Gitiles
Code Review Sign In
gerrit-public.fairphone.software / fp2-dev / platform / external / llvm / 488772f1dd82db3d6da028d6b11a05ddce753e24 / . / docs / LangRef.html
blob: 01fb57a0f15a6b8f9b1410db82486f04b03f8807 [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>
Reid Spencerf234bed2007-07-19 23:13:04 +0000[diff] [blame]802 <dt><tt>signext</tt></dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]803 <dd>This indicates that the parameter should be sign extended just before
804 a call to this function.</dd>
805 <dt><tt>inreg</tt></dt>
806 <dd>This indicates that the parameter should be placed in register (if
807 possible) during assembling function call. Support for this attribute is
808 target-specific</dd>
809 <dt><tt>sret</tt></dt>
810 <dd>This indicates that the parameter specifies the address of a structure
811 that is the return value of the function in the source program.</dd>
812 <dt><tt>noalias</tt></dt>
813 <dd>This indicates that the parameter not alias any other object or any
814 other "noalias" objects during the function call.
815 <dt><tt>noreturn</tt></dt>
816 <dd>This function attribute indicates that the function never returns. This
817 indicates to LLVM that every call to this function should be treated as if
818 an <tt>unreachable</tt> instruction immediately followed the call.</dd>
819 <dt><tt>nounwind</tt></dt>
820 <dd>This function attribute indicates that the function type does not use
821 the unwind instruction and does not allow stack unwinding to propagate
822 through it.</dd>
Duncan Sands4ee46812007-07-27 19:57:41 +0000[diff] [blame]823 <dt><tt>nest</tt></dt>
824 <dd>This indicates that the parameter can be excised using the
825 <a href="#int_trampoline">trampoline intrinsics</a>.</dd>
Duncan Sands13e13f82007-11-22 20:23:04 +0000[diff] [blame]826 <dt><tt>readonly</tt></dt>
Duncan Sandsd69c0e62007-11-14 21:14:02 +0000[diff] [blame]827 <dd>This function attribute indicates that the function has no side-effects
Duncan Sands13e13f82007-11-22 20:23:04 +0000[diff] [blame]828 except for producing a return value or throwing an exception. The value
829 returned must only depend on the function arguments and/or global variables.
830 It may use values obtained by dereferencing pointers.</dd>
831 <dt><tt>readnone</tt></dt>
832 <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]833 function, but in addition it is not allowed to dereference any pointer arguments
834 or global variables.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]835 </dl>
836
837</div>
838
839<!-- ======================================================================= -->
840<div class="doc_subsection">
Gordon Henriksen13fe5e32007-12-10 03:18:06 +0000[diff] [blame]841 <a name="gc">Garbage Collector Names</a>
842</div>
843
844<div class="doc_text">
845<p>Each function may specify a garbage collector name, which is simply a
846string.</p>
847
848<div class="doc_code"><pre
849>define void @f() gc "name" { ...</pre></div>
850
851<p>The compiler declares the supported values of <i>name</i>. Specifying a
852collector which will cause the compiler to alter its output in order to support
853the named garbage collection algorithm.</p>
854</div>
855
856<!-- ======================================================================= -->
857<div class="doc_subsection">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]858 <a name="moduleasm">Module-Level Inline Assembly</a>
859</div>
860
861<div class="doc_text">
862<p>
863Modules may contain "module-level inline asm" blocks, which corresponds to the
864GCC "file scope inline asm" blocks. These blocks are internally concatenated by
865LLVM and treated as a single unit, but may be separated in the .ll file if
866desired. The syntax is very simple:
867</p>
868
869<div class="doc_code">
870<pre>
871module asm "inline asm code goes here"
872module asm "more can go here"
873</pre>
874</div>
875
876<p>The strings can contain any character by escaping non-printable characters.
877 The escape sequence used is simply "\xx" where "xx" is the two digit hex code
878 for the number.
879</p>
880
881<p>
882 The inline asm code is simply printed to the machine code .s file when
883 assembly code is generated.
884</p>
885</div>
886
887<!-- ======================================================================= -->
888<div class="doc_subsection">
889 <a name="datalayout">Data Layout</a>
890</div>
891
892<div class="doc_text">
893<p>A module may specify a target specific data layout string that specifies how
894data is to be laid out in memory. The syntax for the data layout is simply:</p>
895<pre> target datalayout = "<i>layout specification</i>"</pre>
896<p>The <i>layout specification</i> consists of a list of specifications
897separated by the minus sign character ('-'). Each specification starts with a
898letter and may include other information after the letter to define some
899aspect of the data layout. The specifications accepted are as follows: </p>
900<dl>
901 <dt><tt>E</tt></dt>
902 <dd>Specifies that the target lays out data in big-endian form. That is, the
903 bits with the most significance have the lowest address location.</dd>
904 <dt><tt>e</tt></dt>
905 <dd>Specifies that hte target lays out data in little-endian form. That is,
906 the bits with the least significance have the lowest address location.</dd>
907 <dt><tt>p:<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
908 <dd>This specifies the <i>size</i> of a pointer and its <i>abi</i> and
909 <i>preferred</i> alignments. All sizes are in bits. Specifying the <i>pref</i>
910 alignment is optional. If omitted, the preceding <tt>:</tt> should be omitted
911 too.</dd>
912 <dt><tt>i<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
913 <dd>This specifies the alignment for an integer type of a given bit
914 <i>size</i>. The value of <i>size</i> must be in the range [1,2^23).</dd>
915 <dt><tt>v<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
916 <dd>This specifies the alignment for a vector type of a given bit
917 <i>size</i>.</dd>
918 <dt><tt>f<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
919 <dd>This specifies the alignment for a floating point type of a given bit
920 <i>size</i>. The value of <i>size</i> must be either 32 (float) or 64
921 (double).</dd>
922 <dt><tt>a<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
923 <dd>This specifies the alignment for an aggregate type of a given bit
924 <i>size</i>.</dd>
925</dl>
926<p>When constructing the data layout for a given target, LLVM starts with a
927default set of specifications which are then (possibly) overriden by the
928specifications in the <tt>datalayout</tt> keyword. The default specifications
929are given in this list:</p>
930<ul>
931 <li><tt>E</tt> - big endian</li>
932 <li><tt>p:32:64:64</tt> - 32-bit pointers with 64-bit alignment</li>
933 <li><tt>i1:8:8</tt> - i1 is 8-bit (byte) aligned</li>
934 <li><tt>i8:8:8</tt> - i8 is 8-bit (byte) aligned</li>
935 <li><tt>i16:16:16</tt> - i16 is 16-bit aligned</li>
936 <li><tt>i32:32:32</tt> - i32 is 32-bit aligned</li>
937 <li><tt>i64:32:64</tt> - i64 has abi alignment of 32-bits but preferred
938 alignment of 64-bits</li>
939 <li><tt>f32:32:32</tt> - float is 32-bit aligned</li>
940 <li><tt>f64:64:64</tt> - double is 64-bit aligned</li>
941 <li><tt>v64:64:64</tt> - 64-bit vector is 64-bit aligned</li>
942 <li><tt>v128:128:128</tt> - 128-bit vector is 128-bit aligned</li>
943 <li><tt>a0:0:1</tt> - aggregates are 8-bit aligned</li>
944</ul>
945<p>When llvm is determining the alignment for a given type, it uses the
946following rules:
947<ol>
948 <li>If the type sought is an exact match for one of the specifications, that
949 specification is used.</li>
950 <li>If no match is found, and the type sought is an integer type, then the
951 smallest integer type that is larger than the bitwidth of the sought type is
952 used. If none of the specifications are larger than the bitwidth then the the
953 largest integer type is used. For example, given the default specifications
954 above, the i7 type will use the alignment of i8 (next largest) while both
955 i65 and i256 will use the alignment of i64 (largest specified).</li>
956 <li>If no match is found, and the type sought is a vector type, then the
957 largest vector type that is smaller than the sought vector type will be used
958 as a fall back. This happens because <128 x double> can be implemented in
959 terms of 64 <2 x double>, for example.</li>
960</ol>
961</div>
962
963<!-- *********************************************************************** -->
964<div class="doc_section"> <a name="typesystem">Type System</a> </div>
965<!-- *********************************************************************** -->
966
967<div class="doc_text">
968
969<p>The LLVM type system is one of the most important features of the
970intermediate representation. Being typed enables a number of
971optimizations to be performed on the IR directly, without having to do
972extra analyses on the side before the transformation. A strong type
973system makes it easier to read the generated code and enables novel
974analyses and transformations that are not feasible to perform on normal
975three address code representations.</p>
976
977</div>
978
979<!-- ======================================================================= -->
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame^]980<div class="doc_subsection"> <a name="t_classifications">Type
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]981Classifications</a> </div>
982<div class="doc_text">
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame^]983<p>The types fall into a few useful
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]984classifications:</p>
985
986<table border="1" cellspacing="0" cellpadding="4">
987 <tbody>
988 <tr><th>Classification</th><th>Types</th></tr>
989 <tr>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame^]990 <td><a href="#t_integer">integer</a></td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]991 <td><tt>i1, i2, i3, ... i8, ... i16, ... i32, ... i64, ... </tt></td>
992 </tr>
993 <tr>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame^]994 <td><a href="#t_floating">floating point</a></td>
995 <td><tt>float, double, x86_fp80, fp128, ppc_fp128</tt></td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]996 </tr>
997 <tr>
998 <td><a name="t_firstclass">first class</a></td>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame^]999 <td><a href="#t_integer">integer</a>,
1000 <a href="#t_floating">floating point</a>,
1001 <a href="#t_pointer">pointer</a>,
1002 <a href="#t_vector">vector</a>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1003 </td>
1004 </tr>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame^]1005 <tr>
1006 <td><a href="#t_primitive">primitive</a></td>
1007 <td><a href="#t_label">label</a>,
1008 <a href="#t_void">void</a>,
1009 <a href="#t_integer">integer</a>,
1010 <a href="#t_floating">floating point</a>.</td>
1011 </tr>
1012 <tr>
1013 <td><a href="#t_derived">derived</a></td>
1014 <td><a href="#t_integer">integer</a>,
1015 <a href="#t_array">array</a>,
1016 <a href="#t_function">function</a>,
1017 <a href="#t_pointer">pointer</a>,
1018 <a href="#t_struct">structure</a>,
1019 <a href="#t_pstruct">packed structure</a>,
1020 <a href="#t_vector">vector</a>,
1021 <a href="#t_opaque">opaque</a>.
1022 </tr>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1023 </tbody>
1024</table>
1025
1026<p>The <a href="#t_firstclass">first class</a> types are perhaps the
1027most important. Values of these types are the only ones which can be
1028produced by instructions, passed as arguments, or used as operands to
1029instructions. This means that all structures and arrays must be
1030manipulated either by pointer or by component.</p>
1031</div>
1032
1033<!-- ======================================================================= -->
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame^]1034<div class="doc_subsection"> <a name="t_primitive">Primitive Types</a> </div>
1035<div class="doc_text">
1036<p>The primitive types are the fundamental building blocks of the LLVM
1037system.</p>
1038
1039<!-- _______________________________________________________________________ -->
1040<div class="doc_subsubsection"> <a name="t_floating">Floating Point Types</a> </div>
1041
1042<div class="doc_text">
1043 <table>
1044 <tbody>
1045 <tr><th>Type</th><th>Description</th></tr>
1046 <tr><td><tt>float</tt></td><td>32-bit floating point value</td></tr>
1047 <tr><td><tt>double</tt></td><td>64-bit floating point value</td></tr>
1048 <tr><td><tt>fp128</tt></td><td>128-bit floating point value (112-bit mantissa)</td></tr>
1049 <tr><td><tt>x86_fp80</tt></td><td>80-bit floating point value (X87)</td></tr>
1050 <tr><td><tt>ppc_fp128</tt></td><td>128-bit floating point value (two 64-bits)</td></tr>
1051 </tbody>
1052 </table>
1053</div>
1054
1055<!-- _______________________________________________________________________ -->
1056<div class="doc_subsubsection"> <a name="t_void">Void Type</a> </div>
1057
1058<div class="doc_text">
1059<h5>Overview:</h5>
1060<p>The void type does not represent any value and has no size.</p>
1061
1062<h5>Syntax:</h5>
1063
1064<pre>
1065 void
1066</pre>
1067</div>
1068
1069<!-- _______________________________________________________________________ -->
1070<div class="doc_subsubsection"> <a name="t_label">Label Type</a> </div>
1071
1072<div class="doc_text">
1073<h5>Overview:</h5>
1074<p>The label type represents code labels.</p>
1075
1076<h5>Syntax:</h5>
1077
1078<pre>
1079 label
1080</pre>
1081</div>
1082
1083
1084<!-- ======================================================================= -->
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1085<div class="doc_subsection"> <a name="t_derived">Derived Types</a> </div>
1086
1087<div class="doc_text">
1088
1089<p>The real power in LLVM comes from the derived types in the system.
1090This is what allows a programmer to represent arrays, functions,
1091pointers, and other useful types. Note that these derived types may be
1092recursive: For example, it is possible to have a two dimensional array.</p>
1093
1094</div>
1095
1096<!-- _______________________________________________________________________ -->
1097<div class="doc_subsubsection"> <a name="t_integer">Integer Type</a> </div>
1098
1099<div class="doc_text">
1100
1101<h5>Overview:</h5>
1102<p>The integer type is a very simple derived type that simply specifies an
1103arbitrary bit width for the integer type desired. Any bit width from 1 bit to
11042^23-1 (about 8 million) can be specified.</p>
1105
1106<h5>Syntax:</h5>
1107
1108<pre>
1109 iN
1110</pre>
1111
1112<p>The number of bits the integer will occupy is specified by the <tt>N</tt>
1113value.</p>
1114
1115<h5>Examples:</h5>
1116<table class="layout">
Chris Lattner251ab812007-12-18 06:18:21 +0000[diff] [blame]1117 <tbody>
1118 <tr>
1119 <td><tt>i1</tt></td>
1120 <td>a single-bit integer.</td>
1121 </tr><tr>
1122 <td><tt>i32</tt></td>
1123 <td>a 32-bit integer.</td>
1124 </tr><tr>
1125 <td><tt>i1942652</tt></td>
1126 <td>a really big integer of over 1 million bits.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1127 </tr>
Chris Lattner251ab812007-12-18 06:18:21 +0000[diff] [blame]1128 </tbody>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1129</table>
1130</div>
1131
1132<!-- _______________________________________________________________________ -->
1133<div class="doc_subsubsection"> <a name="t_array">Array Type</a> </div>
1134
1135<div class="doc_text">
1136
1137<h5>Overview:</h5>
1138
1139<p>The array type is a very simple derived type that arranges elements
1140sequentially in memory. The array type requires a size (number of
1141elements) and an underlying data type.</p>
1142
1143<h5>Syntax:</h5>
1144
1145<pre>
1146 [&lt;# elements&gt; x &lt;elementtype&gt;]
1147</pre>
1148
1149<p>The number of elements is a constant integer value; elementtype may
1150be any type with a size.</p>
1151
1152<h5>Examples:</h5>
1153<table class="layout">
1154 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1155 <td class="left"><tt>[40 x i32]</tt></td>
1156 <td class="left">Array of 40 32-bit integer values.</td>
1157 </tr>
1158 <tr class="layout">
1159 <td class="left"><tt>[41 x i32]</tt></td>
1160 <td class="left">Array of 41 32-bit integer values.</td>
1161 </tr>
1162 <tr class="layout">
1163 <td class="left"><tt>[4 x i8]</tt></td>
1164 <td class="left">Array of 4 8-bit integer values.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1165 </tr>
1166</table>
1167<p>Here are some examples of multidimensional arrays:</p>
1168<table class="layout">
1169 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1170 <td class="left"><tt>[3 x [4 x i32]]</tt></td>
1171 <td class="left">3x4 array of 32-bit integer values.</td>
1172 </tr>
1173 <tr class="layout">
1174 <td class="left"><tt>[12 x [10 x float]]</tt></td>
1175 <td class="left">12x10 array of single precision floating point values.</td>
1176 </tr>
1177 <tr class="layout">
1178 <td class="left"><tt>[2 x [3 x [4 x i16]]]</tt></td>
1179 <td class="left">2x3x4 array of 16-bit integer values.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1180 </tr>
1181</table>
1182
1183<p>Note that 'variable sized arrays' can be implemented in LLVM with a zero
1184length array. Normally, accesses past the end of an array are undefined in
1185LLVM (e.g. it is illegal to access the 5th element of a 3 element array).
1186As a special case, however, zero length arrays are recognized to be variable
1187length. This allows implementation of 'pascal style arrays' with the LLVM
1188type "{ i32, [0 x float]}", for example.</p>
1189
1190</div>
1191
1192<!-- _______________________________________________________________________ -->
1193<div class="doc_subsubsection"> <a name="t_function">Function Type</a> </div>
1194<div class="doc_text">
1195<h5>Overview:</h5>
1196<p>The function type can be thought of as a function signature. It
1197consists of a return type and a list of formal parameter types.
1198Function types are usually used to build virtual function tables
1199(which are structures of pointers to functions), for indirect function
1200calls, and when defining a function.</p>
1201<p>
1202The return type of a function type cannot be an aggregate type.
1203</p>
1204<h5>Syntax:</h5>
1205<pre> &lt;returntype&gt; (&lt;parameter list&gt;)<br></pre>
1206<p>...where '<tt>&lt;parameter list&gt;</tt>' is a comma-separated list of type
1207specifiers. Optionally, the parameter list may include a type <tt>...</tt>,
1208which indicates that the function takes a variable number of arguments.
1209Variable argument functions can access their arguments with the <a
1210 href="#int_varargs">variable argument handling intrinsic</a> functions.</p>
1211<h5>Examples:</h5>
1212<table class="layout">
1213 <tr class="layout">
1214 <td class="left"><tt>i32 (i32)</tt></td>
1215 <td class="left">function taking an <tt>i32</tt>, returning an <tt>i32</tt>
1216 </td>
1217 </tr><tr class="layout">
Reid Spencerf234bed2007-07-19 23:13:04 +0000[diff] [blame]1218 <td class="left"><tt>float&nbsp;(i16&nbsp;signext,&nbsp;i32&nbsp;*)&nbsp;*
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1219 </tt></td>
1220 <td class="left"><a href="#t_pointer">Pointer</a> to a function that takes
1221 an <tt>i16</tt> that should be sign extended and a
1222 <a href="#t_pointer">pointer</a> to <tt>i32</tt>, returning
1223 <tt>float</tt>.
1224 </td>
1225 </tr><tr class="layout">
1226 <td class="left"><tt>i32 (i8*, ...)</tt></td>
1227 <td class="left">A vararg function that takes at least one
1228 <a href="#t_pointer">pointer</a> to <tt>i8 </tt> (char in C),
1229 which returns an integer. This is the signature for <tt>printf</tt> in
1230 LLVM.
1231 </td>
1232 </tr>
1233</table>
1234
1235</div>
1236<!-- _______________________________________________________________________ -->
1237<div class="doc_subsubsection"> <a name="t_struct">Structure Type</a> </div>
1238<div class="doc_text">
1239<h5>Overview:</h5>
1240<p>The structure type is used to represent a collection of data members
1241together in memory. The packing of the field types is defined to match
1242the ABI of the underlying processor. The elements of a structure may
1243be any type that has a size.</p>
1244<p>Structures are accessed using '<tt><a href="#i_load">load</a></tt>
1245and '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a
1246field with the '<tt><a href="#i_getelementptr">getelementptr</a></tt>'
1247instruction.</p>
1248<h5>Syntax:</h5>
1249<pre> { &lt;type list&gt; }<br></pre>
1250<h5>Examples:</h5>
1251<table class="layout">
1252 <tr class="layout">
1253 <td class="left"><tt>{ i32, i32, i32 }</tt></td>
1254 <td class="left">A triple of three <tt>i32</tt> values</td>
1255 </tr><tr class="layout">
1256 <td class="left"><tt>{&nbsp;float,&nbsp;i32&nbsp;(i32)&nbsp;*&nbsp;}</tt></td>
1257 <td class="left">A pair, where the first element is a <tt>float</tt> and the
1258 second element is a <a href="#t_pointer">pointer</a> to a
1259 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
1260 an <tt>i32</tt>.</td>
1261 </tr>
1262</table>
1263</div>
1264
1265<!-- _______________________________________________________________________ -->
1266<div class="doc_subsubsection"> <a name="t_pstruct">Packed Structure Type</a>
1267</div>
1268<div class="doc_text">
1269<h5>Overview:</h5>
1270<p>The packed structure type is used to represent a collection of data members
1271together in memory. There is no padding between fields. Further, the alignment
1272of a packed structure is 1 byte. The elements of a packed structure may
1273be any type that has a size.</p>
1274<p>Structures are accessed using '<tt><a href="#i_load">load</a></tt>
1275and '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a
1276field with the '<tt><a href="#i_getelementptr">getelementptr</a></tt>'
1277instruction.</p>
1278<h5>Syntax:</h5>
1279<pre> &lt; { &lt;type list&gt; } &gt; <br></pre>
1280<h5>Examples:</h5>
1281<table class="layout">
1282 <tr class="layout">
1283 <td class="left"><tt>&lt; { i32, i32, i32 } &gt;</tt></td>
1284 <td class="left">A triple of three <tt>i32</tt> values</td>
1285 </tr><tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1286 <td class="left"><tt>&lt; { float, i32 (i32)* } &gt;</tt></td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1287 <td class="left">A pair, where the first element is a <tt>float</tt> and the
1288 second element is a <a href="#t_pointer">pointer</a> to a
1289 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
1290 an <tt>i32</tt>.</td>
1291 </tr>
1292</table>
1293</div>
1294
1295<!-- _______________________________________________________________________ -->
1296<div class="doc_subsubsection"> <a name="t_pointer">Pointer Type</a> </div>
1297<div class="doc_text">
1298<h5>Overview:</h5>
1299<p>As in many languages, the pointer type represents a pointer or
Christopher Lambdd0049d2007-12-11 09:31:00 +0000[diff] [blame]1300reference to another object, which must live in memory. Pointer types may have
1301an optional address space attribute defining the target-specific numbered
1302address space where the pointed-to object resides. The default address space is
1303zero.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1304<h5>Syntax:</h5>
1305<pre> &lt;type&gt; *<br></pre>
1306<h5>Examples:</h5>
1307<table class="layout">
1308 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1309 <td class="left"><tt>[4x i32]*</tt></td>
1310 <td class="left">A <a href="#t_pointer">pointer</a> to <a
1311 href="#t_array">array</a> of four <tt>i32</tt> values.</td>
1312 </tr>
1313 <tr class="layout">
1314 <td class="left"><tt>i32 (i32 *) *</tt></td>
1315 <td class="left"> A <a href="#t_pointer">pointer</a> to a <a
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1316 href="#t_function">function</a> that takes an <tt>i32*</tt>, returning an
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1317 <tt>i32</tt>.</td>
1318 </tr>
1319 <tr class="layout">
1320 <td class="left"><tt>i32 addrspace(5)*</tt></td>
1321 <td class="left">A <a href="#t_pointer">pointer</a> to an <tt>i32</tt> value
1322 that resides in address space #5.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1323 </tr>
1324</table>
1325</div>
1326
1327<!-- _______________________________________________________________________ -->
1328<div class="doc_subsubsection"> <a name="t_vector">Vector Type</a> </div>
1329<div class="doc_text">
1330
1331<h5>Overview:</h5>
1332
1333<p>A vector type is a simple derived type that represents a vector
1334of elements. Vector types are used when multiple primitive data
1335are operated in parallel using a single instruction (SIMD).
