofp i8 -1 to double <i>; yields double:-1.0</i>
3726</pre>
3727</div>
3728
3729<!-- _______________________________________________________________________ -->
3730<div class="doc_subsubsection">
3731 <a name="i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a>
3732</div>
3733<div class="doc_text">
3734
3735<h5>Syntax:</h5>
3736<pre>
3737 &lt;result&gt; = ptrtoint &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3738</pre>
3739
3740<h5>Overview:</h5>
3741<p>The '<tt>ptrtoint</tt>' instruction converts the pointer <tt>value</tt> to
3742the integer type <tt>ty2</tt>.</p>
3743
3744<h5>Arguments:</h5>
3745<p>The '<tt>ptrtoint</tt>' instruction takes a <tt>value</tt> to cast, which
3746must be a <a href="#t_pointer">pointer</a> value, and a type to cast it to
3747<tt>ty2</tt>, which must be an <a href="#t_integer">integer</a> type.
3748
3749<h5>Semantics:</h5>
3750<p>The '<tt>ptrtoint</tt>' instruction converts <tt>value</tt> to integer type
3751<tt>ty2</tt> by interpreting the pointer value as an integer and either
3752truncating or zero extending that value to the size of the integer type. If
3753<tt>value</tt> is smaller than <tt>ty2</tt> then a zero extension is done. If
3754<tt>value</tt> is larger than <tt>ty2</tt> then a truncation is done. If they
3755are the same size, then nothing is done (<i>no-op cast</i>) other than a type
3756change.</p>
3757
3758<h5>Example:</h5>
3759<pre>
3760 %X = ptrtoint i32* %X to i8 <i>; yields truncation on 32-bit architecture</i>
3761 %Y = ptrtoint i32* %x to i64 <i>; yields zero extension on 32-bit architecture</i>
3762</pre>
3763</div>
3764
3765<!-- _______________________________________________________________________ -->
3766<div class="doc_subsubsection">
3767 <a name="i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a>
3768</div>
3769<div class="doc_text">
3770
3771<h5>Syntax:</h5>
3772<pre>
3773 &lt;result&gt; = inttoptr &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3774</pre>
3775
3776<h5>Overview:</h5>
3777<p>The '<tt>inttoptr</tt>' instruction converts an integer <tt>value</tt> to
3778a pointer type, <tt>ty2</tt>.</p>
3779
3780<h5>Arguments:</h5>
3781<p>The '<tt>inttoptr</tt>' instruction takes an <a href="#t_integer">integer</a>
3782value to cast, and a type to cast it to, which must be a
3783<a href="#t_pointer">pointer</a> type.
3784
3785<h5>Semantics:</h5>
3786<p>The '<tt>inttoptr</tt>' instruction converts <tt>value</tt> to type
3787<tt>ty2</tt> by applying either a zero extension or a truncation depending on
3788the size of the integer <tt>value</tt>. If <tt>value</tt> is larger than the
3789size of a pointer then a truncation is done. If <tt>value</tt> is smaller than
3790the size of a pointer then a zero extension is done. If they are the same size,
3791nothing is done (<i>no-op cast</i>).</p>
3792
3793<h5>Example:</h5>
3794<pre>
3795 %X = inttoptr i32 255 to i32* <i>; yields zero extension on 64-bit architecture</i>
3796 %X = inttoptr i32 255 to i32* <i>; yields no-op on 32-bit architecture</i>
3797 %Y = inttoptr i64 0 to i32* <i>; yields truncation on 32-bit architecture</i>
3798</pre>
3799</div>
3800
3801<!-- _______________________________________________________________________ -->
3802<div class="doc_subsubsection">
3803 <a name="i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a>
3804</div>
3805<div class="doc_text">
3806
3807<h5>Syntax:</h5>
3808<pre>
3809 &lt;result&gt; = bitcast &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3810</pre>
3811
3812<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3813
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3814<p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
3815<tt>ty2</tt> without changing any bits.</p>
3816
3817<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3818
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3819<p>The '<tt>bitcast</tt>' instruction takes a value to cast, which must be
Dan Gohman7305fa02008-09-08 16:45:59 +0000[diff] [blame]3820a non-aggregate first class value, and a type to cast it to, which must also be
3821a non-aggregate <a href="#t_firstclass">first class</a> type. The bit sizes of
3822<tt>value</tt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3823and the destination type, <tt>ty2</tt>, must be identical. If the source
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3824type is a pointer, the destination type must also be a pointer. This
3825instruction supports bitwise conversion of vectors to integers and to vectors
3826of other types (as long as they have the same size).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3827
3828<h5>Semantics:</h5>
3829<p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
3830<tt>ty2</tt>. It is always a <i>no-op cast</i> because no bits change with
3831this conversion. The conversion is done as if the <tt>value</tt> had been
3832stored to memory and read back as type <tt>ty2</tt>. Pointer types may only be
3833converted to other pointer types with this instruction. To convert pointers to
3834other types, use the <a href="#i_inttoptr">inttoptr</a> or
3835<a href="#i_ptrtoint">ptrtoint</a> instructions first.</p>
3836
3837<h5>Example:</h5>
3838<pre>
3839 %X = bitcast i8 255 to i8 <i>; yields i8 :-1</i>
3840 %Y = bitcast i32* %x to sint* <i>; yields sint*:%x</i>
3841 %Z = bitcast <2xint> %V to i64; <i>; yields i64: %V</i>
3842</pre>
3843</div>
3844
3845<!-- ======================================================================= -->
3846<div class="doc_subsection"> <a name="otherops">Other Operations</a> </div>
3847<div class="doc_text">
3848<p>The instructions in this category are the "miscellaneous"
3849instructions, which defy better classification.</p>
3850</div>
3851
3852<!-- _______________________________________________________________________ -->
3853<div class="doc_subsubsection"><a name="i_icmp">'<tt>icmp</tt>' Instruction</a>
3854</div>
3855<div class="doc_text">
3856<h5>Syntax:</h5>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]3857<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]3858</pre>
3859<h5>Overview:</h5>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]3860<p>The '<tt>icmp</tt>' instruction returns a boolean value or
3861a vector of boolean values based on comparison
3862of its two integer, integer vector, or pointer operands.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3863<h5>Arguments:</h5>
3864<p>The '<tt>icmp</tt>' instruction takes three operands. The first operand is
3865the condition code indicating the kind of comparison to perform. It is not
3866a value, just a keyword. The possible condition code are:
3867<ol>
3868 <li><tt>eq</tt>: equal</li>
3869 <li><tt>ne</tt>: not equal </li>
3870 <li><tt>ugt</tt>: unsigned greater than</li>
3871 <li><tt>uge</tt>: unsigned greater or equal</li>
3872 <li><tt>ult</tt>: unsigned less than</li>
3873 <li><tt>ule</tt>: unsigned less or equal</li>
3874 <li><tt>sgt</tt>: signed greater than</li>
3875 <li><tt>sge</tt>: signed greater or equal</li>
3876 <li><tt>slt</tt>: signed less than</li>
3877 <li><tt>sle</tt>: signed less or equal</li>
3878</ol>
3879<p>The remaining two arguments must be <a href="#t_integer">integer</a> or
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]3880<a href="#t_pointer">pointer</a>
3881or integer <a href="#t_vector">vector</a> typed.
3882They must also be identical types.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3883<h5>Semantics:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3884<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]3885the condition code given as <tt>cond</tt>. The comparison performed always
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]3886yields 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]3887<ol>
3888 <li><tt>eq</tt>: yields <tt>true</tt> if the operands are equal,
3889 <tt>false</tt> otherwise. No sign interpretation is necessary or performed.
3890 </li>
3891 <li><tt>ne</tt>: yields <tt>true</tt> if the operands are unequal,
3892 <tt>false</tt> otherwise. No sign interpretation is necessary or performed.
3893 <li><tt>ugt</tt>: interprets the operands as unsigned values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3894 <tt>true</tt> if <tt>op1</tt> is greater than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3895 <li><tt>uge</tt>: interprets the operands as unsigned values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3896 <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]3897 <li><tt>ult</tt>: interprets the operands as unsigned values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3898 <tt>true</tt> if <tt>op1</tt> is less than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3899 <li><tt>ule</tt>: interprets the operands as unsigned values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3900 <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]3901 <li><tt>sgt</tt>: interprets the operands as signed values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3902 <tt>true</tt> if <tt>op1</tt> is greater than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3903 <li><tt>sge</tt>: interprets the operands as signed values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3904 <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]3905 <li><tt>slt</tt>: interprets the operands as signed values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3906 <tt>true</tt> if <tt>op1</tt> is less than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3907 <li><tt>sle</tt>: interprets the operands as signed values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3908 <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]3909</ol>
3910<p>If the operands are <a href="#t_pointer">pointer</a> typed, the pointer
3911values are compared as if they were integers.</p>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]3912<p>If the operands are integer vectors, then they are compared
3913element by element. The result is an <tt>i1</tt> vector with
3914the same number of elements as the values being compared.
3915Otherwise, the result is an <tt>i1</tt>.
3916</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3917
3918<h5>Example:</h5>
3919<pre> &lt;result&gt; = icmp eq i32 4, 5 <i>; yields: result=false</i>
3920 &lt;result&gt; = icmp ne float* %X, %X <i>; yields: result=false</i>
3921 &lt;result&gt; = icmp ult i16 4, 5 <i>; yields: result=true</i>
3922 &lt;result&gt; = icmp sgt i16 4, 5 <i>; yields: result=false</i>
3923 &lt;result&gt; = icmp ule i16 -4, 5 <i>; yields: result=false</i>
3924 &lt;result&gt; = icmp sge i16 4, 5 <i>; yields: result=false</i>
3925</pre>
3926</div>
3927
3928<!-- _______________________________________________________________________ -->
3929<div class="doc_subsubsection"><a name="i_fcmp">'<tt>fcmp</tt>' Instruction</a>
3930</div>
3931<div class="doc_text">
3932<h5>Syntax:</h5>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]3933<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]3934</pre>
3935<h5>Overview:</h5>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]3936<p>The '<tt>fcmp</tt>' instruction returns a boolean value
3937or vector of boolean values based on comparison
3938of its operands.
3939<p>
3940If the operands are floating point scalars, then the result
3941type is a boolean (<a href="#t_primitive"><tt>i1</tt></a>).
3942</p>
3943<p>If the operands are floating point vectors, then the result type
3944is a vector of boolean with the same number of elements as the
3945operands being compared.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3946<h5>Arguments:</h5>
3947<p>The '<tt>fcmp</tt>' instruction takes three operands. The first operand is
3948the condition code indicating the kind of comparison to perform. It is not
3949a value, just a keyword. The possible condition code are:
3950<ol>
3951 <li><tt>false</tt>: no comparison, always returns false</li>
3952 <li><tt>oeq</tt>: ordered and equal</li>
3953 <li><tt>ogt</tt>: ordered and greater than </li>
3954 <li><tt>oge</tt>: ordered and greater than or equal</li>
3955 <li><tt>olt</tt>: ordered and less than </li>
3956 <li><tt>ole</tt>: ordered and less than or equal</li>
3957 <li><tt>one</tt>: ordered and not equal</li>
3958 <li><tt>ord</tt>: ordered (no nans)</li>
3959 <li><tt>ueq</tt>: unordered or equal</li>
3960 <li><tt>ugt</tt>: unordered or greater than </li>
3961 <li><tt>uge</tt>: unordered or greater than or equal</li>
3962 <li><tt>ult</tt>: unordered or less than </li>
3963 <li><tt>ule</tt>: unordered or less than or equal</li>
3964 <li><tt>une</tt>: unordered or not equal</li>
3965 <li><tt>uno</tt>: unordered (either nans)</li>
3966 <li><tt>true</tt>: no comparison, always returns true</li>
3967</ol>
3968<p><i>Ordered</i> means that neither operand is a QNAN while
3969<i>unordered</i> means that either operand may be a QNAN.</p>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]3970<p>Each of <tt>val1</tt> and <tt>val2</tt> arguments must be
3971either a <a href="#t_floating">floating point</a> type
3972or a <a href="#t_vector">vector</a> of floating point type.
3973They must have identical types.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3974<h5>Semantics:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3975<p>The '<tt>fcmp</tt>' instruction compares <tt>op1</tt> and <tt>op2</tt>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]3976according to the condition code given as <tt>cond</tt>.
3977If the operands are vectors, then the vectors are compared
3978element by element.
