Gitiles
Code Review Sign In
gerrit-public.fairphone.software / fp2-dev / platform / external / llvm / b1aab4ec6121dee018a4aef63db7781597723651 / . / docs / LangRef.html
blob: 01b971bf2f0a9402a06394a4ec68ffff9503b983 [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>
Gordon Henriksen13fe5e32007-12-10 03:18:06 +0000[diff] [blame]29 <li><a href="#gc">Garbage Collector Names</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]30 <li><a href="#moduleasm">Module-Level Inline Assembly</a></li>
31 <li><a href="#datalayout">Data Layout</a></li>
32 </ol>
33 </li>
34 <li><a href="#typesystem">Type System</a>
35 <ol>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]36 <li><a href="#t_classifications">Type Classifications</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]37 <li><a href="#t_primitive">Primitive Types</a>
38 <ol>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]39 <li><a href="#t_floating">Floating Point Types</a></li>
40 <li><a href="#t_void">Void Type</a></li>
41 <li><a href="#t_label">Label Type</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]42 </ol>
43 </li>
44 <li><a href="#t_derived">Derived Types</a>
45 <ol>
Chris Lattner251ab812007-12-18 06:18:21 +0000[diff] [blame]46 <li><a href="#t_integer">Integer Type</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]47 <li><a href="#t_array">Array Type</a></li>
48 <li><a href="#t_function">Function Type</a></li>
49 <li><a href="#t_pointer">Pointer Type</a></li>
50 <li><a href="#t_struct">Structure Type</a></li>
51 <li><a href="#t_pstruct">Packed Structure Type</a></li>
52 <li><a href="#t_vector">Vector Type</a></li>
53 <li><a href="#t_opaque">Opaque Type</a></li>
54 </ol>
55 </li>
56 </ol>
57 </li>
58 <li><a href="#constants">Constants</a>
59 <ol>
60 <li><a href="#simpleconstants">Simple Constants</a>
61 <li><a href="#aggregateconstants">Aggregate Constants</a>
62 <li><a href="#globalconstants">Global Variable and Function Addresses</a>
63 <li><a href="#undefvalues">Undefined Values</a>
64 <li><a href="#constantexprs">Constant Expressions</a>
65 </ol>
66 </li>
67 <li><a href="#othervalues">Other Values</a>
68 <ol>
69 <li><a href="#inlineasm">Inline Assembler Expressions</a>
70 </ol>
71 </li>
72 <li><a href="#instref">Instruction Reference</a>
73 <ol>
74 <li><a href="#terminators">Terminator Instructions</a>
75 <ol>
76 <li><a href="#i_ret">'<tt>ret</tt>' Instruction</a></li>
77 <li><a href="#i_br">'<tt>br</tt>' Instruction</a></li>
78 <li><a href="#i_switch">'<tt>switch</tt>' Instruction</a></li>
79 <li><a href="#i_invoke">'<tt>invoke</tt>' Instruction</a></li>
80 <li><a href="#i_unwind">'<tt>unwind</tt>' Instruction</a></li>
81 <li><a href="#i_unreachable">'<tt>unreachable</tt>' Instruction</a></li>
82 </ol>
83 </li>
84 <li><a href="#binaryops">Binary Operations</a>
85 <ol>
86 <li><a href="#i_add">'<tt>add</tt>' Instruction</a></li>
87 <li><a href="#i_sub">'<tt>sub</tt>' Instruction</a></li>
88 <li><a href="#i_mul">'<tt>mul</tt>' Instruction</a></li>
89 <li><a href="#i_udiv">'<tt>udiv</tt>' Instruction</a></li>
90 <li><a href="#i_sdiv">'<tt>sdiv</tt>' Instruction</a></li>
91 <li><a href="#i_fdiv">'<tt>fdiv</tt>' Instruction</a></li>
92 <li><a href="#i_urem">'<tt>urem</tt>' Instruction</a></li>
93 <li><a href="#i_srem">'<tt>srem</tt>' Instruction</a></li>
94 <li><a href="#i_frem">'<tt>frem</tt>' Instruction</a></li>
95 </ol>
96 </li>
97 <li><a href="#bitwiseops">Bitwise Binary Operations</a>
98 <ol>
99 <li><a href="#i_shl">'<tt>shl</tt>' Instruction</a></li>
100 <li><a href="#i_lshr">'<tt>lshr</tt>' Instruction</a></li>
101 <li><a href="#i_ashr">'<tt>ashr</tt>' Instruction</a></li>
102 <li><a href="#i_and">'<tt>and</tt>' Instruction</a></li>
103 <li><a href="#i_or">'<tt>or</tt>' Instruction</a></li>
104 <li><a href="#i_xor">'<tt>xor</tt>' Instruction</a></li>
105 </ol>
106 </li>
107 <li><a href="#vectorops">Vector Operations</a>
108 <ol>
109 <li><a href="#i_extractelement">'<tt>extractelement</tt>' Instruction</a></li>
110 <li><a href="#i_insertelement">'<tt>insertelement</tt>' Instruction</a></li>
111 <li><a href="#i_shufflevector">'<tt>shufflevector</tt>' Instruction</a></li>
112 </ol>
113 </li>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]114 <li><a href="#aggregateops">Aggregate Operations</a>
115 <ol>
116 <li><a href="#i_extractvalue">'<tt>extractvalue</tt>' Instruction</a></li>
117 <li><a href="#i_insertvalue">'<tt>insertvalue</tt>' Instruction</a></li>
118 </ol>
119 </li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]120 <li><a href="#memoryops">Memory Access and Addressing Operations</a>
121 <ol>
122 <li><a href="#i_malloc">'<tt>malloc</tt>' Instruction</a></li>
123 <li><a href="#i_free">'<tt>free</tt>' Instruction</a></li>
124 <li><a href="#i_alloca">'<tt>alloca</tt>' Instruction</a></li>
125 <li><a href="#i_load">'<tt>load</tt>' Instruction</a></li>
126 <li><a href="#i_store">'<tt>store</tt>' Instruction</a></li>
127 <li><a href="#i_getelementptr">'<tt>getelementptr</tt>' Instruction</a></li>
128 </ol>
129 </li>
130 <li><a href="#convertops">Conversion Operations</a>
131 <ol>
132 <li><a href="#i_trunc">'<tt>trunc .. to</tt>' Instruction</a></li>
133 <li><a href="#i_zext">'<tt>zext .. to</tt>' Instruction</a></li>
134 <li><a href="#i_sext">'<tt>sext .. to</tt>' Instruction</a></li>
135 <li><a href="#i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a></li>
136 <li><a href="#i_fpext">'<tt>fpext .. to</tt>' Instruction</a></li>
137 <li><a href="#i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a></li>
138 <li><a href="#i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a></li>
139 <li><a href="#i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a></li>
140 <li><a href="#i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a></li>
141 <li><a href="#i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a></li>
142 <li><a href="#i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a></li>
143 <li><a href="#i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a></li>
144 </ol>
145 <li><a href="#otherops">Other Operations</a>
146 <ol>
147 <li><a href="#i_icmp">'<tt>icmp</tt>' Instruction</a></li>
148 <li><a href="#i_fcmp">'<tt>fcmp</tt>' Instruction</a></li>
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]149 <li><a href="#i_vicmp">'<tt>vicmp</tt>' Instruction</a></li>
150 <li><a href="#i_vfcmp">'<tt>vfcmp</tt>' Instruction</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]151 <li><a href="#i_phi">'<tt>phi</tt>' Instruction</a></li>
152 <li><a href="#i_select">'<tt>select</tt>' Instruction</a></li>
153 <li><a href="#i_call">'<tt>call</tt>' Instruction</a></li>
154 <li><a href="#i_va_arg">'<tt>va_arg</tt>' Instruction</a></li>
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]155 <li><a href="#i_getresult">'<tt>getresult</tt>' Instruction</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]156 </ol>
157 </li>
158 </ol>
159 </li>
160 <li><a href="#intrinsics">Intrinsic Functions</a>
161 <ol>
162 <li><a href="#int_varargs">Variable Argument Handling Intrinsics</a>
163 <ol>
164 <li><a href="#int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a></li>
165 <li><a href="#int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a></li>
166 <li><a href="#int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a></li>
167 </ol>
168 </li>
169 <li><a href="#int_gc">Accurate Garbage Collection Intrinsics</a>
170 <ol>
171 <li><a href="#int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a></li>
172 <li><a href="#int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a></li>
173 <li><a href="#int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a></li>
174 </ol>
175 </li>
176 <li><a href="#int_codegen">Code Generator Intrinsics</a>
177 <ol>
178 <li><a href="#int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a></li>
179 <li><a href="#int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a></li>
180 <li><a href="#int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a></li>
181 <li><a href="#int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a></li>
182 <li><a href="#int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a></li>
183 <li><a href="#int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a></li>
184 <li><a href="#int_readcyclecounter"><tt>llvm.readcyclecounter</tt>' Intrinsic</a></li>
185 </ol>
186 </li>
187 <li><a href="#int_libc">Standard C Library Intrinsics</a>
188 <ol>
189 <li><a href="#int_memcpy">'<tt>llvm.memcpy.*</tt>' Intrinsic</a></li>
190 <li><a href="#int_memmove">'<tt>llvm.memmove.*</tt>' Intrinsic</a></li>
191 <li><a href="#int_memset">'<tt>llvm.memset.*</tt>' Intrinsic</a></li>
192 <li><a href="#int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a></li>
193 <li><a href="#int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a></li>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]194 <li><a href="#int_sin">'<tt>llvm.sin.*</tt>' Intrinsic</a></li>
195 <li><a href="#int_cos">'<tt>llvm.cos.*</tt>' Intrinsic</a></li>
196 <li><a href="#int_pow">'<tt>llvm.pow.*</tt>' Intrinsic</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]197 </ol>
198 </li>
199 <li><a href="#int_manip">Bit Manipulation Intrinsics</a>
200 <ol>
201 <li><a href="#int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a></li>
202 <li><a href="#int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic </a></li>
203 <li><a href="#int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic </a></li>
204 <li><a href="#int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic </a></li>
205 <li><a href="#int_part_select">'<tt>llvm.part.select.*</tt>' Intrinsic </a></li>
206 <li><a href="#int_part_set">'<tt>llvm.part.set.*</tt>' Intrinsic </a></li>
207 </ol>
208 </li>
209 <li><a href="#int_debugger">Debugger intrinsics</a></li>
210 <li><a href="#int_eh">Exception Handling intrinsics</a></li>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]211 <li><a href="#int_trampoline">Trampoline Intrinsic</a>
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]212 <ol>
213 <li><a href="#int_it">'<tt>llvm.init.trampoline</tt>' Intrinsic</a></li>
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]214 </ol>
215 </li>
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]216 <li><a href="#int_atomics">Atomic intrinsics</a>
217 <ol>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]218 <li><a href="#int_memory_barrier"><tt>llvm.memory_barrier</tt></a></li>
219 <li><a href="#int_atomic_lcs"><tt>llvm.atomic.lcs</tt></a></li>
220 <li><a href="#int_atomic_las"><tt>llvm.atomic.las</tt></a></li>
221 <li><a href="#int_atomic_swap"><tt>llvm.atomic.swap</tt></a></li>
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]222 </ol>
223 </li>
Reid Spencerb043f672007-07-20 19:59:11 +0000[diff] [blame]224 <li><a href="#int_general">General intrinsics</a>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]225 <ol>
Reid Spencerb043f672007-07-20 19:59:11 +0000[diff] [blame]226 <li><a href="#int_var_annotation">
Tanya Lattner51369f32007-09-22 00:01:26 +0000[diff] [blame]227 <tt>llvm.var.annotation</tt>' Intrinsic</a></li>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]228 <li><a href="#int_annotation">
Tanya Lattner51369f32007-09-22 00:01:26 +0000[diff] [blame]229 <tt>llvm.annotation.*</tt>' Intrinsic</a></li>
Anton Korobeynikove6e764f2008-01-15 22:31:34 +0000[diff] [blame]230 <li><a href="#int_trap">
231 <tt>llvm.trap</tt>' Intrinsic</a></li>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]232 </ol>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]233 </li>
234 </ol>
235 </li>
236</ol>
237
238<div class="doc_author">
239 <p>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a>
240 and <a href="mailto:vadve@cs.uiuc.edu">Vikram Adve</a></p>
241</div>
242
243<!-- *********************************************************************** -->
244<div class="doc_section"> <a name="abstract">Abstract </a></div>
245<!-- *********************************************************************** -->
246
247<div class="doc_text">
248<p>This document is a reference manual for the LLVM assembly language.
249LLVM is an SSA based representation that provides type safety,
250low-level operations, flexibility, and the capability of representing
251'all' high-level languages cleanly. It is the common code
252representation used throughout all phases of the LLVM compilation
253strategy.</p>
254</div>
255
256<!-- *********************************************************************** -->
257<div class="doc_section"> <a name="introduction">Introduction</a> </div>
258<!-- *********************************************************************** -->
259
260<div class="doc_text">
261
262<p>The LLVM code representation is designed to be used in three
263different forms: as an in-memory compiler IR, as an on-disk bitcode
264representation (suitable for fast loading by a Just-In-Time compiler),
265and as a human readable assembly language representation. This allows
266LLVM to provide a powerful intermediate representation for efficient
267compiler transformations and analysis, while providing a natural means
268to debug and visualize the transformations. The three different forms
269of LLVM are all equivalent. This document describes the human readable
270representation and notation.</p>
271
272<p>The LLVM representation aims to be light-weight and low-level
273while being expressive, typed, and extensible at the same time. It
274aims to be a "universal IR" of sorts, by being at a low enough level
275that high-level ideas may be cleanly mapped to it (similar to how
276microprocessors are "universal IR's", allowing many source languages to
277be mapped to them). By providing type information, LLVM can be used as
278the target of optimizations: for example, through pointer analysis, it
279can be proven that a C automatic variable is never accessed outside of
280the current function... allowing it to be promoted to a simple SSA
281value instead of a memory location.</p>
282
283</div>
284
285<!-- _______________________________________________________________________ -->
286<div class="doc_subsubsection"> <a name="wellformed">Well-Formedness</a> </div>
287
288<div class="doc_text">
289
290<p>It is important to note that this document describes 'well formed'
291LLVM assembly language. There is a difference between what the parser
292accepts and what is considered 'well formed'. For example, the
293following instruction is syntactically okay, but not well formed:</p>
294
295<div class="doc_code">
296<pre>
297%x = <a href="#i_add">add</a> i32 1, %x
298</pre>
299</div>
300
301<p>...because the definition of <tt>%x</tt> does not dominate all of
302its uses. The LLVM infrastructure provides a verification pass that may
303be used to verify that an LLVM module is well formed. This pass is
304automatically run by the parser after parsing input assembly and by
305the optimizer before it outputs bitcode. The violations pointed out
306by the verifier pass indicate bugs in transformation passes or input to
307the parser.</p>
308</div>
309
Chris Lattnera83fdc02007-10-03 17:34:29 +0000[diff] [blame]310<!-- Describe the typesetting conventions here. -->
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]311
312<!-- *********************************************************************** -->
313<div class="doc_section"> <a name="identifiers">Identifiers</a> </div>
314<!-- *********************************************************************** -->
315
316<div class="doc_text">
317
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]318 <p>LLVM identifiers come in two basic types: global and local. Global
319 identifiers (functions, global variables) begin with the @ character. Local
320 identifiers (register names, types) begin with the % character. Additionally,
321 there are three different formats for identifiers, for different purposes:
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]322
323<ol>
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]324 <li>Named values are represented as a string of characters with their prefix.
325 For example, %foo, @DivisionByZero, %a.really.long.identifier. The actual
326 regular expression used is '<tt>[%@][a-zA-Z$._][a-zA-Z$._0-9]*</tt>'.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]327 Identifiers which require other characters in their names can be surrounded
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]328 with quotes. In this way, anything except a <tt>&quot;</tt> character can
329 be used in a named value.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]330
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]331 <li>Unnamed values are represented as an unsigned numeric value with their
332 prefix. For example, %12, @2, %44.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]333
334 <li>Constants, which are described in a <a href="#constants">section about
335 constants</a>, below.</li>
336</ol>
337
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]338<p>LLVM requires that values start with a prefix for two reasons: Compilers
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]339don't need to worry about name clashes with reserved words, and the set of
340reserved words may be expanded in the future without penalty. Additionally,
341unnamed identifiers allow a compiler to quickly come up with a temporary
342variable without having to avoid symbol table conflicts.</p>
343
344<p>Reserved words in LLVM are very similar to reserved words in other
345languages. There are keywords for different opcodes
346('<tt><a href="#i_add">add</a></tt>',
347 '<tt><a href="#i_bitcast">bitcast</a></tt>',
348 '<tt><a href="#i_ret">ret</a></tt>', etc...), for primitive type names ('<tt><a
349href="#t_void">void</a></tt>', '<tt><a href="#t_primitive">i32</a></tt>', etc...),
350and others. These reserved words cannot conflict with variable names, because
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]351none of them start with a prefix character ('%' or '@').</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]352
353<p>Here is an example of LLVM code to multiply the integer variable
354'<tt>%X</tt>' by 8:</p>
355
356<p>The easy way:</p>
357
358<div class="doc_code">
359<pre>
360%result = <a href="#i_mul">mul</a> i32 %X, 8
361</pre>
362</div>
363
364<p>After strength reduction:</p>
365
366<div class="doc_code">
367<pre>
368%result = <a href="#i_shl">shl</a> i32 %X, i8 3
369</pre>
370</div>
371
372<p>And the hard way:</p>
373
374<div class="doc_code">
375<pre>
376<a href="#i_add">add</a> i32 %X, %X <i>; yields {i32}:%0</i>
377<a href="#i_add">add</a> i32 %0, %0 <i>; yields {i32}:%1</i>
378%result = <a href="#i_add">add</a> i32 %1, %1
379</pre>
380</div>
381
382<p>This last way of multiplying <tt>%X</tt> by 8 illustrates several
383important lexical features of LLVM:</p>
384
385<ol>
386
387 <li>Comments are delimited with a '<tt>;</tt>' and go until the end of
388 line.</li>
389
390 <li>Unnamed temporaries are created when the result of a computation is not
391 assigned to a named value.</li>
392
393 <li>Unnamed temporaries are numbered sequentially</li>
394
395</ol>
396
397<p>...and it also shows a convention that we follow in this document. When
398demonstrating instructions, we will follow an instruction with a comment that
399defines the type and name of value produced. Comments are shown in italic
400text.</p>
401
402</div>
403
404<!-- *********************************************************************** -->
405<div class="doc_section"> <a name="highlevel">High Level Structure</a> </div>
406<!-- *********************************************************************** -->
407
408<!-- ======================================================================= -->
409<div class="doc_subsection"> <a name="modulestructure">Module Structure</a>
410</div>
411
412<div class="doc_text">
413
414<p>LLVM programs are composed of "Module"s, each of which is a
415translation unit of the input programs. Each module consists of
416functions, global variables, and symbol table entries. Modules may be
417combined together with the LLVM linker, which merges function (and
418global variable) definitions, resolves forward declarations, and merges
419symbol table entries. Here is an example of the "hello world" module:</p>
420
421<div class="doc_code">
422<pre><i>; Declare the string constant as a global constant...</i>
423<a href="#identifiers">@.LC0</a> = <a href="#linkage_internal">internal</a> <a
424 href="#globalvars">constant</a> <a href="#t_array">[13 x i8]</a> c"hello world\0A\00" <i>; [13 x i8]*</i>
425
426<i>; External declaration of the puts function</i>
427<a href="#functionstructure">declare</a> i32 @puts(i8 *) <i>; i32(i8 *)* </i>
428
429<i>; Definition of main function</i>
430define i32 @main() { <i>; i32()* </i>
431 <i>; Convert [13x i8 ]* to i8 *...</i>
432 %cast210 = <a
433 href="#i_getelementptr">getelementptr</a> [13 x i8 ]* @.LC0, i64 0, i64 0 <i>; i8 *</i>
434
435 <i>; Call puts function to write out the string to stdout...</i>
436 <a
437 href="#i_call">call</a> i32 @puts(i8 * %cast210) <i>; i32</i>
438 <a
439 href="#i_ret">ret</a> i32 0<br>}<br>
440</pre>
441</div>
442
443<p>This example is made up of a <a href="#globalvars">global variable</a>
444named "<tt>.LC0</tt>", an external declaration of the "<tt>puts</tt>"
445function, and a <a href="#functionstructure">function definition</a>
446for "<tt>main</tt>".</p>
447
448<p>In general, a module is made up of a list of global values,
449where both functions and global variables are global values. Global values are
450represented by a pointer to a memory location (in this case, a pointer to an
451array of char, and a pointer to a function), and have one of the following <a
452href="#linkage">linkage types</a>.</p>
453
454</div>
455
456<!-- ======================================================================= -->
457<div class="doc_subsection">
458 <a name="linkage">Linkage Types</a>
459</div>
460
461<div class="doc_text">
462
463<p>
464All Global Variables and Functions have one of the following types of linkage:
465</p>
466
467<dl>
468
Dale Johannesen96e7e092008-05-23 23:13:41 +0000[diff] [blame]469 <dt><tt><b><a name="linkage_internal">internal</a></b></tt>: </dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]470
471 <dd>Global values with internal linkage are only directly accessible by
472 objects in the current module. In particular, linking code into a module with
473 an internal global value may cause the internal to be renamed as necessary to
474 avoid collisions. Because the symbol is internal to the module, all
475 references can be updated. This corresponds to the notion of the
476 '<tt>static</tt>' keyword in C.
477 </dd>
478
479 <dt><tt><b><a name="linkage_linkonce">linkonce</a></b></tt>: </dt>
480
481 <dd>Globals with "<tt>linkonce</tt>" linkage are merged with other globals of
482 the same name when linkage occurs. This is typically used to implement
483 inline functions, templates, or other code which must be generated in each
484 translation unit that uses it. Unreferenced <tt>linkonce</tt> globals are
485 allowed to be discarded.
486 </dd>
487
Dale Johannesen96e7e092008-05-23 23:13:41 +0000[diff] [blame]488 <dt><tt><b><a name="linkage_common">common</a></b></tt>: </dt>
489
490 <dd>"<tt>common</tt>" linkage is exactly the same as <tt>linkonce</tt>
491 linkage, except that unreferenced <tt>common</tt> globals may not be
492 discarded. This is used for globals that may be emitted in multiple
493 translation units, but that are not guaranteed to be emitted into every
494 translation unit that uses them. One example of this is tentative
495 definitions in C, such as "<tt>int X;</tt>" at global scope.
496 </dd>
497
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]498 <dt><tt><b><a name="linkage_weak">weak</a></b></tt>: </dt>
499
Dale Johannesen96e7e092008-05-23 23:13:41 +0000[diff] [blame]500 <dd>"<tt>weak</tt>" linkage is the same as <tt>common</tt> linkage, except
501 that some targets may choose to emit different assembly sequences for them
502 for target-dependent reasons. This is used for globals that are declared
503 "weak" in C source code.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]504 </dd>
505
506 <dt><tt><b><a name="linkage_appending">appending</a></b></tt>: </dt>
507
508 <dd>"<tt>appending</tt>" linkage may only be applied to global variables of
509 pointer to array type. When two global variables with appending linkage are
510 linked together, the two global arrays are appended together. This is the
511 LLVM, typesafe, equivalent of having the system linker append together
512 "sections" with identical names when .o files are linked.
513 </dd>
514
515 <dt><tt><b><a name="linkage_externweak">extern_weak</a></b></tt>: </dt>
516 <dd>The semantics of this linkage follow the ELF model: the symbol is weak
517 until linked, if not linked, the symbol becomes null instead of being an
518 undefined reference.
519 </dd>
520
521 <dt><tt><b><a name="linkage_external">externally visible</a></b></tt>:</dt>
522
523 <dd>If none of the above identifiers are used, the global is externally
524 visible, meaning that it participates in linkage and can be used to resolve
525 external symbol references.
526 </dd>
527</dl>
528
529 <p>
530 The next two types of linkage are targeted for Microsoft Windows platform
531 only. They are designed to support importing (exporting) symbols from (to)
532 DLLs.
533 </p>
534
535 <dl>
536 <dt><tt><b><a name="linkage_dllimport">dllimport</a></b></tt>: </dt>
537
538 <dd>"<tt>dllimport</tt>" linkage causes the compiler to reference a function
539 or variable via a global pointer to a pointer that is set up by the DLL
540 exporting the symbol. On Microsoft Windows targets, the pointer name is
541 formed by combining <code>_imp__</code> and the function or variable name.
542 </dd>
543
544 <dt><tt><b><a name="linkage_dllexport">dllexport</a></b></tt>: </dt>
545
546 <dd>"<tt>dllexport</tt>" linkage causes the compiler to provide a global
547 pointer to a pointer in a DLL, so that it can be referenced with the
548 <tt>dllimport</tt> attribute. On Microsoft Windows targets, the pointer
549 name is formed by combining <code>_imp__</code> and the function or variable
550 name.
551 </dd>
552
553</dl>
554
555<p><a name="linkage_external"></a>For example, since the "<tt>.LC0</tt>"
556variable is defined to be internal, if another module defined a "<tt>.LC0</tt>"
557variable and was linked with this one, one of the two would be renamed,
558preventing a collision. Since "<tt>main</tt>" and "<tt>puts</tt>" are
559external (i.e., lacking any linkage declarations), they are accessible
560outside of the current module.</p>
561<p>It is illegal for a function <i>declaration</i>
562to have any linkage type other than "externally visible", <tt>dllimport</tt>,
563or <tt>extern_weak</tt>.</p>
564<p>Aliases can have only <tt>external</tt>, <tt>internal</tt> and <tt>weak</tt>
565linkages.
566</div>
567
568<!-- ======================================================================= -->
569<div class="doc_subsection">
570 <a name="callingconv">Calling Conventions</a>
571</div>
572
573<div class="doc_text">
574
575<p>LLVM <a href="#functionstructure">functions</a>, <a href="#i_call">calls</a>
576and <a href="#i_invoke">invokes</a> can all have an optional calling convention
577specified for the call. The calling convention of any pair of dynamic
578caller/callee must match, or the behavior of the program is undefined. The
579following calling conventions are supported by LLVM, and more may be added in
580the future:</p>
581
582<dl>
583 <dt><b>"<tt>ccc</tt>" - The C calling convention</b>:</dt>
584
585 <dd>This calling convention (the default if no other calling convention is
586 specified) matches the target C calling conventions. This calling convention
587 supports varargs function calls and tolerates some mismatch in the declared
588 prototype and implemented declaration of the function (as does normal C).
589 </dd>
590
591 <dt><b>"<tt>fastcc</tt>" - The fast calling convention</b>:</dt>
592
593 <dd>This calling convention attempts to make calls as fast as possible
594 (e.g. by passing things in registers). This calling convention allows the
595 target to use whatever tricks it wants to produce fast code for the target,
596 without having to conform to an externally specified ABI. Implementations of
Arnold Schwaighofer07444922008-05-14 09:17:12 +0000[diff] [blame]597 this convention should allow arbitrary
598 <a href="CodeGenerator.html#tailcallopt">tail call optimization</a> to be
599 supported. This calling convention does not support varargs and requires the
600 prototype of all callees to exactly match the prototype of the function
601 definition.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]602 </dd>
603
604 <dt><b>"<tt>coldcc</tt>" - The cold calling convention</b>:</dt>
605
606 <dd>This calling convention attempts to make code in the caller as efficient
607 as possible under the assumption that the call is not commonly executed. As
608 such, these calls often preserve all registers so that the call does not break
609 any live ranges in the caller side. This calling convention does not support
610 varargs and requires the prototype of all callees to exactly match the
611 prototype of the function definition.
612 </dd>
613
614 <dt><b>"<tt>cc &lt;<em>n</em>&gt;</tt>" - Numbered convention</b>:</dt>
615
616 <dd>Any calling convention may be specified by number, allowing
617 target-specific calling conventions to be used. Target specific calling
618 conventions start at 64.
619 </dd>
620</dl>
621
622<p>More calling conventions can be added/defined on an as-needed basis, to
623support pascal conventions or any other well-known target-independent
624convention.</p>
625
626</div>
627
628<!-- ======================================================================= -->
629<div class="doc_subsection">
630 <a name="visibility">Visibility Styles</a>
631</div>
632
633<div class="doc_text">
634
635<p>
636All Global Variables and Functions have one of the following visibility styles:
637</p>
638
639<dl>
640 <dt><b>"<tt>default</tt>" - Default style</b>:</dt>
641
642 <dd>On ELF, default visibility means that the declaration is visible to other
643 modules and, in shared libraries, means that the declared entity may be
644 overridden. On Darwin, default visibility means that the declaration is
645 visible to other modules. Default visibility corresponds to "external
646 linkage" in the language.
647 </dd>
648
649 <dt><b>"<tt>hidden</tt>" - Hidden style</b>:</dt>
650
651 <dd>Two declarations of an object with hidden visibility refer to the same
652 object if they are in the same shared object. Usually, hidden visibility
653 indicates that the symbol will not be placed into the dynamic symbol table,
654 so no other module (executable or shared library) can reference it
655 directly.
656 </dd>
657
658 <dt><b>"<tt>protected</tt>" - Protected style</b>:</dt>
659
660 <dd>On ELF, protected visibility indicates that the symbol will be placed in
661 the dynamic symbol table, but that references within the defining module will
662 bind to the local symbol. That is, the symbol cannot be overridden by another
663 module.
664 </dd>
665</dl>
666
667</div>
668
669<!-- ======================================================================= -->
670<div class="doc_subsection">
671 <a name="globalvars">Global Variables</a>
672</div>
673
674<div class="doc_text">
675
676<p>Global variables define regions of memory allocated at compilation time
677instead of run-time. Global variables may optionally be initialized, may have
678an explicit section to be placed in, and may have an optional explicit alignment
679specified. A variable may be defined as "thread_local", which means that it
680will not be shared by threads (each thread will have a separated copy of the
681variable). A variable may be defined as a global "constant," which indicates
682that the contents of the variable will <b>never</b> be modified (enabling better
683optimization, allowing the global data to be placed in the read-only section of
684an executable, etc). Note that variables that need runtime initialization
685cannot be marked "constant" as there is a store to the variable.</p>
686
687<p>
688LLVM explicitly allows <em>declarations</em> of global variables to be marked
689constant, even if the final definition of the global is not. This capability
690can be used to enable slightly better optimization of the program, but requires
691the language definition to guarantee that optimizations based on the
692'constantness' are valid for the translation units that do not include the
693definition.
