Gitiles
Code Review Sign In
gerrit-public.fairphone.software / fp2-dev / platform / external / llvm / ef4bf3b063ea20732c4647c6f2d5d77caf907578 / . / docs / LangRef.html
blob: e097c2a84c5d03083246998be8f9f66ab86541c3 [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>
Chris Lattner5b6dc6e2009-01-11 20:53:49 +0000[diff] [blame]25 <li><a href="#namedtypes">Named Types</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]26 <li><a href="#globalvars">Global Variables</a></li>
27 <li><a href="#functionstructure">Functions</a></li>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]28 <li><a href="#aliasstructure">Aliases</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]29 <li><a href="#paramattrs">Parameter Attributes</a></li>
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]30 <li><a href="#fnattrs">Function Attributes</a></li>
Gordon Henriksen13fe5e32007-12-10 03:18:06 +0000[diff] [blame]31 <li><a href="#gc">Garbage Collector Names</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]32 <li><a href="#moduleasm">Module-Level Inline Assembly</a></li>
33 <li><a href="#datalayout">Data Layout</a></li>
34 </ol>
35 </li>
36 <li><a href="#typesystem">Type System</a>
37 <ol>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]38 <li><a href="#t_classifications">Type Classifications</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]39 <li><a href="#t_primitive">Primitive Types</a>
40 <ol>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]41 <li><a href="#t_floating">Floating Point Types</a></li>
42 <li><a href="#t_void">Void Type</a></li>
43 <li><a href="#t_label">Label Type</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]44 </ol>
45 </li>
46 <li><a href="#t_derived">Derived Types</a>
47 <ol>
Chris Lattner251ab812007-12-18 06:18:21 +0000[diff] [blame]48 <li><a href="#t_integer">Integer Type</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]49 <li><a href="#t_array">Array Type</a></li>
50 <li><a href="#t_function">Function Type</a></li>
51 <li><a href="#t_pointer">Pointer Type</a></li>
52 <li><a href="#t_struct">Structure Type</a></li>
53 <li><a href="#t_pstruct">Packed Structure Type</a></li>
54 <li><a href="#t_vector">Vector Type</a></li>
55 <li><a href="#t_opaque">Opaque Type</a></li>
56 </ol>
57 </li>
58 </ol>
59 </li>
60 <li><a href="#constants">Constants</a>
61 <ol>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]62 <li><a href="#simpleconstants">Simple Constants</a></li>
63 <li><a href="#aggregateconstants">Aggregate Constants</a></li>
64 <li><a href="#globalconstants">Global Variable and Function Addresses</a></li>
65 <li><a href="#undefvalues">Undefined Values</a></li>
66 <li><a href="#constantexprs">Constant Expressions</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]67 </ol>
68 </li>
69 <li><a href="#othervalues">Other Values</a>
70 <ol>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]71 <li><a href="#inlineasm">Inline Assembler Expressions</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]72 </ol>
73 </li>
74 <li><a href="#instref">Instruction Reference</a>
75 <ol>
76 <li><a href="#terminators">Terminator Instructions</a>
77 <ol>
78 <li><a href="#i_ret">'<tt>ret</tt>' Instruction</a></li>
79 <li><a href="#i_br">'<tt>br</tt>' Instruction</a></li>
80 <li><a href="#i_switch">'<tt>switch</tt>' Instruction</a></li>
81 <li><a href="#i_invoke">'<tt>invoke</tt>' Instruction</a></li>
82 <li><a href="#i_unwind">'<tt>unwind</tt>' Instruction</a></li>
83 <li><a href="#i_unreachable">'<tt>unreachable</tt>' Instruction</a></li>
84 </ol>
85 </li>
86 <li><a href="#binaryops">Binary Operations</a>
87 <ol>
88 <li><a href="#i_add">'<tt>add</tt>' Instruction</a></li>
89 <li><a href="#i_sub">'<tt>sub</tt>' Instruction</a></li>
90 <li><a href="#i_mul">'<tt>mul</tt>' Instruction</a></li>
91 <li><a href="#i_udiv">'<tt>udiv</tt>' Instruction</a></li>
92 <li><a href="#i_sdiv">'<tt>sdiv</tt>' Instruction</a></li>
93 <li><a href="#i_fdiv">'<tt>fdiv</tt>' Instruction</a></li>
94 <li><a href="#i_urem">'<tt>urem</tt>' Instruction</a></li>
95 <li><a href="#i_srem">'<tt>srem</tt>' Instruction</a></li>
96 <li><a href="#i_frem">'<tt>frem</tt>' Instruction</a></li>
97 </ol>
98 </li>
99 <li><a href="#bitwiseops">Bitwise Binary Operations</a>
100 <ol>
101 <li><a href="#i_shl">'<tt>shl</tt>' Instruction</a></li>
102 <li><a href="#i_lshr">'<tt>lshr</tt>' Instruction</a></li>
103 <li><a href="#i_ashr">'<tt>ashr</tt>' Instruction</a></li>
104 <li><a href="#i_and">'<tt>and</tt>' Instruction</a></li>
105 <li><a href="#i_or">'<tt>or</tt>' Instruction</a></li>
106 <li><a href="#i_xor">'<tt>xor</tt>' Instruction</a></li>
107 </ol>
108 </li>
109 <li><a href="#vectorops">Vector Operations</a>
110 <ol>
111 <li><a href="#i_extractelement">'<tt>extractelement</tt>' Instruction</a></li>
112 <li><a href="#i_insertelement">'<tt>insertelement</tt>' Instruction</a></li>
113 <li><a href="#i_shufflevector">'<tt>shufflevector</tt>' Instruction</a></li>
114 </ol>
115 </li>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]116 <li><a href="#aggregateops">Aggregate Operations</a>
117 <ol>
118 <li><a href="#i_extractvalue">'<tt>extractvalue</tt>' Instruction</a></li>
119 <li><a href="#i_insertvalue">'<tt>insertvalue</tt>' Instruction</a></li>
120 </ol>
121 </li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]122 <li><a href="#memoryops">Memory Access and Addressing Operations</a>
123 <ol>
124 <li><a href="#i_malloc">'<tt>malloc</tt>' Instruction</a></li>
125 <li><a href="#i_free">'<tt>free</tt>' Instruction</a></li>
126 <li><a href="#i_alloca">'<tt>alloca</tt>' Instruction</a></li>
127 <li><a href="#i_load">'<tt>load</tt>' Instruction</a></li>
128 <li><a href="#i_store">'<tt>store</tt>' Instruction</a></li>
129 <li><a href="#i_getelementptr">'<tt>getelementptr</tt>' Instruction</a></li>
130 </ol>
131 </li>
132 <li><a href="#convertops">Conversion Operations</a>
133 <ol>
134 <li><a href="#i_trunc">'<tt>trunc .. to</tt>' Instruction</a></li>
135 <li><a href="#i_zext">'<tt>zext .. to</tt>' Instruction</a></li>
136 <li><a href="#i_sext">'<tt>sext .. to</tt>' Instruction</a></li>
137 <li><a href="#i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a></li>
138 <li><a href="#i_fpext">'<tt>fpext .. to</tt>' Instruction</a></li>
139 <li><a href="#i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a></li>
140 <li><a href="#i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a></li>
141 <li><a href="#i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a></li>
142 <li><a href="#i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a></li>
143 <li><a href="#i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a></li>
144 <li><a href="#i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a></li>
145 <li><a href="#i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a></li>
146 </ol>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]147 </li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]148 <li><a href="#otherops">Other Operations</a>
149 <ol>
150 <li><a href="#i_icmp">'<tt>icmp</tt>' Instruction</a></li>
151 <li><a href="#i_fcmp">'<tt>fcmp</tt>' Instruction</a></li>
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]152 <li><a href="#i_vicmp">'<tt>vicmp</tt>' Instruction</a></li>
153 <li><a href="#i_vfcmp">'<tt>vfcmp</tt>' Instruction</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]154 <li><a href="#i_phi">'<tt>phi</tt>' Instruction</a></li>
155 <li><a href="#i_select">'<tt>select</tt>' Instruction</a></li>
156 <li><a href="#i_call">'<tt>call</tt>' Instruction</a></li>
157 <li><a href="#i_va_arg">'<tt>va_arg</tt>' Instruction</a></li>
158 </ol>
159 </li>
160 </ol>
161 </li>
162 <li><a href="#intrinsics">Intrinsic Functions</a>
163 <ol>
164 <li><a href="#int_varargs">Variable Argument Handling Intrinsics</a>
165 <ol>
166 <li><a href="#int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a></li>
167 <li><a href="#int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a></li>
168 <li><a href="#int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a></li>
169 </ol>
170 </li>
171 <li><a href="#int_gc">Accurate Garbage Collection Intrinsics</a>
172 <ol>
173 <li><a href="#int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a></li>
174 <li><a href="#int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a></li>
175 <li><a href="#int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a></li>
176 </ol>
177 </li>
178 <li><a href="#int_codegen">Code Generator Intrinsics</a>
179 <ol>
180 <li><a href="#int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a></li>
181 <li><a href="#int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a></li>
182 <li><a href="#int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a></li>
183 <li><a href="#int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a></li>
184 <li><a href="#int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a></li>
185 <li><a href="#int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a></li>
186 <li><a href="#int_readcyclecounter"><tt>llvm.readcyclecounter</tt>' Intrinsic</a></li>
187 </ol>
188 </li>
189 <li><a href="#int_libc">Standard C Library Intrinsics</a>
190 <ol>
191 <li><a href="#int_memcpy">'<tt>llvm.memcpy.*</tt>' Intrinsic</a></li>
192 <li><a href="#int_memmove">'<tt>llvm.memmove.*</tt>' Intrinsic</a></li>
193 <li><a href="#int_memset">'<tt>llvm.memset.*</tt>' Intrinsic</a></li>
194 <li><a href="#int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a></li>
195 <li><a href="#int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a></li>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]196 <li><a href="#int_sin">'<tt>llvm.sin.*</tt>' Intrinsic</a></li>
197 <li><a href="#int_cos">'<tt>llvm.cos.*</tt>' Intrinsic</a></li>
198 <li><a href="#int_pow">'<tt>llvm.pow.*</tt>' Intrinsic</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]199 </ol>
200 </li>
201 <li><a href="#int_manip">Bit Manipulation Intrinsics</a>
202 <ol>
203 <li><a href="#int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a></li>
204 <li><a href="#int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic </a></li>
205 <li><a href="#int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic </a></li>
206 <li><a href="#int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic </a></li>
207 <li><a href="#int_part_select">'<tt>llvm.part.select.*</tt>' Intrinsic </a></li>
208 <li><a href="#int_part_set">'<tt>llvm.part.set.*</tt>' Intrinsic </a></li>
209 </ol>
210 </li>
211 <li><a href="#int_debugger">Debugger intrinsics</a></li>
212 <li><a href="#int_eh">Exception Handling intrinsics</a></li>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]213 <li><a href="#int_trampoline">Trampoline Intrinsic</a>
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]214 <ol>
215 <li><a href="#int_it">'<tt>llvm.init.trampoline</tt>' Intrinsic</a></li>
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]216 </ol>
217 </li>
Bill Wendling9127adb2008-11-18 22:10:53 +0000[diff] [blame]218 <li><a href="#int_atomics">Atomic intrinsics</a>
219 <ol>
220 <li><a href="#int_memory_barrier"><tt>llvm.memory_barrier</tt></a></li>
221 <li><a href="#int_atomic_cmp_swap"><tt>llvm.atomic.cmp.swap</tt></a></li>
222 <li><a href="#int_atomic_swap"><tt>llvm.atomic.swap</tt></a></li>
223 <li><a href="#int_atomic_load_add"><tt>llvm.atomic.load.add</tt></a></li>
224 <li><a href="#int_atomic_load_sub"><tt>llvm.atomic.load.sub</tt></a></li>
225 <li><a href="#int_atomic_load_and"><tt>llvm.atomic.load.and</tt></a></li>
226 <li><a href="#int_atomic_load_nand"><tt>llvm.atomic.load.nand</tt></a></li>
227 <li><a href="#int_atomic_load_or"><tt>llvm.atomic.load.or</tt></a></li>
228 <li><a href="#int_atomic_load_xor"><tt>llvm.atomic.load.xor</tt></a></li>
229 <li><a href="#int_atomic_load_max"><tt>llvm.atomic.load.max</tt></a></li>
230 <li><a href="#int_atomic_load_min"><tt>llvm.atomic.load.min</tt></a></li>
231 <li><a href="#int_atomic_load_umax"><tt>llvm.atomic.load.umax</tt></a></li>
232 <li><a href="#int_atomic_load_umin"><tt>llvm.atomic.load.umin</tt></a></li>
233 </ol>
234 </li>
Reid Spencerb043f672007-07-20 19:59:11 +0000[diff] [blame]235 <li><a href="#int_general">General intrinsics</a>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]236 <ol>
Reid Spencerb043f672007-07-20 19:59:11 +0000[diff] [blame]237 <li><a href="#int_var_annotation">
Bill Wendlinge4164592008-11-19 05:56:17 +0000[diff] [blame]238 '<tt>llvm.var.annotation</tt>' Intrinsic</a></li>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]239 <li><a href="#int_annotation">
Bill Wendlinge4164592008-11-19 05:56:17 +0000[diff] [blame]240 '<tt>llvm.annotation.*</tt>' Intrinsic</a></li>
Anton Korobeynikove6e764f2008-01-15 22:31:34 +0000[diff] [blame]241 <li><a href="#int_trap">
Bill Wendlinge4164592008-11-19 05:56:17 +0000[diff] [blame]242 '<tt>llvm.trap</tt>' Intrinsic</a></li>
243 <li><a href="#int_stackprotector">
244 '<tt>llvm.stackprotector</tt>' Intrinsic</a></li>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]245 </ol>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]246 </li>
247 </ol>
248 </li>
249</ol>
250
251<div class="doc_author">
252 <p>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a>
253 and <a href="mailto:vadve@cs.uiuc.edu">Vikram Adve</a></p>
254</div>
255
256<!-- *********************************************************************** -->
257<div class="doc_section"> <a name="abstract">Abstract </a></div>
258<!-- *********************************************************************** -->
259
260<div class="doc_text">
261<p>This document is a reference manual for the LLVM assembly language.
Bill Wendlinge7846a52008-08-05 22:29:16 +0000[diff] [blame]262LLVM is a Static Single Assignment (SSA) based representation that provides
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]263type safety, low-level operations, flexibility, and the capability of
264representing 'all' high-level languages cleanly. It is the common code
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]265representation used throughout all phases of the LLVM compilation
266strategy.</p>
267</div>
268
269<!-- *********************************************************************** -->
270<div class="doc_section"> <a name="introduction">Introduction</a> </div>
271<!-- *********************************************************************** -->
272
273<div class="doc_text">
274
275<p>The LLVM code representation is designed to be used in three
276different forms: as an in-memory compiler IR, as an on-disk bitcode
277representation (suitable for fast loading by a Just-In-Time compiler),
278and as a human readable assembly language representation. This allows
279LLVM to provide a powerful intermediate representation for efficient
280compiler transformations and analysis, while providing a natural means
281to debug and visualize the transformations. The three different forms
282of LLVM are all equivalent. This document describes the human readable
283representation and notation.</p>
284
285<p>The LLVM representation aims to be light-weight and low-level
286while being expressive, typed, and extensible at the same time. It
287aims to be a "universal IR" of sorts, by being at a low enough level
288that high-level ideas may be cleanly mapped to it (similar to how
289microprocessors are "universal IR's", allowing many source languages to
290be mapped to them). By providing type information, LLVM can be used as
291the target of optimizations: for example, through pointer analysis, it
292can be proven that a C automatic variable is never accessed outside of
293the current function... allowing it to be promoted to a simple SSA
294value instead of a memory location.</p>
295
296</div>
297
298<!-- _______________________________________________________________________ -->
299<div class="doc_subsubsection"> <a name="wellformed">Well-Formedness</a> </div>
300
301<div class="doc_text">
302
303<p>It is important to note that this document describes 'well formed'
304LLVM assembly language. There is a difference between what the parser
305accepts and what is considered 'well formed'. For example, the
306following instruction is syntactically okay, but not well formed:</p>
307
308<div class="doc_code">
309<pre>
310%x = <a href="#i_add">add</a> i32 1, %x
311</pre>
312</div>
313
314<p>...because the definition of <tt>%x</tt> does not dominate all of
315its uses. The LLVM infrastructure provides a verification pass that may
316be used to verify that an LLVM module is well formed. This pass is
317automatically run by the parser after parsing input assembly and by
318the optimizer before it outputs bitcode. The violations pointed out
319by the verifier pass indicate bugs in transformation passes or input to
320the parser.</p>
321</div>
322
Chris Lattnera83fdc02007-10-03 17:34:29 +0000[diff] [blame]323<!-- Describe the typesetting conventions here. -->
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]324
325<!-- *********************************************************************** -->
326<div class="doc_section"> <a name="identifiers">Identifiers</a> </div>
327<!-- *********************************************************************** -->
328
329<div class="doc_text">
330
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]331 <p>LLVM identifiers come in two basic types: global and local. Global
332 identifiers (functions, global variables) begin with the @ character. Local
333 identifiers (register names, types) begin with the % character. Additionally,
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]334 there are three different formats for identifiers, for different purposes:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]335
336<ol>
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]337 <li>Named values are represented as a string of characters with their prefix.
338 For example, %foo, @DivisionByZero, %a.really.long.identifier. The actual
339 regular expression used is '<tt>[%@][a-zA-Z$._][a-zA-Z$._0-9]*</tt>'.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]340 Identifiers which require other characters in their names can be surrounded
Daniel Dunbar73f9a772008-10-14 23:51:43 +0000[diff] [blame]341 with quotes. Special characters may be escaped using "\xx" where xx is the
342 ASCII code for the character in hexadecimal. In this way, any character can
343 be used in a name value, even quotes themselves.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]344
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]345 <li>Unnamed values are represented as an unsigned numeric value with their
346 prefix. For example, %12, @2, %44.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]347
348 <li>Constants, which are described in a <a href="#constants">section about
349 constants</a>, below.</li>
350</ol>
351
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]352<p>LLVM requires that values start with a prefix for two reasons: Compilers
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]353don't need to worry about name clashes with reserved words, and the set of
354reserved words may be expanded in the future without penalty. Additionally,
355unnamed identifiers allow a compiler to quickly come up with a temporary
356variable without having to avoid symbol table conflicts.</p>
357
358<p>Reserved words in LLVM are very similar to reserved words in other
359languages. There are keywords for different opcodes
360('<tt><a href="#i_add">add</a></tt>',
361 '<tt><a href="#i_bitcast">bitcast</a></tt>',
362 '<tt><a href="#i_ret">ret</a></tt>', etc...), for primitive type names ('<tt><a
363href="#t_void">void</a></tt>', '<tt><a href="#t_primitive">i32</a></tt>', etc...),
364and others. These reserved words cannot conflict with variable names, because
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]365none of them start with a prefix character ('%' or '@').</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]366
367<p>Here is an example of LLVM code to multiply the integer variable
368'<tt>%X</tt>' by 8:</p>
369
370<p>The easy way:</p>
371
372<div class="doc_code">
373<pre>
374%result = <a href="#i_mul">mul</a> i32 %X, 8
375</pre>
376</div>
377
378<p>After strength reduction:</p>
379
380<div class="doc_code">
381<pre>
382%result = <a href="#i_shl">shl</a> i32 %X, i8 3
383</pre>
384</div>
385
386<p>And the hard way:</p>
387
388<div class="doc_code">
389<pre>
390<a href="#i_add">add</a> i32 %X, %X <i>; yields {i32}:%0</i>
391<a href="#i_add">add</a> i32 %0, %0 <i>; yields {i32}:%1</i>
392%result = <a href="#i_add">add</a> i32 %1, %1
393</pre>
394</div>
395
396<p>This last way of multiplying <tt>%X</tt> by 8 illustrates several
397important lexical features of LLVM:</p>
398
399<ol>
400
401 <li>Comments are delimited with a '<tt>;</tt>' and go until the end of
402 line.</li>
403
404 <li>Unnamed temporaries are created when the result of a computation is not
405 assigned to a named value.</li>
406
407 <li>Unnamed temporaries are numbered sequentially</li>
408
409</ol>
410
411<p>...and it also shows a convention that we follow in this document. When
412demonstrating instructions, we will follow an instruction with a comment that
413defines the type and name of value produced. Comments are shown in italic
414text.</p>
415
416</div>
417
418<!-- *********************************************************************** -->
419<div class="doc_section"> <a name="highlevel">High Level Structure</a> </div>
420<!-- *********************************************************************** -->
421
422<!-- ======================================================================= -->
423<div class="doc_subsection"> <a name="modulestructure">Module Structure</a>
424</div>
425
426<div class="doc_text">
427
428<p>LLVM programs are composed of "Module"s, each of which is a
429translation unit of the input programs. Each module consists of
430functions, global variables, and symbol table entries. Modules may be
431combined together with the LLVM linker, which merges function (and
432global variable) definitions, resolves forward declarations, and merges
433symbol table entries. Here is an example of the "hello world" module:</p>
434
435<div class="doc_code">
436<pre><i>; Declare the string constant as a global constant...</i>
437<a href="#identifiers">@.LC0</a> = <a href="#linkage_internal">internal</a> <a
438 href="#globalvars">constant</a> <a href="#t_array">[13 x i8]</a> c"hello world\0A\00" <i>; [13 x i8]*</i>
439
440<i>; External declaration of the puts function</i>
441<a href="#functionstructure">declare</a> i32 @puts(i8 *) <i>; i32(i8 *)* </i>
442
443<i>; Definition of main function</i>
444define i32 @main() { <i>; i32()* </i>
Dan Gohman01852382009-01-04 23:44:43 +0000[diff] [blame]445 <i>; Convert [13 x i8]* to i8 *...</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]446 %cast210 = <a
Dan Gohman01852382009-01-04 23:44:43 +0000[diff] [blame]447 href="#i_getelementptr">getelementptr</a> [13 x i8]* @.LC0, i64 0, i64 0 <i>; i8 *</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]448
449 <i>; Call puts function to write out the string to stdout...</i>
450 <a
451 href="#i_call">call</a> i32 @puts(i8 * %cast210) <i>; i32</i>
452 <a
453 href="#i_ret">ret</a> i32 0<br>}<br>
454</pre>
455</div>
456
457<p>This example is made up of a <a href="#globalvars">global variable</a>
458named "<tt>.LC0</tt>", an external declaration of the "<tt>puts</tt>"
459function, and a <a href="#functionstructure">function definition</a>
460for "<tt>main</tt>".</p>
461
462<p>In general, a module is made up of a list of global values,
463where both functions and global variables are global values. Global values are
464represented by a pointer to a memory location (in this case, a pointer to an
465array of char, and a pointer to a function), and have one of the following <a
466href="#linkage">linkage types</a>.</p>
467
468</div>
469
470<!-- ======================================================================= -->
471<div class="doc_subsection">
472 <a name="linkage">Linkage Types</a>
473</div>
474
475<div class="doc_text">
476
477<p>
478All Global Variables and Functions have one of the following types of linkage:
479</p>
480
481<dl>
482
Dale Johannesen96e7e092008-05-23 23:13:41 +0000[diff] [blame]483 <dt><tt><b><a name="linkage_internal">internal</a></b></tt>: </dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]484
485 <dd>Global values with internal linkage are only directly accessible by
486 objects in the current module. In particular, linking code into a module with
487 an internal global value may cause the internal to be renamed as necessary to
488 avoid collisions. Because the symbol is internal to the module, all
489 references can be updated. This corresponds to the notion of the
490 '<tt>static</tt>' keyword in C.
491 </dd>
492
493 <dt><tt><b><a name="linkage_linkonce">linkonce</a></b></tt>: </dt>
494
495 <dd>Globals with "<tt>linkonce</tt>" linkage are merged with other globals of
496 the same name when linkage occurs. This is typically used to implement
497 inline functions, templates, or other code which must be generated in each
498 translation unit that uses it. Unreferenced <tt>linkonce</tt> globals are
499 allowed to be discarded.
500 </dd>
501
Dale Johannesen96e7e092008-05-23 23:13:41 +0000[diff] [blame]502 <dt><tt><b><a name="linkage_common">common</a></b></tt>: </dt>
503
504 <dd>"<tt>common</tt>" linkage is exactly the same as <tt>linkonce</tt>
505 linkage, except that unreferenced <tt>common</tt> globals may not be
506 discarded. This is used for globals that may be emitted in multiple
507 translation units, but that are not guaranteed to be emitted into every
508 translation unit that uses them. One example of this is tentative
509 definitions in C, such as "<tt>int X;</tt>" at global scope.
510 </dd>
511
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]512 <dt><tt><b><a name="linkage_weak">weak</a></b></tt>: </dt>
513
Dale Johannesen96e7e092008-05-23 23:13:41 +0000[diff] [blame]514 <dd>"<tt>weak</tt>" linkage is the same as <tt>common</tt> linkage, except
515 that some targets may choose to emit different assembly sequences for them
516 for target-dependent reasons. This is used for globals that are declared
517 "weak" in C source code.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]518 </dd>
519
520 <dt><tt><b><a name="linkage_appending">appending</a></b></tt>: </dt>
521
522 <dd>"<tt>appending</tt>" linkage may only be applied to global variables of
523 pointer to array type. When two global variables with appending linkage are
524 linked together, the two global arrays are appended together. This is the
525 LLVM, typesafe, equivalent of having the system linker append together
526 "sections" with identical names when .o files are linked.
527 </dd>
528
529 <dt><tt><b><a name="linkage_externweak">extern_weak</a></b></tt>: </dt>
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]530 <dd>The semantics of this linkage follow the ELF object file model: the
531 symbol is weak until linked, if not linked, the symbol becomes null instead
532 of being an undefined reference.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]533 </dd>
534
535 <dt><tt><b><a name="linkage_external">externally visible</a></b></tt>:</dt>
536
537 <dd>If none of the above identifiers are used, the global is externally
538 visible, meaning that it participates in linkage and can be used to resolve
539 external symbol references.
540 </dd>
541</dl>
542
543 <p>
544 The next two types of linkage are targeted for Microsoft Windows platform
545 only. They are designed to support importing (exporting) symbols from (to)
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]546 DLLs (Dynamic Link Libraries).
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]547 </p>
548
549 <dl>
550 <dt><tt><b><a name="linkage_dllimport">dllimport</a></b></tt>: </dt>
551
552 <dd>"<tt>dllimport</tt>" linkage causes the compiler to reference a function
553 or variable via a global pointer to a pointer that is set up by the DLL
554 exporting the symbol. On Microsoft Windows targets, the pointer name is
Dan Gohman5ec99832009-01-12 21:35:55 +0000[diff] [blame]555 formed by combining <code>__imp_</code> and the function or variable name.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]556 </dd>
557
558 <dt><tt><b><a name="linkage_dllexport">dllexport</a></b></tt>: </dt>
559
560 <dd>"<tt>dllexport</tt>" linkage causes the compiler to provide a global
561 pointer to a pointer in a DLL, so that it can be referenced with the
562 <tt>dllimport</tt> attribute. On Microsoft Windows targets, the pointer
Dan Gohman5ec99832009-01-12 21:35:55 +0000[diff] [blame]563 name is formed by combining <code>__imp_</code> and the function or variable
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]564 name.
565 </dd>
566
567</dl>
568
Dan Gohman4dfac702008-11-24 17:18:39 +0000[diff] [blame]569<p>For example, since the "<tt>.LC0</tt>"
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]570variable is defined to be internal, if another module defined a "<tt>.LC0</tt>"
571variable and was linked with this one, one of the two would be renamed,
572preventing a collision. Since "<tt>main</tt>" and "<tt>puts</tt>" are
573external (i.e., lacking any linkage declarations), they are accessible
574outside of the current module.</p>
575<p>It is illegal for a function <i>declaration</i>
576to have any linkage type other than "externally visible", <tt>dllimport</tt>,
577or <tt>extern_weak</tt>.</p>
578<p>Aliases can have only <tt>external</tt>, <tt>internal</tt> and <tt>weak</tt>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]579linkages.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]580</div>
581
582<!-- ======================================================================= -->
583<div class="doc_subsection">
584 <a name="callingconv">Calling Conventions</a>
585</div>
586
587<div class="doc_text">
588
589<p>LLVM <a href="#functionstructure">functions</a>, <a href="#i_call">calls</a>
590and <a href="#i_invoke">invokes</a> can all have an optional calling convention
591specified for the call. The calling convention of any pair of dynamic
592caller/callee must match, or the behavior of the program is undefined. The
593following calling conventions are supported by LLVM, and more may be added in
594the future:</p>
595
596<dl>
597 <dt><b>"<tt>ccc</tt>" - The C calling convention</b>:</dt>
598
599 <dd>This calling convention (the default if no other calling convention is
600 specified) matches the target C calling conventions. This calling convention
601 supports varargs function calls and tolerates some mismatch in the declared
602 prototype and implemented declaration of the function (as does normal C).
603 </dd>
604
605 <dt><b>"<tt>fastcc</tt>" - The fast calling convention</b>:</dt>
606
607 <dd>This calling convention attempts to make calls as fast as possible
608 (e.g. by passing things in registers). This calling convention allows the
609 target to use whatever tricks it wants to produce fast code for the target,
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]610 without having to conform to an externally specified ABI (Application Binary
611 Interface). Implementations of this convention should allow arbitrary
Arnold Schwaighofer07444922008-05-14 09:17:12 +0000[diff] [blame]612 <a href="CodeGenerator.html#tailcallopt">tail call optimization</a> to be
613 supported. This calling convention does not support varargs and requires the
614 prototype of all callees to exactly match the prototype of the function
615 definition.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]616 </dd>
617
618 <dt><b>"<tt>coldcc</tt>" - The cold calling convention</b>:</dt>
619
620 <dd>This calling convention attempts to make code in the caller as efficient
621 as possible under the assumption that the call is not commonly executed. As
622 such, these calls often preserve all registers so that the call does not break
623 any live ranges in the caller side. This calling convention does not support
624 varargs and requires the prototype of all callees to exactly match the
625 prototype of the function definition.
626 </dd>
627
628 <dt><b>"<tt>cc &lt;<em>n</em>&gt;</tt>" - Numbered convention</b>:</dt>
629
630 <dd>Any calling convention may be specified by number, allowing
631 target-specific calling conventions to be used. Target specific calling
632 conventions start at 64.
633 </dd>
634</dl>
635
636<p>More calling conventions can be added/defined on an as-needed basis, to
637support pascal conventions or any other well-known target-independent
638convention.</p>
639
640</div>
641
642<!-- ======================================================================= -->
643<div class="doc_subsection">
644 <a name="visibility">Visibility Styles</a>
645</div>
646
647<div class="doc_text">
648
649<p>
650All Global Variables and Functions have one of the following visibility styles:
651</p>
652
653<dl>
654 <dt><b>"<tt>default</tt>" - Default style</b>:</dt>
655
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]656 <dd>On targets that use the ELF object file format, default visibility means
657 that the declaration is visible to other
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]658 modules and, in shared libraries, means that the declared entity may be
659 overridden. On Darwin, default visibility means that the declaration is
660 visible to other modules. Default visibility corresponds to "external
661 linkage" in the language.
662 </dd>
663
664 <dt><b>"<tt>hidden</tt>" - Hidden style</b>:</dt>
665
666 <dd>Two declarations of an object with hidden visibility refer to the same
667 object if they are in the same shared object. Usually, hidden visibility
668 indicates that the symbol will not be placed into the dynamic symbol table,
669 so no other module (executable or shared library) can reference it
670 directly.
671 </dd>
672
673 <dt><b>"<tt>protected</tt>" - Protected style</b>:</dt>
674
675 <dd>On ELF, protected visibility indicates that the symbol will be placed in
676 the dynamic symbol table, but that references within the defining module will
677 bind to the local symbol. That is, the symbol cannot be overridden by another
678 module.
679 </dd>
680</dl>
681
682</div>
683
684<!-- ======================================================================= -->
685<div class="doc_subsection">
Chris Lattner5b6dc6e2009-01-11 20:53:49 +0000[diff] [blame]686 <a name="namedtypes">Named Types</a>
687</div>
688
689<div class="doc_text">
690
691<p>LLVM IR allows you to specify name aliases for certain types. This can make
692it easier to read the IR and make the IR more condensed (particularly when
693recursive types are involved). An example of a name specification is:
694</p>
695
696<div class="doc_code">
697<pre>
698%mytype = type { %mytype*, i32 }
699</pre>
700</div>
701
702<p>You may give a name to any <a href="#typesystem">type</a> except "<a
703href="t_void">void</a>". Type name aliases may be used anywhere a type is
704expected with the syntax "%mytype".</p>
705
706<p>Note that type names are aliases for the structural type that they indicate,
707and that you can therefore specify multiple names for the same type. This often
708leads to confusing behavior when dumping out a .ll file. Since LLVM IR uses
709structural typing, the name is not part of the type. When printing out LLVM IR,
710the printer will pick <em>one name</em> to render all types of a particular
711shape. This means that if you have code where two different source types end up
712having the same LLVM type, that the dumper will sometimes print the "wrong" or
713unexpected type. This is an important design point and isn't going to
714change.</p>
715
716</div>
717
718
719<!-- ======================================================================= -->
720<div class="doc_subsection">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]721 <a name="globalvars">Global Variables</a>
722</div>
723
724<div class="doc_text">
725
726<p>Global variables define regions of memory allocated at compilation time
727instead of run-time. Global variables may optionally be initialized, may have
728an explicit section to be placed in, and may have an optional explicit alignment
729specified. A variable may be defined as "thread_local", which means that it
730will not be shared by threads (each thread will have a separated copy of the
731variable). A variable may be defined as a global "constant," which indicates
732that the contents of the variable will <b>never</b> be modified (enabling better
733optimization, allowing the global data to be placed in the read-only section of
734an executable, etc). Note that variables that need runtime initialization
735cannot be marked "constant" as there is a store to the variable.</p>
736
737<p>
738LLVM explicitly allows <em>declarations</em> of global variables to be marked
739constant, even if the final definition of the global is not. This capability
740can be used to enable slightly better optimization of the program, but requires
741the language definition to guarantee that optimizations based on the
742'constantness' are valid for the translation units that do not include the
743definition.
