Gitiles
Code Review Sign In
gerrit-public.fairphone.software / fp2-dev / platform / external / llvm / a3cc537fb668b8ab2a9c679707f5a57289d23687 / . / docs / LangRef.html
blob: 7796126b62d55fcb8e829029a0c2a6251816697c [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>
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame^]147 <li><a href="#i_getresult">'<tt>getresult</tt>' Instruction</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]148 </ol>
149 </li>
150 </ol>
151 </li>
152 <li><a href="#intrinsics">Intrinsic Functions</a>
153 <ol>
154 <li><a href="#int_varargs">Variable Argument Handling Intrinsics</a>
155 <ol>
156 <li><a href="#int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a></li>
157 <li><a href="#int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a></li>
158 <li><a href="#int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a></li>
159 </ol>
160 </li>
161 <li><a href="#int_gc">Accurate Garbage Collection Intrinsics</a>
162 <ol>
163 <li><a href="#int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a></li>
164 <li><a href="#int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a></li>
165 <li><a href="#int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a></li>
166 </ol>
167 </li>
168 <li><a href="#int_codegen">Code Generator Intrinsics</a>
169 <ol>
170 <li><a href="#int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a></li>
171 <li><a href="#int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a></li>
172 <li><a href="#int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a></li>
173 <li><a href="#int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a></li>
174 <li><a href="#int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a></li>
175 <li><a href="#int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a></li>
176 <li><a href="#int_readcyclecounter"><tt>llvm.readcyclecounter</tt>' Intrinsic</a></li>
177 </ol>
178 </li>
179 <li><a href="#int_libc">Standard C Library Intrinsics</a>
180 <ol>
181 <li><a href="#int_memcpy">'<tt>llvm.memcpy.*</tt>' Intrinsic</a></li>
182 <li><a href="#int_memmove">'<tt>llvm.memmove.*</tt>' Intrinsic</a></li>
183 <li><a href="#int_memset">'<tt>llvm.memset.*</tt>' Intrinsic</a></li>
184 <li><a href="#int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a></li>
185 <li><a href="#int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a></li>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]186 <li><a href="#int_sin">'<tt>llvm.sin.*</tt>' Intrinsic</a></li>
187 <li><a href="#int_cos">'<tt>llvm.cos.*</tt>' Intrinsic</a></li>
188 <li><a href="#int_pow">'<tt>llvm.pow.*</tt>' Intrinsic</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]189 </ol>
190 </li>
191 <li><a href="#int_manip">Bit Manipulation Intrinsics</a>
192 <ol>
193 <li><a href="#int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a></li>
194 <li><a href="#int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic </a></li>
195 <li><a href="#int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic </a></li>
196 <li><a href="#int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic </a></li>
197 <li><a href="#int_part_select">'<tt>llvm.part.select.*</tt>' Intrinsic </a></li>
198 <li><a href="#int_part_set">'<tt>llvm.part.set.*</tt>' Intrinsic </a></li>
199 </ol>
200 </li>
201 <li><a href="#int_debugger">Debugger intrinsics</a></li>
202 <li><a href="#int_eh">Exception Handling intrinsics</a></li>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]203 <li><a href="#int_trampoline">Trampoline Intrinsic</a>
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]204 <ol>
205 <li><a href="#int_it">'<tt>llvm.init.trampoline</tt>' Intrinsic</a></li>
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]206 </ol>
207 </li>
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]208 <li><a href="#int_atomics">Atomic intrinsics</a>
209 <ol>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]210 <li><a href="#int_memory_barrier"><tt>llvm.memory_barrier</tt></a></li>
211 <li><a href="#int_atomic_lcs"><tt>llvm.atomic.lcs</tt></a></li>
212 <li><a href="#int_atomic_las"><tt>llvm.atomic.las</tt></a></li>
213 <li><a href="#int_atomic_swap"><tt>llvm.atomic.swap</tt></a></li>
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]214 </ol>
215 </li>
Reid Spencerb043f672007-07-20 19:59:11 +0000[diff] [blame]216 <li><a href="#int_general">General intrinsics</a>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]217 <ol>
Reid Spencerb043f672007-07-20 19:59:11 +0000[diff] [blame]218 <li><a href="#int_var_annotation">
Tanya Lattner51369f32007-09-22 00:01:26 +0000[diff] [blame]219 <tt>llvm.var.annotation</tt>' Intrinsic</a></li>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]220 <li><a href="#int_annotation">
Tanya Lattner51369f32007-09-22 00:01:26 +0000[diff] [blame]221 <tt>llvm.annotation.*</tt>' Intrinsic</a></li>
Anton Korobeynikove6e764f2008-01-15 22:31:34 +0000[diff] [blame]222 <li><a href="#int_trap">
223 <tt>llvm.trap</tt>' Intrinsic</a></li>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]224 </ol>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]225 </li>
226 </ol>
227 </li>
228</ol>
229
230<div class="doc_author">
231 <p>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a>
232 and <a href="mailto:vadve@cs.uiuc.edu">Vikram Adve</a></p>
233</div>
234
235<!-- *********************************************************************** -->
236<div class="doc_section"> <a name="abstract">Abstract </a></div>
237<!-- *********************************************************************** -->
238
239<div class="doc_text">
240<p>This document is a reference manual for the LLVM assembly language.
241LLVM is an SSA based representation that provides type safety,
242low-level operations, flexibility, and the capability of representing
243'all' high-level languages cleanly. It is the common code
244representation used throughout all phases of the LLVM compilation
245strategy.</p>
246</div>
247
248<!-- *********************************************************************** -->
249<div class="doc_section"> <a name="introduction">Introduction</a> </div>
250<!-- *********************************************************************** -->
251
252<div class="doc_text">
253
254<p>The LLVM code representation is designed to be used in three
255different forms: as an in-memory compiler IR, as an on-disk bitcode
256representation (suitable for fast loading by a Just-In-Time compiler),
257and as a human readable assembly language representation. This allows
258LLVM to provide a powerful intermediate representation for efficient
259compiler transformations and analysis, while providing a natural means
260to debug and visualize the transformations. The three different forms
261of LLVM are all equivalent. This document describes the human readable
262representation and notation.</p>
263
264<p>The LLVM representation aims to be light-weight and low-level
265while being expressive, typed, and extensible at the same time. It
266aims to be a "universal IR" of sorts, by being at a low enough level
267that high-level ideas may be cleanly mapped to it (similar to how
268microprocessors are "universal IR's", allowing many source languages to
269be mapped to them). By providing type information, LLVM can be used as
270the target of optimizations: for example, through pointer analysis, it
271can be proven that a C automatic variable is never accessed outside of
272the current function... allowing it to be promoted to a simple SSA
273value instead of a memory location.</p>
274
275</div>
276
277<!-- _______________________________________________________________________ -->
278<div class="doc_subsubsection"> <a name="wellformed">Well-Formedness</a> </div>
279
280<div class="doc_text">
281
282<p>It is important to note that this document describes 'well formed'
283LLVM assembly language. There is a difference between what the parser
284accepts and what is considered 'well formed'. For example, the
285following instruction is syntactically okay, but not well formed:</p>
286
287<div class="doc_code">
288<pre>
289%x = <a href="#i_add">add</a> i32 1, %x
290</pre>
291</div>
292
293<p>...because the definition of <tt>%x</tt> does not dominate all of
294its uses. The LLVM infrastructure provides a verification pass that may
295be used to verify that an LLVM module is well formed. This pass is
296automatically run by the parser after parsing input assembly and by
297the optimizer before it outputs bitcode. The violations pointed out
298by the verifier pass indicate bugs in transformation passes or input to
299the parser.</p>
300</div>
301
Chris Lattnera83fdc02007-10-03 17:34:29 +0000[diff] [blame]302<!-- Describe the typesetting conventions here. -->
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]303
304<!-- *********************************************************************** -->
305<div class="doc_section"> <a name="identifiers">Identifiers</a> </div>
306<!-- *********************************************************************** -->
307
308<div class="doc_text">
309
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]310 <p>LLVM identifiers come in two basic types: global and local. Global
311 identifiers (functions, global variables) begin with the @ character. Local
312 identifiers (register names, types) begin with the % character. Additionally,
313 there are three different formats for identifiers, for different purposes:
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]314
315<ol>
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]316 <li>Named values are represented as a string of characters with their prefix.
317 For example, %foo, @DivisionByZero, %a.really.long.identifier. The actual
318 regular expression used is '<tt>[%@][a-zA-Z$._][a-zA-Z$._0-9]*</tt>'.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]319 Identifiers which require other characters in their names can be surrounded
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]320 with quotes. In this way, anything except a <tt>&quot;</tt> character can
321 be used in a named value.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]322
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]323 <li>Unnamed values are represented as an unsigned numeric value with their
324 prefix. For example, %12, @2, %44.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]325
326 <li>Constants, which are described in a <a href="#constants">section about
327 constants</a>, below.</li>
328</ol>
329
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]330<p>LLVM requires that values start with a prefix for two reasons: Compilers
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]331don't need to worry about name clashes with reserved words, and the set of
332reserved words may be expanded in the future without penalty. Additionally,
333unnamed identifiers allow a compiler to quickly come up with a temporary
334variable without having to avoid symbol table conflicts.</p>
335
336<p>Reserved words in LLVM are very similar to reserved words in other
337languages. There are keywords for different opcodes
338('<tt><a href="#i_add">add</a></tt>',
339 '<tt><a href="#i_bitcast">bitcast</a></tt>',
340 '<tt><a href="#i_ret">ret</a></tt>', etc...), for primitive type names ('<tt><a
341href="#t_void">void</a></tt>', '<tt><a href="#t_primitive">i32</a></tt>', etc...),
342and others. These reserved words cannot conflict with variable names, because
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]343none of them start with a prefix character ('%' or '@').</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]344
345<p>Here is an example of LLVM code to multiply the integer variable
346'<tt>%X</tt>' by 8:</p>
347
348<p>The easy way:</p>
349
350<div class="doc_code">
351<pre>
352%result = <a href="#i_mul">mul</a> i32 %X, 8
353</pre>
354</div>
355
356<p>After strength reduction:</p>
357
358<div class="doc_code">
359<pre>
360%result = <a href="#i_shl">shl</a> i32 %X, i8 3
361</pre>
362</div>
363
364<p>And the hard way:</p>
365
366<div class="doc_code">
367<pre>
368<a href="#i_add">add</a> i32 %X, %X <i>; yields {i32}:%0</i>
369<a href="#i_add">add</a> i32 %0, %0 <i>; yields {i32}:%1</i>
370%result = <a href="#i_add">add</a> i32 %1, %1
371</pre>
372</div>
373
374<p>This last way of multiplying <tt>%X</tt> by 8 illustrates several
375important lexical features of LLVM:</p>
376
377<ol>
378
379 <li>Comments are delimited with a '<tt>;</tt>' and go until the end of
380 line.</li>
381
382 <li>Unnamed temporaries are created when the result of a computation is not
383 assigned to a named value.</li>
384
385 <li>Unnamed temporaries are numbered sequentially</li>
386
387</ol>
388
389<p>...and it also shows a convention that we follow in this document. When
390demonstrating instructions, we will follow an instruction with a comment that
391defines the type and name of value produced. Comments are shown in italic
392text.</p>
393
394</div>
395
396<!-- *********************************************************************** -->
397<div class="doc_section"> <a name="highlevel">High Level Structure</a> </div>
398<!-- *********************************************************************** -->
399
400<!-- ======================================================================= -->
401<div class="doc_subsection"> <a name="modulestructure">Module Structure</a>
402</div>
403
404<div class="doc_text">
405
406<p>LLVM programs are composed of "Module"s, each of which is a
407translation unit of the input programs. Each module consists of
408functions, global variables, and symbol table entries. Modules may be
409combined together with the LLVM linker, which merges function (and
410global variable) definitions, resolves forward declarations, and merges
411symbol table entries. Here is an example of the "hello world" module:</p>
412
413<div class="doc_code">
414<pre><i>; Declare the string constant as a global constant...</i>
415<a href="#identifiers">@.LC0</a> = <a href="#linkage_internal">internal</a> <a
416 href="#globalvars">constant</a> <a href="#t_array">[13 x i8]</a> c"hello world\0A\00" <i>; [13 x i8]*</i>
417
418<i>; External declaration of the puts function</i>
419<a href="#functionstructure">declare</a> i32 @puts(i8 *) <i>; i32(i8 *)* </i>
420
421<i>; Definition of main function</i>
422define i32 @main() { <i>; i32()* </i>
423 <i>; Convert [13x i8 ]* to i8 *...</i>
424 %cast210 = <a
425 href="#i_getelementptr">getelementptr</a> [13 x i8 ]* @.LC0, i64 0, i64 0 <i>; i8 *</i>
426
427 <i>; Call puts function to write out the string to stdout...</i>
428 <a
429 href="#i_call">call</a> i32 @puts(i8 * %cast210) <i>; i32</i>
430 <a
431 href="#i_ret">ret</a> i32 0<br>}<br>
432</pre>
433</div>
434
435<p>This example is made up of a <a href="#globalvars">global variable</a>
436named "<tt>.LC0</tt>", an external declaration of the "<tt>puts</tt>"
437function, and a <a href="#functionstructure">function definition</a>
438for "<tt>main</tt>".</p>
439
440<p>In general, a module is made up of a list of global values,
441where both functions and global variables are global values. Global values are
442represented by a pointer to a memory location (in this case, a pointer to an
443array of char, and a pointer to a function), and have one of the following <a
444href="#linkage">linkage types</a>.</p>
445
446</div>
447
448<!-- ======================================================================= -->
449<div class="doc_subsection">
450 <a name="linkage">Linkage Types</a>
451</div>
452
453<div class="doc_text">
454
455<p>
456All Global Variables and Functions have one of the following types of linkage:
457</p>
458
459<dl>
460
461 <dt><tt><b><a name="linkage_internal">internal</a></b></tt> </dt>
462
463 <dd>Global values with internal linkage are only directly accessible by
464 objects in the current module. In particular, linking code into a module with
465 an internal global value may cause the internal to be renamed as necessary to
466 avoid collisions. Because the symbol is internal to the module, all
467 references can be updated. This corresponds to the notion of the
468 '<tt>static</tt>' keyword in C.
469 </dd>
470
471 <dt><tt><b><a name="linkage_linkonce">linkonce</a></b></tt>: </dt>
472
473 <dd>Globals with "<tt>linkonce</tt>" linkage are merged with other globals of
474 the same name when linkage occurs. This is typically used to implement
475 inline functions, templates, or other code which must be generated in each
476 translation unit that uses it. Unreferenced <tt>linkonce</tt> globals are
477 allowed to be discarded.
478 </dd>
479
480 <dt><tt><b><a name="linkage_weak">weak</a></b></tt>: </dt>
481
482 <dd>"<tt>weak</tt>" linkage is exactly the same as <tt>linkonce</tt> linkage,
483 except that unreferenced <tt>weak</tt> globals may not be discarded. This is
484 used for globals that may be emitted in multiple translation units, but that
485 are not guaranteed to be emitted into every translation unit that uses them.
486 One example of this are common globals in C, such as "<tt>int X;</tt>" at
487 global scope.
488 </dd>
489
490 <dt><tt><b><a name="linkage_appending">appending</a></b></tt>: </dt>
491
492 <dd>"<tt>appending</tt>" linkage may only be applied to global variables of
493 pointer to array type. When two global variables with appending linkage are
494 linked together, the two global arrays are appended together. This is the
495 LLVM, typesafe, equivalent of having the system linker append together
496 "sections" with identical names when .o files are linked.
497 </dd>
498
499 <dt><tt><b><a name="linkage_externweak">extern_weak</a></b></tt>: </dt>
500 <dd>The semantics of this linkage follow the ELF model: the symbol is weak
501 until linked, if not linked, the symbol becomes null instead of being an
502 undefined reference.
503 </dd>
504
505 <dt><tt><b><a name="linkage_external">externally visible</a></b></tt>:</dt>
506
507 <dd>If none of the above identifiers are used, the global is externally
508 visible, meaning that it participates in linkage and can be used to resolve
509 external symbol references.
510 </dd>
511</dl>
512
513 <p>
514 The next two types of linkage are targeted for Microsoft Windows platform
515 only. They are designed to support importing (exporting) symbols from (to)
516 DLLs.
517 </p>
518
519 <dl>
520 <dt><tt><b><a name="linkage_dllimport">dllimport</a></b></tt>: </dt>
521
522 <dd>"<tt>dllimport</tt>" linkage causes the compiler to reference a function
523 or variable via a global pointer to a pointer that is set up by the DLL
524 exporting the symbol. On Microsoft Windows targets, the pointer name is
525 formed by combining <code>_imp__</code> and the function or variable name.
526 </dd>
527
528 <dt><tt><b><a name="linkage_dllexport">dllexport</a></b></tt>: </dt>
529
530 <dd>"<tt>dllexport</tt>" linkage causes the compiler to provide a global
531 pointer to a pointer in a DLL, so that it can be referenced with the
532 <tt>dllimport</tt> attribute. On Microsoft Windows targets, the pointer
533 name is formed by combining <code>_imp__</code> and the function or variable
534 name.
535 </dd>
536
537</dl>
538
539<p><a name="linkage_external"></a>For example, since the "<tt>.LC0</tt>"
540variable is defined to be internal, if another module defined a "<tt>.LC0</tt>"
541variable and was linked with this one, one of the two would be renamed,
542preventing a collision. Since "<tt>main</tt>" and "<tt>puts</tt>" are
543external (i.e., lacking any linkage declarations), they are accessible
544outside of the current module.</p>
545<p>It is illegal for a function <i>declaration</i>
546to have any linkage type other than "externally visible", <tt>dllimport</tt>,
547or <tt>extern_weak</tt>.</p>
548<p>Aliases can have only <tt>external</tt>, <tt>internal</tt> and <tt>weak</tt>
549linkages.
550</div>
551
552<!-- ======================================================================= -->
553<div class="doc_subsection">
554 <a name="callingconv">Calling Conventions</a>
555</div>
556
557<div class="doc_text">
558
559<p>LLVM <a href="#functionstructure">functions</a>, <a href="#i_call">calls</a>
560and <a href="#i_invoke">invokes</a> can all have an optional calling convention
561specified for the call. The calling convention of any pair of dynamic
562caller/callee must match, or the behavior of the program is undefined. The
563following calling conventions are supported by LLVM, and more may be added in
564the future:</p>
565
566<dl>
567 <dt><b>"<tt>ccc</tt>" - The C calling convention</b>:</dt>
568
569 <dd>This calling convention (the default if no other calling convention is
570 specified) matches the target C calling conventions. This calling convention
571 supports varargs function calls and tolerates some mismatch in the declared
572 prototype and implemented declaration of the function (as does normal C).
573 </dd>
574
575 <dt><b>"<tt>fastcc</tt>" - The fast calling convention</b>:</dt>
576
577 <dd>This calling convention attempts to make calls as fast as possible
578 (e.g. by passing things in registers). This calling convention allows the
579 target to use whatever tricks it wants to produce fast code for the target,
580 without having to conform to an externally specified ABI. Implementations of
581 this convention should allow arbitrary tail call optimization to be supported.
582 This calling convention does not support varargs and requires the prototype of
583 all callees to exactly match the prototype of the function definition.
584 </dd>
585
586 <dt><b>"<tt>coldcc</tt>" - The cold calling convention</b>:</dt>
587
588 <dd>This calling convention attempts to make code in the caller as efficient
589 as possible under the assumption that the call is not commonly executed. As
590 such, these calls often preserve all registers so that the call does not break
591 any live ranges in the caller side. This calling convention does not support
592 varargs and requires the prototype of all callees to exactly match the
593 prototype of the function definition.
594 </dd>
595
596 <dt><b>"<tt>cc &lt;<em>n</em>&gt;</tt>" - Numbered convention</b>:</dt>
597
598 <dd>Any calling convention may be specified by number, allowing
599 target-specific calling conventions to be used. Target specific calling
600 conventions start at 64.
601 </dd>
602</dl>
603
604<p>More calling conventions can be added/defined on an as-needed basis, to
605support pascal conventions or any other well-known target-independent
606convention.</p>
607
608</div>
609
610<!-- ======================================================================= -->
611<div class="doc_subsection">
612 <a name="visibility">Visibility Styles</a>
613</div>
614
615<div class="doc_text">
616
617<p>
618All Global Variables and Functions have one of the following visibility styles:
619</p>
620
621<dl>
622 <dt><b>"<tt>default</tt>" - Default style</b>:</dt>
623
624 <dd>On ELF, default visibility means that the declaration is visible to other
625 modules and, in shared libraries, means that the declared entity may be
626 overridden. On Darwin, default visibility means that the declaration is
627 visible to other modules. Default visibility corresponds to "external
628 linkage" in the language.
629 </dd>
630
631 <dt><b>"<tt>hidden</tt>" - Hidden style</b>:</dt>
632
633 <dd>Two declarations of an object with hidden visibility refer to the same
634 object if they are in the same shared object. Usually, hidden visibility
635 indicates that the symbol will not be placed into the dynamic symbol table,
636 so no other module (executable or shared library) can reference it
637 directly.
638 </dd>
639
640 <dt><b>"<tt>protected</tt>" - Protected style</b>:</dt>
641
642 <dd>On ELF, protected visibility indicates that the symbol will be placed in
643 the dynamic symbol table, but that references within the defining module will
644 bind to the local symbol. That is, the symbol cannot be overridden by another
645 module.
646 </dd>
647</dl>
648
649</div>
650
651<!-- ======================================================================= -->
652<div class="doc_subsection">
653 <a name="globalvars">Global Variables</a>
654</div>
655
656<div class="doc_text">
657
658<p>Global variables define regions of memory allocated at compilation time
659instead of run-time. Global variables may optionally be initialized, may have
660an explicit section to be placed in, and may have an optional explicit alignment
661specified. A variable may be defined as "thread_local", which means that it
662will not be shared by threads (each thread will have a separated copy of the
663variable). A variable may be defined as a global "constant," which indicates
664that the contents of the variable will <b>never</b> be modified (enabling better
665optimization, allowing the global data to be placed in the read-only section of
666an executable, etc). Note that variables that need runtime initialization
667cannot be marked "constant" as there is a store to the variable.</p>
668
669<p>
670LLVM explicitly allows <em>declarations</em> of global variables to be marked
671constant, even if the final definition of the global is not. This capability
672can be used to enable slightly better optimization of the program, but requires
673the language definition to guarantee that optimizations based on the
674'constantness' are valid for the translation units that do not include the
675definition.
676</p>
677
678<p>As SSA values, global variables define pointer values that are in
679scope (i.e. they dominate) all basic blocks in the program. Global
680variables always define a pointer to their "content" type because they
681describe a region of memory, and all memory objects in LLVM are
682accessed through pointers.</p>
683
Christopher Lambdd0049d2007-12-11 09:31:00 +0000[diff] [blame]684<p>A global variable may be declared to reside in a target-specifc numbered
685address space. For targets that support them, address spaces may affect how
686optimizations are performed and/or what target instructions are used to access
Christopher Lamb20a39e92007-12-12 08:44:39 +0000[diff] [blame]687the variable. The default address space is zero. The address space qualifier
688must precede any other attributes.</p>
Christopher Lambdd0049d2007-12-11 09:31:00 +0000[diff] [blame]689
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]690<p>LLVM allows an explicit section to be specified for globals. If the target
691supports it, it will emit globals to the section specified.</p>
692
693<p>An explicit alignment may be specified for a global. If not present, or if
694the alignment is set to zero, the alignment of the global is set by the target
695to whatever it feels convenient. If an explicit alignment is specified, the
696global is forced to have at least that much alignment. All alignments must be
697a power of 2.</p>
698
Christopher Lambdd0049d2007-12-11 09:31:00 +0000[diff] [blame]699<p>For example, the following defines a global in a numbered address space with
700an initializer, section, and alignment:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]701
702<div class="doc_code">
703<pre>
Christopher Lambdd0049d2007-12-11 09:31:00 +0000[diff] [blame]704@G = constant float 1.0 addrspace(5), section "foo", align 4
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]705</pre>
706</div>
707
708</div>
709
710
711<!-- ======================================================================= -->
712<div class="doc_subsection">
713 <a name="functionstructure">Functions</a>
714</div>
715
716<div class="doc_text">
717
718<p>LLVM function definitions consist of the "<tt>define</tt>" keyord,
719an optional <a href="#linkage">linkage type</a>, an optional
720<a href="#visibility">visibility style</a>, an optional
721<a href="#callingconv">calling convention</a>, a return type, an optional
722<a href="#paramattrs">parameter attribute</a> for the return type, a function
723name, a (possibly empty) argument list (each with optional
724<a href="#paramattrs">parameter attributes</a>), an optional section, an
Gordon Henriksend606f5b2007-12-10 03:30:21 +0000[diff] [blame]725optional alignment, an optional <a href="#gc">garbage collector name</a>, an
Gordon Henriksen13fe5e32007-12-10 03:18:06 +0000[diff] [blame]726opening curly brace, a list of basic blocks, and a closing curly brace.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]727
728LLVM function declarations consist of the "<tt>declare</tt>" keyword, an
729optional <a href="#linkage">linkage type</a>, an optional
730<a href="#visibility">visibility style</a>, an optional
731<a href="#callingconv">calling convention</a>, a return type, an optional
732<a href="#paramattrs">parameter attribute</a> for the return type, a function
Gordon Henriksen13fe5e32007-12-10 03:18:06 +0000[diff] [blame]733name, a possibly empty list of arguments, an optional alignment, and an optional
Gordon Henriksend606f5b2007-12-10 03:30:21 +0000[diff] [blame]734<a href="#gc">garbage collector name</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]735
736<p>A function definition contains a list of basic blocks, forming the CFG for
737the function. Each basic block may optionally start with a label (giving the
738basic block a symbol table entry), contains a list of instructions, and ends
739with a <a href="#terminators">terminator</a> instruction (such as a branch or
740function return).</p>
741
742<p>The first basic block in a function is special in two ways: it is immediately
743executed on entrance to the function, and it is not allowed to have predecessor
744basic blocks (i.e. there can not be any branches to the entry block of a
745function). Because the block can have no predecessors, it also cannot have any
746<a href="#i_phi">PHI nodes</a>.</p>
747
748<p>LLVM allows an explicit section to be specified for functions. If the target
749supports it, it will emit functions to the section specified.</p>
750
751<p>An explicit alignment may be specified for a function. If not present, or if
752the alignment is set to zero, the alignment of the function is set by the target
753to whatever it feels convenient. If an explicit alignment is specified, the
754function is forced to have at least that much alignment. All alignments must be
755a power of 2.</p>
756
757</div>
758
759
760<!-- ======================================================================= -->
761<div class="doc_subsection">
762 <a name="aliasstructure">Aliases</a>
763</div>
764<div class="doc_text">
765 <p>Aliases act as "second name" for the aliasee value (which can be either
766 function or global variable or bitcast of global value). Aliases may have an
767 optional <a href="#linkage">linkage type</a>, and an
768 optional <a href="#visibility">visibility style</a>.</p>
769
770 <h5>Syntax:</h5>
771
772<div class="doc_code">
773<pre>
774@&lt;Name&gt; = [Linkage] [Visibility] alias &lt;AliaseeTy&gt; @&lt;Aliasee&gt;
775</pre>
776</div>
777
778</div>
779
780
781
782<!-- ======================================================================= -->
783<div class="doc_subsection"><a name="paramattrs">Parameter Attributes</a></div>
784<div class="doc_text">
785 <p>The return type and each parameter of a function type may have a set of
786 <i>parameter attributes</i> associated with them. Parameter attributes are
787 used to communicate additional information about the result or parameters of
Duncan Sandsf5588dc2007-11-27 13:23:08 +0000[diff] [blame]788 a function. Parameter attributes are considered to be part of the function,
789 not of the function type, so functions with different parameter attributes
790 can have the same function type.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]791
792 <p>Parameter attributes are simple keywords that follow the type specified. If
793 multiple parameter attributes are needed, they are space separated. For
794 example:</p>
795
796<div class="doc_code">
797<pre>
Duncan Sandsf5588dc2007-11-27 13:23:08 +0000[diff] [blame]798declare i32 @printf(i8* noalias , ...) nounwind
799declare i32 @atoi(i8*) nounwind readonly
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]800</pre>
801</div>
802
Duncan Sandsf5588dc2007-11-27 13:23:08 +0000[diff] [blame]803 <p>Note that any attributes for the function result (<tt>nounwind</tt>,
804 <tt>readonly</tt>) come immediately after the argument list.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]805
806 <p>Currently, only the following parameter attributes are defined:</p>
807 <dl>
Reid Spencerf234bed2007-07-19 23:13:04 +0000[diff] [blame]808 <dt><tt>zeroext</tt></dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]809 <dd>This indicates that the parameter should be zero extended just before
810 a call to this function.</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]811
Reid Spencerf234bed2007-07-19 23:13:04 +0000[diff] [blame]812 <dt><tt>signext</tt></dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]813 <dd>This indicates that the parameter should be sign extended just before
814 a call to this function.</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]815
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]816 <dt><tt>inreg</tt></dt>
817 <dd>This indicates that the parameter should be placed in register (if
818 possible) during assembling function call. Support for this attribute is
819 target-specific</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]820
821 <dt><tt>byval</tt></dt>
Chris Lattner04c86182008-01-15 04:34:22 +0000[diff] [blame]822 <dd>This indicates that the pointer parameter should really be passed by
823 value to the function. The attribute implies that a hidden copy of the
824 pointee is made between the caller and the callee, so the callee is unable
825 to modify the value in the callee. This attribute is only valid on llvm
826 pointer arguments. It is generally used to pass structs and arrays by
827 value, but is also valid on scalars (even though this is silly).</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]828
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]829 <dt><tt>sret</tt></dt>
Duncan Sands616cc032008-02-18 04:19:38 +0000[diff] [blame]830 <dd>This indicates that the pointer parameter specifies the address of a
831 structure that is the return value of the function in the source program.