3979Each comparison performed
3980always yields an <a href="#t_primitive">i1</a> result, as follows:
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3981<ol>
3982 <li><tt>false</tt>: always yields <tt>false</tt>, regardless of operands.</li>
3983 <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]3984 <tt>op1</tt> is equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3985 <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]3986 <tt>op1</tt> is greather than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3987 <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]3988 <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3989 <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]3990 <tt>op1</tt> is less than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3991 <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]3992 <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3993 <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]3994 <tt>op1</tt> is not equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3995 <li><tt>ord</tt>: yields <tt>true</tt> if both operands are not a QNAN.</li>
3996 <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]3997 <tt>op1</tt> is equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3998 <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]3999 <tt>op1</tt> is greater than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4000 <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]4001 <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4002 <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]4003 <tt>op1</tt> is less than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4004 <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]4005 <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4006 <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]4007 <tt>op1</tt> is not equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4008 <li><tt>uno</tt>: yields <tt>true</tt> if either operand is a QNAN.</li>
4009 <li><tt>true</tt>: always yields <tt>true</tt>, regardless of operands.</li>
4010</ol>
4011
4012<h5>Example:</h5>
4013<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]4014 &lt;result&gt; = fcmp one float 4.0, 5.0 <i>; yields: result=true</i>
4015 &lt;result&gt; = fcmp olt float 4.0, 5.0 <i>; yields: result=true</i>
4016 &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]4017</pre>
4018</div>
4019
4020<!-- _______________________________________________________________________ -->
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]4021<div class="doc_subsubsection">
4022 <a name="i_vicmp">'<tt>vicmp</tt>' Instruction</a>
4023</div>
4024<div class="doc_text">
4025<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4026<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]4027</pre>
4028<h5>Overview:</h5>
4029<p>The '<tt>vicmp</tt>' instruction returns an integer vector value based on
4030element-wise comparison of its two integer vector operands.</p>
4031<h5>Arguments:</h5>
4032<p>The '<tt>vicmp</tt>' instruction takes three operands. The first operand is
4033the condition code indicating the kind of comparison to perform. It is not
4034a value, just a keyword. The possible condition code are:
4035<ol>
4036 <li><tt>eq</tt>: equal</li>
4037 <li><tt>ne</tt>: not equal </li>
4038 <li><tt>ugt</tt>: unsigned greater than</li>
4039 <li><tt>uge</tt>: unsigned greater or equal</li>
4040 <li><tt>ult</tt>: unsigned less than</li>
4041 <li><tt>ule</tt>: unsigned less or equal</li>
4042 <li><tt>sgt</tt>: signed greater than</li>
4043 <li><tt>sge</tt>: signed greater or equal</li>
4044 <li><tt>slt</tt>: signed less than</li>
4045 <li><tt>sle</tt>: signed less or equal</li>
4046</ol>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4047<p>The remaining two arguments must be <a href="#t_vector">vector</a> or
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]4048<a href="#t_integer">integer</a> typed. They must also be identical types.</p>
4049<h5>Semantics:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4050<p>The '<tt>vicmp</tt>' instruction compares <tt>op1</tt> and <tt>op2</tt>
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]4051according to the condition code given as <tt>cond</tt>. The comparison yields a
4052<a href="#t_vector">vector</a> of <a href="#t_integer">integer</a> result, of
4053identical type as the values being compared. The most significant bit in each
4054element is 1 if the element-wise comparison evaluates to true, and is 0
4055otherwise. All other bits of the result are undefined. The condition codes
4056are evaluated identically to the <a href="#i_icmp">'<tt>icmp</tt>'
4057instruction</a>.
4058
4059<h5>Example:</h5>
4060<pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4061 &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>
4062 &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]4063</pre>
4064</div>
4065
4066<!-- _______________________________________________________________________ -->
4067<div class="doc_subsubsection">
4068 <a name="i_vfcmp">'<tt>vfcmp</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; = vfcmp &lt;cond&gt; &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt;</pre>
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]4073<h5>Overview:</h5>
4074<p>The '<tt>vfcmp</tt>' instruction returns an integer vector value based on
4075element-wise comparison of its two floating point vector operands. The output
4076elements have the same width as the input elements.</p>
4077<h5>Arguments:</h5>
4078<p>The '<tt>vfcmp</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>false</tt>: no comparison, always returns false</li>
4083 <li><tt>oeq</tt>: ordered and equal</li>
4084 <li><tt>ogt</tt>: ordered and greater than </li>
4085 <li><tt>oge</tt>: ordered and greater than or equal</li>
4086 <li><tt>olt</tt>: ordered and less than </li>
4087 <li><tt>ole</tt>: ordered and less than or equal</li>
4088 <li><tt>one</tt>: ordered and not equal</li>
4089 <li><tt>ord</tt>: ordered (no nans)</li>
4090 <li><tt>ueq</tt>: unordered or equal</li>
4091 <li><tt>ugt</tt>: unordered or greater than </li>
4092 <li><tt>uge</tt>: unordered or greater than or equal</li>
4093 <li><tt>ult</tt>: unordered or less than </li>
4094 <li><tt>ule</tt>: unordered or less than or equal</li>
4095 <li><tt>une</tt>: unordered or not equal</li>
4096 <li><tt>uno</tt>: unordered (either nans)</li>
4097 <li><tt>true</tt>: no comparison, always returns true</li>
4098</ol>
4099<p>The remaining two arguments must be <a href="#t_vector">vector</a> of
4100<a href="#t_floating">floating point</a> typed. They must also be identical
4101types.</p>
4102<h5>Semantics:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4103<p>The '<tt>vfcmp</tt>' instruction compares <tt>op1</tt> and <tt>op2</tt>
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]4104according to the condition code given as <tt>cond</tt>. The comparison yields a
4105<a href="#t_vector">vector</a> of <a href="#t_integer">integer</a> result, with
4106an identical number of elements as the values being compared, and each element
4107having identical with to the width of the floating point elements. The most
4108significant bit in each element is 1 if the element-wise comparison evaluates to
4109true, and is 0 otherwise. All other bits of the result are undefined. The
4110condition codes are evaluated identically to the
4111<a href="#i_fcmp">'<tt>fcmp</tt>' instruction</a>.
4112
4113<h5>Example:</h5>
4114<pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4115 &lt;result&gt; = vfcmp oeq &lt;2 x float&gt; &lt; float 4, float 0 &gt;, &lt; float 5, float 0 &gt; <i>; yields: result=&lt;2 x i32&gt; &lt; i32 0, i32 -1 &gt;</i>
4116 &lt;result&gt; = vfcmp ult &lt;2 x double&gt; &lt; double 1, double 2 &gt;, &lt; double 2, double 2&gt; <i>; yields: result=&lt;2 x i64&gt; &lt; i64 -1, i64 0 &gt;</i>
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]4117</pre>
4118</div>
4119
4120<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4121<div class="doc_subsubsection">
4122 <a name="i_phi">'<tt>phi</tt>' Instruction</a>
4123</div>
4124
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4125<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4126
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4127<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4128
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4129<pre> &lt;result&gt; = phi &lt;ty&gt; [ &lt;val0&gt;, &lt;label0&gt;], ...<br></pre>
4130<h5>Overview:</h5>
4131<p>The '<tt>phi</tt>' instruction is used to implement the &#966; node in
4132the SSA graph representing the function.</p>
4133<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4134
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4135<p>The type of the incoming values is specified with the first type
4136field. After this, the '<tt>phi</tt>' instruction takes a list of pairs
4137as arguments, with one pair for each predecessor basic block of the
4138current block. Only values of <a href="#t_firstclass">first class</a>
4139type may be used as the value arguments to the PHI node. Only labels
4140may be used as the label arguments.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4141
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4142<p>There must be no non-phi instructions between the start of a basic
4143block and the PHI instructions: i.e. PHI instructions must be first in
4144a basic block.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4145
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4146<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4147
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4148<p>At runtime, the '<tt>phi</tt>' instruction logically takes on the value
4149specified by the pair corresponding to the predecessor basic block that executed
4150just prior to the current block.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4151
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4152<h5>Example:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4153<pre>
4154Loop: ; Infinite loop that counts from 0 on up...
4155 %indvar = phi i32 [ 0, %LoopHeader ], [ %nextindvar, %Loop ]
4156 %nextindvar = add i32 %indvar, 1
4157 br label %Loop
4158</pre>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4159</div>
4160
4161<!-- _______________________________________________________________________ -->
4162<div class="doc_subsubsection">
4163 <a name="i_select">'<tt>select</tt>' Instruction</a>
4164</div>
4165
4166<div class="doc_text">
4167
4168<h5>Syntax:</h5>
4169
4170<pre>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4171 &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>
4172
4173 <i>selty</i> is either i1 or {&lt;N x i1&gt}
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4174</pre>
4175
4176<h5>Overview:</h5>
4177
4178<p>
4179The '<tt>select</tt>' instruction is used to choose one value based on a
4180condition, without branching.
4181</p>
4182
4183
4184<h5>Arguments:</h5>
4185
4186<p>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4187The '<tt>select</tt>' instruction requires an 'i1' value or
4188a vector of 'i1' values indicating the
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4189condition, and two values of the same <a href="#t_firstclass">first class</a>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4190type. If the val1/val2 are vectors and
4191the condition is a scalar, then entire vectors are selected, not
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4192individual elements.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4193</p>
4194
4195<h5>Semantics:</h5>
4196
4197<p>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4198If 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]4199value argument; otherwise, it returns the second value argument.
4200</p>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4201<p>
4202If the condition is a vector of i1, then the value arguments must
4203be vectors of the same size, and the selection is done element
4204by element.
4205</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4206
4207<h5>Example:</h5>
4208
4209<pre>
4210 %X = select i1 true, i8 17, i8 42 <i>; yields i8:17</i>
4211</pre>
4212</div>
4213
4214
4215<!-- _______________________________________________________________________ -->
4216<div class="doc_subsubsection">
4217 <a name="i_call">'<tt>call</tt>' Instruction</a>
4218</div>
4219
4220<div class="doc_text">
4221
4222<h5>Syntax:</h5>
4223<pre>
Nick Lewycky93082fc2007-09-08 13:57:50 +0000[diff] [blame]4224 &lt;result&gt; = [tail] call [<a href="#callingconv">cconv</a>] &lt;ty&gt; [&lt;fnty&gt;*] &lt;fnptrval&gt;(&lt;param list&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4225</pre>
4226
4227<h5>Overview:</h5>
4228
4229<p>The '<tt>call</tt>' instruction represents a simple function call.</p>
4230
4231<h5>Arguments:</h5>
4232
4233<p>This instruction requires several arguments:</p>
4234
4235<ol>
4236 <li>
4237 <p>The optional "tail" marker indicates whether the callee function accesses
4238 any allocas or varargs in the caller. If the "tail" marker is present, the
4239 function call is eligible for tail call optimization. Note that calls may
4240 be marked "tail" even if they do not occur before a <a
4241 href="#i_ret"><tt>ret</tt></a> instruction.
4242 </li>
4243 <li>
4244 <p>The optional "cconv" marker indicates which <a href="#callingconv">calling
4245 convention</a> the call should use. If none is specified, the call defaults
4246 to using C calling conventions.
4247 </li>
4248 <li>
Nick Lewycky93082fc2007-09-08 13:57:50 +0000[diff] [blame]4249 <p>'<tt>ty</tt>': the type of the call instruction itself which is also
4250 the type of the return value. Functions that return no value are marked
4251 <tt><a href="#t_void">void</a></tt>.</p>
4252 </li>
4253 <li>
4254 <p>'<tt>fnty</tt>': shall be the signature of the pointer to function
4255 value being invoked. The argument types must match the types implied by
4256 this signature. This type can be omitted if the function is not varargs
4257 and if the function type does not return a pointer to a function.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4258 </li>
4259 <li>
4260 <p>'<tt>fnptrval</tt>': An LLVM value containing a pointer to a function to
4261 be invoked. In most cases, this is a direct function invocation, but
4262 indirect <tt>call</tt>s are just as possible, calling an arbitrary pointer
4263 to function value.</p>
4264 </li>
4265 <li>
4266 <p>'<tt>function args</tt>': argument list whose types match the
4267 function signature argument types. All arguments must be of
4268 <a href="#t_firstclass">first class</a> type. If the function signature
4269 indicates the function accepts a variable number of arguments, the extra
4270 arguments can be specified.</p>
4271 </li>
4272</ol>
4273
4274<h5>Semantics:</h5>
4275
4276<p>The '<tt>call</tt>' instruction is used to cause control flow to
4277transfer to a specified function, with its incoming arguments bound to
4278the specified values. Upon a '<tt><a href="#i_ret">ret</a></tt>'
4279instruction in the called function, control flow continues with the
4280instruction after the function call, and the return value of the
Chris Lattner5e893ef2008-03-21 17:24:17 +0000[diff] [blame]4281function is bound to the result argument. If the callee returns multiple
4282values then the return values of the function are only accessible through
4283the '<tt><a href="#i_getresult">getresult</a></tt>' instruction.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4284
4285<h5>Example:</h5>
4286
4287<pre>
Nick Lewycky93082fc2007-09-08 13:57:50 +0000[diff] [blame]4288 %retval = call i32 @test(i32 %argc)
Chris Lattner5e893ef2008-03-21 17:24:17 +0000[diff] [blame]4289 call i32 (i8 *, ...)* @printf(i8 * %msg, i32 12, i8 42) <i>; yields i32</i>
4290 %X = tail call i32 @foo() <i>; yields i32</i>
4291 %Y = tail call <a href="#callingconv">fastcc</a> i32 @foo() <i>; yields i32</i>
4292 call void %foo(i8 97 signext)
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]4293
4294 %struct.A = type { i32, i8 }
Chris Lattner5e893ef2008-03-21 17:24:17 +0000[diff] [blame]4295 %r = call %struct.A @foo() <i>; yields { 32, i8 }</i>
4296 %gr = getresult %struct.A %r, 0 <i>; yields i32</i>
4297 %gr1 = getresult %struct.A %r, 1 <i>; yields i8</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4298</pre>
4299
4300</div>
4301
4302<!-- _______________________________________________________________________ -->
4303<div class="doc_subsubsection">
4304 <a name="i_va_arg">'<tt>va_arg</tt>' Instruction</a>
4305</div>
4306
4307<div class="doc_text">
4308
4309<h5>Syntax:</h5>
4310
4311<pre>
4312 &lt;resultval&gt; = va_arg &lt;va_list*&gt; &lt;arglist&gt;, &lt;argty&gt;
4313</pre>
4314
4315<h5>Overview:</h5>
4316
4317<p>The '<tt>va_arg</tt>' instruction is used to access arguments passed through
4318the "variable argument" area of a function call. It is used to implement the
4319<tt>va_arg</tt> macro in C.</p>
4320
4321<h5>Arguments:</h5>
4322
4323<p>This instruction takes a <tt>va_list*</tt> value and the type of
4324the argument. It returns a value of the specified argument type and
4325increments the <tt>va_list</tt> to point to the next argument. The
4326actual type of <tt>va_list</tt> is target specific.</p>
4327
4328<h5>Semantics:</h5>
4329
4330<p>The '<tt>va_arg</tt>' instruction loads an argument of the specified
4331type from the specified <tt>va_list</tt> and causes the
4332<tt>va_list</tt> to point to the next argument. For more information,
4333see the variable argument handling <a href="#int_varargs">Intrinsic
4334Functions</a>.</p>
4335
4336<p>It is legal for this instruction to be called in a function which does not
4337take a variable number of arguments, for example, the <tt>vfprintf</tt>
4338function.</p>
4339
4340<p><tt>va_arg</tt> is an LLVM instruction instead of an <a
4341href="#intrinsics">intrinsic function</a> because it takes a type as an
4342argument.</p>
4343
4344<h5>Example:</h5>
4345
4346<p>See the <a href="#int_varargs">variable argument processing</a> section.</p>
4347
4348</div>
4349
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]4350<!-- _______________________________________________________________________ -->
4351<div class="doc_subsubsection">
4352 <a name="i_getresult">'<tt>getresult</tt>' Instruction</a>
4353</div>
4354
4355<div class="doc_text">
4356
4357<h5>Syntax:</h5>
4358<pre>
Chris Lattneree9da3f2008-03-21 17:20:51 +0000[diff] [blame]4359 &lt;resultval&gt; = getresult &lt;type&gt; &lt;retval&gt;, &lt;index&gt;
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]4360</pre>
Chris Lattneree9da3f2008-03-21 17:20:51 +0000[diff] [blame]4361
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]4362<h5>Overview:</h5>
4363
4364<p> The '<tt>getresult</tt>' instruction is used to extract individual values
Chris Lattneree9da3f2008-03-21 17:20:51 +0000[diff] [blame]4365from a '<tt><a href="#i_call">call</a></tt>'
4366or '<tt><a href="#i_invoke">invoke</a></tt>' instruction that returns multiple
4367results.</p>
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]4368
4369<h5>Arguments:</h5>
4370
Chris Lattneree9da3f2008-03-21 17:20:51 +0000[diff] [blame]4371<p>The '<tt>getresult</tt>' instruction takes a call or invoke value as its
Chris Lattnerd8dd3522008-04-23 04:06:52 +0000[diff] [blame]4372first argument, or an undef value. The value must have <a
4373href="#t_struct">structure type</a>. The second argument is a constant
4374unsigned index value which must be in range for the number of values returned
4375by the call.</p>
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]4376
4377<h5>Semantics:</h5>
4378
Chris Lattneree9da3f2008-03-21 17:20:51 +0000[diff] [blame]4379<p>The '<tt>getresult</tt>' instruction extracts the element identified by
4380'<tt>index</tt>' from the aggregate value.</p>
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]4381
4382<h5>Example:</h5>
4383
4384<pre>
4385 %struct.A = type { i32, i8 }
4386
4387 %r = call %struct.A @foo()
Chris Lattneree9da3f2008-03-21 17:20:51 +0000[diff] [blame]4388 %gr = getresult %struct.A %r, 0 <i>; yields i32:%gr</i>
4389 %gr1 = getresult %struct.A %r, 1 <i>; yields i8:%gr1</i>
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]4390 add i32 %gr, 42
4391 add i8 %gr1, 41
4392</pre>
4393
4394</div>
4395
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4396<!-- *********************************************************************** -->
4397<div class="doc_section"> <a name="intrinsics">Intrinsic Functions</a> </div>
4398<!-- *********************************************************************** -->
4399
4400<div class="doc_text">
4401
4402<p>LLVM supports the notion of an "intrinsic function". These functions have
4403well known names and semantics and are required to follow certain restrictions.
