blob: 59dacf7ef40d626c8d9ee2f84af6e0ff2e464f47 [file] [log] [blame]
Misha Brukmandaa4cb02004-03-01 17:47:27 +00001<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN"
2 "http://www.w3.org/TR/html4/strict.dtd">
Misha Brukman9d0919f2003-11-08 01:05:38 +00003<html>
4<head>
5 <title>LLVM Assembly Language Reference Manual</title>
Reid Spencer3921c742004-08-26 20:44:00 +00006 <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.">
Misha Brukman9d0919f2003-11-08 01:05:38 +000010 <link rel="stylesheet" href="llvm.css" type="text/css">
11</head>
Chris Lattnerd7923912004-05-23 21:06:01 +000012
Misha Brukman9d0919f2003-11-08 01:05:38 +000013<body>
Chris Lattnerd7923912004-05-23 21:06:01 +000014
Chris Lattner261efe92003-11-25 01:02:51 +000015<div class="doc_title"> LLVM Language Reference Manual </div>
Chris Lattner00950542001-06-06 20:29:01 +000016<ol>
Misha Brukman9d0919f2003-11-08 01:05:38 +000017 <li><a href="#abstract">Abstract</a></li>
18 <li><a href="#introduction">Introduction</a></li>
19 <li><a href="#identifiers">Identifiers</a></li>
Chris Lattnerfa730212004-12-09 16:11:40 +000020 <li><a href="#highlevel">High Level Structure</a>
21 <ol>
22 <li><a href="#modulestructure">Module Structure</a></li>
Chris Lattnere5d947b2004-12-09 16:36:40 +000023 <li><a href="#linkage">Linkage Types</a></li>
Chris Lattnerbad10ee2005-05-06 22:57:40 +000024 <li><a href="#callingconv">Calling Conventions</a></li>
Chris Lattnerfa730212004-12-09 16:11:40 +000025 <li><a href="#globalvars">Global Variables</a></li>
26 <li><a href="#functionstructure">Function Structure</a></li>
27 </ol>
28 </li>
Chris Lattner00950542001-06-06 20:29:01 +000029 <li><a href="#typesystem">Type System</a>
30 <ol>
Chris Lattner261efe92003-11-25 01:02:51 +000031 <li><a href="#t_primitive">Primitive Types</a>
32 <ol>
Misha Brukman9d0919f2003-11-08 01:05:38 +000033 <li><a href="#t_classifications">Type Classifications</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +000034 </ol>
35 </li>
Chris Lattner00950542001-06-06 20:29:01 +000036 <li><a href="#t_derived">Derived Types</a>
37 <ol>
Chris Lattner261efe92003-11-25 01:02:51 +000038 <li><a href="#t_array">Array Type</a></li>
Misha Brukman9d0919f2003-11-08 01:05:38 +000039 <li><a href="#t_function">Function Type</a></li>
40 <li><a href="#t_pointer">Pointer Type</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +000041 <li><a href="#t_struct">Structure Type</a></li>
Chris Lattnera58561b2004-08-12 19:12:28 +000042 <li><a href="#t_packed">Packed Type</a></li>
Chris Lattner69c11bb2005-04-25 17:34:15 +000043 <li><a href="#t_opaque">Opaque Type</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +000044 </ol>
45 </li>
46 </ol>
47 </li>
Chris Lattnerfa730212004-12-09 16:11:40 +000048 <li><a href="#constants">Constants</a>
Chris Lattnerc3f59762004-12-09 17:30:23 +000049 <ol>
50 <li><a href="#simpleconstants">Simple Constants</a>
51 <li><a href="#aggregateconstants">Aggregate Constants</a>
52 <li><a href="#globalconstants">Global Variable and Function Addresses</a>
53 <li><a href="#undefvalues">Undefined Values</a>
54 <li><a href="#constantexprs">Constant Expressions</a>
55 </ol>
Chris Lattner261efe92003-11-25 01:02:51 +000056 </li>
Chris Lattner00950542001-06-06 20:29:01 +000057 <li><a href="#instref">Instruction Reference</a>
58 <ol>
59 <li><a href="#terminators">Terminator Instructions</a>
60 <ol>
Chris Lattner261efe92003-11-25 01:02:51 +000061 <li><a href="#i_ret">'<tt>ret</tt>' Instruction</a></li>
62 <li><a href="#i_br">'<tt>br</tt>' Instruction</a></li>
Misha Brukman9d0919f2003-11-08 01:05:38 +000063 <li><a href="#i_switch">'<tt>switch</tt>' Instruction</a></li>
64 <li><a href="#i_invoke">'<tt>invoke</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +000065 <li><a href="#i_unwind">'<tt>unwind</tt>' Instruction</a></li>
Chris Lattner35eca582004-10-16 18:04:13 +000066 <li><a href="#i_unreachable">'<tt>unreachable</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +000067 </ol>
68 </li>
Chris Lattner00950542001-06-06 20:29:01 +000069 <li><a href="#binaryops">Binary Operations</a>
70 <ol>
Chris Lattner261efe92003-11-25 01:02:51 +000071 <li><a href="#i_add">'<tt>add</tt>' Instruction</a></li>
72 <li><a href="#i_sub">'<tt>sub</tt>' Instruction</a></li>
73 <li><a href="#i_mul">'<tt>mul</tt>' Instruction</a></li>
74 <li><a href="#i_div">'<tt>div</tt>' Instruction</a></li>
75 <li><a href="#i_rem">'<tt>rem</tt>' Instruction</a></li>
Misha Brukman9d0919f2003-11-08 01:05:38 +000076 <li><a href="#i_setcc">'<tt>set<i>cc</i></tt>' Instructions</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +000077 </ol>
78 </li>
Chris Lattner00950542001-06-06 20:29:01 +000079 <li><a href="#bitwiseops">Bitwise Binary Operations</a>
80 <ol>
Misha Brukman9d0919f2003-11-08 01:05:38 +000081 <li><a href="#i_and">'<tt>and</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +000082 <li><a href="#i_or">'<tt>or</tt>' Instruction</a></li>
Misha Brukman9d0919f2003-11-08 01:05:38 +000083 <li><a href="#i_xor">'<tt>xor</tt>' Instruction</a></li>
84 <li><a href="#i_shl">'<tt>shl</tt>' Instruction</a></li>
85 <li><a href="#i_shr">'<tt>shr</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +000086 </ol>
87 </li>
Chris Lattner00950542001-06-06 20:29:01 +000088 <li><a href="#memoryops">Memory Access Operations</a>
89 <ol>
Chris Lattner261efe92003-11-25 01:02:51 +000090 <li><a href="#i_malloc">'<tt>malloc</tt>' Instruction</a></li>
91 <li><a href="#i_free">'<tt>free</tt>' Instruction</a></li>
92 <li><a href="#i_alloca">'<tt>alloca</tt>' Instruction</a></li>
93 <li><a href="#i_load">'<tt>load</tt>' Instruction</a></li>
94 <li><a href="#i_store">'<tt>store</tt>' Instruction</a></li>
95 <li><a href="#i_getelementptr">'<tt>getelementptr</tt>' Instruction</a></li>
96 </ol>
97 </li>
Chris Lattner00950542001-06-06 20:29:01 +000098 <li><a href="#otherops">Other Operations</a>
99 <ol>
Chris Lattner261efe92003-11-25 01:02:51 +0000100 <li><a href="#i_phi">'<tt>phi</tt>' Instruction</a></li>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000101 <li><a href="#i_cast">'<tt>cast .. to</tt>' Instruction</a></li>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000102 <li><a href="#i_select">'<tt>select</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000103 <li><a href="#i_call">'<tt>call</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000104 <li><a href="#i_vaarg">'<tt>vaarg</tt>' Instruction</a></li>
Chris Lattner00950542001-06-06 20:29:01 +0000105 </ol>
Chris Lattner261efe92003-11-25 01:02:51 +0000106 </li>
Chris Lattner00950542001-06-06 20:29:01 +0000107 </ol>
Chris Lattner261efe92003-11-25 01:02:51 +0000108 </li>
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000109 <li><a href="#intrinsics">Intrinsic Functions</a>
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000110 <ol>
Chris Lattner261efe92003-11-25 01:02:51 +0000111 <li><a href="#int_varargs">Variable Argument Handling Intrinsics</a>
112 <ol>
113 <li><a href="#i_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a></li>
114 <li><a href="#i_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a></li>
115 <li><a href="#i_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a></li>
116 </ol>
117 </li>
Chris Lattnerd7923912004-05-23 21:06:01 +0000118 <li><a href="#int_gc">Accurate Garbage Collection Intrinsics</a>
119 <ol>
120 <li><a href="#i_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a></li>
121 <li><a href="#i_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a></li>
122 <li><a href="#i_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a></li>
123 </ol>
124 </li>
Chris Lattner10610642004-02-14 04:08:35 +0000125 <li><a href="#int_codegen">Code Generator Intrinsics</a>
126 <ol>
127 <li><a href="#i_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a></li>
128 <li><a href="#i_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a></li>
Chris Lattner9a9d7ac2005-02-28 19:24:19 +0000129 <li><a href="#i_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a></li>
Andrew Lenharth7f4ec3b2005-03-28 20:05:49 +0000130 <li><a href="#i_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a></li>
John Criswell7123e272004-04-09 16:43:20 +0000131 </ol>
132 </li>
133 <li><a href="#int_os">Operating System Intrinsics</a>
134 <ol>
Chris Lattner32006282004-06-11 02:28:03 +0000135 <li><a href="#i_readport">'<tt>llvm.readport</tt>' Intrinsic</a></li>
136 <li><a href="#i_writeport">'<tt>llvm.writeport</tt>' Intrinsic</a></li>
John Criswell183402a2004-04-12 15:02:16 +0000137 <li><a href="#i_readio">'<tt>llvm.readio</tt>' Intrinsic</a></li>
138 <li><a href="#i_writeio">'<tt>llvm.writeio</tt>' Intrinsic</a></li>
Chris Lattner10610642004-02-14 04:08:35 +0000139 </ol>
Chris Lattner33aec9e2004-02-12 17:01:32 +0000140 <li><a href="#int_libc">Standard C Library Intrinsics</a>
141 <ol>
142 <li><a href="#i_memcpy">'<tt>llvm.memcpy</tt>' Intrinsic</a></li>
Chris Lattner0eb51b42004-02-12 18:10:10 +0000143 <li><a href="#i_memmove">'<tt>llvm.memmove</tt>' Intrinsic</a></li>
Chris Lattner10610642004-02-14 04:08:35 +0000144 <li><a href="#i_memset">'<tt>llvm.memset</tt>' Intrinsic</a></li>
Alkis Evlogimenos96853722004-06-12 19:19:14 +0000145 <li><a href="#i_isunordered">'<tt>llvm.isunordered</tt>' Intrinsic</a></li>
Chris Lattnera4d74142005-07-21 01:29:16 +0000146 <li><a href="#i_sqrt">'<tt>llvm.sqrt</tt>' Intrinsic</a></li>
147
Chris Lattner33aec9e2004-02-12 17:01:32 +0000148 </ol>
149 </li>
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000150 <li><a href="#int_count">Bit counting Intrinsics</a>
151 <ol>
152 <li><a href="#int_ctpop">'<tt>llvm.ctpop</tt>' Intrinsic </a></li>
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000153 <li><a href="#int_ctlz">'<tt>llvm.ctlz</tt>' Intrinsic </a></li>
Chris Lattnereff29ab2005-05-15 19:39:26 +0000154 <li><a href="#int_cttz">'<tt>llvm.cttz</tt>' Intrinsic </a></li>
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000155 </ol>
156 </li>
Chris Lattnerd7923912004-05-23 21:06:01 +0000157 <li><a href="#int_debugger">Debugger intrinsics</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000158 </ol>
159 </li>
Chris Lattner00950542001-06-06 20:29:01 +0000160</ol>
Chris Lattnerd7923912004-05-23 21:06:01 +0000161
162<div class="doc_author">
163 <p>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a>
164 and <a href="mailto:vadve@cs.uiuc.edu">Vikram Adve</a></p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000165</div>
Chris Lattnerd7923912004-05-23 21:06:01 +0000166
Chris Lattner00950542001-06-06 20:29:01 +0000167<!-- *********************************************************************** -->
Chris Lattner261efe92003-11-25 01:02:51 +0000168<div class="doc_section"> <a name="abstract">Abstract </a></div>
169<!-- *********************************************************************** -->
Chris Lattnerd7923912004-05-23 21:06:01 +0000170
Misha Brukman9d0919f2003-11-08 01:05:38 +0000171<div class="doc_text">
Chris Lattner261efe92003-11-25 01:02:51 +0000172<p>This document is a reference manual for the LLVM assembly language.
173LLVM is an SSA based representation that provides type safety,
174low-level operations, flexibility, and the capability of representing
175'all' high-level languages cleanly. It is the common code
176representation used throughout all phases of the LLVM compilation
177strategy.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000178</div>
Chris Lattnerd7923912004-05-23 21:06:01 +0000179
Chris Lattner00950542001-06-06 20:29:01 +0000180<!-- *********************************************************************** -->
Chris Lattner261efe92003-11-25 01:02:51 +0000181<div class="doc_section"> <a name="introduction">Introduction</a> </div>
182<!-- *********************************************************************** -->
Chris Lattnerd7923912004-05-23 21:06:01 +0000183
Misha Brukman9d0919f2003-11-08 01:05:38 +0000184<div class="doc_text">
Chris Lattnerd7923912004-05-23 21:06:01 +0000185
Chris Lattner261efe92003-11-25 01:02:51 +0000186<p>The LLVM code representation is designed to be used in three
187different forms: as an in-memory compiler IR, as an on-disk bytecode
188representation (suitable for fast loading by a Just-In-Time compiler),
189and as a human readable assembly language representation. This allows
190LLVM to provide a powerful intermediate representation for efficient
191compiler transformations and analysis, while providing a natural means
192to debug and visualize the transformations. The three different forms
193of LLVM are all equivalent. This document describes the human readable
194representation and notation.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000195
John Criswellc1f786c2005-05-13 22:25:59 +0000196<p>The LLVM representation aims to be light-weight and low-level
Chris Lattner261efe92003-11-25 01:02:51 +0000197while being expressive, typed, and extensible at the same time. It
198aims to be a "universal IR" of sorts, by being at a low enough level
199that high-level ideas may be cleanly mapped to it (similar to how
200microprocessors are "universal IR's", allowing many source languages to
201be mapped to them). By providing type information, LLVM can be used as
202the target of optimizations: for example, through pointer analysis, it
203can be proven that a C automatic variable is never accessed outside of
204the current function... allowing it to be promoted to a simple SSA
205value instead of a memory location.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000206
Misha Brukman9d0919f2003-11-08 01:05:38 +0000207</div>
Chris Lattnerd7923912004-05-23 21:06:01 +0000208
Chris Lattner00950542001-06-06 20:29:01 +0000209<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000210<div class="doc_subsubsection"> <a name="wellformed">Well-Formedness</a> </div>
Chris Lattnerd7923912004-05-23 21:06:01 +0000211
Misha Brukman9d0919f2003-11-08 01:05:38 +0000212<div class="doc_text">
Chris Lattnerd7923912004-05-23 21:06:01 +0000213
Chris Lattner261efe92003-11-25 01:02:51 +0000214<p>It is important to note that this document describes 'well formed'
215LLVM assembly language. There is a difference between what the parser
216accepts and what is considered 'well formed'. For example, the
217following instruction is syntactically okay, but not well formed:</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000218
219<pre>
220 %x = <a href="#i_add">add</a> int 1, %x
221</pre>
222
Chris Lattner261efe92003-11-25 01:02:51 +0000223<p>...because the definition of <tt>%x</tt> does not dominate all of
224its uses. The LLVM infrastructure provides a verification pass that may
225be used to verify that an LLVM module is well formed. This pass is
John Criswellc1f786c2005-05-13 22:25:59 +0000226automatically run by the parser after parsing input assembly and by
Chris Lattner261efe92003-11-25 01:02:51 +0000227the optimizer before it outputs bytecode. The violations pointed out
228by the verifier pass indicate bugs in transformation passes or input to
229the parser.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000230
Chris Lattner261efe92003-11-25 01:02:51 +0000231<!-- Describe the typesetting conventions here. --> </div>
Chris Lattnerd7923912004-05-23 21:06:01 +0000232
Chris Lattner00950542001-06-06 20:29:01 +0000233<!-- *********************************************************************** -->
Chris Lattner261efe92003-11-25 01:02:51 +0000234<div class="doc_section"> <a name="identifiers">Identifiers</a> </div>
Chris Lattner00950542001-06-06 20:29:01 +0000235<!-- *********************************************************************** -->
Chris Lattnerd7923912004-05-23 21:06:01 +0000236
Misha Brukman9d0919f2003-11-08 01:05:38 +0000237<div class="doc_text">
Chris Lattnerd7923912004-05-23 21:06:01 +0000238
Chris Lattner261efe92003-11-25 01:02:51 +0000239<p>LLVM uses three different forms of identifiers, for different
240purposes:</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000241
Chris Lattner00950542001-06-06 20:29:01 +0000242<ol>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000243 <li>Named values are represented as a string of characters with a '%' prefix.
