Gitiles
Code Review Sign In
gerrit-public.fairphone.software / platform / external / llvm / 275e6beb16b75b57d026a5f45b7645bd60ae8bd9 / . / docs / LangRef.html
blob: fa5019370a4680f322c9e754e18f7e304cc2dd61 [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>
114 <li><a href="#memoryops">Memory Access and Addressing Operations</a>
115 <ol>
116 <li><a href="#i_malloc">'<tt>malloc</tt>' Instruction</a></li>
117 <li><a href="#i_free">'<tt>free</tt>' Instruction</a></li>
118 <li><a href="#i_alloca">'<tt>alloca</tt>' Instruction</a></li>
119 <li><a href="#i_load">'<tt>load</tt>' Instruction</a></li>
120 <li><a href="#i_store">'<tt>store</tt>' Instruction</a></li>
121 <li><a href="#i_getelementptr">'<tt>getelementptr</tt>' Instruction</a></li>
122 </ol>
123 </li>
124 <li><a href="#convertops">Conversion Operations</a>
125 <ol>
126 <li><a href="#i_trunc">'<tt>trunc .. to</tt>' Instruction</a></li>
127 <li><a href="#i_zext">'<tt>zext .. to</tt>' Instruction</a></li>
128 <li><a href="#i_sext">'<tt>sext .. to</tt>' Instruction</a></li>
129 <li><a href="#i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a></li>
130 <li><a href="#i_fpext">'<tt>fpext .. to</tt>' Instruction</a></li>
131 <li><a href="#i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a></li>
132 <li><a href="#i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a></li>
133 <li><a href="#i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a></li>
134 <li><a href="#i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a></li>
135 <li><a href="#i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a></li>
136 <li><a href="#i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a></li>
137 <li><a href="#i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a></li>
138 </ol>
139 <li><a href="#otherops">Other Operations</a>
140 <ol>
141 <li><a href="#i_icmp">'<tt>icmp</tt>' Instruction</a></li>
142 <li><a href="#i_fcmp">'<tt>fcmp</tt>' Instruction</a></li>
143 <li><a href="#i_phi">'<tt>phi</tt>' Instruction</a></li>
144 <li><a href="#i_select">'<tt>select</tt>' Instruction</a></li>
145 <li><a href="#i_call">'<tt>call</tt>' Instruction</a></li>
146 <li><a href="#i_va_arg">'<tt>va_arg</tt>' Instruction</a></li>
147 </ol>
148 </li>
149 </ol>
150 </li>
151 <li><a href="#intrinsics">Intrinsic Functions</a>
152 <ol>
153 <li><a href="#int_varargs">Variable Argument Handling Intrinsics</a>
154 <ol>
155 <li><a href="#int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a></li>
156 <li><a href="#int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a></li>
157 <li><a href="#int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a></li>
158 </ol>
159 </li>
160 <li><a href="#int_gc">Accurate Garbage Collection Intrinsics</a>
161 <ol>
162 <li><a href="#int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a></li>
163 <li><a href="#int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a></li>
164 <li><a href="#int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a></li>
165 </ol>
166 </li>
167 <li><a href="#int_codegen">Code Generator Intrinsics</a>
168 <ol>
169 <li><a href="#int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a></li>
170 <li><a href="#int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a></li>
171 <li><a href="#int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a></li>
172 <li><a href="#int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a></li>
173 <li><a href="#int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a></li>
174 <li><a href="#int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a></li>
175 <li><a href="#int_readcyclecounter"><tt>llvm.readcyclecounter</tt>' Intrinsic</a></li>
176 </ol>
177 </li>
178 <li><a href="#int_libc">Standard C Library Intrinsics</a>
179 <ol>
180 <li><a href="#int_memcpy">'<tt>llvm.memcpy.*</tt>' Intrinsic</a></li>
181 <li><a href="#int_memmove">'<tt>llvm.memmove.*</tt>' Intrinsic</a></li>
182 <li><a href="#int_memset">'<tt>llvm.memset.*</tt>' Intrinsic</a></li>
183 <li><a href="#int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a></li>
184 <li><a href="#int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a></li>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]185 <li><a href="#int_sin">'<tt>llvm.sin.*</tt>' Intrinsic</a></li>
186 <li><a href="#int_cos">'<tt>llvm.cos.*</tt>' Intrinsic</a></li>
187 <li><a href="#int_pow">'<tt>llvm.pow.*</tt>' Intrinsic</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]188 </ol>
189 </li>
190 <li><a href="#int_manip">Bit Manipulation Intrinsics</a>
191 <ol>
192 <li><a href="#int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a></li>
193 <li><a href="#int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic </a></li>
194 <li><a href="#int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic </a></li>
195 <li><a href="#int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic </a></li>
196 <li><a href="#int_part_select">'<tt>llvm.part.select.*</tt>' Intrinsic </a></li>
197 <li><a href="#int_part_set">'<tt>llvm.part.set.*</tt>' Intrinsic </a></li>
198 </ol>
199 </li>
200 <li><a href="#int_debugger">Debugger intrinsics</a></li>
201 <li><a href="#int_eh">Exception Handling intrinsics</a></li>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]202 <li><a href="#int_trampoline">Trampoline Intrinsic</a>
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]203 <ol>
204 <li><a href="#int_it">'<tt>llvm.init.trampoline</tt>' Intrinsic</a></li>
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]205 </ol>
206 </li>
Reid Spencerb043f672007-07-20 19:59:11 +0000[diff] [blame]207 <li><a href="#int_general">General intrinsics</a>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]208 <ol>
Reid Spencerb043f672007-07-20 19:59:11 +0000[diff] [blame]209 <li><a href="#int_var_annotation">
Tanya Lattner51369f32007-09-22 00:01:26 +0000[diff] [blame]210 <tt>llvm.var.annotation</tt>' Intrinsic</a></li>
Reid Spencerb043f672007-07-20 19:59:11 +0000[diff] [blame]211 </ol>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]212 <ol>
213 <li><a href="#int_annotation">
Tanya Lattner51369f32007-09-22 00:01:26 +0000[diff] [blame]214 <tt>llvm.annotation.*</tt>' Intrinsic</a></li>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]215 </ol>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]216 </li>
217 </ol>
218 </li>
219</ol>
220
221<div class="doc_author">
222 <p>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a>
223 and <a href="mailto:vadve@cs.uiuc.edu">Vikram Adve</a></p>
224</div>
225
226<!-- *********************************************************************** -->
227<div class="doc_section"> <a name="abstract">Abstract </a></div>
228<!-- *********************************************************************** -->
229
230<div class="doc_text">
231<p>This document is a reference manual for the LLVM assembly language.
232LLVM is an SSA based representation that provides type safety,
233low-level operations, flexibility, and the capability of representing
234'all' high-level languages cleanly. It is the common code
235representation used throughout all phases of the LLVM compilation
236strategy.</p>
237</div>
238
239<!-- *********************************************************************** -->
240<div class="doc_section"> <a name="introduction">Introduction</a> </div>
241<!-- *********************************************************************** -->
242
243<div class="doc_text">
244
245<p>The LLVM code representation is designed to be used in three
246different forms: as an in-memory compiler IR, as an on-disk bitcode
247representation (suitable for fast loading by a Just-In-Time compiler),
248and as a human readable assembly language representation. This allows
249LLVM to provide a powerful intermediate representation for efficient
250compiler transformations and analysis, while providing a natural means
251to debug and visualize the transformations. The three different forms
252of LLVM are all equivalent. This document describes the human readable
253representation and notation.</p>
254
255<p>The LLVM representation aims to be light-weight and low-level
256while being expressive, typed, and extensible at the same time. It
257aims to be a "universal IR" of sorts, by being at a low enough level
258that high-level ideas may be cleanly mapped to it (similar to how
259microprocessors are "universal IR's", allowing many source languages to
260be mapped to them). By providing type information, LLVM can be used as
261the target of optimizations: for example, through pointer analysis, it
262can be proven that a C automatic variable is never accessed outside of
263the current function... allowing it to be promoted to a simple SSA
264value instead of a memory location.</p>
265
266</div>
267
268<!-- _______________________________________________________________________ -->
269<div class="doc_subsubsection"> <a name="wellformed">Well-Formedness</a> </div>
270
271<div class="doc_text">
272
273<p>It is important to note that this document describes 'well formed'
274LLVM assembly language. There is a difference between what the parser
275accepts and what is considered 'well formed'. For example, the
276following instruction is syntactically okay, but not well formed:</p>
277
278<div class="doc_code">
279<pre>
280%x = <a href="#i_add">add</a> i32 1, %x
281</pre>
282</div>
283
284<p>...because the definition of <tt>%x</tt> does not dominate all of
285its uses. The LLVM infrastructure provides a verification pass that may
286be used to verify that an LLVM module is well formed. This pass is
287automatically run by the parser after parsing input assembly and by
288the optimizer before it outputs bitcode. The violations pointed out
289by the verifier pass indicate bugs in transformation passes or input to
290the parser.</p>
291</div>
292
Chris Lattnera83fdc02007-10-03 17:34:29 +0000[diff] [blame]293<!-- Describe the typesetting conventions here. -->
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]294
295<!-- *********************************************************************** -->
296<div class="doc_section"> <a name="identifiers">Identifiers</a> </div>
297<!-- *********************************************************************** -->
298
299<div class="doc_text">
300
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]301 <p>LLVM identifiers come in two basic types: global and local. Global
302 identifiers (functions, global variables) begin with the @ character. Local
303 identifiers (register names, types) begin with the % character. Additionally,
304 there are three different formats for identifiers, for different purposes:
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]305
306<ol>
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]307 <li>Named values are represented as a string of characters with their prefix.
308 For example, %foo, @DivisionByZero, %a.really.long.identifier. The actual
309 regular expression used is '<tt>[%@][a-zA-Z$._][a-zA-Z$._0-9]*</tt>'.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]310 Identifiers which require other characters in their names can be surrounded
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]311 with quotes. In this way, anything except a <tt>&quot;</tt> character can
312 be used in a named value.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]313
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]314 <li>Unnamed values are represented as an unsigned numeric value with their
315 prefix. For example, %12, @2, %44.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]316
317 <li>Constants, which are described in a <a href="#constants">section about
318 constants</a>, below.</li>
319</ol>
320
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]321<p>LLVM requires that values start with a prefix for two reasons: Compilers
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]322don't need to worry about name clashes with reserved words, and the set of
323reserved words may be expanded in the future without penalty. Additionally,
324unnamed identifiers allow a compiler to quickly come up with a temporary
325variable without having to avoid symbol table conflicts.</p>
326
327<p>Reserved words in LLVM are very similar to reserved words in other
328languages. There are keywords for different opcodes
329('<tt><a href="#i_add">add</a></tt>',
330 '<tt><a href="#i_bitcast">bitcast</a></tt>',
331 '<tt><a href="#i_ret">ret</a></tt>', etc...), for primitive type names ('<tt><a
332href="#t_void">void</a></tt>', '<tt><a href="#t_primitive">i32</a></tt>', etc...),
333and others. These reserved words cannot conflict with variable names, because
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]334none of them start with a prefix character ('%' or '@').</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]335
336<p>Here is an example of LLVM code to multiply the integer variable
337'<tt>%X</tt>' by 8:</p>
338
339<p>The easy way:</p>
340
341<div class="doc_code">
342<pre>
343%result = <a href="#i_mul">mul</a> i32 %X, 8
344</pre>
345</div>
346
347<p>After strength reduction:</p>
348
349<div class="doc_code">
350<pre>
351%result = <a href="#i_shl">shl</a> i32 %X, i8 3
352</pre>
353</div>
354
355<p>And the hard way:</p>
356
357<div class="doc_code">
358<pre>
359<a href="#i_add">add</a> i32 %X, %X <i>; yields {i32}:%0</i>
360<a href="#i_add">add</a> i32 %0, %0 <i>; yields {i32}:%1</i>
361%result = <a href="#i_add">add</a> i32 %1, %1
362</pre>
363</div>
364
365<p>This last way of multiplying <tt>%X</tt> by 8 illustrates several
366important lexical features of LLVM:</p>
367
368<ol>
369
370 <li>Comments are delimited with a '<tt>;</tt>' and go until the end of
371 line.</li>
372
373 <li>Unnamed temporaries are created when the result of a computation is not
374 assigned to a named value.</li>
375
376 <li>Unnamed temporaries are numbered sequentially</li>
377
378</ol>
379
380<p>...and it also shows a convention that we follow in this document. When
381demonstrating instructions, we will follow an instruction with a comment that
382defines the type and name of value produced. Comments are shown in italic
383text.</p>
384
385</div>
386
387<!-- *********************************************************************** -->
388<div class="doc_section"> <a name="highlevel">High Level Structure</a> </div>
389<!-- *********************************************************************** -->
390
391<!-- ======================================================================= -->
392<div class="doc_subsection"> <a name="modulestructure">Module Structure</a>
393</div>
394
395<div class="doc_text">
396
397<p>LLVM programs are composed of "Module"s, each of which is a
398translation unit of the input programs. Each module consists of
399functions, global variables, and symbol table entries. Modules may be
400combined together with the LLVM linker, which merges function (and
401global variable) definitions, resolves forward declarations, and merges
402symbol table entries. Here is an example of the "hello world" module:</p>
403
404<div class="doc_code">
405<pre><i>; Declare the string constant as a global constant...</i>
406<a href="#identifiers">@.LC0</a> = <a href="#linkage_internal">internal</a> <a
407 href="#globalvars">constant</a> <a href="#t_array">[13 x i8]</a> c"hello world\0A\00" <i>; [13 x i8]*</i>
408
409<i>; External declaration of the puts function</i>
410<a href="#functionstructure">declare</a> i32 @puts(i8 *) <i>; i32(i8 *)* </i>
411
412<i>; Definition of main function</i>
413define i32 @main() { <i>; i32()* </i>
414 <i>; Convert [13x i8 ]* to i8 *...</i>
415 %cast210 = <a
416 href="#i_getelementptr">getelementptr</a> [13 x i8 ]* @.LC0, i64 0, i64 0 <i>; i8 *</i>
417
418 <i>; Call puts function to write out the string to stdout...</i>
419 <a
420 href="#i_call">call</a> i32 @puts(i8 * %cast210) <i>; i32</i>
421 <a
422 href="#i_ret">ret</a> i32 0<br>}<br>
423</pre>
424</div>
425
426<p>This example is made up of a <a href="#globalvars">global variable</a>
427named "<tt>.LC0</tt>", an external declaration of the "<tt>puts</tt>"
428function, and a <a href="#functionstructure">function definition</a>
429for "<tt>main</tt>".</p>
430
431<p>In general, a module is made up of a list of global values,
432where both functions and global variables are global values. Global values are
433represented by a pointer to a memory location (in this case, a pointer to an
434array of char, and a pointer to a function), and have one of the following <a
435href="#linkage">linkage types</a>.</p>
436
437</div>
438
439<!-- ======================================================================= -->
440<div class="doc_subsection">
441 <a name="linkage">Linkage Types</a>
442</div>
443
444<div class="doc_text">
445
446<p>
447All Global Variables and Functions have one of the following types of linkage:
448</p>
449
450<dl>
451
452 <dt><tt><b><a name="linkage_internal">internal</a></b></tt> </dt>
453
454 <dd>Global values with internal linkage are only directly accessible by
455 objects in the current module. In particular, linking code into a module with
456 an internal global value may cause the internal to be renamed as necessary to
457 avoid collisions. Because the symbol is internal to the module, all
458 references can be updated. This corresponds to the notion of the
459 '<tt>static</tt>' keyword in C.
460 </dd>
461
462 <dt><tt><b><a name="linkage_linkonce">linkonce</a></b></tt>: </dt>
463
464 <dd>Globals with "<tt>linkonce</tt>" linkage are merged with other globals of
465 the same name when linkage occurs. This is typically used to implement
466 inline functions, templates, or other code which must be generated in each
467 translation unit that uses it. Unreferenced <tt>linkonce</tt> globals are
468 allowed to be discarded.
469 </dd>
470
471 <dt><tt><b><a name="linkage_weak">weak</a></b></tt>: </dt>
472
473 <dd>"<tt>weak</tt>" linkage is exactly the same as <tt>linkonce</tt> linkage,
474 except that unreferenced <tt>weak</tt> globals may not be discarded. This is
475 used for globals that may be emitted in multiple translation units, but that
476 are not guaranteed to be emitted into every translation unit that uses them.
477 One example of this are common globals in C, such as "<tt>int X;</tt>" at
478 global scope.
479 </dd>
480
481 <dt><tt><b><a name="linkage_appending">appending</a></b></tt>: </dt>
482
483 <dd>"<tt>appending</tt>" linkage may only be applied to global variables of
484 pointer to array type. When two global variables with appending linkage are
485 linked together, the two global arrays are appended together. This is the
486 LLVM, typesafe, equivalent of having the system linker append together
487 "sections" with identical names when .o files are linked.
488 </dd>
489
490 <dt><tt><b><a name="linkage_externweak">extern_weak</a></b></tt>: </dt>
491 <dd>The semantics of this linkage follow the ELF model: the symbol is weak
492 until linked, if not linked, the symbol becomes null instead of being an
493 undefined reference.
494 </dd>
495
496 <dt><tt><b><a name="linkage_external">externally visible</a></b></tt>:</dt>
497
498 <dd>If none of the above identifiers are used, the global is externally
499 visible, meaning that it participates in linkage and can be used to resolve
500 external symbol references.
501 </dd>
502</dl>
503
504 <p>
505 The next two types of linkage are targeted for Microsoft Windows platform
506 only. They are designed to support importing (exporting) symbols from (to)
507 DLLs.
508 </p>
509
510 <dl>
511 <dt><tt><b><a name="linkage_dllimport">dllimport</a></b></tt>: </dt>
512
513 <dd>"<tt>dllimport</tt>" linkage causes the compiler to reference a function
514 or variable via a global pointer to a pointer that is set up by the DLL
515 exporting the symbol. On Microsoft Windows targets, the pointer name is
516 formed by combining <code>_imp__</code> and the function or variable name.
517 </dd>
518
519 <dt><tt><b><a name="linkage_dllexport">dllexport</a></b></tt>: </dt>
520
521 <dd>"<tt>dllexport</tt>" linkage causes the compiler to provide a global
522 pointer to a pointer in a DLL, so that it can be referenced with the
523 <tt>dllimport</tt> attribute. On Microsoft Windows targets, the pointer
524 name is formed by combining <code>_imp__</code> and the function or variable
525 name.
526 </dd>
527
528</dl>
529
530<p><a name="linkage_external"></a>For example, since the "<tt>.LC0</tt>"
531variable is defined to be internal, if another module defined a "<tt>.LC0</tt>"
532variable and was linked with this one, one of the two would be renamed,
533preventing a collision. Since "<tt>main</tt>" and "<tt>puts</tt>" are
534external (i.e., lacking any linkage declarations), they are accessible
535outside of the current module.</p>
536<p>It is illegal for a function <i>declaration</i>
537to have any linkage type other than "externally visible", <tt>dllimport</tt>,
538or <tt>extern_weak</tt>.</p>
539<p>Aliases can have only <tt>external</tt>, <tt>internal</tt> and <tt>weak</tt>
540linkages.
541</div>
542
543<!-- ======================================================================= -->
544<div class="doc_subsection">
545 <a name="callingconv">Calling Conventions</a>
546</div>
547
548<div class="doc_text">
549
550<p>LLVM <a href="#functionstructure">functions</a>, <a href="#i_call">calls</a>
551and <a href="#i_invoke">invokes</a> can all have an optional calling convention
552specified for the call. The calling convention of any pair of dynamic
553caller/callee must match, or the behavior of the program is undefined. The
554following calling conventions are supported by LLVM, and more may be added in
555the future:</p>
556
557<dl>
558 <dt><b>"<tt>ccc</tt>" - The C calling convention</b>:</dt>
559
560 <dd>This calling convention (the default if no other calling convention is
561 specified) matches the target C calling conventions. This calling convention
562 supports varargs function calls and tolerates some mismatch in the declared
563 prototype and implemented declaration of the function (as does normal C).
564 </dd>
565
566 <dt><b>"<tt>fastcc</tt>" - The fast calling convention</b>:</dt>
567
568 <dd>This calling convention attempts to make calls as fast as possible
569 (e.g. by passing things in registers). This calling convention allows the
570 target to use whatever tricks it wants to produce fast code for the target,
571 without having to conform to an externally specified ABI. Implementations of
572 this convention should allow arbitrary tail call optimization to be supported.
573 This calling convention does not support varargs and requires the prototype of
574 all callees to exactly match the prototype of the function definition.
575 </dd>
576
577 <dt><b>"<tt>coldcc</tt>" - The cold calling convention</b>:</dt>
578
579 <dd>This calling convention attempts to make code in the caller as efficient
580 as possible under the assumption that the call is not commonly executed. As
581 such, these calls often preserve all registers so that the call does not break
582 any live ranges in the caller side. This calling convention does not support
583 varargs and requires the prototype of all callees to exactly match the
584 prototype of the function definition.
585 </dd>
586
587 <dt><b>"<tt>cc &lt;<em>n</em>&gt;</tt>" - Numbered convention</b>:</dt>
588
589 <dd>Any calling convention may be specified by number, allowing
590 target-specific calling conventions to be used. Target specific calling
591 conventions start at 64.
592 </dd>
593</dl>
594
595<p>More calling conventions can be added/defined on an as-needed basis, to
596support pascal conventions or any other well-known target-independent
597convention.</p>
598
599</div>
600
601<!-- ======================================================================= -->
602<div class="doc_subsection">
603 <a name="visibility">Visibility Styles</a>
604</div>
605
606<div class="doc_text">
607
608<p>
609All Global Variables and Functions have one of the following visibility styles:
610</p>
611
612<dl>
613 <dt><b>"<tt>default</tt>" - Default style</b>:</dt>
614
615 <dd>On ELF, default visibility means that the declaration is visible to other
616 modules and, in shared libraries, means that the declared entity may be
617 overridden. On Darwin, default visibility means that the declaration is
618 visible to other modules. Default visibility corresponds to "external
619 linkage" in the language.
620 </dd>
621
622 <dt><b>"<tt>hidden</tt>" - Hidden style</b>:</dt>
623
624 <dd>Two declarations of an object with hidden visibility refer to the same
625 object if they are in the same shared object. Usually, hidden visibility
626 indicates that the symbol will not be placed into the dynamic symbol table,
627 so no other module (executable or shared library) can reference it
628 directly.
629 </dd>
630
631 <dt><b>"<tt>protected</tt>" - Protected style</b>:</dt>
632
633 <dd>On ELF, protected visibility indicates that the symbol will be placed in
634 the dynamic symbol table, but that references within the defining module will
635 bind to the local symbol. That is, the symbol cannot be overridden by another
636 module.
637 </dd>
638</dl>
639
640</div>
641
642<!-- ======================================================================= -->
643<div class="doc_subsection">
644 <a name="globalvars">Global Variables</a>
645</div>
646
647<div class="doc_text">
648
649<p>Global variables define regions of memory allocated at compilation time
650instead of run-time. Global variables may optionally be initialized, may have
651an explicit section to be placed in, and may have an optional explicit alignment
652specified. A variable may be defined as "thread_local", which means that it
653will not be shared by threads (each thread will have a separated copy of the
654variable). A variable may be defined as a global "constant," which indicates
655that the contents of the variable will <b>never</b> be modified (enabling better
656optimization, allowing the global data to be placed in the read-only section of
657an executable, etc). Note that variables that need runtime initialization
658cannot be marked "constant" as there is a store to the variable.</p>
659
660<p>
661LLVM explicitly allows <em>declarations</em> of global variables to be marked
662constant, even if the final definition of the global is not. This capability
663can be used to enable slightly better optimization of the program, but requires
664the language definition to guarantee that optimizations based on the
665'constantness' are valid for the translation units that do not include the
666definition.