1336A vector type requires a size (number of
1337elements) and an underlying primitive data type. Vectors must have a power
1338of two length (1, 2, 4, 8, 16 ...). Vector types are
1339considered <a href="#t_firstclass">first class</a>.</p>
1340
1341<h5>Syntax:</h5>
1342
1343<pre>
1344 &lt; &lt;# elements&gt; x &lt;elementtype&gt; &gt;
1345</pre>
1346
1347<p>The number of elements is a constant integer value; elementtype may
1348be any integer or floating point type.</p>
1349
1350<h5>Examples:</h5>
1351
1352<table class="layout">
1353 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1354 <td class="left"><tt>&lt;4 x i32&gt;</tt></td>
1355 <td class="left">Vector of 4 32-bit integer values.</td>
1356 </tr>
1357 <tr class="layout">
1358 <td class="left"><tt>&lt;8 x float&gt;</tt></td>
1359 <td class="left">Vector of 8 32-bit floating-point values.</td>
1360 </tr>
1361 <tr class="layout">
1362 <td class="left"><tt>&lt;2 x i64&gt;</tt></td>
1363 <td class="left">Vector of 2 64-bit integer values.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1364 </tr>
1365</table>
1366</div>
1367
1368<!-- _______________________________________________________________________ -->
1369<div class="doc_subsubsection"> <a name="t_opaque">Opaque Type</a> </div>
1370<div class="doc_text">
1371
1372<h5>Overview:</h5>
1373
1374<p>Opaque types are used to represent unknown types in the system. This
Gordon Henriksenda0706e2007-10-14 00:34:53 +0000[diff] [blame]1375corresponds (for example) to the C notion of a forward declared structure type.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1376In LLVM, opaque types can eventually be resolved to any type (not just a
1377structure type).</p>
1378
1379<h5>Syntax:</h5>
1380
1381<pre>
1382 opaque
1383</pre>
1384
1385<h5>Examples:</h5>
1386
1387<table class="layout">
1388 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1389 <td class="left"><tt>opaque</tt></td>
1390 <td class="left">An opaque type.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1391 </tr>
1392</table>
1393</div>
1394
1395
1396<!-- *********************************************************************** -->
1397<div class="doc_section"> <a name="constants">Constants</a> </div>
1398<!-- *********************************************************************** -->
1399
1400<div class="doc_text">
1401
1402<p>LLVM has several different basic types of constants. This section describes
1403them all and their syntax.</p>
1404
1405</div>
1406
1407<!-- ======================================================================= -->
1408<div class="doc_subsection"><a name="simpleconstants">Simple Constants</a></div>
1409
1410<div class="doc_text">
1411
1412<dl>
1413 <dt><b>Boolean constants</b></dt>
1414
1415 <dd>The two strings '<tt>true</tt>' and '<tt>false</tt>' are both valid
1416 constants of the <tt><a href="#t_primitive">i1</a></tt> type.
1417 </dd>
1418
1419 <dt><b>Integer constants</b></dt>
1420
1421 <dd>Standard integers (such as '4') are constants of the <a
1422 href="#t_integer">integer</a> type. Negative numbers may be used with
1423 integer types.
1424 </dd>
1425
1426 <dt><b>Floating point constants</b></dt>
1427
1428 <dd>Floating point constants use standard decimal notation (e.g. 123.421),
1429 exponential notation (e.g. 1.23421e+2), or a more precise hexadecimal
1430 notation (see below). Floating point constants must have a <a
1431 href="#t_floating">floating point</a> type. </dd>
1432
1433 <dt><b>Null pointer constants</b></dt>
1434
1435 <dd>The identifier '<tt>null</tt>' is recognized as a null pointer constant
1436 and must be of <a href="#t_pointer">pointer type</a>.</dd>
1437
1438</dl>
1439
1440<p>The one non-intuitive notation for constants is the optional hexadecimal form
1441of floating point constants. For example, the form '<tt>double
14420x432ff973cafa8000</tt>' is equivalent to (but harder to read than) '<tt>double
14434.5e+15</tt>'. The only time hexadecimal floating point constants are required
1444(and the only time that they are generated by the disassembler) is when a
1445floating point constant must be emitted but it cannot be represented as a
1446decimal floating point number. For example, NaN's, infinities, and other
1447special values are represented in their IEEE hexadecimal format so that
1448assembly and disassembly do not cause any bits to change in the constants.</p>
1449
1450</div>
1451
1452<!-- ======================================================================= -->
1453<div class="doc_subsection"><a name="aggregateconstants">Aggregate Constants</a>
1454</div>
1455
1456<div class="doc_text">
1457<p>Aggregate constants arise from aggregation of simple constants
1458and smaller aggregate constants.</p>
1459
1460<dl>
1461 <dt><b>Structure constants</b></dt>
1462
1463 <dd>Structure constants are represented with notation similar to structure
1464 type definitions (a comma separated list of elements, surrounded by braces
Chris Lattner6c8de962007-12-25 20:34:52 +0000[diff] [blame]1465 (<tt>{}</tt>)). For example: "<tt>{ i32 4, float 17.0, i32* @G }</tt>",
1466 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]1467 must have <a href="#t_struct">structure type</a>, and the number and
1468 types of elements must match those specified by the type.
1469 </dd>
1470
1471 <dt><b>Array constants</b></dt>
1472
1473 <dd>Array constants are represented with notation similar to array type
1474 definitions (a comma separated list of elements, surrounded by square brackets
1475 (<tt>[]</tt>)). For example: "<tt>[ i32 42, i32 11, i32 74 ]</tt>". Array
1476 constants must have <a href="#t_array">array type</a>, and the number and
1477 types of elements must match those specified by the type.
1478 </dd>
1479
1480 <dt><b>Vector constants</b></dt>
1481
1482 <dd>Vector constants are represented with notation similar to vector type
1483 definitions (a comma separated list of elements, surrounded by
1484 less-than/greater-than's (<tt>&lt;&gt;</tt>)). For example: "<tt>&lt; i32 42,
1485 i32 11, i32 74, i32 100 &gt;</tt>". Vector constants must have <a
1486 href="#t_vector">vector type</a>, and the number and types of elements must
1487 match those specified by the type.
1488 </dd>
1489
1490 <dt><b>Zero initialization</b></dt>
1491
1492 <dd>The string '<tt>zeroinitializer</tt>' can be used to zero initialize a
1493 value to zero of <em>any</em> type, including scalar and aggregate types.
1494 This is often used to avoid having to print large zero initializers (e.g. for
1495 large arrays) and is always exactly equivalent to using explicit zero
1496 initializers.
1497 </dd>
1498</dl>
1499
1500</div>
1501
1502<!-- ======================================================================= -->
1503<div class="doc_subsection">
1504 <a name="globalconstants">Global Variable and Function Addresses</a>
1505</div>
1506
1507<div class="doc_text">
1508
1509<p>The addresses of <a href="#globalvars">global variables</a> and <a
1510href="#functionstructure">functions</a> are always implicitly valid (link-time)
1511constants. These constants are explicitly referenced when the <a
1512href="#identifiers">identifier for the global</a> is used and always have <a
1513href="#t_pointer">pointer</a> type. For example, the following is a legal LLVM
1514file:</p>
1515
1516<div class="doc_code">
1517<pre>
1518@X = global i32 17
1519@Y = global i32 42
1520@Z = global [2 x i32*] [ i32* @X, i32* @Y ]
1521</pre>
1522</div>
1523
1524</div>
1525
1526<!-- ======================================================================= -->
1527<div class="doc_subsection"><a name="undefvalues">Undefined Values</a></div>
1528<div class="doc_text">
1529 <p>The string '<tt>undef</tt>' is recognized as a type-less constant that has
1530 no specific value. Undefined values may be of any type and be used anywhere
1531 a constant is permitted.</p>
1532
1533 <p>Undefined values indicate to the compiler that the program is well defined
1534 no matter what value is used, giving the compiler more freedom to optimize.
1535 </p>
1536</div>
1537
1538<!-- ======================================================================= -->
1539<div class="doc_subsection"><a name="constantexprs">Constant Expressions</a>
1540</div>
1541
1542<div class="doc_text">
1543
1544<p>Constant expressions are used to allow expressions involving other constants
1545to be used as constants. Constant expressions may be of any <a
1546href="#t_firstclass">first class</a> type and may involve any LLVM operation
1547that does not have side effects (e.g. load and call are not supported). The
1548following is the syntax for constant expressions:</p>
1549
1550<dl>
1551 <dt><b><tt>trunc ( CST to TYPE )</tt></b></dt>
1552 <dd>Truncate a constant to another type. The bit size of CST must be larger
1553 than the bit size of TYPE. Both types must be integers.</dd>
1554
1555 <dt><b><tt>zext ( CST to TYPE )</tt></b></dt>
1556 <dd>Zero extend a constant to another type. The bit size of CST must be
1557 smaller or equal to the bit size of TYPE. Both types must be integers.</dd>
1558
1559 <dt><b><tt>sext ( CST to TYPE )</tt></b></dt>
1560 <dd>Sign extend a constant to another type. The bit size of CST must be
1561 smaller or equal to the bit size of TYPE. Both types must be integers.</dd>
1562
1563 <dt><b><tt>fptrunc ( CST to TYPE )</tt></b></dt>
1564 <dd>Truncate a floating point constant to another floating point type. The
1565 size of CST must be larger than the size of TYPE. Both types must be
1566 floating point.</dd>
1567
1568 <dt><b><tt>fpext ( CST to TYPE )</tt></b></dt>
1569 <dd>Floating point extend a constant to another type. The size of CST must be
1570 smaller or equal to the size of TYPE. Both types must be floating point.</dd>
1571
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]1572 <dt><b><tt>fptoui ( CST to TYPE )</tt></b></dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1573 <dd>Convert a floating point constant to the corresponding unsigned integer
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]1574 constant. TYPE must be a scalar or vector integer type. CST must be of scalar
1575 or vector floating point type. Both CST and TYPE must be scalars, or vectors
1576 of the same number of elements. If the value won't fit in the integer type,
1577 the results are undefined.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1578
1579 <dt><b><tt>fptosi ( CST to TYPE )</tt></b></dt>
1580 <dd>Convert a floating point constant to the corresponding signed integer
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]1581 constant. TYPE must be a scalar or vector integer type. CST must be of scalar
1582 or vector floating point type. Both CST and TYPE must be scalars, or vectors
1583 of the same number of elements. If the value won't fit in the integer type,
1584 the results are undefined.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1585
1586 <dt><b><tt>uitofp ( CST to TYPE )</tt></b></dt>
1587 <dd>Convert an unsigned integer constant to the corresponding floating point
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]1588 constant. TYPE must be a scalar or vector floating point type. CST must be of
1589 scalar or vector integer type. Both CST and TYPE must be scalars, or vectors
1590 of the same number of elements. If the value won't fit in the floating point
1591 type, the results are undefined.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1592
1593 <dt><b><tt>sitofp ( CST to TYPE )</tt></b></dt>
1594 <dd>Convert a signed integer constant to the corresponding floating point
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]1595 constant. TYPE must be a scalar or vector floating point type. CST must be of
1596 scalar or vector integer type. Both CST and TYPE must be scalars, or vectors
1597 of the same number of elements. If the value won't fit in the floating point
1598 type, the results are undefined.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1599
1600 <dt><b><tt>ptrtoint ( CST to TYPE )</tt></b></dt>
1601 <dd>Convert a pointer typed constant to the corresponding integer constant
1602 TYPE must be an integer type. CST must be of pointer type. The CST value is
1603 zero extended, truncated, or unchanged to make it fit in TYPE.</dd>
1604
1605 <dt><b><tt>inttoptr ( CST to TYPE )</tt></b></dt>
1606 <dd>Convert a integer constant to a pointer constant. TYPE must be a
1607 pointer type. CST must be of integer type. The CST value is zero extended,
1608 truncated, or unchanged to make it fit in a pointer size. This one is
1609 <i>really</i> dangerous!</dd>
1610
1611 <dt><b><tt>bitcast ( CST to TYPE )</tt></b></dt>
1612 <dd>Convert a constant, CST, to another TYPE. The size of CST and TYPE must be
1613 identical (same number of bits). The conversion is done as if the CST value
1614 was stored to memory and read back as TYPE. In other words, no bits change
1615 with this operator, just the type. This can be used for conversion of
1616 vector types to any other type, as long as they have the same bit width. For
1617 pointers it is only valid to cast to another pointer type.
1618 </dd>
1619
1620 <dt><b><tt>getelementptr ( CSTPTR, IDX0, IDX1, ... )</tt></b></dt>
1621
1622 <dd>Perform the <a href="#i_getelementptr">getelementptr operation</a> on
1623 constants. As with the <a href="#i_getelementptr">getelementptr</a>
1624 instruction, the index list may have zero or more indexes, which are required
1625 to make sense for the type of "CSTPTR".</dd>
1626
1627 <dt><b><tt>select ( COND, VAL1, VAL2 )</tt></b></dt>
1628
1629 <dd>Perform the <a href="#i_select">select operation</a> on
1630 constants.</dd>
1631
1632 <dt><b><tt>icmp COND ( VAL1, VAL2 )</tt></b></dt>
1633 <dd>Performs the <a href="#i_icmp">icmp operation</a> on constants.</dd>
1634
1635 <dt><b><tt>fcmp COND ( VAL1, VAL2 )</tt></b></dt>
1636 <dd>Performs the <a href="#i_fcmp">fcmp operation</a> on constants.</dd>
1637
1638 <dt><b><tt>extractelement ( VAL, IDX )</tt></b></dt>
1639
1640 <dd>Perform the <a href="#i_extractelement">extractelement
1641 operation</a> on constants.
1642
1643 <dt><b><tt>insertelement ( VAL, ELT, IDX )</tt></b></dt>
1644
1645 <dd>Perform the <a href="#i_insertelement">insertelement
1646 operation</a> on constants.</dd>
1647
1648
1649 <dt><b><tt>shufflevector ( VEC1, VEC2, IDXMASK )</tt></b></dt>
1650
1651 <dd>Perform the <a href="#i_shufflevector">shufflevector
1652 operation</a> on constants.</dd>
1653
1654 <dt><b><tt>OPCODE ( LHS, RHS )</tt></b></dt>
1655
1656 <dd>Perform the specified operation of the LHS and RHS constants. OPCODE may
1657 be any of the <a href="#binaryops">binary</a> or <a href="#bitwiseops">bitwise
1658 binary</a> operations. The constraints on operands are the same as those for
1659 the corresponding instruction (e.g. no bitwise operations on floating point
1660 values are allowed).</dd>
1661</dl>
1662</div>
1663
1664<!-- *********************************************************************** -->
1665<div class="doc_section"> <a name="othervalues">Other Values</a> </div>
1666<!-- *********************************************************************** -->
1667
1668<!-- ======================================================================= -->
1669<div class="doc_subsection">
1670<a name="inlineasm">Inline Assembler Expressions</a>
1671</div>
1672
1673<div class="doc_text">
1674
1675<p>
1676LLVM supports inline assembler expressions (as opposed to <a href="#moduleasm">
1677Module-Level Inline Assembly</a>) through the use of a special value. This
1678value represents the inline assembler as a string (containing the instructions
1679to emit), a list of operand constraints (stored as a string), and a flag that
1680indicates whether or not the inline asm expression has side effects. An example
1681inline assembler expression is:
1682</p>
1683
1684<div class="doc_code">
1685<pre>
1686i32 (i32) asm "bswap $0", "=r,r"
1687</pre>
1688</div>
1689
1690<p>
1691Inline assembler expressions may <b>only</b> be used as the callee operand of
1692a <a href="#i_call"><tt>call</tt> instruction</a>. Thus, typically we have:
1693</p>
1694
1695<div class="doc_code">
1696<pre>
1697%X = call i32 asm "<a href="#int_bswap">bswap</a> $0", "=r,r"(i32 %Y)
1698</pre>
1699</div>
1700
1701<p>
1702Inline asms with side effects not visible in the constraint list must be marked
1703as having side effects. This is done through the use of the
1704'<tt>sideeffect</tt>' keyword, like so:
1705</p>
1706
1707<div class="doc_code">
1708<pre>
1709call void asm sideeffect "eieio", ""()
1710</pre>
1711</div>
1712
1713<p>TODO: The format of the asm and constraints string still need to be
1714documented here. Constraints on what can be done (e.g. duplication, moving, etc
1715need to be documented).
1716</p>
1717
1718</div>
1719
1720<!-- *********************************************************************** -->
1721<div class="doc_section"> <a name="instref">Instruction Reference</a> </div>
1722<!-- *********************************************************************** -->
1723
1724<div class="doc_text">
1725
1726<p>The LLVM instruction set consists of several different
1727classifications of instructions: <a href="#terminators">terminator
1728instructions</a>, <a href="#binaryops">binary instructions</a>,
1729<a href="#bitwiseops">bitwise binary instructions</a>, <a
1730 href="#memoryops">memory instructions</a>, and <a href="#otherops">other
1731instructions</a>.</p>
1732
1733</div>
1734
1735<!-- ======================================================================= -->
1736<div class="doc_subsection"> <a name="terminators">Terminator
1737Instructions</a> </div>
1738
1739<div class="doc_text">
1740
1741<p>As mentioned <a href="#functionstructure">previously</a>, every
1742basic block in a program ends with a "Terminator" instruction, which
1743indicates which block should be executed after the current block is
1744finished. These terminator instructions typically yield a '<tt>void</tt>'
1745value: they produce control flow, not values (the one exception being
1746the '<a href="#i_invoke"><tt>invoke</tt></a>' instruction).</p>
1747<p>There are six different terminator instructions: the '<a
1748 href="#i_ret"><tt>ret</tt></a>' instruction, the '<a href="#i_br"><tt>br</tt></a>'
1749instruction, the '<a href="#i_switch"><tt>switch</tt></a>' instruction,
1750the '<a href="#i_invoke"><tt>invoke</tt></a>' instruction, the '<a
1751 href="#i_unwind"><tt>unwind</tt></a>' instruction, and the '<a
1752 href="#i_unreachable"><tt>unreachable</tt></a>' instruction.</p>
1753
1754</div>
1755
1756<!-- _______________________________________________________________________ -->
1757<div class="doc_subsubsection"> <a name="i_ret">'<tt>ret</tt>'
1758Instruction</a> </div>
1759<div class="doc_text">
1760<h5>Syntax:</h5>
1761<pre> ret &lt;type&gt; &lt;value&gt; <i>; Return a value from a non-void function</i>
1762 ret void <i>; Return from void function</i>
1763</pre>
1764<h5>Overview:</h5>
1765<p>The '<tt>ret</tt>' instruction is used to return control flow (and a
1766value) from a function back to the caller.</p>
1767<p>There are two forms of the '<tt>ret</tt>' instruction: one that
1768returns a value and then causes control flow, and one that just causes
1769control flow to occur.</p>
1770<h5>Arguments:</h5>
1771<p>The '<tt>ret</tt>' instruction may return any '<a
1772 href="#t_firstclass">first class</a>' type. Notice that a function is
1773not <a href="#wellformed">well formed</a> if there exists a '<tt>ret</tt>'
1774instruction inside of the function that returns a value that does not
1775match the return type of the function.</p>
1776<h5>Semantics:</h5>
1777<p>When the '<tt>ret</tt>' instruction is executed, control flow
1778returns back to the calling function's context. If the caller is a "<a
1779 href="#i_call"><tt>call</tt></a>" instruction, execution continues at
1780the instruction after the call. If the caller was an "<a
1781 href="#i_invoke"><tt>invoke</tt></a>" instruction, execution continues
1782at the beginning of the "normal" destination block. If the instruction
1783returns a value, that value shall set the call or invoke instruction's
1784return value.</p>
1785<h5>Example:</h5>
1786<pre> ret i32 5 <i>; Return an integer value of 5</i>
1787 ret void <i>; Return from a void function</i>
1788</pre>
1789</div>
1790<!-- _______________________________________________________________________ -->
1791<div class="doc_subsubsection"> <a name="i_br">'<tt>br</tt>' Instruction</a> </div>
1792<div class="doc_text">
1793<h5>Syntax:</h5>
1794<pre> br i1 &lt;cond&gt;, label &lt;iftrue&gt;, label &lt;iffalse&gt;<br> br label &lt;dest&gt; <i>; Unconditional branch</i>
1795</pre>
1796<h5>Overview:</h5>
1797<p>The '<tt>br</tt>' instruction is used to cause control flow to
1798transfer to a different basic block in the current function. There are
1799two forms of this instruction, corresponding to a conditional branch
1800and an unconditional branch.</p>
1801<h5>Arguments:</h5>
1802<p>The conditional branch form of the '<tt>br</tt>' instruction takes a
1803single '<tt>i1</tt>' value and two '<tt>label</tt>' values. The
1804unconditional form of the '<tt>br</tt>' instruction takes a single
1805'<tt>label</tt>' value as a target.</p>
1806<h5>Semantics:</h5>
1807<p>Upon execution of a conditional '<tt>br</tt>' instruction, the '<tt>i1</tt>'
1808argument is evaluated. If the value is <tt>true</tt>, control flows
1809to the '<tt>iftrue</tt>' <tt>label</tt> argument. If "cond" is <tt>false</tt>,
1810control flows to the '<tt>iffalse</tt>' <tt>label</tt> argument.</p>
1811<h5>Example:</h5>
1812<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
1813 href="#i_ret">ret</a> i32 1<br>IfUnequal:<br> <a href="#i_ret">ret</a> i32 0<br></pre>
1814</div>
1815<!-- _______________________________________________________________________ -->
1816<div class="doc_subsubsection">
1817 <a name="i_switch">'<tt>switch</tt>' Instruction</a>
1818</div>
1819
1820<div class="doc_text">
1821<h5>Syntax:</h5>
1822
1823<pre>
1824 switch &lt;intty&gt; &lt;value&gt;, label &lt;defaultdest&gt; [ &lt;intty&gt; &lt;val&gt;, label &lt;dest&gt; ... ]
1825</pre>
1826
1827<h5>Overview:</h5>
1828
1829<p>The '<tt>switch</tt>' instruction is used to transfer control flow to one of
1830several different places. It is a generalization of the '<tt>br</tt>'
1831instruction, allowing a branch to occur to one of many possible
1832destinations.</p>
1833
1834
1835<h5>Arguments:</h5>
1836
1837<p>The '<tt>switch</tt>' instruction uses three parameters: an integer
1838comparison value '<tt>value</tt>', a default '<tt>label</tt>' destination, and
1839an array of pairs of comparison value constants and '<tt>label</tt>'s. The
1840table is not allowed to contain duplicate constant entries.</p>
1841
1842<h5>Semantics:</h5>
1843
1844<p>The <tt>switch</tt> instruction specifies a table of values and
1845destinations. When the '<tt>switch</tt>' instruction is executed, this
1846table is searched for the given value. If the value is found, control flow is
1847transfered to the corresponding destination; otherwise, control flow is
1848transfered to the default destination.</p>
1849
1850<h5>Implementation:</h5>
1851
1852<p>Depending on properties of the target machine and the particular
1853<tt>switch</tt> instruction, this instruction may be code generated in different
1854ways. For example, it could be generated as a series of chained conditional
1855branches or with a lookup table.</p>
1856
1857<h5>Example:</h5>
1858
1859<pre>
1860 <i>; Emulate a conditional br instruction</i>
1861 %Val = <a href="#i_zext">zext</a> i1 %value to i32
1862 switch i32 %Val, label %truedest [i32 0, label %falsedest ]
1863
1864 <i>; Emulate an unconditional br instruction</i>
1865 switch i32 0, label %dest [ ]
1866
1867 <i>; Implement a jump table:</i>
1868 switch i32 %val, label %otherwise [ i32 0, label %onzero
1869 i32 1, label %onone
1870 i32 2, label %ontwo ]
1871</pre>
1872</div>
1873
1874<!-- _______________________________________________________________________ -->
1875<div class="doc_subsubsection">
1876 <a name="i_invoke">'<tt>invoke</tt>' Instruction</a>
1877</div>
1878
1879<div class="doc_text">
1880
1881<h5>Syntax:</h5>
1882
1883<pre>
1884 &lt;result&gt; = invoke [<a href="#callingconv">cconv</a>] &lt;ptr to function ty&gt; %&lt;function ptr val&gt;(&lt;function args&gt;)
1885 to label &lt;normal label&gt; unwind label &lt;exception label&gt;
1886</pre>
1887
1888<h5>Overview:</h5>
1889
1890<p>The '<tt>invoke</tt>' instruction causes control to transfer to a specified
1891function, with the possibility of control flow transfer to either the
1892'<tt>normal</tt>' label or the
1893'<tt>exception</tt>' label. If the callee function returns with the
1894"<tt><a href="#i_ret">ret</a></tt>" instruction, control flow will return to the
1895"normal" label. If the callee (or any indirect callees) returns with the "<a
1896href="#i_unwind"><tt>unwind</tt></a>" instruction, control is interrupted and
1897continued at the dynamically nearest "exception" label.</p>
1898
1899<h5>Arguments:</h5>
1900
1901<p>This instruction requires several arguments:</p>
1902
1903<ol>
1904 <li>
1905 The optional "cconv" marker indicates which <a href="#callingconv">calling
1906 convention</a> the call should use. If none is specified, the call defaults
1907 to using C calling conventions.