244 For example, %foo, %DivisionByZero, %a.really.long.identifier. The actual
245 regular expression used is '<tt>%[a-zA-Z$._][a-zA-Z$._0-9]*</tt>'.
246 Identifiers which require other characters in their names can be surrounded
247 with quotes. In this way, anything except a <tt>"</tt> character can be used
248 in a name.</li>
249
250 <li>Unnamed values are represented as an unsigned numeric value with a '%'
251 prefix. For example, %12, %2, %44.</li>
252
Reid Spencercc16dc32004-12-09 18:02:53 +0000253 <li>Constants, which are described in a <a href="#constants">section about
254 constants</a>, below.</li>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000255</ol>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000256
257<p>LLVM requires that values start with a '%' sign for two reasons: Compilers
258don't need to worry about name clashes with reserved words, and the set of
259reserved words may be expanded in the future without penalty. Additionally,
260unnamed identifiers allow a compiler to quickly come up with a temporary
261variable without having to avoid symbol table conflicts.</p>
262
Chris Lattner261efe92003-11-25 01:02:51 +0000263<p>Reserved words in LLVM are very similar to reserved words in other
264languages. There are keywords for different opcodes ('<tt><a
Chris Lattnere5d947b2004-12-09 16:36:40 +0000265href="#i_add">add</a></tt>', '<tt><a href="#i_cast">cast</a></tt>', '<tt><a
266href="#i_ret">ret</a></tt>', etc...), for primitive type names ('<tt><a
267href="#t_void">void</a></tt>', '<tt><a href="#t_uint">uint</a></tt>', etc...),
268and others. These reserved words cannot conflict with variable names, because
269none of them start with a '%' character.</p>
270
271<p>Here is an example of LLVM code to multiply the integer variable
272'<tt>%X</tt>' by 8:</p>
273
Misha Brukman9d0919f2003-11-08 01:05:38 +0000274<p>The easy way:</p>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000275
276<pre>
277 %result = <a href="#i_mul">mul</a> uint %X, 8
278</pre>
279
Misha Brukman9d0919f2003-11-08 01:05:38 +0000280<p>After strength reduction:</p>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000281
282<pre>
283 %result = <a href="#i_shl">shl</a> uint %X, ubyte 3
284</pre>
285
Misha Brukman9d0919f2003-11-08 01:05:38 +0000286<p>And the hard way:</p>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000287
288<pre>
289 <a href="#i_add">add</a> uint %X, %X <i>; yields {uint}:%0</i>
290 <a href="#i_add">add</a> uint %0, %0 <i>; yields {uint}:%1</i>
291 %result = <a href="#i_add">add</a> uint %1, %1
292</pre>
293
Chris Lattner261efe92003-11-25 01:02:51 +0000294<p>This last way of multiplying <tt>%X</tt> by 8 illustrates several
295important lexical features of LLVM:</p>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000296
Chris Lattner00950542001-06-06 20:29:01 +0000297<ol>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000298
299 <li>Comments are delimited with a '<tt>;</tt>' and go until the end of
300 line.</li>
301
302 <li>Unnamed temporaries are created when the result of a computation is not
303 assigned to a named value.</li>
304
Misha Brukman9d0919f2003-11-08 01:05:38 +0000305 <li>Unnamed temporaries are numbered sequentially</li>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000306
Misha Brukman9d0919f2003-11-08 01:05:38 +0000307</ol>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000308
John Criswelle4c57cc2005-05-12 16:52:32 +0000309<p>...and it also shows a convention that we follow in this document. When
Chris Lattnere5d947b2004-12-09 16:36:40 +0000310demonstrating instructions, we will follow an instruction with a comment that
311defines the type and name of value produced. Comments are shown in italic
312text.</p>
313
Misha Brukman9d0919f2003-11-08 01:05:38 +0000314</div>
Chris Lattnerfa730212004-12-09 16:11:40 +0000315
316<!-- *********************************************************************** -->
317<div class="doc_section"> <a name="highlevel">High Level Structure</a> </div>
318<!-- *********************************************************************** -->
319
320<!-- ======================================================================= -->
321<div class="doc_subsection"> <a name="modulestructure">Module Structure</a>
322</div>
323
324<div class="doc_text">
325
326<p>LLVM programs are composed of "Module"s, each of which is a
327translation unit of the input programs. Each module consists of
328functions, global variables, and symbol table entries. Modules may be
329combined together with the LLVM linker, which merges function (and
330global variable) definitions, resolves forward declarations, and merges
331symbol table entries. Here is an example of the "hello world" module:</p>
332
333<pre><i>; Declare the string constant as a global constant...</i>
334<a href="#identifiers">%.LC0</a> = <a href="#linkage_internal">internal</a> <a
335 href="#globalvars">constant</a> <a href="#t_array">[13 x sbyte]</a> c"hello world\0A\00" <i>; [13 x sbyte]*</i>
336
337<i>; External declaration of the puts function</i>
338<a href="#functionstructure">declare</a> int %puts(sbyte*) <i>; int(sbyte*)* </i>
339
340<i>; Definition of main function</i>
341int %main() { <i>; int()* </i>
342 <i>; Convert [13x sbyte]* to sbyte *...</i>
343 %cast210 = <a
344 href="#i_getelementptr">getelementptr</a> [13 x sbyte]* %.LC0, long 0, long 0 <i>; sbyte*</i>
345
346 <i>; Call puts function to write out the string to stdout...</i>
347 <a
348 href="#i_call">call</a> int %puts(sbyte* %cast210) <i>; int</i>
349 <a
350 href="#i_ret">ret</a> int 0<br>}<br></pre>
351
352<p>This example is made up of a <a href="#globalvars">global variable</a>
353named "<tt>.LC0</tt>", an external declaration of the "<tt>puts</tt>"
354function, and a <a href="#functionstructure">function definition</a>
355for "<tt>main</tt>".</p>
356
Chris Lattnere5d947b2004-12-09 16:36:40 +0000357<p>In general, a module is made up of a list of global values,
358where both functions and global variables are global values. Global values are
359represented by a pointer to a memory location (in this case, a pointer to an
360array of char, and a pointer to a function), and have one of the following <a
361href="#linkage">linkage types</a>.</p>
Chris Lattnerfa730212004-12-09 16:11:40 +0000362
Chris Lattnere5d947b2004-12-09 16:36:40 +0000363</div>
364
365<!-- ======================================================================= -->
366<div class="doc_subsection">
367 <a name="linkage">Linkage Types</a>
368</div>
369
370<div class="doc_text">
371
372<p>
373All Global Variables and Functions have one of the following types of linkage:
374</p>
Chris Lattnerfa730212004-12-09 16:11:40 +0000375
376<dl>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000377
Chris Lattnerfa730212004-12-09 16:11:40 +0000378 <dt><tt><b><a name="linkage_internal">internal</a></b></tt> </dt>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000379
380 <dd>Global values with internal linkage are only directly accessible by
381 objects in the current module. In particular, linking code into a module with
382 an internal global value may cause the internal to be renamed as necessary to
383 avoid collisions. Because the symbol is internal to the module, all
384 references can be updated. This corresponds to the notion of the
385 '<tt>static</tt>' keyword in C, or the idea of "anonymous namespaces" in C++.
Chris Lattnerfa730212004-12-09 16:11:40 +0000386 </dd>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000387
Chris Lattnerfa730212004-12-09 16:11:40 +0000388 <dt><tt><b><a name="linkage_linkonce">linkonce</a></b></tt>: </dt>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000389
390 <dd>"<tt>linkonce</tt>" linkage is similar to <tt>internal</tt> linkage, with
391 the twist that linking together two modules defining the same
392 <tt>linkonce</tt> globals will cause one of the globals to be discarded. This
393 is typically used to implement inline functions. Unreferenced
394 <tt>linkonce</tt> globals are allowed to be discarded.
Chris Lattnerfa730212004-12-09 16:11:40 +0000395 </dd>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000396
Chris Lattnerfa730212004-12-09 16:11:40 +0000397 <dt><tt><b><a name="linkage_weak">weak</a></b></tt>: </dt>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000398
399 <dd>"<tt>weak</tt>" linkage is exactly the same as <tt>linkonce</tt> linkage,
400 except that unreferenced <tt>weak</tt> globals may not be discarded. This is
401 used to implement constructs in C such as "<tt>int X;</tt>" at global scope.
Chris Lattnerfa730212004-12-09 16:11:40 +0000402 </dd>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000403
Chris Lattnerfa730212004-12-09 16:11:40 +0000404 <dt><tt><b><a name="linkage_appending">appending</a></b></tt>: </dt>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000405
406 <dd>"<tt>appending</tt>" linkage may only be applied to global variables of
407 pointer to array type. When two global variables with appending linkage are
408 linked together, the two global arrays are appended together. This is the
409 LLVM, typesafe, equivalent of having the system linker append together
410 "sections" with identical names when .o files are linked.
Chris Lattnerfa730212004-12-09 16:11:40 +0000411 </dd>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000412
Chris Lattnerfa730212004-12-09 16:11:40 +0000413 <dt><tt><b><a name="linkage_external">externally visible</a></b></tt>:</dt>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000414
415 <dd>If none of the above identifiers are used, the global is externally
416 visible, meaning that it participates in linkage and can be used to resolve
417 external symbol references.
Chris Lattnerfa730212004-12-09 16:11:40 +0000418 </dd>
419</dl>
420
Chris Lattnerfa730212004-12-09 16:11:40 +0000421<p><a name="linkage_external">For example, since the "<tt>.LC0</tt>"
422variable is defined to be internal, if another module defined a "<tt>.LC0</tt>"
423variable and was linked with this one, one of the two would be renamed,
424preventing a collision. Since "<tt>main</tt>" and "<tt>puts</tt>" are
425external (i.e., lacking any linkage declarations), they are accessible
426outside of the current module. It is illegal for a function <i>declaration</i>
427to have any linkage type other than "externally visible".</a></p>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000428
Chris Lattnerfa730212004-12-09 16:11:40 +0000429</div>
430
431<!-- ======================================================================= -->
432<div class="doc_subsection">
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000433 <a name="callingconv">Calling Conventions</a>
434</div>
435
436<div class="doc_text">
437
438<p>LLVM <a href="#functionstructure">functions</a>, <a href="#i_call">calls</a>
439and <a href="#i_invoke">invokes</a> can all have an optional calling convention
440specified for the call. The calling convention of any pair of dynamic
441caller/callee must match, or the behavior of the program is undefined. The
442following calling conventions are supported by LLVM, and more may be added in
443the future:</p>
444
445<dl>
446 <dt><b>"<tt>ccc</tt>" - The C calling convention</b>:</dt>
447
448 <dd>This calling convention (the default if no other calling convention is
449 specified) matches the target C calling conventions. This calling convention
John Criswelle4c57cc2005-05-12 16:52:32 +0000450 supports varargs function calls and tolerates some mismatch in the declared
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000451 prototype and implemented declaration of the function (as does normal C).
452 </dd>
453
454 <dt><b>"<tt>fastcc</tt>" - The fast calling convention</b>:</dt>
455
456 <dd>This calling convention attempts to make calls as fast as possible
457 (e.g. by passing things in registers). This calling convention allows the
458 target to use whatever tricks it wants to produce fast code for the target,
Chris Lattner8cdc5bc2005-05-06 23:08:23 +0000459 without having to conform to an externally specified ABI. Implementations of
460 this convention should allow arbitrary tail call optimization to be supported.
461 This calling convention does not support varargs and requires the prototype of
462 all callees to exactly match the prototype of the function definition.
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000463 </dd>
464
465 <dt><b>"<tt>coldcc</tt>" - The cold calling convention</b>:</dt>
466
467 <dd>This calling convention attempts to make code in the caller as efficient
468 as possible under the assumption that the call is not commonly executed. As
469 such, these calls often preserve all registers so that the call does not break
470 any live ranges in the caller side. This calling convention does not support
471 varargs and requires the prototype of all callees to exactly match the
472 prototype of the function definition.
473 </dd>
474
Chris Lattnercfe6b372005-05-07 01:46:40 +0000475 <dt><b>"<tt>cc &lt;<em>n</em>&gt;</tt>" - Numbered convention</b>:</dt>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000476
477 <dd>Any calling convention may be specified by number, allowing
478 target-specific calling conventions to be used. Target specific calling
479 conventions start at 64.
480 </dd>
Chris Lattnercfe6b372005-05-07 01:46:40 +0000481</dl>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000482
483<p>More calling conventions can be added/defined on an as-needed basis, to
484support pascal conventions or any other well-known target-independent
485convention.</p>
486
487</div>
488
489<!-- ======================================================================= -->
490<div class="doc_subsection">
Chris Lattnerfa730212004-12-09 16:11:40 +0000491 <a name="globalvars">Global Variables</a>
492</div>
493
494<div class="doc_text">
495
Chris Lattner3689a342005-02-12 19:30:21 +0000496<p>Global variables define regions of memory allocated at compilation time
Chris Lattner2cbdc452005-11-06 08:02:57 +0000497instead of run-time. Global variables may optionally be initialized, and may
498have an optional explicit alignment specified. A
John Criswell0ec250c2005-10-24 16:17:18 +0000499variable may be defined as a global "constant," which indicates that the
Chris Lattner3689a342005-02-12 19:30:21 +0000500contents of the variable will <b>never</b> be modified (enabling better
501optimization, allowing the global data to be placed in the read-only section of
502an executable, etc). Note that variables that need runtime initialization
John Criswell0ec250c2005-10-24 16:17:18 +0000503cannot be marked "constant" as there is a store to the variable.</p>
Chris Lattner3689a342005-02-12 19:30:21 +0000504
505<p>
506LLVM explicitly allows <em>declarations</em> of global variables to be marked
507constant, even if the final definition of the global is not. This capability
508can be used to enable slightly better optimization of the program, but requires
509the language definition to guarantee that optimizations based on the
510'constantness' are valid for the translation units that do not include the
511definition.