667</p>
668
669<p>As SSA values, global variables define pointer values that are in
670scope (i.e. they dominate) all basic blocks in the program. Global
671variables always define a pointer to their "content" type because they
672describe a region of memory, and all memory objects in LLVM are
673accessed through pointers.</p>
674
Christopher Lambdd0049d2007-12-11 09:31:00 +0000[diff] [blame]675<p>A global variable may be declared to reside in a target-specifc numbered
676address space. For targets that support them, address spaces may affect how
677optimizations are performed and/or what target instructions are used to access
Christopher Lamb20a39e92007-12-12 08:44:39 +0000[diff] [blame]678the variable. The default address space is zero. The address space qualifier
679must precede any other attributes.</p>
Christopher Lambdd0049d2007-12-11 09:31:00 +0000[diff] [blame]680
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]681<p>LLVM allows an explicit section to be specified for globals. If the target
682supports it, it will emit globals to the section specified.</p>
683
684<p>An explicit alignment may be specified for a global. If not present, or if
685the alignment is set to zero, the alignment of the global is set by the target
686to whatever it feels convenient. If an explicit alignment is specified, the
687global is forced to have at least that much alignment. All alignments must be
688a power of 2.</p>
689
Christopher Lambdd0049d2007-12-11 09:31:00 +0000[diff] [blame]690<p>For example, the following defines a global in a numbered address space with
691an initializer, section, and alignment:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]692
693<div class="doc_code">
694<pre>
Christopher Lambdd0049d2007-12-11 09:31:00 +0000[diff] [blame]695@G = constant float 1.0 addrspace(5), section "foo", align 4
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]696</pre>
697</div>
698
699</div>
700
701
702<!-- ======================================================================= -->
703<div class="doc_subsection">
704 <a name="functionstructure">Functions</a>
705</div>
706
707<div class="doc_text">
708
709<p>LLVM function definitions consist of the "<tt>define</tt>" keyord,
710an optional <a href="#linkage">linkage type</a>, an optional
711<a href="#visibility">visibility style</a>, an optional
712<a href="#callingconv">calling convention</a>, a return type, an optional
713<a href="#paramattrs">parameter attribute</a> for the return type, a function
714name, a (possibly empty) argument list (each with optional
715<a href="#paramattrs">parameter attributes</a>), an optional section, an
Gordon Henriksend606f5b2007-12-10 03:30:21 +0000[diff] [blame]716optional alignment, an optional <a href="#gc">garbage collector name</a>, an
Gordon Henriksen13fe5e32007-12-10 03:18:06 +0000[diff] [blame]717opening curly brace, a list of basic blocks, and a closing curly brace.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]718
719LLVM function declarations consist of the "<tt>declare</tt>" keyword, an
720optional <a href="#linkage">linkage type</a>, an optional
721<a href="#visibility">visibility style</a>, an optional
722<a href="#callingconv">calling convention</a>, a return type, an optional
723<a href="#paramattrs">parameter attribute</a> for the return type, a function
Gordon Henriksen13fe5e32007-12-10 03:18:06 +0000[diff] [blame]724name, a possibly empty list of arguments, an optional alignment, and an optional
Gordon Henriksend606f5b2007-12-10 03:30:21 +0000[diff] [blame]725<a href="#gc">garbage collector name</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]726
727<p>A function definition contains a list of basic blocks, forming the CFG for
728the function. Each basic block may optionally start with a label (giving the
729basic block a symbol table entry), contains a list of instructions, and ends
730with a <a href="#terminators">terminator</a> instruction (such as a branch or
731function return).</p>
732
733<p>The first basic block in a function is special in two ways: it is immediately
734executed on entrance to the function, and it is not allowed to have predecessor
735basic blocks (i.e. there can not be any branches to the entry block of a
736function). Because the block can have no predecessors, it also cannot have any
737<a href="#i_phi">PHI nodes</a>.</p>
738
739<p>LLVM allows an explicit section to be specified for functions. If the target
740supports it, it will emit functions to the section specified.</p>
741
742<p>An explicit alignment may be specified for a function. If not present, or if
743the alignment is set to zero, the alignment of the function is set by the target
744to whatever it feels convenient. If an explicit alignment is specified, the
745function is forced to have at least that much alignment. All alignments must be
746a power of 2.</p>
747
748</div>
749
750
751<!-- ======================================================================= -->
752<div class="doc_subsection">
753 <a name="aliasstructure">Aliases</a>
754</div>
755<div class="doc_text">
756 <p>Aliases act as "second name" for the aliasee value (which can be either
757 function or global variable or bitcast of global value). Aliases may have an
758 optional <a href="#linkage">linkage type</a>, and an
759 optional <a href="#visibility">visibility style</a>.</p>
760
761 <h5>Syntax:</h5>
762
763<div class="doc_code">
764<pre>
765@&lt;Name&gt; = [Linkage] [Visibility] alias &lt;AliaseeTy&gt; @&lt;Aliasee&gt;
766</pre>
767</div>
768
769</div>
770
771
772
773<!-- ======================================================================= -->
774<div class="doc_subsection"><a name="paramattrs">Parameter Attributes</a></div>
775<div class="doc_text">
776 <p>The return type and each parameter of a function type may have a set of
777 <i>parameter attributes</i> associated with them. Parameter attributes are
778 used to communicate additional information about the result or parameters of
Duncan Sandsf5588dc2007-11-27 13:23:08 +0000[diff] [blame]779 a function. Parameter attributes are considered to be part of the function,
780 not of the function type, so functions with different parameter attributes
781 can have the same function type.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]782
783 <p>Parameter attributes are simple keywords that follow the type specified. If
784 multiple parameter attributes are needed, they are space separated. For
785 example:</p>
786
787<div class="doc_code">
788<pre>
Duncan Sandsf5588dc2007-11-27 13:23:08 +0000[diff] [blame]789declare i32 @printf(i8* noalias , ...) nounwind
790declare i32 @atoi(i8*) nounwind readonly
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]791</pre>
792</div>
793
Duncan Sandsf5588dc2007-11-27 13:23:08 +0000[diff] [blame]794 <p>Note that any attributes for the function result (<tt>nounwind</tt>,
795 <tt>readonly</tt>) come immediately after the argument list.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]796
797 <p>Currently, only the following parameter attributes are defined:</p>
798 <dl>
Reid Spencerf234bed2007-07-19 23:13:04 +0000[diff] [blame]799 <dt><tt>zeroext</tt></dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]800 <dd>This indicates that the parameter should be zero extended just before
801 a call to this function.</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame^]802
Reid Spencerf234bed2007-07-19 23:13:04 +0000[diff] [blame]803 <dt><tt>signext</tt></dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]804 <dd>This indicates that the parameter should be sign extended just before
805 a call to this function.</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame^]806
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]807 <dt><tt>inreg</tt></dt>
808 <dd>This indicates that the parameter should be placed in register (if
809 possible) during assembling function call. Support for this attribute is
810 target-specific</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame^]811
812 <dt><tt>byval</tt></dt>
813 <dd>This indicates that the pointer parameter is really an aggregate that
814 was passed by value to the function. The attribute implies that a hidden
815 copy of the struct is made between the caller and the callee, so the
816 callee is unable to modify the struct in the callee. This attribute is only
817 valid on llvm pointer arguments.</dd>
818
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]819 <dt><tt>sret</tt></dt>
820 <dd>This indicates that the parameter specifies the address of a structure
821 that is the return value of the function in the source program.</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame^]822
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]823 <dt><tt>noalias</tt></dt>
824 <dd>This indicates that the parameter not alias any other object or any
825 other "noalias" objects during the function call.
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame^]826
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]827 <dt><tt>noreturn</tt></dt>
828 <dd>This function attribute indicates that the function never returns. This
829 indicates to LLVM that every call to this function should be treated as if
830 an <tt>unreachable</tt> instruction immediately followed the call.</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame^]831
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]832 <dt><tt>nounwind</tt></dt>
833 <dd>This function attribute indicates that the function type does not use
834 the unwind instruction and does not allow stack unwinding to propagate
835 through it.</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame^]836
Duncan Sands4ee46812007-07-27 19:57:41 +0000[diff] [blame]837 <dt><tt>nest</tt></dt>
838 <dd>This indicates that the parameter can be excised using the
839 <a href="#int_trampoline">trampoline intrinsics</a>.</dd>
Duncan Sands13e13f82007-11-22 20:23:04 +0000[diff] [blame]840 <dt><tt>readonly</tt></dt>
Duncan Sandsd69c0e62007-11-14 21:14:02 +0000[diff] [blame]841 <dd>This function attribute indicates that the function has no side-effects
Duncan Sands13e13f82007-11-22 20:23:04 +0000[diff] [blame]842 except for producing a return value or throwing an exception. The value
843 returned must only depend on the function arguments and/or global variables.
844 It may use values obtained by dereferencing pointers.</dd>
845 <dt><tt>readnone</tt></dt>
846 <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]847 function, but in addition it is not allowed to dereference any pointer arguments
848 or global variables.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]849 </dl>
850
851</div>
852
853<!-- ======================================================================= -->
854<div class="doc_subsection">
Gordon Henriksen13fe5e32007-12-10 03:18:06 +0000[diff] [blame]855 <a name="gc">Garbage Collector Names</a>
856</div>
857
858<div class="doc_text">
859<p>Each function may specify a garbage collector name, which is simply a
860string.</p>
861
862<div class="doc_code"><pre
863>define void @f() gc "name" { ...</pre></div>
864
865<p>The compiler declares the supported values of <i>name</i>. Specifying a
866collector which will cause the compiler to alter its output in order to support
867the named garbage collection algorithm.</p>
868</div>
869
870<!-- ======================================================================= -->
871<div class="doc_subsection">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]872 <a name="moduleasm">Module-Level Inline Assembly</a>
873</div>
874
875<div class="doc_text">
876<p>
877Modules may contain "module-level inline asm" blocks, which corresponds to the
878GCC "file scope inline asm" blocks. These blocks are internally concatenated by
879LLVM and treated as a single unit, but may be separated in the .ll file if
880desired. The syntax is very simple:
881</p>
882
883<div class="doc_code">
884<pre>
885module asm "inline asm code goes here"
886module asm "more can go here"
887</pre>
888</div>
889
890<p>The strings can contain any character by escaping non-printable characters.
891 The escape sequence used is simply "\xx" where "xx" is the two digit hex code
892 for the number.
893</p>
894
895<p>
896 The inline asm code is simply printed to the machine code .s file when
897 assembly code is generated.
898</p>
899</div>
900
901<!-- ======================================================================= -->
902<div class="doc_subsection">
903 <a name="datalayout">Data Layout</a>
904</div>
905
906<div class="doc_text">
907<p>A module may specify a target specific data layout string that specifies how
908data is to be laid out in memory. The syntax for the data layout is simply:</p>
909<pre> target datalayout = "<i>layout specification</i>"</pre>
910<p>The <i>layout specification</i> consists of a list of specifications
911separated by the minus sign character ('-'). Each specification starts with a
912letter and may include other information after the letter to define some
913aspect of the data layout. The specifications accepted are as follows: </p>
914<dl>
915 <dt><tt>E</tt></dt>
916 <dd>Specifies that the target lays out data in big-endian form. That is, the
917 bits with the most significance have the lowest address location.</dd>
918 <dt><tt>e</tt></dt>
919 <dd>Specifies that hte target lays out data in little-endian form. That is,
920 the bits with the least significance have the lowest address location.</dd>
921 <dt><tt>p:<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
922 <dd>This specifies the <i>size</i> of a pointer and its <i>abi</i> and
923 <i>preferred</i> alignments. All sizes are in bits. Specifying the <i>pref</i>
924 alignment is optional. If omitted, the preceding <tt>:</tt> should be omitted
925 too.</dd>
926 <dt><tt>i<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
927 <dd>This specifies the alignment for an integer type of a given bit
928 <i>size</i>. The value of <i>size</i> must be in the range [1,2^23).</dd>
929 <dt><tt>v<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
930 <dd>This specifies the alignment for a vector type of a given bit
931 <i>size</i>.</dd>
932 <dt><tt>f<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
933 <dd>This specifies the alignment for a floating point type of a given bit
934 <i>size</i>. The value of <i>size</i> must be either 32 (float) or 64
935 (double).</dd>
936 <dt><tt>a<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
937 <dd>This specifies the alignment for an aggregate type of a given bit
938 <i>size</i>.</dd>
939</dl>
940<p>When constructing the data layout for a given target, LLVM starts with a
941default set of specifications which are then (possibly) overriden by the
942specifications in the <tt>datalayout</tt> keyword. The default specifications
943are given in this list:</p>
944<ul>
945 <li><tt>E</tt> - big endian</li>
946 <li><tt>p:32:64:64</tt> - 32-bit pointers with 64-bit alignment</li>
947 <li><tt>i1:8:8</tt> - i1 is 8-bit (byte) aligned</li>
948 <li><tt>i8:8:8</tt> - i8 is 8-bit (byte) aligned</li>
949 <li><tt>i16:16:16</tt> - i16 is 16-bit aligned</li>
950 <li><tt>i32:32:32</tt> - i32 is 32-bit aligned</li>
951 <li><tt>i64:32:64</tt> - i64 has abi alignment of 32-bits but preferred
952 alignment of 64-bits</li>
953 <li><tt>f32:32:32</tt> - float is 32-bit aligned</li>
954 <li><tt>f64:64:64</tt> - double is 64-bit aligned</li>
955 <li><tt>v64:64:64</tt> - 64-bit vector is 64-bit aligned</li>
956 <li><tt>v128:128:128</tt> - 128-bit vector is 128-bit aligned</li>
957 <li><tt>a0:0:1</tt> - aggregates are 8-bit aligned</li>
958</ul>
959<p>When llvm is determining the alignment for a given type, it uses the
960following rules:
961<ol>
962 <li>If the type sought is an exact match for one of the specifications, that
963 specification is used.</li>
964 <li>If no match is found, and the type sought is an integer type, then the
965 smallest integer type that is larger than the bitwidth of the sought type is
966 used. If none of the specifications are larger than the bitwidth then the the
967 largest integer type is used. For example, given the default specifications
968 above, the i7 type will use the alignment of i8 (next largest) while both
969 i65 and i256 will use the alignment of i64 (largest specified).</li>
970 <li>If no match is found, and the type sought is a vector type, then the
971 largest vector type that is smaller than the sought vector type will be used
972 as a fall back. This happens because <128 x double> can be implemented in
973 terms of 64 <2 x double>, for example.</li>
974</ol>
975</div>
976
977<!-- *********************************************************************** -->
978<div class="doc_section"> <a name="typesystem">Type System</a> </div>
979<!-- *********************************************************************** -->
980
981<div class="doc_text">
982
983<p>The LLVM type system is one of the most important features of the
984intermediate representation. Being typed enables a number of
985optimizations to be performed on the IR directly, without having to do
986extra analyses on the side before the transformation. A strong type
987system makes it easier to read the generated code and enables novel
988analyses and transformations that are not feasible to perform on normal
989three address code representations.</p>
990
991</div>
992
993<!-- ======================================================================= -->
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]994<div class="doc_subsection"> <a name="t_classifications">Type
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]995Classifications</a> </div>
996<div class="doc_text">
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]997<p>The types fall into a few useful
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]998classifications:</p>
999
1000<table border="1" cellspacing="0" cellpadding="4">
1001 <tbody>
1002 <tr><th>Classification</th><th>Types</th></tr>
1003 <tr>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1004 <td><a href="#t_integer">integer</a></td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1005 <td><tt>i1, i2, i3, ... i8, ... i16, ... i32, ... i64, ... </tt></td>
1006 </tr>
1007 <tr>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1008 <td><a href="#t_floating">floating point</a></td>
1009 <td><tt>float, double, x86_fp80, fp128, ppc_fp128</tt></td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1010 </tr>
1011 <tr>
1012 <td><a name="t_firstclass">first class</a></td>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1013 <td><a href="#t_integer">integer</a>,
1014 <a href="#t_floating">floating point</a>,
1015 <a href="#t_pointer">pointer</a>,
1016 <a href="#t_vector">vector</a>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1017 </td>
1018 </tr>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1019 <tr>
1020 <td><a href="#t_primitive">primitive</a></td>
1021 <td><a href="#t_label">label</a>,
1022 <a href="#t_void">void</a>,
1023 <a href="#t_integer">integer</a>,
1024 <a href="#t_floating">floating point</a>.</td>
1025 </tr>
1026 <tr>
1027 <td><a href="#t_derived">derived</a></td>
1028 <td><a href="#t_integer">integer</a>,
1029 <a href="#t_array">array</a>,
1030 <a href="#t_function">function</a>,
1031 <a href="#t_pointer">pointer</a>,
1032 <a href="#t_struct">structure</a>,
1033 <a href="#t_pstruct">packed structure</a>,
1034 <a href="#t_vector">vector</a>,
1035 <a href="#t_opaque">opaque</a>.
1036 </tr>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1037 </tbody>
1038</table>
1039
1040<p>The <a href="#t_firstclass">first class</a> types are perhaps the
1041most important. Values of these types are the only ones which can be
1042produced by instructions, passed as arguments, or used as operands to
1043instructions. This means that all structures and arrays must be
1044manipulated either by pointer or by component.</p>
1045</div>
1046
1047<!-- ======================================================================= -->
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1048<div class="doc_subsection"> <a name="t_primitive">Primitive Types</a> </div>
Chris Lattner86437612008-01-04 04:34:14 +0000[diff] [blame]1049
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1050<div class="doc_text">
1051<p>The primitive types are the fundamental building blocks of the LLVM
1052system.</p>
1053
Chris Lattner86437612008-01-04 04:34:14 +0000[diff] [blame]1054</div>
1055
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1056<!-- _______________________________________________________________________ -->
1057<div class="doc_subsubsection"> <a name="t_floating">Floating Point Types</a> </div>
1058
1059<div class="doc_text">
1060 <table>
1061 <tbody>
1062 <tr><th>Type</th><th>Description</th></tr>
1063 <tr><td><tt>float</tt></td><td>32-bit floating point value</td></tr>
1064 <tr><td><tt>double</tt></td><td>64-bit floating point value</td></tr>
1065 <tr><td><tt>fp128</tt></td><td>128-bit floating point value (112-bit mantissa)</td></tr>
1066 <tr><td><tt>x86_fp80</tt></td><td>80-bit floating point value (X87)</td></tr>
1067 <tr><td><tt>ppc_fp128</tt></td><td>128-bit floating point value (two 64-bits)</td></tr>
1068 </tbody>
1069 </table>
1070</div>
1071
1072<!-- _______________________________________________________________________ -->
1073<div class="doc_subsubsection"> <a name="t_void">Void Type</a> </div>
1074
1075<div class="doc_text">
1076<h5>Overview:</h5>
1077<p>The void type does not represent any value and has no size.</p>
1078
1079<h5>Syntax:</h5>
1080
1081<pre>
1082 void
1083</pre>
1084</div>
1085
1086<!-- _______________________________________________________________________ -->
1087<div class="doc_subsubsection"> <a name="t_label">Label Type</a> </div>
1088
1089<div class="doc_text">
1090<h5>Overview:</h5>
1091<p>The label type represents code labels.</p>
1092
1093<h5>Syntax:</h5>
1094
1095<pre>
1096 label
1097</pre>
1098</div>
1099
1100
1101<!-- ======================================================================= -->
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1102<div class="doc_subsection"> <a name="t_derived">Derived Types</a> </div>
1103
1104<div class="doc_text">
1105
1106<p>The real power in LLVM comes from the derived types in the system.
1107This is what allows a programmer to represent arrays, functions,
1108pointers, and other useful types. Note that these derived types may be
1109recursive: For example, it is possible to have a two dimensional array.</p>
1110
1111</div>
1112
1113<!-- _______________________________________________________________________ -->
1114<div class="doc_subsubsection"> <a name="t_integer">Integer Type</a> </div>
1115
1116<div class="doc_text">
1117
1118<h5>Overview:</h5>
1119<p>The integer type is a very simple derived type that simply specifies an
1120arbitrary bit width for the integer type desired. Any bit width from 1 bit to
11212^23-1 (about 8 million) can be specified.</p>
1122
1123<h5>Syntax:</h5>
1124
1125<pre>
1126 iN
1127</pre>
1128
1129<p>The number of bits the integer will occupy is specified by the <tt>N</tt>
1130value.</p>
1131
1132<h5>Examples:</h5>
1133<table class="layout">
Chris Lattner251ab812007-12-18 06:18:21 +0000[diff] [blame]1134 <tbody>
1135 <tr>
1136 <td><tt>i1</tt></td>
1137 <td>a single-bit integer.</td>
1138 </tr><tr>
1139 <td><tt>i32</tt></td>
1140 <td>a 32-bit integer.</td>
1141 </tr><tr>
1142 <td><tt>i1942652</tt></td>
1143 <td>a really big integer of over 1 million bits.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1144 </tr>
Chris Lattner251ab812007-12-18 06:18:21 +0000[diff] [blame]1145 </tbody>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1146</table>
1147</div>
1148
1149<!-- _______________________________________________________________________ -->
1150<div class="doc_subsubsection"> <a name="t_array">Array Type</a> </div>
1151
1152<div class="doc_text">
1153
1154<h5>Overview:</h5>
1155
1156<p>The array type is a very simple derived type that arranges elements
1157sequentially in memory. The array type requires a size (number of
1158elements) and an underlying data type.</p>
1159
1160<h5>Syntax:</h5>
1161
1162<pre>
1163 [&lt;# elements&gt; x &lt;elementtype&gt;]
1164</pre>
1165
1166<p>The number of elements is a constant integer value; elementtype may
1167be any type with a size.</p>
1168
1169<h5>Examples:</h5>
1170<table class="layout">
1171 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1172 <td class="left"><tt>[40 x i32]</tt></td>
1173 <td class="left">Array of 40 32-bit integer values.</td>
1174 </tr>
1175 <tr class="layout">
1176 <td class="left"><tt>[41 x i32]</tt></td>
1177 <td class="left">Array of 41 32-bit integer values.</td>
1178 </tr>
1179 <tr class="layout">
1180 <td class="left"><tt>[4 x i8]</tt></td>
1181 <td class="left">Array of 4 8-bit integer values.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1182 </tr>
1183</table>
1184<p>Here are some examples of multidimensional arrays:</p>
1185<table class="layout">
1186 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1187 <td class="left"><tt>[3 x [4 x i32]]</tt></td>
1188 <td class="left">3x4 array of 32-bit integer values.</td>
1189 </tr>
1190 <tr class="layout">
1191 <td class="left"><tt>[12 x [10 x float]]</tt></td>
1192 <td class="left">12x10 array of single precision floating point values.</td>
1193 </tr>
1194 <tr class="layout">
1195 <td class="left"><tt>[2 x [3 x [4 x i16]]]</tt></td>
1196 <td class="left">2x3x4 array of 16-bit integer values.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1197 </tr>
1198</table>
1199
1200<p>Note that 'variable sized arrays' can be implemented in LLVM with a zero
1201length array. Normally, accesses past the end of an array are undefined in
1202LLVM (e.g. it is illegal to access the 5th element of a 3 element array).
1203As a special case, however, zero length arrays are recognized to be variable
1204length. This allows implementation of 'pascal style arrays' with the LLVM
1205type "{ i32, [0 x float]}", for example.</p>
1206
1207</div>
1208
1209<!-- _______________________________________________________________________ -->
1210<div class="doc_subsubsection"> <a name="t_function">Function Type</a> </div>
1211<div class="doc_text">
1212<h5>Overview:</h5>
1213<p>The function type can be thought of as a function signature. It
1214consists of a return type and a list of formal parameter types.
1215Function types are usually used to build virtual function tables
1216(which are structures of pointers to functions), for indirect function
1217calls, and when defining a function.</p>
1218<p>
1219The return type of a function type cannot be an aggregate type.
1220</p>
1221<h5>Syntax:</h5>
1222<pre> &lt;returntype&gt; (&lt;parameter list&gt;)<br></pre>
1223<p>...where '<tt>&lt;parameter list&gt;</tt>' is a comma-separated list of type
1224specifiers. Optionally, the parameter list may include a type <tt>...</tt>,
1225which indicates that the function takes a variable number of arguments.
1226Variable argument functions can access their arguments with the <a
1227 href="#int_varargs">variable argument handling intrinsic</a> functions.</p>
1228<h5>Examples:</h5>
1229<table class="layout">
1230 <tr class="layout">
1231 <td class="left"><tt>i32 (i32)</tt></td>
1232 <td class="left">function taking an <tt>i32</tt>, returning an <tt>i32</tt>
1233 </td>
1234 </tr><tr class="layout">
Reid Spencerf234bed2007-07-19 23:13:04 +0000[diff] [blame]1235 <td class="left"><tt>float&nbsp;(i16&nbsp;signext,&nbsp;i32&nbsp;*)&nbsp;*
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1236 </tt></td>
1237 <td class="left"><a href="#t_pointer">Pointer</a> to a function that takes
1238 an <tt>i16</tt> that should be sign extended and a
1239 <a href="#t_pointer">pointer</a> to <tt>i32</tt>, returning
1240 <tt>float</tt>.
1241 </td>
1242 </tr><tr class="layout">
1243 <td class="left"><tt>i32 (i8*, ...)</tt></td>
1244 <td class="left">A vararg function that takes at least one
1245 <a href="#t_pointer">pointer</a> to <tt>i8 </tt> (char in C),
1246 which returns an integer. This is the signature for <tt>printf</tt> in
1247 LLVM.
1248 </td>
1249 </tr>
1250</table>
1251
1252</div>
1253<!-- _______________________________________________________________________ -->
1254<div class="doc_subsubsection"> <a name="t_struct">Structure Type</a> </div>
1255<div class="doc_text">
1256<h5>Overview:</h5>
1257<p>The structure type is used to represent a collection of data members
1258together in memory. The packing of the field types is defined to match
1259the ABI of the underlying processor. The elements of a structure may
1260be any type that has a size.</p>
1261<p>Structures are accessed using '<tt><a href="#i_load">load</a></tt>
1262and '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a
1263field with the '<tt><a href="#i_getelementptr">getelementptr</a></tt>'
1264instruction.</p>
1265<h5>Syntax:</h5>
1266<pre> { &lt;type list&gt; }<br></pre>
1267<h5>Examples:</h5>
1268<table class="layout">
1269 <tr class="layout">
1270 <td class="left"><tt>{ i32, i32, i32 }</tt></td>
1271 <td class="left">A triple of three <tt>i32</tt> values</td>
1272 </tr><tr class="layout">
1273 <td class="left"><tt>{&nbsp;float,&nbsp;i32&nbsp;(i32)&nbsp;*&nbsp;}</tt></td>
1274 <td class="left">A pair, where the first element is a <tt>float</tt> and the
1275 second element is a <a href="#t_pointer">pointer</a> to a
1276 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
1277 an <tt>i32</tt>.</td>
1278 </tr>
1279</table>
1280</div>
1281
1282<!-- _______________________________________________________________________ -->
1283<div class="doc_subsubsection"> <a name="t_pstruct">Packed Structure Type</a>
1284</div>
1285<div class="doc_text">
1286<h5>Overview:</h5>
1287<p>The packed structure type is used to represent a collection of data members
1288together in memory. There is no padding between fields. Further, the alignment
1289of a packed structure is 1 byte. The elements of a packed structure may
1290be any type that has a size.</p>
1291<p>Structures are accessed using '<tt><a href="#i_load">load</a></tt>
1292and '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a
1293field with the '<tt><a href="#i_getelementptr">getelementptr</a></tt>'
1294instruction.</p>
1295<h5>Syntax:</h5>
1296<pre> &lt; { &lt;type list&gt; } &gt; <br></pre>
1297<h5>Examples:</h5>
1298<table class="layout">
1299 <tr class="layout">
1300 <td class="left"><tt>&lt; { i32, i32, i32 } &gt;</tt></td>
1301 <td class="left">A triple of three <tt>i32</tt> values</td>
1302 </tr><tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1303 <td class="left"><tt>&lt; { float, i32 (i32)* } &gt;</tt></td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1304 <td class="left">A pair, where the first element is a <tt>float</tt> and the
1305 second element is a <a href="#t_pointer">pointer</a> to a
1306 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
1307 an <tt>i32</tt>.</td>
1308 </tr>
1309</table>
1310</div>
1311
1312<!-- _______________________________________________________________________ -->
1313<div class="doc_subsubsection"> <a name="t_pointer">Pointer Type</a> </div>
1314<div class="doc_text">
1315<h5>Overview:</h5>
1316<p>As in many languages, the pointer type represents a pointer or
Christopher Lambdd0049d2007-12-11 09:31:00 +0000[diff] [blame]1317reference to another object, which must live in memory. Pointer types may have
1318an optional address space attribute defining the target-specific numbered
1319address space where the pointed-to object resides. The default address space is
1320zero.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1321<h5>Syntax:</h5>
1322<pre> &lt;type&gt; *<br></pre>
1323<h5>Examples:</h5>
1324<table class="layout">
1325 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1326 <td class="left"><tt>[4x i32]*</tt></td>
1327 <td class="left">A <a href="#t_pointer">pointer</a> to <a
1328 href="#t_array">array</a> of four <tt>i32</tt> values.</td>
1329 </tr>
1330 <tr class="layout">
1331 <td class="left"><tt>i32 (i32 *) *</tt></td>
1332 <td class="left"> A <a href="#t_pointer">pointer</a> to a <a
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1333 href="#t_function">function</a> that takes an <tt>i32*</tt>, returning an
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1334 <tt>i32</tt>.</td>
1335 </tr>
1336 <tr class="layout">
1337 <td class="left"><tt>i32 addrspace(5)*</tt></td>
1338 <td class="left">A <a href="#t_pointer">pointer</a> to an <tt>i32</tt> value
1339 that resides in address space #5.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1340 </tr>
1341</table>
1342</div>
1343
1344<!-- _______________________________________________________________________ -->
1345<div class="doc_subsubsection"> <a name="t_vector">Vector Type</a> </div>
1346<div class="doc_text">
1347
1348<h5>Overview:</h5>
1349
1350<p>A vector type is a simple derived type that represents a vector
1351of elements. Vector types are used when multiple primitive data
1352are operated in parallel using a single instruction (SIMD).