744</p>
745
746<p>As SSA values, global variables define pointer values that are in
747scope (i.e. they dominate) all basic blocks in the program. Global
748variables always define a pointer to their "content" type because they
749describe a region of memory, and all memory objects in LLVM are
750accessed through pointers.</p>
751
Christopher Lambdd0049d2007-12-11 09:31:00 +0000[diff] [blame]752<p>A global variable may be declared to reside in a target-specifc numbered
753address space. For targets that support them, address spaces may affect how
754optimizations are performed and/or what target instructions are used to access
Christopher Lamb20a39e92007-12-12 08:44:39 +0000[diff] [blame]755the variable. The default address space is zero. The address space qualifier
756must precede any other attributes.</p>
Christopher Lambdd0049d2007-12-11 09:31:00 +0000[diff] [blame]757
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]758<p>LLVM allows an explicit section to be specified for globals. If the target
759supports it, it will emit globals to the section specified.</p>
760
761<p>An explicit alignment may be specified for a global. If not present, or if
762the alignment is set to zero, the alignment of the global is set by the target
763to whatever it feels convenient. If an explicit alignment is specified, the
764global is forced to have at least that much alignment. All alignments must be
765a power of 2.</p>
766
Christopher Lambdd0049d2007-12-11 09:31:00 +0000[diff] [blame]767<p>For example, the following defines a global in a numbered address space with
768an initializer, section, and alignment:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]769
770<div class="doc_code">
771<pre>
Dan Gohman21ef02c2009-01-11 00:40:00 +0000[diff] [blame]772@G = addrspace(5) constant float 1.0, section "foo", align 4
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]773</pre>
774</div>
775
776</div>
777
778
779<!-- ======================================================================= -->
780<div class="doc_subsection">
781 <a name="functionstructure">Functions</a>
782</div>
783
784<div class="doc_text">
785
786<p>LLVM function definitions consist of the "<tt>define</tt>" keyord,
787an optional <a href="#linkage">linkage type</a>, an optional
788<a href="#visibility">visibility style</a>, an optional
789<a href="#callingconv">calling convention</a>, a return type, an optional
790<a href="#paramattrs">parameter attribute</a> for the return type, a function
791name, a (possibly empty) argument list (each with optional
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]792<a href="#paramattrs">parameter attributes</a>), optional
793<a href="#fnattrs">function attributes</a>, an optional section,
794an optional alignment, an optional <a href="#gc">garbage collector name</a>,
Chris Lattneree202542008-10-04 18:10:21 +0000[diff] [blame]795an opening curly brace, a list of basic blocks, and a closing curly brace.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]796
797LLVM function declarations consist of the "<tt>declare</tt>" keyword, an
798optional <a href="#linkage">linkage type</a>, an optional
799<a href="#visibility">visibility style</a>, an optional
800<a href="#callingconv">calling convention</a>, a return type, an optional
801<a href="#paramattrs">parameter attribute</a> for the return type, a function
Gordon Henriksen13fe5e32007-12-10 03:18:06 +0000[diff] [blame]802name, a possibly empty list of arguments, an optional alignment, and an optional
Gordon Henriksend606f5b2007-12-10 03:30:21 +0000[diff] [blame]803<a href="#gc">garbage collector name</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]804
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]805<p>A function definition contains a list of basic blocks, forming the CFG
806(Control Flow Graph) for
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]807the function. Each basic block may optionally start with a label (giving the
808basic block a symbol table entry), contains a list of instructions, and ends
809with a <a href="#terminators">terminator</a> instruction (such as a branch or
810function return).</p>
811
812<p>The first basic block in a function is special in two ways: it is immediately
813executed on entrance to the function, and it is not allowed to have predecessor
814basic blocks (i.e. there can not be any branches to the entry block of a
815function). Because the block can have no predecessors, it also cannot have any
816<a href="#i_phi">PHI nodes</a>.</p>
817
818<p>LLVM allows an explicit section to be specified for functions. If the target
819supports it, it will emit functions to the section specified.</p>
820
821<p>An explicit alignment may be specified for a function. If not present, or if
822the alignment is set to zero, the alignment of the function is set by the target
823to whatever it feels convenient. If an explicit alignment is specified, the
824function is forced to have at least that much alignment. All alignments must be
825a power of 2.</p>
826
Devang Pateld0bfcc72008-10-07 17:48:33 +0000[diff] [blame]827 <h5>Syntax:</h5>
828
829<div class="doc_code">
Chris Lattner1e5c5cd02008-10-13 16:55:18 +0000[diff] [blame]830<tt>
831define [<a href="#linkage">linkage</a>] [<a href="#visibility">visibility</a>]
832 [<a href="#callingconv">cconv</a>] [<a href="#paramattrs">ret attrs</a>]
833 &lt;ResultType&gt; @&lt;FunctionName&gt; ([argument list])
834 [<a href="#fnattrs">fn Attrs</a>] [section "name"] [align N]
835 [<a href="#gc">gc</a>] { ... }
836</tt>
Devang Pateld0bfcc72008-10-07 17:48:33 +0000[diff] [blame]837</div>
838
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]839</div>
840
841
842<!-- ======================================================================= -->
843<div class="doc_subsection">
844 <a name="aliasstructure">Aliases</a>
845</div>
846<div class="doc_text">
847 <p>Aliases act as "second name" for the aliasee value (which can be either
Anton Korobeynikov96822812008-03-22 08:36:14 +0000[diff] [blame]848 function, global variable, another alias or bitcast of global value). Aliases
849 may have an optional <a href="#linkage">linkage type</a>, and an
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]850 optional <a href="#visibility">visibility style</a>.</p>
851
852 <h5>Syntax:</h5>
853
854<div class="doc_code">
855<pre>
Duncan Sandsd7bfabf2008-09-12 20:48:21 +0000[diff] [blame]856@&lt;Name&gt; = alias [Linkage] [Visibility] &lt;AliaseeTy&gt; @&lt;Aliasee&gt;
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]857</pre>
858</div>
859
860</div>
861
862
863
864<!-- ======================================================================= -->
865<div class="doc_subsection"><a name="paramattrs">Parameter Attributes</a></div>
866<div class="doc_text">
867 <p>The return type and each parameter of a function type may have a set of
868 <i>parameter attributes</i> associated with them. Parameter attributes are
869 used to communicate additional information about the result or parameters of
Duncan Sandsf5588dc2007-11-27 13:23:08 +0000[diff] [blame]870 a function. Parameter attributes are considered to be part of the function,
871 not of the function type, so functions with different parameter attributes
872 can have the same function type.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]873
874 <p>Parameter attributes are simple keywords that follow the type specified. If
875 multiple parameter attributes are needed, they are space separated. For
876 example:</p>
877
878<div class="doc_code">
879<pre>
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]880declare i32 @printf(i8* noalias , ...)
Chris Lattnerf33b8452008-10-04 18:33:34 +0000[diff] [blame]881declare i32 @atoi(i8 zeroext)
882declare signext i8 @returns_signed_char()
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]883</pre>
884</div>
885
Duncan Sandsf5588dc2007-11-27 13:23:08 +0000[diff] [blame]886 <p>Note that any attributes for the function result (<tt>nounwind</tt>,
887 <tt>readonly</tt>) come immediately after the argument list.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]888
889 <p>Currently, only the following parameter attributes are defined:</p>
890 <dl>
Reid Spencerf234bed2007-07-19 23:13:04 +0000[diff] [blame]891 <dt><tt>zeroext</tt></dt>
Chris Lattnerf33b8452008-10-04 18:33:34 +0000[diff] [blame]892 <dd>This indicates to the code generator that the parameter or return value
893 should be zero-extended to a 32-bit value by the caller (for a parameter)
894 or the callee (for a return value).</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]895
Reid Spencerf234bed2007-07-19 23:13:04 +0000[diff] [blame]896 <dt><tt>signext</tt></dt>
Chris Lattnerf33b8452008-10-04 18:33:34 +0000[diff] [blame]897 <dd>This indicates to the code generator that the parameter or return value
898 should be sign-extended to a 32-bit value by the caller (for a parameter)
899 or the callee (for a return value).</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]900
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]901 <dt><tt>inreg</tt></dt>
Dale Johannesenc08a0e22008-09-25 20:47:45 +0000[diff] [blame]902 <dd>This indicates that this parameter or return value should be treated
903 in a special target-dependent fashion during while emitting code for a
904 function call or return (usually, by putting it in a register as opposed
Chris Lattnerf33b8452008-10-04 18:33:34 +0000[diff] [blame]905 to memory, though some targets use it to distinguish between two different
906 kinds of registers). Use of this attribute is target-specific.</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]907
Duncan Sands7d94e5a2008-10-06 08:14:18 +0000[diff] [blame]908 <dt><tt><a name="byval">byval</a></tt></dt>
Chris Lattner04c86182008-01-15 04:34:22 +0000[diff] [blame]909 <dd>This indicates that the pointer parameter should really be passed by
910 value to the function. The attribute implies that a hidden copy of the
911 pointee is made between the caller and the callee, so the callee is unable
Chris Lattner6a9f3c42008-08-05 18:21:08 +0000[diff] [blame]912 to modify the value in the callee. This attribute is only valid on LLVM
Chris Lattner04c86182008-01-15 04:34:22 +0000[diff] [blame]913 pointer arguments. It is generally used to pass structs and arrays by
Duncan Sands7d94e5a2008-10-06 08:14:18 +0000[diff] [blame]914 value, but is also valid on pointers to scalars. The copy is considered to
915 belong to the caller not the callee (for example,
916 <tt><a href="#readonly">readonly</a></tt> functions should not write to
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]917 <tt>byval</tt> parameters). This is not a valid attribute for return
918 values. </dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]919
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]920 <dt><tt>sret</tt></dt>
Duncan Sands616cc032008-02-18 04:19:38 +0000[diff] [blame]921 <dd>This indicates that the pointer parameter specifies the address of a
922 structure that is the return value of the function in the source program.
Chris Lattnerf33b8452008-10-04 18:33:34 +0000[diff] [blame]923 This pointer must be guaranteed by the caller to be valid: loads and stores
924 to the structure may be assumed by the callee to not to trap. This may only
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]925 be applied to the first parameter. This is not a valid attribute for
926 return values. </dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]927
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]928 <dt><tt>noalias</tt></dt>
Nick Lewyckyff384472008-11-24 03:41:24 +0000[diff] [blame]929 <dd>This indicates that the pointer does not alias any global or any other
930 parameter. The caller is responsible for ensuring that this is the
Nick Lewyckya604a942008-11-24 05:00:44 +0000[diff] [blame]931 case. On a function return value, <tt>noalias</tt> additionally indicates
932 that the pointer does not alias any other pointers visible to the
Nick Lewycky40fb6f22008-12-19 06:39:12 +0000[diff] [blame]933 caller. For further details, please see the discussion of the NoAlias
934 response in
935 <a href="http://llvm.org/docs/AliasAnalysis.html#MustMayNo">alias
936 analysis</a>.</dd>
937
938 <dt><tt>nocapture</tt></dt>
939 <dd>This indicates that the callee does not make any copies of the pointer
940 that outlive the callee itself. This is not a valid attribute for return
941 values.</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]942
Duncan Sands4ee46812007-07-27 19:57:41 +0000[diff] [blame]943 <dt><tt>nest</tt></dt>
Duncan Sandsf1a7d4c2008-07-08 09:27:25 +0000[diff] [blame]944 <dd>This indicates that the pointer parameter can be excised using the
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]945 <a href="#int_trampoline">trampoline intrinsics</a>. This is not a valid
946 attribute for return values.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]947 </dl>
948
949</div>
950
951<!-- ======================================================================= -->
952<div class="doc_subsection">
Gordon Henriksen13fe5e32007-12-10 03:18:06 +0000[diff] [blame]953 <a name="gc">Garbage Collector Names</a>
954</div>
955
956<div class="doc_text">
957<p>Each function may specify a garbage collector name, which is simply a
958string.</p>
959
960<div class="doc_code"><pre
961>define void @f() gc "name" { ...</pre></div>
962
963<p>The compiler declares the supported values of <i>name</i>. Specifying a
964collector which will cause the compiler to alter its output in order to support
965the named garbage collection algorithm.</p>
966</div>
967
968<!-- ======================================================================= -->
969<div class="doc_subsection">
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]970 <a name="fnattrs">Function Attributes</a>
Devang Pateld468f1c2008-09-04 23:05:13 +0000[diff] [blame]971</div>
972
973<div class="doc_text">
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]974
975<p>Function attributes are set to communicate additional information about
976 a function. Function attributes are considered to be part of the function,
977 not of the function type, so functions with different parameter attributes
978 can have the same function type.</p>
979
980 <p>Function attributes are simple keywords that follow the type specified. If
981 multiple attributes are needed, they are space separated. For
982 example:</p>
Devang Pateld468f1c2008-09-04 23:05:13 +0000[diff] [blame]983
984<div class="doc_code">
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]985<pre>
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]986define void @f() noinline { ... }
987define void @f() alwaysinline { ... }
988define void @f() alwaysinline optsize { ... }
989define void @f() optsize
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]990</pre>
Devang Pateld468f1c2008-09-04 23:05:13 +0000[diff] [blame]991</div>
992
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]993<dl>
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]994<dt><tt>alwaysinline</tt></dt>
Chris Lattner82d70f92008-10-04 18:23:17 +0000[diff] [blame]995<dd>This attribute indicates that the inliner should attempt to inline this
996function into callers whenever possible, ignoring any active inlining size
997threshold for this caller.</dd>
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]998
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]999<dt><tt>noinline</tt></dt>
Chris Lattner82d70f92008-10-04 18:23:17 +0000[diff] [blame]1000<dd>This attribute indicates that the inliner should never inline this function
Chris Lattner377f3ae2008-10-05 17:14:59 +0000[diff] [blame]1001in any situation. This attribute may not be used together with the
Chris Lattner82d70f92008-10-04 18:23:17 +0000[diff] [blame]1002<tt>alwaysinline</tt> attribute.</dd>
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]1003
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]1004<dt><tt>optsize</tt></dt>
Devang Patelc29099b2008-09-29 18:34:44 +0000[diff] [blame]1005<dd>This attribute suggests that optimization passes and code generator passes
Chris Lattner82d70f92008-10-04 18:23:17 +0000[diff] [blame]1006make choices that keep the code size of this function low, and otherwise do
1007optimizations specifically to reduce code size.</dd>
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]1008
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]1009<dt><tt>noreturn</tt></dt>
Chris Lattner82d70f92008-10-04 18:23:17 +0000[diff] [blame]1010<dd>This function attribute indicates that the function never returns normally.
1011This produces undefined behavior at runtime if the function ever does
1012dynamically return.</dd>
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]1013
1014<dt><tt>nounwind</tt></dt>
Chris Lattner82d70f92008-10-04 18:23:17 +0000[diff] [blame]1015<dd>This function attribute indicates that the function never returns with an
1016unwind or exceptional control flow. If the function does unwind, its runtime
1017behavior is undefined.</dd>
1018
1019<dt><tt>readnone</tt></dt>
Duncan Sands7d94e5a2008-10-06 08:14:18 +0000[diff] [blame]1020<dd>This attribute indicates that the function computes its result (or the
1021exception it throws) based strictly on its arguments, without dereferencing any
1022pointer arguments or otherwise accessing any mutable state (e.g. memory, control
1023registers, etc) visible to caller functions. It does not write through any
1024pointer arguments (including <tt><a href="#byval">byval</a></tt> arguments) and
1025never changes any state visible to callers.</dd>
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]1026
Duncan Sands7d94e5a2008-10-06 08:14:18 +0000[diff] [blame]1027<dt><tt><a name="readonly">readonly</a></tt></dt>
1028<dd>This attribute indicates that the function does not write through any
1029pointer arguments (including <tt><a href="#byval">byval</a></tt> arguments)
1030or otherwise modify any state (e.g. memory, control registers, etc) visible to
1031caller functions. It may dereference pointer arguments and read state that may
1032be set in the caller. A readonly function always returns the same value (or
1033throws the same exception) when called with the same set of arguments and global
1034state.</dd>
Bill Wendlingdfaabba2008-11-13 01:02:51 +0000[diff] [blame]1035
1036<dt><tt><a name="ssp">ssp</a></tt></dt>
Bill Wendling34f70d82008-11-26 19:19:05 +0000[diff] [blame]1037<dd>This attribute indicates that the function should emit a stack smashing
Bill Wendlingdfaabba2008-11-13 01:02:51 +0000[diff] [blame]1038protector. It is in the form of a "canary"&mdash;a random value placed on the
1039stack before the local variables that's checked upon return from the function to
1040see if it has been overwritten. A heuristic is used to determine if a function
Bill Wendling34f70d82008-11-26 19:19:05 +0000[diff] [blame]1041needs stack protectors or not.
Bill Wendlingdfaabba2008-11-13 01:02:51 +0000[diff] [blame]1042
Bill Wendlingbe9ec3f2008-11-26 19:07:40 +0000[diff] [blame]1043<p>If a function that has an <tt>ssp</tt> attribute is inlined into a function
1044that doesn't have an <tt>ssp</tt> attribute, then the resulting function will
1045have an <tt>ssp</tt> attribute.</p></dd>
1046
1047<dt><tt>sspreq</tt></dt>
Bill Wendling34f70d82008-11-26 19:19:05 +0000[diff] [blame]1048<dd>This attribute indicates that the function should <em>always</em> emit a
Bill Wendlingdfaabba2008-11-13 01:02:51 +0000[diff] [blame]1049stack smashing protector. This overrides the <tt><a href="#ssp">ssp</a></tt>
Bill Wendling34f70d82008-11-26 19:19:05 +0000[diff] [blame]1050function attribute.
Bill Wendlingbe9ec3f2008-11-26 19:07:40 +0000[diff] [blame]1051
1052<p>If a function that has an <tt>sspreq</tt> attribute is inlined into a
1053function that doesn't have an <tt>sspreq</tt> attribute or which has
1054an <tt>ssp</tt> attribute, then the resulting function will have
1055an <tt>sspreq</tt> attribute.</p></dd>
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]1056</dl>
1057
Devang Pateld468f1c2008-09-04 23:05:13 +0000[diff] [blame]1058</div>
1059
1060<!-- ======================================================================= -->
1061<div class="doc_subsection">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1062 <a name="moduleasm">Module-Level Inline Assembly</a>
1063</div>
1064
1065<div class="doc_text">
1066<p>
1067Modules may contain "module-level inline asm" blocks, which corresponds to the
1068GCC "file scope inline asm" blocks. These blocks are internally concatenated by
1069LLVM and treated as a single unit, but may be separated in the .ll file if
1070desired. The syntax is very simple:
1071</p>
1072
1073<div class="doc_code">
1074<pre>
1075module asm "inline asm code goes here"
1076module asm "more can go here"
1077</pre>
1078</div>
1079
1080<p>The strings can contain any character by escaping non-printable characters.
1081 The escape sequence used is simply "\xx" where "xx" is the two digit hex code
1082 for the number.
1083</p>
1084
1085<p>
1086 The inline asm code is simply printed to the machine code .s file when
1087 assembly code is generated.
1088</p>
1089</div>
1090
1091<!-- ======================================================================= -->
1092<div class="doc_subsection">
1093 <a name="datalayout">Data Layout</a>
1094</div>
1095
1096<div class="doc_text">
1097<p>A module may specify a target specific data layout string that specifies how
1098data is to be laid out in memory. The syntax for the data layout is simply:</p>
1099<pre> target datalayout = "<i>layout specification</i>"</pre>
1100<p>The <i>layout specification</i> consists of a list of specifications
1101separated by the minus sign character ('-'). Each specification starts with a
1102letter and may include other information after the letter to define some
1103aspect of the data layout. The specifications accepted are as follows: </p>
1104<dl>
1105 <dt><tt>E</tt></dt>
1106 <dd>Specifies that the target lays out data in big-endian form. That is, the
1107 bits with the most significance have the lowest address location.</dd>
1108 <dt><tt>e</tt></dt>
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]1109 <dd>Specifies that the target lays out data in little-endian form. That is,
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1110 the bits with the least significance have the lowest address location.</dd>
1111 <dt><tt>p:<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1112 <dd>This specifies the <i>size</i> of a pointer and its <i>abi</i> and
1113 <i>preferred</i> alignments. All sizes are in bits. Specifying the <i>pref</i>
1114 alignment is optional. If omitted, the preceding <tt>:</tt> should be omitted
1115 too.</dd>
1116 <dt><tt>i<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1117 <dd>This specifies the alignment for an integer type of a given bit
1118 <i>size</i>. The value of <i>size</i> must be in the range [1,2^23).</dd>
1119 <dt><tt>v<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1120 <dd>This specifies the alignment for a vector type of a given bit
1121 <i>size</i>.</dd>
1122 <dt><tt>f<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1123 <dd>This specifies the alignment for a floating point type of a given bit
1124 <i>size</i>. The value of <i>size</i> must be either 32 (float) or 64
1125 (double).</dd>
1126 <dt><tt>a<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1127 <dd>This specifies the alignment for an aggregate type of a given bit
1128 <i>size</i>.</dd>
1129</dl>
1130<p>When constructing the data layout for a given target, LLVM starts with a
1131default set of specifications which are then (possibly) overriden by the
1132specifications in the <tt>datalayout</tt> keyword. The default specifications
1133are given in this list:</p>
1134<ul>
1135 <li><tt>E</tt> - big endian</li>
1136 <li><tt>p:32:64:64</tt> - 32-bit pointers with 64-bit alignment</li>
1137 <li><tt>i1:8:8</tt> - i1 is 8-bit (byte) aligned</li>
1138 <li><tt>i8:8:8</tt> - i8 is 8-bit (byte) aligned</li>
1139 <li><tt>i16:16:16</tt> - i16 is 16-bit aligned</li>
1140 <li><tt>i32:32:32</tt> - i32 is 32-bit aligned</li>
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]1141 <li><tt>i64:32:64</tt> - i64 has ABI alignment of 32-bits but preferred
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1142 alignment of 64-bits</li>
1143 <li><tt>f32:32:32</tt> - float is 32-bit aligned</li>
1144 <li><tt>f64:64:64</tt> - double is 64-bit aligned</li>
1145 <li><tt>v64:64:64</tt> - 64-bit vector is 64-bit aligned</li>
1146 <li><tt>v128:128:128</tt> - 128-bit vector is 128-bit aligned</li>
1147 <li><tt>a0:0:1</tt> - aggregates are 8-bit aligned</li>
1148</ul>
Chris Lattner6a9f3c42008-08-05 18:21:08 +0000[diff] [blame]1149<p>When LLVM is determining the alignment for a given type, it uses the
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]1150following rules:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1151<ol>
1152 <li>If the type sought is an exact match for one of the specifications, that
1153 specification is used.</li>
1154 <li>If no match is found, and the type sought is an integer type, then the
1155 smallest integer type that is larger than the bitwidth of the sought type is
1156 used. If none of the specifications are larger than the bitwidth then the the
1157 largest integer type is used. For example, given the default specifications
1158 above, the i7 type will use the alignment of i8 (next largest) while both
1159 i65 and i256 will use the alignment of i64 (largest specified).</li>
1160 <li>If no match is found, and the type sought is a vector type, then the
1161 largest vector type that is smaller than the sought vector type will be used
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]1162 as a fall back. This happens because &lt;128 x double&gt; can be implemented
1163 in terms of 64 &lt;2 x double&gt;, for example.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1164</ol>
1165</div>
1166
1167<!-- *********************************************************************** -->
1168<div class="doc_section"> <a name="typesystem">Type System</a> </div>
1169<!-- *********************************************************************** -->
1170
1171<div class="doc_text">
1172
1173<p>The LLVM type system is one of the most important features of the
1174intermediate representation. Being typed enables a number of
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]1175optimizations to be performed on the intermediate representation directly,
1176without having to do
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1177extra analyses on the side before the transformation. A strong type
1178system makes it easier to read the generated code and enables novel
1179analyses and transformations that are not feasible to perform on normal
1180three address code representations.</p>
1181
1182</div>
1183
1184<!-- ======================================================================= -->
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1185<div class="doc_subsection"> <a name="t_classifications">Type
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1186Classifications</a> </div>
1187<div class="doc_text">
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1188<p>The types fall into a few useful
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1189classifications:</p>
1190
1191<table border="1" cellspacing="0" cellpadding="4">
1192 <tbody>
1193 <tr><th>Classification</th><th>Types</th></tr>
1194 <tr>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1195 <td><a href="#t_integer">integer</a></td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1196 <td><tt>i1, i2, i3, ... i8, ... i16, ... i32, ... i64, ... </tt></td>
1197 </tr>
1198 <tr>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1199 <td><a href="#t_floating">floating point</a></td>
1200 <td><tt>float, double, x86_fp80, fp128, ppc_fp128</tt></td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1201 </tr>
1202 <tr>
1203 <td><a name="t_firstclass">first class</a></td>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1204 <td><a href="#t_integer">integer</a>,
1205 <a href="#t_floating">floating point</a>,
1206 <a href="#t_pointer">pointer</a>,
Dan Gohmanf6237db2008-06-18 18:42:13 +0000[diff] [blame]1207 <a href="#t_vector">vector</a>,
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]1208 <a href="#t_struct">structure</a>,
1209 <a href="#t_array">array</a>,
Dan Gohmand13951e2008-05-23 22:50:26 +0000[diff] [blame]1210 <a href="#t_label">label</a>.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1211 </td>
1212 </tr>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1213 <tr>
1214 <td><a href="#t_primitive">primitive</a></td>
1215 <td><a href="#t_label">label</a>,
1216 <a href="#t_void">void</a>,
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1217 <a href="#t_floating">floating point</a>.</td>
1218 </tr>
1219 <tr>
1220 <td><a href="#t_derived">derived</a></td>
1221 <td><a href="#t_integer">integer</a>,
1222 <a href="#t_array">array</a>,
1223 <a href="#t_function">function</a>,
1224 <a href="#t_pointer">pointer</a>,
1225 <a href="#t_struct">structure</a>,
1226 <a href="#t_pstruct">packed structure</a>,
1227 <a href="#t_vector">vector</a>,
1228 <a href="#t_opaque">opaque</a>.
Dan Gohman032ba852008-10-14 16:32:04 +0000[diff] [blame]1229 </td>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1230 </tr>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1231 </tbody>
1232</table>
1233
1234<p>The <a href="#t_firstclass">first class</a> types are perhaps the
1235most important. Values of these types are the only ones which can be
1236produced by instructions, passed as arguments, or used as operands to
Dan Gohman4f29e422008-05-23 21:53:15 +0000[diff] [blame]1237instructions.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1238</div>
1239
1240<!-- ======================================================================= -->
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1241<div class="doc_subsection"> <a name="t_primitive">Primitive Types</a> </div>
Chris Lattner86437612008-01-04 04:34:14 +0000[diff] [blame]1242
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1243<div class="doc_text">
1244<p>The primitive types are the fundamental building blocks of the LLVM
1245system.</p>
1246
Chris Lattner86437612008-01-04 04:34:14 +0000[diff] [blame]1247</div>
1248
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1249<!-- _______________________________________________________________________ -->
1250<div class="doc_subsubsection"> <a name="t_floating">Floating Point Types</a> </div>
1251
1252<div class="doc_text">
1253 <table>
1254 <tbody>
1255 <tr><th>Type</th><th>Description</th></tr>
1256 <tr><td><tt>float</tt></td><td>32-bit floating point value</td></tr>
1257 <tr><td><tt>double</tt></td><td>64-bit floating point value</td></tr>
1258 <tr><td><tt>fp128</tt></td><td>128-bit floating point value (112-bit mantissa)</td></tr>
1259 <tr><td><tt>x86_fp80</tt></td><td>80-bit floating point value (X87)</td></tr>
1260 <tr><td><tt>ppc_fp128</tt></td><td>128-bit floating point value (two 64-bits)</td></tr>
1261 </tbody>
1262 </table>
1263</div>
1264
1265<!-- _______________________________________________________________________ -->
1266<div class="doc_subsubsection"> <a name="t_void">Void Type</a> </div>
1267
1268<div class="doc_text">
1269<h5>Overview:</h5>
1270<p>The void type does not represent any value and has no size.</p>
1271
1272<h5>Syntax:</h5>
1273
1274<pre>
1275 void
1276</pre>
1277</div>
1278
1279<!-- _______________________________________________________________________ -->
1280<div class="doc_subsubsection"> <a name="t_label">Label Type</a> </div>
1281
1282<div class="doc_text">
1283<h5>Overview:</h5>
1284<p>The label type represents code labels.</p>
1285
1286<h5>Syntax:</h5>
1287
1288<pre>
1289 label
1290</pre>
1291</div>
1292
1293
1294<!-- ======================================================================= -->
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1295<div class="doc_subsection"> <a name="t_derived">Derived Types</a> </div>
1296
1297<div class="doc_text">
1298
1299<p>The real power in LLVM comes from the derived types in the system.
1300This is what allows a programmer to represent arrays, functions,
1301pointers, and other useful types. Note that these derived types may be
1302recursive: For example, it is possible to have a two dimensional array.</p>
1303
1304</div>
1305
1306<!-- _______________________________________________________________________ -->
1307<div class="doc_subsubsection"> <a name="t_integer">Integer Type</a> </div>
1308
1309<div class="doc_text">
1310
1311<h5>Overview:</h5>
1312<p>The integer type is a very simple derived type that simply specifies an
1313arbitrary bit width for the integer type desired. Any bit width from 1 bit to
13142^23-1 (about 8 million) can be specified.</p>
1315
1316<h5>Syntax:</h5>
1317
1318<pre>
1319 iN
1320</pre>
1321
1322<p>The number of bits the integer will occupy is specified by the <tt>N</tt>
1323value.</p>
1324
1325<h5>Examples:</h5>
1326<table class="layout">
Chris Lattner251ab812007-12-18 06:18:21 +0000[diff] [blame]1327 <tbody>
1328 <tr>
1329 <td><tt>i1</tt></td>
1330 <td>a single-bit integer.</td>
1331 </tr><tr>
1332 <td><tt>i32</tt></td>
1333 <td>a 32-bit integer.</td>
1334 </tr><tr>
1335 <td><tt>i1942652</tt></td>
1336 <td>a really big integer of over 1 million bits.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1337 </tr>
Chris Lattner251ab812007-12-18 06:18:21 +0000[diff] [blame]1338 </tbody>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1339</table>
1340</div>
1341
1342<!-- _______________________________________________________________________ -->
1343<div class="doc_subsubsection"> <a name="t_array">Array Type</a> </div>
1344
1345<div class="doc_text">
1346
1347<h5>Overview:</h5>
1348
1349<p>The array type is a very simple derived type that arranges elements
1350sequentially in memory. The array type requires a size (number of
1351elements) and an underlying data type.</p>
1352
1353<h5>Syntax:</h5>
1354
1355<pre>
1356 [&lt;# elements&gt; x &lt;elementtype&gt;]
1357</pre>
1358
1359<p>The number of elements is a constant integer value; elementtype may
1360be any type with a size.</p>
1361
1362<h5>Examples:</h5>
1363<table class="layout">
1364 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1365 <td class="left"><tt>[40 x i32]</tt></td>
1366 <td class="left">Array of 40 32-bit integer values.</td>
1367 </tr>
1368 <tr class="layout">
1369 <td class="left"><tt>[41 x i32]</tt></td>
1370 <td class="left">Array of 41 32-bit integer values.</td>
1371 </tr>
1372 <tr class="layout">
1373 <td class="left"><tt>[4 x i8]</tt></td>
1374 <td class="left">Array of 4 8-bit integer values.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1375 </tr>
1376</table>
1377<p>Here are some examples of multidimensional arrays:</p>
1378<table class="layout">
1379 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1380 <td class="left"><tt>[3 x [4 x i32]]</tt></td>
1381 <td class="left">3x4 array of 32-bit integer values.</td>
1382 </tr>
1383 <tr class="layout">
1384 <td class="left"><tt>[12 x [10 x float]]</tt></td>
1385 <td class="left">12x10 array of single precision floating point values.</td>
1386 </tr>
1387 <tr class="layout">
1388 <td class="left"><tt>[2 x [3 x [4 x i16]]]</tt></td>
1389 <td class="left">2x3x4 array of 16-bit integer values.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1390 </tr>
1391</table>
1392
1393<p>Note that 'variable sized arrays' can be implemented in LLVM with a zero
1394length array. Normally, accesses past the end of an array are undefined in
1395LLVM (e.g. it is illegal to access the 5th element of a 3 element array).
1396As a special case, however, zero length arrays are recognized to be variable
1397length. This allows implementation of 'pascal style arrays' with the LLVM
1398type "{ i32, [0 x float]}", for example.</p>
1399
1400</div>
1401
1402<!-- _______________________________________________________________________ -->
1403<div class="doc_subsubsection"> <a name="t_function">Function Type</a> </div>
1404<div class="doc_text">
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1405
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1406<h5>Overview:</h5>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1407
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1408<p>The function type can be thought of as a function signature. It
Devang Pateld4ba41d2008-03-24 05:35:41 +0000[diff] [blame]1409consists of a return type and a list of formal parameter types. The
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1410return type of a function type is a scalar type, a void type, or a struct type.
Devang Pateld5404c02008-03-24 20:52:42 +0000[diff] [blame]1411If the return type is a struct type then all struct elements must be of first
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1412class types, and the struct must have at least one element.</p>
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]1413
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1414<h5>Syntax:</h5>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1415
1416<pre>
1417 &lt;returntype list&gt; (&lt;parameter list&gt;)
1418</pre>
1419
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1420<p>...where '<tt>&lt;parameter list&gt;</tt>' is a comma-separated list of type
1421specifiers. Optionally, the parameter list may include a type <tt>...</tt>,
1422which indicates that the function takes a variable number of arguments.
1423Variable argument functions can access their arguments with the <a
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]1424 href="#int_varargs">variable argument handling intrinsic</a> functions.
1425'<tt>&lt;returntype list&gt;</tt>' is a comma-separated list of
1426<a href="#t_firstclass">first class</a> type specifiers.</p>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1427
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1428<h5>Examples:</h5>
1429<table class="layout">
1430 <tr class="layout">
1431 <td class="left"><tt>i32 (i32)</tt></td>
1432 <td class="left">function taking an <tt>i32</tt>, returning an <tt>i32</tt>
1433 </td>
1434 </tr><tr class="layout">
Reid Spencerf234bed2007-07-19 23:13:04 +0000[diff] [blame]1435 <td class="left"><tt>float&nbsp;(i16&nbsp;signext,&nbsp;i32&nbsp;*)&nbsp;*
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1436 </tt></td>
1437 <td class="left"><a href="#t_pointer">Pointer</a> to a function that takes
1438 an <tt>i16</tt> that should be sign extended and a
1439 <a href="#t_pointer">pointer</a> to <tt>i32</tt>, returning
1440 <tt>float</tt>.
1441 </td>
1442 </tr><tr class="layout">
1443 <td class="left"><tt>i32 (i8*, ...)</tt></td>
1444 <td class="left">A vararg function that takes at least one
1445 <a href="#t_pointer">pointer</a> to <tt>i8 </tt> (char in C),
1446 which returns an integer. This is the signature for <tt>printf</tt> in
1447 LLVM.
1448 </td>
Devang Pateld4ba41d2008-03-24 05:35:41 +0000[diff] [blame]1449 </tr><tr class="layout">
1450 <td class="left"><tt>{i32, i32} (i32)</tt></td>
Misha Brukmanafc88b02008-11-27 06:41:20 +0000[diff] [blame]1451 <td class="left">A function taking an <tt>i32</tt>, returning two
1452 <tt>i32</tt> values as an aggregate of type <tt>{ i32, i32 }</tt>
Devang Pateld4ba41d2008-03-24 05:35:41 +0000[diff] [blame]1453 </td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1454 </tr>
1455</table>
1456
1457</div>
1458<!-- _______________________________________________________________________ -->
1459<div class="doc_subsubsection"> <a name="t_struct">Structure Type</a> </div>
1460<div class="doc_text">
1461<h5>Overview:</h5>
1462<p>The structure type is used to represent a collection of data members
1463together in memory. The packing of the field types is defined to match
1464the ABI of the underlying processor. The elements of a structure may
1465be any type that has a size.</p>
1466<p>Structures are accessed using '<tt><a href="#i_load">load</a></tt>
1467and '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a
1468field with the '<tt><a href="#i_getelementptr">getelementptr</a></tt>'
1469instruction.</p>
1470<h5>Syntax:</h5>
1471<pre> { &lt;type list&gt; }<br></pre>
1472<h5>Examples:</h5>
1473<table class="layout">
1474 <tr class="layout">
1475 <td class="left"><tt>{ i32, i32, i32 }</tt></td>
1476 <td class="left">A triple of three <tt>i32</tt> values</td>
1477 </tr><tr class="layout">
1478 <td class="left"><tt>{&nbsp;float,&nbsp;i32&nbsp;(i32)&nbsp;*&nbsp;}</tt></td>
1479 <td class="left">A pair, where the first element is a <tt>float</tt> and the
1480 second element is a <a href="#t_pointer">pointer</a> to a
1481 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
1482 an <tt>i32</tt>.</td>
1483 </tr>
1484</table>
1485</div>
1486
1487<!-- _______________________________________________________________________ -->
1488<div class="doc_subsubsection"> <a name="t_pstruct">Packed Structure Type</a>
1489</div>
1490<div class="doc_text">
1491<h5>Overview:</h5>
1492<p>The packed structure type is used to represent a collection of data members
1493together in memory. There is no padding between fields. Further, the alignment
1494of a packed structure is 1 byte. The elements of a packed structure may
1495be any type that has a size.</p>
1496<p>Structures are accessed using '<tt><a href="#i_load">load</a></tt>
1497and '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a
1498field with the '<tt><a href="#i_getelementptr">getelementptr</a></tt>'
1499instruction.</p>
1500<h5>Syntax:</h5>
1501<pre> &lt; { &lt;type list&gt; } &gt; <br></pre>
1502<h5>Examples:</h5>
1503<table class="layout">
1504 <tr class="layout">
1505 <td class="left"><tt>&lt; { i32, i32, i32 } &gt;</tt></td>
1506 <td class="left">A triple of three <tt>i32</tt> values</td>
1507 </tr><tr class="layout">
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]1508 <td class="left">
1509<tt>&lt;&nbsp;{&nbsp;float,&nbsp;i32&nbsp;(i32)*&nbsp;}&nbsp;&gt;</tt></td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1510 <td class="left">A pair, where the first element is a <tt>float</tt> and the
1511 second element is a <a href="#t_pointer">pointer</a> to a
1512 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
1513 an <tt>i32</tt>.</td>
1514 </tr>
1515</table>
1516</div>
1517
1518<!-- _______________________________________________________________________ -->
1519<div class="doc_subsubsection"> <a name="t_pointer">Pointer Type</a> </div>
1520<div class="doc_text">
1521<h5>Overview:</h5>
1522<p>As in many languages, the pointer type represents a pointer or
Christopher Lambdd0049d2007-12-11 09:31:00 +0000[diff] [blame]1523reference to another object, which must live in memory. Pointer types may have
1524an optional address space attribute defining the target-specific numbered
1525address space where the pointed-to object resides. The default address space is
1526zero.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1527<h5>Syntax:</h5>
1528<pre> &lt;type&gt; *<br></pre>
1529<h5>Examples:</h5>
1530<table class="layout">
1531 <tr class="layout">
Dan Gohman01852382009-01-04 23:44:43 +0000[diff] [blame]1532 <td class="left"><tt>[4 x i32]*</tt></td>
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1533 <td class="left">A <a href="#t_pointer">pointer</a> to <a
1534 href="#t_array">array</a> of four <tt>i32</tt> values.</td>
1535 </tr>
1536 <tr class="layout">
1537 <td class="left"><tt>i32 (i32 *) *</tt></td>
1538 <td class="left"> A <a href="#t_pointer">pointer</a> to a <a
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1539 href="#t_function">function</a> that takes an <tt>i32*</tt>, returning an
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1540 <tt>i32</tt>.</td>
1541 </tr>
1542 <tr class="layout">
1543 <td class="left"><tt>i32 addrspace(5)*</tt></td>
1544 <td class="left">A <a href="#t_pointer">pointer</a> to an <tt>i32</tt> value
1545 that resides in address space #5.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1546 </tr>
1547</table>
1548</div>
1549
1550<!-- _______________________________________________________________________ -->
1551<div class="doc_subsubsection"> <a name="t_vector">Vector Type</a> </div>
1552<div class="doc_text">
1553
1554<h5>Overview:</h5>
1555
1556<p>A vector type is a simple derived type that represents a vector
1557of elements. Vector types are used when multiple primitive data
1558are operated in parallel using a single instruction (SIMD).