4404Overall, these intrinsics represent an extension mechanism for the LLVM
4405language that does not require changing all of the transformations in LLVM when
4406adding to the language (or the bitcode reader/writer, the parser, etc...).</p>
4407
4408<p>Intrinsic function names must all start with an "<tt>llvm.</tt>" prefix. This
4409prefix is reserved in LLVM for intrinsic names; thus, function names may not
4410begin with this prefix. Intrinsic functions must always be external functions:
4411you cannot define the body of intrinsic functions. Intrinsic functions may
4412only be used in call or invoke instructions: it is illegal to take the address
4413of an intrinsic function. Additionally, because intrinsic functions are part
4414of the LLVM language, it is required if any are added that they be documented
4415here.</p>
4416
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]4417<p>Some intrinsic functions can be overloaded, i.e., the intrinsic represents
4418a family of functions that perform the same operation but on different data
4419types. Because LLVM can represent over 8 million different integer types,
4420overloading is used commonly to allow an intrinsic function to operate on any
4421integer type. One or more of the argument types or the result type can be
4422overloaded to accept any integer type. Argument types may also be defined as
4423exactly matching a previous argument's type or the result type. This allows an
4424intrinsic function which accepts multiple arguments, but needs all of them to
4425be of the same type, to only be overloaded with respect to a single argument or
4426the result.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4427
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]4428<p>Overloaded intrinsics will have the names of its overloaded argument types
4429encoded into its function name, each preceded by a period. Only those types
4430which are overloaded result in a name suffix. Arguments whose type is matched
4431against another type do not. For example, the <tt>llvm.ctpop</tt> function can
4432take an integer of any width and returns an integer of exactly the same integer
4433width. This leads to a family of functions such as
4434<tt>i8 @llvm.ctpop.i8(i8 %val)</tt> and <tt>i29 @llvm.ctpop.i29(i29 %val)</tt>.
4435Only one type, the return type, is overloaded, and only one type suffix is
4436required. Because the argument's type is matched against the return type, it
4437does not require its own name suffix.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4438
4439<p>To learn how to add an intrinsic function, please see the
4440<a href="ExtendingLLVM.html">Extending LLVM Guide</a>.
4441</p>
4442
4443</div>
4444
4445<!-- ======================================================================= -->
4446<div class="doc_subsection">
4447 <a name="int_varargs">Variable Argument Handling Intrinsics</a>
4448</div>
4449
4450<div class="doc_text">
4451
4452<p>Variable argument support is defined in LLVM with the <a
4453 href="#i_va_arg"><tt>va_arg</tt></a> instruction and these three
4454intrinsic functions. These functions are related to the similarly
4455named macros defined in the <tt>&lt;stdarg.h&gt;</tt> header file.</p>
4456
4457<p>All of these functions operate on arguments that use a
4458target-specific value type "<tt>va_list</tt>". The LLVM assembly
4459language reference manual does not define what this type is, so all
4460transformations should be prepared to handle these functions regardless of
4461the type used.</p>
4462
4463<p>This example shows how the <a href="#i_va_arg"><tt>va_arg</tt></a>
4464instruction and the variable argument handling intrinsic functions are
4465used.</p>
4466
4467<div class="doc_code">
4468<pre>
4469define i32 @test(i32 %X, ...) {
4470 ; Initialize variable argument processing
4471 %ap = alloca i8*
4472 %ap2 = bitcast i8** %ap to i8*
4473 call void @llvm.va_start(i8* %ap2)
4474
4475 ; Read a single integer argument
4476 %tmp = va_arg i8** %ap, i32
4477
4478 ; Demonstrate usage of llvm.va_copy and llvm.va_end
4479 %aq = alloca i8*
4480 %aq2 = bitcast i8** %aq to i8*
4481 call void @llvm.va_copy(i8* %aq2, i8* %ap2)
4482 call void @llvm.va_end(i8* %aq2)
4483
4484 ; Stop processing of arguments.
4485 call void @llvm.va_end(i8* %ap2)
4486 ret i32 %tmp
4487}
4488
4489declare void @llvm.va_start(i8*)
4490declare void @llvm.va_copy(i8*, i8*)
4491declare void @llvm.va_end(i8*)
4492</pre>
4493</div>
4494
4495</div>
4496
4497<!-- _______________________________________________________________________ -->
4498<div class="doc_subsubsection">
4499 <a name="int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a>
4500</div>
4501
4502
4503<div class="doc_text">
4504<h5>Syntax:</h5>
4505<pre> declare void %llvm.va_start(i8* &lt;arglist&gt;)<br></pre>
4506<h5>Overview:</h5>
4507<P>The '<tt>llvm.va_start</tt>' intrinsic initializes
4508<tt>*&lt;arglist&gt;</tt> for subsequent use by <tt><a
4509href="#i_va_arg">va_arg</a></tt>.</p>
4510
4511<h5>Arguments:</h5>
4512
4513<P>The argument is a pointer to a <tt>va_list</tt> element to initialize.</p>
4514
4515<h5>Semantics:</h5>
4516
4517<P>The '<tt>llvm.va_start</tt>' intrinsic works just like the <tt>va_start</tt>
4518macro available in C. In a target-dependent way, it initializes the
4519<tt>va_list</tt> element to which the argument points, so that the next call to
4520<tt>va_arg</tt> will produce the first variable argument passed to the function.
4521Unlike the C <tt>va_start</tt> macro, this intrinsic does not need to know the
4522last argument of the function as the compiler can figure that out.</p>
4523
4524</div>
4525
4526<!-- _______________________________________________________________________ -->
4527<div class="doc_subsubsection">
4528 <a name="int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a>
4529</div>
4530
4531<div class="doc_text">
4532<h5>Syntax:</h5>
4533<pre> declare void @llvm.va_end(i8* &lt;arglist&gt;)<br></pre>
4534<h5>Overview:</h5>
4535
4536<p>The '<tt>llvm.va_end</tt>' intrinsic destroys <tt>*&lt;arglist&gt;</tt>,
4537which has been initialized previously with <tt><a href="#int_va_start">llvm.va_start</a></tt>
4538or <tt><a href="#i_va_copy">llvm.va_copy</a></tt>.</p>
4539
4540<h5>Arguments:</h5>
4541
4542<p>The argument is a pointer to a <tt>va_list</tt> to destroy.</p>
4543
4544<h5>Semantics:</h5>
4545
4546<p>The '<tt>llvm.va_end</tt>' intrinsic works just like the <tt>va_end</tt>
4547macro available in C. In a target-dependent way, it destroys the
4548<tt>va_list</tt> element to which the argument points. Calls to <a
4549href="#int_va_start"><tt>llvm.va_start</tt></a> and <a href="#int_va_copy">
4550<tt>llvm.va_copy</tt></a> must be matched exactly with calls to
4551<tt>llvm.va_end</tt>.</p>
4552
4553</div>
4554
4555<!-- _______________________________________________________________________ -->
4556<div class="doc_subsubsection">
4557 <a name="int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a>
4558</div>
4559
4560<div class="doc_text">
4561
4562<h5>Syntax:</h5>
4563
4564<pre>
4565 declare void @llvm.va_copy(i8* &lt;destarglist&gt;, i8* &lt;srcarglist&gt;)
4566</pre>
4567
4568<h5>Overview:</h5>
4569
4570<p>The '<tt>llvm.va_copy</tt>' intrinsic copies the current argument position
4571from the source argument list to the destination argument list.</p>
4572
4573<h5>Arguments:</h5>
4574
4575<p>The first argument is a pointer to a <tt>va_list</tt> element to initialize.
4576The second argument is a pointer to a <tt>va_list</tt> element to copy from.</p>
4577
4578
4579<h5>Semantics:</h5>
4580
4581<p>The '<tt>llvm.va_copy</tt>' intrinsic works just like the <tt>va_copy</tt>
4582macro available in C. In a target-dependent way, it copies the source
4583<tt>va_list</tt> element into the destination <tt>va_list</tt> element. This
4584intrinsic is necessary because the <tt><a href="#int_va_start">
4585llvm.va_start</a></tt> intrinsic may be arbitrarily complex and require, for
4586example, memory allocation.</p>
4587
4588</div>
4589
4590<!-- ======================================================================= -->
4591<div class="doc_subsection">
4592 <a name="int_gc">Accurate Garbage Collection Intrinsics</a>
4593</div>
4594
4595<div class="doc_text">
4596
4597<p>
4598LLVM support for <a href="GarbageCollection.html">Accurate Garbage
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]4599Collection</a> (GC) requires the implementation and generation of these
4600intrinsics.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4601These intrinsics allow identification of <a href="#int_gcroot">GC roots on the
4602stack</a>, as well as garbage collector implementations that require <a
4603href="#int_gcread">read</a> and <a href="#int_gcwrite">write</a> barriers.
4604Front-ends for type-safe garbage collected languages should generate these
4605intrinsics to make use of the LLVM garbage collectors. For more details, see <a
4606href="GarbageCollection.html">Accurate Garbage Collection with LLVM</a>.