1353A vector type requires a size (number of
1354elements) and an underlying primitive data type. Vectors must have a power
1355of two length (1, 2, 4, 8, 16 ...). Vector types are
1356considered <a href="#t_firstclass">first class</a>.</p>
1357
1358<h5>Syntax:</h5>
1359
1360<pre>
1361 &lt; &lt;# elements&gt; x &lt;elementtype&gt; &gt;
1362</pre>
1363
1364<p>The number of elements is a constant integer value; elementtype may
1365be any integer or floating point type.</p>
1366
1367<h5>Examples:</h5>
1368
1369<table class="layout">
1370 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1371 <td class="left"><tt>&lt;4 x i32&gt;</tt></td>
1372 <td class="left">Vector of 4 32-bit integer values.</td>
1373 </tr>
1374 <tr class="layout">
1375 <td class="left"><tt>&lt;8 x float&gt;</tt></td>
1376 <td class="left">Vector of 8 32-bit floating-point values.</td>
1377 </tr>
1378 <tr class="layout">
1379 <td class="left"><tt>&lt;2 x i64&gt;</tt></td>
1380 <td class="left">Vector of 2 64-bit integer values.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1381 </tr>
1382</table>
1383</div>
1384
1385<!-- _______________________________________________________________________ -->
1386<div class="doc_subsubsection"> <a name="t_opaque">Opaque Type</a> </div>
1387<div class="doc_text">
1388
1389<h5>Overview:</h5>
1390
1391<p>Opaque types are used to represent unknown types in the system. This
Gordon Henriksenda0706e2007-10-14 00:34:53 +0000[diff] [blame]1392corresponds (for example) to the C notion of a forward declared structure type.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1393In LLVM, opaque types can eventually be resolved to any type (not just a
1394structure type).</p>
1395
1396<h5>Syntax:</h5>
1397
1398<pre>
1399 opaque
1400</pre>
1401
1402<h5>Examples:</h5>
1403
1404<table class="layout">
1405 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1406 <td class="left"><tt>opaque</tt></td>
1407 <td class="left">An opaque type.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1408 </tr>
1409</table>
1410</div>
1411
1412
1413<!-- *********************************************************************** -->
1414<div class="doc_section"> <a name="constants">Constants</a> </div>
1415<!-- *********************************************************************** -->
1416
1417<div class="doc_text">
1418
1419<p>LLVM has several different basic types of constants. This section describes
1420them all and their syntax.</p>
1421
1422</div>
1423
1424<!-- ======================================================================= -->
1425<div class="doc_subsection"><a name="simpleconstants">Simple Constants</a></div>
1426
1427<div class="doc_text">
1428
1429<dl>
1430 <dt><b>Boolean constants</b></dt>
1431
1432 <dd>The two strings '<tt>true</tt>' and '<tt>false</tt>' are both valid
1433 constants of the <tt><a href="#t_primitive">i1</a></tt> type.
1434 </dd>
1435
1436 <dt><b>Integer constants</b></dt>
1437
1438 <dd>Standard integers (such as '4') are constants of the <a
1439 href="#t_integer">integer</a> type. Negative numbers may be used with
1440 integer types.
1441 </dd>
1442
1443 <dt><b>Floating point constants</b></dt>
1444
1445 <dd>Floating point constants use standard decimal notation (e.g. 123.421),
1446 exponential notation (e.g. 1.23421e+2), or a more precise hexadecimal
1447 notation (see below). Floating point constants must have a <a
1448 href="#t_floating">floating point</a> type. </dd>
1449
1450 <dt><b>Null pointer constants</b></dt>
1451
1452 <dd>The identifier '<tt>null</tt>' is recognized as a null pointer constant
1453 and must be of <a href="#t_pointer">pointer type</a>.</dd>
1454
1455</dl>
1456
1457<p>The one non-intuitive notation for constants is the optional hexadecimal form
1458of floating point constants. For example, the form '<tt>double
14590x432ff973cafa8000</tt>' is equivalent to (but harder to read than) '<tt>double
14604.5e+15</tt>'. The only time hexadecimal floating point constants are required
1461(and the only time that they are generated by the disassembler) is when a
1462floating point constant must be emitted but it cannot be represented as a
1463decimal floating point number. For example, NaN's, infinities, and other
1464special values are represented in their IEEE hexadecimal format so that
1465assembly and disassembly do not cause any bits to change in the constants.</p>
1466
1467</div>
1468
1469<!-- ======================================================================= -->
1470<div class="doc_subsection"><a name="aggregateconstants">Aggregate Constants</a>
1471</div>
1472
1473<div class="doc_text">
1474<p>Aggregate constants arise from aggregation of simple constants
1475and smaller aggregate constants.</p>
1476
1477<dl>
1478 <dt><b>Structure constants</b></dt>
1479
1480 <dd>Structure constants are represented with notation similar to structure
1481 type definitions (a comma separated list of elements, surrounded by braces
Chris Lattner6c8de962007-12-25 20:34:52 +0000[diff] [blame]1482 (<tt>{}</tt>)). For example: "<tt>{ i32 4, float 17.0, i32* @G }</tt>",
1483 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]1484 must have <a href="#t_struct">structure type</a>, and the number and
1485 types of elements must match those specified by the type.
1486 </dd>
1487
1488 <dt><b>Array constants</b></dt>
1489
1490 <dd>Array constants are represented with notation similar to array type
1491 definitions (a comma separated list of elements, surrounded by square brackets
1492 (<tt>[]</tt>)). For example: "<tt>[ i32 42, i32 11, i32 74 ]</tt>". Array
1493 constants must have <a href="#t_array">array type</a>, and the number and
1494 types of elements must match those specified by the type.
1495 </dd>
1496
1497 <dt><b>Vector constants</b></dt>
1498
1499 <dd>Vector constants are represented with notation similar to vector type
1500 definitions (a comma separated list of elements, surrounded by
1501 less-than/greater-than's (<tt>&lt;&gt;</tt>)). For example: "<tt>&lt; i32 42,
1502 i32 11, i32 74, i32 100 &gt;</tt>". Vector constants must have <a
1503 href="#t_vector">vector type</a>, and the number and types of elements must
1504 match those specified by the type.
1505 </dd>
1506
1507 <dt><b>Zero initialization</b></dt>
1508
1509 <dd>The string '<tt>zeroinitializer</tt>' can be used to zero initialize a
1510 value to zero of <em>any</em> type, including scalar and aggregate types.
1511 This is often used to avoid having to print large zero initializers (e.g. for
1512 large arrays) and is always exactly equivalent to using explicit zero
1513 initializers.
1514 </dd>
1515</dl>
1516
1517</div>
1518
1519<!-- ======================================================================= -->
1520<div class="doc_subsection">
1521 <a name="globalconstants">Global Variable and Function Addresses</a>
1522</div>
1523
1524<div class="doc_text">
1525
1526<p>The addresses of <a href="#globalvars">global variables</a> and <a
1527href="#functionstructure">functions</a> are always implicitly valid (link-time)
1528constants. These constants are explicitly referenced when the <a
1529href="#identifiers">identifier for the global</a> is used and always have <a
1530href="#t_pointer">pointer</a> type. For example, the following is a legal LLVM
1531file:</p>
1532
1533<div class="doc_code">
1534<pre>
1535@X = global i32 17
1536@Y = global i32 42
1537@Z = global [2 x i32*] [ i32* @X, i32* @Y ]
1538</pre>
1539</div>
1540
1541</div>
1542
1543<!-- ======================================================================= -->
1544<div class="doc_subsection"><a name="undefvalues">Undefined Values</a></div>
1545<div class="doc_text">
1546 <p>The string '<tt>undef</tt>' is recognized as a type-less constant that has
1547 no specific value. Undefined values may be of any type and be used anywhere
1548 a constant is permitted.</p>
1549
1550 <p>Undefined values indicate to the compiler that the program is well defined
1551 no matter what value is used, giving the compiler more freedom to optimize.
1552 </p>
1553</div>
1554
1555<!-- ======================================================================= -->
1556<div class="doc_subsection"><a name="constantexprs">Constant Expressions</a>
1557</div>
1558
1559<div class="doc_text">
1560
1561<p>Constant expressions are used to allow expressions involving other constants
1562to be used as constants. Constant expressions may be of any <a
1563href="#t_firstclass">first class</a> type and may involve any LLVM operation
1564that does not have side effects (e.g. load and call are not supported). The
1565following is the syntax for constant expressions:</p>
1566
1567<dl>
1568 <dt><b><tt>trunc ( CST to TYPE )</tt></b></dt>
1569 <dd>Truncate a constant to another type. The bit size of CST must be larger
1570 than the bit size of TYPE. Both types must be integers.</dd>
1571
1572 <dt><b><tt>zext ( CST to TYPE )</tt></b></dt>
1573 <dd>Zero extend a constant to another type. The bit size of CST must be
1574 smaller or equal to the bit size of TYPE. Both types must be integers.</dd>
1575
1576 <dt><b><tt>sext ( CST to TYPE )</tt></b></dt>
1577 <dd>Sign extend a constant to another type. The bit size of CST must be
1578 smaller or equal to the bit size of TYPE. Both types must be integers.</dd>
1579
1580 <dt><b><tt>fptrunc ( CST to TYPE )</tt></b></dt>
1581 <dd>Truncate a floating point constant to another floating point type. The
1582 size of CST must be larger than the size of TYPE. Both types must be
1583 floating point.</dd>
1584
1585 <dt><b><tt>fpext ( CST to TYPE )</tt></b></dt>
1586 <dd>Floating point extend a constant to another type. The size of CST must be
1587 smaller or equal to the size of TYPE. Both types must be floating point.</dd>
1588
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]1589 <dt><b><tt>fptoui ( CST to TYPE )</tt></b></dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1590 <dd>Convert a floating point constant to the corresponding unsigned integer
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]1591 constant. TYPE must be a scalar or vector integer type. CST must be of scalar
1592 or vector floating point type. Both CST and TYPE must be scalars, or vectors
1593 of the same number of elements. If the value won't fit in the integer type,
1594 the results are undefined.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1595
1596 <dt><b><tt>fptosi ( CST to TYPE )</tt></b></dt>
1597 <dd>Convert a floating point constant to the corresponding signed integer
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]1598 constant. TYPE must be a scalar or vector integer type. CST must be of scalar
1599 or vector floating point type. Both CST and TYPE must be scalars, or vectors
1600 of the same number of elements. If the value won't fit in the integer type,
1601 the results are undefined.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1602
1603 <dt><b><tt>uitofp ( CST to TYPE )</tt></b></dt>
1604 <dd>Convert an unsigned integer constant to the corresponding floating point
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]1605 constant. TYPE must be a scalar or vector floating point type. CST must be of
1606 scalar or vector integer type. Both CST and TYPE must be scalars, or vectors
1607 of the same number of elements. If the value won't fit in the floating point
1608 type, the results are undefined.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1609
1610 <dt><b><tt>sitofp ( CST to TYPE )</tt></b></dt>
1611 <dd>Convert a signed integer constant to the corresponding floating point
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]1612 constant. TYPE must be a scalar or vector floating point type. CST must be of
1613 scalar or vector integer type. Both CST and TYPE must be scalars, or vectors
1614 of the same number of elements. If the value won't fit in the floating point
1615 type, the results are undefined.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1616
1617 <dt><b><tt>ptrtoint ( CST to TYPE )</tt></b></dt>
1618 <dd>Convert a pointer typed constant to the corresponding integer constant
1619 TYPE must be an integer type. CST must be of pointer type. The CST value is
1620 zero extended, truncated, or unchanged to make it fit in TYPE.</dd>
1621
1622 <dt><b><tt>inttoptr ( CST to TYPE )</tt></b></dt>
1623 <dd>Convert a integer constant to a pointer constant. TYPE must be a
1624 pointer type. CST must be of integer type. The CST value is zero extended,
1625 truncated, or unchanged to make it fit in a pointer size. This one is
1626 <i>really</i> dangerous!</dd>
1627
1628 <dt><b><tt>bitcast ( CST to TYPE )</tt></b></dt>
1629 <dd>Convert a constant, CST, to another TYPE. The size of CST and TYPE must be
1630 identical (same number of bits). The conversion is done as if the CST value
1631 was stored to memory and read back as TYPE. In other words, no bits change
1632 with this operator, just the type. This can be used for conversion of
1633 vector types to any other type, as long as they have the same bit width. For
1634 pointers it is only valid to cast to another pointer type.
1635 </dd>
1636
1637 <dt><b><tt>getelementptr ( CSTPTR, IDX0, IDX1, ... )</tt></b></dt>
1638
1639 <dd>Perform the <a href="#i_getelementptr">getelementptr operation</a> on
1640 constants. As with the <a href="#i_getelementptr">getelementptr</a>
1641 instruction, the index list may have zero or more indexes, which are required
1642 to make sense for the type of "CSTPTR".</dd>
1643
1644 <dt><b><tt>select ( COND, VAL1, VAL2 )</tt></b></dt>
1645
1646 <dd>Perform the <a href="#i_select">select operation</a> on
1647 constants.</dd>
1648
1649 <dt><b><tt>icmp COND ( VAL1, VAL2 )</tt></b></dt>
1650 <dd>Performs the <a href="#i_icmp">icmp operation</a> on constants.</dd>
1651
1652 <dt><b><tt>fcmp COND ( VAL1, VAL2 )</tt></b></dt>
1653 <dd>Performs the <a href="#i_fcmp">fcmp operation</a> on constants.</dd>
1654
1655 <dt><b><tt>extractelement ( VAL, IDX )</tt></b></dt>
1656
1657 <dd>Perform the <a href="#i_extractelement">extractelement
1658 operation</a> on constants.
1659
1660 <dt><b><tt>insertelement ( VAL, ELT, IDX )</tt></b></dt>
1661
1662 <dd>Perform the <a href="#i_insertelement">insertelement
1663 operation</a> on constants.</dd>
1664
1665
1666 <dt><b><tt>shufflevector ( VEC1, VEC2, IDXMASK )</tt></b></dt>
1667
1668 <dd>Perform the <a href="#i_shufflevector">shufflevector
1669 operation</a> on constants.</dd>
1670
1671 <dt><b><tt>OPCODE ( LHS, RHS )</tt></b></dt>
1672
1673 <dd>Perform the specified operation of the LHS and RHS constants. OPCODE may
1674 be any of the <a href="#binaryops">binary</a> or <a href="#bitwiseops">bitwise
1675 binary</a> operations. The constraints on operands are the same as those for
1676 the corresponding instruction (e.g. no bitwise operations on floating point
1677 values are allowed).</dd>
1678</dl>
1679</div>
1680
1681<!-- *********************************************************************** -->
1682<div class="doc_section"> <a name="othervalues">Other Values</a> </div>
1683<!-- *********************************************************************** -->
1684
1685<!-- ======================================================================= -->
1686<div class="doc_subsection">
1687<a name="inlineasm">Inline Assembler Expressions</a>
1688</div>
1689
1690<div class="doc_text">
1691
1692<p>
1693LLVM supports inline assembler expressions (as opposed to <a href="#moduleasm">
1694Module-Level Inline Assembly</a>) through the use of a special value. This
1695value represents the inline assembler as a string (containing the instructions
1696to emit), a list of operand constraints (stored as a string), and a flag that
1697indicates whether or not the inline asm expression has side effects. An example
1698inline assembler expression is:
1699</p>
1700
1701<div class="doc_code">
1702<pre>
1703i32 (i32) asm "bswap $0", "=r,r"
1704</pre>
1705</div>
1706
1707<p>
1708Inline assembler expressions may <b>only</b> be used as the callee operand of
1709a <a href="#i_call"><tt>call</tt> instruction</a>. Thus, typically we have:
1710</p>
1711
1712<div class="doc_code">
1713<pre>
1714%X = call i32 asm "<a href="#int_bswap">bswap</a> $0", "=r,r"(i32 %Y)
1715</pre>
1716</div>
1717
1718<p>
1719Inline asms with side effects not visible in the constraint list must be marked
1720as having side effects. This is done through the use of the
1721'<tt>sideeffect</tt>' keyword, like so:
1722</p>
1723
1724<div class="doc_code">
1725<pre>
1726call void asm sideeffect "eieio", ""()
1727</pre>
1728</div>
1729
1730<p>TODO: The format of the asm and constraints string still need to be
1731documented here. Constraints on what can be done (e.g. duplication, moving, etc
1732need to be documented).
1733</p>
1734
1735</div>
1736
1737<!-- *********************************************************************** -->
1738<div class="doc_section"> <a name="instref">Instruction Reference</a> </div>
1739<!-- *********************************************************************** -->
1740
1741<div class="doc_text">
1742
1743<p>The LLVM instruction set consists of several different
1744classifications of instructions: <a href="#terminators">terminator
1745instructions</a>, <a href="#binaryops">binary instructions</a>,
1746<a href="#bitwiseops">bitwise binary instructions</a>, <a
1747 href="#memoryops">memory instructions</a>, and <a href="#otherops">other
1748instructions</a>.</p>
1749
1750</div>
1751
1752<!-- ======================================================================= -->
1753<div class="doc_subsection"> <a name="terminators">Terminator
1754Instructions</a> </div>
1755
1756<div class="doc_text">
1757
1758<p>As mentioned <a href="#functionstructure">previously</a>, every
1759basic block in a program ends with a "Terminator" instruction, which
1760indicates which block should be executed after the current block is
1761finished. These terminator instructions typically yield a '<tt>void</tt>'
1762value: they produce control flow, not values (the one exception being
1763the '<a href="#i_invoke"><tt>invoke</tt></a>' instruction).</p>
1764<p>There are six different terminator instructions: the '<a
1765 href="#i_ret"><tt>ret</tt></a>' instruction, the '<a href="#i_br"><tt>br</tt></a>'
1766instruction, the '<a href="#i_switch"><tt>switch</tt></a>' instruction,
1767the '<a href="#i_invoke"><tt>invoke</tt></a>' instruction, the '<a
1768 href="#i_unwind"><tt>unwind</tt></a>' instruction, and the '<a
1769 href="#i_unreachable"><tt>unreachable</tt></a>' instruction.</p>
1770
1771</div>
1772
1773<!-- _______________________________________________________________________ -->
1774<div class="doc_subsubsection"> <a name="i_ret">'<tt>ret</tt>'
1775Instruction</a> </div>
1776<div class="doc_text">
1777<h5>Syntax:</h5>
1778<pre> ret &lt;type&gt; &lt;value&gt; <i>; Return a value from a non-void function</i>
1779 ret void <i>; Return from void function</i>
1780</pre>
1781<h5>Overview:</h5>
1782<p>The '<tt>ret</tt>' instruction is used to return control flow (and a
1783value) from a function back to the caller.</p>
1784<p>There are two forms of the '<tt>ret</tt>' instruction: one that
1785returns a value and then causes control flow, and one that just causes
1786control flow to occur.</p>
1787<h5>Arguments:</h5>
1788<p>The '<tt>ret</tt>' instruction may return any '<a
1789 href="#t_firstclass">first class</a>' type. Notice that a function is
1790not <a href="#wellformed">well formed</a> if there exists a '<tt>ret</tt>'
1791instruction inside of the function that returns a value that does not
1792match the return type of the function.</p>
1793<h5>Semantics:</h5>
1794<p>When the '<tt>ret</tt>' instruction is executed, control flow
1795returns back to the calling function's context. If the caller is a "<a
1796 href="#i_call"><tt>call</tt></a>" instruction, execution continues at
1797the instruction after the call. If the caller was an "<a
1798 href="#i_invoke"><tt>invoke</tt></a>" instruction, execution continues
1799at the beginning of the "normal" destination block. If the instruction
1800returns a value, that value shall set the call or invoke instruction's
1801return value.</p>
1802<h5>Example:</h5>
1803<pre> ret i32 5 <i>; Return an integer value of 5</i>
1804 ret void <i>; Return from a void function</i>
1805</pre>
1806</div>
1807<!-- _______________________________________________________________________ -->
1808<div class="doc_subsubsection"> <a name="i_br">'<tt>br</tt>' Instruction</a> </div>
1809<div class="doc_text">
1810<h5>Syntax:</h5>
1811<pre> br i1 &lt;cond&gt;, label &lt;iftrue&gt;, label &lt;iffalse&gt;<br> br label &lt;dest&gt; <i>; Unconditional branch</i>
1812</pre>
1813<h5>Overview:</h5>
1814<p>The '<tt>br</tt>' instruction is used to cause control flow to
1815transfer to a different basic block in the current function. There are
1816two forms of this instruction, corresponding to a conditional branch
1817and an unconditional branch.</p>
1818<h5>Arguments:</h5>
1819<p>The conditional branch form of the '<tt>br</tt>' instruction takes a
1820single '<tt>i1</tt>' value and two '<tt>label</tt>' values. The
1821unconditional form of the '<tt>br</tt>' instruction takes a single
1822'<tt>label</tt>' value as a target.</p>
1823<h5>Semantics:</h5>
1824<p>Upon execution of a conditional '<tt>br</tt>' instruction, the '<tt>i1</tt>'
1825argument is evaluated. If the value is <tt>true</tt>, control flows
1826to the '<tt>iftrue</tt>' <tt>label</tt> argument. If "cond" is <tt>false</tt>,
1827control flows to the '<tt>iffalse</tt>' <tt>label</tt> argument.</p>
1828<h5>Example:</h5>
1829<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
1830 href="#i_ret">ret</a> i32 1<br>IfUnequal:<br> <a href="#i_ret">ret</a> i32 0<br></pre>
1831</div>
1832<!-- _______________________________________________________________________ -->
1833<div class="doc_subsubsection">
1834 <a name="i_switch">'<tt>switch</tt>' Instruction</a>
1835</div>
1836
1837<div class="doc_text">
1838<h5>Syntax:</h5>
1839
1840<pre>
1841 switch &lt;intty&gt; &lt;value&gt;, label &lt;defaultdest&gt; [ &lt;intty&gt; &lt;val&gt;, label &lt;dest&gt; ... ]
1842</pre>
1843
1844<h5>Overview:</h5>
1845
1846<p>The '<tt>switch</tt>' instruction is used to transfer control flow to one of
1847several different places. It is a generalization of the '<tt>br</tt>'
1848instruction, allowing a branch to occur to one of many possible
1849destinations.</p>
1850
1851
1852<h5>Arguments:</h5>
1853
1854<p>The '<tt>switch</tt>' instruction uses three parameters: an integer
1855comparison value '<tt>value</tt>', a default '<tt>label</tt>' destination, and
1856an array of pairs of comparison value constants and '<tt>label</tt>'s. The
1857table is not allowed to contain duplicate constant entries.</p>
1858
1859<h5>Semantics:</h5>
1860
1861<p>The <tt>switch</tt> instruction specifies a table of values and
1862destinations. When the '<tt>switch</tt>' instruction is executed, this
1863table is searched for the given value. If the value is found, control flow is
1864transfered to the corresponding destination; otherwise, control flow is
1865transfered to the default destination.</p>
1866
1867<h5>Implementation:</h5>
1868
1869<p>Depending on properties of the target machine and the particular
1870<tt>switch</tt> instruction, this instruction may be code generated in different
1871ways. For example, it could be generated as a series of chained conditional
1872branches or with a lookup table.</p>
1873
1874<h5>Example:</h5>
1875
1876<pre>
1877 <i>; Emulate a conditional br instruction</i>
1878 %Val = <a href="#i_zext">zext</a> i1 %value to i32
1879 switch i32 %Val, label %truedest [i32 0, label %falsedest ]
1880
1881 <i>; Emulate an unconditional br instruction</i>
1882 switch i32 0, label %dest [ ]
1883
1884 <i>; Implement a jump table:</i>
1885 switch i32 %val, label %otherwise [ i32 0, label %onzero
1886 i32 1, label %onone
1887 i32 2, label %ontwo ]
1888</pre>
1889</div>
1890
1891<!-- _______________________________________________________________________ -->
1892<div class="doc_subsubsection">
1893 <a name="i_invoke">'<tt>invoke</tt>' Instruction</a>
1894</div>
1895
1896<div class="doc_text">
1897
1898<h5>Syntax:</h5>
1899
1900<pre>
1901 &lt;result&gt; = invoke [<a href="#callingconv">cconv</a>] &lt;ptr to function ty&gt; %&lt;function ptr val&gt;(&lt;function args&gt;)
1902 to label &lt;normal label&gt; unwind label &lt;exception label&gt;
1903</pre>
1904
1905<h5>Overview:</h5>
1906
1907<p>The '<tt>invoke</tt>' instruction causes control to transfer to a specified
1908function, with the possibility of control flow transfer to either the
1909'<tt>normal</tt>' label or the
1910'<tt>exception</tt>' label. If the callee function returns with the
1911"<tt><a href="#i_ret">ret</a></tt>" instruction, control flow will return to the
1912"normal" label. If the callee (or any indirect callees) returns with the "<a
1913href="#i_unwind"><tt>unwind</tt></a>" instruction, control is interrupted and
1914continued at the dynamically nearest "exception" label.</p>
1915
1916<h5>Arguments:</h5>
1917
1918<p>This instruction requires several arguments:</p>
1919
1920<ol>
1921 <li>
1922 The optional "cconv" marker indicates which <a href="#callingconv">calling
1923 convention</a> the call should use. If none is specified, the call defaults
1924 to using C calling conventions.
1925 </li>
1926 <li>'<tt>ptr to function ty</tt>': shall be the signature of the pointer to
1927 function value being invoked. In most cases, this is a direct function
1928 invocation, but indirect <tt>invoke</tt>s are just as possible, branching off
1929 an arbitrary pointer to function value.
1930 </li>
1931
1932 <li>'<tt>function ptr val</tt>': An LLVM value containing a pointer to a
1933 function to be invoked. </li>
1934
1935 <li>'<tt>function args</tt>': argument list whose types match the function
1936 signature argument types. If the function signature indicates the function
1937 accepts a variable number of arguments, the extra arguments can be
1938 specified. </li>
1939
1940 <li>'<tt>normal label</tt>': the label reached when the called function
1941 executes a '<tt><a href="#i_ret">ret</a></tt>' instruction. </li>
1942
1943 <li>'<tt>exception label</tt>': the label reached when a callee returns with
1944 the <a href="#i_unwind"><tt>unwind</tt></a> instruction. </li>
1945
1946</ol>
1947
1948<h5>Semantics:</h5>
1949
1950<p>This instruction is designed to operate as a standard '<tt><a
1951href="#i_call">call</a></tt>' instruction in most regards. The primary
1952difference is that it establishes an association with a label, which is used by
1953the runtime library to unwind the stack.</p>
1954
1955<p>This instruction is used in languages with destructors to ensure that proper
1956cleanup is performed in the case of either a <tt>longjmp</tt> or a thrown
1957exception. Additionally, this is important for implementation of
1958'<tt>catch</tt>' clauses in high-level languages that support them.</p>
1959
1960<h5>Example:</h5>
1961<pre>
1962 %retval = invoke i32 %Test(i32 15) to label %Continue
1963 unwind label %TestCleanup <i>; {i32}:retval set</i>
1964 %retval = invoke <a href="#callingconv">coldcc</a> i32 %Test(i32 15) to label %Continue
1965 unwind label %TestCleanup <i>; {i32}:retval set</i>
1966</pre>
1967</div>
1968
1969
1970<!-- _______________________________________________________________________ -->
1971
1972<div class="doc_subsubsection"> <a name="i_unwind">'<tt>unwind</tt>'
1973Instruction</a> </div>
1974
1975<div class="doc_text">
1976
1977<h5>Syntax:</h5>
1978<pre>
1979 unwind
1980</pre>
1981
1982<h5>Overview:</h5>
1983
1984<p>The '<tt>unwind</tt>' instruction unwinds the stack, continuing control flow
1985at the first callee in the dynamic call stack which used an <a
1986href="#i_invoke"><tt>invoke</tt></a> instruction to perform the call. This is
1987primarily used to implement exception handling.</p>
1988
1989<h5>Semantics:</h5>
1990
1991<p>The '<tt>unwind</tt>' intrinsic causes execution of the current function to
1992immediately halt. The dynamic call stack is then searched for the first <a
1993href="#i_invoke"><tt>invoke</tt></a> instruction on the call stack. Once found,
1994execution continues at the "exceptional" destination block specified by the
1995<tt>invoke</tt> instruction. If there is no <tt>invoke</tt> instruction in the
1996dynamic call chain, undefined behavior results.</p>
1997</div>
1998
1999<!-- _______________________________________________________________________ -->
2000
2001<div class="doc_subsubsection"> <a name="i_unreachable">'<tt>unreachable</tt>'
2002Instruction</a> </div>
2003
2004<div class="doc_text">
2005
2006<h5>Syntax:</h5>
2007<pre>
2008 unreachable
2009</pre>
2010
2011<h5>Overview:</h5>
2012
2013<p>The '<tt>unreachable</tt>' instruction has no defined semantics. This
2014instruction is used to inform the optimizer that a particular portion of the
2015code is not reachable. This can be used to indicate that the code after a
2016no-return function cannot be reached, and other facts.</p>
2017
2018<h5>Semantics:</h5>
2019
2020<p>The '<tt>unreachable</tt>' instruction has no defined semantics.</p>
2021</div>
2022
2023
2024
2025<!-- ======================================================================= -->
2026<div class="doc_subsection"> <a name="binaryops">Binary Operations</a> </div>
2027<div class="doc_text">
2028<p>Binary operators are used to do most of the computation in a
2029program. They require two operands, execute an operation on them, and
2030produce a single value. The operands might represent
2031multiple data, as is the case with the <a href="#t_vector">vector</a> data type.