1559A vector type requires a size (number of
1560elements) and an underlying primitive data type. Vectors must have a power
1561of two length (1, 2, 4, 8, 16 ...). Vector types are
1562considered <a href="#t_firstclass">first class</a>.</p>
1563
1564<h5>Syntax:</h5>
1565
1566<pre>
1567 &lt; &lt;# elements&gt; x &lt;elementtype&gt; &gt;
1568</pre>
1569
1570<p>The number of elements is a constant integer value; elementtype may
1571be any integer or floating point type.</p>
1572
1573<h5>Examples:</h5>
1574
1575<table class="layout">
1576 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1577 <td class="left"><tt>&lt;4 x i32&gt;</tt></td>
1578 <td class="left">Vector of 4 32-bit integer values.</td>
1579 </tr>
1580 <tr class="layout">
1581 <td class="left"><tt>&lt;8 x float&gt;</tt></td>
1582 <td class="left">Vector of 8 32-bit floating-point values.</td>
1583 </tr>
1584 <tr class="layout">
1585 <td class="left"><tt>&lt;2 x i64&gt;</tt></td>
1586 <td class="left">Vector of 2 64-bit integer values.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1587 </tr>
1588</table>
1589</div>
1590
1591<!-- _______________________________________________________________________ -->
1592<div class="doc_subsubsection"> <a name="t_opaque">Opaque Type</a> </div>
1593<div class="doc_text">
1594
1595<h5>Overview:</h5>
1596
1597<p>Opaque types are used to represent unknown types in the system. This
Gordon Henriksenda0706e2007-10-14 00:34:53 +0000[diff] [blame]1598corresponds (for example) to the C notion of a forward declared structure type.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1599In LLVM, opaque types can eventually be resolved to any type (not just a
1600structure type).</p>
1601
1602<h5>Syntax:</h5>
1603
1604<pre>
1605 opaque
1606</pre>
1607
1608<h5>Examples:</h5>
1609
1610<table class="layout">
1611 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1612 <td class="left"><tt>opaque</tt></td>
1613 <td class="left">An opaque type.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1614 </tr>
1615</table>
1616</div>
1617
1618
1619<!-- *********************************************************************** -->
1620<div class="doc_section"> <a name="constants">Constants</a> </div>
1621<!-- *********************************************************************** -->
1622
1623<div class="doc_text">
1624
1625<p>LLVM has several different basic types of constants. This section describes
1626them all and their syntax.</p>
1627
1628</div>
1629
1630<!-- ======================================================================= -->
1631<div class="doc_subsection"><a name="simpleconstants">Simple Constants</a></div>
1632
1633<div class="doc_text">
1634
1635<dl>
1636 <dt><b>Boolean constants</b></dt>
1637
1638 <dd>The two strings '<tt>true</tt>' and '<tt>false</tt>' are both valid
1639 constants of the <tt><a href="#t_primitive">i1</a></tt> type.
1640 </dd>
1641
1642 <dt><b>Integer constants</b></dt>
1643
1644 <dd>Standard integers (such as '4') are constants of the <a
1645 href="#t_integer">integer</a> type. Negative numbers may be used with
1646 integer types.
1647 </dd>
1648
1649 <dt><b>Floating point constants</b></dt>
1650
1651 <dd>Floating point constants use standard decimal notation (e.g. 123.421),
1652 exponential notation (e.g. 1.23421e+2), or a more precise hexadecimal
Chris Lattnerdc15b1d2008-04-01 18:45:27 +0000[diff] [blame]1653 notation (see below). The assembler requires the exact decimal value of
1654 a floating-point constant. For example, the assembler accepts 1.25 but
1655 rejects 1.3 because 1.3 is a repeating decimal in binary. Floating point
1656 constants must have a <a href="#t_floating">floating point</a> type. </dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1657
1658 <dt><b>Null pointer constants</b></dt>
1659
1660 <dd>The identifier '<tt>null</tt>' is recognized as a null pointer constant
1661 and must be of <a href="#t_pointer">pointer type</a>.</dd>
1662
1663</dl>
1664
1665<p>The one non-intuitive notation for constants is the optional hexadecimal form
1666of floating point constants. For example, the form '<tt>double
16670x432ff973cafa8000</tt>' is equivalent to (but harder to read than) '<tt>double
16684.5e+15</tt>'. The only time hexadecimal floating point constants are required
1669(and the only time that they are generated by the disassembler) is when a
1670floating point constant must be emitted but it cannot be represented as a
1671decimal floating point number. For example, NaN's, infinities, and other
1672special values are represented in their IEEE hexadecimal format so that
1673assembly and disassembly do not cause any bits to change in the constants.</p>
1674
1675</div>
1676
1677<!-- ======================================================================= -->
1678<div class="doc_subsection"><a name="aggregateconstants">Aggregate Constants</a>
1679</div>
1680
1681<div class="doc_text">
1682<p>Aggregate constants arise from aggregation of simple constants
1683and smaller aggregate constants.</p>
1684
1685<dl>
1686 <dt><b>Structure constants</b></dt>
1687
1688 <dd>Structure constants are represented with notation similar to structure
1689 type definitions (a comma separated list of elements, surrounded by braces
Chris Lattner6c8de962007-12-25 20:34:52 +0000[diff] [blame]1690 (<tt>{}</tt>)). For example: "<tt>{ i32 4, float 17.0, i32* @G }</tt>",
1691 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]1692 must have <a href="#t_struct">structure type</a>, and the number and
1693 types of elements must match those specified by the type.
1694 </dd>
1695
1696 <dt><b>Array constants</b></dt>
1697
1698 <dd>Array constants are represented with notation similar to array type
1699 definitions (a comma separated list of elements, surrounded by square brackets
1700 (<tt>[]</tt>)). For example: "<tt>[ i32 42, i32 11, i32 74 ]</tt>". Array
1701 constants must have <a href="#t_array">array type</a>, and the number and
1702 types of elements must match those specified by the type.
1703 </dd>
1704
1705 <dt><b>Vector constants</b></dt>
1706
1707 <dd>Vector constants are represented with notation similar to vector type
1708 definitions (a comma separated list of elements, surrounded by
1709 less-than/greater-than's (<tt>&lt;&gt;</tt>)). For example: "<tt>&lt; i32 42,
1710 i32 11, i32 74, i32 100 &gt;</tt>". Vector constants must have <a
1711 href="#t_vector">vector type</a>, and the number and types of elements must
1712 match those specified by the type.
1713 </dd>
1714
1715 <dt><b>Zero initialization</b></dt>
1716
1717 <dd>The string '<tt>zeroinitializer</tt>' can be used to zero initialize a
1718 value to zero of <em>any</em> type, including scalar and aggregate types.
1719 This is often used to avoid having to print large zero initializers (e.g. for
1720 large arrays) and is always exactly equivalent to using explicit zero
1721 initializers.
1722 </dd>
1723</dl>
1724
1725</div>
1726
1727<!-- ======================================================================= -->
1728<div class="doc_subsection">
1729 <a name="globalconstants">Global Variable and Function Addresses</a>
1730</div>
1731
1732<div class="doc_text">
1733
1734<p>The addresses of <a href="#globalvars">global variables</a> and <a
1735href="#functionstructure">functions</a> are always implicitly valid (link-time)
1736constants. These constants are explicitly referenced when the <a
1737href="#identifiers">identifier for the global</a> is used and always have <a
1738href="#t_pointer">pointer</a> type. For example, the following is a legal LLVM
1739file:</p>
1740
1741<div class="doc_code">
1742<pre>
1743@X = global i32 17
1744@Y = global i32 42
1745@Z = global [2 x i32*] [ i32* @X, i32* @Y ]
1746</pre>
1747</div>
1748
1749</div>
1750
1751<!-- ======================================================================= -->
1752<div class="doc_subsection"><a name="undefvalues">Undefined Values</a></div>
1753<div class="doc_text">
1754 <p>The string '<tt>undef</tt>' is recognized as a type-less constant that has
1755 no specific value. Undefined values may be of any type and be used anywhere
1756 a constant is permitted.</p>
1757
1758 <p>Undefined values indicate to the compiler that the program is well defined
1759 no matter what value is used, giving the compiler more freedom to optimize.
1760 </p>
1761</div>
1762
1763<!-- ======================================================================= -->
1764<div class="doc_subsection"><a name="constantexprs">Constant Expressions</a>
1765</div>
1766
1767<div class="doc_text">
1768
1769<p>Constant expressions are used to allow expressions involving other constants
1770to be used as constants. Constant expressions may be of any <a
1771href="#t_firstclass">first class</a> type and may involve any LLVM operation
1772that does not have side effects (e.g. load and call are not supported). The
1773following is the syntax for constant expressions:</p>
1774
1775<dl>
1776 <dt><b><tt>trunc ( CST to TYPE )</tt></b></dt>
1777 <dd>Truncate a constant to another type. The bit size of CST must be larger
1778 than the bit size of TYPE. Both types must be integers.</dd>
1779
1780 <dt><b><tt>zext ( CST to TYPE )</tt></b></dt>
1781 <dd>Zero extend a constant to another type. The bit size of CST must be
1782 smaller or equal to the bit size of TYPE. Both types must be integers.</dd>
1783
1784 <dt><b><tt>sext ( CST to TYPE )</tt></b></dt>
1785 <dd>Sign extend a constant to another type. The bit size of CST must be
1786 smaller or equal to the bit size of TYPE. Both types must be integers.</dd>
1787
1788 <dt><b><tt>fptrunc ( CST to TYPE )</tt></b></dt>
1789 <dd>Truncate a floating point constant to another floating point type. The
1790 size of CST must be larger than the size of TYPE. Both types must be
1791 floating point.</dd>
1792
1793 <dt><b><tt>fpext ( CST to TYPE )</tt></b></dt>
1794 <dd>Floating point extend a constant to another type. The size of CST must be
1795 smaller or equal to the size of TYPE. Both types must be floating point.</dd>
1796
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]1797 <dt><b><tt>fptoui ( CST to TYPE )</tt></b></dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1798 <dd>Convert a floating point constant to the corresponding unsigned integer
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]1799 constant. TYPE must be a scalar or vector integer type. CST must be of scalar
1800 or vector floating point type. Both CST and TYPE must be scalars, or vectors
1801 of the same number of elements. If the value won't fit in the integer type,
1802 the results are undefined.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1803
1804 <dt><b><tt>fptosi ( CST to TYPE )</tt></b></dt>
1805 <dd>Convert a floating point constant to the corresponding signed integer
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]1806 constant. TYPE must be a scalar or vector integer type. CST must be of scalar
1807 or vector floating point type. Both CST and TYPE must be scalars, or vectors
1808 of the same number of elements. If the value won't fit in the integer type,
1809 the results are undefined.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1810
1811 <dt><b><tt>uitofp ( CST to TYPE )</tt></b></dt>
1812 <dd>Convert an unsigned integer constant to the corresponding floating point
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]1813 constant. TYPE must be a scalar or vector floating point type. CST must be of
1814 scalar or vector integer type. Both CST and TYPE must be scalars, or vectors
1815 of the same number of elements. If the value won't fit in the floating point
1816 type, the results are undefined.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1817
1818 <dt><b><tt>sitofp ( CST to TYPE )</tt></b></dt>
1819 <dd>Convert a signed integer constant to the corresponding floating point
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]1820 constant. TYPE must be a scalar or vector floating point type. CST must be of
1821 scalar or vector integer type. Both CST and TYPE must be scalars, or vectors
1822 of the same number of elements. If the value won't fit in the floating point
1823 type, the results are undefined.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1824
1825 <dt><b><tt>ptrtoint ( CST to TYPE )</tt></b></dt>
1826 <dd>Convert a pointer typed constant to the corresponding integer constant
1827 TYPE must be an integer type. CST must be of pointer type. The CST value is
1828 zero extended, truncated, or unchanged to make it fit in TYPE.</dd>
1829
1830 <dt><b><tt>inttoptr ( CST to TYPE )</tt></b></dt>
1831 <dd>Convert a integer constant to a pointer constant. TYPE must be a
1832 pointer type. CST must be of integer type. The CST value is zero extended,
1833 truncated, or unchanged to make it fit in a pointer size. This one is
1834 <i>really</i> dangerous!</dd>
1835
1836 <dt><b><tt>bitcast ( CST to TYPE )</tt></b></dt>
1837 <dd>Convert a constant, CST, to another TYPE. The size of CST and TYPE must be
1838 identical (same number of bits). The conversion is done as if the CST value
1839 was stored to memory and read back as TYPE. In other words, no bits change
1840 with this operator, just the type. This can be used for conversion of
1841 vector types to any other type, as long as they have the same bit width. For
Dan Gohman7305fa02008-09-08 16:45:59 +0000[diff] [blame]1842 pointers it is only valid to cast to another pointer type. It is not valid
1843 to bitcast to or from an aggregate type.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1844 </dd>
1845
1846 <dt><b><tt>getelementptr ( CSTPTR, IDX0, IDX1, ... )</tt></b></dt>
1847
1848 <dd>Perform the <a href="#i_getelementptr">getelementptr operation</a> on
1849 constants. As with the <a href="#i_getelementptr">getelementptr</a>
1850 instruction, the index list may have zero or more indexes, which are required
1851 to make sense for the type of "CSTPTR".</dd>
1852
1853 <dt><b><tt>select ( COND, VAL1, VAL2 )</tt></b></dt>
1854
1855 <dd>Perform the <a href="#i_select">select operation</a> on
1856 constants.</dd>
1857
1858 <dt><b><tt>icmp COND ( VAL1, VAL2 )</tt></b></dt>
1859 <dd>Performs the <a href="#i_icmp">icmp operation</a> on constants.</dd>
1860
1861 <dt><b><tt>fcmp COND ( VAL1, VAL2 )</tt></b></dt>
1862 <dd>Performs the <a href="#i_fcmp">fcmp operation</a> on constants.</dd>
1863
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]1864 <dt><b><tt>vicmp COND ( VAL1, VAL2 )</tt></b></dt>
1865 <dd>Performs the <a href="#i_vicmp">vicmp operation</a> on constants.</dd>
1866
1867 <dt><b><tt>vfcmp COND ( VAL1, VAL2 )</tt></b></dt>
1868 <dd>Performs the <a href="#i_vfcmp">vfcmp operation</a> on constants.</dd>
1869
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1870 <dt><b><tt>extractelement ( VAL, IDX )</tt></b></dt>
1871
1872 <dd>Perform the <a href="#i_extractelement">extractelement
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]1873 operation</a> on constants.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1874
1875 <dt><b><tt>insertelement ( VAL, ELT, IDX )</tt></b></dt>
1876
1877 <dd>Perform the <a href="#i_insertelement">insertelement
1878 operation</a> on constants.</dd>
1879
1880
1881 <dt><b><tt>shufflevector ( VEC1, VEC2, IDXMASK )</tt></b></dt>
1882
1883 <dd>Perform the <a href="#i_shufflevector">shufflevector
1884 operation</a> on constants.</dd>
1885
1886 <dt><b><tt>OPCODE ( LHS, RHS )</tt></b></dt>
1887
1888 <dd>Perform the specified operation of the LHS and RHS constants. OPCODE may
1889 be any of the <a href="#binaryops">binary</a> or <a href="#bitwiseops">bitwise
1890 binary</a> operations. The constraints on operands are the same as those for
1891 the corresponding instruction (e.g. no bitwise operations on floating point
1892 values are allowed).</dd>
1893</dl>
1894</div>
1895
1896<!-- *********************************************************************** -->
1897<div class="doc_section"> <a name="othervalues">Other Values</a> </div>
1898<!-- *********************************************************************** -->
1899
1900<!-- ======================================================================= -->
1901<div class="doc_subsection">
1902<a name="inlineasm">Inline Assembler Expressions</a>
1903</div>
1904
1905<div class="doc_text">
1906
1907<p>
1908LLVM supports inline assembler expressions (as opposed to <a href="#moduleasm">
1909Module-Level Inline Assembly</a>) through the use of a special value. This
1910value represents the inline assembler as a string (containing the instructions
1911to emit), a list of operand constraints (stored as a string), and a flag that
1912indicates whether or not the inline asm expression has side effects. An example
1913inline assembler expression is:
1914</p>
1915
1916<div class="doc_code">
1917<pre>
1918i32 (i32) asm "bswap $0", "=r,r"
1919</pre>
1920</div>
1921
1922<p>
1923Inline assembler expressions may <b>only</b> be used as the callee operand of
1924a <a href="#i_call"><tt>call</tt> instruction</a>. Thus, typically we have:
1925</p>
1926
1927<div class="doc_code">
1928<pre>
1929%X = call i32 asm "<a href="#int_bswap">bswap</a> $0", "=r,r"(i32 %Y)
1930</pre>
1931</div>
1932
1933<p>
1934Inline asms with side effects not visible in the constraint list must be marked
1935as having side effects. This is done through the use of the
1936'<tt>sideeffect</tt>' keyword, like so:
1937</p>
1938
1939<div class="doc_code">
1940<pre>
1941call void asm sideeffect "eieio", ""()
1942</pre>
1943</div>
1944
1945<p>TODO: The format of the asm and constraints string still need to be
1946documented here. Constraints on what can be done (e.g. duplication, moving, etc
Chris Lattner1cdd64e2008-10-04 18:36:02 +0000[diff] [blame]1947need to be documented). This is probably best done by reference to another
1948document that covers inline asm from a holistic perspective.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1949</p>
1950
1951</div>
1952
1953<!-- *********************************************************************** -->
1954<div class="doc_section"> <a name="instref">Instruction Reference</a> </div>
1955<!-- *********************************************************************** -->
1956
1957<div class="doc_text">
1958
1959<p>The LLVM instruction set consists of several different
1960classifications of instructions: <a href="#terminators">terminator
1961instructions</a>, <a href="#binaryops">binary instructions</a>,
1962<a href="#bitwiseops">bitwise binary instructions</a>, <a
1963 href="#memoryops">memory instructions</a>, and <a href="#otherops">other
1964instructions</a>.</p>
1965
1966</div>
1967
1968<!-- ======================================================================= -->
1969<div class="doc_subsection"> <a name="terminators">Terminator
1970Instructions</a> </div>
1971
1972<div class="doc_text">
1973
1974<p>As mentioned <a href="#functionstructure">previously</a>, every
1975basic block in a program ends with a "Terminator" instruction, which
1976indicates which block should be executed after the current block is
1977finished. These terminator instructions typically yield a '<tt>void</tt>'
1978value: they produce control flow, not values (the one exception being
1979the '<a href="#i_invoke"><tt>invoke</tt></a>' instruction).</p>
1980<p>There are six different terminator instructions: the '<a
1981 href="#i_ret"><tt>ret</tt></a>' instruction, the '<a href="#i_br"><tt>br</tt></a>'
1982instruction, the '<a href="#i_switch"><tt>switch</tt></a>' instruction,
1983the '<a href="#i_invoke"><tt>invoke</tt></a>' instruction, the '<a
1984 href="#i_unwind"><tt>unwind</tt></a>' instruction, and the '<a
1985 href="#i_unreachable"><tt>unreachable</tt></a>' instruction.</p>
1986
1987</div>
1988
1989<!-- _______________________________________________________________________ -->
1990<div class="doc_subsubsection"> <a name="i_ret">'<tt>ret</tt>'
1991Instruction</a> </div>
1992<div class="doc_text">
1993<h5>Syntax:</h5>
Dan Gohman3e700032008-10-04 19:00:07 +0000[diff] [blame]1994<pre>
1995 ret &lt;type&gt; &lt;value&gt; <i>; Return a value from a non-void function</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1996 ret void <i>; Return from void function</i>
1997</pre>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1998
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1999<h5>Overview:</h5>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]2000
Dan Gohman3e700032008-10-04 19:00:07 +0000[diff] [blame]2001<p>The '<tt>ret</tt>' instruction is used to return control flow (and
2002optionally a value) from a function back to the caller.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2003<p>There are two forms of the '<tt>ret</tt>' instruction: one that
Dan Gohman3e700032008-10-04 19:00:07 +0000[diff] [blame]2004returns a value and then causes control flow, and one that just causes
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2005control flow to occur.</p>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]2006
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2007<h5>Arguments:</h5>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]2008
Dan Gohman3e700032008-10-04 19:00:07 +0000[diff] [blame]2009<p>The '<tt>ret</tt>' instruction optionally accepts a single argument,
2010the return value. The type of the return value must be a
2011'<a href="#t_firstclass">first class</a>' type.</p>
2012
2013<p>A function is not <a href="#wellformed">well formed</a> if
2014it it has a non-void return type and contains a '<tt>ret</tt>'
2015instruction with no return value or a return value with a type that
2016does not match its type, or if it has a void return type and contains
2017a '<tt>ret</tt>' instruction with a return value.</p>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]2018
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2019<h5>Semantics:</h5>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]2020
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2021<p>When the '<tt>ret</tt>' instruction is executed, control flow
2022returns back to the calling function's context. If the caller is a "<a
2023 href="#i_call"><tt>call</tt></a>" instruction, execution continues at
2024the instruction after the call. If the caller was an "<a
2025 href="#i_invoke"><tt>invoke</tt></a>" instruction, execution continues
2026at the beginning of the "normal" destination block. If the instruction
2027returns a value, that value shall set the call or invoke instruction's
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]2028return value.</p>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]2029
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2030<h5>Example:</h5>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]2031
2032<pre>
2033 ret i32 5 <i>; Return an integer value of 5</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2034 ret void <i>; Return from a void function</i>
Dan Gohman3e700032008-10-04 19:00:07 +0000[diff] [blame]2035 ret { i32, i8 } { i32 4, i8 2 } <i>; Return an aggregate of values 4 and 2</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2036</pre>
Dan Gohman60967192009-01-12 23:12:39 +0000[diff] [blame]2037
2038<p>Note that the code generator does not yet fully support larger
2039 aggregate return values.</p>
2040
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2041</div>
2042<!-- _______________________________________________________________________ -->
2043<div class="doc_subsubsection"> <a name="i_br">'<tt>br</tt>' Instruction</a> </div>
2044<div class="doc_text">
2045<h5>Syntax:</h5>
2046<pre> br i1 &lt;cond&gt;, label &lt;iftrue&gt;, label &lt;iffalse&gt;<br> br label &lt;dest&gt; <i>; Unconditional branch</i>
2047</pre>
2048<h5>Overview:</h5>
2049<p>The '<tt>br</tt>' instruction is used to cause control flow to
2050transfer to a different basic block in the current function. There are
2051two forms of this instruction, corresponding to a conditional branch
2052and an unconditional branch.</p>
2053<h5>Arguments:</h5>
2054<p>The conditional branch form of the '<tt>br</tt>' instruction takes a
2055single '<tt>i1</tt>' value and two '<tt>label</tt>' values. The
2056unconditional form of the '<tt>br</tt>' instruction takes a single
2057'<tt>label</tt>' value as a target.</p>
2058<h5>Semantics:</h5>
2059<p>Upon execution of a conditional '<tt>br</tt>' instruction, the '<tt>i1</tt>'
2060argument is evaluated. If the value is <tt>true</tt>, control flows
2061to the '<tt>iftrue</tt>' <tt>label</tt> argument. If "cond" is <tt>false</tt>,
2062control flows to the '<tt>iffalse</tt>' <tt>label</tt> argument.</p>
2063<h5>Example:</h5>
2064<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
2065 href="#i_ret">ret</a> i32 1<br>IfUnequal:<br> <a href="#i_ret">ret</a> i32 0<br></pre>
2066</div>
2067<!-- _______________________________________________________________________ -->
2068<div class="doc_subsubsection">
2069 <a name="i_switch">'<tt>switch</tt>' Instruction</a>
2070</div>
2071
2072<div class="doc_text">
2073<h5>Syntax:</h5>
2074
2075<pre>
2076 switch &lt;intty&gt; &lt;value&gt;, label &lt;defaultdest&gt; [ &lt;intty&gt; &lt;val&gt;, label &lt;dest&gt; ... ]
2077</pre>
2078
2079<h5>Overview:</h5>
2080
2081<p>The '<tt>switch</tt>' instruction is used to transfer control flow to one of
2082several different places. It is a generalization of the '<tt>br</tt>'
2083instruction, allowing a branch to occur to one of many possible
2084destinations.</p>
2085
2086
2087<h5>Arguments:</h5>
2088
2089<p>The '<tt>switch</tt>' instruction uses three parameters: an integer
2090comparison value '<tt>value</tt>', a default '<tt>label</tt>' destination, and
2091an array of pairs of comparison value constants and '<tt>label</tt>'s. The
2092table is not allowed to contain duplicate constant entries.</p>
2093
2094<h5>Semantics:</h5>
2095
2096<p>The <tt>switch</tt> instruction specifies a table of values and
2097destinations. When the '<tt>switch</tt>' instruction is executed, this
2098table is searched for the given value. If the value is found, control flow is
2099transfered to the corresponding destination; otherwise, control flow is
2100transfered to the default destination.</p>
2101
2102<h5>Implementation:</h5>
2103
2104<p>Depending on properties of the target machine and the particular
2105<tt>switch</tt> instruction, this instruction may be code generated in different
2106ways. For example, it could be generated as a series of chained conditional
2107branches or with a lookup table.</p>
2108
2109<h5>Example:</h5>
2110
2111<pre>
2112 <i>; Emulate a conditional br instruction</i>
2113 %Val = <a href="#i_zext">zext</a> i1 %value to i32
Dan Gohman01852382009-01-04 23:44:43 +0000[diff] [blame]2114 switch i32 %Val, label %truedest [ i32 0, label %falsedest ]
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2115
2116 <i>; Emulate an unconditional br instruction</i>
2117 switch i32 0, label %dest [ ]
2118
2119 <i>; Implement a jump table:</i>
Dan Gohman01852382009-01-04 23:44:43 +0000[diff] [blame]2120 switch i32 %val, label %otherwise [ i32 0, label %onzero
2121 i32 1, label %onone
2122 i32 2, label %ontwo ]
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2123</pre>
2124</div>
2125
2126<!-- _______________________________________________________________________ -->
2127<div class="doc_subsubsection">
2128 <a name="i_invoke">'<tt>invoke</tt>' Instruction</a>
2129</div>
2130
2131<div class="doc_text">
2132
2133<h5>Syntax:</h5>
2134
2135<pre>
Devang Pateld0bfcc72008-10-07 17:48:33 +0000[diff] [blame]2136 &lt;result&gt; = invoke [<a href="#callingconv">cconv</a>] [<a href="#paramattrs">ret attrs</a>] &lt;ptr to function ty&gt; &lt;function ptr val&gt;(&lt;function args&gt;) [<a href="#fnattrs">fn attrs</a>]
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2137 to label &lt;normal label&gt; unwind label &lt;exception label&gt;
2138</pre>
2139
2140<h5>Overview:</h5>
2141
2142<p>The '<tt>invoke</tt>' instruction causes control to transfer to a specified
2143function, with the possibility of control flow transfer to either the
2144'<tt>normal</tt>' label or the
2145'<tt>exception</tt>' label. If the callee function returns with the
2146"<tt><a href="#i_ret">ret</a></tt>" instruction, control flow will return to the
2147"normal" label. If the callee (or any indirect callees) returns with the "<a
2148href="#i_unwind"><tt>unwind</tt></a>" instruction, control is interrupted and
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]2149continued at the dynamically nearest "exception" label.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2150
2151<h5>Arguments:</h5>
2152
2153<p>This instruction requires several arguments:</p>
2154
2155<ol>
2156 <li>
2157 The optional "cconv" marker indicates which <a href="#callingconv">calling
2158 convention</a> the call should use. If none is specified, the call defaults
2159 to using C calling conventions.
2160 </li>
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]2161
2162 <li>The optional <a href="#paramattrs">Parameter Attributes</a> list for
2163 return values. Only '<tt>zeroext</tt>', '<tt>signext</tt>',
2164 and '<tt>inreg</tt>' attributes are valid here.</li>
2165
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2166 <li>'<tt>ptr to function ty</tt>': shall be the signature of the pointer to
2167 function value being invoked. In most cases, this is a direct function
2168 invocation, but indirect <tt>invoke</tt>s are just as possible, branching off
2169 an arbitrary pointer to function value.
2170 </li>
2171
2172 <li>'<tt>function ptr val</tt>': An LLVM value containing a pointer to a
2173 function to be invoked. </li>
2174
2175 <li>'<tt>function args</tt>': argument list whose types match the function
2176 signature argument types. If the function signature indicates the function
2177 accepts a variable number of arguments, the extra arguments can be
2178 specified. </li>
2179
2180 <li>'<tt>normal label</tt>': the label reached when the called function
2181 executes a '<tt><a href="#i_ret">ret</a></tt>' instruction. </li>
2182
2183 <li>'<tt>exception label</tt>': the label reached when a callee returns with
2184 the <a href="#i_unwind"><tt>unwind</tt></a> instruction. </li>
2185
Devang Pateld0bfcc72008-10-07 17:48:33 +0000[diff] [blame]2186 <li>The optional <a href="#fnattrs">function attributes</a> list. Only
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]2187 '<tt>noreturn</tt>', '<tt>nounwind</tt>', '<tt>readonly</tt>' and
2188 '<tt>readnone</tt>' attributes are valid here.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2189</ol>
2190
2191<h5>Semantics:</h5>
2192
2193<p>This instruction is designed to operate as a standard '<tt><a
2194href="#i_call">call</a></tt>' instruction in most regards. The primary
2195difference is that it establishes an association with a label, which is used by
2196the runtime library to unwind the stack.</p>
2197
2198<p>This instruction is used in languages with destructors to ensure that proper
2199cleanup is performed in the case of either a <tt>longjmp</tt> or a thrown
2200exception. Additionally, this is important for implementation of
2201'<tt>catch</tt>' clauses in high-level languages that support them.</p>
2202
2203<h5>Example:</h5>
2204<pre>
Nick Lewyckya1c11a12008-03-16 07:18:12 +0000[diff] [blame]2205 %retval = invoke i32 @Test(i32 15) to label %Continue
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2206 unwind label %TestCleanup <i>; {i32}:retval set</i>
Nick Lewyckya1c11a12008-03-16 07:18:12 +0000[diff] [blame]2207 %retval = invoke <a href="#callingconv">coldcc</a> i32 %Testfnptr(i32 15) to label %Continue
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2208 unwind label %TestCleanup <i>; {i32}:retval set</i>
2209</pre>
2210</div>
2211
2212
2213<!-- _______________________________________________________________________ -->
2214
2215<div class="doc_subsubsection"> <a name="i_unwind">'<tt>unwind</tt>'
2216Instruction</a> </div>
2217
2218<div class="doc_text">
2219
2220<h5>Syntax:</h5>
2221<pre>
2222 unwind
2223</pre>
2224
2225<h5>Overview:</h5>
2226
2227<p>The '<tt>unwind</tt>' instruction unwinds the stack, continuing control flow
2228at the first callee in the dynamic call stack which used an <a
2229href="#i_invoke"><tt>invoke</tt></a> instruction to perform the call. This is
2230primarily used to implement exception handling.</p>
2231
2232<h5>Semantics:</h5>
2233
Chris Lattner8b094fc2008-04-19 21:01:16 +0000[diff] [blame]2234<p>The '<tt>unwind</tt>' instruction causes execution of the current function to
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2235immediately halt. The dynamic call stack is then searched for the first <a
2236href="#i_invoke"><tt>invoke</tt></a> instruction on the call stack. Once found,
2237execution continues at the "exceptional" destination block specified by the
2238<tt>invoke</tt> instruction. If there is no <tt>invoke</tt> instruction in the
2239dynamic call chain, undefined behavior results.</p>
2240</div>
2241
2242<!-- _______________________________________________________________________ -->
2243
2244<div class="doc_subsubsection"> <a name="i_unreachable">'<tt>unreachable</tt>'
2245Instruction</a> </div>
2246
2247<div class="doc_text">
2248
2249<h5>Syntax:</h5>
2250<pre>
2251 unreachable
2252</pre>
2253
2254<h5>Overview:</h5>
2255
2256<p>The '<tt>unreachable</tt>' instruction has no defined semantics. This
2257instruction is used to inform the optimizer that a particular portion of the
2258code is not reachable. This can be used to indicate that the code after a
2259no-return function cannot be reached, and other facts.</p>
2260
2261<h5>Semantics:</h5>
2262
2263<p>The '<tt>unreachable</tt>' instruction has no defined semantics.</p>
2264</div>
2265
2266
2267
2268<!-- ======================================================================= -->
2269<div class="doc_subsection"> <a name="binaryops">Binary Operations</a> </div>
2270<div class="doc_text">
2271<p>Binary operators are used to do most of the computation in a
Chris Lattnerab596d92008-04-01 18:47:32 +0000[diff] [blame]2272program. They require two operands of the same type, execute an operation on them, and
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2273produce a single value. The operands might represent
2274multiple data, as is the case with the <a href="#t_vector">vector</a> data type.