832 May only be applied to the first parameter.</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]833
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]834 <dt><tt>noalias</tt></dt>
Owen Andersonc4fc4cd2008-02-18 04:09:01 +0000[diff] [blame]835 <dd>This indicates that the parameter does not alias any global or any other
836 parameter. The caller is responsible for ensuring that this is the case,
837 usually by placing the value in a stack allocation.</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]838
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]839 <dt><tt>noreturn</tt></dt>
840 <dd>This function attribute indicates that the function never returns. This
841 indicates to LLVM that every call to this function should be treated as if
842 an <tt>unreachable</tt> instruction immediately followed the call.</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]843
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]844 <dt><tt>nounwind</tt></dt>
845 <dd>This function attribute indicates that the function type does not use
846 the unwind instruction and does not allow stack unwinding to propagate
847 through it.</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]848
Duncan Sands4ee46812007-07-27 19:57:41 +0000[diff] [blame]849 <dt><tt>nest</tt></dt>
850 <dd>This indicates that the parameter can be excised using the
851 <a href="#int_trampoline">trampoline intrinsics</a>.</dd>
Duncan Sands13e13f82007-11-22 20:23:04 +0000[diff] [blame]852 <dt><tt>readonly</tt></dt>
Duncan Sandsd69c0e62007-11-14 21:14:02 +0000[diff] [blame]853 <dd>This function attribute indicates that the function has no side-effects
Duncan Sands13e13f82007-11-22 20:23:04 +0000[diff] [blame]854 except for producing a return value or throwing an exception. The value
855 returned must only depend on the function arguments and/or global variables.
856 It may use values obtained by dereferencing pointers.</dd>
857 <dt><tt>readnone</tt></dt>
858 <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]859 function, but in addition it is not allowed to dereference any pointer arguments
860 or global variables.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]861 </dl>
862
863</div>
864
865<!-- ======================================================================= -->
866<div class="doc_subsection">
Gordon Henriksen13fe5e32007-12-10 03:18:06 +0000[diff] [blame]867 <a name="gc">Garbage Collector Names</a>
868</div>
869
870<div class="doc_text">
871<p>Each function may specify a garbage collector name, which is simply a
872string.</p>
873
874<div class="doc_code"><pre
875>define void @f() gc "name" { ...</pre></div>
876
877<p>The compiler declares the supported values of <i>name</i>. Specifying a
878collector which will cause the compiler to alter its output in order to support
879the named garbage collection algorithm.</p>
880</div>
881
882<!-- ======================================================================= -->
883<div class="doc_subsection">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]884 <a name="moduleasm">Module-Level Inline Assembly</a>
885</div>
886
887<div class="doc_text">
888<p>
889Modules may contain "module-level inline asm" blocks, which corresponds to the
890GCC "file scope inline asm" blocks. These blocks are internally concatenated by
891LLVM and treated as a single unit, but may be separated in the .ll file if
892desired. The syntax is very simple:
893</p>
894
895<div class="doc_code">
896<pre>
897module asm "inline asm code goes here"
898module asm "more can go here"
899</pre>
900</div>
901
902<p>The strings can contain any character by escaping non-printable characters.
903 The escape sequence used is simply "\xx" where "xx" is the two digit hex code
904 for the number.
905</p>
906
907<p>
908 The inline asm code is simply printed to the machine code .s file when
909 assembly code is generated.
910</p>
911</div>
912
913<!-- ======================================================================= -->
914<div class="doc_subsection">
915 <a name="datalayout">Data Layout</a>
916</div>
917
918<div class="doc_text">
919<p>A module may specify a target specific data layout string that specifies how
920data is to be laid out in memory. The syntax for the data layout is simply:</p>
921<pre> target datalayout = "<i>layout specification</i>"</pre>
922<p>The <i>layout specification</i> consists of a list of specifications
923separated by the minus sign character ('-'). Each specification starts with a
924letter and may include other information after the letter to define some
925aspect of the data layout. The specifications accepted are as follows: </p>
926<dl>
927 <dt><tt>E</tt></dt>
928 <dd>Specifies that the target lays out data in big-endian form. That is, the
929 bits with the most significance have the lowest address location.</dd>
930 <dt><tt>e</tt></dt>
931 <dd>Specifies that hte target lays out data in little-endian form. That is,
932 the bits with the least significance have the lowest address location.</dd>
933 <dt><tt>p:<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
934 <dd>This specifies the <i>size</i> of a pointer and its <i>abi</i> and
935 <i>preferred</i> alignments. All sizes are in bits. Specifying the <i>pref</i>
936 alignment is optional. If omitted, the preceding <tt>:</tt> should be omitted
937 too.</dd>
938 <dt><tt>i<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
939 <dd>This specifies the alignment for an integer type of a given bit
940 <i>size</i>. The value of <i>size</i> must be in the range [1,2^23).</dd>
941 <dt><tt>v<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
942 <dd>This specifies the alignment for a vector type of a given bit
943 <i>size</i>.</dd>
944 <dt><tt>f<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
945 <dd>This specifies the alignment for a floating point type of a given bit
946 <i>size</i>. The value of <i>size</i> must be either 32 (float) or 64
947 (double).</dd>
948 <dt><tt>a<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
949 <dd>This specifies the alignment for an aggregate type of a given bit
950 <i>size</i>.</dd>
951</dl>
952<p>When constructing the data layout for a given target, LLVM starts with a
953default set of specifications which are then (possibly) overriden by the
954specifications in the <tt>datalayout</tt> keyword. The default specifications
955are given in this list:</p>
956<ul>
957 <li><tt>E</tt> - big endian</li>
958 <li><tt>p:32:64:64</tt> - 32-bit pointers with 64-bit alignment</li>
959 <li><tt>i1:8:8</tt> - i1 is 8-bit (byte) aligned</li>
960 <li><tt>i8:8:8</tt> - i8 is 8-bit (byte) aligned</li>
961 <li><tt>i16:16:16</tt> - i16 is 16-bit aligned</li>
962 <li><tt>i32:32:32</tt> - i32 is 32-bit aligned</li>
963 <li><tt>i64:32:64</tt> - i64 has abi alignment of 32-bits but preferred
964 alignment of 64-bits</li>
965 <li><tt>f32:32:32</tt> - float is 32-bit aligned</li>
966 <li><tt>f64:64:64</tt> - double is 64-bit aligned</li>
967 <li><tt>v64:64:64</tt> - 64-bit vector is 64-bit aligned</li>
968 <li><tt>v128:128:128</tt> - 128-bit vector is 128-bit aligned</li>
969 <li><tt>a0:0:1</tt> - aggregates are 8-bit aligned</li>
970</ul>
971<p>When llvm is determining the alignment for a given type, it uses the
972following rules:
973<ol>
974 <li>If the type sought is an exact match for one of the specifications, that
975 specification is used.</li>
976 <li>If no match is found, and the type sought is an integer type, then the
977 smallest integer type that is larger than the bitwidth of the sought type is
978 used. If none of the specifications are larger than the bitwidth then the the
979 largest integer type is used. For example, given the default specifications
980 above, the i7 type will use the alignment of i8 (next largest) while both
981 i65 and i256 will use the alignment of i64 (largest specified).</li>
982 <li>If no match is found, and the type sought is a vector type, then the
983 largest vector type that is smaller than the sought vector type will be used
984 as a fall back. This happens because <128 x double> can be implemented in
985 terms of 64 <2 x double>, for example.</li>
986</ol>
987</div>
988
989<!-- *********************************************************************** -->
990<div class="doc_section"> <a name="typesystem">Type System</a> </div>
991<!-- *********************************************************************** -->
992
993<div class="doc_text">
994
995<p>The LLVM type system is one of the most important features of the
996intermediate representation. Being typed enables a number of
997optimizations to be performed on the IR directly, without having to do
998extra analyses on the side before the transformation. A strong type
999system makes it easier to read the generated code and enables novel
1000analyses and transformations that are not feasible to perform on normal
1001three address code representations.</p>
1002
1003</div>
1004
1005<!-- ======================================================================= -->
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1006<div class="doc_subsection"> <a name="t_classifications">Type
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1007Classifications</a> </div>
1008<div class="doc_text">
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1009<p>The types fall into a few useful
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1010classifications:</p>
1011
1012<table border="1" cellspacing="0" cellpadding="4">
1013 <tbody>
1014 <tr><th>Classification</th><th>Types</th></tr>
1015 <tr>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1016 <td><a href="#t_integer">integer</a></td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1017 <td><tt>i1, i2, i3, ... i8, ... i16, ... i32, ... i64, ... </tt></td>
1018 </tr>
1019 <tr>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1020 <td><a href="#t_floating">floating point</a></td>
1021 <td><tt>float, double, x86_fp80, fp128, ppc_fp128</tt></td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1022 </tr>
1023 <tr>
1024 <td><a name="t_firstclass">first class</a></td>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1025 <td><a href="#t_integer">integer</a>,
1026 <a href="#t_floating">floating point</a>,
1027 <a href="#t_pointer">pointer</a>,
1028 <a href="#t_vector">vector</a>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1029 </td>
1030 </tr>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1031 <tr>
1032 <td><a href="#t_primitive">primitive</a></td>
1033 <td><a href="#t_label">label</a>,
1034 <a href="#t_void">void</a>,
1035 <a href="#t_integer">integer</a>,
1036 <a href="#t_floating">floating point</a>.</td>
1037 </tr>
1038 <tr>
1039 <td><a href="#t_derived">derived</a></td>
1040 <td><a href="#t_integer">integer</a>,
1041 <a href="#t_array">array</a>,
1042 <a href="#t_function">function</a>,
1043 <a href="#t_pointer">pointer</a>,
1044 <a href="#t_struct">structure</a>,
1045 <a href="#t_pstruct">packed structure</a>,
1046 <a href="#t_vector">vector</a>,
1047 <a href="#t_opaque">opaque</a>.
1048 </tr>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1049 </tbody>
1050</table>
1051
1052<p>The <a href="#t_firstclass">first class</a> types are perhaps the
1053most important. Values of these types are the only ones which can be
1054produced by instructions, passed as arguments, or used as operands to
1055instructions. This means that all structures and arrays must be
1056manipulated either by pointer or by component.</p>
1057</div>
1058
1059<!-- ======================================================================= -->
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1060<div class="doc_subsection"> <a name="t_primitive">Primitive Types</a> </div>
Chris Lattner86437612008-01-04 04:34:14 +0000[diff] [blame]1061
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1062<div class="doc_text">
1063<p>The primitive types are the fundamental building blocks of the LLVM
1064system.</p>
1065
Chris Lattner86437612008-01-04 04:34:14 +0000[diff] [blame]1066</div>
1067
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1068<!-- _______________________________________________________________________ -->
1069<div class="doc_subsubsection"> <a name="t_floating">Floating Point Types</a> </div>
1070
1071<div class="doc_text">
1072 <table>
1073 <tbody>
1074 <tr><th>Type</th><th>Description</th></tr>
1075 <tr><td><tt>float</tt></td><td>32-bit floating point value</td></tr>
1076 <tr><td><tt>double</tt></td><td>64-bit floating point value</td></tr>
1077 <tr><td><tt>fp128</tt></td><td>128-bit floating point value (112-bit mantissa)</td></tr>
1078 <tr><td><tt>x86_fp80</tt></td><td>80-bit floating point value (X87)</td></tr>
1079 <tr><td><tt>ppc_fp128</tt></td><td>128-bit floating point value (two 64-bits)</td></tr>
1080 </tbody>
1081 </table>
1082</div>
1083
1084<!-- _______________________________________________________________________ -->
1085<div class="doc_subsubsection"> <a name="t_void">Void Type</a> </div>
1086
1087<div class="doc_text">
1088<h5>Overview:</h5>
1089<p>The void type does not represent any value and has no size.</p>
1090
1091<h5>Syntax:</h5>
1092
1093<pre>
1094 void
1095</pre>
1096</div>
1097
1098<!-- _______________________________________________________________________ -->
1099<div class="doc_subsubsection"> <a name="t_label">Label Type</a> </div>
1100
1101<div class="doc_text">
1102<h5>Overview:</h5>
1103<p>The label type represents code labels.</p>
1104
1105<h5>Syntax:</h5>
1106
1107<pre>
1108 label
1109</pre>
1110</div>
1111
1112
1113<!-- ======================================================================= -->
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1114<div class="doc_subsection"> <a name="t_derived">Derived Types</a> </div>
1115
1116<div class="doc_text">
1117
1118<p>The real power in LLVM comes from the derived types in the system.
1119This is what allows a programmer to represent arrays, functions,
1120pointers, and other useful types. Note that these derived types may be
1121recursive: For example, it is possible to have a two dimensional array.</p>
1122
1123</div>
1124
1125<!-- _______________________________________________________________________ -->
1126<div class="doc_subsubsection"> <a name="t_integer">Integer Type</a> </div>
1127
1128<div class="doc_text">
1129
1130<h5>Overview:</h5>
1131<p>The integer type is a very simple derived type that simply specifies an
1132arbitrary bit width for the integer type desired. Any bit width from 1 bit to
11332^23-1 (about 8 million) can be specified.</p>
1134
1135<h5>Syntax:</h5>
1136
1137<pre>
1138 iN
1139</pre>
1140
1141<p>The number of bits the integer will occupy is specified by the <tt>N</tt>
1142value.</p>
1143
1144<h5>Examples:</h5>
1145<table class="layout">
Chris Lattner251ab812007-12-18 06:18:21 +0000[diff] [blame]1146 <tbody>
1147 <tr>
1148 <td><tt>i1</tt></td>
1149 <td>a single-bit integer.</td>
1150 </tr><tr>
1151 <td><tt>i32</tt></td>
1152 <td>a 32-bit integer.</td>
1153 </tr><tr>
1154 <td><tt>i1942652</tt></td>
1155 <td>a really big integer of over 1 million bits.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1156 </tr>
Chris Lattner251ab812007-12-18 06:18:21 +0000[diff] [blame]1157 </tbody>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1158</table>
1159</div>
1160
1161<!-- _______________________________________________________________________ -->
1162<div class="doc_subsubsection"> <a name="t_array">Array Type</a> </div>
1163
1164<div class="doc_text">
1165
1166<h5>Overview:</h5>
1167
1168<p>The array type is a very simple derived type that arranges elements
1169sequentially in memory. The array type requires a size (number of
1170elements) and an underlying data type.</p>
1171
1172<h5>Syntax:</h5>
1173
1174<pre>
1175 [&lt;# elements&gt; x &lt;elementtype&gt;]
1176</pre>
1177
1178<p>The number of elements is a constant integer value; elementtype may
1179be any type with a size.</p>
1180
1181<h5>Examples:</h5>
1182<table class="layout">
1183 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1184 <td class="left"><tt>[40 x i32]</tt></td>
1185 <td class="left">Array of 40 32-bit integer values.</td>
1186 </tr>
1187 <tr class="layout">
1188 <td class="left"><tt>[41 x i32]</tt></td>
1189 <td class="left">Array of 41 32-bit integer values.</td>
1190 </tr>
1191 <tr class="layout">
1192 <td class="left"><tt>[4 x i8]</tt></td>
1193 <td class="left">Array of 4 8-bit integer values.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1194 </tr>
1195</table>
1196<p>Here are some examples of multidimensional arrays:</p>
1197<table class="layout">
1198 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1199 <td class="left"><tt>[3 x [4 x i32]]</tt></td>
1200 <td class="left">3x4 array of 32-bit integer values.</td>
1201 </tr>
1202 <tr class="layout">
1203 <td class="left"><tt>[12 x [10 x float]]</tt></td>
1204 <td class="left">12x10 array of single precision floating point values.</td>
1205 </tr>
1206 <tr class="layout">
1207 <td class="left"><tt>[2 x [3 x [4 x i16]]]</tt></td>
1208 <td class="left">2x3x4 array of 16-bit integer values.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1209 </tr>
1210</table>
1211
1212<p>Note that 'variable sized arrays' can be implemented in LLVM with a zero
1213length array. Normally, accesses past the end of an array are undefined in
1214LLVM (e.g. it is illegal to access the 5th element of a 3 element array).
1215As a special case, however, zero length arrays are recognized to be variable
1216length. This allows implementation of 'pascal style arrays' with the LLVM
1217type "{ i32, [0 x float]}", for example.</p>
1218
1219</div>
1220
1221<!-- _______________________________________________________________________ -->
1222<div class="doc_subsubsection"> <a name="t_function">Function Type</a> </div>
1223<div class="doc_text">
1224<h5>Overview:</h5>
1225<p>The function type can be thought of as a function signature. It
1226consists of a return type and a list of formal parameter types.
1227Function types are usually used to build virtual function tables
1228(which are structures of pointers to functions), for indirect function
1229calls, and when defining a function.</p>
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame^]1230
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1231<h5>Syntax:</h5>
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame^]1232<pre> &lt;returntype list&gt; (&lt;parameter list&gt;)<br></pre>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1233<p>...where '<tt>&lt;parameter list&gt;</tt>' is a comma-separated list of type
1234specifiers. Optionally, the parameter list may include a type <tt>...</tt>,
1235which indicates that the function takes a variable number of arguments.
1236Variable argument functions can access their arguments with the <a
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame^]1237 href="#int_varargs">variable argument handling intrinsic</a> functions.
1238'<tt>&lt;returntype list&gt;</tt>' is a comma-separated list of
1239<a href="#t_firstclass">first class</a> type specifiers.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1240<h5>Examples:</h5>
1241<table class="layout">
1242 <tr class="layout">
1243 <td class="left"><tt>i32 (i32)</tt></td>
1244 <td class="left">function taking an <tt>i32</tt>, returning an <tt>i32</tt>
1245 </td>
1246 </tr><tr class="layout">
Reid Spencerf234bed2007-07-19 23:13:04 +0000[diff] [blame]1247 <td class="left"><tt>float&nbsp;(i16&nbsp;signext,&nbsp;i32&nbsp;*)&nbsp;*
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1248 </tt></td>
1249 <td class="left"><a href="#t_pointer">Pointer</a> to a function that takes
1250 an <tt>i16</tt> that should be sign extended and a
1251 <a href="#t_pointer">pointer</a> to <tt>i32</tt>, returning
1252 <tt>float</tt>.
1253 </td>
1254 </tr><tr class="layout">
1255 <td class="left"><tt>i32 (i8*, ...)</tt></td>
1256 <td class="left">A vararg function that takes at least one
1257 <a href="#t_pointer">pointer</a> to <tt>i8 </tt> (char in C),
1258 which returns an integer. This is the signature for <tt>printf</tt> in
1259 LLVM.
1260 </td>
1261 </tr>
1262</table>
1263
1264</div>
1265<!-- _______________________________________________________________________ -->
1266<div class="doc_subsubsection"> <a name="t_struct">Structure Type</a> </div>
1267<div class="doc_text">
1268<h5>Overview:</h5>
1269<p>The structure type is used to represent a collection of data members
1270together in memory. The packing of the field types is defined to match
1271the ABI of the underlying processor. The elements of a structure may
1272be any type that has a size.</p>
1273<p>Structures are accessed using '<tt><a href="#i_load">load</a></tt>
1274and '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a
1275field with the '<tt><a href="#i_getelementptr">getelementptr</a></tt>'
1276instruction.</p>
1277<h5>Syntax:</h5>
1278<pre> { &lt;type list&gt; }<br></pre>
1279<h5>Examples:</h5>
1280<table class="layout">
1281 <tr class="layout">
1282 <td class="left"><tt>{ i32, i32, i32 }</tt></td>
1283 <td class="left">A triple of three <tt>i32</tt> values</td>
1284 </tr><tr class="layout">
1285 <td class="left"><tt>{&nbsp;float,&nbsp;i32&nbsp;(i32)&nbsp;*&nbsp;}</tt></td>
1286 <td class="left">A pair, where the first element is a <tt>float</tt> and the
1287 second element is a <a href="#t_pointer">pointer</a> to a
1288 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
1289 an <tt>i32</tt>.</td>
1290 </tr>
1291</table>
1292</div>
1293
1294<!-- _______________________________________________________________________ -->
1295<div class="doc_subsubsection"> <a name="t_pstruct">Packed Structure Type</a>
1296</div>
1297<div class="doc_text">
1298<h5>Overview:</h5>
1299<p>The packed structure type is used to represent a collection of data members
1300together in memory. There is no padding between fields. Further, the alignment
1301of a packed structure is 1 byte. The elements of a packed structure may
1302be any type that has a size.</p>
1303<p>Structures are accessed using '<tt><a href="#i_load">load</a></tt>
1304and '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a
1305field with the '<tt><a href="#i_getelementptr">getelementptr</a></tt>'
1306instruction.</p>
1307<h5>Syntax:</h5>
1308<pre> &lt; { &lt;type list&gt; } &gt; <br></pre>
1309<h5>Examples:</h5>
1310<table class="layout">
1311 <tr class="layout">
1312 <td class="left"><tt>&lt; { i32, i32, i32 } &gt;</tt></td>
1313 <td class="left">A triple of three <tt>i32</tt> values</td>
1314 </tr><tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1315 <td class="left"><tt>&lt; { float, i32 (i32)* } &gt;</tt></td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1316 <td class="left">A pair, where the first element is a <tt>float</tt> and the
1317 second element is a <a href="#t_pointer">pointer</a> to a
1318 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
1319 an <tt>i32</tt>.</td>
1320 </tr>
1321</table>
1322</div>
1323
1324<!-- _______________________________________________________________________ -->
1325<div class="doc_subsubsection"> <a name="t_pointer">Pointer Type</a> </div>
1326<div class="doc_text">
1327<h5>Overview:</h5>
1328<p>As in many languages, the pointer type represents a pointer or
Christopher Lambdd0049d2007-12-11 09:31:00 +0000[diff] [blame]1329reference to another object, which must live in memory. Pointer types may have
1330an optional address space attribute defining the target-specific numbered
1331address space where the pointed-to object resides. The default address space is
1332zero.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1333<h5>Syntax:</h5>
1334<pre> &lt;type&gt; *<br></pre>
1335<h5>Examples:</h5>
1336<table class="layout">
1337 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1338 <td class="left"><tt>[4x i32]*</tt></td>
1339 <td class="left">A <a href="#t_pointer">pointer</a> to <a
1340 href="#t_array">array</a> of four <tt>i32</tt> values.</td>
1341 </tr>
1342 <tr class="layout">
1343 <td class="left"><tt>i32 (i32 *) *</tt></td>
1344 <td class="left"> A <a href="#t_pointer">pointer</a> to a <a
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1345 href="#t_function">function</a> that takes an <tt>i32*</tt>, returning an
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1346 <tt>i32</tt>.</td>
1347 </tr>
1348 <tr class="layout">
1349 <td class="left"><tt>i32 addrspace(5)*</tt></td>
1350 <td class="left">A <a href="#t_pointer">pointer</a> to an <tt>i32</tt> value
1351 that resides in address space #5.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1352 </tr>
1353</table>
1354</div>
1355
1356<!-- _______________________________________________________________________ -->
1357<div class="doc_subsubsection"> <a name="t_vector">Vector Type</a> </div>
1358<div class="doc_text">
1359
1360<h5>Overview:</h5>
1361
1362<p>A vector type is a simple derived type that represents a vector
1363of elements. Vector types are used when multiple primitive data
1364are operated in parallel using a single instruction (SIMD).
1365A vector type requires a size (number of
1366elements) and an underlying primitive data type. Vectors must have a power
1367of two length (1, 2, 4, 8, 16 ...). Vector types are
1368considered <a href="#t_firstclass">first class</a>.</p>
1369
1370<h5>Syntax:</h5>
1371
1372<pre>
1373 &lt; &lt;# elements&gt; x &lt;elementtype&gt; &gt;
1374</pre>
1375
1376<p>The number of elements is a constant integer value; elementtype may
1377be any integer or floating point type.</p>
1378
1379<h5>Examples:</h5>
1380
1381<table class="layout">
1382 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1383 <td class="left"><tt>&lt;4 x i32&gt;</tt></td>
1384 <td class="left">Vector of 4 32-bit integer values.</td>
1385 </tr>
1386 <tr class="layout">
1387 <td class="left"><tt>&lt;8 x float&gt;</tt></td>
1388 <td class="left">Vector of 8 32-bit floating-point values.</td>
1389 </tr>
1390 <tr class="layout">
1391 <td class="left"><tt>&lt;2 x i64&gt;</tt></td>
1392 <td class="left">Vector of 2 64-bit integer values.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1393 </tr>
1394</table>
1395</div>
1396
1397<!-- _______________________________________________________________________ -->
1398<div class="doc_subsubsection"> <a name="t_opaque">Opaque Type</a> </div>
1399<div class="doc_text">
1400
1401<h5>Overview:</h5>
1402
1403<p>Opaque types are used to represent unknown types in the system. This
Gordon Henriksenda0706e2007-10-14 00:34:53 +0000[diff] [blame]1404corresponds (for example) to the C notion of a forward declared structure type.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1405In LLVM, opaque types can eventually be resolved to any type (not just a
1406structure type).</p>
1407
1408<h5>Syntax:</h5>
1409
1410<pre>
1411 opaque
1412</pre>
1413
1414<h5>Examples:</h5>
1415
1416<table class="layout">
1417 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1418 <td class="left"><tt>opaque</tt></td>
1419 <td class="left">An opaque type.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1420 </tr>
1421</table>
1422</div>
1423
1424
1425<!-- *********************************************************************** -->
1426<div class="doc_section"> <a name="constants">Constants</a> </div>
1427<!-- *********************************************************************** -->
1428
1429<div class="doc_text">
1430
1431<p>LLVM has several different basic types of constants. This section describes
1432them all and their syntax.</p>
1433
1434</div>
1435
1436<!-- ======================================================================= -->
1437<div class="doc_subsection"><a name="simpleconstants">Simple Constants</a></div>
1438
1439<div class="doc_text">
1440
1441<dl>
1442 <dt><b>Boolean constants</b></dt>
1443
1444 <dd>The two strings '<tt>true</tt>' and '<tt>false</tt>' are both valid
1445 constants of the <tt><a href="#t_primitive">i1</a></tt> type.
1446 </dd>
1447
1448 <dt><b>Integer constants</b></dt>
1449
1450 <dd>Standard integers (such as '4') are constants of the <a
1451 href="#t_integer">integer</a> type. Negative numbers may be used with
1452 integer types.
1453 </dd>
1454
1455 <dt><b>Floating point constants</b></dt>
1456
1457 <dd>Floating point constants use standard decimal notation (e.g. 123.421),
1458 exponential notation (e.g. 1.23421e+2), or a more precise hexadecimal
1459 notation (see below). Floating point constants must have a <a
1460 href="#t_floating">floating point</a> type. </dd>
1461
1462 <dt><b>Null pointer constants</b></dt>
1463
1464 <dd>The identifier '<tt>null</tt>' is recognized as a null pointer constant
1465 and must be of <a href="#t_pointer">pointer type</a>.</dd>
1466
1467</dl>
1468
1469<p>The one non-intuitive notation for constants is the optional hexadecimal form
1470of floating point constants. For example, the form '<tt>double
14710x432ff973cafa8000</tt>' is equivalent to (but harder to read than) '<tt>double
14724.5e+15</tt>'. The only time hexadecimal floating point constants are required
1473(and the only time that they are generated by the disassembler) is when a
1474floating point constant must be emitted but it cannot be represented as a
1475decimal floating point number. For example, NaN's, infinities, and other
1476special values are represented in their IEEE hexadecimal format so that
1477assembly and disassembly do not cause any bits to change in the constants.</p>
1478
1479</div>
1480
1481<!-- ======================================================================= -->
1482<div class="doc_subsection"><a name="aggregateconstants">Aggregate Constants</a>
1483</div>
1484
1485<div class="doc_text">
1486<p>Aggregate constants arise from aggregation of simple constants
1487and smaller aggregate constants.</p>
1488
1489<dl>
1490 <dt><b>Structure constants</b></dt>
1491
1492 <dd>Structure constants are represented with notation similar to structure
1493 type definitions (a comma separated list of elements, surrounded by braces
Chris Lattner6c8de962007-12-25 20:34:52 +0000[diff] [blame]1494 (<tt>{}</tt>)). For example: "<tt>{ i32 4, float 17.0, i32* @G }</tt>",
1495 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]1496 must have <a href="#t_struct">structure type</a>, and the number and
1497 types of elements must match those specified by the type.
1498 </dd>
1499
1500 <dt><b>Array constants</b></dt>
1501
1502 <dd>Array constants are represented with notation similar to array type
1503 definitions (a comma separated list of elements, surrounded by square brackets
1504 (<tt>[]</tt>)). For example: "<tt>[ i32 42, i32 11, i32 74 ]</tt>". Array
1505 constants must have <a href="#t_array">array type</a>, and the number and
1506 types of elements must match those specified by the type.
1507 </dd>
1508
1509 <dt><b>Vector constants</b></dt>
1510
1511 <dd>Vector constants are represented with notation similar to vector type
1512 definitions (a comma separated list of elements, surrounded by
1513 less-than/greater-than's (<tt>&lt;&gt;</tt>)). For example: "<tt>&lt; i32 42,
1514 i32 11, i32 74, i32 100 &gt;</tt>". Vector constants must have <a
1515 href="#t_vector">vector type</a>, and the number and types of elements must
1516 match those specified by the type.
1517 </dd>
1518
1519 <dt><b>Zero initialization</b></dt>
1520
1521 <dd>The string '<tt>zeroinitializer</tt>' can be used to zero initialize a
1522 value to zero of <em>any</em> type, including scalar and aggregate types.