2032The result value of a binary operator is not
2033necessarily the same type as its operands.</p>
2034<p>There are several different binary operators:</p>
2035</div>
2036<!-- _______________________________________________________________________ -->
2037<div class="doc_subsubsection"> <a name="i_add">'<tt>add</tt>'
2038Instruction</a> </div>
2039<div class="doc_text">
2040<h5>Syntax:</h5>
2041<pre> &lt;result&gt; = add &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
2042</pre>
2043<h5>Overview:</h5>
2044<p>The '<tt>add</tt>' instruction returns the sum of its two operands.</p>
2045<h5>Arguments:</h5>
2046<p>The two arguments to the '<tt>add</tt>' instruction must be either <a
2047 href="#t_integer">integer</a> or <a href="#t_floating">floating point</a> values.
2048 This instruction can also take <a href="#t_vector">vector</a> versions of the values.
2049Both arguments must have identical types.</p>
2050<h5>Semantics:</h5>
2051<p>The value produced is the integer or floating point sum of the two
2052operands.</p>
2053<h5>Example:</h5>
2054<pre> &lt;result&gt; = add i32 4, %var <i>; yields {i32}:result = 4 + %var</i>
2055</pre>
2056</div>
2057<!-- _______________________________________________________________________ -->
2058<div class="doc_subsubsection"> <a name="i_sub">'<tt>sub</tt>'
2059Instruction</a> </div>
2060<div class="doc_text">
2061<h5>Syntax:</h5>
2062<pre> &lt;result&gt; = sub &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
2063</pre>
2064<h5>Overview:</h5>
2065<p>The '<tt>sub</tt>' instruction returns the difference of its two
2066operands.</p>
2067<p>Note that the '<tt>sub</tt>' instruction is used to represent the '<tt>neg</tt>'
2068instruction present in most other intermediate representations.</p>
2069<h5>Arguments:</h5>
2070<p>The two arguments to the '<tt>sub</tt>' instruction must be either <a
2071 href="#t_integer">integer</a> or <a href="#t_floating">floating point</a>
2072values.
2073This instruction can also take <a href="#t_vector">vector</a> versions of the values.
2074Both arguments must have identical types.</p>
2075<h5>Semantics:</h5>
2076<p>The value produced is the integer or floating point difference of
2077the two operands.</p>
2078<h5>Example:</h5>
2079<pre>
2080 &lt;result&gt; = sub i32 4, %var <i>; yields {i32}:result = 4 - %var</i>
2081 &lt;result&gt; = sub i32 0, %val <i>; yields {i32}:result = -%var</i>
2082</pre>
2083</div>
2084<!-- _______________________________________________________________________ -->
2085<div class="doc_subsubsection"> <a name="i_mul">'<tt>mul</tt>'
2086Instruction</a> </div>
2087<div class="doc_text">
2088<h5>Syntax:</h5>
2089<pre> &lt;result&gt; = mul &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
2090</pre>
2091<h5>Overview:</h5>
2092<p>The '<tt>mul</tt>' instruction returns the product of its two
2093operands.</p>
2094<h5>Arguments:</h5>
2095<p>The two arguments to the '<tt>mul</tt>' instruction must be either <a
2096 href="#t_integer">integer</a> or <a href="#t_floating">floating point</a>
2097values.
2098This instruction can also take <a href="#t_vector">vector</a> versions of the values.
2099Both arguments must have identical types.</p>
2100<h5>Semantics:</h5>
2101<p>The value produced is the integer or floating point product of the
2102two operands.</p>
2103<p>Because the operands are the same width, the result of an integer
2104multiplication is the same whether the operands should be deemed unsigned or
2105signed.</p>
2106<h5>Example:</h5>
2107<pre> &lt;result&gt; = mul i32 4, %var <i>; yields {i32}:result = 4 * %var</i>
2108</pre>
2109</div>
2110<!-- _______________________________________________________________________ -->
2111<div class="doc_subsubsection"> <a name="i_udiv">'<tt>udiv</tt>' Instruction
2112</a></div>
2113<div class="doc_text">
2114<h5>Syntax:</h5>
2115<pre> &lt;result&gt; = udiv &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
2116</pre>
2117<h5>Overview:</h5>
2118<p>The '<tt>udiv</tt>' instruction returns the quotient of its two
2119operands.</p>
2120<h5>Arguments:</h5>
2121<p>The two arguments to the '<tt>udiv</tt>' instruction must be
2122<a href="#t_integer">integer</a> values. Both arguments must have identical
2123types. This instruction can also take <a href="#t_vector">vector</a> versions
2124of the values in which case the elements must be integers.</p>
2125<h5>Semantics:</h5>
2126<p>The value produced is the unsigned integer quotient of the two operands. This
2127instruction always performs an unsigned division operation, regardless of
2128whether the arguments are unsigned or not.</p>
2129<h5>Example:</h5>
2130<pre> &lt;result&gt; = udiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
2131</pre>
2132</div>
2133<!-- _______________________________________________________________________ -->
2134<div class="doc_subsubsection"> <a name="i_sdiv">'<tt>sdiv</tt>' Instruction
2135</a> </div>
2136<div class="doc_text">
2137<h5>Syntax:</h5>
2138<pre> &lt;result&gt; = sdiv &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
2139</pre>
2140<h5>Overview:</h5>
2141<p>The '<tt>sdiv</tt>' instruction returns the quotient of its two
2142operands.</p>
2143<h5>Arguments:</h5>
2144<p>The two arguments to the '<tt>sdiv</tt>' instruction must be
2145<a href="#t_integer">integer</a> values. Both arguments must have identical
2146types. This instruction can also take <a href="#t_vector">vector</a> versions
2147of the values in which case the elements must be integers.</p>
2148<h5>Semantics:</h5>
2149<p>The value produced is the signed integer quotient of the two operands. This
2150instruction always performs a signed division operation, regardless of whether
2151the arguments are signed or not.</p>
2152<h5>Example:</h5>
2153<pre> &lt;result&gt; = sdiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
2154</pre>
2155</div>
2156<!-- _______________________________________________________________________ -->
2157<div class="doc_subsubsection"> <a name="i_fdiv">'<tt>fdiv</tt>'
2158Instruction</a> </div>
2159<div class="doc_text">
2160<h5>Syntax:</h5>
2161<pre> &lt;result&gt; = fdiv &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
2162</pre>
2163<h5>Overview:</h5>
2164<p>The '<tt>fdiv</tt>' instruction returns the quotient of its two
2165operands.</p>
2166<h5>Arguments:</h5>
2167<p>The two arguments to the '<tt>fdiv</tt>' instruction must be
2168<a href="#t_floating">floating point</a> values. Both arguments must have
2169identical types. This instruction can also take <a href="#t_vector">vector</a>
2170versions of floating point values.</p>
2171<h5>Semantics:</h5>
2172<p>The value produced is the floating point quotient of the two operands.</p>
2173<h5>Example:</h5>
2174<pre> &lt;result&gt; = fdiv float 4.0, %var <i>; yields {float}:result = 4.0 / %var</i>
2175</pre>
2176</div>
2177<!-- _______________________________________________________________________ -->
2178<div class="doc_subsubsection"> <a name="i_urem">'<tt>urem</tt>' Instruction</a>
2179</div>
2180<div class="doc_text">
2181<h5>Syntax:</h5>
2182<pre> &lt;result&gt; = urem &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
2183</pre>
2184<h5>Overview:</h5>
2185<p>The '<tt>urem</tt>' instruction returns the remainder from the
2186unsigned division of its two arguments.</p>
2187<h5>Arguments:</h5>
2188<p>The two arguments to the '<tt>urem</tt>' instruction must be
2189<a href="#t_integer">integer</a> values. Both arguments must have identical
Dan Gohman3e3fd8c2007-11-05 23:35:22 +0000[diff] [blame]2190types. This instruction can also take <a href="#t_vector">vector</a> versions
2191of the values in which case the elements must be integers.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2192<h5>Semantics:</h5>
2193<p>This instruction returns the unsigned integer <i>remainder</i> of a division.
2194This instruction always performs an unsigned division to get the remainder,
2195regardless of whether the arguments are unsigned or not.</p>
2196<h5>Example:</h5>
2197<pre> &lt;result&gt; = urem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
2198</pre>
2199
2200</div>
2201<!-- _______________________________________________________________________ -->
2202<div class="doc_subsubsection"> <a name="i_srem">'<tt>srem</tt>'
2203Instruction</a> </div>
2204<div class="doc_text">
2205<h5>Syntax:</h5>
2206<pre> &lt;result&gt; = srem &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
2207</pre>
2208<h5>Overview:</h5>
2209<p>The '<tt>srem</tt>' instruction returns the remainder from the
Dan Gohman3e3fd8c2007-11-05 23:35:22 +0000[diff] [blame]2210signed division of its two operands. This instruction can also take
2211<a href="#t_vector">vector</a> versions of the values in which case
2212the elements must be integers.</p>
Chris Lattner08497ce2008-01-04 04:33:49 +0000[diff] [blame]2213
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2214<h5>Arguments:</h5>
2215<p>The two arguments to the '<tt>srem</tt>' instruction must be
2216<a href="#t_integer">integer</a> values. Both arguments must have identical
2217types.</p>
2218<h5>Semantics:</h5>
2219<p>This instruction returns the <i>remainder</i> of a division (where the result
2220has the same sign as the dividend, <tt>var1</tt>), not the <i>modulo</i>
2221operator (where the result has the same sign as the divisor, <tt>var2</tt>) of
2222a value. For more information about the difference, see <a
2223 href="http://mathforum.org/dr.math/problems/anne.4.28.99.html">The
2224Math Forum</a>. For a table of how this is implemented in various languages,
2225please see <a href="http://en.wikipedia.org/wiki/Modulo_operation">
2226Wikipedia: modulo operation</a>.</p>
2227<h5>Example:</h5>
2228<pre> &lt;result&gt; = srem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
2229</pre>
2230
2231</div>
2232<!-- _______________________________________________________________________ -->
2233<div class="doc_subsubsection"> <a name="i_frem">'<tt>frem</tt>'
2234Instruction</a> </div>
2235<div class="doc_text">
2236<h5>Syntax:</h5>
2237<pre> &lt;result&gt; = frem &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
2238</pre>
2239<h5>Overview:</h5>
2240<p>The '<tt>frem</tt>' instruction returns the remainder from the
2241division of its two operands.</p>
2242<h5>Arguments:</h5>
2243<p>The two arguments to the '<tt>frem</tt>' instruction must be
2244<a href="#t_floating">floating point</a> values. Both arguments must have
Dan Gohman3e3fd8c2007-11-05 23:35:22 +0000[diff] [blame]2245identical types. This instruction can also take <a href="#t_vector">vector</a>
2246versions of floating point values.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2247<h5>Semantics:</h5>
2248<p>This instruction returns the <i>remainder</i> of a division.</p>
2249<h5>Example:</h5>
2250<pre> &lt;result&gt; = frem float 4.0, %var <i>; yields {float}:result = 4.0 % %var</i>
2251</pre>
2252</div>
2253
2254<!-- ======================================================================= -->
2255<div class="doc_subsection"> <a name="bitwiseops">Bitwise Binary
2256Operations</a> </div>
2257<div class="doc_text">
2258<p>Bitwise binary operators are used to do various forms of
2259bit-twiddling in a program. They are generally very efficient
2260instructions and can commonly be strength reduced from other
2261instructions. They require two operands, execute an operation on them,
2262and produce a single value. The resulting value of the bitwise binary
2263operators is always the same type as its first operand.</p>
2264</div>
2265
2266<!-- _______________________________________________________________________ -->
2267<div class="doc_subsubsection"> <a name="i_shl">'<tt>shl</tt>'
2268Instruction</a> </div>
2269<div class="doc_text">
2270<h5>Syntax:</h5>
2271<pre> &lt;result&gt; = shl &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
2272</pre>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2273
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2274<h5>Overview:</h5>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2275
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2276<p>The '<tt>shl</tt>' instruction returns the first operand shifted to
2277the left a specified number of bits.</p>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2278
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2279<h5>Arguments:</h5>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2280
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2281<p>Both arguments to the '<tt>shl</tt>' instruction must be the same <a
2282 href="#t_integer">integer</a> type.</p>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2283
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2284<h5>Semantics:</h5>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2285
2286<p>The value produced is <tt>var1</tt> * 2<sup><tt>var2</tt></sup>. If
2287<tt>var2</tt> is (statically or dynamically) equal to or larger than the number
2288of bits in <tt>var1</tt>, the result is undefined.</p>
2289
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2290<h5>Example:</h5><pre>
2291 &lt;result&gt; = shl i32 4, %var <i>; yields {i32}: 4 &lt;&lt; %var</i>
2292 &lt;result&gt; = shl i32 4, 2 <i>; yields {i32}: 16</i>
2293 &lt;result&gt; = shl i32 1, 10 <i>; yields {i32}: 1024</i>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2294 &lt;result&gt; = shl i32 1, 32 <i>; undefined</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2295</pre>
2296</div>
2297<!-- _______________________________________________________________________ -->
2298<div class="doc_subsubsection"> <a name="i_lshr">'<tt>lshr</tt>'
2299Instruction</a> </div>
2300<div class="doc_text">
2301<h5>Syntax:</h5>
2302<pre> &lt;result&gt; = lshr &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
2303</pre>
2304
2305<h5>Overview:</h5>
2306<p>The '<tt>lshr</tt>' instruction (logical shift right) returns the first
2307operand shifted to the right a specified number of bits with zero fill.</p>
2308
2309<h5>Arguments:</h5>
2310<p>Both arguments to the '<tt>lshr</tt>' instruction must be the same
2311<a href="#t_integer">integer</a> type.</p>
2312
2313<h5>Semantics:</h5>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2314
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2315<p>This instruction always performs a logical shift right operation. The most
2316significant bits of the result will be filled with zero bits after the
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2317shift. If <tt>var2</tt> is (statically or dynamically) equal to or larger than
2318the number of bits in <tt>var1</tt>, the result is undefined.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2319
2320<h5>Example:</h5>
2321<pre>
2322 &lt;result&gt; = lshr i32 4, 1 <i>; yields {i32}:result = 2</i>
2323 &lt;result&gt; = lshr i32 4, 2 <i>; yields {i32}:result = 1</i>
2324 &lt;result&gt; = lshr i8 4, 3 <i>; yields {i8}:result = 0</i>
2325 &lt;result&gt; = lshr i8 -2, 1 <i>; yields {i8}:result = 0x7FFFFFFF </i>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2326 &lt;result&gt; = lshr i32 1, 32 <i>; undefined</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2327</pre>
2328</div>
2329
2330<!-- _______________________________________________________________________ -->
2331<div class="doc_subsubsection"> <a name="i_ashr">'<tt>ashr</tt>'
2332Instruction</a> </div>
2333<div class="doc_text">
2334
2335<h5>Syntax:</h5>
2336<pre> &lt;result&gt; = ashr &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
2337</pre>
2338
2339<h5>Overview:</h5>
2340<p>The '<tt>ashr</tt>' instruction (arithmetic shift right) returns the first
2341operand shifted to the right a specified number of bits with sign extension.</p>
2342
2343<h5>Arguments:</h5>
2344<p>Both arguments to the '<tt>ashr</tt>' instruction must be the same
2345<a href="#t_integer">integer</a> type.</p>
2346
2347<h5>Semantics:</h5>
2348<p>This instruction always performs an arithmetic shift right operation,
2349The most significant bits of the result will be filled with the sign bit
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2350of <tt>var1</tt>. If <tt>var2</tt> is (statically or dynamically) equal to or
2351larger than the number of bits in <tt>var1</tt>, the result is undefined.
2352</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2353
2354<h5>Example:</h5>
2355<pre>
2356 &lt;result&gt; = ashr i32 4, 1 <i>; yields {i32}:result = 2</i>
2357 &lt;result&gt; = ashr i32 4, 2 <i>; yields {i32}:result = 1</i>
2358 &lt;result&gt; = ashr i8 4, 3 <i>; yields {i8}:result = 0</i>
2359 &lt;result&gt; = ashr i8 -2, 1 <i>; yields {i8}:result = -1</i>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2360 &lt;result&gt; = ashr i32 1, 32 <i>; undefined</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2361</pre>
2362</div>
2363
2364<!-- _______________________________________________________________________ -->
2365<div class="doc_subsubsection"> <a name="i_and">'<tt>and</tt>'
2366Instruction</a> </div>
2367<div class="doc_text">
2368<h5>Syntax:</h5>
2369<pre> &lt;result&gt; = and &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
2370</pre>
2371<h5>Overview:</h5>
2372<p>The '<tt>and</tt>' instruction returns the bitwise logical and of
2373its two operands.</p>
2374<h5>Arguments:</h5>
2375<p>The two arguments to the '<tt>and</tt>' instruction must be <a
2376 href="#t_integer">integer</a> values. Both arguments must have
2377identical types.</p>
2378<h5>Semantics:</h5>
2379<p>The truth table used for the '<tt>and</tt>' instruction is:</p>
2380<p> </p>
2381<div style="align: center">
2382<table border="1" cellspacing="0" cellpadding="4">
2383 <tbody>
2384 <tr>
2385 <td>In0</td>
2386 <td>In1</td>
2387 <td>Out</td>
2388 </tr>
2389 <tr>
2390 <td>0</td>
2391 <td>0</td>
2392 <td>0</td>
2393 </tr>
2394 <tr>
2395 <td>0</td>
2396 <td>1</td>
2397 <td>0</td>
2398 </tr>
2399 <tr>
2400 <td>1</td>
2401 <td>0</td>
2402 <td>0</td>
2403 </tr>
2404 <tr>
2405 <td>1</td>
2406 <td>1</td>
2407 <td>1</td>
2408 </tr>
2409 </tbody>
2410</table>
2411</div>
2412<h5>Example:</h5>
2413<pre> &lt;result&gt; = and i32 4, %var <i>; yields {i32}:result = 4 &amp; %var</i>
2414 &lt;result&gt; = and i32 15, 40 <i>; yields {i32}:result = 8</i>
2415 &lt;result&gt; = and i32 4, 8 <i>; yields {i32}:result = 0</i>
2416</pre>
2417</div>
2418<!-- _______________________________________________________________________ -->
2419<div class="doc_subsubsection"> <a name="i_or">'<tt>or</tt>' Instruction</a> </div>
2420<div class="doc_text">
2421<h5>Syntax:</h5>
2422<pre> &lt;result&gt; = or &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
2423</pre>
2424<h5>Overview:</h5>
2425<p>The '<tt>or</tt>' instruction returns the bitwise logical inclusive
2426or of its two operands.</p>
2427<h5>Arguments:</h5>
2428<p>The two arguments to the '<tt>or</tt>' instruction must be <a
2429 href="#t_integer">integer</a> values. Both arguments must have
2430identical types.</p>
2431<h5>Semantics:</h5>
2432<p>The truth table used for the '<tt>or</tt>' instruction is:</p>
2433<p> </p>
2434<div style="align: center">
2435<table border="1" cellspacing="0" cellpadding="4">
2436 <tbody>
2437 <tr>
2438 <td>In0</td>
2439 <td>In1</td>
2440 <td>Out</td>
2441 </tr>
2442 <tr>
2443 <td>0</td>
2444 <td>0</td>
2445 <td>0</td>
2446 </tr>
2447 <tr>
2448 <td>0</td>
2449 <td>1</td>
2450 <td>1</td>
2451 </tr>
2452 <tr>
2453 <td>1</td>
2454 <td>0</td>
2455 <td>1</td>
2456 </tr>
2457 <tr>
2458 <td>1</td>
2459 <td>1</td>
2460 <td>1</td>
2461 </tr>
2462 </tbody>
2463</table>
2464</div>
2465<h5>Example:</h5>
2466<pre> &lt;result&gt; = or i32 4, %var <i>; yields {i32}:result = 4 | %var</i>
2467 &lt;result&gt; = or i32 15, 40 <i>; yields {i32}:result = 47</i>
2468 &lt;result&gt; = or i32 4, 8 <i>; yields {i32}:result = 12</i>
2469</pre>
2470</div>
2471<!-- _______________________________________________________________________ -->
2472<div class="doc_subsubsection"> <a name="i_xor">'<tt>xor</tt>'
2473Instruction</a> </div>
2474<div class="doc_text">
2475<h5>Syntax:</h5>
2476<pre> &lt;result&gt; = xor &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
2477</pre>
2478<h5>Overview:</h5>
2479<p>The '<tt>xor</tt>' instruction returns the bitwise logical exclusive
2480or of its two operands. The <tt>xor</tt> is used to implement the
2481"one's complement" operation, which is the "~" operator in C.</p>
2482<h5>Arguments:</h5>
2483<p>The two arguments to the '<tt>xor</tt>' instruction must be <a
2484 href="#t_integer">integer</a> values. Both arguments must have
2485identical types.</p>
2486<h5>Semantics:</h5>
2487<p>The truth table used for the '<tt>xor</tt>' instruction is:</p>
2488<p> </p>
2489<div style="align: center">
2490<table border="1" cellspacing="0" cellpadding="4">
2491 <tbody>
2492 <tr>
2493 <td>In0</td>
2494 <td>In1</td>
2495 <td>Out</td>
2496 </tr>
2497 <tr>
2498 <td>0</td>
2499 <td>0</td>
2500 <td>0</td>
2501 </tr>
2502 <tr>
2503 <td>0</td>
2504 <td>1</td>
2505 <td>1</td>
2506 </tr>
2507 <tr>
2508 <td>1</td>
2509 <td>0</td>
2510 <td>1</td>
2511 </tr>
2512 <tr>
2513 <td>1</td>
2514 <td>1</td>
2515 <td>0</td>
2516 </tr>
2517 </tbody>
2518</table>
2519</div>
2520<p> </p>
2521<h5>Example:</h5>
2522<pre> &lt;result&gt; = xor i32 4, %var <i>; yields {i32}:result = 4 ^ %var</i>
2523 &lt;result&gt; = xor i32 15, 40 <i>; yields {i32}:result = 39</i>
2524 &lt;result&gt; = xor i32 4, 8 <i>; yields {i32}:result = 12</i>
2525 &lt;result&gt; = xor i32 %V, -1 <i>; yields {i32}:result = ~%V</i>
2526</pre>
2527</div>
2528
2529<!-- ======================================================================= -->
2530<div class="doc_subsection">
2531 <a name="vectorops">Vector Operations</a>
2532</div>
2533
2534<div class="doc_text">
2535
2536<p>LLVM supports several instructions to represent vector operations in a
2537target-independent manner. These instructions cover the element-access and
2538vector-specific operations needed to process vectors effectively. While LLVM
2539does directly support these vector operations, many sophisticated algorithms
2540will want to use target-specific intrinsics to take full advantage of a specific
2541target.</p>
2542
2543</div>
2544
2545<!-- _______________________________________________________________________ -->
2546<div class="doc_subsubsection">
2547 <a name="i_extractelement">'<tt>extractelement</tt>' Instruction</a>
2548</div>
2549
2550<div class="doc_text">
2551
2552<h5>Syntax:</h5>
2553
2554<pre>
2555 &lt;result&gt; = extractelement &lt;n x &lt;ty&gt;&gt; &lt;val&gt;, i32 &lt;idx&gt; <i>; yields &lt;ty&gt;</i>
2556</pre>
2557
2558<h5>Overview:</h5>
2559
2560<p>
2561The '<tt>extractelement</tt>' instruction extracts a single scalar
2562element from a vector at a specified index.
2563</p>
2564
2565
2566<h5>Arguments:</h5>
2567
2568<p>
2569The first operand of an '<tt>extractelement</tt>' instruction is a
2570value of <a href="#t_vector">vector</a> type. The second operand is
2571an index indicating the position from which to extract the element.
2572The index may be a variable.</p>
2573
2574<h5>Semantics:</h5>
2575
2576<p>
2577The result is a scalar of the same type as the element type of
2578<tt>val</tt>. Its value is the value at position <tt>idx</tt> of
2579<tt>val</tt>. If <tt>idx</tt> exceeds the length of <tt>val</tt>, the
2580results are undefined.
2581</p>
2582
2583<h5>Example:</h5>
2584
2585<pre>
2586 %result = extractelement &lt;4 x i32&gt; %vec, i32 0 <i>; yields i32</i>
2587</pre>
2588</div>
2589
2590
2591<!-- _______________________________________________________________________ -->
2592<div class="doc_subsubsection">
2593 <a name="i_insertelement">'<tt>insertelement</tt>' Instruction</a>
2594</div>
2595
2596<div class="doc_text">
2597
2598<h5>Syntax:</h5>
2599
2600<pre>
2601 &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>
2602</pre>
2603
2604<h5>Overview:</h5>
2605
2606<p>
2607The '<tt>insertelement</tt>' instruction inserts a scalar
2608element into a vector at a specified index.
2609</p>
2610
2611
2612<h5>Arguments:</h5>
2613
2614<p>
2615The first operand of an '<tt>insertelement</tt>' instruction is a
2616value of <a href="#t_vector">vector</a> type. The second operand is a
2617scalar value whose type must equal the element type of the first
2618operand. The third operand is an index indicating the position at
2619which to insert the value. The index may be a variable.</p>
2620
2621<h5>Semantics:</h5>
2622
2623<p>
2624The result is a vector of the same type as <tt>val</tt>. Its
2625element values are those of <tt>val</tt> except at position
2626<tt>idx</tt>, where it gets the value <tt>elt</tt>. If <tt>idx</tt>
2627exceeds the length of <tt>val</tt>, the results are undefined.
2628</p>
2629
2630<h5>Example:</h5>
2631
2632<pre>
2633 %result = insertelement &lt;4 x i32&gt; %vec, i32 1, i32 0 <i>; yields &lt;4 x i32&gt;</i>
2634</pre>
2635</div>
2636
2637<!-- _______________________________________________________________________ -->
2638<div class="doc_subsubsection">
2639 <a name="i_shufflevector">'<tt>shufflevector</tt>' Instruction</a>
2640</div>
2641
2642<div class="doc_text">
2643
2644<h5>Syntax:</h5>
2645
2646<pre>
2647 &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>
2648</pre>
2649
2650<h5>Overview:</h5>
2651
2652<p>
2653The '<tt>shufflevector</tt>' instruction constructs a permutation of elements
2654from two input vectors, returning a vector of the same type.