1908 </li>
1909 <li>'<tt>ptr to function ty</tt>': shall be the signature of the pointer to
1910 function value being invoked. In most cases, this is a direct function
1911 invocation, but indirect <tt>invoke</tt>s are just as possible, branching off
1912 an arbitrary pointer to function value.
1913 </li>
1914
1915 <li>'<tt>function ptr val</tt>': An LLVM value containing a pointer to a
1916 function to be invoked. </li>
1917
1918 <li>'<tt>function args</tt>': argument list whose types match the function
1919 signature argument types. If the function signature indicates the function
1920 accepts a variable number of arguments, the extra arguments can be
1921 specified. </li>
1922
1923 <li>'<tt>normal label</tt>': the label reached when the called function
1924 executes a '<tt><a href="#i_ret">ret</a></tt>' instruction. </li>
1925
1926 <li>'<tt>exception label</tt>': the label reached when a callee returns with
1927 the <a href="#i_unwind"><tt>unwind</tt></a> instruction. </li>
1928
1929</ol>
1930
1931<h5>Semantics:</h5>
1932
1933<p>This instruction is designed to operate as a standard '<tt><a
1934href="#i_call">call</a></tt>' instruction in most regards. The primary
1935difference is that it establishes an association with a label, which is used by
1936the runtime library to unwind the stack.</p>
1937
1938<p>This instruction is used in languages with destructors to ensure that proper
1939cleanup is performed in the case of either a <tt>longjmp</tt> or a thrown
1940exception. Additionally, this is important for implementation of
1941'<tt>catch</tt>' clauses in high-level languages that support them.</p>
1942
1943<h5>Example:</h5>
1944<pre>
1945 %retval = invoke i32 %Test(i32 15) to label %Continue
1946 unwind label %TestCleanup <i>; {i32}:retval set</i>
1947 %retval = invoke <a href="#callingconv">coldcc</a> i32 %Test(i32 15) to label %Continue
1948 unwind label %TestCleanup <i>; {i32}:retval set</i>
1949</pre>
1950</div>
1951
1952
1953<!-- _______________________________________________________________________ -->
1954
1955<div class="doc_subsubsection"> <a name="i_unwind">'<tt>unwind</tt>'
1956Instruction</a> </div>
1957
1958<div class="doc_text">
1959
1960<h5>Syntax:</h5>
1961<pre>
1962 unwind
1963</pre>
1964
1965<h5>Overview:</h5>
1966
1967<p>The '<tt>unwind</tt>' instruction unwinds the stack, continuing control flow
1968at the first callee in the dynamic call stack which used an <a
1969href="#i_invoke"><tt>invoke</tt></a> instruction to perform the call. This is
1970primarily used to implement exception handling.</p>
1971
1972<h5>Semantics:</h5>
1973
1974<p>The '<tt>unwind</tt>' intrinsic causes execution of the current function to
1975immediately halt. The dynamic call stack is then searched for the first <a
1976href="#i_invoke"><tt>invoke</tt></a> instruction on the call stack. Once found,
1977execution continues at the "exceptional" destination block specified by the
1978<tt>invoke</tt> instruction. If there is no <tt>invoke</tt> instruction in the
1979dynamic call chain, undefined behavior results.</p>
1980</div>
1981
1982<!-- _______________________________________________________________________ -->
1983
1984<div class="doc_subsubsection"> <a name="i_unreachable">'<tt>unreachable</tt>'
1985Instruction</a> </div>
1986
1987<div class="doc_text">
1988
1989<h5>Syntax:</h5>
1990<pre>
1991 unreachable
1992</pre>
1993
1994<h5>Overview:</h5>
1995
1996<p>The '<tt>unreachable</tt>' instruction has no defined semantics. This
1997instruction is used to inform the optimizer that a particular portion of the
1998code is not reachable. This can be used to indicate that the code after a
1999no-return function cannot be reached, and other facts.</p>
2000
2001<h5>Semantics:</h5>
2002
2003<p>The '<tt>unreachable</tt>' instruction has no defined semantics.</p>
2004</div>
2005
2006
2007
2008<!-- ======================================================================= -->
2009<div class="doc_subsection"> <a name="binaryops">Binary Operations</a> </div>
2010<div class="doc_text">
2011<p>Binary operators are used to do most of the computation in a
2012program. They require two operands, execute an operation on them, and
2013produce a single value. The operands might represent
2014multiple data, as is the case with the <a href="#t_vector">vector</a> data type.
2015The result value of a binary operator is not
2016necessarily the same type as its operands.</p>
2017<p>There are several different binary operators:</p>
2018</div>
2019<!-- _______________________________________________________________________ -->
2020<div class="doc_subsubsection"> <a name="i_add">'<tt>add</tt>'
2021Instruction</a> </div>
2022<div class="doc_text">
2023<h5>Syntax:</h5>
2024<pre> &lt;result&gt; = add &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
2025</pre>
2026<h5>Overview:</h5>
2027<p>The '<tt>add</tt>' instruction returns the sum of its two operands.</p>
2028<h5>Arguments:</h5>
2029<p>The two arguments to the '<tt>add</tt>' instruction must be either <a
2030 href="#t_integer">integer</a> or <a href="#t_floating">floating point</a> values.
2031 This instruction can also take <a href="#t_vector">vector</a> versions of the values.
2032Both arguments must have identical types.</p>
2033<h5>Semantics:</h5>
2034<p>The value produced is the integer or floating point sum of the two
2035operands.</p>
2036<h5>Example:</h5>
2037<pre> &lt;result&gt; = add i32 4, %var <i>; yields {i32}:result = 4 + %var</i>
2038</pre>
2039</div>
2040<!-- _______________________________________________________________________ -->
2041<div class="doc_subsubsection"> <a name="i_sub">'<tt>sub</tt>'
2042Instruction</a> </div>
2043<div class="doc_text">
2044<h5>Syntax:</h5>
2045<pre> &lt;result&gt; = sub &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
2046</pre>
2047<h5>Overview:</h5>
2048<p>The '<tt>sub</tt>' instruction returns the difference of its two
2049operands.</p>
2050<p>Note that the '<tt>sub</tt>' instruction is used to represent the '<tt>neg</tt>'
2051instruction present in most other intermediate representations.</p>
2052<h5>Arguments:</h5>
2053<p>The two arguments to the '<tt>sub</tt>' instruction must be either <a
2054 href="#t_integer">integer</a> or <a href="#t_floating">floating point</a>
2055values.
2056This instruction can also take <a href="#t_vector">vector</a> versions of the values.
2057Both arguments must have identical types.</p>
2058<h5>Semantics:</h5>
2059<p>The value produced is the integer or floating point difference of
2060the two operands.</p>
2061<h5>Example:</h5>
2062<pre>
2063 &lt;result&gt; = sub i32 4, %var <i>; yields {i32}:result = 4 - %var</i>
2064 &lt;result&gt; = sub i32 0, %val <i>; yields {i32}:result = -%var</i>
2065</pre>
2066</div>
2067<!-- _______________________________________________________________________ -->
2068<div class="doc_subsubsection"> <a name="i_mul">'<tt>mul</tt>'
2069Instruction</a> </div>
2070<div class="doc_text">
2071<h5>Syntax:</h5>
2072<pre> &lt;result&gt; = mul &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
2073</pre>
2074<h5>Overview:</h5>
2075<p>The '<tt>mul</tt>' instruction returns the product of its two
2076operands.</p>
2077<h5>Arguments:</h5>
2078<p>The two arguments to the '<tt>mul</tt>' instruction must be either <a
2079 href="#t_integer">integer</a> or <a href="#t_floating">floating point</a>
2080values.
2081This instruction can also take <a href="#t_vector">vector</a> versions of the values.
2082Both arguments must have identical types.</p>
2083<h5>Semantics:</h5>
2084<p>The value produced is the integer or floating point product of the
2085two operands.</p>
2086<p>Because the operands are the same width, the result of an integer
2087multiplication is the same whether the operands should be deemed unsigned or
2088signed.</p>
2089<h5>Example:</h5>
2090<pre> &lt;result&gt; = mul i32 4, %var <i>; yields {i32}:result = 4 * %var</i>
2091</pre>
2092</div>
2093<!-- _______________________________________________________________________ -->
2094<div class="doc_subsubsection"> <a name="i_udiv">'<tt>udiv</tt>' Instruction
2095</a></div>
2096<div class="doc_text">
2097<h5>Syntax:</h5>
2098<pre> &lt;result&gt; = udiv &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
2099</pre>
2100<h5>Overview:</h5>
2101<p>The '<tt>udiv</tt>' instruction returns the quotient of its two
2102operands.</p>
2103<h5>Arguments:</h5>
2104<p>The two arguments to the '<tt>udiv</tt>' instruction must be
2105<a href="#t_integer">integer</a> values. Both arguments must have identical
2106types. This instruction can also take <a href="#t_vector">vector</a> versions
2107of the values in which case the elements must be integers.</p>
2108<h5>Semantics:</h5>
2109<p>The value produced is the unsigned integer quotient of the two operands. This
2110instruction always performs an unsigned division operation, regardless of
2111whether the arguments are unsigned or not.</p>
2112<h5>Example:</h5>
2113<pre> &lt;result&gt; = udiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
2114</pre>
2115</div>
2116<!-- _______________________________________________________________________ -->
2117<div class="doc_subsubsection"> <a name="i_sdiv">'<tt>sdiv</tt>' Instruction
2118</a> </div>
2119<div class="doc_text">
2120<h5>Syntax:</h5>
2121<pre> &lt;result&gt; = sdiv &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
2122</pre>
2123<h5>Overview:</h5>
2124<p>The '<tt>sdiv</tt>' instruction returns the quotient of its two
2125operands.</p>
2126<h5>Arguments:</h5>
2127<p>The two arguments to the '<tt>sdiv</tt>' instruction must be
2128<a href="#t_integer">integer</a> values. Both arguments must have identical
2129types. This instruction can also take <a href="#t_vector">vector</a> versions
2130of the values in which case the elements must be integers.</p>
2131<h5>Semantics:</h5>
2132<p>The value produced is the signed integer quotient of the two operands. This
2133instruction always performs a signed division operation, regardless of whether
2134the arguments are signed or not.</p>
2135<h5>Example:</h5>
2136<pre> &lt;result&gt; = sdiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
2137</pre>
2138</div>
2139<!-- _______________________________________________________________________ -->
2140<div class="doc_subsubsection"> <a name="i_fdiv">'<tt>fdiv</tt>'
2141Instruction</a> </div>
2142<div class="doc_text">
2143<h5>Syntax:</h5>
2144<pre> &lt;result&gt; = fdiv &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
2145</pre>
2146<h5>Overview:</h5>
2147<p>The '<tt>fdiv</tt>' instruction returns the quotient of its two
2148operands.</p>
2149<h5>Arguments:</h5>
2150<p>The two arguments to the '<tt>fdiv</tt>' instruction must be
2151<a href="#t_floating">floating point</a> values. Both arguments must have
2152identical types. This instruction can also take <a href="#t_vector">vector</a>
2153versions of floating point values.</p>
2154<h5>Semantics:</h5>
2155<p>The value produced is the floating point quotient of the two operands.</p>
2156<h5>Example:</h5>
2157<pre> &lt;result&gt; = fdiv float 4.0, %var <i>; yields {float}:result = 4.0 / %var</i>
2158</pre>
2159</div>
2160<!-- _______________________________________________________________________ -->
2161<div class="doc_subsubsection"> <a name="i_urem">'<tt>urem</tt>' Instruction</a>
2162</div>
2163<div class="doc_text">
2164<h5>Syntax:</h5>
2165<pre> &lt;result&gt; = urem &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
2166</pre>
2167<h5>Overview:</h5>
2168<p>The '<tt>urem</tt>' instruction returns the remainder from the
2169unsigned division of its two arguments.</p>
2170<h5>Arguments:</h5>
2171<p>The two arguments to the '<tt>urem</tt>' instruction must be
2172<a href="#t_integer">integer</a> values. Both arguments must have identical
Dan Gohman3e3fd8c2007-11-05 23:35:22 +0000[diff] [blame]2173types. This instruction can also take <a href="#t_vector">vector</a> versions
2174of the values in which case the elements must be integers.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2175<h5>Semantics:</h5>
2176<p>This instruction returns the unsigned integer <i>remainder</i> of a division.
2177This instruction always performs an unsigned division to get the remainder,
2178regardless of whether the arguments are unsigned or not.</p>
2179<h5>Example:</h5>
2180<pre> &lt;result&gt; = urem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
2181</pre>
2182
2183</div>
2184<!-- _______________________________________________________________________ -->
2185<div class="doc_subsubsection"> <a name="i_srem">'<tt>srem</tt>'
2186Instruction</a> </div>
2187<div class="doc_text">
2188<h5>Syntax:</h5>
2189<pre> &lt;result&gt; = srem &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
2190</pre>
2191<h5>Overview:</h5>
2192<p>The '<tt>srem</tt>' instruction returns the remainder from the
Dan Gohman3e3fd8c2007-11-05 23:35:22 +0000[diff] [blame]2193signed division of its two operands. This instruction can also take
2194<a href="#t_vector">vector</a> versions of the values in which case
2195the elements must be integers.</p>
2196</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2197<h5>Arguments:</h5>
2198<p>The two arguments to the '<tt>srem</tt>' instruction must be
2199<a href="#t_integer">integer</a> values. Both arguments must have identical
2200types.</p>
2201<h5>Semantics:</h5>
2202<p>This instruction returns the <i>remainder</i> of a division (where the result
2203has the same sign as the dividend, <tt>var1</tt>), not the <i>modulo</i>
2204operator (where the result has the same sign as the divisor, <tt>var2</tt>) of
2205a value. For more information about the difference, see <a
2206 href="http://mathforum.org/dr.math/problems/anne.4.28.99.html">The
2207Math Forum</a>. For a table of how this is implemented in various languages,
2208please see <a href="http://en.wikipedia.org/wiki/Modulo_operation">
2209Wikipedia: modulo operation</a>.</p>
2210<h5>Example:</h5>
2211<pre> &lt;result&gt; = srem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
2212</pre>
2213
2214</div>
2215<!-- _______________________________________________________________________ -->
2216<div class="doc_subsubsection"> <a name="i_frem">'<tt>frem</tt>'
2217Instruction</a> </div>
2218<div class="doc_text">
2219<h5>Syntax:</h5>
2220<pre> &lt;result&gt; = frem &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
2221</pre>
2222<h5>Overview:</h5>
2223<p>The '<tt>frem</tt>' instruction returns the remainder from the
2224division of its two operands.</p>
2225<h5>Arguments:</h5>
2226<p>The two arguments to the '<tt>frem</tt>' instruction must be
2227<a href="#t_floating">floating point</a> values. Both arguments must have
Dan Gohman3e3fd8c2007-11-05 23:35:22 +0000[diff] [blame]2228identical types. This instruction can also take <a href="#t_vector">vector</a>
2229versions of floating point values.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2230<h5>Semantics:</h5>
2231<p>This instruction returns the <i>remainder</i> of a division.</p>
2232<h5>Example:</h5>
2233<pre> &lt;result&gt; = frem float 4.0, %var <i>; yields {float}:result = 4.0 % %var</i>
2234</pre>
2235</div>
2236
2237<!-- ======================================================================= -->
2238<div class="doc_subsection"> <a name="bitwiseops">Bitwise Binary
2239Operations</a> </div>
2240<div class="doc_text">
2241<p>Bitwise binary operators are used to do various forms of
2242bit-twiddling in a program. They are generally very efficient
2243instructions and can commonly be strength reduced from other
2244instructions. They require two operands, execute an operation on them,
2245and produce a single value. The resulting value of the bitwise binary
2246operators is always the same type as its first operand.</p>
2247</div>
2248
2249<!-- _______________________________________________________________________ -->
2250<div class="doc_subsubsection"> <a name="i_shl">'<tt>shl</tt>'
2251Instruction</a> </div>
2252<div class="doc_text">
2253<h5>Syntax:</h5>
2254<pre> &lt;result&gt; = shl &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
2255</pre>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2256
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2257<h5>Overview:</h5>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2258
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2259<p>The '<tt>shl</tt>' instruction returns the first operand shifted to
2260the left a specified number of bits.</p>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2261
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2262<h5>Arguments:</h5>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2263
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2264<p>Both arguments to the '<tt>shl</tt>' instruction must be the same <a
2265 href="#t_integer">integer</a> type.</p>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2266
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2267<h5>Semantics:</h5>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2268
2269<p>The value produced is <tt>var1</tt> * 2<sup><tt>var2</tt></sup>. If
2270<tt>var2</tt> is (statically or dynamically) equal to or larger than the number
2271of bits in <tt>var1</tt>, the result is undefined.</p>
2272
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2273<h5>Example:</h5><pre>
2274 &lt;result&gt; = shl i32 4, %var <i>; yields {i32}: 4 &lt;&lt; %var</i>
2275 &lt;result&gt; = shl i32 4, 2 <i>; yields {i32}: 16</i>
2276 &lt;result&gt; = shl i32 1, 10 <i>; yields {i32}: 1024</i>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2277 &lt;result&gt; = shl i32 1, 32 <i>; undefined</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2278</pre>
2279</div>
2280<!-- _______________________________________________________________________ -->
2281<div class="doc_subsubsection"> <a name="i_lshr">'<tt>lshr</tt>'
2282Instruction</a> </div>
2283<div class="doc_text">
2284<h5>Syntax:</h5>
2285<pre> &lt;result&gt; = lshr &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
2286</pre>
2287
2288<h5>Overview:</h5>
2289<p>The '<tt>lshr</tt>' instruction (logical shift right) returns the first
2290operand shifted to the right a specified number of bits with zero fill.</p>
2291
2292<h5>Arguments:</h5>
2293<p>Both arguments to the '<tt>lshr</tt>' instruction must be the same
2294<a href="#t_integer">integer</a> type.</p>
2295
2296<h5>Semantics:</h5>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2297
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2298<p>This instruction always performs a logical shift right operation. The most
2299significant bits of the result will be filled with zero bits after the
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2300shift. If <tt>var2</tt> is (statically or dynamically) equal to or larger than
2301the number of bits in <tt>var1</tt>, the result is undefined.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2302
2303<h5>Example:</h5>
2304<pre>
2305 &lt;result&gt; = lshr i32 4, 1 <i>; yields {i32}:result = 2</i>
2306 &lt;result&gt; = lshr i32 4, 2 <i>; yields {i32}:result = 1</i>
2307 &lt;result&gt; = lshr i8 4, 3 <i>; yields {i8}:result = 0</i>
2308 &lt;result&gt; = lshr i8 -2, 1 <i>; yields {i8}:result = 0x7FFFFFFF </i>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2309 &lt;result&gt; = lshr i32 1, 32 <i>; undefined</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2310</pre>
2311</div>
2312
2313<!-- _______________________________________________________________________ -->
2314<div class="doc_subsubsection"> <a name="i_ashr">'<tt>ashr</tt>'
2315Instruction</a> </div>
2316<div class="doc_text">
2317
2318<h5>Syntax:</h5>
2319<pre> &lt;result&gt; = ashr &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
2320</pre>
2321
2322<h5>Overview:</h5>
2323<p>The '<tt>ashr</tt>' instruction (arithmetic shift right) returns the first
2324operand shifted to the right a specified number of bits with sign extension.</p>
2325
2326<h5>Arguments:</h5>
2327<p>Both arguments to the '<tt>ashr</tt>' instruction must be the same
2328<a href="#t_integer">integer</a> type.</p>
2329
2330<h5>Semantics:</h5>
2331<p>This instruction always performs an arithmetic shift right operation,
2332The most significant bits of the result will be filled with the sign bit
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2333of <tt>var1</tt>. If <tt>var2</tt> is (statically or dynamically) equal to or
2334larger than the number of bits in <tt>var1</tt>, the result is undefined.