1523 This is often used to avoid having to print large zero initializers (e.g. for
1524 large arrays) and is always exactly equivalent to using explicit zero
1525 initializers.
1526 </dd>
1527</dl>
1528
1529</div>
1530
1531<!-- ======================================================================= -->
1532<div class="doc_subsection">
1533 <a name="globalconstants">Global Variable and Function Addresses</a>
1534</div>
1535
1536<div class="doc_text">
1537
1538<p>The addresses of <a href="#globalvars">global variables</a> and <a
1539href="#functionstructure">functions</a> are always implicitly valid (link-time)
1540constants. These constants are explicitly referenced when the <a
1541href="#identifiers">identifier for the global</a> is used and always have <a
1542href="#t_pointer">pointer</a> type. For example, the following is a legal LLVM
1543file:</p>
1544
1545<div class="doc_code">
1546<pre>
1547@X = global i32 17
1548@Y = global i32 42
1549@Z = global [2 x i32*] [ i32* @X, i32* @Y ]
1550</pre>
1551</div>
1552
1553</div>
1554
1555<!-- ======================================================================= -->
1556<div class="doc_subsection"><a name="undefvalues">Undefined Values</a></div>
1557<div class="doc_text">
1558 <p>The string '<tt>undef</tt>' is recognized as a type-less constant that has
1559 no specific value. Undefined values may be of any type and be used anywhere
1560 a constant is permitted.</p>
1561
1562 <p>Undefined values indicate to the compiler that the program is well defined
1563 no matter what value is used, giving the compiler more freedom to optimize.
1564 </p>
1565</div>
1566
1567<!-- ======================================================================= -->
1568<div class="doc_subsection"><a name="constantexprs">Constant Expressions</a>
1569</div>
1570
1571<div class="doc_text">
1572
1573<p>Constant expressions are used to allow expressions involving other constants
1574to be used as constants. Constant expressions may be of any <a
1575href="#t_firstclass">first class</a> type and may involve any LLVM operation
1576that does not have side effects (e.g. load and call are not supported). The
1577following is the syntax for constant expressions:</p>
1578
1579<dl>
1580 <dt><b><tt>trunc ( CST to TYPE )</tt></b></dt>
1581 <dd>Truncate a constant to another type. The bit size of CST must be larger
1582 than the bit size of TYPE. Both types must be integers.</dd>
1583
1584 <dt><b><tt>zext ( CST to TYPE )</tt></b></dt>
1585 <dd>Zero extend a constant to another type. The bit size of CST must be
1586 smaller or equal to the bit size of TYPE. Both types must be integers.</dd>
1587
1588 <dt><b><tt>sext ( CST to TYPE )</tt></b></dt>
1589 <dd>Sign extend a constant to another type. The bit size of CST must be
1590 smaller or equal to the bit size of TYPE. Both types must be integers.</dd>
1591
1592 <dt><b><tt>fptrunc ( CST to TYPE )</tt></b></dt>
1593 <dd>Truncate a floating point constant to another floating point type. The
1594 size of CST must be larger than the size of TYPE. Both types must be
1595 floating point.</dd>
1596
1597 <dt><b><tt>fpext ( CST to TYPE )</tt></b></dt>
1598 <dd>Floating point extend a constant to another type. The size of CST must be
1599 smaller or equal to the size of TYPE. Both types must be floating point.</dd>
1600
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]1601 <dt><b><tt>fptoui ( CST to TYPE )</tt></b></dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1602 <dd>Convert a floating point constant to the corresponding unsigned integer
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]1603 constant. TYPE must be a scalar or vector integer type. CST must be of scalar
1604 or vector floating point type. Both CST and TYPE must be scalars, or vectors
1605 of the same number of elements. If the value won't fit in the integer type,
1606 the results are undefined.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1607
1608 <dt><b><tt>fptosi ( CST to TYPE )</tt></b></dt>
1609 <dd>Convert a floating point constant to the corresponding signed integer
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]1610 constant. TYPE must be a scalar or vector integer type. CST must be of scalar
1611 or vector floating point type. Both CST and TYPE must be scalars, or vectors
1612 of the same number of elements. If the value won't fit in the integer type,
1613 the results are undefined.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1614
1615 <dt><b><tt>uitofp ( CST to TYPE )</tt></b></dt>
1616 <dd>Convert an unsigned integer constant to the corresponding floating point
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]1617 constant. TYPE must be a scalar or vector floating point type. CST must be of
1618 scalar or vector integer type. Both CST and TYPE must be scalars, or vectors
1619 of the same number of elements. If the value won't fit in the floating point
1620 type, the results are undefined.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1621
1622 <dt><b><tt>sitofp ( CST to TYPE )</tt></b></dt>
1623 <dd>Convert a signed integer constant to the corresponding floating point
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]1624 constant. TYPE must be a scalar or vector floating point type. CST must be of
1625 scalar or vector integer type. Both CST and TYPE must be scalars, or vectors
1626 of the same number of elements. If the value won't fit in the floating point
1627 type, the results are undefined.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1628
1629 <dt><b><tt>ptrtoint ( CST to TYPE )</tt></b></dt>
1630 <dd>Convert a pointer typed constant to the corresponding integer constant
1631 TYPE must be an integer type. CST must be of pointer type. The CST value is
1632 zero extended, truncated, or unchanged to make it fit in TYPE.</dd>
1633
1634 <dt><b><tt>inttoptr ( CST to TYPE )</tt></b></dt>
1635 <dd>Convert a integer constant to a pointer constant. TYPE must be a
1636 pointer type. CST must be of integer type. The CST value is zero extended,
1637 truncated, or unchanged to make it fit in a pointer size. This one is
1638 <i>really</i> dangerous!</dd>
1639
1640 <dt><b><tt>bitcast ( CST to TYPE )</tt></b></dt>
1641 <dd>Convert a constant, CST, to another TYPE. The size of CST and TYPE must be
1642 identical (same number of bits). The conversion is done as if the CST value
1643 was stored to memory and read back as TYPE. In other words, no bits change
1644 with this operator, just the type. This can be used for conversion of
1645 vector types to any other type, as long as they have the same bit width. For
1646 pointers it is only valid to cast to another pointer type.
1647 </dd>
1648
1649 <dt><b><tt>getelementptr ( CSTPTR, IDX0, IDX1, ... )</tt></b></dt>
1650
1651 <dd>Perform the <a href="#i_getelementptr">getelementptr operation</a> on
1652 constants. As with the <a href="#i_getelementptr">getelementptr</a>
1653 instruction, the index list may have zero or more indexes, which are required
1654 to make sense for the type of "CSTPTR".</dd>
1655
1656 <dt><b><tt>select ( COND, VAL1, VAL2 )</tt></b></dt>
1657
1658 <dd>Perform the <a href="#i_select">select operation</a> on
1659 constants.</dd>
1660
1661 <dt><b><tt>icmp COND ( VAL1, VAL2 )</tt></b></dt>
1662 <dd>Performs the <a href="#i_icmp">icmp operation</a> on constants.</dd>
1663
1664 <dt><b><tt>fcmp COND ( VAL1, VAL2 )</tt></b></dt>
1665 <dd>Performs the <a href="#i_fcmp">fcmp operation</a> on constants.</dd>
1666
1667 <dt><b><tt>extractelement ( VAL, IDX )</tt></b></dt>
1668
1669 <dd>Perform the <a href="#i_extractelement">extractelement
1670 operation</a> on constants.
1671
1672 <dt><b><tt>insertelement ( VAL, ELT, IDX )</tt></b></dt>
1673
1674 <dd>Perform the <a href="#i_insertelement">insertelement
1675 operation</a> on constants.</dd>
1676
1677
1678 <dt><b><tt>shufflevector ( VEC1, VEC2, IDXMASK )</tt></b></dt>
1679
1680 <dd>Perform the <a href="#i_shufflevector">shufflevector
1681 operation</a> on constants.</dd>
1682
1683 <dt><b><tt>OPCODE ( LHS, RHS )</tt></b></dt>
1684
1685 <dd>Perform the specified operation of the LHS and RHS constants. OPCODE may
1686 be any of the <a href="#binaryops">binary</a> or <a href="#bitwiseops">bitwise
1687 binary</a> operations. The constraints on operands are the same as those for
1688 the corresponding instruction (e.g. no bitwise operations on floating point
1689 values are allowed).</dd>
1690</dl>
1691</div>
1692
1693<!-- *********************************************************************** -->
1694<div class="doc_section"> <a name="othervalues">Other Values</a> </div>
1695<!-- *********************************************************************** -->
1696
1697<!-- ======================================================================= -->
1698<div class="doc_subsection">
1699<a name="inlineasm">Inline Assembler Expressions</a>
1700</div>
1701
1702<div class="doc_text">
1703
1704<p>
1705LLVM supports inline assembler expressions (as opposed to <a href="#moduleasm">
1706Module-Level Inline Assembly</a>) through the use of a special value. This
1707value represents the inline assembler as a string (containing the instructions
1708to emit), a list of operand constraints (stored as a string), and a flag that
1709indicates whether or not the inline asm expression has side effects. An example
1710inline assembler expression is:
1711</p>
1712
1713<div class="doc_code">
1714<pre>
1715i32 (i32) asm "bswap $0", "=r,r"
1716</pre>
1717</div>
1718
1719<p>
1720Inline assembler expressions may <b>only</b> be used as the callee operand of
1721a <a href="#i_call"><tt>call</tt> instruction</a>. Thus, typically we have:
1722</p>
1723
1724<div class="doc_code">
1725<pre>
1726%X = call i32 asm "<a href="#int_bswap">bswap</a> $0", "=r,r"(i32 %Y)
1727</pre>
1728</div>
1729
1730<p>
1731Inline asms with side effects not visible in the constraint list must be marked
1732as having side effects. This is done through the use of the
1733'<tt>sideeffect</tt>' keyword, like so:
1734</p>
1735
1736<div class="doc_code">
1737<pre>
1738call void asm sideeffect "eieio", ""()
1739</pre>
1740</div>
1741
1742<p>TODO: The format of the asm and constraints string still need to be
1743documented here. Constraints on what can be done (e.g. duplication, moving, etc
1744need to be documented).
1745</p>
1746
1747</div>
1748
1749<!-- *********************************************************************** -->
1750<div class="doc_section"> <a name="instref">Instruction Reference</a> </div>
1751<!-- *********************************************************************** -->
1752
1753<div class="doc_text">
1754
1755<p>The LLVM instruction set consists of several different
1756classifications of instructions: <a href="#terminators">terminator
1757instructions</a>, <a href="#binaryops">binary instructions</a>,
1758<a href="#bitwiseops">bitwise binary instructions</a>, <a
1759 href="#memoryops">memory instructions</a>, and <a href="#otherops">other
1760instructions</a>.</p>
1761
1762</div>
1763
1764<!-- ======================================================================= -->
1765<div class="doc_subsection"> <a name="terminators">Terminator
1766Instructions</a> </div>
1767
1768<div class="doc_text">
1769
1770<p>As mentioned <a href="#functionstructure">previously</a>, every
1771basic block in a program ends with a "Terminator" instruction, which
1772indicates which block should be executed after the current block is
1773finished. These terminator instructions typically yield a '<tt>void</tt>'
1774value: they produce control flow, not values (the one exception being
1775the '<a href="#i_invoke"><tt>invoke</tt></a>' instruction).</p>
1776<p>There are six different terminator instructions: the '<a
1777 href="#i_ret"><tt>ret</tt></a>' instruction, the '<a href="#i_br"><tt>br</tt></a>'
1778instruction, the '<a href="#i_switch"><tt>switch</tt></a>' instruction,
1779the '<a href="#i_invoke"><tt>invoke</tt></a>' instruction, the '<a
1780 href="#i_unwind"><tt>unwind</tt></a>' instruction, and the '<a
1781 href="#i_unreachable"><tt>unreachable</tt></a>' instruction.</p>
1782
1783</div>
1784
1785<!-- _______________________________________________________________________ -->
1786<div class="doc_subsubsection"> <a name="i_ret">'<tt>ret</tt>'
1787Instruction</a> </div>
1788<div class="doc_text">
1789<h5>Syntax:</h5>
1790<pre> ret &lt;type&gt; &lt;value&gt; <i>; Return a value from a non-void function</i>
1791 ret void <i>; Return from void function</i>
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame^]1792 ret &lt;type&gt; &lt;value&gt;, &lt;type&gt; &lt;value&gt; <i>; Return two values from a non-void function </i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1793</pre>
1794<h5>Overview:</h5>
1795<p>The '<tt>ret</tt>' instruction is used to return control flow (and a
1796value) from a function back to the caller.</p>
1797<p>There are two forms of the '<tt>ret</tt>' instruction: one that
1798returns a value and then causes control flow, and one that just causes
1799control flow to occur.</p>
1800<h5>Arguments:</h5>
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame^]1801<p>The '<tt>ret</tt>' instruction may return one or multiple values. The
1802type of each return value must be '<a href="#t_firstclass">first class</a>'
1803 type. Notice that a function is not <a href="#wellformed">well formed</a>
1804if there exists a '<tt>ret</tt>' instruction inside of the function that
1805returns values that does not match the return type of the function.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1806<h5>Semantics:</h5>
1807<p>When the '<tt>ret</tt>' instruction is executed, control flow
1808returns back to the calling function's context. If the caller is a "<a
1809 href="#i_call"><tt>call</tt></a>" instruction, execution continues at
1810the instruction after the call. If the caller was an "<a
1811 href="#i_invoke"><tt>invoke</tt></a>" instruction, execution continues
1812at the beginning of the "normal" destination block. If the instruction
1813returns a value, that value shall set the call or invoke instruction's
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame^]1814return value. If the instruction returns multiple values then these
1815value can only be accessed through '<a href="#i_getresult"><tt>getresult</tt>
1816</a>' insctruction.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1817<h5>Example:</h5>
1818<pre> ret i32 5 <i>; Return an integer value of 5</i>
1819 ret void <i>; Return from a void function</i>
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame^]1820 ret i32 4, i8 2 <i>; Return two values 4 and 2 </i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1821</pre>
1822</div>
1823<!-- _______________________________________________________________________ -->
1824<div class="doc_subsubsection"> <a name="i_br">'<tt>br</tt>' Instruction</a> </div>
1825<div class="doc_text">
1826<h5>Syntax:</h5>
1827<pre> br i1 &lt;cond&gt;, label &lt;iftrue&gt;, label &lt;iffalse&gt;<br> br label &lt;dest&gt; <i>; Unconditional branch</i>
1828</pre>
1829<h5>Overview:</h5>
1830<p>The '<tt>br</tt>' instruction is used to cause control flow to
1831transfer to a different basic block in the current function. There are
1832two forms of this instruction, corresponding to a conditional branch
1833and an unconditional branch.</p>
1834<h5>Arguments:</h5>
1835<p>The conditional branch form of the '<tt>br</tt>' instruction takes a
1836single '<tt>i1</tt>' value and two '<tt>label</tt>' values. The
1837unconditional form of the '<tt>br</tt>' instruction takes a single
1838'<tt>label</tt>' value as a target.</p>
1839<h5>Semantics:</h5>
1840<p>Upon execution of a conditional '<tt>br</tt>' instruction, the '<tt>i1</tt>'
1841argument is evaluated. If the value is <tt>true</tt>, control flows
1842to the '<tt>iftrue</tt>' <tt>label</tt> argument. If "cond" is <tt>false</tt>,
1843control flows to the '<tt>iffalse</tt>' <tt>label</tt> argument.</p>
1844<h5>Example:</h5>
1845<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
1846 href="#i_ret">ret</a> i32 1<br>IfUnequal:<br> <a href="#i_ret">ret</a> i32 0<br></pre>
1847</div>
1848<!-- _______________________________________________________________________ -->
1849<div class="doc_subsubsection">
1850 <a name="i_switch">'<tt>switch</tt>' Instruction</a>
1851</div>
1852
1853<div class="doc_text">
1854<h5>Syntax:</h5>
1855
1856<pre>
1857 switch &lt;intty&gt; &lt;value&gt;, label &lt;defaultdest&gt; [ &lt;intty&gt; &lt;val&gt;, label &lt;dest&gt; ... ]
1858</pre>
1859
1860<h5>Overview:</h5>
1861
1862<p>The '<tt>switch</tt>' instruction is used to transfer control flow to one of
1863several different places. It is a generalization of the '<tt>br</tt>'
1864instruction, allowing a branch to occur to one of many possible
1865destinations.</p>
1866
1867
1868<h5>Arguments:</h5>
1869
1870<p>The '<tt>switch</tt>' instruction uses three parameters: an integer
1871comparison value '<tt>value</tt>', a default '<tt>label</tt>' destination, and
1872an array of pairs of comparison value constants and '<tt>label</tt>'s. The
1873table is not allowed to contain duplicate constant entries.</p>
1874
1875<h5>Semantics:</h5>
1876
1877<p>The <tt>switch</tt> instruction specifies a table of values and
1878destinations. When the '<tt>switch</tt>' instruction is executed, this
1879table is searched for the given value. If the value is found, control flow is
1880transfered to the corresponding destination; otherwise, control flow is
1881transfered to the default destination.</p>
1882
1883<h5>Implementation:</h5>
1884
1885<p>Depending on properties of the target machine and the particular
1886<tt>switch</tt> instruction, this instruction may be code generated in different
1887ways. For example, it could be generated as a series of chained conditional
1888branches or with a lookup table.</p>
1889
1890<h5>Example:</h5>
1891
1892<pre>
1893 <i>; Emulate a conditional br instruction</i>
1894 %Val = <a href="#i_zext">zext</a> i1 %value to i32
1895 switch i32 %Val, label %truedest [i32 0, label %falsedest ]
1896
1897 <i>; Emulate an unconditional br instruction</i>
1898 switch i32 0, label %dest [ ]
1899
1900 <i>; Implement a jump table:</i>
1901 switch i32 %val, label %otherwise [ i32 0, label %onzero
1902 i32 1, label %onone
1903 i32 2, label %ontwo ]
1904</pre>
1905</div>
1906
1907<!-- _______________________________________________________________________ -->
1908<div class="doc_subsubsection">
1909 <a name="i_invoke">'<tt>invoke</tt>' Instruction</a>
1910</div>
1911
1912<div class="doc_text">
1913
1914<h5>Syntax:</h5>
1915
1916<pre>
1917 &lt;result&gt; = invoke [<a href="#callingconv">cconv</a>] &lt;ptr to function ty&gt; %&lt;function ptr val&gt;(&lt;function args&gt;)
1918 to label &lt;normal label&gt; unwind label &lt;exception label&gt;
1919</pre>
1920
1921<h5>Overview:</h5>
1922
1923<p>The '<tt>invoke</tt>' instruction causes control to transfer to a specified
1924function, with the possibility of control flow transfer to either the
1925'<tt>normal</tt>' label or the
1926'<tt>exception</tt>' label. If the callee function returns with the
1927"<tt><a href="#i_ret">ret</a></tt>" instruction, control flow will return to the
1928"normal" label. If the callee (or any indirect callees) returns with the "<a
1929href="#i_unwind"><tt>unwind</tt></a>" instruction, control is interrupted and
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame^]1930continued at the dynamically nearest "exception" label. If the callee function
1931returns multiple values then individual return values are accessed only through
1932'<tt><a href="#i_getresult">getresult</a></tt>' instruction.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1933
1934<h5>Arguments:</h5>
1935
1936<p>This instruction requires several arguments:</p>
1937
1938<ol>
1939 <li>
1940 The optional "cconv" marker indicates which <a href="#callingconv">calling
1941 convention</a> the call should use. If none is specified, the call defaults
1942 to using C calling conventions.
1943 </li>
1944 <li>'<tt>ptr to function ty</tt>': shall be the signature of the pointer to
1945 function value being invoked. In most cases, this is a direct function
1946 invocation, but indirect <tt>invoke</tt>s are just as possible, branching off
1947 an arbitrary pointer to function value.
1948 </li>
1949
1950 <li>'<tt>function ptr val</tt>': An LLVM value containing a pointer to a
1951 function to be invoked. </li>
1952
1953 <li>'<tt>function args</tt>': argument list whose types match the function
1954 signature argument types. If the function signature indicates the function
1955 accepts a variable number of arguments, the extra arguments can be
1956 specified. </li>
1957
1958 <li>'<tt>normal label</tt>': the label reached when the called function
1959 executes a '<tt><a href="#i_ret">ret</a></tt>' instruction. </li>
1960
1961 <li>'<tt>exception label</tt>': the label reached when a callee returns with
1962 the <a href="#i_unwind"><tt>unwind</tt></a> instruction. </li>
1963
1964</ol>
1965
1966<h5>Semantics:</h5>
1967
1968<p>This instruction is designed to operate as a standard '<tt><a
1969href="#i_call">call</a></tt>' instruction in most regards. The primary
1970difference is that it establishes an association with a label, which is used by
1971the runtime library to unwind the stack.</p>
1972
1973<p>This instruction is used in languages with destructors to ensure that proper
1974cleanup is performed in the case of either a <tt>longjmp</tt> or a thrown
1975exception. Additionally, this is important for implementation of
1976'<tt>catch</tt>' clauses in high-level languages that support them.</p>
1977
1978<h5>Example:</h5>
1979<pre>
1980 %retval = invoke i32 %Test(i32 15) to label %Continue
1981 unwind label %TestCleanup <i>; {i32}:retval set</i>
1982 %retval = invoke <a href="#callingconv">coldcc</a> i32 %Test(i32 15) to label %Continue
1983 unwind label %TestCleanup <i>; {i32}:retval set</i>
1984</pre>
1985</div>
1986
1987
1988<!-- _______________________________________________________________________ -->
1989
1990<div class="doc_subsubsection"> <a name="i_unwind">'<tt>unwind</tt>'
1991Instruction</a> </div>
1992
1993<div class="doc_text">
1994
1995<h5>Syntax:</h5>
1996<pre>
1997 unwind
1998</pre>
1999
2000<h5>Overview:</h5>
2001
2002<p>The '<tt>unwind</tt>' instruction unwinds the stack, continuing control flow
2003at the first callee in the dynamic call stack which used an <a
2004href="#i_invoke"><tt>invoke</tt></a> instruction to perform the call. This is
2005primarily used to implement exception handling.</p>
2006
2007<h5>Semantics:</h5>
2008
2009<p>The '<tt>unwind</tt>' intrinsic causes execution of the current function to
2010immediately halt. The dynamic call stack is then searched for the first <a
2011href="#i_invoke"><tt>invoke</tt></a> instruction on the call stack. Once found,
2012execution continues at the "exceptional" destination block specified by the
2013<tt>invoke</tt> instruction. If there is no <tt>invoke</tt> instruction in the
2014dynamic call chain, undefined behavior results.</p>
2015</div>
2016
2017<!-- _______________________________________________________________________ -->
2018
2019<div class="doc_subsubsection"> <a name="i_unreachable">'<tt>unreachable</tt>'
2020Instruction</a> </div>
2021
2022<div class="doc_text">
2023
2024<h5>Syntax:</h5>
2025<pre>
2026 unreachable
2027</pre>
2028
2029<h5>Overview:</h5>
2030
2031<p>The '<tt>unreachable</tt>' instruction has no defined semantics. This
2032instruction is used to inform the optimizer that a particular portion of the
2033code is not reachable. This can be used to indicate that the code after a
2034no-return function cannot be reached, and other facts.</p>
2035
2036<h5>Semantics:</h5>
2037
2038<p>The '<tt>unreachable</tt>' instruction has no defined semantics.</p>
2039</div>
2040
2041
2042
2043<!-- ======================================================================= -->
2044<div class="doc_subsection"> <a name="binaryops">Binary Operations</a> </div>
2045<div class="doc_text">
2046<p>Binary operators are used to do most of the computation in a
2047program. They require two operands, execute an operation on them, and
2048produce a single value. The operands might represent
2049multiple data, as is the case with the <a href="#t_vector">vector</a> data type.
2050The result value of a binary operator is not
2051necessarily the same type as its operands.</p>
2052<p>There are several different binary operators:</p>
2053</div>
2054<!-- _______________________________________________________________________ -->
2055<div class="doc_subsubsection"> <a name="i_add">'<tt>add</tt>'
2056Instruction</a> </div>
2057<div class="doc_text">
2058<h5>Syntax:</h5>
2059<pre> &lt;result&gt; = add &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
2060</pre>
2061<h5>Overview:</h5>
2062<p>The '<tt>add</tt>' instruction returns the sum of its two operands.</p>
2063<h5>Arguments:</h5>
2064<p>The two arguments to the '<tt>add</tt>' instruction must be either <a
2065 href="#t_integer">integer</a> or <a href="#t_floating">floating point</a> values.
2066 This instruction can also take <a href="#t_vector">vector</a> versions of the values.
2067Both arguments must have identical types.</p>
2068<h5>Semantics:</h5>
2069<p>The value produced is the integer or floating point sum of the two
2070operands.</p>
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2071<p>If an integer sum has unsigned overflow, the result returned is the
2072mathematical result modulo 2<sup>n</sup>, where n is the bit width of
2073the result.</p>
2074<p>Because LLVM integers use a two's complement representation, this
2075instruction is appropriate for both signed and unsigned integers.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2076<h5>Example:</h5>
2077<pre> &lt;result&gt; = add i32 4, %var <i>; yields {i32}:result = 4 + %var</i>
2078</pre>
2079</div>
2080<!-- _______________________________________________________________________ -->
2081<div class="doc_subsubsection"> <a name="i_sub">'<tt>sub</tt>'
2082Instruction</a> </div>
2083<div class="doc_text">
2084<h5>Syntax:</h5>
2085<pre> &lt;result&gt; = sub &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
2086</pre>
2087<h5>Overview:</h5>
2088<p>The '<tt>sub</tt>' instruction returns the difference of its two
2089operands.</p>
2090<p>Note that the '<tt>sub</tt>' instruction is used to represent the '<tt>neg</tt>'
2091instruction present in most other intermediate representations.</p>
2092<h5>Arguments:</h5>
2093<p>The two arguments to the '<tt>sub</tt>' instruction must be either <a
2094 href="#t_integer">integer</a> or <a href="#t_floating">floating point</a>
2095values.
2096This instruction can also take <a href="#t_vector">vector</a> versions of the values.
2097Both arguments must have identical types.</p>
2098<h5>Semantics:</h5>
2099<p>The value produced is the integer or floating point difference of
2100the two operands.</p>
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2101<p>If an integer difference has unsigned overflow, the result returned is the
2102mathematical result modulo 2<sup>n</sup>, where n is the bit width of
2103the result.</p>
2104<p>Because LLVM integers use a two's complement representation, this
2105instruction is appropriate for both signed and unsigned integers.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2106<h5>Example:</h5>
2107<pre>
2108 &lt;result&gt; = sub i32 4, %var <i>; yields {i32}:result = 4 - %var</i>
2109 &lt;result&gt; = sub i32 0, %val <i>; yields {i32}:result = -%var</i>
2110</pre>
2111</div>
2112<!-- _______________________________________________________________________ -->
2113<div class="doc_subsubsection"> <a name="i_mul">'<tt>mul</tt>'
2114Instruction</a> </div>
2115<div class="doc_text">
2116<h5>Syntax:</h5>
2117<pre> &lt;result&gt; = mul &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
2118</pre>
2119<h5>Overview:</h5>
2120<p>The '<tt>mul</tt>' instruction returns the product of its two
2121operands.</p>
2122<h5>Arguments:</h5>
2123<p>The two arguments to the '<tt>mul</tt>' instruction must be either <a
2124 href="#t_integer">integer</a> or <a href="#t_floating">floating point</a>
2125values.
2126This instruction can also take <a href="#t_vector">vector</a> versions of the values.