4607</p>
Christopher Lambcfe00962007-12-17 01:00:21 +0000[diff] [blame]4608
4609<p>The garbage collection intrinsics only operate on objects in the generic
4610 address space (address space zero).</p>
4611
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4612</div>
4613
4614<!-- _______________________________________________________________________ -->
4615<div class="doc_subsubsection">
4616 <a name="int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a>
4617</div>
4618
4619<div class="doc_text">
4620
4621<h5>Syntax:</h5>
4622
4623<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4624 declare void @llvm.gcroot(i8** %ptrloc, i8* %metadata)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4625</pre>
4626
4627<h5>Overview:</h5>
4628
4629<p>The '<tt>llvm.gcroot</tt>' intrinsic declares the existence of a GC root to
4630the code generator, and allows some metadata to be associated with it.</p>
4631
4632<h5>Arguments:</h5>
4633
4634<p>The first argument specifies the address of a stack object that contains the
4635root pointer. The second pointer (which must be either a constant or a global
4636value address) contains the meta-data to be associated with the root.</p>
4637
4638<h5>Semantics:</h5>
4639
Chris Lattnera7d94ba2008-04-24 05:59:56 +0000[diff] [blame]4640<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]4641location. At compile-time, the code generator generates information to allow
Gordon Henriksen40393542007-12-25 02:31:26 +0000[diff] [blame]4642the runtime to find the pointer at GC safe points. The '<tt>llvm.gcroot</tt>'
4643intrinsic may only be used in a function which <a href="#gc">specifies a GC
4644algorithm</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4645
4646</div>
4647
4648
4649<!-- _______________________________________________________________________ -->
4650<div class="doc_subsubsection">
4651 <a name="int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a>
4652</div>
4653
4654<div class="doc_text">
4655
4656<h5>Syntax:</h5>
4657
4658<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4659 declare i8* @llvm.gcread(i8* %ObjPtr, i8** %Ptr)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4660</pre>
4661
4662<h5>Overview:</h5>
4663
4664<p>The '<tt>llvm.gcread</tt>' intrinsic identifies reads of references from heap
4665locations, allowing garbage collector implementations that require read
4666barriers.</p>
4667
4668<h5>Arguments:</h5>
4669
4670<p>The second argument is the address to read from, which should be an address
4671allocated from the garbage collector. The first object is a pointer to the
4672start of the referenced object, if needed by the language runtime (otherwise
4673null).</p>
4674
4675<h5>Semantics:</h5>
4676
4677<p>The '<tt>llvm.gcread</tt>' intrinsic has the same semantics as a load
4678instruction, but may be replaced with substantially more complex code by the
Gordon Henriksen40393542007-12-25 02:31:26 +0000[diff] [blame]4679garbage collector runtime, as needed. The '<tt>llvm.gcread</tt>' intrinsic
4680may only be used in a function which <a href="#gc">specifies a GC
4681algorithm</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4682
4683</div>
4684
4685
4686<!-- _______________________________________________________________________ -->
4687<div class="doc_subsubsection">
4688 <a name="int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a>
4689</div>
4690
4691<div class="doc_text">
4692
4693<h5>Syntax:</h5>
4694
4695<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4696 declare void @llvm.gcwrite(i8* %P1, i8* %Obj, i8** %P2)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4697</pre>
4698
4699<h5>Overview:</h5>
4700
4701<p>The '<tt>llvm.gcwrite</tt>' intrinsic identifies writes of references to heap
4702locations, allowing garbage collector implementations that require write
4703barriers (such as generational or reference counting collectors).</p>
4704
4705<h5>Arguments:</h5>
4706
4707<p>The first argument is the reference to store, the second is the start of the
4708object to store it to, and the third is the address of the field of Obj to
4709store to. If the runtime does not require a pointer to the object, Obj may be
4710null.</p>
4711
4712<h5>Semantics:</h5>
4713
4714<p>The '<tt>llvm.gcwrite</tt>' intrinsic has the same semantics as a store
4715instruction, but may be replaced with substantially more complex code by the
Gordon Henriksen40393542007-12-25 02:31:26 +0000[diff] [blame]4716garbage collector runtime, as needed. The '<tt>llvm.gcwrite</tt>' intrinsic
4717may only be used in a function which <a href="#gc">specifies a GC
4718algorithm</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4719
4720</div>
4721
4722
4723
4724<!-- ======================================================================= -->
4725<div class="doc_subsection">
4726 <a name="int_codegen">Code Generator Intrinsics</a>
4727</div>
4728
4729<div class="doc_text">
4730<p>
4731These intrinsics are provided by LLVM to expose special features that may only
4732be implemented with code generator support.
4733</p>
4734
4735</div>
4736
4737<!-- _______________________________________________________________________ -->
4738<div class="doc_subsubsection">
4739 <a name="int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a>
4740</div>
4741
4742<div class="doc_text">
4743
4744<h5>Syntax:</h5>
4745<pre>
4746 declare i8 *@llvm.returnaddress(i32 &lt;level&gt;)
4747</pre>
4748
4749<h5>Overview:</h5>
4750
4751<p>
4752The '<tt>llvm.returnaddress</tt>' intrinsic attempts to compute a
4753target-specific value indicating the return address of the current function
4754or one of its callers.
4755</p>
4756
4757<h5>Arguments:</h5>
4758
4759<p>
4760The argument to this intrinsic indicates which function to return the address
4761for. Zero indicates the calling function, one indicates its caller, etc. The
4762argument is <b>required</b> to be a constant integer value.
4763</p>
4764
4765<h5>Semantics:</h5>
4766
4767<p>
4768The '<tt>llvm.returnaddress</tt>' intrinsic either returns a pointer indicating
4769the return address of the specified call frame, or zero if it cannot be
4770identified. The value returned by this intrinsic is likely to be incorrect or 0
4771for arguments other than zero, so it should only be used for debugging purposes.
4772</p>
4773
4774<p>
4775Note that calling this intrinsic does not prevent function inlining or other
4776aggressive transformations, so the value returned may not be that of the obvious
4777source-language caller.
4778</p>
4779</div>
4780
4781
4782<!-- _______________________________________________________________________ -->
4783<div class="doc_subsubsection">
4784 <a name="int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a>
4785</div>
4786
4787<div class="doc_text">
4788
4789<h5>Syntax:</h5>
4790<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4791 declare i8 *@llvm.frameaddress(i32 &lt;level&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4792</pre>
4793
4794<h5>Overview:</h5>
4795
4796<p>
4797The '<tt>llvm.frameaddress</tt>' intrinsic attempts to return the
4798target-specific frame pointer value for the specified stack frame.
4799</p>
4800
4801<h5>Arguments:</h5>
4802
4803<p>
4804The argument to this intrinsic indicates which function to return the frame
4805pointer for. Zero indicates the calling function, one indicates its caller,
4806etc. The argument is <b>required</b> to be a constant integer value.
4807</p>
4808
4809<h5>Semantics:</h5>
4810
4811<p>
4812The '<tt>llvm.frameaddress</tt>' intrinsic either returns a pointer indicating
4813the frame address of the specified call frame, or zero if it cannot be
4814identified. The value returned by this intrinsic is likely to be incorrect or 0
4815for arguments other than zero, so it should only be used for debugging purposes.
4816</p>
4817
4818<p>
4819Note that calling this intrinsic does not prevent function inlining or other
4820aggressive transformations, so the value returned may not be that of the obvious
4821source-language caller.
4822</p>
4823</div>
4824
4825<!-- _______________________________________________________________________ -->
4826<div class="doc_subsubsection">
4827 <a name="int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a>
4828</div>
4829
4830<div class="doc_text">
4831
4832<h5>Syntax:</h5>
4833<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4834 declare i8 *@llvm.stacksave()
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4835</pre>
4836
4837<h5>Overview:</h5>
4838
4839<p>
4840The '<tt>llvm.stacksave</tt>' intrinsic is used to remember the current state of
4841the function stack, for use with <a href="#int_stackrestore">
4842<tt>llvm.stackrestore</tt></a>. This is useful for implementing language
4843features like scoped automatic variable sized arrays in C99.
4844</p>
4845
4846<h5>Semantics:</h5>
4847
4848<p>
4849This intrinsic returns a opaque pointer value that can be passed to <a
4850href="#int_stackrestore"><tt>llvm.stackrestore</tt></a>. When an
4851<tt>llvm.stackrestore</tt> intrinsic is executed with a value saved from
4852<tt>llvm.stacksave</tt>, it effectively restores the state of the stack to the
4853state it was in when the <tt>llvm.stacksave</tt> intrinsic executed. In
4854practice, this pops any <a href="#i_alloca">alloca</a> blocks from the stack
4855that were allocated after the <tt>llvm.stacksave</tt> was executed.
4856</p>
4857
4858</div>
4859
4860<!-- _______________________________________________________________________ -->
4861<div class="doc_subsubsection">
4862 <a name="int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a>
4863</div>
4864
4865<div class="doc_text">
4866
4867<h5>Syntax:</h5>
4868<pre>
4869 declare void @llvm.stackrestore(i8 * %ptr)
4870</pre>
4871
4872<h5>Overview:</h5>
4873
4874<p>
4875The '<tt>llvm.stackrestore</tt>' intrinsic is used to restore the state of
4876the function stack to the state it was in when the corresponding <a
4877href="#int_stacksave"><tt>llvm.stacksave</tt></a> intrinsic executed. This is
4878useful for implementing language features like scoped automatic variable sized
4879arrays in C99.
4880</p>
4881
4882<h5>Semantics:</h5>
4883
4884<p>
4885See the description for <a href="#int_stacksave"><tt>llvm.stacksave</tt></a>.
4886</p>
4887
4888</div>
4889
4890
4891<!-- _______________________________________________________________________ -->
4892<div class="doc_subsubsection">
4893 <a name="int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a>
4894</div>
4895
4896<div class="doc_text">
4897
4898<h5>Syntax:</h5>
4899<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4900 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]4901</pre>
4902
4903<h5>Overview:</h5>
4904
4905
4906<p>
4907The '<tt>llvm.prefetch</tt>' intrinsic is a hint to the code generator to insert
4908a prefetch instruction if supported; otherwise, it is a noop. Prefetches have
4909no
4910effect on the behavior of the program but can change its performance
4911characteristics.
4912</p>
4913
4914<h5>Arguments:</h5>
4915
4916<p>
4917<tt>address</tt> is the address to be prefetched, <tt>rw</tt> is the specifier
4918determining if the fetch should be for a read (0) or write (1), and
4919<tt>locality</tt> is a temporal locality specifier ranging from (0) - no
4920locality, to (3) - extremely local keep in cache. The <tt>rw</tt> and
4921<tt>locality</tt> arguments must be constant integers.
4922</p>
4923
4924<h5>Semantics:</h5>
4925
4926<p>
4927This intrinsic does not modify the behavior of the program. In particular,
4928prefetches cannot trap and do not produce a value. On targets that support this
4929intrinsic, the prefetch can provide hints to the processor cache for better
4930performance.
4931</p>
4932
4933</div>
4934
4935<!-- _______________________________________________________________________ -->
4936<div class="doc_subsubsection">
4937 <a name="int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a>
4938</div>
4939
4940<div class="doc_text">
4941
4942<h5>Syntax:</h5>
4943<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4944 declare void @llvm.pcmarker(i32 &lt;id&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4945</pre>
4946
4947<h5>Overview:</h5>
4948
4949
4950<p>
4951The '<tt>llvm.pcmarker</tt>' intrinsic is a method to export a Program Counter
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]4952(PC) in a region of
4953code to simulators and other tools. The method is target specific, but it is
4954expected that the marker will use exported symbols to transmit the PC of the
4955marker.
4956The marker makes no guarantees that it will remain with any specific instruction
4957after optimizations. It is possible that the presence of a marker will inhibit
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4958optimizations. The intended use is to be inserted after optimizations to allow
4959correlations of simulation runs.
4960</p>
4961
4962<h5>Arguments:</h5>
4963
4964<p>
4965<tt>id</tt> is a numerical id identifying the marker.
4966</p>
4967
4968<h5>Semantics:</h5>
4969
4970<p>
4971This intrinsic does not modify the behavior of the program. Backends that do not
4972support this intrinisic may ignore it.
4973</p>
4974
4975</div>
4976
4977<!-- _______________________________________________________________________ -->
4978<div class="doc_subsubsection">
4979 <a name="int_readcyclecounter">'<tt>llvm.readcyclecounter</tt>' Intrinsic</a>
4980</div>
4981
4982<div class="doc_text">
4983
4984<h5>Syntax:</h5>
4985<pre>
4986 declare i64 @llvm.readcyclecounter( )
4987</pre>
4988
4989<h5>Overview:</h5>
4990
4991
4992<p>
4993The '<tt>llvm.readcyclecounter</tt>' intrinsic provides access to the cycle
4994counter register (or similar low latency, high accuracy clocks) on those targets
4995that support it. On X86, it should map to RDTSC. On Alpha, it should map to RPCC.
4996As the backing counters overflow quickly (on the order of 9 seconds on alpha), this
4997should only be used for small timings.
4998</p>
4999
5000<h5>Semantics:</h5>
5001
5002<p>
5003When directly supported, reading the cycle counter should not modify any memory.
5004Implementations are allowed to either return a application specific value or a
5005system wide value. On backends without support, this is lowered to a constant 0.
5006</p>
5007
5008</div>
5009
5010<!-- ======================================================================= -->
5011<div class="doc_subsection">
5012 <a name="int_libc">Standard C Library Intrinsics</a>
5013</div>
5014
5015<div class="doc_text">
5016<p>
5017LLVM provides intrinsics for a few important standard C library functions.
5018These intrinsics allow source-language front-ends to pass information about the
5019alignment of the pointer arguments to the code generator, providing opportunity
5020for more efficient code generation.
5021</p>
5022
5023</div>
5024
5025<!-- _______________________________________________________________________ -->
5026<div class="doc_subsubsection">
5027 <a name="int_memcpy">'<tt>llvm.memcpy</tt>' Intrinsic</a>
5028</div>
5029
5030<div class="doc_text">
5031
5032<h5>Syntax:</h5>
5033<pre>
5034 declare void @llvm.memcpy.i32(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5035 i32 &lt;len&gt;, i32 &lt;align&gt;)
5036 declare void @llvm.memcpy.i64(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5037 i64 &lt;len&gt;, i32 &lt;align&gt;)
5038</pre>
5039
5040<h5>Overview:</h5>
5041
5042<p>
5043The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the source
5044location to the destination location.
5045</p>
5046
5047<p>
5048Note that, unlike the standard libc function, the <tt>llvm.memcpy.*</tt>
5049intrinsics do not return a value, and takes an extra alignment argument.
5050</p>
5051
5052<h5>Arguments:</h5>
5053
5054<p>
5055The first argument is a pointer to the destination, the second is a pointer to
5056the source. The third argument is an integer argument
5057specifying the number of bytes to copy, and the fourth argument is the alignment
5058of the source and destination locations.
5059</p>
5060
5061<p>
5062If the call to this intrinisic has an alignment value that is not 0 or 1, then
5063the caller guarantees that both the source and destination pointers are aligned
5064to that boundary.
5065</p>
5066
5067<h5>Semantics:</h5>
5068
5069<p>
5070The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the source
5071location to the destination location, which are not allowed to overlap. It
5072copies "len" bytes of memory over. If the argument is known to be aligned to
5073some boundary, this can be specified as the fourth argument, otherwise it should
5074be set to 0 or 1.