2335</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2336
2337<h5>Example:</h5>
2338<pre>
2339 &lt;result&gt; = ashr i32 4, 1 <i>; yields {i32}:result = 2</i>
2340 &lt;result&gt; = ashr i32 4, 2 <i>; yields {i32}:result = 1</i>
2341 &lt;result&gt; = ashr i8 4, 3 <i>; yields {i8}:result = 0</i>
2342 &lt;result&gt; = ashr i8 -2, 1 <i>; yields {i8}:result = -1</i>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2343 &lt;result&gt; = ashr i32 1, 32 <i>; undefined</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2344</pre>
2345</div>
2346
2347<!-- _______________________________________________________________________ -->
2348<div class="doc_subsubsection"> <a name="i_and">'<tt>and</tt>'
2349Instruction</a> </div>
2350<div class="doc_text">
2351<h5>Syntax:</h5>
2352<pre> &lt;result&gt; = and &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
2353</pre>
2354<h5>Overview:</h5>
2355<p>The '<tt>and</tt>' instruction returns the bitwise logical and of
2356its two operands.</p>
2357<h5>Arguments:</h5>
2358<p>The two arguments to the '<tt>and</tt>' instruction must be <a
2359 href="#t_integer">integer</a> values. Both arguments must have
2360identical types.</p>
2361<h5>Semantics:</h5>
2362<p>The truth table used for the '<tt>and</tt>' instruction is:</p>
2363<p> </p>
2364<div style="align: center">
2365<table border="1" cellspacing="0" cellpadding="4">
2366 <tbody>
2367 <tr>
2368 <td>In0</td>
2369 <td>In1</td>
2370 <td>Out</td>
2371 </tr>
2372 <tr>
2373 <td>0</td>
2374 <td>0</td>
2375 <td>0</td>
2376 </tr>
2377 <tr>
2378 <td>0</td>
2379 <td>1</td>
2380 <td>0</td>
2381 </tr>
2382 <tr>
2383 <td>1</td>
2384 <td>0</td>
2385 <td>0</td>
2386 </tr>
2387 <tr>
2388 <td>1</td>
2389 <td>1</td>
2390 <td>1</td>
2391 </tr>
2392 </tbody>
2393</table>
2394</div>
2395<h5>Example:</h5>
2396<pre> &lt;result&gt; = and i32 4, %var <i>; yields {i32}:result = 4 &amp; %var</i>
2397 &lt;result&gt; = and i32 15, 40 <i>; yields {i32}:result = 8</i>
2398 &lt;result&gt; = and i32 4, 8 <i>; yields {i32}:result = 0</i>
2399</pre>
2400</div>
2401<!-- _______________________________________________________________________ -->
2402<div class="doc_subsubsection"> <a name="i_or">'<tt>or</tt>' Instruction</a> </div>
2403<div class="doc_text">
2404<h5>Syntax:</h5>
2405<pre> &lt;result&gt; = or &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
2406</pre>
2407<h5>Overview:</h5>
2408<p>The '<tt>or</tt>' instruction returns the bitwise logical inclusive
2409or of its two operands.</p>
2410<h5>Arguments:</h5>
2411<p>The two arguments to the '<tt>or</tt>' instruction must be <a
2412 href="#t_integer">integer</a> values. Both arguments must have
2413identical types.</p>
2414<h5>Semantics:</h5>
2415<p>The truth table used for the '<tt>or</tt>' instruction is:</p>
2416<p> </p>
2417<div style="align: center">
2418<table border="1" cellspacing="0" cellpadding="4">
2419 <tbody>
2420 <tr>
2421 <td>In0</td>
2422 <td>In1</td>
2423 <td>Out</td>
2424 </tr>
2425 <tr>
2426 <td>0</td>
2427 <td>0</td>
2428 <td>0</td>
2429 </tr>
2430 <tr>
2431 <td>0</td>
2432 <td>1</td>
2433 <td>1</td>
2434 </tr>
2435 <tr>
2436 <td>1</td>
2437 <td>0</td>
2438 <td>1</td>
2439 </tr>
2440 <tr>
2441 <td>1</td>
2442 <td>1</td>
2443 <td>1</td>
2444 </tr>
2445 </tbody>
2446</table>
2447</div>
2448<h5>Example:</h5>
2449<pre> &lt;result&gt; = or i32 4, %var <i>; yields {i32}:result = 4 | %var</i>
2450 &lt;result&gt; = or i32 15, 40 <i>; yields {i32}:result = 47</i>
2451 &lt;result&gt; = or i32 4, 8 <i>; yields {i32}:result = 12</i>
2452</pre>
2453</div>
2454<!-- _______________________________________________________________________ -->
2455<div class="doc_subsubsection"> <a name="i_xor">'<tt>xor</tt>'
2456Instruction</a> </div>
2457<div class="doc_text">
2458<h5>Syntax:</h5>
2459<pre> &lt;result&gt; = xor &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
2460</pre>
2461<h5>Overview:</h5>
2462<p>The '<tt>xor</tt>' instruction returns the bitwise logical exclusive
2463or of its two operands. The <tt>xor</tt> is used to implement the
2464"one's complement" operation, which is the "~" operator in C.</p>
2465<h5>Arguments:</h5>
2466<p>The two arguments to the '<tt>xor</tt>' instruction must be <a
2467 href="#t_integer">integer</a> values. Both arguments must have
2468identical types.</p>
2469<h5>Semantics:</h5>
2470<p>The truth table used for the '<tt>xor</tt>' instruction is:</p>
2471<p> </p>
2472<div style="align: center">
2473<table border="1" cellspacing="0" cellpadding="4">
2474 <tbody>
2475 <tr>
2476 <td>In0</td>
2477 <td>In1</td>
2478 <td>Out</td>
2479 </tr>
2480 <tr>
2481 <td>0</td>
2482 <td>0</td>
2483 <td>0</td>
2484 </tr>
2485 <tr>
2486 <td>0</td>
2487 <td>1</td>
2488 <td>1</td>
2489 </tr>
2490 <tr>
2491 <td>1</td>
2492 <td>0</td>
2493 <td>1</td>
2494 </tr>
2495 <tr>
2496 <td>1</td>
2497 <td>1</td>
2498 <td>0</td>
2499 </tr>
2500 </tbody>
2501</table>
2502</div>
2503<p> </p>
2504<h5>Example:</h5>
2505<pre> &lt;result&gt; = xor i32 4, %var <i>; yields {i32}:result = 4 ^ %var</i>
2506 &lt;result&gt; = xor i32 15, 40 <i>; yields {i32}:result = 39</i>
2507 &lt;result&gt; = xor i32 4, 8 <i>; yields {i32}:result = 12</i>
2508 &lt;result&gt; = xor i32 %V, -1 <i>; yields {i32}:result = ~%V</i>
2509</pre>
2510</div>
2511
2512<!-- ======================================================================= -->
2513<div class="doc_subsection">
2514 <a name="vectorops">Vector Operations</a>
2515</div>
2516
2517<div class="doc_text">
2518
2519<p>LLVM supports several instructions to represent vector operations in a
2520target-independent manner. These instructions cover the element-access and
2521vector-specific operations needed to process vectors effectively. While LLVM
2522does directly support these vector operations, many sophisticated algorithms
2523will want to use target-specific intrinsics to take full advantage of a specific
2524target.</p>
2525
2526</div>
2527
2528<!-- _______________________________________________________________________ -->
2529<div class="doc_subsubsection">
2530 <a name="i_extractelement">'<tt>extractelement</tt>' Instruction</a>
2531</div>
2532
2533<div class="doc_text">
2534
2535<h5>Syntax:</h5>
2536
2537<pre>
2538 &lt;result&gt; = extractelement &lt;n x &lt;ty&gt;&gt; &lt;val&gt;, i32 &lt;idx&gt; <i>; yields &lt;ty&gt;</i>
2539</pre>
2540
2541<h5>Overview:</h5>
2542
2543<p>
2544The '<tt>extractelement</tt>' instruction extracts a single scalar
2545element from a vector at a specified index.
2546</p>
2547
2548
2549<h5>Arguments:</h5>
2550
2551<p>
2552The first operand of an '<tt>extractelement</tt>' instruction is a
2553value of <a href="#t_vector">vector</a> type. The second operand is
2554an index indicating the position from which to extract the element.
2555The index may be a variable.</p>
2556
2557<h5>Semantics:</h5>
2558
2559<p>
2560The result is a scalar of the same type as the element type of
2561<tt>val</tt>. Its value is the value at position <tt>idx</tt> of
2562<tt>val</tt>. If <tt>idx</tt> exceeds the length of <tt>val</tt>, the
2563results are undefined.
2564</p>
2565
2566<h5>Example:</h5>
2567
2568<pre>
2569 %result = extractelement &lt;4 x i32&gt; %vec, i32 0 <i>; yields i32</i>
2570</pre>
2571</div>
2572
2573
2574<!-- _______________________________________________________________________ -->
2575<div class="doc_subsubsection">
2576 <a name="i_insertelement">'<tt>insertelement</tt>' Instruction</a>
2577</div>
2578
2579<div class="doc_text">
2580
2581<h5>Syntax:</h5>
2582
2583<pre>
2584 &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>
2585</pre>
2586
2587<h5>Overview:</h5>
2588
2589<p>
2590The '<tt>insertelement</tt>' instruction inserts a scalar
2591element into a vector at a specified index.
2592</p>
2593
2594
2595<h5>Arguments:</h5>
2596
2597<p>
2598The first operand of an '<tt>insertelement</tt>' instruction is a
2599value of <a href="#t_vector">vector</a> type. The second operand is a
2600scalar value whose type must equal the element type of the first
2601operand. The third operand is an index indicating the position at
2602which to insert the value. The index may be a variable.</p>
2603
2604<h5>Semantics:</h5>
2605
2606<p>
2607The result is a vector of the same type as <tt>val</tt>. Its
2608element values are those of <tt>val</tt> except at position
2609<tt>idx</tt>, where it gets the value <tt>elt</tt>. If <tt>idx</tt>
2610exceeds the length of <tt>val</tt>, the results are undefined.
2611</p>
2612
2613<h5>Example:</h5>
2614
2615<pre>
2616 %result = insertelement &lt;4 x i32&gt; %vec, i32 1, i32 0 <i>; yields &lt;4 x i32&gt;</i>
2617</pre>
2618</div>
2619
2620<!-- _______________________________________________________________________ -->
2621<div class="doc_subsubsection">
2622 <a name="i_shufflevector">'<tt>shufflevector</tt>' Instruction</a>
2623</div>
2624
2625<div class="doc_text">
2626
2627<h5>Syntax:</h5>
2628
2629<pre>
2630 &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>
2631</pre>
2632
2633<h5>Overview:</h5>
2634
2635<p>
2636The '<tt>shufflevector</tt>' instruction constructs a permutation of elements
2637from two input vectors, returning a vector of the same type.
2638</p>
2639
2640<h5>Arguments:</h5>
2641
2642<p>
2643The first two operands of a '<tt>shufflevector</tt>' instruction are vectors
2644with types that match each other and types that match the result of the
2645instruction. The third argument is a shuffle mask, which has the same number
2646of elements as the other vector type, but whose element type is always 'i32'.
2647</p>
2648
2649<p>
2650The shuffle mask operand is required to be a constant vector with either
2651constant integer or undef values.
2652</p>
2653
2654<h5>Semantics:</h5>
2655
2656<p>
2657The elements of the two input vectors are numbered from left to right across
2658both of the vectors. The shuffle mask operand specifies, for each element of
2659the result vector, which element of the two input registers the result element
2660gets. The element selector may be undef (meaning "don't care") and the second
2661operand may be undef if performing a shuffle from only one vector.
2662</p>
2663
2664<h5>Example:</h5>
2665
2666<pre>
2667 %result = shufflevector &lt;4 x i32&gt; %v1, &lt;4 x i32&gt; %v2,
2668 &lt;4 x i32&gt; &lt;i32 0, i32 4, i32 1, i32 5&gt; <i>; yields &lt;4 x i32&gt;</i>
2669 %result = shufflevector &lt;4 x i32&gt; %v1, &lt;4 x i32&gt; undef,
2670 &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.
2671</pre>
2672</div>
2673
2674
2675<!-- ======================================================================= -->
2676<div class="doc_subsection">
2677 <a name="memoryops">Memory Access and Addressing Operations</a>
2678</div>
2679
2680<div class="doc_text">
2681
2682<p>A key design point of an SSA-based representation is how it
2683represents memory. In LLVM, no memory locations are in SSA form, which
2684makes things very simple. This section describes how to read, write,
2685allocate, and free memory in LLVM.</p>
2686
2687</div>
2688
2689<!-- _______________________________________________________________________ -->
2690<div class="doc_subsubsection">
2691 <a name="i_malloc">'<tt>malloc</tt>' Instruction</a>
2692</div>
2693
2694<div class="doc_text">
2695
2696<h5>Syntax:</h5>
2697
2698<pre>
2699 &lt;result&gt; = malloc &lt;type&gt;[, i32 &lt;NumElements&gt;][, align &lt;alignment&gt;] <i>; yields {type*}:result</i>
2700</pre>
2701
2702<h5>Overview:</h5>
2703
2704<p>The '<tt>malloc</tt>' instruction allocates memory from the system
Christopher Lambcfe00962007-12-17 01:00:21 +0000[diff] [blame]2705heap and returns a pointer to it. The object is always allocated in the generic
2706address space (address space zero).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2707
2708<h5>Arguments:</h5>
2709
2710<p>The '<tt>malloc</tt>' instruction allocates
2711<tt>sizeof(&lt;type&gt;)*NumElements</tt>
2712bytes of memory from the operating system and returns a pointer of the
2713appropriate type to the program. If "NumElements" is specified, it is the
2714number of elements allocated. If an alignment is specified, the value result
2715of the allocation is guaranteed to be aligned to at least that boundary. If
2716not specified, or if zero, the target can choose to align the allocation on any
2717convenient boundary.</p>
2718
2719<p>'<tt>type</tt>' must be a sized type.</p>
2720
2721<h5>Semantics:</h5>
2722
2723<p>Memory is allocated using the system "<tt>malloc</tt>" function, and
2724a pointer is returned.</p>
2725
2726<h5>Example:</h5>
2727
2728<pre>
2729 %array = malloc [4 x i8 ] <i>; yields {[%4 x i8]*}:array</i>
2730
2731 %size = <a href="#i_add">add</a> i32 2, 2 <i>; yields {i32}:size = i32 4</i>
2732 %array1 = malloc i8, i32 4 <i>; yields {i8*}:array1</i>
2733 %array2 = malloc [12 x i8], i32 %size <i>; yields {[12 x i8]*}:array2</i>
2734 %array3 = malloc i32, i32 4, align 1024 <i>; yields {i32*}:array3</i>
2735 %array4 = malloc i32, align 1024 <i>; yields {i32*}:array4</i>
2736</pre>
2737</div>
2738
2739<!-- _______________________________________________________________________ -->
2740<div class="doc_subsubsection">
2741 <a name="i_free">'<tt>free</tt>' Instruction</a>
2742</div>
2743
2744<div class="doc_text">
2745
2746<h5>Syntax:</h5>
2747
2748<pre>
2749 free &lt;type&gt; &lt;value&gt; <i>; yields {void}</i>
2750</pre>
2751
2752<h5>Overview:</h5>
2753
2754<p>The '<tt>free</tt>' instruction returns memory back to the unused
2755memory heap to be reallocated in the future.</p>
2756
2757<h5>Arguments:</h5>
2758
2759<p>'<tt>value</tt>' shall be a pointer value that points to a value
2760that was allocated with the '<tt><a href="#i_malloc">malloc</a></tt>'
2761instruction.</p>
2762
2763<h5>Semantics:</h5>
2764
2765<p>Access to the memory pointed to by the pointer is no longer defined
2766after this instruction executes.</p>
2767
2768<h5>Example:</h5>
2769
2770<pre>
2771 %array = <a href="#i_malloc">malloc</a> [4 x i8] <i>; yields {[4 x i8]*}:array</i>
2772 free [4 x i8]* %array
2773</pre>
2774</div>
2775
2776<!-- _______________________________________________________________________ -->
2777<div class="doc_subsubsection">
2778 <a name="i_alloca">'<tt>alloca</tt>' Instruction</a>
2779</div>
2780
2781<div class="doc_text">
2782
2783<h5>Syntax:</h5>
2784
2785<pre>
2786 &lt;result&gt; = alloca &lt;type&gt;[, i32 &lt;NumElements&gt;][, align &lt;alignment&gt;] <i>; yields {type*}:result</i>
2787</pre>
2788
2789<h5>Overview:</h5>
2790
2791<p>The '<tt>alloca</tt>' instruction allocates memory on the stack frame of the
2792currently executing function, to be automatically released when this function
Christopher Lambcfe00962007-12-17 01:00:21 +0000[diff] [blame]2793returns to its caller. The object is always allocated in the generic address
2794space (address space zero).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2795
2796<h5>Arguments:</h5>
2797
2798<p>The '<tt>alloca</tt>' instruction allocates <tt>sizeof(&lt;type&gt;)*NumElements</tt>
2799bytes of memory on the runtime stack, returning a pointer of the
2800appropriate type to the program. If "NumElements" is specified, it is the
2801number of elements allocated. If an alignment is specified, the value result
2802of the allocation is guaranteed to be aligned to at least that boundary. If
2803not specified, or if zero, the target can choose to align the allocation on any
2804convenient boundary.</p>
2805
2806<p>'<tt>type</tt>' may be any sized type.</p>
2807
2808<h5>Semantics:</h5>
2809
2810<p>Memory is allocated; a pointer is returned. '<tt>alloca</tt>'d
2811memory is automatically released when the function returns. The '<tt>alloca</tt>'
2812instruction is commonly used to represent automatic variables that must
2813have an address available. When the function returns (either with the <tt><a
2814 href="#i_ret">ret</a></tt> or <tt><a href="#i_unwind">unwind</a></tt>
2815instructions), the memory is reclaimed.</p>
2816
2817<h5>Example:</h5>
2818
2819<pre>
2820 %ptr = alloca i32 <i>; yields {i32*}:ptr</i>
2821 %ptr = alloca i32, i32 4 <i>; yields {i32*}:ptr</i>
2822 %ptr = alloca i32, i32 4, align 1024 <i>; yields {i32*}:ptr</i>
2823 %ptr = alloca i32, align 1024 <i>; yields {i32*}:ptr</i>
2824</pre>
2825</div>
2826
2827<!-- _______________________________________________________________________ -->
2828<div class="doc_subsubsection"> <a name="i_load">'<tt>load</tt>'
2829Instruction</a> </div>
2830<div class="doc_text">
2831<h5>Syntax:</h5>
2832<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>
2833<h5>Overview:</h5>
2834<p>The '<tt>load</tt>' instruction is used to read from memory.</p>
2835<h5>Arguments:</h5>
2836<p>The argument to the '<tt>load</tt>' instruction specifies the memory
2837address from which to load. The pointer must point to a <a
2838 href="#t_firstclass">first class</a> type. If the <tt>load</tt> is
2839marked as <tt>volatile</tt>, then the optimizer is not allowed to modify
2840the number or order of execution of this <tt>load</tt> with other
2841volatile <tt>load</tt> and <tt><a href="#i_store">store</a></tt>
2842instructions. </p>
2843<h5>Semantics:</h5>
2844<p>The location of memory pointed to is loaded.</p>
2845<h5>Examples:</h5>
2846<pre> %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
2847 <a
2848 href="#i_store">store</a> i32 3, i32* %ptr <i>; yields {void}</i>
2849 %val = load i32* %ptr <i>; yields {i32}:val = i32 3</i>
2850</pre>
2851</div>
2852<!-- _______________________________________________________________________ -->
2853<div class="doc_subsubsection"> <a name="i_store">'<tt>store</tt>'
2854Instruction</a> </div>
2855<div class="doc_text">
2856<h5>Syntax:</h5>
2857<pre> store &lt;ty&gt; &lt;value&gt;, &lt;ty&gt;* &lt;pointer&gt;[, align &lt;alignment&gt;] <i>; yields {void}</i>
2858 volatile store &lt;ty&gt; &lt;value&gt;, &lt;ty&gt;* &lt;pointer&gt;[, align &lt;alignment&gt;] <i>; yields {void}</i>
2859</pre>
2860<h5>Overview:</h5>
2861<p>The '<tt>store</tt>' instruction is used to write to memory.</p>
2862<h5>Arguments:</h5>
2863<p>There are two arguments to the '<tt>store</tt>' instruction: a value
2864to store and an address at which to store it. The type of the '<tt>&lt;pointer&gt;</tt>'
2865operand must be a pointer to the type of the '<tt>&lt;value&gt;</tt>'
2866operand. If the <tt>store</tt> is marked as <tt>volatile</tt>, then the
2867optimizer is not allowed to modify the number or order of execution of
2868this <tt>store</tt> with other volatile <tt>load</tt> and <tt><a
2869 href="#i_store">store</a></tt> instructions.</p>
2870<h5>Semantics:</h5>
2871<p>The contents of memory are updated to contain '<tt>&lt;value&gt;</tt>'
2872at the location specified by the '<tt>&lt;pointer&gt;</tt>' operand.</p>
2873<h5>Example:</h5>
2874<pre> %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
Bill Wendling63ffa142007-10-22 05:10:05 +0000[diff] [blame]2875 store i32 3, i32* %ptr <i>; yields {void}</i>
2876 %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]2877</pre>
2878</div>
2879
2880<!-- _______________________________________________________________________ -->
2881<div class="doc_subsubsection">
2882 <a name="i_getelementptr">'<tt>getelementptr</tt>' Instruction</a>
2883</div>
2884
2885<div class="doc_text">
2886<h5>Syntax:</h5>
2887<pre>
2888 &lt;result&gt; = getelementptr &lt;ty&gt;* &lt;ptrval&gt;{, &lt;ty&gt; &lt;idx&gt;}*
2889</pre>
2890
2891<h5>Overview:</h5>
2892
2893<p>
2894The '<tt>getelementptr</tt>' instruction is used to get the address of a
2895subelement of an aggregate data structure.</p>
2896
2897<h5>Arguments:</h5>
2898
2899<p>This instruction takes a list of integer operands that indicate what
2900elements of the aggregate object to index to. The actual types of the arguments
2901provided depend on the type of the first pointer argument. The
2902'<tt>getelementptr</tt>' instruction is used to index down through the type
2903levels of a structure or to a specific index in an array. When indexing into a
2904structure, only <tt>i32</tt> integer constants are allowed. When indexing
2905into an array or pointer, only integers of 32 or 64 bits are allowed, and will
2906be sign extended to 64-bit values.</p>
2907
2908<p>For example, let's consider a C code fragment and how it gets
2909compiled to LLVM:</p>
2910
2911<div class="doc_code">
2912<pre>
2913struct RT {
2914 char A;
2915 int B[10][20];
2916 char C;
2917};
2918struct ST {
2919 int X;
2920 double Y;
2921 struct RT Z;
2922};
2923
2924int *foo(struct ST *s) {
2925 return &amp;s[1].Z.B[5][13];
2926}
2927</pre>
2928</div>
2929
2930<p>The LLVM code generated by the GCC frontend is:</p>
2931
2932<div class="doc_code">
2933<pre>
2934%RT = type { i8 , [10 x [20 x i32]], i8 }
2935%ST = type { i32, double, %RT }
2936
2937define i32* %foo(%ST* %s) {
2938entry:
2939 %reg = getelementptr %ST* %s, i32 1, i32 2, i32 1, i32 5, i32 13
2940 ret i32* %reg
2941}
2942</pre>
2943</div>
2944
2945<h5>Semantics:</h5>
2946
2947<p>The index types specified for the '<tt>getelementptr</tt>' instruction depend
2948on the pointer type that is being indexed into. <a href="#t_pointer">Pointer</a>
2949and <a href="#t_array">array</a> types can use a 32-bit or 64-bit
2950<a href="#t_integer">integer</a> type but the value will always be sign extended
2951to 64-bits. <a href="#t_struct">Structure</a> types require <tt>i32</tt>
2952<b>constants</b>.</p>
2953
2954<p>In the example above, the first index is indexing into the '<tt>%ST*</tt>'
2955type, which is a pointer, yielding a '<tt>%ST</tt>' = '<tt>{ i32, double, %RT
2956}</tt>' type, a structure. The second index indexes into the third element of
2957the structure, yielding a '<tt>%RT</tt>' = '<tt>{ i8 , [10 x [20 x i32]],
2958i8 }</tt>' type, another structure. The third index indexes into the second
2959element of the structure, yielding a '<tt>[10 x [20 x i32]]</tt>' type, an
2960array. The two dimensions of the array are subscripted into, yielding an
2961'<tt>i32</tt>' type. The '<tt>getelementptr</tt>' instruction returns a pointer
2962to this element, thus computing a value of '<tt>i32*</tt>' type.</p>
2963
2964<p>Note that it is perfectly legal to index partially through a
2965structure, returning a pointer to an inner element. Because of this,
2966the LLVM code for the given testcase is equivalent to:</p>
2967
2968<pre>
2969 define i32* %foo(%ST* %s) {
2970 %t1 = getelementptr %ST* %s, i32 1 <i>; yields %ST*:%t1</i>
2971 %t2 = getelementptr %ST* %t1, i32 0, i32 2 <i>; yields %RT*:%t2</i>
2972 %t3 = getelementptr %RT* %t2, i32 0, i32 1 <i>; yields [10 x [20 x i32]]*:%t3</i>
2973 %t4 = getelementptr [10 x [20 x i32]]* %t3, i32 0, i32 5 <i>; yields [20 x i32]*:%t4</i>
2974 %t5 = getelementptr [20 x i32]* %t4, i32 0, i32 13 <i>; yields i32*:%t5</i>
2975 ret i32* %t5
2976 }
2977</pre>
2978
2979<p>Note that it is undefined to access an array out of bounds: array and
2980pointer indexes must always be within the defined bounds of the array type.