694</p>
695
696<p>As SSA values, global variables define pointer values that are in
697scope (i.e. they dominate) all basic blocks in the program. Global
698variables always define a pointer to their "content" type because they
699describe a region of memory, and all memory objects in LLVM are
700accessed through pointers.</p>
701
Christopher Lambdd0049d2007-12-11 09:31:00 +0000[diff] [blame]702<p>A global variable may be declared to reside in a target-specifc numbered
703address space. For targets that support them, address spaces may affect how
704optimizations are performed and/or what target instructions are used to access
Christopher Lamb20a39e92007-12-12 08:44:39 +0000[diff] [blame]705the variable. The default address space is zero. The address space qualifier
706must precede any other attributes.</p>
Christopher Lambdd0049d2007-12-11 09:31:00 +0000[diff] [blame]707
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]708<p>LLVM allows an explicit section to be specified for globals. If the target
709supports it, it will emit globals to the section specified.</p>
710
711<p>An explicit alignment may be specified for a global. If not present, or if
712the alignment is set to zero, the alignment of the global is set by the target
713to whatever it feels convenient. If an explicit alignment is specified, the
714global is forced to have at least that much alignment. All alignments must be
715a power of 2.</p>
716
Christopher Lambdd0049d2007-12-11 09:31:00 +0000[diff] [blame]717<p>For example, the following defines a global in a numbered address space with
718an initializer, section, and alignment:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]719
720<div class="doc_code">
721<pre>
Christopher Lambdd0049d2007-12-11 09:31:00 +0000[diff] [blame]722@G = constant float 1.0 addrspace(5), section "foo", align 4
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]723</pre>
724</div>
725
726</div>
727
728
729<!-- ======================================================================= -->
730<div class="doc_subsection">
731 <a name="functionstructure">Functions</a>
732</div>
733
734<div class="doc_text">
735
736<p>LLVM function definitions consist of the "<tt>define</tt>" keyord,
737an optional <a href="#linkage">linkage type</a>, an optional
738<a href="#visibility">visibility style</a>, an optional
739<a href="#callingconv">calling convention</a>, a return type, an optional
740<a href="#paramattrs">parameter attribute</a> for the return type, a function
741name, a (possibly empty) argument list (each with optional
742<a href="#paramattrs">parameter attributes</a>), an optional section, an
Gordon Henriksend606f5b2007-12-10 03:30:21 +0000[diff] [blame]743optional alignment, an optional <a href="#gc">garbage collector name</a>, an
Gordon Henriksen13fe5e32007-12-10 03:18:06 +0000[diff] [blame]744opening curly brace, a list of basic blocks, and a closing curly brace.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]745
746LLVM function declarations consist of the "<tt>declare</tt>" keyword, an
747optional <a href="#linkage">linkage type</a>, an optional
748<a href="#visibility">visibility style</a>, an optional
749<a href="#callingconv">calling convention</a>, a return type, an optional
750<a href="#paramattrs">parameter attribute</a> for the return type, a function
Gordon Henriksen13fe5e32007-12-10 03:18:06 +0000[diff] [blame]751name, a possibly empty list of arguments, an optional alignment, and an optional
Gordon Henriksend606f5b2007-12-10 03:30:21 +0000[diff] [blame]752<a href="#gc">garbage collector name</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]753
754<p>A function definition contains a list of basic blocks, forming the CFG for
755the function. Each basic block may optionally start with a label (giving the
756basic block a symbol table entry), contains a list of instructions, and ends
757with a <a href="#terminators">terminator</a> instruction (such as a branch or
758function return).</p>
759
760<p>The first basic block in a function is special in two ways: it is immediately
761executed on entrance to the function, and it is not allowed to have predecessor
762basic blocks (i.e. there can not be any branches to the entry block of a
763function). Because the block can have no predecessors, it also cannot have any
764<a href="#i_phi">PHI nodes</a>.</p>
765
766<p>LLVM allows an explicit section to be specified for functions. If the target
767supports it, it will emit functions to the section specified.</p>
768
769<p>An explicit alignment may be specified for a function. If not present, or if
770the alignment is set to zero, the alignment of the function is set by the target
771to whatever it feels convenient. If an explicit alignment is specified, the
772function is forced to have at least that much alignment. All alignments must be
773a power of 2.</p>
774
775</div>
776
777
778<!-- ======================================================================= -->
779<div class="doc_subsection">
780 <a name="aliasstructure">Aliases</a>
781</div>
782<div class="doc_text">
783 <p>Aliases act as "second name" for the aliasee value (which can be either
Anton Korobeynikov96822812008-03-22 08:36:14 +0000[diff] [blame]784 function, global variable, another alias or bitcast of global value). Aliases
785 may have an optional <a href="#linkage">linkage type</a>, and an
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]786 optional <a href="#visibility">visibility style</a>.</p>
787
788 <h5>Syntax:</h5>
789
790<div class="doc_code">
791<pre>
792@&lt;Name&gt; = [Linkage] [Visibility] alias &lt;AliaseeTy&gt; @&lt;Aliasee&gt;
793</pre>
794</div>
795
796</div>
797
798
799
800<!-- ======================================================================= -->
801<div class="doc_subsection"><a name="paramattrs">Parameter Attributes</a></div>
802<div class="doc_text">
803 <p>The return type and each parameter of a function type may have a set of
804 <i>parameter attributes</i> associated with them. Parameter attributes are
805 used to communicate additional information about the result or parameters of
Duncan Sandsf5588dc2007-11-27 13:23:08 +0000[diff] [blame]806 a function. Parameter attributes are considered to be part of the function,
807 not of the function type, so functions with different parameter attributes
808 can have the same function type.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]809
810 <p>Parameter attributes are simple keywords that follow the type specified. If
811 multiple parameter attributes are needed, they are space separated. For
812 example:</p>
813
814<div class="doc_code">
815<pre>
Duncan Sandsf5588dc2007-11-27 13:23:08 +0000[diff] [blame]816declare i32 @printf(i8* noalias , ...) nounwind
817declare i32 @atoi(i8*) nounwind readonly
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]818</pre>
819</div>
820
Duncan Sandsf5588dc2007-11-27 13:23:08 +0000[diff] [blame]821 <p>Note that any attributes for the function result (<tt>nounwind</tt>,
822 <tt>readonly</tt>) come immediately after the argument list.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]823
824 <p>Currently, only the following parameter attributes are defined:</p>
825 <dl>
Reid Spencerf234bed2007-07-19 23:13:04 +0000[diff] [blame]826 <dt><tt>zeroext</tt></dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]827 <dd>This indicates that the parameter should be zero extended just before
828 a call to this function.</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]829
Reid Spencerf234bed2007-07-19 23:13:04 +0000[diff] [blame]830 <dt><tt>signext</tt></dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]831 <dd>This indicates that the parameter should be sign extended just before
832 a call to this function.</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]833
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]834 <dt><tt>inreg</tt></dt>
835 <dd>This indicates that the parameter should be placed in register (if
836 possible) during assembling function call. Support for this attribute is
837 target-specific</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]838
839 <dt><tt>byval</tt></dt>
Chris Lattner04c86182008-01-15 04:34:22 +0000[diff] [blame]840 <dd>This indicates that the pointer parameter should really be passed by
841 value to the function. The attribute implies that a hidden copy of the
842 pointee is made between the caller and the callee, so the callee is unable
843 to modify the value in the callee. This attribute is only valid on llvm
844 pointer arguments. It is generally used to pass structs and arrays by
845 value, but is also valid on scalars (even though this is silly).</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]846
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]847 <dt><tt>sret</tt></dt>
Duncan Sands616cc032008-02-18 04:19:38 +0000[diff] [blame]848 <dd>This indicates that the pointer parameter specifies the address of a
849 structure that is the return value of the function in the source program.
Duncan Sandsef0e9e42008-03-17 12:17:41 +0000[diff] [blame]850 Loads and stores to the structure are assumed not to trap.
Duncan Sands616cc032008-02-18 04:19:38 +0000[diff] [blame]851 May only be applied to the first parameter.</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]852
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]853 <dt><tt>noalias</tt></dt>
Owen Andersonc4fc4cd2008-02-18 04:09:01 +0000[diff] [blame]854 <dd>This indicates that the parameter does not alias any global or any other
855 parameter. The caller is responsible for ensuring that this is the case,
856 usually by placing the value in a stack allocation.</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]857
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]858 <dt><tt>noreturn</tt></dt>
859 <dd>This function attribute indicates that the function never returns. This
860 indicates to LLVM that every call to this function should be treated as if
861 an <tt>unreachable</tt> instruction immediately followed the call.</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]862
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]863 <dt><tt>nounwind</tt></dt>
Duncan Sandsef0e9e42008-03-17 12:17:41 +0000[diff] [blame]864 <dd>This function attribute indicates that no exceptions unwind out of the
865 function. Usually this is because the function makes no use of exceptions,
866 but it may also be that the function catches any exceptions thrown when
867 executing it.</dd>
868
Duncan Sands4ee46812007-07-27 19:57:41 +0000[diff] [blame]869 <dt><tt>nest</tt></dt>
870 <dd>This indicates that the parameter can be excised using the
871 <a href="#int_trampoline">trampoline intrinsics</a>.</dd>
Duncan Sands13e13f82007-11-22 20:23:04 +0000[diff] [blame]872 <dt><tt>readonly</tt></dt>
Duncan Sandsd69c0e62007-11-14 21:14:02 +0000[diff] [blame]873 <dd>This function attribute indicates that the function has no side-effects
Duncan Sands13e13f82007-11-22 20:23:04 +0000[diff] [blame]874 except for producing a return value or throwing an exception. The value
875 returned must only depend on the function arguments and/or global variables.
876 It may use values obtained by dereferencing pointers.</dd>
877 <dt><tt>readnone</tt></dt>
878 <dd>A <tt>readnone</tt> function has the same restrictions as a <tt>readonly</tt>
Duncan Sandsd69c0e62007-11-14 21:14:02 +0000[diff] [blame]879 function, but in addition it is not allowed to dereference any pointer arguments
880 or global variables.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]881 </dl>
882
883</div>
884
885<!-- ======================================================================= -->
886<div class="doc_subsection">
Gordon Henriksen13fe5e32007-12-10 03:18:06 +0000[diff] [blame]887 <a name="gc">Garbage Collector Names</a>
888</div>
889
890<div class="doc_text">
891<p>Each function may specify a garbage collector name, which is simply a
892string.</p>
893
894<div class="doc_code"><pre
895>define void @f() gc "name" { ...</pre></div>
896
897<p>The compiler declares the supported values of <i>name</i>. Specifying a
898collector which will cause the compiler to alter its output in order to support
899the named garbage collection algorithm.</p>
900</div>
901
902<!-- ======================================================================= -->
903<div class="doc_subsection">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]904 <a name="moduleasm">Module-Level Inline Assembly</a>
905</div>
906
907<div class="doc_text">
908<p>
909Modules may contain "module-level inline asm" blocks, which corresponds to the
910GCC "file scope inline asm" blocks. These blocks are internally concatenated by
911LLVM and treated as a single unit, but may be separated in the .ll file if
912desired. The syntax is very simple:
913</p>
914
915<div class="doc_code">
916<pre>
917module asm "inline asm code goes here"
918module asm "more can go here"
919</pre>
920</div>
921
922<p>The strings can contain any character by escaping non-printable characters.
923 The escape sequence used is simply "\xx" where "xx" is the two digit hex code
924 for the number.
925</p>
926
927<p>
928 The inline asm code is simply printed to the machine code .s file when
929 assembly code is generated.
930</p>
931</div>
932
933<!-- ======================================================================= -->
934<div class="doc_subsection">
935 <a name="datalayout">Data Layout</a>
936</div>
937
938<div class="doc_text">
939<p>A module may specify a target specific data layout string that specifies how
940data is to be laid out in memory. The syntax for the data layout is simply:</p>
941<pre> target datalayout = "<i>layout specification</i>"</pre>
942<p>The <i>layout specification</i> consists of a list of specifications
943separated by the minus sign character ('-'). Each specification starts with a
944letter and may include other information after the letter to define some
945aspect of the data layout. The specifications accepted are as follows: </p>
946<dl>
947 <dt><tt>E</tt></dt>
948 <dd>Specifies that the target lays out data in big-endian form. That is, the
949 bits with the most significance have the lowest address location.</dd>
950 <dt><tt>e</tt></dt>
951 <dd>Specifies that hte target lays out data in little-endian form. That is,
952 the bits with the least significance have the lowest address location.</dd>
953 <dt><tt>p:<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
954 <dd>This specifies the <i>size</i> of a pointer and its <i>abi</i> and
955 <i>preferred</i> alignments. All sizes are in bits. Specifying the <i>pref</i>
956 alignment is optional. If omitted, the preceding <tt>:</tt> should be omitted
957 too.</dd>
958 <dt><tt>i<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
959 <dd>This specifies the alignment for an integer type of a given bit
960 <i>size</i>. The value of <i>size</i> must be in the range [1,2^23).</dd>
961 <dt><tt>v<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
962 <dd>This specifies the alignment for a vector type of a given bit
963 <i>size</i>.</dd>
964 <dt><tt>f<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
965 <dd>This specifies the alignment for a floating point type of a given bit
966 <i>size</i>. The value of <i>size</i> must be either 32 (float) or 64
967 (double).</dd>
968 <dt><tt>a<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
969 <dd>This specifies the alignment for an aggregate type of a given bit
970 <i>size</i>.</dd>
971</dl>
972<p>When constructing the data layout for a given target, LLVM starts with a
973default set of specifications which are then (possibly) overriden by the
974specifications in the <tt>datalayout</tt> keyword. The default specifications
975are given in this list:</p>
976<ul>
977 <li><tt>E</tt> - big endian</li>
978 <li><tt>p:32:64:64</tt> - 32-bit pointers with 64-bit alignment</li>
979 <li><tt>i1:8:8</tt> - i1 is 8-bit (byte) aligned</li>
980 <li><tt>i8:8:8</tt> - i8 is 8-bit (byte) aligned</li>
981 <li><tt>i16:16:16</tt> - i16 is 16-bit aligned</li>
982 <li><tt>i32:32:32</tt> - i32 is 32-bit aligned</li>
983 <li><tt>i64:32:64</tt> - i64 has abi alignment of 32-bits but preferred
984 alignment of 64-bits</li>
985 <li><tt>f32:32:32</tt> - float is 32-bit aligned</li>
986 <li><tt>f64:64:64</tt> - double is 64-bit aligned</li>
987 <li><tt>v64:64:64</tt> - 64-bit vector is 64-bit aligned</li>
988 <li><tt>v128:128:128</tt> - 128-bit vector is 128-bit aligned</li>
989 <li><tt>a0:0:1</tt> - aggregates are 8-bit aligned</li>
990</ul>
991<p>When llvm is determining the alignment for a given type, it uses the
992following rules:
993<ol>
994 <li>If the type sought is an exact match for one of the specifications, that
995 specification is used.</li>
996 <li>If no match is found, and the type sought is an integer type, then the
997 smallest integer type that is larger than the bitwidth of the sought type is
998 used. If none of the specifications are larger than the bitwidth then the the
999 largest integer type is used. For example, given the default specifications
1000 above, the i7 type will use the alignment of i8 (next largest) while both
1001 i65 and i256 will use the alignment of i64 (largest specified).</li>
1002 <li>If no match is found, and the type sought is a vector type, then the
1003 largest vector type that is smaller than the sought vector type will be used
1004 as a fall back. This happens because <128 x double> can be implemented in
1005 terms of 64 <2 x double>, for example.</li>
1006</ol>
1007</div>
1008
1009<!-- *********************************************************************** -->
1010<div class="doc_section"> <a name="typesystem">Type System</a> </div>
1011<!-- *********************************************************************** -->
1012
1013<div class="doc_text">
1014
1015<p>The LLVM type system is one of the most important features of the
1016intermediate representation. Being typed enables a number of
1017optimizations to be performed on the IR directly, without having to do
1018extra analyses on the side before the transformation. A strong type
1019system makes it easier to read the generated code and enables novel
1020analyses and transformations that are not feasible to perform on normal
1021three address code representations.</p>
1022
1023</div>
1024
1025<!-- ======================================================================= -->
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1026<div class="doc_subsection"> <a name="t_classifications">Type
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1027Classifications</a> </div>
1028<div class="doc_text">
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1029<p>The types fall into a few useful
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1030classifications:</p>
1031
1032<table border="1" cellspacing="0" cellpadding="4">
1033 <tbody>
1034 <tr><th>Classification</th><th>Types</th></tr>
1035 <tr>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1036 <td><a href="#t_integer">integer</a></td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1037 <td><tt>i1, i2, i3, ... i8, ... i16, ... i32, ... i64, ... </tt></td>
1038 </tr>
1039 <tr>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1040 <td><a href="#t_floating">floating point</a></td>
1041 <td><tt>float, double, x86_fp80, fp128, ppc_fp128</tt></td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1042 </tr>
1043 <tr>
1044 <td><a name="t_firstclass">first class</a></td>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1045 <td><a href="#t_integer">integer</a>,
1046 <a href="#t_floating">floating point</a>,
1047 <a href="#t_pointer">pointer</a>,
Dan Gohmanf6237db2008-06-18 18:42:13 +0000[diff] [blame]1048 <a href="#t_vector">vector</a>,
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]1049 <a href="#t_struct">structure</a>,
1050 <a href="#t_array">array</a>,
Dan Gohmand13951e2008-05-23 22:50:26 +0000[diff] [blame]1051 <a href="#t_label">label</a>.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1052 </td>
1053 </tr>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1054 <tr>
1055 <td><a href="#t_primitive">primitive</a></td>
1056 <td><a href="#t_label">label</a>,
1057 <a href="#t_void">void</a>,
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1058 <a href="#t_floating">floating point</a>.</td>
1059 </tr>
1060 <tr>
1061 <td><a href="#t_derived">derived</a></td>
1062 <td><a href="#t_integer">integer</a>,
1063 <a href="#t_array">array</a>,
1064 <a href="#t_function">function</a>,
1065 <a href="#t_pointer">pointer</a>,
1066 <a href="#t_struct">structure</a>,
1067 <a href="#t_pstruct">packed structure</a>,
1068 <a href="#t_vector">vector</a>,
1069 <a href="#t_opaque">opaque</a>.
1070 </tr>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1071 </tbody>
1072</table>
1073
1074<p>The <a href="#t_firstclass">first class</a> types are perhaps the
1075most important. Values of these types are the only ones which can be
1076produced by instructions, passed as arguments, or used as operands to
Dan Gohman4f29e422008-05-23 21:53:15 +0000[diff] [blame]1077instructions.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1078</div>
1079
1080<!-- ======================================================================= -->
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1081<div class="doc_subsection"> <a name="t_primitive">Primitive Types</a> </div>
Chris Lattner86437612008-01-04 04:34:14 +0000[diff] [blame]1082
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1083<div class="doc_text">
1084<p>The primitive types are the fundamental building blocks of the LLVM
1085system.</p>
1086
Chris Lattner86437612008-01-04 04:34:14 +0000[diff] [blame]1087</div>
1088
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1089<!-- _______________________________________________________________________ -->
1090<div class="doc_subsubsection"> <a name="t_floating">Floating Point Types</a> </div>
1091
1092<div class="doc_text">
1093 <table>
1094 <tbody>
1095 <tr><th>Type</th><th>Description</th></tr>
1096 <tr><td><tt>float</tt></td><td>32-bit floating point value</td></tr>
1097 <tr><td><tt>double</tt></td><td>64-bit floating point value</td></tr>
1098 <tr><td><tt>fp128</tt></td><td>128-bit floating point value (112-bit mantissa)</td></tr>
1099 <tr><td><tt>x86_fp80</tt></td><td>80-bit floating point value (X87)</td></tr>
1100 <tr><td><tt>ppc_fp128</tt></td><td>128-bit floating point value (two 64-bits)</td></tr>
1101 </tbody>
1102 </table>
1103</div>
1104
1105<!-- _______________________________________________________________________ -->
1106<div class="doc_subsubsection"> <a name="t_void">Void Type</a> </div>
1107
1108<div class="doc_text">
1109<h5>Overview:</h5>
1110<p>The void type does not represent any value and has no size.</p>
1111
1112<h5>Syntax:</h5>
1113
1114<pre>
1115 void
1116</pre>
1117</div>
1118
1119<!-- _______________________________________________________________________ -->
1120<div class="doc_subsubsection"> <a name="t_label">Label Type</a> </div>
1121
1122<div class="doc_text">
1123<h5>Overview:</h5>
1124<p>The label type represents code labels.</p>
1125
1126<h5>Syntax:</h5>
1127
1128<pre>
1129 label
1130</pre>
1131</div>
1132
1133
1134<!-- ======================================================================= -->
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1135<div class="doc_subsection"> <a name="t_derived">Derived Types</a> </div>
1136
1137<div class="doc_text">
1138
1139<p>The real power in LLVM comes from the derived types in the system.
1140This is what allows a programmer to represent arrays, functions,
1141pointers, and other useful types. Note that these derived types may be
1142recursive: For example, it is possible to have a two dimensional array.</p>
1143
1144</div>
1145
1146<!-- _______________________________________________________________________ -->
1147<div class="doc_subsubsection"> <a name="t_integer">Integer Type</a> </div>
1148
1149<div class="doc_text">
1150
1151<h5>Overview:</h5>
1152<p>The integer type is a very simple derived type that simply specifies an
1153arbitrary bit width for the integer type desired. Any bit width from 1 bit to
11542^23-1 (about 8 million) can be specified.</p>
1155
1156<h5>Syntax:</h5>
1157
1158<pre>
1159 iN
1160</pre>
1161
1162<p>The number of bits the integer will occupy is specified by the <tt>N</tt>
1163value.</p>
1164
1165<h5>Examples:</h5>
1166<table class="layout">
Chris Lattner251ab812007-12-18 06:18:21 +0000[diff] [blame]1167 <tbody>
1168 <tr>
1169 <td><tt>i1</tt></td>
1170 <td>a single-bit integer.</td>
1171 </tr><tr>
1172 <td><tt>i32</tt></td>
1173 <td>a 32-bit integer.</td>
1174 </tr><tr>
1175 <td><tt>i1942652</tt></td>
1176 <td>a really big integer of over 1 million bits.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1177 </tr>
Chris Lattner251ab812007-12-18 06:18:21 +0000[diff] [blame]1178 </tbody>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1179</table>
1180</div>
1181
1182<!-- _______________________________________________________________________ -->
1183<div class="doc_subsubsection"> <a name="t_array">Array Type</a> </div>
1184
1185<div class="doc_text">
1186
1187<h5>Overview:</h5>
1188
1189<p>The array type is a very simple derived type that arranges elements
1190sequentially in memory. The array type requires a size (number of
1191elements) and an underlying data type.</p>
1192
1193<h5>Syntax:</h5>
1194
1195<pre>
1196 [&lt;# elements&gt; x &lt;elementtype&gt;]
1197</pre>
1198
1199<p>The number of elements is a constant integer value; elementtype may
1200be any type with a size.</p>
1201
1202<h5>Examples:</h5>
1203<table class="layout">
1204 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1205 <td class="left"><tt>[40 x i32]</tt></td>
1206 <td class="left">Array of 40 32-bit integer values.</td>
1207 </tr>
1208 <tr class="layout">
1209 <td class="left"><tt>[41 x i32]</tt></td>
1210 <td class="left">Array of 41 32-bit integer values.</td>
1211 </tr>
1212 <tr class="layout">
1213 <td class="left"><tt>[4 x i8]</tt></td>
1214 <td class="left">Array of 4 8-bit integer values.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1215 </tr>
1216</table>
1217<p>Here are some examples of multidimensional arrays:</p>
1218<table class="layout">
1219 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1220 <td class="left"><tt>[3 x [4 x i32]]</tt></td>
1221 <td class="left">3x4 array of 32-bit integer values.</td>
1222 </tr>
1223 <tr class="layout">
1224 <td class="left"><tt>[12 x [10 x float]]</tt></td>
1225 <td class="left">12x10 array of single precision floating point values.</td>
1226 </tr>
1227 <tr class="layout">
1228 <td class="left"><tt>[2 x [3 x [4 x i16]]]</tt></td>
1229 <td class="left">2x3x4 array of 16-bit integer values.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1230 </tr>
1231</table>
1232
1233<p>Note that 'variable sized arrays' can be implemented in LLVM with a zero
1234length array. Normally, accesses past the end of an array are undefined in
1235LLVM (e.g. it is illegal to access the 5th element of a 3 element array).
1236As a special case, however, zero length arrays are recognized to be variable
1237length. This allows implementation of 'pascal style arrays' with the LLVM
1238type "{ i32, [0 x float]}", for example.</p>
1239
1240</div>
1241
1242<!-- _______________________________________________________________________ -->
1243<div class="doc_subsubsection"> <a name="t_function">Function Type</a> </div>
1244<div class="doc_text">
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1245
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1246<h5>Overview:</h5>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1247
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1248<p>The function type can be thought of as a function signature. It
Devang Pateld4ba41d2008-03-24 05:35:41 +0000[diff] [blame]1249consists of a return type and a list of formal parameter types. The
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1250return 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]1251If 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]1252class types, and the struct must have at least one element.</p>
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]1253
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1254<h5>Syntax:</h5>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1255
1256<pre>
1257 &lt;returntype list&gt; (&lt;parameter list&gt;)
1258</pre>
1259
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1260<p>...where '<tt>&lt;parameter list&gt;</tt>' is a comma-separated list of type
1261specifiers. Optionally, the parameter list may include a type <tt>...</tt>,
1262which indicates that the function takes a variable number of arguments.
1263Variable argument functions can access their arguments with the <a
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]1264 href="#int_varargs">variable argument handling intrinsic</a> functions.
1265'<tt>&lt;returntype list&gt;</tt>' is a comma-separated list of
1266<a href="#t_firstclass">first class</a> type specifiers.</p>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1267
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1268<h5>Examples:</h5>
1269<table class="layout">
1270 <tr class="layout">
1271 <td class="left"><tt>i32 (i32)</tt></td>
1272 <td class="left">function taking an <tt>i32</tt>, returning an <tt>i32</tt>
1273 </td>
1274 </tr><tr class="layout">
Reid Spencerf234bed2007-07-19 23:13:04 +0000[diff] [blame]1275 <td class="left"><tt>float&nbsp;(i16&nbsp;signext,&nbsp;i32&nbsp;*)&nbsp;*
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1276 </tt></td>
1277 <td class="left"><a href="#t_pointer">Pointer</a> to a function that takes
1278 an <tt>i16</tt> that should be sign extended and a
1279 <a href="#t_pointer">pointer</a> to <tt>i32</tt>, returning
1280 <tt>float</tt>.
1281 </td>
1282 </tr><tr class="layout">
1283 <td class="left"><tt>i32 (i8*, ...)</tt></td>
1284 <td class="left">A vararg function that takes at least one
1285 <a href="#t_pointer">pointer</a> to <tt>i8 </tt> (char in C),
1286 which returns an integer. This is the signature for <tt>printf</tt> in
1287 LLVM.
1288 </td>
Devang Pateld4ba41d2008-03-24 05:35:41 +0000[diff] [blame]1289 </tr><tr class="layout">
1290 <td class="left"><tt>{i32, i32} (i32)</tt></td>
Devang Patel6dddba22008-03-24 18:10:52 +0000[diff] [blame]1291 <td class="left">A function taking an <tt>i32></tt>, returning two
1292 <tt> i32 </tt> values as an aggregate of type <tt>{ i32, i32 }</tt>
Devang Pateld4ba41d2008-03-24 05:35:41 +0000[diff] [blame]1293 </td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1294 </tr>
1295</table>
1296
1297</div>
1298<!-- _______________________________________________________________________ -->
1299<div class="doc_subsubsection"> <a name="t_struct">Structure Type</a> </div>
1300<div class="doc_text">
1301<h5>Overview:</h5>
1302<p>The structure type is used to represent a collection of data members
1303together in memory. The packing of the field types is defined to match
1304the ABI of the underlying processor. The elements of a structure may
1305be any type that has a size.</p>
1306<p>Structures are accessed using '<tt><a href="#i_load">load</a></tt>
1307and '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a
1308field with the '<tt><a href="#i_getelementptr">getelementptr</a></tt>'
1309instruction.</p>
1310<h5>Syntax:</h5>
1311<pre> { &lt;type list&gt; }<br></pre>
1312<h5>Examples:</h5>
1313<table class="layout">
1314 <tr class="layout">
1315 <td class="left"><tt>{ i32, i32, i32 }</tt></td>
1316 <td class="left">A triple of three <tt>i32</tt> values</td>
1317 </tr><tr class="layout">
1318 <td class="left"><tt>{&nbsp;float,&nbsp;i32&nbsp;(i32)&nbsp;*&nbsp;}</tt></td>
1319 <td class="left">A pair, where the first element is a <tt>float</tt> and the
1320 second element is a <a href="#t_pointer">pointer</a> to a
1321 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
1322 an <tt>i32</tt>.</td>
1323 </tr>
1324</table>
1325</div>
1326
1327<!-- _______________________________________________________________________ -->
1328<div class="doc_subsubsection"> <a name="t_pstruct">Packed Structure Type</a>
1329</div>
1330<div class="doc_text">
1331<h5>Overview:</h5>
1332<p>The packed structure type is used to represent a collection of data members
1333together in memory. There is no padding between fields. Further, the alignment
1334of a packed structure is 1 byte. The elements of a packed structure may
1335be any type that has a size.</p>
1336<p>Structures are accessed using '<tt><a href="#i_load">load</a></tt>
1337and '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a
1338field with the '<tt><a href="#i_getelementptr">getelementptr</a></tt>'
1339instruction.</p>
1340<h5>Syntax:</h5>
1341<pre> &lt; { &lt;type list&gt; } &gt; <br></pre>
1342<h5>Examples:</h5>
1343<table class="layout">
1344 <tr class="layout">
1345 <td class="left"><tt>&lt; { i32, i32, i32 } &gt;</tt></td>
1346 <td class="left">A triple of three <tt>i32</tt> values</td>
1347 </tr><tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1348 <td class="left"><tt>&lt; { float, i32 (i32)* } &gt;</tt></td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1349 <td class="left">A pair, where the first element is a <tt>float</tt> and the
1350 second element is a <a href="#t_pointer">pointer</a> to a
1351 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
1352 an <tt>i32</tt>.</td>
1353 </tr>
1354</table>
1355</div>
1356
1357<!-- _______________________________________________________________________ -->
1358<div class="doc_subsubsection"> <a name="t_pointer">Pointer Type</a> </div>
1359<div class="doc_text">
1360<h5>Overview:</h5>
1361<p>As in many languages, the pointer type represents a pointer or
Christopher Lambdd0049d2007-12-11 09:31:00 +0000[diff] [blame]1362reference to another object, which must live in memory. Pointer types may have
1363an optional address space attribute defining the target-specific numbered
1364address space where the pointed-to object resides. The default address space is
1365zero.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1366<h5>Syntax:</h5>
1367<pre> &lt;type&gt; *<br></pre>
1368<h5>Examples:</h5>
1369<table class="layout">
1370 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1371 <td class="left"><tt>[4x i32]*</tt></td>
1372 <td class="left">A <a href="#t_pointer">pointer</a> to <a
1373 href="#t_array">array</a> of four <tt>i32</tt> values.</td>
1374 </tr>
1375 <tr class="layout">
1376 <td class="left"><tt>i32 (i32 *) *</tt></td>
1377 <td class="left"> A <a href="#t_pointer">pointer</a> to a <a
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1378 href="#t_function">function</a> that takes an <tt>i32*</tt>, returning an
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1379 <tt>i32</tt>.</td>
1380 </tr>
1381 <tr class="layout">
1382 <td class="left"><tt>i32 addrspace(5)*</tt></td>
1383 <td class="left">A <a href="#t_pointer">pointer</a> to an <tt>i32</tt> value
1384 that resides in address space #5.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1385 </tr>
1386</table>
1387</div>
1388
1389<!-- _______________________________________________________________________ -->
1390<div class="doc_subsubsection"> <a name="t_vector">Vector Type</a> </div>
1391<div class="doc_text">
1392
1393<h5>Overview:</h5>
1394
1395<p>A vector type is a simple derived type that represents a vector
1396of elements. Vector types are used when multiple primitive data
1397are operated in parallel using a single instruction (SIMD).
1398A vector type requires a size (number of
1399elements) and an underlying primitive data type. Vectors must have a power
1400of two length (1, 2, 4, 8, 16 ...). Vector types are
1401considered <a href="#t_firstclass">first class</a>.</p>
1402
1403<h5>Syntax:</h5>
1404
1405<pre>
1406 &lt; &lt;# elements&gt; x &lt;elementtype&gt; &gt;
1407</pre>
1408
1409<p>The number of elements is a constant integer value; elementtype may
1410be any integer or floating point type.</p>
1411
1412<h5>Examples:</h5>
1413
1414<table class="layout">
1415 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1416 <td class="left"><tt>&lt;4 x i32&gt;</tt></td>
1417 <td class="left">Vector of 4 32-bit integer values.</td>
1418 </tr>
1419 <tr class="layout">
1420 <td class="left"><tt>&lt;8 x float&gt;</tt></td>
1421 <td class="left">Vector of 8 32-bit floating-point values.</td>
1422 </tr>
1423 <tr class="layout">
1424 <td class="left"><tt>&lt;2 x i64&gt;</tt></td>
1425 <td class="left">Vector of 2 64-bit integer values.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1426 </tr>
1427</table>
1428</div>
1429
1430<!-- _______________________________________________________________________ -->
1431<div class="doc_subsubsection"> <a name="t_opaque">Opaque Type</a> </div>
1432<div class="doc_text">
1433
1434<h5>Overview:</h5>
1435
1436<p>Opaque types are used to represent unknown types in the system. This
Gordon Henriksenda0706e2007-10-14 00:34:53 +0000[diff] [blame]1437corresponds (for example) to the C notion of a forward declared structure type.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1438In LLVM, opaque types can eventually be resolved to any type (not just a
1439structure type).</p>
1440
1441<h5>Syntax:</h5>
1442
1443<pre>
1444 opaque
1445</pre>
1446
1447<h5>Examples:</h5>
1448
1449<table class="layout">
1450 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1451 <td class="left"><tt>opaque</tt></td>
1452 <td class="left">An opaque type.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1453 </tr>
1454</table>
1455</div>
1456
1457
1458<!-- *********************************************************************** -->
1459<div class="doc_section"> <a name="constants">Constants</a> </div>
1460<!-- *********************************************************************** -->
1461
1462<div class="doc_text">
1463
1464<p>LLVM has several different basic types of constants. This section describes
1465them all and their syntax.</p>
1466
1467</div>
1468
1469<!-- ======================================================================= -->
1470<div class="doc_subsection"><a name="simpleconstants">Simple Constants</a></div>
1471
1472<div class="doc_text">
1473
1474<dl>
1475 <dt><b>Boolean constants</b></dt>
1476
1477 <dd>The two strings '<tt>true</tt>' and '<tt>false</tt>' are both valid
1478 constants of the <tt><a href="#t_primitive">i1</a></tt> type.