2655</p>
2656
2657<h5>Arguments:</h5>
2658
2659<p>
2660The first two operands of a '<tt>shufflevector</tt>' instruction are vectors
2661with types that match each other and types that match the result of the
2662instruction. The third argument is a shuffle mask, which has the same number
2663of elements as the other vector type, but whose element type is always 'i32'.
2664</p>
2665
2666<p>
2667The shuffle mask operand is required to be a constant vector with either
2668constant integer or undef values.
2669</p>
2670
2671<h5>Semantics:</h5>
2672
2673<p>
2674The elements of the two input vectors are numbered from left to right across
2675both of the vectors. The shuffle mask operand specifies, for each element of
2676the result vector, which element of the two input registers the result element
2677gets. The element selector may be undef (meaning "don't care") and the second
2678operand may be undef if performing a shuffle from only one vector.
2679</p>
2680
2681<h5>Example:</h5>
2682
2683<pre>
2684 %result = shufflevector &lt;4 x i32&gt; %v1, &lt;4 x i32&gt; %v2,
2685 &lt;4 x i32&gt; &lt;i32 0, i32 4, i32 1, i32 5&gt; <i>; yields &lt;4 x i32&gt;</i>
2686 %result = shufflevector &lt;4 x i32&gt; %v1, &lt;4 x i32&gt; undef,
2687 &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.
2688</pre>
2689</div>
2690
2691
2692<!-- ======================================================================= -->
2693<div class="doc_subsection">
2694 <a name="memoryops">Memory Access and Addressing Operations</a>
2695</div>
2696
2697<div class="doc_text">
2698
2699<p>A key design point of an SSA-based representation is how it
2700represents memory. In LLVM, no memory locations are in SSA form, which
2701makes things very simple. This section describes how to read, write,
2702allocate, and free memory in LLVM.</p>
2703
2704</div>
2705
2706<!-- _______________________________________________________________________ -->
2707<div class="doc_subsubsection">
2708 <a name="i_malloc">'<tt>malloc</tt>' Instruction</a>
2709</div>
2710
2711<div class="doc_text">
2712
2713<h5>Syntax:</h5>
2714
2715<pre>
2716 &lt;result&gt; = malloc &lt;type&gt;[, i32 &lt;NumElements&gt;][, align &lt;alignment&gt;] <i>; yields {type*}:result</i>
2717</pre>
2718
2719<h5>Overview:</h5>
2720
2721<p>The '<tt>malloc</tt>' instruction allocates memory from the system
Christopher Lambcfe00962007-12-17 01:00:21 +0000[diff] [blame]2722heap and returns a pointer to it. The object is always allocated in the generic
2723address space (address space zero).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2724
2725<h5>Arguments:</h5>
2726
2727<p>The '<tt>malloc</tt>' instruction allocates
2728<tt>sizeof(&lt;type&gt;)*NumElements</tt>
2729bytes of memory from the operating system and returns a pointer of the
2730appropriate type to the program. If "NumElements" is specified, it is the
2731number of elements allocated. If an alignment is specified, the value result
2732of the allocation is guaranteed to be aligned to at least that boundary. If
2733not specified, or if zero, the target can choose to align the allocation on any
2734convenient boundary.</p>
2735
2736<p>'<tt>type</tt>' must be a sized type.</p>
2737
2738<h5>Semantics:</h5>
2739
2740<p>Memory is allocated using the system "<tt>malloc</tt>" function, and
2741a pointer is returned.</p>
2742
2743<h5>Example:</h5>
2744
2745<pre>
2746 %array = malloc [4 x i8 ] <i>; yields {[%4 x i8]*}:array</i>
2747
2748 %size = <a href="#i_add">add</a> i32 2, 2 <i>; yields {i32}:size = i32 4</i>
2749 %array1 = malloc i8, i32 4 <i>; yields {i8*}:array1</i>
2750 %array2 = malloc [12 x i8], i32 %size <i>; yields {[12 x i8]*}:array2</i>
2751 %array3 = malloc i32, i32 4, align 1024 <i>; yields {i32*}:array3</i>
2752 %array4 = malloc i32, align 1024 <i>; yields {i32*}:array4</i>
2753</pre>
2754</div>
2755
2756<!-- _______________________________________________________________________ -->
2757<div class="doc_subsubsection">
2758 <a name="i_free">'<tt>free</tt>' Instruction</a>
2759</div>
2760
2761<div class="doc_text">
2762
2763<h5>Syntax:</h5>
2764
2765<pre>
2766 free &lt;type&gt; &lt;value&gt; <i>; yields {void}</i>
2767</pre>
2768
2769<h5>Overview:</h5>
2770
2771<p>The '<tt>free</tt>' instruction returns memory back to the unused
2772memory heap to be reallocated in the future.</p>
2773
2774<h5>Arguments:</h5>
2775
2776<p>'<tt>value</tt>' shall be a pointer value that points to a value
2777that was allocated with the '<tt><a href="#i_malloc">malloc</a></tt>'
2778instruction.</p>
2779
2780<h5>Semantics:</h5>
2781
2782<p>Access to the memory pointed to by the pointer is no longer defined
2783after this instruction executes.</p>
2784
2785<h5>Example:</h5>
2786
2787<pre>
2788 %array = <a href="#i_malloc">malloc</a> [4 x i8] <i>; yields {[4 x i8]*}:array</i>
2789 free [4 x i8]* %array
2790</pre>
2791</div>
2792
2793<!-- _______________________________________________________________________ -->
2794<div class="doc_subsubsection">
2795 <a name="i_alloca">'<tt>alloca</tt>' Instruction</a>
2796</div>
2797
2798<div class="doc_text">
2799
2800<h5>Syntax:</h5>
2801
2802<pre>
2803 &lt;result&gt; = alloca &lt;type&gt;[, i32 &lt;NumElements&gt;][, align &lt;alignment&gt;] <i>; yields {type*}:result</i>
2804</pre>
2805
2806<h5>Overview:</h5>
2807
2808<p>The '<tt>alloca</tt>' instruction allocates memory on the stack frame of the
2809currently executing function, to be automatically released when this function
Christopher Lambcfe00962007-12-17 01:00:21 +0000[diff] [blame]2810returns to its caller. The object is always allocated in the generic address
2811space (address space zero).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2812
2813<h5>Arguments:</h5>
2814
2815<p>The '<tt>alloca</tt>' instruction allocates <tt>sizeof(&lt;type&gt;)*NumElements</tt>
2816bytes of memory on the runtime stack, returning a pointer of the
2817appropriate type to the program. If "NumElements" is specified, it is the
2818number of elements allocated. If an alignment is specified, the value result
2819of the allocation is guaranteed to be aligned to at least that boundary. If
2820not specified, or if zero, the target can choose to align the allocation on any
2821convenient boundary.</p>
2822
2823<p>'<tt>type</tt>' may be any sized type.</p>
2824
2825<h5>Semantics:</h5>
2826
2827<p>Memory is allocated; a pointer is returned. '<tt>alloca</tt>'d
2828memory is automatically released when the function returns. The '<tt>alloca</tt>'
2829instruction is commonly used to represent automatic variables that must
2830have an address available. When the function returns (either with the <tt><a
2831 href="#i_ret">ret</a></tt> or <tt><a href="#i_unwind">unwind</a></tt>
2832instructions), the memory is reclaimed.</p>
2833
2834<h5>Example:</h5>
2835
2836<pre>
2837 %ptr = alloca i32 <i>; yields {i32*}:ptr</i>
2838 %ptr = alloca i32, i32 4 <i>; yields {i32*}:ptr</i>
2839 %ptr = alloca i32, i32 4, align 1024 <i>; yields {i32*}:ptr</i>
2840 %ptr = alloca i32, align 1024 <i>; yields {i32*}:ptr</i>
2841</pre>
2842</div>
2843
2844<!-- _______________________________________________________________________ -->
2845<div class="doc_subsubsection"> <a name="i_load">'<tt>load</tt>'
2846Instruction</a> </div>
2847<div class="doc_text">
2848<h5>Syntax:</h5>
2849<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>
2850<h5>Overview:</h5>
2851<p>The '<tt>load</tt>' instruction is used to read from memory.</p>
2852<h5>Arguments:</h5>
2853<p>The argument to the '<tt>load</tt>' instruction specifies the memory
2854address from which to load. The pointer must point to a <a
2855 href="#t_firstclass">first class</a> type. If the <tt>load</tt> is
2856marked as <tt>volatile</tt>, then the optimizer is not allowed to modify
2857the number or order of execution of this <tt>load</tt> with other
2858volatile <tt>load</tt> and <tt><a href="#i_store">store</a></tt>
2859instructions. </p>
Chris Lattner21003c82008-01-06 21:04:43 +0000[diff] [blame]2860<p>
2861The optional "align" argument specifies the alignment of the operation
2862(that is, the alignment of the memory address). A value of 0 or an
2863omitted "align" argument means that the operation has the preferential
2864alignment for the target. It is the responsibility of the code emitter
2865to ensure that the alignment information is correct. Overestimating
2866the alignment results in an undefined behavior. Underestimating the
2867alignment may produce less efficient code. An alignment of 1 is always
2868safe.
2869</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2870<h5>Semantics:</h5>
2871<p>The location of memory pointed to is loaded.</p>
2872<h5>Examples:</h5>
2873<pre> %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
2874 <a
2875 href="#i_store">store</a> i32 3, i32* %ptr <i>; yields {void}</i>
2876 %val = load i32* %ptr <i>; yields {i32}:val = i32 3</i>
2877</pre>
2878</div>
2879<!-- _______________________________________________________________________ -->
2880<div class="doc_subsubsection"> <a name="i_store">'<tt>store</tt>'
2881Instruction</a> </div>
2882<div class="doc_text">
2883<h5>Syntax:</h5>
2884<pre> store &lt;ty&gt; &lt;value&gt;, &lt;ty&gt;* &lt;pointer&gt;[, align &lt;alignment&gt;] <i>; yields {void}</i>
2885 volatile store &lt;ty&gt; &lt;value&gt;, &lt;ty&gt;* &lt;pointer&gt;[, align &lt;alignment&gt;] <i>; yields {void}</i>
2886</pre>
2887<h5>Overview:</h5>
2888<p>The '<tt>store</tt>' instruction is used to write to memory.</p>
2889<h5>Arguments:</h5>
2890<p>There are two arguments to the '<tt>store</tt>' instruction: a value
2891to store and an address at which to store it. The type of the '<tt>&lt;pointer&gt;</tt>'
2892operand must be a pointer to the type of the '<tt>&lt;value&gt;</tt>'
2893operand. If the <tt>store</tt> is marked as <tt>volatile</tt>, then the
2894optimizer is not allowed to modify the number or order of execution of
2895this <tt>store</tt> with other volatile <tt>load</tt> and <tt><a
2896 href="#i_store">store</a></tt> instructions.</p>
Chris Lattner21003c82008-01-06 21:04:43 +0000[diff] [blame]2897<p>
2898The optional "align" argument specifies the alignment of the operation
2899(that is, the alignment of the memory address). A value of 0 or an
2900omitted "align" argument means that the operation has the preferential
2901alignment for the target. It is the responsibility of the code emitter
2902to ensure that the alignment information is correct. Overestimating
2903the alignment results in an undefined behavior. Underestimating the
2904alignment may produce less efficient code. An alignment of 1 is always
2905safe.
2906</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2907<h5>Semantics:</h5>
2908<p>The contents of memory are updated to contain '<tt>&lt;value&gt;</tt>'
2909at the location specified by the '<tt>&lt;pointer&gt;</tt>' operand.</p>
2910<h5>Example:</h5>
2911<pre> %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
Bill Wendling63ffa142007-10-22 05:10:05 +0000[diff] [blame]2912 store i32 3, i32* %ptr <i>; yields {void}</i>
2913 %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]2914</pre>
2915</div>
2916
2917<!-- _______________________________________________________________________ -->
2918<div class="doc_subsubsection">
2919 <a name="i_getelementptr">'<tt>getelementptr</tt>' Instruction</a>
2920</div>
2921
2922<div class="doc_text">
2923<h5>Syntax:</h5>
2924<pre>
2925 &lt;result&gt; = getelementptr &lt;ty&gt;* &lt;ptrval&gt;{, &lt;ty&gt; &lt;idx&gt;}*
2926</pre>
2927
2928<h5>Overview:</h5>
2929
2930<p>
2931The '<tt>getelementptr</tt>' instruction is used to get the address of a
2932subelement of an aggregate data structure.</p>
2933
2934<h5>Arguments:</h5>
2935
2936<p>This instruction takes a list of integer operands that indicate what
2937elements of the aggregate object to index to. The actual types of the arguments
2938provided depend on the type of the first pointer argument. The
2939'<tt>getelementptr</tt>' instruction is used to index down through the type
2940levels of a structure or to a specific index in an array. When indexing into a
2941structure, only <tt>i32</tt> integer constants are allowed. When indexing
2942into an array or pointer, only integers of 32 or 64 bits are allowed, and will
2943be sign extended to 64-bit values.</p>
2944
2945<p>For example, let's consider a C code fragment and how it gets
2946compiled to LLVM:</p>
2947
2948<div class="doc_code">
2949<pre>
2950struct RT {
2951 char A;
2952 int B[10][20];
2953 char C;
2954};
2955struct ST {
2956 int X;
2957 double Y;
2958 struct RT Z;
2959};
2960
2961int *foo(struct ST *s) {
2962 return &amp;s[1].Z.B[5][13];
2963}
2964</pre>
2965</div>
2966
2967<p>The LLVM code generated by the GCC frontend is:</p>
2968
2969<div class="doc_code">
2970<pre>
2971%RT = type { i8 , [10 x [20 x i32]], i8 }
2972%ST = type { i32, double, %RT }
2973
2974define i32* %foo(%ST* %s) {
2975entry:
2976 %reg = getelementptr %ST* %s, i32 1, i32 2, i32 1, i32 5, i32 13
2977 ret i32* %reg
2978}
2979</pre>
2980</div>
2981
2982<h5>Semantics:</h5>
2983
2984<p>The index types specified for the '<tt>getelementptr</tt>' instruction depend
2985on the pointer type that is being indexed into. <a href="#t_pointer">Pointer</a>
2986and <a href="#t_array">array</a> types can use a 32-bit or 64-bit
2987<a href="#t_integer">integer</a> type but the value will always be sign extended
2988to 64-bits. <a href="#t_struct">Structure</a> types require <tt>i32</tt>
2989<b>constants</b>.</p>
2990
2991<p>In the example above, the first index is indexing into the '<tt>%ST*</tt>'
2992type, which is a pointer, yielding a '<tt>%ST</tt>' = '<tt>{ i32, double, %RT
2993}</tt>' type, a structure. The second index indexes into the third element of
2994the structure, yielding a '<tt>%RT</tt>' = '<tt>{ i8 , [10 x [20 x i32]],
2995i8 }</tt>' type, another structure. The third index indexes into the second
2996element of the structure, yielding a '<tt>[10 x [20 x i32]]</tt>' type, an
2997array. The two dimensions of the array are subscripted into, yielding an
2998'<tt>i32</tt>' type. The '<tt>getelementptr</tt>' instruction returns a pointer
2999to this element, thus computing a value of '<tt>i32*</tt>' type.</p>
3000
3001<p>Note that it is perfectly legal to index partially through a
3002structure, returning a pointer to an inner element. Because of this,
3003the LLVM code for the given testcase is equivalent to:</p>
3004
3005<pre>
3006 define i32* %foo(%ST* %s) {
3007 %t1 = getelementptr %ST* %s, i32 1 <i>; yields %ST*:%t1</i>
3008 %t2 = getelementptr %ST* %t1, i32 0, i32 2 <i>; yields %RT*:%t2</i>
3009 %t3 = getelementptr %RT* %t2, i32 0, i32 1 <i>; yields [10 x [20 x i32]]*:%t3</i>
3010 %t4 = getelementptr [10 x [20 x i32]]* %t3, i32 0, i32 5 <i>; yields [20 x i32]*:%t4</i>
3011 %t5 = getelementptr [20 x i32]* %t4, i32 0, i32 13 <i>; yields i32*:%t5</i>
3012 ret i32* %t5
3013 }
3014</pre>
3015
3016<p>Note that it is undefined to access an array out of bounds: array and
3017pointer indexes must always be within the defined bounds of the array type.
3018The one exception for this rules is zero length arrays. These arrays are
3019defined to be accessible as variable length arrays, which requires access
3020beyond the zero'th element.</p>
3021
3022<p>The getelementptr instruction is often confusing. For some more insight
3023into how it works, see <a href="GetElementPtr.html">the getelementptr
3024FAQ</a>.</p>
3025
3026<h5>Example:</h5>
3027
3028<pre>
3029 <i>; yields [12 x i8]*:aptr</i>
3030 %aptr = getelementptr {i32, [12 x i8]}* %sptr, i64 0, i32 1
3031</pre>
3032</div>
3033
3034<!-- ======================================================================= -->
3035<div class="doc_subsection"> <a name="convertops">Conversion Operations</a>
3036</div>
3037<div class="doc_text">
3038<p>The instructions in this category are the conversion instructions (casting)
3039which all take a single operand and a type. They perform various bit conversions
3040on the operand.</p>
3041</div>
3042
3043<!-- _______________________________________________________________________ -->
3044<div class="doc_subsubsection">
3045 <a name="i_trunc">'<tt>trunc .. to</tt>' Instruction</a>
3046</div>
3047<div class="doc_text">
3048
3049<h5>Syntax:</h5>
3050<pre>
3051 &lt;result&gt; = trunc &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3052</pre>
3053
3054<h5>Overview:</h5>
3055<p>
3056The '<tt>trunc</tt>' instruction truncates its operand to the type <tt>ty2</tt>.
3057</p>
3058
3059<h5>Arguments:</h5>
3060<p>
3061The '<tt>trunc</tt>' instruction takes a <tt>value</tt> to trunc, which must
3062be an <a href="#t_integer">integer</a> type, and a type that specifies the size
3063and type of the result, which must be an <a href="#t_integer">integer</a>
3064type. The bit size of <tt>value</tt> must be larger than the bit size of
3065<tt>ty2</tt>. Equal sized types are not allowed.</p>
3066
3067<h5>Semantics:</h5>
3068<p>
3069The '<tt>trunc</tt>' instruction truncates the high order bits in <tt>value</tt>
3070and converts the remaining bits to <tt>ty2</tt>. Since the source size must be
3071larger than the destination size, <tt>trunc</tt> cannot be a <i>no-op cast</i>.
3072It will always truncate bits.</p>
3073
3074<h5>Example:</h5>
3075<pre>
3076 %X = trunc i32 257 to i8 <i>; yields i8:1</i>
3077 %Y = trunc i32 123 to i1 <i>; yields i1:true</i>
3078 %Y = trunc i32 122 to i1 <i>; yields i1:false</i>
3079</pre>
3080</div>
3081
3082<!-- _______________________________________________________________________ -->
3083<div class="doc_subsubsection">
3084 <a name="i_zext">'<tt>zext .. to</tt>' Instruction</a>
3085</div>
3086<div class="doc_text">
3087
3088<h5>Syntax:</h5>
3089<pre>
3090 &lt;result&gt; = zext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3091</pre>
3092
3093<h5>Overview:</h5>
3094<p>The '<tt>zext</tt>' instruction zero extends its operand to type
3095<tt>ty2</tt>.</p>
3096
3097
3098<h5>Arguments:</h5>
3099<p>The '<tt>zext</tt>' instruction takes a value to cast, which must be of
3100<a href="#t_integer">integer</a> type, and a type to cast it to, which must
3101also be of <a href="#t_integer">integer</a> type. The bit size of the
3102<tt>value</tt> must be smaller than the bit size of the destination type,
3103<tt>ty2</tt>.</p>
3104
3105<h5>Semantics:</h5>
3106<p>The <tt>zext</tt> fills the high order bits of the <tt>value</tt> with zero
3107bits until it reaches the size of the destination type, <tt>ty2</tt>.</p>
3108
3109<p>When zero extending from i1, the result will always be either 0 or 1.</p>
3110
3111<h5>Example:</h5>
3112<pre>
3113 %X = zext i32 257 to i64 <i>; yields i64:257</i>
3114 %Y = zext i1 true to i32 <i>; yields i32:1</i>
3115</pre>
3116</div>
3117
3118<!-- _______________________________________________________________________ -->
3119<div class="doc_subsubsection">
3120 <a name="i_sext">'<tt>sext .. to</tt>' Instruction</a>
3121</div>
3122<div class="doc_text">
3123
3124<h5>Syntax:</h5>
3125<pre>
3126 &lt;result&gt; = sext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3127</pre>
3128
3129<h5>Overview:</h5>
3130<p>The '<tt>sext</tt>' sign extends <tt>value</tt> to the type <tt>ty2</tt>.</p>
3131
3132<h5>Arguments:</h5>
3133<p>
3134The '<tt>sext</tt>' instruction takes a value to cast, which must be of
3135<a href="#t_integer">integer</a> type, and a type to cast it to, which must
3136also be of <a href="#t_integer">integer</a> type. The bit size of the
3137<tt>value</tt> must be smaller than the bit size of the destination type,
3138<tt>ty2</tt>.</p>
3139
3140<h5>Semantics:</h5>
3141<p>
3142The '<tt>sext</tt>' instruction performs a sign extension by copying the sign
3143bit (highest order bit) of the <tt>value</tt> until it reaches the bit size of
3144the type <tt>ty2</tt>.</p>
3145
3146<p>When sign extending from i1, the extension always results in -1 or 0.</p>
3147
3148<h5>Example:</h5>
3149<pre>
3150 %X = sext i8 -1 to i16 <i>; yields i16 :65535</i>
3151 %Y = sext i1 true to i32 <i>; yields i32:-1</i>
3152</pre>
3153</div>
3154
3155<!-- _______________________________________________________________________ -->
3156<div class="doc_subsubsection">
3157 <a name="i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a>
3158</div>
3159
3160<div class="doc_text">
3161
3162<h5>Syntax:</h5>
3163
3164<pre>
3165 &lt;result&gt; = fptrunc &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3166</pre>
3167
3168<h5>Overview:</h5>
3169<p>The '<tt>fptrunc</tt>' instruction truncates <tt>value</tt> to type
3170<tt>ty2</tt>.</p>
3171
3172
3173<h5>Arguments:</h5>
3174<p>The '<tt>fptrunc</tt>' instruction takes a <a href="#t_floating">floating
3175 point</a> value to cast and a <a href="#t_floating">floating point</a> type to
3176cast it to. The size of <tt>value</tt> must be larger than the size of
3177<tt>ty2</tt>. This implies that <tt>fptrunc</tt> cannot be used to make a
3178<i>no-op cast</i>.</p>
3179
3180<h5>Semantics:</h5>
3181<p> The '<tt>fptrunc</tt>' instruction truncates a <tt>value</tt> from a larger
3182<a href="#t_floating">floating point</a> type to a smaller
3183<a href="#t_floating">floating point</a> type. If the value cannot fit within
3184the destination type, <tt>ty2</tt>, then the results are undefined.</p>
3185
3186<h5>Example:</h5>
3187<pre>
3188 %X = fptrunc double 123.0 to float <i>; yields float:123.0</i>
3189 %Y = fptrunc double 1.0E+300 to float <i>; yields undefined</i>
3190</pre>
3191</div>
3192
3193<!-- _______________________________________________________________________ -->
3194<div class="doc_subsubsection">
3195 <a name="i_fpext">'<tt>fpext .. to</tt>' Instruction</a>
3196</div>
3197<div class="doc_text">
3198
3199<h5>Syntax:</h5>
3200<pre>
3201 &lt;result&gt; = fpext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3202</pre>
3203
3204<h5>Overview:</h5>
3205<p>The '<tt>fpext</tt>' extends a floating point <tt>value</tt> to a larger
3206floating point value.</p>
3207
3208<h5>Arguments:</h5>
3209<p>The '<tt>fpext</tt>' instruction takes a
3210<a href="#t_floating">floating point</a> <tt>value</tt> to cast,
3211and a <a href="#t_floating">floating point</a> type to cast it to. The source
3212type must be smaller than the destination type.</p>
3213
3214<h5>Semantics:</h5>
3215<p>The '<tt>fpext</tt>' instruction extends the <tt>value</tt> from a smaller
3216<a href="#t_floating">floating point</a> type to a larger
3217<a href="#t_floating">floating point</a> type. The <tt>fpext</tt> cannot be
3218used to make a <i>no-op cast</i> because it always changes bits. Use
3219<tt>bitcast</tt> to make a <i>no-op cast</i> for a floating point cast.</p>
3220
3221<h5>Example:</h5>
3222<pre>
3223 %X = fpext float 3.1415 to double <i>; yields double:3.1415</i>
3224 %Y = fpext float 1.0 to float <i>; yields float:1.0 (no-op)</i>
3225</pre>
3226</div>
3227
3228<!-- _______________________________________________________________________ -->
3229<div class="doc_subsubsection">
3230 <a name="i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a>
3231</div>
3232<div class="doc_text">
3233
3234<h5>Syntax:</h5>
3235<pre>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3236 &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]3237</pre>
3238
3239<h5>Overview:</h5>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3240<p>The '<tt>fptoui</tt>' converts a floating point <tt>value</tt> to its
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3241unsigned integer equivalent of type <tt>ty2</tt>.
3242</p>
3243
3244<h5>Arguments:</h5>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3245<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]3246scalar or vector <a href="#t_floating">floating point</a> value, and a type
3247to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
3248type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
3249vector integer type with the same number of elements as <tt>ty</tt></p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3250
3251<h5>Semantics:</h5>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3252<p> The '<tt>fptoui</tt>' instruction converts its
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3253<a href="#t_floating">floating point</a> operand into the nearest (rounding
3254towards zero) unsigned integer value. If the value cannot fit in <tt>ty2</tt>,
3255the results are undefined.</p>
3256
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3257<h5>Example:</h5>
3258<pre>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3259 %X = fptoui double 123.0 to i32 <i>; yields i32:123</i>
Chris Lattner681f1e82007-09-22 03:17:52 +0000[diff] [blame]3260 %Y = fptoui float 1.0E+300 to i1 <i>; yields undefined:1</i>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3261 %X = fptoui float 1.04E+17 to i8 <i>; yields undefined:1</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3262</pre>
3263</div>
3264
3265<!-- _______________________________________________________________________ -->
3266<div class="doc_subsubsection">
3267 <a name="i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a>
3268</div>
3269<div class="doc_text">
3270
3271<h5>Syntax:</h5>
3272<pre>
3273 &lt;result&gt; = fptosi &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3274</pre>
3275
3276<h5>Overview:</h5>
3277<p>The '<tt>fptosi</tt>' instruction converts
3278<a href="#t_floating">floating point</a> <tt>value</tt> to type <tt>ty2</tt>.