2981The one exception for this rules is zero length arrays. These arrays are
2982defined to be accessible as variable length arrays, which requires access
2983beyond the zero'th element.</p>
2984
2985<p>The getelementptr instruction is often confusing. For some more insight
2986into how it works, see <a href="GetElementPtr.html">the getelementptr
2987FAQ</a>.</p>
2988
2989<h5>Example:</h5>
2990
2991<pre>
2992 <i>; yields [12 x i8]*:aptr</i>
2993 %aptr = getelementptr {i32, [12 x i8]}* %sptr, i64 0, i32 1
2994</pre>
2995</div>
2996
2997<!-- ======================================================================= -->
2998<div class="doc_subsection"> <a name="convertops">Conversion Operations</a>
2999</div>
3000<div class="doc_text">
3001<p>The instructions in this category are the conversion instructions (casting)
3002which all take a single operand and a type. They perform various bit conversions
3003on the operand.</p>
3004</div>
3005
3006<!-- _______________________________________________________________________ -->
3007<div class="doc_subsubsection">
3008 <a name="i_trunc">'<tt>trunc .. to</tt>' Instruction</a>
3009</div>
3010<div class="doc_text">
3011
3012<h5>Syntax:</h5>
3013<pre>
3014 &lt;result&gt; = trunc &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3015</pre>
3016
3017<h5>Overview:</h5>
3018<p>
3019The '<tt>trunc</tt>' instruction truncates its operand to the type <tt>ty2</tt>.
3020</p>
3021
3022<h5>Arguments:</h5>
3023<p>
3024The '<tt>trunc</tt>' instruction takes a <tt>value</tt> to trunc, which must
3025be an <a href="#t_integer">integer</a> type, and a type that specifies the size
3026and type of the result, which must be an <a href="#t_integer">integer</a>
3027type. The bit size of <tt>value</tt> must be larger than the bit size of
3028<tt>ty2</tt>. Equal sized types are not allowed.</p>
3029
3030<h5>Semantics:</h5>
3031<p>
3032The '<tt>trunc</tt>' instruction truncates the high order bits in <tt>value</tt>
3033and converts the remaining bits to <tt>ty2</tt>. Since the source size must be
3034larger than the destination size, <tt>trunc</tt> cannot be a <i>no-op cast</i>.
3035It will always truncate bits.</p>
3036
3037<h5>Example:</h5>
3038<pre>
3039 %X = trunc i32 257 to i8 <i>; yields i8:1</i>
3040 %Y = trunc i32 123 to i1 <i>; yields i1:true</i>
3041 %Y = trunc i32 122 to i1 <i>; yields i1:false</i>
3042</pre>
3043</div>
3044
3045<!-- _______________________________________________________________________ -->
3046<div class="doc_subsubsection">
3047 <a name="i_zext">'<tt>zext .. to</tt>' Instruction</a>
3048</div>
3049<div class="doc_text">
3050
3051<h5>Syntax:</h5>
3052<pre>
3053 &lt;result&gt; = zext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3054</pre>
3055
3056<h5>Overview:</h5>
3057<p>The '<tt>zext</tt>' instruction zero extends its operand to type
3058<tt>ty2</tt>.</p>
3059
3060
3061<h5>Arguments:</h5>
3062<p>The '<tt>zext</tt>' instruction takes a value to cast, which must be of
3063<a href="#t_integer">integer</a> type, and a type to cast it to, which must
3064also be of <a href="#t_integer">integer</a> type. The bit size of the
3065<tt>value</tt> must be smaller than the bit size of the destination type,
3066<tt>ty2</tt>.</p>
3067
3068<h5>Semantics:</h5>
3069<p>The <tt>zext</tt> fills the high order bits of the <tt>value</tt> with zero
3070bits until it reaches the size of the destination type, <tt>ty2</tt>.</p>
3071
3072<p>When zero extending from i1, the result will always be either 0 or 1.</p>
3073
3074<h5>Example:</h5>
3075<pre>
3076 %X = zext i32 257 to i64 <i>; yields i64:257</i>
3077 %Y = zext i1 true to i32 <i>; yields i32:1</i>
3078</pre>
3079</div>
3080
3081<!-- _______________________________________________________________________ -->
3082<div class="doc_subsubsection">
3083 <a name="i_sext">'<tt>sext .. to</tt>' Instruction</a>
3084</div>
3085<div class="doc_text">
3086
3087<h5>Syntax:</h5>
3088<pre>
3089 &lt;result&gt; = sext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3090</pre>
3091
3092<h5>Overview:</h5>
3093<p>The '<tt>sext</tt>' sign extends <tt>value</tt> to the type <tt>ty2</tt>.</p>
3094
3095<h5>Arguments:</h5>
3096<p>
3097The '<tt>sext</tt>' instruction takes a value to cast, which must be of
3098<a href="#t_integer">integer</a> type, and a type to cast it to, which must
3099also be of <a href="#t_integer">integer</a> type. The bit size of the
3100<tt>value</tt> must be smaller than the bit size of the destination type,
3101<tt>ty2</tt>.</p>
3102
3103<h5>Semantics:</h5>
3104<p>
3105The '<tt>sext</tt>' instruction performs a sign extension by copying the sign
3106bit (highest order bit) of the <tt>value</tt> until it reaches the bit size of
3107the type <tt>ty2</tt>.</p>
3108
3109<p>When sign extending from i1, the extension always results in -1 or 0.</p>
3110
3111<h5>Example:</h5>
3112<pre>
3113 %X = sext i8 -1 to i16 <i>; yields i16 :65535</i>
3114 %Y = sext i1 true to i32 <i>; yields i32:-1</i>
3115</pre>
3116</div>
3117
3118<!-- _______________________________________________________________________ -->
3119<div class="doc_subsubsection">
3120 <a name="i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a>
3121</div>
3122
3123<div class="doc_text">
3124
3125<h5>Syntax:</h5>
3126
3127<pre>
3128 &lt;result&gt; = fptrunc &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3129</pre>
3130
3131<h5>Overview:</h5>
3132<p>The '<tt>fptrunc</tt>' instruction truncates <tt>value</tt> to type
3133<tt>ty2</tt>.</p>
3134
3135
3136<h5>Arguments:</h5>
3137<p>The '<tt>fptrunc</tt>' instruction takes a <a href="#t_floating">floating
3138 point</a> value to cast and a <a href="#t_floating">floating point</a> type to
3139cast it to. The size of <tt>value</tt> must be larger than the size of
3140<tt>ty2</tt>. This implies that <tt>fptrunc</tt> cannot be used to make a
3141<i>no-op cast</i>.</p>
3142
3143<h5>Semantics:</h5>
3144<p> The '<tt>fptrunc</tt>' instruction truncates a <tt>value</tt> from a larger
3145<a href="#t_floating">floating point</a> type to a smaller
3146<a href="#t_floating">floating point</a> type. If the value cannot fit within
3147the destination type, <tt>ty2</tt>, then the results are undefined.</p>
3148
3149<h5>Example:</h5>
3150<pre>
3151 %X = fptrunc double 123.0 to float <i>; yields float:123.0</i>
3152 %Y = fptrunc double 1.0E+300 to float <i>; yields undefined</i>
3153</pre>
3154</div>
3155
3156<!-- _______________________________________________________________________ -->
3157<div class="doc_subsubsection">
3158 <a name="i_fpext">'<tt>fpext .. to</tt>' Instruction</a>
3159</div>
3160<div class="doc_text">
3161
3162<h5>Syntax:</h5>
3163<pre>
3164 &lt;result&gt; = fpext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3165</pre>
3166
3167<h5>Overview:</h5>
3168<p>The '<tt>fpext</tt>' extends a floating point <tt>value</tt> to a larger
3169floating point value.</p>
3170
3171<h5>Arguments:</h5>
3172<p>The '<tt>fpext</tt>' instruction takes a
3173<a href="#t_floating">floating point</a> <tt>value</tt> to cast,
3174and a <a href="#t_floating">floating point</a> type to cast it to. The source
3175type must be smaller than the destination type.</p>
3176
3177<h5>Semantics:</h5>
3178<p>The '<tt>fpext</tt>' instruction extends the <tt>value</tt> from a smaller
3179<a href="#t_floating">floating point</a> type to a larger
3180<a href="#t_floating">floating point</a> type. The <tt>fpext</tt> cannot be
3181used to make a <i>no-op cast</i> because it always changes bits. Use
3182<tt>bitcast</tt> to make a <i>no-op cast</i> for a floating point cast.</p>
3183
3184<h5>Example:</h5>
3185<pre>
3186 %X = fpext float 3.1415 to double <i>; yields double:3.1415</i>
3187 %Y = fpext float 1.0 to float <i>; yields float:1.0 (no-op)</i>
3188</pre>
3189</div>
3190
3191<!-- _______________________________________________________________________ -->
3192<div class="doc_subsubsection">
3193 <a name="i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a>
3194</div>
3195<div class="doc_text">
3196
3197<h5>Syntax:</h5>
3198<pre>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3199 &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]3200</pre>
3201
3202<h5>Overview:</h5>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3203<p>The '<tt>fptoui</tt>' converts a floating point <tt>value</tt> to its
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3204unsigned integer equivalent of type <tt>ty2</tt>.
3205</p>
3206
3207<h5>Arguments:</h5>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3208<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]3209scalar or vector <a href="#t_floating">floating point</a> value, and a type
3210to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
3211type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
3212vector integer type with the same number of elements as <tt>ty</tt></p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3213
3214<h5>Semantics:</h5>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3215<p> The '<tt>fptoui</tt>' instruction converts its
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3216<a href="#t_floating">floating point</a> operand into the nearest (rounding
3217towards zero) unsigned integer value. If the value cannot fit in <tt>ty2</tt>,
3218the results are undefined.</p>
3219
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3220<h5>Example:</h5>
3221<pre>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3222 %X = fptoui double 123.0 to i32 <i>; yields i32:123</i>
Chris Lattner681f1e82007-09-22 03:17:52 +0000[diff] [blame]3223 %Y = fptoui float 1.0E+300 to i1 <i>; yields undefined:1</i>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3224 %X = fptoui float 1.04E+17 to i8 <i>; yields undefined:1</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3225</pre>
3226</div>
3227
3228<!-- _______________________________________________________________________ -->
3229<div class="doc_subsubsection">
3230 <a name="i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a>
3231</div>
3232<div class="doc_text">
3233
3234<h5>Syntax:</h5>
3235<pre>
3236 &lt;result&gt; = fptosi &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3237</pre>
3238
3239<h5>Overview:</h5>
3240<p>The '<tt>fptosi</tt>' instruction converts
3241<a href="#t_floating">floating point</a> <tt>value</tt> to type <tt>ty2</tt>.
3242</p>
3243
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3244<h5>Arguments:</h5>
3245<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]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>
3252<p>The '<tt>fptosi</tt>' instruction converts its
3253<a href="#t_floating">floating point</a> operand into the nearest (rounding
3254towards zero) signed 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>
3259 %X = fptosi double -123.0 to i32 <i>; yields i32:-123</i>
Chris Lattner681f1e82007-09-22 03:17:52 +0000[diff] [blame]3260 %Y = fptosi float 1.0E-247 to i1 <i>; yields undefined:1</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3261 %X = fptosi float 1.04E+17 to i8 <i>; yields undefined:1</i>
3262</pre>
3263</div>
3264
3265<!-- _______________________________________________________________________ -->
3266<div class="doc_subsubsection">
3267 <a name="i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a>
3268</div>
3269<div class="doc_text">
3270
3271<h5>Syntax:</h5>
3272<pre>
3273 &lt;result&gt; = uitofp &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>uitofp</tt>' instruction regards <tt>value</tt> as an unsigned
3278integer and converts that value to the <tt>ty2</tt> type.</p>
3279
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3280<h5>Arguments:</h5>
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]3281<p>The '<tt>uitofp</tt>' instruction takes a value to cast, which must be a
3282scalar or vector <a href="#t_integer">integer</a> value, and a type to cast it
3283to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
3284type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
3285floating point type with the same number of elements as <tt>ty</tt></p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3286
3287<h5>Semantics:</h5>
3288<p>The '<tt>uitofp</tt>' instruction interprets its operand as an unsigned
3289integer quantity and converts it to the corresponding floating point value. If
3290the value cannot fit in the floating point value, the results are undefined.</p>
3291
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3292<h5>Example:</h5>
3293<pre>
3294 %X = uitofp i32 257 to float <i>; yields float:257.0</i>
3295 %Y = uitofp i8 -1 to double <i>; yields double:255.0</i>
3296</pre>
3297</div>
3298
3299<!-- _______________________________________________________________________ -->
3300<div class="doc_subsubsection">
3301 <a name="i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a>
3302</div>
3303<div class="doc_text">
3304
3305<h5>Syntax:</h5>
3306<pre>
3307 &lt;result&gt; = sitofp &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3308</pre>
3309
3310<h5>Overview:</h5>
3311<p>The '<tt>sitofp</tt>' instruction regards <tt>value</tt> as a signed
3312integer and converts that value to the <tt>ty2</tt> type.</p>
3313
3314<h5>Arguments:</h5>
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]3315<p>The '<tt>sitofp</tt>' instruction takes a value to cast, which must be a
3316scalar or vector <a href="#t_integer">integer</a> value, and a type to cast it
3317to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
3318type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
3319floating point type with the same number of elements as <tt>ty</tt></p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3320
3321<h5>Semantics:</h5>
3322<p>The '<tt>sitofp</tt>' instruction interprets its operand as a signed
3323integer quantity and converts it to the corresponding floating point value. If
3324the value cannot fit in the floating point value, the results are undefined.</p>
3325
3326<h5>Example:</h5>
3327<pre>
3328 %X = sitofp i32 257 to float <i>; yields float:257.0</i>
3329 %Y = sitofp i8 -1 to double <i>; yields double:-1.0</i>
3330</pre>
3331</div>
3332
3333<!-- _______________________________________________________________________ -->
3334<div class="doc_subsubsection">
3335 <a name="i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a>
3336</div>
3337<div class="doc_text">
3338
3339<h5>Syntax:</h5>
3340<pre>
3341 &lt;result&gt; = ptrtoint &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3342</pre>
3343
3344<h5>Overview:</h5>
3345<p>The '<tt>ptrtoint</tt>' instruction converts the pointer <tt>value</tt> to
3346the integer type <tt>ty2</tt>.</p>
3347
3348<h5>Arguments:</h5>
3349<p>The '<tt>ptrtoint</tt>' instruction takes a <tt>value</tt> to cast, which
3350must be a <a href="#t_pointer">pointer</a> value, and a type to cast it to
3351<tt>ty2</tt>, which must be an <a href="#t_integer">integer</a> type.
3352
3353<h5>Semantics:</h5>
3354<p>The '<tt>ptrtoint</tt>' instruction converts <tt>value</tt> to integer type
3355<tt>ty2</tt> by interpreting the pointer value as an integer and either
3356truncating or zero extending that value to the size of the integer type. If
3357<tt>value</tt> is smaller than <tt>ty2</tt> then a zero extension is done. If
3358<tt>value</tt> is larger than <tt>ty2</tt> then a truncation is done. If they
3359are the same size, then nothing is done (<i>no-op cast</i>) other than a type
3360change.</p>
3361
3362<h5>Example:</h5>
3363<pre>
3364 %X = ptrtoint i32* %X to i8 <i>; yields truncation on 32-bit architecture</i>
3365 %Y = ptrtoint i32* %x to i64 <i>; yields zero extension on 32-bit architecture</i>
3366</pre>
3367</div>
3368
3369<!-- _______________________________________________________________________ -->
3370<div class="doc_subsubsection">
3371 <a name="i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a>
3372</div>
3373<div class="doc_text">
3374
3375<h5>Syntax:</h5>
3376<pre>
3377 &lt;result&gt; = inttoptr &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3378</pre>
3379
3380<h5>Overview:</h5>
3381<p>The '<tt>inttoptr</tt>' instruction converts an integer <tt>value</tt> to
3382a pointer type, <tt>ty2</tt>.</p>
3383
3384<h5>Arguments:</h5>
3385<p>The '<tt>inttoptr</tt>' instruction takes an <a href="#t_integer">integer</a>
3386value to cast, and a type to cast it to, which must be a
3387<a href="#t_pointer">pointer</a> type.
3388
3389<h5>Semantics:</h5>
3390<p>The '<tt>inttoptr</tt>' instruction converts <tt>value</tt> to type
3391<tt>ty2</tt> by applying either a zero extension or a truncation depending on
3392the size of the integer <tt>value</tt>. If <tt>value</tt> is larger than the
3393size of a pointer then a truncation is done. If <tt>value</tt> is smaller than
3394the size of a pointer then a zero extension is done. If they are the same size,
3395nothing is done (<i>no-op cast</i>).</p>
3396
3397<h5>Example:</h5>
3398<pre>
3399 %X = inttoptr i32 255 to i32* <i>; yields zero extension on 64-bit architecture</i>
3400 %X = inttoptr i32 255 to i32* <i>; yields no-op on 32-bit architecture</i>
3401 %Y = inttoptr i64 0 to i32* <i>; yields truncation on 32-bit architecture</i>
3402</pre>
3403</div>
3404
3405<!-- _______________________________________________________________________ -->
3406<div class="doc_subsubsection">
3407 <a name="i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a>
3408</div>
3409<div class="doc_text">
3410
3411<h5>Syntax:</h5>
3412<pre>
3413 &lt;result&gt; = bitcast &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3414</pre>
3415
3416<h5>Overview:</h5>
3417<p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
3418<tt>ty2</tt> without changing any bits.</p>
3419
3420<h5>Arguments:</h5>
3421<p>The '<tt>bitcast</tt>' instruction takes a value to cast, which must be
3422a first class value, and a type to cast it to, which must also be a <a
3423 href="#t_firstclass">first class</a> type. The bit sizes of <tt>value</tt>
3424and the destination type, <tt>ty2</tt>, must be identical. If the source
3425type is a pointer, the destination type must also be a pointer.</p>
3426
3427<h5>Semantics:</h5>
3428<p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
3429<tt>ty2</tt>. It is always a <i>no-op cast</i> because no bits change with
3430this conversion. The conversion is done as if the <tt>value</tt> had been
3431stored to memory and read back as type <tt>ty2</tt>. Pointer types may only be
3432converted to other pointer types with this instruction. To convert pointers to
3433other types, use the <a href="#i_inttoptr">inttoptr</a> or
3434<a href="#i_ptrtoint">ptrtoint</a> instructions first.</p>
3435
3436<h5>Example:</h5>
3437<pre>
3438 %X = bitcast i8 255 to i8 <i>; yields i8 :-1</i>
3439 %Y = bitcast i32* %x to sint* <i>; yields sint*:%x</i>
3440 %Z = bitcast <2xint> %V to i64; <i>; yields i64: %V</i>
3441</pre>
3442</div>
3443
3444<!-- ======================================================================= -->
3445<div class="doc_subsection"> <a name="otherops">Other Operations</a> </div>
3446<div class="doc_text">
3447<p>The instructions in this category are the "miscellaneous"
3448instructions, which defy better classification.</p>
3449</div>
3450
3451<!-- _______________________________________________________________________ -->
3452<div class="doc_subsubsection"><a name="i_icmp">'<tt>icmp</tt>' Instruction</a>
3453</div>
3454<div class="doc_text">
3455<h5>Syntax:</h5>
3456<pre> &lt;result&gt; = icmp &lt;cond&gt; &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {i1}:result</i>
3457</pre>
3458<h5>Overview:</h5>
3459<p>The '<tt>icmp</tt>' instruction returns a boolean value based on comparison
3460of its two integer operands.</p>
3461<h5>Arguments:</h5>
3462<p>The '<tt>icmp</tt>' instruction takes three operands. The first operand is
3463the condition code indicating the kind of comparison to perform. It is not
3464a value, just a keyword. The possible condition code are:
3465<ol>
3466 <li><tt>eq</tt>: equal</li>
3467 <li><tt>ne</tt>: not equal </li>
3468 <li><tt>ugt</tt>: unsigned greater than</li>
3469 <li><tt>uge</tt>: unsigned greater or equal</li>
3470 <li><tt>ult</tt>: unsigned less than</li>
3471 <li><tt>ule</tt>: unsigned less or equal</li>
3472 <li><tt>sgt</tt>: signed greater than</li>
3473 <li><tt>sge</tt>: signed greater or equal</li>
3474 <li><tt>slt</tt>: signed less than</li>
3475 <li><tt>sle</tt>: signed less or equal</li>
3476</ol>
3477<p>The remaining two arguments must be <a href="#t_integer">integer</a> or
3478<a href="#t_pointer">pointer</a> typed. They must also be identical types.</p>
3479<h5>Semantics:</h5>
3480<p>The '<tt>icmp</tt>' compares <tt>var1</tt> and <tt>var2</tt> according to
3481the condition code given as <tt>cond</tt>. The comparison performed always
3482yields a <a href="#t_primitive">i1</a> result, as follows:
3483<ol>
3484 <li><tt>eq</tt>: yields <tt>true</tt> if the operands are equal,
3485 <tt>false</tt> otherwise. No sign interpretation is necessary or performed.
3486 </li>
3487 <li><tt>ne</tt>: yields <tt>true</tt> if the operands are unequal,
3488 <tt>false</tt> otherwise. No sign interpretation is necessary or performed.