Chris Lattnerab596d92008-04-01 18:47:32 +0000[diff] [blame]2275The result value has the same type as its operands.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2276<p>There are several different binary operators:</p>
2277</div>
2278<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2279<div class="doc_subsubsection">
2280 <a name="i_add">'<tt>add</tt>' Instruction</a>
2281</div>
2282
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2283<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2284
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2285<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2286
2287<pre>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2288 &lt;result&gt; = add &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2289</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2290
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2291<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2292
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2293<p>The '<tt>add</tt>' instruction returns the sum of its two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2294
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2295<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2296
2297<p>The two arguments to the '<tt>add</tt>' instruction must be <a
2298 href="#t_integer">integer</a>, <a href="#t_floating">floating point</a>, or
2299 <a href="#t_vector">vector</a> values. Both arguments must have identical
2300 types.</p>
2301
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2302<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2303
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2304<p>The value produced is the integer or floating point sum of the two
2305operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2306
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2307<p>If an integer sum has unsigned overflow, the result returned is the
2308mathematical result modulo 2<sup>n</sup>, where n is the bit width of
2309the result.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2310
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2311<p>Because LLVM integers use a two's complement representation, this
2312instruction is appropriate for both signed and unsigned integers.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2313
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2314<h5>Example:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2315
2316<pre>
2317 &lt;result&gt; = add i32 4, %var <i>; yields {i32}:result = 4 + %var</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2318</pre>
2319</div>
2320<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2321<div class="doc_subsubsection">
2322 <a name="i_sub">'<tt>sub</tt>' Instruction</a>
2323</div>
2324
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2325<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2326
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2327<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2328
2329<pre>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2330 &lt;result&gt; = sub &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2331</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2332
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2333<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2334
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2335<p>The '<tt>sub</tt>' instruction returns the difference of its two
2336operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2337
2338<p>Note that the '<tt>sub</tt>' instruction is used to represent the
2339'<tt>neg</tt>' instruction present in most other intermediate
2340representations.</p>
2341
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2342<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2343
2344<p>The two arguments to the '<tt>sub</tt>' instruction must be <a
2345 href="#t_integer">integer</a>, <a href="#t_floating">floating point</a>,
2346 or <a href="#t_vector">vector</a> values. Both arguments must have identical
2347 types.</p>
2348
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2349<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2350
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2351<p>The value produced is the integer or floating point difference of
2352the two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2353
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2354<p>If an integer difference has unsigned overflow, the result returned is the
2355mathematical result modulo 2<sup>n</sup>, where n is the bit width of
2356the result.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2357
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2358<p>Because LLVM integers use a two's complement representation, this
2359instruction is appropriate for both signed and unsigned integers.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2360
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2361<h5>Example:</h5>
2362<pre>
2363 &lt;result&gt; = sub i32 4, %var <i>; yields {i32}:result = 4 - %var</i>
2364 &lt;result&gt; = sub i32 0, %val <i>; yields {i32}:result = -%var</i>
2365</pre>
2366</div>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2367
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2368<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2369<div class="doc_subsubsection">
2370 <a name="i_mul">'<tt>mul</tt>' Instruction</a>
2371</div>
2372
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2373<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2374
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2375<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2376<pre> &lt;result&gt; = mul &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2377</pre>
2378<h5>Overview:</h5>
2379<p>The '<tt>mul</tt>' instruction returns the product of its two
2380operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2381
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2382<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2383
2384<p>The two arguments to the '<tt>mul</tt>' instruction must be <a
2385href="#t_integer">integer</a>, <a href="#t_floating">floating point</a>,
2386or <a href="#t_vector">vector</a> values. Both arguments must have identical
2387types.</p>
2388
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2389<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2390
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2391<p>The value produced is the integer or floating point product of the
2392two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2393
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2394<p>If the result of an integer multiplication has unsigned overflow,
2395the result returned is the mathematical result modulo
23962<sup>n</sup>, where n is the bit width of the result.</p>
2397<p>Because LLVM integers use a two's complement representation, and the
2398result is the same width as the operands, this instruction returns the
2399correct result for both signed and unsigned integers. If a full product
2400(e.g. <tt>i32</tt>x<tt>i32</tt>-><tt>i64</tt>) is needed, the operands
2401should be sign-extended or zero-extended as appropriate to the
2402width of the full product.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2403<h5>Example:</h5>
2404<pre> &lt;result&gt; = mul i32 4, %var <i>; yields {i32}:result = 4 * %var</i>
2405</pre>
2406</div>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2407
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2408<!-- _______________________________________________________________________ -->
2409<div class="doc_subsubsection"> <a name="i_udiv">'<tt>udiv</tt>' Instruction
2410</a></div>
2411<div class="doc_text">
2412<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2413<pre> &lt;result&gt; = udiv &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2414</pre>
2415<h5>Overview:</h5>
2416<p>The '<tt>udiv</tt>' instruction returns the quotient of its two
2417operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2418
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2419<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2420
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2421<p>The two arguments to the '<tt>udiv</tt>' instruction must be
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2422<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2423values. Both arguments must have identical types.</p>
2424
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2425<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2426
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2427<p>The value produced is the unsigned integer quotient of the two operands.</p>
2428<p>Note that unsigned integer division and signed integer division are distinct
2429operations; for signed integer division, use '<tt>sdiv</tt>'.</p>
2430<p>Division by zero leads to undefined behavior.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2431<h5>Example:</h5>
2432<pre> &lt;result&gt; = udiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
2433</pre>
2434</div>
2435<!-- _______________________________________________________________________ -->
2436<div class="doc_subsubsection"> <a name="i_sdiv">'<tt>sdiv</tt>' Instruction
2437</a> </div>
2438<div class="doc_text">
2439<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2440<pre>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2441 &lt;result&gt; = sdiv &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2442</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2443
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2444<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2445
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2446<p>The '<tt>sdiv</tt>' instruction returns the quotient of its two
2447operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2448
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2449<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2450
2451<p>The two arguments to the '<tt>sdiv</tt>' instruction must be
2452<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2453values. Both arguments must have identical types.</p>
2454
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2455<h5>Semantics:</h5>
Chris Lattnerdc15b1d2008-04-01 18:45:27 +0000[diff] [blame]2456<p>The value produced is the signed integer quotient of the two operands rounded towards zero.</p>
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2457<p>Note that signed integer division and unsigned integer division are distinct
2458operations; for unsigned integer division, use '<tt>udiv</tt>'.</p>
2459<p>Division by zero leads to undefined behavior. Overflow also leads to
2460undefined behavior; this is a rare case, but can occur, for example,
2461by doing a 32-bit division of -2147483648 by -1.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2462<h5>Example:</h5>
2463<pre> &lt;result&gt; = sdiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
2464</pre>
2465</div>
2466<!-- _______________________________________________________________________ -->
2467<div class="doc_subsubsection"> <a name="i_fdiv">'<tt>fdiv</tt>'
2468Instruction</a> </div>
2469<div class="doc_text">
2470<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2471<pre>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2472 &lt;result&gt; = fdiv &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2473</pre>
2474<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2475
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2476<p>The '<tt>fdiv</tt>' instruction returns the quotient of its two
2477operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2478
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2479<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2480
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2481<p>The two arguments to the '<tt>fdiv</tt>' instruction must be
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2482<a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a>
2483of floating point values. Both arguments must have identical types.</p>
2484
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2485<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2486
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2487<p>The value produced is the floating point quotient of the two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2488
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2489<h5>Example:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2490
2491<pre>
2492 &lt;result&gt; = fdiv float 4.0, %var <i>; yields {float}:result = 4.0 / %var</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2493</pre>
2494</div>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2495
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2496<!-- _______________________________________________________________________ -->
2497<div class="doc_subsubsection"> <a name="i_urem">'<tt>urem</tt>' Instruction</a>
2498</div>
2499<div class="doc_text">
2500<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2501<pre> &lt;result&gt; = urem &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2502</pre>
2503<h5>Overview:</h5>
2504<p>The '<tt>urem</tt>' instruction returns the remainder from the
2505unsigned division of its two arguments.</p>
2506<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2507<p>The two arguments to the '<tt>urem</tt>' instruction must be
2508<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2509values. Both arguments must have identical types.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2510<h5>Semantics:</h5>
2511<p>This instruction returns the unsigned integer <i>remainder</i> of a division.
Chris Lattnerdc15b1d2008-04-01 18:45:27 +0000[diff] [blame]2512This instruction always performs an unsigned division to get the remainder.</p>
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2513<p>Note that unsigned integer remainder and signed integer remainder are
2514distinct operations; for signed integer remainder, use '<tt>srem</tt>'.</p>
2515<p>Taking the remainder of a division by zero leads to undefined behavior.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2516<h5>Example:</h5>
2517<pre> &lt;result&gt; = urem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
2518</pre>
2519
2520</div>
2521<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2522<div class="doc_subsubsection">
2523 <a name="i_srem">'<tt>srem</tt>' Instruction</a>
2524</div>
2525
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2526<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2527
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2528<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2529
2530<pre>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2531 &lt;result&gt; = srem &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2532</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2533
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2534<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2535
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2536<p>The '<tt>srem</tt>' instruction returns the remainder from the
Dan Gohman3e3fd8c2007-11-05 23:35:22 +0000[diff] [blame]2537signed division of its two operands. This instruction can also take
2538<a href="#t_vector">vector</a> versions of the values in which case
2539the elements must be integers.</p>
Chris Lattner08497ce2008-01-04 04:33:49 +0000[diff] [blame]2540
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2541<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2542
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2543<p>The two arguments to the '<tt>srem</tt>' instruction must be
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2544<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2545values. Both arguments must have identical types.</p>
2546
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2547<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2548
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2549<p>This instruction returns the <i>remainder</i> of a division (where the result
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2550has the same sign as the dividend, <tt>op1</tt>), not the <i>modulo</i>
2551operator (where the result has the same sign as the divisor, <tt>op2</tt>) of
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2552a value. For more information about the difference, see <a
2553 href="http://mathforum.org/dr.math/problems/anne.4.28.99.html">The
2554Math Forum</a>. For a table of how this is implemented in various languages,
2555please see <a href="http://en.wikipedia.org/wiki/Modulo_operation">
2556Wikipedia: modulo operation</a>.</p>
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2557<p>Note that signed integer remainder and unsigned integer remainder are
2558distinct operations; for unsigned integer remainder, use '<tt>urem</tt>'.</p>
2559<p>Taking the remainder of a division by zero leads to undefined behavior.
2560Overflow also leads to undefined behavior; this is a rare case, but can occur,
2561for example, by taking the remainder of a 32-bit division of -2147483648 by -1.
2562(The remainder doesn't actually overflow, but this rule lets srem be
2563implemented using instructions that return both the result of the division
2564and the remainder.)</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2565<h5>Example:</h5>
2566<pre> &lt;result&gt; = srem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
2567</pre>
2568
2569</div>
2570<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2571<div class="doc_subsubsection">
2572 <a name="i_frem">'<tt>frem</tt>' Instruction</a> </div>
2573
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2574<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2575
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2576<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2577<pre> &lt;result&gt; = frem &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2578</pre>
2579<h5>Overview:</h5>
2580<p>The '<tt>frem</tt>' instruction returns the remainder from the
2581division of its two operands.</p>
2582<h5>Arguments:</h5>
2583<p>The two arguments to the '<tt>frem</tt>' instruction must be
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2584<a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a>
2585of floating point values. Both arguments must have identical types.</p>
2586
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2587<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2588
Chris Lattnerdc15b1d2008-04-01 18:45:27 +0000[diff] [blame]2589<p>This instruction returns the <i>remainder</i> of a division.
2590The remainder has the same sign as the dividend.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2591
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2592<h5>Example:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2593
2594<pre>
2595 &lt;result&gt; = frem float 4.0, %var <i>; yields {float}:result = 4.0 % %var</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2596</pre>
2597</div>
2598
2599<!-- ======================================================================= -->
2600<div class="doc_subsection"> <a name="bitwiseops">Bitwise Binary
2601Operations</a> </div>
2602<div class="doc_text">
2603<p>Bitwise binary operators are used to do various forms of
2604bit-twiddling in a program. They are generally very efficient
2605instructions and can commonly be strength reduced from other
Chris Lattnerdc15b1d2008-04-01 18:45:27 +0000[diff] [blame]2606instructions. They require two operands of the same type, execute an operation on them,
2607and produce a single value. The resulting value is the same type as its operands.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2608</div>
2609
2610<!-- _______________________________________________________________________ -->
2611<div class="doc_subsubsection"> <a name="i_shl">'<tt>shl</tt>'
2612Instruction</a> </div>
2613<div class="doc_text">
2614<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2615<pre> &lt;result&gt; = shl &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2616</pre>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2617
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2618<h5>Overview:</h5>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2619
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2620<p>The '<tt>shl</tt>' instruction returns the first operand shifted to
2621the left a specified number of bits.</p>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2622
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2623<h5>Arguments:</h5>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2624
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2625<p>Both arguments to the '<tt>shl</tt>' instruction must be the same <a
Nate Begemanbb1ce942008-07-29 15:49:41 +0000[diff] [blame]2626 href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2627type. '<tt>op2</tt>' is treated as an unsigned value.</p>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2628
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2629<h5>Semantics:</h5>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2630
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2631<p>The value produced is <tt>op1</tt> * 2<sup><tt>op2</tt></sup> mod 2<sup>n</sup>,
2632where n is the width of the result. If <tt>op2</tt> is (statically or dynamically) negative or
Mon P Wangb0f51822008-12-10 08:55:09 +0000[diff] [blame]2633equal to or larger than the number of bits in <tt>op1</tt>, the result is undefined.
2634If the arguments are vectors, each vector element of <tt>op1</tt> is shifted by the
2635corresponding shift amount in <tt>op2</tt>.</p>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2636
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2637<h5>Example:</h5><pre>
2638 &lt;result&gt; = shl i32 4, %var <i>; yields {i32}: 4 &lt;&lt; %var</i>
2639 &lt;result&gt; = shl i32 4, 2 <i>; yields {i32}: 16</i>
2640 &lt;result&gt; = shl i32 1, 10 <i>; yields {i32}: 1024</i>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2641 &lt;result&gt; = shl i32 1, 32 <i>; undefined</i>
Mon P Wang9901e732008-12-09 05:46:39 +0000[diff] [blame]2642 &lt;result&gt; = shl &lt;2 x i32&gt; &lt; i32 1, i32 1&gt;, &lt; i32 1, i32 2&gt; <i>; yields: result=&lt;2 x i32&gt; &lt; i32 2, i32 4&gt;</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2643</pre>
2644</div>
2645<!-- _______________________________________________________________________ -->
2646<div class="doc_subsubsection"> <a name="i_lshr">'<tt>lshr</tt>'
2647Instruction</a> </div>
2648<div class="doc_text">
2649<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2650<pre> &lt;result&gt; = lshr &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2651</pre>
2652
2653<h5>Overview:</h5>
2654<p>The '<tt>lshr</tt>' instruction (logical shift right) returns the first
2655operand shifted to the right a specified number of bits with zero fill.</p>
2656
2657<h5>Arguments:</h5>
2658<p>Both arguments to the '<tt>lshr</tt>' instruction must be the same
Nate Begemanbb1ce942008-07-29 15:49:41 +0000[diff] [blame]2659<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2660type. '<tt>op2</tt>' is treated as an unsigned value.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2661
2662<h5>Semantics:</h5>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2663
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2664<p>This instruction always performs a logical shift right operation. The most
2665significant bits of the result will be filled with zero bits after the
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2666shift. If <tt>op2</tt> is (statically or dynamically) equal to or larger than
Mon P Wangb0f51822008-12-10 08:55:09 +0000[diff] [blame]2667the number of bits in <tt>op1</tt>, the result is undefined. If the arguments are
2668vectors, each vector element of <tt>op1</tt> is shifted by the corresponding shift
2669amount in <tt>op2</tt>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2670
2671<h5>Example:</h5>
2672<pre>
2673 &lt;result&gt; = lshr i32 4, 1 <i>; yields {i32}:result = 2</i>
2674 &lt;result&gt; = lshr i32 4, 2 <i>; yields {i32}:result = 1</i>
2675 &lt;result&gt; = lshr i8 4, 3 <i>; yields {i8}:result = 0</i>
2676 &lt;result&gt; = lshr i8 -2, 1 <i>; yields {i8}:result = 0x7FFFFFFF </i>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2677 &lt;result&gt; = lshr i32 1, 32 <i>; undefined</i>
Mon P Wang9901e732008-12-09 05:46:39 +0000[diff] [blame]2678 &lt;result&gt; = lshr &lt;2 x i32&gt; &lt; i32 -2, i32 4&gt;, &lt; i32 1, i32 2&gt; <i>; yields: result=&lt;2 x i32&gt; &lt; i32 0x7FFFFFFF, i32 1&gt;</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2679</pre>
2680</div>
2681
2682<!-- _______________________________________________________________________ -->
2683<div class="doc_subsubsection"> <a name="i_ashr">'<tt>ashr</tt>'
2684Instruction</a> </div>
2685<div class="doc_text">
2686
2687<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2688<pre> &lt;result&gt; = ashr &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2689</pre>
2690
2691<h5>Overview:</h5>
2692<p>The '<tt>ashr</tt>' instruction (arithmetic shift right) returns the first
2693operand shifted to the right a specified number of bits with sign extension.</p>
2694
2695<h5>Arguments:</h5>
2696<p>Both arguments to the '<tt>ashr</tt>' instruction must be the same
Nate Begemanbb1ce942008-07-29 15:49:41 +0000[diff] [blame]2697<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2698type. '<tt>op2</tt>' is treated as an unsigned value.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2699
2700<h5>Semantics:</h5>
2701<p>This instruction always performs an arithmetic shift right operation,
2702The most significant bits of the result will be filled with the sign bit
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2703of <tt>op1</tt>. If <tt>op2</tt> is (statically or dynamically) equal to or
Mon P Wangb0f51822008-12-10 08:55:09 +0000[diff] [blame]2704larger than the number of bits in <tt>op1</tt>, the result is undefined. If the
2705arguments are vectors, each vector element of <tt>op1</tt> is shifted by the
2706corresponding shift amount in <tt>op2</tt>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2707
2708<h5>Example:</h5>
2709<pre>
2710 &lt;result&gt; = ashr i32 4, 1 <i>; yields {i32}:result = 2</i>
2711 &lt;result&gt; = ashr i32 4, 2 <i>; yields {i32}:result = 1</i>
2712 &lt;result&gt; = ashr i8 4, 3 <i>; yields {i8}:result = 0</i>
2713 &lt;result&gt; = ashr i8 -2, 1 <i>; yields {i8}:result = -1</i>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2714 &lt;result&gt; = ashr i32 1, 32 <i>; undefined</i>
Mon P Wang9901e732008-12-09 05:46:39 +0000[diff] [blame]2715 &lt;result&gt; = ashr &lt;2 x i32&gt; &lt; i32 -2, i32 4&gt;, &lt; i32 1, i32 3&gt; <i>; yields: result=&lt;2 x i32&gt; &lt; i32 -1, i32 0&gt;</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2716</pre>
2717</div>
2718
2719<!-- _______________________________________________________________________ -->
2720<div class="doc_subsubsection"> <a name="i_and">'<tt>and</tt>'
2721Instruction</a> </div>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2722
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2723<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2724
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2725<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2726
2727<pre>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2728 &lt;result&gt; = and &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2729</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2730
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2731<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2732
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2733<p>The '<tt>and</tt>' instruction returns the bitwise logical and of
2734its two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2735
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2736<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2737
2738<p>The two arguments to the '<tt>and</tt>' instruction must be
2739<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2740values. Both arguments must have identical types.</p>
2741
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2742<h5>Semantics:</h5>
2743<p>The truth table used for the '<tt>and</tt>' instruction is:</p>
2744<p> </p>
Bill Wendling1e19eac2008-09-07 10:29:20 +0000[diff] [blame]2745<div>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2746<table border="1" cellspacing="0" cellpadding="4">
2747 <tbody>
2748 <tr>
2749 <td>In0</td>
2750 <td>In1</td>
2751 <td>Out</td>
2752 </tr>
2753 <tr>
2754 <td>0</td>
2755 <td>0</td>
2756 <td>0</td>
2757 </tr>
2758 <tr>
2759 <td>0</td>
2760 <td>1</td>
2761 <td>0</td>
2762 </tr>
2763 <tr>
2764 <td>1</td>
2765 <td>0</td>
2766 <td>0</td>
2767 </tr>
2768 <tr>
2769 <td>1</td>
2770 <td>1</td>
2771 <td>1</td>
2772 </tr>
2773 </tbody>
2774</table>
2775</div>
2776<h5>Example:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2777<pre>
2778 &lt;result&gt; = and i32 4, %var <i>; yields {i32}:result = 4 &amp; %var</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2779 &lt;result&gt; = and i32 15, 40 <i>; yields {i32}:result = 8</i>
2780 &lt;result&gt; = and i32 4, 8 <i>; yields {i32}:result = 0</i>
2781</pre>
2782</div>
2783<!-- _______________________________________________________________________ -->
2784<div class="doc_subsubsection"> <a name="i_or">'<tt>or</tt>' Instruction</a> </div>
2785<div class="doc_text">
2786<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2787<pre> &lt;result&gt; = or &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2788</pre>
2789<h5>Overview:</h5>
2790<p>The '<tt>or</tt>' instruction returns the bitwise logical inclusive
2791or of its two operands.</p>
2792<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2793
2794<p>The two arguments to the '<tt>or</tt>' instruction must be
2795<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2796values. Both arguments must have identical types.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2797<h5>Semantics:</h5>
2798<p>The truth table used for the '<tt>or</tt>' instruction is:</p>
2799<p> </p>
Bill Wendling1e19eac2008-09-07 10:29:20 +0000[diff] [blame]2800<div>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2801<table border="1" cellspacing="0" cellpadding="4">
2802 <tbody>
2803 <tr>
2804 <td>In0</td>
2805 <td>In1</td>
2806 <td>Out</td>
2807 </tr>
2808 <tr>
2809 <td>0</td>
2810 <td>0</td>
2811 <td>0</td>
2812 </tr>
2813 <tr>
2814 <td>0</td>
2815 <td>1</td>
2816 <td>1</td>
2817 </tr>
2818 <tr>
2819 <td>1</td>
2820 <td>0</td>
2821 <td>1</td>
2822 </tr>
2823 <tr>
2824 <td>1</td>
2825 <td>1</td>
2826 <td>1</td>
2827 </tr>
2828 </tbody>
2829</table>
2830</div>
2831<h5>Example:</h5>
2832<pre> &lt;result&gt; = or i32 4, %var <i>; yields {i32}:result = 4 | %var</i>
2833 &lt;result&gt; = or i32 15, 40 <i>; yields {i32}:result = 47</i>
2834 &lt;result&gt; = or i32 4, 8 <i>; yields {i32}:result = 12</i>
2835</pre>
2836</div>
2837<!-- _______________________________________________________________________ -->
2838<div class="doc_subsubsection"> <a name="i_xor">'<tt>xor</tt>'
2839Instruction</a> </div>
2840<div class="doc_text">
2841<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2842<pre> &lt;result&gt; = xor &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2843</pre>
2844<h5>Overview:</h5>
2845<p>The '<tt>xor</tt>' instruction returns the bitwise logical exclusive
2846or of its two operands. The <tt>xor</tt> is used to implement the
2847"one's complement" operation, which is the "~" operator in C.</p>
2848<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2849<p>The two arguments to the '<tt>xor</tt>' instruction must be
2850<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2851values. Both arguments must have identical types.</p>
2852
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2853<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2854
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2855<p>The truth table used for the '<tt>xor</tt>' instruction is:</p>
2856<p> </p>
Bill Wendling1e19eac2008-09-07 10:29:20 +0000[diff] [blame]2857<div>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2858<table border="1" cellspacing="0" cellpadding="4">
2859 <tbody>
2860 <tr>
2861 <td>In0</td>
2862 <td>In1</td>
2863 <td>Out</td>
2864 </tr>
2865 <tr>
2866 <td>0</td>
2867 <td>0</td>
2868 <td>0</td>
2869 </tr>
2870 <tr>
2871 <td>0</td>
2872 <td>1</td>
2873 <td>1</td>
2874 </tr>
2875 <tr>
2876 <td>1</td>
2877 <td>0</td>
2878 <td>1</td>
2879 </tr>
2880 <tr>
2881 <td>1</td>
2882 <td>1</td>
2883 <td>0</td>
2884 </tr>
2885 </tbody>
2886</table>
2887</div>
2888<p> </p>
2889<h5>Example:</h5>
2890<pre> &lt;result&gt; = xor i32 4, %var <i>; yields {i32}:result = 4 ^ %var</i>
2891 &lt;result&gt; = xor i32 15, 40 <i>; yields {i32}:result = 39</i>
2892 &lt;result&gt; = xor i32 4, 8 <i>; yields {i32}:result = 12</i>
2893 &lt;result&gt; = xor i32 %V, -1 <i>; yields {i32}:result = ~%V</i>
2894</pre>
2895</div>
2896
2897<!-- ======================================================================= -->
2898<div class="doc_subsection">
2899 <a name="vectorops">Vector Operations</a>
2900</div>
2901
2902<div class="doc_text">
2903
2904<p>LLVM supports several instructions to represent vector operations in a
2905target-independent manner. These instructions cover the element-access and
2906vector-specific operations needed to process vectors effectively. While LLVM
2907does directly support these vector operations, many sophisticated algorithms
2908will want to use target-specific intrinsics to take full advantage of a specific
2909target.</p>
2910
2911</div>
2912
2913<!-- _______________________________________________________________________ -->
2914<div class="doc_subsubsection">
2915 <a name="i_extractelement">'<tt>extractelement</tt>' Instruction</a>
2916</div>
2917
2918<div class="doc_text">
2919
2920<h5>Syntax:</h5>
2921
2922<pre>
2923 &lt;result&gt; = extractelement &lt;n x &lt;ty&gt;&gt; &lt;val&gt;, i32 &lt;idx&gt; <i>; yields &lt;ty&gt;</i>
2924</pre>
2925
2926<h5>Overview:</h5>
2927
2928<p>
2929The '<tt>extractelement</tt>' instruction extracts a single scalar
2930element from a vector at a specified index.
2931</p>
2932
2933
2934<h5>Arguments:</h5>
2935
2936<p>
2937The first operand of an '<tt>extractelement</tt>' instruction is a
2938value of <a href="#t_vector">vector</a> type. The second operand is
2939an index indicating the position from which to extract the element.
2940The index may be a variable.</p>
2941
2942<h5>Semantics:</h5>
2943
2944<p>
2945The result is a scalar of the same type as the element type of
2946<tt>val</tt>. Its value is the value at position <tt>idx</tt> of
2947<tt>val</tt>. If <tt>idx</tt> exceeds the length of <tt>val</tt>, the
2948results are undefined.
2949</p>
2950
2951<h5>Example:</h5>
2952
2953<pre>
2954 %result = extractelement &lt;4 x i32&gt; %vec, i32 0 <i>; yields i32</i>
2955</pre>
2956</div>
2957
2958
2959<!-- _______________________________________________________________________ -->
2960<div class="doc_subsubsection">
2961 <a name="i_insertelement">'<tt>insertelement</tt>' Instruction</a>
2962</div>
2963
2964<div class="doc_text">
2965
2966<h5>Syntax:</h5>
2967
2968<pre>
Dan Gohmanbcc3c502008-05-12 23:38:42 +0000[diff] [blame]2969 &lt;result&gt; = insertelement &lt;n x &lt;ty&gt;&gt; &lt;val&gt;, &lt;ty&gt; &lt;elt&gt;, i32 &lt;idx&gt; <i>; yields &lt;n x &lt;ty&gt;&gt;</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2970</pre>
2971
2972<h5>Overview:</h5>
2973
2974<p>
2975The '<tt>insertelement</tt>' instruction inserts a scalar
2976element into a vector at a specified index.
2977</p>
2978
2979
2980<h5>Arguments:</h5>
2981
2982<p>
2983The first operand of an '<tt>insertelement</tt>' instruction is a
2984value of <a href="#t_vector">vector</a> type. The second operand is a
2985scalar value whose type must equal the element type of the first
2986operand. The third operand is an index indicating the position at
2987which to insert the value. The index may be a variable.</p>
2988
2989<h5>Semantics:</h5>
2990
2991<p>
2992The result is a vector of the same type as <tt>val</tt>. Its
2993element values are those of <tt>val</tt> except at position
2994<tt>idx</tt>, where it gets the value <tt>elt</tt>. If <tt>idx</tt>
2995exceeds the length of <tt>val</tt>, the results are undefined.
2996</p>
2997
2998<h5>Example:</h5>
2999
3000<pre>
3001 %result = insertelement &lt;4 x i32&gt; %vec, i32 1, i32 0 <i>; yields &lt;4 x i32&gt;</i>
3002</pre>
3003</div>
3004
3005<!-- _______________________________________________________________________ -->
3006<div class="doc_subsubsection">
3007 <a name="i_shufflevector">'<tt>shufflevector</tt>' Instruction</a>
3008</div>
3009
3010<div class="doc_text">
3011
3012<h5>Syntax:</h5>
3013
3014<pre>
Mon P Wangbff5d9c2008-11-10 04:46:22 +0000[diff] [blame]3015 &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;m x i32&gt; &lt;mask&gt; <i>; yields &lt;m x &lt;ty&gt;&gt;</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3016</pre>
3017
3018<h5>Overview:</h5>
3019
3020<p>
3021The '<tt>shufflevector</tt>' instruction constructs a permutation of elements
Mon P Wangbff5d9c2008-11-10 04:46:22 +0000[diff] [blame]3022from two input vectors, returning a vector with the same element type as
3023the input and length that is the same as the shuffle mask.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3024</p>
3025
3026<h5>Arguments:</h5>
3027
3028<p>
Mon P Wangbff5d9c2008-11-10 04:46:22 +0000[diff] [blame]3029The first two operands of a '<tt>shufflevector</tt>' instruction are vectors
3030with types that match each other. The third argument is a shuffle mask whose
3031element type is always 'i32'. The result of the instruction is a vector whose
3032length is the same as the shuffle mask and whose element type is the same as
3033the element type of the first two operands.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3034</p>
3035
3036<p>
3037The shuffle mask operand is required to be a constant vector with either
3038constant integer or undef values.
3039</p>
3040
3041<h5>Semantics:</h5>
3042
3043<p>
3044The elements of the two input vectors are numbered from left to right across
3045both of the vectors. The shuffle mask operand specifies, for each element of
Mon P Wangbff5d9c2008-11-10 04:46:22 +0000[diff] [blame]3046the result vector, which element of the two input vectors the result element
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3047gets. The element selector may be undef (meaning "don't care") and the second
3048operand may be undef if performing a shuffle from only one vector.
3049</p>
3050
3051<h5>Example:</h5>
3052
3053<pre>
3054 %result = shufflevector &lt;4 x i32&gt; %v1, &lt;4 x i32&gt; %v2,
3055 &lt;4 x i32&gt; &lt;i32 0, i32 4, i32 1, i32 5&gt; <i>; yields &lt;4 x i32&gt;</i>
3056 %result = shufflevector &lt;4 x i32&gt; %v1, &lt;4 x i32&gt; undef,
3057 &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.
Mon P Wangbff5d9c2008-11-10 04:46:22 +0000[diff] [blame]3058 %result = shufflevector &lt;8 x i32&gt; %v1, &lt;8 x i32&gt; undef,
3059 &lt;4 x i32&gt; &lt;i32 0, i32 1, i32 2, i32 3&gt; <i>; yields &lt;4 x i32&gt;</i>
3060 %result = shufflevector &lt;4 x i32&gt; %v1, &lt;4 x i32&gt; %v2,
3061 &lt;8 x i32&gt; &lt;i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7 &gt; <i>; yields &lt;8 x i32&gt;</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3062</pre>
3063</div>
3064
3065
3066<!-- ======================================================================= -->
3067<div class="doc_subsection">
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3068 <a name="aggregateops">Aggregate Operations</a>
3069</div>
3070
3071<div class="doc_text">
3072
3073<p>LLVM supports several instructions for working with aggregate values.
3074</p>
3075
3076</div>
3077
3078<!-- _______________________________________________________________________ -->
3079<div class="doc_subsubsection">
3080 <a name="i_extractvalue">'<tt>extractvalue</tt>' Instruction</a>
3081</div>
3082
3083<div class="doc_text">
3084
3085<h5>Syntax:</h5>
3086
3087<pre>
3088 &lt;result&gt; = extractvalue &lt;aggregate type&gt; &lt;val&gt;, &lt;idx&gt;{, &lt;idx&gt;}*
3089</pre>
3090
3091<h5>Overview:</h5>
3092
3093<p>
Dan Gohman6c6dea02008-05-13 18:16:06 +0000[diff] [blame]3094The '<tt>extractvalue</tt>' instruction extracts the value of a struct field
3095or array element from an aggregate value.
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3096</p>
3097
3098
3099<h5>Arguments:</h5>
3100
3101<p>
3102The first operand of an '<tt>extractvalue</tt>' instruction is a
3103value of <a href="#t_struct">struct</a> or <a href="#t_array">array</a>
Dan Gohman6c6dea02008-05-13 18:16:06 +0000[diff] [blame]3104type. The operands are constant indices to specify which value to extract
Dan Gohmane5febe42008-05-31 00:58:22 +0000[diff] [blame]3105in a similar manner as indices in a
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3106'<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.
3107</p>
3108
3109<h5>Semantics:</h5>
3110
3111<p>
3112The result is the value at the position in the aggregate specified by
3113the index operands.
3114</p>
3115
3116<h5>Example:</h5>
3117
3118<pre>
Dan Gohmane5febe42008-05-31 00:58:22 +0000[diff] [blame]3119 %result = extractvalue {i32, float} %agg, 0 <i>; yields i32</i>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3120</pre>
3121</div>
3122
3123
3124<!-- _______________________________________________________________________ -->
3125<div class="doc_subsubsection">
3126 <a name="i_insertvalue">'<tt>insertvalue</tt>' Instruction</a>
3127</div>
3128
3129<div class="doc_text">
3130
3131<h5>Syntax:</h5>
3132
3133<pre>
Dan Gohmane5febe42008-05-31 00:58:22 +0000[diff] [blame]3134 &lt;result&gt; = insertvalue &lt;aggregate type&gt; &lt;val&gt;, &lt;ty&gt; &lt;val&gt;, &lt;idx&gt; <i>; yields &lt;n x &lt;ty&gt;&gt;</i>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3135</pre>
3136
3137<h5>Overview:</h5>
3138
3139<p>
3140The '<tt>insertvalue</tt>' instruction inserts a value
Dan Gohman6c6dea02008-05-13 18:16:06 +0000[diff] [blame]3141into a struct field or array element in an aggregate.
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3142</p>
3143
3144
3145<h5>Arguments:</h5>
3146
3147<p>
3148The first operand of an '<tt>insertvalue</tt>' instruction is a
3149value of <a href="#t_struct">struct</a> or <a href="#t_array">array</a> type.
3150The second operand is a first-class value to insert.
Dan Gohman4f29e422008-05-23 21:53:15 +0000[diff] [blame]3151The following operands are constant indices
Dan Gohmane5febe42008-05-31 00:58:22 +0000[diff] [blame]3152indicating the position at which to insert the value in a similar manner as
Dan Gohman6c6dea02008-05-13 18:16:06 +0000[diff] [blame]3153indices in a
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3154'<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.
3155The value to insert must have the same type as the value identified
Dan Gohman6c6dea02008-05-13 18:16:06 +0000[diff] [blame]3156by the indices.
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]3157</p>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3158
3159<h5>Semantics:</h5>
3160
3161<p>
3162The result is an aggregate of the same type as <tt>val</tt>. Its
3163value is that of <tt>val</tt> except that the value at the position
Dan Gohman6c6dea02008-05-13 18:16:06 +0000[diff] [blame]3164specified by the indices is that of <tt>elt</tt>.
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3165</p>
3166
3167<h5>Example:</h5>
3168
3169<pre>
Dan Gohmanb1aab4e2008-06-23 15:26:37 +0000[diff] [blame]3170 %result = insertvalue {i32, float} %agg, i32 1, 0 <i>; yields {i32, float}</i>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3171</pre>
3172</div>
3173
3174
3175<!-- ======================================================================= -->
3176<div class="doc_subsection">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3177 <a name="memoryops">Memory Access and Addressing Operations</a>
3178</div>
3179
3180<div class="doc_text">
3181
3182<p>A key design point of an SSA-based representation is how it
3183represents memory. In LLVM, no memory locations are in SSA form, which
3184makes things very simple. This section describes how to read, write,
3185allocate, and free memory in LLVM.</p>
3186
3187</div>
3188
3189<!-- _______________________________________________________________________ -->
3190<div class="doc_subsubsection">
3191 <a name="i_malloc">'<tt>malloc</tt>' Instruction</a>
3192</div>
3193
3194<div class="doc_text">
3195
3196<h5>Syntax:</h5>
3197
3198<pre>
3199 &lt;result&gt; = malloc &lt;type&gt;[, i32 &lt;NumElements&gt;][, align &lt;alignment&gt;] <i>; yields {type*}:result</i>
3200</pre>
3201
3202<h5>Overview:</h5>
3203
3204<p>The '<tt>malloc</tt>' instruction allocates memory from the system
Christopher Lambcfe00962007-12-17 01:00:21 +0000[diff] [blame]3205heap and returns a pointer to it. The object is always allocated in the generic
3206address space (address space zero).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3207
3208<h5>Arguments:</h5>
3209
3210<p>The '<tt>malloc</tt>' instruction allocates
3211<tt>sizeof(&lt;type&gt;)*NumElements</tt>
3212bytes of memory from the operating system and returns a pointer of the
3213appropriate type to the program. If "NumElements" is specified, it is the
Gabor Greif5082cf42008-02-09 22:24:34 +0000[diff] [blame]3214number of elements allocated, otherwise "NumElements" is defaulted to be one.