1479 </dd>
1480
1481 <dt><b>Integer constants</b></dt>
1482
1483 <dd>Standard integers (such as '4') are constants of the <a
1484 href="#t_integer">integer</a> type. Negative numbers may be used with
1485 integer types.
1486 </dd>
1487
1488 <dt><b>Floating point constants</b></dt>
1489
1490 <dd>Floating point constants use standard decimal notation (e.g. 123.421),
1491 exponential notation (e.g. 1.23421e+2), or a more precise hexadecimal
Chris Lattnerdc15b1d2008-04-01 18:45:27 +0000[diff] [blame]1492 notation (see below). The assembler requires the exact decimal value of
1493 a floating-point constant. For example, the assembler accepts 1.25 but
1494 rejects 1.3 because 1.3 is a repeating decimal in binary. Floating point
1495 constants must have a <a href="#t_floating">floating point</a> type. </dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1496
1497 <dt><b>Null pointer constants</b></dt>
1498
1499 <dd>The identifier '<tt>null</tt>' is recognized as a null pointer constant
1500 and must be of <a href="#t_pointer">pointer type</a>.</dd>
1501
1502</dl>
1503
1504<p>The one non-intuitive notation for constants is the optional hexadecimal form
1505of floating point constants. For example, the form '<tt>double
15060x432ff973cafa8000</tt>' is equivalent to (but harder to read than) '<tt>double
15074.5e+15</tt>'. The only time hexadecimal floating point constants are required
1508(and the only time that they are generated by the disassembler) is when a
1509floating point constant must be emitted but it cannot be represented as a
1510decimal floating point number. For example, NaN's, infinities, and other
1511special values are represented in their IEEE hexadecimal format so that
1512assembly and disassembly do not cause any bits to change in the constants.</p>
1513
1514</div>
1515
1516<!-- ======================================================================= -->
1517<div class="doc_subsection"><a name="aggregateconstants">Aggregate Constants</a>
1518</div>
1519
1520<div class="doc_text">
1521<p>Aggregate constants arise from aggregation of simple constants
1522and smaller aggregate constants.</p>
1523
1524<dl>
1525 <dt><b>Structure constants</b></dt>
1526
1527 <dd>Structure constants are represented with notation similar to structure
1528 type definitions (a comma separated list of elements, surrounded by braces
Chris Lattner6c8de962007-12-25 20:34:52 +0000[diff] [blame]1529 (<tt>{}</tt>)). For example: "<tt>{ i32 4, float 17.0, i32* @G }</tt>",
1530 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]1531 must have <a href="#t_struct">structure type</a>, and the number and
1532 types of elements must match those specified by the type.
1533 </dd>
1534
1535 <dt><b>Array constants</b></dt>
1536
1537 <dd>Array constants are represented with notation similar to array type
1538 definitions (a comma separated list of elements, surrounded by square brackets
1539 (<tt>[]</tt>)). For example: "<tt>[ i32 42, i32 11, i32 74 ]</tt>". Array
1540 constants must have <a href="#t_array">array type</a>, and the number and
1541 types of elements must match those specified by the type.
1542 </dd>
1543
1544 <dt><b>Vector constants</b></dt>
1545
1546 <dd>Vector constants are represented with notation similar to vector type
1547 definitions (a comma separated list of elements, surrounded by
1548 less-than/greater-than's (<tt>&lt;&gt;</tt>)). For example: "<tt>&lt; i32 42,
1549 i32 11, i32 74, i32 100 &gt;</tt>". Vector constants must have <a
1550 href="#t_vector">vector type</a>, and the number and types of elements must
1551 match those specified by the type.
1552 </dd>
1553
1554 <dt><b>Zero initialization</b></dt>
1555
1556 <dd>The string '<tt>zeroinitializer</tt>' can be used to zero initialize a
1557 value to zero of <em>any</em> type, including scalar and aggregate types.
1558 This is often used to avoid having to print large zero initializers (e.g. for
1559 large arrays) and is always exactly equivalent to using explicit zero
1560 initializers.
1561 </dd>
1562</dl>
1563
1564</div>
1565
1566<!-- ======================================================================= -->
1567<div class="doc_subsection">
1568 <a name="globalconstants">Global Variable and Function Addresses</a>
1569</div>
1570
1571<div class="doc_text">
1572
1573<p>The addresses of <a href="#globalvars">global variables</a> and <a
1574href="#functionstructure">functions</a> are always implicitly valid (link-time)
1575constants. These constants are explicitly referenced when the <a
1576href="#identifiers">identifier for the global</a> is used and always have <a
1577href="#t_pointer">pointer</a> type. For example, the following is a legal LLVM
1578file:</p>
1579
1580<div class="doc_code">
1581<pre>
1582@X = global i32 17
1583@Y = global i32 42
1584@Z = global [2 x i32*] [ i32* @X, i32* @Y ]
1585</pre>
1586</div>
1587
1588</div>
1589
1590<!-- ======================================================================= -->
1591<div class="doc_subsection"><a name="undefvalues">Undefined Values</a></div>
1592<div class="doc_text">
1593 <p>The string '<tt>undef</tt>' is recognized as a type-less constant that has
1594 no specific value. Undefined values may be of any type and be used anywhere
1595 a constant is permitted.</p>
1596
1597 <p>Undefined values indicate to the compiler that the program is well defined
1598 no matter what value is used, giving the compiler more freedom to optimize.
1599 </p>
1600</div>
1601
1602<!-- ======================================================================= -->
1603<div class="doc_subsection"><a name="constantexprs">Constant Expressions</a>
1604</div>
1605
1606<div class="doc_text">
1607
1608<p>Constant expressions are used to allow expressions involving other constants
1609to be used as constants. Constant expressions may be of any <a
1610href="#t_firstclass">first class</a> type and may involve any LLVM operation
1611that does not have side effects (e.g. load and call are not supported). The
1612following is the syntax for constant expressions:</p>
1613
1614<dl>
1615 <dt><b><tt>trunc ( CST to TYPE )</tt></b></dt>
1616 <dd>Truncate a constant to another type. The bit size of CST must be larger
1617 than the bit size of TYPE. Both types must be integers.</dd>
1618
1619 <dt><b><tt>zext ( CST to TYPE )</tt></b></dt>
1620 <dd>Zero extend a constant to another type. The bit size of CST must be
1621 smaller or equal to the bit size of TYPE. Both types must be integers.</dd>
1622
1623 <dt><b><tt>sext ( CST to TYPE )</tt></b></dt>
1624 <dd>Sign extend a constant to another type. The bit size of CST must be
1625 smaller or equal to the bit size of TYPE. Both types must be integers.</dd>
1626
1627 <dt><b><tt>fptrunc ( CST to TYPE )</tt></b></dt>
1628 <dd>Truncate a floating point constant to another floating point type. The
1629 size of CST must be larger than the size of TYPE. Both types must be
1630 floating point.</dd>
1631
1632 <dt><b><tt>fpext ( CST to TYPE )</tt></b></dt>
1633 <dd>Floating point extend a constant to another type. The size of CST must be
1634 smaller or equal to the size of TYPE. Both types must be floating point.</dd>
1635
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]1636 <dt><b><tt>fptoui ( CST to TYPE )</tt></b></dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1637 <dd>Convert a floating point constant to the corresponding unsigned integer
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]1638 constant. TYPE must be a scalar or vector integer type. CST must be of scalar
1639 or vector floating point type. Both CST and TYPE must be scalars, or vectors
1640 of the same number of elements. If the value won't fit in the integer type,
1641 the results are undefined.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1642
1643 <dt><b><tt>fptosi ( CST to TYPE )</tt></b></dt>
1644 <dd>Convert a floating point constant to the corresponding signed integer
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]1645 constant. TYPE must be a scalar or vector integer type. CST must be of scalar
1646 or vector floating point type. Both CST and TYPE must be scalars, or vectors
1647 of the same number of elements. If the value won't fit in the integer type,
1648 the results are undefined.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1649
1650 <dt><b><tt>uitofp ( CST to TYPE )</tt></b></dt>
1651 <dd>Convert an unsigned integer constant to the corresponding floating point
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]1652 constant. TYPE must be a scalar or vector floating point type. CST must be of
1653 scalar or vector integer type. Both CST and TYPE must be scalars, or vectors
1654 of the same number of elements. If the value won't fit in the floating point
1655 type, the results are undefined.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1656
1657 <dt><b><tt>sitofp ( CST to TYPE )</tt></b></dt>
1658 <dd>Convert a signed integer constant to the corresponding floating point
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]1659 constant. TYPE must be a scalar or vector floating point type. CST must be of
1660 scalar or vector integer type. Both CST and TYPE must be scalars, or vectors
1661 of the same number of elements. If the value won't fit in the floating point
1662 type, the results are undefined.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1663
1664 <dt><b><tt>ptrtoint ( CST to TYPE )</tt></b></dt>
1665 <dd>Convert a pointer typed constant to the corresponding integer constant
1666 TYPE must be an integer type. CST must be of pointer type. The CST value is
1667 zero extended, truncated, or unchanged to make it fit in TYPE.</dd>
1668
1669 <dt><b><tt>inttoptr ( CST to TYPE )</tt></b></dt>
1670 <dd>Convert a integer constant to a pointer constant. TYPE must be a
1671 pointer type. CST must be of integer type. The CST value is zero extended,
1672 truncated, or unchanged to make it fit in a pointer size. This one is
1673 <i>really</i> dangerous!</dd>
1674
1675 <dt><b><tt>bitcast ( CST to TYPE )</tt></b></dt>
1676 <dd>Convert a constant, CST, to another TYPE. The size of CST and TYPE must be
1677 identical (same number of bits). The conversion is done as if the CST value
1678 was stored to memory and read back as TYPE. In other words, no bits change
1679 with this operator, just the type. This can be used for conversion of
1680 vector types to any other type, as long as they have the same bit width. For
1681 pointers it is only valid to cast to another pointer type.
1682 </dd>
1683
1684 <dt><b><tt>getelementptr ( CSTPTR, IDX0, IDX1, ... )</tt></b></dt>
1685
1686 <dd>Perform the <a href="#i_getelementptr">getelementptr operation</a> on
1687 constants. As with the <a href="#i_getelementptr">getelementptr</a>
1688 instruction, the index list may have zero or more indexes, which are required
1689 to make sense for the type of "CSTPTR".</dd>
1690
1691 <dt><b><tt>select ( COND, VAL1, VAL2 )</tt></b></dt>
1692
1693 <dd>Perform the <a href="#i_select">select operation</a> on
1694 constants.</dd>
1695
1696 <dt><b><tt>icmp COND ( VAL1, VAL2 )</tt></b></dt>
1697 <dd>Performs the <a href="#i_icmp">icmp operation</a> on constants.</dd>
1698
1699 <dt><b><tt>fcmp COND ( VAL1, VAL2 )</tt></b></dt>
1700 <dd>Performs the <a href="#i_fcmp">fcmp operation</a> on constants.</dd>
1701
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]1702 <dt><b><tt>vicmp COND ( VAL1, VAL2 )</tt></b></dt>
1703 <dd>Performs the <a href="#i_vicmp">vicmp operation</a> on constants.</dd>
1704
1705 <dt><b><tt>vfcmp COND ( VAL1, VAL2 )</tt></b></dt>
1706 <dd>Performs the <a href="#i_vfcmp">vfcmp operation</a> on constants.</dd>
1707
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1708 <dt><b><tt>extractelement ( VAL, IDX )</tt></b></dt>
1709
1710 <dd>Perform the <a href="#i_extractelement">extractelement
1711 operation</a> on constants.
1712
1713 <dt><b><tt>insertelement ( VAL, ELT, IDX )</tt></b></dt>
1714
1715 <dd>Perform the <a href="#i_insertelement">insertelement
1716 operation</a> on constants.</dd>
1717
1718
1719 <dt><b><tt>shufflevector ( VEC1, VEC2, IDXMASK )</tt></b></dt>
1720
1721 <dd>Perform the <a href="#i_shufflevector">shufflevector
1722 operation</a> on constants.</dd>
1723
1724 <dt><b><tt>OPCODE ( LHS, RHS )</tt></b></dt>
1725
1726 <dd>Perform the specified operation of the LHS and RHS constants. OPCODE may
1727 be any of the <a href="#binaryops">binary</a> or <a href="#bitwiseops">bitwise
1728 binary</a> operations. The constraints on operands are the same as those for
1729 the corresponding instruction (e.g. no bitwise operations on floating point
1730 values are allowed).</dd>
1731</dl>
1732</div>
1733
1734<!-- *********************************************************************** -->
1735<div class="doc_section"> <a name="othervalues">Other Values</a> </div>
1736<!-- *********************************************************************** -->
1737
1738<!-- ======================================================================= -->
1739<div class="doc_subsection">
1740<a name="inlineasm">Inline Assembler Expressions</a>
1741</div>
1742
1743<div class="doc_text">
1744
1745<p>
1746LLVM supports inline assembler expressions (as opposed to <a href="#moduleasm">
1747Module-Level Inline Assembly</a>) through the use of a special value. This
1748value represents the inline assembler as a string (containing the instructions
1749to emit), a list of operand constraints (stored as a string), and a flag that
1750indicates whether or not the inline asm expression has side effects. An example
1751inline assembler expression is:
1752</p>
1753
1754<div class="doc_code">
1755<pre>
1756i32 (i32) asm "bswap $0", "=r,r"
1757</pre>
1758</div>
1759
1760<p>
1761Inline assembler expressions may <b>only</b> be used as the callee operand of
1762a <a href="#i_call"><tt>call</tt> instruction</a>. Thus, typically we have:
1763</p>
1764
1765<div class="doc_code">
1766<pre>
1767%X = call i32 asm "<a href="#int_bswap">bswap</a> $0", "=r,r"(i32 %Y)
1768</pre>
1769</div>
1770
1771<p>
1772Inline asms with side effects not visible in the constraint list must be marked
1773as having side effects. This is done through the use of the
1774'<tt>sideeffect</tt>' keyword, like so:
1775</p>
1776
1777<div class="doc_code">
1778<pre>
1779call void asm sideeffect "eieio", ""()
1780</pre>
1781</div>
1782
1783<p>TODO: The format of the asm and constraints string still need to be
1784documented here. Constraints on what can be done (e.g. duplication, moving, etc
1785need to be documented).
1786</p>
1787
1788</div>
1789
1790<!-- *********************************************************************** -->
1791<div class="doc_section"> <a name="instref">Instruction Reference</a> </div>
1792<!-- *********************************************************************** -->
1793
1794<div class="doc_text">
1795
1796<p>The LLVM instruction set consists of several different
1797classifications of instructions: <a href="#terminators">terminator
1798instructions</a>, <a href="#binaryops">binary instructions</a>,
1799<a href="#bitwiseops">bitwise binary instructions</a>, <a
1800 href="#memoryops">memory instructions</a>, and <a href="#otherops">other
1801instructions</a>.</p>
1802
1803</div>
1804
1805<!-- ======================================================================= -->
1806<div class="doc_subsection"> <a name="terminators">Terminator
1807Instructions</a> </div>
1808
1809<div class="doc_text">
1810
1811<p>As mentioned <a href="#functionstructure">previously</a>, every
1812basic block in a program ends with a "Terminator" instruction, which
1813indicates which block should be executed after the current block is
1814finished. These terminator instructions typically yield a '<tt>void</tt>'
1815value: they produce control flow, not values (the one exception being
1816the '<a href="#i_invoke"><tt>invoke</tt></a>' instruction).</p>
1817<p>There are six different terminator instructions: the '<a
1818 href="#i_ret"><tt>ret</tt></a>' instruction, the '<a href="#i_br"><tt>br</tt></a>'
1819instruction, the '<a href="#i_switch"><tt>switch</tt></a>' instruction,
1820the '<a href="#i_invoke"><tt>invoke</tt></a>' instruction, the '<a
1821 href="#i_unwind"><tt>unwind</tt></a>' instruction, and the '<a
1822 href="#i_unreachable"><tt>unreachable</tt></a>' instruction.</p>
1823
1824</div>
1825
1826<!-- _______________________________________________________________________ -->
1827<div class="doc_subsubsection"> <a name="i_ret">'<tt>ret</tt>'
1828Instruction</a> </div>
1829<div class="doc_text">
1830<h5>Syntax:</h5>
1831<pre> ret &lt;type&gt; &lt;value&gt; <i>; Return a value from a non-void function</i>
1832 ret void <i>; Return from void function</i>
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]1833 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]1834</pre>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1835
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1836<h5>Overview:</h5>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1837
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1838<p>The '<tt>ret</tt>' instruction is used to return control flow (and a
1839value) from a function back to the caller.</p>
1840<p>There are two forms of the '<tt>ret</tt>' instruction: one that
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1841returns value(s) and then causes control flow, and one that just causes
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1842control flow to occur.</p>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1843
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1844<h5>Arguments:</h5>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1845
1846<p>The '<tt>ret</tt>' instruction may return zero, one or multiple values.
1847The type of each return value must be a '<a href="#t_firstclass">first
1848class</a>' type. Note that a function is not <a href="#wellformed">well
1849formed</a> if there exists a '<tt>ret</tt>' instruction inside of the
1850function that returns values that do not match the return type of the
1851function.</p>
1852
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1853<h5>Semantics:</h5>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1854
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1855<p>When the '<tt>ret</tt>' instruction is executed, control flow
1856returns back to the calling function's context. If the caller is a "<a
1857 href="#i_call"><tt>call</tt></a>" instruction, execution continues at
1858the instruction after the call. If the caller was an "<a
1859 href="#i_invoke"><tt>invoke</tt></a>" instruction, execution continues
1860at the beginning of the "normal" destination block. If the instruction
1861returns a value, that value shall set the call or invoke instruction's
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]1862return value. If the instruction returns multiple values then these
Devang Patelec8a5b02008-03-11 05:51:59 +0000[diff] [blame]1863values can only be accessed through a '<a href="#i_getresult"><tt>getresult</tt>
1864</a>' instruction.</p>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1865
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1866<h5>Example:</h5>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1867
1868<pre>
1869 ret i32 5 <i>; Return an integer value of 5</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1870 ret void <i>; Return from a void function</i>
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]1871 ret i32 4, i8 2 <i>; Return two values 4 and 2 </i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1872</pre>
1873</div>
1874<!-- _______________________________________________________________________ -->
1875<div class="doc_subsubsection"> <a name="i_br">'<tt>br</tt>' Instruction</a> </div>
1876<div class="doc_text">
1877<h5>Syntax:</h5>
1878<pre> br i1 &lt;cond&gt;, label &lt;iftrue&gt;, label &lt;iffalse&gt;<br> br label &lt;dest&gt; <i>; Unconditional branch</i>
1879</pre>
1880<h5>Overview:</h5>
1881<p>The '<tt>br</tt>' instruction is used to cause control flow to
1882transfer to a different basic block in the current function. There are
1883two forms of this instruction, corresponding to a conditional branch
1884and an unconditional branch.</p>
1885<h5>Arguments:</h5>
1886<p>The conditional branch form of the '<tt>br</tt>' instruction takes a
1887single '<tt>i1</tt>' value and two '<tt>label</tt>' values. The
1888unconditional form of the '<tt>br</tt>' instruction takes a single
1889'<tt>label</tt>' value as a target.</p>
1890<h5>Semantics:</h5>
1891<p>Upon execution of a conditional '<tt>br</tt>' instruction, the '<tt>i1</tt>'
1892argument is evaluated. If the value is <tt>true</tt>, control flows
1893to the '<tt>iftrue</tt>' <tt>label</tt> argument. If "cond" is <tt>false</tt>,
1894control flows to the '<tt>iffalse</tt>' <tt>label</tt> argument.</p>
1895<h5>Example:</h5>
1896<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
1897 href="#i_ret">ret</a> i32 1<br>IfUnequal:<br> <a href="#i_ret">ret</a> i32 0<br></pre>
1898</div>
1899<!-- _______________________________________________________________________ -->
1900<div class="doc_subsubsection">
1901 <a name="i_switch">'<tt>switch</tt>' Instruction</a>
1902</div>
1903
1904<div class="doc_text">
1905<h5>Syntax:</h5>
1906
1907<pre>
1908 switch &lt;intty&gt; &lt;value&gt;, label &lt;defaultdest&gt; [ &lt;intty&gt; &lt;val&gt;, label &lt;dest&gt; ... ]
1909</pre>
1910
1911<h5>Overview:</h5>
1912
1913<p>The '<tt>switch</tt>' instruction is used to transfer control flow to one of
1914several different places. It is a generalization of the '<tt>br</tt>'
1915instruction, allowing a branch to occur to one of many possible
1916destinations.</p>
1917
1918
1919<h5>Arguments:</h5>
1920
1921<p>The '<tt>switch</tt>' instruction uses three parameters: an integer
1922comparison value '<tt>value</tt>', a default '<tt>label</tt>' destination, and
1923an array of pairs of comparison value constants and '<tt>label</tt>'s. The
1924table is not allowed to contain duplicate constant entries.</p>
1925
1926<h5>Semantics:</h5>
1927
1928<p>The <tt>switch</tt> instruction specifies a table of values and
1929destinations. When the '<tt>switch</tt>' instruction is executed, this
1930table is searched for the given value. If the value is found, control flow is
1931transfered to the corresponding destination; otherwise, control flow is
1932transfered to the default destination.</p>
1933
1934<h5>Implementation:</h5>
1935
1936<p>Depending on properties of the target machine and the particular
1937<tt>switch</tt> instruction, this instruction may be code generated in different
1938ways. For example, it could be generated as a series of chained conditional
1939branches or with a lookup table.</p>
1940
1941<h5>Example:</h5>
1942
1943<pre>
1944 <i>; Emulate a conditional br instruction</i>
1945 %Val = <a href="#i_zext">zext</a> i1 %value to i32
1946 switch i32 %Val, label %truedest [i32 0, label %falsedest ]
1947
1948 <i>; Emulate an unconditional br instruction</i>
1949 switch i32 0, label %dest [ ]
1950
1951 <i>; Implement a jump table:</i>
1952 switch i32 %val, label %otherwise [ i32 0, label %onzero
1953 i32 1, label %onone
1954 i32 2, label %ontwo ]
1955</pre>
1956</div>
1957
1958<!-- _______________________________________________________________________ -->
1959<div class="doc_subsubsection">
1960 <a name="i_invoke">'<tt>invoke</tt>' Instruction</a>
1961</div>
1962
1963<div class="doc_text">
1964
1965<h5>Syntax:</h5>
1966
1967<pre>
Nick Lewyckya1c11a12008-03-16 07:18:12 +0000[diff] [blame]1968 &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]1969 to label &lt;normal label&gt; unwind label &lt;exception label&gt;
1970</pre>
1971
1972<h5>Overview:</h5>
1973
1974<p>The '<tt>invoke</tt>' instruction causes control to transfer to a specified
1975function, with the possibility of control flow transfer to either the
1976'<tt>normal</tt>' label or the
1977'<tt>exception</tt>' label. If the callee function returns with the
1978"<tt><a href="#i_ret">ret</a></tt>" instruction, control flow will return to the
1979"normal" label. If the callee (or any indirect callees) returns with the "<a
1980href="#i_unwind"><tt>unwind</tt></a>" instruction, control is interrupted and
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]1981continued at the dynamically nearest "exception" label. If the callee function
Devang Patelec8a5b02008-03-11 05:51:59 +0000[diff] [blame]1982returns multiple values then individual return values are only accessible through
1983a '<tt><a href="#i_getresult">getresult</a></tt>' instruction.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1984
1985<h5>Arguments:</h5>
1986
1987<p>This instruction requires several arguments:</p>
1988
1989<ol>
1990 <li>
1991 The optional "cconv" marker indicates which <a href="#callingconv">calling
1992 convention</a> the call should use. If none is specified, the call defaults
1993 to using C calling conventions.
1994 </li>
1995 <li>'<tt>ptr to function ty</tt>': shall be the signature of the pointer to
1996 function value being invoked. In most cases, this is a direct function
1997 invocation, but indirect <tt>invoke</tt>s are just as possible, branching off
1998 an arbitrary pointer to function value.
1999 </li>
2000
2001 <li>'<tt>function ptr val</tt>': An LLVM value containing a pointer to a
2002 function to be invoked. </li>
2003
2004 <li>'<tt>function args</tt>': argument list whose types match the function
2005 signature argument types. If the function signature indicates the function
2006 accepts a variable number of arguments, the extra arguments can be
2007 specified. </li>
2008
2009 <li>'<tt>normal label</tt>': the label reached when the called function
2010 executes a '<tt><a href="#i_ret">ret</a></tt>' instruction. </li>
2011
2012 <li>'<tt>exception label</tt>': the label reached when a callee returns with
2013 the <a href="#i_unwind"><tt>unwind</tt></a> instruction. </li>
2014
2015</ol>
2016
2017<h5>Semantics:</h5>
2018
2019<p>This instruction is designed to operate as a standard '<tt><a
2020href="#i_call">call</a></tt>' instruction in most regards. The primary
2021difference is that it establishes an association with a label, which is used by
2022the runtime library to unwind the stack.</p>
2023
2024<p>This instruction is used in languages with destructors to ensure that proper
2025cleanup is performed in the case of either a <tt>longjmp</tt> or a thrown
2026exception. Additionally, this is important for implementation of
2027'<tt>catch</tt>' clauses in high-level languages that support them.</p>
2028
2029<h5>Example:</h5>
2030<pre>
Nick Lewyckya1c11a12008-03-16 07:18:12 +0000[diff] [blame]2031 %retval = invoke i32 @Test(i32 15) to label %Continue
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2032 unwind label %TestCleanup <i>; {i32}:retval set</i>
Nick Lewyckya1c11a12008-03-16 07:18:12 +0000[diff] [blame]2033 %retval = invoke <a href="#callingconv">coldcc</a> i32 %Testfnptr(i32 15) to label %Continue
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2034 unwind label %TestCleanup <i>; {i32}:retval set</i>
2035</pre>
2036</div>
2037
2038
2039<!-- _______________________________________________________________________ -->
2040
2041<div class="doc_subsubsection"> <a name="i_unwind">'<tt>unwind</tt>'
2042Instruction</a> </div>
2043
2044<div class="doc_text">
2045
2046<h5>Syntax:</h5>
2047<pre>
2048 unwind
2049</pre>
2050
2051<h5>Overview:</h5>
2052
2053<p>The '<tt>unwind</tt>' instruction unwinds the stack, continuing control flow
2054at the first callee in the dynamic call stack which used an <a
2055href="#i_invoke"><tt>invoke</tt></a> instruction to perform the call. This is
2056primarily used to implement exception handling.</p>
2057
2058<h5>Semantics:</h5>
2059
Chris Lattner8b094fc2008-04-19 21:01:16 +0000[diff] [blame]2060<p>The '<tt>unwind</tt>' instruction causes execution of the current function to
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2061immediately halt. The dynamic call stack is then searched for the first <a
2062href="#i_invoke"><tt>invoke</tt></a> instruction on the call stack. Once found,
2063execution continues at the "exceptional" destination block specified by the
2064<tt>invoke</tt> instruction. If there is no <tt>invoke</tt> instruction in the
2065dynamic call chain, undefined behavior results.</p>
2066</div>
2067
2068<!-- _______________________________________________________________________ -->
2069
2070<div class="doc_subsubsection"> <a name="i_unreachable">'<tt>unreachable</tt>'
2071Instruction</a> </div>
2072
2073<div class="doc_text">
2074
2075<h5>Syntax:</h5>
2076<pre>
2077 unreachable
2078</pre>
2079
2080<h5>Overview:</h5>
2081
2082<p>The '<tt>unreachable</tt>' instruction has no defined semantics. This
2083instruction is used to inform the optimizer that a particular portion of the
2084code is not reachable. This can be used to indicate that the code after a
2085no-return function cannot be reached, and other facts.</p>
2086
2087<h5>Semantics:</h5>
2088
2089<p>The '<tt>unreachable</tt>' instruction has no defined semantics.</p>
2090</div>
2091
2092
2093
2094<!-- ======================================================================= -->
2095<div class="doc_subsection"> <a name="binaryops">Binary Operations</a> </div>
2096<div class="doc_text">
2097<p>Binary operators are used to do most of the computation in a
Chris Lattnerab596d92008-04-01 18:47:32 +0000[diff] [blame]2098program. They require two operands of the same type, execute an operation on them, and
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2099produce a single value. The operands might represent
2100multiple 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]2101The result value has the same type as its operands.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2102<p>There are several different binary operators:</p>
2103</div>
2104<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2105<div class="doc_subsubsection">
2106 <a name="i_add">'<tt>add</tt>' Instruction</a>
2107</div>
2108
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2109<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2110
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2111<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2112
2113<pre>
2114 &lt;result&gt; = add &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2115</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2116
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2117<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2118
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2119<p>The '<tt>add</tt>' instruction returns the sum of its two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2120
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2121<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2122
2123<p>The two arguments to the '<tt>add</tt>' instruction must be <a
2124 href="#t_integer">integer</a>, <a href="#t_floating">floating point</a>, or
2125 <a href="#t_vector">vector</a> values. Both arguments must have identical
2126 types.</p>
2127
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2128<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2129
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2130<p>The value produced is the integer or floating point sum of the two
2131operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2132
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2133<p>If an integer sum has unsigned overflow, the result returned is the
2134mathematical result modulo 2<sup>n</sup>, where n is the bit width of
2135the result.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2136
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2137<p>Because LLVM integers use a two's complement representation, this
2138instruction is appropriate for both signed and unsigned integers.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2139
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2140<h5>Example:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2141
2142<pre>
2143 &lt;result&gt; = add i32 4, %var <i>; yields {i32}:result = 4 + %var</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2144</pre>
2145</div>
2146<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2147<div class="doc_subsubsection">
2148 <a name="i_sub">'<tt>sub</tt>' Instruction</a>
2149</div>
2150
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2151<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2152
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2153<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2154
2155<pre>
2156 &lt;result&gt; = sub &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2157</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2158
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2159<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2160
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2161<p>The '<tt>sub</tt>' instruction returns the difference of its two
2162operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2163
2164<p>Note that the '<tt>sub</tt>' instruction is used to represent the
2165'<tt>neg</tt>' instruction present in most other intermediate
2166representations.</p>
2167
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2168<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2169
2170<p>The two arguments to the '<tt>sub</tt>' instruction must be <a
2171 href="#t_integer">integer</a>, <a href="#t_floating">floating point</a>,
2172 or <a href="#t_vector">vector</a> values. Both arguments must have identical
2173 types.</p>
2174
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2175<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2176
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2177<p>The value produced is the integer or floating point difference of
2178the two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2179
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2180<p>If an integer difference has unsigned overflow, the result returned is the
2181mathematical result modulo 2<sup>n</sup>, where n is the bit width of
2182the result.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2183
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2184<p>Because LLVM integers use a two's complement representation, this
2185instruction is appropriate for both signed and unsigned integers.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2186
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2187<h5>Example:</h5>
2188<pre>
2189 &lt;result&gt; = sub i32 4, %var <i>; yields {i32}:result = 4 - %var</i>
2190 &lt;result&gt; = sub i32 0, %val <i>; yields {i32}:result = -%var</i>
2191</pre>
2192</div>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2193
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2194<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2195<div class="doc_subsubsection">
2196 <a name="i_mul">'<tt>mul</tt>' Instruction</a>
2197</div>
2198
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2199<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2200
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2201<h5>Syntax:</h5>
2202<pre> &lt;result&gt; = mul &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
2203</pre>
2204<h5>Overview:</h5>
2205<p>The '<tt>mul</tt>' instruction returns the product of its two
2206operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2207
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2208<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2209
2210<p>The two arguments to the '<tt>mul</tt>' instruction must be <a
2211href="#t_integer">integer</a>, <a href="#t_floating">floating point</a>,
2212or <a href="#t_vector">vector</a> values. Both arguments must have identical
2213types.</p>
2214
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2215<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2216
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2217<p>The value produced is the integer or floating point product of the
2218two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2219
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2220<p>If the result of an integer multiplication has unsigned overflow,
2221the result returned is the mathematical result modulo
22222<sup>n</sup>, where n is the bit width of the result.</p>
2223<p>Because LLVM integers use a two's complement representation, and the
2224result is the same width as the operands, this instruction returns the
2225correct result for both signed and unsigned integers. If a full product
2226(e.g. <tt>i32</tt>x<tt>i32</tt>-><tt>i64</tt>) is needed, the operands
2227should be sign-extended or zero-extended as appropriate to the
2228width of the full product.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2229<h5>Example:</h5>
2230<pre> &lt;result&gt; = mul i32 4, %var <i>; yields {i32}:result = 4 * %var</i>
2231</pre>
2232</div>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2233
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2234<!-- _______________________________________________________________________ -->
2235<div class="doc_subsubsection"> <a name="i_udiv">'<tt>udiv</tt>' Instruction
2236</a></div>
2237<div class="doc_text">
2238<h5>Syntax:</h5>
2239<pre> &lt;result&gt; = udiv &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
2240</pre>
2241<h5>Overview:</h5>
2242<p>The '<tt>udiv</tt>' instruction returns the quotient of its two
2243operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2244
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2245<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2246
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2247<p>The two arguments to the '<tt>udiv</tt>' instruction must be
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2248<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2249values. Both arguments must have identical types.</p>
2250
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2251<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2252
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2253<p>The value produced is the unsigned integer quotient of the two operands.</p>
2254<p>Note that unsigned integer division and signed integer division are distinct
2255operations; for signed integer division, use '<tt>sdiv</tt>'.</p>
2256<p>Division by zero leads to undefined behavior.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2257<h5>Example:</h5>
2258<pre> &lt;result&gt; = udiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
2259</pre>
2260</div>
2261<!-- _______________________________________________________________________ -->
2262<div class="doc_subsubsection"> <a name="i_sdiv">'<tt>sdiv</tt>' Instruction
2263</a> </div>
2264<div class="doc_text">
2265<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2266<pre>
2267 &lt;result&gt; = sdiv &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2268</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2269
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2270<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2271
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2272<p>The '<tt>sdiv</tt>' instruction returns the quotient of its two
2273operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2274
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2275<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2276
2277<p>The two arguments to the '<tt>sdiv</tt>' instruction must be
2278<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2279values. Both arguments must have identical types.</p>
2280
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2281<h5>Semantics:</h5>
Chris Lattnerdc15b1d2008-04-01 18:45:27 +0000[diff] [blame]2282<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]2283<p>Note that signed integer division and unsigned integer division are distinct
2284operations; for unsigned integer division, use '<tt>udiv</tt>'.</p>
2285<p>Division by zero leads to undefined behavior. Overflow also leads to
2286undefined behavior; this is a rare case, but can occur, for example,
2287by doing a 32-bit division of -2147483648 by -1.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2288<h5>Example:</h5>
2289<pre> &lt;result&gt; = sdiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
2290</pre>
2291</div>
2292<!-- _______________________________________________________________________ -->
2293<div class="doc_subsubsection"> <a name="i_fdiv">'<tt>fdiv</tt>'
2294Instruction</a> </div>
2295<div class="doc_text">
2296<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2297<pre>
2298 &lt;result&gt; = fdiv &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2299</pre>
2300<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2301
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2302<p>The '<tt>fdiv</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>fdiv</tt>' instruction must be
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2308<a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a>
2309of floating point values. 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
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2313<p>The value produced is the floating point quotient of the two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2314
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2315<h5>Example:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2316
2317<pre>
2318 &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]2319</pre>
2320</div>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2321
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2322<!-- _______________________________________________________________________ -->
2323<div class="doc_subsubsection"> <a name="i_urem">'<tt>urem</tt>' Instruction</a>
2324</div>
2325<div class="doc_text">
2326<h5>Syntax:</h5>
2327<pre> &lt;result&gt; = urem &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
2328</pre>
2329<h5>Overview:</h5>
2330<p>The '<tt>urem</tt>' instruction returns the remainder from the
2331unsigned division of its two arguments.</p>
2332<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2333<p>The two arguments to the '<tt>urem</tt>' instruction must be
2334<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2335values. Both arguments must have identical types.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2336<h5>Semantics:</h5>
2337<p>This instruction returns the unsigned integer <i>remainder</i> of a division.