3489 <li><tt>ugt</tt>: interprets the operands as unsigned values and yields
3490 <tt>true</tt> if <tt>var1</tt> is greater than <tt>var2</tt>.</li>
3491 <li><tt>uge</tt>: interprets the operands as unsigned values and yields
3492 <tt>true</tt> if <tt>var1</tt> is greater than or equal to <tt>var2</tt>.</li>
3493 <li><tt>ult</tt>: interprets the operands as unsigned values and yields
3494 <tt>true</tt> if <tt>var1</tt> is less than <tt>var2</tt>.</li>
3495 <li><tt>ule</tt>: interprets the operands as unsigned values and yields
3496 <tt>true</tt> if <tt>var1</tt> is less than or equal to <tt>var2</tt>.</li>
3497 <li><tt>sgt</tt>: interprets the operands as signed values and yields
3498 <tt>true</tt> if <tt>var1</tt> is greater than <tt>var2</tt>.</li>
3499 <li><tt>sge</tt>: interprets the operands as signed values and yields
3500 <tt>true</tt> if <tt>var1</tt> is greater than or equal to <tt>var2</tt>.</li>
3501 <li><tt>slt</tt>: interprets the operands as signed values and yields
3502 <tt>true</tt> if <tt>var1</tt> is less than <tt>var2</tt>.</li>
3503 <li><tt>sle</tt>: interprets the operands as signed values and yields
3504 <tt>true</tt> if <tt>var1</tt> is less than or equal to <tt>var2</tt>.</li>
3505</ol>
3506<p>If the operands are <a href="#t_pointer">pointer</a> typed, the pointer
3507values are compared as if they were integers.</p>
3508
3509<h5>Example:</h5>
3510<pre> &lt;result&gt; = icmp eq i32 4, 5 <i>; yields: result=false</i>
3511 &lt;result&gt; = icmp ne float* %X, %X <i>; yields: result=false</i>
3512 &lt;result&gt; = icmp ult i16 4, 5 <i>; yields: result=true</i>
3513 &lt;result&gt; = icmp sgt i16 4, 5 <i>; yields: result=false</i>
3514 &lt;result&gt; = icmp ule i16 -4, 5 <i>; yields: result=false</i>
3515 &lt;result&gt; = icmp sge i16 4, 5 <i>; yields: result=false</i>
3516</pre>
3517</div>
3518
3519<!-- _______________________________________________________________________ -->
3520<div class="doc_subsubsection"><a name="i_fcmp">'<tt>fcmp</tt>' Instruction</a>
3521</div>
3522<div class="doc_text">
3523<h5>Syntax:</h5>
3524<pre> &lt;result&gt; = fcmp &lt;cond&gt; &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {i1}:result</i>
3525</pre>
3526<h5>Overview:</h5>
3527<p>The '<tt>fcmp</tt>' instruction returns a boolean value based on comparison
3528of its floating point operands.</p>
3529<h5>Arguments:</h5>
3530<p>The '<tt>fcmp</tt>' instruction takes three operands. The first operand is
3531the condition code indicating the kind of comparison to perform. It is not
3532a value, just a keyword. The possible condition code are:
3533<ol>
3534 <li><tt>false</tt>: no comparison, always returns false</li>
3535 <li><tt>oeq</tt>: ordered and equal</li>
3536 <li><tt>ogt</tt>: ordered and greater than </li>
3537 <li><tt>oge</tt>: ordered and greater than or equal</li>
3538 <li><tt>olt</tt>: ordered and less than </li>
3539 <li><tt>ole</tt>: ordered and less than or equal</li>
3540 <li><tt>one</tt>: ordered and not equal</li>
3541 <li><tt>ord</tt>: ordered (no nans)</li>
3542 <li><tt>ueq</tt>: unordered or equal</li>
3543 <li><tt>ugt</tt>: unordered or greater than </li>
3544 <li><tt>uge</tt>: unordered or greater than or equal</li>
3545 <li><tt>ult</tt>: unordered or less than </li>
3546 <li><tt>ule</tt>: unordered or less than or equal</li>
3547 <li><tt>une</tt>: unordered or not equal</li>
3548 <li><tt>uno</tt>: unordered (either nans)</li>
3549 <li><tt>true</tt>: no comparison, always returns true</li>
3550</ol>
3551<p><i>Ordered</i> means that neither operand is a QNAN while
3552<i>unordered</i> means that either operand may be a QNAN.</p>
3553<p>The <tt>val1</tt> and <tt>val2</tt> arguments must be
3554<a href="#t_floating">floating point</a> typed. They must have identical
3555types.</p>
3556<h5>Semantics:</h5>
3557<p>The '<tt>fcmp</tt>' compares <tt>var1</tt> and <tt>var2</tt> according to
3558the condition code given as <tt>cond</tt>. The comparison performed always
3559yields a <a href="#t_primitive">i1</a> result, as follows:
3560<ol>
3561 <li><tt>false</tt>: always yields <tt>false</tt>, regardless of operands.</li>
3562 <li><tt>oeq</tt>: yields <tt>true</tt> if both operands are not a QNAN and
3563 <tt>var1</tt> is equal to <tt>var2</tt>.</li>
3564 <li><tt>ogt</tt>: yields <tt>true</tt> if both operands are not a QNAN and
3565 <tt>var1</tt> is greather than <tt>var2</tt>.</li>
3566 <li><tt>oge</tt>: yields <tt>true</tt> if both operands are not a QNAN and
3567 <tt>var1</tt> is greater than or equal to <tt>var2</tt>.</li>
3568 <li><tt>olt</tt>: yields <tt>true</tt> if both operands are not a QNAN and
3569 <tt>var1</tt> is less than <tt>var2</tt>.</li>
3570 <li><tt>ole</tt>: yields <tt>true</tt> if both operands are not a QNAN and
3571 <tt>var1</tt> is less than or equal to <tt>var2</tt>.</li>
3572 <li><tt>one</tt>: yields <tt>true</tt> if both operands are not a QNAN and
3573 <tt>var1</tt> is not equal to <tt>var2</tt>.</li>
3574 <li><tt>ord</tt>: yields <tt>true</tt> if both operands are not a QNAN.</li>
3575 <li><tt>ueq</tt>: yields <tt>true</tt> if either operand is a QNAN or
3576 <tt>var1</tt> is equal to <tt>var2</tt>.</li>
3577 <li><tt>ugt</tt>: yields <tt>true</tt> if either operand is a QNAN or
3578 <tt>var1</tt> is greater than <tt>var2</tt>.</li>
3579 <li><tt>uge</tt>: yields <tt>true</tt> if either operand is a QNAN or
3580 <tt>var1</tt> is greater than or equal to <tt>var2</tt>.</li>
3581 <li><tt>ult</tt>: yields <tt>true</tt> if either operand is a QNAN or
3582 <tt>var1</tt> is less than <tt>var2</tt>.</li>
3583 <li><tt>ule</tt>: yields <tt>true</tt> if either operand is a QNAN or
3584 <tt>var1</tt> is less than or equal to <tt>var2</tt>.</li>
3585 <li><tt>une</tt>: yields <tt>true</tt> if either operand is a QNAN or
3586 <tt>var1</tt> is not equal to <tt>var2</tt>.</li>
3587 <li><tt>uno</tt>: yields <tt>true</tt> if either operand is a QNAN.</li>
3588 <li><tt>true</tt>: always yields <tt>true</tt>, regardless of operands.</li>
3589</ol>
3590
3591<h5>Example:</h5>
3592<pre> &lt;result&gt; = fcmp oeq float 4.0, 5.0 <i>; yields: result=false</i>
3593 &lt;result&gt; = icmp one float 4.0, 5.0 <i>; yields: result=true</i>
3594 &lt;result&gt; = icmp olt float 4.0, 5.0 <i>; yields: result=true</i>
3595 &lt;result&gt; = icmp ueq double 1.0, 2.0 <i>; yields: result=false</i>
3596</pre>
3597</div>
3598
3599<!-- _______________________________________________________________________ -->
3600<div class="doc_subsubsection"> <a name="i_phi">'<tt>phi</tt>'
3601Instruction</a> </div>
3602<div class="doc_text">
3603<h5>Syntax:</h5>
3604<pre> &lt;result&gt; = phi &lt;ty&gt; [ &lt;val0&gt;, &lt;label0&gt;], ...<br></pre>
3605<h5>Overview:</h5>
3606<p>The '<tt>phi</tt>' instruction is used to implement the &#966; node in
3607the SSA graph representing the function.</p>
3608<h5>Arguments:</h5>
3609<p>The type of the incoming values is specified with the first type
3610field. After this, the '<tt>phi</tt>' instruction takes a list of pairs
3611as arguments, with one pair for each predecessor basic block of the
3612current block. Only values of <a href="#t_firstclass">first class</a>
3613type may be used as the value arguments to the PHI node. Only labels
3614may be used as the label arguments.</p>
3615<p>There must be no non-phi instructions between the start of a basic
3616block and the PHI instructions: i.e. PHI instructions must be first in
3617a basic block.</p>
3618<h5>Semantics:</h5>
3619<p>At runtime, the '<tt>phi</tt>' instruction logically takes on the value
3620specified by the pair corresponding to the predecessor basic block that executed
3621just prior to the current block.</p>
3622<h5>Example:</h5>
3623<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>
3624</div>
3625
3626<!-- _______________________________________________________________________ -->
3627<div class="doc_subsubsection">
3628 <a name="i_select">'<tt>select</tt>' Instruction</a>
3629</div>
3630
3631<div class="doc_text">
3632
3633<h5>Syntax:</h5>
3634
3635<pre>
3636 &lt;result&gt; = select i1 &lt;cond&gt;, &lt;ty&gt; &lt;val1&gt;, &lt;ty&gt; &lt;val2&gt; <i>; yields ty</i>
3637</pre>
3638
3639<h5>Overview:</h5>
3640
3641<p>
3642The '<tt>select</tt>' instruction is used to choose one value based on a
3643condition, without branching.
3644</p>
3645
3646
3647<h5>Arguments:</h5>
3648
3649<p>
3650The '<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.
3651</p>
3652
3653<h5>Semantics:</h5>
3654
3655<p>
3656If the boolean condition evaluates to true, the instruction returns the first
3657value argument; otherwise, it returns the second value argument.
3658</p>
3659
3660<h5>Example:</h5>
3661
3662<pre>
3663 %X = select i1 true, i8 17, i8 42 <i>; yields i8:17</i>
3664</pre>
3665</div>
3666
3667
3668<!-- _______________________________________________________________________ -->
3669<div class="doc_subsubsection">
3670 <a name="i_call">'<tt>call</tt>' Instruction</a>
3671</div>
3672
3673<div class="doc_text">
3674
3675<h5>Syntax:</h5>
3676<pre>
Nick Lewycky93082fc2007-09-08 13:57:50 +0000[diff] [blame]3677 &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]3678</pre>
3679
3680<h5>Overview:</h5>
3681
3682<p>The '<tt>call</tt>' instruction represents a simple function call.</p>
3683
3684<h5>Arguments:</h5>
3685
3686<p>This instruction requires several arguments:</p>
3687
3688<ol>
3689 <li>
3690 <p>The optional "tail" marker indicates whether the callee function accesses
3691 any allocas or varargs in the caller. If the "tail" marker is present, the
3692 function call is eligible for tail call optimization. Note that calls may
3693 be marked "tail" even if they do not occur before a <a
3694 href="#i_ret"><tt>ret</tt></a> instruction.
3695 </li>
3696 <li>
3697 <p>The optional "cconv" marker indicates which <a href="#callingconv">calling
3698 convention</a> the call should use. If none is specified, the call defaults
3699 to using C calling conventions.
3700 </li>
3701 <li>
Nick Lewycky93082fc2007-09-08 13:57:50 +0000[diff] [blame]3702 <p>'<tt>ty</tt>': the type of the call instruction itself which is also
3703 the type of the return value. Functions that return no value are marked
3704 <tt><a href="#t_void">void</a></tt>.</p>
3705 </li>
3706 <li>
3707 <p>'<tt>fnty</tt>': shall be the signature of the pointer to function
3708 value being invoked. The argument types must match the types implied by
3709 this signature. This type can be omitted if the function is not varargs
3710 and if the function type does not return a pointer to a function.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3711 </li>
3712 <li>
3713 <p>'<tt>fnptrval</tt>': An LLVM value containing a pointer to a function to
3714 be invoked. In most cases, this is a direct function invocation, but
3715 indirect <tt>call</tt>s are just as possible, calling an arbitrary pointer
3716 to function value.</p>
3717 </li>
3718 <li>
3719 <p>'<tt>function args</tt>': argument list whose types match the
3720 function signature argument types. All arguments must be of
3721 <a href="#t_firstclass">first class</a> type. If the function signature
3722 indicates the function accepts a variable number of arguments, the extra
3723 arguments can be specified.</p>
3724 </li>
3725</ol>
3726
3727<h5>Semantics:</h5>
3728
3729<p>The '<tt>call</tt>' instruction is used to cause control flow to
3730transfer to a specified function, with its incoming arguments bound to
3731the specified values. Upon a '<tt><a href="#i_ret">ret</a></tt>'
3732instruction in the called function, control flow continues with the
3733instruction after the function call, and the return value of the
3734function is bound to the result argument. This is a simpler case of
3735the <a href="#i_invoke">invoke</a> instruction.</p>
3736
3737<h5>Example:</h5>
3738
3739<pre>
Nick Lewycky93082fc2007-09-08 13:57:50 +0000[diff] [blame]3740 %retval = call i32 @test(i32 %argc)
3741 call i32 (i8 *, ...)* @printf(i8 * %msg, i32 12, i8 42);
3742 %X = tail call i32 @foo()
3743 %Y = tail call <a href="#callingconv">fastcc</a> i32 @foo()
3744 %Z = call void %foo(i8 97 signext)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3745</pre>
3746
3747</div>
3748
3749<!-- _______________________________________________________________________ -->
3750<div class="doc_subsubsection">
3751 <a name="i_va_arg">'<tt>va_arg</tt>' Instruction</a>
3752</div>
3753
3754<div class="doc_text">
3755
3756<h5>Syntax:</h5>
3757
3758<pre>
3759 &lt;resultval&gt; = va_arg &lt;va_list*&gt; &lt;arglist&gt;, &lt;argty&gt;
3760</pre>
3761
3762<h5>Overview:</h5>
3763
3764<p>The '<tt>va_arg</tt>' instruction is used to access arguments passed through
3765the "variable argument" area of a function call. It is used to implement the
3766<tt>va_arg</tt> macro in C.</p>
3767
3768<h5>Arguments:</h5>
3769
3770<p>This instruction takes a <tt>va_list*</tt> value and the type of
3771the argument. It returns a value of the specified argument type and
3772increments the <tt>va_list</tt> to point to the next argument. The
3773actual type of <tt>va_list</tt> is target specific.</p>
3774
3775<h5>Semantics:</h5>
3776
3777<p>The '<tt>va_arg</tt>' instruction loads an argument of the specified
3778type from the specified <tt>va_list</tt> and causes the
3779<tt>va_list</tt> to point to the next argument. For more information,
3780see the variable argument handling <a href="#int_varargs">Intrinsic
3781Functions</a>.</p>
3782
3783<p>It is legal for this instruction to be called in a function which does not
3784take a variable number of arguments, for example, the <tt>vfprintf</tt>
3785function.</p>
3786
3787<p><tt>va_arg</tt> is an LLVM instruction instead of an <a
3788href="#intrinsics">intrinsic function</a> because it takes a type as an
3789argument.</p>
3790
3791<h5>Example:</h5>
3792
3793<p>See the <a href="#int_varargs">variable argument processing</a> section.</p>
3794
3795</div>
3796
3797<!-- *********************************************************************** -->
3798<div class="doc_section"> <a name="intrinsics">Intrinsic Functions</a> </div>
3799<!-- *********************************************************************** -->
3800
3801<div class="doc_text">
3802
3803<p>LLVM supports the notion of an "intrinsic function". These functions have
3804well known names and semantics and are required to follow certain restrictions.
3805Overall, these intrinsics represent an extension mechanism for the LLVM
3806language that does not require changing all of the transformations in LLVM when
3807adding to the language (or the bitcode reader/writer, the parser, etc...).</p>
3808
3809<p>Intrinsic function names must all start with an "<tt>llvm.</tt>" prefix. This
3810prefix is reserved in LLVM for intrinsic names; thus, function names may not
3811begin with this prefix. Intrinsic functions must always be external functions:
3812you cannot define the body of intrinsic functions. Intrinsic functions may
3813only be used in call or invoke instructions: it is illegal to take the address
3814of an intrinsic function. Additionally, because intrinsic functions are part
3815of the LLVM language, it is required if any are added that they be documented
3816here.</p>
3817
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]3818<p>Some intrinsic functions can be overloaded, i.e., the intrinsic represents
3819a family of functions that perform the same operation but on different data
3820types. Because LLVM can represent over 8 million different integer types,
3821overloading is used commonly to allow an intrinsic function to operate on any
3822integer type. One or more of the argument types or the result type can be
3823overloaded to accept any integer type. Argument types may also be defined as
3824exactly matching a previous argument's type or the result type. This allows an
3825intrinsic function which accepts multiple arguments, but needs all of them to
3826be of the same type, to only be overloaded with respect to a single argument or
3827the result.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3828
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]3829<p>Overloaded intrinsics will have the names of its overloaded argument types
3830encoded into its function name, each preceded by a period. Only those types
3831which are overloaded result in a name suffix. Arguments whose type is matched
3832against another type do not. For example, the <tt>llvm.ctpop</tt> function can
3833take an integer of any width and returns an integer of exactly the same integer
3834width. This leads to a family of functions such as
3835<tt>i8 @llvm.ctpop.i8(i8 %val)</tt> and <tt>i29 @llvm.ctpop.i29(i29 %val)</tt>.
3836Only one type, the return type, is overloaded, and only one type suffix is
3837required. Because the argument's type is matched against the return type, it
3838does not require its own name suffix.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3839
3840<p>To learn how to add an intrinsic function, please see the
3841<a href="ExtendingLLVM.html">Extending LLVM Guide</a>.
3842</p>
3843
3844</div>
3845
3846<!-- ======================================================================= -->
3847<div class="doc_subsection">
3848 <a name="int_varargs">Variable Argument Handling Intrinsics</a>
3849</div>
3850
3851<div class="doc_text">
3852
3853<p>Variable argument support is defined in LLVM with the <a
3854 href="#i_va_arg"><tt>va_arg</tt></a> instruction and these three
3855intrinsic functions. These functions are related to the similarly
3856named macros defined in the <tt>&lt;stdarg.h&gt;</tt> header file.</p>
3857
3858<p>All of these functions operate on arguments that use a
3859target-specific value type "<tt>va_list</tt>". The LLVM assembly
3860language reference manual does not define what this type is, so all
3861transformations should be prepared to handle these functions regardless of
3862the type used.</p>
3863
3864<p>This example shows how the <a href="#i_va_arg"><tt>va_arg</tt></a>
3865instruction and the variable argument handling intrinsic functions are
3866used.</p>
3867
3868<div class="doc_code">
3869<pre>
3870define i32 @test(i32 %X, ...) {
3871 ; Initialize variable argument processing
3872 %ap = alloca i8*
3873 %ap2 = bitcast i8** %ap to i8*
3874 call void @llvm.va_start(i8* %ap2)
3875
3876 ; Read a single integer argument
3877 %tmp = va_arg i8** %ap, i32
3878
3879 ; Demonstrate usage of llvm.va_copy and llvm.va_end
3880 %aq = alloca i8*
3881 %aq2 = bitcast i8** %aq to i8*
3882 call void @llvm.va_copy(i8* %aq2, i8* %ap2)
3883 call void @llvm.va_end(i8* %aq2)
3884
3885 ; Stop processing of arguments.
3886 call void @llvm.va_end(i8* %ap2)
3887 ret i32 %tmp
3888}
3889
3890declare void @llvm.va_start(i8*)
3891declare void @llvm.va_copy(i8*, i8*)
3892declare void @llvm.va_end(i8*)
3893</pre>
3894</div>
3895
3896</div>
3897
3898<!-- _______________________________________________________________________ -->
3899<div class="doc_subsubsection">
3900 <a name="int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a>
3901</div>
3902
3903
3904<div class="doc_text">
3905<h5>Syntax:</h5>
3906<pre> declare void %llvm.va_start(i8* &lt;arglist&gt;)<br></pre>
3907<h5>Overview:</h5>
3908<P>The '<tt>llvm.va_start</tt>' intrinsic initializes
3909<tt>*&lt;arglist&gt;</tt> for subsequent use by <tt><a
3910href="#i_va_arg">va_arg</a></tt>.</p>
3911
3912<h5>Arguments:</h5>
3913
3914<P>The argument is a pointer to a <tt>va_list</tt> element to initialize.</p>
3915
3916<h5>Semantics:</h5>
3917
3918<P>The '<tt>llvm.va_start</tt>' intrinsic works just like the <tt>va_start</tt>
3919macro available in C. In a target-dependent way, it initializes the
3920<tt>va_list</tt> element to which the argument points, so that the next call to
3921<tt>va_arg</tt> will produce the first variable argument passed to the function.
3922Unlike the C <tt>va_start</tt> macro, this intrinsic does not need to know the
3923last argument of the function as the compiler can figure that out.</p>
3924
3925</div>
3926
3927<!-- _______________________________________________________________________ -->
3928<div class="doc_subsubsection">
3929 <a name="int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a>
3930</div>
3931
3932<div class="doc_text">
3933<h5>Syntax:</h5>
3934<pre> declare void @llvm.va_end(i8* &lt;arglist&gt;)<br></pre>
3935<h5>Overview:</h5>
3936
3937<p>The '<tt>llvm.va_end</tt>' intrinsic destroys <tt>*&lt;arglist&gt;</tt>,
3938which has been initialized previously with <tt><a href="#int_va_start">llvm.va_start</a></tt>
3939or <tt><a href="#i_va_copy">llvm.va_copy</a></tt>.</p>
3940
3941<h5>Arguments:</h5>
3942
3943<p>The argument is a pointer to a <tt>va_list</tt> to destroy.</p>
3944
3945<h5>Semantics:</h5>
3946
3947<p>The '<tt>llvm.va_end</tt>' intrinsic works just like the <tt>va_end</tt>
3948macro available in C. In a target-dependent way, it destroys the
3949<tt>va_list</tt> element to which the argument points. Calls to <a
3950href="#int_va_start"><tt>llvm.va_start</tt></a> and <a href="#int_va_copy">
3951<tt>llvm.va_copy</tt></a> must be matched exactly with calls to
3952<tt>llvm.va_end</tt>.</p>
3953
3954</div>
3955
3956<!-- _______________________________________________________________________ -->
3957<div class="doc_subsubsection">
3958 <a name="int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a>
3959</div>
3960
3961<div class="doc_text">
3962
3963<h5>Syntax:</h5>
3964
3965<pre>
3966 declare void @llvm.va_copy(i8* &lt;destarglist&gt;, i8* &lt;srcarglist&gt;)
3967</pre>
3968
3969<h5>Overview:</h5>
3970
3971<p>The '<tt>llvm.va_copy</tt>' intrinsic copies the current argument position
3972from the source argument list to the destination argument list.</p>
3973
3974<h5>Arguments:</h5>
3975
3976<p>The first argument is a pointer to a <tt>va_list</tt> element to initialize.
3977The second argument is a pointer to a <tt>va_list</tt> element to copy from.</p>
3978
3979
3980<h5>Semantics:</h5>
3981
3982<p>The '<tt>llvm.va_copy</tt>' intrinsic works just like the <tt>va_copy</tt>
3983macro available in C. In a target-dependent way, it copies the source
3984<tt>va_list</tt> element into the destination <tt>va_list</tt> element. This
3985intrinsic is necessary because the <tt><a href="#int_va_start">
3986llvm.va_start</a></tt> intrinsic may be arbitrarily complex and require, for
3987example, memory allocation.</p>
3988
3989</div>
3990
3991<!-- ======================================================================= -->
3992<div class="doc_subsection">
3993 <a name="int_gc">Accurate Garbage Collection Intrinsics</a>
3994</div>
3995
3996<div class="doc_text">
3997
3998<p>
3999LLVM support for <a href="GarbageCollection.html">Accurate Garbage
4000Collection</a> requires the implementation and generation of these intrinsics.