2127Both arguments must have identical types.</p>
2128<h5>Semantics:</h5>
2129<p>The value produced is the integer or floating point product of the
2130two operands.</p>
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2131<p>If the result of an integer multiplication has unsigned overflow,
2132the result returned is the mathematical result modulo
21332<sup>n</sup>, where n is the bit width of the result.</p>
2134<p>Because LLVM integers use a two's complement representation, and the
2135result is the same width as the operands, this instruction returns the
2136correct result for both signed and unsigned integers. If a full product
2137(e.g. <tt>i32</tt>x<tt>i32</tt>-><tt>i64</tt>) is needed, the operands
2138should be sign-extended or zero-extended as appropriate to the
2139width of the full product.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2140<h5>Example:</h5>
2141<pre> &lt;result&gt; = mul i32 4, %var <i>; yields {i32}:result = 4 * %var</i>
2142</pre>
2143</div>
2144<!-- _______________________________________________________________________ -->
2145<div class="doc_subsubsection"> <a name="i_udiv">'<tt>udiv</tt>' Instruction
2146</a></div>
2147<div class="doc_text">
2148<h5>Syntax:</h5>
2149<pre> &lt;result&gt; = udiv &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
2150</pre>
2151<h5>Overview:</h5>
2152<p>The '<tt>udiv</tt>' instruction returns the quotient of its two
2153operands.</p>
2154<h5>Arguments:</h5>
2155<p>The two arguments to the '<tt>udiv</tt>' instruction must be
2156<a href="#t_integer">integer</a> values. Both arguments must have identical
2157types. This instruction can also take <a href="#t_vector">vector</a> versions
2158of the values in which case the elements must be integers.</p>
2159<h5>Semantics:</h5>
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2160<p>The value produced is the unsigned integer quotient of the two operands.</p>
2161<p>Note that unsigned integer division and signed integer division are distinct
2162operations; for signed integer division, use '<tt>sdiv</tt>'.</p>
2163<p>Division by zero leads to undefined behavior.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2164<h5>Example:</h5>
2165<pre> &lt;result&gt; = udiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
2166</pre>
2167</div>
2168<!-- _______________________________________________________________________ -->
2169<div class="doc_subsubsection"> <a name="i_sdiv">'<tt>sdiv</tt>' Instruction
2170</a> </div>
2171<div class="doc_text">
2172<h5>Syntax:</h5>
2173<pre> &lt;result&gt; = sdiv &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
2174</pre>
2175<h5>Overview:</h5>
2176<p>The '<tt>sdiv</tt>' instruction returns the quotient of its two
2177operands.</p>
2178<h5>Arguments:</h5>
2179<p>The two arguments to the '<tt>sdiv</tt>' instruction must be
2180<a href="#t_integer">integer</a> values. Both arguments must have identical
2181types. This instruction can also take <a href="#t_vector">vector</a> versions
2182of the values in which case the elements must be integers.</p>
2183<h5>Semantics:</h5>
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2184<p>The value produced is the signed integer quotient of the two operands.</p>
2185<p>Note that signed integer division and unsigned integer division are distinct
2186operations; for unsigned integer division, use '<tt>udiv</tt>'.</p>
2187<p>Division by zero leads to undefined behavior. Overflow also leads to
2188undefined behavior; this is a rare case, but can occur, for example,
2189by doing a 32-bit division of -2147483648 by -1.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2190<h5>Example:</h5>
2191<pre> &lt;result&gt; = sdiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
2192</pre>
2193</div>
2194<!-- _______________________________________________________________________ -->
2195<div class="doc_subsubsection"> <a name="i_fdiv">'<tt>fdiv</tt>'
2196Instruction</a> </div>
2197<div class="doc_text">
2198<h5>Syntax:</h5>
2199<pre> &lt;result&gt; = fdiv &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
2200</pre>
2201<h5>Overview:</h5>
2202<p>The '<tt>fdiv</tt>' instruction returns the quotient of its two
2203operands.</p>
2204<h5>Arguments:</h5>
2205<p>The two arguments to the '<tt>fdiv</tt>' instruction must be
2206<a href="#t_floating">floating point</a> values. Both arguments must have
2207identical types. This instruction can also take <a href="#t_vector">vector</a>
2208versions of floating point values.</p>
2209<h5>Semantics:</h5>
2210<p>The value produced is the floating point quotient of the two operands.</p>
2211<h5>Example:</h5>
2212<pre> &lt;result&gt; = fdiv float 4.0, %var <i>; yields {float}:result = 4.0 / %var</i>
2213</pre>
2214</div>
2215<!-- _______________________________________________________________________ -->
2216<div class="doc_subsubsection"> <a name="i_urem">'<tt>urem</tt>' Instruction</a>
2217</div>
2218<div class="doc_text">
2219<h5>Syntax:</h5>
2220<pre> &lt;result&gt; = urem &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
2221</pre>
2222<h5>Overview:</h5>
2223<p>The '<tt>urem</tt>' instruction returns the remainder from the
2224unsigned division of its two arguments.</p>
2225<h5>Arguments:</h5>
2226<p>The two arguments to the '<tt>urem</tt>' instruction must be
2227<a href="#t_integer">integer</a> values. Both arguments must have identical
Dan Gohman3e3fd8c2007-11-05 23:35:22 +0000[diff] [blame]2228types. This instruction can also take <a href="#t_vector">vector</a> versions
2229of the values in which case the elements must be integers.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2230<h5>Semantics:</h5>
2231<p>This instruction returns the unsigned integer <i>remainder</i> of a division.
2232This instruction always performs an unsigned division to get the remainder,
2233regardless of whether the arguments are unsigned or not.</p>
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2234<p>Note that unsigned integer remainder and signed integer remainder are
2235distinct operations; for signed integer remainder, use '<tt>srem</tt>'.</p>
2236<p>Taking the remainder of a division by zero leads to undefined behavior.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2237<h5>Example:</h5>
2238<pre> &lt;result&gt; = urem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
2239</pre>
2240
2241</div>
2242<!-- _______________________________________________________________________ -->
2243<div class="doc_subsubsection"> <a name="i_srem">'<tt>srem</tt>'
2244Instruction</a> </div>
2245<div class="doc_text">
2246<h5>Syntax:</h5>
2247<pre> &lt;result&gt; = srem &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
2248</pre>
2249<h5>Overview:</h5>
2250<p>The '<tt>srem</tt>' instruction returns the remainder from the
Dan Gohman3e3fd8c2007-11-05 23:35:22 +0000[diff] [blame]2251signed division of its two operands. This instruction can also take
2252<a href="#t_vector">vector</a> versions of the values in which case
2253the elements must be integers.</p>
Chris Lattner08497ce2008-01-04 04:33:49 +0000[diff] [blame]2254
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2255<h5>Arguments:</h5>
2256<p>The two arguments to the '<tt>srem</tt>' instruction must be
2257<a href="#t_integer">integer</a> values. Both arguments must have identical
2258types.</p>
2259<h5>Semantics:</h5>
2260<p>This instruction returns the <i>remainder</i> of a division (where the result
2261has the same sign as the dividend, <tt>var1</tt>), not the <i>modulo</i>
2262operator (where the result has the same sign as the divisor, <tt>var2</tt>) of
2263a value. For more information about the difference, see <a
2264 href="http://mathforum.org/dr.math/problems/anne.4.28.99.html">The
2265Math Forum</a>. For a table of how this is implemented in various languages,
2266please see <a href="http://en.wikipedia.org/wiki/Modulo_operation">
2267Wikipedia: modulo operation</a>.</p>
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2268<p>Note that signed integer remainder and unsigned integer remainder are
2269distinct operations; for unsigned integer remainder, use '<tt>urem</tt>'.</p>
2270<p>Taking the remainder of a division by zero leads to undefined behavior.
2271Overflow also leads to undefined behavior; this is a rare case, but can occur,
2272for example, by taking the remainder of a 32-bit division of -2147483648 by -1.
2273(The remainder doesn't actually overflow, but this rule lets srem be
2274implemented using instructions that return both the result of the division
2275and the remainder.)</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2276<h5>Example:</h5>
2277<pre> &lt;result&gt; = srem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
2278</pre>
2279
2280</div>
2281<!-- _______________________________________________________________________ -->
2282<div class="doc_subsubsection"> <a name="i_frem">'<tt>frem</tt>'
2283Instruction</a> </div>
2284<div class="doc_text">
2285<h5>Syntax:</h5>
2286<pre> &lt;result&gt; = frem &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
2287</pre>
2288<h5>Overview:</h5>
2289<p>The '<tt>frem</tt>' instruction returns the remainder from the
2290division of its two operands.</p>
2291<h5>Arguments:</h5>
2292<p>The two arguments to the '<tt>frem</tt>' instruction must be
2293<a href="#t_floating">floating point</a> values. Both arguments must have
Dan Gohman3e3fd8c2007-11-05 23:35:22 +0000[diff] [blame]2294identical types. This instruction can also take <a href="#t_vector">vector</a>
2295versions of floating point values.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2296<h5>Semantics:</h5>
2297<p>This instruction returns the <i>remainder</i> of a division.</p>
2298<h5>Example:</h5>
2299<pre> &lt;result&gt; = frem float 4.0, %var <i>; yields {float}:result = 4.0 % %var</i>
2300</pre>
2301</div>
2302
2303<!-- ======================================================================= -->
2304<div class="doc_subsection"> <a name="bitwiseops">Bitwise Binary
2305Operations</a> </div>
2306<div class="doc_text">
2307<p>Bitwise binary operators are used to do various forms of
2308bit-twiddling in a program. They are generally very efficient
2309instructions and can commonly be strength reduced from other
2310instructions. They require two operands, execute an operation on them,
2311and produce a single value. The resulting value of the bitwise binary
2312operators is always the same type as its first operand.</p>
2313</div>
2314
2315<!-- _______________________________________________________________________ -->
2316<div class="doc_subsubsection"> <a name="i_shl">'<tt>shl</tt>'
2317Instruction</a> </div>
2318<div class="doc_text">
2319<h5>Syntax:</h5>
2320<pre> &lt;result&gt; = shl &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
2321</pre>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2322
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2323<h5>Overview:</h5>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2324
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2325<p>The '<tt>shl</tt>' instruction returns the first operand shifted to
2326the left a specified number of bits.</p>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2327
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2328<h5>Arguments:</h5>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2329
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2330<p>Both arguments to the '<tt>shl</tt>' instruction must be the same <a
2331 href="#t_integer">integer</a> type.</p>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2332
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2333<h5>Semantics:</h5>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2334
2335<p>The value produced is <tt>var1</tt> * 2<sup><tt>var2</tt></sup>. If
2336<tt>var2</tt> is (statically or dynamically) equal to or larger than the number
2337of bits in <tt>var1</tt>, the result is undefined.</p>
2338
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2339<h5>Example:</h5><pre>
2340 &lt;result&gt; = shl i32 4, %var <i>; yields {i32}: 4 &lt;&lt; %var</i>
2341 &lt;result&gt; = shl i32 4, 2 <i>; yields {i32}: 16</i>
2342 &lt;result&gt; = shl i32 1, 10 <i>; yields {i32}: 1024</i>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2343 &lt;result&gt; = shl i32 1, 32 <i>; undefined</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2344</pre>
2345</div>
2346<!-- _______________________________________________________________________ -->
2347<div class="doc_subsubsection"> <a name="i_lshr">'<tt>lshr</tt>'
2348Instruction</a> </div>
2349<div class="doc_text">
2350<h5>Syntax:</h5>
2351<pre> &lt;result&gt; = lshr &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
2352</pre>
2353
2354<h5>Overview:</h5>
2355<p>The '<tt>lshr</tt>' instruction (logical shift right) returns the first
2356operand shifted to the right a specified number of bits with zero fill.</p>
2357
2358<h5>Arguments:</h5>
2359<p>Both arguments to the '<tt>lshr</tt>' instruction must be the same
2360<a href="#t_integer">integer</a> type.</p>
2361
2362<h5>Semantics:</h5>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2363
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2364<p>This instruction always performs a logical shift right operation. The most
2365significant bits of the result will be filled with zero bits after the
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2366shift. If <tt>var2</tt> is (statically or dynamically) equal to or larger than
2367the number of bits in <tt>var1</tt>, the result is undefined.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2368
2369<h5>Example:</h5>
2370<pre>
2371 &lt;result&gt; = lshr i32 4, 1 <i>; yields {i32}:result = 2</i>
2372 &lt;result&gt; = lshr i32 4, 2 <i>; yields {i32}:result = 1</i>
2373 &lt;result&gt; = lshr i8 4, 3 <i>; yields {i8}:result = 0</i>
2374 &lt;result&gt; = lshr i8 -2, 1 <i>; yields {i8}:result = 0x7FFFFFFF </i>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2375 &lt;result&gt; = lshr i32 1, 32 <i>; undefined</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2376</pre>
2377</div>
2378
2379<!-- _______________________________________________________________________ -->
2380<div class="doc_subsubsection"> <a name="i_ashr">'<tt>ashr</tt>'
2381Instruction</a> </div>
2382<div class="doc_text">
2383
2384<h5>Syntax:</h5>
2385<pre> &lt;result&gt; = ashr &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
2386</pre>
2387
2388<h5>Overview:</h5>
2389<p>The '<tt>ashr</tt>' instruction (arithmetic shift right) returns the first
2390operand shifted to the right a specified number of bits with sign extension.</p>
2391
2392<h5>Arguments:</h5>
2393<p>Both arguments to the '<tt>ashr</tt>' instruction must be the same
2394<a href="#t_integer">integer</a> type.</p>
2395
2396<h5>Semantics:</h5>
2397<p>This instruction always performs an arithmetic shift right operation,
2398The most significant bits of the result will be filled with the sign bit
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2399of <tt>var1</tt>. If <tt>var2</tt> is (statically or dynamically) equal to or
2400larger than the number of bits in <tt>var1</tt>, the result is undefined.
2401</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2402
2403<h5>Example:</h5>
2404<pre>
2405 &lt;result&gt; = ashr i32 4, 1 <i>; yields {i32}:result = 2</i>
2406 &lt;result&gt; = ashr i32 4, 2 <i>; yields {i32}:result = 1</i>
2407 &lt;result&gt; = ashr i8 4, 3 <i>; yields {i8}:result = 0</i>
2408 &lt;result&gt; = ashr i8 -2, 1 <i>; yields {i8}:result = -1</i>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2409 &lt;result&gt; = ashr i32 1, 32 <i>; undefined</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2410</pre>
2411</div>
2412
2413<!-- _______________________________________________________________________ -->
2414<div class="doc_subsubsection"> <a name="i_and">'<tt>and</tt>'
2415Instruction</a> </div>
2416<div class="doc_text">
2417<h5>Syntax:</h5>
2418<pre> &lt;result&gt; = and &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
2419</pre>
2420<h5>Overview:</h5>
2421<p>The '<tt>and</tt>' instruction returns the bitwise logical and of
2422its two operands.</p>
2423<h5>Arguments:</h5>
2424<p>The two arguments to the '<tt>and</tt>' instruction must be <a
2425 href="#t_integer">integer</a> values. Both arguments must have
2426identical types.</p>
2427<h5>Semantics:</h5>
2428<p>The truth table used for the '<tt>and</tt>' instruction is:</p>
2429<p> </p>
2430<div style="align: center">
2431<table border="1" cellspacing="0" cellpadding="4">
2432 <tbody>
2433 <tr>
2434 <td>In0</td>
2435 <td>In1</td>
2436 <td>Out</td>
2437 </tr>
2438 <tr>
2439 <td>0</td>
2440 <td>0</td>
2441 <td>0</td>
2442 </tr>
2443 <tr>
2444 <td>0</td>
2445 <td>1</td>
2446 <td>0</td>
2447 </tr>
2448 <tr>
2449 <td>1</td>
2450 <td>0</td>
2451 <td>0</td>
2452 </tr>
2453 <tr>
2454 <td>1</td>
2455 <td>1</td>
2456 <td>1</td>
2457 </tr>
2458 </tbody>
2459</table>
2460</div>
2461<h5>Example:</h5>
2462<pre> &lt;result&gt; = and i32 4, %var <i>; yields {i32}:result = 4 &amp; %var</i>
2463 &lt;result&gt; = and i32 15, 40 <i>; yields {i32}:result = 8</i>
2464 &lt;result&gt; = and i32 4, 8 <i>; yields {i32}:result = 0</i>
2465</pre>
2466</div>
2467<!-- _______________________________________________________________________ -->
2468<div class="doc_subsubsection"> <a name="i_or">'<tt>or</tt>' Instruction</a> </div>
2469<div class="doc_text">
2470<h5>Syntax:</h5>
2471<pre> &lt;result&gt; = or &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
2472</pre>
2473<h5>Overview:</h5>
2474<p>The '<tt>or</tt>' instruction returns the bitwise logical inclusive
2475or of its two operands.</p>
2476<h5>Arguments:</h5>
2477<p>The two arguments to the '<tt>or</tt>' instruction must be <a
2478 href="#t_integer">integer</a> values. Both arguments must have
2479identical types.</p>
2480<h5>Semantics:</h5>
2481<p>The truth table used for the '<tt>or</tt>' instruction is:</p>
2482<p> </p>
2483<div style="align: center">
2484<table border="1" cellspacing="0" cellpadding="4">
2485 <tbody>
2486 <tr>
2487 <td>In0</td>
2488 <td>In1</td>
2489 <td>Out</td>
2490 </tr>
2491 <tr>
2492 <td>0</td>
2493 <td>0</td>
2494 <td>0</td>
2495 </tr>
2496 <tr>
2497 <td>0</td>
2498 <td>1</td>
2499 <td>1</td>
2500 </tr>
2501 <tr>
2502 <td>1</td>
2503 <td>0</td>
2504 <td>1</td>
2505 </tr>
2506 <tr>
2507 <td>1</td>
2508 <td>1</td>
2509 <td>1</td>
2510 </tr>
2511 </tbody>
2512</table>
2513</div>
2514<h5>Example:</h5>
2515<pre> &lt;result&gt; = or i32 4, %var <i>; yields {i32}:result = 4 | %var</i>
2516 &lt;result&gt; = or i32 15, 40 <i>; yields {i32}:result = 47</i>
2517 &lt;result&gt; = or i32 4, 8 <i>; yields {i32}:result = 12</i>
2518</pre>
2519</div>
2520<!-- _______________________________________________________________________ -->
2521<div class="doc_subsubsection"> <a name="i_xor">'<tt>xor</tt>'
2522Instruction</a> </div>
2523<div class="doc_text">
2524<h5>Syntax:</h5>
2525<pre> &lt;result&gt; = xor &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
2526</pre>
2527<h5>Overview:</h5>
2528<p>The '<tt>xor</tt>' instruction returns the bitwise logical exclusive
2529or of its two operands. The <tt>xor</tt> is used to implement the
2530"one's complement" operation, which is the "~" operator in C.</p>
2531<h5>Arguments:</h5>
2532<p>The two arguments to the '<tt>xor</tt>' instruction must be <a
2533 href="#t_integer">integer</a> values. Both arguments must have
2534identical types.</p>
2535<h5>Semantics:</h5>
2536<p>The truth table used for the '<tt>xor</tt>' instruction is:</p>
2537<p> </p>
2538<div style="align: center">
2539<table border="1" cellspacing="0" cellpadding="4">
2540 <tbody>
2541 <tr>
2542 <td>In0</td>
2543 <td>In1</td>
2544 <td>Out</td>
2545 </tr>
2546 <tr>
2547 <td>0</td>
2548 <td>0</td>
2549 <td>0</td>
2550 </tr>
2551 <tr>
2552 <td>0</td>
2553 <td>1</td>
2554 <td>1</td>
2555 </tr>
2556 <tr>
2557 <td>1</td>
2558 <td>0</td>
2559 <td>1</td>
2560 </tr>
2561 <tr>
2562 <td>1</td>
2563 <td>1</td>
2564 <td>0</td>
2565 </tr>
2566 </tbody>
2567</table>
2568</div>
2569<p> </p>
2570<h5>Example:</h5>
2571<pre> &lt;result&gt; = xor i32 4, %var <i>; yields {i32}:result = 4 ^ %var</i>
2572 &lt;result&gt; = xor i32 15, 40 <i>; yields {i32}:result = 39</i>
2573 &lt;result&gt; = xor i32 4, 8 <i>; yields {i32}:result = 12</i>
2574 &lt;result&gt; = xor i32 %V, -1 <i>; yields {i32}:result = ~%V</i>
2575</pre>
2576</div>
2577
2578<!-- ======================================================================= -->
2579<div class="doc_subsection">
2580 <a name="vectorops">Vector Operations</a>
2581</div>
2582
2583<div class="doc_text">
2584
2585<p>LLVM supports several instructions to represent vector operations in a
2586target-independent manner. These instructions cover the element-access and
2587vector-specific operations needed to process vectors effectively. While LLVM
2588does directly support these vector operations, many sophisticated algorithms
2589will want to use target-specific intrinsics to take full advantage of a specific
2590target.</p>
2591
2592</div>
2593
2594<!-- _______________________________________________________________________ -->
2595<div class="doc_subsubsection">
2596 <a name="i_extractelement">'<tt>extractelement</tt>' Instruction</a>
2597</div>
2598
2599<div class="doc_text">
2600
2601<h5>Syntax:</h5>
2602
2603<pre>
2604 &lt;result&gt; = extractelement &lt;n x &lt;ty&gt;&gt; &lt;val&gt;, i32 &lt;idx&gt; <i>; yields &lt;ty&gt;</i>
2605</pre>
2606
2607<h5>Overview:</h5>
2608
2609<p>
2610The '<tt>extractelement</tt>' instruction extracts a single scalar
2611element from a vector at a specified index.
2612</p>
2613
2614
2615<h5>Arguments:</h5>
2616
2617<p>
2618The first operand of an '<tt>extractelement</tt>' instruction is a
2619value of <a href="#t_vector">vector</a> type. The second operand is
2620an index indicating the position from which to extract the element.
2621The index may be a variable.</p>
2622
2623<h5>Semantics:</h5>
2624
2625<p>
2626The result is a scalar of the same type as the element type of
2627<tt>val</tt>. Its value is the value at position <tt>idx</tt> of
2628<tt>val</tt>. If <tt>idx</tt> exceeds the length of <tt>val</tt>, the
2629results are undefined.
2630</p>
2631
2632<h5>Example:</h5>
2633
2634<pre>
2635 %result = extractelement &lt;4 x i32&gt; %vec, i32 0 <i>; yields i32</i>
2636</pre>
2637</div>
2638
2639
2640<!-- _______________________________________________________________________ -->
2641<div class="doc_subsubsection">
2642 <a name="i_insertelement">'<tt>insertelement</tt>' Instruction</a>
2643</div>
2644
2645<div class="doc_text">
2646
2647<h5>Syntax:</h5>
2648
2649<pre>
2650 &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>
2651</pre>
2652
2653<h5>Overview:</h5>
2654
2655<p>
2656The '<tt>insertelement</tt>' instruction inserts a scalar
2657element into a vector at a specified index.
2658</p>
2659
2660
2661<h5>Arguments:</h5>
2662
2663<p>
2664The first operand of an '<tt>insertelement</tt>' instruction is a
2665value of <a href="#t_vector">vector</a> type. The second operand is a
2666scalar value whose type must equal the element type of the first
2667operand. The third operand is an index indicating the position at
2668which to insert the value. The index may be a variable.</p>
2669
2670<h5>Semantics:</h5>
2671
2672<p>
2673The result is a vector of the same type as <tt>val</tt>. Its
2674element values are those of <tt>val</tt> except at position
2675<tt>idx</tt>, where it gets the value <tt>elt</tt>. If <tt>idx</tt>
2676exceeds the length of <tt>val</tt>, the results are undefined.
2677</p>
2678
2679<h5>Example:</h5>
2680
2681<pre>
2682 %result = insertelement &lt;4 x i32&gt; %vec, i32 1, i32 0 <i>; yields &lt;4 x i32&gt;</i>
2683</pre>
2684</div>
2685
2686<!-- _______________________________________________________________________ -->
2687<div class="doc_subsubsection">
2688 <a name="i_shufflevector">'<tt>shufflevector</tt>' Instruction</a>
2689</div>
2690
2691<div class="doc_text">
2692
2693<h5>Syntax:</h5>
2694
2695<pre>
2696 &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>
2697</pre>
2698
2699<h5>Overview:</h5>
2700
2701<p>
2702The '<tt>shufflevector</tt>' instruction constructs a permutation of elements
2703from two input vectors, returning a vector of the same type.
2704</p>
2705
2706<h5>Arguments:</h5>
2707
2708<p>
2709The first two operands of a '<tt>shufflevector</tt>' instruction are vectors
2710with types that match each other and types that match the result of the
2711instruction. The third argument is a shuffle mask, which has the same number
2712of elements as the other vector type, but whose element type is always 'i32'.
2713</p>
2714
2715<p>
2716The shuffle mask operand is required to be a constant vector with either
2717constant integer or undef values.
2718</p>
2719
2720<h5>Semantics:</h5>
2721
2722<p>
2723The elements of the two input vectors are numbered from left to right across
2724both of the vectors. The shuffle mask operand specifies, for each element of
2725the result vector, which element of the two input registers the result element
2726gets. The element selector may be undef (meaning "don't care") and the second
2727operand may be undef if performing a shuffle from only one vector.
2728</p>
2729
2730<h5>Example:</h5>
2731
2732<pre>
2733 %result = shufflevector &lt;4 x i32&gt; %v1, &lt;4 x i32&gt; %v2,
2734 &lt;4 x i32&gt; &lt;i32 0, i32 4, i32 1, i32 5&gt; <i>; yields &lt;4 x i32&gt;</i>
2735 %result = shufflevector &lt;4 x i32&gt; %v1, &lt;4 x i32&gt; undef,
2736 &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.
2737</pre>
2738</div>
2739
2740
2741<!-- ======================================================================= -->
2742<div class="doc_subsection">
2743 <a name="memoryops">Memory Access and Addressing Operations</a>
2744</div>
2745
2746<div class="doc_text">
2747
2748<p>A key design point of an SSA-based representation is how it
2749represents memory. In LLVM, no memory locations are in SSA form, which
2750makes things very simple. This section describes how to read, write,
2751allocate, and free memory in LLVM.</p>
2752
2753</div>
2754
2755<!-- _______________________________________________________________________ -->
2756<div class="doc_subsubsection">
2757 <a name="i_malloc">'<tt>malloc</tt>' Instruction</a>
2758</div>
2759
2760<div class="doc_text">
2761
2762<h5>Syntax:</h5>
2763
2764<pre>
2765 &lt;result&gt; = malloc &lt;type&gt;[, i32 &lt;NumElements&gt;][, align &lt;alignment&gt;] <i>; yields {type*}:result</i>
2766</pre>
2767
2768<h5>Overview:</h5>
2769
2770<p>The '<tt>malloc</tt>' instruction allocates memory from the system
Christopher Lambcfe00962007-12-17 01:00:21 +0000[diff] [blame]2771heap and returns a pointer to it. The object is always allocated in the generic
2772address space (address space zero).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2773
2774<h5>Arguments:</h5>
2775
2776<p>The '<tt>malloc</tt>' instruction allocates
2777<tt>sizeof(&lt;type&gt;)*NumElements</tt>
2778bytes of memory from the operating system and returns a pointer of the
2779appropriate type to the program. If "NumElements" is specified, it is the
Gabor Greif5082cf42008-02-09 22:24:34 +0000[diff] [blame]2780number of elements allocated, otherwise "NumElements" is defaulted to be one.
2781If an alignment is specified, the value result of the allocation is guaranteed to
2782be aligned to at least that boundary. If not specified, or if zero, the target can
2783choose to align the allocation on any convenient boundary.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2784
2785<p>'<tt>type</tt>' must be a sized type.</p>
2786
2787<h5>Semantics:</h5>
2788
2789<p>Memory is allocated using the system "<tt>malloc</tt>" function, and
2790a pointer is returned.</p>
2791
2792<h5>Example:</h5>
2793
2794<pre>
2795 %array = malloc [4 x i8 ] <i>; yields {[%4 x i8]*}:array</i>
2796
2797 %size = <a href="#i_add">add</a> i32 2, 2 <i>; yields {i32}:size = i32 4</i>
2798 %array1 = malloc i8, i32 4 <i>; yields {i8*}:array1</i>
2799 %array2 = malloc [12 x i8], i32 %size <i>; yields {[12 x i8]*}:array2</i>
2800 %array3 = malloc i32, i32 4, align 1024 <i>; yields {i32*}:array3</i>
2801 %array4 = malloc i32, align 1024 <i>; yields {i32*}:array4</i>
2802</pre>
2803</div>
2804
2805<!-- _______________________________________________________________________ -->
2806<div class="doc_subsubsection">
2807 <a name="i_free">'<tt>free</tt>' Instruction</a>
2808</div>
2809
2810<div class="doc_text">
2811
2812<h5>Syntax:</h5>
2813
2814<pre>
2815 free &lt;type&gt; &lt;value&gt; <i>; yields {void}</i>
2816</pre>
2817
2818<h5>Overview:</h5>
2819
2820<p>The '<tt>free</tt>' instruction returns memory back to the unused
2821memory heap to be reallocated in the future.</p>
2822
2823<h5>Arguments:</h5>
2824
2825<p>'<tt>value</tt>' shall be a pointer value that points to a value
2826that was allocated with the '<tt><a href="#i_malloc">malloc</a></tt>'
2827instruction.</p>
2828
2829<h5>Semantics:</h5>
2830
2831<p>Access to the memory pointed to by the pointer is no longer defined
2832after this instruction executes.</p>
2833
2834<h5>Example:</h5>
2835
2836<pre>
2837 %array = <a href="#i_malloc">malloc</a> [4 x i8] <i>; yields {[4 x i8]*}:array</i>
2838 free [4 x i8]* %array
2839</pre>
2840</div>
2841
2842<!-- _______________________________________________________________________ -->
2843<div class="doc_subsubsection">
2844 <a name="i_alloca">'<tt>alloca</tt>' Instruction</a>
2845</div>
2846
2847<div class="doc_text">
2848
2849<h5>Syntax:</h5>
2850
2851<pre>
2852 &lt;result&gt; = alloca &lt;type&gt;[, i32 &lt;NumElements&gt;][, align &lt;alignment&gt;] <i>; yields {type*}:result</i>
2853</pre>
2854
2855<h5>Overview:</h5>
2856
2857<p>The '<tt>alloca</tt>' instruction allocates memory on the stack frame of the
2858currently executing function, to be automatically released when this function
Christopher Lambcfe00962007-12-17 01:00:21 +0000[diff] [blame]2859returns to its caller. The object is always allocated in the generic address
2860space (address space zero).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2861
2862<h5>Arguments:</h5>
2863
2864<p>The '<tt>alloca</tt>' instruction allocates <tt>sizeof(&lt;type&gt;)*NumElements</tt>
2865bytes of memory on the runtime stack, returning a pointer of the
Gabor Greif5082cf42008-02-09 22:24:34 +0000[diff] [blame]2866appropriate type to the program. If "NumElements" is specified, it is the
2867number of elements allocated, otherwise "NumElements" is defaulted to be one.