Chris Lattner10368b62008-04-02 00:38:26 +0000[diff] [blame]3215If a constant alignment is specified, the value result of the allocation is guaranteed to
Gabor Greif5082cf42008-02-09 22:24:34 +0000[diff] [blame]3216be aligned to at least that boundary. If not specified, or if zero, the target can
3217choose to align the allocation on any convenient boundary.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3218
3219<p>'<tt>type</tt>' must be a sized type.</p>
3220
3221<h5>Semantics:</h5>
3222
3223<p>Memory is allocated using the system "<tt>malloc</tt>" function, and
Nick Lewyckyff384472008-11-24 03:41:24 +0000[diff] [blame]3224a pointer is returned. The result of a zero byte allocation is undefined. The
Chris Lattner8b094fc2008-04-19 21:01:16 +0000[diff] [blame]3225result is null if there is insufficient memory available.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3226
3227<h5>Example:</h5>
3228
3229<pre>
Dan Gohmanf54f50a2009-01-04 23:49:44 +0000[diff] [blame]3230 %array = malloc [4 x i8] <i>; yields {[%4 x i8]*}:array</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3231
3232 %size = <a href="#i_add">add</a> i32 2, 2 <i>; yields {i32}:size = i32 4</i>
3233 %array1 = malloc i8, i32 4 <i>; yields {i8*}:array1</i>
3234 %array2 = malloc [12 x i8], i32 %size <i>; yields {[12 x i8]*}:array2</i>
3235 %array3 = malloc i32, i32 4, align 1024 <i>; yields {i32*}:array3</i>
3236 %array4 = malloc i32, align 1024 <i>; yields {i32*}:array4</i>
3237</pre>
Dan Gohman60967192009-01-12 23:12:39 +0000[diff] [blame]3238
3239<p>Note that the code generator does not yet respect the
3240 alignment value.</p>
3241
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3242</div>
3243
3244<!-- _______________________________________________________________________ -->
3245<div class="doc_subsubsection">
3246 <a name="i_free">'<tt>free</tt>' Instruction</a>
3247</div>
3248
3249<div class="doc_text">
3250
3251<h5>Syntax:</h5>
3252
3253<pre>
Dan Gohmanf54f50a2009-01-04 23:49:44 +0000[diff] [blame]3254 free &lt;type&gt; &lt;value&gt; <i>; yields {void}</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3255</pre>
3256
3257<h5>Overview:</h5>
3258
3259<p>The '<tt>free</tt>' instruction returns memory back to the unused
3260memory heap to be reallocated in the future.</p>
3261
3262<h5>Arguments:</h5>
3263
3264<p>'<tt>value</tt>' shall be a pointer value that points to a value
3265that was allocated with the '<tt><a href="#i_malloc">malloc</a></tt>'
3266instruction.</p>
3267
3268<h5>Semantics:</h5>
3269
3270<p>Access to the memory pointed to by the pointer is no longer defined
Chris Lattner329d6532008-04-19 22:41:32 +0000[diff] [blame]3271after this instruction executes. If the pointer is null, the operation
3272is a noop.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3273
3274<h5>Example:</h5>
3275
3276<pre>
Dan Gohmanf54f50a2009-01-04 23:49:44 +0000[diff] [blame]3277 %array = <a href="#i_malloc">malloc</a> [4 x i8] <i>; yields {[4 x i8]*}:array</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3278 free [4 x i8]* %array
3279</pre>
3280</div>
3281
3282<!-- _______________________________________________________________________ -->
3283<div class="doc_subsubsection">
3284 <a name="i_alloca">'<tt>alloca</tt>' Instruction</a>
3285</div>
3286
3287<div class="doc_text">
3288
3289<h5>Syntax:</h5>
3290
3291<pre>
3292 &lt;result&gt; = alloca &lt;type&gt;[, i32 &lt;NumElements&gt;][, align &lt;alignment&gt;] <i>; yields {type*}:result</i>
3293</pre>
3294
3295<h5>Overview:</h5>
3296
3297<p>The '<tt>alloca</tt>' instruction allocates memory on the stack frame of the
3298currently executing function, to be automatically released when this function
Christopher Lambcfe00962007-12-17 01:00:21 +0000[diff] [blame]3299returns to its caller. The object is always allocated in the generic address
3300space (address space zero).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3301
3302<h5>Arguments:</h5>
3303
3304<p>The '<tt>alloca</tt>' instruction allocates <tt>sizeof(&lt;type&gt;)*NumElements</tt>
3305bytes of memory on the runtime stack, returning a pointer of the
Gabor Greif5082cf42008-02-09 22:24:34 +0000[diff] [blame]3306appropriate type to the program. If "NumElements" is specified, it is the
3307number of elements allocated, otherwise "NumElements" is defaulted to be one.
Chris Lattner10368b62008-04-02 00:38:26 +0000[diff] [blame]3308If a constant alignment is specified, the value result of the allocation is guaranteed
Gabor Greif5082cf42008-02-09 22:24:34 +0000[diff] [blame]3309to be aligned to at least that boundary. If not specified, or if zero, the target
3310can choose to align the allocation on any convenient boundary.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3311
3312<p>'<tt>type</tt>' may be any sized type.</p>
3313
3314<h5>Semantics:</h5>
3315
Chris Lattner8b094fc2008-04-19 21:01:16 +0000[diff] [blame]3316<p>Memory is allocated; a pointer is returned. The operation is undefiend if
3317there is insufficient stack space for the allocation. '<tt>alloca</tt>'d
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3318memory is automatically released when the function returns. The '<tt>alloca</tt>'
3319instruction is commonly used to represent automatic variables that must
3320have an address available. When the function returns (either with the <tt><a
3321 href="#i_ret">ret</a></tt> or <tt><a href="#i_unwind">unwind</a></tt>
Chris Lattner10368b62008-04-02 00:38:26 +0000[diff] [blame]3322instructions), the memory is reclaimed. Allocating zero bytes
3323is legal, but the result is undefined.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3324
3325<h5>Example:</h5>
3326
3327<pre>
Dan Gohmanf54f50a2009-01-04 23:49:44 +0000[diff] [blame]3328 %ptr = alloca i32 <i>; yields {i32*}:ptr</i>
3329 %ptr = alloca i32, i32 4 <i>; yields {i32*}:ptr</i>
3330 %ptr = alloca i32, i32 4, align 1024 <i>; yields {i32*}:ptr</i>
3331 %ptr = alloca i32, align 1024 <i>; yields {i32*}:ptr</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3332</pre>
3333</div>
3334
3335<!-- _______________________________________________________________________ -->
3336<div class="doc_subsubsection"> <a name="i_load">'<tt>load</tt>'
3337Instruction</a> </div>
3338<div class="doc_text">
3339<h5>Syntax:</h5>
3340<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>
3341<h5>Overview:</h5>
3342<p>The '<tt>load</tt>' instruction is used to read from memory.</p>
3343<h5>Arguments:</h5>
3344<p>The argument to the '<tt>load</tt>' instruction specifies the memory
3345address from which to load. The pointer must point to a <a
3346 href="#t_firstclass">first class</a> type. If the <tt>load</tt> is
3347marked as <tt>volatile</tt>, then the optimizer is not allowed to modify
3348the number or order of execution of this <tt>load</tt> with other
3349volatile <tt>load</tt> and <tt><a href="#i_store">store</a></tt>
3350instructions. </p>
Chris Lattner21003c82008-01-06 21:04:43 +0000[diff] [blame]3351<p>
Chris Lattner10368b62008-04-02 00:38:26 +0000[diff] [blame]3352The optional constant "align" argument specifies the alignment of the operation
Chris Lattner21003c82008-01-06 21:04:43 +0000[diff] [blame]3353(that is, the alignment of the memory address). A value of 0 or an
3354omitted "align" argument means that the operation has the preferential
3355alignment for the target. It is the responsibility of the code emitter
3356to ensure that the alignment information is correct. Overestimating
3357the alignment results in an undefined behavior. Underestimating the
3358alignment may produce less efficient code. An alignment of 1 is always
3359safe.
3360</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3361<h5>Semantics:</h5>
3362<p>The location of memory pointed to is loaded.</p>
3363<h5>Examples:</h5>
3364<pre> %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
3365 <a
3366 href="#i_store">store</a> i32 3, i32* %ptr <i>; yields {void}</i>
3367 %val = load i32* %ptr <i>; yields {i32}:val = i32 3</i>
3368</pre>
3369</div>
3370<!-- _______________________________________________________________________ -->
3371<div class="doc_subsubsection"> <a name="i_store">'<tt>store</tt>'
3372Instruction</a> </div>
3373<div class="doc_text">
3374<h5>Syntax:</h5>
3375<pre> store &lt;ty&gt; &lt;value&gt;, &lt;ty&gt;* &lt;pointer&gt;[, align &lt;alignment&gt;] <i>; yields {void}</i>
3376 volatile store &lt;ty&gt; &lt;value&gt;, &lt;ty&gt;* &lt;pointer&gt;[, align &lt;alignment&gt;] <i>; yields {void}</i>
3377</pre>
3378<h5>Overview:</h5>
3379<p>The '<tt>store</tt>' instruction is used to write to memory.</p>
3380<h5>Arguments:</h5>
3381<p>There are two arguments to the '<tt>store</tt>' instruction: a value
3382to store and an address at which to store it. The type of the '<tt>&lt;pointer&gt;</tt>'
Chris Lattner10368b62008-04-02 00:38:26 +0000[diff] [blame]3383operand must be a pointer to the <a href="#t_firstclass">first class</a> type
3384of the '<tt>&lt;value&gt;</tt>'
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3385operand. If the <tt>store</tt> is marked as <tt>volatile</tt>, then the
3386optimizer is not allowed to modify the number or order of execution of
3387this <tt>store</tt> with other volatile <tt>load</tt> and <tt><a
3388 href="#i_store">store</a></tt> instructions.</p>
Chris Lattner21003c82008-01-06 21:04:43 +0000[diff] [blame]3389<p>
Chris Lattner10368b62008-04-02 00:38:26 +0000[diff] [blame]3390The optional constant "align" argument specifies the alignment of the operation
Chris Lattner21003c82008-01-06 21:04:43 +0000[diff] [blame]3391(that is, the alignment of the memory address). A value of 0 or an
3392omitted "align" argument means that the operation has the preferential
3393alignment for the target. It is the responsibility of the code emitter
3394to ensure that the alignment information is correct. Overestimating
3395the alignment results in an undefined behavior. Underestimating the
3396alignment may produce less efficient code. An alignment of 1 is always
3397safe.
3398</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3399<h5>Semantics:</h5>
3400<p>The contents of memory are updated to contain '<tt>&lt;value&gt;</tt>'
3401at the location specified by the '<tt>&lt;pointer&gt;</tt>' operand.</p>
3402<h5>Example:</h5>
3403<pre> %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
Bill Wendling63ffa142007-10-22 05:10:05 +0000[diff] [blame]3404 store i32 3, i32* %ptr <i>; yields {void}</i>
3405 %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]3406</pre>
3407</div>
3408
3409<!-- _______________________________________________________________________ -->
3410<div class="doc_subsubsection">
3411 <a name="i_getelementptr">'<tt>getelementptr</tt>' Instruction</a>
3412</div>
3413
3414<div class="doc_text">
3415<h5>Syntax:</h5>
3416<pre>
Matthijs Kooijman32a080e2008-10-13 13:44:15 +0000[diff] [blame]3417 &lt;result&gt; = getelementptr &lt;pty&gt;* &lt;ptrval&gt;{, &lt;ty&gt; &lt;idx&gt;}*
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3418</pre>
3419
3420<h5>Overview:</h5>
3421
3422<p>
3423The '<tt>getelementptr</tt>' instruction is used to get the address of a
Matthijs Kooijman32a080e2008-10-13 13:44:15 +0000[diff] [blame]3424subelement of an aggregate data structure. It performs address calculation only
3425and does not access memory.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3426
3427<h5>Arguments:</h5>
3428
Matthijs Kooijman32a080e2008-10-13 13:44:15 +0000[diff] [blame]3429<p>The first argument is always a pointer, and forms the basis of the
3430calculation. The remaining arguments are indices, that indicate which of the
3431elements of the aggregate object are indexed. The interpretation of each index
3432is dependent on the type being indexed into. The first index always indexes the
3433pointer value given as the first argument, the second index indexes a value of
3434the type pointed to (not necessarily the value directly pointed to, since the
3435first index can be non-zero), etc. The first type indexed into must be a pointer
3436value, subsequent types can be arrays, vectors and structs. Note that subsequent
3437types being indexed into can never be pointers, since that would require loading
3438the pointer before continuing calculation.</p>
3439
3440<p>The type of each index argument depends on the type it is indexing into.
3441When indexing into a (packed) structure, only <tt>i32</tt> integer
3442<b>constants</b> are allowed. When indexing into an array, pointer or vector,
3443only integers of 32 or 64 bits are allowed (also non-constants). 32-bit values
3444will be sign extended to 64-bits if required.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3445
3446<p>For example, let's consider a C code fragment and how it gets
3447compiled to LLVM:</p>
3448
3449<div class="doc_code">
3450<pre>
3451struct RT {
3452 char A;
3453 int B[10][20];
3454 char C;
3455};
3456struct ST {
3457 int X;
3458 double Y;
3459 struct RT Z;
3460};
3461
3462int *foo(struct ST *s) {
3463 return &amp;s[1].Z.B[5][13];
3464}
3465</pre>
3466</div>
3467
3468<p>The LLVM code generated by the GCC frontend is:</p>
3469
3470<div class="doc_code">
3471<pre>
Chris Lattner5b6dc6e2009-01-11 20:53:49 +0000[diff] [blame]3472%RT = <a href="#namedtypes">type</a> { i8 , [10 x [20 x i32]], i8 }
3473%ST = <a href="#namedtypes">type</a> { i32, double, %RT }
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3474
3475define i32* %foo(%ST* %s) {
3476entry:
3477 %reg = getelementptr %ST* %s, i32 1, i32 2, i32 1, i32 5, i32 13
3478 ret i32* %reg
3479}
3480</pre>
3481</div>
3482
3483<h5>Semantics:</h5>
3484
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3485<p>In the example above, the first index is indexing into the '<tt>%ST*</tt>'
3486type, which is a pointer, yielding a '<tt>%ST</tt>' = '<tt>{ i32, double, %RT
3487}</tt>' type, a structure. The second index indexes into the third element of
3488the structure, yielding a '<tt>%RT</tt>' = '<tt>{ i8 , [10 x [20 x i32]],
3489i8 }</tt>' type, another structure. The third index indexes into the second
3490element of the structure, yielding a '<tt>[10 x [20 x i32]]</tt>' type, an
3491array. The two dimensions of the array are subscripted into, yielding an
3492'<tt>i32</tt>' type. The '<tt>getelementptr</tt>' instruction returns a pointer
3493to this element, thus computing a value of '<tt>i32*</tt>' type.</p>
3494
3495<p>Note that it is perfectly legal to index partially through a
3496structure, returning a pointer to an inner element. Because of this,
3497the LLVM code for the given testcase is equivalent to:</p>
3498
3499<pre>
3500 define i32* %foo(%ST* %s) {
3501 %t1 = getelementptr %ST* %s, i32 1 <i>; yields %ST*:%t1</i>
3502 %t2 = getelementptr %ST* %t1, i32 0, i32 2 <i>; yields %RT*:%t2</i>
3503 %t3 = getelementptr %RT* %t2, i32 0, i32 1 <i>; yields [10 x [20 x i32]]*:%t3</i>
3504 %t4 = getelementptr [10 x [20 x i32]]* %t3, i32 0, i32 5 <i>; yields [20 x i32]*:%t4</i>
3505 %t5 = getelementptr [20 x i32]* %t4, i32 0, i32 13 <i>; yields i32*:%t5</i>
3506 ret i32* %t5
3507 }
3508</pre>
3509
3510<p>Note that it is undefined to access an array out of bounds: array and
3511pointer indexes must always be within the defined bounds of the array type.
Chris Lattnera7d94ba2008-04-24 05:59:56 +0000[diff] [blame]3512The one exception for this rule is zero length arrays. These arrays are
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3513defined to be accessible as variable length arrays, which requires access
3514beyond the zero'th element.</p>
3515
3516<p>The getelementptr instruction is often confusing. For some more insight
3517into how it works, see <a href="GetElementPtr.html">the getelementptr
3518FAQ</a>.</p>
3519
3520<h5>Example:</h5>
3521
3522<pre>
3523 <i>; yields [12 x i8]*:aptr</i>
Matthijs Kooijman32a080e2008-10-13 13:44:15 +0000[diff] [blame]3524 %aptr = getelementptr {i32, [12 x i8]}* %saptr, i64 0, i32 1
3525 <i>; yields i8*:vptr</i>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]3526 %vptr = getelementptr {i32, &lt;2 x i8&gt;}* %svptr, i64 0, i32 1, i32 1
Matthijs Kooijman32a080e2008-10-13 13:44:15 +0000[diff] [blame]3527 <i>; yields i8*:eptr</i>
3528 %eptr = getelementptr [12 x i8]* %aptr, i64 0, i32 1
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3529</pre>
3530</div>
3531
3532<!-- ======================================================================= -->
3533<div class="doc_subsection"> <a name="convertops">Conversion Operations</a>
3534</div>
3535<div class="doc_text">
3536<p>The instructions in this category are the conversion instructions (casting)
3537which all take a single operand and a type. They perform various bit conversions
3538on the operand.</p>
3539</div>
3540
3541<!-- _______________________________________________________________________ -->
3542<div class="doc_subsubsection">
3543 <a name="i_trunc">'<tt>trunc .. to</tt>' Instruction</a>
3544</div>
3545<div class="doc_text">
3546
3547<h5>Syntax:</h5>
3548<pre>
3549 &lt;result&gt; = trunc &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3550</pre>
3551
3552<h5>Overview:</h5>
3553<p>
3554The '<tt>trunc</tt>' instruction truncates its operand to the type <tt>ty2</tt>.
3555</p>
3556
3557<h5>Arguments:</h5>
3558<p>
3559The '<tt>trunc</tt>' instruction takes a <tt>value</tt> to trunc, which must
3560be an <a href="#t_integer">integer</a> type, and a type that specifies the size
3561and type of the result, which must be an <a href="#t_integer">integer</a>
3562type. The bit size of <tt>value</tt> must be larger than the bit size of
3563<tt>ty2</tt>. Equal sized types are not allowed.</p>
3564
3565<h5>Semantics:</h5>
3566<p>
3567The '<tt>trunc</tt>' instruction truncates the high order bits in <tt>value</tt>
3568and converts the remaining bits to <tt>ty2</tt>. Since the source size must be
3569larger than the destination size, <tt>trunc</tt> cannot be a <i>no-op cast</i>.
3570It will always truncate bits.</p>
3571
3572<h5>Example:</h5>
3573<pre>
3574 %X = trunc i32 257 to i8 <i>; yields i8:1</i>
3575 %Y = trunc i32 123 to i1 <i>; yields i1:true</i>
3576 %Y = trunc i32 122 to i1 <i>; yields i1:false</i>
3577</pre>
3578</div>
3579
3580<!-- _______________________________________________________________________ -->
3581<div class="doc_subsubsection">
3582 <a name="i_zext">'<tt>zext .. to</tt>' Instruction</a>
3583</div>
3584<div class="doc_text">
3585
3586<h5>Syntax:</h5>
3587<pre>
3588 &lt;result&gt; = zext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3589</pre>
3590
3591<h5>Overview:</h5>
3592<p>The '<tt>zext</tt>' instruction zero extends its operand to type
3593<tt>ty2</tt>.</p>
3594
3595
3596<h5>Arguments:</h5>
3597<p>The '<tt>zext</tt>' instruction takes a value to cast, which must be of
3598<a href="#t_integer">integer</a> type, and a type to cast it to, which must
3599also be of <a href="#t_integer">integer</a> type. The bit size of the
3600<tt>value</tt> must be smaller than the bit size of the destination type,
3601<tt>ty2</tt>.</p>
3602
3603<h5>Semantics:</h5>
3604<p>The <tt>zext</tt> fills the high order bits of the <tt>value</tt> with zero
3605bits until it reaches the size of the destination type, <tt>ty2</tt>.</p>
3606
3607<p>When zero extending from i1, the result will always be either 0 or 1.</p>
3608
3609<h5>Example:</h5>
3610<pre>
3611 %X = zext i32 257 to i64 <i>; yields i64:257</i>
3612 %Y = zext i1 true to i32 <i>; yields i32:1</i>
3613</pre>
3614</div>
3615
3616<!-- _______________________________________________________________________ -->
3617<div class="doc_subsubsection">
3618 <a name="i_sext">'<tt>sext .. to</tt>' Instruction</a>
3619</div>
3620<div class="doc_text">
3621
3622<h5>Syntax:</h5>
3623<pre>
3624 &lt;result&gt; = sext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3625</pre>
3626
3627<h5>Overview:</h5>
3628<p>The '<tt>sext</tt>' sign extends <tt>value</tt> to the type <tt>ty2</tt>.</p>
3629
3630<h5>Arguments:</h5>
3631<p>
3632The '<tt>sext</tt>' instruction takes a value to cast, which must be of
3633<a href="#t_integer">integer</a> type, and a type to cast it to, which must
3634also be of <a href="#t_integer">integer</a> type. The bit size of the
3635<tt>value</tt> must be smaller than the bit size of the destination type,
3636<tt>ty2</tt>.</p>
3637
3638<h5>Semantics:</h5>
3639<p>
3640The '<tt>sext</tt>' instruction performs a sign extension by copying the sign
3641bit (highest order bit) of the <tt>value</tt> until it reaches the bit size of
3642the type <tt>ty2</tt>.</p>
3643
3644<p>When sign extending from i1, the extension always results in -1 or 0.</p>
3645
3646<h5>Example:</h5>
3647<pre>
3648 %X = sext i8 -1 to i16 <i>; yields i16 :65535</i>
3649 %Y = sext i1 true to i32 <i>; yields i32:-1</i>
3650</pre>
3651</div>
3652
3653<!-- _______________________________________________________________________ -->
3654<div class="doc_subsubsection">
3655 <a name="i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a>
3656</div>
3657
3658<div class="doc_text">
3659
3660<h5>Syntax:</h5>
3661
3662<pre>
3663 &lt;result&gt; = fptrunc &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3664</pre>
3665
3666<h5>Overview:</h5>
3667<p>The '<tt>fptrunc</tt>' instruction truncates <tt>value</tt> to type
3668<tt>ty2</tt>.</p>
3669
3670
3671<h5>Arguments:</h5>
3672<p>The '<tt>fptrunc</tt>' instruction takes a <a href="#t_floating">floating
3673 point</a> value to cast and a <a href="#t_floating">floating point</a> type to
3674cast it to. The size of <tt>value</tt> must be larger than the size of
3675<tt>ty2</tt>. This implies that <tt>fptrunc</tt> cannot be used to make a
3676<i>no-op cast</i>.</p>
3677
3678<h5>Semantics:</h5>
3679<p> The '<tt>fptrunc</tt>' instruction truncates a <tt>value</tt> from a larger
3680<a href="#t_floating">floating point</a> type to a smaller
3681<a href="#t_floating">floating point</a> type. If the value cannot fit within
3682the destination type, <tt>ty2</tt>, then the results are undefined.</p>
3683
3684<h5>Example:</h5>
3685<pre>
3686 %X = fptrunc double 123.0 to float <i>; yields float:123.0</i>
3687 %Y = fptrunc double 1.0E+300 to float <i>; yields undefined</i>
3688</pre>
3689</div>
3690
3691<!-- _______________________________________________________________________ -->
3692<div class="doc_subsubsection">
3693 <a name="i_fpext">'<tt>fpext .. to</tt>' Instruction</a>
3694</div>
3695<div class="doc_text">
3696
3697<h5>Syntax:</h5>
3698<pre>
3699 &lt;result&gt; = fpext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3700</pre>
3701
3702<h5>Overview:</h5>
3703<p>The '<tt>fpext</tt>' extends a floating point <tt>value</tt> to a larger
3704floating point value.</p>
3705
3706<h5>Arguments:</h5>
3707<p>The '<tt>fpext</tt>' instruction takes a
3708<a href="#t_floating">floating point</a> <tt>value</tt> to cast,
3709and a <a href="#t_floating">floating point</a> type to cast it to. The source
3710type must be smaller than the destination type.</p>
3711
3712<h5>Semantics:</h5>
3713<p>The '<tt>fpext</tt>' instruction extends the <tt>value</tt> from a smaller
3714<a href="#t_floating">floating point</a> type to a larger
3715<a href="#t_floating">floating point</a> type. The <tt>fpext</tt> cannot be
3716used to make a <i>no-op cast</i> because it always changes bits. Use
3717<tt>bitcast</tt> to make a <i>no-op cast</i> for a floating point cast.</p>
3718
3719<h5>Example:</h5>
3720<pre>
3721 %X = fpext float 3.1415 to double <i>; yields double:3.1415</i>
3722 %Y = fpext float 1.0 to float <i>; yields float:1.0 (no-op)</i>
3723</pre>
3724</div>
3725
3726<!-- _______________________________________________________________________ -->
3727<div class="doc_subsubsection">
3728 <a name="i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a>
3729</div>
3730<div class="doc_text">
3731
3732<h5>Syntax:</h5>
3733<pre>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3734 &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]3735</pre>
3736
3737<h5>Overview:</h5>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3738<p>The '<tt>fptoui</tt>' converts a floating point <tt>value</tt> to its
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3739unsigned integer equivalent of type <tt>ty2</tt>.
3740</p>
3741
3742<h5>Arguments:</h5>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3743<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]3744scalar or vector <a href="#t_floating">floating point</a> value, and a type
3745to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
3746type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
3747vector integer type with the same number of elements as <tt>ty</tt></p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3748
3749<h5>Semantics:</h5>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3750<p> The '<tt>fptoui</tt>' instruction converts its
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3751<a href="#t_floating">floating point</a> operand into the nearest (rounding
3752towards zero) unsigned integer value. If the value cannot fit in <tt>ty2</tt>,
3753the results are undefined.</p>
3754
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3755<h5>Example:</h5>
3756<pre>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3757 %X = fptoui double 123.0 to i32 <i>; yields i32:123</i>
Chris Lattner681f1e82007-09-22 03:17:52 +0000[diff] [blame]3758 %Y = fptoui float 1.0E+300 to i1 <i>; yields undefined:1</i>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]3759 %X = fptoui float 1.04E+17 to i8 <i>; yields undefined:1</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3760</pre>
3761</div>
3762
3763<!-- _______________________________________________________________________ -->
3764<div class="doc_subsubsection">
3765 <a name="i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a>
3766</div>
3767<div class="doc_text">
3768
3769<h5>Syntax:</h5>
3770<pre>
3771 &lt;result&gt; = fptosi &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3772</pre>
3773
3774<h5>Overview:</h5>
3775<p>The '<tt>fptosi</tt>' instruction converts
3776<a href="#t_floating">floating point</a> <tt>value</tt> to type <tt>ty2</tt>.
3777</p>
3778
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3779<h5>Arguments:</h5>
3780<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]3781scalar or vector <a href="#t_floating">floating point</a> value, and a type
3782to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
3783type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
3784vector integer type with the same number of elements as <tt>ty</tt></p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3785
3786<h5>Semantics:</h5>
3787<p>The '<tt>fptosi</tt>' instruction converts its
3788<a href="#t_floating">floating point</a> operand into the nearest (rounding
3789towards zero) signed integer value. If the value cannot fit in <tt>ty2</tt>,
3790the results are undefined.</p>
3791
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3792<h5>Example:</h5>
3793<pre>
3794 %X = fptosi double -123.0 to i32 <i>; yields i32:-123</i>
Chris Lattner681f1e82007-09-22 03:17:52 +0000[diff] [blame]3795 %Y = fptosi float 1.0E-247 to i1 <i>; yields undefined:1</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3796 %X = fptosi float 1.04E+17 to i8 <i>; yields undefined:1</i>
3797</pre>
3798</div>
3799
3800<!-- _______________________________________________________________________ -->
3801<div class="doc_subsubsection">
3802 <a name="i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a>
3803</div>
3804<div class="doc_text">
3805
3806<h5>Syntax:</h5>
3807<pre>
3808 &lt;result&gt; = uitofp &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3809</pre>
3810
3811<h5>Overview:</h5>
3812<p>The '<tt>uitofp</tt>' instruction regards <tt>value</tt> as an unsigned
3813integer and converts that value to the <tt>ty2</tt> type.</p>
3814
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3815<h5>Arguments:</h5>
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]3816<p>The '<tt>uitofp</tt>' instruction takes a value to cast, which must be a
3817scalar or vector <a href="#t_integer">integer</a> value, and a type to cast it
3818to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
3819type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
3820floating point type with the same number of elements as <tt>ty</tt></p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3821
3822<h5>Semantics:</h5>
3823<p>The '<tt>uitofp</tt>' instruction interprets its operand as an unsigned
3824integer quantity and converts it to the corresponding floating point value. If
3825the value cannot fit in the floating point value, the results are undefined.</p>
3826
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3827<h5>Example:</h5>
3828<pre>
3829 %X = uitofp i32 257 to float <i>; yields float:257.0</i>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]3830 %Y = uitofp i8 -1 to double <i>; yields double:255.0</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3831</pre>
3832</div>
3833
3834<!-- _______________________________________________________________________ -->
3835<div class="doc_subsubsection">
3836 <a name="i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a>
3837</div>
3838<div class="doc_text">
3839
3840<h5>Syntax:</h5>
3841<pre>
3842 &lt;result&gt; = sitofp &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3843</pre>
3844
3845<h5>Overview:</h5>
3846<p>The '<tt>sitofp</tt>' instruction regards <tt>value</tt> as a signed
3847integer and converts that value to the <tt>ty2</tt> type.</p>
3848
3849<h5>Arguments:</h5>
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]3850<p>The '<tt>sitofp</tt>' instruction takes a value to cast, which must be a
3851scalar or vector <a href="#t_integer">integer</a> value, and a type to cast it
3852to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
3853type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
3854floating point type with the same number of elements as <tt>ty</tt></p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3855
3856<h5>Semantics:</h5>
3857<p>The '<tt>sitofp</tt>' instruction interprets its operand as a signed
3858integer quantity and converts it to the corresponding floating point value. If
3859the value cannot fit in the floating point value, the results are undefined.</p>
3860
3861<h5>Example:</h5>
3862<pre>
3863 %X = sitofp i32 257 to float <i>; yields float:257.0</i>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]3864 %Y = sitofp i8 -1 to double <i>; yields double:-1.0</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3865</pre>
3866</div>
3867
3868<!-- _______________________________________________________________________ -->
3869<div class="doc_subsubsection">
3870 <a name="i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a>
3871</div>
3872<div class="doc_text">
3873
3874<h5>Syntax:</h5>
3875<pre>
3876 &lt;result&gt; = ptrtoint &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3877</pre>
3878
3879<h5>Overview:</h5>
3880<p>The '<tt>ptrtoint</tt>' instruction converts the pointer <tt>value</tt> to
3881the integer type <tt>ty2</tt>.</p>
3882
3883<h5>Arguments:</h5>
3884<p>The '<tt>ptrtoint</tt>' instruction takes a <tt>value</tt> to cast, which
3885must be a <a href="#t_pointer">pointer</a> value, and a type to cast it to
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]3886<tt>ty2</tt>, which must be an <a href="#t_integer">integer</a> type.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3887
3888<h5>Semantics:</h5>
3889<p>The '<tt>ptrtoint</tt>' instruction converts <tt>value</tt> to integer type
3890<tt>ty2</tt> by interpreting the pointer value as an integer and either
3891truncating or zero extending that value to the size of the integer type. If
3892<tt>value</tt> is smaller than <tt>ty2</tt> then a zero extension is done. If
3893<tt>value</tt> is larger than <tt>ty2</tt> then a truncation is done. If they
3894are the same size, then nothing is done (<i>no-op cast</i>) other than a type
3895change.</p>
3896
3897<h5>Example:</h5>
3898<pre>
3899 %X = ptrtoint i32* %X to i8 <i>; yields truncation on 32-bit architecture</i>
3900 %Y = ptrtoint i32* %x to i64 <i>; yields zero extension on 32-bit architecture</i>
3901</pre>
3902</div>
3903
3904<!-- _______________________________________________________________________ -->
3905<div class="doc_subsubsection">
3906 <a name="i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a>
3907</div>
3908<div class="doc_text">
3909
3910<h5>Syntax:</h5>
3911<pre>
3912 &lt;result&gt; = inttoptr &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3913</pre>
3914
3915<h5>Overview:</h5>
3916<p>The '<tt>inttoptr</tt>' instruction converts an integer <tt>value</tt> to
3917a pointer type, <tt>ty2</tt>.</p>
3918
3919<h5>Arguments:</h5>
3920<p>The '<tt>inttoptr</tt>' instruction takes an <a href="#t_integer">integer</a>
3921value to cast, and a type to cast it to, which must be a
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]3922<a href="#t_pointer">pointer</a> type.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3923
3924<h5>Semantics:</h5>
3925<p>The '<tt>inttoptr</tt>' instruction converts <tt>value</tt> to type
3926<tt>ty2</tt> by applying either a zero extension or a truncation depending on
3927the size of the integer <tt>value</tt>. If <tt>value</tt> is larger than the
3928size of a pointer then a truncation is done. If <tt>value</tt> is smaller than
3929the size of a pointer then a zero extension is done. If they are the same size,
3930nothing is done (<i>no-op cast</i>).</p>
3931
3932<h5>Example:</h5>
3933<pre>
3934 %X = inttoptr i32 255 to i32* <i>; yields zero extension on 64-bit architecture</i>
3935 %X = inttoptr i32 255 to i32* <i>; yields no-op on 32-bit architecture</i>
3936 %Y = inttoptr i64 0 to i32* <i>; yields truncation on 32-bit architecture</i>
3937</pre>
3938</div>
3939
3940<!-- _______________________________________________________________________ -->
3941<div class="doc_subsubsection">
3942 <a name="i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a>
3943</div>
3944<div class="doc_text">
3945
3946<h5>Syntax:</h5>
3947<pre>
3948 &lt;result&gt; = bitcast &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3949</pre>
3950
3951<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3952
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3953<p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
3954<tt>ty2</tt> without changing any bits.</p>
3955
3956<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3957
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3958<p>The '<tt>bitcast</tt>' instruction takes a value to cast, which must be
Dan Gohman7305fa02008-09-08 16:45:59 +0000[diff] [blame]3959a non-aggregate first class value, and a type to cast it to, which must also be
3960a non-aggregate <a href="#t_firstclass">first class</a> type. The bit sizes of
3961<tt>value</tt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3962and the destination type, <tt>ty2</tt>, must be identical. If the source
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3963type is a pointer, the destination type must also be a pointer. This
3964instruction supports bitwise conversion of vectors to integers and to vectors
3965of other types (as long as they have the same size).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3966
3967<h5>Semantics:</h5>
3968<p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
3969<tt>ty2</tt>. It is always a <i>no-op cast</i> because no bits change with
3970this conversion. The conversion is done as if the <tt>value</tt> had been
3971stored to memory and read back as type <tt>ty2</tt>. Pointer types may only be
3972converted to other pointer types with this instruction. To convert pointers to
3973other types, use the <a href="#i_inttoptr">inttoptr</a> or
3974<a href="#i_ptrtoint">ptrtoint</a> instructions first.</p>
3975
3976<h5>Example:</h5>
3977<pre>
3978 %X = bitcast i8 255 to i8 <i>; yields i8 :-1</i>
3979 %Y = bitcast i32* %x to sint* <i>; yields sint*:%x</i>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]3980 %Z = bitcast &lt;2 x int&gt; %V to i64; <i>; yields i64: %V</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3981</pre>
3982</div>
3983
3984<!-- ======================================================================= -->
3985<div class="doc_subsection"> <a name="otherops">Other Operations</a> </div>
3986<div class="doc_text">
3987<p>The instructions in this category are the "miscellaneous"
3988instructions, which defy better classification.</p>
3989</div>
3990
3991<!-- _______________________________________________________________________ -->
3992<div class="doc_subsubsection"><a name="i_icmp">'<tt>icmp</tt>' Instruction</a>
3993</div>
3994<div class="doc_text">
3995<h5>Syntax:</h5>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]3996<pre> &lt;result&gt; = icmp &lt;cond&gt; &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {i1} or {&lt;N x i1&gt;}:result</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3997</pre>
3998<h5>Overview:</h5>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]3999<p>The '<tt>icmp</tt>' instruction returns a boolean value or
4000a vector of boolean values based on comparison
4001of its two integer, integer vector, or pointer operands.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4002<h5>Arguments:</h5>
4003<p>The '<tt>icmp</tt>' instruction takes three operands. The first operand is
4004the condition code indicating the kind of comparison to perform. It is not
4005a value, just a keyword. The possible condition code are:
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4006</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4007<ol>
4008 <li><tt>eq</tt>: equal</li>
4009 <li><tt>ne</tt>: not equal </li>
4010 <li><tt>ugt</tt>: unsigned greater than</li>
4011 <li><tt>uge</tt>: unsigned greater or equal</li>
4012 <li><tt>ult</tt>: unsigned less than</li>
4013 <li><tt>ule</tt>: unsigned less or equal</li>
4014 <li><tt>sgt</tt>: signed greater than</li>
4015 <li><tt>sge</tt>: signed greater or equal</li>
4016 <li><tt>slt</tt>: signed less than</li>
4017 <li><tt>sle</tt>: signed less or equal</li>
4018</ol>
4019<p>The remaining two arguments must be <a href="#t_integer">integer</a> or
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4020<a href="#t_pointer">pointer</a>
4021or integer <a href="#t_vector">vector</a> typed.