512</p>
Chris Lattnerfa730212004-12-09 16:11:40 +0000513
514<p>As SSA values, global variables define pointer values that are in
515scope (i.e. they dominate) all basic blocks in the program. Global
516variables always define a pointer to their "content" type because they
517describe a region of memory, and all memory objects in LLVM are
518accessed through pointers.</p>
519
Chris Lattner2cbdc452005-11-06 08:02:57 +0000520<p>An explicit alignment may be specified for a global. If not present, or if
521the alignment is set to zero, the alignment of the global is set by the target
522to whatever it feels convenient. If an explicit alignment is specified, the
523global is forced to have at least that much alignment. All alignments must be
524a power of 2.</p>
525
Chris Lattnerfa730212004-12-09 16:11:40 +0000526</div>
527
528
529<!-- ======================================================================= -->
530<div class="doc_subsection">
531 <a name="functionstructure">Functions</a>
532</div>
533
534<div class="doc_text">
535
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000536<p>LLVM function definitions consist of an optional <a href="#linkage">linkage
537type</a>, an optional <a href="#callingconv">calling convention</a>, a return
Chris Lattner2cbdc452005-11-06 08:02:57 +0000538type, a function name, a (possibly empty) argument list, an optional alignment,
539an opening curly brace,
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000540a list of basic blocks, and a closing curly brace. LLVM function declarations
541are defined with the "<tt>declare</tt>" keyword, an optional <a
Chris Lattner2cbdc452005-11-06 08:02:57 +0000542href="#callingconv">calling convention</a>, a return type, a function name,
543a possibly empty list of arguments, and an optional alignment.</p>
Chris Lattnerfa730212004-12-09 16:11:40 +0000544
545<p>A function definition contains a list of basic blocks, forming the CFG for
546the function. Each basic block may optionally start with a label (giving the
547basic block a symbol table entry), contains a list of instructions, and ends
548with a <a href="#terminators">terminator</a> instruction (such as a branch or
549function return).</p>
550
John Criswelle4c57cc2005-05-12 16:52:32 +0000551<p>The first basic block in a program is special in two ways: it is immediately
Chris Lattnerfa730212004-12-09 16:11:40 +0000552executed on entrance to the function, and it is not allowed to have predecessor
553basic blocks (i.e. there can not be any branches to the entry block of a
554function). Because the block can have no predecessors, it also cannot have any
555<a href="#i_phi">PHI nodes</a>.</p>
556
557<p>LLVM functions are identified by their name and type signature. Hence, two
558functions with the same name but different parameter lists or return values are
Chris Lattnerd4f6b172005-03-07 22:13:59 +0000559considered different functions, and LLVM will resolve references to each
Chris Lattnerfa730212004-12-09 16:11:40 +0000560appropriately.</p>
561
Chris Lattner2cbdc452005-11-06 08:02:57 +0000562<p>An explicit alignment may be specified for a function. If not present, or if
563the alignment is set to zero, the alignment of the function is set by the target
564to whatever it feels convenient. If an explicit alignment is specified, the
565function is forced to have at least that much alignment. All alignments must be
566a power of 2.</p>
567
Chris Lattnerfa730212004-12-09 16:11:40 +0000568</div>
569
570
571
Chris Lattner00950542001-06-06 20:29:01 +0000572<!-- *********************************************************************** -->
Chris Lattner261efe92003-11-25 01:02:51 +0000573<div class="doc_section"> <a name="typesystem">Type System</a> </div>
574<!-- *********************************************************************** -->
Chris Lattnerfa730212004-12-09 16:11:40 +0000575
Misha Brukman9d0919f2003-11-08 01:05:38 +0000576<div class="doc_text">
Chris Lattnerfa730212004-12-09 16:11:40 +0000577
Misha Brukman9d0919f2003-11-08 01:05:38 +0000578<p>The LLVM type system is one of the most important features of the
Chris Lattner261efe92003-11-25 01:02:51 +0000579intermediate representation. Being typed enables a number of
580optimizations to be performed on the IR directly, without having to do
581extra analyses on the side before the transformation. A strong type
582system makes it easier to read the generated code and enables novel
583analyses and transformations that are not feasible to perform on normal
584three address code representations.</p>
Chris Lattnerfa730212004-12-09 16:11:40 +0000585
586</div>
587
Chris Lattner00950542001-06-06 20:29:01 +0000588<!-- ======================================================================= -->
Chris Lattner261efe92003-11-25 01:02:51 +0000589<div class="doc_subsection"> <a name="t_primitive">Primitive Types</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000590<div class="doc_text">
John Criswell4457dc92004-04-09 16:48:45 +0000591<p>The primitive types are the fundamental building blocks of the LLVM
Chris Lattnerd4f6b172005-03-07 22:13:59 +0000592system. The current set of primitive types is as follows:</p>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000593
Reid Spencerd3f876c2004-11-01 08:19:36 +0000594<table class="layout">
595 <tr class="layout">
596 <td class="left">
597 <table>
Chris Lattner261efe92003-11-25 01:02:51 +0000598 <tbody>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000599 <tr><th>Type</th><th>Description</th></tr>
600 <tr><td><tt>void</tt></td><td>No value</td></tr>
Misha Brukmancfa87bc2005-04-22 18:02:52 +0000601 <tr><td><tt>ubyte</tt></td><td>Unsigned 8-bit value</td></tr>
602 <tr><td><tt>ushort</tt></td><td>Unsigned 16-bit value</td></tr>
603 <tr><td><tt>uint</tt></td><td>Unsigned 32-bit value</td></tr>
604 <tr><td><tt>ulong</tt></td><td>Unsigned 64-bit value</td></tr>
605 <tr><td><tt>float</tt></td><td>32-bit floating point value</td></tr>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000606 <tr><td><tt>label</tt></td><td>Branch destination</td></tr>
Chris Lattner261efe92003-11-25 01:02:51 +0000607 </tbody>
608 </table>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000609 </td>
610 <td class="right">
611 <table>
Chris Lattner261efe92003-11-25 01:02:51 +0000612 <tbody>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000613 <tr><th>Type</th><th>Description</th></tr>
614 <tr><td><tt>bool</tt></td><td>True or False value</td></tr>
Misha Brukmancfa87bc2005-04-22 18:02:52 +0000615 <tr><td><tt>sbyte</tt></td><td>Signed 8-bit value</td></tr>
616 <tr><td><tt>short</tt></td><td>Signed 16-bit value</td></tr>
617 <tr><td><tt>int</tt></td><td>Signed 32-bit value</td></tr>
618 <tr><td><tt>long</tt></td><td>Signed 64-bit value</td></tr>
619 <tr><td><tt>double</tt></td><td>64-bit floating point value</td></tr>
Chris Lattner261efe92003-11-25 01:02:51 +0000620 </tbody>
621 </table>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000622 </td>
623 </tr>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000624</table>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000625</div>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000626
Chris Lattner00950542001-06-06 20:29:01 +0000627<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000628<div class="doc_subsubsection"> <a name="t_classifications">Type
629Classifications</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000630<div class="doc_text">
Chris Lattner261efe92003-11-25 01:02:51 +0000631<p>These different primitive types fall into a few useful
632classifications:</p>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000633
634<table border="1" cellspacing="0" cellpadding="4">
Chris Lattner261efe92003-11-25 01:02:51 +0000635 <tbody>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000636 <tr><th>Classification</th><th>Types</th></tr>
Chris Lattner261efe92003-11-25 01:02:51 +0000637 <tr>
638 <td><a name="t_signed">signed</a></td>
639 <td><tt>sbyte, short, int, long, float, double</tt></td>
640 </tr>
641 <tr>
642 <td><a name="t_unsigned">unsigned</a></td>
643 <td><tt>ubyte, ushort, uint, ulong</tt></td>
644 </tr>
645 <tr>
646 <td><a name="t_integer">integer</a></td>
647 <td><tt>ubyte, sbyte, ushort, short, uint, int, ulong, long</tt></td>
648 </tr>
649 <tr>
650 <td><a name="t_integral">integral</a></td>
Misha Brukmanc24b7582004-08-12 20:16:08 +0000651 <td><tt>bool, ubyte, sbyte, ushort, short, uint, int, ulong, long</tt>
652 </td>
Chris Lattner261efe92003-11-25 01:02:51 +0000653 </tr>
654 <tr>
655 <td><a name="t_floating">floating point</a></td>
656 <td><tt>float, double</tt></td>
657 </tr>
658 <tr>
659 <td><a name="t_firstclass">first class</a></td>
Misha Brukmanc24b7582004-08-12 20:16:08 +0000660 <td><tt>bool, ubyte, sbyte, ushort, short, uint, int, ulong, long,<br>
661 float, double, <a href="#t_pointer">pointer</a>,
662 <a href="#t_packed">packed</a></tt></td>
Chris Lattner261efe92003-11-25 01:02:51 +0000663 </tr>
664 </tbody>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000665</table>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000666
Chris Lattner261efe92003-11-25 01:02:51 +0000667<p>The <a href="#t_firstclass">first class</a> types are perhaps the
668most important. Values of these types are the only ones which can be
669produced by instructions, passed as arguments, or used as operands to
670instructions. This means that all structures and arrays must be
671manipulated either by pointer or by component.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000672</div>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000673
Chris Lattner00950542001-06-06 20:29:01 +0000674<!-- ======================================================================= -->
Chris Lattner261efe92003-11-25 01:02:51 +0000675<div class="doc_subsection"> <a name="t_derived">Derived Types</a> </div>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000676
Misha Brukman9d0919f2003-11-08 01:05:38 +0000677<div class="doc_text">
Chris Lattnerc3f59762004-12-09 17:30:23 +0000678
Chris Lattner261efe92003-11-25 01:02:51 +0000679<p>The real power in LLVM comes from the derived types in the system.
680This is what allows a programmer to represent arrays, functions,
681pointers, and other useful types. Note that these derived types may be
682recursive: For example, it is possible to have a two dimensional array.</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000683
Misha Brukman9d0919f2003-11-08 01:05:38 +0000684</div>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000685
Chris Lattner00950542001-06-06 20:29:01 +0000686<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000687<div class="doc_subsubsection"> <a name="t_array">Array Type</a> </div>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000688
Misha Brukman9d0919f2003-11-08 01:05:38 +0000689<div class="doc_text">
Chris Lattnerc3f59762004-12-09 17:30:23 +0000690
Chris Lattner00950542001-06-06 20:29:01 +0000691<h5>Overview:</h5>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000692
Misha Brukman9d0919f2003-11-08 01:05:38 +0000693<p>The array type is a very simple derived type that arranges elements
Chris Lattner261efe92003-11-25 01:02:51 +0000694sequentially in memory. The array type requires a size (number of
695elements) and an underlying data type.</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000696
Chris Lattner7faa8832002-04-14 06:13:44 +0000697<h5>Syntax:</h5>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000698
699<pre>
700 [&lt;# elements&gt; x &lt;elementtype&gt;]
701</pre>
702
John Criswelle4c57cc2005-05-12 16:52:32 +0000703<p>The number of elements is a constant integer value; elementtype may
Chris Lattner261efe92003-11-25 01:02:51 +0000704be any type with a size.</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000705
Chris Lattner7faa8832002-04-14 06:13:44 +0000706<h5>Examples:</h5>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000707<table class="layout">
708 <tr class="layout">
709 <td class="left">
710 <tt>[40 x int ]</tt><br/>
711 <tt>[41 x int ]</tt><br/>
712 <tt>[40 x uint]</tt><br/>
713 </td>
714 <td class="left">
715 Array of 40 integer values.<br/>
716 Array of 41 integer values.<br/>
717 Array of 40 unsigned integer values.<br/>
718 </td>
719 </tr>
Chris Lattner00950542001-06-06 20:29:01 +0000720</table>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000721<p>Here are some examples of multidimensional arrays:</p>
722<table class="layout">
723 <tr class="layout">
724 <td class="left">
725 <tt>[3 x [4 x int]]</tt><br/>
726 <tt>[12 x [10 x float]]</tt><br/>
727 <tt>[2 x [3 x [4 x uint]]]</tt><br/>
728 </td>
729 <td class="left">
John Criswellc1f786c2005-05-13 22:25:59 +0000730 3x4 array of integer values.<br/>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000731 12x10 array of single precision floating point values.<br/>
732 2x3x4 array of unsigned integer values.<br/>
733 </td>
734 </tr>
735</table>
Chris Lattnere67a9512005-06-24 17:22:57 +0000736
John Criswell0ec250c2005-10-24 16:17:18 +0000737<p>Note that 'variable sized arrays' can be implemented in LLVM with a zero
738length array. Normally, accesses past the end of an array are undefined in
Chris Lattnere67a9512005-06-24 17:22:57 +0000739LLVM (e.g. it is illegal to access the 5th element of a 3 element array).
740As a special case, however, zero length arrays are recognized to be variable
741length. This allows implementation of 'pascal style arrays' with the LLVM
742type "{ int, [0 x float]}", for example.</p>
743
Misha Brukman9d0919f2003-11-08 01:05:38 +0000744</div>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000745
Chris Lattner00950542001-06-06 20:29:01 +0000746<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000747<div class="doc_subsubsection"> <a name="t_function">Function Type</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000748<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +0000749<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000750<p>The function type can be thought of as a function signature. It
751consists of a return type and a list of formal parameter types.
John Criswell009900b2003-11-25 21:45:46 +0000752Function types are usually used to build virtual function tables
Chris Lattner261efe92003-11-25 01:02:51 +0000753(which are structures of pointers to functions), for indirect function
754calls, and when defining a function.</p>
John Criswell009900b2003-11-25 21:45:46 +0000755<p>
756The return type of a function type cannot be an aggregate type.
757</p>
Chris Lattner00950542001-06-06 20:29:01 +0000758<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000759<pre> &lt;returntype&gt; (&lt;parameter list&gt;)<br></pre>
John Criswell0ec250c2005-10-24 16:17:18 +0000760<p>...where '<tt>&lt;parameter list&gt;</tt>' is a comma-separated list of type
Misha Brukmanc24b7582004-08-12 20:16:08 +0000761specifiers. Optionally, the parameter list may include a type <tt>...</tt>,
Chris Lattner27f71f22003-09-03 00:41:47 +0000762which indicates that the function takes a variable number of arguments.