Chris Lattnerdc15b1d2008-04-01 18:45:27 +0000[diff] [blame]2338This instruction always performs an unsigned division to get the remainder.</p>
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2339<p>Note that unsigned integer remainder and signed integer remainder are
2340distinct operations; for signed integer remainder, use '<tt>srem</tt>'.</p>
2341<p>Taking the remainder of a division by zero leads to undefined behavior.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2342<h5>Example:</h5>
2343<pre> &lt;result&gt; = urem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
2344</pre>
2345
2346</div>
2347<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2348<div class="doc_subsubsection">
2349 <a name="i_srem">'<tt>srem</tt>' Instruction</a>
2350</div>
2351
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2352<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2353
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2354<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2355
2356<pre>
2357 &lt;result&gt; = srem &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2358</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2359
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]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>srem</tt>' instruction returns the remainder from the
Dan Gohman3e3fd8c2007-11-05 23:35:22 +0000[diff] [blame]2363signed division of its two operands. This instruction can also take
2364<a href="#t_vector">vector</a> versions of the values in which case
2365the elements must be integers.</p>
Chris Lattner08497ce2008-01-04 04:33:49 +0000[diff] [blame]2366
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2367<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2368
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2369<p>The two arguments to the '<tt>srem</tt>' instruction must be
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2370<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2371values. Both arguments must have identical types.</p>
2372
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2373<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2374
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2375<p>This instruction returns the <i>remainder</i> of a division (where the result
2376has the same sign as the dividend, <tt>var1</tt>), not the <i>modulo</i>
2377operator (where the result has the same sign as the divisor, <tt>var2</tt>) of
2378a value. For more information about the difference, see <a
2379 href="http://mathforum.org/dr.math/problems/anne.4.28.99.html">The
2380Math Forum</a>. For a table of how this is implemented in various languages,
2381please see <a href="http://en.wikipedia.org/wiki/Modulo_operation">
2382Wikipedia: modulo operation</a>.</p>
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2383<p>Note that signed integer remainder and unsigned integer remainder are
2384distinct operations; for unsigned integer remainder, use '<tt>urem</tt>'.</p>
2385<p>Taking the remainder of a division by zero leads to undefined behavior.
2386Overflow also leads to undefined behavior; this is a rare case, but can occur,
2387for example, by taking the remainder of a 32-bit division of -2147483648 by -1.
2388(The remainder doesn't actually overflow, but this rule lets srem be
2389implemented using instructions that return both the result of the division
2390and the remainder.)</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2391<h5>Example:</h5>
2392<pre> &lt;result&gt; = srem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
2393</pre>
2394
2395</div>
2396<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2397<div class="doc_subsubsection">
2398 <a name="i_frem">'<tt>frem</tt>' Instruction</a> </div>
2399
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2400<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2401
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2402<h5>Syntax:</h5>
2403<pre> &lt;result&gt; = frem &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
2404</pre>
2405<h5>Overview:</h5>
2406<p>The '<tt>frem</tt>' instruction returns the remainder from the
2407division of its two operands.</p>
2408<h5>Arguments:</h5>
2409<p>The two arguments to the '<tt>frem</tt>' instruction must be
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2410<a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a>
2411of floating point values. Both arguments must have identical types.</p>
2412
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2413<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2414
Chris Lattnerdc15b1d2008-04-01 18:45:27 +0000[diff] [blame]2415<p>This instruction returns the <i>remainder</i> of a division.
2416The remainder has the same sign as the dividend.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2417
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2418<h5>Example:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2419
2420<pre>
2421 &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]2422</pre>
2423</div>
2424
2425<!-- ======================================================================= -->
2426<div class="doc_subsection"> <a name="bitwiseops">Bitwise Binary
2427Operations</a> </div>
2428<div class="doc_text">
2429<p>Bitwise binary operators are used to do various forms of
2430bit-twiddling in a program. They are generally very efficient
2431instructions and can commonly be strength reduced from other
Chris Lattnerdc15b1d2008-04-01 18:45:27 +0000[diff] [blame]2432instructions. They require two operands of the same type, execute an operation on them,
2433and 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]2434</div>
2435
2436<!-- _______________________________________________________________________ -->
2437<div class="doc_subsubsection"> <a name="i_shl">'<tt>shl</tt>'
2438Instruction</a> </div>
2439<div class="doc_text">
2440<h5>Syntax:</h5>
2441<pre> &lt;result&gt; = shl &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
2442</pre>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2443
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2444<h5>Overview:</h5>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2445
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2446<p>The '<tt>shl</tt>' instruction returns the first operand shifted to
2447the left a specified number of bits.</p>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2448
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2449<h5>Arguments:</h5>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2450
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2451<p>Both arguments to the '<tt>shl</tt>' instruction must be the same <a
Chris Lattner8b094fc2008-04-19 21:01:16 +0000[diff] [blame]2452 href="#t_integer">integer</a> type. '<tt>var2</tt>' is treated as an
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2453unsigned value. This instruction does not support
2454<a href="#t_vector">vector</a> operands.</p>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2455
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2456<h5>Semantics:</h5>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2457
Chris Lattnerdc15b1d2008-04-01 18:45:27 +0000[diff] [blame]2458<p>The value produced is <tt>var1</tt> * 2<sup><tt>var2</tt></sup> mod 2<sup>n</sup>,
2459where n is the width of the result. If <tt>var2</tt> is (statically or dynamically) negative or
2460equal to or larger than the number of bits in <tt>var1</tt>, the result is undefined.</p>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2461
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2462<h5>Example:</h5><pre>
2463 &lt;result&gt; = shl i32 4, %var <i>; yields {i32}: 4 &lt;&lt; %var</i>
2464 &lt;result&gt; = shl i32 4, 2 <i>; yields {i32}: 16</i>
2465 &lt;result&gt; = shl i32 1, 10 <i>; yields {i32}: 1024</i>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2466 &lt;result&gt; = shl i32 1, 32 <i>; undefined</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2467</pre>
2468</div>
2469<!-- _______________________________________________________________________ -->
2470<div class="doc_subsubsection"> <a name="i_lshr">'<tt>lshr</tt>'
2471Instruction</a> </div>
2472<div class="doc_text">
2473<h5>Syntax:</h5>
2474<pre> &lt;result&gt; = lshr &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
2475</pre>
2476
2477<h5>Overview:</h5>
2478<p>The '<tt>lshr</tt>' instruction (logical shift right) returns the first
2479operand shifted to the right a specified number of bits with zero fill.</p>
2480
2481<h5>Arguments:</h5>
2482<p>Both arguments to the '<tt>lshr</tt>' instruction must be the same
Chris Lattner8b094fc2008-04-19 21:01:16 +0000[diff] [blame]2483<a href="#t_integer">integer</a> type. '<tt>var2</tt>' is treated as an
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2484unsigned value. This instruction does not support
2485<a href="#t_vector">vector</a> operands.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2486
2487<h5>Semantics:</h5>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2488
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2489<p>This instruction always performs a logical shift right operation. The most
2490significant bits of the result will be filled with zero bits after the
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2491shift. If <tt>var2</tt> is (statically or dynamically) equal to or larger than
2492the number of bits in <tt>var1</tt>, the result is undefined.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2493
2494<h5>Example:</h5>
2495<pre>
2496 &lt;result&gt; = lshr i32 4, 1 <i>; yields {i32}:result = 2</i>
2497 &lt;result&gt; = lshr i32 4, 2 <i>; yields {i32}:result = 1</i>
2498 &lt;result&gt; = lshr i8 4, 3 <i>; yields {i8}:result = 0</i>
2499 &lt;result&gt; = lshr i8 -2, 1 <i>; yields {i8}:result = 0x7FFFFFFF </i>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2500 &lt;result&gt; = lshr i32 1, 32 <i>; undefined</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2501</pre>
2502</div>
2503
2504<!-- _______________________________________________________________________ -->
2505<div class="doc_subsubsection"> <a name="i_ashr">'<tt>ashr</tt>'
2506Instruction</a> </div>
2507<div class="doc_text">
2508
2509<h5>Syntax:</h5>
2510<pre> &lt;result&gt; = ashr &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
2511</pre>
2512
2513<h5>Overview:</h5>
2514<p>The '<tt>ashr</tt>' instruction (arithmetic shift right) returns the first
2515operand shifted to the right a specified number of bits with sign extension.</p>
2516
2517<h5>Arguments:</h5>
2518<p>Both arguments to the '<tt>ashr</tt>' instruction must be the same
Chris Lattner8b094fc2008-04-19 21:01:16 +0000[diff] [blame]2519<a href="#t_integer">integer</a> type. '<tt>var2</tt>' is treated as an
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2520unsigned value. This instruction does not support
2521<a href="#t_vector">vector</a> operands.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2522
2523<h5>Semantics:</h5>
2524<p>This instruction always performs an arithmetic shift right operation,
2525The most significant bits of the result will be filled with the sign bit
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2526of <tt>var1</tt>. If <tt>var2</tt> is (statically or dynamically) equal to or
2527larger than the number of bits in <tt>var1</tt>, the result is undefined.
2528</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2529
2530<h5>Example:</h5>
2531<pre>
2532 &lt;result&gt; = ashr i32 4, 1 <i>; yields {i32}:result = 2</i>
2533 &lt;result&gt; = ashr i32 4, 2 <i>; yields {i32}:result = 1</i>
2534 &lt;result&gt; = ashr i8 4, 3 <i>; yields {i8}:result = 0</i>
2535 &lt;result&gt; = ashr i8 -2, 1 <i>; yields {i8}:result = -1</i>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2536 &lt;result&gt; = ashr i32 1, 32 <i>; undefined</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2537</pre>
2538</div>
2539
2540<!-- _______________________________________________________________________ -->
2541<div class="doc_subsubsection"> <a name="i_and">'<tt>and</tt>'
2542Instruction</a> </div>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2543
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2544<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2545
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2546<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2547
2548<pre>
2549 &lt;result&gt; = and &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2550</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2551
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2552<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2553
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2554<p>The '<tt>and</tt>' instruction returns the bitwise logical and of
2555its two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2556
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2557<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2558
2559<p>The two arguments to the '<tt>and</tt>' instruction must be
2560<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2561values. Both arguments must have identical types.</p>
2562
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2563<h5>Semantics:</h5>
2564<p>The truth table used for the '<tt>and</tt>' instruction is:</p>
2565<p> </p>
2566<div style="align: center">
2567<table border="1" cellspacing="0" cellpadding="4">
2568 <tbody>
2569 <tr>
2570 <td>In0</td>
2571 <td>In1</td>
2572 <td>Out</td>
2573 </tr>
2574 <tr>
2575 <td>0</td>
2576 <td>0</td>
2577 <td>0</td>
2578 </tr>
2579 <tr>
2580 <td>0</td>
2581 <td>1</td>
2582 <td>0</td>
2583 </tr>
2584 <tr>
2585 <td>1</td>
2586 <td>0</td>
2587 <td>0</td>
2588 </tr>
2589 <tr>
2590 <td>1</td>
2591 <td>1</td>
2592 <td>1</td>
2593 </tr>
2594 </tbody>
2595</table>
2596</div>
2597<h5>Example:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2598<pre>
2599 &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]2600 &lt;result&gt; = and i32 15, 40 <i>; yields {i32}:result = 8</i>
2601 &lt;result&gt; = and i32 4, 8 <i>; yields {i32}:result = 0</i>
2602</pre>
2603</div>
2604<!-- _______________________________________________________________________ -->
2605<div class="doc_subsubsection"> <a name="i_or">'<tt>or</tt>' Instruction</a> </div>
2606<div class="doc_text">
2607<h5>Syntax:</h5>
2608<pre> &lt;result&gt; = or &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
2609</pre>
2610<h5>Overview:</h5>
2611<p>The '<tt>or</tt>' instruction returns the bitwise logical inclusive
2612or of its two operands.</p>
2613<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2614
2615<p>The two arguments to the '<tt>or</tt>' instruction must be
2616<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2617values. Both arguments must have identical types.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2618<h5>Semantics:</h5>
2619<p>The truth table used for the '<tt>or</tt>' instruction is:</p>
2620<p> </p>
2621<div style="align: center">
2622<table border="1" cellspacing="0" cellpadding="4">
2623 <tbody>
2624 <tr>
2625 <td>In0</td>
2626 <td>In1</td>
2627 <td>Out</td>
2628 </tr>
2629 <tr>
2630 <td>0</td>
2631 <td>0</td>
2632 <td>0</td>
2633 </tr>
2634 <tr>
2635 <td>0</td>
2636 <td>1</td>
2637 <td>1</td>
2638 </tr>
2639 <tr>
2640 <td>1</td>
2641 <td>0</td>
2642 <td>1</td>
2643 </tr>
2644 <tr>
2645 <td>1</td>
2646 <td>1</td>
2647 <td>1</td>
2648 </tr>
2649 </tbody>
2650</table>
2651</div>
2652<h5>Example:</h5>
2653<pre> &lt;result&gt; = or i32 4, %var <i>; yields {i32}:result = 4 | %var</i>
2654 &lt;result&gt; = or i32 15, 40 <i>; yields {i32}:result = 47</i>
2655 &lt;result&gt; = or i32 4, 8 <i>; yields {i32}:result = 12</i>
2656</pre>
2657</div>
2658<!-- _______________________________________________________________________ -->
2659<div class="doc_subsubsection"> <a name="i_xor">'<tt>xor</tt>'
2660Instruction</a> </div>
2661<div class="doc_text">
2662<h5>Syntax:</h5>
2663<pre> &lt;result&gt; = xor &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
2664</pre>
2665<h5>Overview:</h5>
2666<p>The '<tt>xor</tt>' instruction returns the bitwise logical exclusive
2667or of its two operands. The <tt>xor</tt> is used to implement the
2668"one's complement" operation, which is the "~" operator in C.</p>
2669<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2670<p>The two arguments to the '<tt>xor</tt>' instruction must be
2671<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2672values. Both arguments must have identical types.</p>
2673
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2674<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2675
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2676<p>The truth table used for the '<tt>xor</tt>' instruction is:</p>
2677<p> </p>
2678<div style="align: center">
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>0</td>
2705 </tr>
2706 </tbody>
2707</table>
2708</div>
2709<p> </p>
2710<h5>Example:</h5>
2711<pre> &lt;result&gt; = xor i32 4, %var <i>; yields {i32}:result = 4 ^ %var</i>
2712 &lt;result&gt; = xor i32 15, 40 <i>; yields {i32}:result = 39</i>
2713 &lt;result&gt; = xor i32 4, 8 <i>; yields {i32}:result = 12</i>
2714 &lt;result&gt; = xor i32 %V, -1 <i>; yields {i32}:result = ~%V</i>
2715</pre>
2716</div>
2717
2718<!-- ======================================================================= -->
2719<div class="doc_subsection">
2720 <a name="vectorops">Vector Operations</a>
2721</div>
2722
2723<div class="doc_text">
2724
2725<p>LLVM supports several instructions to represent vector operations in a
2726target-independent manner. These instructions cover the element-access and
2727vector-specific operations needed to process vectors effectively. While LLVM
2728does directly support these vector operations, many sophisticated algorithms
2729will want to use target-specific intrinsics to take full advantage of a specific
2730target.</p>
2731
2732</div>
2733
2734<!-- _______________________________________________________________________ -->
2735<div class="doc_subsubsection">
2736 <a name="i_extractelement">'<tt>extractelement</tt>' Instruction</a>
2737</div>
2738
2739<div class="doc_text">
2740
2741<h5>Syntax:</h5>
2742
2743<pre>
2744 &lt;result&gt; = extractelement &lt;n x &lt;ty&gt;&gt; &lt;val&gt;, i32 &lt;idx&gt; <i>; yields &lt;ty&gt;</i>
2745</pre>
2746
2747<h5>Overview:</h5>
2748
2749<p>
2750The '<tt>extractelement</tt>' instruction extracts a single scalar
2751element from a vector at a specified index.
2752</p>
2753
2754
2755<h5>Arguments:</h5>
2756
2757<p>
2758The first operand of an '<tt>extractelement</tt>' instruction is a
2759value of <a href="#t_vector">vector</a> type. The second operand is
2760an index indicating the position from which to extract the element.
2761The index may be a variable.</p>
2762
2763<h5>Semantics:</h5>
2764
2765<p>
2766The result is a scalar of the same type as the element type of
2767<tt>val</tt>. Its value is the value at position <tt>idx</tt> of
2768<tt>val</tt>. If <tt>idx</tt> exceeds the length of <tt>val</tt>, the
2769results are undefined.
2770</p>
2771
2772<h5>Example:</h5>
2773
2774<pre>
2775 %result = extractelement &lt;4 x i32&gt; %vec, i32 0 <i>; yields i32</i>
2776</pre>
2777</div>
2778
2779
2780<!-- _______________________________________________________________________ -->
2781<div class="doc_subsubsection">
2782 <a name="i_insertelement">'<tt>insertelement</tt>' Instruction</a>
2783</div>
2784
2785<div class="doc_text">
2786
2787<h5>Syntax:</h5>
2788
2789<pre>
Dan Gohmanbcc3c502008-05-12 23:38:42 +0000[diff] [blame]2790 &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]2791</pre>
2792
2793<h5>Overview:</h5>
2794
2795<p>
2796The '<tt>insertelement</tt>' instruction inserts a scalar
2797element into a vector at a specified index.
2798</p>
2799
2800
2801<h5>Arguments:</h5>
2802
2803<p>
2804The first operand of an '<tt>insertelement</tt>' instruction is a
2805value of <a href="#t_vector">vector</a> type. The second operand is a
2806scalar value whose type must equal the element type of the first
2807operand. The third operand is an index indicating the position at
2808which to insert the value. The index may be a variable.</p>
2809
2810<h5>Semantics:</h5>
2811
2812<p>
2813The result is a vector of the same type as <tt>val</tt>. Its
2814element values are those of <tt>val</tt> except at position
2815<tt>idx</tt>, where it gets the value <tt>elt</tt>. If <tt>idx</tt>
2816exceeds the length of <tt>val</tt>, the results are undefined.
2817</p>
2818
2819<h5>Example:</h5>
2820
2821<pre>
2822 %result = insertelement &lt;4 x i32&gt; %vec, i32 1, i32 0 <i>; yields &lt;4 x i32&gt;</i>
2823</pre>
2824</div>
2825
2826<!-- _______________________________________________________________________ -->
2827<div class="doc_subsubsection">
2828 <a name="i_shufflevector">'<tt>shufflevector</tt>' Instruction</a>
2829</div>
2830
2831<div class="doc_text">
2832
2833<h5>Syntax:</h5>
2834
2835<pre>
2836 &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>
2837</pre>
2838
2839<h5>Overview:</h5>
2840
2841<p>
2842The '<tt>shufflevector</tt>' instruction constructs a permutation of elements
2843from two input vectors, returning a vector of the same type.
2844</p>
2845
2846<h5>Arguments:</h5>
2847
2848<p>
2849The first two operands of a '<tt>shufflevector</tt>' instruction are vectors
2850with types that match each other and types that match the result of the
2851instruction. The third argument is a shuffle mask, which has the same number
2852of elements as the other vector type, but whose element type is always 'i32'.
2853</p>
2854
2855<p>
2856The shuffle mask operand is required to be a constant vector with either
2857constant integer or undef values.
2858</p>
2859
2860<h5>Semantics:</h5>
2861
2862<p>
2863The elements of the two input vectors are numbered from left to right across
2864both of the vectors. The shuffle mask operand specifies, for each element of
2865the result vector, which element of the two input registers the result element
2866gets. The element selector may be undef (meaning "don't care") and the second
2867operand may be undef if performing a shuffle from only one vector.
2868</p>
2869
2870<h5>Example:</h5>
2871
2872<pre>
2873 %result = shufflevector &lt;4 x i32&gt; %v1, &lt;4 x i32&gt; %v2,
2874 &lt;4 x i32&gt; &lt;i32 0, i32 4, i32 1, i32 5&gt; <i>; yields &lt;4 x i32&gt;</i>
2875 %result = shufflevector &lt;4 x i32&gt; %v1, &lt;4 x i32&gt; undef,
2876 &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.
2877</pre>
2878</div>
2879
2880
2881<!-- ======================================================================= -->
2882<div class="doc_subsection">
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]2883 <a name="aggregateops">Aggregate Operations</a>
2884</div>
2885
2886<div class="doc_text">
2887
2888<p>LLVM supports several instructions for working with aggregate values.
2889</p>
2890
2891</div>
2892
2893<!-- _______________________________________________________________________ -->
2894<div class="doc_subsubsection">
2895 <a name="i_extractvalue">'<tt>extractvalue</tt>' Instruction</a>
2896</div>
2897
2898<div class="doc_text">
2899
2900<h5>Syntax:</h5>
2901
2902<pre>
2903 &lt;result&gt; = extractvalue &lt;aggregate type&gt; &lt;val&gt;, &lt;idx&gt;{, &lt;idx&gt;}*
2904</pre>
2905
2906<h5>Overview:</h5>
2907
2908<p>
Dan Gohman6c6dea02008-05-13 18:16:06 +0000[diff] [blame]2909The '<tt>extractvalue</tt>' instruction extracts the value of a struct field
2910or array element from an aggregate value.
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]2911</p>
2912
2913
2914<h5>Arguments:</h5>
2915
2916<p>
2917The first operand of an '<tt>extractvalue</tt>' instruction is a
2918value of <a href="#t_struct">struct</a> or <a href="#t_array">array</a>
Dan Gohman6c6dea02008-05-13 18:16:06 +0000[diff] [blame]2919type. The operands are constant indices to specify which value to extract
Dan Gohmane5febe42008-05-31 00:58:22 +0000[diff] [blame]2920in a similar manner as indices in a
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]2921'<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.
2922</p>
2923
2924<h5>Semantics:</h5>
2925
2926<p>
2927The result is the value at the position in the aggregate specified by
2928the index operands.
2929</p>
2930
2931<h5>Example:</h5>
2932
2933<pre>
Dan Gohmane5febe42008-05-31 00:58:22 +0000[diff] [blame]2934 %result = extractvalue {i32, float} %agg, 0 <i>; yields i32</i>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]2935</pre>
2936</div>
2937
2938
2939<!-- _______________________________________________________________________ -->
2940<div class="doc_subsubsection">
2941 <a name="i_insertvalue">'<tt>insertvalue</tt>' Instruction</a>
2942</div>
2943
2944<div class="doc_text">
2945
2946<h5>Syntax:</h5>
2947
2948<pre>
Dan Gohmane5febe42008-05-31 00:58:22 +0000[diff] [blame]2949 &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]2950</pre>
2951
2952<h5>Overview:</h5>
2953
2954<p>
2955The '<tt>insertvalue</tt>' instruction inserts a value
Dan Gohman6c6dea02008-05-13 18:16:06 +0000[diff] [blame]2956into a struct field or array element in an aggregate.
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]2957</p>
2958
2959
2960<h5>Arguments:</h5>
2961
2962<p>
2963The first operand of an '<tt>insertvalue</tt>' instruction is a
2964value of <a href="#t_struct">struct</a> or <a href="#t_array">array</a> type.
2965The second operand is a first-class value to insert.
Dan Gohman4f29e422008-05-23 21:53:15 +0000[diff] [blame]2966The following operands are constant indices
Dan Gohmane5febe42008-05-31 00:58:22 +0000[diff] [blame]2967indicating the position at which to insert the value in a similar manner as
Dan Gohman6c6dea02008-05-13 18:16:06 +0000[diff] [blame]2968indices in a
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]2969'<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.
2970The value to insert must have the same type as the value identified
Dan Gohman6c6dea02008-05-13 18:16:06 +0000[diff] [blame]2971by the indices.
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]2972
2973<h5>Semantics:</h5>
2974
2975<p>
2976The result is an aggregate of the same type as <tt>val</tt>. Its
2977value is that of <tt>val</tt> except that the value at the position
Dan Gohman6c6dea02008-05-13 18:16:06 +0000[diff] [blame]2978specified by the indices is that of <tt>elt</tt>.
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]2979</p>
2980
2981<h5>Example:</h5>
2982
2983<pre>
Dan Gohmanb1aab4e2008-06-23 15:26:37 +0000[diff] [blame^]2984 %result = insertvalue {i32, float} %agg, i32 1, 0 <i>; yields {i32, float}</i>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]2985</pre>
2986</div>
2987
2988
2989<!-- ======================================================================= -->
2990<div class="doc_subsection">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2991 <a name="memoryops">Memory Access and Addressing Operations</a>
2992</div>
2993
2994<div class="doc_text">
2995
2996<p>A key design point of an SSA-based representation is how it
2997represents memory. In LLVM, no memory locations are in SSA form, which
2998makes things very simple. This section describes how to read, write,
2999allocate, and free memory in LLVM.</p>
3000
3001</div>
3002
3003<!-- _______________________________________________________________________ -->
3004<div class="doc_subsubsection">
3005 <a name="i_malloc">'<tt>malloc</tt>' Instruction</a>
3006</div>
3007
3008<div class="doc_text">
3009
3010<h5>Syntax:</h5>
3011
3012<pre>
3013 &lt;result&gt; = malloc &lt;type&gt;[, i32 &lt;NumElements&gt;][, align &lt;alignment&gt;] <i>; yields {type*}:result</i>
3014</pre>
3015
3016<h5>Overview:</h5>
3017
3018<p>The '<tt>malloc</tt>' instruction allocates memory from the system
Christopher Lambcfe00962007-12-17 01:00:21 +0000[diff] [blame]3019heap and returns a pointer to it. The object is always allocated in the generic
3020address space (address space zero).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3021
3022<h5>Arguments:</h5>
3023
3024<p>The '<tt>malloc</tt>' instruction allocates
3025<tt>sizeof(&lt;type&gt;)*NumElements</tt>
3026bytes of memory from the operating system and returns a pointer of the
3027appropriate type to the program. If "NumElements" is specified, it is the
Gabor Greif5082cf42008-02-09 22:24:34 +0000[diff] [blame]3028number of elements allocated, otherwise "NumElements" is defaulted to be one.