3279</p>
3280
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3281<h5>Arguments:</h5>
3282<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]3283scalar or vector <a href="#t_floating">floating point</a> value, and a type
3284to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
3285type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
3286vector integer type with the same number of elements as <tt>ty</tt></p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3287
3288<h5>Semantics:</h5>
3289<p>The '<tt>fptosi</tt>' instruction converts its
3290<a href="#t_floating">floating point</a> operand into the nearest (rounding
3291towards zero) signed integer value. If the value cannot fit in <tt>ty2</tt>,
3292the results are undefined.</p>
3293
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3294<h5>Example:</h5>
3295<pre>
3296 %X = fptosi double -123.0 to i32 <i>; yields i32:-123</i>
Chris Lattner681f1e82007-09-22 03:17:52 +0000[diff] [blame]3297 %Y = fptosi float 1.0E-247 to i1 <i>; yields undefined:1</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3298 %X = fptosi float 1.04E+17 to i8 <i>; yields undefined:1</i>
3299</pre>
3300</div>
3301
3302<!-- _______________________________________________________________________ -->
3303<div class="doc_subsubsection">
3304 <a name="i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a>
3305</div>
3306<div class="doc_text">
3307
3308<h5>Syntax:</h5>
3309<pre>
3310 &lt;result&gt; = uitofp &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3311</pre>
3312
3313<h5>Overview:</h5>
3314<p>The '<tt>uitofp</tt>' instruction regards <tt>value</tt> as an unsigned
3315integer and converts that value to the <tt>ty2</tt> type.</p>
3316
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3317<h5>Arguments:</h5>
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]3318<p>The '<tt>uitofp</tt>' instruction takes a value to cast, which must be a
3319scalar or vector <a href="#t_integer">integer</a> value, and a type to cast it
3320to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
3321type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
3322floating point type with the same number of elements as <tt>ty</tt></p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3323
3324<h5>Semantics:</h5>
3325<p>The '<tt>uitofp</tt>' instruction interprets its operand as an unsigned
3326integer quantity and converts it to the corresponding floating point value. If
3327the value cannot fit in the floating point value, the results are undefined.</p>
3328
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3329<h5>Example:</h5>
3330<pre>
3331 %X = uitofp i32 257 to float <i>; yields float:257.0</i>
3332 %Y = uitofp i8 -1 to double <i>; yields double:255.0</i>
3333</pre>
3334</div>
3335
3336<!-- _______________________________________________________________________ -->
3337<div class="doc_subsubsection">
3338 <a name="i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a>
3339</div>
3340<div class="doc_text">
3341
3342<h5>Syntax:</h5>
3343<pre>
3344 &lt;result&gt; = sitofp &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3345</pre>
3346
3347<h5>Overview:</h5>
3348<p>The '<tt>sitofp</tt>' instruction regards <tt>value</tt> as a signed
3349integer and converts that value to the <tt>ty2</tt> type.</p>
3350
3351<h5>Arguments:</h5>
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]3352<p>The '<tt>sitofp</tt>' instruction takes a value to cast, which must be a
3353scalar or vector <a href="#t_integer">integer</a> value, and a type to cast it
3354to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
3355type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
3356floating point type with the same number of elements as <tt>ty</tt></p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3357
3358<h5>Semantics:</h5>
3359<p>The '<tt>sitofp</tt>' instruction interprets its operand as a signed
3360integer quantity and converts it to the corresponding floating point value. If
3361the value cannot fit in the floating point value, the results are undefined.</p>
3362
3363<h5>Example:</h5>
3364<pre>
3365 %X = sitofp i32 257 to float <i>; yields float:257.0</i>
3366 %Y = sitofp i8 -1 to double <i>; yields double:-1.0</i>
3367</pre>
3368</div>
3369
3370<!-- _______________________________________________________________________ -->
3371<div class="doc_subsubsection">
3372 <a name="i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a>
3373</div>
3374<div class="doc_text">
3375
3376<h5>Syntax:</h5>
3377<pre>
3378 &lt;result&gt; = ptrtoint &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3379</pre>
3380
3381<h5>Overview:</h5>
3382<p>The '<tt>ptrtoint</tt>' instruction converts the pointer <tt>value</tt> to
3383the integer type <tt>ty2</tt>.</p>
3384
3385<h5>Arguments:</h5>
3386<p>The '<tt>ptrtoint</tt>' instruction takes a <tt>value</tt> to cast, which
3387must be a <a href="#t_pointer">pointer</a> value, and a type to cast it to
3388<tt>ty2</tt>, which must be an <a href="#t_integer">integer</a> type.
3389
3390<h5>Semantics:</h5>
3391<p>The '<tt>ptrtoint</tt>' instruction converts <tt>value</tt> to integer type
3392<tt>ty2</tt> by interpreting the pointer value as an integer and either
3393truncating or zero extending that value to the size of the integer type. If
3394<tt>value</tt> is smaller than <tt>ty2</tt> then a zero extension is done. If
3395<tt>value</tt> is larger than <tt>ty2</tt> then a truncation is done. If they
3396are the same size, then nothing is done (<i>no-op cast</i>) other than a type
3397change.</p>
3398
3399<h5>Example:</h5>
3400<pre>
3401 %X = ptrtoint i32* %X to i8 <i>; yields truncation on 32-bit architecture</i>
3402 %Y = ptrtoint i32* %x to i64 <i>; yields zero extension on 32-bit architecture</i>
3403</pre>
3404</div>
3405
3406<!-- _______________________________________________________________________ -->
3407<div class="doc_subsubsection">
3408 <a name="i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a>
3409</div>
3410<div class="doc_text">
3411
3412<h5>Syntax:</h5>
3413<pre>
3414 &lt;result&gt; = inttoptr &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3415</pre>
3416
3417<h5>Overview:</h5>
3418<p>The '<tt>inttoptr</tt>' instruction converts an integer <tt>value</tt> to
3419a pointer type, <tt>ty2</tt>.</p>
3420
3421<h5>Arguments:</h5>
3422<p>The '<tt>inttoptr</tt>' instruction takes an <a href="#t_integer">integer</a>
3423value to cast, and a type to cast it to, which must be a
3424<a href="#t_pointer">pointer</a> type.
3425
3426<h5>Semantics:</h5>
3427<p>The '<tt>inttoptr</tt>' instruction converts <tt>value</tt> to type
3428<tt>ty2</tt> by applying either a zero extension or a truncation depending on
3429the size of the integer <tt>value</tt>. If <tt>value</tt> is larger than the
3430size of a pointer then a truncation is done. If <tt>value</tt> is smaller than
3431the size of a pointer then a zero extension is done. If they are the same size,
3432nothing is done (<i>no-op cast</i>).</p>
3433
3434<h5>Example:</h5>
3435<pre>
3436 %X = inttoptr i32 255 to i32* <i>; yields zero extension on 64-bit architecture</i>
3437 %X = inttoptr i32 255 to i32* <i>; yields no-op on 32-bit architecture</i>
3438 %Y = inttoptr i64 0 to i32* <i>; yields truncation on 32-bit architecture</i>
3439</pre>
3440</div>
3441
3442<!-- _______________________________________________________________________ -->
3443<div class="doc_subsubsection">
3444 <a name="i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a>
3445</div>
3446<div class="doc_text">
3447
3448<h5>Syntax:</h5>
3449<pre>
3450 &lt;result&gt; = bitcast &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3451</pre>
3452
3453<h5>Overview:</h5>
3454<p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
3455<tt>ty2</tt> without changing any bits.</p>
3456
3457<h5>Arguments:</h5>
3458<p>The '<tt>bitcast</tt>' instruction takes a value to cast, which must be
3459a first class value, and a type to cast it to, which must also be a <a
3460 href="#t_firstclass">first class</a> type. The bit sizes of <tt>value</tt>
3461and the destination type, <tt>ty2</tt>, must be identical. If the source
3462type is a pointer, the destination type must also be a pointer.</p>
3463
3464<h5>Semantics:</h5>
3465<p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
3466<tt>ty2</tt>. It is always a <i>no-op cast</i> because no bits change with
3467this conversion. The conversion is done as if the <tt>value</tt> had been
3468stored to memory and read back as type <tt>ty2</tt>. Pointer types may only be
3469converted to other pointer types with this instruction. To convert pointers to
3470other types, use the <a href="#i_inttoptr">inttoptr</a> or
3471<a href="#i_ptrtoint">ptrtoint</a> instructions first.</p>
3472
3473<h5>Example:</h5>
3474<pre>
3475 %X = bitcast i8 255 to i8 <i>; yields i8 :-1</i>
3476 %Y = bitcast i32* %x to sint* <i>; yields sint*:%x</i>
3477 %Z = bitcast <2xint> %V to i64; <i>; yields i64: %V</i>
3478</pre>
3479</div>
3480
3481<!-- ======================================================================= -->
3482<div class="doc_subsection"> <a name="otherops">Other Operations</a> </div>
3483<div class="doc_text">
3484<p>The instructions in this category are the "miscellaneous"
3485instructions, which defy better classification.</p>
3486</div>
3487
3488<!-- _______________________________________________________________________ -->
3489<div class="doc_subsubsection"><a name="i_icmp">'<tt>icmp</tt>' Instruction</a>
3490</div>
3491<div class="doc_text">
3492<h5>Syntax:</h5>
3493<pre> &lt;result&gt; = icmp &lt;cond&gt; &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {i1}:result</i>
3494</pre>
3495<h5>Overview:</h5>
3496<p>The '<tt>icmp</tt>' instruction returns a boolean value based on comparison
3497of its two integer operands.</p>
3498<h5>Arguments:</h5>
3499<p>The '<tt>icmp</tt>' instruction takes three operands. The first operand is
3500the condition code indicating the kind of comparison to perform. It is not
3501a value, just a keyword. The possible condition code are:
3502<ol>
3503 <li><tt>eq</tt>: equal</li>
3504 <li><tt>ne</tt>: not equal </li>
3505 <li><tt>ugt</tt>: unsigned greater than</li>
3506 <li><tt>uge</tt>: unsigned greater or equal</li>
3507 <li><tt>ult</tt>: unsigned less than</li>
3508 <li><tt>ule</tt>: unsigned less or equal</li>
3509 <li><tt>sgt</tt>: signed greater than</li>
3510 <li><tt>sge</tt>: signed greater or equal</li>
3511 <li><tt>slt</tt>: signed less than</li>
3512 <li><tt>sle</tt>: signed less or equal</li>
3513</ol>
3514<p>The remaining two arguments must be <a href="#t_integer">integer</a> or
3515<a href="#t_pointer">pointer</a> typed. They must also be identical types.</p>
3516<h5>Semantics:</h5>
3517<p>The '<tt>icmp</tt>' compares <tt>var1</tt> and <tt>var2</tt> according to
3518the condition code given as <tt>cond</tt>. The comparison performed always
3519yields a <a href="#t_primitive">i1</a> result, as follows:
3520<ol>
3521 <li><tt>eq</tt>: yields <tt>true</tt> if the operands are equal,
3522 <tt>false</tt> otherwise. No sign interpretation is necessary or performed.
3523 </li>
3524 <li><tt>ne</tt>: yields <tt>true</tt> if the operands are unequal,
3525 <tt>false</tt> otherwise. No sign interpretation is necessary or performed.
3526 <li><tt>ugt</tt>: interprets the operands as unsigned values and yields
3527 <tt>true</tt> if <tt>var1</tt> is greater than <tt>var2</tt>.</li>
3528 <li><tt>uge</tt>: interprets the operands as unsigned values and yields
3529 <tt>true</tt> if <tt>var1</tt> is greater than or equal to <tt>var2</tt>.</li>
3530 <li><tt>ult</tt>: interprets the operands as unsigned values and yields
3531 <tt>true</tt> if <tt>var1</tt> is less than <tt>var2</tt>.</li>
3532 <li><tt>ule</tt>: interprets the operands as unsigned values and yields
3533 <tt>true</tt> if <tt>var1</tt> is less than or equal to <tt>var2</tt>.</li>
3534 <li><tt>sgt</tt>: interprets the operands as signed values and yields
3535 <tt>true</tt> if <tt>var1</tt> is greater than <tt>var2</tt>.</li>
3536 <li><tt>sge</tt>: interprets the operands as signed values and yields
3537 <tt>true</tt> if <tt>var1</tt> is greater than or equal to <tt>var2</tt>.</li>
3538 <li><tt>slt</tt>: interprets the operands as signed values and yields
3539 <tt>true</tt> if <tt>var1</tt> is less than <tt>var2</tt>.</li>
3540 <li><tt>sle</tt>: interprets the operands as signed values and yields
3541 <tt>true</tt> if <tt>var1</tt> is less than or equal to <tt>var2</tt>.</li>
3542</ol>
3543<p>If the operands are <a href="#t_pointer">pointer</a> typed, the pointer
3544values are compared as if they were integers.</p>
3545
3546<h5>Example:</h5>
3547<pre> &lt;result&gt; = icmp eq i32 4, 5 <i>; yields: result=false</i>
3548 &lt;result&gt; = icmp ne float* %X, %X <i>; yields: result=false</i>
3549 &lt;result&gt; = icmp ult i16 4, 5 <i>; yields: result=true</i>
3550 &lt;result&gt; = icmp sgt i16 4, 5 <i>; yields: result=false</i>
3551 &lt;result&gt; = icmp ule i16 -4, 5 <i>; yields: result=false</i>
3552 &lt;result&gt; = icmp sge i16 4, 5 <i>; yields: result=false</i>
3553</pre>
3554</div>
3555
3556<!-- _______________________________________________________________________ -->
3557<div class="doc_subsubsection"><a name="i_fcmp">'<tt>fcmp</tt>' Instruction</a>
3558</div>
3559<div class="doc_text">
3560<h5>Syntax:</h5>
3561<pre> &lt;result&gt; = fcmp &lt;cond&gt; &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {i1}:result</i>
3562</pre>
3563<h5>Overview:</h5>
3564<p>The '<tt>fcmp</tt>' instruction returns a boolean value based on comparison
3565of its floating point operands.</p>
3566<h5>Arguments:</h5>
3567<p>The '<tt>fcmp</tt>' instruction takes three operands. The first operand is
3568the condition code indicating the kind of comparison to perform. It is not
3569a value, just a keyword. The possible condition code are:
3570<ol>
3571 <li><tt>false</tt>: no comparison, always returns false</li>
3572 <li><tt>oeq</tt>: ordered and equal</li>
3573 <li><tt>ogt</tt>: ordered and greater than </li>
3574 <li><tt>oge</tt>: ordered and greater than or equal</li>
3575 <li><tt>olt</tt>: ordered and less than </li>
3576 <li><tt>ole</tt>: ordered and less than or equal</li>
3577 <li><tt>one</tt>: ordered and not equal</li>
3578 <li><tt>ord</tt>: ordered (no nans)</li>
3579 <li><tt>ueq</tt>: unordered or equal</li>
3580 <li><tt>ugt</tt>: unordered or greater than </li>
3581 <li><tt>uge</tt>: unordered or greater than or equal</li>
3582 <li><tt>ult</tt>: unordered or less than </li>
3583 <li><tt>ule</tt>: unordered or less than or equal</li>
3584 <li><tt>une</tt>: unordered or not equal</li>
3585 <li><tt>uno</tt>: unordered (either nans)</li>
3586 <li><tt>true</tt>: no comparison, always returns true</li>
3587</ol>
3588<p><i>Ordered</i> means that neither operand is a QNAN while
3589<i>unordered</i> means that either operand may be a QNAN.</p>
3590<p>The <tt>val1</tt> and <tt>val2</tt> arguments must be
3591<a href="#t_floating">floating point</a> typed. They must have identical
3592types.</p>
3593<h5>Semantics:</h5>
3594<p>The '<tt>fcmp</tt>' compares <tt>var1</tt> and <tt>var2</tt> according to
3595the condition code given as <tt>cond</tt>. The comparison performed always
3596yields a <a href="#t_primitive">i1</a> result, as follows:
3597<ol>
3598 <li><tt>false</tt>: always yields <tt>false</tt>, regardless of operands.</li>
3599 <li><tt>oeq</tt>: yields <tt>true</tt> if both operands are not a QNAN and
3600 <tt>var1</tt> is equal to <tt>var2</tt>.</li>
3601 <li><tt>ogt</tt>: yields <tt>true</tt> if both operands are not a QNAN and
3602 <tt>var1</tt> is greather than <tt>var2</tt>.</li>
3603 <li><tt>oge</tt>: yields <tt>true</tt> if both operands are not a QNAN and
3604 <tt>var1</tt> is greater than or equal to <tt>var2</tt>.</li>
3605 <li><tt>olt</tt>: yields <tt>true</tt> if both operands are not a QNAN and
3606 <tt>var1</tt> is less than <tt>var2</tt>.</li>
3607 <li><tt>ole</tt>: yields <tt>true</tt> if both operands are not a QNAN and
3608 <tt>var1</tt> is less than or equal to <tt>var2</tt>.</li>
3609 <li><tt>one</tt>: yields <tt>true</tt> if both operands are not a QNAN and
3610 <tt>var1</tt> is not equal to <tt>var2</tt>.</li>
3611 <li><tt>ord</tt>: yields <tt>true</tt> if both operands are not a QNAN.</li>
3612 <li><tt>ueq</tt>: yields <tt>true</tt> if either operand is a QNAN or
3613 <tt>var1</tt> is equal to <tt>var2</tt>.</li>
3614 <li><tt>ugt</tt>: yields <tt>true</tt> if either operand is a QNAN or
3615 <tt>var1</tt> is greater than <tt>var2</tt>.</li>
3616 <li><tt>uge</tt>: yields <tt>true</tt> if either operand is a QNAN or
3617 <tt>var1</tt> is greater than or equal to <tt>var2</tt>.</li>
3618 <li><tt>ult</tt>: yields <tt>true</tt> if either operand is a QNAN or
3619 <tt>var1</tt> is less than <tt>var2</tt>.</li>
3620 <li><tt>ule</tt>: yields <tt>true</tt> if either operand is a QNAN or
3621 <tt>var1</tt> is less than or equal to <tt>var2</tt>.</li>
3622 <li><tt>une</tt>: yields <tt>true</tt> if either operand is a QNAN or
3623 <tt>var1</tt> is not equal to <tt>var2</tt>.</li>
3624 <li><tt>uno</tt>: yields <tt>true</tt> if either operand is a QNAN.</li>
3625 <li><tt>true</tt>: always yields <tt>true</tt>, regardless of operands.</li>
3626</ol>
3627
3628<h5>Example:</h5>
3629<pre> &lt;result&gt; = fcmp oeq float 4.0, 5.0 <i>; yields: result=false</i>
3630 &lt;result&gt; = icmp one float 4.0, 5.0 <i>; yields: result=true</i>
3631 &lt;result&gt; = icmp olt float 4.0, 5.0 <i>; yields: result=true</i>
3632 &lt;result&gt; = icmp ueq double 1.0, 2.0 <i>; yields: result=false</i>
3633</pre>
3634</div>
3635
3636<!-- _______________________________________________________________________ -->
3637<div class="doc_subsubsection"> <a name="i_phi">'<tt>phi</tt>'
3638Instruction</a> </div>
3639<div class="doc_text">
3640<h5>Syntax:</h5>
3641<pre> &lt;result&gt; = phi &lt;ty&gt; [ &lt;val0&gt;, &lt;label0&gt;], ...<br></pre>
3642<h5>Overview:</h5>
3643<p>The '<tt>phi</tt>' instruction is used to implement the &#966; node in
3644the SSA graph representing the function.</p>
3645<h5>Arguments:</h5>
3646<p>The type of the incoming values is specified with the first type
3647field. After this, the '<tt>phi</tt>' instruction takes a list of pairs
3648as arguments, with one pair for each predecessor basic block of the
3649current block. Only values of <a href="#t_firstclass">first class</a>
3650type may be used as the value arguments to the PHI node. Only labels
3651may be used as the label arguments.</p>
3652<p>There must be no non-phi instructions between the start of a basic
3653block and the PHI instructions: i.e. PHI instructions must be first in
3654a basic block.</p>
3655<h5>Semantics:</h5>
3656<p>At runtime, the '<tt>phi</tt>' instruction logically takes on the value
3657specified by the pair corresponding to the predecessor basic block that executed
3658just prior to the current block.</p>
3659<h5>Example:</h5>
3660<pre>Loop: ; Infinite loop that counts from 0 on up...<br> %indvar = phi i32 [ 0, %LoopHeader ], [ %nextindvar, %Loop ]<br> %nextindvar = add i32 %indvar, 1<br> br label %Loop<br></pre>
3661</div>
3662
3663<!-- _______________________________________________________________________ -->
3664<div class="doc_subsubsection">
3665 <a name="i_select">'<tt>select</tt>' Instruction</a>
3666</div>
3667
3668<div class="doc_text">
3669
3670<h5>Syntax:</h5>
3671
3672<pre>
3673 &lt;result&gt; = select i1 &lt;cond&gt;, &lt;ty&gt; &lt;val1&gt;, &lt;ty&gt; &lt;val2&gt; <i>; yields ty</i>
3674</pre>
3675
3676<h5>Overview:</h5>
3677
3678<p>
3679The '<tt>select</tt>' instruction is used to choose one value based on a
3680condition, without branching.
3681</p>
3682
3683
3684<h5>Arguments:</h5>
3685
3686<p>
3687The '<tt>select</tt>' instruction requires a boolean value indicating the condition, and two values of the same <a href="#t_firstclass">first class</a> type.
3688</p>
3689
3690<h5>Semantics:</h5>
3691
3692<p>
3693If the boolean condition evaluates to true, the instruction returns the first
3694value argument; otherwise, it returns the second value argument.
3695</p>
3696
3697<h5>Example:</h5>
3698
3699<pre>
3700 %X = select i1 true, i8 17, i8 42 <i>; yields i8:17</i>
3701</pre>
3702</div>
3703
3704
3705<!-- _______________________________________________________________________ -->
3706<div class="doc_subsubsection">
3707 <a name="i_call">'<tt>call</tt>' Instruction</a>
3708</div>
3709
3710<div class="doc_text">
3711
3712<h5>Syntax:</h5>
3713<pre>
Nick Lewycky93082fc2007-09-08 13:57:50 +0000[diff] [blame]3714 &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]3715</pre>
3716
3717<h5>Overview:</h5>
3718
3719<p>The '<tt>call</tt>' instruction represents a simple function call.</p>
3720
3721<h5>Arguments:</h5>
3722
3723<p>This instruction requires several arguments:</p>
3724
3725<ol>
3726 <li>
3727 <p>The optional "tail" marker indicates whether the callee function accesses
3728 any allocas or varargs in the caller. If the "tail" marker is present, the
3729 function call is eligible for tail call optimization. Note that calls may
3730 be marked "tail" even if they do not occur before a <a
3731 href="#i_ret"><tt>ret</tt></a> instruction.
3732 </li>
3733 <li>
3734 <p>The optional "cconv" marker indicates which <a href="#callingconv">calling
3735 convention</a> the call should use. If none is specified, the call defaults
3736 to using C calling conventions.
3737 </li>
3738 <li>
Nick Lewycky93082fc2007-09-08 13:57:50 +0000[diff] [blame]3739 <p>'<tt>ty</tt>': the type of the call instruction itself which is also
3740 the type of the return value. Functions that return no value are marked
3741 <tt><a href="#t_void">void</a></tt>.</p>
3742 </li>
3743 <li>
3744 <p>'<tt>fnty</tt>': shall be the signature of the pointer to function
3745 value being invoked. The argument types must match the types implied by
3746 this signature. This type can be omitted if the function is not varargs
3747 and if the function type does not return a pointer to a function.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3748 </li>
3749 <li>
3750 <p>'<tt>fnptrval</tt>': An LLVM value containing a pointer to a function to
3751 be invoked. In most cases, this is a direct function invocation, but
3752 indirect <tt>call</tt>s are just as possible, calling an arbitrary pointer
3753 to function value.</p>
3754 </li>
3755 <li>
3756 <p>'<tt>function args</tt>': argument list whose types match the
3757 function signature argument types. All arguments must be of
3758 <a href="#t_firstclass">first class</a> type. If the function signature
3759 indicates the function accepts a variable number of arguments, the extra
3760 arguments can be specified.</p>
3761 </li>
3762</ol>
3763
3764<h5>Semantics:</h5>
3765
3766<p>The '<tt>call</tt>' instruction is used to cause control flow to
3767transfer to a specified function, with its incoming arguments bound to
3768the specified values. Upon a '<tt><a href="#i_ret">ret</a></tt>'
3769instruction in the called function, control flow continues with the
3770instruction after the function call, and the return value of the
3771function is bound to the result argument. This is a simpler case of
3772the <a href="#i_invoke">invoke</a> instruction.</p>
3773
3774<h5>Example:</h5>
3775
3776<pre>
Nick Lewycky93082fc2007-09-08 13:57:50 +0000[diff] [blame]3777 %retval = call i32 @test(i32 %argc)
3778 call i32 (i8 *, ...)* @printf(i8 * %msg, i32 12, i8 42);
3779 %X = tail call i32 @foo()
3780 %Y = tail call <a href="#callingconv">fastcc</a> i32 @foo()
3781 %Z = call void %foo(i8 97 signext)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3782</pre>
3783
3784</div>
3785
3786<!-- _______________________________________________________________________ -->
3787<div class="doc_subsubsection">
3788 <a name="i_va_arg">'<tt>va_arg</tt>' Instruction</a>
3789</div>
3790
3791<div class="doc_text">
3792
3793<h5>Syntax:</h5>
3794
3795<pre>
3796 &lt;resultval&gt; = va_arg &lt;va_list*&gt; &lt;arglist&gt;, &lt;argty&gt;
3797</pre>
3798
3799<h5>Overview:</h5>
3800
3801<p>The '<tt>va_arg</tt>' instruction is used to access arguments passed through
3802the "variable argument" area of a function call. It is used to implement the
3803<tt>va_arg</tt> macro in C.</p>
3804
3805<h5>Arguments:</h5>
3806
3807<p>This instruction takes a <tt>va_list*</tt> value and the type of
3808the argument. It returns a value of the specified argument type and
3809increments the <tt>va_list</tt> to point to the next argument. The
3810actual type of <tt>va_list</tt> is target specific.</p>
3811
3812<h5>Semantics:</h5>
3813
3814<p>The '<tt>va_arg</tt>' instruction loads an argument of the specified
3815type from the specified <tt>va_list</tt> and causes the
3816<tt>va_list</tt> to point to the next argument. For more information,
3817see the variable argument handling <a href="#int_varargs">Intrinsic
3818Functions</a>.</p>
3819
3820<p>It is legal for this instruction to be called in a function which does not
3821take a variable number of arguments, for example, the <tt>vfprintf</tt>
3822function.</p>
3823
3824<p><tt>va_arg</tt> is an LLVM instruction instead of an <a
3825href="#intrinsics">intrinsic function</a> because it takes a type as an
3826argument.</p>
3827
3828<h5>Example:</h5>
3829
3830<p>See the <a href="#int_varargs">variable argument processing</a> section.</p>
3831
3832</div>
3833
3834<!-- *********************************************************************** -->
3835<div class="doc_section"> <a name="intrinsics">Intrinsic Functions</a> </div>
3836<!-- *********************************************************************** -->
3837
3838<div class="doc_text">
3839
3840<p>LLVM supports the notion of an "intrinsic function". These functions have
3841well known names and semantics and are required to follow certain restrictions.
3842Overall, these intrinsics represent an extension mechanism for the LLVM
3843language that does not require changing all of the transformations in LLVM when
3844adding to the language (or the bitcode reader/writer, the parser, etc...).</p>
3845
3846<p>Intrinsic function names must all start with an "<tt>llvm.</tt>" prefix. This
3847prefix is reserved in LLVM for intrinsic names; thus, function names may not
3848begin with this prefix. Intrinsic functions must always be external functions:
3849you cannot define the body of intrinsic functions. Intrinsic functions may
3850only be used in call or invoke instructions: it is illegal to take the address
3851of an intrinsic function. Additionally, because intrinsic functions are part
3852of the LLVM language, it is required if any are added that they be documented
3853here.</p>
3854
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]3855<p>Some intrinsic functions can be overloaded, i.e., the intrinsic represents
3856a family of functions that perform the same operation but on different data
3857types. Because LLVM can represent over 8 million different integer types,
3858overloading is used commonly to allow an intrinsic function to operate on any
3859integer type. One or more of the argument types or the result type can be
3860overloaded to accept any integer type. Argument types may also be defined as
3861exactly matching a previous argument's type or the result type. This allows an
3862intrinsic function which accepts multiple arguments, but needs all of them to
3863be of the same type, to only be overloaded with respect to a single argument or
3864the result.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3865
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]3866<p>Overloaded intrinsics will have the names of its overloaded argument types
3867encoded into its function name, each preceded by a period. Only those types
3868which are overloaded result in a name suffix. Arguments whose type is matched
3869against another type do not. For example, the <tt>llvm.ctpop</tt> function can
3870take an integer of any width and returns an integer of exactly the same integer
3871width. This leads to a family of functions such as
3872<tt>i8 @llvm.ctpop.i8(i8 %val)</tt> and <tt>i29 @llvm.ctpop.i29(i29 %val)</tt>.