763Variable argument functions can access their arguments with the <a
Chris Lattner261efe92003-11-25 01:02:51 +0000764 href="#int_varargs">variable argument handling intrinsic</a> functions.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000765<h5>Examples:</h5>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000766<table class="layout">
767 <tr class="layout">
768 <td class="left">
769 <tt>int (int)</tt> <br/>
770 <tt>float (int, int *) *</tt><br/>
771 <tt>int (sbyte *, ...)</tt><br/>
772 </td>
773 <td class="left">
774 function taking an <tt>int</tt>, returning an <tt>int</tt><br/>
775 <a href="#t_pointer">Pointer</a> to a function that takes an
Misha Brukmanc24b7582004-08-12 20:16:08 +0000776 <tt>int</tt> and a <a href="#t_pointer">pointer</a> to <tt>int</tt>,
Reid Spencerd3f876c2004-11-01 08:19:36 +0000777 returning <tt>float</tt>.<br/>
778 A vararg function that takes at least one <a href="#t_pointer">pointer</a>
779 to <tt>sbyte</tt> (signed char in C), which returns an integer. This is
780 the signature for <tt>printf</tt> in LLVM.<br/>
781 </td>
782 </tr>
Chris Lattner00950542001-06-06 20:29:01 +0000783</table>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000784
Misha Brukman9d0919f2003-11-08 01:05:38 +0000785</div>
Chris Lattner00950542001-06-06 20:29:01 +0000786<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000787<div class="doc_subsubsection"> <a name="t_struct">Structure Type</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000788<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +0000789<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000790<p>The structure type is used to represent a collection of data members
791together in memory. The packing of the field types is defined to match
792the ABI of the underlying processor. The elements of a structure may
793be any type that has a size.</p>
794<p>Structures are accessed using '<tt><a href="#i_load">load</a></tt>
795and '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a
796field with the '<tt><a href="#i_getelementptr">getelementptr</a></tt>'
797instruction.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000798<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000799<pre> { &lt;type list&gt; }<br></pre>
Chris Lattner00950542001-06-06 20:29:01 +0000800<h5>Examples:</h5>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000801<table class="layout">
802 <tr class="layout">
803 <td class="left">
804 <tt>{ int, int, int }</tt><br/>
805 <tt>{ float, int (int) * }</tt><br/>
806 </td>
807 <td class="left">
808 a triple of three <tt>int</tt> values<br/>
809 A pair, where the first element is a <tt>float</tt> and the second element
810 is a <a href="#t_pointer">pointer</a> to a <a href="#t_function">function</a>
811 that takes an <tt>int</tt>, returning an <tt>int</tt>.<br/>
812 </td>
813 </tr>
Chris Lattner00950542001-06-06 20:29:01 +0000814</table>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000815</div>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000816
Chris Lattner00950542001-06-06 20:29:01 +0000817<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000818<div class="doc_subsubsection"> <a name="t_pointer">Pointer Type</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000819<div class="doc_text">
Chris Lattner7faa8832002-04-14 06:13:44 +0000820<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000821<p>As in many languages, the pointer type represents a pointer or
822reference to another object, which must live in memory.</p>
Chris Lattner7faa8832002-04-14 06:13:44 +0000823<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000824<pre> &lt;type&gt; *<br></pre>
Chris Lattner7faa8832002-04-14 06:13:44 +0000825<h5>Examples:</h5>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000826<table class="layout">
827 <tr class="layout">
828 <td class="left">
829 <tt>[4x int]*</tt><br/>
830 <tt>int (int *) *</tt><br/>
831 </td>
832 <td class="left">
833 A <a href="#t_pointer">pointer</a> to <a href="#t_array">array</a> of
834 four <tt>int</tt> values<br/>
835 A <a href="#t_pointer">pointer</a> to a <a
Chris Lattnera977c482005-02-19 02:22:14 +0000836 href="#t_function">function</a> that takes an <tt>int*</tt>, returning an
Reid Spencerd3f876c2004-11-01 08:19:36 +0000837 <tt>int</tt>.<br/>
838 </td>
839 </tr>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000840</table>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000841</div>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000842
Chris Lattnera58561b2004-08-12 19:12:28 +0000843<!-- _______________________________________________________________________ -->
844<div class="doc_subsubsection"> <a name="t_packed">Packed Type</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000845<div class="doc_text">
Chris Lattner69c11bb2005-04-25 17:34:15 +0000846
Chris Lattnera58561b2004-08-12 19:12:28 +0000847<h5>Overview:</h5>
Chris Lattner69c11bb2005-04-25 17:34:15 +0000848
Chris Lattnera58561b2004-08-12 19:12:28 +0000849<p>A packed type is a simple derived type that represents a vector
850of elements. Packed types are used when multiple primitive data
851are operated in parallel using a single instruction (SIMD).
852A packed type requires a size (number of
Chris Lattnerb8d172f2005-11-10 01:44:22 +0000853elements) and an underlying primitive data type. Vectors must have a power
854of two length (1, 2, 4, 8, 16 ...). Packed types are
Chris Lattnera58561b2004-08-12 19:12:28 +0000855considered <a href="#t_firstclass">first class</a>.</p>
Chris Lattner69c11bb2005-04-25 17:34:15 +0000856
Chris Lattnera58561b2004-08-12 19:12:28 +0000857<h5>Syntax:</h5>
Chris Lattner69c11bb2005-04-25 17:34:15 +0000858
859<pre>
860 &lt; &lt;# elements&gt; x &lt;elementtype&gt; &gt;
861</pre>
862
John Criswellc1f786c2005-05-13 22:25:59 +0000863<p>The number of elements is a constant integer value; elementtype may
Chris Lattnera58561b2004-08-12 19:12:28 +0000864be any integral or floating point type.</p>
Chris Lattner69c11bb2005-04-25 17:34:15 +0000865
Chris Lattnera58561b2004-08-12 19:12:28 +0000866<h5>Examples:</h5>
Chris Lattner69c11bb2005-04-25 17:34:15 +0000867
Reid Spencerd3f876c2004-11-01 08:19:36 +0000868<table class="layout">
869 <tr class="layout">
870 <td class="left">
871 <tt>&lt;4 x int&gt;</tt><br/>
872 <tt>&lt;8 x float&gt;</tt><br/>
873 <tt>&lt;2 x uint&gt;</tt><br/>
874 </td>
875 <td class="left">
876 Packed vector of 4 integer values.<br/>
877 Packed vector of 8 floating-point values.<br/>
878 Packed vector of 2 unsigned integer values.<br/>
879 </td>
880 </tr>
881</table>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000882</div>
883
Chris Lattner69c11bb2005-04-25 17:34:15 +0000884<!-- _______________________________________________________________________ -->
885<div class="doc_subsubsection"> <a name="t_opaque">Opaque Type</a> </div>
886<div class="doc_text">
887
888<h5>Overview:</h5>
889
890<p>Opaque types are used to represent unknown types in the system. This
891corresponds (for example) to the C notion of a foward declared structure type.
892In LLVM, opaque types can eventually be resolved to any type (not just a
893structure type).</p>
894
895<h5>Syntax:</h5>
896
897<pre>
898 opaque
899</pre>
900
901<h5>Examples:</h5>
902
903<table class="layout">
904 <tr class="layout">
905 <td class="left">
906 <tt>opaque</tt>
907 </td>
908 <td class="left">
909 An opaque type.<br/>
910 </td>
911 </tr>
912</table>
913</div>
914
915
Chris Lattnerc3f59762004-12-09 17:30:23 +0000916<!-- *********************************************************************** -->
917<div class="doc_section"> <a name="constants">Constants</a> </div>
918<!-- *********************************************************************** -->
919
920<div class="doc_text">
921
922<p>LLVM has several different basic types of constants. This section describes
923them all and their syntax.</p>
924
925</div>
926
927<!-- ======================================================================= -->
Reid Spencercc16dc32004-12-09 18:02:53 +0000928<div class="doc_subsection"><a name="simpleconstants">Simple Constants</a></div>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000929
930<div class="doc_text">
931
932<dl>
933 <dt><b>Boolean constants</b></dt>
934
935 <dd>The two strings '<tt>true</tt>' and '<tt>false</tt>' are both valid
936 constants of the <tt><a href="#t_primitive">bool</a></tt> type.
937 </dd>
938
939 <dt><b>Integer constants</b></dt>
940
Reid Spencercc16dc32004-12-09 18:02:53 +0000941 <dd>Standard integers (such as '4') are constants of the <a
Chris Lattnerc3f59762004-12-09 17:30:23 +0000942 href="#t_integer">integer</a> type. Negative numbers may be used with signed
943 integer types.
944 </dd>
945
946 <dt><b>Floating point constants</b></dt>
947
948 <dd>Floating point constants use standard decimal notation (e.g. 123.421),
949 exponential notation (e.g. 1.23421e+2), or a more precise hexadecimal
Chris Lattnerc3f59762004-12-09 17:30:23 +0000950 notation (see below). Floating point constants must have a <a
951 href="#t_floating">floating point</a> type. </dd>
952
953 <dt><b>Null pointer constants</b></dt>
954
John Criswell9e2485c2004-12-10 15:51:16 +0000955 <dd>The identifier '<tt>null</tt>' is recognized as a null pointer constant
Chris Lattnerc3f59762004-12-09 17:30:23 +0000956 and must be of <a href="#t_pointer">pointer type</a>.</dd>
957
958</dl>
959
John Criswell9e2485c2004-12-10 15:51:16 +0000960<p>The one non-intuitive notation for constants is the optional hexadecimal form
Chris Lattnerc3f59762004-12-09 17:30:23 +0000961of floating point constants. For example, the form '<tt>double
9620x432ff973cafa8000</tt>' is equivalent to (but harder to read than) '<tt>double
9634.5e+15</tt>'. The only time hexadecimal floating point constants are required
Reid Spencercc16dc32004-12-09 18:02:53 +0000964(and the only time that they are generated by the disassembler) is when a
965floating point constant must be emitted but it cannot be represented as a
966decimal floating point number. For example, NaN's, infinities, and other
967special values are represented in their IEEE hexadecimal format so that
968assembly and disassembly do not cause any bits to change in the constants.</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000969
970</div>
971
972<!-- ======================================================================= -->
973<div class="doc_subsection"><a name="aggregateconstants">Aggregate Constants</a>
974</div>
975
976<div class="doc_text">
Chris Lattnerd4f6b172005-03-07 22:13:59 +0000977<p>Aggregate constants arise from aggregation of simple constants
978and smaller aggregate constants.</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000979
980<dl>
981 <dt><b>Structure constants</b></dt>
982
983 <dd>Structure constants are represented with notation similar to structure
984 type definitions (a comma separated list of elements, surrounded by braces
Chris Lattnerd4f6b172005-03-07 22:13:59 +0000985 (<tt>{}</tt>)). For example: "<tt>{ int 4, float 17.0, int* %G }</tt>",
986 where "<tt>%G</tt>" is declared as "<tt>%G = external global int</tt>". Structure constants
987 must have <a href="#t_struct">structure type</a>, and the number and
Chris Lattnerc3f59762004-12-09 17:30:23 +0000988 types of elements must match those specified by the type.
989 </dd>
990
991 <dt><b>Array constants</b></dt>
992
993 <dd>Array constants are represented with notation similar to array type
994 definitions (a comma separated list of elements, surrounded by square brackets
John Criswell9e2485c2004-12-10 15:51:16 +0000995 (<tt>[]</tt>)). For example: "<tt>[ int 42, int 11, int 74 ]</tt>". Array
Chris Lattnerc3f59762004-12-09 17:30:23 +0000996 constants must have <a href="#t_array">array type</a>, and the number and
997 types of elements must match those specified by the type.
998 </dd>
999
1000 <dt><b>Packed constants</b></dt>
1001
1002 <dd>Packed constants are represented with notation similar to packed type
1003 definitions (a comma separated list of elements, surrounded by
John Criswell9e2485c2004-12-10 15:51:16 +00001004 less-than/greater-than's (<tt>&lt;&gt;</tt>)). For example: "<tt>&lt; int 42,
Chris Lattnerc3f59762004-12-09 17:30:23 +00001005 int 11, int 74, int 100 &gt;</tt>". Packed constants must have <a
1006 href="#t_packed">packed type</a>, and the number and types of elements must
1007 match those specified by the type.
1008 </dd>
1009
1010 <dt><b>Zero initialization</b></dt>
1011
1012 <dd>The string '<tt>zeroinitializer</tt>' can be used to zero initialize a
1013 value to zero of <em>any</em> type, including scalar and aggregate types.
1014 This is often used to avoid having to print large zero initializers (e.g. for
John Criswell0ec250c2005-10-24 16:17:18 +00001015 large arrays) and is always exactly equivalent to using explicit zero
Chris Lattnerc3f59762004-12-09 17:30:23 +00001016 initializers.
1017 </dd>
1018</dl>
1019
1020</div>
1021
1022<!-- ======================================================================= -->
1023<div class="doc_subsection">
1024 <a name="globalconstants">Global Variable and Function Addresses</a>
1025</div>
1026
1027<div class="doc_text">
1028
1029<p>The addresses of <a href="#globalvars">global variables</a> and <a
1030href="#functionstructure">functions</a> are always implicitly valid (link-time)
John Criswell9e2485c2004-12-10 15:51:16 +00001031constants. These constants are explicitly referenced when the <a
1032href="#identifiers">identifier for the global</a> is used and always have <a
Chris Lattnerc3f59762004-12-09 17:30:23 +00001033href="#t_pointer">pointer</a> type. For example, the following is a legal LLVM
1034file:</p>
1035
1036<pre>
1037 %X = global int 17
1038 %Y = global int 42
1039 %Z = global [2 x int*] [ int* %X, int* %Y ]
1040</pre>
1041
1042</div>
1043
1044<!-- ======================================================================= -->
Reid Spencer2dc45b82004-12-09 18:13:12 +00001045<div class="doc_subsection"><a name="undefvalues">Undefined Values</a></div>
Chris Lattnerc3f59762004-12-09 17:30:23 +00001046<div class="doc_text">
Reid Spencer2dc45b82004-12-09 18:13:12 +00001047 <p>The string '<tt>undef</tt>' is recognized as a type-less constant that has
John Criswellc1f786c2005-05-13 22:25:59 +00001048 no specific value. Undefined values may be of any type and be used anywhere
Reid Spencer2dc45b82004-12-09 18:13:12 +00001049 a constant is permitted.</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +00001050
Reid Spencer2dc45b82004-12-09 18:13:12 +00001051 <p>Undefined values indicate to the compiler that the program is well defined
1052 no matter what value is used, giving the compiler more freedom to optimize.