2868If an alignment is specified, the value result of the allocation is guaranteed
2869to be aligned to at least that boundary. If not specified, or if zero, the target
2870can choose to align the allocation on any convenient boundary.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2871
2872<p>'<tt>type</tt>' may be any sized type.</p>
2873
2874<h5>Semantics:</h5>
2875
2876<p>Memory is allocated; a pointer is returned. '<tt>alloca</tt>'d
2877memory is automatically released when the function returns. The '<tt>alloca</tt>'
2878instruction is commonly used to represent automatic variables that must
2879have an address available. When the function returns (either with the <tt><a
2880 href="#i_ret">ret</a></tt> or <tt><a href="#i_unwind">unwind</a></tt>
2881instructions), the memory is reclaimed.</p>
2882
2883<h5>Example:</h5>
2884
2885<pre>
2886 %ptr = alloca i32 <i>; yields {i32*}:ptr</i>
2887 %ptr = alloca i32, i32 4 <i>; yields {i32*}:ptr</i>
2888 %ptr = alloca i32, i32 4, align 1024 <i>; yields {i32*}:ptr</i>
2889 %ptr = alloca i32, align 1024 <i>; yields {i32*}:ptr</i>
2890</pre>
2891</div>
2892
2893<!-- _______________________________________________________________________ -->
2894<div class="doc_subsubsection"> <a name="i_load">'<tt>load</tt>'
2895Instruction</a> </div>
2896<div class="doc_text">
2897<h5>Syntax:</h5>
2898<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>
2899<h5>Overview:</h5>
2900<p>The '<tt>load</tt>' instruction is used to read from memory.</p>
2901<h5>Arguments:</h5>
2902<p>The argument to the '<tt>load</tt>' instruction specifies the memory
2903address from which to load. The pointer must point to a <a
2904 href="#t_firstclass">first class</a> type. If the <tt>load</tt> is
2905marked as <tt>volatile</tt>, then the optimizer is not allowed to modify
2906the number or order of execution of this <tt>load</tt> with other
2907volatile <tt>load</tt> and <tt><a href="#i_store">store</a></tt>
2908instructions. </p>
Chris Lattner21003c82008-01-06 21:04:43 +0000[diff] [blame]2909<p>
2910The optional "align" argument specifies the alignment of the operation
2911(that is, the alignment of the memory address). A value of 0 or an
2912omitted "align" argument means that the operation has the preferential
2913alignment for the target. It is the responsibility of the code emitter
2914to ensure that the alignment information is correct. Overestimating
2915the alignment results in an undefined behavior. Underestimating the
2916alignment may produce less efficient code. An alignment of 1 is always
2917safe.
2918</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2919<h5>Semantics:</h5>
2920<p>The location of memory pointed to is loaded.</p>
2921<h5>Examples:</h5>
2922<pre> %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
2923 <a
2924 href="#i_store">store</a> i32 3, i32* %ptr <i>; yields {void}</i>
2925 %val = load i32* %ptr <i>; yields {i32}:val = i32 3</i>
2926</pre>
2927</div>
2928<!-- _______________________________________________________________________ -->
2929<div class="doc_subsubsection"> <a name="i_store">'<tt>store</tt>'
2930Instruction</a> </div>
2931<div class="doc_text">
2932<h5>Syntax:</h5>
2933<pre> store &lt;ty&gt; &lt;value&gt;, &lt;ty&gt;* &lt;pointer&gt;[, align &lt;alignment&gt;] <i>; yields {void}</i>
2934 volatile store &lt;ty&gt; &lt;value&gt;, &lt;ty&gt;* &lt;pointer&gt;[, align &lt;alignment&gt;] <i>; yields {void}</i>
2935</pre>
2936<h5>Overview:</h5>
2937<p>The '<tt>store</tt>' instruction is used to write to memory.</p>
2938<h5>Arguments:</h5>
2939<p>There are two arguments to the '<tt>store</tt>' instruction: a value
2940to store and an address at which to store it. The type of the '<tt>&lt;pointer&gt;</tt>'
2941operand must be a pointer to the type of the '<tt>&lt;value&gt;</tt>'
2942operand. If the <tt>store</tt> is marked as <tt>volatile</tt>, then the
2943optimizer is not allowed to modify the number or order of execution of
2944this <tt>store</tt> with other volatile <tt>load</tt> and <tt><a
2945 href="#i_store">store</a></tt> instructions.</p>
Chris Lattner21003c82008-01-06 21:04:43 +0000[diff] [blame]2946<p>
2947The optional "align" argument specifies the alignment of the operation
2948(that is, the alignment of the memory address). A value of 0 or an
2949omitted "align" argument means that the operation has the preferential
2950alignment for the target. It is the responsibility of the code emitter
2951to ensure that the alignment information is correct. Overestimating
2952the alignment results in an undefined behavior. Underestimating the
2953alignment may produce less efficient code. An alignment of 1 is always
2954safe.
2955</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2956<h5>Semantics:</h5>
2957<p>The contents of memory are updated to contain '<tt>&lt;value&gt;</tt>'
2958at the location specified by the '<tt>&lt;pointer&gt;</tt>' operand.</p>
2959<h5>Example:</h5>
2960<pre> %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
Bill Wendling63ffa142007-10-22 05:10:05 +0000[diff] [blame]2961 store i32 3, i32* %ptr <i>; yields {void}</i>
2962 %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]2963</pre>
2964</div>
2965
2966<!-- _______________________________________________________________________ -->
2967<div class="doc_subsubsection">
2968 <a name="i_getelementptr">'<tt>getelementptr</tt>' Instruction</a>
2969</div>
2970
2971<div class="doc_text">
2972<h5>Syntax:</h5>
2973<pre>
2974 &lt;result&gt; = getelementptr &lt;ty&gt;* &lt;ptrval&gt;{, &lt;ty&gt; &lt;idx&gt;}*
2975</pre>
2976
2977<h5>Overview:</h5>
2978
2979<p>
2980The '<tt>getelementptr</tt>' instruction is used to get the address of a
2981subelement of an aggregate data structure.</p>
2982
2983<h5>Arguments:</h5>
2984
2985<p>This instruction takes a list of integer operands that indicate what
2986elements of the aggregate object to index to. The actual types of the arguments
2987provided depend on the type of the first pointer argument. The
2988'<tt>getelementptr</tt>' instruction is used to index down through the type
2989levels of a structure or to a specific index in an array. When indexing into a
2990structure, only <tt>i32</tt> integer constants are allowed. When indexing
2991into an array or pointer, only integers of 32 or 64 bits are allowed, and will
2992be sign extended to 64-bit values.</p>
2993
2994<p>For example, let's consider a C code fragment and how it gets
2995compiled to LLVM:</p>
2996
2997<div class="doc_code">
2998<pre>
2999struct RT {
3000 char A;
3001 int B[10][20];
3002 char C;
3003};
3004struct ST {
3005 int X;
3006 double Y;
3007 struct RT Z;
3008};
3009
3010int *foo(struct ST *s) {
3011 return &amp;s[1].Z.B[5][13];
3012}
3013</pre>
3014</div>
3015
3016<p>The LLVM code generated by the GCC frontend is:</p>
3017
3018<div class="doc_code">
3019<pre>
3020%RT = type { i8 , [10 x [20 x i32]], i8 }
3021%ST = type { i32, double, %RT }
3022
3023define i32* %foo(%ST* %s) {
3024entry:
3025 %reg = getelementptr %ST* %s, i32 1, i32 2, i32 1, i32 5, i32 13
3026 ret i32* %reg
3027}
3028</pre>
3029</div>
3030
3031<h5>Semantics:</h5>
3032
3033<p>The index types specified for the '<tt>getelementptr</tt>' instruction depend
3034on the pointer type that is being indexed into. <a href="#t_pointer">Pointer</a>
3035and <a href="#t_array">array</a> types can use a 32-bit or 64-bit
3036<a href="#t_integer">integer</a> type but the value will always be sign extended
3037to 64-bits. <a href="#t_struct">Structure</a> types require <tt>i32</tt>
3038<b>constants</b>.</p>
3039
3040<p>In the example above, the first index is indexing into the '<tt>%ST*</tt>'
3041type, which is a pointer, yielding a '<tt>%ST</tt>' = '<tt>{ i32, double, %RT
3042}</tt>' type, a structure. The second index indexes into the third element of
3043the structure, yielding a '<tt>%RT</tt>' = '<tt>{ i8 , [10 x [20 x i32]],
3044i8 }</tt>' type, another structure. The third index indexes into the second
3045element of the structure, yielding a '<tt>[10 x [20 x i32]]</tt>' type, an
3046array. The two dimensions of the array are subscripted into, yielding an
3047'<tt>i32</tt>' type. The '<tt>getelementptr</tt>' instruction returns a pointer
3048to this element, thus computing a value of '<tt>i32*</tt>' type.</p>
3049
3050<p>Note that it is perfectly legal to index partially through a
3051structure, returning a pointer to an inner element. Because of this,
3052the LLVM code for the given testcase is equivalent to:</p>
3053
3054<pre>
3055 define i32* %foo(%ST* %s) {
3056 %t1 = getelementptr %ST* %s, i32 1 <i>; yields %ST*:%t1</i>
3057 %t2 = getelementptr %ST* %t1, i32 0, i32 2 <i>; yields %RT*:%t2</i>
3058 %t3 = getelementptr %RT* %t2, i32 0, i32 1 <i>; yields [10 x [20 x i32]]*:%t3</i>
3059 %t4 = getelementptr [10 x [20 x i32]]* %t3, i32 0, i32 5 <i>; yields [20 x i32]*:%t4</i>
3060 %t5 = getelementptr [20 x i32]* %t4, i32 0, i32 13 <i>; yields i32*:%t5</i>
3061 ret i32* %t5
3062 }
3063</pre>
3064
3065<p>Note that it is undefined to access an array out of bounds: array and
3066pointer indexes must always be within the defined bounds of the array type.
3067The one exception for this rules is zero length arrays. These arrays are
3068defined to be accessible as variable length arrays, which requires access
3069beyond the zero'th element.</p>
3070
3071<p>The getelementptr instruction is often confusing. For some more insight
3072into how it works, see <a href="GetElementPtr.html">the getelementptr
3073FAQ</a>.</p>
3074
3075<h5>Example:</h5>
3076
3077<pre>
3078 <i>; yields [12 x i8]*:aptr</i>
3079 %aptr = getelementptr {i32, [12 x i8]}* %sptr, i64 0, i32 1
3080</pre>
3081</div>
3082
3083<!-- ======================================================================= -->
3084<div class="doc_subsection"> <a name="convertops">Conversion Operations</a>
3085</div>
3086<div class="doc_text">
3087<p>The instructions in this category are the conversion instructions (casting)
3088which all take a single operand and a type. They perform various bit conversions
3089on the operand.</p>
3090</div>
3091
3092<!-- _______________________________________________________________________ -->
3093<div class="doc_subsubsection">
3094 <a name="i_trunc">'<tt>trunc .. to</tt>' Instruction</a>
3095</div>
3096<div class="doc_text">
3097
3098<h5>Syntax:</h5>
3099<pre>
3100 &lt;result&gt; = trunc &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3101</pre>
3102
3103<h5>Overview:</h5>
3104<p>
3105The '<tt>trunc</tt>' instruction truncates its operand to the type <tt>ty2</tt>.
3106</p>
3107
3108<h5>Arguments:</h5>
3109<p>
3110The '<tt>trunc</tt>' instruction takes a <tt>value</tt> to trunc, which must
3111be an <a href="#t_integer">integer</a> type, and a type that specifies the size
3112and type of the result, which must be an <a href="#t_integer">integer</a>
3113type. The bit size of <tt>value</tt> must be larger than the bit size of
3114<tt>ty2</tt>. Equal sized types are not allowed.</p>
3115
3116<h5>Semantics:</h5>
3117<p>
3118The '<tt>trunc</tt>' instruction truncates the high order bits in <tt>value</tt>
3119and converts the remaining bits to <tt>ty2</tt>. Since the source size must be
3120larger than the destination size, <tt>trunc</tt> cannot be a <i>no-op cast</i>.
3121It will always truncate bits.</p>
3122
3123<h5>Example:</h5>
3124<pre>
3125 %X = trunc i32 257 to i8 <i>; yields i8:1</i>
3126 %Y = trunc i32 123 to i1 <i>; yields i1:true</i>
3127 %Y = trunc i32 122 to i1 <i>; yields i1:false</i>
3128</pre>
3129</div>
3130
3131<!-- _______________________________________________________________________ -->
3132<div class="doc_subsubsection">
3133 <a name="i_zext">'<tt>zext .. to</tt>' Instruction</a>
3134</div>
3135<div class="doc_text">
3136
3137<h5>Syntax:</h5>
3138<pre>
3139 &lt;result&gt; = zext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3140</pre>
3141
3142<h5>Overview:</h5>
3143<p>The '<tt>zext</tt>' instruction zero extends its operand to type
3144<tt>ty2</tt>.</p>
3145
3146
3147<h5>Arguments:</h5>
3148<p>The '<tt>zext</tt>' instruction takes a value to cast, which must be of
3149<a href="#t_integer">integer</a> type, and a type to cast it to, which must
3150also be of <a href="#t_integer">integer</a> type. The bit size of the
3151<tt>value</tt> must be smaller than the bit size of the destination type,
3152<tt>ty2</tt>.</p>
3153
3154<h5>Semantics:</h5>
3155<p>The <tt>zext</tt> fills the high order bits of the <tt>value</tt> with zero
3156bits until it reaches the size of the destination type, <tt>ty2</tt>.</p>
3157
3158<p>When zero extending from i1, the result will always be either 0 or 1.</p>
3159
3160<h5>Example:</h5>
3161<pre>
3162 %X = zext i32 257 to i64 <i>; yields i64:257</i>
3163 %Y = zext i1 true to i32 <i>; yields i32:1</i>
3164</pre>
3165</div>
3166
3167<!-- _______________________________________________________________________ -->
3168<div class="doc_subsubsection">
3169 <a name="i_sext">'<tt>sext .. to</tt>' Instruction</a>
3170</div>
3171<div class="doc_text">
3172
3173<h5>Syntax:</h5>
3174<pre>
3175 &lt;result&gt; = sext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3176</pre>
3177
3178<h5>Overview:</h5>
3179<p>The '<tt>sext</tt>' sign extends <tt>value</tt> to the type <tt>ty2</tt>.</p>
3180
3181<h5>Arguments:</h5>
3182<p>
3183The '<tt>sext</tt>' instruction takes a value to cast, which must be of
3184<a href="#t_integer">integer</a> type, and a type to cast it to, which must
3185also be of <a href="#t_integer">integer</a> type. The bit size of the
3186<tt>value</tt> must be smaller than the bit size of the destination type,
3187<tt>ty2</tt>.</p>
3188
3189<h5>Semantics:</h5>
3190<p>
3191The '<tt>sext</tt>' instruction performs a sign extension by copying the sign
3192bit (highest order bit) of the <tt>value</tt> until it reaches the bit size of
3193the type <tt>ty2</tt>.</p>
3194
3195<p>When sign extending from i1, the extension always results in -1 or 0.</p>
3196
3197<h5>Example:</h5>
3198<pre>
3199 %X = sext i8 -1 to i16 <i>; yields i16 :65535</i>
3200 %Y = sext i1 true to i32 <i>; yields i32:-1</i>
3201</pre>
3202</div>
3203
3204<!-- _______________________________________________________________________ -->
3205<div class="doc_subsubsection">
3206 <a name="i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a>
3207</div>
3208
3209<div class="doc_text">
3210
3211<h5>Syntax:</h5>
3212
3213<pre>
3214 &lt;result&gt; = fptrunc &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3215</pre>
3216
3217<h5>Overview:</h5>
3218<p>The '<tt>fptrunc</tt>' instruction truncates <tt>value</tt> to type
3219<tt>ty2</tt>.</p>
3220
3221
3222<h5>Arguments:</h5>
3223<p>The '<tt>fptrunc</tt>' instruction takes a <a href="#t_floating">floating
3224 point</a> value to cast and a <a href="#t_floating">floating point</a> type to
3225cast it to. The size of <tt>value</tt> must be larger than the size of
3226<tt>ty2</tt>. This implies that <tt>fptrunc</tt> cannot be used to make a
3227<i>no-op cast</i>.</p>
3228
3229<h5>Semantics:</h5>
3230<p> The '<tt>fptrunc</tt>' instruction truncates a <tt>value</tt> from a larger
3231<a href="#t_floating">floating point</a> type to a smaller
3232<a href="#t_floating">floating point</a> type. If the value cannot fit within
3233the destination type, <tt>ty2</tt>, then the results are undefined.</p>
3234
3235<h5>Example:</h5>
3236<pre>
3237 %X = fptrunc double 123.0 to float <i>; yields float:123.0</i>
3238 %Y = fptrunc double 1.0E+300 to float <i>; yields undefined</i>
3239</pre>
3240</div>
3241
3242<!-- _______________________________________________________________________ -->
3243<div class="doc_subsubsection">
3244 <a name="i_fpext">'<tt>fpext .. to</tt>' Instruction</a>
3245</div>
3246<div class="doc_text">
3247
3248<h5>Syntax:</h5>
3249<pre>
3250 &lt;result&gt; = fpext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3251</pre>
3252
3253<h5>Overview:</h5>
3254<p>The '<tt>fpext</tt>' extends a floating point <tt>value</tt> to a larger
3255floating point value.</p>
3256
3257<h5>Arguments:</h5>
3258<p>The '<tt>fpext</tt>' instruction takes a
3259<a href="#t_floating">floating point</a> <tt>value</tt> to cast,
3260and a <a href="#t_floating">floating point</a> type to cast it to. The source
3261type must be smaller than the destination type.</p>
3262
3263<h5>Semantics:</h5>
3264<p>The '<tt>fpext</tt>' instruction extends the <tt>value</tt> from a smaller
3265<a href="#t_floating">floating point</a> type to a larger
3266<a href="#t_floating">floating point</a> type. The <tt>fpext</tt> cannot be
3267used to make a <i>no-op cast</i> because it always changes bits. Use
3268<tt>bitcast</tt> to make a <i>no-op cast</i> for a floating point cast.</p>
3269
3270<h5>Example:</h5>
3271<pre>
3272 %X = fpext float 3.1415 to double <i>; yields double:3.1415</i>
3273 %Y = fpext float 1.0 to float <i>; yields float:1.0 (no-op)</i>
3274</pre>
3275</div>
3276
3277<!-- _______________________________________________________________________ -->
3278<div class="doc_subsubsection">
3279 <a name="i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a>
3280</div>
3281<div class="doc_text">
3282
3283<h5>Syntax:</h5>
3284<pre>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3285 &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]3286</pre>
3287
3288<h5>Overview:</h5>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3289<p>The '<tt>fptoui</tt>' converts a floating point <tt>value</tt> to its
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3290unsigned integer equivalent of type <tt>ty2</tt>.
3291</p>
3292
3293<h5>Arguments:</h5>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3294<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]3295scalar or vector <a href="#t_floating">floating point</a> value, and a type
3296to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
3297type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
3298vector integer type with the same number of elements as <tt>ty</tt></p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3299
3300<h5>Semantics:</h5>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3301<p> The '<tt>fptoui</tt>' instruction converts its
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3302<a href="#t_floating">floating point</a> operand into the nearest (rounding
3303towards zero) unsigned integer value. If the value cannot fit in <tt>ty2</tt>,
3304the results are undefined.</p>
3305
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3306<h5>Example:</h5>
3307<pre>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3308 %X = fptoui double 123.0 to i32 <i>; yields i32:123</i>
Chris Lattner681f1e82007-09-22 03:17:52 +0000[diff] [blame]3309 %Y = fptoui float 1.0E+300 to i1 <i>; yields undefined:1</i>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3310 %X = fptoui float 1.04E+17 to i8 <i>; yields undefined:1</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3311</pre>
3312</div>
3313
3314<!-- _______________________________________________________________________ -->
3315<div class="doc_subsubsection">
3316 <a name="i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a>
3317</div>
3318<div class="doc_text">
3319
3320<h5>Syntax:</h5>
3321<pre>
3322 &lt;result&gt; = fptosi &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3323</pre>
3324
3325<h5>Overview:</h5>
3326<p>The '<tt>fptosi</tt>' instruction converts
3327<a href="#t_floating">floating point</a> <tt>value</tt> to type <tt>ty2</tt>.
3328</p>
3329
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3330<h5>Arguments:</h5>
3331<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]3332scalar or vector <a href="#t_floating">floating point</a> value, and a type
3333to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
3334type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
3335vector integer type with the same number of elements as <tt>ty</tt></p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3336
3337<h5>Semantics:</h5>
3338<p>The '<tt>fptosi</tt>' instruction converts its
3339<a href="#t_floating">floating point</a> operand into the nearest (rounding
3340towards zero) signed integer value. If the value cannot fit in <tt>ty2</tt>,
3341the results are undefined.</p>
3342
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3343<h5>Example:</h5>
3344<pre>
3345 %X = fptosi double -123.0 to i32 <i>; yields i32:-123</i>
Chris Lattner681f1e82007-09-22 03:17:52 +0000[diff] [blame]3346 %Y = fptosi float 1.0E-247 to i1 <i>; yields undefined:1</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3347 %X = fptosi float 1.04E+17 to i8 <i>; yields undefined:1</i>
3348</pre>
3349</div>
3350
3351<!-- _______________________________________________________________________ -->
3352<div class="doc_subsubsection">
3353 <a name="i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a>
3354</div>
3355<div class="doc_text">
3356
3357<h5>Syntax:</h5>
3358<pre>
3359 &lt;result&gt; = uitofp &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3360</pre>
3361
3362<h5>Overview:</h5>
3363<p>The '<tt>uitofp</tt>' instruction regards <tt>value</tt> as an unsigned
3364integer and converts that value to the <tt>ty2</tt> type.</p>
3365
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3366<h5>Arguments:</h5>
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]3367<p>The '<tt>uitofp</tt>' instruction takes a value to cast, which must be a
3368scalar or vector <a href="#t_integer">integer</a> value, and a type to cast it
3369to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
3370type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
3371floating point type with the same number of elements as <tt>ty</tt></p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3372
3373<h5>Semantics:</h5>
3374<p>The '<tt>uitofp</tt>' instruction interprets its operand as an unsigned
3375integer quantity and converts it to the corresponding floating point value. If
3376the value cannot fit in the floating point value, the results are undefined.</p>
3377
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3378<h5>Example:</h5>
3379<pre>
3380 %X = uitofp i32 257 to float <i>; yields float:257.0</i>
3381 %Y = uitofp i8 -1 to double <i>; yields double:255.0</i>
3382</pre>
3383</div>
3384
3385<!-- _______________________________________________________________________ -->
3386<div class="doc_subsubsection">
3387 <a name="i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a>
3388</div>
3389<div class="doc_text">
3390
3391<h5>Syntax:</h5>
3392<pre>
3393 &lt;result&gt; = sitofp &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3394</pre>
3395
3396<h5>Overview:</h5>
3397<p>The '<tt>sitofp</tt>' instruction regards <tt>value</tt> as a signed
3398integer and converts that value to the <tt>ty2</tt> type.</p>
3399
3400<h5>Arguments:</h5>
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]3401<p>The '<tt>sitofp</tt>' instruction takes a value to cast, which must be a
3402scalar or vector <a href="#t_integer">integer</a> value, and a type to cast it
3403to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
3404type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
3405floating point type with the same number of elements as <tt>ty</tt></p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3406
3407<h5>Semantics:</h5>
3408<p>The '<tt>sitofp</tt>' instruction interprets its operand as a signed
3409integer quantity and converts it to the corresponding floating point value. If
3410the value cannot fit in the floating point value, the results are undefined.</p>
3411
3412<h5>Example:</h5>
3413<pre>
3414 %X = sitofp i32 257 to float <i>; yields float:257.0</i>
3415 %Y = sitofp i8 -1 to double <i>; yields double:-1.0</i>
3416</pre>
3417</div>
3418
3419<!-- _______________________________________________________________________ -->
3420<div class="doc_subsubsection">
3421 <a name="i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a>
3422</div>
3423<div class="doc_text">
3424
3425<h5>Syntax:</h5>
3426<pre>
3427 &lt;result&gt; = ptrtoint &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3428</pre>
3429
3430<h5>Overview:</h5>
3431<p>The '<tt>ptrtoint</tt>' instruction converts the pointer <tt>value</tt> to
3432the integer type <tt>ty2</tt>.</p>
3433
3434<h5>Arguments:</h5>
3435<p>The '<tt>ptrtoint</tt>' instruction takes a <tt>value</tt> to cast, which
3436must be a <a href="#t_pointer">pointer</a> value, and a type to cast it to
3437<tt>ty2</tt>, which must be an <a href="#t_integer">integer</a> type.
3438
3439<h5>Semantics:</h5>
3440<p>The '<tt>ptrtoint</tt>' instruction converts <tt>value</tt> to integer type
3441<tt>ty2</tt> by interpreting the pointer value as an integer and either
3442truncating or zero extending that value to the size of the integer type. If
3443<tt>value</tt> is smaller than <tt>ty2</tt> then a zero extension is done. If
3444<tt>value</tt> is larger than <tt>ty2</tt> then a truncation is done. If they
3445are the same size, then nothing is done (<i>no-op cast</i>) other than a type
3446change.</p>
3447
3448<h5>Example:</h5>
3449<pre>
3450 %X = ptrtoint i32* %X to i8 <i>; yields truncation on 32-bit architecture</i>
3451 %Y = ptrtoint i32* %x to i64 <i>; yields zero extension on 32-bit architecture</i>
3452</pre>
3453</div>
3454
3455<!-- _______________________________________________________________________ -->
3456<div class="doc_subsubsection">
3457 <a name="i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a>
3458</div>
3459<div class="doc_text">
3460
3461<h5>Syntax:</h5>
3462<pre>
3463 &lt;result&gt; = inttoptr &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3464</pre>
3465
3466<h5>Overview:</h5>
3467<p>The '<tt>inttoptr</tt>' instruction converts an integer <tt>value</tt> to
3468a pointer type, <tt>ty2</tt>.</p>
3469
3470<h5>Arguments:</h5>
3471<p>The '<tt>inttoptr</tt>' instruction takes an <a href="#t_integer">integer</a>
3472value to cast, and a type to cast it to, which must be a
3473<a href="#t_pointer">pointer</a> type.
3474
3475<h5>Semantics:</h5>
3476<p>The '<tt>inttoptr</tt>' instruction converts <tt>value</tt> to type
3477<tt>ty2</tt> by applying either a zero extension or a truncation depending on
3478the size of the integer <tt>value</tt>. If <tt>value</tt> is larger than the
3479size of a pointer then a truncation is done. If <tt>value</tt> is smaller than
3480the size of a pointer then a zero extension is done. If they are the same size,
3481nothing is done (<i>no-op cast</i>).</p>
3482
3483<h5>Example:</h5>
3484<pre>
3485 %X = inttoptr i32 255 to i32* <i>; yields zero extension on 64-bit architecture</i>
3486 %X = inttoptr i32 255 to i32* <i>; yields no-op on 32-bit architecture</i>
3487 %Y = inttoptr i64 0 to i32* <i>; yields truncation on 32-bit architecture</i>
3488</pre>
3489</div>
3490
3491<!-- _______________________________________________________________________ -->
3492<div class="doc_subsubsection">
3493 <a name="i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a>
3494</div>
3495<div class="doc_text">
3496
3497<h5>Syntax:</h5>
3498<pre>
3499 &lt;result&gt; = bitcast &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3500</pre>
3501
3502<h5>Overview:</h5>
3503<p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
3504<tt>ty2</tt> without changing any bits.</p>
3505
3506<h5>Arguments:</h5>
3507<p>The '<tt>bitcast</tt>' instruction takes a value to cast, which must be
3508a first class value, and a type to cast it to, which must also be a <a
3509 href="#t_firstclass">first class</a> type. The bit sizes of <tt>value</tt>
3510and the destination type, <tt>ty2</tt>, must be identical. If the source
3511type is a pointer, the destination type must also be a pointer.</p>
3512
3513<h5>Semantics:</h5>
3514<p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
3515<tt>ty2</tt>. It is always a <i>no-op cast</i> because no bits change with
3516this conversion. The conversion is done as if the <tt>value</tt> had been
3517stored to memory and read back as type <tt>ty2</tt>. Pointer types may only be
3518converted to other pointer types with this instruction. To convert pointers to
3519other types, use the <a href="#i_inttoptr">inttoptr</a> or
3520<a href="#i_ptrtoint">ptrtoint</a> instructions first.</p>
3521
3522<h5>Example:</h5>
3523<pre>
3524 %X = bitcast i8 255 to i8 <i>; yields i8 :-1</i>
3525 %Y = bitcast i32* %x to sint* <i>; yields sint*:%x</i>
3526 %Z = bitcast <2xint> %V to i64; <i>; yields i64: %V</i>
3527</pre>
3528</div>
3529
3530<!-- ======================================================================= -->
3531<div class="doc_subsection"> <a name="otherops">Other Operations</a> </div>
3532<div class="doc_text">
3533<p>The instructions in this category are the "miscellaneous"
3534instructions, which defy better classification.</p>
3535</div>
3536
3537<!-- _______________________________________________________________________ -->
3538<div class="doc_subsubsection"><a name="i_icmp">'<tt>icmp</tt>' Instruction</a>
3539</div>
3540<div class="doc_text">
3541<h5>Syntax:</h5>
3542<pre> &lt;result&gt; = icmp &lt;cond&gt; &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {i1}:result</i>
3543</pre>
3544<h5>Overview:</h5>
3545<p>The '<tt>icmp</tt>' instruction returns a boolean value based on comparison
3546of its two integer operands.</p>
3547<h5>Arguments:</h5>
3548<p>The '<tt>icmp</tt>' instruction takes three operands. The first operand is
3549the condition code indicating the kind of comparison to perform. It is not
3550a value, just a keyword. The possible condition code are:
3551<ol>
3552 <li><tt>eq</tt>: equal</li>
3553 <li><tt>ne</tt>: not equal </li>
3554 <li><tt>ugt</tt>: unsigned greater than</li>
3555 <li><tt>uge</tt>: unsigned greater or equal</li>
3556 <li><tt>ult</tt>: unsigned less than</li>
3557 <li><tt>ule</tt>: unsigned less or equal</li>
3558 <li><tt>sgt</tt>: signed greater than</li>
3559 <li><tt>sge</tt>: signed greater or equal</li>
3560 <li><tt>slt</tt>: signed less than</li>
3561 <li><tt>sle</tt>: signed less or equal</li>
3562</ol>
3563<p>The remaining two arguments must be <a href="#t_integer">integer</a> or
3564<a href="#t_pointer">pointer</a> typed. They must also be identical types.</p>
3565<h5>Semantics:</h5>
3566<p>The '<tt>icmp</tt>' compares <tt>var1</tt> and <tt>var2</tt> according to
3567the condition code given as <tt>cond</tt>. The comparison performed always
3568yields a <a href="#t_primitive">i1</a> result, as follows:
3569<ol>
3570 <li><tt>eq</tt>: yields <tt>true</tt> if the operands are equal,
3571 <tt>false</tt> otherwise. No sign interpretation is necessary or performed.