5075</p>
5076</div>
5077
5078
5079<!-- _______________________________________________________________________ -->
5080<div class="doc_subsubsection">
5081 <a name="int_memmove">'<tt>llvm.memmove</tt>' Intrinsic</a>
5082</div>
5083
5084<div class="doc_text">
5085
5086<h5>Syntax:</h5>
5087<pre>
5088 declare void @llvm.memmove.i32(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5089 i32 &lt;len&gt;, i32 &lt;align&gt;)
5090 declare void @llvm.memmove.i64(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5091 i64 &lt;len&gt;, i32 &lt;align&gt;)
5092</pre>
5093
5094<h5>Overview:</h5>
5095
5096<p>
5097The '<tt>llvm.memmove.*</tt>' intrinsics move a block of memory from the source
5098location to the destination location. It is similar to the
Chris Lattnerdba16ea2008-01-06 19:51:52 +0000[diff] [blame]5099'<tt>llvm.memcpy</tt>' intrinsic but allows the two memory locations to overlap.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5100</p>
5101
5102<p>
5103Note that, unlike the standard libc function, the <tt>llvm.memmove.*</tt>
5104intrinsics do not return a value, and takes an extra alignment argument.
5105</p>
5106
5107<h5>Arguments:</h5>
5108
5109<p>
5110The first argument is a pointer to the destination, the second is a pointer to
5111the source. The third argument is an integer argument
5112specifying the number of bytes to copy, and the fourth argument is the alignment
5113of the source and destination locations.
5114</p>
5115
5116<p>
5117If the call to this intrinisic has an alignment value that is not 0 or 1, then
5118the caller guarantees that the source and destination pointers are aligned to
5119that boundary.
5120</p>
5121
5122<h5>Semantics:</h5>
5123
5124<p>
5125The '<tt>llvm.memmove.*</tt>' intrinsics copy a block of memory from the source
5126location to the destination location, which may overlap. It
5127copies "len" bytes of memory over. If the argument is known to be aligned to
5128some boundary, this can be specified as the fourth argument, otherwise it should
5129be set to 0 or 1.
5130</p>
5131</div>
5132
5133
5134<!-- _______________________________________________________________________ -->
5135<div class="doc_subsubsection">
5136 <a name="int_memset">'<tt>llvm.memset.*</tt>' Intrinsics</a>
5137</div>
5138
5139<div class="doc_text">
5140
5141<h5>Syntax:</h5>
5142<pre>
5143 declare void @llvm.memset.i32(i8 * &lt;dest&gt;, i8 &lt;val&gt;,
5144 i32 &lt;len&gt;, i32 &lt;align&gt;)
5145 declare void @llvm.memset.i64(i8 * &lt;dest&gt;, i8 &lt;val&gt;,
5146 i64 &lt;len&gt;, i32 &lt;align&gt;)
5147</pre>
5148
5149<h5>Overview:</h5>
5150
5151<p>
5152The '<tt>llvm.memset.*</tt>' intrinsics fill a block of memory with a particular
5153byte value.
5154</p>
5155
5156<p>
5157Note that, unlike the standard libc function, the <tt>llvm.memset</tt> intrinsic
5158does not return a value, and takes an extra alignment argument.
5159</p>
5160
5161<h5>Arguments:</h5>
5162
5163<p>
5164The first argument is a pointer to the destination to fill, the second is the
5165byte value to fill it with, the third argument is an integer
5166argument specifying the number of bytes to fill, and the fourth argument is the
5167known alignment of destination location.
5168</p>
5169
5170<p>
5171If the call to this intrinisic has an alignment value that is not 0 or 1, then
5172the caller guarantees that the destination pointer is aligned to that boundary.
5173</p>
5174
5175<h5>Semantics:</h5>
5176
5177<p>
5178The '<tt>llvm.memset.*</tt>' intrinsics fill "len" bytes of memory starting at
5179the
5180destination location. If the argument is known to be aligned to some boundary,
5181this can be specified as the fourth argument, otherwise it should be set to 0 or
51821.
5183</p>
5184</div>
5185
5186
5187<!-- _______________________________________________________________________ -->
5188<div class="doc_subsubsection">
5189 <a name="int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a>
5190</div>
5191
5192<div class="doc_text">
5193
5194<h5>Syntax:</h5>
Dale Johannesenf9adbb62007-10-02 17:47:38 +0000[diff] [blame]5195<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]5196floating point or vector of floating point type. Not all targets support all
5197types however.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5198<pre>
Dale Johannesenf9adbb62007-10-02 17:47:38 +0000[diff] [blame]5199 declare float @llvm.sqrt.f32(float %Val)
5200 declare double @llvm.sqrt.f64(double %Val)
5201 declare x86_fp80 @llvm.sqrt.f80(x86_fp80 %Val)
5202 declare fp128 @llvm.sqrt.f128(fp128 %Val)
5203 declare ppc_fp128 @llvm.sqrt.ppcf128(ppc_fp128 %Val)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5204</pre>
5205
5206<h5>Overview:</h5>
5207
5208<p>
5209The '<tt>llvm.sqrt</tt>' intrinsics return the sqrt of the specified operand,
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5210returning the same value as the libm '<tt>sqrt</tt>' functions would. Unlike
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5211<tt>sqrt</tt> in libm, however, <tt>llvm.sqrt</tt> has undefined behavior for
Chris Lattnerf914a002008-01-29 07:00:44 +0000[diff] [blame]5212negative numbers other than -0.0 (which allows for better optimization, because
5213there is no need to worry about errno being set). <tt>llvm.sqrt(-0.0)</tt> is
5214defined to return -0.0 like IEEE sqrt.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5215</p>
5216
5217<h5>Arguments:</h5>
5218
5219<p>
5220The argument and return value are floating point numbers of the same type.
5221</p>
5222
5223<h5>Semantics:</h5>
5224
5225<p>
5226This function returns the sqrt of the specified operand if it is a nonnegative
5227floating point number.
5228</p>
5229</div>
5230
5231<!-- _______________________________________________________________________ -->
5232<div class="doc_subsubsection">
5233 <a name="int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a>
5234</div>
5235
5236<div class="doc_text">
5237
5238<h5>Syntax:</h5>
Dale Johannesenf9adbb62007-10-02 17:47:38 +0000[diff] [blame]5239<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]5240floating point or vector of floating point type. Not all targets support all
5241types however.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5242<pre>
Dale Johannesenf9adbb62007-10-02 17:47:38 +0000[diff] [blame]5243 declare float @llvm.powi.f32(float %Val, i32 %power)
5244 declare double @llvm.powi.f64(double %Val, i32 %power)
5245 declare x86_fp80 @llvm.powi.f80(x86_fp80 %Val, i32 %power)
5246 declare fp128 @llvm.powi.f128(fp128 %Val, i32 %power)
5247 declare ppc_fp128 @llvm.powi.ppcf128(ppc_fp128 %Val, i32 %power)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5248</pre>
5249
5250<h5>Overview:</h5>
5251
5252<p>
5253The '<tt>llvm.powi.*</tt>' intrinsics return the first operand raised to the
5254specified (positive or negative) power. The order of evaluation of
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5255multiplications is not defined. When a vector of floating point type is
5256used, the second argument remains a scalar integer value.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5257</p>
5258
5259<h5>Arguments:</h5>
5260
5261<p>
5262The second argument is an integer power, and the first is a value to raise to
5263that power.
5264</p>
5265
5266<h5>Semantics:</h5>
5267
5268<p>
5269This function returns the first value raised to the second power with an
5270unspecified sequence of rounding operations.</p>
5271</div>
5272
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5273<!-- _______________________________________________________________________ -->
5274<div class="doc_subsubsection">
5275 <a name="int_sin">'<tt>llvm.sin.*</tt>' Intrinsic</a>
5276</div>
5277
5278<div class="doc_text">
5279
5280<h5>Syntax:</h5>
5281<p>This is an overloaded intrinsic. You can use <tt>llvm.sin</tt> on any
5282floating point or vector of floating point type. Not all targets support all
5283types however.
5284<pre>
5285 declare float @llvm.sin.f32(float %Val)
5286 declare double @llvm.sin.f64(double %Val)
5287 declare x86_fp80 @llvm.sin.f80(x86_fp80 %Val)
5288 declare fp128 @llvm.sin.f128(fp128 %Val)
5289 declare ppc_fp128 @llvm.sin.ppcf128(ppc_fp128 %Val)
5290</pre>
5291
5292<h5>Overview:</h5>
5293
5294<p>
5295The '<tt>llvm.sin.*</tt>' intrinsics return the sine of the operand.
5296</p>
5297
5298<h5>Arguments:</h5>
5299
5300<p>
5301The argument and return value are floating point numbers of the same type.
5302</p>
5303
5304<h5>Semantics:</h5>
5305
5306<p>
5307This function returns the sine of the specified operand, returning the
5308same values as the libm <tt>sin</tt> functions would, and handles error
Dan Gohmaneaba92e2007-10-17 18:05:13 +0000[diff] [blame]5309conditions in the same way.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5310</div>
5311
5312<!-- _______________________________________________________________________ -->
5313<div class="doc_subsubsection">
5314 <a name="int_cos">'<tt>llvm.cos.*</tt>' Intrinsic</a>
5315</div>
5316
5317<div class="doc_text">
5318
5319<h5>Syntax:</h5>
5320<p>This is an overloaded intrinsic. You can use <tt>llvm.cos</tt> on any
5321floating point or vector of floating point type. Not all targets support all
5322types however.
5323<pre>
5324 declare float @llvm.cos.f32(float %Val)
5325 declare double @llvm.cos.f64(double %Val)
5326 declare x86_fp80 @llvm.cos.f80(x86_fp80 %Val)
5327 declare fp128 @llvm.cos.f128(fp128 %Val)
5328 declare ppc_fp128 @llvm.cos.ppcf128(ppc_fp128 %Val)
5329</pre>
5330
5331<h5>Overview:</h5>
5332
5333<p>
5334The '<tt>llvm.cos.*</tt>' intrinsics return the cosine of the operand.
5335</p>
5336
5337<h5>Arguments:</h5>
5338
5339<p>
5340The argument and return value are floating point numbers of the same type.
5341</p>
5342
5343<h5>Semantics:</h5>
5344
5345<p>
5346This function returns the cosine of the specified operand, returning the
5347same values as the libm <tt>cos</tt> functions would, and handles error
Dan Gohmaneaba92e2007-10-17 18:05:13 +0000[diff] [blame]5348conditions in the same way.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5349</div>
5350
5351<!-- _______________________________________________________________________ -->
5352<div class="doc_subsubsection">
5353 <a name="int_pow">'<tt>llvm.pow.*</tt>' Intrinsic</a>
5354</div>
5355
5356<div class="doc_text">
5357
5358<h5>Syntax:</h5>
5359<p>This is an overloaded intrinsic. You can use <tt>llvm.pow</tt> on any
5360floating point or vector of floating point type. Not all targets support all
5361types however.
5362<pre>
5363 declare float @llvm.pow.f32(float %Val, float %Power)
5364 declare double @llvm.pow.f64(double %Val, double %Power)
5365 declare x86_fp80 @llvm.pow.f80(x86_fp80 %Val, x86_fp80 %Power)
5366 declare fp128 @llvm.pow.f128(fp128 %Val, fp128 %Power)
5367 declare ppc_fp128 @llvm.pow.ppcf128(ppc_fp128 %Val, ppc_fp128 Power)
5368</pre>
5369
5370<h5>Overview:</h5>
5371
5372<p>
5373The '<tt>llvm.pow.*</tt>' intrinsics return the first operand raised to the
5374specified (positive or negative) power.
5375</p>
5376
5377<h5>Arguments:</h5>
5378
5379<p>
5380The second argument is a floating point power, and the first is a value to
5381raise to that power.
5382</p>
5383
5384<h5>Semantics:</h5>
5385
5386<p>
5387This function returns the first value raised to the second power,
5388returning the
5389same values as the libm <tt>pow</tt> functions would, and handles error
Dan Gohmaneaba92e2007-10-17 18:05:13 +0000[diff] [blame]5390conditions in the same way.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5391</div>
5392
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5393
5394<!-- ======================================================================= -->
5395<div class="doc_subsection">
5396 <a name="int_manip">Bit Manipulation Intrinsics</a>
5397</div>
5398
5399<div class="doc_text">
5400<p>
5401LLVM provides intrinsics for a few important bit manipulation operations.
5402These allow efficient code generation for some algorithms.
5403</p>
5404
5405</div>
5406
5407<!-- _______________________________________________________________________ -->
5408<div class="doc_subsubsection">
5409 <a name="int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a>
5410</div>
5411
5412<div class="doc_text">
5413
5414<h5>Syntax:</h5>
5415<p>This is an overloaded intrinsic function. You can use bswap on any integer
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5416type that is an even number of bytes (i.e. BitWidth % 16 == 0).
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5417<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5418 declare i16 @llvm.bswap.i16(i16 &lt;id&gt;)
5419 declare i32 @llvm.bswap.i32(i32 &lt;id&gt;)
5420 declare i64 @llvm.bswap.i64(i64 &lt;id&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5421</pre>
5422
5423<h5>Overview:</h5>
5424
5425<p>
5426The '<tt>llvm.bswap</tt>' family of intrinsics is used to byte swap integer
5427values with an even number of bytes (positive multiple of 16 bits). These are
5428useful for performing operations on data that is not in the target's native
5429byte order.
5430</p>
5431
5432<h5>Semantics:</h5>
5433
5434<p>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5435The <tt>llvm.bswap.i16</tt> intrinsic returns an i16 value that has the high
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5436and low byte of the input i16 swapped. Similarly, the <tt>llvm.bswap.i32</tt>
5437intrinsic returns an i32 value that has the four bytes of the input i32
5438swapped, 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]5439i32 will have its bytes in 3, 2, 1, 0 order. The <tt>llvm.bswap.i48</tt>,
5440<tt>llvm.bswap.i64</tt> and other intrinsics extend this concept to
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5441additional even-byte lengths (6 bytes, 8 bytes and more, respectively).