1053 </p>
Chris Lattnerc3f59762004-12-09 17:30:23 +00001054</div>
1055
1056<!-- ======================================================================= -->
1057<div class="doc_subsection"><a name="constantexprs">Constant Expressions</a>
1058</div>
1059
1060<div class="doc_text">
1061
1062<p>Constant expressions are used to allow expressions involving other constants
1063to be used as constants. Constant expressions may be of any <a
John Criswellc1f786c2005-05-13 22:25:59 +00001064href="#t_firstclass">first class</a> type and may involve any LLVM operation
Chris Lattnerc3f59762004-12-09 17:30:23 +00001065that does not have side effects (e.g. load and call are not supported). The
1066following is the syntax for constant expressions:</p>
1067
1068<dl>
1069 <dt><b><tt>cast ( CST to TYPE )</tt></b></dt>
1070
1071 <dd>Cast a constant to another type.</dd>
1072
1073 <dt><b><tt>getelementptr ( CSTPTR, IDX0, IDX1, ... )</tt></b></dt>
1074
1075 <dd>Perform the <a href="#i_getelementptr">getelementptr operation</a> on
1076 constants. As with the <a href="#i_getelementptr">getelementptr</a>
1077 instruction, the index list may have zero or more indexes, which are required
1078 to make sense for the type of "CSTPTR".</dd>
1079
1080 <dt><b><tt>OPCODE ( LHS, RHS )</tt></b></dt>
1081
Reid Spencer2dc45b82004-12-09 18:13:12 +00001082 <dd>Perform the specified operation of the LHS and RHS constants. OPCODE may
1083 be any of the <a href="#binaryops">binary</a> or <a href="#bitwiseops">bitwise
Chris Lattnerc3f59762004-12-09 17:30:23 +00001084 binary</a> operations. The constraints on operands are the same as those for
1085 the corresponding instruction (e.g. no bitwise operations on floating point
John Criswelle4c57cc2005-05-12 16:52:32 +00001086 values are allowed).</dd>
Chris Lattnerc3f59762004-12-09 17:30:23 +00001087</dl>
Chris Lattnerc3f59762004-12-09 17:30:23 +00001088</div>
Chris Lattner9ee5d222004-03-08 16:49:10 +00001089
Chris Lattner00950542001-06-06 20:29:01 +00001090<!-- *********************************************************************** -->
Chris Lattner261efe92003-11-25 01:02:51 +00001091<div class="doc_section"> <a name="instref">Instruction Reference</a> </div>
1092<!-- *********************************************************************** -->
Chris Lattnerc3f59762004-12-09 17:30:23 +00001093
Misha Brukman9d0919f2003-11-08 01:05:38 +00001094<div class="doc_text">
Chris Lattnerc3f59762004-12-09 17:30:23 +00001095
Chris Lattner261efe92003-11-25 01:02:51 +00001096<p>The LLVM instruction set consists of several different
1097classifications of instructions: <a href="#terminators">terminator
John Criswellc1f786c2005-05-13 22:25:59 +00001098instructions</a>, <a href="#binaryops">binary instructions</a>,
1099<a href="#bitwiseops">bitwise binary instructions</a>, <a
Chris Lattner261efe92003-11-25 01:02:51 +00001100 href="#memoryops">memory instructions</a>, and <a href="#otherops">other
1101instructions</a>.</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +00001102
Misha Brukman9d0919f2003-11-08 01:05:38 +00001103</div>
Chris Lattnerc3f59762004-12-09 17:30:23 +00001104
Chris Lattner00950542001-06-06 20:29:01 +00001105<!-- ======================================================================= -->
Chris Lattner261efe92003-11-25 01:02:51 +00001106<div class="doc_subsection"> <a name="terminators">Terminator
1107Instructions</a> </div>
Chris Lattnerc3f59762004-12-09 17:30:23 +00001108
Misha Brukman9d0919f2003-11-08 01:05:38 +00001109<div class="doc_text">
Chris Lattnerc3f59762004-12-09 17:30:23 +00001110
Chris Lattner261efe92003-11-25 01:02:51 +00001111<p>As mentioned <a href="#functionstructure">previously</a>, every
1112basic block in a program ends with a "Terminator" instruction, which
1113indicates which block should be executed after the current block is
1114finished. These terminator instructions typically yield a '<tt>void</tt>'
1115value: they produce control flow, not values (the one exception being
1116the '<a href="#i_invoke"><tt>invoke</tt></a>' instruction).</p>
John Criswell9e2485c2004-12-10 15:51:16 +00001117<p>There are six different terminator instructions: the '<a
Chris Lattner261efe92003-11-25 01:02:51 +00001118 href="#i_ret"><tt>ret</tt></a>' instruction, the '<a href="#i_br"><tt>br</tt></a>'
1119instruction, the '<a href="#i_switch"><tt>switch</tt></a>' instruction,
Chris Lattner35eca582004-10-16 18:04:13 +00001120the '<a href="#i_invoke"><tt>invoke</tt></a>' instruction, the '<a
1121 href="#i_unwind"><tt>unwind</tt></a>' instruction, and the '<a
1122 href="#i_unreachable"><tt>unreachable</tt></a>' instruction.</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +00001123
Misha Brukman9d0919f2003-11-08 01:05:38 +00001124</div>
Chris Lattnerc3f59762004-12-09 17:30:23 +00001125
Chris Lattner00950542001-06-06 20:29:01 +00001126<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +00001127<div class="doc_subsubsection"> <a name="i_ret">'<tt>ret</tt>'
1128Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001129<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +00001130<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001131<pre> ret &lt;type&gt; &lt;value&gt; <i>; Return a value from a non-void function</i>
Chris Lattner7faa8832002-04-14 06:13:44 +00001132 ret void <i>; Return from void function</i>
Chris Lattner00950542001-06-06 20:29:01 +00001133</pre>
Chris Lattner00950542001-06-06 20:29:01 +00001134<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001135<p>The '<tt>ret</tt>' instruction is used to return control flow (and a
John Criswellc1f786c2005-05-13 22:25:59 +00001136value) from a function back to the caller.</p>
John Criswell4457dc92004-04-09 16:48:45 +00001137<p>There are two forms of the '<tt>ret</tt>' instruction: one that
Chris Lattner261efe92003-11-25 01:02:51 +00001138returns a value and then causes control flow, and one that just causes
1139control flow to occur.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001140<h5>Arguments:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001141<p>The '<tt>ret</tt>' instruction may return any '<a
1142 href="#t_firstclass">first class</a>' type. Notice that a function is
1143not <a href="#wellformed">well formed</a> if there exists a '<tt>ret</tt>'
1144instruction inside of the function that returns a value that does not
1145match the return type of the function.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001146<h5>Semantics:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001147<p>When the '<tt>ret</tt>' instruction is executed, control flow
1148returns back to the calling function's context. If the caller is a "<a
John Criswellfa081872004-06-25 15:16:57 +00001149 href="#i_call"><tt>call</tt></a>" instruction, execution continues at
Chris Lattner261efe92003-11-25 01:02:51 +00001150the instruction after the call. If the caller was an "<a
1151 href="#i_invoke"><tt>invoke</tt></a>" instruction, execution continues
John Criswelle4c57cc2005-05-12 16:52:32 +00001152at the beginning of the "normal" destination block. If the instruction
Chris Lattner261efe92003-11-25 01:02:51 +00001153returns a value, that value shall set the call or invoke instruction's
1154return value.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001155<h5>Example:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001156<pre> ret int 5 <i>; Return an integer value of 5</i>
Chris Lattner7faa8832002-04-14 06:13:44 +00001157 ret void <i>; Return from a void function</i>
Chris Lattner00950542001-06-06 20:29:01 +00001158</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001159</div>
Chris Lattner00950542001-06-06 20:29:01 +00001160<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +00001161<div class="doc_subsubsection"> <a name="i_br">'<tt>br</tt>' Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001162<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +00001163<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001164<pre> br bool &lt;cond&gt;, label &lt;iftrue&gt;, label &lt;iffalse&gt;<br> br label &lt;dest&gt; <i>; Unconditional branch</i>
Chris Lattner00950542001-06-06 20:29:01 +00001165</pre>
Chris Lattner00950542001-06-06 20:29:01 +00001166<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001167<p>The '<tt>br</tt>' instruction is used to cause control flow to
1168transfer to a different basic block in the current function. There are
1169two forms of this instruction, corresponding to a conditional branch
1170and an unconditional branch.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001171<h5>Arguments:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001172<p>The conditional branch form of the '<tt>br</tt>' instruction takes a
1173single '<tt>bool</tt>' value and two '<tt>label</tt>' values. The
1174unconditional form of the '<tt>br</tt>' instruction takes a single '<tt>label</tt>'
1175value as a target.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001176<h5>Semantics:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001177<p>Upon execution of a conditional '<tt>br</tt>' instruction, the '<tt>bool</tt>'
1178argument is evaluated. If the value is <tt>true</tt>, control flows
1179to the '<tt>iftrue</tt>' <tt>label</tt> argument. If "cond" is <tt>false</tt>,
1180control flows to the '<tt>iffalse</tt>' <tt>label</tt> argument.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001181<h5>Example:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001182<pre>Test:<br> %cond = <a href="#i_setcc">seteq</a> int %a, %b<br> br bool %cond, label %IfEqual, label %IfUnequal<br>IfEqual:<br> <a
1183 href="#i_ret">ret</a> int 1<br>IfUnequal:<br> <a href="#i_ret">ret</a> int 0<br></pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001184</div>
Chris Lattner00950542001-06-06 20:29:01 +00001185<!-- _______________________________________________________________________ -->
Chris Lattnerc88c17b2004-02-24 04:54:45 +00001186<div class="doc_subsubsection">
1187 <a name="i_switch">'<tt>switch</tt>' Instruction</a>
1188</div>
1189
Misha Brukman9d0919f2003-11-08 01:05:38 +00001190<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +00001191<h5>Syntax:</h5>
Chris Lattnerc88c17b2004-02-24 04:54:45 +00001192
1193<pre>
1194 switch &lt;intty&gt; &lt;value&gt;, label &lt;defaultdest&gt; [ &lt;intty&gt; &lt;val&gt;, label &lt;dest&gt; ... ]
1195</pre>
1196
Chris Lattner00950542001-06-06 20:29:01 +00001197<h5>Overview:</h5>
Chris Lattnerc88c17b2004-02-24 04:54:45 +00001198
1199<p>The '<tt>switch</tt>' instruction is used to transfer control flow to one of
1200several different places. It is a generalization of the '<tt>br</tt>'
Misha Brukman9d0919f2003-11-08 01:05:38 +00001201instruction, allowing a branch to occur to one of many possible
1202destinations.</p>
Chris Lattnerc88c17b2004-02-24 04:54:45 +00001203
1204
Chris Lattner00950542001-06-06 20:29:01 +00001205<h5>Arguments:</h5>
Chris Lattnerc88c17b2004-02-24 04:54:45 +00001206
1207<p>The '<tt>switch</tt>' instruction uses three parameters: an integer
1208comparison value '<tt>value</tt>', a default '<tt>label</tt>' destination, and
1209an array of pairs of comparison value constants and '<tt>label</tt>'s. The
1210table is not allowed to contain duplicate constant entries.</p>
1211
Chris Lattner00950542001-06-06 20:29:01 +00001212<h5>Semantics:</h5>
Chris Lattnerc88c17b2004-02-24 04:54:45 +00001213
Chris Lattner261efe92003-11-25 01:02:51 +00001214<p>The <tt>switch</tt> instruction specifies a table of values and
1215destinations. When the '<tt>switch</tt>' instruction is executed, this
John Criswell84114752004-06-25 16:05:06 +00001216table is searched for the given value. If the value is found, control flow is
1217transfered to the corresponding destination; otherwise, control flow is
1218transfered to the default destination.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001219
Chris Lattnerc88c17b2004-02-24 04:54:45 +00001220<h5>Implementation:</h5>
1221
1222<p>Depending on properties of the target machine and the particular
1223<tt>switch</tt> instruction, this instruction may be code generated in different
John Criswell84114752004-06-25 16:05:06 +00001224ways. For example, it could be generated as a series of chained conditional
1225branches or with a lookup table.</p>
Chris Lattnerc88c17b2004-02-24 04:54:45 +00001226
1227<h5>Example:</h5>
1228
1229<pre>
1230 <i>; Emulate a conditional br instruction</i>
1231 %Val = <a href="#i_cast">cast</a> bool %value to int
1232 switch int %Val, label %truedest [int 0, label %falsedest ]
1233
1234 <i>; Emulate an unconditional br instruction</i>
1235 switch uint 0, label %dest [ ]
1236
1237 <i>; Implement a jump table:</i>
1238 switch uint %val, label %otherwise [ uint 0, label %onzero
1239 uint 1, label %onone
1240 uint 2, label %ontwo ]
Chris Lattner00950542001-06-06 20:29:01 +00001241</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001242</div>
Chris Lattnerbad10ee2005-05-06 22:57:40 +00001243
Chris Lattner00950542001-06-06 20:29:01 +00001244<!-- _______________________________________________________________________ -->
Chris Lattnerbad10ee2005-05-06 22:57:40 +00001245<div class="doc_subsubsection">
1246 <a name="i_invoke">'<tt>invoke</tt>' Instruction</a>
1247</div>
1248
Misha Brukman9d0919f2003-11-08 01:05:38 +00001249<div class="doc_text">
Chris Lattnerbad10ee2005-05-06 22:57:40 +00001250
Chris Lattner00950542001-06-06 20:29:01 +00001251<h5>Syntax:</h5>
Chris Lattnerbad10ee2005-05-06 22:57:40 +00001252
1253<pre>
1254 &lt;result&gt; = invoke [<a href="#callingconv">cconv</a>] &lt;ptr to function ty&gt; %&lt;function ptr val&gt;(&lt;function args&gt;)
1255 to label &lt;normal label&gt; except label &lt;exception label&gt;
1256</pre>
1257
Chris Lattner6536cfe2002-05-06 22:08:29 +00001258<h5>Overview:</h5>
Chris Lattnerbad10ee2005-05-06 22:57:40 +00001259
1260<p>The '<tt>invoke</tt>' instruction causes control to transfer to a specified
1261function, with the possibility of control flow transfer to either the
John Criswelle4c57cc2005-05-12 16:52:32 +00001262'<tt>normal</tt>' label or the
1263'<tt>exception</tt>' label. If the callee function returns with the
Chris Lattnerbad10ee2005-05-06 22:57:40 +00001264"<tt><a href="#i_ret">ret</a></tt>" instruction, control flow will return to the
1265"normal" label. If the callee (or any indirect callees) returns with the "<a
John Criswelle4c57cc2005-05-12 16:52:32 +00001266href="#i_unwind"><tt>unwind</tt></a>" instruction, control is interrupted and
1267continued at the dynamically nearest "exception" label.</p>
Chris Lattnerbad10ee2005-05-06 22:57:40 +00001268
Chris Lattner00950542001-06-06 20:29:01 +00001269<h5>Arguments:</h5>
Chris Lattnerbad10ee2005-05-06 22:57:40 +00001270
Misha Brukman9d0919f2003-11-08 01:05:38 +00001271<p>This instruction requires several arguments:</p>
Chris Lattnerbad10ee2005-05-06 22:57:40 +00001272
Chris Lattner00950542001-06-06 20:29:01 +00001273<ol>
Chris Lattnerbad10ee2005-05-06 22:57:40 +00001274 <li>
John Criswellc1f786c2005-05-13 22:25:59 +00001275 The optional "cconv" marker indicates which <a href="callingconv">calling
Chris Lattnerbad10ee2005-05-06 22:57:40 +00001276 convention</a> the call should use. If none is specified, the call defaults
1277 to using C calling conventions.
1278 </li>
1279 <li>'<tt>ptr to function ty</tt>': shall be the signature of the pointer to
1280 function value being invoked. In most cases, this is a direct function
1281 invocation, but indirect <tt>invoke</tt>s are just as possible, branching off
1282 an arbitrary pointer to function value.