4001These intrinsics allow identification of <a href="#int_gcroot">GC roots on the
4002stack</a>, as well as garbage collector implementations that require <a
4003href="#int_gcread">read</a> and <a href="#int_gcwrite">write</a> barriers.
4004Front-ends for type-safe garbage collected languages should generate these
4005intrinsics to make use of the LLVM garbage collectors. For more details, see <a
4006href="GarbageCollection.html">Accurate Garbage Collection with LLVM</a>.
4007</p>
Christopher Lambcfe00962007-12-17 01:00:21 +0000[diff] [blame]4008
4009<p>The garbage collection intrinsics only operate on objects in the generic
4010 address space (address space zero).</p>
4011
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4012</div>
4013
4014<!-- _______________________________________________________________________ -->
4015<div class="doc_subsubsection">
4016 <a name="int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a>
4017</div>
4018
4019<div class="doc_text">
4020
4021<h5>Syntax:</h5>
4022
4023<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4024 declare void @llvm.gcroot(i8** %ptrloc, i8* %metadata)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4025</pre>
4026
4027<h5>Overview:</h5>
4028
4029<p>The '<tt>llvm.gcroot</tt>' intrinsic declares the existence of a GC root to
4030the code generator, and allows some metadata to be associated with it.</p>
4031
4032<h5>Arguments:</h5>
4033
4034<p>The first argument specifies the address of a stack object that contains the
4035root pointer. The second pointer (which must be either a constant or a global
4036value address) contains the meta-data to be associated with the root.</p>
4037
4038<h5>Semantics:</h5>
4039
4040<p>At runtime, a call to this intrinsics stores a null pointer into the "ptrloc"
4041location. At compile-time, the code generator generates information to allow
Gordon Henriksen40393542007-12-25 02:31:26 +0000[diff] [blame]4042the runtime to find the pointer at GC safe points. The '<tt>llvm.gcroot</tt>'
4043intrinsic may only be used in a function which <a href="#gc">specifies a GC
4044algorithm</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4045
4046</div>
4047
4048
4049<!-- _______________________________________________________________________ -->
4050<div class="doc_subsubsection">
4051 <a name="int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a>
4052</div>
4053
4054<div class="doc_text">
4055
4056<h5>Syntax:</h5>
4057
4058<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4059 declare i8* @llvm.gcread(i8* %ObjPtr, i8** %Ptr)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4060</pre>
4061
4062<h5>Overview:</h5>
4063
4064<p>The '<tt>llvm.gcread</tt>' intrinsic identifies reads of references from heap
4065locations, allowing garbage collector implementations that require read
4066barriers.</p>
4067
4068<h5>Arguments:</h5>
4069
4070<p>The second argument is the address to read from, which should be an address
4071allocated from the garbage collector. The first object is a pointer to the
4072start of the referenced object, if needed by the language runtime (otherwise
4073null).</p>
4074
4075<h5>Semantics:</h5>
4076
4077<p>The '<tt>llvm.gcread</tt>' intrinsic has the same semantics as a load
4078instruction, but may be replaced with substantially more complex code by the
Gordon Henriksen40393542007-12-25 02:31:26 +0000[diff] [blame]4079garbage collector runtime, as needed. The '<tt>llvm.gcread</tt>' intrinsic
4080may 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_gcwrite">'<tt>llvm.gcwrite</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 void @llvm.gcwrite(i8* %P1, i8* %Obj, i8** %P2)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4097</pre>
4098
4099<h5>Overview:</h5>
4100
4101<p>The '<tt>llvm.gcwrite</tt>' intrinsic identifies writes of references to heap
4102locations, allowing garbage collector implementations that require write
4103barriers (such as generational or reference counting collectors).</p>
4104
4105<h5>Arguments:</h5>
4106
4107<p>The first argument is the reference to store, the second is the start of the
4108object to store it to, and the third is the address of the field of Obj to
4109store to. If the runtime does not require a pointer to the object, Obj may be
4110null.</p>
4111
4112<h5>Semantics:</h5>
4113
4114<p>The '<tt>llvm.gcwrite</tt>' intrinsic has the same semantics as a store
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.gcwrite</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<!-- ======================================================================= -->
4125<div class="doc_subsection">
4126 <a name="int_codegen">Code Generator Intrinsics</a>
4127</div>
4128
4129<div class="doc_text">
4130<p>
4131These intrinsics are provided by LLVM to expose special features that may only
4132be implemented with code generator support.
4133</p>
4134
4135</div>
4136
4137<!-- _______________________________________________________________________ -->
4138<div class="doc_subsubsection">
4139 <a name="int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a>
4140</div>
4141
4142<div class="doc_text">
4143
4144<h5>Syntax:</h5>
4145<pre>
4146 declare i8 *@llvm.returnaddress(i32 &lt;level&gt;)
4147</pre>
4148
4149<h5>Overview:</h5>
4150
4151<p>
4152The '<tt>llvm.returnaddress</tt>' intrinsic attempts to compute a
4153target-specific value indicating the return address of the current function
4154or one of its callers.
4155</p>
4156
4157<h5>Arguments:</h5>
4158
4159<p>
4160The argument to this intrinsic indicates which function to return the address
4161for. Zero indicates the calling function, one indicates its caller, etc. The
4162argument is <b>required</b> to be a constant integer value.
4163</p>
4164
4165<h5>Semantics:</h5>
4166
4167<p>
4168The '<tt>llvm.returnaddress</tt>' intrinsic either returns a pointer indicating
4169the return address of the specified call frame, or zero if it cannot be
4170identified. The value returned by this intrinsic is likely to be incorrect or 0
4171for arguments other than zero, so it should only be used for debugging purposes.
4172</p>
4173
4174<p>
4175Note that calling this intrinsic does not prevent function inlining or other
4176aggressive transformations, so the value returned may not be that of the obvious
4177source-language caller.
4178</p>
4179</div>
4180
4181
4182<!-- _______________________________________________________________________ -->
4183<div class="doc_subsubsection">
4184 <a name="int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a>
4185</div>
4186
4187<div class="doc_text">
4188
4189<h5>Syntax:</h5>
4190<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4191 declare i8 *@llvm.frameaddress(i32 &lt;level&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4192</pre>
4193
4194<h5>Overview:</h5>
4195
4196<p>
4197The '<tt>llvm.frameaddress</tt>' intrinsic attempts to return the
4198target-specific frame pointer value for the specified stack frame.
4199</p>
4200
4201<h5>Arguments:</h5>
4202
4203<p>
4204The argument to this intrinsic indicates which function to return the frame
4205pointer for. Zero indicates the calling function, one indicates its caller,
4206etc. The argument is <b>required</b> to be a constant integer value.
4207</p>
4208
4209<h5>Semantics:</h5>
4210
4211<p>
4212The '<tt>llvm.frameaddress</tt>' intrinsic either returns a pointer indicating
4213the frame address of the specified call frame, or zero if it cannot be
4214identified. The value returned by this intrinsic is likely to be incorrect or 0
4215for arguments other than zero, so it should only be used for debugging purposes.
4216</p>
4217
4218<p>
4219Note that calling this intrinsic does not prevent function inlining or other
4220aggressive transformations, so the value returned may not be that of the obvious
4221source-language caller.
4222</p>
4223</div>
4224
4225<!-- _______________________________________________________________________ -->
4226<div class="doc_subsubsection">
4227 <a name="int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a>
4228</div>
4229
4230<div class="doc_text">
4231
4232<h5>Syntax:</h5>
4233<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4234 declare i8 *@llvm.stacksave()
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4235</pre>
4236
4237<h5>Overview:</h5>
4238
4239<p>
4240The '<tt>llvm.stacksave</tt>' intrinsic is used to remember the current state of
4241the function stack, for use with <a href="#int_stackrestore">
4242<tt>llvm.stackrestore</tt></a>. This is useful for implementing language
4243features like scoped automatic variable sized arrays in C99.
4244</p>
4245
4246<h5>Semantics:</h5>
4247
4248<p>
4249This intrinsic returns a opaque pointer value that can be passed to <a
4250href="#int_stackrestore"><tt>llvm.stackrestore</tt></a>. When an
4251<tt>llvm.stackrestore</tt> intrinsic is executed with a value saved from
4252<tt>llvm.stacksave</tt>, it effectively restores the state of the stack to the
4253state it was in when the <tt>llvm.stacksave</tt> intrinsic executed. In
4254practice, this pops any <a href="#i_alloca">alloca</a> blocks from the stack
4255that were allocated after the <tt>llvm.stacksave</tt> was executed.
4256</p>
4257
4258</div>
4259
4260<!-- _______________________________________________________________________ -->
4261<div class="doc_subsubsection">
4262 <a name="int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a>
4263</div>
4264
4265<div class="doc_text">
4266
4267<h5>Syntax:</h5>
4268<pre>
4269 declare void @llvm.stackrestore(i8 * %ptr)
4270</pre>
4271
4272<h5>Overview:</h5>
4273
4274<p>
4275The '<tt>llvm.stackrestore</tt>' intrinsic is used to restore the state of
4276the function stack to the state it was in when the corresponding <a
4277href="#int_stacksave"><tt>llvm.stacksave</tt></a> intrinsic executed. This is
4278useful for implementing language features like scoped automatic variable sized
4279arrays in C99.
4280</p>
4281
4282<h5>Semantics:</h5>
4283
4284<p>
4285See the description for <a href="#int_stacksave"><tt>llvm.stacksave</tt></a>.
4286</p>
4287
4288</div>
4289
4290
4291<!-- _______________________________________________________________________ -->
4292<div class="doc_subsubsection">
4293 <a name="int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a>
4294</div>
4295
4296<div class="doc_text">
4297
4298<h5>Syntax:</h5>
4299<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4300 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]4301</pre>
4302
4303<h5>Overview:</h5>
4304
4305
4306<p>
4307The '<tt>llvm.prefetch</tt>' intrinsic is a hint to the code generator to insert
4308a prefetch instruction if supported; otherwise, it is a noop. Prefetches have
4309no
4310effect on the behavior of the program but can change its performance
4311characteristics.
4312</p>
4313
4314<h5>Arguments:</h5>
4315
4316<p>
4317<tt>address</tt> is the address to be prefetched, <tt>rw</tt> is the specifier
4318determining if the fetch should be for a read (0) or write (1), and
4319<tt>locality</tt> is a temporal locality specifier ranging from (0) - no
4320locality, to (3) - extremely local keep in cache. The <tt>rw</tt> and
4321<tt>locality</tt> arguments must be constant integers.
4322</p>
4323
4324<h5>Semantics:</h5>
4325
4326<p>
4327This intrinsic does not modify the behavior of the program. In particular,
4328prefetches cannot trap and do not produce a value. On targets that support this
4329intrinsic, the prefetch can provide hints to the processor cache for better
4330performance.
4331</p>
4332
4333</div>
4334
4335<!-- _______________________________________________________________________ -->
4336<div class="doc_subsubsection">
4337 <a name="int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a>
4338</div>
4339
4340<div class="doc_text">
4341
4342<h5>Syntax:</h5>
4343<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4344 declare void @llvm.pcmarker(i32 &lt;id&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4345</pre>
4346
4347<h5>Overview:</h5>
4348
4349
4350<p>
4351The '<tt>llvm.pcmarker</tt>' intrinsic is a method to export a Program Counter
4352(PC) in a region of
4353code to simulators and other tools. The method is target specific, but it is
4354expected that the marker will use exported symbols to transmit the PC of the marker.
4355The marker makes no guarantees that it will remain with any specific instruction
4356after optimizations. It is possible that the presence of a marker will inhibit
4357optimizations. The intended use is to be inserted after optimizations to allow
4358correlations of simulation runs.
4359</p>
4360
4361<h5>Arguments:</h5>
4362
4363<p>
4364<tt>id</tt> is a numerical id identifying the marker.
4365</p>
4366
4367<h5>Semantics:</h5>
4368
4369<p>
4370This intrinsic does not modify the behavior of the program. Backends that do not
4371support this intrinisic may ignore it.
4372</p>
4373
4374</div>
4375
4376<!-- _______________________________________________________________________ -->
4377<div class="doc_subsubsection">
4378 <a name="int_readcyclecounter">'<tt>llvm.readcyclecounter</tt>' Intrinsic</a>
4379</div>
4380
4381<div class="doc_text">
4382
4383<h5>Syntax:</h5>
4384<pre>
4385 declare i64 @llvm.readcyclecounter( )
4386</pre>
4387
4388<h5>Overview:</h5>
4389
4390
4391<p>
4392The '<tt>llvm.readcyclecounter</tt>' intrinsic provides access to the cycle
4393counter register (or similar low latency, high accuracy clocks) on those targets
4394that support it. On X86, it should map to RDTSC. On Alpha, it should map to RPCC.
4395As the backing counters overflow quickly (on the order of 9 seconds on alpha), this
4396should only be used for small timings.
4397</p>
4398
4399<h5>Semantics:</h5>
4400
4401<p>
4402When directly supported, reading the cycle counter should not modify any memory.
4403Implementations are allowed to either return a application specific value or a
4404system wide value. On backends without support, this is lowered to a constant 0.
4405</p>
4406
4407</div>
4408
4409<!-- ======================================================================= -->
4410<div class="doc_subsection">
4411 <a name="int_libc">Standard C Library Intrinsics</a>
4412</div>
4413
4414<div class="doc_text">
4415<p>
4416LLVM provides intrinsics for a few important standard C library functions.
4417These intrinsics allow source-language front-ends to pass information about the
4418alignment of the pointer arguments to the code generator, providing opportunity
4419for more efficient code generation.
4420</p>
4421
4422</div>
4423
4424<!-- _______________________________________________________________________ -->
4425<div class="doc_subsubsection">
4426 <a name="int_memcpy">'<tt>llvm.memcpy</tt>' Intrinsic</a>
4427</div>
4428
4429<div class="doc_text">
4430
4431<h5>Syntax:</h5>
4432<pre>
4433 declare void @llvm.memcpy.i32(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
4434 i32 &lt;len&gt;, i32 &lt;align&gt;)
4435 declare void @llvm.memcpy.i64(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
4436 i64 &lt;len&gt;, i32 &lt;align&gt;)
4437</pre>
4438
4439<h5>Overview:</h5>
4440
4441<p>
4442The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the source
4443location to the destination location.
4444</p>
4445
4446<p>
4447Note that, unlike the standard libc function, the <tt>llvm.memcpy.*</tt>
4448intrinsics do not return a value, and takes an extra alignment argument.
4449</p>
4450
4451<h5>Arguments:</h5>
4452
4453<p>
4454The first argument is a pointer to the destination, the second is a pointer to
4455the source. The third argument is an integer argument
4456specifying the number of bytes to copy, and the fourth argument is the alignment
4457of the source and destination locations.
4458</p>
4459
4460<p>
4461If the call to this intrinisic has an alignment value that is not 0 or 1, then
4462the caller guarantees that both the source and destination pointers are aligned
4463to that boundary.
4464</p>
4465
4466<h5>Semantics:</h5>
4467
4468<p>
4469The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the source
4470location to the destination location, which are not allowed to overlap. It
4471copies "len" bytes of memory over. If the argument is known to be aligned to
4472some boundary, this can be specified as the fourth argument, otherwise it should
4473be set to 0 or 1.
4474</p>
4475</div>
4476
4477
4478<!-- _______________________________________________________________________ -->
4479<div class="doc_subsubsection">
4480 <a name="int_memmove">'<tt>llvm.memmove</tt>' Intrinsic</a>
4481</div>
4482
4483<div class="doc_text">
4484
4485<h5>Syntax:</h5>
4486<pre>
4487 declare void @llvm.memmove.i32(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
4488 i32 &lt;len&gt;, i32 &lt;align&gt;)
4489 declare void @llvm.memmove.i64(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
4490 i64 &lt;len&gt;, i32 &lt;align&gt;)
4491</pre>
4492
4493<h5>Overview:</h5>
4494
4495<p>
4496The '<tt>llvm.memmove.*</tt>' intrinsics move a block of memory from the source
4497location to the destination location. It is similar to the
4498'<tt>llvm.memcmp</tt>' intrinsic but allows the two memory locations to overlap.
4499</p>
4500
4501<p>
4502Note that, unlike the standard libc function, the <tt>llvm.memmove.*</tt>
4503intrinsics do not return a value, and takes an extra alignment argument.
4504</p>
4505
4506<h5>Arguments:</h5>
4507
4508<p>
4509The first argument is a pointer to the destination, the second is a pointer to
4510the source. The third argument is an integer argument
4511specifying the number of bytes to copy, and the fourth argument is the alignment
4512of the source and destination locations.
4513</p>
4514
4515<p>
4516If the call to this intrinisic has an alignment value that is not 0 or 1, then
4517the caller guarantees that the source and destination pointers are aligned to
4518that boundary.
4519</p>
4520
4521<h5>Semantics:</h5>
4522
4523<p>
4524The '<tt>llvm.memmove.*</tt>' intrinsics copy a block of memory from the source
4525location to the destination location, which may overlap. It
4526copies "len" bytes of memory over. If the argument is known to be aligned to
4527some boundary, this can be specified as the fourth argument, otherwise it should
4528be set to 0 or 1.
4529</p>
4530</div>
4531
4532
4533<!-- _______________________________________________________________________ -->
4534<div class="doc_subsubsection">
4535 <a name="int_memset">'<tt>llvm.memset.*</tt>' Intrinsics</a>
4536</div>
4537
4538<div class="doc_text">
4539
4540<h5>Syntax:</h5>
4541<pre>
4542 declare void @llvm.memset.i32(i8 * &lt;dest&gt;, i8 &lt;val&gt;,
4543 i32 &lt;len&gt;, i32 &lt;align&gt;)
4544 declare void @llvm.memset.i64(i8 * &lt;dest&gt;, i8 &lt;val&gt;,
4545 i64 &lt;len&gt;, i32 &lt;align&gt;)
4546</pre>
4547
4548<h5>Overview:</h5>
4549
4550<p>
4551The '<tt>llvm.memset.*</tt>' intrinsics fill a block of memory with a particular
4552byte value.
4553</p>
4554
4555<p>
4556Note that, unlike the standard libc function, the <tt>llvm.memset</tt> intrinsic
4557does not return a value, and takes an extra alignment argument.
4558</p>
4559
4560<h5>Arguments:</h5>
4561
4562<p>
4563The first argument is a pointer to the destination to fill, the second is the
4564byte value to fill it with, the third argument is an integer
4565argument specifying the number of bytes to fill, and the fourth argument is the
4566known alignment of destination location.
4567</p>
4568
4569<p>
4570If the call to this intrinisic has an alignment value that is not 0 or 1, then
4571the caller guarantees that the destination pointer is aligned to that boundary.
4572</p>
4573
4574<h5>Semantics:</h5>
4575
4576<p>
4577The '<tt>llvm.memset.*</tt>' intrinsics fill "len" bytes of memory starting at
4578the
4579destination location. If the argument is known to be aligned to some boundary,
4580this can be specified as the fourth argument, otherwise it should be set to 0 or
45811.
4582</p>
4583</div>
4584
4585
4586<!-- _______________________________________________________________________ -->
4587<div class="doc_subsubsection">
4588 <a name="int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a>
4589</div>
4590
4591<div class="doc_text">
4592
4593<h5>Syntax:</h5>
Dale Johannesenf9adbb62007-10-02 17:47:38 +0000[diff] [blame]4594<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]4595floating point or vector of floating point type. Not all targets support all
4596types however.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4597<pre>
Dale Johannesenf9adbb62007-10-02 17:47:38 +0000[diff] [blame]4598 declare float @llvm.sqrt.f32(float %Val)
4599 declare double @llvm.sqrt.f64(double %Val)
4600 declare x86_fp80 @llvm.sqrt.f80(x86_fp80 %Val)
4601 declare fp128 @llvm.sqrt.f128(fp128 %Val)
4602 declare ppc_fp128 @llvm.sqrt.ppcf128(ppc_fp128 %Val)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4603</pre>
4604
4605<h5>Overview:</h5>
4606
4607<p>
4608The '<tt>llvm.sqrt</tt>' intrinsics return the sqrt of the specified operand,
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]4609returning the same value as the libm '<tt>sqrt</tt>' functions would. Unlike
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4610<tt>sqrt</tt> in libm, however, <tt>llvm.sqrt</tt> has undefined behavior for
4611negative numbers (which allows for better optimization).
4612</p>
4613
4614<h5>Arguments:</h5>
4615
4616<p>
4617The argument and return value are floating point numbers of the same type.
4618</p>
4619
4620<h5>Semantics:</h5>
4621
4622<p>
4623This function returns the sqrt of the specified operand if it is a nonnegative
4624floating point number.
4625</p>
4626</div>
4627
4628<!-- _______________________________________________________________________ -->
4629<div class="doc_subsubsection">
4630 <a name="int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a>
4631</div>
4632
4633<div class="doc_text">
4634
4635<h5>Syntax:</h5>
Dale Johannesenf9adbb62007-10-02 17:47:38 +0000[diff] [blame]4636<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]4637floating point or vector of floating point type. Not all targets support all
4638types however.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4639<pre>
Dale Johannesenf9adbb62007-10-02 17:47:38 +0000[diff] [blame]4640 declare float @llvm.powi.f32(float %Val, i32 %power)
4641 declare double @llvm.powi.f64(double %Val, i32 %power)
4642 declare x86_fp80 @llvm.powi.f80(x86_fp80 %Val, i32 %power)
4643 declare fp128 @llvm.powi.f128(fp128 %Val, i32 %power)
4644 declare ppc_fp128 @llvm.powi.ppcf128(ppc_fp128 %Val, i32 %power)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4645</pre>
4646
4647<h5>Overview:</h5>
4648
4649<p>
4650The '<tt>llvm.powi.*</tt>' intrinsics return the first operand raised to the
4651specified (positive or negative) power. The order of evaluation of
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]4652multiplications is not defined. When a vector of floating point type is
4653used, the second argument remains a scalar integer value.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4654</p>
4655
4656<h5>Arguments:</h5>
4657
4658<p>
4659The second argument is an integer power, and the first is a value to raise to
4660that power.
4661</p>
4662
4663<h5>Semantics:</h5>
4664
4665<p>
4666This function returns the first value raised to the second power with an
4667unspecified sequence of rounding operations.</p>
4668</div>
4669
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]4670<!-- _______________________________________________________________________ -->
4671<div class="doc_subsubsection">
4672 <a name="int_sin">'<tt>llvm.sin.*</tt>' Intrinsic</a>
4673</div>
4674
4675<div class="doc_text">
4676
4677<h5>Syntax:</h5>
4678<p>This is an overloaded intrinsic. You can use <tt>llvm.sin</tt> on any
4679floating point or vector of floating point type. Not all targets support all
4680types however.
4681<pre>
4682 declare float @llvm.sin.f32(float %Val)
4683 declare double @llvm.sin.f64(double %Val)
4684 declare x86_fp80 @llvm.sin.f80(x86_fp80 %Val)
4685 declare fp128 @llvm.sin.f128(fp128 %Val)
4686 declare ppc_fp128 @llvm.sin.ppcf128(ppc_fp128 %Val)
4687</pre>
4688
4689<h5>Overview:</h5>
4690
4691<p>
4692The '<tt>llvm.sin.*</tt>' intrinsics return the sine of the operand.
4693</p>
4694
4695<h5>Arguments:</h5>
4696
4697<p>
4698The argument and return value are floating point numbers of the same type.