3572 </li>
3573 <li><tt>ne</tt>: yields <tt>true</tt> if the operands are unequal,
3574 <tt>false</tt> otherwise. No sign interpretation is necessary or performed.
3575 <li><tt>ugt</tt>: interprets the operands as unsigned values and yields
3576 <tt>true</tt> if <tt>var1</tt> is greater than <tt>var2</tt>.</li>
3577 <li><tt>uge</tt>: interprets the operands as unsigned values and yields
3578 <tt>true</tt> if <tt>var1</tt> is greater than or equal to <tt>var2</tt>.</li>
3579 <li><tt>ult</tt>: interprets the operands as unsigned values and yields
3580 <tt>true</tt> if <tt>var1</tt> is less than <tt>var2</tt>.</li>
3581 <li><tt>ule</tt>: interprets the operands as unsigned values and yields
3582 <tt>true</tt> if <tt>var1</tt> is less than or equal to <tt>var2</tt>.</li>
3583 <li><tt>sgt</tt>: interprets the operands as signed values and yields
3584 <tt>true</tt> if <tt>var1</tt> is greater than <tt>var2</tt>.</li>
3585 <li><tt>sge</tt>: interprets the operands as signed values and yields
3586 <tt>true</tt> if <tt>var1</tt> is greater than or equal to <tt>var2</tt>.</li>
3587 <li><tt>slt</tt>: interprets the operands as signed values and yields
3588 <tt>true</tt> if <tt>var1</tt> is less than <tt>var2</tt>.</li>
3589 <li><tt>sle</tt>: interprets the operands as signed values and yields
3590 <tt>true</tt> if <tt>var1</tt> is less than or equal to <tt>var2</tt>.</li>
3591</ol>
3592<p>If the operands are <a href="#t_pointer">pointer</a> typed, the pointer
3593values are compared as if they were integers.</p>
3594
3595<h5>Example:</h5>
3596<pre> &lt;result&gt; = icmp eq i32 4, 5 <i>; yields: result=false</i>
3597 &lt;result&gt; = icmp ne float* %X, %X <i>; yields: result=false</i>
3598 &lt;result&gt; = icmp ult i16 4, 5 <i>; yields: result=true</i>
3599 &lt;result&gt; = icmp sgt i16 4, 5 <i>; yields: result=false</i>
3600 &lt;result&gt; = icmp ule i16 -4, 5 <i>; yields: result=false</i>
3601 &lt;result&gt; = icmp sge i16 4, 5 <i>; yields: result=false</i>
3602</pre>
3603</div>
3604
3605<!-- _______________________________________________________________________ -->
3606<div class="doc_subsubsection"><a name="i_fcmp">'<tt>fcmp</tt>' Instruction</a>
3607</div>
3608<div class="doc_text">
3609<h5>Syntax:</h5>
3610<pre> &lt;result&gt; = fcmp &lt;cond&gt; &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {i1}:result</i>
3611</pre>
3612<h5>Overview:</h5>
3613<p>The '<tt>fcmp</tt>' instruction returns a boolean value based on comparison
3614of its floating point operands.</p>
3615<h5>Arguments:</h5>
3616<p>The '<tt>fcmp</tt>' instruction takes three operands. The first operand is
3617the condition code indicating the kind of comparison to perform. It is not
3618a value, just a keyword. The possible condition code are:
3619<ol>
3620 <li><tt>false</tt>: no comparison, always returns false</li>
3621 <li><tt>oeq</tt>: ordered and equal</li>
3622 <li><tt>ogt</tt>: ordered and greater than </li>
3623 <li><tt>oge</tt>: ordered and greater than or equal</li>
3624 <li><tt>olt</tt>: ordered and less than </li>
3625 <li><tt>ole</tt>: ordered and less than or equal</li>
3626 <li><tt>one</tt>: ordered and not equal</li>
3627 <li><tt>ord</tt>: ordered (no nans)</li>
3628 <li><tt>ueq</tt>: unordered or equal</li>
3629 <li><tt>ugt</tt>: unordered or greater than </li>
3630 <li><tt>uge</tt>: unordered or greater than or equal</li>
3631 <li><tt>ult</tt>: unordered or less than </li>
3632 <li><tt>ule</tt>: unordered or less than or equal</li>
3633 <li><tt>une</tt>: unordered or not equal</li>
3634 <li><tt>uno</tt>: unordered (either nans)</li>
3635 <li><tt>true</tt>: no comparison, always returns true</li>
3636</ol>
3637<p><i>Ordered</i> means that neither operand is a QNAN while
3638<i>unordered</i> means that either operand may be a QNAN.</p>
3639<p>The <tt>val1</tt> and <tt>val2</tt> arguments must be
3640<a href="#t_floating">floating point</a> typed. They must have identical
3641types.</p>
3642<h5>Semantics:</h5>
3643<p>The '<tt>fcmp</tt>' compares <tt>var1</tt> and <tt>var2</tt> according to
3644the condition code given as <tt>cond</tt>. The comparison performed always
3645yields a <a href="#t_primitive">i1</a> result, as follows:
3646<ol>
3647 <li><tt>false</tt>: always yields <tt>false</tt>, regardless of operands.</li>
3648 <li><tt>oeq</tt>: yields <tt>true</tt> if both operands are not a QNAN and
3649 <tt>var1</tt> is equal to <tt>var2</tt>.</li>
3650 <li><tt>ogt</tt>: yields <tt>true</tt> if both operands are not a QNAN and
3651 <tt>var1</tt> is greather than <tt>var2</tt>.</li>
3652 <li><tt>oge</tt>: yields <tt>true</tt> if both operands are not a QNAN and
3653 <tt>var1</tt> is greater than or equal to <tt>var2</tt>.</li>
3654 <li><tt>olt</tt>: yields <tt>true</tt> if both operands are not a QNAN and
3655 <tt>var1</tt> is less than <tt>var2</tt>.</li>
3656 <li><tt>ole</tt>: yields <tt>true</tt> if both operands are not a QNAN and
3657 <tt>var1</tt> is less than or equal to <tt>var2</tt>.</li>
3658 <li><tt>one</tt>: yields <tt>true</tt> if both operands are not a QNAN and
3659 <tt>var1</tt> is not equal to <tt>var2</tt>.</li>
3660 <li><tt>ord</tt>: yields <tt>true</tt> if both operands are not a QNAN.</li>
3661 <li><tt>ueq</tt>: yields <tt>true</tt> if either operand is a QNAN or
3662 <tt>var1</tt> is equal to <tt>var2</tt>.</li>
3663 <li><tt>ugt</tt>: yields <tt>true</tt> if either operand is a QNAN or
3664 <tt>var1</tt> is greater than <tt>var2</tt>.</li>
3665 <li><tt>uge</tt>: yields <tt>true</tt> if either operand is a QNAN or
3666 <tt>var1</tt> is greater than or equal to <tt>var2</tt>.</li>
3667 <li><tt>ult</tt>: yields <tt>true</tt> if either operand is a QNAN or
3668 <tt>var1</tt> is less than <tt>var2</tt>.</li>
3669 <li><tt>ule</tt>: yields <tt>true</tt> if either operand is a QNAN or
3670 <tt>var1</tt> is less than or equal to <tt>var2</tt>.</li>
3671 <li><tt>une</tt>: yields <tt>true</tt> if either operand is a QNAN or
3672 <tt>var1</tt> is not equal to <tt>var2</tt>.</li>
3673 <li><tt>uno</tt>: yields <tt>true</tt> if either operand is a QNAN.</li>
3674 <li><tt>true</tt>: always yields <tt>true</tt>, regardless of operands.</li>
3675</ol>
3676
3677<h5>Example:</h5>
3678<pre> &lt;result&gt; = fcmp oeq float 4.0, 5.0 <i>; yields: result=false</i>
3679 &lt;result&gt; = icmp one float 4.0, 5.0 <i>; yields: result=true</i>
3680 &lt;result&gt; = icmp olt float 4.0, 5.0 <i>; yields: result=true</i>
3681 &lt;result&gt; = icmp ueq double 1.0, 2.0 <i>; yields: result=false</i>
3682</pre>
3683</div>
3684
3685<!-- _______________________________________________________________________ -->
3686<div class="doc_subsubsection"> <a name="i_phi">'<tt>phi</tt>'
3687Instruction</a> </div>
3688<div class="doc_text">
3689<h5>Syntax:</h5>
3690<pre> &lt;result&gt; = phi &lt;ty&gt; [ &lt;val0&gt;, &lt;label0&gt;], ...<br></pre>
3691<h5>Overview:</h5>
3692<p>The '<tt>phi</tt>' instruction is used to implement the &#966; node in
3693the SSA graph representing the function.</p>
3694<h5>Arguments:</h5>
3695<p>The type of the incoming values is specified with the first type
3696field. After this, the '<tt>phi</tt>' instruction takes a list of pairs
3697as arguments, with one pair for each predecessor basic block of the
3698current block. Only values of <a href="#t_firstclass">first class</a>
3699type may be used as the value arguments to the PHI node. Only labels
3700may be used as the label arguments.</p>
3701<p>There must be no non-phi instructions between the start of a basic
3702block and the PHI instructions: i.e. PHI instructions must be first in
3703a basic block.</p>
3704<h5>Semantics:</h5>
3705<p>At runtime, the '<tt>phi</tt>' instruction logically takes on the value
3706specified by the pair corresponding to the predecessor basic block that executed
3707just prior to the current block.</p>
3708<h5>Example:</h5>
3709<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>
3710</div>
3711
3712<!-- _______________________________________________________________________ -->
3713<div class="doc_subsubsection">
3714 <a name="i_select">'<tt>select</tt>' Instruction</a>
3715</div>
3716
3717<div class="doc_text">
3718
3719<h5>Syntax:</h5>
3720
3721<pre>
3722 &lt;result&gt; = select i1 &lt;cond&gt;, &lt;ty&gt; &lt;val1&gt;, &lt;ty&gt; &lt;val2&gt; <i>; yields ty</i>
3723</pre>
3724
3725<h5>Overview:</h5>
3726
3727<p>
3728The '<tt>select</tt>' instruction is used to choose one value based on a
3729condition, without branching.
3730</p>
3731
3732
3733<h5>Arguments:</h5>
3734
3735<p>
3736The '<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.
3737</p>
3738
3739<h5>Semantics:</h5>
3740
3741<p>
3742If the boolean condition evaluates to true, the instruction returns the first
3743value argument; otherwise, it returns the second value argument.
3744</p>
3745
3746<h5>Example:</h5>
3747
3748<pre>
3749 %X = select i1 true, i8 17, i8 42 <i>; yields i8:17</i>
3750</pre>
3751</div>
3752
3753
3754<!-- _______________________________________________________________________ -->
3755<div class="doc_subsubsection">
3756 <a name="i_call">'<tt>call</tt>' Instruction</a>
3757</div>
3758
3759<div class="doc_text">
3760
3761<h5>Syntax:</h5>
3762<pre>
Nick Lewycky93082fc2007-09-08 13:57:50 +0000[diff] [blame]3763 &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]3764</pre>
3765
3766<h5>Overview:</h5>
3767
3768<p>The '<tt>call</tt>' instruction represents a simple function call.</p>
3769
3770<h5>Arguments:</h5>
3771
3772<p>This instruction requires several arguments:</p>
3773
3774<ol>
3775 <li>
3776 <p>The optional "tail" marker indicates whether the callee function accesses
3777 any allocas or varargs in the caller. If the "tail" marker is present, the
3778 function call is eligible for tail call optimization. Note that calls may
3779 be marked "tail" even if they do not occur before a <a
3780 href="#i_ret"><tt>ret</tt></a> instruction.
3781 </li>
3782 <li>
3783 <p>The optional "cconv" marker indicates which <a href="#callingconv">calling
3784 convention</a> the call should use. If none is specified, the call defaults
3785 to using C calling conventions.
3786 </li>
3787 <li>
Nick Lewycky93082fc2007-09-08 13:57:50 +0000[diff] [blame]3788 <p>'<tt>ty</tt>': the type of the call instruction itself which is also
3789 the type of the return value. Functions that return no value are marked
3790 <tt><a href="#t_void">void</a></tt>.</p>
3791 </li>
3792 <li>
3793 <p>'<tt>fnty</tt>': shall be the signature of the pointer to function
3794 value being invoked. The argument types must match the types implied by
3795 this signature. This type can be omitted if the function is not varargs
3796 and if the function type does not return a pointer to a function.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3797 </li>
3798 <li>
3799 <p>'<tt>fnptrval</tt>': An LLVM value containing a pointer to a function to
3800 be invoked. In most cases, this is a direct function invocation, but
3801 indirect <tt>call</tt>s are just as possible, calling an arbitrary pointer
3802 to function value.</p>
3803 </li>
3804 <li>
3805 <p>'<tt>function args</tt>': argument list whose types match the
3806 function signature argument types. All arguments must be of
3807 <a href="#t_firstclass">first class</a> type. If the function signature
3808 indicates the function accepts a variable number of arguments, the extra
3809 arguments can be specified.</p>
3810 </li>
3811</ol>
3812
3813<h5>Semantics:</h5>
3814
3815<p>The '<tt>call</tt>' instruction is used to cause control flow to
3816transfer to a specified function, with its incoming arguments bound to
3817the specified values. Upon a '<tt><a href="#i_ret">ret</a></tt>'
3818instruction in the called function, control flow continues with the
3819instruction after the function call, and the return value of the
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame^]3820function is bound to the result argument. If the '<tt><a href="#i_ret">ret</a>
3821</tt>' instruction returns multiple values then the return value of the
3822function is only accessed through '<tt><a href="#i_getresult">getresult</a>
3823</tt>' instruction. This is a simpler case of
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3824the <a href="#i_invoke">invoke</a> instruction.</p>
3825
3826<h5>Example:</h5>
3827
3828<pre>
Nick Lewycky93082fc2007-09-08 13:57:50 +0000[diff] [blame]3829 %retval = call i32 @test(i32 %argc)
3830 call i32 (i8 *, ...)* @printf(i8 * %msg, i32 12, i8 42);
3831 %X = tail call i32 @foo()
3832 %Y = tail call <a href="#callingconv">fastcc</a> i32 @foo()
3833 %Z = call void %foo(i8 97 signext)
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame^]3834
3835 %struct.A = type { i32, i8 }
3836 %r = call %struct.A @foo()
3837 %gr = getresult %struct.A %r, 0
3838 %gr1 = getresult %struct.A %r, 1
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3839</pre>
3840
3841</div>
3842
3843<!-- _______________________________________________________________________ -->
3844<div class="doc_subsubsection">
3845 <a name="i_va_arg">'<tt>va_arg</tt>' Instruction</a>
3846</div>
3847
3848<div class="doc_text">
3849
3850<h5>Syntax:</h5>
3851
3852<pre>
3853 &lt;resultval&gt; = va_arg &lt;va_list*&gt; &lt;arglist&gt;, &lt;argty&gt;
3854</pre>
3855
3856<h5>Overview:</h5>
3857
3858<p>The '<tt>va_arg</tt>' instruction is used to access arguments passed through
3859the "variable argument" area of a function call. It is used to implement the
3860<tt>va_arg</tt> macro in C.</p>
3861
3862<h5>Arguments:</h5>
3863
3864<p>This instruction takes a <tt>va_list*</tt> value and the type of
3865the argument. It returns a value of the specified argument type and
3866increments the <tt>va_list</tt> to point to the next argument. The
3867actual type of <tt>va_list</tt> is target specific.</p>
3868
3869<h5>Semantics:</h5>
3870
3871<p>The '<tt>va_arg</tt>' instruction loads an argument of the specified
3872type from the specified <tt>va_list</tt> and causes the
3873<tt>va_list</tt> to point to the next argument. For more information,
3874see the variable argument handling <a href="#int_varargs">Intrinsic
3875Functions</a>.</p>
3876
3877<p>It is legal for this instruction to be called in a function which does not
3878take a variable number of arguments, for example, the <tt>vfprintf</tt>
3879function.</p>
3880
3881<p><tt>va_arg</tt> is an LLVM instruction instead of an <a
3882href="#intrinsics">intrinsic function</a> because it takes a type as an
3883argument.</p>
3884
3885<h5>Example:</h5>
3886
3887<p>See the <a href="#int_varargs">variable argument processing</a> section.</p>
3888
3889</div>
3890
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame^]3891<!-- _______________________________________________________________________ -->
3892<div class="doc_subsubsection">
3893 <a name="i_getresult">'<tt>getresult</tt>' Instruction</a>
3894</div>
3895
3896<div class="doc_text">
3897
3898<h5>Syntax:</h5>
3899<pre>
3900 &lt;resultval&gt; = getresult &lt;type&gt; &lt;retval&gt;, &lt;index&gt;
3901</pre>
3902<h5>Overview:</h5>
3903
3904<p> The '<tt>getresult</tt>' instruction is used to extract individual values
3905from multiple values returned by '<tt><a href="#i_call">call</a></tt>'
3906or '<tt><a href="#i_invoke">invoke</a></tt>' instruction.
3907
3908<h5>Arguments:</h5>
3909
3910The '<tt>getresult</tt>' instruction takes return value as first argument. The
3911value must have <a href="#t_struct">structure type</a>. The second argument
3912is unsigned index value.
3913
3914<h5>Semantics:</h5>
3915
3916The '<tt>getresult</tt>' instruction extracts element identified by
3917'<tt>index</tt>' from the aggregate value.
3918
3919<h5>Example:</h5>
3920
3921<pre>
3922 %struct.A = type { i32, i8 }
3923
3924 %r = call %struct.A @foo()
3925 %gr = getresult %struct.A %r, 0
3926 %gr1 = getresult %struct.A %r, 1
3927 add i32 %gr, 42
3928 add i8 %gr1, 41
3929</pre>
3930
3931</div>
3932
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3933<!-- *********************************************************************** -->
3934<div class="doc_section"> <a name="intrinsics">Intrinsic Functions</a> </div>
3935<!-- *********************************************************************** -->
3936
3937<div class="doc_text">
3938
3939<p>LLVM supports the notion of an "intrinsic function". These functions have
3940well known names and semantics and are required to follow certain restrictions.
3941Overall, these intrinsics represent an extension mechanism for the LLVM
3942language that does not require changing all of the transformations in LLVM when
3943adding to the language (or the bitcode reader/writer, the parser, etc...).</p>
3944
3945<p>Intrinsic function names must all start with an "<tt>llvm.</tt>" prefix. This
3946prefix is reserved in LLVM for intrinsic names; thus, function names may not
3947begin with this prefix. Intrinsic functions must always be external functions:
3948you cannot define the body of intrinsic functions. Intrinsic functions may
3949only be used in call or invoke instructions: it is illegal to take the address
3950of an intrinsic function. Additionally, because intrinsic functions are part
3951of the LLVM language, it is required if any are added that they be documented
3952here.</p>
3953
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]3954<p>Some intrinsic functions can be overloaded, i.e., the intrinsic represents
3955a family of functions that perform the same operation but on different data
3956types. Because LLVM can represent over 8 million different integer types,
3957overloading is used commonly to allow an intrinsic function to operate on any
3958integer type. One or more of the argument types or the result type can be
3959overloaded to accept any integer type. Argument types may also be defined as
3960exactly matching a previous argument's type or the result type. This allows an
3961intrinsic function which accepts multiple arguments, but needs all of them to
3962be of the same type, to only be overloaded with respect to a single argument or
3963the result.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3964
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]3965<p>Overloaded intrinsics will have the names of its overloaded argument types
3966encoded into its function name, each preceded by a period. Only those types
3967which are overloaded result in a name suffix. Arguments whose type is matched
3968against another type do not. For example, the <tt>llvm.ctpop</tt> function can
3969take an integer of any width and returns an integer of exactly the same integer
3970width. This leads to a family of functions such as
3971<tt>i8 @llvm.ctpop.i8(i8 %val)</tt> and <tt>i29 @llvm.ctpop.i29(i29 %val)</tt>.
3972Only one type, the return type, is overloaded, and only one type suffix is
3973required. Because the argument's type is matched against the return type, it
3974does not require its own name suffix.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3975
3976<p>To learn how to add an intrinsic function, please see the
3977<a href="ExtendingLLVM.html">Extending LLVM Guide</a>.
3978</p>
3979
3980</div>
3981
3982<!-- ======================================================================= -->
3983<div class="doc_subsection">
3984 <a name="int_varargs">Variable Argument Handling Intrinsics</a>
3985</div>
3986
3987<div class="doc_text">
3988
3989<p>Variable argument support is defined in LLVM with the <a
3990 href="#i_va_arg"><tt>va_arg</tt></a> instruction and these three
3991intrinsic functions. These functions are related to the similarly
3992named macros defined in the <tt>&lt;stdarg.h&gt;</tt> header file.</p>
3993
3994<p>All of these functions operate on arguments that use a
3995target-specific value type "<tt>va_list</tt>". The LLVM assembly
3996language reference manual does not define what this type is, so all
3997transformations should be prepared to handle these functions regardless of
3998the type used.</p>
3999
4000<p>This example shows how the <a href="#i_va_arg"><tt>va_arg</tt></a>
4001instruction and the variable argument handling intrinsic functions are
4002used.</p>
4003
4004<div class="doc_code">
4005<pre>
4006define i32 @test(i32 %X, ...) {
4007 ; Initialize variable argument processing
4008 %ap = alloca i8*
4009 %ap2 = bitcast i8** %ap to i8*
4010 call void @llvm.va_start(i8* %ap2)
4011
4012 ; Read a single integer argument
4013 %tmp = va_arg i8** %ap, i32
4014
4015 ; Demonstrate usage of llvm.va_copy and llvm.va_end
4016 %aq = alloca i8*
4017 %aq2 = bitcast i8** %aq to i8*
4018 call void @llvm.va_copy(i8* %aq2, i8* %ap2)
4019 call void @llvm.va_end(i8* %aq2)
4020
4021 ; Stop processing of arguments.
4022 call void @llvm.va_end(i8* %ap2)
4023 ret i32 %tmp
4024}
4025
4026declare void @llvm.va_start(i8*)
4027declare void @llvm.va_copy(i8*, i8*)
4028declare void @llvm.va_end(i8*)
4029</pre>
4030</div>
4031
4032</div>
4033
4034<!-- _______________________________________________________________________ -->
4035<div class="doc_subsubsection">
4036 <a name="int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a>
4037</div>
4038
4039
4040<div class="doc_text">
4041<h5>Syntax:</h5>
4042<pre> declare void %llvm.va_start(i8* &lt;arglist&gt;)<br></pre>
4043<h5>Overview:</h5>
4044<P>The '<tt>llvm.va_start</tt>' intrinsic initializes
4045<tt>*&lt;arglist&gt;</tt> for subsequent use by <tt><a
4046href="#i_va_arg">va_arg</a></tt>.</p>
4047
4048<h5>Arguments:</h5>
4049
4050<P>The argument is a pointer to a <tt>va_list</tt> element to initialize.</p>
4051
4052<h5>Semantics:</h5>
4053
4054<P>The '<tt>llvm.va_start</tt>' intrinsic works just like the <tt>va_start</tt>
4055macro available in C. In a target-dependent way, it initializes the
4056<tt>va_list</tt> element to which the argument points, so that the next call to
4057<tt>va_arg</tt> will produce the first variable argument passed to the function.
4058Unlike the C <tt>va_start</tt> macro, this intrinsic does not need to know the
4059last argument of the function as the compiler can figure that out.</p>
4060
4061</div>
4062
4063<!-- _______________________________________________________________________ -->
4064<div class="doc_subsubsection">
4065 <a name="int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a>
4066</div>
4067
4068<div class="doc_text">
4069<h5>Syntax:</h5>
4070<pre> declare void @llvm.va_end(i8* &lt;arglist&gt;)<br></pre>
4071<h5>Overview:</h5>
4072
4073<p>The '<tt>llvm.va_end</tt>' intrinsic destroys <tt>*&lt;arglist&gt;</tt>,
4074which has been initialized previously with <tt><a href="#int_va_start">llvm.va_start</a></tt>
4075or <tt><a href="#i_va_copy">llvm.va_copy</a></tt>.</p>
4076
4077<h5>Arguments:</h5>
4078
4079<p>The argument is a pointer to a <tt>va_list</tt> to destroy.</p>
4080
4081<h5>Semantics:</h5>
4082
4083<p>The '<tt>llvm.va_end</tt>' intrinsic works just like the <tt>va_end</tt>
4084macro available in C. In a target-dependent way, it destroys the
4085<tt>va_list</tt> element to which the argument points. Calls to <a
4086href="#int_va_start"><tt>llvm.va_start</tt></a> and <a href="#int_va_copy">
4087<tt>llvm.va_copy</tt></a> must be matched exactly with calls to
4088<tt>llvm.va_end</tt>.</p>
4089
4090</div>
4091
4092<!-- _______________________________________________________________________ -->
4093<div class="doc_subsubsection">
4094 <a name="int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a>
4095</div>
4096
4097<div class="doc_text">
4098
4099<h5>Syntax:</h5>
4100
4101<pre>
4102 declare void @llvm.va_copy(i8* &lt;destarglist&gt;, i8* &lt;srcarglist&gt;)
4103</pre>
4104
4105<h5>Overview:</h5>
4106
4107<p>The '<tt>llvm.va_copy</tt>' intrinsic copies the current argument position
4108from the source argument list to the destination argument list.</p>
4109
4110<h5>Arguments:</h5>
4111
4112<p>The first argument is a pointer to a <tt>va_list</tt> element to initialize.
4113The second argument is a pointer to a <tt>va_list</tt> element to copy from.</p>
4114
4115
4116<h5>Semantics:</h5>
4117
4118<p>The '<tt>llvm.va_copy</tt>' intrinsic works just like the <tt>va_copy</tt>
4119macro available in C. In a target-dependent way, it copies the source
4120<tt>va_list</tt> element into the destination <tt>va_list</tt> element. This
4121intrinsic is necessary because the <tt><a href="#int_va_start">
4122llvm.va_start</a></tt> intrinsic may be arbitrarily complex and require, for
4123example, memory allocation.</p>
4124
4125</div>
4126
4127<!-- ======================================================================= -->
4128<div class="doc_subsection">
4129 <a name="int_gc">Accurate Garbage Collection Intrinsics</a>
4130</div>
4131
4132<div class="doc_text">
4133
4134<p>
4135LLVM support for <a href="GarbageCollection.html">Accurate Garbage
4136Collection</a> requires the implementation and generation of these intrinsics.
4137These intrinsics allow identification of <a href="#int_gcroot">GC roots on the
4138stack</a>, as well as garbage collector implementations that require <a
4139href="#int_gcread">read</a> and <a href="#int_gcwrite">write</a> barriers.
4140Front-ends for type-safe garbage collected languages should generate these
4141intrinsics to make use of the LLVM garbage collectors. For more details, see <a
4142href="GarbageCollection.html">Accurate Garbage Collection with LLVM</a>.