1283 </li>
1284
1285 <li>'<tt>function ptr val</tt>': An LLVM value containing a pointer to a
1286 function to be invoked. </li>
1287
1288 <li>'<tt>function args</tt>': argument list whose types match the function
1289 signature argument types. If the function signature indicates the function
1290 accepts a variable number of arguments, the extra arguments can be
1291 specified. </li>
1292
1293 <li>'<tt>normal label</tt>': the label reached when the called function
1294 executes a '<tt><a href="#i_ret">ret</a></tt>' instruction. </li>
1295
1296 <li>'<tt>exception label</tt>': the label reached when a callee returns with
1297 the <a href="#i_unwind"><tt>unwind</tt></a> instruction. </li>
1298
Chris Lattner00950542001-06-06 20:29:01 +00001299</ol>
Chris Lattnerbad10ee2005-05-06 22:57:40 +00001300
Chris Lattner00950542001-06-06 20:29:01 +00001301<h5>Semantics:</h5>
Chris Lattnerbad10ee2005-05-06 22:57:40 +00001302
Misha Brukman9d0919f2003-11-08 01:05:38 +00001303<p>This instruction is designed to operate as a standard '<tt><a
Chris Lattnerbad10ee2005-05-06 22:57:40 +00001304href="#i_call">call</a></tt>' instruction in most regards. The primary
1305difference is that it establishes an association with a label, which is used by
1306the runtime library to unwind the stack.</p>
1307
1308<p>This instruction is used in languages with destructors to ensure that proper
1309cleanup is performed in the case of either a <tt>longjmp</tt> or a thrown
1310exception. Additionally, this is important for implementation of
1311'<tt>catch</tt>' clauses in high-level languages that support them.</p>
1312
Chris Lattner00950542001-06-06 20:29:01 +00001313<h5>Example:</h5>
Chris Lattnerbad10ee2005-05-06 22:57:40 +00001314<pre>
1315 %retval = invoke int %Test(int 15) to label %Continue
1316 except label %TestCleanup <i>; {int}:retval set</i>
1317 %retval = invoke <a href="#callingconv">coldcc</a> int %Test(int 15) to label %Continue
1318 except label %TestCleanup <i>; {int}:retval set</i>
Chris Lattner00950542001-06-06 20:29:01 +00001319</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001320</div>
Chris Lattner35eca582004-10-16 18:04:13 +00001321
1322
Chris Lattner27f71f22003-09-03 00:41:47 +00001323<!-- _______________________________________________________________________ -->
Chris Lattner35eca582004-10-16 18:04:13 +00001324
Chris Lattner261efe92003-11-25 01:02:51 +00001325<div class="doc_subsubsection"> <a name="i_unwind">'<tt>unwind</tt>'
1326Instruction</a> </div>
Chris Lattner35eca582004-10-16 18:04:13 +00001327
Misha Brukman9d0919f2003-11-08 01:05:38 +00001328<div class="doc_text">
Chris Lattner35eca582004-10-16 18:04:13 +00001329
Chris Lattner27f71f22003-09-03 00:41:47 +00001330<h5>Syntax:</h5>
Chris Lattner35eca582004-10-16 18:04:13 +00001331<pre>
1332 unwind
1333</pre>
1334
Chris Lattner27f71f22003-09-03 00:41:47 +00001335<h5>Overview:</h5>
Chris Lattner35eca582004-10-16 18:04:13 +00001336
1337<p>The '<tt>unwind</tt>' instruction unwinds the stack, continuing control flow
1338at the first callee in the dynamic call stack which used an <a
1339href="#i_invoke"><tt>invoke</tt></a> instruction to perform the call. This is
1340primarily used to implement exception handling.</p>
1341
Chris Lattner27f71f22003-09-03 00:41:47 +00001342<h5>Semantics:</h5>
Chris Lattner35eca582004-10-16 18:04:13 +00001343
1344<p>The '<tt>unwind</tt>' intrinsic causes execution of the current function to
1345immediately halt. The dynamic call stack is then searched for the first <a
1346href="#i_invoke"><tt>invoke</tt></a> instruction on the call stack. Once found,
1347execution continues at the "exceptional" destination block specified by the
1348<tt>invoke</tt> instruction. If there is no <tt>invoke</tt> instruction in the
1349dynamic call chain, undefined behavior results.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001350</div>
Chris Lattner35eca582004-10-16 18:04:13 +00001351
1352<!-- _______________________________________________________________________ -->
1353
1354<div class="doc_subsubsection"> <a name="i_unreachable">'<tt>unreachable</tt>'
1355Instruction</a> </div>
1356
1357<div class="doc_text">
1358
1359<h5>Syntax:</h5>
1360<pre>
1361 unreachable
1362</pre>
1363
1364<h5>Overview:</h5>
1365
1366<p>The '<tt>unreachable</tt>' instruction has no defined semantics. This
1367instruction is used to inform the optimizer that a particular portion of the
1368code is not reachable. This can be used to indicate that the code after a
1369no-return function cannot be reached, and other facts.</p>
1370
1371<h5>Semantics:</h5>
1372
1373<p>The '<tt>unreachable</tt>' instruction has no defined semantics.</p>
1374</div>
1375
1376
1377
Chris Lattner00950542001-06-06 20:29:01 +00001378<!-- ======================================================================= -->
Chris Lattner261efe92003-11-25 01:02:51 +00001379<div class="doc_subsection"> <a name="binaryops">Binary Operations</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001380<div class="doc_text">
Chris Lattner261efe92003-11-25 01:02:51 +00001381<p>Binary operators are used to do most of the computation in a
1382program. They require two operands, execute an operation on them, and
John Criswell9e2485c2004-12-10 15:51:16 +00001383produce a single value. The operands might represent
Chris Lattnera58561b2004-08-12 19:12:28 +00001384multiple data, as is the case with the <a href="#t_packed">packed</a> data type.
1385The result value of a binary operator is not
Chris Lattner261efe92003-11-25 01:02:51 +00001386necessarily the same type as its operands.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001387<p>There are several different binary operators:</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001388</div>
Chris Lattner00950542001-06-06 20:29:01 +00001389<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +00001390<div class="doc_subsubsection"> <a name="i_add">'<tt>add</tt>'
1391Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001392<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +00001393<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001394<pre> &lt;result&gt; = add &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
Chris Lattner00950542001-06-06 20:29:01 +00001395</pre>
Chris Lattner00950542001-06-06 20:29:01 +00001396<h5>Overview:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001397<p>The '<tt>add</tt>' instruction returns the sum of its two operands.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001398<h5>Arguments:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001399<p>The two arguments to the '<tt>add</tt>' instruction must be either <a
Chris Lattnera58561b2004-08-12 19:12:28 +00001400 href="#t_integer">integer</a> or <a href="#t_floating">floating point</a> values.
1401 This instruction can also take <a href="#t_packed">packed</a> versions of the values.
1402Both arguments must have identical types.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001403<h5>Semantics:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001404<p>The value produced is the integer or floating point sum of the two
1405operands.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001406<h5>Example:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001407<pre> &lt;result&gt; = add int 4, %var <i>; yields {int}:result = 4 + %var</i>
Chris Lattner00950542001-06-06 20:29:01 +00001408</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001409</div>
Chris Lattner00950542001-06-06 20:29:01 +00001410<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +00001411<div class="doc_subsubsection"> <a name="i_sub">'<tt>sub</tt>'
1412Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001413<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +00001414<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001415<pre> &lt;result&gt; = sub &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
Chris Lattner00950542001-06-06 20:29:01 +00001416</pre>
Chris Lattner00950542001-06-06 20:29:01 +00001417<h5>Overview:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001418<p>The '<tt>sub</tt>' instruction returns the difference of its two
1419operands.</p>
Chris Lattner261efe92003-11-25 01:02:51 +00001420<p>Note that the '<tt>sub</tt>' instruction is used to represent the '<tt>neg</tt>'
1421instruction present in most other intermediate representations.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001422<h5>Arguments:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001423<p>The two arguments to the '<tt>sub</tt>' instruction must be either <a
Chris Lattner261efe92003-11-25 01:02:51 +00001424 href="#t_integer">integer</a> or <a href="#t_floating">floating point</a>
Chris Lattnera58561b2004-08-12 19:12:28 +00001425values.
1426This instruction can also take <a href="#t_packed">packed</a> versions of the values.
1427Both arguments must have identical types.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001428<h5>Semantics:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001429<p>The value produced is the integer or floating point difference of
1430the two operands.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001431<h5>Example:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001432<pre> &lt;result&gt; = sub int 4, %var <i>; yields {int}:result = 4 - %var</i>
Chris Lattner00950542001-06-06 20:29:01 +00001433 &lt;result&gt; = sub int 0, %val <i>; yields {int}:result = -%var</i>
1434</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001435</div>
Chris Lattner00950542001-06-06 20:29:01 +00001436<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +00001437<div class="doc_subsubsection"> <a name="i_mul">'<tt>mul</tt>'
1438Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001439<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +00001440<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001441<pre> &lt;result&gt; = mul &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
Chris Lattner00950542001-06-06 20:29:01 +00001442</pre>
Chris Lattner00950542001-06-06 20:29:01 +00001443<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001444<p>The '<tt>mul</tt>' instruction returns the product of its two
1445operands.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001446<h5>Arguments:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001447<p>The two arguments to the '<tt>mul</tt>' instruction must be either <a
Chris Lattner261efe92003-11-25 01:02:51 +00001448 href="#t_integer">integer</a> or <a href="#t_floating">floating point</a>
Chris Lattnera58561b2004-08-12 19:12:28 +00001449values.
1450This instruction can also take <a href="#t_packed">packed</a> versions of the values.
1451Both arguments must have identical types.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001452<h5>Semantics:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001453<p>The value produced is the integer or floating point product of the
Misha Brukman9d0919f2003-11-08 01:05:38 +00001454two operands.</p>
Chris Lattner261efe92003-11-25 01:02:51 +00001455<p>There is no signed vs unsigned multiplication. The appropriate
1456action is taken based on the type of the operand.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001457<h5>Example:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001458<pre> &lt;result&gt; = mul int 4, %var <i>; yields {int}:result = 4 * %var</i>
Chris Lattner00950542001-06-06 20:29:01 +00001459</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001460</div>
Chris Lattner00950542001-06-06 20:29:01 +00001461<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +00001462<div class="doc_subsubsection"> <a name="i_div">'<tt>div</tt>'
1463Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001464<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +00001465<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001466<pre> &lt;result&gt; = div &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
1467</pre>
1468<h5>Overview:</h5>
1469<p>The '<tt>div</tt>' instruction returns the quotient of its two
1470operands.</p>
1471<h5>Arguments:</h5>
1472<p>The two arguments to the '<tt>div</tt>' instruction must be either <a
1473 href="#t_integer">integer</a> or <a href="#t_floating">floating point</a>
Chris Lattnera58561b2004-08-12 19:12:28 +00001474values.
1475This instruction can also take <a href="#t_packed">packed</a> versions of the values.
1476Both arguments must have identical types.</p>
Chris Lattner261efe92003-11-25 01:02:51 +00001477<h5>Semantics:</h5>
1478<p>The value produced is the integer or floating point quotient of the
1479two operands.</p>
1480<h5>Example:</h5>
1481<pre> &lt;result&gt; = div int 4, %var <i>; yields {int}:result = 4 / %var</i>
1482</pre>
1483</div>
1484<!-- _______________________________________________________________________ -->
1485<div class="doc_subsubsection"> <a name="i_rem">'<tt>rem</tt>'
1486Instruction</a> </div>
1487<div class="doc_text">
1488<h5>Syntax:</h5>
1489<pre> &lt;result&gt; = rem &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
1490</pre>
1491<h5>Overview:</h5>
1492<p>The '<tt>rem</tt>' instruction returns the remainder from the
1493division of its two operands.</p>
1494<h5>Arguments:</h5>
1495<p>The two arguments to the '<tt>rem</tt>' instruction must be either <a
1496 href="#t_integer">integer</a> or <a href="#t_floating">floating point</a>
Chris Lattnera58561b2004-08-12 19:12:28 +00001497values.
1498This instruction can also take <a href="#t_packed">packed</a> versions of the values.