4022They must also be identical types.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4023<h5>Semantics:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4024<p>The '<tt>icmp</tt>' compares <tt>op1</tt> and <tt>op2</tt> according to
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4025the condition code given as <tt>cond</tt>. The comparison performed always
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4026yields either an <a href="#t_primitive"><tt>i1</tt></a> or vector of <tt>i1</tt> result, as follows:
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4027</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4028<ol>
4029 <li><tt>eq</tt>: yields <tt>true</tt> if the operands are equal,
4030 <tt>false</tt> otherwise. No sign interpretation is necessary or performed.
4031 </li>
4032 <li><tt>ne</tt>: yields <tt>true</tt> if the operands are unequal,
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4033 <tt>false</tt> otherwise. No sign interpretation is necessary or performed.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4034 <li><tt>ugt</tt>: interprets the operands as unsigned values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4035 <tt>true</tt> if <tt>op1</tt> is greater than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4036 <li><tt>uge</tt>: interprets the operands as unsigned values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4037 <tt>true</tt> if <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4038 <li><tt>ult</tt>: interprets the operands as unsigned values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4039 <tt>true</tt> if <tt>op1</tt> is less than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4040 <li><tt>ule</tt>: interprets the operands as unsigned values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4041 <tt>true</tt> if <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4042 <li><tt>sgt</tt>: interprets the operands as signed values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4043 <tt>true</tt> if <tt>op1</tt> is greater than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4044 <li><tt>sge</tt>: interprets the operands as signed values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4045 <tt>true</tt> if <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4046 <li><tt>slt</tt>: interprets the operands as signed values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4047 <tt>true</tt> if <tt>op1</tt> is less than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4048 <li><tt>sle</tt>: interprets the operands as signed values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4049 <tt>true</tt> if <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4050</ol>
4051<p>If the operands are <a href="#t_pointer">pointer</a> typed, the pointer
4052values are compared as if they were integers.</p>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4053<p>If the operands are integer vectors, then they are compared
4054element by element. The result is an <tt>i1</tt> vector with
4055the same number of elements as the values being compared.
4056Otherwise, the result is an <tt>i1</tt>.
4057</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4058
4059<h5>Example:</h5>
4060<pre> &lt;result&gt; = icmp eq i32 4, 5 <i>; yields: result=false</i>
4061 &lt;result&gt; = icmp ne float* %X, %X <i>; yields: result=false</i>
4062 &lt;result&gt; = icmp ult i16 4, 5 <i>; yields: result=true</i>
4063 &lt;result&gt; = icmp sgt i16 4, 5 <i>; yields: result=false</i>
4064 &lt;result&gt; = icmp ule i16 -4, 5 <i>; yields: result=false</i>
4065 &lt;result&gt; = icmp sge i16 4, 5 <i>; yields: result=false</i>
4066</pre>
4067</div>
4068
4069<!-- _______________________________________________________________________ -->
4070<div class="doc_subsubsection"><a name="i_fcmp">'<tt>fcmp</tt>' Instruction</a>
4071</div>
4072<div class="doc_text">
4073<h5>Syntax:</h5>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4074<pre> &lt;result&gt; = fcmp &lt;cond&gt; &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {i1} or {&lt;N x i1&gt;}:result</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4075</pre>
4076<h5>Overview:</h5>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4077<p>The '<tt>fcmp</tt>' instruction returns a boolean value
4078or vector of boolean values based on comparison
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4079of its operands.</p>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4080<p>
4081If the operands are floating point scalars, then the result
4082type is a boolean (<a href="#t_primitive"><tt>i1</tt></a>).
4083</p>
4084<p>If the operands are floating point vectors, then the result type
4085is a vector of boolean with the same number of elements as the
4086operands being compared.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4087<h5>Arguments:</h5>
4088<p>The '<tt>fcmp</tt>' instruction takes three operands. The first operand is
4089the condition code indicating the kind of comparison to perform. It is not
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4090a value, just a keyword. The possible condition code are:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4091<ol>
4092 <li><tt>false</tt>: no comparison, always returns false</li>
4093 <li><tt>oeq</tt>: ordered and equal</li>
4094 <li><tt>ogt</tt>: ordered and greater than </li>
4095 <li><tt>oge</tt>: ordered and greater than or equal</li>
4096 <li><tt>olt</tt>: ordered and less than </li>
4097 <li><tt>ole</tt>: ordered and less than or equal</li>
4098 <li><tt>one</tt>: ordered and not equal</li>
4099 <li><tt>ord</tt>: ordered (no nans)</li>
4100 <li><tt>ueq</tt>: unordered or equal</li>
4101 <li><tt>ugt</tt>: unordered or greater than </li>
4102 <li><tt>uge</tt>: unordered or greater than or equal</li>
4103 <li><tt>ult</tt>: unordered or less than </li>
4104 <li><tt>ule</tt>: unordered or less than or equal</li>
4105 <li><tt>une</tt>: unordered or not equal</li>
4106 <li><tt>uno</tt>: unordered (either nans)</li>
4107 <li><tt>true</tt>: no comparison, always returns true</li>
4108</ol>
4109<p><i>Ordered</i> means that neither operand is a QNAN while
4110<i>unordered</i> means that either operand may be a QNAN.</p>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4111<p>Each of <tt>val1</tt> and <tt>val2</tt> arguments must be
4112either a <a href="#t_floating">floating point</a> type
4113or a <a href="#t_vector">vector</a> of floating point type.
4114They must have identical types.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4115<h5>Semantics:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4116<p>The '<tt>fcmp</tt>' instruction compares <tt>op1</tt> and <tt>op2</tt>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4117according to the condition code given as <tt>cond</tt>.
4118If the operands are vectors, then the vectors are compared
4119element by element.
4120Each comparison performed
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4121always yields an <a href="#t_primitive">i1</a> result, as follows:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4122<ol>
4123 <li><tt>false</tt>: always yields <tt>false</tt>, regardless of operands.</li>
4124 <li><tt>oeq</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4125 <tt>op1</tt> is equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4126 <li><tt>ogt</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4127 <tt>op1</tt> is greather than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4128 <li><tt>oge</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4129 <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4130 <li><tt>olt</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4131 <tt>op1</tt> is less than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4132 <li><tt>ole</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4133 <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4134 <li><tt>one</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4135 <tt>op1</tt> is not equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4136 <li><tt>ord</tt>: yields <tt>true</tt> if both operands are not a QNAN.</li>
4137 <li><tt>ueq</tt>: yields <tt>true</tt> if either operand is a QNAN or
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4138 <tt>op1</tt> is equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4139 <li><tt>ugt</tt>: yields <tt>true</tt> if either operand is a QNAN or
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4140 <tt>op1</tt> is greater than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4141 <li><tt>uge</tt>: yields <tt>true</tt> if either operand is a QNAN or
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4142 <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4143 <li><tt>ult</tt>: yields <tt>true</tt> if either operand is a QNAN or
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4144 <tt>op1</tt> is less than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4145 <li><tt>ule</tt>: yields <tt>true</tt> if either operand is a QNAN or
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4146 <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4147 <li><tt>une</tt>: yields <tt>true</tt> if either operand is a QNAN or
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4148 <tt>op1</tt> is not equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4149 <li><tt>uno</tt>: yields <tt>true</tt> if either operand is a QNAN.</li>
4150 <li><tt>true</tt>: always yields <tt>true</tt>, regardless of operands.</li>
4151</ol>
4152
4153<h5>Example:</h5>
4154<pre> &lt;result&gt; = fcmp oeq float 4.0, 5.0 <i>; yields: result=false</i>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4155 &lt;result&gt; = fcmp one float 4.0, 5.0 <i>; yields: result=true</i>
4156 &lt;result&gt; = fcmp olt float 4.0, 5.0 <i>; yields: result=true</i>
4157 &lt;result&gt; = fcmp ueq double 1.0, 2.0 <i>; yields: result=false</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4158</pre>
4159</div>
4160
4161<!-- _______________________________________________________________________ -->
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]4162<div class="doc_subsubsection">
4163 <a name="i_vicmp">'<tt>vicmp</tt>' Instruction</a>
4164</div>
4165<div class="doc_text">
4166<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4167<pre> &lt;result&gt; = vicmp &lt;cond&gt; &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]4168</pre>
4169<h5>Overview:</h5>
4170<p>The '<tt>vicmp</tt>' instruction returns an integer vector value based on
4171element-wise comparison of its two integer vector operands.</p>
4172<h5>Arguments:</h5>
4173<p>The '<tt>vicmp</tt>' instruction takes three operands. The first operand is
4174the condition code indicating the kind of comparison to perform. It is not
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4175a value, just a keyword. The possible condition code are:</p>
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]4176<ol>
4177 <li><tt>eq</tt>: equal</li>
4178 <li><tt>ne</tt>: not equal </li>
4179 <li><tt>ugt</tt>: unsigned greater than</li>
4180 <li><tt>uge</tt>: unsigned greater or equal</li>
4181 <li><tt>ult</tt>: unsigned less than</li>
4182 <li><tt>ule</tt>: unsigned less or equal</li>
4183 <li><tt>sgt</tt>: signed greater than</li>
4184 <li><tt>sge</tt>: signed greater or equal</li>
4185 <li><tt>slt</tt>: signed less than</li>
4186 <li><tt>sle</tt>: signed less or equal</li>
4187</ol>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4188<p>The remaining two arguments must be <a href="#t_vector">vector</a> or
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]4189<a href="#t_integer">integer</a> typed. They must also be identical types.</p>
4190<h5>Semantics:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4191<p>The '<tt>vicmp</tt>' instruction compares <tt>op1</tt> and <tt>op2</tt>
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]4192according to the condition code given as <tt>cond</tt>. The comparison yields a
4193<a href="#t_vector">vector</a> of <a href="#t_integer">integer</a> result, of
4194identical type as the values being compared. The most significant bit in each
4195element is 1 if the element-wise comparison evaluates to true, and is 0
4196otherwise. All other bits of the result are undefined. The condition codes
4197are evaluated identically to the <a href="#i_icmp">'<tt>icmp</tt>'
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4198instruction</a>.</p>
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]4199
4200<h5>Example:</h5>
4201<pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4202 &lt;result&gt; = vicmp eq &lt;2 x i32&gt; &lt; i32 4, i32 0&gt;, &lt; i32 5, i32 0&gt; <i>; yields: result=&lt;2 x i32&gt; &lt; i32 0, i32 -1 &gt;</i>
4203 &lt;result&gt; = vicmp ult &lt;2 x i8 &gt; &lt; i8 1, i8 2&gt;, &lt; i8 2, i8 2 &gt; <i>; yields: result=&lt;2 x i8&gt; &lt; i8 -1, i8 0 &gt;</i>
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]4204</pre>
4205</div>
4206
4207<!-- _______________________________________________________________________ -->
4208<div class="doc_subsubsection">
4209 <a name="i_vfcmp">'<tt>vfcmp</tt>' Instruction</a>
4210</div>
4211<div class="doc_text">
4212<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4213<pre> &lt;result&gt; = vfcmp &lt;cond&gt; &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt;</pre>
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]4214<h5>Overview:</h5>
4215<p>The '<tt>vfcmp</tt>' instruction returns an integer vector value based on
4216element-wise comparison of its two floating point vector operands. The output
4217elements have the same width as the input elements.</p>
4218<h5>Arguments:</h5>
4219<p>The '<tt>vfcmp</tt>' instruction takes three operands. The first operand is
4220the condition code indicating the kind of comparison to perform. It is not
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4221a value, just a keyword. The possible condition code are:</p>
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]4222<ol>
4223 <li><tt>false</tt>: no comparison, always returns false</li>
4224 <li><tt>oeq</tt>: ordered and equal</li>
4225 <li><tt>ogt</tt>: ordered and greater than </li>
4226 <li><tt>oge</tt>: ordered and greater than or equal</li>
4227 <li><tt>olt</tt>: ordered and less than </li>
4228 <li><tt>ole</tt>: ordered and less than or equal</li>
4229 <li><tt>one</tt>: ordered and not equal</li>
4230 <li><tt>ord</tt>: ordered (no nans)</li>
4231 <li><tt>ueq</tt>: unordered or equal</li>
4232 <li><tt>ugt</tt>: unordered or greater than </li>
4233 <li><tt>uge</tt>: unordered or greater than or equal</li>
4234 <li><tt>ult</tt>: unordered or less than </li>
4235 <li><tt>ule</tt>: unordered or less than or equal</li>
4236 <li><tt>une</tt>: unordered or not equal</li>
4237 <li><tt>uno</tt>: unordered (either nans)</li>
4238 <li><tt>true</tt>: no comparison, always returns true</li>
4239</ol>
4240<p>The remaining two arguments must be <a href="#t_vector">vector</a> of
4241<a href="#t_floating">floating point</a> typed. They must also be identical
4242types.</p>
4243<h5>Semantics:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4244<p>The '<tt>vfcmp</tt>' instruction compares <tt>op1</tt> and <tt>op2</tt>
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]4245according to the condition code given as <tt>cond</tt>. The comparison yields a
4246<a href="#t_vector">vector</a> of <a href="#t_integer">integer</a> result, with
4247an identical number of elements as the values being compared, and each element
4248having identical with to the width of the floating point elements. The most
4249significant bit in each element is 1 if the element-wise comparison evaluates to
4250true, and is 0 otherwise. All other bits of the result are undefined. The
4251condition codes are evaluated identically to the
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4252<a href="#i_fcmp">'<tt>fcmp</tt>' instruction</a>.</p>
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]4253
4254<h5>Example:</h5>
4255<pre>
Chris Lattner1e5c5cd02008-10-13 16:55:18 +0000[diff] [blame]4256 <i>; yields: result=&lt;2 x i32&gt; &lt; i32 0, i32 -1 &gt;</i>
4257 &lt;result&gt; = vfcmp oeq &lt;2 x float&gt; &lt; float 4, float 0 &gt;, &lt; float 5, float 0 &gt;
4258
4259 <i>; yields: result=&lt;2 x i64&gt; &lt; i64 -1, i64 0 &gt;</i>
4260 &lt;result&gt; = vfcmp ult &lt;2 x double&gt; &lt; double 1, double 2 &gt;, &lt; double 2, double 2&gt;
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]4261</pre>
4262</div>
4263
4264<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4265<div class="doc_subsubsection">
4266 <a name="i_phi">'<tt>phi</tt>' Instruction</a>
4267</div>
4268
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4269<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4270
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4271<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4272
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4273<pre> &lt;result&gt; = phi &lt;ty&gt; [ &lt;val0&gt;, &lt;label0&gt;], ...<br></pre>
4274<h5>Overview:</h5>
4275<p>The '<tt>phi</tt>' instruction is used to implement the &#966; node in
4276the SSA graph representing the function.</p>
4277<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4278
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4279<p>The type of the incoming values is specified with the first type
4280field. After this, the '<tt>phi</tt>' instruction takes a list of pairs
4281as arguments, with one pair for each predecessor basic block of the
4282current block. Only values of <a href="#t_firstclass">first class</a>
4283type may be used as the value arguments to the PHI node. Only labels
4284may be used as the label arguments.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4285
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4286<p>There must be no non-phi instructions between the start of a basic
4287block and the PHI instructions: i.e. PHI instructions must be first in
4288a basic block.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4289
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4290<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4291
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4292<p>At runtime, the '<tt>phi</tt>' instruction logically takes on the value
4293specified by the pair corresponding to the predecessor basic block that executed
4294just prior to the current block.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4295
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4296<h5>Example:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4297<pre>
4298Loop: ; Infinite loop that counts from 0 on up...
4299 %indvar = phi i32 [ 0, %LoopHeader ], [ %nextindvar, %Loop ]
4300 %nextindvar = add i32 %indvar, 1
4301 br label %Loop
4302</pre>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4303</div>
4304
4305<!-- _______________________________________________________________________ -->
4306<div class="doc_subsubsection">
4307 <a name="i_select">'<tt>select</tt>' Instruction</a>
4308</div>
4309
4310<div class="doc_text">
4311
4312<h5>Syntax:</h5>
4313
4314<pre>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4315 &lt;result&gt; = select <i>selty</i> &lt;cond&gt;, &lt;ty&gt; &lt;val1&gt;, &lt;ty&gt; &lt;val2&gt; <i>; yields ty</i>
4316
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4317 <i>selty</i> is either i1 or {&lt;N x i1&gt;}
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4318</pre>
4319
4320<h5>Overview:</h5>
4321
4322<p>
4323The '<tt>select</tt>' instruction is used to choose one value based on a
4324condition, without branching.
4325</p>
4326
4327
4328<h5>Arguments:</h5>
4329
4330<p>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4331The '<tt>select</tt>' instruction requires an 'i1' value or
4332a vector of 'i1' values indicating the
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4333condition, and two values of the same <a href="#t_firstclass">first class</a>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4334type. If the val1/val2 are vectors and
4335the condition is a scalar, then entire vectors are selected, not
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4336individual elements.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4337</p>
4338
4339<h5>Semantics:</h5>
4340
4341<p>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4342If the condition is an i1 and it evaluates to 1, the instruction returns the first
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4343value argument; otherwise, it returns the second value argument.
4344</p>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4345<p>
4346If the condition is a vector of i1, then the value arguments must
4347be vectors of the same size, and the selection is done element
4348by element.
4349</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4350
4351<h5>Example:</h5>
4352
4353<pre>
4354 %X = select i1 true, i8 17, i8 42 <i>; yields i8:17</i>
4355</pre>
4356</div>
4357
4358
4359<!-- _______________________________________________________________________ -->
4360<div class="doc_subsubsection">
4361 <a name="i_call">'<tt>call</tt>' Instruction</a>
4362</div>
4363
4364<div class="doc_text">
4365
4366<h5>Syntax:</h5>
4367<pre>
Devang Pateld0bfcc72008-10-07 17:48:33 +0000[diff] [blame]4368 &lt;result&gt; = [tail] call [<a href="#callingconv">cconv</a>] [<a href="#paramattrs">ret attrs</a>] &lt;ty&gt; [&lt;fnty&gt;*] &lt;fnptrval&gt;(&lt;function args&gt;) [<a href="#fnattrs">fn attrs</a>]
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4369</pre>
4370
4371<h5>Overview:</h5>
4372
4373<p>The '<tt>call</tt>' instruction represents a simple function call.</p>
4374
4375<h5>Arguments:</h5>
4376
4377<p>This instruction requires several arguments:</p>
4378
4379<ol>
4380 <li>
4381 <p>The optional "tail" marker indicates whether the callee function accesses
4382 any allocas or varargs in the caller. If the "tail" marker is present, the
4383 function call is eligible for tail call optimization. Note that calls may
4384 be marked "tail" even if they do not occur before a <a
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4385 href="#i_ret"><tt>ret</tt></a> instruction.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4386 </li>
4387 <li>
4388 <p>The optional "cconv" marker indicates which <a href="#callingconv">calling
4389 convention</a> the call should use. If none is specified, the call defaults
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4390 to using C calling conventions.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4391 </li>
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]4392
4393 <li>
4394 <p>The optional <a href="#paramattrs">Parameter Attributes</a> list for
4395 return values. Only '<tt>zeroext</tt>', '<tt>signext</tt>',
4396 and '<tt>inreg</tt>' attributes are valid here.</p>
4397 </li>
4398
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4399 <li>
Nick Lewycky93082fc2007-09-08 13:57:50 +0000[diff] [blame]4400 <p>'<tt>ty</tt>': the type of the call instruction itself which is also
4401 the type of the return value. Functions that return no value are marked
4402 <tt><a href="#t_void">void</a></tt>.</p>
4403 </li>
4404 <li>
4405 <p>'<tt>fnty</tt>': shall be the signature of the pointer to function
4406 value being invoked. The argument types must match the types implied by
4407 this signature. This type can be omitted if the function is not varargs
4408 and if the function type does not return a pointer to a function.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4409 </li>
4410 <li>
4411 <p>'<tt>fnptrval</tt>': An LLVM value containing a pointer to a function to
4412 be invoked. In most cases, this is a direct function invocation, but
4413 indirect <tt>call</tt>s are just as possible, calling an arbitrary pointer
4414 to function value.</p>
4415 </li>
4416 <li>
4417 <p>'<tt>function args</tt>': argument list whose types match the
4418 function signature argument types. All arguments must be of
4419 <a href="#t_firstclass">first class</a> type. If the function signature
4420 indicates the function accepts a variable number of arguments, the extra
4421 arguments can be specified.</p>
4422 </li>
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]4423 <li>
Devang Pateld0bfcc72008-10-07 17:48:33 +0000[diff] [blame]4424 <p>The optional <a href="#fnattrs">function attributes</a> list. Only
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]4425 '<tt>noreturn</tt>', '<tt>nounwind</tt>', '<tt>readonly</tt>' and
4426 '<tt>readnone</tt>' attributes are valid here.</p>
4427 </li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4428</ol>
4429
4430<h5>Semantics:</h5>
4431
4432<p>The '<tt>call</tt>' instruction is used to cause control flow to
4433transfer to a specified function, with its incoming arguments bound to
4434the specified values. Upon a '<tt><a href="#i_ret">ret</a></tt>'
4435instruction in the called function, control flow continues with the
4436instruction after the function call, and the return value of the
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4437function is bound to the result argument.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4438
4439<h5>Example:</h5>
4440
4441<pre>
Nick Lewycky93082fc2007-09-08 13:57:50 +0000[diff] [blame]4442 %retval = call i32 @test(i32 %argc)
Chris Lattner5e893ef2008-03-21 17:24:17 +0000[diff] [blame]4443 call i32 (i8 *, ...)* @printf(i8 * %msg, i32 12, i8 42) <i>; yields i32</i>
4444 %X = tail call i32 @foo() <i>; yields i32</i>
4445 %Y = tail call <a href="#callingconv">fastcc</a> i32 @foo() <i>; yields i32</i>
4446 call void %foo(i8 97 signext)
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]4447
4448 %struct.A = type { i32, i8 }
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]4449 %r = call %struct.A @foo() <i>; yields { 32, i8 }</i>
Dan Gohman3e700032008-10-04 19:00:07 +0000[diff] [blame]4450 %gr = extractvalue %struct.A %r, 0 <i>; yields i32</i>
4451 %gr1 = extractvalue %struct.A %r, 1 <i>; yields i8</i>
Chris Lattnerac454b32008-10-08 06:26:11 +0000[diff] [blame]4452 %Z = call void @foo() noreturn <i>; indicates that %foo never returns normally</i>
Matthijs Kooijman2c4e05a2008-10-07 10:03:45 +0000[diff] [blame]4453 %ZZ = call zeroext i32 @bar() <i>; Return value is %zero extended</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4454</pre>
4455
4456</div>
4457
4458<!-- _______________________________________________________________________ -->
4459<div class="doc_subsubsection">
4460 <a name="i_va_arg">'<tt>va_arg</tt>' Instruction</a>
4461</div>
4462
4463<div class="doc_text">
4464
4465<h5>Syntax:</h5>
4466
4467<pre>
4468 &lt;resultval&gt; = va_arg &lt;va_list*&gt; &lt;arglist&gt;, &lt;argty&gt;
4469</pre>
4470
4471<h5>Overview:</h5>
4472
4473<p>The '<tt>va_arg</tt>' instruction is used to access arguments passed through
4474the "variable argument" area of a function call. It is used to implement the
4475<tt>va_arg</tt> macro in C.</p>
4476
4477<h5>Arguments:</h5>
4478
4479<p>This instruction takes a <tt>va_list*</tt> value and the type of
4480the argument. It returns a value of the specified argument type and
4481increments the <tt>va_list</tt> to point to the next argument. The
4482actual type of <tt>va_list</tt> is target specific.</p>
4483
4484<h5>Semantics:</h5>
4485
4486<p>The '<tt>va_arg</tt>' instruction loads an argument of the specified
4487type from the specified <tt>va_list</tt> and causes the
4488<tt>va_list</tt> to point to the next argument. For more information,
4489see the variable argument handling <a href="#int_varargs">Intrinsic
4490Functions</a>.</p>
4491
4492<p>It is legal for this instruction to be called in a function which does not
4493take a variable number of arguments, for example, the <tt>vfprintf</tt>
4494function.</p>
4495
4496<p><tt>va_arg</tt> is an LLVM instruction instead of an <a
4497href="#intrinsics">intrinsic function</a> because it takes a type as an
4498argument.</p>
4499
4500<h5>Example:</h5>
4501
4502<p>See the <a href="#int_varargs">variable argument processing</a> section.</p>
4503
Dan Gohman60967192009-01-12 23:12:39 +0000[diff] [blame]4504<p>Note that the code generator does not yet fully support va_arg
4505 on many targets. Also, it does not currently support va_arg with
4506 aggregate types on any target.</p>
4507
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4508</div>
4509
4510<!-- *********************************************************************** -->
4511<div class="doc_section"> <a name="intrinsics">Intrinsic Functions</a> </div>
4512<!-- *********************************************************************** -->
4513
4514<div class="doc_text">
4515
4516<p>LLVM supports the notion of an "intrinsic function". These functions have
4517well known names and semantics and are required to follow certain restrictions.
4518Overall, these intrinsics represent an extension mechanism for the LLVM
4519language that does not require changing all of the transformations in LLVM when
4520adding to the language (or the bitcode reader/writer, the parser, etc...).</p>
4521
4522<p>Intrinsic function names must all start with an "<tt>llvm.</tt>" prefix. This
4523prefix is reserved in LLVM for intrinsic names; thus, function names may not
4524begin with this prefix. Intrinsic functions must always be external functions:
4525you cannot define the body of intrinsic functions. Intrinsic functions may
4526only be used in call or invoke instructions: it is illegal to take the address
4527of an intrinsic function. Additionally, because intrinsic functions are part
4528of the LLVM language, it is required if any are added that they be documented
4529here.</p>
4530
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]4531<p>Some intrinsic functions can be overloaded, i.e., the intrinsic represents
4532a family of functions that perform the same operation but on different data
4533types. Because LLVM can represent over 8 million different integer types,
4534overloading is used commonly to allow an intrinsic function to operate on any
4535integer type. One or more of the argument types or the result type can be
4536overloaded to accept any integer type. Argument types may also be defined as
4537exactly matching a previous argument's type or the result type. This allows an
4538intrinsic function which accepts multiple arguments, but needs all of them to
4539be of the same type, to only be overloaded with respect to a single argument or
4540the result.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4541
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]4542<p>Overloaded intrinsics will have the names of its overloaded argument types
4543encoded into its function name, each preceded by a period. Only those types
4544which are overloaded result in a name suffix. Arguments whose type is matched
4545against another type do not. For example, the <tt>llvm.ctpop</tt> function can
4546take an integer of any width and returns an integer of exactly the same integer
4547width. This leads to a family of functions such as
4548<tt>i8 @llvm.ctpop.i8(i8 %val)</tt> and <tt>i29 @llvm.ctpop.i29(i29 %val)</tt>.
4549Only one type, the return type, is overloaded, and only one type suffix is
4550required. Because the argument's type is matched against the return type, it
4551does not require its own name suffix.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4552
4553<p>To learn how to add an intrinsic function, please see the
4554<a href="ExtendingLLVM.html">Extending LLVM Guide</a>.
4555</p>
4556
4557</div>
4558
4559<!-- ======================================================================= -->
4560<div class="doc_subsection">
4561 <a name="int_varargs">Variable Argument Handling Intrinsics</a>
4562</div>
4563
4564<div class="doc_text">
4565
4566<p>Variable argument support is defined in LLVM with the <a
4567 href="#i_va_arg"><tt>va_arg</tt></a> instruction and these three
4568intrinsic functions. These functions are related to the similarly
4569named macros defined in the <tt>&lt;stdarg.h&gt;</tt> header file.</p>
4570
4571<p>All of these functions operate on arguments that use a
4572target-specific value type "<tt>va_list</tt>". The LLVM assembly
4573language reference manual does not define what this type is, so all
4574transformations should be prepared to handle these functions regardless of
4575the type used.</p>
4576
4577<p>This example shows how the <a href="#i_va_arg"><tt>va_arg</tt></a>
4578instruction and the variable argument handling intrinsic functions are
4579used.</p>
4580
4581<div class="doc_code">
4582<pre>
4583define i32 @test(i32 %X, ...) {
4584 ; Initialize variable argument processing
4585 %ap = alloca i8*
4586 %ap2 = bitcast i8** %ap to i8*
4587 call void @llvm.va_start(i8* %ap2)
4588
4589 ; Read a single integer argument
4590 %tmp = va_arg i8** %ap, i32
4591
4592 ; Demonstrate usage of llvm.va_copy and llvm.va_end
4593 %aq = alloca i8*
4594 %aq2 = bitcast i8** %aq to i8*
4595 call void @llvm.va_copy(i8* %aq2, i8* %ap2)
4596 call void @llvm.va_end(i8* %aq2)
4597
4598 ; Stop processing of arguments.
4599 call void @llvm.va_end(i8* %ap2)
4600 ret i32 %tmp
4601}
4602
4603declare void @llvm.va_start(i8*)
4604declare void @llvm.va_copy(i8*, i8*)
4605declare void @llvm.va_end(i8*)
4606</pre>
4607</div>
4608
4609</div>
4610
4611<!-- _______________________________________________________________________ -->
4612<div class="doc_subsubsection">
4613 <a name="int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a>
4614</div>
4615
4616
4617<div class="doc_text">
4618<h5>Syntax:</h5>
4619<pre> declare void %llvm.va_start(i8* &lt;arglist&gt;)<br></pre>
4620<h5>Overview:</h5>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4621<p>The '<tt>llvm.va_start</tt>' intrinsic initializes
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4622<tt>*&lt;arglist&gt;</tt> for subsequent use by <tt><a
4623href="#i_va_arg">va_arg</a></tt>.</p>
4624
4625<h5>Arguments:</h5>
4626
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4627<p>The argument is a pointer to a <tt>va_list</tt> element to initialize.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4628
4629<h5>Semantics:</h5>
4630
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4631<p>The '<tt>llvm.va_start</tt>' intrinsic works just like the <tt>va_start</tt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4632macro available in C. In a target-dependent way, it initializes the
4633<tt>va_list</tt> element to which the argument points, so that the next call to
4634<tt>va_arg</tt> will produce the first variable argument passed to the function.
4635Unlike the C <tt>va_start</tt> macro, this intrinsic does not need to know the
4636last argument of the function as the compiler can figure that out.</p>
4637
4638</div>
4639
4640<!-- _______________________________________________________________________ -->
4641<div class="doc_subsubsection">
4642 <a name="int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a>
4643</div>
4644
4645<div class="doc_text">
4646<h5>Syntax:</h5>
4647<pre> declare void @llvm.va_end(i8* &lt;arglist&gt;)<br></pre>
4648<h5>Overview:</h5>
4649
4650<p>The '<tt>llvm.va_end</tt>' intrinsic destroys <tt>*&lt;arglist&gt;</tt>,
4651which has been initialized previously with <tt><a href="#int_va_start">llvm.va_start</a></tt>
4652or <tt><a href="#i_va_copy">llvm.va_copy</a></tt>.</p>
4653
4654<h5>Arguments:</h5>
4655
4656<p>The argument is a pointer to a <tt>va_list</tt> to destroy.</p>
4657
4658<h5>Semantics:</h5>
4659
4660<p>The '<tt>llvm.va_end</tt>' intrinsic works just like the <tt>va_end</tt>
4661macro available in C. In a target-dependent way, it destroys the
4662<tt>va_list</tt> element to which the argument points. Calls to <a
4663href="#int_va_start"><tt>llvm.va_start</tt></a> and <a href="#int_va_copy">
4664<tt>llvm.va_copy</tt></a> must be matched exactly with calls to
4665<tt>llvm.va_end</tt>.</p>
4666
4667</div>
4668
4669<!-- _______________________________________________________________________ -->
4670<div class="doc_subsubsection">
4671 <a name="int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a>
4672</div>
4673
4674<div class="doc_text">
4675
4676<h5>Syntax:</h5>
4677
4678<pre>
4679 declare void @llvm.va_copy(i8* &lt;destarglist&gt;, i8* &lt;srcarglist&gt;)
4680</pre>
4681
4682<h5>Overview:</h5>
4683
4684<p>The '<tt>llvm.va_copy</tt>' intrinsic copies the current argument position
4685from the source argument list to the destination argument list.</p>
4686
4687<h5>Arguments:</h5>
4688
4689<p>The first argument is a pointer to a <tt>va_list</tt> element to initialize.
4690The second argument is a pointer to a <tt>va_list</tt> element to copy from.</p>
4691
4692
4693<h5>Semantics:</h5>
4694
4695<p>The '<tt>llvm.va_copy</tt>' intrinsic works just like the <tt>va_copy</tt>
4696macro available in C. In a target-dependent way, it copies the source
4697<tt>va_list</tt> element into the destination <tt>va_list</tt> element. This
4698intrinsic is necessary because the <tt><a href="#int_va_start">
4699llvm.va_start</a></tt> intrinsic may be arbitrarily complex and require, for
4700example, memory allocation.</p>
4701
4702</div>
4703
4704<!-- ======================================================================= -->
4705<div class="doc_subsection">
4706 <a name="int_gc">Accurate Garbage Collection Intrinsics</a>
4707</div>
4708
4709<div class="doc_text">
4710
4711<p>
4712LLVM support for <a href="GarbageCollection.html">Accurate Garbage
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]4713Collection</a> (GC) requires the implementation and generation of these
4714intrinsics.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4715These intrinsics allow identification of <a href="#int_gcroot">GC roots on the
4716stack</a>, as well as garbage collector implementations that require <a
4717href="#int_gcread">read</a> and <a href="#int_gcwrite">write</a> barriers.
4718Front-ends for type-safe garbage collected languages should generate these
4719intrinsics to make use of the LLVM garbage collectors. For more details, see <a
4720href="GarbageCollection.html">Accurate Garbage Collection with LLVM</a>.
4721</p>
Christopher Lambcfe00962007-12-17 01:00:21 +0000[diff] [blame]4722
4723<p>The garbage collection intrinsics only operate on objects in the generic
4724 address space (address space zero).</p>
4725
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4726</div>
4727
4728<!-- _______________________________________________________________________ -->
4729<div class="doc_subsubsection">
4730 <a name="int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a>
4731</div>
4732
4733<div class="doc_text">
4734
4735<h5>Syntax:</h5>
4736
4737<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4738 declare void @llvm.gcroot(i8** %ptrloc, i8* %metadata)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4739</pre>
4740
4741<h5>Overview:</h5>
4742
4743<p>The '<tt>llvm.gcroot</tt>' intrinsic declares the existence of a GC root to
4744the code generator, and allows some metadata to be associated with it.</p>
4745
4746<h5>Arguments:</h5>
4747
4748<p>The first argument specifies the address of a stack object that contains the
4749root pointer. The second pointer (which must be either a constant or a global
4750value address) contains the meta-data to be associated with the root.</p>
4751
4752<h5>Semantics:</h5>
4753
Chris Lattnera7d94ba2008-04-24 05:59:56 +0000[diff] [blame]4754<p>At runtime, a call to this intrinsic stores a null pointer into the "ptrloc"
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4755location. At compile-time, the code generator generates information to allow
Gordon Henriksen40393542007-12-25 02:31:26 +0000[diff] [blame]4756the runtime to find the pointer at GC safe points. The '<tt>llvm.gcroot</tt>'
4757intrinsic may only be used in a function which <a href="#gc">specifies a GC
4758algorithm</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4759
4760</div>
4761
4762
4763<!-- _______________________________________________________________________ -->
4764<div class="doc_subsubsection">
4765 <a name="int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a>
4766</div>
4767
4768<div class="doc_text">
4769
4770<h5>Syntax:</h5>
4771
4772<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4773 declare i8* @llvm.gcread(i8* %ObjPtr, i8** %Ptr)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4774</pre>
4775
4776<h5>Overview:</h5>
4777
4778<p>The '<tt>llvm.gcread</tt>' intrinsic identifies reads of references from heap
4779locations, allowing garbage collector implementations that require read
4780barriers.</p>
4781
4782<h5>Arguments:</h5>
4783
4784<p>The second argument is the address to read from, which should be an address
4785allocated from the garbage collector. The first object is a pointer to the
4786start of the referenced object, if needed by the language runtime (otherwise
4787null).</p>
4788
4789<h5>Semantics:</h5>
4790
4791<p>The '<tt>llvm.gcread</tt>' intrinsic has the same semantics as a load
4792instruction, but may be replaced with substantially more complex code by the
Gordon Henriksen40393542007-12-25 02:31:26 +0000[diff] [blame]4793garbage collector runtime, as needed. The '<tt>llvm.gcread</tt>' intrinsic
4794may only be used in a function which <a href="#gc">specifies a GC
4795algorithm</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4796
4797</div>
4798
4799
4800<!-- _______________________________________________________________________ -->
4801<div class="doc_subsubsection">
4802 <a name="int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a>
4803</div>
4804
4805<div class="doc_text">
4806
4807<h5>Syntax:</h5>
4808
4809<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4810 declare void @llvm.gcwrite(i8* %P1, i8* %Obj, i8** %P2)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4811</pre>
4812
4813<h5>Overview:</h5>
4814
4815<p>The '<tt>llvm.gcwrite</tt>' intrinsic identifies writes of references to heap
4816locations, allowing garbage collector implementations that require write
4817barriers (such as generational or reference counting collectors).</p>
4818
4819<h5>Arguments:</h5>
4820
4821<p>The first argument is the reference to store, the second is the start of the
4822object to store it to, and the third is the address of the field of Obj to
4823store to. If the runtime does not require a pointer to the object, Obj may be
4824null.</p>
4825
4826<h5>Semantics:</h5>
4827
4828<p>The '<tt>llvm.gcwrite</tt>' intrinsic has the same semantics as a store
4829instruction, but may be replaced with substantially more complex code by the
Gordon Henriksen40393542007-12-25 02:31:26 +0000[diff] [blame]4830garbage collector runtime, as needed. The '<tt>llvm.gcwrite</tt>' intrinsic
4831may only be used in a function which <a href="#gc">specifies a GC
4832algorithm</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4833
4834</div>
4835
4836
4837
4838<!-- ======================================================================= -->
4839<div class="doc_subsection">
4840 <a name="int_codegen">Code Generator Intrinsics</a>
4841</div>
4842
4843<div class="doc_text">
4844<p>
4845These intrinsics are provided by LLVM to expose special features that may only
4846be implemented with code generator support.