4143</p>
Christopher Lambcfe00962007-12-17 01:00:21 +0000[diff] [blame]4144
4145<p>The garbage collection intrinsics only operate on objects in the generic
4146 address space (address space zero).</p>
4147
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4148</div>
4149
4150<!-- _______________________________________________________________________ -->
4151<div class="doc_subsubsection">
4152 <a name="int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a>
4153</div>
4154
4155<div class="doc_text">
4156
4157<h5>Syntax:</h5>
4158
4159<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4160 declare void @llvm.gcroot(i8** %ptrloc, i8* %metadata)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4161</pre>
4162
4163<h5>Overview:</h5>
4164
4165<p>The '<tt>llvm.gcroot</tt>' intrinsic declares the existence of a GC root to
4166the code generator, and allows some metadata to be associated with it.</p>
4167
4168<h5>Arguments:</h5>
4169
4170<p>The first argument specifies the address of a stack object that contains the
4171root pointer. The second pointer (which must be either a constant or a global
4172value address) contains the meta-data to be associated with the root.</p>
4173
4174<h5>Semantics:</h5>
4175
4176<p>At runtime, a call to this intrinsics stores a null pointer into the "ptrloc"
4177location. At compile-time, the code generator generates information to allow
Gordon Henriksen40393542007-12-25 02:31:26 +0000[diff] [blame]4178the runtime to find the pointer at GC safe points. The '<tt>llvm.gcroot</tt>'
4179intrinsic may only be used in a function which <a href="#gc">specifies a GC
4180algorithm</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4181
4182</div>
4183
4184
4185<!-- _______________________________________________________________________ -->
4186<div class="doc_subsubsection">
4187 <a name="int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a>
4188</div>
4189
4190<div class="doc_text">
4191
4192<h5>Syntax:</h5>
4193
4194<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4195 declare i8* @llvm.gcread(i8* %ObjPtr, i8** %Ptr)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4196</pre>
4197
4198<h5>Overview:</h5>
4199
4200<p>The '<tt>llvm.gcread</tt>' intrinsic identifies reads of references from heap
4201locations, allowing garbage collector implementations that require read
4202barriers.</p>
4203
4204<h5>Arguments:</h5>
4205
4206<p>The second argument is the address to read from, which should be an address
4207allocated from the garbage collector. The first object is a pointer to the
4208start of the referenced object, if needed by the language runtime (otherwise
4209null).</p>
4210
4211<h5>Semantics:</h5>
4212
4213<p>The '<tt>llvm.gcread</tt>' intrinsic has the same semantics as a load
4214instruction, but may be replaced with substantially more complex code by the
Gordon Henriksen40393542007-12-25 02:31:26 +0000[diff] [blame]4215garbage collector runtime, as needed. The '<tt>llvm.gcread</tt>' intrinsic
4216may only be used in a function which <a href="#gc">specifies a GC
4217algorithm</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4218
4219</div>
4220
4221
4222<!-- _______________________________________________________________________ -->
4223<div class="doc_subsubsection">
4224 <a name="int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a>
4225</div>
4226
4227<div class="doc_text">
4228
4229<h5>Syntax:</h5>
4230
4231<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4232 declare void @llvm.gcwrite(i8* %P1, i8* %Obj, i8** %P2)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4233</pre>
4234
4235<h5>Overview:</h5>
4236
4237<p>The '<tt>llvm.gcwrite</tt>' intrinsic identifies writes of references to heap
4238locations, allowing garbage collector implementations that require write
4239barriers (such as generational or reference counting collectors).</p>
4240
4241<h5>Arguments:</h5>
4242
4243<p>The first argument is the reference to store, the second is the start of the
4244object to store it to, and the third is the address of the field of Obj to
4245store to. If the runtime does not require a pointer to the object, Obj may be
4246null.</p>
4247
4248<h5>Semantics:</h5>
4249
4250<p>The '<tt>llvm.gcwrite</tt>' intrinsic has the same semantics as a store
4251instruction, but may be replaced with substantially more complex code by the
Gordon Henriksen40393542007-12-25 02:31:26 +0000[diff] [blame]4252garbage collector runtime, as needed. The '<tt>llvm.gcwrite</tt>' intrinsic
4253may only be used in a function which <a href="#gc">specifies a GC
4254algorithm</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4255
4256</div>
4257
4258
4259
4260<!-- ======================================================================= -->
4261<div class="doc_subsection">
4262 <a name="int_codegen">Code Generator Intrinsics</a>
4263</div>
4264
4265<div class="doc_text">
4266<p>
4267These intrinsics are provided by LLVM to expose special features that may only
4268be implemented with code generator support.
4269</p>
4270
4271</div>
4272
4273<!-- _______________________________________________________________________ -->
4274<div class="doc_subsubsection">
4275 <a name="int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a>
4276</div>
4277
4278<div class="doc_text">
4279
4280<h5>Syntax:</h5>
4281<pre>
4282 declare i8 *@llvm.returnaddress(i32 &lt;level&gt;)
4283</pre>
4284
4285<h5>Overview:</h5>
4286
4287<p>
4288The '<tt>llvm.returnaddress</tt>' intrinsic attempts to compute a
4289target-specific value indicating the return address of the current function
4290or one of its callers.
4291</p>
4292
4293<h5>Arguments:</h5>
4294
4295<p>
4296The argument to this intrinsic indicates which function to return the address
4297for. Zero indicates the calling function, one indicates its caller, etc. The
4298argument is <b>required</b> to be a constant integer value.
4299</p>
4300
4301<h5>Semantics:</h5>
4302
4303<p>
4304The '<tt>llvm.returnaddress</tt>' intrinsic either returns a pointer indicating
4305the return address of the specified call frame, or zero if it cannot be
4306identified. The value returned by this intrinsic is likely to be incorrect or 0
4307for arguments other than zero, so it should only be used for debugging purposes.
4308</p>
4309
4310<p>
4311Note that calling this intrinsic does not prevent function inlining or other
4312aggressive transformations, so the value returned may not be that of the obvious
4313source-language caller.
4314</p>
4315</div>
4316
4317
4318<!-- _______________________________________________________________________ -->
4319<div class="doc_subsubsection">
4320 <a name="int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a>
4321</div>
4322
4323<div class="doc_text">
4324
4325<h5>Syntax:</h5>
4326<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4327 declare i8 *@llvm.frameaddress(i32 &lt;level&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4328</pre>
4329
4330<h5>Overview:</h5>
4331
4332<p>
4333The '<tt>llvm.frameaddress</tt>' intrinsic attempts to return the
4334target-specific frame pointer value for the specified stack frame.
4335</p>
4336
4337<h5>Arguments:</h5>
4338
4339<p>
4340The argument to this intrinsic indicates which function to return the frame
4341pointer for. Zero indicates the calling function, one indicates its caller,
4342etc. The argument is <b>required</b> to be a constant integer value.
4343</p>
4344
4345<h5>Semantics:</h5>
4346
4347<p>
4348The '<tt>llvm.frameaddress</tt>' intrinsic either returns a pointer indicating
4349the frame address of the specified call frame, or zero if it cannot be
4350identified. The value returned by this intrinsic is likely to be incorrect or 0
4351for arguments other than zero, so it should only be used for debugging purposes.
4352</p>
4353
4354<p>
4355Note that calling this intrinsic does not prevent function inlining or other
4356aggressive transformations, so the value returned may not be that of the obvious
4357source-language caller.
4358</p>
4359</div>
4360
4361<!-- _______________________________________________________________________ -->
4362<div class="doc_subsubsection">
4363 <a name="int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a>
4364</div>
4365
4366<div class="doc_text">
4367
4368<h5>Syntax:</h5>
4369<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4370 declare i8 *@llvm.stacksave()
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4371</pre>
4372
4373<h5>Overview:</h5>
4374
4375<p>
4376The '<tt>llvm.stacksave</tt>' intrinsic is used to remember the current state of
4377the function stack, for use with <a href="#int_stackrestore">
4378<tt>llvm.stackrestore</tt></a>. This is useful for implementing language
4379features like scoped automatic variable sized arrays in C99.
4380</p>
4381
4382<h5>Semantics:</h5>
4383
4384<p>
4385This intrinsic returns a opaque pointer value that can be passed to <a
4386href="#int_stackrestore"><tt>llvm.stackrestore</tt></a>. When an
4387<tt>llvm.stackrestore</tt> intrinsic is executed with a value saved from
4388<tt>llvm.stacksave</tt>, it effectively restores the state of the stack to the
4389state it was in when the <tt>llvm.stacksave</tt> intrinsic executed. In
4390practice, this pops any <a href="#i_alloca">alloca</a> blocks from the stack
4391that were allocated after the <tt>llvm.stacksave</tt> was executed.
4392</p>
4393
4394</div>
4395
4396<!-- _______________________________________________________________________ -->
4397<div class="doc_subsubsection">
4398 <a name="int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a>
4399</div>
4400
4401<div class="doc_text">
4402
4403<h5>Syntax:</h5>
4404<pre>
4405 declare void @llvm.stackrestore(i8 * %ptr)
4406</pre>
4407
4408<h5>Overview:</h5>
4409
4410<p>
4411The '<tt>llvm.stackrestore</tt>' intrinsic is used to restore the state of
4412the function stack to the state it was in when the corresponding <a
4413href="#int_stacksave"><tt>llvm.stacksave</tt></a> intrinsic executed. This is
4414useful for implementing language features like scoped automatic variable sized
4415arrays in C99.
4416</p>
4417
4418<h5>Semantics:</h5>
4419
4420<p>
4421See the description for <a href="#int_stacksave"><tt>llvm.stacksave</tt></a>.
4422</p>
4423
4424</div>
4425
4426
4427<!-- _______________________________________________________________________ -->
4428<div class="doc_subsubsection">
4429 <a name="int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a>
4430</div>
4431
4432<div class="doc_text">
4433
4434<h5>Syntax:</h5>
4435<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4436 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]4437</pre>
4438
4439<h5>Overview:</h5>
4440
4441
4442<p>
4443The '<tt>llvm.prefetch</tt>' intrinsic is a hint to the code generator to insert
4444a prefetch instruction if supported; otherwise, it is a noop. Prefetches have
4445no
4446effect on the behavior of the program but can change its performance
4447characteristics.
4448</p>
4449
4450<h5>Arguments:</h5>
4451
4452<p>
4453<tt>address</tt> is the address to be prefetched, <tt>rw</tt> is the specifier
4454determining if the fetch should be for a read (0) or write (1), and
4455<tt>locality</tt> is a temporal locality specifier ranging from (0) - no
4456locality, to (3) - extremely local keep in cache. The <tt>rw</tt> and
4457<tt>locality</tt> arguments must be constant integers.
4458</p>
4459
4460<h5>Semantics:</h5>
4461
4462<p>
4463This intrinsic does not modify the behavior of the program. In particular,
4464prefetches cannot trap and do not produce a value. On targets that support this
4465intrinsic, the prefetch can provide hints to the processor cache for better
4466performance.
4467</p>
4468
4469</div>
4470
4471<!-- _______________________________________________________________________ -->
4472<div class="doc_subsubsection">
4473 <a name="int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a>
4474</div>
4475
4476<div class="doc_text">
4477
4478<h5>Syntax:</h5>
4479<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4480 declare void @llvm.pcmarker(i32 &lt;id&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4481</pre>
4482
4483<h5>Overview:</h5>
4484
4485
4486<p>
4487The '<tt>llvm.pcmarker</tt>' intrinsic is a method to export a Program Counter
4488(PC) in a region of
4489code to simulators and other tools. The method is target specific, but it is
4490expected that the marker will use exported symbols to transmit the PC of the marker.
4491The marker makes no guarantees that it will remain with any specific instruction
4492after optimizations. It is possible that the presence of a marker will inhibit
4493optimizations. The intended use is to be inserted after optimizations to allow
4494correlations of simulation runs.
4495</p>
4496
4497<h5>Arguments:</h5>
4498
4499<p>
4500<tt>id</tt> is a numerical id identifying the marker.
4501</p>
4502
4503<h5>Semantics:</h5>
4504
4505<p>
4506This intrinsic does not modify the behavior of the program. Backends that do not
4507support this intrinisic may ignore it.
4508</p>
4509
4510</div>
4511
4512<!-- _______________________________________________________________________ -->
4513<div class="doc_subsubsection">
4514 <a name="int_readcyclecounter">'<tt>llvm.readcyclecounter</tt>' Intrinsic</a>
4515</div>
4516
4517<div class="doc_text">
4518
4519<h5>Syntax:</h5>
4520<pre>
4521 declare i64 @llvm.readcyclecounter( )
4522</pre>
4523
4524<h5>Overview:</h5>
4525
4526
4527<p>
4528The '<tt>llvm.readcyclecounter</tt>' intrinsic provides access to the cycle
4529counter register (or similar low latency, high accuracy clocks) on those targets
4530that support it. On X86, it should map to RDTSC. On Alpha, it should map to RPCC.
4531As the backing counters overflow quickly (on the order of 9 seconds on alpha), this
4532should only be used for small timings.
4533</p>
4534
4535<h5>Semantics:</h5>
4536
4537<p>
4538When directly supported, reading the cycle counter should not modify any memory.
4539Implementations are allowed to either return a application specific value or a
4540system wide value. On backends without support, this is lowered to a constant 0.
4541</p>
4542
4543</div>
4544
4545<!-- ======================================================================= -->
4546<div class="doc_subsection">
4547 <a name="int_libc">Standard C Library Intrinsics</a>
4548</div>
4549
4550<div class="doc_text">
4551<p>
4552LLVM provides intrinsics for a few important standard C library functions.
4553These intrinsics allow source-language front-ends to pass information about the
4554alignment of the pointer arguments to the code generator, providing opportunity
4555for more efficient code generation.
4556</p>
4557
4558</div>
4559
4560<!-- _______________________________________________________________________ -->
4561<div class="doc_subsubsection">
4562 <a name="int_memcpy">'<tt>llvm.memcpy</tt>' Intrinsic</a>
4563</div>
4564
4565<div class="doc_text">
4566
4567<h5>Syntax:</h5>
4568<pre>
4569 declare void @llvm.memcpy.i32(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
4570 i32 &lt;len&gt;, i32 &lt;align&gt;)
4571 declare void @llvm.memcpy.i64(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
4572 i64 &lt;len&gt;, i32 &lt;align&gt;)
4573</pre>
4574
4575<h5>Overview:</h5>
4576
4577<p>
4578The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the source
4579location to the destination location.
4580</p>
4581
4582<p>
4583Note that, unlike the standard libc function, the <tt>llvm.memcpy.*</tt>
4584intrinsics do not return a value, and takes an extra alignment argument.
4585</p>
4586
4587<h5>Arguments:</h5>
4588
4589<p>
4590The first argument is a pointer to the destination, the second is a pointer to
4591the source. The third argument is an integer argument
4592specifying the number of bytes to copy, and the fourth argument is the alignment
4593of the source and destination locations.
4594</p>
4595
4596<p>
4597If the call to this intrinisic has an alignment value that is not 0 or 1, then
4598the caller guarantees that both the source and destination pointers are aligned
4599to that boundary.
4600</p>
4601
4602<h5>Semantics:</h5>
4603
4604<p>
4605The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the source
4606location to the destination location, which are not allowed to overlap. It
4607copies "len" bytes of memory over. If the argument is known to be aligned to
4608some boundary, this can be specified as the fourth argument, otherwise it should
4609be set to 0 or 1.
4610</p>
4611</div>
4612
4613
4614<!-- _______________________________________________________________________ -->
4615<div class="doc_subsubsection">
4616 <a name="int_memmove">'<tt>llvm.memmove</tt>' Intrinsic</a>
4617</div>
4618
4619<div class="doc_text">
4620
4621<h5>Syntax:</h5>
4622<pre>
4623 declare void @llvm.memmove.i32(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
4624 i32 &lt;len&gt;, i32 &lt;align&gt;)
4625 declare void @llvm.memmove.i64(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
4626 i64 &lt;len&gt;, i32 &lt;align&gt;)
4627</pre>
4628
4629<h5>Overview:</h5>
4630
4631<p>
4632The '<tt>llvm.memmove.*</tt>' intrinsics move a block of memory from the source
4633location to the destination location. It is similar to the
Chris Lattnerdba16ea2008-01-06 19:51:52 +0000[diff] [blame]4634'<tt>llvm.memcpy</tt>' intrinsic but allows the two memory locations to overlap.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4635</p>
4636
4637<p>
4638Note that, unlike the standard libc function, the <tt>llvm.memmove.*</tt>
4639intrinsics do not return a value, and takes an extra alignment argument.
4640</p>
4641
4642<h5>Arguments:</h5>
4643
4644<p>
4645The first argument is a pointer to the destination, the second is a pointer to
4646the source. The third argument is an integer argument
4647specifying the number of bytes to copy, and the fourth argument is the alignment
4648of the source and destination locations.
4649</p>
4650
4651<p>
4652If the call to this intrinisic has an alignment value that is not 0 or 1, then
4653the caller guarantees that the source and destination pointers are aligned to
4654that boundary.
4655</p>
4656
4657<h5>Semantics:</h5>
4658
4659<p>
4660The '<tt>llvm.memmove.*</tt>' intrinsics copy a block of memory from the source
4661location to the destination location, which may overlap. It
4662copies "len" bytes of memory over. If the argument is known to be aligned to
4663some boundary, this can be specified as the fourth argument, otherwise it should
4664be set to 0 or 1.
4665</p>
4666</div>
4667
4668
4669<!-- _______________________________________________________________________ -->
4670<div class="doc_subsubsection">
4671 <a name="int_memset">'<tt>llvm.memset.*</tt>' Intrinsics</a>
4672</div>
4673
4674<div class="doc_text">
4675
4676<h5>Syntax:</h5>
4677<pre>
4678 declare void @llvm.memset.i32(i8 * &lt;dest&gt;, i8 &lt;val&gt;,
4679 i32 &lt;len&gt;, i32 &lt;align&gt;)
4680 declare void @llvm.memset.i64(i8 * &lt;dest&gt;, i8 &lt;val&gt;,
4681 i64 &lt;len&gt;, i32 &lt;align&gt;)
4682</pre>
4683
4684<h5>Overview:</h5>
4685
4686<p>
4687The '<tt>llvm.memset.*</tt>' intrinsics fill a block of memory with a particular
4688byte value.
4689</p>
4690
4691<p>
4692Note that, unlike the standard libc function, the <tt>llvm.memset</tt> intrinsic
4693does not return a value, and takes an extra alignment argument.
4694</p>
4695
4696<h5>Arguments:</h5>
4697
4698<p>
4699The first argument is a pointer to the destination to fill, the second is the
4700byte value to fill it with, the third argument is an integer
4701argument specifying the number of bytes to fill, and the fourth argument is the
4702known alignment of destination location.
4703</p>
4704
4705<p>
4706If the call to this intrinisic has an alignment value that is not 0 or 1, then
4707the caller guarantees that the destination pointer is aligned to that boundary.
4708</p>
4709
4710<h5>Semantics:</h5>
4711
4712<p>
4713The '<tt>llvm.memset.*</tt>' intrinsics fill "len" bytes of memory starting at
4714the
4715destination location. If the argument is known to be aligned to some boundary,
4716this can be specified as the fourth argument, otherwise it should be set to 0 or
47171.
4718</p>
4719</div>
4720
4721
4722<!-- _______________________________________________________________________ -->
4723<div class="doc_subsubsection">
4724 <a name="int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a>
4725</div>
4726
4727<div class="doc_text">
4728
4729<h5>Syntax:</h5>
Dale Johannesenf9adbb62007-10-02 17:47:38 +0000[diff] [blame]4730<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]4731floating point or vector of floating point type. Not all targets support all
4732types however.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4733<pre>
Dale Johannesenf9adbb62007-10-02 17:47:38 +0000[diff] [blame]4734 declare float @llvm.sqrt.f32(float %Val)
4735 declare double @llvm.sqrt.f64(double %Val)
4736 declare x86_fp80 @llvm.sqrt.f80(x86_fp80 %Val)
4737 declare fp128 @llvm.sqrt.f128(fp128 %Val)
4738 declare ppc_fp128 @llvm.sqrt.ppcf128(ppc_fp128 %Val)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4739</pre>
4740
4741<h5>Overview:</h5>
4742
4743<p>
4744The '<tt>llvm.sqrt</tt>' intrinsics return the sqrt of the specified operand,
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]4745returning the same value as the libm '<tt>sqrt</tt>' functions would. Unlike
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4746<tt>sqrt</tt> in libm, however, <tt>llvm.sqrt</tt> has undefined behavior for
Chris Lattnerf914a002008-01-29 07:00:44 +0000[diff] [blame]4747negative numbers other than -0.0 (which allows for better optimization, because
4748there is no need to worry about errno being set). <tt>llvm.sqrt(-0.0)</tt> is
4749defined to return -0.0 like IEEE sqrt.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4750</p>
4751
4752<h5>Arguments:</h5>
4753
4754<p>
4755The argument and return value are floating point numbers of the same type.
4756</p>
4757
4758<h5>Semantics:</h5>
4759
4760<p>
4761This function returns the sqrt of the specified operand if it is a nonnegative
4762floating point number.
4763</p>
4764</div>
4765
4766<!-- _______________________________________________________________________ -->
4767<div class="doc_subsubsection">
4768 <a name="int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a>
4769</div>
4770
4771<div class="doc_text">
4772
4773<h5>Syntax:</h5>
Dale Johannesenf9adbb62007-10-02 17:47:38 +0000[diff] [blame]4774<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]4775floating point or vector of floating point type. Not all targets support all
4776types however.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4777<pre>
Dale Johannesenf9adbb62007-10-02 17:47:38 +0000[diff] [blame]4778 declare float @llvm.powi.f32(float %Val, i32 %power)
4779 declare double @llvm.powi.f64(double %Val, i32 %power)
4780 declare x86_fp80 @llvm.powi.f80(x86_fp80 %Val, i32 %power)
4781 declare fp128 @llvm.powi.f128(fp128 %Val, i32 %power)
4782 declare ppc_fp128 @llvm.powi.ppcf128(ppc_fp128 %Val, i32 %power)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4783</pre>
4784
4785<h5>Overview:</h5>
4786
4787<p>
4788The '<tt>llvm.powi.*</tt>' intrinsics return the first operand raised to the
4789specified (positive or negative) power. The order of evaluation of
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]4790multiplications is not defined. When a vector of floating point type is
4791used, the second argument remains a scalar integer value.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4792</p>
4793
4794<h5>Arguments:</h5>
4795
4796<p>
4797The second argument is an integer power, and the first is a value to raise to
4798that power.
4799</p>
4800
4801<h5>Semantics:</h5>
4802
4803<p>
4804This function returns the first value raised to the second power with an
4805unspecified sequence of rounding operations.</p>
4806</div>
4807
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]4808<!-- _______________________________________________________________________ -->
4809<div class="doc_subsubsection">
4810 <a name="int_sin">'<tt>llvm.sin.*</tt>' Intrinsic</a>
4811</div>
4812
4813<div class="doc_text">
4814
4815<h5>Syntax:</h5>
4816<p>This is an overloaded intrinsic. You can use <tt>llvm.sin</tt> on any
4817floating point or vector of floating point type. Not all targets support all
4818types however.
4819<pre>
4820 declare float @llvm.sin.f32(float %Val)
4821 declare double @llvm.sin.f64(double %Val)
4822 declare x86_fp80 @llvm.sin.f80(x86_fp80 %Val)
4823 declare fp128 @llvm.sin.f128(fp128 %Val)
4824 declare ppc_fp128 @llvm.sin.ppcf128(ppc_fp128 %Val)
4825</pre>
4826
4827<h5>Overview:</h5>
4828
4829<p>
4830The '<tt>llvm.sin.*</tt>' intrinsics return the sine of the operand.
4831</p>
4832
4833<h5>Arguments:</h5>
4834
4835<p>
4836The argument and return value are floating point numbers of the same type.
4837</p>
4838
4839<h5>Semantics:</h5>
4840
4841<p>
4842This function returns the sine of the specified operand, returning the
4843same values as the libm <tt>sin</tt> functions would, and handles error
Dan Gohmaneaba92e2007-10-17 18:05:13 +0000[diff] [blame]4844conditions in the same way.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]4845</div>
4846
4847<!-- _______________________________________________________________________ -->
4848<div class="doc_subsubsection">
4849 <a name="int_cos">'<tt>llvm.cos.*</tt>' Intrinsic</a>
4850</div>
4851
4852<div class="doc_text">
4853
4854<h5>Syntax:</h5>
4855<p>This is an overloaded intrinsic. You can use <tt>llvm.cos</tt> on any
4856floating point or vector of floating point type. Not all targets support all
4857types however.
4858<pre>
4859 declare float @llvm.cos.f32(float %Val)
4860 declare double @llvm.cos.f64(double %Val)
4861 declare x86_fp80 @llvm.cos.f80(x86_fp80 %Val)
4862 declare fp128 @llvm.cos.f128(fp128 %Val)
4863 declare ppc_fp128 @llvm.cos.ppcf128(ppc_fp128 %Val)
4864</pre>
4865
4866<h5>Overview:</h5>
4867
4868<p>
4869The '<tt>llvm.cos.*</tt>' intrinsics return the cosine of the operand.
4870</p>
4871
4872<h5>Arguments:</h5>
4873
4874<p>
4875The argument and return value are floating point numbers of the same type.
4876</p>
4877
4878<h5>Semantics:</h5>
4879
4880<p>
4881This function returns the cosine of the specified operand, returning the
4882same values as the libm <tt>cos</tt> functions would, and handles error
Dan Gohmaneaba92e2007-10-17 18:05:13 +0000[diff] [blame]4883conditions in the same way.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]4884</div>
4885
4886<!-- _______________________________________________________________________ -->
4887<div class="doc_subsubsection">
4888 <a name="int_pow">'<tt>llvm.pow.*</tt>' Intrinsic</a>
4889</div>
4890
4891<div class="doc_text">
4892
4893<h5>Syntax:</h5>
4894<p>This is an overloaded intrinsic. You can use <tt>llvm.pow</tt> on any
4895floating point or vector of floating point type. Not all targets support all
4896types however.
4897<pre>
4898 declare float @llvm.pow.f32(float %Val, float %Power)
4899 declare double @llvm.pow.f64(double %Val, double %Power)
4900 declare x86_fp80 @llvm.pow.f80(x86_fp80 %Val, x86_fp80 %Power)
4901 declare fp128 @llvm.pow.f128(fp128 %Val, fp128 %Power)
4902 declare ppc_fp128 @llvm.pow.ppcf128(ppc_fp128 %Val, ppc_fp128 Power)
4903</pre>
4904
4905<h5>Overview:</h5>
4906
4907<p>
4908The '<tt>llvm.pow.*</tt>' intrinsics return the first operand raised to the
4909specified (positive or negative) power.
4910</p>
4911
4912<h5>Arguments:</h5>
4913
4914<p>
4915The second argument is a floating point power, and the first is a value to
4916raise to that power.
4917</p>
4918
4919<h5>Semantics:</h5>
4920
4921<p>
4922This function returns the first value raised to the second power,
4923returning the
4924same values as the libm <tt>pow</tt> functions would, and handles error
Dan Gohmaneaba92e2007-10-17 18:05:13 +0000[diff] [blame]4925conditions in the same way.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]4926</div>
4927
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4928
4929<!-- ======================================================================= -->
4930<div class="doc_subsection">
4931 <a name="int_manip">Bit Manipulation Intrinsics</a>
4932</div>
4933
4934<div class="doc_text">
4935<p>
4936LLVM provides intrinsics for a few important bit manipulation operations.
4937These allow efficient code generation for some algorithms.
4938</p>
4939
4940</div>
4941
4942<!-- _______________________________________________________________________ -->
4943<div class="doc_subsubsection">
4944 <a name="int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a>
4945</div>
4946
4947<div class="doc_text">
4948
4949<h5>Syntax:</h5>
4950<p>This is an overloaded intrinsic function. You can use bswap on any integer
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]4951type that is an even number of bytes (i.e. BitWidth % 16 == 0).
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4952<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]4953 declare i16 @llvm.bswap.i16(i16 &lt;id&gt;)
4954 declare i32 @llvm.bswap.i32(i32 &lt;id&gt;)
4955 declare i64 @llvm.bswap.i64(i64 &lt;id&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4956</pre>
4957
4958<h5>Overview:</h5>
4959
4960<p>
4961The '<tt>llvm.bswap</tt>' family of intrinsics is used to byte swap integer
4962values with an even number of bytes (positive multiple of 16 bits). These are
4963useful for performing operations on data that is not in the target's native
4964byte order.
4965</p>
4966
4967<h5>Semantics:</h5>
4968
4969<p>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]4970The <tt>llvm.bswap.i16</tt> intrinsic returns an i16 value that has the high
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4971and low byte of the input i16 swapped. Similarly, the <tt>llvm.bswap.i32</tt>
4972intrinsic returns an i32 value that has the four bytes of the input i32
4973swapped, 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]4974i32 will have its bytes in 3, 2, 1, 0 order. The <tt>llvm.bswap.i48</tt>,
4975<tt>llvm.bswap.i64</tt> and other intrinsics extend this concept to
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4976additional even-byte lengths (6 bytes, 8 bytes and more, respectively).
4977</p>
4978
4979</div>
4980
4981<!-- _______________________________________________________________________ -->
4982<div class="doc_subsubsection">
4983 <a name="int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic</a>
4984</div>
4985
4986<div class="doc_text">
4987
4988<h5>Syntax:</h5>
4989<p>This is an overloaded intrinsic. You can use llvm.ctpop on any integer bit
4990width. Not all targets support all bit widths however.
4991<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]4992 declare i8 @llvm.ctpop.i8 (i8 &lt;src&gt;)
4993 declare i16 @llvm.ctpop.i16(i16 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4994 declare i32 @llvm.ctpop.i32(i32 &lt;src&gt;)
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]4995 declare i64 @llvm.ctpop.i64(i64 &lt;src&gt;)
4996 declare i256 @llvm.ctpop.i256(i256 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4997</pre>
4998
4999<h5>Overview:</h5>
5000
5001<p>
5002The '<tt>llvm.ctpop</tt>' family of intrinsics counts the number of bits set in a
5003value.
5004</p>
5005
5006<h5>Arguments:</h5>
5007
5008<p>
5009The only argument is the value to be counted. The argument may be of any
5010integer type. The return type must match the argument type.
5011</p>
5012
5013<h5>Semantics:</h5>
5014
5015<p>
5016The '<tt>llvm.ctpop</tt>' intrinsic counts the 1's in a variable.
5017</p>
5018</div>
5019
5020<!-- _______________________________________________________________________ -->
5021<div class="doc_subsubsection">
5022 <a name="int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic</a>
5023</div>
5024
5025<div class="doc_text">
5026
5027<h5>Syntax:</h5>
5028<p>This is an overloaded intrinsic. You can use <tt>llvm.ctlz</tt> on any
5029integer bit width. Not all targets support all bit widths however.
5030<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5031 declare i8 @llvm.ctlz.i8 (i8 &lt;src&gt;)
5032 declare i16 @llvm.ctlz.i16(i16 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5033 declare i32 @llvm.ctlz.i32(i32 &lt;src&gt;)
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5034 declare i64 @llvm.ctlz.i64(i64 &lt;src&gt;)
5035 declare i256 @llvm.ctlz.i256(i256 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5036</pre>
5037
5038<h5>Overview:</h5>
5039
5040<p>
5041The '<tt>llvm.ctlz</tt>' family of intrinsic functions counts the number of
5042leading zeros in a variable.