5442</p>
5443
5444</div>
5445
5446<!-- _______________________________________________________________________ -->
5447<div class="doc_subsubsection">
5448 <a name="int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic</a>
5449</div>
5450
5451<div class="doc_text">
5452
5453<h5>Syntax:</h5>
5454<p>This is an overloaded intrinsic. You can use llvm.ctpop on any integer bit
5455width. Not all targets support all bit widths however.
5456<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5457 declare i8 @llvm.ctpop.i8 (i8 &lt;src&gt;)
5458 declare i16 @llvm.ctpop.i16(i16 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5459 declare i32 @llvm.ctpop.i32(i32 &lt;src&gt;)
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5460 declare i64 @llvm.ctpop.i64(i64 &lt;src&gt;)
5461 declare i256 @llvm.ctpop.i256(i256 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5462</pre>
5463
5464<h5>Overview:</h5>
5465
5466<p>
5467The '<tt>llvm.ctpop</tt>' family of intrinsics counts the number of bits set in a
5468value.
5469</p>
5470
5471<h5>Arguments:</h5>
5472
5473<p>
5474The only argument is the value to be counted. The argument may be of any
5475integer type. The return type must match the argument type.
5476</p>
5477
5478<h5>Semantics:</h5>
5479
5480<p>
5481The '<tt>llvm.ctpop</tt>' intrinsic counts the 1's in a variable.
5482</p>
5483</div>
5484
5485<!-- _______________________________________________________________________ -->
5486<div class="doc_subsubsection">
5487 <a name="int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic</a>
5488</div>
5489
5490<div class="doc_text">
5491
5492<h5>Syntax:</h5>
5493<p>This is an overloaded intrinsic. You can use <tt>llvm.ctlz</tt> on any
5494integer bit width. Not all targets support all bit widths however.
5495<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5496 declare i8 @llvm.ctlz.i8 (i8 &lt;src&gt;)
5497 declare i16 @llvm.ctlz.i16(i16 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5498 declare i32 @llvm.ctlz.i32(i32 &lt;src&gt;)
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5499 declare i64 @llvm.ctlz.i64(i64 &lt;src&gt;)
5500 declare i256 @llvm.ctlz.i256(i256 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5501</pre>
5502
5503<h5>Overview:</h5>
5504
5505<p>
5506The '<tt>llvm.ctlz</tt>' family of intrinsic functions counts the number of
5507leading zeros in a variable.
5508</p>
5509
5510<h5>Arguments:</h5>
5511
5512<p>
5513The only argument is the value to be counted. The argument may be of any
5514integer type. The return type must match the argument type.
5515</p>
5516
5517<h5>Semantics:</h5>
5518
5519<p>
5520The '<tt>llvm.ctlz</tt>' intrinsic counts the leading (most significant) zeros
5521in a variable. If the src == 0 then the result is the size in bits of the type
5522of src. For example, <tt>llvm.ctlz(i32 2) = 30</tt>.
5523</p>
5524</div>
5525
5526
5527
5528<!-- _______________________________________________________________________ -->
5529<div class="doc_subsubsection">
5530 <a name="int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic</a>
5531</div>
5532
5533<div class="doc_text">
5534
5535<h5>Syntax:</h5>
5536<p>This is an overloaded intrinsic. You can use <tt>llvm.cttz</tt> on any
5537integer bit width. Not all targets support all bit widths however.
5538<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5539 declare i8 @llvm.cttz.i8 (i8 &lt;src&gt;)
5540 declare i16 @llvm.cttz.i16(i16 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5541 declare i32 @llvm.cttz.i32(i32 &lt;src&gt;)
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5542 declare i64 @llvm.cttz.i64(i64 &lt;src&gt;)
5543 declare i256 @llvm.cttz.i256(i256 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5544</pre>
5545
5546<h5>Overview:</h5>
5547
5548<p>
5549The '<tt>llvm.cttz</tt>' family of intrinsic functions counts the number of
5550trailing zeros.
5551</p>
5552
5553<h5>Arguments:</h5>
5554
5555<p>
5556The only argument is the value to be counted. The argument may be of any
5557integer type. The return type must match the argument type.
5558</p>
5559
5560<h5>Semantics:</h5>
5561
5562<p>
5563The '<tt>llvm.cttz</tt>' intrinsic counts the trailing (least significant) zeros
5564in a variable. If the src == 0 then the result is the size in bits of the type
5565of src. For example, <tt>llvm.cttz(2) = 1</tt>.
5566</p>
5567</div>
5568
5569<!-- _______________________________________________________________________ -->
5570<div class="doc_subsubsection">
5571 <a name="int_part_select">'<tt>llvm.part.select.*</tt>' Intrinsic</a>
5572</div>
5573
5574<div class="doc_text">
5575
5576<h5>Syntax:</h5>
5577<p>This is an overloaded intrinsic. You can use <tt>llvm.part.select</tt>
5578on any integer bit width.
5579<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5580 declare i17 @llvm.part.select.i17 (i17 %val, i32 %loBit, i32 %hiBit)
5581 declare i29 @llvm.part.select.i29 (i29 %val, i32 %loBit, i32 %hiBit)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5582</pre>
5583
5584<h5>Overview:</h5>
5585<p>The '<tt>llvm.part.select</tt>' family of intrinsic functions selects a
5586range of bits from an integer value and returns them in the same bit width as
5587the original value.</p>
5588
5589<h5>Arguments:</h5>
5590<p>The first argument, <tt>%val</tt> and the result may be integer types of
5591any bit width but they must have the same bit width. The second and third
5592arguments must be <tt>i32</tt> type since they specify only a bit index.</p>
5593
5594<h5>Semantics:</h5>
5595<p>The operation of the '<tt>llvm.part.select</tt>' intrinsic has two modes
5596of operation: forwards and reverse. If <tt>%loBit</tt> is greater than
5597<tt>%hiBits</tt> then the intrinsic operates in reverse mode. Otherwise it
5598operates in forward mode.</p>
5599<p>In forward mode, this intrinsic is the equivalent of shifting <tt>%val</tt>
5600right by <tt>%loBit</tt> bits and then ANDing it with a mask with
5601only the <tt>%hiBit - %loBit</tt> bits set, as follows:</p>
5602<ol>
5603 <li>The <tt>%val</tt> is shifted right (LSHR) by the number of bits specified
5604 by <tt>%loBits</tt>. This normalizes the value to the low order bits.</li>
5605 <li>The <tt>%loBits</tt> value is subtracted from the <tt>%hiBits</tt> value
5606 to determine the number of bits to retain.</li>
5607 <li>A mask of the retained bits is created by shifting a -1 value.</li>
5608 <li>The mask is ANDed with <tt>%val</tt> to produce the result.
5609</ol>
5610<p>In reverse mode, a similar computation is made except that the bits are
5611returned in the reverse order. So, for example, if <tt>X</tt> has the value
5612<tt>i16 0x0ACF (101011001111)</tt> and we apply
5613<tt>part.select(i16 X, 8, 3)</tt> to it, we get back the value
5614<tt>i16 0x0026 (000000100110)</tt>.</p>
5615</div>
5616
5617<div class="doc_subsubsection">
5618 <a name="int_part_set">'<tt>llvm.part.set.*</tt>' Intrinsic</a>
5619</div>
5620
5621<div class="doc_text">
5622
5623<h5>Syntax:</h5>
5624<p>This is an overloaded intrinsic. You can use <tt>llvm.part.set</tt>
5625on any integer bit width.
5626<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5627 declare i17 @llvm.part.set.i17.i9 (i17 %val, i9 %repl, i32 %lo, i32 %hi)
5628 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]5629</pre>
5630
5631<h5>Overview:</h5>
5632<p>The '<tt>llvm.part.set</tt>' family of intrinsic functions replaces a range
5633of bits in an integer value with another integer value. It returns the integer
5634with the replaced bits.</p>
5635
5636<h5>Arguments:</h5>
5637<p>The first argument, <tt>%val</tt> and the result may be integer types of
5638any bit width but they must have the same bit width. <tt>%val</tt> is the value
5639whose bits will be replaced. The second argument, <tt>%repl</tt> may be an
5640integer of any bit width. The third and fourth arguments must be <tt>i32</tt>
5641type since they specify only a bit index.</p>
5642
5643<h5>Semantics:</h5>
5644<p>The operation of the '<tt>llvm.part.set</tt>' intrinsic has two modes
5645of operation: forwards and reverse. If <tt>%lo</tt> is greater than
5646<tt>%hi</tt> then the intrinsic operates in reverse mode. Otherwise it
5647operates in forward mode.</p>
5648<p>For both modes, the <tt>%repl</tt> value is prepared for use by either
5649truncating it down to the size of the replacement area or zero extending it
5650up to that size.</p>
5651<p>In forward mode, the bits between <tt>%lo</tt> and <tt>%hi</tt> (inclusive)
5652are replaced with corresponding bits from <tt>%repl</tt>. That is the 0th bit
5653in <tt>%repl</tt> replaces the <tt>%lo</tt>th bit in <tt>%val</tt> and etc. up
5654to the <tt>%hi</tt>th bit.
5655<p>In reverse mode, a similar computation is made except that the bits are
5656reversed. That is, the <tt>0</tt>th bit in <tt>%repl</tt> replaces the
5657<tt>%hi</tt> bit in <tt>%val</tt> and etc. down to the <tt>%lo</tt>th bit.
5658<h5>Examples:</h5>
5659<pre>
5660 llvm.part.set(0xFFFF, 0, 4, 7) -&gt; 0xFF0F
5661 llvm.part.set(0xFFFF, 0, 7, 4) -&gt; 0xFF0F
5662 llvm.part.set(0xFFFF, 1, 7, 4) -&gt; 0xFF8F
5663 llvm.part.set(0xFFFF, F, 8, 3) -&gt; 0xFFE7
5664 llvm.part.set(0xFFFF, 0, 3, 8) -&gt; 0xFE07
5665</pre>
5666</div>
5667
5668<!-- ======================================================================= -->
5669<div class="doc_subsection">
5670 <a name="int_debugger">Debugger Intrinsics</a>
5671</div>
5672
5673<div class="doc_text">
5674<p>
5675The LLVM debugger intrinsics (which all start with <tt>llvm.dbg.</tt> prefix),
5676are described in the <a
5677href="SourceLevelDebugging.html#format_common_intrinsics">LLVM Source Level
5678Debugging</a> document.
5679</p>
5680</div>
5681
5682
5683<!-- ======================================================================= -->
5684<div class="doc_subsection">
5685 <a name="int_eh">Exception Handling Intrinsics</a>
5686</div>
5687
5688<div class="doc_text">
5689<p> The LLVM exception handling intrinsics (which all start with
5690<tt>llvm.eh.</tt> prefix), are described in the <a
5691href="ExceptionHandling.html#format_common_intrinsics">LLVM Exception
5692Handling</a> document. </p>
5693</div>
5694
5695<!-- ======================================================================= -->
5696<div class="doc_subsection">
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5697 <a name="int_trampoline">Trampoline Intrinsic</a>
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5698</div>
5699
5700<div class="doc_text">
5701<p>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5702 This intrinsic makes it possible to excise one parameter, marked with
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5703 the <tt>nest</tt> attribute, from a function. The result is a callable
5704 function pointer lacking the nest parameter - the caller does not need
5705 to provide a value for it. Instead, the value to use is stored in
5706 advance in a "trampoline", a block of memory usually allocated
5707 on the stack, which also contains code to splice the nest value into the
5708 argument list. This is used to implement the GCC nested function address
5709 extension.
5710</p>
5711<p>
5712 For example, if the function is
5713 <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]5714 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]5715<pre>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5716 %tramp = alloca [10 x i8], align 4 ; size and alignment only correct for X86
5717 %tramp1 = getelementptr [10 x i8]* %tramp, i32 0, i32 0
5718 %p = call i8* @llvm.init.trampoline( i8* %tramp1, i8* bitcast (i32 (i8* nest , i32, i32)* @f to i8*), i8* %nval )
5719 %fp = bitcast i8* %p to i32 (i32, i32)*
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5720</pre>
Bill Wendlinge262b4c2007-09-22 09:23:55 +0000[diff] [blame]5721 <p>The call <tt>%val = call i32 %fp( i32 %x, i32 %y )</tt> is then equivalent
5722 to <tt>%val = call i32 %f( i8* %nval, i32 %x, i32 %y )</tt>.</p>
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5723</div>
5724
5725<!-- _______________________________________________________________________ -->
5726<div class="doc_subsubsection">
5727 <a name="int_it">'<tt>llvm.init.trampoline</tt>' Intrinsic</a>
5728</div>
5729<div class="doc_text">
5730<h5>Syntax:</h5>
5731<pre>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5732declare 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]5733</pre>
5734<h5>Overview:</h5>
5735<p>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5736 This fills the memory pointed to by <tt>tramp</tt> with code
5737 and returns a function pointer suitable for executing it.
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5738</p>
5739<h5>Arguments:</h5>
5740<p>
5741 The <tt>llvm.init.trampoline</tt> intrinsic takes three arguments, all
5742 pointers. The <tt>tramp</tt> argument must point to a sufficiently large
5743 and sufficiently aligned block of memory; this memory is written to by the
Duncan Sands35012212007-08-22 23:39:54 +0000[diff] [blame]5744 intrinsic. Note that the size and the alignment are target-specific - LLVM
5745 currently provides no portable way of determining them, so a front-end that
5746 generates this intrinsic needs to have some target-specific knowledge.
5747 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]5748</p>
5749<h5>Semantics:</h5>
5750<p>
5751 The block of memory pointed to by <tt>tramp</tt> is filled with target
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5752 dependent code, turning it into a function. A pointer to this function is
5753 returned, but needs to be bitcast to an
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5754 <a href="#int_trampoline">appropriate function pointer type</a>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5755 before being called. The new function's signature is the same as that of
5756 <tt>func</tt> with any arguments marked with the <tt>nest</tt> attribute
5757 removed. At most one such <tt>nest</tt> argument is allowed, and it must be
5758 of pointer type. Calling the new function is equivalent to calling
5759 <tt>func</tt> with the same argument list, but with <tt>nval</tt> used for the
5760 missing <tt>nest</tt> argument. If, after calling
5761 <tt>llvm.init.trampoline</tt>, the memory pointed to by <tt>tramp</tt> is
5762 modified, then the effect of any later call to the returned function pointer is
5763 undefined.