3873Only one type, the return type, is overloaded, and only one type suffix is
3874required. Because the argument's type is matched against the return type, it
3875does not require its own name suffix.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3876
3877<p>To learn how to add an intrinsic function, please see the
3878<a href="ExtendingLLVM.html">Extending LLVM Guide</a>.
3879</p>
3880
3881</div>
3882
3883<!-- ======================================================================= -->
3884<div class="doc_subsection">
3885 <a name="int_varargs">Variable Argument Handling Intrinsics</a>
3886</div>
3887
3888<div class="doc_text">
3889
3890<p>Variable argument support is defined in LLVM with the <a
3891 href="#i_va_arg"><tt>va_arg</tt></a> instruction and these three
3892intrinsic functions. These functions are related to the similarly
3893named macros defined in the <tt>&lt;stdarg.h&gt;</tt> header file.</p>
3894
3895<p>All of these functions operate on arguments that use a
3896target-specific value type "<tt>va_list</tt>". The LLVM assembly
3897language reference manual does not define what this type is, so all
3898transformations should be prepared to handle these functions regardless of
3899the type used.</p>
3900
3901<p>This example shows how the <a href="#i_va_arg"><tt>va_arg</tt></a>
3902instruction and the variable argument handling intrinsic functions are
3903used.</p>
3904
3905<div class="doc_code">
3906<pre>
3907define i32 @test(i32 %X, ...) {
3908 ; Initialize variable argument processing
3909 %ap = alloca i8*
3910 %ap2 = bitcast i8** %ap to i8*
3911 call void @llvm.va_start(i8* %ap2)
3912
3913 ; Read a single integer argument
3914 %tmp = va_arg i8** %ap, i32
3915
3916 ; Demonstrate usage of llvm.va_copy and llvm.va_end
3917 %aq = alloca i8*
3918 %aq2 = bitcast i8** %aq to i8*
3919 call void @llvm.va_copy(i8* %aq2, i8* %ap2)
3920 call void @llvm.va_end(i8* %aq2)
3921
3922 ; Stop processing of arguments.
3923 call void @llvm.va_end(i8* %ap2)
3924 ret i32 %tmp
3925}
3926
3927declare void @llvm.va_start(i8*)
3928declare void @llvm.va_copy(i8*, i8*)
3929declare void @llvm.va_end(i8*)
3930</pre>
3931</div>
3932
3933</div>
3934
3935<!-- _______________________________________________________________________ -->
3936<div class="doc_subsubsection">
3937 <a name="int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a>
3938</div>
3939
3940
3941<div class="doc_text">
3942<h5>Syntax:</h5>
3943<pre> declare void %llvm.va_start(i8* &lt;arglist&gt;)<br></pre>
3944<h5>Overview:</h5>
3945<P>The '<tt>llvm.va_start</tt>' intrinsic initializes
3946<tt>*&lt;arglist&gt;</tt> for subsequent use by <tt><a
3947href="#i_va_arg">va_arg</a></tt>.</p>
3948
3949<h5>Arguments:</h5>
3950
3951<P>The argument is a pointer to a <tt>va_list</tt> element to initialize.</p>
3952
3953<h5>Semantics:</h5>
3954
3955<P>The '<tt>llvm.va_start</tt>' intrinsic works just like the <tt>va_start</tt>
3956macro available in C. In a target-dependent way, it initializes the
3957<tt>va_list</tt> element to which the argument points, so that the next call to
3958<tt>va_arg</tt> will produce the first variable argument passed to the function.
3959Unlike the C <tt>va_start</tt> macro, this intrinsic does not need to know the
3960last argument of the function as the compiler can figure that out.</p>
3961
3962</div>
3963
3964<!-- _______________________________________________________________________ -->
3965<div class="doc_subsubsection">
3966 <a name="int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a>
3967</div>
3968
3969<div class="doc_text">
3970<h5>Syntax:</h5>
3971<pre> declare void @llvm.va_end(i8* &lt;arglist&gt;)<br></pre>
3972<h5>Overview:</h5>
3973
3974<p>The '<tt>llvm.va_end</tt>' intrinsic destroys <tt>*&lt;arglist&gt;</tt>,
3975which has been initialized previously with <tt><a href="#int_va_start">llvm.va_start</a></tt>
3976or <tt><a href="#i_va_copy">llvm.va_copy</a></tt>.</p>
3977
3978<h5>Arguments:</h5>
3979
3980<p>The argument is a pointer to a <tt>va_list</tt> to destroy.</p>
3981
3982<h5>Semantics:</h5>
3983
3984<p>The '<tt>llvm.va_end</tt>' intrinsic works just like the <tt>va_end</tt>
3985macro available in C. In a target-dependent way, it destroys the
3986<tt>va_list</tt> element to which the argument points. Calls to <a
3987href="#int_va_start"><tt>llvm.va_start</tt></a> and <a href="#int_va_copy">
3988<tt>llvm.va_copy</tt></a> must be matched exactly with calls to
3989<tt>llvm.va_end</tt>.</p>
3990
3991</div>
3992
3993<!-- _______________________________________________________________________ -->
3994<div class="doc_subsubsection">
3995 <a name="int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a>
3996</div>
3997
3998<div class="doc_text">
3999
4000<h5>Syntax:</h5>
4001
4002<pre>
4003 declare void @llvm.va_copy(i8* &lt;destarglist&gt;, i8* &lt;srcarglist&gt;)
4004</pre>
4005
4006<h5>Overview:</h5>
4007
4008<p>The '<tt>llvm.va_copy</tt>' intrinsic copies the current argument position
4009from the source argument list to the destination argument list.</p>
4010
4011<h5>Arguments:</h5>
4012
4013<p>The first argument is a pointer to a <tt>va_list</tt> element to initialize.
4014The second argument is a pointer to a <tt>va_list</tt> element to copy from.</p>
4015
4016
4017<h5>Semantics:</h5>
4018
4019<p>The '<tt>llvm.va_copy</tt>' intrinsic works just like the <tt>va_copy</tt>
4020macro available in C. In a target-dependent way, it copies the source
4021<tt>va_list</tt> element into the destination <tt>va_list</tt> element. This
4022intrinsic is necessary because the <tt><a href="#int_va_start">
4023llvm.va_start</a></tt> intrinsic may be arbitrarily complex and require, for
4024example, memory allocation.</p>
4025
4026</div>
4027
4028<!-- ======================================================================= -->
4029<div class="doc_subsection">
4030 <a name="int_gc">Accurate Garbage Collection Intrinsics</a>
4031</div>
4032
4033<div class="doc_text">
4034
4035<p>
4036LLVM support for <a href="GarbageCollection.html">Accurate Garbage
4037Collection</a> requires the implementation and generation of these intrinsics.
4038These intrinsics allow identification of <a href="#int_gcroot">GC roots on the
4039stack</a>, as well as garbage collector implementations that require <a
4040href="#int_gcread">read</a> and <a href="#int_gcwrite">write</a> barriers.
4041Front-ends for type-safe garbage collected languages should generate these
4042intrinsics to make use of the LLVM garbage collectors. For more details, see <a
4043href="GarbageCollection.html">Accurate Garbage Collection with LLVM</a>.
4044</p>
Christopher Lambcfe00962007-12-17 01:00:21 +0000[diff] [blame]4045
4046<p>The garbage collection intrinsics only operate on objects in the generic
4047 address space (address space zero).</p>
4048
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4049</div>
4050
4051<!-- _______________________________________________________________________ -->
4052<div class="doc_subsubsection">
4053 <a name="int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a>
4054</div>
4055
4056<div class="doc_text">
4057
4058<h5>Syntax:</h5>
4059
4060<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4061 declare void @llvm.gcroot(i8** %ptrloc, i8* %metadata)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4062</pre>
4063
4064<h5>Overview:</h5>
4065
4066<p>The '<tt>llvm.gcroot</tt>' intrinsic declares the existence of a GC root to
4067the code generator, and allows some metadata to be associated with it.</p>
4068
4069<h5>Arguments:</h5>
4070
4071<p>The first argument specifies the address of a stack object that contains the
4072root pointer. The second pointer (which must be either a constant or a global
4073value address) contains the meta-data to be associated with the root.</p>
4074
4075<h5>Semantics:</h5>
4076
4077<p>At runtime, a call to this intrinsics stores a null pointer into the "ptrloc"
4078location. At compile-time, the code generator generates information to allow
Gordon Henriksen40393542007-12-25 02:31:26 +0000[diff] [blame]4079the runtime to find the pointer at GC safe points. The '<tt>llvm.gcroot</tt>'
4080intrinsic may only be used in a function which <a href="#gc">specifies a GC
4081algorithm</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4082
4083</div>
4084
4085
4086<!-- _______________________________________________________________________ -->
4087<div class="doc_subsubsection">
4088 <a name="int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a>
4089</div>
4090
4091<div class="doc_text">
4092
4093<h5>Syntax:</h5>
4094
4095<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4096 declare i8* @llvm.gcread(i8* %ObjPtr, i8** %Ptr)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4097</pre>
4098
4099<h5>Overview:</h5>
4100
4101<p>The '<tt>llvm.gcread</tt>' intrinsic identifies reads of references from heap
4102locations, allowing garbage collector implementations that require read
4103barriers.</p>
4104
4105<h5>Arguments:</h5>
4106
4107<p>The second argument is the address to read from, which should be an address
4108allocated from the garbage collector. The first object is a pointer to the
4109start of the referenced object, if needed by the language runtime (otherwise
4110null).</p>
4111
4112<h5>Semantics:</h5>
4113
4114<p>The '<tt>llvm.gcread</tt>' intrinsic has the same semantics as a load
4115instruction, but may be replaced with substantially more complex code by the
Gordon Henriksen40393542007-12-25 02:31:26 +0000[diff] [blame]4116garbage collector runtime, as needed. The '<tt>llvm.gcread</tt>' intrinsic
4117may only be used in a function which <a href="#gc">specifies a GC
4118algorithm</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4119
4120</div>
4121
4122
4123<!-- _______________________________________________________________________ -->
4124<div class="doc_subsubsection">
4125 <a name="int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a>
4126</div>
4127
4128<div class="doc_text">
4129
4130<h5>Syntax:</h5>
4131
4132<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4133 declare void @llvm.gcwrite(i8* %P1, i8* %Obj, i8** %P2)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4134</pre>
4135
4136<h5>Overview:</h5>
4137
4138<p>The '<tt>llvm.gcwrite</tt>' intrinsic identifies writes of references to heap
4139locations, allowing garbage collector implementations that require write
4140barriers (such as generational or reference counting collectors).</p>
4141
4142<h5>Arguments:</h5>
4143
4144<p>The first argument is the reference to store, the second is the start of the
4145object to store it to, and the third is the address of the field of Obj to
4146store to. If the runtime does not require a pointer to the object, Obj may be
4147null.</p>
4148
4149<h5>Semantics:</h5>
4150
4151<p>The '<tt>llvm.gcwrite</tt>' intrinsic has the same semantics as a store
4152instruction, but may be replaced with substantially more complex code by the
Gordon Henriksen40393542007-12-25 02:31:26 +0000[diff] [blame]4153garbage collector runtime, as needed. The '<tt>llvm.gcwrite</tt>' intrinsic
4154may only be used in a function which <a href="#gc">specifies a GC
4155algorithm</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4156
4157</div>
4158
4159
4160
4161<!-- ======================================================================= -->
4162<div class="doc_subsection">
4163 <a name="int_codegen">Code Generator Intrinsics</a>
4164</div>
4165
4166<div class="doc_text">
4167<p>
4168These intrinsics are provided by LLVM to expose special features that may only
4169be implemented with code generator support.
4170</p>
4171
4172</div>
4173
4174<!-- _______________________________________________________________________ -->
4175<div class="doc_subsubsection">
4176 <a name="int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a>
4177</div>
4178
4179<div class="doc_text">
4180
4181<h5>Syntax:</h5>
4182<pre>
4183 declare i8 *@llvm.returnaddress(i32 &lt;level&gt;)
4184</pre>
4185
4186<h5>Overview:</h5>
4187
4188<p>
4189The '<tt>llvm.returnaddress</tt>' intrinsic attempts to compute a
4190target-specific value indicating the return address of the current function
4191or one of its callers.
4192</p>
4193
4194<h5>Arguments:</h5>
4195
4196<p>
4197The argument to this intrinsic indicates which function to return the address
4198for. Zero indicates the calling function, one indicates its caller, etc. The
4199argument is <b>required</b> to be a constant integer value.
4200</p>
4201
4202<h5>Semantics:</h5>
4203
4204<p>
4205The '<tt>llvm.returnaddress</tt>' intrinsic either returns a pointer indicating
4206the return address of the specified call frame, or zero if it cannot be
4207identified. The value returned by this intrinsic is likely to be incorrect or 0
4208for arguments other than zero, so it should only be used for debugging purposes.
4209</p>
4210
4211<p>
4212Note that calling this intrinsic does not prevent function inlining or other
4213aggressive transformations, so the value returned may not be that of the obvious
4214source-language caller.
4215</p>
4216</div>
4217
4218
4219<!-- _______________________________________________________________________ -->
4220<div class="doc_subsubsection">
4221 <a name="int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a>
4222</div>
4223
4224<div class="doc_text">
4225
4226<h5>Syntax:</h5>
4227<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4228 declare i8 *@llvm.frameaddress(i32 &lt;level&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4229</pre>
4230
4231<h5>Overview:</h5>
4232
4233<p>
4234The '<tt>llvm.frameaddress</tt>' intrinsic attempts to return the
4235target-specific frame pointer value for the specified stack frame.
4236</p>
4237
4238<h5>Arguments:</h5>
4239
4240<p>
4241The argument to this intrinsic indicates which function to return the frame
4242pointer for. Zero indicates the calling function, one indicates its caller,
4243etc. The argument is <b>required</b> to be a constant integer value.
4244</p>
4245
4246<h5>Semantics:</h5>
4247
4248<p>
4249The '<tt>llvm.frameaddress</tt>' intrinsic either returns a pointer indicating
4250the frame address of the specified call frame, or zero if it cannot be
4251identified. The value returned by this intrinsic is likely to be incorrect or 0
4252for arguments other than zero, so it should only be used for debugging purposes.
4253</p>
4254
4255<p>
4256Note that calling this intrinsic does not prevent function inlining or other
4257aggressive transformations, so the value returned may not be that of the obvious
4258source-language caller.
4259</p>
4260</div>
4261
4262<!-- _______________________________________________________________________ -->
4263<div class="doc_subsubsection">
4264 <a name="int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a>
4265</div>
4266
4267<div class="doc_text">
4268
4269<h5>Syntax:</h5>
4270<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4271 declare i8 *@llvm.stacksave()
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4272</pre>
4273
4274<h5>Overview:</h5>
4275
4276<p>
4277The '<tt>llvm.stacksave</tt>' intrinsic is used to remember the current state of
4278the function stack, for use with <a href="#int_stackrestore">
4279<tt>llvm.stackrestore</tt></a>. This is useful for implementing language
4280features like scoped automatic variable sized arrays in C99.
4281</p>
4282
4283<h5>Semantics:</h5>
4284
4285<p>
4286This intrinsic returns a opaque pointer value that can be passed to <a
4287href="#int_stackrestore"><tt>llvm.stackrestore</tt></a>. When an
4288<tt>llvm.stackrestore</tt> intrinsic is executed with a value saved from
4289<tt>llvm.stacksave</tt>, it effectively restores the state of the stack to the
4290state it was in when the <tt>llvm.stacksave</tt> intrinsic executed. In
4291practice, this pops any <a href="#i_alloca">alloca</a> blocks from the stack
4292that were allocated after the <tt>llvm.stacksave</tt> was executed.
4293</p>
4294
4295</div>
4296
4297<!-- _______________________________________________________________________ -->
4298<div class="doc_subsubsection">
4299 <a name="int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a>
4300</div>
4301
4302<div class="doc_text">
4303
4304<h5>Syntax:</h5>
4305<pre>
4306 declare void @llvm.stackrestore(i8 * %ptr)
4307</pre>
4308
4309<h5>Overview:</h5>
4310
4311<p>
4312The '<tt>llvm.stackrestore</tt>' intrinsic is used to restore the state of
4313the function stack to the state it was in when the corresponding <a
4314href="#int_stacksave"><tt>llvm.stacksave</tt></a> intrinsic executed. This is
4315useful for implementing language features like scoped automatic variable sized
4316arrays in C99.
4317</p>
4318
4319<h5>Semantics:</h5>
4320
4321<p>
4322See the description for <a href="#int_stacksave"><tt>llvm.stacksave</tt></a>.
4323</p>
4324
4325</div>
4326
4327
4328<!-- _______________________________________________________________________ -->
4329<div class="doc_subsubsection">
4330 <a name="int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a>
4331</div>
4332
4333<div class="doc_text">
4334
4335<h5>Syntax:</h5>
4336<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4337 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]4338</pre>
4339
4340<h5>Overview:</h5>
4341
4342
4343<p>
4344The '<tt>llvm.prefetch</tt>' intrinsic is a hint to the code generator to insert
4345a prefetch instruction if supported; otherwise, it is a noop. Prefetches have
4346no
4347effect on the behavior of the program but can change its performance
4348characteristics.
4349</p>
4350
4351<h5>Arguments:</h5>
4352
4353<p>
4354<tt>address</tt> is the address to be prefetched, <tt>rw</tt> is the specifier
4355determining if the fetch should be for a read (0) or write (1), and
4356<tt>locality</tt> is a temporal locality specifier ranging from (0) - no
4357locality, to (3) - extremely local keep in cache. The <tt>rw</tt> and
4358<tt>locality</tt> arguments must be constant integers.
4359</p>
4360
4361<h5>Semantics:</h5>
4362
4363<p>
4364This intrinsic does not modify the behavior of the program. In particular,
4365prefetches cannot trap and do not produce a value. On targets that support this
4366intrinsic, the prefetch can provide hints to the processor cache for better
4367performance.
4368</p>
4369
4370</div>
4371
4372<!-- _______________________________________________________________________ -->
4373<div class="doc_subsubsection">
4374 <a name="int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a>
4375</div>
4376
4377<div class="doc_text">
4378
4379<h5>Syntax:</h5>
4380<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4381 declare void @llvm.pcmarker(i32 &lt;id&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4382</pre>
4383
4384<h5>Overview:</h5>
4385
4386
4387<p>
4388The '<tt>llvm.pcmarker</tt>' intrinsic is a method to export a Program Counter
4389(PC) in a region of
4390code to simulators and other tools. The method is target specific, but it is
4391expected that the marker will use exported symbols to transmit the PC of the marker.
4392The marker makes no guarantees that it will remain with any specific instruction
4393after optimizations. It is possible that the presence of a marker will inhibit
4394optimizations. The intended use is to be inserted after optimizations to allow
4395correlations of simulation runs.
4396</p>
4397
4398<h5>Arguments:</h5>
4399
4400<p>
4401<tt>id</tt> is a numerical id identifying the marker.
4402</p>
4403
4404<h5>Semantics:</h5>
4405
4406<p>
4407This intrinsic does not modify the behavior of the program. Backends that do not
4408support this intrinisic may ignore it.
4409</p>
4410
4411</div>
4412
4413<!-- _______________________________________________________________________ -->
4414<div class="doc_subsubsection">
4415 <a name="int_readcyclecounter">'<tt>llvm.readcyclecounter</tt>' Intrinsic</a>
4416</div>
4417
4418<div class="doc_text">
4419
4420<h5>Syntax:</h5>
4421<pre>
4422 declare i64 @llvm.readcyclecounter( )
4423</pre>
4424
4425<h5>Overview:</h5>
4426
4427
4428<p>
4429The '<tt>llvm.readcyclecounter</tt>' intrinsic provides access to the cycle
4430counter register (or similar low latency, high accuracy clocks) on those targets
4431that support it. On X86, it should map to RDTSC. On Alpha, it should map to RPCC.
4432As the backing counters overflow quickly (on the order of 9 seconds on alpha), this
4433should only be used for small timings.
4434</p>
4435
4436<h5>Semantics:</h5>
4437
4438<p>
4439When directly supported, reading the cycle counter should not modify any memory.
4440Implementations are allowed to either return a application specific value or a
4441system wide value. On backends without support, this is lowered to a constant 0.
4442</p>
4443
4444</div>
4445
4446<!-- ======================================================================= -->
4447<div class="doc_subsection">
4448 <a name="int_libc">Standard C Library Intrinsics</a>
4449</div>
4450
4451<div class="doc_text">
4452<p>
4453LLVM provides intrinsics for a few important standard C library functions.
4454These intrinsics allow source-language front-ends to pass information about the
4455alignment of the pointer arguments to the code generator, providing opportunity
4456for more efficient code generation.
4457</p>
4458
4459</div>
4460
4461<!-- _______________________________________________________________________ -->
4462<div class="doc_subsubsection">
4463 <a name="int_memcpy">'<tt>llvm.memcpy</tt>' Intrinsic</a>
4464</div>
4465
4466<div class="doc_text">
4467
4468<h5>Syntax:</h5>
4469<pre>
4470 declare void @llvm.memcpy.i32(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
4471 i32 &lt;len&gt;, i32 &lt;align&gt;)
4472 declare void @llvm.memcpy.i64(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
4473 i64 &lt;len&gt;, i32 &lt;align&gt;)
4474</pre>
4475
4476<h5>Overview:</h5>
4477
4478<p>
4479The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the source
4480location to the destination location.
4481</p>
4482
4483<p>
4484Note that, unlike the standard libc function, the <tt>llvm.memcpy.*</tt>
4485intrinsics do not return a value, and takes an extra alignment argument.
4486</p>
4487
4488<h5>Arguments:</h5>
4489
4490<p>
4491The first argument is a pointer to the destination, the second is a pointer to
4492the source. The third argument is an integer argument
4493specifying the number of bytes to copy, and the fourth argument is the alignment
4494of the source and destination locations.
4495</p>
4496
4497<p>
4498If the call to this intrinisic has an alignment value that is not 0 or 1, then
4499the caller guarantees that both the source and destination pointers are aligned
4500to that boundary.
4501</p>
4502
4503<h5>Semantics:</h5>
4504
4505<p>
4506The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the source
4507location to the destination location, which are not allowed to overlap. It
4508copies "len" bytes of memory over. If the argument is known to be aligned to
4509some boundary, this can be specified as the fourth argument, otherwise it should
4510be set to 0 or 1.
4511</p>
4512</div>
4513
4514
4515<!-- _______________________________________________________________________ -->
4516<div class="doc_subsubsection">
4517 <a name="int_memmove">'<tt>llvm.memmove</tt>' Intrinsic</a>
4518</div>
4519
4520<div class="doc_text">
4521
4522<h5>Syntax:</h5>
4523<pre>
4524 declare void @llvm.memmove.i32(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
4525 i32 &lt;len&gt;, i32 &lt;align&gt;)
4526 declare void @llvm.memmove.i64(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
4527 i64 &lt;len&gt;, i32 &lt;align&gt;)
4528</pre>
4529
4530<h5>Overview:</h5>
4531
4532<p>
4533The '<tt>llvm.memmove.*</tt>' intrinsics move a block of memory from the source
4534location to the destination location. It is similar to the
Chris Lattnerdba16ea2008-01-06 19:51:52 +0000[diff] [blame]4535'<tt>llvm.memcpy</tt>' intrinsic but allows the two memory locations to overlap.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4536</p>
4537
4538<p>
4539Note that, unlike the standard libc function, the <tt>llvm.memmove.*</tt>
4540intrinsics do not return a value, and takes an extra alignment argument.
4541</p>
4542
4543<h5>Arguments:</h5>
4544
4545<p>
4546The first argument is a pointer to the destination, the second is a pointer to
4547the source. The third argument is an integer argument
4548specifying the number of bytes to copy, and the fourth argument is the alignment
4549of the source and destination locations.
4550</p>
4551
4552<p>
4553If the call to this intrinisic has an alignment value that is not 0 or 1, then
4554the caller guarantees that the source and destination pointers are aligned to
4555that boundary.
4556</p>
4557
4558<h5>Semantics:</h5>
4559
4560<p>
4561The '<tt>llvm.memmove.*</tt>' intrinsics copy a block of memory from the source
4562location to the destination location, which may overlap. It
4563copies "len" bytes of memory over. If the argument is known to be aligned to
4564some boundary, this can be specified as the fourth argument, otherwise it should
4565be set to 0 or 1.
4566</p>
4567</div>
4568
4569
4570<!-- _______________________________________________________________________ -->
4571<div class="doc_subsubsection">
4572 <a name="int_memset">'<tt>llvm.memset.*</tt>' Intrinsics</a>
4573</div>
4574
4575<div class="doc_text">
4576
4577<h5>Syntax:</h5>
4578<pre>
4579 declare void @llvm.memset.i32(i8 * &lt;dest&gt;, i8 &lt;val&gt;,
4580 i32 &lt;len&gt;, i32 &lt;align&gt;)
4581 declare void @llvm.memset.i64(i8 * &lt;dest&gt;, i8 &lt;val&gt;,
4582 i64 &lt;len&gt;, i32 &lt;align&gt;)
4583</pre>
4584
4585<h5>Overview:</h5>
4586
4587<p>
4588The '<tt>llvm.memset.*</tt>' intrinsics fill a block of memory with a particular
4589byte value.
4590</p>
4591
4592<p>
4593Note that, unlike the standard libc function, the <tt>llvm.memset</tt> intrinsic
4594does not return a value, and takes an extra alignment argument.
4595</p>
4596
4597<h5>Arguments:</h5>
4598
4599<p>
4600The first argument is a pointer to the destination to fill, the second is the
4601byte value to fill it with, the third argument is an integer
4602argument specifying the number of bytes to fill, and the fourth argument is the
4603known alignment of destination location.
4604</p>
4605
4606<p>
4607If the call to this intrinisic has an alignment value that is not 0 or 1, then
4608the caller guarantees that the destination pointer is aligned to that boundary.
4609</p>
4610
4611<h5>Semantics:</h5>
4612
4613<p>
4614The '<tt>llvm.memset.*</tt>' intrinsics fill "len" bytes of memory starting at
4615the
4616destination location. If the argument is known to be aligned to some boundary,
4617this can be specified as the fourth argument, otherwise it should be set to 0 or
46181.