5043</p>
5044
5045<h5>Arguments:</h5>
5046
5047<p>
5048The only argument is the value to be counted. The argument may be of any
5049integer type. The return type must match the argument type.
5050</p>
5051
5052<h5>Semantics:</h5>
5053
5054<p>
5055The '<tt>llvm.ctlz</tt>' intrinsic counts the leading (most significant) zeros
5056in a variable. If the src == 0 then the result is the size in bits of the type
5057of src. For example, <tt>llvm.ctlz(i32 2) = 30</tt>.
5058</p>
5059</div>
5060
5061
5062
5063<!-- _______________________________________________________________________ -->
5064<div class="doc_subsubsection">
5065 <a name="int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic</a>
5066</div>
5067
5068<div class="doc_text">
5069
5070<h5>Syntax:</h5>
5071<p>This is an overloaded intrinsic. You can use <tt>llvm.cttz</tt> on any
5072integer bit width. Not all targets support all bit widths however.
5073<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5074 declare i8 @llvm.cttz.i8 (i8 &lt;src&gt;)
5075 declare i16 @llvm.cttz.i16(i16 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5076 declare i32 @llvm.cttz.i32(i32 &lt;src&gt;)
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5077 declare i64 @llvm.cttz.i64(i64 &lt;src&gt;)
5078 declare i256 @llvm.cttz.i256(i256 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5079</pre>
5080
5081<h5>Overview:</h5>
5082
5083<p>
5084The '<tt>llvm.cttz</tt>' family of intrinsic functions counts the number of
5085trailing zeros.
5086</p>
5087
5088<h5>Arguments:</h5>
5089
5090<p>
5091The only argument is the value to be counted. The argument may be of any
5092integer type. The return type must match the argument type.
5093</p>
5094
5095<h5>Semantics:</h5>
5096
5097<p>
5098The '<tt>llvm.cttz</tt>' intrinsic counts the trailing (least significant) zeros
5099in a variable. If the src == 0 then the result is the size in bits of the type
5100of src. For example, <tt>llvm.cttz(2) = 1</tt>.
5101</p>
5102</div>
5103
5104<!-- _______________________________________________________________________ -->
5105<div class="doc_subsubsection">
5106 <a name="int_part_select">'<tt>llvm.part.select.*</tt>' Intrinsic</a>
5107</div>
5108
5109<div class="doc_text">
5110
5111<h5>Syntax:</h5>
5112<p>This is an overloaded intrinsic. You can use <tt>llvm.part.select</tt>
5113on any integer bit width.
5114<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5115 declare i17 @llvm.part.select.i17 (i17 %val, i32 %loBit, i32 %hiBit)
5116 declare i29 @llvm.part.select.i29 (i29 %val, i32 %loBit, i32 %hiBit)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5117</pre>
5118
5119<h5>Overview:</h5>
5120<p>The '<tt>llvm.part.select</tt>' family of intrinsic functions selects a
5121range of bits from an integer value and returns them in the same bit width as
5122the original value.</p>
5123
5124<h5>Arguments:</h5>
5125<p>The first argument, <tt>%val</tt> and the result may be integer types of
5126any bit width but they must have the same bit width. The second and third
5127arguments must be <tt>i32</tt> type since they specify only a bit index.</p>
5128
5129<h5>Semantics:</h5>
5130<p>The operation of the '<tt>llvm.part.select</tt>' intrinsic has two modes
5131of operation: forwards and reverse. If <tt>%loBit</tt> is greater than
5132<tt>%hiBits</tt> then the intrinsic operates in reverse mode. Otherwise it
5133operates in forward mode.</p>
5134<p>In forward mode, this intrinsic is the equivalent of shifting <tt>%val</tt>
5135right by <tt>%loBit</tt> bits and then ANDing it with a mask with
5136only the <tt>%hiBit - %loBit</tt> bits set, as follows:</p>
5137<ol>
5138 <li>The <tt>%val</tt> is shifted right (LSHR) by the number of bits specified
5139 by <tt>%loBits</tt>. This normalizes the value to the low order bits.</li>
5140 <li>The <tt>%loBits</tt> value is subtracted from the <tt>%hiBits</tt> value
5141 to determine the number of bits to retain.</li>
5142 <li>A mask of the retained bits is created by shifting a -1 value.</li>
5143 <li>The mask is ANDed with <tt>%val</tt> to produce the result.
5144</ol>
5145<p>In reverse mode, a similar computation is made except that the bits are
5146returned in the reverse order. So, for example, if <tt>X</tt> has the value
5147<tt>i16 0x0ACF (101011001111)</tt> and we apply
5148<tt>part.select(i16 X, 8, 3)</tt> to it, we get back the value
5149<tt>i16 0x0026 (000000100110)</tt>.</p>
5150</div>
5151
5152<div class="doc_subsubsection">
5153 <a name="int_part_set">'<tt>llvm.part.set.*</tt>' Intrinsic</a>
5154</div>
5155
5156<div class="doc_text">
5157
5158<h5>Syntax:</h5>
5159<p>This is an overloaded intrinsic. You can use <tt>llvm.part.set</tt>
5160on any integer bit width.
5161<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5162 declare i17 @llvm.part.set.i17.i9 (i17 %val, i9 %repl, i32 %lo, i32 %hi)
5163 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]5164</pre>
5165
5166<h5>Overview:</h5>
5167<p>The '<tt>llvm.part.set</tt>' family of intrinsic functions replaces a range
5168of bits in an integer value with another integer value. It returns the integer
5169with the replaced bits.</p>
5170
5171<h5>Arguments:</h5>
5172<p>The first argument, <tt>%val</tt> and the result may be integer types of
5173any bit width but they must have the same bit width. <tt>%val</tt> is the value
5174whose bits will be replaced. The second argument, <tt>%repl</tt> may be an
5175integer of any bit width. The third and fourth arguments must be <tt>i32</tt>
5176type since they specify only a bit index.</p>
5177
5178<h5>Semantics:</h5>
5179<p>The operation of the '<tt>llvm.part.set</tt>' intrinsic has two modes
5180of operation: forwards and reverse. If <tt>%lo</tt> is greater than
5181<tt>%hi</tt> then the intrinsic operates in reverse mode. Otherwise it
5182operates in forward mode.</p>
5183<p>For both modes, the <tt>%repl</tt> value is prepared for use by either
5184truncating it down to the size of the replacement area or zero extending it
5185up to that size.</p>
5186<p>In forward mode, the bits between <tt>%lo</tt> and <tt>%hi</tt> (inclusive)
5187are replaced with corresponding bits from <tt>%repl</tt>. That is the 0th bit
5188in <tt>%repl</tt> replaces the <tt>%lo</tt>th bit in <tt>%val</tt> and etc. up
5189to the <tt>%hi</tt>th bit.
5190<p>In reverse mode, a similar computation is made except that the bits are
5191reversed. That is, the <tt>0</tt>th bit in <tt>%repl</tt> replaces the
5192<tt>%hi</tt> bit in <tt>%val</tt> and etc. down to the <tt>%lo</tt>th bit.
5193<h5>Examples:</h5>
5194<pre>
5195 llvm.part.set(0xFFFF, 0, 4, 7) -&gt; 0xFF0F
5196 llvm.part.set(0xFFFF, 0, 7, 4) -&gt; 0xFF0F
5197 llvm.part.set(0xFFFF, 1, 7, 4) -&gt; 0xFF8F
5198 llvm.part.set(0xFFFF, F, 8, 3) -&gt; 0xFFE7
5199 llvm.part.set(0xFFFF, 0, 3, 8) -&gt; 0xFE07
5200</pre>
5201</div>
5202
5203<!-- ======================================================================= -->
5204<div class="doc_subsection">
5205 <a name="int_debugger">Debugger Intrinsics</a>
5206</div>
5207
5208<div class="doc_text">
5209<p>
5210The LLVM debugger intrinsics (which all start with <tt>llvm.dbg.</tt> prefix),
5211are described in the <a
5212href="SourceLevelDebugging.html#format_common_intrinsics">LLVM Source Level
5213Debugging</a> document.
5214</p>
5215</div>
5216
5217
5218<!-- ======================================================================= -->
5219<div class="doc_subsection">
5220 <a name="int_eh">Exception Handling Intrinsics</a>
5221</div>
5222
5223<div class="doc_text">
5224<p> The LLVM exception handling intrinsics (which all start with
5225<tt>llvm.eh.</tt> prefix), are described in the <a
5226href="ExceptionHandling.html#format_common_intrinsics">LLVM Exception
5227Handling</a> document. </p>
5228</div>
5229
5230<!-- ======================================================================= -->
5231<div class="doc_subsection">
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5232 <a name="int_trampoline">Trampoline Intrinsic</a>
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5233</div>
5234
5235<div class="doc_text">
5236<p>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5237 This intrinsic makes it possible to excise one parameter, marked with
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5238 the <tt>nest</tt> attribute, from a function. The result is a callable
5239 function pointer lacking the nest parameter - the caller does not need
5240 to provide a value for it. Instead, the value to use is stored in
5241 advance in a "trampoline", a block of memory usually allocated
5242 on the stack, which also contains code to splice the nest value into the
5243 argument list. This is used to implement the GCC nested function address
5244 extension.
5245</p>
5246<p>
5247 For example, if the function is
5248 <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]5249 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]5250<pre>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5251 %tramp = alloca [10 x i8], align 4 ; size and alignment only correct for X86
5252 %tramp1 = getelementptr [10 x i8]* %tramp, i32 0, i32 0
5253 %p = call i8* @llvm.init.trampoline( i8* %tramp1, i8* bitcast (i32 (i8* nest , i32, i32)* @f to i8*), i8* %nval )
5254 %fp = bitcast i8* %p to i32 (i32, i32)*
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5255</pre>
Bill Wendlinge262b4c2007-09-22 09:23:55 +0000[diff] [blame]5256 <p>The call <tt>%val = call i32 %fp( i32 %x, i32 %y )</tt> is then equivalent
5257 to <tt>%val = call i32 %f( i8* %nval, i32 %x, i32 %y )</tt>.</p>
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5258</div>
5259
5260<!-- _______________________________________________________________________ -->
5261<div class="doc_subsubsection">
5262 <a name="int_it">'<tt>llvm.init.trampoline</tt>' Intrinsic</a>
5263</div>
5264<div class="doc_text">
5265<h5>Syntax:</h5>
5266<pre>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5267declare 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]5268</pre>
5269<h5>Overview:</h5>
5270<p>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5271 This fills the memory pointed to by <tt>tramp</tt> with code
5272 and returns a function pointer suitable for executing it.
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5273</p>
5274<h5>Arguments:</h5>
5275<p>
5276 The <tt>llvm.init.trampoline</tt> intrinsic takes three arguments, all
5277 pointers. The <tt>tramp</tt> argument must point to a sufficiently large
5278 and sufficiently aligned block of memory; this memory is written to by the
Duncan Sands35012212007-08-22 23:39:54 +0000[diff] [blame]5279 intrinsic. Note that the size and the alignment are target-specific - LLVM
5280 currently provides no portable way of determining them, so a front-end that
5281 generates this intrinsic needs to have some target-specific knowledge.
5282 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]5283</p>
5284<h5>Semantics:</h5>
5285<p>
5286 The block of memory pointed to by <tt>tramp</tt> is filled with target
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5287 dependent code, turning it into a function. A pointer to this function is
5288 returned, but needs to be bitcast to an
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5289 <a href="#int_trampoline">appropriate function pointer type</a>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5290 before being called. The new function's signature is the same as that of
5291 <tt>func</tt> with any arguments marked with the <tt>nest</tt> attribute
5292 removed. At most one such <tt>nest</tt> argument is allowed, and it must be
5293 of pointer type. Calling the new function is equivalent to calling
5294 <tt>func</tt> with the same argument list, but with <tt>nval</tt> used for the
5295 missing <tt>nest</tt> argument. If, after calling
5296 <tt>llvm.init.trampoline</tt>, the memory pointed to by <tt>tramp</tt> is
5297 modified, then the effect of any later call to the returned function pointer is
5298 undefined.
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5299</p>
5300</div>
5301
5302<!-- ======================================================================= -->
5303<div class="doc_subsection">
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]5304 <a name="int_atomics">Atomic Operations and Synchronization Intrinsics</a>
5305</div>
5306
5307<div class="doc_text">
5308<p>
5309 These intrinsic functions expand the "universal IR" of LLVM to represent
5310 hardware constructs for atomic operations and memory synchronization. This
5311 provides an interface to the hardware, not an interface to the programmer. It
5312 is aimed at a low enough level to allow any programming models or APIs which
5313 need atomic behaviors to map cleanly onto it. It is also modeled primarily on
5314 hardware behavior. Just as hardware provides a "universal IR" for source
5315 languages, it also provides a starting point for developing a "universal"
5316 atomic operation and synchronization IR.
5317</p>
5318<p>
5319 These do <em>not</em> form an API such as high-level threading libraries,
5320 software transaction memory systems, atomic primitives, and intrinsic
5321 functions as found in BSD, GNU libc, atomic_ops, APR, and other system and
5322 application libraries. The hardware interface provided by LLVM should allow
5323 a clean implementation of all of these APIs and parallel programming models.
5324 No one model or paradigm should be selected above others unless the hardware
5325 itself ubiquitously does so.
5326
5327</p>
5328</div>
5329
5330<!-- _______________________________________________________________________ -->
5331<div class="doc_subsubsection">
5332 <a name="int_memory_barrier">'<tt>llvm.memory.barrier</tt>' Intrinsic</a>
5333</div>
5334<div class="doc_text">
5335<h5>Syntax:</h5>
5336<pre>
5337declare void @llvm.memory.barrier( i1 &lt;ll&gt;, i1 &lt;ls&gt;, i1 &lt;sl&gt;, i1 &lt;ss&gt;,
5338i1 &lt;device&gt; )
5339
5340</pre>
5341<h5>Overview:</h5>
5342<p>
5343 The <tt>llvm.memory.barrier</tt> intrinsic guarantees ordering between
5344 specific pairs of memory access types.
5345</p>
5346<h5>Arguments:</h5>
5347<p>
5348 The <tt>llvm.memory.barrier</tt> intrinsic requires five boolean arguments.
5349 The first four arguments enables a specific barrier as listed below. The fith
5350 argument specifies that the barrier applies to io or device or uncached memory.
5351
5352</p>
5353 <ul>
5354 <li><tt>ll</tt>: load-load barrier</li>
5355 <li><tt>ls</tt>: load-store barrier</li>
5356 <li><tt>sl</tt>: store-load barrier</li>
5357 <li><tt>ss</tt>: store-store barrier</li>
5358 <li><tt>device</tt>: barrier applies to device and uncached memory also.
5359 </ul>
5360<h5>Semantics:</h5>
5361<p>
5362 This intrinsic causes the system to enforce some ordering constraints upon
5363 the loads and stores of the program. This barrier does not indicate
5364 <em>when</em> any events will occur, it only enforces an <em>order</em> in
5365 which they occur. For any of the specified pairs of load and store operations
5366 (f.ex. load-load, or store-load), all of the first operations preceding the
5367 barrier will complete before any of the second operations succeeding the
5368 barrier begin. Specifically the semantics for each pairing is as follows:
5369</p>
5370 <ul>
5371 <li><tt>ll</tt>: All loads before the barrier must complete before any load
5372 after the barrier begins.</li>
5373
5374 <li><tt>ls</tt>: All loads before the barrier must complete before any
5375 store after the barrier begins.</li>
5376 <li><tt>ss</tt>: All stores before the barrier must complete before any
5377 store after the barrier begins.</li>
5378 <li><tt>sl</tt>: All stores before the barrier must complete before any
5379 load after the barrier begins.</li>
5380 </ul>
5381<p>
5382 These semantics are applied with a logical "and" behavior when more than one
5383 is enabled in a single memory barrier intrinsic.
5384</p>
5385<p>
5386 Backends may implement stronger barriers than those requested when they do not
5387 support as fine grained a barrier as requested. Some architectures do not
5388 need all types of barriers and on such architectures, these become noops.
5389</p>
5390<h5>Example:</h5>
5391<pre>
5392%ptr = malloc i32
5393 store i32 4, %ptr
5394
5395%result1 = load i32* %ptr <i>; yields {i32}:result1 = 4</i>
5396 call void @llvm.memory.barrier( i1 false, i1 true, i1 false, i1 false )
5397 <i>; guarantee the above finishes</i>
5398 store i32 8, %ptr <i>; before this begins</i>
5399</pre>
5400</div>
5401
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]5402<!-- _______________________________________________________________________ -->
5403<div class="doc_subsubsection">
5404 <a name="int_atomic_lcs">'<tt>llvm.atomic.lcs.*</tt>' Intrinsic</a>
5405</div>
5406<div class="doc_text">
5407<h5>Syntax:</h5>
5408<p>
5409 This is an overloaded intrinsic. You can use <tt>llvm.atomic.lcs</tt> on any
5410 integer bit width. Not all targets support all bit widths however.</p>
5411
5412<pre>
5413declare i8 @llvm.atomic.lcs.i8( i8* &lt;ptr&gt;, i8 &lt;cmp&gt;, i8 &lt;val&gt; )
5414declare i16 @llvm.atomic.lcs.i16( i16* &lt;ptr&gt;, i16 &lt;cmp&gt;, i16 &lt;val&gt; )
5415declare i32 @llvm.atomic.lcs.i32( i32* &lt;ptr&gt;, i32 &lt;cmp&gt;, i32 &lt;val&gt; )
5416declare i64 @llvm.atomic.lcs.i64( i64* &lt;ptr&gt;, i64 &lt;cmp&gt;, i64 &lt;val&gt; )
5417
5418</pre>
5419<h5>Overview:</h5>
5420<p>
5421 This loads a value in memory and compares it to a given value. If they are
5422 equal, it stores a new value into the memory.
5423</p>
5424<h5>Arguments:</h5>
5425<p>
5426 The <tt>llvm.atomic.lcs</tt> intrinsic takes three arguments. The result as
5427 well as both <tt>cmp</tt> and <tt>val</tt> must be integer values with the
5428 same bit width. The <tt>ptr</tt> argument must be a pointer to a value of
5429 this integer type. While any bit width integer may be used, targets may only
5430 lower representations they support in hardware.
5431
5432</p>
5433<h5>Semantics:</h5>
5434<p>
5435 This entire intrinsic must be executed atomically. It first loads the value
5436 in memory pointed to by <tt>ptr</tt> and compares it with the value
5437 <tt>cmp</tt>. If they are equal, <tt>val</tt> is stored into the memory. The
5438 loaded value is yielded in all cases. This provides the equivalent of an
5439 atomic compare-and-swap operation within the SSA framework.
5440</p>
5441<h5>Examples:</h5>
5442
5443<pre>
5444%ptr = malloc i32
5445 store i32 4, %ptr
5446
5447%val1 = add i32 4, 4
5448%result1 = call i32 @llvm.atomic.lcs.i32( i32* %ptr, i32 4, %val1 )
5449 <i>; yields {i32}:result1 = 4</i>
5450%stored1 = icmp eq i32 %result1, 4 <i>; yields {i1}:stored1 = true</i>
5451%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 8</i>
5452
5453%val2 = add i32 1, 1
5454%result2 = call i32 @llvm.atomic.lcs.i32( i32* %ptr, i32 5, %val2 )
5455 <i>; yields {i32}:result2 = 8</i>
5456%stored2 = icmp eq i32 %result2, 5 <i>; yields {i1}:stored2 = false</i>
5457
5458%memval2 = load i32* %ptr <i>; yields {i32}:memval2 = 8</i>
5459</pre>
5460</div>
5461
5462<!-- _______________________________________________________________________ -->
5463<div class="doc_subsubsection">
5464 <a name="int_atomic_swap">'<tt>llvm.atomic.swap.*</tt>' Intrinsic</a>
5465</div>
5466<div class="doc_text">
5467<h5>Syntax:</h5>
5468
5469<p>
5470 This is an overloaded intrinsic. You can use <tt>llvm.atomic.swap</tt> on any
5471 integer bit width. Not all targets support all bit widths however.</p>
5472<pre>
5473declare i8 @llvm.atomic.swap.i8( i8* &lt;ptr&gt;, i8 &lt;val&gt; )
5474declare i16 @llvm.atomic.swap.i16( i16* &lt;ptr&gt;, i16 &lt;val&gt; )
5475declare i32 @llvm.atomic.swap.i32( i32* &lt;ptr&gt;, i32 &lt;val&gt; )
5476declare i64 @llvm.atomic.swap.i64( i64* &lt;ptr&gt;, i64 &lt;val&gt; )
5477
5478</pre>
5479<h5>Overview:</h5>
5480<p>
5481 This intrinsic loads the value stored in memory at <tt>ptr</tt> and yields
5482 the value from memory. It then stores the value in <tt>val</tt> in the memory
5483 at <tt>ptr</tt>.
5484</p>
5485<h5>Arguments:</h5>
5486
5487<p>
5488 The <tt>llvm.atomic.ls</tt> intrinsic takes two arguments. Both the
5489 <tt>val</tt> argument and the result must be integers of the same bit width.
5490 The first argument, <tt>ptr</tt>, must be a pointer to a value of this
5491 integer type. The targets may only lower integer representations they
5492 support.
5493</p>
5494<h5>Semantics:</h5>
5495<p>
5496 This intrinsic loads the value pointed to by <tt>ptr</tt>, yields it, and
5497 stores <tt>val</tt> back into <tt>ptr</tt> atomically. This provides the
5498 equivalent of an atomic swap operation within the SSA framework.
5499
5500</p>
5501<h5>Examples:</h5>
5502<pre>
5503%ptr = malloc i32
5504 store i32 4, %ptr
5505
5506%val1 = add i32 4, 4
5507%result1 = call i32 @llvm.atomic.swap.i32( i32* %ptr, i32 %val1 )
5508 <i>; yields {i32}:result1 = 4</i>
5509%stored1 = icmp eq i32 %result1, 4 <i>; yields {i1}:stored1 = true</i>
5510%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 8</i>
5511
5512%val2 = add i32 1, 1
5513%result2 = call i32 @llvm.atomic.swap.i32( i32* %ptr, i32 %val2 )
5514 <i>; yields {i32}:result2 = 8</i>
5515
5516%stored2 = icmp eq i32 %result2, 8 <i>; yields {i1}:stored2 = true</i>
5517%memval2 = load i32* %ptr <i>; yields {i32}:memval2 = 2</i>
5518</pre>
5519</div>
5520
5521<!-- _______________________________________________________________________ -->
5522<div class="doc_subsubsection">
5523 <a name="int_atomic_las">'<tt>llvm.atomic.las.*</tt>' Intrinsic</a>
5524
5525</div>
5526<div class="doc_text">
5527<h5>Syntax:</h5>
5528<p>
5529 This is an overloaded intrinsic. You can use <tt>llvm.atomic.las</tt> on any
5530 integer bit width. Not all targets support all bit widths however.</p>
5531<pre>
5532declare i8 @llvm.atomic.las.i8.( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
5533declare i16 @llvm.atomic.las.i16.( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
5534declare i32 @llvm.atomic.las.i32.( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
5535declare i64 @llvm.atomic.las.i64.( i64* &lt;ptr&gt;, i64 &lt;delta&gt; )
5536
5537</pre>
5538<h5>Overview:</h5>
5539<p>
5540 This intrinsic adds <tt>delta</tt> to the value stored in memory at
5541 <tt>ptr</tt>. It yields the original value at <tt>ptr</tt>.
5542</p>
5543<h5>Arguments:</h5>
5544<p>
5545
5546 The intrinsic takes two arguments, the first a pointer to an integer value
5547 and the second an integer value. The result is also an integer value. These
5548 integer types can have any bit width, but they must all have the same bit
5549 width. The targets may only lower integer representations they support.
5550</p>
5551<h5>Semantics:</h5>
5552<p>
5553 This intrinsic does a series of operations atomically. It first loads the
5554 value stored at <tt>ptr</tt>. It then adds <tt>delta</tt>, stores the result
5555 to <tt>ptr</tt>. It yields the original value stored at <tt>ptr</tt>.
5556</p>
5557
5558<h5>Examples:</h5>
5559<pre>
5560%ptr = malloc i32
5561 store i32 4, %ptr
5562%result1 = call i32 @llvm.atomic.las.i32( i32* %ptr, i32 4 )
5563 <i>; yields {i32}:result1 = 4</i>
5564%result2 = call i32 @llvm.atomic.las.i32( i32* %ptr, i32 2 )
5565 <i>; yields {i32}:result2 = 8</i>
5566%result3 = call i32 @llvm.atomic.las.i32( i32* %ptr, i32 5 )
5567 <i>; yields {i32}:result3 = 10</i>
5568%memval = load i32* %ptr <i>; yields {i32}:memval1 = 15</i>
5569</pre>
5570</div>
5571
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]5572
5573<!-- ======================================================================= -->
5574<div class="doc_subsection">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5575 <a name="int_general">General Intrinsics</a>
5576</div>
5577
5578<div class="doc_text">
5579<p> This class of intrinsics is designed to be generic and has
5580no specific purpose. </p>
5581</div>
5582
5583<!-- _______________________________________________________________________ -->
5584<div class="doc_subsubsection">
5585 <a name="int_var_annotation">'<tt>llvm.var.annotation</tt>' Intrinsic</a>
5586</div>
5587
5588<div class="doc_text">
5589
5590<h5>Syntax:</h5>
5591<pre>
5592 declare void @llvm.var.annotation(i8* &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
5593</pre>
5594
5595<h5>Overview:</h5>
5596
5597<p>
5598The '<tt>llvm.var.annotation</tt>' intrinsic
5599</p>
5600
5601<h5>Arguments:</h5>
5602
5603<p>
5604The first argument is a pointer to a value, the second is a pointer to a
5605global string, the third is a pointer to a global string which is the source
5606file name, and the last argument is the line number.
5607</p>
5608
5609<h5>Semantics:</h5>
5610
5611<p>
Anton Korobeynikove6e764f2008-01-15 22:31:34 +0000[diff] [blame]5612This intrinsic allows annotation of local variables with arbitrary strings.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5613This can be useful for special purpose optimizations that want to look for these
Anton Korobeynikove6e764f2008-01-15 22:31:34 +0000[diff] [blame]5614annotations. These have no other defined use, they are ignored by code
5615generation and optimization.
5616</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5617</div>
5618
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]5619<!-- _______________________________________________________________________ -->
5620<div class="doc_subsubsection">
Tanya Lattnerc9869b12007-09-21 23:57:59 +0000[diff] [blame]5621 <a name="int_annotation">'<tt>llvm.annotation.*</tt>' Intrinsic</a>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]5622</div>
5623
5624<div class="doc_text">
5625
5626<h5>Syntax:</h5>
Tanya Lattnere545be72007-09-21 23:56:27 +0000[diff] [blame]5627<p>This is an overloaded intrinsic. You can use '<tt>llvm.annotation</tt>' on
5628any integer bit width.
5629</p>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]5630<pre>
Tanya Lattner09161fe2007-09-22 00:03:01 +0000[diff] [blame]5631 declare i8 @llvm.annotation.i8(i8 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
5632 declare i16 @llvm.annotation.i16(i16 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
5633 declare i32 @llvm.annotation.i32(i32 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
5634 declare i64 @llvm.annotation.i64(i64 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
5635 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]5636</pre>
5637
5638<h5>Overview:</h5>
Tanya Lattnere545be72007-09-21 23:56:27 +0000[diff] [blame]5639
5640<p>
5641The '<tt>llvm.annotation</tt>' intrinsic.
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]5642</p>
5643
5644<h5>Arguments:</h5>
5645
5646<p>
5647The first argument is an integer value (result of some expression),
5648the second is a pointer to a global string, the third is a pointer to a global
5649string which is the source file name, and the last argument is the line number.
Tanya Lattnere545be72007-09-21 23:56:27 +0000[diff] [blame]5650It returns the value of the first argument.
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]5651</p>
5652
5653<h5>Semantics:</h5>
5654
5655<p>
5656This intrinsic allows annotations to be put on arbitrary expressions
5657with arbitrary strings. This can be useful for special purpose optimizations
5658that want to look for these annotations. These have no other defined use, they
5659are ignored by code generation and optimization.
5660</div>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5661
Anton Korobeynikove6e764f2008-01-15 22:31:34 +0000[diff] [blame]5662<!-- _______________________________________________________________________ -->
5663<div class="doc_subsubsection">
5664 <a name="int_trap">'<tt>llvm.trap</tt>' Intrinsic</a>
5665</div>
5666
5667<div class="doc_text">
5668
5669<h5>Syntax:</h5>
5670<pre>
5671 declare void @llvm.trap()
5672</pre>
5673
5674<h5>Overview:</h5>
5675
5676<p>
5677The '<tt>llvm.trap</tt>' intrinsic
5678</p>
5679
5680<h5>Arguments:</h5>
5681
5682<p>
5683None
5684</p>
5685
5686<h5>Semantics:</h5>
5687
5688<p>
5689This intrinsics is lowered to the target dependent trap instruction. If the
5690target does not have a trap instruction, this intrinsic will be lowered to the
5691call of the abort() function.
5692</p>
5693</div>
5694
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5695<!-- *********************************************************************** -->
5696<hr>
5697<address>
5698 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
5699 src="http://jigsaw.w3.org/css-validator/images/vcss" alt="Valid CSS!"></a>
5700 <a href="http://validator.w3.org/check/referer"><img
Chris Lattner08497ce2008-01-04 04:33:49 +0000[diff] [blame]5701 src="http://www.w3.org/Icons/valid-html401" alt="Valid HTML 4.01!"></a>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5702
5703 <a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
5704 <a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br>
5705 Last modified: $Date$
5706</address>
Chris Lattner08497ce2008-01-04 04:33:49 +0000[diff] [blame]5707
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5708</body>
5709</html>