4699</p>
4700
4701<h5>Semantics:</h5>
4702
4703<p>
4704This function returns the sine of the specified operand, returning the
4705same values as the libm <tt>sin</tt> functions would, and handles error
Dan Gohmaneaba92e2007-10-17 18:05:13 +0000[diff] [blame]4706conditions in the same way.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]4707</div>
4708
4709<!-- _______________________________________________________________________ -->
4710<div class="doc_subsubsection">
4711 <a name="int_cos">'<tt>llvm.cos.*</tt>' Intrinsic</a>
4712</div>
4713
4714<div class="doc_text">
4715
4716<h5>Syntax:</h5>
4717<p>This is an overloaded intrinsic. You can use <tt>llvm.cos</tt> on any
4718floating point or vector of floating point type. Not all targets support all
4719types however.
4720<pre>
4721 declare float @llvm.cos.f32(float %Val)
4722 declare double @llvm.cos.f64(double %Val)
4723 declare x86_fp80 @llvm.cos.f80(x86_fp80 %Val)
4724 declare fp128 @llvm.cos.f128(fp128 %Val)
4725 declare ppc_fp128 @llvm.cos.ppcf128(ppc_fp128 %Val)
4726</pre>
4727
4728<h5>Overview:</h5>
4729
4730<p>
4731The '<tt>llvm.cos.*</tt>' intrinsics return the cosine of the operand.
4732</p>
4733
4734<h5>Arguments:</h5>
4735
4736<p>
4737The argument and return value are floating point numbers of the same type.
4738</p>
4739
4740<h5>Semantics:</h5>
4741
4742<p>
4743This function returns the cosine of the specified operand, returning the
4744same values as the libm <tt>cos</tt> functions would, and handles error
Dan Gohmaneaba92e2007-10-17 18:05:13 +0000[diff] [blame]4745conditions in the same way.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]4746</div>
4747
4748<!-- _______________________________________________________________________ -->
4749<div class="doc_subsubsection">
4750 <a name="int_pow">'<tt>llvm.pow.*</tt>' Intrinsic</a>
4751</div>
4752
4753<div class="doc_text">
4754
4755<h5>Syntax:</h5>
4756<p>This is an overloaded intrinsic. You can use <tt>llvm.pow</tt> on any
4757floating point or vector of floating point type. Not all targets support all
4758types however.
4759<pre>
4760 declare float @llvm.pow.f32(float %Val, float %Power)
4761 declare double @llvm.pow.f64(double %Val, double %Power)
4762 declare x86_fp80 @llvm.pow.f80(x86_fp80 %Val, x86_fp80 %Power)
4763 declare fp128 @llvm.pow.f128(fp128 %Val, fp128 %Power)
4764 declare ppc_fp128 @llvm.pow.ppcf128(ppc_fp128 %Val, ppc_fp128 Power)
4765</pre>
4766
4767<h5>Overview:</h5>
4768
4769<p>
4770The '<tt>llvm.pow.*</tt>' intrinsics return the first operand raised to the
4771specified (positive or negative) power.
4772</p>
4773
4774<h5>Arguments:</h5>
4775
4776<p>
4777The second argument is a floating point power, and the first is a value to
4778raise to that power.
4779</p>
4780
4781<h5>Semantics:</h5>
4782
4783<p>
4784This function returns the first value raised to the second power,
4785returning the
4786same values as the libm <tt>pow</tt> functions would, and handles error
Dan Gohmaneaba92e2007-10-17 18:05:13 +0000[diff] [blame]4787conditions in the same way.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]4788</div>
4789
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4790
4791<!-- ======================================================================= -->
4792<div class="doc_subsection">
4793 <a name="int_manip">Bit Manipulation Intrinsics</a>
4794</div>
4795
4796<div class="doc_text">
4797<p>
4798LLVM provides intrinsics for a few important bit manipulation operations.
4799These allow efficient code generation for some algorithms.
4800</p>
4801
4802</div>
4803
4804<!-- _______________________________________________________________________ -->
4805<div class="doc_subsubsection">
4806 <a name="int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a>
4807</div>
4808
4809<div class="doc_text">
4810
4811<h5>Syntax:</h5>
4812<p>This is an overloaded intrinsic function. You can use bswap on any integer
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]4813type that is an even number of bytes (i.e. BitWidth % 16 == 0).
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4814<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]4815 declare i16 @llvm.bswap.i16(i16 &lt;id&gt;)
4816 declare i32 @llvm.bswap.i32(i32 &lt;id&gt;)
4817 declare i64 @llvm.bswap.i64(i64 &lt;id&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4818</pre>
4819
4820<h5>Overview:</h5>
4821
4822<p>
4823The '<tt>llvm.bswap</tt>' family of intrinsics is used to byte swap integer
4824values with an even number of bytes (positive multiple of 16 bits). These are
4825useful for performing operations on data that is not in the target's native
4826byte order.
4827</p>
4828
4829<h5>Semantics:</h5>
4830
4831<p>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]4832The <tt>llvm.bswap.i16</tt> intrinsic returns an i16 value that has the high
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4833and low byte of the input i16 swapped. Similarly, the <tt>llvm.bswap.i32</tt>
4834intrinsic returns an i32 value that has the four bytes of the input i32
4835swapped, 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]4836i32 will have its bytes in 3, 2, 1, 0 order. The <tt>llvm.bswap.i48</tt>,
4837<tt>llvm.bswap.i64</tt> and other intrinsics extend this concept to
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4838additional even-byte lengths (6 bytes, 8 bytes and more, respectively).
4839</p>
4840
4841</div>
4842
4843<!-- _______________________________________________________________________ -->
4844<div class="doc_subsubsection">
4845 <a name="int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic</a>
4846</div>
4847
4848<div class="doc_text">
4849
4850<h5>Syntax:</h5>
4851<p>This is an overloaded intrinsic. You can use llvm.ctpop on any integer bit
4852width. Not all targets support all bit widths however.
4853<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]4854 declare i8 @llvm.ctpop.i8 (i8 &lt;src&gt;)
4855 declare i16 @llvm.ctpop.i16(i16 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4856 declare i32 @llvm.ctpop.i32(i32 &lt;src&gt;)
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]4857 declare i64 @llvm.ctpop.i64(i64 &lt;src&gt;)
4858 declare i256 @llvm.ctpop.i256(i256 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4859</pre>
4860
4861<h5>Overview:</h5>
4862
4863<p>
4864The '<tt>llvm.ctpop</tt>' family of intrinsics counts the number of bits set in a
4865value.
4866</p>
4867
4868<h5>Arguments:</h5>
4869
4870<p>
4871The only argument is the value to be counted. The argument may be of any
4872integer type. The return type must match the argument type.
4873</p>
4874
4875<h5>Semantics:</h5>
4876
4877<p>
4878The '<tt>llvm.ctpop</tt>' intrinsic counts the 1's in a variable.
4879</p>
4880</div>
4881
4882<!-- _______________________________________________________________________ -->
4883<div class="doc_subsubsection">
4884 <a name="int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic</a>
4885</div>
4886
4887<div class="doc_text">
4888
4889<h5>Syntax:</h5>
4890<p>This is an overloaded intrinsic. You can use <tt>llvm.ctlz</tt> on any
4891integer bit width. Not all targets support all bit widths however.
4892<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]4893 declare i8 @llvm.ctlz.i8 (i8 &lt;src&gt;)
4894 declare i16 @llvm.ctlz.i16(i16 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4895 declare i32 @llvm.ctlz.i32(i32 &lt;src&gt;)
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]4896 declare i64 @llvm.ctlz.i64(i64 &lt;src&gt;)
4897 declare i256 @llvm.ctlz.i256(i256 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4898</pre>
4899
4900<h5>Overview:</h5>
4901
4902<p>
4903The '<tt>llvm.ctlz</tt>' family of intrinsic functions counts the number of
4904leading zeros in a variable.
4905</p>
4906
4907<h5>Arguments:</h5>
4908
4909<p>
4910The only argument is the value to be counted. The argument may be of any
4911integer type. The return type must match the argument type.
4912</p>
4913
4914<h5>Semantics:</h5>
4915
4916<p>
4917The '<tt>llvm.ctlz</tt>' intrinsic counts the leading (most significant) zeros
4918in a variable. If the src == 0 then the result is the size in bits of the type
4919of src. For example, <tt>llvm.ctlz(i32 2) = 30</tt>.
4920</p>
4921</div>
4922
4923
4924
4925<!-- _______________________________________________________________________ -->
4926<div class="doc_subsubsection">
4927 <a name="int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic</a>
4928</div>
4929
4930<div class="doc_text">
4931
4932<h5>Syntax:</h5>
4933<p>This is an overloaded intrinsic. You can use <tt>llvm.cttz</tt> on any
4934integer bit width. Not all targets support all bit widths however.
4935<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]4936 declare i8 @llvm.cttz.i8 (i8 &lt;src&gt;)
4937 declare i16 @llvm.cttz.i16(i16 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4938 declare i32 @llvm.cttz.i32(i32 &lt;src&gt;)
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]4939 declare i64 @llvm.cttz.i64(i64 &lt;src&gt;)
4940 declare i256 @llvm.cttz.i256(i256 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4941</pre>
4942
4943<h5>Overview:</h5>
4944
4945<p>
4946The '<tt>llvm.cttz</tt>' family of intrinsic functions counts the number of
4947trailing zeros.
4948</p>
4949
4950<h5>Arguments:</h5>
4951
4952<p>
4953The only argument is the value to be counted. The argument may be of any
4954integer type. The return type must match the argument type.
4955</p>
4956
4957<h5>Semantics:</h5>
4958
4959<p>
4960The '<tt>llvm.cttz</tt>' intrinsic counts the trailing (least significant) zeros
4961in a variable. If the src == 0 then the result is the size in bits of the type
4962of src. For example, <tt>llvm.cttz(2) = 1</tt>.
4963</p>
4964</div>
4965
4966<!-- _______________________________________________________________________ -->
4967<div class="doc_subsubsection">
4968 <a name="int_part_select">'<tt>llvm.part.select.*</tt>' Intrinsic</a>
4969</div>
4970
4971<div class="doc_text">
4972
4973<h5>Syntax:</h5>
4974<p>This is an overloaded intrinsic. You can use <tt>llvm.part.select</tt>
4975on any integer bit width.
4976<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]4977 declare i17 @llvm.part.select.i17 (i17 %val, i32 %loBit, i32 %hiBit)
4978 declare i29 @llvm.part.select.i29 (i29 %val, i32 %loBit, i32 %hiBit)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4979</pre>
4980
4981<h5>Overview:</h5>
4982<p>The '<tt>llvm.part.select</tt>' family of intrinsic functions selects a
4983range of bits from an integer value and returns them in the same bit width as
4984the original value.</p>
4985
4986<h5>Arguments:</h5>
4987<p>The first argument, <tt>%val</tt> and the result may be integer types of
4988any bit width but they must have the same bit width. The second and third
4989arguments must be <tt>i32</tt> type since they specify only a bit index.</p>
4990
4991<h5>Semantics:</h5>
4992<p>The operation of the '<tt>llvm.part.select</tt>' intrinsic has two modes
4993of operation: forwards and reverse. If <tt>%loBit</tt> is greater than
4994<tt>%hiBits</tt> then the intrinsic operates in reverse mode. Otherwise it
4995operates in forward mode.</p>
4996<p>In forward mode, this intrinsic is the equivalent of shifting <tt>%val</tt>
4997right by <tt>%loBit</tt> bits and then ANDing it with a mask with
4998only the <tt>%hiBit - %loBit</tt> bits set, as follows:</p>
4999<ol>
5000 <li>The <tt>%val</tt> is shifted right (LSHR) by the number of bits specified
5001 by <tt>%loBits</tt>. This normalizes the value to the low order bits.</li>
5002 <li>The <tt>%loBits</tt> value is subtracted from the <tt>%hiBits</tt> value
5003 to determine the number of bits to retain.</li>
5004 <li>A mask of the retained bits is created by shifting a -1 value.</li>
5005 <li>The mask is ANDed with <tt>%val</tt> to produce the result.
5006</ol>
5007<p>In reverse mode, a similar computation is made except that the bits are
5008returned in the reverse order. So, for example, if <tt>X</tt> has the value
5009<tt>i16 0x0ACF (101011001111)</tt> and we apply
5010<tt>part.select(i16 X, 8, 3)</tt> to it, we get back the value
5011<tt>i16 0x0026 (000000100110)</tt>.</p>
5012</div>
5013
5014<div class="doc_subsubsection">
5015 <a name="int_part_set">'<tt>llvm.part.set.*</tt>' Intrinsic</a>
5016</div>
5017
5018<div class="doc_text">
5019
5020<h5>Syntax:</h5>
5021<p>This is an overloaded intrinsic. You can use <tt>llvm.part.set</tt>
5022on any integer bit width.
5023<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5024 declare i17 @llvm.part.set.i17.i9 (i17 %val, i9 %repl, i32 %lo, i32 %hi)
5025 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]5026</pre>
5027
5028<h5>Overview:</h5>
5029<p>The '<tt>llvm.part.set</tt>' family of intrinsic functions replaces a range
5030of bits in an integer value with another integer value. It returns the integer
5031with the replaced bits.</p>
5032
5033<h5>Arguments:</h5>
5034<p>The first argument, <tt>%val</tt> and the result may be integer types of
5035any bit width but they must have the same bit width. <tt>%val</tt> is the value
5036whose bits will be replaced. The second argument, <tt>%repl</tt> may be an
5037integer of any bit width. The third and fourth arguments must be <tt>i32</tt>
5038type since they specify only a bit index.</p>
5039
5040<h5>Semantics:</h5>
5041<p>The operation of the '<tt>llvm.part.set</tt>' intrinsic has two modes
5042of operation: forwards and reverse. If <tt>%lo</tt> is greater than
5043<tt>%hi</tt> then the intrinsic operates in reverse mode. Otherwise it
5044operates in forward mode.</p>
5045<p>For both modes, the <tt>%repl</tt> value is prepared for use by either
5046truncating it down to the size of the replacement area or zero extending it
5047up to that size.</p>
5048<p>In forward mode, the bits between <tt>%lo</tt> and <tt>%hi</tt> (inclusive)
5049are replaced with corresponding bits from <tt>%repl</tt>. That is the 0th bit
5050in <tt>%repl</tt> replaces the <tt>%lo</tt>th bit in <tt>%val</tt> and etc. up
5051to the <tt>%hi</tt>th bit.
5052<p>In reverse mode, a similar computation is made except that the bits are
5053reversed. That is, the <tt>0</tt>th bit in <tt>%repl</tt> replaces the
5054<tt>%hi</tt> bit in <tt>%val</tt> and etc. down to the <tt>%lo</tt>th bit.
5055<h5>Examples:</h5>
5056<pre>
5057 llvm.part.set(0xFFFF, 0, 4, 7) -&gt; 0xFF0F
5058 llvm.part.set(0xFFFF, 0, 7, 4) -&gt; 0xFF0F
5059 llvm.part.set(0xFFFF, 1, 7, 4) -&gt; 0xFF8F
5060 llvm.part.set(0xFFFF, F, 8, 3) -&gt; 0xFFE7
5061 llvm.part.set(0xFFFF, 0, 3, 8) -&gt; 0xFE07
5062</pre>
5063</div>
5064
5065<!-- ======================================================================= -->
5066<div class="doc_subsection">
5067 <a name="int_debugger">Debugger Intrinsics</a>
5068</div>
5069
5070<div class="doc_text">
5071<p>
5072The LLVM debugger intrinsics (which all start with <tt>llvm.dbg.</tt> prefix),
5073are described in the <a
5074href="SourceLevelDebugging.html#format_common_intrinsics">LLVM Source Level
5075Debugging</a> document.
5076</p>
5077</div>
5078
5079
5080<!-- ======================================================================= -->
5081<div class="doc_subsection">
5082 <a name="int_eh">Exception Handling Intrinsics</a>
5083</div>
5084
5085<div class="doc_text">
5086<p> The LLVM exception handling intrinsics (which all start with
5087<tt>llvm.eh.</tt> prefix), are described in the <a
5088href="ExceptionHandling.html#format_common_intrinsics">LLVM Exception
5089Handling</a> document. </p>
5090</div>
5091
5092<!-- ======================================================================= -->
5093<div class="doc_subsection">
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5094 <a name="int_trampoline">Trampoline Intrinsic</a>
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5095</div>
5096
5097<div class="doc_text">
5098<p>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5099 This intrinsic makes it possible to excise one parameter, marked with
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5100 the <tt>nest</tt> attribute, from a function. The result is a callable
5101 function pointer lacking the nest parameter - the caller does not need
5102 to provide a value for it. Instead, the value to use is stored in
5103 advance in a "trampoline", a block of memory usually allocated
5104 on the stack, which also contains code to splice the nest value into the
5105 argument list. This is used to implement the GCC nested function address
5106 extension.
5107</p>
5108<p>
5109 For example, if the function is
5110 <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]5111 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]5112<pre>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5113 %tramp = alloca [10 x i8], align 4 ; size and alignment only correct for X86
5114 %tramp1 = getelementptr [10 x i8]* %tramp, i32 0, i32 0
5115 %p = call i8* @llvm.init.trampoline( i8* %tramp1, i8* bitcast (i32 (i8* nest , i32, i32)* @f to i8*), i8* %nval )
5116 %fp = bitcast i8* %p to i32 (i32, i32)*
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5117</pre>
Bill Wendlinge262b4c2007-09-22 09:23:55 +0000[diff] [blame]5118 <p>The call <tt>%val = call i32 %fp( i32 %x, i32 %y )</tt> is then equivalent
5119 to <tt>%val = call i32 %f( i8* %nval, i32 %x, i32 %y )</tt>.</p>
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5120</div>
5121
5122<!-- _______________________________________________________________________ -->
5123<div class="doc_subsubsection">
5124 <a name="int_it">'<tt>llvm.init.trampoline</tt>' Intrinsic</a>
5125</div>
5126<div class="doc_text">
5127<h5>Syntax:</h5>
5128<pre>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5129declare 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]5130</pre>
5131<h5>Overview:</h5>
5132<p>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5133 This fills the memory pointed to by <tt>tramp</tt> with code
5134 and returns a function pointer suitable for executing it.
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5135</p>
5136<h5>Arguments:</h5>
5137<p>
5138 The <tt>llvm.init.trampoline</tt> intrinsic takes three arguments, all
5139 pointers. The <tt>tramp</tt> argument must point to a sufficiently large
5140 and sufficiently aligned block of memory; this memory is written to by the
Duncan Sands35012212007-08-22 23:39:54 +0000[diff] [blame]5141 intrinsic. Note that the size and the alignment are target-specific - LLVM
5142 currently provides no portable way of determining them, so a front-end that
5143 generates this intrinsic needs to have some target-specific knowledge.
5144 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]5145</p>
5146<h5>Semantics:</h5>
5147<p>
5148 The block of memory pointed to by <tt>tramp</tt> is filled with target
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5149 dependent code, turning it into a function. A pointer to this function is
5150 returned, but needs to be bitcast to an
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5151 <a href="#int_trampoline">appropriate function pointer type</a>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5152 before being called. The new function's signature is the same as that of
5153 <tt>func</tt> with any arguments marked with the <tt>nest</tt> attribute
5154 removed. At most one such <tt>nest</tt> argument is allowed, and it must be
5155 of pointer type. Calling the new function is equivalent to calling
5156 <tt>func</tt> with the same argument list, but with <tt>nval</tt> used for the
5157 missing <tt>nest</tt> argument. If, after calling
5158 <tt>llvm.init.trampoline</tt>, the memory pointed to by <tt>tramp</tt> is
5159 modified, then the effect of any later call to the returned function pointer is
5160 undefined.
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5161</p>
5162</div>
5163
5164<!-- ======================================================================= -->
5165<div class="doc_subsection">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5166 <a name="int_general">General Intrinsics</a>
5167</div>
5168
5169<div class="doc_text">
5170<p> This class of intrinsics is designed to be generic and has
5171no specific purpose. </p>
5172</div>
5173
5174<!-- _______________________________________________________________________ -->
5175<div class="doc_subsubsection">
5176 <a name="int_var_annotation">'<tt>llvm.var.annotation</tt>' Intrinsic</a>
5177</div>
5178
5179<div class="doc_text">
5180
5181<h5>Syntax:</h5>
5182<pre>
5183 declare void @llvm.var.annotation(i8* &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
5184</pre>
5185
5186<h5>Overview:</h5>
5187
5188<p>
5189The '<tt>llvm.var.annotation</tt>' intrinsic
5190</p>
5191
5192<h5>Arguments:</h5>
5193
5194<p>
5195The first argument is a pointer to a value, the second is a pointer to a
5196global string, the third is a pointer to a global string which is the source
5197file name, and the last argument is the line number.
5198</p>
5199
5200<h5>Semantics:</h5>
5201
5202<p>
5203This intrinsic allows annotation of local variables with arbitrary strings.
5204This can be useful for special purpose optimizations that want to look for these
5205 annotations. These have no other defined use, they are ignored by code
5206 generation and optimization.
5207</div>
5208
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]5209<!-- _______________________________________________________________________ -->
5210<div class="doc_subsubsection">
Tanya Lattnerc9869b12007-09-21 23:57:59 +0000[diff] [blame]5211 <a name="int_annotation">'<tt>llvm.annotation.*</tt>' Intrinsic</a>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]5212</div>
5213
5214<div class="doc_text">
5215
5216<h5>Syntax:</h5>
Tanya Lattnere545be72007-09-21 23:56:27 +0000[diff] [blame]5217<p>This is an overloaded intrinsic. You can use '<tt>llvm.annotation</tt>' on
5218any integer bit width.
5219</p>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]5220<pre>
Tanya Lattner09161fe2007-09-22 00:03:01 +0000[diff] [blame]5221 declare i8 @llvm.annotation.i8(i8 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
5222 declare i16 @llvm.annotation.i16(i16 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
5223 declare i32 @llvm.annotation.i32(i32 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
5224 declare i64 @llvm.annotation.i64(i64 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
5225 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]5226</pre>
5227
5228<h5>Overview:</h5>
Tanya Lattnere545be72007-09-21 23:56:27 +0000[diff] [blame]5229
5230<p>
5231The '<tt>llvm.annotation</tt>' intrinsic.
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]5232</p>
5233
5234<h5>Arguments:</h5>
5235
5236<p>
5237The first argument is an integer value (result of some expression),
5238the second is a pointer to a global string, the third is a pointer to a global
5239string which is the source file name, and the last argument is the line number.
Tanya Lattnere545be72007-09-21 23:56:27 +0000[diff] [blame]5240It returns the value of the first argument.
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]5241</p>
5242
5243<h5>Semantics:</h5>
5244
5245<p>
5246This intrinsic allows annotations to be put on arbitrary expressions
5247with arbitrary strings. This can be useful for special purpose optimizations
5248that want to look for these annotations. These have no other defined use, they
5249are ignored by code generation and optimization.
5250</div>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5251
5252<!-- *********************************************************************** -->
5253<hr>
5254<address>
5255 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
5256 src="http://jigsaw.w3.org/css-validator/images/vcss" alt="Valid CSS!"></a>
5257 <a href="http://validator.w3.org/check/referer"><img
5258 src="http://www.w3.org/Icons/valid-html401" alt="Valid HTML 4.01!" /></a>
5259
5260 <a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
5261 <a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br>
5262 Last modified: $Date$
5263</address>
5264</body>
5265</html>