4619</p>
4620</div>
4621
4622
4623<!-- _______________________________________________________________________ -->
4624<div class="doc_subsubsection">
4625 <a name="int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a>
4626</div>
4627
4628<div class="doc_text">
4629
4630<h5>Syntax:</h5>
Dale Johannesenf9adbb62007-10-02 17:47:38 +0000[diff] [blame]4631<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]4632floating point or vector of floating point type. Not all targets support all
4633types however.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4634<pre>
Dale Johannesenf9adbb62007-10-02 17:47:38 +0000[diff] [blame]4635 declare float @llvm.sqrt.f32(float %Val)
4636 declare double @llvm.sqrt.f64(double %Val)
4637 declare x86_fp80 @llvm.sqrt.f80(x86_fp80 %Val)
4638 declare fp128 @llvm.sqrt.f128(fp128 %Val)
4639 declare ppc_fp128 @llvm.sqrt.ppcf128(ppc_fp128 %Val)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4640</pre>
4641
4642<h5>Overview:</h5>
4643
4644<p>
4645The '<tt>llvm.sqrt</tt>' intrinsics return the sqrt of the specified operand,
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]4646returning the same value as the libm '<tt>sqrt</tt>' functions would. Unlike
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4647<tt>sqrt</tt> in libm, however, <tt>llvm.sqrt</tt> has undefined behavior for
4648negative numbers (which allows for better optimization).
4649</p>
4650
4651<h5>Arguments:</h5>
4652
4653<p>
4654The argument and return value are floating point numbers of the same type.
4655</p>
4656
4657<h5>Semantics:</h5>
4658
4659<p>
4660This function returns the sqrt of the specified operand if it is a nonnegative
4661floating point number.
4662</p>
4663</div>
4664
4665<!-- _______________________________________________________________________ -->
4666<div class="doc_subsubsection">
4667 <a name="int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a>
4668</div>
4669
4670<div class="doc_text">
4671
4672<h5>Syntax:</h5>
Dale Johannesenf9adbb62007-10-02 17:47:38 +0000[diff] [blame]4673<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]4674floating point or vector of floating point type. Not all targets support all
4675types however.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4676<pre>
Dale Johannesenf9adbb62007-10-02 17:47:38 +0000[diff] [blame]4677 declare float @llvm.powi.f32(float %Val, i32 %power)
4678 declare double @llvm.powi.f64(double %Val, i32 %power)
4679 declare x86_fp80 @llvm.powi.f80(x86_fp80 %Val, i32 %power)
4680 declare fp128 @llvm.powi.f128(fp128 %Val, i32 %power)
4681 declare ppc_fp128 @llvm.powi.ppcf128(ppc_fp128 %Val, i32 %power)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4682</pre>
4683
4684<h5>Overview:</h5>
4685
4686<p>
4687The '<tt>llvm.powi.*</tt>' intrinsics return the first operand raised to the
4688specified (positive or negative) power. The order of evaluation of
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]4689multiplications is not defined. When a vector of floating point type is
4690used, the second argument remains a scalar integer value.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4691</p>
4692
4693<h5>Arguments:</h5>
4694
4695<p>
4696The second argument is an integer power, and the first is a value to raise to
4697that power.
4698</p>
4699
4700<h5>Semantics:</h5>
4701
4702<p>
4703This function returns the first value raised to the second power with an
4704unspecified sequence of rounding operations.</p>
4705</div>
4706
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]4707<!-- _______________________________________________________________________ -->
4708<div class="doc_subsubsection">
4709 <a name="int_sin">'<tt>llvm.sin.*</tt>' Intrinsic</a>
4710</div>
4711
4712<div class="doc_text">
4713
4714<h5>Syntax:</h5>
4715<p>This is an overloaded intrinsic. You can use <tt>llvm.sin</tt> on any
4716floating point or vector of floating point type. Not all targets support all
4717types however.
4718<pre>
4719 declare float @llvm.sin.f32(float %Val)
4720 declare double @llvm.sin.f64(double %Val)
4721 declare x86_fp80 @llvm.sin.f80(x86_fp80 %Val)
4722 declare fp128 @llvm.sin.f128(fp128 %Val)
4723 declare ppc_fp128 @llvm.sin.ppcf128(ppc_fp128 %Val)
4724</pre>
4725
4726<h5>Overview:</h5>
4727
4728<p>
4729The '<tt>llvm.sin.*</tt>' intrinsics return the sine of the operand.
4730</p>
4731
4732<h5>Arguments:</h5>
4733
4734<p>
4735The argument and return value are floating point numbers of the same type.
4736</p>
4737
4738<h5>Semantics:</h5>
4739
4740<p>
4741This function returns the sine of the specified operand, returning the
4742same values as the libm <tt>sin</tt> functions would, and handles error
Dan Gohmaneaba92e2007-10-17 18:05:13 +0000[diff] [blame]4743conditions in the same way.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]4744</div>
4745
4746<!-- _______________________________________________________________________ -->
4747<div class="doc_subsubsection">
4748 <a name="int_cos">'<tt>llvm.cos.*</tt>' Intrinsic</a>
4749</div>
4750
4751<div class="doc_text">
4752
4753<h5>Syntax:</h5>
4754<p>This is an overloaded intrinsic. You can use <tt>llvm.cos</tt> on any
4755floating point or vector of floating point type. Not all targets support all
4756types however.
4757<pre>
4758 declare float @llvm.cos.f32(float %Val)
4759 declare double @llvm.cos.f64(double %Val)
4760 declare x86_fp80 @llvm.cos.f80(x86_fp80 %Val)
4761 declare fp128 @llvm.cos.f128(fp128 %Val)
4762 declare ppc_fp128 @llvm.cos.ppcf128(ppc_fp128 %Val)
4763</pre>
4764
4765<h5>Overview:</h5>
4766
4767<p>
4768The '<tt>llvm.cos.*</tt>' intrinsics return the cosine of the operand.
4769</p>
4770
4771<h5>Arguments:</h5>
4772
4773<p>
4774The argument and return value are floating point numbers of the same type.
4775</p>
4776
4777<h5>Semantics:</h5>
4778
4779<p>
4780This function returns the cosine of the specified operand, returning the
4781same values as the libm <tt>cos</tt> functions would, and handles error
Dan Gohmaneaba92e2007-10-17 18:05:13 +0000[diff] [blame]4782conditions in the same way.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]4783</div>
4784
4785<!-- _______________________________________________________________________ -->
4786<div class="doc_subsubsection">
4787 <a name="int_pow">'<tt>llvm.pow.*</tt>' Intrinsic</a>
4788</div>
4789
4790<div class="doc_text">
4791
4792<h5>Syntax:</h5>
4793<p>This is an overloaded intrinsic. You can use <tt>llvm.pow</tt> on any
4794floating point or vector of floating point type. Not all targets support all
4795types however.
4796<pre>
4797 declare float @llvm.pow.f32(float %Val, float %Power)
4798 declare double @llvm.pow.f64(double %Val, double %Power)
4799 declare x86_fp80 @llvm.pow.f80(x86_fp80 %Val, x86_fp80 %Power)
4800 declare fp128 @llvm.pow.f128(fp128 %Val, fp128 %Power)
4801 declare ppc_fp128 @llvm.pow.ppcf128(ppc_fp128 %Val, ppc_fp128 Power)
4802</pre>
4803
4804<h5>Overview:</h5>
4805
4806<p>
4807The '<tt>llvm.pow.*</tt>' intrinsics return the first operand raised to the
4808specified (positive or negative) power.
4809</p>
4810
4811<h5>Arguments:</h5>
4812
4813<p>
4814The second argument is a floating point power, and the first is a value to
4815raise to that power.
4816</p>
4817
4818<h5>Semantics:</h5>
4819
4820<p>
4821This function returns the first value raised to the second power,
4822returning the
4823same values as the libm <tt>pow</tt> functions would, and handles error
Dan Gohmaneaba92e2007-10-17 18:05:13 +0000[diff] [blame]4824conditions in the same way.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]4825</div>
4826
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4827
4828<!-- ======================================================================= -->
4829<div class="doc_subsection">
4830 <a name="int_manip">Bit Manipulation Intrinsics</a>
4831</div>
4832
4833<div class="doc_text">
4834<p>
4835LLVM provides intrinsics for a few important bit manipulation operations.
4836These allow efficient code generation for some algorithms.
4837</p>
4838
4839</div>
4840
4841<!-- _______________________________________________________________________ -->
4842<div class="doc_subsubsection">
4843 <a name="int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a>
4844</div>
4845
4846<div class="doc_text">
4847
4848<h5>Syntax:</h5>
4849<p>This is an overloaded intrinsic function. You can use bswap on any integer
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]4850type that is an even number of bytes (i.e. BitWidth % 16 == 0).
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4851<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]4852 declare i16 @llvm.bswap.i16(i16 &lt;id&gt;)
4853 declare i32 @llvm.bswap.i32(i32 &lt;id&gt;)
4854 declare i64 @llvm.bswap.i64(i64 &lt;id&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4855</pre>
4856
4857<h5>Overview:</h5>
4858
4859<p>
4860The '<tt>llvm.bswap</tt>' family of intrinsics is used to byte swap integer
4861values with an even number of bytes (positive multiple of 16 bits). These are
4862useful for performing operations on data that is not in the target's native
4863byte order.
4864</p>
4865
4866<h5>Semantics:</h5>
4867
4868<p>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]4869The <tt>llvm.bswap.i16</tt> intrinsic returns an i16 value that has the high
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4870and low byte of the input i16 swapped. Similarly, the <tt>llvm.bswap.i32</tt>
4871intrinsic returns an i32 value that has the four bytes of the input i32
4872swapped, 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]4873i32 will have its bytes in 3, 2, 1, 0 order. The <tt>llvm.bswap.i48</tt>,
4874<tt>llvm.bswap.i64</tt> and other intrinsics extend this concept to
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4875additional even-byte lengths (6 bytes, 8 bytes and more, respectively).
4876</p>
4877
4878</div>
4879
4880<!-- _______________________________________________________________________ -->
4881<div class="doc_subsubsection">
4882 <a name="int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic</a>
4883</div>
4884
4885<div class="doc_text">
4886
4887<h5>Syntax:</h5>
4888<p>This is an overloaded intrinsic. You can use llvm.ctpop on any integer bit
4889width. Not all targets support all bit widths however.
4890<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]4891 declare i8 @llvm.ctpop.i8 (i8 &lt;src&gt;)
4892 declare i16 @llvm.ctpop.i16(i16 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4893 declare i32 @llvm.ctpop.i32(i32 &lt;src&gt;)
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]4894 declare i64 @llvm.ctpop.i64(i64 &lt;src&gt;)
4895 declare i256 @llvm.ctpop.i256(i256 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4896</pre>
4897
4898<h5>Overview:</h5>
4899
4900<p>
4901The '<tt>llvm.ctpop</tt>' family of intrinsics counts the number of bits set in a
4902value.
4903</p>
4904
4905<h5>Arguments:</h5>
4906
4907<p>
4908The only argument is the value to be counted. The argument may be of any
4909integer type. The return type must match the argument type.
4910</p>
4911
4912<h5>Semantics:</h5>
4913
4914<p>
4915The '<tt>llvm.ctpop</tt>' intrinsic counts the 1's in a variable.
4916</p>
4917</div>
4918
4919<!-- _______________________________________________________________________ -->
4920<div class="doc_subsubsection">
4921 <a name="int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic</a>
4922</div>
4923
4924<div class="doc_text">
4925
4926<h5>Syntax:</h5>
4927<p>This is an overloaded intrinsic. You can use <tt>llvm.ctlz</tt> on any
4928integer bit width. Not all targets support all bit widths however.
4929<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]4930 declare i8 @llvm.ctlz.i8 (i8 &lt;src&gt;)
4931 declare i16 @llvm.ctlz.i16(i16 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4932 declare i32 @llvm.ctlz.i32(i32 &lt;src&gt;)
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]4933 declare i64 @llvm.ctlz.i64(i64 &lt;src&gt;)
4934 declare i256 @llvm.ctlz.i256(i256 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4935</pre>
4936
4937<h5>Overview:</h5>
4938
4939<p>
4940The '<tt>llvm.ctlz</tt>' family of intrinsic functions counts the number of
4941leading zeros in a variable.
4942</p>
4943
4944<h5>Arguments:</h5>
4945
4946<p>
4947The only argument is the value to be counted. The argument may be of any
4948integer type. The return type must match the argument type.
4949</p>
4950
4951<h5>Semantics:</h5>
4952
4953<p>
4954The '<tt>llvm.ctlz</tt>' intrinsic counts the leading (most significant) zeros
4955in a variable. If the src == 0 then the result is the size in bits of the type
4956of src. For example, <tt>llvm.ctlz(i32 2) = 30</tt>.
4957</p>
4958</div>
4959
4960
4961
4962<!-- _______________________________________________________________________ -->
4963<div class="doc_subsubsection">
4964 <a name="int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic</a>
4965</div>
4966
4967<div class="doc_text">
4968
4969<h5>Syntax:</h5>
4970<p>This is an overloaded intrinsic. You can use <tt>llvm.cttz</tt> on any
4971integer bit width. Not all targets support all bit widths however.
4972<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]4973 declare i8 @llvm.cttz.i8 (i8 &lt;src&gt;)
4974 declare i16 @llvm.cttz.i16(i16 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4975 declare i32 @llvm.cttz.i32(i32 &lt;src&gt;)
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]4976 declare i64 @llvm.cttz.i64(i64 &lt;src&gt;)
4977 declare i256 @llvm.cttz.i256(i256 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4978</pre>
4979
4980<h5>Overview:</h5>
4981
4982<p>
4983The '<tt>llvm.cttz</tt>' family of intrinsic functions counts the number of
4984trailing zeros.
4985</p>
4986
4987<h5>Arguments:</h5>
4988
4989<p>
4990The only argument is the value to be counted. The argument may be of any
4991integer type. The return type must match the argument type.
4992</p>
4993
4994<h5>Semantics:</h5>
4995
4996<p>
4997The '<tt>llvm.cttz</tt>' intrinsic counts the trailing (least significant) zeros
4998in a variable. If the src == 0 then the result is the size in bits of the type
4999of src. For example, <tt>llvm.cttz(2) = 1</tt>.
5000</p>
5001</div>
5002
5003<!-- _______________________________________________________________________ -->
5004<div class="doc_subsubsection">
5005 <a name="int_part_select">'<tt>llvm.part.select.*</tt>' Intrinsic</a>
5006</div>
5007
5008<div class="doc_text">
5009
5010<h5>Syntax:</h5>
5011<p>This is an overloaded intrinsic. You can use <tt>llvm.part.select</tt>
5012on any integer bit width.
5013<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5014 declare i17 @llvm.part.select.i17 (i17 %val, i32 %loBit, i32 %hiBit)
5015 declare i29 @llvm.part.select.i29 (i29 %val, i32 %loBit, i32 %hiBit)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5016</pre>
5017
5018<h5>Overview:</h5>
5019<p>The '<tt>llvm.part.select</tt>' family of intrinsic functions selects a
5020range of bits from an integer value and returns them in the same bit width as
5021the original value.</p>
5022
5023<h5>Arguments:</h5>
5024<p>The first argument, <tt>%val</tt> and the result may be integer types of
5025any bit width but they must have the same bit width. The second and third
5026arguments must be <tt>i32</tt> type since they specify only a bit index.</p>
5027
5028<h5>Semantics:</h5>
5029<p>The operation of the '<tt>llvm.part.select</tt>' intrinsic has two modes
5030of operation: forwards and reverse. If <tt>%loBit</tt> is greater than
5031<tt>%hiBits</tt> then the intrinsic operates in reverse mode. Otherwise it
5032operates in forward mode.</p>
5033<p>In forward mode, this intrinsic is the equivalent of shifting <tt>%val</tt>
5034right by <tt>%loBit</tt> bits and then ANDing it with a mask with
5035only the <tt>%hiBit - %loBit</tt> bits set, as follows:</p>
5036<ol>
5037 <li>The <tt>%val</tt> is shifted right (LSHR) by the number of bits specified
5038 by <tt>%loBits</tt>. This normalizes the value to the low order bits.</li>
5039 <li>The <tt>%loBits</tt> value is subtracted from the <tt>%hiBits</tt> value
5040 to determine the number of bits to retain.</li>
5041 <li>A mask of the retained bits is created by shifting a -1 value.</li>
5042 <li>The mask is ANDed with <tt>%val</tt> to produce the result.
5043</ol>
5044<p>In reverse mode, a similar computation is made except that the bits are
5045returned in the reverse order. So, for example, if <tt>X</tt> has the value
5046<tt>i16 0x0ACF (101011001111)</tt> and we apply
5047<tt>part.select(i16 X, 8, 3)</tt> to it, we get back the value
5048<tt>i16 0x0026 (000000100110)</tt>.</p>
5049</div>
5050
5051<div class="doc_subsubsection">
5052 <a name="int_part_set">'<tt>llvm.part.set.*</tt>' Intrinsic</a>
5053</div>
5054
5055<div class="doc_text">
5056
5057<h5>Syntax:</h5>
5058<p>This is an overloaded intrinsic. You can use <tt>llvm.part.set</tt>
5059on any integer bit width.
5060<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5061 declare i17 @llvm.part.set.i17.i9 (i17 %val, i9 %repl, i32 %lo, i32 %hi)
5062 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]5063</pre>
5064
5065<h5>Overview:</h5>
5066<p>The '<tt>llvm.part.set</tt>' family of intrinsic functions replaces a range
5067of bits in an integer value with another integer value. It returns the integer
5068with the replaced bits.</p>
5069
5070<h5>Arguments:</h5>
5071<p>The first argument, <tt>%val</tt> and the result may be integer types of
5072any bit width but they must have the same bit width. <tt>%val</tt> is the value
5073whose bits will be replaced. The second argument, <tt>%repl</tt> may be an
5074integer of any bit width. The third and fourth arguments must be <tt>i32</tt>
5075type since they specify only a bit index.</p>
5076
5077<h5>Semantics:</h5>
5078<p>The operation of the '<tt>llvm.part.set</tt>' intrinsic has two modes
5079of operation: forwards and reverse. If <tt>%lo</tt> is greater than
5080<tt>%hi</tt> then the intrinsic operates in reverse mode. Otherwise it
5081operates in forward mode.</p>
5082<p>For both modes, the <tt>%repl</tt> value is prepared for use by either
5083truncating it down to the size of the replacement area or zero extending it
5084up to that size.</p>
5085<p>In forward mode, the bits between <tt>%lo</tt> and <tt>%hi</tt> (inclusive)
5086are replaced with corresponding bits from <tt>%repl</tt>. That is the 0th bit
5087in <tt>%repl</tt> replaces the <tt>%lo</tt>th bit in <tt>%val</tt> and etc. up
5088to the <tt>%hi</tt>th bit.
5089<p>In reverse mode, a similar computation is made except that the bits are
5090reversed. That is, the <tt>0</tt>th bit in <tt>%repl</tt> replaces the
5091<tt>%hi</tt> bit in <tt>%val</tt> and etc. down to the <tt>%lo</tt>th bit.
5092<h5>Examples:</h5>
5093<pre>
5094 llvm.part.set(0xFFFF, 0, 4, 7) -&gt; 0xFF0F
5095 llvm.part.set(0xFFFF, 0, 7, 4) -&gt; 0xFF0F
5096 llvm.part.set(0xFFFF, 1, 7, 4) -&gt; 0xFF8F
5097 llvm.part.set(0xFFFF, F, 8, 3) -&gt; 0xFFE7
5098 llvm.part.set(0xFFFF, 0, 3, 8) -&gt; 0xFE07
5099</pre>
5100</div>
5101
5102<!-- ======================================================================= -->
5103<div class="doc_subsection">
5104 <a name="int_debugger">Debugger Intrinsics</a>
5105</div>
5106
5107<div class="doc_text">
5108<p>
5109The LLVM debugger intrinsics (which all start with <tt>llvm.dbg.</tt> prefix),
5110are described in the <a
5111href="SourceLevelDebugging.html#format_common_intrinsics">LLVM Source Level
5112Debugging</a> document.
5113</p>
5114</div>
5115
5116
5117<!-- ======================================================================= -->
5118<div class="doc_subsection">
5119 <a name="int_eh">Exception Handling Intrinsics</a>
5120</div>
5121
5122<div class="doc_text">
5123<p> The LLVM exception handling intrinsics (which all start with
5124<tt>llvm.eh.</tt> prefix), are described in the <a
5125href="ExceptionHandling.html#format_common_intrinsics">LLVM Exception
5126Handling</a> document. </p>
5127</div>
5128
5129<!-- ======================================================================= -->
5130<div class="doc_subsection">
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5131 <a name="int_trampoline">Trampoline Intrinsic</a>
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5132</div>
5133
5134<div class="doc_text">
5135<p>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5136 This intrinsic makes it possible to excise one parameter, marked with
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5137 the <tt>nest</tt> attribute, from a function. The result is a callable
5138 function pointer lacking the nest parameter - the caller does not need
5139 to provide a value for it. Instead, the value to use is stored in
5140 advance in a "trampoline", a block of memory usually allocated
5141 on the stack, which also contains code to splice the nest value into the
5142 argument list. This is used to implement the GCC nested function address
5143 extension.
5144</p>
5145<p>
5146 For example, if the function is
5147 <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]5148 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]5149<pre>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5150 %tramp = alloca [10 x i8], align 4 ; size and alignment only correct for X86
5151 %tramp1 = getelementptr [10 x i8]* %tramp, i32 0, i32 0
5152 %p = call i8* @llvm.init.trampoline( i8* %tramp1, i8* bitcast (i32 (i8* nest , i32, i32)* @f to i8*), i8* %nval )
5153 %fp = bitcast i8* %p to i32 (i32, i32)*
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5154</pre>
Bill Wendlinge262b4c2007-09-22 09:23:55 +0000[diff] [blame]5155 <p>The call <tt>%val = call i32 %fp( i32 %x, i32 %y )</tt> is then equivalent
5156 to <tt>%val = call i32 %f( i8* %nval, i32 %x, i32 %y )</tt>.</p>
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5157</div>
5158
5159<!-- _______________________________________________________________________ -->
5160<div class="doc_subsubsection">
5161 <a name="int_it">'<tt>llvm.init.trampoline</tt>' Intrinsic</a>
5162</div>
5163<div class="doc_text">
5164<h5>Syntax:</h5>
5165<pre>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5166declare 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]5167</pre>
5168<h5>Overview:</h5>
5169<p>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5170 This fills the memory pointed to by <tt>tramp</tt> with code
5171 and returns a function pointer suitable for executing it.
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5172</p>
5173<h5>Arguments:</h5>
5174<p>
5175 The <tt>llvm.init.trampoline</tt> intrinsic takes three arguments, all
5176 pointers. The <tt>tramp</tt> argument must point to a sufficiently large
5177 and sufficiently aligned block of memory; this memory is written to by the
Duncan Sands35012212007-08-22 23:39:54 +0000[diff] [blame]5178 intrinsic. Note that the size and the alignment are target-specific - LLVM
5179 currently provides no portable way of determining them, so a front-end that
5180 generates this intrinsic needs to have some target-specific knowledge.
5181 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]5182</p>
5183<h5>Semantics:</h5>
5184<p>
5185 The block of memory pointed to by <tt>tramp</tt> is filled with target
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5186 dependent code, turning it into a function. A pointer to this function is
5187 returned, but needs to be bitcast to an
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5188 <a href="#int_trampoline">appropriate function pointer type</a>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5189 before being called. The new function's signature is the same as that of
5190 <tt>func</tt> with any arguments marked with the <tt>nest</tt> attribute
5191 removed. At most one such <tt>nest</tt> argument is allowed, and it must be
5192 of pointer type. Calling the new function is equivalent to calling
5193 <tt>func</tt> with the same argument list, but with <tt>nval</tt> used for the
5194 missing <tt>nest</tt> argument. If, after calling
5195 <tt>llvm.init.trampoline</tt>, the memory pointed to by <tt>tramp</tt> is
5196 modified, then the effect of any later call to the returned function pointer is
5197 undefined.
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5198</p>
5199</div>
5200
5201<!-- ======================================================================= -->
5202<div class="doc_subsection">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5203 <a name="int_general">General Intrinsics</a>
5204</div>
5205
5206<div class="doc_text">
5207<p> This class of intrinsics is designed to be generic and has
5208no specific purpose. </p>
5209</div>
5210
5211<!-- _______________________________________________________________________ -->
5212<div class="doc_subsubsection">
5213 <a name="int_var_annotation">'<tt>llvm.var.annotation</tt>' Intrinsic</a>
5214</div>
5215
5216<div class="doc_text">
5217
5218<h5>Syntax:</h5>
5219<pre>
5220 declare void @llvm.var.annotation(i8* &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
5221</pre>
5222
5223<h5>Overview:</h5>
5224
5225<p>
5226The '<tt>llvm.var.annotation</tt>' intrinsic
5227</p>
5228
5229<h5>Arguments:</h5>
5230
5231<p>
5232The first argument is a pointer to a value, the second is a pointer to a
5233global string, the third is a pointer to a global string which is the source
5234file name, and the last argument is the line number.
5235</p>
5236
5237<h5>Semantics:</h5>
5238
5239<p>
5240This intrinsic allows annotation of local variables with arbitrary strings.
5241This can be useful for special purpose optimizations that want to look for these
5242 annotations. These have no other defined use, they are ignored by code
5243 generation and optimization.
5244</div>
5245
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]5246<!-- _______________________________________________________________________ -->
5247<div class="doc_subsubsection">
Tanya Lattnerc9869b12007-09-21 23:57:59 +0000[diff] [blame]5248 <a name="int_annotation">'<tt>llvm.annotation.*</tt>' Intrinsic</a>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]5249</div>
5250
5251<div class="doc_text">
5252
5253<h5>Syntax:</h5>
Tanya Lattnere545be72007-09-21 23:56:27 +0000[diff] [blame]5254<p>This is an overloaded intrinsic. You can use '<tt>llvm.annotation</tt>' on
5255any integer bit width.
5256</p>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]5257<pre>
Tanya Lattner09161fe2007-09-22 00:03:01 +0000[diff] [blame]5258 declare i8 @llvm.annotation.i8(i8 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
5259 declare i16 @llvm.annotation.i16(i16 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
5260 declare i32 @llvm.annotation.i32(i32 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
5261 declare i64 @llvm.annotation.i64(i64 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
5262 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]5263</pre>
5264
5265<h5>Overview:</h5>
Tanya Lattnere545be72007-09-21 23:56:27 +0000[diff] [blame]5266
5267<p>
5268The '<tt>llvm.annotation</tt>' intrinsic.
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]5269</p>
5270
5271<h5>Arguments:</h5>
5272
5273<p>
5274The first argument is an integer value (result of some expression),
5275the second is a pointer to a global string, the third is a pointer to a global
5276string which is the source file name, and the last argument is the line number.
Tanya Lattnere545be72007-09-21 23:56:27 +0000[diff] [blame]5277It returns the value of the first argument.
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]5278</p>
5279
5280<h5>Semantics:</h5>
5281
5282<p>
5283This intrinsic allows annotations to be put on arbitrary expressions
5284with arbitrary strings. This can be useful for special purpose optimizations
5285that want to look for these annotations. These have no other defined use, they
5286are ignored by code generation and optimization.
5287</div>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5288
5289<!-- *********************************************************************** -->
5290<hr>
5291<address>
5292 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
5293 src="http://jigsaw.w3.org/css-validator/images/vcss" alt="Valid CSS!"></a>
5294 <a href="http://validator.w3.org/check/referer"><img
Chris Lattner08497ce2008-01-04 04:33:49 +0000[diff] [blame]5295 src="http://www.w3.org/Icons/valid-html401" alt="Valid HTML 4.01!"></a>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5296
5297 <a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
5298 <a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br>
5299 Last modified: $Date$
5300</address>
Chris Lattner08497ce2008-01-04 04:33:49 +0000[diff] [blame]5301
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5302</body>
5303</html>