Chris Lattner10368b62008-04-02 00:38:26 +0000[diff] [blame]3029If a constant alignment is specified, the value result of the allocation is guaranteed to
Gabor Greif5082cf42008-02-09 22:24:34 +0000[diff] [blame]3030be aligned to at least that boundary. If not specified, or if zero, the target can
3031choose to align the allocation on any convenient boundary.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3032
3033<p>'<tt>type</tt>' must be a sized type.</p>
3034
3035<h5>Semantics:</h5>
3036
3037<p>Memory is allocated using the system "<tt>malloc</tt>" function, and
Chris Lattner8b094fc2008-04-19 21:01:16 +0000[diff] [blame]3038a pointer is returned. The result of a zero byte allocattion is undefined. The
3039result is null if there is insufficient memory available.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3040
3041<h5>Example:</h5>
3042
3043<pre>
3044 %array = malloc [4 x i8 ] <i>; yields {[%4 x i8]*}:array</i>
3045
3046 %size = <a href="#i_add">add</a> i32 2, 2 <i>; yields {i32}:size = i32 4</i>
3047 %array1 = malloc i8, i32 4 <i>; yields {i8*}:array1</i>
3048 %array2 = malloc [12 x i8], i32 %size <i>; yields {[12 x i8]*}:array2</i>
3049 %array3 = malloc i32, i32 4, align 1024 <i>; yields {i32*}:array3</i>
3050 %array4 = malloc i32, align 1024 <i>; yields {i32*}:array4</i>
3051</pre>
3052</div>
3053
3054<!-- _______________________________________________________________________ -->
3055<div class="doc_subsubsection">
3056 <a name="i_free">'<tt>free</tt>' Instruction</a>
3057</div>
3058
3059<div class="doc_text">
3060
3061<h5>Syntax:</h5>
3062
3063<pre>
3064 free &lt;type&gt; &lt;value&gt; <i>; yields {void}</i>
3065</pre>
3066
3067<h5>Overview:</h5>
3068
3069<p>The '<tt>free</tt>' instruction returns memory back to the unused
3070memory heap to be reallocated in the future.</p>
3071
3072<h5>Arguments:</h5>
3073
3074<p>'<tt>value</tt>' shall be a pointer value that points to a value
3075that was allocated with the '<tt><a href="#i_malloc">malloc</a></tt>'
3076instruction.</p>
3077
3078<h5>Semantics:</h5>
3079
3080<p>Access to the memory pointed to by the pointer is no longer defined
Chris Lattner329d6532008-04-19 22:41:32 +0000[diff] [blame]3081after this instruction executes. If the pointer is null, the operation
3082is a noop.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3083
3084<h5>Example:</h5>
3085
3086<pre>
3087 %array = <a href="#i_malloc">malloc</a> [4 x i8] <i>; yields {[4 x i8]*}:array</i>
3088 free [4 x i8]* %array
3089</pre>
3090</div>
3091
3092<!-- _______________________________________________________________________ -->
3093<div class="doc_subsubsection">
3094 <a name="i_alloca">'<tt>alloca</tt>' Instruction</a>
3095</div>
3096
3097<div class="doc_text">
3098
3099<h5>Syntax:</h5>
3100
3101<pre>
3102 &lt;result&gt; = alloca &lt;type&gt;[, i32 &lt;NumElements&gt;][, align &lt;alignment&gt;] <i>; yields {type*}:result</i>
3103</pre>
3104
3105<h5>Overview:</h5>
3106
3107<p>The '<tt>alloca</tt>' instruction allocates memory on the stack frame of the
3108currently executing function, to be automatically released when this function
Christopher Lambcfe00962007-12-17 01:00:21 +0000[diff] [blame]3109returns to its caller. The object is always allocated in the generic address
3110space (address space zero).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3111
3112<h5>Arguments:</h5>
3113
3114<p>The '<tt>alloca</tt>' instruction allocates <tt>sizeof(&lt;type&gt;)*NumElements</tt>
3115bytes of memory on the runtime stack, returning a pointer of the
Gabor Greif5082cf42008-02-09 22:24:34 +0000[diff] [blame]3116appropriate type to the program. If "NumElements" is specified, it is the
3117number of elements allocated, otherwise "NumElements" is defaulted to be one.
Chris Lattner10368b62008-04-02 00:38:26 +0000[diff] [blame]3118If a constant alignment is specified, the value result of the allocation is guaranteed
Gabor Greif5082cf42008-02-09 22:24:34 +0000[diff] [blame]3119to be aligned to at least that boundary. If not specified, or if zero, the target
3120can choose to align the allocation on any convenient boundary.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3121
3122<p>'<tt>type</tt>' may be any sized type.</p>
3123
3124<h5>Semantics:</h5>
3125
Chris Lattner8b094fc2008-04-19 21:01:16 +0000[diff] [blame]3126<p>Memory is allocated; a pointer is returned. The operation is undefiend if
3127there is insufficient stack space for the allocation. '<tt>alloca</tt>'d
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3128memory is automatically released when the function returns. The '<tt>alloca</tt>'
3129instruction is commonly used to represent automatic variables that must
3130have an address available. When the function returns (either with the <tt><a
3131 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]3132instructions), the memory is reclaimed. Allocating zero bytes
3133is legal, but the result is undefined.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3134
3135<h5>Example:</h5>
3136
3137<pre>
3138 %ptr = alloca i32 <i>; yields {i32*}:ptr</i>
3139 %ptr = alloca i32, i32 4 <i>; yields {i32*}:ptr</i>
3140 %ptr = alloca i32, i32 4, align 1024 <i>; yields {i32*}:ptr</i>
3141 %ptr = alloca i32, align 1024 <i>; yields {i32*}:ptr</i>
3142</pre>
3143</div>
3144
3145<!-- _______________________________________________________________________ -->
3146<div class="doc_subsubsection"> <a name="i_load">'<tt>load</tt>'
3147Instruction</a> </div>
3148<div class="doc_text">
3149<h5>Syntax:</h5>
3150<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>
3151<h5>Overview:</h5>
3152<p>The '<tt>load</tt>' instruction is used to read from memory.</p>
3153<h5>Arguments:</h5>
3154<p>The argument to the '<tt>load</tt>' instruction specifies the memory
3155address from which to load. The pointer must point to a <a
3156 href="#t_firstclass">first class</a> type. If the <tt>load</tt> is
3157marked as <tt>volatile</tt>, then the optimizer is not allowed to modify
3158the number or order of execution of this <tt>load</tt> with other
3159volatile <tt>load</tt> and <tt><a href="#i_store">store</a></tt>
3160instructions. </p>
Chris Lattner21003c82008-01-06 21:04:43 +0000[diff] [blame]3161<p>
Chris Lattner10368b62008-04-02 00:38:26 +0000[diff] [blame]3162The optional constant "align" argument specifies the alignment of the operation
Chris Lattner21003c82008-01-06 21:04:43 +0000[diff] [blame]3163(that is, the alignment of the memory address). A value of 0 or an
3164omitted "align" argument means that the operation has the preferential
3165alignment for the target. It is the responsibility of the code emitter
3166to ensure that the alignment information is correct. Overestimating
3167the alignment results in an undefined behavior. Underestimating the
3168alignment may produce less efficient code. An alignment of 1 is always
3169safe.
3170</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3171<h5>Semantics:</h5>
3172<p>The location of memory pointed to is loaded.</p>
3173<h5>Examples:</h5>
3174<pre> %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
3175 <a
3176 href="#i_store">store</a> i32 3, i32* %ptr <i>; yields {void}</i>
3177 %val = load i32* %ptr <i>; yields {i32}:val = i32 3</i>
3178</pre>
3179</div>
3180<!-- _______________________________________________________________________ -->
3181<div class="doc_subsubsection"> <a name="i_store">'<tt>store</tt>'
3182Instruction</a> </div>
3183<div class="doc_text">
3184<h5>Syntax:</h5>
3185<pre> store &lt;ty&gt; &lt;value&gt;, &lt;ty&gt;* &lt;pointer&gt;[, align &lt;alignment&gt;] <i>; yields {void}</i>
3186 volatile store &lt;ty&gt; &lt;value&gt;, &lt;ty&gt;* &lt;pointer&gt;[, align &lt;alignment&gt;] <i>; yields {void}</i>
3187</pre>
3188<h5>Overview:</h5>
3189<p>The '<tt>store</tt>' instruction is used to write to memory.</p>
3190<h5>Arguments:</h5>
3191<p>There are two arguments to the '<tt>store</tt>' instruction: a value
3192to 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]3193operand must be a pointer to the <a href="#t_firstclass">first class</a> type
3194of the '<tt>&lt;value&gt;</tt>'
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3195operand. If the <tt>store</tt> is marked as <tt>volatile</tt>, then the
3196optimizer is not allowed to modify the number or order of execution of
3197this <tt>store</tt> with other volatile <tt>load</tt> and <tt><a
3198 href="#i_store">store</a></tt> instructions.</p>
Chris Lattner21003c82008-01-06 21:04:43 +0000[diff] [blame]3199<p>
Chris Lattner10368b62008-04-02 00:38:26 +0000[diff] [blame]3200The optional constant "align" argument specifies the alignment of the operation
Chris Lattner21003c82008-01-06 21:04:43 +0000[diff] [blame]3201(that is, the alignment of the memory address). A value of 0 or an
3202omitted "align" argument means that the operation has the preferential
3203alignment for the target. It is the responsibility of the code emitter
3204to ensure that the alignment information is correct. Overestimating
3205the alignment results in an undefined behavior. Underestimating the
3206alignment may produce less efficient code. An alignment of 1 is always
3207safe.
3208</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3209<h5>Semantics:</h5>
3210<p>The contents of memory are updated to contain '<tt>&lt;value&gt;</tt>'
3211at the location specified by the '<tt>&lt;pointer&gt;</tt>' operand.</p>
3212<h5>Example:</h5>
3213<pre> %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
Bill Wendling63ffa142007-10-22 05:10:05 +0000[diff] [blame]3214 store i32 3, i32* %ptr <i>; yields {void}</i>
3215 %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]3216</pre>
3217</div>
3218
3219<!-- _______________________________________________________________________ -->
3220<div class="doc_subsubsection">
3221 <a name="i_getelementptr">'<tt>getelementptr</tt>' Instruction</a>
3222</div>
3223
3224<div class="doc_text">
3225<h5>Syntax:</h5>
3226<pre>
3227 &lt;result&gt; = getelementptr &lt;ty&gt;* &lt;ptrval&gt;{, &lt;ty&gt; &lt;idx&gt;}*
3228</pre>
3229
3230<h5>Overview:</h5>
3231
3232<p>
3233The '<tt>getelementptr</tt>' instruction is used to get the address of a
3234subelement of an aggregate data structure.</p>
3235
3236<h5>Arguments:</h5>
3237
3238<p>This instruction takes a list of integer operands that indicate what
3239elements of the aggregate object to index to. The actual types of the arguments
3240provided depend on the type of the first pointer argument. The
3241'<tt>getelementptr</tt>' instruction is used to index down through the type
3242levels of a structure or to a specific index in an array. When indexing into a
3243structure, only <tt>i32</tt> integer constants are allowed. When indexing
Chris Lattnera7d94ba2008-04-24 05:59:56 +0000[diff] [blame]3244into an array or pointer, only integers of 32 or 64 bits are allowed; 32-bit
3245values will be sign extended to 64-bits if required.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3246
3247<p>For example, let's consider a C code fragment and how it gets
3248compiled to LLVM:</p>
3249
3250<div class="doc_code">
3251<pre>
3252struct RT {
3253 char A;
3254 int B[10][20];
3255 char C;
3256};
3257struct ST {
3258 int X;
3259 double Y;
3260 struct RT Z;
3261};
3262
3263int *foo(struct ST *s) {
3264 return &amp;s[1].Z.B[5][13];
3265}
3266</pre>
3267</div>
3268
3269<p>The LLVM code generated by the GCC frontend is:</p>
3270
3271<div class="doc_code">
3272<pre>
3273%RT = type { i8 , [10 x [20 x i32]], i8 }
3274%ST = type { i32, double, %RT }
3275
3276define i32* %foo(%ST* %s) {
3277entry:
3278 %reg = getelementptr %ST* %s, i32 1, i32 2, i32 1, i32 5, i32 13
3279 ret i32* %reg
3280}
3281</pre>
3282</div>
3283
3284<h5>Semantics:</h5>
3285
3286<p>The index types specified for the '<tt>getelementptr</tt>' instruction depend
3287on the pointer type that is being indexed into. <a href="#t_pointer">Pointer</a>
3288and <a href="#t_array">array</a> types can use a 32-bit or 64-bit
3289<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]3290to 64-bits. <a href="#t_struct">Structure</a> and <a href="#t_pstruct">packed
3291structure</a> types require <tt>i32</tt> <b>constants</b>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3292
3293<p>In the example above, the first index is indexing into the '<tt>%ST*</tt>'
3294type, which is a pointer, yielding a '<tt>%ST</tt>' = '<tt>{ i32, double, %RT
3295}</tt>' type, a structure. The second index indexes into the third element of
3296the structure, yielding a '<tt>%RT</tt>' = '<tt>{ i8 , [10 x [20 x i32]],
3297i8 }</tt>' type, another structure. The third index indexes into the second
3298element of the structure, yielding a '<tt>[10 x [20 x i32]]</tt>' type, an
3299array. The two dimensions of the array are subscripted into, yielding an
3300'<tt>i32</tt>' type. The '<tt>getelementptr</tt>' instruction returns a pointer
3301to this element, thus computing a value of '<tt>i32*</tt>' type.</p>
3302
3303<p>Note that it is perfectly legal to index partially through a
3304structure, returning a pointer to an inner element. Because of this,
3305the LLVM code for the given testcase is equivalent to:</p>
3306
3307<pre>
3308 define i32* %foo(%ST* %s) {
3309 %t1 = getelementptr %ST* %s, i32 1 <i>; yields %ST*:%t1</i>
3310 %t2 = getelementptr %ST* %t1, i32 0, i32 2 <i>; yields %RT*:%t2</i>
3311 %t3 = getelementptr %RT* %t2, i32 0, i32 1 <i>; yields [10 x [20 x i32]]*:%t3</i>
3312 %t4 = getelementptr [10 x [20 x i32]]* %t3, i32 0, i32 5 <i>; yields [20 x i32]*:%t4</i>
3313 %t5 = getelementptr [20 x i32]* %t4, i32 0, i32 13 <i>; yields i32*:%t5</i>
3314 ret i32* %t5
3315 }
3316</pre>
3317
3318<p>Note that it is undefined to access an array out of bounds: array and
3319pointer indexes must always be within the defined bounds of the array type.
Chris Lattnera7d94ba2008-04-24 05:59:56 +0000[diff] [blame]3320The one exception for this rule is zero length arrays. These arrays are
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3321defined to be accessible as variable length arrays, which requires access
3322beyond the zero'th element.</p>
3323
3324<p>The getelementptr instruction is often confusing. For some more insight
3325into how it works, see <a href="GetElementPtr.html">the getelementptr
3326FAQ</a>.</p>
3327
3328<h5>Example:</h5>
3329
3330<pre>
3331 <i>; yields [12 x i8]*:aptr</i>
3332 %aptr = getelementptr {i32, [12 x i8]}* %sptr, i64 0, i32 1
3333</pre>
3334</div>
3335
3336<!-- ======================================================================= -->
3337<div class="doc_subsection"> <a name="convertops">Conversion Operations</a>
3338</div>
3339<div class="doc_text">
3340<p>The instructions in this category are the conversion instructions (casting)
3341which all take a single operand and a type. They perform various bit conversions
3342on the operand.</p>
3343</div>
3344
3345<!-- _______________________________________________________________________ -->
3346<div class="doc_subsubsection">
3347 <a name="i_trunc">'<tt>trunc .. to</tt>' Instruction</a>
3348</div>
3349<div class="doc_text">
3350
3351<h5>Syntax:</h5>
3352<pre>
3353 &lt;result&gt; = trunc &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3354</pre>
3355
3356<h5>Overview:</h5>
3357<p>
3358The '<tt>trunc</tt>' instruction truncates its operand to the type <tt>ty2</tt>.
3359</p>
3360
3361<h5>Arguments:</h5>
3362<p>
3363The '<tt>trunc</tt>' instruction takes a <tt>value</tt> to trunc, which must
3364be an <a href="#t_integer">integer</a> type, and a type that specifies the size
3365and type of the result, which must be an <a href="#t_integer">integer</a>
3366type. The bit size of <tt>value</tt> must be larger than the bit size of
3367<tt>ty2</tt>. Equal sized types are not allowed.</p>
3368
3369<h5>Semantics:</h5>
3370<p>
3371The '<tt>trunc</tt>' instruction truncates the high order bits in <tt>value</tt>
3372and converts the remaining bits to <tt>ty2</tt>. Since the source size must be
3373larger than the destination size, <tt>trunc</tt> cannot be a <i>no-op cast</i>.
3374It will always truncate bits.</p>
3375
3376<h5>Example:</h5>
3377<pre>
3378 %X = trunc i32 257 to i8 <i>; yields i8:1</i>
3379 %Y = trunc i32 123 to i1 <i>; yields i1:true</i>
3380 %Y = trunc i32 122 to i1 <i>; yields i1:false</i>
3381</pre>
3382</div>
3383
3384<!-- _______________________________________________________________________ -->
3385<div class="doc_subsubsection">
3386 <a name="i_zext">'<tt>zext .. to</tt>' Instruction</a>
3387</div>
3388<div class="doc_text">
3389
3390<h5>Syntax:</h5>
3391<pre>
3392 &lt;result&gt; = zext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3393</pre>
3394
3395<h5>Overview:</h5>
3396<p>The '<tt>zext</tt>' instruction zero extends its operand to type
3397<tt>ty2</tt>.</p>
3398
3399
3400<h5>Arguments:</h5>
3401<p>The '<tt>zext</tt>' instruction takes a value to cast, which must be of
3402<a href="#t_integer">integer</a> type, and a type to cast it to, which must
3403also be of <a href="#t_integer">integer</a> type. The bit size of the
3404<tt>value</tt> must be smaller than the bit size of the destination type,
3405<tt>ty2</tt>.</p>
3406
3407<h5>Semantics:</h5>
3408<p>The <tt>zext</tt> fills the high order bits of the <tt>value</tt> with zero
3409bits until it reaches the size of the destination type, <tt>ty2</tt>.</p>
3410
3411<p>When zero extending from i1, the result will always be either 0 or 1.</p>
3412
3413<h5>Example:</h5>
3414<pre>
3415 %X = zext i32 257 to i64 <i>; yields i64:257</i>
3416 %Y = zext i1 true to i32 <i>; yields i32:1</i>
3417</pre>
3418</div>
3419
3420<!-- _______________________________________________________________________ -->
3421<div class="doc_subsubsection">
3422 <a name="i_sext">'<tt>sext .. to</tt>' Instruction</a>
3423</div>
3424<div class="doc_text">
3425
3426<h5>Syntax:</h5>
3427<pre>
3428 &lt;result&gt; = sext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3429</pre>
3430
3431<h5>Overview:</h5>
3432<p>The '<tt>sext</tt>' sign extends <tt>value</tt> to the type <tt>ty2</tt>.</p>
3433
3434<h5>Arguments:</h5>
3435<p>
3436The '<tt>sext</tt>' instruction takes a value to cast, which must be of
3437<a href="#t_integer">integer</a> type, and a type to cast it to, which must
3438also be of <a href="#t_integer">integer</a> type. The bit size of the
3439<tt>value</tt> must be smaller than the bit size of the destination type,
3440<tt>ty2</tt>.</p>
3441
3442<h5>Semantics:</h5>
3443<p>
3444The '<tt>sext</tt>' instruction performs a sign extension by copying the sign
3445bit (highest order bit) of the <tt>value</tt> until it reaches the bit size of
3446the type <tt>ty2</tt>.</p>
3447
3448<p>When sign extending from i1, the extension always results in -1 or 0.</p>
3449
3450<h5>Example:</h5>
3451<pre>
3452 %X = sext i8 -1 to i16 <i>; yields i16 :65535</i>
3453 %Y = sext i1 true to i32 <i>; yields i32:-1</i>
3454</pre>
3455</div>
3456
3457<!-- _______________________________________________________________________ -->
3458<div class="doc_subsubsection">
3459 <a name="i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a>
3460</div>
3461
3462<div class="doc_text">
3463
3464<h5>Syntax:</h5>
3465
3466<pre>
3467 &lt;result&gt; = fptrunc &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3468</pre>
3469
3470<h5>Overview:</h5>
3471<p>The '<tt>fptrunc</tt>' instruction truncates <tt>value</tt> to type
3472<tt>ty2</tt>.</p>
3473
3474
3475<h5>Arguments:</h5>
3476<p>The '<tt>fptrunc</tt>' instruction takes a <a href="#t_floating">floating
3477 point</a> value to cast and a <a href="#t_floating">floating point</a> type to
3478cast it to. The size of <tt>value</tt> must be larger than the size of
3479<tt>ty2</tt>. This implies that <tt>fptrunc</tt> cannot be used to make a
3480<i>no-op cast</i>.</p>
3481
3482<h5>Semantics:</h5>
3483<p> The '<tt>fptrunc</tt>' instruction truncates a <tt>value</tt> from a larger
3484<a href="#t_floating">floating point</a> type to a smaller
3485<a href="#t_floating">floating point</a> type. If the value cannot fit within
3486the destination type, <tt>ty2</tt>, then the results are undefined.</p>
3487
3488<h5>Example:</h5>
3489<pre>
3490 %X = fptrunc double 123.0 to float <i>; yields float:123.0</i>
3491 %Y = fptrunc double 1.0E+300 to float <i>; yields undefined</i>
3492</pre>
3493</div>
3494
3495<!-- _______________________________________________________________________ -->
3496<div class="doc_subsubsection">
3497 <a name="i_fpext">'<tt>fpext .. to</tt>' Instruction</a>
3498</div>
3499<div class="doc_text">
3500
3501<h5>Syntax:</h5>
3502<pre>
3503 &lt;result&gt; = fpext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3504</pre>
3505
3506<h5>Overview:</h5>
3507<p>The '<tt>fpext</tt>' extends a floating point <tt>value</tt> to a larger
3508floating point value.</p>
3509
3510<h5>Arguments:</h5>
3511<p>The '<tt>fpext</tt>' instruction takes a
3512<a href="#t_floating">floating point</a> <tt>value</tt> to cast,
3513and a <a href="#t_floating">floating point</a> type to cast it to. The source
3514type must be smaller than the destination type.</p>
3515
3516<h5>Semantics:</h5>
3517<p>The '<tt>fpext</tt>' instruction extends the <tt>value</tt> from a smaller
3518<a href="#t_floating">floating point</a> type to a larger
3519<a href="#t_floating">floating point</a> type. The <tt>fpext</tt> cannot be
3520used to make a <i>no-op cast</i> because it always changes bits. Use
3521<tt>bitcast</tt> to make a <i>no-op cast</i> for a floating point cast.</p>
3522
3523<h5>Example:</h5>
3524<pre>
3525 %X = fpext float 3.1415 to double <i>; yields double:3.1415</i>
3526 %Y = fpext float 1.0 to float <i>; yields float:1.0 (no-op)</i>
3527</pre>
3528</div>
3529
3530<!-- _______________________________________________________________________ -->
3531<div class="doc_subsubsection">
3532 <a name="i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a>
3533</div>
3534<div class="doc_text">
3535
3536<h5>Syntax:</h5>
3537<pre>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3538 &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]3539</pre>
3540
3541<h5>Overview:</h5>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3542<p>The '<tt>fptoui</tt>' converts a floating point <tt>value</tt> to its
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3543unsigned integer equivalent of type <tt>ty2</tt>.
3544</p>
3545
3546<h5>Arguments:</h5>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3547<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]3548scalar or vector <a href="#t_floating">floating point</a> value, and a type
3549to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
3550type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
3551vector integer type with the same number of elements as <tt>ty</tt></p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3552
3553<h5>Semantics:</h5>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3554<p> The '<tt>fptoui</tt>' instruction converts its
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3555<a href="#t_floating">floating point</a> operand into the nearest (rounding
3556towards zero) unsigned integer value. If the value cannot fit in <tt>ty2</tt>,
3557the results are undefined.</p>
3558
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3559<h5>Example:</h5>
3560<pre>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3561 %X = fptoui double 123.0 to i32 <i>; yields i32:123</i>
Chris Lattner681f1e82007-09-22 03:17:52 +0000[diff] [blame]3562 %Y = fptoui float 1.0E+300 to i1 <i>; yields undefined:1</i>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3563 %X = fptoui float 1.04E+17 to i8 <i>; yields undefined:1</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3564</pre>
3565</div>
3566
3567<!-- _______________________________________________________________________ -->
3568<div class="doc_subsubsection">
3569 <a name="i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a>
3570</div>
3571<div class="doc_text">
3572
3573<h5>Syntax:</h5>
3574<pre>
3575 &lt;result&gt; = fptosi &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3576</pre>
3577
3578<h5>Overview:</h5>
3579<p>The '<tt>fptosi</tt>' instruction converts
3580<a href="#t_floating">floating point</a> <tt>value</tt> to type <tt>ty2</tt>.
3581</p>
3582
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3583<h5>Arguments:</h5>
3584<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]3585scalar or vector <a href="#t_floating">floating point</a> value, and a type
3586to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
3587type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
3588vector integer type with the same number of elements as <tt>ty</tt></p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3589
3590<h5>Semantics:</h5>
3591<p>The '<tt>fptosi</tt>' instruction converts its
3592<a href="#t_floating">floating point</a> operand into the nearest (rounding
3593towards zero) signed integer value. If the value cannot fit in <tt>ty2</tt>,
3594the results are undefined.</p>
3595
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3596<h5>Example:</h5>
3597<pre>
3598 %X = fptosi double -123.0 to i32 <i>; yields i32:-123</i>
Chris Lattner681f1e82007-09-22 03:17:52 +0000[diff] [blame]3599 %Y = fptosi float 1.0E-247 to i1 <i>; yields undefined:1</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3600 %X = fptosi float 1.04E+17 to i8 <i>; yields undefined:1</i>
3601</pre>
3602</div>
3603
3604<!-- _______________________________________________________________________ -->
3605<div class="doc_subsubsection">
3606 <a name="i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a>
3607</div>
3608<div class="doc_text">
3609
3610<h5>Syntax:</h5>
3611<pre>
3612 &lt;result&gt; = uitofp &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3613</pre>
3614
3615<h5>Overview:</h5>
3616<p>The '<tt>uitofp</tt>' instruction regards <tt>value</tt> as an unsigned
3617integer and converts that value to the <tt>ty2</tt> type.</p>
3618
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3619<h5>Arguments:</h5>
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]3620<p>The '<tt>uitofp</tt>' instruction takes a value to cast, which must be a
3621scalar or vector <a href="#t_integer">integer</a> value, and a type to cast it
3622to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
3623type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
3624floating point type with the same number of elements as <tt>ty</tt></p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3625
3626<h5>Semantics:</h5>
3627<p>The '<tt>uitofp</tt>' instruction interprets its operand as an unsigned
3628integer quantity and converts it to the corresponding floating point value. If
3629the value cannot fit in the floating point value, the results are undefined.</p>
3630
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3631<h5>Example:</h5>
3632<pre>
3633 %X = uitofp i32 257 to float <i>; yields float:257.0</i>
3634 %Y = uitofp i8 -1 to double <i>; yields double:255.0</i>
3635</pre>
3636</div>
3637
3638<!-- _______________________________________________________________________ -->
3639<div class="doc_subsubsection">
3640 <a name="i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a>
3641</div>
3642<div class="doc_text">
3643
3644<h5>Syntax:</h5>
3645<pre>
3646 &lt;result&gt; = sitofp &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3647</pre>
3648
3649<h5>Overview:</h5>
3650<p>The '<tt>sitofp</tt>' instruction regards <tt>value</tt> as a signed
3651integer and converts that value to the <tt>ty2</tt> type.</p>
3652
3653<h5>Arguments:</h5>
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]3654<p>The '<tt>sitofp</tt>' instruction takes a value to cast, which must be a
3655scalar or vector <a href="#t_integer">integer</a> value, and a type to cast it
3656to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
3657type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
3658floating point type with the same number of elements as <tt>ty</tt></p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3659
3660<h5>Semantics:</h5>
3661<p>The '<tt>sitofp</tt>' instruction interprets its operand as a signed
3662integer quantity and converts it to the corresponding floating point value. If
3663the value cannot fit in the floating point value, the results are undefined.</p>
3664
3665<h5>Example:</h5>
3666<pre>
3667 %X = sitofp i32 257 to float <i>; yields float:257.0</i>
3668 %Y = sitofp i8 -1 to double <i>; yields double:-1.0</i>
3669</pre>
3670</div>
3671
3672<!-- _______________________________________________________________________ -->
3673<div class="doc_subsubsection">
3674 <a name="i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a>
3675</div>
3676<div class="doc_text">
3677
3678<h5>Syntax:</h5>
3679<pre>
3680 &lt;result&gt; = ptrtoint &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3681</pre>
3682
3683<h5>Overview:</h5>
3684<p>The '<tt>ptrtoint</tt>' instruction converts the pointer <tt>value</tt> to
3685the integer type <tt>ty2</tt>.</p>
3686
3687<h5>Arguments:</h5>
3688<p>The '<tt>ptrtoint</tt>' instruction takes a <tt>value</tt> to cast, which
3689must be a <a href="#t_pointer">pointer</a> value, and a type to cast it to
3690<tt>ty2</tt>, which must be an <a href="#t_integer">integer</a> type.
3691
3692<h5>Semantics:</h5>
3693<p>The '<tt>ptrtoint</tt>' instruction converts <tt>value</tt> to integer type
3694<tt>ty2</tt> by interpreting the pointer value as an integer and either
3695truncating or zero extending that value to the size of the integer type. If
3696<tt>value</tt> is smaller than <tt>ty2</tt> then a zero extension is done. If
3697<tt>value</tt> is larger than <tt>ty2</tt> then a truncation is done. If they
3698are the same size, then nothing is done (<i>no-op cast</i>) other than a type
3699change.</p>
3700
3701<h5>Example:</h5>
3702<pre>
3703 %X = ptrtoint i32* %X to i8 <i>; yields truncation on 32-bit architecture</i>
3704 %Y = ptrtoint i32* %x to i64 <i>; yields zero extension on 32-bit architecture</i>
3705</pre>
3706</div>
3707
3708<!-- _______________________________________________________________________ -->
3709<div class="doc_subsubsection">
3710 <a name="i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a>
3711</div>
3712<div class="doc_text">
3713
3714<h5>Syntax:</h5>
3715<pre>
3716 &lt;result&gt; = inttoptr &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3717</pre>
3718
3719<h5>Overview:</h5>
3720<p>The '<tt>inttoptr</tt>' instruction converts an integer <tt>value</tt> to
3721a pointer type, <tt>ty2</tt>.</p>
3722
3723<h5>Arguments:</h5>
3724<p>The '<tt>inttoptr</tt>' instruction takes an <a href="#t_integer">integer</a>
3725value to cast, and a type to cast it to, which must be a
3726<a href="#t_pointer">pointer</a> type.
3727
3728<h5>Semantics:</h5>
3729<p>The '<tt>inttoptr</tt>' instruction converts <tt>value</tt> to type
3730<tt>ty2</tt> by applying either a zero extension or a truncation depending on
3731the size of the integer <tt>value</tt>. If <tt>value</tt> is larger than the
3732size of a pointer then a truncation is done. If <tt>value</tt> is smaller than
3733the size of a pointer then a zero extension is done. If they are the same size,
3734nothing is done (<i>no-op cast</i>).</p>
3735
3736<h5>Example:</h5>
3737<pre>
3738 %X = inttoptr i32 255 to i32* <i>; yields zero extension on 64-bit architecture</i>
3739 %X = inttoptr i32 255 to i32* <i>; yields no-op on 32-bit architecture</i>
3740 %Y = inttoptr i64 0 to i32* <i>; yields truncation on 32-bit architecture</i>
3741</pre>
3742</div>
3743
3744<!-- _______________________________________________________________________ -->
3745<div class="doc_subsubsection">
3746 <a name="i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a>
3747</div>
3748<div class="doc_text">
3749
3750<h5>Syntax:</h5>
3751<pre>
3752 &lt;result&gt; = bitcast &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3753</pre>
3754
3755<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3756
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3757<p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
3758<tt>ty2</tt> without changing any bits.</p>
3759
3760<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3761
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3762<p>The '<tt>bitcast</tt>' instruction takes a value to cast, which must be
3763a first class value, and a type to cast it to, which must also be a <a
3764 href="#t_firstclass">first class</a> type. The bit sizes of <tt>value</tt>
3765and the destination type, <tt>ty2</tt>, must be identical. If the source
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3766type is a pointer, the destination type must also be a pointer. This
3767instruction supports bitwise conversion of vectors to integers and to vectors
3768of other types (as long as they have the same size).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3769
3770<h5>Semantics:</h5>
3771<p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
3772<tt>ty2</tt>. It is always a <i>no-op cast</i> because no bits change with
3773this conversion. The conversion is done as if the <tt>value</tt> had been
3774stored to memory and read back as type <tt>ty2</tt>. Pointer types may only be
3775converted to other pointer types with this instruction. To convert pointers to
3776other types, use the <a href="#i_inttoptr">inttoptr</a> or
3777<a href="#i_ptrtoint">ptrtoint</a> instructions first.</p>
3778
3779<h5>Example:</h5>
3780<pre>
3781 %X = bitcast i8 255 to i8 <i>; yields i8 :-1</i>
3782 %Y = bitcast i32* %x to sint* <i>; yields sint*:%x</i>
3783 %Z = bitcast <2xint> %V to i64; <i>; yields i64: %V</i>
3784</pre>
3785</div>
3786
3787<!-- ======================================================================= -->
3788<div class="doc_subsection"> <a name="otherops">Other Operations</a> </div>
3789<div class="doc_text">
3790<p>The instructions in this category are the "miscellaneous"
3791instructions, which defy better classification.</p>
3792</div>
3793
3794<!-- _______________________________________________________________________ -->
3795<div class="doc_subsubsection"><a name="i_icmp">'<tt>icmp</tt>' Instruction</a>
3796</div>
3797<div class="doc_text">
3798<h5>Syntax:</h5>
3799<pre> &lt;result&gt; = icmp &lt;cond&gt; &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {i1}:result</i>
3800</pre>
3801<h5>Overview:</h5>
3802<p>The '<tt>icmp</tt>' instruction returns a boolean value based on comparison
Chris Lattner10368b62008-04-02 00:38:26 +0000[diff] [blame]3803of its two integer or pointer operands.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3804<h5>Arguments:</h5>
3805<p>The '<tt>icmp</tt>' instruction takes three operands. The first operand is
3806the condition code indicating the kind of comparison to perform. It is not
3807a value, just a keyword. The possible condition code are:
3808<ol>
3809 <li><tt>eq</tt>: equal</li>
3810 <li><tt>ne</tt>: not equal </li>
3811 <li><tt>ugt</tt>: unsigned greater than</li>
3812 <li><tt>uge</tt>: unsigned greater or equal</li>
3813 <li><tt>ult</tt>: unsigned less than</li>
3814 <li><tt>ule</tt>: unsigned less or equal</li>
3815 <li><tt>sgt</tt>: signed greater than</li>
3816 <li><tt>sge</tt>: signed greater or equal</li>
3817 <li><tt>slt</tt>: signed less than</li>
3818 <li><tt>sle</tt>: signed less or equal</li>
3819</ol>
3820<p>The remaining two arguments must be <a href="#t_integer">integer</a> or
3821<a href="#t_pointer">pointer</a> typed. They must also be identical types.</p>
3822<h5>Semantics:</h5>
3823<p>The '<tt>icmp</tt>' compares <tt>var1</tt> and <tt>var2</tt> according to
3824the condition code given as <tt>cond</tt>. The comparison performed always
3825yields a <a href="#t_primitive">i1</a> result, as follows:
3826<ol>
3827 <li><tt>eq</tt>: yields <tt>true</tt> if the operands are equal,
3828 <tt>false</tt> otherwise. No sign interpretation is necessary or performed.