1499Both arguments must have identical types.</p>
Chris Lattner261efe92003-11-25 01:02:51 +00001500<h5>Semantics:</h5>
1501<p>This returns the <i>remainder</i> of a division (where the result
1502has the same sign as the divisor), not the <i>modulus</i> (where the
1503result has the same sign as the dividend) of a value. For more
John Criswell0ec250c2005-10-24 16:17:18 +00001504information about the difference, see <a
Chris Lattner261efe92003-11-25 01:02:51 +00001505 href="http://mathforum.org/dr.math/problems/anne.4.28.99.html">The
1506Math Forum</a>.</p>
1507<h5>Example:</h5>
1508<pre> &lt;result&gt; = rem int 4, %var <i>; yields {int}:result = 4 % %var</i>
1509</pre>
1510</div>
1511<!-- _______________________________________________________________________ -->
1512<div class="doc_subsubsection"> <a name="i_setcc">'<tt>set<i>cc</i></tt>'
1513Instructions</a> </div>
1514<div class="doc_text">
1515<h5>Syntax:</h5>
1516<pre> &lt;result&gt; = seteq &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {bool}:result</i>
Chris Lattner00950542001-06-06 20:29:01 +00001517 &lt;result&gt; = setne &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {bool}:result</i>
1518 &lt;result&gt; = setlt &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {bool}:result</i>
1519 &lt;result&gt; = setgt &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {bool}:result</i>
1520 &lt;result&gt; = setle &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {bool}:result</i>
1521 &lt;result&gt; = setge &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {bool}:result</i>
1522</pre>
Chris Lattner261efe92003-11-25 01:02:51 +00001523<h5>Overview:</h5>
1524<p>The '<tt>set<i>cc</i></tt>' family of instructions returns a boolean
1525value based on a comparison of their two operands.</p>
1526<h5>Arguments:</h5>
1527<p>The two arguments to the '<tt>set<i>cc</i></tt>' instructions must
1528be of <a href="#t_firstclass">first class</a> type (it is not possible
1529to compare '<tt>label</tt>'s, '<tt>array</tt>'s, '<tt>structure</tt>'
1530or '<tt>void</tt>' values, etc...). Both arguments must have identical
1531types.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001532<h5>Semantics:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001533<p>The '<tt>seteq</tt>' instruction yields a <tt>true</tt> '<tt>bool</tt>'
1534value if both operands are equal.<br>
1535The '<tt>setne</tt>' instruction yields a <tt>true</tt> '<tt>bool</tt>'
1536value if both operands are unequal.<br>
1537The '<tt>setlt</tt>' instruction yields a <tt>true</tt> '<tt>bool</tt>'
1538value if the first operand is less than the second operand.<br>
1539The '<tt>setgt</tt>' instruction yields a <tt>true</tt> '<tt>bool</tt>'
1540value if the first operand is greater than the second operand.<br>
1541The '<tt>setle</tt>' instruction yields a <tt>true</tt> '<tt>bool</tt>'
1542value if the first operand is less than or equal to the second operand.<br>
1543The '<tt>setge</tt>' instruction yields a <tt>true</tt> '<tt>bool</tt>'
1544value if the first operand is greater than or equal to the second
1545operand.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001546<h5>Example:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001547<pre> &lt;result&gt; = seteq int 4, 5 <i>; yields {bool}:result = false</i>
Chris Lattner00950542001-06-06 20:29:01 +00001548 &lt;result&gt; = setne float 4, 5 <i>; yields {bool}:result = true</i>
1549 &lt;result&gt; = setlt uint 4, 5 <i>; yields {bool}:result = true</i>
1550 &lt;result&gt; = setgt sbyte 4, 5 <i>; yields {bool}:result = false</i>
1551 &lt;result&gt; = setle sbyte 4, 5 <i>; yields {bool}:result = true</i>
1552 &lt;result&gt; = setge sbyte 4, 5 <i>; yields {bool}:result = false</i>
1553</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001554</div>
Chris Lattner00950542001-06-06 20:29:01 +00001555<!-- ======================================================================= -->
Chris Lattner261efe92003-11-25 01:02:51 +00001556<div class="doc_subsection"> <a name="bitwiseops">Bitwise Binary
1557Operations</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001558<div class="doc_text">
Chris Lattner261efe92003-11-25 01:02:51 +00001559<p>Bitwise binary operators are used to do various forms of
1560bit-twiddling in a program. They are generally very efficient
John Criswell9e2485c2004-12-10 15:51:16 +00001561instructions and can commonly be strength reduced from other
Chris Lattner261efe92003-11-25 01:02:51 +00001562instructions. They require two operands, execute an operation on them,
1563and produce a single value. The resulting value of the bitwise binary
1564operators is always the same type as its first operand.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001565</div>
Chris Lattner00950542001-06-06 20:29:01 +00001566<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +00001567<div class="doc_subsubsection"> <a name="i_and">'<tt>and</tt>'
1568Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001569<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +00001570<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001571<pre> &lt;result&gt; = and &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
Chris Lattner00950542001-06-06 20:29:01 +00001572</pre>
Chris Lattner00950542001-06-06 20:29:01 +00001573<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001574<p>The '<tt>and</tt>' instruction returns the bitwise logical and of
1575its two operands.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001576<h5>Arguments:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001577<p>The two arguments to the '<tt>and</tt>' instruction must be <a
Chris Lattner261efe92003-11-25 01:02:51 +00001578 href="#t_integral">integral</a> values. Both arguments must have
1579identical types.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001580<h5>Semantics:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001581<p>The truth table used for the '<tt>and</tt>' instruction is:</p>
Chris Lattner261efe92003-11-25 01:02:51 +00001582<p> </p>
Misha Brukmandaa4cb02004-03-01 17:47:27 +00001583<div style="align: center">
Misha Brukman9d0919f2003-11-08 01:05:38 +00001584<table border="1" cellspacing="0" cellpadding="4">
Chris Lattner261efe92003-11-25 01:02:51 +00001585 <tbody>
1586 <tr>
1587 <td>In0</td>
1588 <td>In1</td>
1589 <td>Out</td>
1590 </tr>
1591 <tr>
1592 <td>0</td>
1593 <td>0</td>
1594 <td>0</td>
1595 </tr>
1596 <tr>
1597 <td>0</td>
1598 <td>1</td>
1599 <td>0</td>
1600 </tr>
1601 <tr>
1602 <td>1</td>
1603 <td>0</td>
1604 <td>0</td>
1605 </tr>
1606 <tr>
1607 <td>1</td>
1608 <td>1</td>
1609 <td>1</td>
1610 </tr>
1611 </tbody>
1612</table>
Misha Brukmandaa4cb02004-03-01 17:47:27 +00001613</div>
Chris Lattner00950542001-06-06 20:29:01 +00001614<h5>Example:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001615<pre> &lt;result&gt; = and int 4, %var <i>; yields {int}:result = 4 &amp; %var</i>
Chris Lattner00950542001-06-06 20:29:01 +00001616 &lt;result&gt; = and int 15, 40 <i>; yields {int}:result = 8</i>
1617 &lt;result&gt; = and int 4, 8 <i>; yields {int}:result = 0</i>
1618</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001619</div>
Chris Lattner00950542001-06-06 20:29:01 +00001620<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +00001621<div class="doc_subsubsection"> <a name="i_or">'<tt>or</tt>' Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001622<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +00001623<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001624<pre> &lt;result&gt; = or &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
Chris Lattner00950542001-06-06 20:29:01 +00001625</pre>
Chris Lattner261efe92003-11-25 01:02:51 +00001626<h5>Overview:</h5>
1627<p>The '<tt>or</tt>' instruction returns the bitwise logical inclusive
1628or of its two operands.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001629<h5>Arguments:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001630<p>The two arguments to the '<tt>or</tt>' instruction must be <a
Chris Lattner261efe92003-11-25 01:02:51 +00001631 href="#t_integral">integral</a> values. Both arguments must have
1632identical types.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001633<h5>Semantics:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001634<p>The truth table used for the '<tt>or</tt>' instruction is:</p>
Chris Lattner261efe92003-11-25 01:02:51 +00001635<p> </p>
Misha Brukmandaa4cb02004-03-01 17:47:27 +00001636<div style="align: center">
Chris Lattner261efe92003-11-25 01:02:51 +00001637<table border="1" cellspacing="0" cellpadding="4">
1638 <tbody>
1639 <tr>
1640 <td>In0</td>
1641 <td>In1</td>
1642 <td>Out</td>
1643 </tr>
1644 <tr>
1645 <td>0</td>
1646 <td>0</td>
1647 <td>0</td>
1648 </tr>
1649 <tr>
1650 <td>0</td>
1651 <td>1</td>
1652 <td>1</td>
1653 </tr>
1654 <tr>
1655 <td>1</td>
1656 <td>0</td>
1657 <td>1</td>
1658 </tr>
1659 <tr>
1660 <td>1</td>
1661 <td>1</td>
1662 <td>1</td>
1663 </tr>
1664 </tbody>
1665</table>
Misha Brukmandaa4cb02004-03-01 17:47:27 +00001666</div>
Chris Lattner00950542001-06-06 20:29:01 +00001667<h5>Example:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001668<pre> &lt;result&gt; = or int 4, %var <i>; yields {int}:result = 4 | %var</i>
Chris Lattner00950542001-06-06 20:29:01 +00001669 &lt;result&gt; = or int 15, 40 <i>; yields {int}:result = 47</i>
1670 &lt;result&gt; = or int 4, 8 <i>; yields {int}:result = 12</i>
1671</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001672</div>
Chris Lattner00950542001-06-06 20:29:01 +00001673<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +00001674<div class="doc_subsubsection"> <a name="i_xor">'<tt>xor</tt>'
1675Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001676<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +00001677<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001678<pre> &lt;result&gt; = xor &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
Chris Lattner00950542001-06-06 20:29:01 +00001679</pre>
Chris Lattner00950542001-06-06 20:29:01 +00001680<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001681<p>The '<tt>xor</tt>' instruction returns the bitwise logical exclusive
1682or of its two operands. The <tt>xor</tt> is used to implement the
1683"one's complement" operation, which is the "~" operator in C.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001684<h5>Arguments:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001685<p>The two arguments to the '<tt>xor</tt>' instruction must be <a
Chris Lattner261efe92003-11-25 01:02:51 +00001686 href="#t_integral">integral</a> values. Both arguments must have
1687identical types.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001688<h5>Semantics:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001689<p>The truth table used for the '<tt>xor</tt>' instruction is:</p>
Chris Lattner261efe92003-11-25 01:02:51 +00001690<p> </p>
Misha Brukmandaa4cb02004-03-01 17:47:27 +00001691<div style="align: center">
Chris Lattner261efe92003-11-25 01:02:51 +00001692<table border="1" cellspacing="0" cellpadding="4">
1693 <tbody>
1694 <tr>
1695 <td>In0</td>
1696 <td>In1</td>
1697 <td>Out</td>
1698 </tr>
1699 <tr>
1700 <td>0</td>
1701 <td>0</td>
1702 <td>0</td>
1703 </tr>
1704 <tr>
1705 <td>0</td>
1706 <td>1</td>
1707 <td>1</td>
1708 </tr>
1709 <tr>
1710 <td>1</td>
1711 <td>0</td>
1712 <td>1</td>
1713 </tr>
1714 <tr>
1715 <td>1</td>
1716 <td>1</td>
1717 <td>0</td>
1718 </tr>
1719 </tbody>
1720</table>
Misha Brukmandaa4cb02004-03-01 17:47:27 +00001721</div>
Chris Lattner261efe92003-11-25 01:02:51 +00001722<p> </p>
Chris Lattner00950542001-06-06 20:29:01 +00001723<h5>Example:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001724<pre> &lt;result&gt; = xor int 4, %var <i>; yields {int}:result = 4 ^ %var</i>
Chris Lattner00950542001-06-06 20:29:01 +00001725 &lt;result&gt; = xor int 15, 40 <i>; yields {int}:result = 39</i>
1726 &lt;result&gt; = xor int 4, 8 <i>; yields {int}:result = 12</i>
Chris Lattner27f71f22003-09-03 00:41:47 +00001727 &lt;result&gt; = xor int %V, -1 <i>; yields {int}:result = ~%V</i>
Chris Lattner00950542001-06-06 20:29:01 +00001728</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001729</div>
Chris Lattner00950542001-06-06 20:29:01 +00001730<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +00001731<div class="doc_subsubsection"> <a name="i_shl">'<tt>shl</tt>'
1732Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001733<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +00001734<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001735<pre> &lt;result&gt; = shl &lt;ty&gt; &lt;var1&gt;, ubyte &lt;var2&gt; <i>; yields {ty}:result</i>
Chris Lattner00950542001-06-06 20:29:01 +00001736</pre>
Chris Lattner00950542001-06-06 20:29:01 +00001737<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001738<p>The '<tt>shl</tt>' instruction returns the first operand shifted to
1739the left a specified number of bits.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001740<h5>Arguments:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001741<p>The first argument to the '<tt>shl</tt>' instruction must be an <a
Chris Lattner261efe92003-11-25 01:02:51 +00001742 href="#t_integer">integer</a> type. The second argument must be an '<tt>ubyte</tt>'
1743type.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001744<h5>Semantics:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001745<p>The value produced is <tt>var1</tt> * 2<sup><tt>var2</tt></sup>.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001746<h5>Example:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001747<pre> &lt;result&gt; = shl int 4, ubyte %var <i>; yields {int}:result = 4 &lt;&lt; %var</i>
Chris Lattner00950542001-06-06 20:29:01 +00001748 &lt;result&gt; = shl int 4, ubyte 2 <i>; yields {int}:result = 16</i>
1749 &lt;result&gt; = shl int 1, ubyte 10 <i>; yields {int}:result = 1024</i>
1750</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001751</div>
Chris Lattner00950542001-06-06 20:29:01 +00001752<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +00001753<div class="doc_subsubsection"> <a name="i_shr">'<tt>shr</tt>'
1754Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001755<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +00001756<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001757<pre> &lt;result&gt; = shr &lt;ty&gt; &lt;var1&gt;, ubyte &lt;var2&gt; <i>; yields {ty}:result</i>
Chris Lattner00950542001-06-06 20:29:01 +00001758</pre>
Chris Lattner00950542001-06-06 20:29:01 +00001759<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001760<p>The '<tt>shr</tt>' instruction returns the first operand shifted to
1761the right a specified number of bits.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001762<h5>Arguments:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001763<p>The first argument to the '<tt>shr</tt>' instruction must be an <a
Chris Lattner261efe92003-11-25 01:02:51 +00001764 href="#t_integer">integer</a> type. The second argument must be an '<tt>ubyte</tt>'
1765type.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001766<h5>Semantics:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001767<p>If the first argument is a <a href="#t_signed">signed</a> type, the
1768most significant bit is duplicated in the newly free'd bit positions.