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5764</p>
5765</div>
5766
5767<!-- ======================================================================= -->
5768<div class="doc_subsection">
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]5769 <a name="int_atomics">Atomic Operations and Synchronization Intrinsics</a>
5770</div>
5771
5772<div class="doc_text">
5773<p>
5774 These intrinsic functions expand the "universal IR" of LLVM to represent
5775 hardware constructs for atomic operations and memory synchronization. This
5776 provides an interface to the hardware, not an interface to the programmer. It
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]5777 is aimed at a low enough level to allow any programming models or APIs
5778 (Application Programming Interfaces) which
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]5779 need atomic behaviors to map cleanly onto it. It is also modeled primarily on
5780 hardware behavior. Just as hardware provides a "universal IR" for source
5781 languages, it also provides a starting point for developing a "universal"
5782 atomic operation and synchronization IR.
5783</p>
5784<p>
5785 These do <em>not</em> form an API such as high-level threading libraries,
5786 software transaction memory systems, atomic primitives, and intrinsic
5787 functions as found in BSD, GNU libc, atomic_ops, APR, and other system and
5788 application libraries. The hardware interface provided by LLVM should allow
5789 a clean implementation of all of these APIs and parallel programming models.
5790 No one model or paradigm should be selected above others unless the hardware
5791 itself ubiquitously does so.
5792
5793</p>
5794</div>
5795
5796<!-- _______________________________________________________________________ -->
5797<div class="doc_subsubsection">
5798 <a name="int_memory_barrier">'<tt>llvm.memory.barrier</tt>' Intrinsic</a>
5799</div>
5800<div class="doc_text">
5801<h5>Syntax:</h5>
5802<pre>
5803declare void @llvm.memory.barrier( i1 &lt;ll&gt;, i1 &lt;ls&gt;, i1 &lt;sl&gt;, i1 &lt;ss&gt;,
5804i1 &lt;device&gt; )
5805
5806</pre>
5807<h5>Overview:</h5>
5808<p>
5809 The <tt>llvm.memory.barrier</tt> intrinsic guarantees ordering between
5810 specific pairs of memory access types.
5811</p>
5812<h5>Arguments:</h5>
5813<p>
5814 The <tt>llvm.memory.barrier</tt> intrinsic requires five boolean arguments.
5815 The first four arguments enables a specific barrier as listed below. The fith
5816 argument specifies that the barrier applies to io or device or uncached memory.
5817
5818</p>
5819 <ul>
5820 <li><tt>ll</tt>: load-load barrier</li>
5821 <li><tt>ls</tt>: load-store barrier</li>
5822 <li><tt>sl</tt>: store-load barrier</li>
5823 <li><tt>ss</tt>: store-store barrier</li>
5824 <li><tt>device</tt>: barrier applies to device and uncached memory also.
5825 </ul>
5826<h5>Semantics:</h5>
5827<p>
5828 This intrinsic causes the system to enforce some ordering constraints upon
5829 the loads and stores of the program. This barrier does not indicate
5830 <em>when</em> any events will occur, it only enforces an <em>order</em> in
5831 which they occur. For any of the specified pairs of load and store operations
5832 (f.ex. load-load, or store-load), all of the first operations preceding the
5833 barrier will complete before any of the second operations succeeding the
5834 barrier begin. Specifically the semantics for each pairing is as follows:
5835</p>
5836 <ul>
5837 <li><tt>ll</tt>: All loads before the barrier must complete before any load
5838 after the barrier begins.</li>
5839
5840 <li><tt>ls</tt>: All loads before the barrier must complete before any
5841 store after the barrier begins.</li>
5842 <li><tt>ss</tt>: All stores before the barrier must complete before any
5843 store after the barrier begins.</li>
5844 <li><tt>sl</tt>: All stores before the barrier must complete before any
5845 load after the barrier begins.</li>
5846 </ul>
5847<p>
5848 These semantics are applied with a logical "and" behavior when more than one
5849 is enabled in a single memory barrier intrinsic.
5850</p>
5851<p>
5852 Backends may implement stronger barriers than those requested when they do not
5853 support as fine grained a barrier as requested. Some architectures do not
5854 need all types of barriers and on such architectures, these become noops.
5855</p>
5856<h5>Example:</h5>
5857<pre>
5858%ptr = malloc i32
5859 store i32 4, %ptr
5860
5861%result1 = load i32* %ptr <i>; yields {i32}:result1 = 4</i>
5862 call void @llvm.memory.barrier( i1 false, i1 true, i1 false, i1 false )
5863 <i>; guarantee the above finishes</i>
5864 store i32 8, %ptr <i>; before this begins</i>
5865</pre>
5866</div>
5867
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]5868<!-- _______________________________________________________________________ -->
5869<div class="doc_subsubsection">
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]5870 <a name="int_atomic_cmp_swap">'<tt>llvm.atomic.cmp.swap.*</tt>' Intrinsic</a>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]5871</div>
5872<div class="doc_text">
5873<h5>Syntax:</h5>
5874<p>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]5875 This is an overloaded intrinsic. You can use <tt>llvm.atomic.cmp.swap</tt> on
5876 any integer bit width and for different address spaces. Not all targets
5877 support all bit widths however.</p>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]5878
5879<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]5880declare i8 @llvm.atomic.cmp.swap.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;cmp&gt;, i8 &lt;val&gt; )
5881declare i16 @llvm.atomic.cmp.swap.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;cmp&gt;, i16 &lt;val&gt; )
5882declare i32 @llvm.atomic.cmp.swap.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;cmp&gt;, i32 &lt;val&gt; )
5883declare 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]5884
5885</pre>
5886<h5>Overview:</h5>
5887<p>
5888 This loads a value in memory and compares it to a given value. If they are
5889 equal, it stores a new value into the memory.
5890</p>
5891<h5>Arguments:</h5>
5892<p>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]5893 The <tt>llvm.atomic.cmp.swap</tt> intrinsic takes three arguments. The result as
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]5894 well as both <tt>cmp</tt> and <tt>val</tt> must be integer values with the
5895 same bit width. The <tt>ptr</tt> argument must be a pointer to a value of
5896 this integer type. While any bit width integer may be used, targets may only
5897 lower representations they support in hardware.
5898
5899</p>
5900<h5>Semantics:</h5>
5901<p>
5902 This entire intrinsic must be executed atomically. It first loads the value
5903 in memory pointed to by <tt>ptr</tt> and compares it with the value
5904 <tt>cmp</tt>. If they are equal, <tt>val</tt> is stored into the memory. The
5905 loaded value is yielded in all cases. This provides the equivalent of an
5906 atomic compare-and-swap operation within the SSA framework.
5907</p>
5908<h5>Examples:</h5>
5909
5910<pre>
5911%ptr = malloc i32
5912 store i32 4, %ptr
5913
5914%val1 = add i32 4, 4
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]5915%result1 = call i32 @llvm.atomic.cmp.swap.i32.p0i32( i32* %ptr, i32 4, %val1 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]5916 <i>; yields {i32}:result1 = 4</i>
5917%stored1 = icmp eq i32 %result1, 4 <i>; yields {i1}:stored1 = true</i>
5918%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 8</i>
5919
5920%val2 = add i32 1, 1
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]5921%result2 = call i32 @llvm.atomic.cmp.swap.i32.p0i32( i32* %ptr, i32 5, %val2 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]5922 <i>; yields {i32}:result2 = 8</i>
5923%stored2 = icmp eq i32 %result2, 5 <i>; yields {i1}:stored2 = false</i>
5924
5925%memval2 = load i32* %ptr <i>; yields {i32}:memval2 = 8</i>
5926</pre>
5927</div>
5928
5929<!-- _______________________________________________________________________ -->
5930<div class="doc_subsubsection">
5931 <a name="int_atomic_swap">'<tt>llvm.atomic.swap.*</tt>' Intrinsic</a>
5932</div>
5933<div class="doc_text">
5934<h5>Syntax:</h5>
5935
5936<p>
5937 This is an overloaded intrinsic. You can use <tt>llvm.atomic.swap</tt> on any
5938 integer bit width. Not all targets support all bit widths however.</p>
5939<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]5940declare i8 @llvm.atomic.swap.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;val&gt; )
5941declare i16 @llvm.atomic.swap.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;val&gt; )
5942declare i32 @llvm.atomic.swap.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;val&gt; )
5943declare i64 @llvm.atomic.swap.i64.p0i64( i64* &lt;ptr&gt;, i64 &lt;val&gt; )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]5944
5945</pre>
5946<h5>Overview:</h5>
5947<p>
5948 This intrinsic loads the value stored in memory at <tt>ptr</tt> and yields
5949 the value from memory. It then stores the value in <tt>val</tt> in the memory
5950 at <tt>ptr</tt>.
5951</p>
5952<h5>Arguments:</h5>
5953
5954<p>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]5955 The <tt>llvm.atomic.swap</tt> intrinsic takes two arguments. Both the
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]5956 <tt>val</tt> argument and the result must be integers of the same bit width.
5957 The first argument, <tt>ptr</tt>, must be a pointer to a value of this
5958 integer type. The targets may only lower integer representations they
5959 support.
5960</p>
5961<h5>Semantics:</h5>
5962<p>
5963 This intrinsic loads the value pointed to by <tt>ptr</tt>, yields it, and
5964 stores <tt>val</tt> back into <tt>ptr</tt> atomically. This provides the
5965 equivalent of an atomic swap operation within the SSA framework.
5966
5967</p>
5968<h5>Examples:</h5>
5969<pre>
5970%ptr = malloc i32
5971 store i32 4, %ptr
5972
5973%val1 = add i32 4, 4
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]5974%result1 = call i32 @llvm.atomic.swap.i32.p0i32( i32* %ptr, i32 %val1 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]5975 <i>; yields {i32}:result1 = 4</i>
5976%stored1 = icmp eq i32 %result1, 4 <i>; yields {i1}:stored1 = true</i>
5977%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 8</i>
5978
5979%val2 = add i32 1, 1
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]5980%result2 = call i32 @llvm.atomic.swap.i32.p0i32( i32* %ptr, i32 %val2 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]5981 <i>; yields {i32}:result2 = 8</i>
5982
5983%stored2 = icmp eq i32 %result2, 8 <i>; yields {i1}:stored2 = true</i>
5984%memval2 = load i32* %ptr <i>; yields {i32}:memval2 = 2</i>
5985</pre>
5986</div>
5987
5988<!-- _______________________________________________________________________ -->
5989<div class="doc_subsubsection">
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]5990 <a name="int_atomic_load_add">'<tt>llvm.atomic.load.add.*</tt>' Intrinsic</a>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]5991
5992</div>
5993<div class="doc_text">
5994<h5>Syntax:</h5>
5995<p>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]5996 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]5997 integer bit width. Not all targets support all bit widths however.</p>
5998<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]5999declare i8 @llvm.atomic.load.add.i8..p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6000declare i16 @llvm.atomic.load.add.i16..p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6001declare i32 @llvm.atomic.load.add.i32..p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6002declare 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]6003
6004</pre>
6005<h5>Overview:</h5>
6006<p>
6007 This intrinsic adds <tt>delta</tt> to the value stored in memory at
6008 <tt>ptr</tt>. It yields the original value at <tt>ptr</tt>.
6009</p>
6010<h5>Arguments:</h5>
6011<p>
6012
6013 The intrinsic takes two arguments, the first a pointer to an integer value
6014 and the second an integer value. The result is also an integer value. These
6015 integer types can have any bit width, but they must all have the same bit
6016 width. The targets may only lower integer representations they support.
6017</p>
6018<h5>Semantics:</h5>
6019<p>
6020 This intrinsic does a series of operations atomically. It first loads the
6021 value stored at <tt>ptr</tt>. It then adds <tt>delta</tt>, stores the result
6022 to <tt>ptr</tt>. It yields the original value stored at <tt>ptr</tt>.
6023</p>
6024
6025<h5>Examples:</h5>
6026<pre>
6027%ptr = malloc i32
6028 store i32 4, %ptr
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6029%result1 = call i32 @llvm.atomic.load.add.i32.p0i32( i32* %ptr, i32 4 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6030 <i>; yields {i32}:result1 = 4</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6031%result2 = call i32 @llvm.atomic.load.add.i32.p0i32( i32* %ptr, i32 2 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6032 <i>; yields {i32}:result2 = 8</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6033%result3 = call i32 @llvm.atomic.load.add.i32.p0i32( i32* %ptr, i32 5 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6034 <i>; yields {i32}:result3 = 10</i>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6035%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 15</i>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6036</pre>
6037</div>
6038
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6039<!-- _______________________________________________________________________ -->
6040<div class="doc_subsubsection">
6041 <a name="int_atomic_load_sub">'<tt>llvm.atomic.load.sub.*</tt>' Intrinsic</a>
6042
6043</div>
6044<div class="doc_text">
6045<h5>Syntax:</h5>
6046<p>
6047 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]6048 any integer bit width and for different address spaces. Not all targets
6049 support all bit widths however.</p>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6050<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6051declare i8 @llvm.atomic.load.sub.i8.p0i32( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6052declare i16 @llvm.atomic.load.sub.i16.p0i32( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6053declare i32 @llvm.atomic.load.sub.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6054declare 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]6055
6056</pre>
6057<h5>Overview:</h5>
6058<p>
6059 This intrinsic subtracts <tt>delta</tt> to the value stored in memory at
6060 <tt>ptr</tt>. It yields the original value at <tt>ptr</tt>.
6061</p>
6062<h5>Arguments:</h5>
6063<p>
6064
6065 The intrinsic takes two arguments, the first a pointer to an integer value
6066 and the second an integer value. The result is also an integer value. These
6067 integer types can have any bit width, but they must all have the same bit
6068 width. The targets may only lower integer representations they support.