3829 </li>
3830 <li><tt>ne</tt>: yields <tt>true</tt> if the operands are unequal,
3831 <tt>false</tt> otherwise. No sign interpretation is necessary or performed.
3832 <li><tt>ugt</tt>: interprets the operands as unsigned values and yields
3833 <tt>true</tt> if <tt>var1</tt> is greater than <tt>var2</tt>.</li>
3834 <li><tt>uge</tt>: interprets the operands as unsigned values and yields
3835 <tt>true</tt> if <tt>var1</tt> is greater than or equal to <tt>var2</tt>.</li>
3836 <li><tt>ult</tt>: interprets the operands as unsigned values and yields
3837 <tt>true</tt> if <tt>var1</tt> is less than <tt>var2</tt>.</li>
3838 <li><tt>ule</tt>: interprets the operands as unsigned values and yields
3839 <tt>true</tt> if <tt>var1</tt> is less than or equal to <tt>var2</tt>.</li>
3840 <li><tt>sgt</tt>: interprets the operands as signed values and yields
3841 <tt>true</tt> if <tt>var1</tt> is greater than <tt>var2</tt>.</li>
3842 <li><tt>sge</tt>: interprets the operands as signed values and yields
3843 <tt>true</tt> if <tt>var1</tt> is greater than or equal to <tt>var2</tt>.</li>
3844 <li><tt>slt</tt>: interprets the operands as signed values and yields
3845 <tt>true</tt> if <tt>var1</tt> is less than <tt>var2</tt>.</li>
3846 <li><tt>sle</tt>: interprets the operands as signed values and yields
3847 <tt>true</tt> if <tt>var1</tt> is less than or equal to <tt>var2</tt>.</li>
3848</ol>
3849<p>If the operands are <a href="#t_pointer">pointer</a> typed, the pointer
3850values are compared as if they were integers.</p>
3851
3852<h5>Example:</h5>
3853<pre> &lt;result&gt; = icmp eq i32 4, 5 <i>; yields: result=false</i>
3854 &lt;result&gt; = icmp ne float* %X, %X <i>; yields: result=false</i>
3855 &lt;result&gt; = icmp ult i16 4, 5 <i>; yields: result=true</i>
3856 &lt;result&gt; = icmp sgt i16 4, 5 <i>; yields: result=false</i>
3857 &lt;result&gt; = icmp ule i16 -4, 5 <i>; yields: result=false</i>
3858 &lt;result&gt; = icmp sge i16 4, 5 <i>; yields: result=false</i>
3859</pre>
3860</div>
3861
3862<!-- _______________________________________________________________________ -->
3863<div class="doc_subsubsection"><a name="i_fcmp">'<tt>fcmp</tt>' Instruction</a>
3864</div>
3865<div class="doc_text">
3866<h5>Syntax:</h5>
3867<pre> &lt;result&gt; = fcmp &lt;cond&gt; &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {i1}:result</i>
3868</pre>
3869<h5>Overview:</h5>
3870<p>The '<tt>fcmp</tt>' instruction returns a boolean value based on comparison
3871of its floating point operands.</p>
3872<h5>Arguments:</h5>
3873<p>The '<tt>fcmp</tt>' instruction takes three operands. The first operand is
3874the condition code indicating the kind of comparison to perform. It is not
3875a value, just a keyword. The possible condition code are:
3876<ol>
3877 <li><tt>false</tt>: no comparison, always returns false</li>
3878 <li><tt>oeq</tt>: ordered and equal</li>
3879 <li><tt>ogt</tt>: ordered and greater than </li>
3880 <li><tt>oge</tt>: ordered and greater than or equal</li>
3881 <li><tt>olt</tt>: ordered and less than </li>
3882 <li><tt>ole</tt>: ordered and less than or equal</li>
3883 <li><tt>one</tt>: ordered and not equal</li>
3884 <li><tt>ord</tt>: ordered (no nans)</li>
3885 <li><tt>ueq</tt>: unordered or equal</li>
3886 <li><tt>ugt</tt>: unordered or greater than </li>
3887 <li><tt>uge</tt>: unordered or greater than or equal</li>
3888 <li><tt>ult</tt>: unordered or less than </li>
3889 <li><tt>ule</tt>: unordered or less than or equal</li>
3890 <li><tt>une</tt>: unordered or not equal</li>
3891 <li><tt>uno</tt>: unordered (either nans)</li>
3892 <li><tt>true</tt>: no comparison, always returns true</li>
3893</ol>
3894<p><i>Ordered</i> means that neither operand is a QNAN while
3895<i>unordered</i> means that either operand may be a QNAN.</p>
3896<p>The <tt>val1</tt> and <tt>val2</tt> arguments must be
3897<a href="#t_floating">floating point</a> typed. They must have identical
3898types.</p>
3899<h5>Semantics:</h5>
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]3900<p>The '<tt>fcmp</tt>' instruction compares <tt>var1</tt> and <tt>var2</tt>
3901according to the condition code given as <tt>cond</tt>. The comparison performed
3902always yields a <a href="#t_primitive">i1</a> result, as follows:
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3903<ol>
3904 <li><tt>false</tt>: always yields <tt>false</tt>, regardless of operands.</li>
3905 <li><tt>oeq</tt>: yields <tt>true</tt> if both operands are not a QNAN and
3906 <tt>var1</tt> is equal to <tt>var2</tt>.</li>
3907 <li><tt>ogt</tt>: yields <tt>true</tt> if both operands are not a QNAN and
3908 <tt>var1</tt> is greather than <tt>var2</tt>.</li>
3909 <li><tt>oge</tt>: yields <tt>true</tt> if both operands are not a QNAN and
3910 <tt>var1</tt> is greater than or equal to <tt>var2</tt>.</li>
3911 <li><tt>olt</tt>: yields <tt>true</tt> if both operands are not a QNAN and
3912 <tt>var1</tt> is less than <tt>var2</tt>.</li>
3913 <li><tt>ole</tt>: yields <tt>true</tt> if both operands are not a QNAN and
3914 <tt>var1</tt> is less than or equal to <tt>var2</tt>.</li>
3915 <li><tt>one</tt>: yields <tt>true</tt> if both operands are not a QNAN and
3916 <tt>var1</tt> is not equal to <tt>var2</tt>.</li>
3917 <li><tt>ord</tt>: yields <tt>true</tt> if both operands are not a QNAN.</li>
3918 <li><tt>ueq</tt>: yields <tt>true</tt> if either operand is a QNAN or
3919 <tt>var1</tt> is equal to <tt>var2</tt>.</li>
3920 <li><tt>ugt</tt>: yields <tt>true</tt> if either operand is a QNAN or
3921 <tt>var1</tt> is greater than <tt>var2</tt>.</li>
3922 <li><tt>uge</tt>: yields <tt>true</tt> if either operand is a QNAN or
3923 <tt>var1</tt> is greater than or equal to <tt>var2</tt>.</li>
3924 <li><tt>ult</tt>: yields <tt>true</tt> if either operand is a QNAN or
3925 <tt>var1</tt> is less than <tt>var2</tt>.</li>
3926 <li><tt>ule</tt>: yields <tt>true</tt> if either operand is a QNAN or
3927 <tt>var1</tt> is less than or equal to <tt>var2</tt>.</li>
3928 <li><tt>une</tt>: yields <tt>true</tt> if either operand is a QNAN or
3929 <tt>var1</tt> is not equal to <tt>var2</tt>.</li>
3930 <li><tt>uno</tt>: yields <tt>true</tt> if either operand is a QNAN.</li>
3931 <li><tt>true</tt>: always yields <tt>true</tt>, regardless of operands.</li>
3932</ol>
3933
3934<h5>Example:</h5>
3935<pre> &lt;result&gt; = fcmp oeq float 4.0, 5.0 <i>; yields: result=false</i>
3936 &lt;result&gt; = icmp one float 4.0, 5.0 <i>; yields: result=true</i>
3937 &lt;result&gt; = icmp olt float 4.0, 5.0 <i>; yields: result=true</i>
3938 &lt;result&gt; = icmp ueq double 1.0, 2.0 <i>; yields: result=false</i>
3939</pre>
3940</div>
3941
3942<!-- _______________________________________________________________________ -->
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]3943<div class="doc_subsubsection">
3944 <a name="i_vicmp">'<tt>vicmp</tt>' Instruction</a>
3945</div>
3946<div class="doc_text">
3947<h5>Syntax:</h5>
3948<pre> &lt;result&gt; = vicmp &lt;cond&gt; &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
3949</pre>
3950<h5>Overview:</h5>
3951<p>The '<tt>vicmp</tt>' instruction returns an integer vector value based on
3952element-wise comparison of its two integer vector operands.</p>
3953<h5>Arguments:</h5>
3954<p>The '<tt>vicmp</tt>' instruction takes three operands. The first operand is
3955the condition code indicating the kind of comparison to perform. It is not
3956a value, just a keyword. The possible condition code are:
3957<ol>
3958 <li><tt>eq</tt>: equal</li>
3959 <li><tt>ne</tt>: not equal </li>
3960 <li><tt>ugt</tt>: unsigned greater than</li>
3961 <li><tt>uge</tt>: unsigned greater or equal</li>
3962 <li><tt>ult</tt>: unsigned less than</li>
3963 <li><tt>ule</tt>: unsigned less or equal</li>
3964 <li><tt>sgt</tt>: signed greater than</li>
3965 <li><tt>sge</tt>: signed greater or equal</li>
3966 <li><tt>slt</tt>: signed less than</li>
3967 <li><tt>sle</tt>: signed less or equal</li>
3968</ol>
3969<p>The remaining two arguments must be <a href="#t_vector">vector</a> of
3970<a href="#t_integer">integer</a> typed. They must also be identical types.</p>
3971<h5>Semantics:</h5>
3972<p>The '<tt>vicmp</tt>' instruction compares <tt>var1</tt> and <tt>var2</tt>
3973according to the condition code given as <tt>cond</tt>. The comparison yields a
3974<a href="#t_vector">vector</a> of <a href="#t_integer">integer</a> result, of
3975identical type as the values being compared. The most significant bit in each
3976element is 1 if the element-wise comparison evaluates to true, and is 0
3977otherwise. All other bits of the result are undefined. The condition codes
3978are evaluated identically to the <a href="#i_icmp">'<tt>icmp</tt>'
3979instruction</a>.
3980
3981<h5>Example:</h5>
3982<pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3983 &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>
3984 &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]3985</pre>
3986</div>
3987
3988<!-- _______________________________________________________________________ -->
3989<div class="doc_subsubsection">
3990 <a name="i_vfcmp">'<tt>vfcmp</tt>' Instruction</a>
3991</div>
3992<div class="doc_text">
3993<h5>Syntax:</h5>
3994<pre> &lt;result&gt; = vfcmp &lt;cond&gt; &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt;</pre>
3995<h5>Overview:</h5>
3996<p>The '<tt>vfcmp</tt>' instruction returns an integer vector value based on
3997element-wise comparison of its two floating point vector operands. The output
3998elements have the same width as the input elements.</p>
3999<h5>Arguments:</h5>
4000<p>The '<tt>vfcmp</tt>' instruction takes three operands. The first operand is
4001the condition code indicating the kind of comparison to perform. It is not
4002a value, just a keyword. The possible condition code are:
4003<ol>
4004 <li><tt>false</tt>: no comparison, always returns false</li>
4005 <li><tt>oeq</tt>: ordered and equal</li>
4006 <li><tt>ogt</tt>: ordered and greater than </li>
4007 <li><tt>oge</tt>: ordered and greater than or equal</li>
4008 <li><tt>olt</tt>: ordered and less than </li>
4009 <li><tt>ole</tt>: ordered and less than or equal</li>
4010 <li><tt>one</tt>: ordered and not equal</li>
4011 <li><tt>ord</tt>: ordered (no nans)</li>
4012 <li><tt>ueq</tt>: unordered or equal</li>
4013 <li><tt>ugt</tt>: unordered or greater than </li>
4014 <li><tt>uge</tt>: unordered or greater than or equal</li>
4015 <li><tt>ult</tt>: unordered or less than </li>
4016 <li><tt>ule</tt>: unordered or less than or equal</li>
4017 <li><tt>une</tt>: unordered or not equal</li>
4018 <li><tt>uno</tt>: unordered (either nans)</li>
4019 <li><tt>true</tt>: no comparison, always returns true</li>
4020</ol>
4021<p>The remaining two arguments must be <a href="#t_vector">vector</a> of
4022<a href="#t_floating">floating point</a> typed. They must also be identical
4023types.</p>
4024<h5>Semantics:</h5>
4025<p>The '<tt>vfcmp</tt>' instruction compares <tt>var1</tt> and <tt>var2</tt>
4026according to the condition code given as <tt>cond</tt>. The comparison yields a
4027<a href="#t_vector">vector</a> of <a href="#t_integer">integer</a> result, with
4028an identical number of elements as the values being compared, and each element
4029having identical with to the width of the floating point elements. The most
4030significant bit in each element is 1 if the element-wise comparison evaluates to
4031true, and is 0 otherwise. All other bits of the result are undefined. The
4032condition codes are evaluated identically to the
4033<a href="#i_fcmp">'<tt>fcmp</tt>' instruction</a>.
4034
4035<h5>Example:</h5>
4036<pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4037 &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>
4038 &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]4039</pre>
4040</div>
4041
4042<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4043<div class="doc_subsubsection">
4044 <a name="i_phi">'<tt>phi</tt>' Instruction</a>
4045</div>
4046
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4047<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4048
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4049<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4050
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4051<pre> &lt;result&gt; = phi &lt;ty&gt; [ &lt;val0&gt;, &lt;label0&gt;], ...<br></pre>
4052<h5>Overview:</h5>
4053<p>The '<tt>phi</tt>' instruction is used to implement the &#966; node in
4054the SSA graph representing the function.</p>
4055<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4056
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4057<p>The type of the incoming values is specified with the first type
4058field. After this, the '<tt>phi</tt>' instruction takes a list of pairs
4059as arguments, with one pair for each predecessor basic block of the
4060current block. Only values of <a href="#t_firstclass">first class</a>
4061type may be used as the value arguments to the PHI node. Only labels
4062may be used as the label arguments.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4063
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4064<p>There must be no non-phi instructions between the start of a basic
4065block and the PHI instructions: i.e. PHI instructions must be first in
4066a basic block.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4067
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4068<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4069
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4070<p>At runtime, the '<tt>phi</tt>' instruction logically takes on the value
4071specified by the pair corresponding to the predecessor basic block that executed
4072just prior to the current block.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4073
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4074<h5>Example:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4075<pre>
4076Loop: ; Infinite loop that counts from 0 on up...
4077 %indvar = phi i32 [ 0, %LoopHeader ], [ %nextindvar, %Loop ]
4078 %nextindvar = add i32 %indvar, 1
4079 br label %Loop
4080</pre>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4081</div>
4082
4083<!-- _______________________________________________________________________ -->
4084<div class="doc_subsubsection">
4085 <a name="i_select">'<tt>select</tt>' Instruction</a>
4086</div>
4087
4088<div class="doc_text">
4089
4090<h5>Syntax:</h5>
4091
4092<pre>
4093 &lt;result&gt; = select i1 &lt;cond&gt;, &lt;ty&gt; &lt;val1&gt;, &lt;ty&gt; &lt;val2&gt; <i>; yields ty</i>
4094</pre>
4095
4096<h5>Overview:</h5>
4097
4098<p>
4099The '<tt>select</tt>' instruction is used to choose one value based on a
4100condition, without branching.
4101</p>
4102
4103
4104<h5>Arguments:</h5>
4105
4106<p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4107The '<tt>select</tt>' instruction requires an 'i1' value indicating the
4108condition, and two values of the same <a href="#t_firstclass">first class</a>
4109type. If the val1/val2 are vectors, the entire vectors are selected, not
4110individual elements.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4111</p>
4112
4113<h5>Semantics:</h5>
4114
4115<p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4116If the i1 condition evaluates is 1, the instruction returns the first
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4117value argument; otherwise, it returns the second value argument.
4118</p>
4119
4120<h5>Example:</h5>
4121
4122<pre>
4123 %X = select i1 true, i8 17, i8 42 <i>; yields i8:17</i>
4124</pre>
4125</div>
4126
4127
4128<!-- _______________________________________________________________________ -->
4129<div class="doc_subsubsection">
4130 <a name="i_call">'<tt>call</tt>' Instruction</a>
4131</div>
4132
4133<div class="doc_text">
4134
4135<h5>Syntax:</h5>
4136<pre>
Nick Lewycky93082fc2007-09-08 13:57:50 +0000[diff] [blame]4137 &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]4138</pre>
4139
4140<h5>Overview:</h5>
4141
4142<p>The '<tt>call</tt>' instruction represents a simple function call.</p>
4143
4144<h5>Arguments:</h5>
4145
4146<p>This instruction requires several arguments:</p>
4147
4148<ol>
4149 <li>
4150 <p>The optional "tail" marker indicates whether the callee function accesses
4151 any allocas or varargs in the caller. If the "tail" marker is present, the
4152 function call is eligible for tail call optimization. Note that calls may
4153 be marked "tail" even if they do not occur before a <a
4154 href="#i_ret"><tt>ret</tt></a> instruction.
4155 </li>
4156 <li>
4157 <p>The optional "cconv" marker indicates which <a href="#callingconv">calling
4158 convention</a> the call should use. If none is specified, the call defaults
4159 to using C calling conventions.
4160 </li>
4161 <li>
Nick Lewycky93082fc2007-09-08 13:57:50 +0000[diff] [blame]4162 <p>'<tt>ty</tt>': the type of the call instruction itself which is also
4163 the type of the return value. Functions that return no value are marked
4164 <tt><a href="#t_void">void</a></tt>.</p>
4165 </li>
4166 <li>
4167 <p>'<tt>fnty</tt>': shall be the signature of the pointer to function
4168 value being invoked. The argument types must match the types implied by
4169 this signature. This type can be omitted if the function is not varargs
4170 and if the function type does not return a pointer to a function.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4171 </li>
4172 <li>
4173 <p>'<tt>fnptrval</tt>': An LLVM value containing a pointer to a function to
4174 be invoked. In most cases, this is a direct function invocation, but
4175 indirect <tt>call</tt>s are just as possible, calling an arbitrary pointer
4176 to function value.</p>
4177 </li>
4178 <li>
4179 <p>'<tt>function args</tt>': argument list whose types match the
4180 function signature argument types. All arguments must be of
4181 <a href="#t_firstclass">first class</a> type. If the function signature
4182 indicates the function accepts a variable number of arguments, the extra
4183 arguments can be specified.</p>
4184 </li>
4185</ol>
4186
4187<h5>Semantics:</h5>
4188
4189<p>The '<tt>call</tt>' instruction is used to cause control flow to
4190transfer to a specified function, with its incoming arguments bound to
4191the specified values. Upon a '<tt><a href="#i_ret">ret</a></tt>'
4192instruction in the called function, control flow continues with the
4193instruction after the function call, and the return value of the
Chris Lattner5e893ef2008-03-21 17:24:17 +0000[diff] [blame]4194function is bound to the result argument. If the callee returns multiple
4195values then the return values of the function are only accessible through
4196the '<tt><a href="#i_getresult">getresult</a></tt>' instruction.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4197
4198<h5>Example:</h5>
4199
4200<pre>
Nick Lewycky93082fc2007-09-08 13:57:50 +0000[diff] [blame]4201 %retval = call i32 @test(i32 %argc)
Chris Lattner5e893ef2008-03-21 17:24:17 +0000[diff] [blame]4202 call i32 (i8 *, ...)* @printf(i8 * %msg, i32 12, i8 42) <i>; yields i32</i>
4203 %X = tail call i32 @foo() <i>; yields i32</i>
4204 %Y = tail call <a href="#callingconv">fastcc</a> i32 @foo() <i>; yields i32</i>
4205 call void %foo(i8 97 signext)
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]4206
4207 %struct.A = type { i32, i8 }
Chris Lattner5e893ef2008-03-21 17:24:17 +0000[diff] [blame]4208 %r = call %struct.A @foo() <i>; yields { 32, i8 }</i>
4209 %gr = getresult %struct.A %r, 0 <i>; yields i32</i>
4210 %gr1 = getresult %struct.A %r, 1 <i>; yields i8</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4211</pre>
4212
4213</div>
4214
4215<!-- _______________________________________________________________________ -->
4216<div class="doc_subsubsection">
4217 <a name="i_va_arg">'<tt>va_arg</tt>' Instruction</a>
4218</div>
4219
4220<div class="doc_text">
4221
4222<h5>Syntax:</h5>
4223
4224<pre>
4225 &lt;resultval&gt; = va_arg &lt;va_list*&gt; &lt;arglist&gt;, &lt;argty&gt;
4226</pre>
4227
4228<h5>Overview:</h5>
4229
4230<p>The '<tt>va_arg</tt>' instruction is used to access arguments passed through
4231the "variable argument" area of a function call. It is used to implement the
4232<tt>va_arg</tt> macro in C.</p>
4233
4234<h5>Arguments:</h5>
4235
4236<p>This instruction takes a <tt>va_list*</tt> value and the type of
4237the argument. It returns a value of the specified argument type and
4238increments the <tt>va_list</tt> to point to the next argument. The
4239actual type of <tt>va_list</tt> is target specific.</p>
4240
4241<h5>Semantics:</h5>
4242
4243<p>The '<tt>va_arg</tt>' instruction loads an argument of the specified
4244type from the specified <tt>va_list</tt> and causes the
4245<tt>va_list</tt> to point to the next argument. For more information,
4246see the variable argument handling <a href="#int_varargs">Intrinsic
4247Functions</a>.</p>
4248
4249<p>It is legal for this instruction to be called in a function which does not
4250take a variable number of arguments, for example, the <tt>vfprintf</tt>
4251function.</p>
4252
4253<p><tt>va_arg</tt> is an LLVM instruction instead of an <a
4254href="#intrinsics">intrinsic function</a> because it takes a type as an
4255argument.</p>
4256
4257<h5>Example:</h5>
4258
4259<p>See the <a href="#int_varargs">variable argument processing</a> section.</p>
4260
4261</div>
4262
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]4263<!-- _______________________________________________________________________ -->
4264<div class="doc_subsubsection">
4265 <a name="i_getresult">'<tt>getresult</tt>' Instruction</a>
4266</div>
4267
4268<div class="doc_text">
4269
4270<h5>Syntax:</h5>
4271<pre>
Chris Lattneree9da3f2008-03-21 17:20:51 +0000[diff] [blame]4272 &lt;resultval&gt; = getresult &lt;type&gt; &lt;retval&gt;, &lt;index&gt;
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]4273</pre>
Chris Lattneree9da3f2008-03-21 17:20:51 +0000[diff] [blame]4274
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]4275<h5>Overview:</h5>
4276
4277<p> The '<tt>getresult</tt>' instruction is used to extract individual values
Chris Lattneree9da3f2008-03-21 17:20:51 +0000[diff] [blame]4278from a '<tt><a href="#i_call">call</a></tt>'
4279or '<tt><a href="#i_invoke">invoke</a></tt>' instruction that returns multiple
4280results.</p>
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]4281
4282<h5>Arguments:</h5>
4283
Chris Lattneree9da3f2008-03-21 17:20:51 +0000[diff] [blame]4284<p>The '<tt>getresult</tt>' instruction takes a call or invoke value as its
Chris Lattnerd8dd3522008-04-23 04:06:52 +0000[diff] [blame]4285first argument, or an undef value. The value must have <a
4286href="#t_struct">structure type</a>. The second argument is a constant
4287unsigned index value which must be in range for the number of values returned
4288by the call.</p>
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]4289
4290<h5>Semantics:</h5>
4291
Chris Lattneree9da3f2008-03-21 17:20:51 +0000[diff] [blame]4292<p>The '<tt>getresult</tt>' instruction extracts the element identified by
4293'<tt>index</tt>' from the aggregate value.</p>
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]4294
4295<h5>Example:</h5>
4296
4297<pre>
4298 %struct.A = type { i32, i8 }
4299
4300 %r = call %struct.A @foo()
Chris Lattneree9da3f2008-03-21 17:20:51 +0000[diff] [blame]4301 %gr = getresult %struct.A %r, 0 <i>; yields i32:%gr</i>
4302 %gr1 = getresult %struct.A %r, 1 <i>; yields i8:%gr1</i>
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]4303 add i32 %gr, 42
4304 add i8 %gr1, 41
4305</pre>
4306
4307</div>
4308
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4309<!-- *********************************************************************** -->
4310<div class="doc_section"> <a name="intrinsics">Intrinsic Functions</a> </div>
4311<!-- *********************************************************************** -->
4312
4313<div class="doc_text">
4314
4315<p>LLVM supports the notion of an "intrinsic function". These functions have
4316well known names and semantics and are required to follow certain restrictions.
4317Overall, these intrinsics represent an extension mechanism for the LLVM
4318language that does not require changing all of the transformations in LLVM when
4319adding to the language (or the bitcode reader/writer, the parser, etc...).</p>
4320
4321<p>Intrinsic function names must all start with an "<tt>llvm.</tt>" prefix. This
4322prefix is reserved in LLVM for intrinsic names; thus, function names may not
4323begin with this prefix. Intrinsic functions must always be external functions:
4324you cannot define the body of intrinsic functions. Intrinsic functions may
4325only be used in call or invoke instructions: it is illegal to take the address
4326of an intrinsic function. Additionally, because intrinsic functions are part
4327of the LLVM language, it is required if any are added that they be documented
4328here.</p>
4329
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]4330<p>Some intrinsic functions can be overloaded, i.e., the intrinsic represents
4331a family of functions that perform the same operation but on different data
4332types. Because LLVM can represent over 8 million different integer types,
4333overloading is used commonly to allow an intrinsic function to operate on any
4334integer type. One or more of the argument types or the result type can be
4335overloaded to accept any integer type. Argument types may also be defined as
4336exactly matching a previous argument's type or the result type. This allows an
4337intrinsic function which accepts multiple arguments, but needs all of them to
4338be of the same type, to only be overloaded with respect to a single argument or
4339the result.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4340
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]4341<p>Overloaded intrinsics will have the names of its overloaded argument types
4342encoded into its function name, each preceded by a period. Only those types
4343which are overloaded result in a name suffix. Arguments whose type is matched
4344against another type do not. For example, the <tt>llvm.ctpop</tt> function can
4345take an integer of any width and returns an integer of exactly the same integer
4346width. This leads to a family of functions such as
4347<tt>i8 @llvm.ctpop.i8(i8 %val)</tt> and <tt>i29 @llvm.ctpop.i29(i29 %val)</tt>.
4348Only one type, the return type, is overloaded, and only one type suffix is
4349required. Because the argument's type is matched against the return type, it
4350does not require its own name suffix.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4351
4352<p>To learn how to add an intrinsic function, please see the
4353<a href="ExtendingLLVM.html">Extending LLVM Guide</a>.
4354</p>
4355
4356</div>
4357
4358<!-- ======================================================================= -->
4359<div class="doc_subsection">
4360 <a name="int_varargs">Variable Argument Handling Intrinsics</a>
4361</div>
4362
4363<div class="doc_text">
4364
4365<p>Variable argument support is defined in LLVM with the <a
4366 href="#i_va_arg"><tt>va_arg</tt></a> instruction and these three
4367intrinsic functions. These functions are related to the similarly
4368named macros defined in the <tt>&lt;stdarg.h&gt;</tt> header file.</p>
4369
4370<p>All of these functions operate on arguments that use a
4371target-specific value type "<tt>va_list</tt>". The LLVM assembly
4372language reference manual does not define what this type is, so all
4373transformations should be prepared to handle these functions regardless of
4374the type used.</p>
4375
4376<p>This example shows how the <a href="#i_va_arg"><tt>va_arg</tt></a>
4377instruction and the variable argument handling intrinsic functions are
4378used.</p>
4379
4380<div class="doc_code">
4381<pre>
4382define i32 @test(i32 %X, ...) {
4383 ; Initialize variable argument processing
4384 %ap = alloca i8*
4385 %ap2 = bitcast i8** %ap to i8*
4386 call void @llvm.va_start(i8* %ap2)
4387
4388 ; Read a single integer argument
4389 %tmp = va_arg i8** %ap, i32
4390
4391 ; Demonstrate usage of llvm.va_copy and llvm.va_end
4392 %aq = alloca i8*
4393 %aq2 = bitcast i8** %aq to i8*
4394 call void @llvm.va_copy(i8* %aq2, i8* %ap2)
4395 call void @llvm.va_end(i8* %aq2)
4396
4397 ; Stop processing of arguments.
4398 call void @llvm.va_end(i8* %ap2)
4399 ret i32 %tmp
4400}
4401
4402declare void @llvm.va_start(i8*)
4403declare void @llvm.va_copy(i8*, i8*)
4404declare void @llvm.va_end(i8*)
4405</pre>
4406</div>
4407
4408</div>
4409
4410<!-- _______________________________________________________________________ -->
4411<div class="doc_subsubsection">
4412 <a name="int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a>
4413</div>
4414
4415
4416<div class="doc_text">
4417<h5>Syntax:</h5>
4418<pre> declare void %llvm.va_start(i8* &lt;arglist&gt;)<br></pre>
4419<h5>Overview:</h5>
4420<P>The '<tt>llvm.va_start</tt>' intrinsic initializes
4421<tt>*&lt;arglist&gt;</tt> for subsequent use by <tt><a
4422href="#i_va_arg">va_arg</a></tt>.</p>
4423
4424<h5>Arguments:</h5>
4425
4426<P>The argument is a pointer to a <tt>va_list</tt> element to initialize.</p>
4427
4428<h5>Semantics:</h5>
4429
4430<P>The '<tt>llvm.va_start</tt>' intrinsic works just like the <tt>va_start</tt>
4431macro available in C. In a target-dependent way, it initializes the
4432<tt>va_list</tt> element to which the argument points, so that the next call to
4433<tt>va_arg</tt> will produce the first variable argument passed to the function.