4847</p>
4848
4849</div>
4850
4851<!-- _______________________________________________________________________ -->
4852<div class="doc_subsubsection">
4853 <a name="int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a>
4854</div>
4855
4856<div class="doc_text">
4857
4858<h5>Syntax:</h5>
4859<pre>
4860 declare i8 *@llvm.returnaddress(i32 &lt;level&gt;)
4861</pre>
4862
4863<h5>Overview:</h5>
4864
4865<p>
4866The '<tt>llvm.returnaddress</tt>' intrinsic attempts to compute a
4867target-specific value indicating the return address of the current function
4868or one of its callers.
4869</p>
4870
4871<h5>Arguments:</h5>
4872
4873<p>
4874The argument to this intrinsic indicates which function to return the address
4875for. Zero indicates the calling function, one indicates its caller, etc. The
4876argument is <b>required</b> to be a constant integer value.
4877</p>
4878
4879<h5>Semantics:</h5>
4880
4881<p>
4882The '<tt>llvm.returnaddress</tt>' intrinsic either returns a pointer indicating
4883the return address of the specified call frame, or zero if it cannot be
4884identified. The value returned by this intrinsic is likely to be incorrect or 0
4885for arguments other than zero, so it should only be used for debugging purposes.
4886</p>
4887
4888<p>
4889Note that calling this intrinsic does not prevent function inlining or other
4890aggressive transformations, so the value returned may not be that of the obvious
4891source-language caller.
4892</p>
4893</div>
4894
4895
4896<!-- _______________________________________________________________________ -->
4897<div class="doc_subsubsection">
4898 <a name="int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a>
4899</div>
4900
4901<div class="doc_text">
4902
4903<h5>Syntax:</h5>
4904<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4905 declare i8 *@llvm.frameaddress(i32 &lt;level&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4906</pre>
4907
4908<h5>Overview:</h5>
4909
4910<p>
4911The '<tt>llvm.frameaddress</tt>' intrinsic attempts to return the
4912target-specific frame pointer value for the specified stack frame.
4913</p>
4914
4915<h5>Arguments:</h5>
4916
4917<p>
4918The argument to this intrinsic indicates which function to return the frame
4919pointer for. Zero indicates the calling function, one indicates its caller,
4920etc. The argument is <b>required</b> to be a constant integer value.
4921</p>
4922
4923<h5>Semantics:</h5>
4924
4925<p>
4926The '<tt>llvm.frameaddress</tt>' intrinsic either returns a pointer indicating
4927the frame address of the specified call frame, or zero if it cannot be
4928identified. The value returned by this intrinsic is likely to be incorrect or 0
4929for arguments other than zero, so it should only be used for debugging purposes.
4930</p>
4931
4932<p>
4933Note that calling this intrinsic does not prevent function inlining or other
4934aggressive transformations, so the value returned may not be that of the obvious
4935source-language caller.
4936</p>
4937</div>
4938
4939<!-- _______________________________________________________________________ -->
4940<div class="doc_subsubsection">
4941 <a name="int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a>
4942</div>
4943
4944<div class="doc_text">
4945
4946<h5>Syntax:</h5>
4947<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4948 declare i8 *@llvm.stacksave()
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4949</pre>
4950
4951<h5>Overview:</h5>
4952
4953<p>
4954The '<tt>llvm.stacksave</tt>' intrinsic is used to remember the current state of
4955the function stack, for use with <a href="#int_stackrestore">
4956<tt>llvm.stackrestore</tt></a>. This is useful for implementing language
4957features like scoped automatic variable sized arrays in C99.
4958</p>
4959
4960<h5>Semantics:</h5>
4961
4962<p>
4963This intrinsic returns a opaque pointer value that can be passed to <a
4964href="#int_stackrestore"><tt>llvm.stackrestore</tt></a>. When an
4965<tt>llvm.stackrestore</tt> intrinsic is executed with a value saved from
4966<tt>llvm.stacksave</tt>, it effectively restores the state of the stack to the
4967state it was in when the <tt>llvm.stacksave</tt> intrinsic executed. In
4968practice, this pops any <a href="#i_alloca">alloca</a> blocks from the stack
4969that were allocated after the <tt>llvm.stacksave</tt> was executed.
4970</p>
4971
4972</div>
4973
4974<!-- _______________________________________________________________________ -->
4975<div class="doc_subsubsection">
4976 <a name="int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a>
4977</div>
4978
4979<div class="doc_text">
4980
4981<h5>Syntax:</h5>
4982<pre>
4983 declare void @llvm.stackrestore(i8 * %ptr)
4984</pre>
4985
4986<h5>Overview:</h5>
4987
4988<p>
4989The '<tt>llvm.stackrestore</tt>' intrinsic is used to restore the state of
4990the function stack to the state it was in when the corresponding <a
4991href="#int_stacksave"><tt>llvm.stacksave</tt></a> intrinsic executed. This is
4992useful for implementing language features like scoped automatic variable sized
4993arrays in C99.
4994</p>
4995
4996<h5>Semantics:</h5>
4997
4998<p>
4999See the description for <a href="#int_stacksave"><tt>llvm.stacksave</tt></a>.
5000</p>
5001
5002</div>
5003
5004
5005<!-- _______________________________________________________________________ -->
5006<div class="doc_subsubsection">
5007 <a name="int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a>
5008</div>
5009
5010<div class="doc_text">
5011
5012<h5>Syntax:</h5>
5013<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]5014 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]5015</pre>
5016
5017<h5>Overview:</h5>
5018
5019
5020<p>
5021The '<tt>llvm.prefetch</tt>' intrinsic is a hint to the code generator to insert
5022a prefetch instruction if supported; otherwise, it is a noop. Prefetches have
5023no
5024effect on the behavior of the program but can change its performance
5025characteristics.
5026</p>
5027
5028<h5>Arguments:</h5>
5029
5030<p>
5031<tt>address</tt> is the address to be prefetched, <tt>rw</tt> is the specifier
5032determining if the fetch should be for a read (0) or write (1), and
5033<tt>locality</tt> is a temporal locality specifier ranging from (0) - no
5034locality, to (3) - extremely local keep in cache. The <tt>rw</tt> and
5035<tt>locality</tt> arguments must be constant integers.
5036</p>
5037
5038<h5>Semantics:</h5>
5039
5040<p>
5041This intrinsic does not modify the behavior of the program. In particular,
5042prefetches cannot trap and do not produce a value. On targets that support this
5043intrinsic, the prefetch can provide hints to the processor cache for better
5044performance.
5045</p>
5046
5047</div>
5048
5049<!-- _______________________________________________________________________ -->
5050<div class="doc_subsubsection">
5051 <a name="int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a>
5052</div>
5053
5054<div class="doc_text">
5055
5056<h5>Syntax:</h5>
5057<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]5058 declare void @llvm.pcmarker(i32 &lt;id&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5059</pre>
5060
5061<h5>Overview:</h5>
5062
5063
5064<p>
5065The '<tt>llvm.pcmarker</tt>' intrinsic is a method to export a Program Counter
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]5066(PC) in a region of
5067code to simulators and other tools. The method is target specific, but it is
5068expected that the marker will use exported symbols to transmit the PC of the
5069marker.
5070The marker makes no guarantees that it will remain with any specific instruction
5071after optimizations. It is possible that the presence of a marker will inhibit
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5072optimizations. The intended use is to be inserted after optimizations to allow
5073correlations of simulation runs.
5074</p>
5075
5076<h5>Arguments:</h5>
5077
5078<p>
5079<tt>id</tt> is a numerical id identifying the marker.
5080</p>
5081
5082<h5>Semantics:</h5>
5083
5084<p>
5085This intrinsic does not modify the behavior of the program. Backends that do not
5086support this intrinisic may ignore it.
5087</p>
5088
5089</div>
5090
5091<!-- _______________________________________________________________________ -->
5092<div class="doc_subsubsection">
5093 <a name="int_readcyclecounter">'<tt>llvm.readcyclecounter</tt>' Intrinsic</a>
5094</div>
5095
5096<div class="doc_text">
5097
5098<h5>Syntax:</h5>
5099<pre>
5100 declare i64 @llvm.readcyclecounter( )
5101</pre>
5102
5103<h5>Overview:</h5>
5104
5105
5106<p>
5107The '<tt>llvm.readcyclecounter</tt>' intrinsic provides access to the cycle
5108counter register (or similar low latency, high accuracy clocks) on those targets
5109that support it. On X86, it should map to RDTSC. On Alpha, it should map to RPCC.
5110As the backing counters overflow quickly (on the order of 9 seconds on alpha), this
5111should only be used for small timings.
5112</p>
5113
5114<h5>Semantics:</h5>
5115
5116<p>
5117When directly supported, reading the cycle counter should not modify any memory.
5118Implementations are allowed to either return a application specific value or a
5119system wide value. On backends without support, this is lowered to a constant 0.
5120</p>
5121
5122</div>
5123
5124<!-- ======================================================================= -->
5125<div class="doc_subsection">
5126 <a name="int_libc">Standard C Library Intrinsics</a>
5127</div>
5128
5129<div class="doc_text">
5130<p>
5131LLVM provides intrinsics for a few important standard C library functions.
5132These intrinsics allow source-language front-ends to pass information about the
5133alignment of the pointer arguments to the code generator, providing opportunity
5134for more efficient code generation.
5135</p>
5136
5137</div>
5138
5139<!-- _______________________________________________________________________ -->
5140<div class="doc_subsubsection">
5141 <a name="int_memcpy">'<tt>llvm.memcpy</tt>' Intrinsic</a>
5142</div>
5143
5144<div class="doc_text">
5145
5146<h5>Syntax:</h5>
Chris Lattner82c2e432008-11-21 16:42:48 +0000[diff] [blame]5147<p>This is an overloaded intrinsic. You can use llvm.memcpy on any integer bit
5148width. Not all targets support all bit widths however.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5149<pre>
Chris Lattner82c2e432008-11-21 16:42:48 +0000[diff] [blame]5150 declare void @llvm.memcpy.i8(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5151 i8 &lt;len&gt;, i32 &lt;align&gt;)
5152 declare void @llvm.memcpy.i16(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5153 i16 &lt;len&gt;, i32 &lt;align&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5154 declare void @llvm.memcpy.i32(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5155 i32 &lt;len&gt;, i32 &lt;align&gt;)
5156 declare void @llvm.memcpy.i64(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5157 i64 &lt;len&gt;, i32 &lt;align&gt;)
5158</pre>
5159
5160<h5>Overview:</h5>
5161
5162<p>
5163The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the source
5164location to the destination location.
5165</p>
5166
5167<p>
5168Note that, unlike the standard libc function, the <tt>llvm.memcpy.*</tt>
5169intrinsics do not return a value, and takes an extra alignment argument.
5170</p>
5171
5172<h5>Arguments:</h5>
5173
5174<p>
5175The first argument is a pointer to the destination, the second is a pointer to
5176the source. The third argument is an integer argument
5177specifying the number of bytes to copy, and the fourth argument is the alignment
5178of the source and destination locations.
5179</p>
5180
5181<p>
5182If the call to this intrinisic has an alignment value that is not 0 or 1, then
5183the caller guarantees that both the source and destination pointers are aligned
5184to that boundary.
5185</p>
5186
5187<h5>Semantics:</h5>
5188
5189<p>
5190The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the source
5191location to the destination location, which are not allowed to overlap. It
5192copies "len" bytes of memory over. If the argument is known to be aligned to
5193some boundary, this can be specified as the fourth argument, otherwise it should
5194be set to 0 or 1.
5195</p>
5196</div>
5197
5198
5199<!-- _______________________________________________________________________ -->
5200<div class="doc_subsubsection">
5201 <a name="int_memmove">'<tt>llvm.memmove</tt>' Intrinsic</a>
5202</div>
5203
5204<div class="doc_text">
5205
5206<h5>Syntax:</h5>
Chris Lattner82c2e432008-11-21 16:42:48 +0000[diff] [blame]5207<p>This is an overloaded intrinsic. You can use llvm.memmove on any integer bit
5208width. Not all targets support all bit widths however.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5209<pre>
Chris Lattner82c2e432008-11-21 16:42:48 +0000[diff] [blame]5210 declare void @llvm.memmove.i8(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5211 i8 &lt;len&gt;, i32 &lt;align&gt;)
5212 declare void @llvm.memmove.i16(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5213 i16 &lt;len&gt;, i32 &lt;align&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5214 declare void @llvm.memmove.i32(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5215 i32 &lt;len&gt;, i32 &lt;align&gt;)
5216 declare void @llvm.memmove.i64(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5217 i64 &lt;len&gt;, i32 &lt;align&gt;)
5218</pre>
5219
5220<h5>Overview:</h5>
5221
5222<p>
5223The '<tt>llvm.memmove.*</tt>' intrinsics move a block of memory from the source
5224location to the destination location. It is similar to the
Chris Lattnerdba16ea2008-01-06 19:51:52 +0000[diff] [blame]5225'<tt>llvm.memcpy</tt>' intrinsic but allows the two memory locations to overlap.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5226</p>
5227
5228<p>
5229Note that, unlike the standard libc function, the <tt>llvm.memmove.*</tt>
5230intrinsics do not return a value, and takes an extra alignment argument.
5231</p>
5232
5233<h5>Arguments:</h5>
5234
5235<p>
5236The first argument is a pointer to the destination, the second is a pointer to
5237the source. The third argument is an integer argument
5238specifying the number of bytes to copy, and the fourth argument is the alignment
5239of the source and destination locations.
5240</p>
5241
5242<p>
5243If the call to this intrinisic has an alignment value that is not 0 or 1, then
5244the caller guarantees that the source and destination pointers are aligned to
5245that boundary.
5246</p>
5247
5248<h5>Semantics:</h5>
5249
5250<p>
5251The '<tt>llvm.memmove.*</tt>' intrinsics copy a block of memory from the source
5252location to the destination location, which may overlap. It
5253copies "len" bytes of memory over. If the argument is known to be aligned to
5254some boundary, this can be specified as the fourth argument, otherwise it should
5255be set to 0 or 1.
5256</p>
5257</div>
5258
5259
5260<!-- _______________________________________________________________________ -->
5261<div class="doc_subsubsection">
5262 <a name="int_memset">'<tt>llvm.memset.*</tt>' Intrinsics</a>
5263</div>
5264
5265<div class="doc_text">
5266
5267<h5>Syntax:</h5>
Chris Lattner82c2e432008-11-21 16:42:48 +0000[diff] [blame]5268<p>This is an overloaded intrinsic. You can use llvm.memset on any integer bit
5269width. Not all targets support all bit widths however.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5270<pre>
Chris Lattner82c2e432008-11-21 16:42:48 +0000[diff] [blame]5271 declare void @llvm.memset.i8(i8 * &lt;dest&gt;, i8 &lt;val&gt;,
5272 i8 &lt;len&gt;, i32 &lt;align&gt;)
5273 declare void @llvm.memset.i16(i8 * &lt;dest&gt;, i8 &lt;val&gt;,
5274 i16 &lt;len&gt;, i32 &lt;align&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5275 declare void @llvm.memset.i32(i8 * &lt;dest&gt;, i8 &lt;val&gt;,
5276 i32 &lt;len&gt;, i32 &lt;align&gt;)
5277 declare void @llvm.memset.i64(i8 * &lt;dest&gt;, i8 &lt;val&gt;,
5278 i64 &lt;len&gt;, i32 &lt;align&gt;)
5279</pre>
5280
5281<h5>Overview:</h5>
5282
5283<p>
5284The '<tt>llvm.memset.*</tt>' intrinsics fill a block of memory with a particular
5285byte value.
5286</p>
5287
5288<p>
5289Note that, unlike the standard libc function, the <tt>llvm.memset</tt> intrinsic
5290does not return a value, and takes an extra alignment argument.
5291</p>
5292
5293<h5>Arguments:</h5>
5294
5295<p>
5296The first argument is a pointer to the destination to fill, the second is the
5297byte value to fill it with, the third argument is an integer
5298argument specifying the number of bytes to fill, and the fourth argument is the
5299known alignment of destination location.
5300</p>
5301
5302<p>
5303If the call to this intrinisic has an alignment value that is not 0 or 1, then
5304the caller guarantees that the destination pointer is aligned to that boundary.
5305</p>
5306
5307<h5>Semantics:</h5>
5308
5309<p>
5310The '<tt>llvm.memset.*</tt>' intrinsics fill "len" bytes of memory starting at
5311the
5312destination location. If the argument is known to be aligned to some boundary,
5313this can be specified as the fourth argument, otherwise it should be set to 0 or
53141.
5315</p>
5316</div>
5317
5318
5319<!-- _______________________________________________________________________ -->
5320<div class="doc_subsubsection">
5321 <a name="int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a>
5322</div>
5323
5324<div class="doc_text">
5325
5326<h5>Syntax:</h5>
Dale Johannesenf9adbb62007-10-02 17:47:38 +0000[diff] [blame]5327<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]5328floating point or vector of floating point type. Not all targets support all
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5329types however.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5330<pre>
Dale Johannesenf9adbb62007-10-02 17:47:38 +0000[diff] [blame]5331 declare float @llvm.sqrt.f32(float %Val)
5332 declare double @llvm.sqrt.f64(double %Val)
5333 declare x86_fp80 @llvm.sqrt.f80(x86_fp80 %Val)
5334 declare fp128 @llvm.sqrt.f128(fp128 %Val)
5335 declare ppc_fp128 @llvm.sqrt.ppcf128(ppc_fp128 %Val)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5336</pre>
5337
5338<h5>Overview:</h5>
5339
5340<p>
5341The '<tt>llvm.sqrt</tt>' intrinsics return the sqrt of the specified operand,
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5342returning the same value as the libm '<tt>sqrt</tt>' functions would. Unlike
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5343<tt>sqrt</tt> in libm, however, <tt>llvm.sqrt</tt> has undefined behavior for
Chris Lattnerf914a002008-01-29 07:00:44 +0000[diff] [blame]5344negative numbers other than -0.0 (which allows for better optimization, because
5345there is no need to worry about errno being set). <tt>llvm.sqrt(-0.0)</tt> is
5346defined to return -0.0 like IEEE sqrt.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5347</p>
5348
5349<h5>Arguments:</h5>
5350
5351<p>
5352The argument and return value are floating point numbers of the same type.
5353</p>
5354
5355<h5>Semantics:</h5>
5356
5357<p>
5358This function returns the sqrt of the specified operand if it is a nonnegative
5359floating point number.
5360</p>
5361</div>
5362
5363<!-- _______________________________________________________________________ -->
5364<div class="doc_subsubsection">
5365 <a name="int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a>
5366</div>
5367
5368<div class="doc_text">
5369
5370<h5>Syntax:</h5>
Dale Johannesenf9adbb62007-10-02 17:47:38 +0000[diff] [blame]5371<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]5372floating point or vector of floating point type. Not all targets support all
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5373types however.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5374<pre>
Dale Johannesenf9adbb62007-10-02 17:47:38 +0000[diff] [blame]5375 declare float @llvm.powi.f32(float %Val, i32 %power)
5376 declare double @llvm.powi.f64(double %Val, i32 %power)
5377 declare x86_fp80 @llvm.powi.f80(x86_fp80 %Val, i32 %power)
5378 declare fp128 @llvm.powi.f128(fp128 %Val, i32 %power)
5379 declare ppc_fp128 @llvm.powi.ppcf128(ppc_fp128 %Val, i32 %power)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5380</pre>
5381
5382<h5>Overview:</h5>
5383
5384<p>
5385The '<tt>llvm.powi.*</tt>' intrinsics return the first operand raised to the
5386specified (positive or negative) power. The order of evaluation of
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5387multiplications is not defined. When a vector of floating point type is
5388used, the second argument remains a scalar integer value.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5389</p>
5390
5391<h5>Arguments:</h5>
5392
5393<p>
5394The second argument is an integer power, and the first is a value to raise to
5395that power.
5396</p>
5397
5398<h5>Semantics:</h5>
5399
5400<p>
5401This function returns the first value raised to the second power with an
5402unspecified sequence of rounding operations.</p>
5403</div>
5404
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5405<!-- _______________________________________________________________________ -->
5406<div class="doc_subsubsection">
5407 <a name="int_sin">'<tt>llvm.sin.*</tt>' Intrinsic</a>
5408</div>
5409
5410<div class="doc_text">
5411
5412<h5>Syntax:</h5>
5413<p>This is an overloaded intrinsic. You can use <tt>llvm.sin</tt> on any
5414floating point or vector of floating point type. Not all targets support all
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5415types however.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5416<pre>
5417 declare float @llvm.sin.f32(float %Val)
5418 declare double @llvm.sin.f64(double %Val)
5419 declare x86_fp80 @llvm.sin.f80(x86_fp80 %Val)
5420 declare fp128 @llvm.sin.f128(fp128 %Val)
5421 declare ppc_fp128 @llvm.sin.ppcf128(ppc_fp128 %Val)
5422</pre>
5423
5424<h5>Overview:</h5>
5425
5426<p>
5427The '<tt>llvm.sin.*</tt>' intrinsics return the sine of the operand.
5428</p>
5429
5430<h5>Arguments:</h5>
5431
5432<p>
5433The argument and return value are floating point numbers of the same type.
5434</p>
5435
5436<h5>Semantics:</h5>
5437
5438<p>
5439This function returns the sine of the specified operand, returning the
5440same values as the libm <tt>sin</tt> functions would, and handles error
Dan Gohmaneaba92e2007-10-17 18:05:13 +0000[diff] [blame]5441conditions in the same way.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5442</div>
5443
5444<!-- _______________________________________________________________________ -->
5445<div class="doc_subsubsection">
5446 <a name="int_cos">'<tt>llvm.cos.*</tt>' Intrinsic</a>
5447</div>
5448
5449<div class="doc_text">
5450
5451<h5>Syntax:</h5>
5452<p>This is an overloaded intrinsic. You can use <tt>llvm.cos</tt> on any
5453floating point or vector of floating point type. Not all targets support all
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5454types however.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5455<pre>
5456 declare float @llvm.cos.f32(float %Val)
5457 declare double @llvm.cos.f64(double %Val)
5458 declare x86_fp80 @llvm.cos.f80(x86_fp80 %Val)
5459 declare fp128 @llvm.cos.f128(fp128 %Val)
5460 declare ppc_fp128 @llvm.cos.ppcf128(ppc_fp128 %Val)
5461</pre>
5462
5463<h5>Overview:</h5>
5464
5465<p>
5466The '<tt>llvm.cos.*</tt>' intrinsics return the cosine of the operand.
5467</p>
5468
5469<h5>Arguments:</h5>
5470
5471<p>
5472The argument and return value are floating point numbers of the same type.
5473</p>
5474
5475<h5>Semantics:</h5>
5476
5477<p>
5478This function returns the cosine of the specified operand, returning the
5479same values as the libm <tt>cos</tt> functions would, and handles error
Dan Gohmaneaba92e2007-10-17 18:05:13 +0000[diff] [blame]5480conditions in the same way.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5481</div>
5482
5483<!-- _______________________________________________________________________ -->
5484<div class="doc_subsubsection">
5485 <a name="int_pow">'<tt>llvm.pow.*</tt>' Intrinsic</a>
5486</div>
5487
5488<div class="doc_text">
5489
5490<h5>Syntax:</h5>
5491<p>This is an overloaded intrinsic. You can use <tt>llvm.pow</tt> on any
5492floating point or vector of floating point type. Not all targets support all
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5493types however.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5494<pre>
5495 declare float @llvm.pow.f32(float %Val, float %Power)
5496 declare double @llvm.pow.f64(double %Val, double %Power)
5497 declare x86_fp80 @llvm.pow.f80(x86_fp80 %Val, x86_fp80 %Power)
5498 declare fp128 @llvm.pow.f128(fp128 %Val, fp128 %Power)
5499 declare ppc_fp128 @llvm.pow.ppcf128(ppc_fp128 %Val, ppc_fp128 Power)
5500</pre>
5501
5502<h5>Overview:</h5>
5503
5504<p>
5505The '<tt>llvm.pow.*</tt>' intrinsics return the first operand raised to the
5506specified (positive or negative) power.
5507</p>
5508
5509<h5>Arguments:</h5>
5510
5511<p>
5512The second argument is a floating point power, and the first is a value to
5513raise to that power.
5514</p>
5515
5516<h5>Semantics:</h5>
5517
5518<p>
5519This function returns the first value raised to the second power,
5520returning the
5521same values as the libm <tt>pow</tt> functions would, and handles error
Dan Gohmaneaba92e2007-10-17 18:05:13 +0000[diff] [blame]5522conditions in the same way.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5523</div>
5524
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5525
5526<!-- ======================================================================= -->
5527<div class="doc_subsection">
5528 <a name="int_manip">Bit Manipulation Intrinsics</a>
5529</div>
5530
5531<div class="doc_text">
5532<p>
5533LLVM provides intrinsics for a few important bit manipulation operations.
5534These allow efficient code generation for some algorithms.
5535</p>
5536
5537</div>
5538
5539<!-- _______________________________________________________________________ -->
5540<div class="doc_subsubsection">
5541 <a name="int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a>
5542</div>
5543
5544<div class="doc_text">
5545
5546<h5>Syntax:</h5>
5547<p>This is an overloaded intrinsic function. You can use bswap on any integer
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5548type that is an even number of bytes (i.e. BitWidth % 16 == 0).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5549<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5550 declare i16 @llvm.bswap.i16(i16 &lt;id&gt;)
5551 declare i32 @llvm.bswap.i32(i32 &lt;id&gt;)
5552 declare i64 @llvm.bswap.i64(i64 &lt;id&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5553</pre>
5554
5555<h5>Overview:</h5>
5556
5557<p>
5558The '<tt>llvm.bswap</tt>' family of intrinsics is used to byte swap integer
5559values with an even number of bytes (positive multiple of 16 bits). These are
5560useful for performing operations on data that is not in the target's native
5561byte order.
5562</p>
5563
5564<h5>Semantics:</h5>
5565
5566<p>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5567The <tt>llvm.bswap.i16</tt> intrinsic returns an i16 value that has the high
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5568and low byte of the input i16 swapped. Similarly, the <tt>llvm.bswap.i32</tt>
5569intrinsic returns an i32 value that has the four bytes of the input i32
5570swapped, 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]5571i32 will have its bytes in 3, 2, 1, 0 order. The <tt>llvm.bswap.i48</tt>,
5572<tt>llvm.bswap.i64</tt> and other intrinsics extend this concept to
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5573additional even-byte lengths (6 bytes, 8 bytes and more, respectively).
5574</p>
5575
5576</div>
5577
5578<!-- _______________________________________________________________________ -->
5579<div class="doc_subsubsection">
5580 <a name="int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic</a>
5581</div>
5582
5583<div class="doc_text">
5584
5585<h5>Syntax:</h5>
5586<p>This is an overloaded intrinsic. You can use llvm.ctpop on any integer bit
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5587width. Not all targets support all bit widths however.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5588<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5589 declare i8 @llvm.ctpop.i8 (i8 &lt;src&gt;)
5590 declare i16 @llvm.ctpop.i16(i16 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5591 declare i32 @llvm.ctpop.i32(i32 &lt;src&gt;)
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5592 declare i64 @llvm.ctpop.i64(i64 &lt;src&gt;)
5593 declare i256 @llvm.ctpop.i256(i256 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5594</pre>
5595
5596<h5>Overview:</h5>
5597
5598<p>
5599The '<tt>llvm.ctpop</tt>' family of intrinsics counts the number of bits set in a
5600value.
5601</p>
5602
5603<h5>Arguments:</h5>
5604
5605<p>
5606The only argument is the value to be counted. The argument may be of any
5607integer type. The return type must match the argument type.
5608</p>
5609
5610<h5>Semantics:</h5>
5611
5612<p>
5613The '<tt>llvm.ctpop</tt>' intrinsic counts the 1's in a variable.
5614</p>
5615</div>
5616
5617<!-- _______________________________________________________________________ -->
5618<div class="doc_subsubsection">
5619 <a name="int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic</a>
5620</div>
5621
5622<div class="doc_text">
5623
5624<h5>Syntax:</h5>
5625<p>This is an overloaded intrinsic. You can use <tt>llvm.ctlz</tt> on any
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5626integer bit width. Not all targets support all bit widths however.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5627<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5628 declare i8 @llvm.ctlz.i8 (i8 &lt;src&gt;)
5629 declare i16 @llvm.ctlz.i16(i16 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5630 declare i32 @llvm.ctlz.i32(i32 &lt;src&gt;)
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5631 declare i64 @llvm.ctlz.i64(i64 &lt;src&gt;)
5632 declare i256 @llvm.ctlz.i256(i256 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5633</pre>
5634
5635<h5>Overview:</h5>
5636
5637<p>
5638The '<tt>llvm.ctlz</tt>' family of intrinsic functions counts the number of
5639leading zeros in a variable.
5640</p>
5641
5642<h5>Arguments:</h5>
5643
5644<p>
5645The only argument is the value to be counted. The argument may be of any
5646integer type. The return type must match the argument type.
5647</p>
5648
5649<h5>Semantics:</h5>
5650
5651<p>
5652The '<tt>llvm.ctlz</tt>' intrinsic counts the leading (most significant) zeros
5653in a variable. If the src == 0 then the result is the size in bits of the type
5654of src. For example, <tt>llvm.ctlz(i32 2) = 30</tt>.
5655</p>
5656</div>
5657
5658
5659
5660<!-- _______________________________________________________________________ -->
5661<div class="doc_subsubsection">
5662 <a name="int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic</a>
5663</div>
5664
5665<div class="doc_text">
5666
5667<h5>Syntax:</h5>
5668<p>This is an overloaded intrinsic. You can use <tt>llvm.cttz</tt> on any
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5669integer bit width. Not all targets support all bit widths however.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5670<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5671 declare i8 @llvm.cttz.i8 (i8 &lt;src&gt;)
5672 declare i16 @llvm.cttz.i16(i16 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5673 declare i32 @llvm.cttz.i32(i32 &lt;src&gt;)
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5674 declare i64 @llvm.cttz.i64(i64 &lt;src&gt;)
5675 declare i256 @llvm.cttz.i256(i256 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5676</pre>
5677
5678<h5>Overview:</h5>
5679
5680<p>
5681The '<tt>llvm.cttz</tt>' family of intrinsic functions counts the number of
5682trailing zeros.
5683</p>
5684
5685<h5>Arguments:</h5>
5686
5687<p>
5688The only argument is the value to be counted. The argument may be of any
5689integer type. The return type must match the argument type.
5690</p>
5691
5692<h5>Semantics:</h5>
5693
5694<p>
5695The '<tt>llvm.cttz</tt>' intrinsic counts the trailing (least significant) zeros
5696in a variable. If the src == 0 then the result is the size in bits of the type
5697of src. For example, <tt>llvm.cttz(2) = 1</tt>.
5698</p>
5699</div>
5700
5701<!-- _______________________________________________________________________ -->
5702<div class="doc_subsubsection">
5703 <a name="int_part_select">'<tt>llvm.part.select.*</tt>' Intrinsic</a>
5704</div>
5705
5706<div class="doc_text">
5707
5708<h5>Syntax:</h5>
5709<p>This is an overloaded intrinsic. You can use <tt>llvm.part.select</tt>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5710on any integer bit width.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5711<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5712 declare i17 @llvm.part.select.i17 (i17 %val, i32 %loBit, i32 %hiBit)
5713 declare i29 @llvm.part.select.i29 (i29 %val, i32 %loBit, i32 %hiBit)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5714</pre>
5715
5716<h5>Overview:</h5>
5717<p>The '<tt>llvm.part.select</tt>' family of intrinsic functions selects a
5718range of bits from an integer value and returns them in the same bit width as
5719the original value.</p>
5720
5721<h5>Arguments:</h5>
5722<p>The first argument, <tt>%val</tt> and the result may be integer types of
5723any bit width but they must have the same bit width. The second and third
5724arguments must be <tt>i32</tt> type since they specify only a bit index.</p>
5725
5726<h5>Semantics:</h5>
5727<p>The operation of the '<tt>llvm.part.select</tt>' intrinsic has two modes
5728of operation: forwards and reverse. If <tt>%loBit</tt> is greater than
5729<tt>%hiBits</tt> then the intrinsic operates in reverse mode. Otherwise it
5730operates in forward mode.</p>
5731<p>In forward mode, this intrinsic is the equivalent of shifting <tt>%val</tt>
5732right by <tt>%loBit</tt> bits and then ANDing it with a mask with
5733only the <tt>%hiBit - %loBit</tt> bits set, as follows:</p>
5734<ol>
5735 <li>The <tt>%val</tt> is shifted right (LSHR) by the number of bits specified
5736 by <tt>%loBits</tt>. This normalizes the value to the low order bits.</li>
5737 <li>The <tt>%loBits</tt> value is subtracted from the <tt>%hiBits</tt> value
5738 to determine the number of bits to retain.</li>
5739 <li>A mask of the retained bits is created by shifting a -1 value.</li>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5740 <li>The mask is ANDed with <tt>%val</tt> to produce the result.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5741</ol>
5742<p>In reverse mode, a similar computation is made except that the bits are
5743returned in the reverse order. So, for example, if <tt>X</tt> has the value
5744<tt>i16 0x0ACF (101011001111)</tt> and we apply
5745<tt>part.select(i16 X, 8, 3)</tt> to it, we get back the value
5746<tt>i16 0x0026 (000000100110)</tt>.</p>
5747</div>
5748
5749<div class="doc_subsubsection">
5750 <a name="int_part_set">'<tt>llvm.part.set.*</tt>' Intrinsic</a>
5751</div>
5752
5753<div class="doc_text">
5754
5755<h5>Syntax:</h5>
5756<p>This is an overloaded intrinsic. You can use <tt>llvm.part.set</tt>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5757on any integer bit width.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5758<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5759 declare i17 @llvm.part.set.i17.i9 (i17 %val, i9 %repl, i32 %lo, i32 %hi)
5760 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]5761</pre>
5762
5763<h5>Overview:</h5>
5764<p>The '<tt>llvm.part.set</tt>' family of intrinsic functions replaces a range
5765of bits in an integer value with another integer value. It returns the integer
5766with the replaced bits.</p>
5767
5768<h5>Arguments:</h5>
5769<p>The first argument, <tt>%val</tt> and the result may be integer types of
5770any bit width but they must have the same bit width. <tt>%val</tt> is the value
5771whose bits will be replaced. The second argument, <tt>%repl</tt> may be an
5772integer of any bit width. The third and fourth arguments must be <tt>i32</tt>
5773type since they specify only a bit index.</p>
5774
5775<h5>Semantics:</h5>
5776<p>The operation of the '<tt>llvm.part.set</tt>' intrinsic has two modes
5777of operation: forwards and reverse. If <tt>%lo</tt> is greater than
5778<tt>%hi</tt> then the intrinsic operates in reverse mode. Otherwise it
5779operates in forward mode.</p>
5780<p>For both modes, the <tt>%repl</tt> value is prepared for use by either
5781truncating it down to the size of the replacement area or zero extending it
5782up to that size.</p>
5783<p>In forward mode, the bits between <tt>%lo</tt> and <tt>%hi</tt> (inclusive)
5784are replaced with corresponding bits from <tt>%repl</tt>. That is the 0th bit
5785in <tt>%repl</tt> replaces the <tt>%lo</tt>th bit in <tt>%val</tt> and etc. up
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5786to the <tt>%hi</tt>th bit.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5787<p>In reverse mode, a similar computation is made except that the bits are
5788reversed. That is, the <tt>0</tt>th bit in <tt>%repl</tt> replaces the
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5789<tt>%hi</tt> bit in <tt>%val</tt> and etc. down to the <tt>%lo</tt>th bit.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5790<h5>Examples:</h5>
5791<pre>
5792 llvm.part.set(0xFFFF, 0, 4, 7) -&gt; 0xFF0F
5793 llvm.part.set(0xFFFF, 0, 7, 4) -&gt; 0xFF0F
5794 llvm.part.set(0xFFFF, 1, 7, 4) -&gt; 0xFF8F
5795 llvm.part.set(0xFFFF, F, 8, 3) -&gt; 0xFFE7
5796 llvm.part.set(0xFFFF, 0, 3, 8) -&gt; 0xFE07
5797</pre>
5798</div>
5799
5800<!-- ======================================================================= -->
5801<div class="doc_subsection">
5802 <a name="int_debugger">Debugger Intrinsics</a>
5803</div>
5804
5805<div class="doc_text">
5806<p>
5807The LLVM debugger intrinsics (which all start with <tt>llvm.dbg.</tt> prefix),
5808are described in the <a
5809href="SourceLevelDebugging.html#format_common_intrinsics">LLVM Source Level
5810Debugging</a> document.