6069</p>
6070<h5>Semantics:</h5>
6071<p>
6072 This intrinsic does a series of operations atomically. It first loads the
6073 value stored at <tt>ptr</tt>. It then subtracts <tt>delta</tt>, stores the
6074 result to <tt>ptr</tt>. It yields the original value stored at <tt>ptr</tt>.
6075</p>
6076
6077<h5>Examples:</h5>
6078<pre>
6079%ptr = malloc i32
6080 store i32 8, %ptr
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6081%result1 = call i32 @llvm.atomic.load.sub.i32.p0i32( i32* %ptr, i32 4 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6082 <i>; yields {i32}:result1 = 8</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6083%result2 = call i32 @llvm.atomic.load.sub.i32.p0i32( i32* %ptr, i32 2 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6084 <i>; yields {i32}:result2 = 4</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6085%result3 = call i32 @llvm.atomic.load.sub.i32.p0i32( i32* %ptr, i32 5 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6086 <i>; yields {i32}:result3 = 2</i>
6087%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = -3</i>
6088</pre>
6089</div>
6090
6091<!-- _______________________________________________________________________ -->
6092<div class="doc_subsubsection">
6093 <a name="int_atomic_load_and">'<tt>llvm.atomic.load.and.*</tt>' Intrinsic</a><br>
6094 <a name="int_atomic_load_nand">'<tt>llvm.atomic.load.nand.*</tt>' Intrinsic</a><br>
6095 <a name="int_atomic_load_or">'<tt>llvm.atomic.load.or.*</tt>' Intrinsic</a><br>
6096 <a name="int_atomic_load_xor">'<tt>llvm.atomic.load.xor.*</tt>' Intrinsic</a><br>
6097
6098</div>
6099<div class="doc_text">
6100<h5>Syntax:</h5>
6101<p>
6102 These are overloaded intrinsics. You can use <tt>llvm.atomic.load_and</tt>,
6103 <tt>llvm.atomic.load_nand</tt>, <tt>llvm.atomic.load_or</tt>, and
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6104 <tt>llvm.atomic.load_xor</tt> on any integer bit width and for different
6105 address spaces. Not all targets support all bit widths however.</p>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6106<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6107declare i8 @llvm.atomic.load.and.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6108declare i16 @llvm.atomic.load.and.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6109declare i32 @llvm.atomic.load.and.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6110declare 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]6111
6112</pre>
6113
6114<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6115declare i8 @llvm.atomic.load.or.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6116declare i16 @llvm.atomic.load.or.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6117declare i32 @llvm.atomic.load.or.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6118declare 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]6119
6120</pre>
6121
6122<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6123declare i8 @llvm.atomic.load.nand.i8.p0i32( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6124declare i16 @llvm.atomic.load.nand.i16.p0i32( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6125declare i32 @llvm.atomic.load.nand.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6126declare 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]6127
6128</pre>
6129
6130<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6131declare i8 @llvm.atomic.load.xor.i8.p0i32( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6132declare i16 @llvm.atomic.load.xor.i16.p0i32( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6133declare i32 @llvm.atomic.load.xor.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6134declare 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]6135
6136</pre>
6137<h5>Overview:</h5>
6138<p>
6139 These intrinsics bitwise the operation (and, nand, or, xor) <tt>delta</tt> to
6140 the value stored in memory at <tt>ptr</tt>. It yields the original value
6141 at <tt>ptr</tt>.
6142</p>
6143<h5>Arguments:</h5>
6144<p>
6145
6146 These intrinsics take two arguments, the first a pointer to an integer value
6147 and the second an integer value. The result is also an integer value. These
6148 integer types can have any bit width, but they must all have the same bit
6149 width. The targets may only lower integer representations they support.
6150</p>
6151<h5>Semantics:</h5>
6152<p>
6153 These intrinsics does a series of operations atomically. They first load the
6154 value stored at <tt>ptr</tt>. They then do the bitwise operation
6155 <tt>delta</tt>, store the result to <tt>ptr</tt>. They yield the original
6156 value stored at <tt>ptr</tt>.
6157</p>
6158
6159<h5>Examples:</h5>
6160<pre>
6161%ptr = malloc i32
6162 store i32 0x0F0F, %ptr
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6163%result0 = call i32 @llvm.atomic.load.nand.i32.p0i32( i32* %ptr, i32 0xFF )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6164 <i>; yields {i32}:result0 = 0x0F0F</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6165%result1 = call i32 @llvm.atomic.load.and.i32.p0i32( i32* %ptr, i32 0xFF )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6166 <i>; yields {i32}:result1 = 0xFFFFFFF0</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6167%result2 = call i32 @llvm.atomic.load.or.i32.p0i32( i32* %ptr, i32 0F )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6168 <i>; yields {i32}:result2 = 0xF0</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6169%result3 = call i32 @llvm.atomic.load.xor.i32.p0i32( i32* %ptr, i32 0F )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6170 <i>; yields {i32}:result3 = FF</i>
6171%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = F0</i>
6172</pre>
6173</div>
6174
6175
6176<!-- _______________________________________________________________________ -->
6177<div class="doc_subsubsection">
6178 <a name="int_atomic_load_max">'<tt>llvm.atomic.load.max.*</tt>' Intrinsic</a><br>
6179 <a name="int_atomic_load_min">'<tt>llvm.atomic.load.min.*</tt>' Intrinsic</a><br>
6180 <a name="int_atomic_load_umax">'<tt>llvm.atomic.load.umax.*</tt>' Intrinsic</a><br>
6181 <a name="int_atomic_load_umin">'<tt>llvm.atomic.load.umin.*</tt>' Intrinsic</a><br>
6182
6183</div>
6184<div class="doc_text">
6185<h5>Syntax:</h5>
6186<p>
6187 These are overloaded intrinsics. You can use <tt>llvm.atomic.load_max</tt>,
6188 <tt>llvm.atomic.load_min</tt>, <tt>llvm.atomic.load_umax</tt>, and
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6189 <tt>llvm.atomic.load_umin</tt> on any integer bit width and for different
6190 address spaces. Not all targets
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6191 support all bit widths however.</p>
6192<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6193declare i8 @llvm.atomic.load.max.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6194declare i16 @llvm.atomic.load.max.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6195declare i32 @llvm.atomic.load.max.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6196declare 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]6197
6198</pre>
6199
6200<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6201declare i8 @llvm.atomic.load.min.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6202declare i16 @llvm.atomic.load.min.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6203declare i32 @llvm.atomic.load.min.i32..p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6204declare 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]6205
6206</pre>
6207
6208<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6209declare i8 @llvm.atomic.load.umax.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6210declare i16 @llvm.atomic.load.umax.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6211declare i32 @llvm.atomic.load.umax.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6212declare 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]6213
6214</pre>
6215
6216<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6217declare i8 @llvm.atomic.load.umin.i8..p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6218declare i16 @llvm.atomic.load.umin.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6219declare i32 @llvm.atomic.load.umin.i32..p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6220declare 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]6221
6222</pre>
6223<h5>Overview:</h5>
6224<p>
6225 These intrinsics takes the signed or unsigned minimum or maximum of
6226 <tt>delta</tt> and the value stored in memory at <tt>ptr</tt>. It yields the
6227 original value at <tt>ptr</tt>.
6228</p>
6229<h5>Arguments:</h5>
6230<p>
6231
6232 These intrinsics take two arguments, the first a pointer to an integer value
6233 and the second an integer value. The result is also an integer value. These
6234 integer types can have any bit width, but they must all have the same bit
6235 width. The targets may only lower integer representations they support.
6236</p>
6237<h5>Semantics:</h5>
6238<p>
6239 These intrinsics does a series of operations atomically. They first load the
6240 value stored at <tt>ptr</tt>. They then do the signed or unsigned min or max
6241 <tt>delta</tt> and the value, store the result to <tt>ptr</tt>. They yield
6242 the original value stored at <tt>ptr</tt>.
6243</p>
6244
6245<h5>Examples:</h5>
6246<pre>
6247%ptr = malloc i32
6248 store i32 7, %ptr
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6249%result0 = call i32 @llvm.atomic.load.min.i32.p0i32( i32* %ptr, i32 -2 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6250 <i>; yields {i32}:result0 = 7</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6251%result1 = call i32 @llvm.atomic.load.max.i32.p0i32( i32* %ptr, i32 8 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6252 <i>; yields {i32}:result1 = -2</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6253%result2 = call i32 @llvm.atomic.load.umin.i32.p0i32( i32* %ptr, i32 10 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6254 <i>; yields {i32}:result2 = 8</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6255%result3 = call i32 @llvm.atomic.load.umax.i32.p0i32( i32* %ptr, i32 30 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6256 <i>; yields {i32}:result3 = 8</i>
6257%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 30</i>
6258</pre>
6259</div>
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]6260
6261<!-- ======================================================================= -->
6262<div class="doc_subsection">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6263 <a name="int_general">General Intrinsics</a>
6264</div>
6265
6266<div class="doc_text">
6267<p> This class of intrinsics is designed to be generic and has
6268no specific purpose. </p>
6269</div>
6270
6271<!-- _______________________________________________________________________ -->
6272<div class="doc_subsubsection">
6273 <a name="int_var_annotation">'<tt>llvm.var.annotation</tt>' Intrinsic</a>
6274</div>
6275
6276<div class="doc_text">
6277
6278<h5>Syntax:</h5>
6279<pre>
6280 declare void @llvm.var.annotation(i8* &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
6281</pre>
6282
6283<h5>Overview:</h5>
6284
6285<p>
6286The '<tt>llvm.var.annotation</tt>' intrinsic
6287</p>
6288
6289<h5>Arguments:</h5>
6290
6291<p>
6292The first argument is a pointer to a value, the second is a pointer to a
6293global string, the third is a pointer to a global string which is the source
6294file name, and the last argument is the line number.
6295</p>
6296
6297<h5>Semantics:</h5>
6298
6299<p>
Anton Korobeynikove6e764f2008-01-15 22:31:34 +0000[diff] [blame]6300This intrinsic allows annotation of local variables with arbitrary strings.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6301This can be useful for special purpose optimizations that want to look for these
Anton Korobeynikove6e764f2008-01-15 22:31:34 +0000[diff] [blame]6302annotations. These have no other defined use, they are ignored by code
6303generation and optimization.
6304</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6305</div>
6306
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]6307<!-- _______________________________________________________________________ -->
6308<div class="doc_subsubsection">
Tanya Lattnerc9869b12007-09-21 23:57:59 +0000[diff] [blame]6309 <a name="int_annotation">'<tt>llvm.annotation.*</tt>' Intrinsic</a>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]6310</div>
6311
6312<div class="doc_text">
6313
6314<h5>Syntax:</h5>
Tanya Lattnere545be72007-09-21 23:56:27 +0000[diff] [blame]6315<p>This is an overloaded intrinsic. You can use '<tt>llvm.annotation</tt>' on
6316any integer bit width.
6317</p>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]6318<pre>
Tanya Lattner09161fe2007-09-22 00:03:01 +0000[diff] [blame]6319 declare i8 @llvm.annotation.i8(i8 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
6320 declare i16 @llvm.annotation.i16(i16 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
6321 declare i32 @llvm.annotation.i32(i32 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
6322 declare i64 @llvm.annotation.i64(i64 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
6323 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]6324</pre>
6325
6326<h5>Overview:</h5>
Tanya Lattnere545be72007-09-21 23:56:27 +0000[diff] [blame]6327
6328<p>
6329The '<tt>llvm.annotation</tt>' intrinsic.
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]6330</p>
6331
6332<h5>Arguments:</h5>
6333
6334<p>
6335The first argument is an integer value (result of some expression),
6336the second is a pointer to a global string, the third is a pointer to a global
6337string which is the source file name, and the last argument is the line number.
Tanya Lattnere545be72007-09-21 23:56:27 +0000[diff] [blame]6338It returns the value of the first argument.
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]6339</p>
6340
6341<h5>Semantics:</h5>
6342
6343<p>
6344This intrinsic allows annotations to be put on arbitrary expressions
6345with arbitrary strings. This can be useful for special purpose optimizations
6346that want to look for these annotations. These have no other defined use, they
6347are ignored by code generation and optimization.
6348</div>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6349
Anton Korobeynikove6e764f2008-01-15 22:31:34 +0000[diff] [blame]6350<!-- _______________________________________________________________________ -->
6351<div class="doc_subsubsection">
6352 <a name="int_trap">'<tt>llvm.trap</tt>' Intrinsic</a>
6353</div>
6354
6355<div class="doc_text">
6356
6357<h5>Syntax:</h5>
6358<pre>
6359 declare void @llvm.trap()
6360</pre>
6361
6362<h5>Overview:</h5>
6363
6364<p>
6365The '<tt>llvm.trap</tt>' intrinsic
6366</p>
6367
6368<h5>Arguments:</h5>
6369
6370<p>
6371None
6372</p>
6373
6374<h5>Semantics:</h5>
6375
6376<p>
6377This intrinsics is lowered to the target dependent trap instruction. If the
6378target does not have a trap instruction, this intrinsic will be lowered to the
6379call of the abort() function.
6380</p>
6381</div>
6382
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6383<!-- *********************************************************************** -->
6384<hr>
6385<address>
6386 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
6387 src="http://jigsaw.w3.org/css-validator/images/vcss" alt="Valid CSS!"></a>
6388 <a href="http://validator.w3.org/check/referer"><img
Chris Lattner08497ce2008-01-04 04:33:49 +0000[diff] [blame]6389 src="http://www.w3.org/Icons/valid-html401" alt="Valid HTML 4.01!"></a>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6390
6391 <a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
6392 <a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br>
6393 Last modified: $Date$
6394</address>
Chris Lattner08497ce2008-01-04 04:33:49 +0000[diff] [blame]6395
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6396</body>
6397</html>