4434Unlike the C <tt>va_start</tt> macro, this intrinsic does not need to know the
4435last argument of the function as the compiler can figure that out.</p>
4436
4437</div>
4438
4439<!-- _______________________________________________________________________ -->
4440<div class="doc_subsubsection">
4441 <a name="int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a>
4442</div>
4443
4444<div class="doc_text">
4445<h5>Syntax:</h5>
4446<pre> declare void @llvm.va_end(i8* &lt;arglist&gt;)<br></pre>
4447<h5>Overview:</h5>
4448
4449<p>The '<tt>llvm.va_end</tt>' intrinsic destroys <tt>*&lt;arglist&gt;</tt>,
4450which has been initialized previously with <tt><a href="#int_va_start">llvm.va_start</a></tt>
4451or <tt><a href="#i_va_copy">llvm.va_copy</a></tt>.</p>
4452
4453<h5>Arguments:</h5>
4454
4455<p>The argument is a pointer to a <tt>va_list</tt> to destroy.</p>
4456
4457<h5>Semantics:</h5>
4458
4459<p>The '<tt>llvm.va_end</tt>' intrinsic works just like the <tt>va_end</tt>
4460macro available in C. In a target-dependent way, it destroys the
4461<tt>va_list</tt> element to which the argument points. Calls to <a
4462href="#int_va_start"><tt>llvm.va_start</tt></a> and <a href="#int_va_copy">
4463<tt>llvm.va_copy</tt></a> must be matched exactly with calls to
4464<tt>llvm.va_end</tt>.</p>
4465
4466</div>
4467
4468<!-- _______________________________________________________________________ -->
4469<div class="doc_subsubsection">
4470 <a name="int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a>
4471</div>
4472
4473<div class="doc_text">
4474
4475<h5>Syntax:</h5>
4476
4477<pre>
4478 declare void @llvm.va_copy(i8* &lt;destarglist&gt;, i8* &lt;srcarglist&gt;)
4479</pre>
4480
4481<h5>Overview:</h5>
4482
4483<p>The '<tt>llvm.va_copy</tt>' intrinsic copies the current argument position
4484from the source argument list to the destination argument list.</p>
4485
4486<h5>Arguments:</h5>
4487
4488<p>The first argument is a pointer to a <tt>va_list</tt> element to initialize.
4489The second argument is a pointer to a <tt>va_list</tt> element to copy from.</p>
4490
4491
4492<h5>Semantics:</h5>
4493
4494<p>The '<tt>llvm.va_copy</tt>' intrinsic works just like the <tt>va_copy</tt>
4495macro available in C. In a target-dependent way, it copies the source
4496<tt>va_list</tt> element into the destination <tt>va_list</tt> element. This
4497intrinsic is necessary because the <tt><a href="#int_va_start">
4498llvm.va_start</a></tt> intrinsic may be arbitrarily complex and require, for
4499example, memory allocation.</p>
4500
4501</div>
4502
4503<!-- ======================================================================= -->
4504<div class="doc_subsection">
4505 <a name="int_gc">Accurate Garbage Collection Intrinsics</a>
4506</div>
4507
4508<div class="doc_text">
4509
4510<p>
4511LLVM support for <a href="GarbageCollection.html">Accurate Garbage
4512Collection</a> requires the implementation and generation of these intrinsics.
4513These intrinsics allow identification of <a href="#int_gcroot">GC roots on the
4514stack</a>, as well as garbage collector implementations that require <a
4515href="#int_gcread">read</a> and <a href="#int_gcwrite">write</a> barriers.
4516Front-ends for type-safe garbage collected languages should generate these
4517intrinsics to make use of the LLVM garbage collectors. For more details, see <a
4518href="GarbageCollection.html">Accurate Garbage Collection with LLVM</a>.
4519</p>
Christopher Lambcfe00962007-12-17 01:00:21 +0000[diff] [blame]4520
4521<p>The garbage collection intrinsics only operate on objects in the generic
4522 address space (address space zero).</p>
4523
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4524</div>
4525
4526<!-- _______________________________________________________________________ -->
4527<div class="doc_subsubsection">
4528 <a name="int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a>
4529</div>
4530
4531<div class="doc_text">
4532
4533<h5>Syntax:</h5>
4534
4535<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4536 declare void @llvm.gcroot(i8** %ptrloc, i8* %metadata)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4537</pre>
4538
4539<h5>Overview:</h5>
4540
4541<p>The '<tt>llvm.gcroot</tt>' intrinsic declares the existence of a GC root to
4542the code generator, and allows some metadata to be associated with it.</p>
4543
4544<h5>Arguments:</h5>
4545
4546<p>The first argument specifies the address of a stack object that contains the
4547root pointer. The second pointer (which must be either a constant or a global
4548value address) contains the meta-data to be associated with the root.</p>
4549
4550<h5>Semantics:</h5>
4551
Chris Lattnera7d94ba2008-04-24 05:59:56 +0000[diff] [blame]4552<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]4553location. At compile-time, the code generator generates information to allow
Gordon Henriksen40393542007-12-25 02:31:26 +0000[diff] [blame]4554the runtime to find the pointer at GC safe points. The '<tt>llvm.gcroot</tt>'
4555intrinsic may only be used in a function which <a href="#gc">specifies a GC
4556algorithm</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4557
4558</div>
4559
4560
4561<!-- _______________________________________________________________________ -->
4562<div class="doc_subsubsection">
4563 <a name="int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a>
4564</div>
4565
4566<div class="doc_text">
4567
4568<h5>Syntax:</h5>
4569
4570<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4571 declare i8* @llvm.gcread(i8* %ObjPtr, i8** %Ptr)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4572</pre>
4573
4574<h5>Overview:</h5>
4575
4576<p>The '<tt>llvm.gcread</tt>' intrinsic identifies reads of references from heap
4577locations, allowing garbage collector implementations that require read
4578barriers.</p>
4579
4580<h5>Arguments:</h5>
4581
4582<p>The second argument is the address to read from, which should be an address
4583allocated from the garbage collector. The first object is a pointer to the
4584start of the referenced object, if needed by the language runtime (otherwise
4585null).</p>
4586
4587<h5>Semantics:</h5>
4588
4589<p>The '<tt>llvm.gcread</tt>' intrinsic has the same semantics as a load
4590instruction, but may be replaced with substantially more complex code by the
Gordon Henriksen40393542007-12-25 02:31:26 +0000[diff] [blame]4591garbage collector runtime, as needed. The '<tt>llvm.gcread</tt>' intrinsic
4592may only be used in a function which <a href="#gc">specifies a GC
4593algorithm</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4594
4595</div>
4596
4597
4598<!-- _______________________________________________________________________ -->
4599<div class="doc_subsubsection">
4600 <a name="int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a>
4601</div>
4602
4603<div class="doc_text">
4604
4605<h5>Syntax:</h5>
4606
4607<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4608 declare void @llvm.gcwrite(i8* %P1, i8* %Obj, i8** %P2)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4609</pre>
4610
4611<h5>Overview:</h5>
4612
4613<p>The '<tt>llvm.gcwrite</tt>' intrinsic identifies writes of references to heap
4614locations, allowing garbage collector implementations that require write
4615barriers (such as generational or reference counting collectors).</p>
4616
4617<h5>Arguments:</h5>
4618
4619<p>The first argument is the reference to store, the second is the start of the
4620object to store it to, and the third is the address of the field of Obj to
4621store to. If the runtime does not require a pointer to the object, Obj may be
4622null.</p>
4623
4624<h5>Semantics:</h5>
4625
4626<p>The '<tt>llvm.gcwrite</tt>' intrinsic has the same semantics as a store
4627instruction, but may be replaced with substantially more complex code by the
Gordon Henriksen40393542007-12-25 02:31:26 +0000[diff] [blame]4628garbage collector runtime, as needed. The '<tt>llvm.gcwrite</tt>' intrinsic
4629may only be used in a function which <a href="#gc">specifies a GC
4630algorithm</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4631
4632</div>
4633
4634
4635
4636<!-- ======================================================================= -->
4637<div class="doc_subsection">
4638 <a name="int_codegen">Code Generator Intrinsics</a>
4639</div>
4640
4641<div class="doc_text">
4642<p>
4643These intrinsics are provided by LLVM to expose special features that may only
4644be implemented with code generator support.
4645</p>
4646
4647</div>
4648
4649<!-- _______________________________________________________________________ -->
4650<div class="doc_subsubsection">
4651 <a name="int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a>
4652</div>
4653
4654<div class="doc_text">
4655
4656<h5>Syntax:</h5>
4657<pre>
4658 declare i8 *@llvm.returnaddress(i32 &lt;level&gt;)
4659</pre>
4660
4661<h5>Overview:</h5>
4662
4663<p>
4664The '<tt>llvm.returnaddress</tt>' intrinsic attempts to compute a
4665target-specific value indicating the return address of the current function
4666or one of its callers.
4667</p>
4668
4669<h5>Arguments:</h5>
4670
4671<p>
4672The argument to this intrinsic indicates which function to return the address
4673for. Zero indicates the calling function, one indicates its caller, etc. The
4674argument is <b>required</b> to be a constant integer value.
4675</p>
4676
4677<h5>Semantics:</h5>
4678
4679<p>
4680The '<tt>llvm.returnaddress</tt>' intrinsic either returns a pointer indicating
4681the return address of the specified call frame, or zero if it cannot be
4682identified. The value returned by this intrinsic is likely to be incorrect or 0
4683for arguments other than zero, so it should only be used for debugging purposes.
4684</p>
4685
4686<p>
4687Note that calling this intrinsic does not prevent function inlining or other
4688aggressive transformations, so the value returned may not be that of the obvious
4689source-language caller.
4690</p>
4691</div>
4692
4693
4694<!-- _______________________________________________________________________ -->
4695<div class="doc_subsubsection">
4696 <a name="int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a>
4697</div>
4698
4699<div class="doc_text">
4700
4701<h5>Syntax:</h5>
4702<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4703 declare i8 *@llvm.frameaddress(i32 &lt;level&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4704</pre>
4705
4706<h5>Overview:</h5>
4707
4708<p>
4709The '<tt>llvm.frameaddress</tt>' intrinsic attempts to return the
4710target-specific frame pointer value for the specified stack frame.
4711</p>
4712
4713<h5>Arguments:</h5>
4714
4715<p>
4716The argument to this intrinsic indicates which function to return the frame
4717pointer for. Zero indicates the calling function, one indicates its caller,
4718etc. The argument is <b>required</b> to be a constant integer value.
4719</p>
4720
4721<h5>Semantics:</h5>
4722
4723<p>
4724The '<tt>llvm.frameaddress</tt>' intrinsic either returns a pointer indicating
4725the frame address of the specified call frame, or zero if it cannot be
4726identified. The value returned by this intrinsic is likely to be incorrect or 0
4727for arguments other than zero, so it should only be used for debugging purposes.
4728</p>
4729
4730<p>
4731Note that calling this intrinsic does not prevent function inlining or other
4732aggressive transformations, so the value returned may not be that of the obvious
4733source-language caller.
4734</p>
4735</div>
4736
4737<!-- _______________________________________________________________________ -->
4738<div class="doc_subsubsection">
4739 <a name="int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a>
4740</div>
4741
4742<div class="doc_text">
4743
4744<h5>Syntax:</h5>
4745<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4746 declare i8 *@llvm.stacksave()
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4747</pre>
4748
4749<h5>Overview:</h5>
4750
4751<p>
4752The '<tt>llvm.stacksave</tt>' intrinsic is used to remember the current state of
4753the function stack, for use with <a href="#int_stackrestore">
4754<tt>llvm.stackrestore</tt></a>. This is useful for implementing language
4755features like scoped automatic variable sized arrays in C99.
4756</p>
4757
4758<h5>Semantics:</h5>
4759
4760<p>
4761This intrinsic returns a opaque pointer value that can be passed to <a
4762href="#int_stackrestore"><tt>llvm.stackrestore</tt></a>. When an
4763<tt>llvm.stackrestore</tt> intrinsic is executed with a value saved from
4764<tt>llvm.stacksave</tt>, it effectively restores the state of the stack to the
4765state it was in when the <tt>llvm.stacksave</tt> intrinsic executed. In
4766practice, this pops any <a href="#i_alloca">alloca</a> blocks from the stack
4767that were allocated after the <tt>llvm.stacksave</tt> was executed.
4768</p>
4769
4770</div>
4771
4772<!-- _______________________________________________________________________ -->
4773<div class="doc_subsubsection">
4774 <a name="int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a>
4775</div>
4776
4777<div class="doc_text">
4778
4779<h5>Syntax:</h5>
4780<pre>
4781 declare void @llvm.stackrestore(i8 * %ptr)
4782</pre>
4783
4784<h5>Overview:</h5>
4785
4786<p>
4787The '<tt>llvm.stackrestore</tt>' intrinsic is used to restore the state of
4788the function stack to the state it was in when the corresponding <a
4789href="#int_stacksave"><tt>llvm.stacksave</tt></a> intrinsic executed. This is
4790useful for implementing language features like scoped automatic variable sized
4791arrays in C99.
4792</p>
4793
4794<h5>Semantics:</h5>
4795
4796<p>
4797See the description for <a href="#int_stacksave"><tt>llvm.stacksave</tt></a>.
4798</p>
4799
4800</div>
4801
4802
4803<!-- _______________________________________________________________________ -->
4804<div class="doc_subsubsection">
4805 <a name="int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a>
4806</div>
4807
4808<div class="doc_text">
4809
4810<h5>Syntax:</h5>
4811<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4812 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]4813</pre>
4814
4815<h5>Overview:</h5>
4816
4817
4818<p>
4819The '<tt>llvm.prefetch</tt>' intrinsic is a hint to the code generator to insert
4820a prefetch instruction if supported; otherwise, it is a noop. Prefetches have
4821no
4822effect on the behavior of the program but can change its performance
4823characteristics.
4824</p>
4825
4826<h5>Arguments:</h5>
4827
4828<p>
4829<tt>address</tt> is the address to be prefetched, <tt>rw</tt> is the specifier
4830determining if the fetch should be for a read (0) or write (1), and
4831<tt>locality</tt> is a temporal locality specifier ranging from (0) - no
4832locality, to (3) - extremely local keep in cache. The <tt>rw</tt> and
4833<tt>locality</tt> arguments must be constant integers.
4834</p>
4835
4836<h5>Semantics:</h5>
4837
4838<p>
4839This intrinsic does not modify the behavior of the program. In particular,
4840prefetches cannot trap and do not produce a value. On targets that support this
4841intrinsic, the prefetch can provide hints to the processor cache for better
4842performance.
4843</p>
4844
4845</div>
4846
4847<!-- _______________________________________________________________________ -->
4848<div class="doc_subsubsection">
4849 <a name="int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a>
4850</div>
4851
4852<div class="doc_text">
4853
4854<h5>Syntax:</h5>
4855<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4856 declare void @llvm.pcmarker(i32 &lt;id&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4857</pre>
4858
4859<h5>Overview:</h5>
4860
4861
4862<p>
4863The '<tt>llvm.pcmarker</tt>' intrinsic is a method to export a Program Counter
4864(PC) in a region of
4865code to simulators and other tools. The method is target specific, but it is
4866expected that the marker will use exported symbols to transmit the PC of the marker.
4867The marker makes no guarantees that it will remain with any specific instruction
4868after optimizations. It is possible that the presence of a marker will inhibit
4869optimizations. The intended use is to be inserted after optimizations to allow
4870correlations of simulation runs.
4871</p>
4872
4873<h5>Arguments:</h5>
4874
4875<p>
4876<tt>id</tt> is a numerical id identifying the marker.
4877</p>
4878
4879<h5>Semantics:</h5>
4880
4881<p>
4882This intrinsic does not modify the behavior of the program. Backends that do not
4883support this intrinisic may ignore it.
4884</p>
4885
4886</div>
4887
4888<!-- _______________________________________________________________________ -->
4889<div class="doc_subsubsection">
4890 <a name="int_readcyclecounter">'<tt>llvm.readcyclecounter</tt>' Intrinsic</a>
4891</div>
4892
4893<div class="doc_text">
4894
4895<h5>Syntax:</h5>
4896<pre>
4897 declare i64 @llvm.readcyclecounter( )
4898</pre>
4899
4900<h5>Overview:</h5>
4901
4902
4903<p>
4904The '<tt>llvm.readcyclecounter</tt>' intrinsic provides access to the cycle
4905counter register (or similar low latency, high accuracy clocks) on those targets
4906that support it. On X86, it should map to RDTSC. On Alpha, it should map to RPCC.
4907As the backing counters overflow quickly (on the order of 9 seconds on alpha), this
4908should only be used for small timings.
4909</p>
4910
4911<h5>Semantics:</h5>
4912
4913<p>
4914When directly supported, reading the cycle counter should not modify any memory.
4915Implementations are allowed to either return a application specific value or a
4916system wide value. On backends without support, this is lowered to a constant 0.
4917</p>
4918
4919</div>
4920
4921<!-- ======================================================================= -->
4922<div class="doc_subsection">
4923 <a name="int_libc">Standard C Library Intrinsics</a>
4924</div>
4925
4926<div class="doc_text">
4927<p>
4928LLVM provides intrinsics for a few important standard C library functions.
4929These intrinsics allow source-language front-ends to pass information about the
4930alignment of the pointer arguments to the code generator, providing opportunity
4931for more efficient code generation.
4932</p>
4933
4934</div>
4935
4936<!-- _______________________________________________________________________ -->
4937<div class="doc_subsubsection">
4938 <a name="int_memcpy">'<tt>llvm.memcpy</tt>' Intrinsic</a>
4939</div>
4940
4941<div class="doc_text">
4942
4943<h5>Syntax:</h5>
4944<pre>
4945 declare void @llvm.memcpy.i32(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
4946 i32 &lt;len&gt;, i32 &lt;align&gt;)
4947 declare void @llvm.memcpy.i64(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
4948 i64 &lt;len&gt;, i32 &lt;align&gt;)
4949</pre>
4950
4951<h5>Overview:</h5>
4952
4953<p>
4954The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the source
4955location to the destination location.
4956</p>
4957
4958<p>
4959Note that, unlike the standard libc function, the <tt>llvm.memcpy.*</tt>
4960intrinsics do not return a value, and takes an extra alignment argument.
4961</p>
4962
4963<h5>Arguments:</h5>
4964
4965<p>
4966The first argument is a pointer to the destination, the second is a pointer to
4967the source. The third argument is an integer argument
4968specifying the number of bytes to copy, and the fourth argument is the alignment
4969of the source and destination locations.
4970</p>
4971
4972<p>
4973If the call to this intrinisic has an alignment value that is not 0 or 1, then
4974the caller guarantees that both the source and destination pointers are aligned
4975to that boundary.
4976</p>
4977
4978<h5>Semantics:</h5>
4979
4980<p>
4981The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the source
4982location to the destination location, which are not allowed to overlap. It
4983copies "len" bytes of memory over. If the argument is known to be aligned to
4984some boundary, this can be specified as the fourth argument, otherwise it should
4985be set to 0 or 1.
4986</p>
4987</div>
4988
4989
4990<!-- _______________________________________________________________________ -->
4991<div class="doc_subsubsection">
4992 <a name="int_memmove">'<tt>llvm.memmove</tt>' Intrinsic</a>
4993</div>
4994
4995<div class="doc_text">
4996
4997<h5>Syntax:</h5>
4998<pre>
4999 declare void @llvm.memmove.i32(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5000 i32 &lt;len&gt;, i32 &lt;align&gt;)
5001 declare void @llvm.memmove.i64(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5002 i64 &lt;len&gt;, i32 &lt;align&gt;)
5003</pre>
5004
5005<h5>Overview:</h5>
5006
5007<p>
5008The '<tt>llvm.memmove.*</tt>' intrinsics move a block of memory from the source
5009location to the destination location. It is similar to the
Chris Lattnerdba16ea2008-01-06 19:51:52 +0000[diff] [blame]5010'<tt>llvm.memcpy</tt>' intrinsic but allows the two memory locations to overlap.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5011</p>
5012
5013<p>
5014Note that, unlike the standard libc function, the <tt>llvm.memmove.*</tt>
5015intrinsics do not return a value, and takes an extra alignment argument.
5016</p>
5017
5018<h5>Arguments:</h5>
5019
5020<p>
5021The first argument is a pointer to the destination, the second is a pointer to
5022the source. The third argument is an integer argument
5023specifying the number of bytes to copy, and the fourth argument is the alignment
5024of the source and destination locations.
5025</p>
5026
5027<p>
5028If the call to this intrinisic has an alignment value that is not 0 or 1, then
5029the caller guarantees that the source and destination pointers are aligned to
5030that boundary.
5031</p>
5032
5033<h5>Semantics:</h5>
5034
5035<p>
5036The '<tt>llvm.memmove.*</tt>' intrinsics copy a block of memory from the source
5037location to the destination location, which may overlap. It
5038copies "len" bytes of memory over. If the argument is known to be aligned to
5039some boundary, this can be specified as the fourth argument, otherwise it should
5040be set to 0 or 1.
5041</p>
5042</div>
5043
5044
5045<!-- _______________________________________________________________________ -->
5046<div class="doc_subsubsection">
5047 <a name="int_memset">'<tt>llvm.memset.*</tt>' Intrinsics</a>
5048</div>
5049
5050<div class="doc_text">
5051
5052<h5>Syntax:</h5>
5053<pre>
5054 declare void @llvm.memset.i32(i8 * &lt;dest&gt;, i8 &lt;val&gt;,
5055 i32 &lt;len&gt;, i32 &lt;align&gt;)
5056 declare void @llvm.memset.i64(i8 * &lt;dest&gt;, i8 &lt;val&gt;,
5057 i64 &lt;len&gt;, i32 &lt;align&gt;)
5058</pre>
5059
5060<h5>Overview:</h5>
5061
5062<p>
5063The '<tt>llvm.memset.*</tt>' intrinsics fill a block of memory with a particular
5064byte value.
5065</p>
5066
5067<p>
5068Note that, unlike the standard libc function, the <tt>llvm.memset</tt> intrinsic
5069does not return a value, and takes an extra alignment argument.
5070</p>
5071
5072<h5>Arguments:</h5>
5073
5074<p>
5075The first argument is a pointer to the destination to fill, the second is the
5076byte value to fill it with, the third argument is an integer
5077argument specifying the number of bytes to fill, and the fourth argument is the
5078known alignment of destination location.
5079</p>
5080
5081<p>
5082If the call to this intrinisic has an alignment value that is not 0 or 1, then
5083the caller guarantees that the destination pointer is aligned to that boundary.
5084</p>
5085
5086<h5>Semantics:</h5>
5087
5088<p>
5089The '<tt>llvm.memset.*</tt>' intrinsics fill "len" bytes of memory starting at
5090the
5091destination location. If the argument is known to be aligned to some boundary,
5092this can be specified as the fourth argument, otherwise it should be set to 0 or
50931.
5094</p>
5095</div>
5096
5097
5098<!-- _______________________________________________________________________ -->
5099<div class="doc_subsubsection">
5100 <a name="int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a>
5101</div>
5102
5103<div class="doc_text">
5104
5105<h5>Syntax:</h5>
Dale Johannesenf9adbb62007-10-02 17:47:38 +0000[diff] [blame]5106<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]5107floating point or vector of floating point type. Not all targets support all
5108types however.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5109<pre>
Dale Johannesenf9adbb62007-10-02 17:47:38 +0000[diff] [blame]5110 declare float @llvm.sqrt.f32(float %Val)
5111 declare double @llvm.sqrt.f64(double %Val)
5112 declare x86_fp80 @llvm.sqrt.f80(x86_fp80 %Val)
5113 declare fp128 @llvm.sqrt.f128(fp128 %Val)
5114 declare ppc_fp128 @llvm.sqrt.ppcf128(ppc_fp128 %Val)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5115</pre>
5116
5117<h5>Overview:</h5>
5118
5119<p>
5120The '<tt>llvm.sqrt</tt>' intrinsics return the sqrt of the specified operand,
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5121returning the same value as the libm '<tt>sqrt</tt>' functions would. Unlike
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5122<tt>sqrt</tt> in libm, however, <tt>llvm.sqrt</tt> has undefined behavior for
Chris Lattnerf914a002008-01-29 07:00:44 +0000[diff] [blame]5123negative numbers other than -0.0 (which allows for better optimization, because
5124there is no need to worry about errno being set). <tt>llvm.sqrt(-0.0)</tt> is
5125defined to return -0.0 like IEEE sqrt.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5126</p>
5127
5128<h5>Arguments:</h5>
5129
5130<p>
5131The argument and return value are floating point numbers of the same type.
5132</p>
5133
5134<h5>Semantics:</h5>
5135
5136<p>
5137This function returns the sqrt of the specified operand if it is a nonnegative
5138floating point number.
5139</p>
5140</div>
5141
5142<!-- _______________________________________________________________________ -->
5143<div class="doc_subsubsection">
5144 <a name="int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a>
5145</div>
5146
5147<div class="doc_text">
5148
5149<h5>Syntax:</h5>
Dale Johannesenf9adbb62007-10-02 17:47:38 +0000[diff] [blame]5150<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]5151floating point or vector of floating point type. Not all targets support all
5152types however.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5153<pre>
Dale Johannesenf9adbb62007-10-02 17:47:38 +0000[diff] [blame]5154 declare float @llvm.powi.f32(float %Val, i32 %power)
5155 declare double @llvm.powi.f64(double %Val, i32 %power)
5156 declare x86_fp80 @llvm.powi.f80(x86_fp80 %Val, i32 %power)
5157 declare fp128 @llvm.powi.f128(fp128 %Val, i32 %power)
5158 declare ppc_fp128 @llvm.powi.ppcf128(ppc_fp128 %Val, i32 %power)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5159</pre>
5160
5161<h5>Overview:</h5>
5162
5163<p>
5164The '<tt>llvm.powi.*</tt>' intrinsics return the first operand raised to the
5165specified (positive or negative) power. The order of evaluation of
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5166multiplications is not defined. When a vector of floating point type is
5167used, the second argument remains a scalar integer value.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5168</p>
5169
5170<h5>Arguments:</h5>
5171
5172<p>
5173The second argument is an integer power, and the first is a value to raise to
5174that power.
5175</p>
5176
5177<h5>Semantics:</h5>
5178
5179<p>
5180This function returns the first value raised to the second power with an
5181unspecified sequence of rounding operations.</p>
5182</div>
5183
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5184<!-- _______________________________________________________________________ -->
5185<div class="doc_subsubsection">
5186 <a name="int_sin">'<tt>llvm.sin.*</tt>' Intrinsic</a>
5187</div>
5188
5189<div class="doc_text">
5190
5191<h5>Syntax:</h5>
5192<p>This is an overloaded intrinsic. You can use <tt>llvm.sin</tt> on any
5193floating point or vector of floating point type. Not all targets support all
5194types however.
5195<pre>
5196 declare float @llvm.sin.f32(float %Val)
5197 declare double @llvm.sin.f64(double %Val)
5198 declare x86_fp80 @llvm.sin.f80(x86_fp80 %Val)
5199 declare fp128 @llvm.sin.f128(fp128 %Val)
5200 declare ppc_fp128 @llvm.sin.ppcf128(ppc_fp128 %Val)
5201</pre>
5202
5203<h5>Overview:</h5>
5204
5205<p>
5206The '<tt>llvm.sin.*</tt>' intrinsics return the sine of the operand.
5207</p>
5208
5209<h5>Arguments:</h5>
5210
5211<p>
5212The argument and return value are floating point numbers of the same type.
5213</p>
5214
5215<h5>Semantics:</h5>
5216
5217<p>
5218This function returns the sine of the specified operand, returning the
5219same values as the libm <tt>sin</tt> functions would, and handles error
Dan Gohmaneaba92e2007-10-17 18:05:13 +0000[diff] [blame]5220conditions in the same way.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5221</div>
5222
5223<!-- _______________________________________________________________________ -->
5224<div class="doc_subsubsection">
5225 <a name="int_cos">'<tt>llvm.cos.*</tt>' Intrinsic</a>
5226</div>
5227
5228<div class="doc_text">
5229
5230<h5>Syntax:</h5>
5231<p>This is an overloaded intrinsic. You can use <tt>llvm.cos</tt> on any
5232floating point or vector of floating point type. Not all targets support all
5233types however.
5234<pre>
5235 declare float @llvm.cos.f32(float %Val)
5236 declare double @llvm.cos.f64(double %Val)
5237 declare x86_fp80 @llvm.cos.f80(x86_fp80 %Val)
5238 declare fp128 @llvm.cos.f128(fp128 %Val)
5239 declare ppc_fp128 @llvm.cos.ppcf128(ppc_fp128 %Val)
5240</pre>
5241
5242<h5>Overview:</h5>
5243
5244<p>
5245The '<tt>llvm.cos.*</tt>' intrinsics return the cosine of the operand.
5246</p>
5247
5248<h5>Arguments:</h5>
5249
5250<p>
5251The argument and return value are floating point numbers of the same type.
5252</p>
5253
5254<h5>Semantics:</h5>
5255
5256<p>
5257This function returns the cosine of the specified operand, returning the
5258same values as the libm <tt>cos</tt> functions would, and handles error
Dan Gohmaneaba92e2007-10-17 18:05:13 +0000[diff] [blame]5259conditions in the same way.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5260</div>
5261
5262<!-- _______________________________________________________________________ -->
5263<div class="doc_subsubsection">
5264 <a name="int_pow">'<tt>llvm.pow.*</tt>' Intrinsic</a>
5265</div>
5266
5267<div class="doc_text">
5268
5269<h5>Syntax:</h5>
5270<p>This is an overloaded intrinsic. You can use <tt>llvm.pow</tt> on any
5271floating point or vector of floating point type. Not all targets support all
5272types however.
5273<pre>
5274 declare float @llvm.pow.f32(float %Val, float %Power)
5275 declare double @llvm.pow.f64(double %Val, double %Power)
5276 declare x86_fp80 @llvm.pow.f80(x86_fp80 %Val, x86_fp80 %Power)
5277 declare fp128 @llvm.pow.f128(fp128 %Val, fp128 %Power)
5278 declare ppc_fp128 @llvm.pow.ppcf128(ppc_fp128 %Val, ppc_fp128 Power)
5279</pre>
5280
5281<h5>Overview:</h5>
5282
5283<p>
5284The '<tt>llvm.pow.*</tt>' intrinsics return the first operand raised to the
5285specified (positive or negative) power.
5286</p>
5287
5288<h5>Arguments:</h5>
5289
5290<p>
5291The second argument is a floating point power, and the first is a value to
5292raise to that power.
5293</p>
5294
5295<h5>Semantics:</h5>
5296
5297<p>
5298This function returns the first value raised to the second power,
5299returning the
5300same values as the libm <tt>pow</tt> functions would, and handles error
Dan Gohmaneaba92e2007-10-17 18:05:13 +0000[diff] [blame]5301conditions in the same way.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5302</div>
5303
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5304
5305<!-- ======================================================================= -->
5306<div class="doc_subsection">
5307 <a name="int_manip">Bit Manipulation Intrinsics</a>
5308</div>
5309
5310<div class="doc_text">
5311<p>
5312LLVM provides intrinsics for a few important bit manipulation operations.
5313These allow efficient code generation for some algorithms.
5314</p>
5315
5316</div>
5317
5318<!-- _______________________________________________________________________ -->
5319<div class="doc_subsubsection">
5320 <a name="int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a>
5321</div>
5322
5323<div class="doc_text">
5324
5325<h5>Syntax:</h5>
5326<p>This is an overloaded intrinsic function. You can use bswap on any integer
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5327type that is an even number of bytes (i.e. BitWidth % 16 == 0).
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5328<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5329 declare i16 @llvm.bswap.i16(i16 &lt;id&gt;)
5330 declare i32 @llvm.bswap.i32(i32 &lt;id&gt;)
5331 declare i64 @llvm.bswap.i64(i64 &lt;id&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5332</pre>
5333
5334<h5>Overview:</h5>
5335
5336<p>
5337The '<tt>llvm.bswap</tt>' family of intrinsics is used to byte swap integer
5338values with an even number of bytes (positive multiple of 16 bits). These are
5339useful for performing operations on data that is not in the target's native
5340byte order.
5341</p>
5342
5343<h5>Semantics:</h5>
5344
5345<p>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5346The <tt>llvm.bswap.i16</tt> intrinsic returns an i16 value that has the high
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5347and low byte of the input i16 swapped. Similarly, the <tt>llvm.bswap.i32</tt>
5348intrinsic returns an i32 value that has the four bytes of the input i32
5349swapped, 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]5350i32 will have its bytes in 3, 2, 1, 0 order. The <tt>llvm.bswap.i48</tt>,
5351<tt>llvm.bswap.i64</tt> and other intrinsics extend this concept to
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5352additional even-byte lengths (6 bytes, 8 bytes and more, respectively).