5811</p>
5812</div>
5813
5814
5815<!-- ======================================================================= -->
5816<div class="doc_subsection">
5817 <a name="int_eh">Exception Handling Intrinsics</a>
5818</div>
5819
5820<div class="doc_text">
5821<p> The LLVM exception handling intrinsics (which all start with
5822<tt>llvm.eh.</tt> prefix), are described in the <a
5823href="ExceptionHandling.html#format_common_intrinsics">LLVM Exception
5824Handling</a> document. </p>
5825</div>
5826
5827<!-- ======================================================================= -->
5828<div class="doc_subsection">
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5829 <a name="int_trampoline">Trampoline Intrinsic</a>
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5830</div>
5831
5832<div class="doc_text">
5833<p>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5834 This intrinsic makes it possible to excise one parameter, marked with
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5835 the <tt>nest</tt> attribute, from a function. The result is a callable
5836 function pointer lacking the nest parameter - the caller does not need
5837 to provide a value for it. Instead, the value to use is stored in
5838 advance in a "trampoline", a block of memory usually allocated
5839 on the stack, which also contains code to splice the nest value into the
5840 argument list. This is used to implement the GCC nested function address
5841 extension.
5842</p>
5843<p>
5844 For example, if the function is
5845 <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]5846 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]5847<pre>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5848 %tramp = alloca [10 x i8], align 4 ; size and alignment only correct for X86
5849 %tramp1 = getelementptr [10 x i8]* %tramp, i32 0, i32 0
5850 %p = call i8* @llvm.init.trampoline( i8* %tramp1, i8* bitcast (i32 (i8* nest , i32, i32)* @f to i8*), i8* %nval )
5851 %fp = bitcast i8* %p to i32 (i32, i32)*
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5852</pre>
Bill Wendlinge262b4c2007-09-22 09:23:55 +0000[diff] [blame]5853 <p>The call <tt>%val = call i32 %fp( i32 %x, i32 %y )</tt> is then equivalent
5854 to <tt>%val = call i32 %f( i8* %nval, i32 %x, i32 %y )</tt>.</p>
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5855</div>
5856
5857<!-- _______________________________________________________________________ -->
5858<div class="doc_subsubsection">
5859 <a name="int_it">'<tt>llvm.init.trampoline</tt>' Intrinsic</a>
5860</div>
5861<div class="doc_text">
5862<h5>Syntax:</h5>
5863<pre>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5864declare 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]5865</pre>
5866<h5>Overview:</h5>
5867<p>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5868 This fills the memory pointed to by <tt>tramp</tt> with code
5869 and returns a function pointer suitable for executing it.
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5870</p>
5871<h5>Arguments:</h5>
5872<p>
5873 The <tt>llvm.init.trampoline</tt> intrinsic takes three arguments, all
5874 pointers. The <tt>tramp</tt> argument must point to a sufficiently large
5875 and sufficiently aligned block of memory; this memory is written to by the
Duncan Sands35012212007-08-22 23:39:54 +0000[diff] [blame]5876 intrinsic. Note that the size and the alignment are target-specific - LLVM
5877 currently provides no portable way of determining them, so a front-end that
5878 generates this intrinsic needs to have some target-specific knowledge.
5879 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]5880</p>
5881<h5>Semantics:</h5>
5882<p>
5883 The block of memory pointed to by <tt>tramp</tt> is filled with target
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5884 dependent code, turning it into a function. A pointer to this function is
5885 returned, but needs to be bitcast to an
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5886 <a href="#int_trampoline">appropriate function pointer type</a>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]5887 before being called. The new function's signature is the same as that of
5888 <tt>func</tt> with any arguments marked with the <tt>nest</tt> attribute
5889 removed. At most one such <tt>nest</tt> argument is allowed, and it must be
5890 of pointer type. Calling the new function is equivalent to calling
5891 <tt>func</tt> with the same argument list, but with <tt>nval</tt> used for the
5892 missing <tt>nest</tt> argument. If, after calling
5893 <tt>llvm.init.trampoline</tt>, the memory pointed to by <tt>tramp</tt> is
5894 modified, then the effect of any later call to the returned function pointer is
5895 undefined.
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]5896</p>
5897</div>
5898
5899<!-- ======================================================================= -->
5900<div class="doc_subsection">
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]5901 <a name="int_atomics">Atomic Operations and Synchronization Intrinsics</a>
5902</div>
5903
5904<div class="doc_text">
5905<p>
5906 These intrinsic functions expand the "universal IR" of LLVM to represent
5907 hardware constructs for atomic operations and memory synchronization. This
5908 provides an interface to the hardware, not an interface to the programmer. It
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]5909 is aimed at a low enough level to allow any programming models or APIs
5910 (Application Programming Interfaces) which
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]5911 need atomic behaviors to map cleanly onto it. It is also modeled primarily on
5912 hardware behavior. Just as hardware provides a "universal IR" for source
5913 languages, it also provides a starting point for developing a "universal"
5914 atomic operation and synchronization IR.
5915</p>
5916<p>
5917 These do <em>not</em> form an API such as high-level threading libraries,
5918 software transaction memory systems, atomic primitives, and intrinsic
5919 functions as found in BSD, GNU libc, atomic_ops, APR, and other system and
5920 application libraries. The hardware interface provided by LLVM should allow
5921 a clean implementation of all of these APIs and parallel programming models.
5922 No one model or paradigm should be selected above others unless the hardware
5923 itself ubiquitously does so.
5924
5925</p>
5926</div>
5927
5928<!-- _______________________________________________________________________ -->
5929<div class="doc_subsubsection">
5930 <a name="int_memory_barrier">'<tt>llvm.memory.barrier</tt>' Intrinsic</a>
5931</div>
5932<div class="doc_text">
5933<h5>Syntax:</h5>
5934<pre>
5935declare void @llvm.memory.barrier( i1 &lt;ll&gt;, i1 &lt;ls&gt;, i1 &lt;sl&gt;, i1 &lt;ss&gt;,
5936i1 &lt;device&gt; )
5937
5938</pre>
5939<h5>Overview:</h5>
5940<p>
5941 The <tt>llvm.memory.barrier</tt> intrinsic guarantees ordering between
5942 specific pairs of memory access types.
5943</p>
5944<h5>Arguments:</h5>
5945<p>
5946 The <tt>llvm.memory.barrier</tt> intrinsic requires five boolean arguments.
5947 The first four arguments enables a specific barrier as listed below. The fith
5948 argument specifies that the barrier applies to io or device or uncached memory.
5949
5950</p>
5951 <ul>
5952 <li><tt>ll</tt>: load-load barrier</li>
5953 <li><tt>ls</tt>: load-store barrier</li>
5954 <li><tt>sl</tt>: store-load barrier</li>
5955 <li><tt>ss</tt>: store-store barrier</li>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5956 <li><tt>device</tt>: barrier applies to device and uncached memory also.</li>
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]5957 </ul>
5958<h5>Semantics:</h5>
5959<p>
5960 This intrinsic causes the system to enforce some ordering constraints upon
5961 the loads and stores of the program. This barrier does not indicate
5962 <em>when</em> any events will occur, it only enforces an <em>order</em> in
5963 which they occur. For any of the specified pairs of load and store operations
5964 (f.ex. load-load, or store-load), all of the first operations preceding the
5965 barrier will complete before any of the second operations succeeding the
5966 barrier begin. Specifically the semantics for each pairing is as follows:
5967</p>
5968 <ul>
5969 <li><tt>ll</tt>: All loads before the barrier must complete before any load
5970 after the barrier begins.</li>
5971
5972 <li><tt>ls</tt>: All loads before the barrier must complete before any
5973 store after the barrier begins.</li>
5974 <li><tt>ss</tt>: All stores before the barrier must complete before any
5975 store after the barrier begins.</li>
5976 <li><tt>sl</tt>: All stores before the barrier must complete before any
5977 load after the barrier begins.</li>
5978 </ul>
5979<p>
5980 These semantics are applied with a logical "and" behavior when more than one
5981 is enabled in a single memory barrier intrinsic.
5982</p>
5983<p>
5984 Backends may implement stronger barriers than those requested when they do not
5985 support as fine grained a barrier as requested. Some architectures do not
5986 need all types of barriers and on such architectures, these become noops.
5987</p>
5988<h5>Example:</h5>
5989<pre>
5990%ptr = malloc i32
5991 store i32 4, %ptr
5992
5993%result1 = load i32* %ptr <i>; yields {i32}:result1 = 4</i>
5994 call void @llvm.memory.barrier( i1 false, i1 true, i1 false, i1 false )
5995 <i>; guarantee the above finishes</i>
5996 store i32 8, %ptr <i>; before this begins</i>
5997</pre>
5998</div>
5999
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6000<!-- _______________________________________________________________________ -->
6001<div class="doc_subsubsection">
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6002 <a name="int_atomic_cmp_swap">'<tt>llvm.atomic.cmp.swap.*</tt>' Intrinsic</a>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6003</div>
6004<div class="doc_text">
6005<h5>Syntax:</h5>
6006<p>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6007 This is an overloaded intrinsic. You can use <tt>llvm.atomic.cmp.swap</tt> on
6008 any integer bit width and for different address spaces. Not all targets
6009 support all bit widths however.</p>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6010
6011<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6012declare i8 @llvm.atomic.cmp.swap.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;cmp&gt;, i8 &lt;val&gt; )
6013declare i16 @llvm.atomic.cmp.swap.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;cmp&gt;, i16 &lt;val&gt; )
6014declare i32 @llvm.atomic.cmp.swap.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;cmp&gt;, i32 &lt;val&gt; )
6015declare i64 @llvm.atomic.cmp.swap.i64.p0i64( i64* &lt;ptr&gt;, i64 &lt;cmp&gt;, i64 &lt;val&gt; )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6016
6017</pre>
6018<h5>Overview:</h5>
6019<p>
6020 This loads a value in memory and compares it to a given value. If they are
6021 equal, it stores a new value into the memory.
6022</p>
6023<h5>Arguments:</h5>
6024<p>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6025 The <tt>llvm.atomic.cmp.swap</tt> intrinsic takes three arguments. The result as
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6026 well as both <tt>cmp</tt> and <tt>val</tt> must be integer values with the
6027 same bit width. The <tt>ptr</tt> argument must be a pointer to a value of
6028 this integer type. While any bit width integer may be used, targets may only
6029 lower representations they support in hardware.
6030
6031</p>
6032<h5>Semantics:</h5>
6033<p>
6034 This entire intrinsic must be executed atomically. It first loads the value
6035 in memory pointed to by <tt>ptr</tt> and compares it with the value
6036 <tt>cmp</tt>. If they are equal, <tt>val</tt> is stored into the memory. The
6037 loaded value is yielded in all cases. This provides the equivalent of an
6038 atomic compare-and-swap operation within the SSA framework.
6039</p>
6040<h5>Examples:</h5>
6041
6042<pre>
6043%ptr = malloc i32
6044 store i32 4, %ptr
6045
6046%val1 = add i32 4, 4
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6047%result1 = call i32 @llvm.atomic.cmp.swap.i32.p0i32( i32* %ptr, i32 4, %val1 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6048 <i>; yields {i32}:result1 = 4</i>
6049%stored1 = icmp eq i32 %result1, 4 <i>; yields {i1}:stored1 = true</i>
6050%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 8</i>
6051
6052%val2 = add i32 1, 1
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6053%result2 = call i32 @llvm.atomic.cmp.swap.i32.p0i32( i32* %ptr, i32 5, %val2 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6054 <i>; yields {i32}:result2 = 8</i>
6055%stored2 = icmp eq i32 %result2, 5 <i>; yields {i1}:stored2 = false</i>
6056
6057%memval2 = load i32* %ptr <i>; yields {i32}:memval2 = 8</i>
6058</pre>
6059</div>
6060
6061<!-- _______________________________________________________________________ -->
6062<div class="doc_subsubsection">
6063 <a name="int_atomic_swap">'<tt>llvm.atomic.swap.*</tt>' Intrinsic</a>
6064</div>
6065<div class="doc_text">
6066<h5>Syntax:</h5>
6067
6068<p>
6069 This is an overloaded intrinsic. You can use <tt>llvm.atomic.swap</tt> on any
6070 integer bit width. Not all targets support all bit widths however.</p>
6071<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6072declare i8 @llvm.atomic.swap.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;val&gt; )
6073declare i16 @llvm.atomic.swap.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;val&gt; )
6074declare i32 @llvm.atomic.swap.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;val&gt; )
6075declare i64 @llvm.atomic.swap.i64.p0i64( i64* &lt;ptr&gt;, i64 &lt;val&gt; )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6076
6077</pre>
6078<h5>Overview:</h5>
6079<p>
6080 This intrinsic loads the value stored in memory at <tt>ptr</tt> and yields
6081 the value from memory. It then stores the value in <tt>val</tt> in the memory
6082 at <tt>ptr</tt>.
6083</p>
6084<h5>Arguments:</h5>
6085
6086<p>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6087 The <tt>llvm.atomic.swap</tt> intrinsic takes two arguments. Both the
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6088 <tt>val</tt> argument and the result must be integers of the same bit width.
6089 The first argument, <tt>ptr</tt>, must be a pointer to a value of this
6090 integer type. The targets may only lower integer representations they
6091 support.
6092</p>
6093<h5>Semantics:</h5>
6094<p>
6095 This intrinsic loads the value pointed to by <tt>ptr</tt>, yields it, and
6096 stores <tt>val</tt> back into <tt>ptr</tt> atomically. This provides the
6097 equivalent of an atomic swap operation within the SSA framework.
6098
6099</p>
6100<h5>Examples:</h5>
6101<pre>
6102%ptr = malloc i32
6103 store i32 4, %ptr
6104
6105%val1 = add i32 4, 4
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6106%result1 = call i32 @llvm.atomic.swap.i32.p0i32( i32* %ptr, i32 %val1 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6107 <i>; yields {i32}:result1 = 4</i>
6108%stored1 = icmp eq i32 %result1, 4 <i>; yields {i1}:stored1 = true</i>
6109%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 8</i>
6110
6111%val2 = add i32 1, 1
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6112%result2 = call i32 @llvm.atomic.swap.i32.p0i32( i32* %ptr, i32 %val2 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6113 <i>; yields {i32}:result2 = 8</i>
6114
6115%stored2 = icmp eq i32 %result2, 8 <i>; yields {i1}:stored2 = true</i>
6116%memval2 = load i32* %ptr <i>; yields {i32}:memval2 = 2</i>
6117</pre>
6118</div>
6119
6120<!-- _______________________________________________________________________ -->
6121<div class="doc_subsubsection">
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6122 <a name="int_atomic_load_add">'<tt>llvm.atomic.load.add.*</tt>' Intrinsic</a>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6123
6124</div>
6125<div class="doc_text">
6126<h5>Syntax:</h5>
6127<p>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6128 This is an overloaded intrinsic. You can use <tt>llvm.atomic.load.add</tt> on any
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6129 integer bit width. Not all targets support all bit widths however.</p>
6130<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6131declare i8 @llvm.atomic.load.add.i8..p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6132declare i16 @llvm.atomic.load.add.i16..p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6133declare i32 @llvm.atomic.load.add.i32..p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6134declare i64 @llvm.atomic.load.add.i64..p0i64( i64* &lt;ptr&gt;, i64 &lt;delta&gt; )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6135
6136</pre>
6137<h5>Overview:</h5>
6138<p>
6139 This intrinsic adds <tt>delta</tt> to the value stored in memory at
6140 <tt>ptr</tt>. It yields the original value at <tt>ptr</tt>.
6141</p>
6142<h5>Arguments:</h5>
6143<p>
6144
6145 The intrinsic takes two arguments, the first a pointer to an integer value
6146 and the second an integer value. The result is also an integer value. These
6147 integer types can have any bit width, but they must all have the same bit
6148 width. The targets may only lower integer representations they support.
6149</p>
6150<h5>Semantics:</h5>
6151<p>
6152 This intrinsic does a series of operations atomically. It first loads the
6153 value stored at <tt>ptr</tt>. It then adds <tt>delta</tt>, stores the result
6154 to <tt>ptr</tt>. It yields the original value stored at <tt>ptr</tt>.
6155</p>
6156
6157<h5>Examples:</h5>
6158<pre>
6159%ptr = malloc i32
6160 store i32 4, %ptr
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6161%result1 = call i32 @llvm.atomic.load.add.i32.p0i32( i32* %ptr, i32 4 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6162 <i>; yields {i32}:result1 = 4</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6163%result2 = call i32 @llvm.atomic.load.add.i32.p0i32( i32* %ptr, i32 2 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6164 <i>; yields {i32}:result2 = 8</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6165%result3 = call i32 @llvm.atomic.load.add.i32.p0i32( i32* %ptr, i32 5 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6166 <i>; yields {i32}:result3 = 10</i>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6167%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 15</i>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6168</pre>
6169</div>
6170
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6171<!-- _______________________________________________________________________ -->
6172<div class="doc_subsubsection">
6173 <a name="int_atomic_load_sub">'<tt>llvm.atomic.load.sub.*</tt>' Intrinsic</a>
6174
6175</div>
6176<div class="doc_text">
6177<h5>Syntax:</h5>
6178<p>
6179 This is an overloaded intrinsic. You can use <tt>llvm.atomic.load.sub</tt> on
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6180 any integer bit width and for different address spaces. Not all targets
6181 support all bit widths however.</p>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6182<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6183declare i8 @llvm.atomic.load.sub.i8.p0i32( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6184declare i16 @llvm.atomic.load.sub.i16.p0i32( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6185declare i32 @llvm.atomic.load.sub.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6186declare i64 @llvm.atomic.load.sub.i64.p0i32( i64* &lt;ptr&gt;, i64 &lt;delta&gt; )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6187
6188</pre>
6189<h5>Overview:</h5>
6190<p>
6191 This intrinsic subtracts <tt>delta</tt> to the value stored in memory at
6192 <tt>ptr</tt>. It yields the original value at <tt>ptr</tt>.
6193</p>
6194<h5>Arguments:</h5>
6195<p>
6196
6197 The intrinsic takes two arguments, the first a pointer to an integer value
6198 and the second an integer value. The result is also an integer value. These
6199 integer types can have any bit width, but they must all have the same bit
6200 width. The targets may only lower integer representations they support.
6201</p>
6202<h5>Semantics:</h5>
6203<p>
6204 This intrinsic does a series of operations atomically. It first loads the
6205 value stored at <tt>ptr</tt>. It then subtracts <tt>delta</tt>, stores the
6206 result to <tt>ptr</tt>. It yields the original value stored at <tt>ptr</tt>.
6207</p>
6208
6209<h5>Examples:</h5>
6210<pre>
6211%ptr = malloc i32
6212 store i32 8, %ptr
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6213%result1 = call i32 @llvm.atomic.load.sub.i32.p0i32( i32* %ptr, i32 4 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6214 <i>; yields {i32}:result1 = 8</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6215%result2 = call i32 @llvm.atomic.load.sub.i32.p0i32( i32* %ptr, i32 2 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6216 <i>; yields {i32}:result2 = 4</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6217%result3 = call i32 @llvm.atomic.load.sub.i32.p0i32( i32* %ptr, i32 5 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6218 <i>; yields {i32}:result3 = 2</i>
6219%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = -3</i>
6220</pre>
6221</div>
6222
6223<!-- _______________________________________________________________________ -->
6224<div class="doc_subsubsection">
6225 <a name="int_atomic_load_and">'<tt>llvm.atomic.load.and.*</tt>' Intrinsic</a><br>
6226 <a name="int_atomic_load_nand">'<tt>llvm.atomic.load.nand.*</tt>' Intrinsic</a><br>
6227 <a name="int_atomic_load_or">'<tt>llvm.atomic.load.or.*</tt>' Intrinsic</a><br>
6228 <a name="int_atomic_load_xor">'<tt>llvm.atomic.load.xor.*</tt>' Intrinsic</a><br>
6229
6230</div>
6231<div class="doc_text">
6232<h5>Syntax:</h5>
6233<p>
6234 These are overloaded intrinsics. You can use <tt>llvm.atomic.load_and</tt>,
6235 <tt>llvm.atomic.load_nand</tt>, <tt>llvm.atomic.load_or</tt>, and
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6236 <tt>llvm.atomic.load_xor</tt> on any integer bit width and for different
6237 address spaces. Not all targets support all bit widths however.</p>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6238<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6239declare i8 @llvm.atomic.load.and.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6240declare i16 @llvm.atomic.load.and.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6241declare i32 @llvm.atomic.load.and.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6242declare i64 @llvm.atomic.load.and.i64.p0i64( i64* &lt;ptr&gt;, i64 &lt;delta&gt; )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6243
6244</pre>
6245
6246<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6247declare i8 @llvm.atomic.load.or.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6248declare i16 @llvm.atomic.load.or.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6249declare i32 @llvm.atomic.load.or.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6250declare i64 @llvm.atomic.load.or.i64.p0i64( i64* &lt;ptr&gt;, i64 &lt;delta&gt; )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6251
6252</pre>
6253
6254<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6255declare i8 @llvm.atomic.load.nand.i8.p0i32( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6256declare i16 @llvm.atomic.load.nand.i16.p0i32( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6257declare i32 @llvm.atomic.load.nand.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6258declare i64 @llvm.atomic.load.nand.i64.p0i32( i64* &lt;ptr&gt;, i64 &lt;delta&gt; )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6259
6260</pre>
6261
6262<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6263declare i8 @llvm.atomic.load.xor.i8.p0i32( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6264declare i16 @llvm.atomic.load.xor.i16.p0i32( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6265declare i32 @llvm.atomic.load.xor.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6266declare i64 @llvm.atomic.load.xor.i64.p0i32( i64* &lt;ptr&gt;, i64 &lt;delta&gt; )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6267
6268</pre>
6269<h5>Overview:</h5>
6270<p>
6271 These intrinsics bitwise the operation (and, nand, or, xor) <tt>delta</tt> to
6272 the value stored in memory at <tt>ptr</tt>. It yields the original value
6273 at <tt>ptr</tt>.
6274</p>
6275<h5>Arguments:</h5>
6276<p>
6277
6278 These intrinsics take two arguments, the first a pointer to an integer value
6279 and the second an integer value. The result is also an integer value. These
6280 integer types can have any bit width, but they must all have the same bit
6281 width. The targets may only lower integer representations they support.
6282</p>
6283<h5>Semantics:</h5>
6284<p>
6285 These intrinsics does a series of operations atomically. They first load the
6286 value stored at <tt>ptr</tt>. They then do the bitwise operation
6287 <tt>delta</tt>, store the result to <tt>ptr</tt>. They yield the original
6288 value stored at <tt>ptr</tt>.
6289</p>
6290
6291<h5>Examples:</h5>
6292<pre>
6293%ptr = malloc i32
6294 store i32 0x0F0F, %ptr
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6295%result0 = call i32 @llvm.atomic.load.nand.i32.p0i32( i32* %ptr, i32 0xFF )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6296 <i>; yields {i32}:result0 = 0x0F0F</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6297%result1 = call i32 @llvm.atomic.load.and.i32.p0i32( i32* %ptr, i32 0xFF )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6298 <i>; yields {i32}:result1 = 0xFFFFFFF0</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6299%result2 = call i32 @llvm.atomic.load.or.i32.p0i32( i32* %ptr, i32 0F )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6300 <i>; yields {i32}:result2 = 0xF0</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6301%result3 = call i32 @llvm.atomic.load.xor.i32.p0i32( i32* %ptr, i32 0F )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6302 <i>; yields {i32}:result3 = FF</i>
6303%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = F0</i>
6304</pre>
6305</div>
6306
6307
6308<!-- _______________________________________________________________________ -->
6309<div class="doc_subsubsection">
6310 <a name="int_atomic_load_max">'<tt>llvm.atomic.load.max.*</tt>' Intrinsic</a><br>
6311 <a name="int_atomic_load_min">'<tt>llvm.atomic.load.min.*</tt>' Intrinsic</a><br>
6312 <a name="int_atomic_load_umax">'<tt>llvm.atomic.load.umax.*</tt>' Intrinsic</a><br>
6313 <a name="int_atomic_load_umin">'<tt>llvm.atomic.load.umin.*</tt>' Intrinsic</a><br>
6314
6315</div>
6316<div class="doc_text">
6317<h5>Syntax:</h5>
6318<p>
6319 These are overloaded intrinsics. You can use <tt>llvm.atomic.load_max</tt>,
6320 <tt>llvm.atomic.load_min</tt>, <tt>llvm.atomic.load_umax</tt>, and
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6321 <tt>llvm.atomic.load_umin</tt> on any integer bit width and for different
6322 address spaces. Not all targets
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6323 support all bit widths however.</p>
6324<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6325declare i8 @llvm.atomic.load.max.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6326declare i16 @llvm.atomic.load.max.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6327declare i32 @llvm.atomic.load.max.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6328declare i64 @llvm.atomic.load.max.i64.p0i64( i64* &lt;ptr&gt;, i64 &lt;delta&gt; )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6329
6330</pre>
6331
6332<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6333declare i8 @llvm.atomic.load.min.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6334declare i16 @llvm.atomic.load.min.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6335declare i32 @llvm.atomic.load.min.i32..p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6336declare i64 @llvm.atomic.load.min.i64..p0i64( i64* &lt;ptr&gt;, i64 &lt;delta&gt; )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6337
6338</pre>
6339
6340<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6341declare i8 @llvm.atomic.load.umax.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6342declare i16 @llvm.atomic.load.umax.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6343declare i32 @llvm.atomic.load.umax.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6344declare i64 @llvm.atomic.load.umax.i64.p0i64( i64* &lt;ptr&gt;, i64 &lt;delta&gt; )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6345
6346</pre>
6347
6348<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6349declare i8 @llvm.atomic.load.umin.i8..p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6350declare i16 @llvm.atomic.load.umin.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6351declare i32 @llvm.atomic.load.umin.i32..p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6352declare i64 @llvm.atomic.load.umin.i64..p0i64( i64* &lt;ptr&gt;, i64 &lt;delta&gt; )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6353
6354</pre>
6355<h5>Overview:</h5>
6356<p>
6357 These intrinsics takes the signed or unsigned minimum or maximum of
6358 <tt>delta</tt> and the value stored in memory at <tt>ptr</tt>. It yields the
6359 original value at <tt>ptr</tt>.
6360</p>
6361<h5>Arguments:</h5>
6362<p>
6363
6364 These intrinsics take two arguments, the first a pointer to an integer value
6365 and the second an integer value. The result is also an integer value. These
6366 integer types can have any bit width, but they must all have the same bit
6367 width. The targets may only lower integer representations they support.
6368</p>
6369<h5>Semantics:</h5>
6370<p>
6371 These intrinsics does a series of operations atomically. They first load the
6372 value stored at <tt>ptr</tt>. They then do the signed or unsigned min or max
6373 <tt>delta</tt> and the value, store the result to <tt>ptr</tt>. They yield
6374 the original value stored at <tt>ptr</tt>.
6375</p>
6376
6377<h5>Examples:</h5>
6378<pre>
6379%ptr = malloc i32
6380 store i32 7, %ptr
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6381%result0 = call i32 @llvm.atomic.load.min.i32.p0i32( i32* %ptr, i32 -2 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6382 <i>; yields {i32}:result0 = 7</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6383%result1 = call i32 @llvm.atomic.load.max.i32.p0i32( i32* %ptr, i32 8 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6384 <i>; yields {i32}:result1 = -2</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6385%result2 = call i32 @llvm.atomic.load.umin.i32.p0i32( i32* %ptr, i32 10 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6386 <i>; yields {i32}:result2 = 8</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6387%result3 = call i32 @llvm.atomic.load.umax.i32.p0i32( i32* %ptr, i32 30 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6388 <i>; yields {i32}:result3 = 8</i>
6389%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 30</i>
6390</pre>
6391</div>
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]6392
6393<!-- ======================================================================= -->
6394<div class="doc_subsection">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6395 <a name="int_general">General Intrinsics</a>
6396</div>
6397
6398<div class="doc_text">
6399<p> This class of intrinsics is designed to be generic and has
6400no specific purpose. </p>
6401</div>
6402
6403<!-- _______________________________________________________________________ -->
6404<div class="doc_subsubsection">
6405 <a name="int_var_annotation">'<tt>llvm.var.annotation</tt>' Intrinsic</a>
6406</div>
6407
6408<div class="doc_text">
6409
6410<h5>Syntax:</h5>
6411<pre>
6412 declare void @llvm.var.annotation(i8* &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
6413</pre>
6414
6415<h5>Overview:</h5>
6416
6417<p>
6418The '<tt>llvm.var.annotation</tt>' intrinsic
6419</p>
6420
6421<h5>Arguments:</h5>
6422
6423<p>
6424The first argument is a pointer to a value, the second is a pointer to a
6425global string, the third is a pointer to a global string which is the source
6426file name, and the last argument is the line number.
6427</p>
6428
6429<h5>Semantics:</h5>
6430
6431<p>
Anton Korobeynikove6e764f2008-01-15 22:31:34 +0000[diff] [blame]6432This intrinsic allows annotation of local variables with arbitrary strings.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6433This can be useful for special purpose optimizations that want to look for these
Anton Korobeynikove6e764f2008-01-15 22:31:34 +0000[diff] [blame]6434annotations. These have no other defined use, they are ignored by code
6435generation and optimization.
6436</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6437</div>
6438
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]6439<!-- _______________________________________________________________________ -->
6440<div class="doc_subsubsection">
Tanya Lattnerc9869b12007-09-21 23:57:59 +0000[diff] [blame]6441 <a name="int_annotation">'<tt>llvm.annotation.*</tt>' Intrinsic</a>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]6442</div>
6443
6444<div class="doc_text">
6445
6446<h5>Syntax:</h5>
Tanya Lattnere545be72007-09-21 23:56:27 +0000[diff] [blame]6447<p>This is an overloaded intrinsic. You can use '<tt>llvm.annotation</tt>' on
6448any integer bit width.
6449</p>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]6450<pre>
Tanya Lattner09161fe2007-09-22 00:03:01 +0000[diff] [blame]6451 declare i8 @llvm.annotation.i8(i8 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
6452 declare i16 @llvm.annotation.i16(i16 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
6453 declare i32 @llvm.annotation.i32(i32 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
6454 declare i64 @llvm.annotation.i64(i64 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
6455 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]6456</pre>
6457
6458<h5>Overview:</h5>
Tanya Lattnere545be72007-09-21 23:56:27 +0000[diff] [blame]6459
6460<p>
6461The '<tt>llvm.annotation</tt>' intrinsic.
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]6462</p>
6463
6464<h5>Arguments:</h5>
6465
6466<p>
6467The first argument is an integer value (result of some expression),
6468the second is a pointer to a global string, the third is a pointer to a global
6469string which is the source file name, and the last argument is the line number.
Tanya Lattnere545be72007-09-21 23:56:27 +0000[diff] [blame]6470It returns the value of the first argument.
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]6471</p>
6472
6473<h5>Semantics:</h5>
6474
6475<p>
6476This intrinsic allows annotations to be put on arbitrary expressions
6477with arbitrary strings. This can be useful for special purpose optimizations
6478that want to look for these annotations. These have no other defined use, they
6479are ignored by code generation and optimization.
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]6480</p>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]6481</div>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6482
Anton Korobeynikove6e764f2008-01-15 22:31:34 +0000[diff] [blame]6483<!-- _______________________________________________________________________ -->
6484<div class="doc_subsubsection">
6485 <a name="int_trap">'<tt>llvm.trap</tt>' Intrinsic</a>
6486</div>
6487
6488<div class="doc_text">
6489
6490<h5>Syntax:</h5>
6491<pre>
6492 declare void @llvm.trap()
6493</pre>
6494
6495<h5>Overview:</h5>
6496
6497<p>
6498The '<tt>llvm.trap</tt>' intrinsic
6499</p>
6500
6501<h5>Arguments:</h5>
6502
6503<p>
6504None
6505</p>
6506
6507<h5>Semantics:</h5>
6508
6509<p>
6510This intrinsics is lowered to the target dependent trap instruction. If the
6511target does not have a trap instruction, this intrinsic will be lowered to the
6512call of the abort() function.
6513</p>
6514</div>
6515
Bill Wendlinge4164592008-11-19 05:56:17 +0000[diff] [blame]6516<!-- _______________________________________________________________________ -->
6517<div class="doc_subsubsection">
Misha Brukman5dd7f4d2008-11-22 23:55:29 +0000[diff] [blame]6518 <a name="int_stackprotector">'<tt>llvm.stackprotector</tt>' Intrinsic</a>
Bill Wendlinge4164592008-11-19 05:56:17 +0000[diff] [blame]6519</div>
6520<div class="doc_text">
6521<h5>Syntax:</h5>
6522<pre>
6523declare void @llvm.stackprotector( i8* &lt;guard&gt;, i8** &lt;slot&gt; )
6524
6525</pre>
6526<h5>Overview:</h5>
6527<p>
6528 The <tt>llvm.stackprotector</tt> intrinsic takes the <tt>guard</tt> and stores
6529 it onto the stack at <tt>slot</tt>. The stack slot is adjusted to ensure that
6530 it is placed on the stack before local variables.
6531</p>
6532<h5>Arguments:</h5>
6533<p>
6534 The <tt>llvm.stackprotector</tt> intrinsic requires two pointer arguments. The
6535 first argument is the value loaded from the stack guard
6536 <tt>@__stack_chk_guard</tt>. The second variable is an <tt>alloca</tt> that
6537 has enough space to hold the value of the guard.
6538</p>
6539<h5>Semantics:</h5>
6540<p>
6541 This intrinsic causes the prologue/epilogue inserter to force the position of
6542 the <tt>AllocaInst</tt> stack slot to be before local variables on the
6543 stack. This is to ensure that if a local variable on the stack is overwritten,
6544 it will destroy the value of the guard. When the function exits, the guard on
6545 the stack is checked against the original guard. If they're different, then
6546 the program aborts by calling the <tt>__stack_chk_fail()</tt> function.
6547</p>
6548</div>
6549
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6550<!-- *********************************************************************** -->
6551<hr>
6552<address>
6553 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
Misha Brukman947321d2008-12-11 17:34:48 +0000[diff] [blame]6554 src="http://jigsaw.w3.org/css-validator/images/vcss-blue" alt="Valid CSS"></a>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6555 <a href="http://validator.w3.org/check/referer"><img
Misha Brukman947321d2008-12-11 17:34:48 +0000[diff] [blame]6556 src="http://www.w3.org/Icons/valid-html401-blue" alt="Valid HTML 4.01"></a>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6557
6558 <a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
6559 <a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br>
6560 Last modified: $Date$
6561</address>
Chris Lattner08497ce2008-01-04 04:33:49 +0000[diff] [blame]6562
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6563</body>
6564</html>