5353</p>
5354
5355</div>
5356
5357<!-- _______________________________________________________________________ -->
5358<div class="doc_subsubsection">
5359 <a name="int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic</a>
5360</div>
5361
5362<div class="doc_text">
5363
5364<h5>Syntax:</h5>
5365<p>This is an overloaded intrinsic. You can use llvm.ctpop on any integer bit
5366width. Not all targets support all bit widths however.
5367<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5368 declare i8 @llvm.ctpop.i8 (i8 &lt;src&gt;)
5369 declare i16 @llvm.ctpop.i16(i16 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5370 declare i32 @llvm.ctpop.i32(i32 &lt;src&gt;)
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5371 declare i64 @llvm.ctpop.i64(i64 &lt;src&gt;)
5372 declare i256 @llvm.ctpop.i256(i256 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5373</pre>
5374
5375<h5>Overview:</h5>
5376
5377<p>
5378The '<tt>llvm.ctpop</tt>' family of intrinsics counts the number of bits set in a
5379value.
5380</p>
5381
5382<h5>Arguments:</h5>
5383
5384<p>
5385The only argument is the value to be counted. The argument may be of any
5386integer type. The return type must match the argument type.
5387</p>
5388
5389<h5>Semantics:</h5>
5390
5391<p>
5392The '<tt>llvm.ctpop</tt>' intrinsic counts the 1's in a variable.
5393</p>
5394</div>
5395
5396<!-- _______________________________________________________________________ -->
5397<div class="doc_subsubsection">
5398 <a name="int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic</a>
5399</div>
5400
5401<div class="doc_text">
5402
5403<h5>Syntax:</h5>
5404<p>This is an overloaded intrinsic. You can use <tt>llvm.ctlz</tt> on any
5405integer bit width. Not all targets support all bit widths however.
5406<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5407 declare i8 @llvm.ctlz.i8 (i8 &lt;src&gt;)
5408 declare i16 @llvm.ctlz.i16(i16 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5409 declare i32 @llvm.ctlz.i32(i32 &lt;src&gt;)
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5410 declare i64 @llvm.ctlz.i64(i64 &lt;src&gt;)
5411 declare i256 @llvm.ctlz.i256(i256 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5412</pre>
5413
5414<h5>Overview:</h5>
5415
5416<p>
5417The '<tt>llvm.ctlz</tt>' family of intrinsic functions counts the number of
5418leading zeros in a variable.
5419</p>
5420
5421<h5>Arguments:</h5>
5422
5423<p>
5424The only argument is the value to be counted. The argument may be of any
5425integer type. The return type must match the argument type.
5426</p>
5427
5428<h5>Semantics:</h5>
5429
5430<p>
5431The '<tt>llvm.ctlz</tt>' intrinsic counts the leading (most significant) zeros
5432in a variable. If the src == 0 then the result is the size in bits of the type
5433of src. For example, <tt>llvm.ctlz(i32 2) = 30</tt>.
5434</p>
5435</div>
5436
5437
5438
5439<!-- _______________________________________________________________________ -->
5440<div class="doc_subsubsection">
5441 <a name="int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic</a>
5442</div>
5443
5444<div class="doc_text">
5445
5446<h5>Syntax:</h5>
5447<p>This is an overloaded intrinsic. You can use <tt>llvm.cttz</tt> on any
5448integer bit width. Not all targets support all bit widths however.
5449<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5450 declare i8 @llvm.cttz.i8 (i8 &lt;src&gt;)
5451 declare i16 @llvm.cttz.i16(i16 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5452 declare i32 @llvm.cttz.i32(i32 &lt;src&gt;)
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5453 declare i64 @llvm.cttz.i64(i64 &lt;src&gt;)
5454 declare i256 @llvm.cttz.i256(i256 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5455</pre>
5456
5457<h5>Overview:</h5>
5458
5459<p>
5460The '<tt>llvm.cttz</tt>' family of intrinsic functions counts the number of
5461trailing zeros.
5462</p>
5463
5464<h5>Arguments:</h5>
5465
5466<p>
5467The only argument is the value to be counted. The argument may be of any
5468integer type. The return type must match the argument type.
5469</p>
5470
5471<h5>Semantics:</h5>
5472
5473<p>
5474The '<tt>llvm.cttz</tt>' intrinsic counts the trailing (least significant) zeros
5475in a variable. If the src == 0 then the result is the size in bits of the type
5476of src. For example, <tt>llvm.cttz(2) = 1</tt>.
5477</p>
5478</div>
5479
5480<!-- _______________________________________________________________________ -->
5481<div class="doc_subsubsection">
5482 <a name="int_part_select">'<tt>llvm.part.select.*</tt>' Intrinsic</a>
5483</div>
5484
5485<div class="doc_text">
5486
5487<h5>Syntax:</h5>
5488<p>This is an overloaded intrinsic. You can use <tt>llvm.part.select</tt>
5489on any integer bit width.
5490<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5491 declare i17 @llvm.part.select.i17 (i17 %val, i32 %loBit, i32 %hiBit)
5492 declare i29 @llvm.part.select.i29 (i29 %val, i32 %loBit, i32 %hiBit)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5493</pre>
5494
5495<h5>Overview:</h5>
5496<p>The '<tt>llvm.part.select</tt>' family of intrinsic functions selects a
5497range of bits from an integer value and returns them in the same bit width as
5498the original value.</p>
5499
5500<h5>Arguments:</h5>
5501<p>The first argument, <tt>%val</tt> and the result may be integer types of
5502any bit width but they must have the same bit width. The second and third
5503arguments must be <tt>i32</tt> type since they specify only a bit index.</p>
5504
5505<h5>Semantics:</h5>
5506<p>The operation of the '<tt>llvm.part.select</tt>' intrinsic has two modes
5507of operation: forwards and reverse. If <tt>%loBit</tt> is greater than
5508<tt>%hiBits</tt> then the intrinsic operates in reverse mode. Otherwise it
5509operates in forward mode.</p>
5510<p>In forward mode, this intrinsic is the equivalent of shifting <tt>%val</tt>
5511right by <tt>%loBit</tt> bits and then ANDing it with a mask with
5512only the <tt>%hiBit - %loBit</tt> bits set, as follows:</p>
5513<ol>
5514 <li>The <tt>%val</tt> is shifted right (LSHR) by the number of bits specified
5515 by <tt>%loBits</tt>. This normalizes the value to the low order bits.</li>
5516 <li>The <tt>%loBits</tt> value is subtracted from the <tt>%hiBits</tt> value
5517 to determine the number of bits to retain.</li>
5518 <li>A mask of the retained bits is created by shifting a -1 value.</li>
5519 <li>The mask is ANDed with <tt>%val</tt> to produce the result.
5520</ol>
5521<p>In reverse mode, a similar computation is made except that the bits are
5522returned in the reverse order. So, for example, if <tt>X</tt> has the value
5523<tt>i16 0x0ACF (101011001111)</tt> and we apply
5524<tt>part.select(i16 X, 8, 3)</tt> to it, we get back the value
5525<tt>i16 0x0026 (000000100110)</tt>.</p>
5526</div>
5527
5528<div class="doc_subsubsection">
5529 <a name="int_part_set">'<tt>llvm.part.set.*</tt>' Intrinsic</a>
5530</div>
5531
5532<div class="doc_text">
5533
5534<h5>Syntax:</h5>
5535<p>This is an overloaded intrinsic. You can use <tt>llvm.part.set</tt>
5536on any integer bit width.
5537<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5538 declare i17 @llvm.part.set.i17.i9 (i17 %val, i9 %repl, i32 %lo, i32 %hi)
5539 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]5540</pre>
5541
5542<h5>Overview:</h5>
5543<p>The '<tt>llvm.part.set</tt>' family of intrinsic functions replaces a range
5544of bits in an integer value with another integer value. It returns the integer
5545with the replaced bits.</p>
5546
5547<h5>Arguments:</h5>
5548<p>The first argument, <tt>%val</tt> and the result may be integer types of
5549any bit width but they must have the same bit width. <tt>%val</tt> is the value
5550whose bits will be replaced. The second argument, <tt>%repl</tt> may be an
5551integer of any bit width. The third and fourth arguments must be <tt>i32</tt>
5552type since they specify only a bit index.</p>
5553
5554<h5>Semantics:</h5>
5555<p>The operation of the '<tt>llvm.part.set</tt>' intrinsic has two modes
5556of operation: forwards and reverse. If <tt>%lo</tt> is greater than
5557<tt>%hi</tt> then the intrinsic operates in reverse mode. Otherwise it
5558operates in forward mode.</p>
5559<p>For both modes, the <tt>%repl</tt> value is prepared for use by either
5560truncating it down to the size of the replacement area or zero extending it
5561up to that size.</p>
5562<p>In forward mode, the bits between <tt>%lo</tt> and <tt>%hi</tt> (inclusive)
5563are replaced with corresponding bits from <tt>%repl</tt>. That is the 0th bit
5564in <tt>%repl</tt> replaces the <tt>%lo</tt>th bit in <tt>%val</tt> and etc. up
5565to the <tt>%hi</tt>th bit.
5566<p>In reverse mode, a similar computation is made except that the bits are
5567reversed. That is, the <tt>0</tt>th bit in <tt>%repl</tt> replaces the
5568<tt>%hi</tt> bit in <tt>%val</tt> and etc. down to the <tt>%lo</tt>th bit.
5569<h5>Examples:</h5>
5570<pre>
5571 llvm.part.set(0xFFFF, 0, 4, 7) -&gt; 0xFF0F
5572 llvm.part.set(0xFFFF, 0, 7, 4) -&gt; 0xFF0F
5573 llvm.part.set(0xFFFF, 1, 7, 4) -&gt; 0xFF8F
5574 llvm.part.set(0xFFFF, F, 8, 3) -&gt; 0xFFE7
5575 llvm.part.set(0xFFFF, 0, 3, 8) -&gt; 0xFE07
5576</pre>
5577</div>
5578
5579<!-- ======================================================================= -->
5580<div class="doc_subsection">
5581 <a name="int_debugger">Debugger Intrinsics</a>
5582</div>
5583
5584<div class="doc_text">
5585<p>
5586The LLVM debugger intrinsics (which all start with <tt>llvm.dbg.</tt> prefix),
5587are described in the <a
5588href="SourceLevelDebugging.html#format_common_intrinsics">LLVM Source Level
5589Debugging</a> document.
5590</p>
5591</div>
5592
5593
5594<!-- ======================================================================= -->
5595<div class="doc_subsection">
5596 <a name="int_eh">Exception Handling Intrinsics</a>
5597</div>
5598
5599<div class="doc_text">
5600<p> The LLVM exception handling intrinsics (which all start with
5601<tt>llvm.eh.</tt> prefix), are described in the <a
5602href="ExceptionHandling.html#format_common_intrinsics">LLVM Exception
5603Handling</a> document. </p>
5604</div>
5605
5606<!-- ======================================================================= -->
5607<div class="doc_subsection">
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5608 <a name="int_trampoline">Trampoline Intrinsic</a>
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5609</div>
5610
5611<div class="doc_text">
5612<p>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5613 This intrinsic makes it possible to excise one parameter, marked with
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5614 the <tt>nest</tt> attribute, from a function. The result is a callable
5615 function pointer lacking the nest parameter - the caller does not need
5616 to provide a value for it. Instead, the value to use is stored in
5617 advance in a "trampoline", a block of memory usually allocated
5618 on the stack, which also contains code to splice the nest value into the
5619 argument list. This is used to implement the GCC nested function address
5620 extension.
5621</p>
5622<p>
5623 For example, if the function is
5624 <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]5625 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]5626<pre>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5627 %tramp = alloca [10 x i8], align 4 ; size and alignment only correct for X86
5628 %tramp1 = getelementptr [10 x i8]* %tramp, i32 0, i32 0
5629 %p = call i8* @llvm.init.trampoline( i8* %tramp1, i8* bitcast (i32 (i8* nest , i32, i32)* @f to i8*), i8* %nval )
5630 %fp = bitcast i8* %p to i32 (i32, i32)*
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5631</pre>
Bill Wendlinge262b4c2007-09-22 09:23:55 +0000[diff] [blame]5632 <p>The call <tt>%val = call i32 %fp( i32 %x, i32 %y )</tt> is then equivalent
5633 to <tt>%val = call i32 %f( i8* %nval, i32 %x, i32 %y )</tt>.</p>
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5634</div>
5635
5636<!-- _______________________________________________________________________ -->
5637<div class="doc_subsubsection">
5638 <a name="int_it">'<tt>llvm.init.trampoline</tt>' Intrinsic</a>
5639</div>
5640<div class="doc_text">
5641<h5>Syntax:</h5>
5642<pre>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5643declare 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]5644</pre>
5645<h5>Overview:</h5>
5646<p>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5647 This fills the memory pointed to by <tt>tramp</tt> with code
5648 and returns a function pointer suitable for executing it.
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5649</p>
5650<h5>Arguments:</h5>
5651<p>
5652 The <tt>llvm.init.trampoline</tt> intrinsic takes three arguments, all
5653 pointers. The <tt>tramp</tt> argument must point to a sufficiently large
5654 and sufficiently aligned block of memory; this memory is written to by the
Duncan Sands35012212007-08-22 23:39:54 +0000[diff] [blame]5655 intrinsic. Note that the size and the alignment are target-specific - LLVM
5656 currently provides no portable way of determining them, so a front-end that
5657 generates this intrinsic needs to have some target-specific knowledge.
5658 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]5659</p>
5660<h5>Semantics:</h5>
5661<p>
5662 The block of memory pointed to by <tt>tramp</tt> is filled with target
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5663 dependent code, turning it into a function. A pointer to this function is
5664 returned, but needs to be bitcast to an
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5665 <a href="#int_trampoline">appropriate function pointer type</a>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5666 before being called. The new function's signature is the same as that of
5667 <tt>func</tt> with any arguments marked with the <tt>nest</tt> attribute
5668 removed. At most one such <tt>nest</tt> argument is allowed, and it must be
5669 of pointer type. Calling the new function is equivalent to calling
5670 <tt>func</tt> with the same argument list, but with <tt>nval</tt> used for the
5671 missing <tt>nest</tt> argument. If, after calling
5672 <tt>llvm.init.trampoline</tt>, the memory pointed to by <tt>tramp</tt> is
5673 modified, then the effect of any later call to the returned function pointer is
5674 undefined.
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5675</p>
5676</div>
5677
5678<!-- ======================================================================= -->
5679<div class="doc_subsection">
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]5680 <a name="int_atomics">Atomic Operations and Synchronization Intrinsics</a>
5681</div>
5682
5683<div class="doc_text">
5684<p>
5685 These intrinsic functions expand the "universal IR" of LLVM to represent
5686 hardware constructs for atomic operations and memory synchronization. This
5687 provides an interface to the hardware, not an interface to the programmer. It
5688 is aimed at a low enough level to allow any programming models or APIs which
5689 need atomic behaviors to map cleanly onto it. It is also modeled primarily on
5690 hardware behavior. Just as hardware provides a "universal IR" for source
5691 languages, it also provides a starting point for developing a "universal"
5692 atomic operation and synchronization IR.
5693</p>
5694<p>
5695 These do <em>not</em> form an API such as high-level threading libraries,
5696 software transaction memory systems, atomic primitives, and intrinsic
5697 functions as found in BSD, GNU libc, atomic_ops, APR, and other system and
5698 application libraries. The hardware interface provided by LLVM should allow
5699 a clean implementation of all of these APIs and parallel programming models.
5700 No one model or paradigm should be selected above others unless the hardware
5701 itself ubiquitously does so.
5702
5703</p>
5704</div>
5705
5706<!-- _______________________________________________________________________ -->
5707<div class="doc_subsubsection">
5708 <a name="int_memory_barrier">'<tt>llvm.memory.barrier</tt>' Intrinsic</a>
5709</div>
5710<div class="doc_text">
5711<h5>Syntax:</h5>
5712<pre>
5713declare void @llvm.memory.barrier( i1 &lt;ll&gt;, i1 &lt;ls&gt;, i1 &lt;sl&gt;, i1 &lt;ss&gt;,
5714i1 &lt;device&gt; )
5715
5716</pre>
5717<h5>Overview:</h5>
5718<p>
5719 The <tt>llvm.memory.barrier</tt> intrinsic guarantees ordering between
5720 specific pairs of memory access types.
5721</p>
5722<h5>Arguments:</h5>
5723<p>
5724 The <tt>llvm.memory.barrier</tt> intrinsic requires five boolean arguments.
5725 The first four arguments enables a specific barrier as listed below. The fith
5726 argument specifies that the barrier applies to io or device or uncached memory.
5727
5728</p>
5729 <ul>
5730 <li><tt>ll</tt>: load-load barrier</li>
5731 <li><tt>ls</tt>: load-store barrier</li>
5732 <li><tt>sl</tt>: store-load barrier</li>
5733 <li><tt>ss</tt>: store-store barrier</li>
5734 <li><tt>device</tt>: barrier applies to device and uncached memory also.
5735 </ul>
5736<h5>Semantics:</h5>
5737<p>
5738 This intrinsic causes the system to enforce some ordering constraints upon
5739 the loads and stores of the program. This barrier does not indicate
5740 <em>when</em> any events will occur, it only enforces an <em>order</em> in
5741 which they occur. For any of the specified pairs of load and store operations
5742 (f.ex. load-load, or store-load), all of the first operations preceding the
5743 barrier will complete before any of the second operations succeeding the
5744 barrier begin. Specifically the semantics for each pairing is as follows:
5745</p>
5746 <ul>
5747 <li><tt>ll</tt>: All loads before the barrier must complete before any load
5748 after the barrier begins.</li>
5749
5750 <li><tt>ls</tt>: All loads before the barrier must complete before any
5751 store after the barrier begins.</li>
5752 <li><tt>ss</tt>: All stores before the barrier must complete before any
5753 store after the barrier begins.</li>
5754 <li><tt>sl</tt>: All stores before the barrier must complete before any
5755 load after the barrier begins.</li>
5756 </ul>
5757<p>
5758 These semantics are applied with a logical "and" behavior when more than one
5759 is enabled in a single memory barrier intrinsic.
5760</p>
5761<p>
5762 Backends may implement stronger barriers than those requested when they do not
5763 support as fine grained a barrier as requested. Some architectures do not
5764 need all types of barriers and on such architectures, these become noops.
5765</p>
5766<h5>Example:</h5>
5767<pre>
5768%ptr = malloc i32
5769 store i32 4, %ptr
5770
5771%result1 = load i32* %ptr <i>; yields {i32}:result1 = 4</i>
5772 call void @llvm.memory.barrier( i1 false, i1 true, i1 false, i1 false )
5773 <i>; guarantee the above finishes</i>
5774 store i32 8, %ptr <i>; before this begins</i>
5775</pre>
5776</div>
5777
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]5778<!-- _______________________________________________________________________ -->
5779<div class="doc_subsubsection">
5780 <a name="int_atomic_lcs">'<tt>llvm.atomic.lcs.*</tt>' Intrinsic</a>
5781</div>
5782<div class="doc_text">
5783<h5>Syntax:</h5>
5784<p>
5785 This is an overloaded intrinsic. You can use <tt>llvm.atomic.lcs</tt> on any
5786 integer bit width. Not all targets support all bit widths however.</p>
5787
5788<pre>
5789declare i8 @llvm.atomic.lcs.i8( i8* &lt;ptr&gt;, i8 &lt;cmp&gt;, i8 &lt;val&gt; )
5790declare i16 @llvm.atomic.lcs.i16( i16* &lt;ptr&gt;, i16 &lt;cmp&gt;, i16 &lt;val&gt; )
5791declare i32 @llvm.atomic.lcs.i32( i32* &lt;ptr&gt;, i32 &lt;cmp&gt;, i32 &lt;val&gt; )
5792declare i64 @llvm.atomic.lcs.i64( i64* &lt;ptr&gt;, i64 &lt;cmp&gt;, i64 &lt;val&gt; )
5793
5794</pre>
5795<h5>Overview:</h5>
5796<p>
5797 This loads a value in memory and compares it to a given value. If they are
5798 equal, it stores a new value into the memory.
5799</p>
5800<h5>Arguments:</h5>
5801<p>
5802 The <tt>llvm.atomic.lcs</tt> intrinsic takes three arguments. The result as
5803 well as both <tt>cmp</tt> and <tt>val</tt> must be integer values with the
5804 same bit width. The <tt>ptr</tt> argument must be a pointer to a value of
5805 this integer type. While any bit width integer may be used, targets may only
5806 lower representations they support in hardware.
5807
5808</p>
5809<h5>Semantics:</h5>
5810<p>
5811 This entire intrinsic must be executed atomically. It first loads the value
5812 in memory pointed to by <tt>ptr</tt> and compares it with the value
5813 <tt>cmp</tt>. If they are equal, <tt>val</tt> is stored into the memory. The
5814 loaded value is yielded in all cases. This provides the equivalent of an
5815 atomic compare-and-swap operation within the SSA framework.
5816</p>
5817<h5>Examples:</h5>
5818
5819<pre>
5820%ptr = malloc i32
5821 store i32 4, %ptr
5822
5823%val1 = add i32 4, 4
5824%result1 = call i32 @llvm.atomic.lcs.i32( i32* %ptr, i32 4, %val1 )
5825 <i>; yields {i32}:result1 = 4</i>
5826%stored1 = icmp eq i32 %result1, 4 <i>; yields {i1}:stored1 = true</i>
5827%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 8</i>
5828
5829%val2 = add i32 1, 1
5830%result2 = call i32 @llvm.atomic.lcs.i32( i32* %ptr, i32 5, %val2 )
5831 <i>; yields {i32}:result2 = 8</i>
5832%stored2 = icmp eq i32 %result2, 5 <i>; yields {i1}:stored2 = false</i>
5833
5834%memval2 = load i32* %ptr <i>; yields {i32}:memval2 = 8</i>
5835</pre>
5836</div>
5837
5838<!-- _______________________________________________________________________ -->
5839<div class="doc_subsubsection">
5840 <a name="int_atomic_swap">'<tt>llvm.atomic.swap.*</tt>' Intrinsic</a>
5841</div>
5842<div class="doc_text">
5843<h5>Syntax:</h5>
5844
5845<p>
5846 This is an overloaded intrinsic. You can use <tt>llvm.atomic.swap</tt> on any
5847 integer bit width. Not all targets support all bit widths however.</p>
5848<pre>
5849declare i8 @llvm.atomic.swap.i8( i8* &lt;ptr&gt;, i8 &lt;val&gt; )
5850declare i16 @llvm.atomic.swap.i16( i16* &lt;ptr&gt;, i16 &lt;val&gt; )
5851declare i32 @llvm.atomic.swap.i32( i32* &lt;ptr&gt;, i32 &lt;val&gt; )
5852declare i64 @llvm.atomic.swap.i64( i64* &lt;ptr&gt;, i64 &lt;val&gt; )
5853
5854</pre>
5855<h5>Overview:</h5>
5856<p>
5857 This intrinsic loads the value stored in memory at <tt>ptr</tt> and yields
5858 the value from memory. It then stores the value in <tt>val</tt> in the memory
5859 at <tt>ptr</tt>.
5860</p>
5861<h5>Arguments:</h5>
5862
5863<p>
5864 The <tt>llvm.atomic.ls</tt> intrinsic takes two arguments. Both the
5865 <tt>val</tt> argument and the result must be integers of the same bit width.
5866 The first argument, <tt>ptr</tt>, must be a pointer to a value of this
5867 integer type. The targets may only lower integer representations they
5868 support.
5869</p>
5870<h5>Semantics:</h5>
5871<p>
5872 This intrinsic loads the value pointed to by <tt>ptr</tt>, yields it, and
5873 stores <tt>val</tt> back into <tt>ptr</tt> atomically. This provides the
5874 equivalent of an atomic swap operation within the SSA framework.
5875
5876</p>
5877<h5>Examples:</h5>
5878<pre>
5879%ptr = malloc i32
5880 store i32 4, %ptr
5881
5882%val1 = add i32 4, 4
5883%result1 = call i32 @llvm.atomic.swap.i32( i32* %ptr, i32 %val1 )
5884 <i>; yields {i32}:result1 = 4</i>
5885%stored1 = icmp eq i32 %result1, 4 <i>; yields {i1}:stored1 = true</i>
5886%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 8</i>
5887
5888%val2 = add i32 1, 1
5889%result2 = call i32 @llvm.atomic.swap.i32( i32* %ptr, i32 %val2 )
5890 <i>; yields {i32}:result2 = 8</i>
5891
5892%stored2 = icmp eq i32 %result2, 8 <i>; yields {i1}:stored2 = true</i>
5893%memval2 = load i32* %ptr <i>; yields {i32}:memval2 = 2</i>
5894</pre>
5895</div>
5896
5897<!-- _______________________________________________________________________ -->
5898<div class="doc_subsubsection">
5899 <a name="int_atomic_las">'<tt>llvm.atomic.las.*</tt>' Intrinsic</a>
5900
5901</div>
5902<div class="doc_text">
5903<h5>Syntax:</h5>
5904<p>
5905 This is an overloaded intrinsic. You can use <tt>llvm.atomic.las</tt> on any
5906 integer bit width. Not all targets support all bit widths however.</p>
5907<pre>
5908declare i8 @llvm.atomic.las.i8.( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
5909declare i16 @llvm.atomic.las.i16.( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
5910declare i32 @llvm.atomic.las.i32.( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
5911declare i64 @llvm.atomic.las.i64.( i64* &lt;ptr&gt;, i64 &lt;delta&gt; )
5912
5913</pre>
5914<h5>Overview:</h5>
5915<p>
5916 This intrinsic adds <tt>delta</tt> to the value stored in memory at
5917 <tt>ptr</tt>. It yields the original value at <tt>ptr</tt>.
5918</p>
5919<h5>Arguments:</h5>
5920<p>
5921
5922 The intrinsic takes two arguments, the first a pointer to an integer value
5923 and the second an integer value. The result is also an integer value. These
5924 integer types can have any bit width, but they must all have the same bit
5925 width. The targets may only lower integer representations they support.
5926</p>
5927<h5>Semantics:</h5>
5928<p>
5929 This intrinsic does a series of operations atomically. It first loads the
5930 value stored at <tt>ptr</tt>. It then adds <tt>delta</tt>, stores the result
5931 to <tt>ptr</tt>. It yields the original value stored at <tt>ptr</tt>.
5932</p>
5933
5934<h5>Examples:</h5>
5935<pre>
5936%ptr = malloc i32
5937 store i32 4, %ptr
5938%result1 = call i32 @llvm.atomic.las.i32( i32* %ptr, i32 4 )
5939 <i>; yields {i32}:result1 = 4</i>
5940%result2 = call i32 @llvm.atomic.las.i32( i32* %ptr, i32 2 )
5941 <i>; yields {i32}:result2 = 8</i>
5942%result3 = call i32 @llvm.atomic.las.i32( i32* %ptr, i32 5 )
5943 <i>; yields {i32}:result3 = 10</i>
5944%memval = load i32* %ptr <i>; yields {i32}:memval1 = 15</i>
5945</pre>
5946</div>
5947
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]5948
5949<!-- ======================================================================= -->
5950<div class="doc_subsection">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5951 <a name="int_general">General Intrinsics</a>
5952</div>
5953
5954<div class="doc_text">
5955<p> This class of intrinsics is designed to be generic and has
5956no specific purpose. </p>
5957</div>
5958
5959<!-- _______________________________________________________________________ -->
5960<div class="doc_subsubsection">
5961 <a name="int_var_annotation">'<tt>llvm.var.annotation</tt>' Intrinsic</a>
5962</div>
5963
5964<div class="doc_text">
5965
5966<h5>Syntax:</h5>
5967<pre>
5968 declare void @llvm.var.annotation(i8* &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
5969</pre>
5970
5971<h5>Overview:</h5>
5972
5973<p>
5974The '<tt>llvm.var.annotation</tt>' intrinsic
5975</p>
5976
5977<h5>Arguments:</h5>
5978
5979<p>
5980The first argument is a pointer to a value, the second is a pointer to a
5981global string, the third is a pointer to a global string which is the source
5982file name, and the last argument is the line number.
5983</p>
5984
5985<h5>Semantics:</h5>
5986
5987<p>
Anton Korobeynikove6e764f2008-01-15 22:31:34 +0000[diff] [blame]5988This intrinsic allows annotation of local variables with arbitrary strings.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5989This can be useful for special purpose optimizations that want to look for these
Anton Korobeynikove6e764f2008-01-15 22:31:34 +0000[diff] [blame]5990annotations. These have no other defined use, they are ignored by code
5991generation and optimization.
5992</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5993</div>
5994
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]5995<!-- _______________________________________________________________________ -->
5996<div class="doc_subsubsection">
Tanya Lattnerc9869b12007-09-21 23:57:59 +0000[diff] [blame]5997 <a name="int_annotation">'<tt>llvm.annotation.*</tt>' Intrinsic</a>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]5998</div>
5999
6000<div class="doc_text">
6001
6002<h5>Syntax:</h5>
Tanya Lattnere545be72007-09-21 23:56:27 +0000[diff] [blame]6003<p>This is an overloaded intrinsic. You can use '<tt>llvm.annotation</tt>' on
6004any integer bit width.
6005</p>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]6006<pre>
Tanya Lattner09161fe2007-09-22 00:03:01 +0000[diff] [blame]6007 declare i8 @llvm.annotation.i8(i8 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
6008 declare i16 @llvm.annotation.i16(i16 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
6009 declare i32 @llvm.annotation.i32(i32 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
6010 declare i64 @llvm.annotation.i64(i64 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
6011 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]6012</pre>
6013
6014<h5>Overview:</h5>
Tanya Lattnere545be72007-09-21 23:56:27 +0000[diff] [blame]6015
6016<p>
6017The '<tt>llvm.annotation</tt>' intrinsic.
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]6018</p>
6019
6020<h5>Arguments:</h5>
6021
6022<p>
6023The first argument is an integer value (result of some expression),
6024the second is a pointer to a global string, the third is a pointer to a global
6025string which is the source file name, and the last argument is the line number.
Tanya Lattnere545be72007-09-21 23:56:27 +0000[diff] [blame]6026It returns the value of the first argument.
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]6027</p>
6028
6029<h5>Semantics:</h5>
6030
6031<p>
6032This intrinsic allows annotations to be put on arbitrary expressions
6033with arbitrary strings. This can be useful for special purpose optimizations
6034that want to look for these annotations. These have no other defined use, they
6035are ignored by code generation and optimization.
6036</div>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6037
Anton Korobeynikove6e764f2008-01-15 22:31:34 +0000[diff] [blame]6038<!-- _______________________________________________________________________ -->
6039<div class="doc_subsubsection">
6040 <a name="int_trap">'<tt>llvm.trap</tt>' Intrinsic</a>
6041</div>
6042
6043<div class="doc_text">
6044
6045<h5>Syntax:</h5>
6046<pre>
6047 declare void @llvm.trap()
6048</pre>
6049
6050<h5>Overview:</h5>
6051
6052<p>
6053The '<tt>llvm.trap</tt>' intrinsic
6054</p>
6055
6056<h5>Arguments:</h5>
6057
6058<p>
6059None
6060</p>
6061
6062<h5>Semantics:</h5>
6063
6064<p>
6065This intrinsics is lowered to the target dependent trap instruction. If the
6066target does not have a trap instruction, this intrinsic will be lowered to the
6067call of the abort() function.
6068</p>
6069</div>
6070
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6071<!-- *********************************************************************** -->
6072<hr>
6073<address>
6074 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
6075 src="http://jigsaw.w3.org/css-validator/images/vcss" alt="Valid CSS!"></a>
6076 <a href="http://validator.w3.org/check/referer"><img
Chris Lattner08497ce2008-01-04 04:33:49 +0000[diff] [blame]6077 src="http://www.w3.org/Icons/valid-html401" alt="Valid HTML 4.01!"></a>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6078
6079 <a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
6080 <a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br>
6081 Last modified: $Date$
6082</address>
Chris Lattner08497ce2008-01-04 04:33:49 +0000[diff] [blame]6083
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6084</body>
6085</html>