1769If the first argument is unsigned, zero bits shall fill the empty
1770positions.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001771<h5>Example:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001772<pre> &lt;result&gt; = shr int 4, ubyte %var <i>; yields {int}:result = 4 &gt;&gt; %var</i>
Chris Lattner8c6bb902003-06-18 21:30:51 +00001773 &lt;result&gt; = shr uint 4, ubyte 1 <i>; yields {uint}:result = 2</i>
Chris Lattner00950542001-06-06 20:29:01 +00001774 &lt;result&gt; = shr int 4, ubyte 2 <i>; yields {int}:result = 1</i>
Chris Lattner8c6bb902003-06-18 21:30:51 +00001775 &lt;result&gt; = shr sbyte 4, ubyte 3 <i>; yields {sbyte}:result = 0</i>
1776 &lt;result&gt; = shr sbyte -2, ubyte 1 <i>; yields {sbyte}:result = -1</i>
Chris Lattner00950542001-06-06 20:29:01 +00001777</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001778</div>
Chris Lattner2cbdc452005-11-06 08:02:57 +00001779
Chris Lattner00950542001-06-06 20:29:01 +00001780<!-- ======================================================================= -->
Chris Lattner2cbdc452005-11-06 08:02:57 +00001781<div class="doc_subsection">
1782 <a name="memoryops">Memory Access Operations</a>
1783</div>
1784
Misha Brukman9d0919f2003-11-08 01:05:38 +00001785<div class="doc_text">
Chris Lattner2cbdc452005-11-06 08:02:57 +00001786
Chris Lattner261efe92003-11-25 01:02:51 +00001787<p>A key design point of an SSA-based representation is how it
1788represents memory. In LLVM, no memory locations are in SSA form, which
1789makes things very simple. This section describes how to read, write,
John Criswell9e2485c2004-12-10 15:51:16 +00001790allocate, and free memory in LLVM.</p>
Chris Lattner2cbdc452005-11-06 08:02:57 +00001791
Misha Brukman9d0919f2003-11-08 01:05:38 +00001792</div>
Chris Lattner2cbdc452005-11-06 08:02:57 +00001793
Chris Lattner00950542001-06-06 20:29:01 +00001794<!-- _______________________________________________________________________ -->
Chris Lattner2cbdc452005-11-06 08:02:57 +00001795<div class="doc_subsubsection">
1796 <a name="i_malloc">'<tt>malloc</tt>' Instruction</a>
1797</div>
1798
Misha Brukman9d0919f2003-11-08 01:05:38 +00001799<div class="doc_text">
Chris Lattner2cbdc452005-11-06 08:02:57 +00001800
Chris Lattner00950542001-06-06 20:29:01 +00001801<h5>Syntax:</h5>
Chris Lattner2cbdc452005-11-06 08:02:57 +00001802
1803<pre>
1804 &lt;result&gt; = malloc &lt;type&gt;[, uint &lt;NumElements&gt;][, align &lt;alignment&gt;] <i>; yields {type*}:result</i>
Chris Lattner00950542001-06-06 20:29:01 +00001805</pre>
Chris Lattner2cbdc452005-11-06 08:02:57 +00001806
Chris Lattner00950542001-06-06 20:29:01 +00001807<h5>Overview:</h5>
Chris Lattner2cbdc452005-11-06 08:02:57 +00001808
Chris Lattner261efe92003-11-25 01:02:51 +00001809<p>The '<tt>malloc</tt>' instruction allocates memory from the system
1810heap and returns a pointer to it.</p>
Chris Lattner2cbdc452005-11-06 08:02:57 +00001811
Chris Lattner00950542001-06-06 20:29:01 +00001812<h5>Arguments:</h5>
Chris Lattner2cbdc452005-11-06 08:02:57 +00001813
1814<p>The '<tt>malloc</tt>' instruction allocates
1815<tt>sizeof(&lt;type&gt;)*NumElements</tt>
John Criswell6e4ca612004-02-24 16:13:56 +00001816bytes of memory from the operating system and returns a pointer of the
Chris Lattner2cbdc452005-11-06 08:02:57 +00001817appropriate type to the program. If "NumElements" is specified, it is the
1818number of elements allocated. If an alignment is specified, the value result
1819of the allocation is guaranteed to be aligned to at least that boundary. If
1820not specified, or if zero, the target can choose to align the allocation on any
1821convenient boundary.</p>
1822
Misha Brukman9d0919f2003-11-08 01:05:38 +00001823<p>'<tt>type</tt>' must be a sized type.</p>
Chris Lattner2cbdc452005-11-06 08:02:57 +00001824
Chris Lattner00950542001-06-06 20:29:01 +00001825<h5>Semantics:</h5>
Chris Lattner2cbdc452005-11-06 08:02:57 +00001826
Chris Lattner261efe92003-11-25 01:02:51 +00001827<p>Memory is allocated using the system "<tt>malloc</tt>" function, and
1828a pointer is returned.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001829
Chris Lattner2cbdc452005-11-06 08:02:57 +00001830<h5>Example:</h5>
1831
1832<pre>
1833 %array = malloc [4 x ubyte ] <i>; yields {[%4 x ubyte]*}:array</i>
1834
1835 %size = <a href="#i_add">add</a> uint 2, 2 <i>; yields {uint}:size = uint 4</i>
Chris Lattner7faa8832002-04-14 06:13:44 +00001836 %array1 = malloc ubyte, uint 4 <i>; yields {ubyte*}:array1</i>
1837 %array2 = malloc [12 x ubyte], uint %size <i>; yields {[12 x ubyte]*}:array2</i>
Chris Lattner2cbdc452005-11-06 08:02:57 +00001838 %array3 = malloc int, uint 4, align 1024 <i>; yields {int*}:array3</i>
1839 %array4 = malloc int, align 1024 <i>; yields {int*}:array4</i>
Chris Lattner00950542001-06-06 20:29:01 +00001840</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001841</div>
Chris Lattner2cbdc452005-11-06 08:02:57 +00001842
Chris Lattner00950542001-06-06 20:29:01 +00001843<!-- _______________________________________________________________________ -->
Chris Lattner2cbdc452005-11-06 08:02:57 +00001844<div class="doc_subsubsection">
1845 <a name="i_free">'<tt>free</tt>' Instruction</a>
1846</div>
1847
Misha Brukman9d0919f2003-11-08 01:05:38 +00001848<div class="doc_text">
Chris Lattner2cbdc452005-11-06 08:02:57 +00001849
Chris Lattner00950542001-06-06 20:29:01 +00001850<h5>Syntax:</h5>
Chris Lattner2cbdc452005-11-06 08:02:57 +00001851
1852<pre>
1853 free &lt;type&gt; &lt;value&gt; <i>; yields {void}</i>
Chris Lattner00950542001-06-06 20:29:01 +00001854</pre>
Chris Lattner2cbdc452005-11-06 08:02:57 +00001855
Chris Lattner00950542001-06-06 20:29:01 +00001856<h5>Overview:</h5>
Chris Lattner2cbdc452005-11-06 08:02:57 +00001857
Chris Lattner261efe92003-11-25 01:02:51 +00001858<p>The '<tt>free</tt>' instruction returns memory back to the unused
John Criswellc1f786c2005-05-13 22:25:59 +00001859memory heap to be reallocated in the future.</p>
Chris Lattner2cbdc452005-11-06 08:02:57 +00001860
Chris Lattner00950542001-06-06 20:29:01 +00001861<h5>Arguments:</h5>
Chris Lattner2cbdc452005-11-06 08:02:57 +00001862
Chris Lattner261efe92003-11-25 01:02:51 +00001863<p>'<tt>value</tt>' shall be a pointer value that points to a value
1864that was allocated with the '<tt><a href="#i_malloc">malloc</a></tt>'
1865instruction.</p>
Chris Lattner2cbdc452005-11-06 08:02:57 +00001866
Chris Lattner00950542001-06-06 20:29:01 +00001867<h5>Semantics:</h5>
Chris Lattner2cbdc452005-11-06 08:02:57 +00001868
John Criswell9e2485c2004-12-10 15:51:16 +00001869<p>Access to the memory pointed to by the pointer is no longer defined
Chris Lattner261efe92003-11-25 01:02:51 +00001870after this instruction executes.</p>
Chris Lattner2cbdc452005-11-06 08:02:57 +00001871
Chris Lattner00950542001-06-06 20:29:01 +00001872<h5>Example:</h5>
Chris Lattner2cbdc452005-11-06 08:02:57 +00001873
1874<pre>
1875 %array = <a href="#i_malloc">malloc</a> [4 x ubyte] <i>; yields {[4 x ubyte]*}:array</i>
Chris Lattner00950542001-06-06 20:29:01 +00001876 free [4 x ubyte]* %array
1877</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001878</div>
Chris Lattner2cbdc452005-11-06 08:02:57 +00001879
Chris Lattner00950542001-06-06 20:29:01 +00001880<!-- _______________________________________________________________________ -->
Chris Lattner2cbdc452005-11-06 08:02:57 +00001881<div class="doc_subsubsection">
1882 <a name="i_alloca">'<tt>alloca</tt>' Instruction</a>
1883</div>
1884
Misha Brukman9d0919f2003-11-08 01:05:38 +00001885<div class="doc_text">
Chris Lattner2cbdc452005-11-06 08:02:57 +00001886
Chris Lattner00950542001-06-06 20:29:01 +00001887<h5>Syntax:</h5>
Chris Lattner2cbdc452005-11-06 08:02:57 +00001888
1889<pre>
1890 &lt;result&gt; = alloca &lt;type&gt;[, uint &lt;NumElements&gt;][, align &lt;alignment&gt;] <i>; yields {type*}:result</i>
Chris Lattner00950542001-06-06 20:29:01 +00001891</pre>
Chris Lattner2cbdc452005-11-06 08:02:57 +00001892
Chris Lattner00950542001-06-06 20:29:01 +00001893<h5>Overview:</h5>
Chris Lattner2cbdc452005-11-06 08:02:57 +00001894
Chris Lattner261efe92003-11-25 01:02:51 +00001895<p>The '<tt>alloca</tt>' instruction allocates memory on the current
1896stack frame of the procedure that is live until the current function
1897returns to its caller.</p>
Chris Lattner2cbdc452005-11-06 08:02:57 +00001898
Chris Lattner00950542001-06-06 20:29:01 +00001899<h5>Arguments:</h5>
Chris Lattner2cbdc452005-11-06 08:02:57 +00001900
John Criswell9e2485c2004-12-10 15:51:16 +00001901<p>The '<tt>alloca</tt>' instruction allocates <tt>sizeof(&lt;type&gt;)*NumElements</tt>
Chris Lattner261efe92003-11-25 01:02:51 +00001902bytes of memory on the runtime stack, returning a pointer of the
Chris Lattner2cbdc452005-11-06 08:02:57 +00001903appropriate type to the program. If "NumElements" is specified, it is the
1904number of elements allocated. If an alignment is specified, the value result
1905of the allocation is guaranteed to be aligned to at least that boundary. If
1906not specified, or if zero, the target can choose to align the allocation on any
1907convenient boundary.</p>
1908
Misha Brukman9d0919f2003-11-08 01:05:38 +00001909<p>'<tt>type</tt>' may be any sized type.</p>
Chris Lattner2cbdc452005-11-06 08:02:57 +00001910
Chris Lattner00950542001-06-06 20:29:01 +00001911<h5>Semantics:</h5>
Chris Lattner2cbdc452005-11-06 08:02:57 +00001912
John Criswellc1f786c2005-05-13 22:25:59 +00001913<p>Memory is allocated; a pointer is returned. '<tt>alloca</tt>'d
Chris Lattner261efe92003-11-25 01:02:51 +00001914memory is automatically released when the function returns. The '<tt>alloca</tt>'
1915instruction is commonly used to represent automatic variables that must
1916have an address available. When the function returns (either with the <tt><a
John Criswelldae2e932005-05-12 16:55:34 +00001917 href="#i_ret">ret</a></tt> or <tt><a href="#i_unwind">unwind</a></tt>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001918instructions), the memory is reclaimed.</p>
Chris Lattner2cbdc452005-11-06 08:02:57 +00001919
Chris Lattner00950542001-06-06 20:29:01 +00001920<h5>Example:</h5>
Chris Lattner2cbdc452005-11-06 08:02:57 +00001921
1922<pre>
1923 %ptr = alloca int <i>; yields {int*}:ptr</i>
Chris Lattner7faa8832002-04-14 06:13:44 +00001924 %ptr = alloca int, uint 4 <i>; yields {int*}:ptr</i>
Chris Lattner2cbdc452005-11-06 08:02:57 +00001925 %ptr = alloca int, uint 4, align 1024 <i>; yields {int*}:ptr</i>
1926 %ptr = alloca int, align 1024 <i>; yields {int*}:ptr</i>
Chris Lattner00950542001-06-06 20:29:01 +00001927</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001928</div>
Chris Lattner2cbdc452005-11-06 08:02:57 +00001929
Chris Lattner00950542001-06-06 20:29:01 +00001930<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +00001931<div class="doc_subsubsection"> <a name="i_load">'<tt>load</tt>'
1932Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001933<div class="doc_text">
Chris Lattner2b7d3202002-05-06 03:03:22 +00001934<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001935<pre> &lt;result&gt; = load &lt;ty&gt;* &lt;pointer&gt;<br> &lt;result&gt; = volatile load &lt;ty&gt;* &lt;pointer&gt;<br></pre>
Chris Lattner2b7d3202002-05-06 03:03:22 +00001936<h5>Overview:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001937<p>The '<tt>load</tt>' instruction is used to read from memory.</p>
Chris Lattner2b7d3202002-05-06 03:03:22 +00001938<h5>Arguments:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001939<p>The argument to the '<tt>load</tt>' instruction specifies the memory
John Criswell0ec250c2005-10-24 16:17:18 +00001940address from which to load. The pointer must point to a <a
Chris Lattnere53e5082004-06-03 22:57:15 +00001941 href="#t_firstclass">first class</a> type. If the <tt>load</tt> is
John Criswell0ec250c2005-10-24 16:17:18 +00001942marked as <tt>volatile</tt>, then the optimizer is not allowed to modify
Chris Lattner261efe92003-11-25 01:02:51 +00001943the number or order of execution of this <tt>load</tt> with other
1944volatile <tt>load</tt> and <tt><a href="#i_store">store</a></tt>
1945instructions. </p>
Chris Lattner2b7d3202002-05-06 03:03:22 +00001946<h5>Semantics:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001947<p>The location of memory pointed to is loaded.</p>
Chris Lattner2b7d3202002-05-06 03:03:22 +00001948<h5>Examples:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001949<pre> %ptr = <a href="#i_alloca">alloca</a> int <i>; yields {int*}:ptr</i>
1950 <a
1951 href="#i_store">store</a> int 3, int* %ptr <i>; yields {void}</i>
Chris Lattner2b7d3202002-05-06 03:03:22 +00001952 %val = load int* %ptr <i>; yields {int}:val = int 3</i>
1953</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001954</div>
Chris Lattner2b7d3202002-05-06 03:03:22 +00001955<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +00001956<div class="doc_subsubsection"> <a name="i_store">'<tt>store</tt>'
1957Instruction</a> </div>
Chris Lattner2b7d3202002-05-06 03:03:22 +00001958<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001959<pre> store &lt;ty&gt; &lt;value&gt;, &lt;ty&gt;* &lt;pointer&gt; <i>; yields {void}</i>
Chris Lattnerf0651072003-09-08 18:27:49 +00001960 volatile store &lt;ty&gt; &lt;value&gt;, &lt;ty&gt;* &lt;pointer&gt; <i>; yields {void}</i>
Chris Lattner2b7d3202002-05-06 03:03:22 +00001961</pre>
Chris Lattner2b7d3202002-05-06 03:03:22 +00001962<h5>Overview:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001963<p>The '<tt>store</tt>' instruction is used to write to memory.</p>
Chris Lattner2b7d3202002-05-06 03:03:22 +00001964<h5>Arguments:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001965<p>There are two arguments to the '<tt>store</tt>' instruction: a value
John Criswell0ec250c2005-10-24 16:17:18 +00001966to store and an address in which to store it. The type of the '<tt>&lt;pointer&gt;</tt>'
Chris Lattner261efe92003-11-25 01:02:51 +00001967operand must be a pointer to the type of the '<tt>&lt;value&gt;</tt>'
John Criswellc1f786c2005-05-13 22:25:59 +00001968operand. If the <tt>store</tt> is marked as <tt>volatile</tt>, then the
Chris Lattner261efe92003-11-25 01:02:51 +00001969optimizer is not allowed to modify the number or order of execution of
1970this <tt>store</tt> with other volatile <tt>load</tt> and <tt><a
1971 href="#i_store">store</a></tt> instructions.</p>
1972<h5>Semantics:</h5>
1973<p>The contents of memory are updated to contain '<tt>&lt;value&gt;</tt>'
1974at the location specified by the '<tt>&lt;pointer&gt;</tt>' operand.</p>
Chris Lattner2b7d3202002-05-06 03:03:22 +00001975<h5>Example:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001976<pre> %ptr = <a href="#i_alloca">alloca</a> int <i>; yields {int*}:ptr</i>
1977 <a
1978 href="#i_store">store</a> int 3, int* %ptr <i>; yields {void}</i>
Chris Lattner2b7d3202002-05-06 03:03:22 +00001979 %val = load int* %ptr <i>; yields {int}:val = int 3</i>
1980</pre>
Chris Lattner2b7d3202002-05-06 03:03:22 +00001981<!-- _______________________________________________________________________ -->
Chris Lattnerf74d5c72004-04-05 01:30:49 +00001982<div class="doc_subsubsection">
1983 <a name="i_getelementptr">'<tt>getelementptr</tt>' Instruction</a>
1984</div>
1985
Misha Brukman9d0919f2003-11-08 01:05:38 +00001986<div class="doc_text">
Chris Lattner7faa8832002-04-14 06:13:44 +00001987<h5>Syntax:</h5>
Chris Lattnerf74d5c72004-04-05 01:30:49 +00001988<pre>
1989 &lt;result&gt; = getelementptr &lt;ty&gt;* &lt;ptrval&gt;{, &lt;ty&gt; &lt;idx&gt;}*
1990</pre>
1991
Chris Lattner7faa8832002-04-14 06:13:44 +00001992<h5>Overview:</h5>
Chris Lattnerf74d5c72004-04-05 01:30:49 +00001993
1994<p>
1995The '<tt>getelementptr</tt>' instruction is used to get the address of a
1996subelement of an aggregate data structure.</p>
1997
Chris Lattner7faa8832002-04-14 06:13:44 +00001998<h5>Arguments:</h5>
Chris Lattnerf74d5c72004-04-05 01:30:49 +00001999
2000<p>This instruction takes a list of integer constants that indicate what
2001elements of the aggregate object to index to. The actual types of the arguments
2002provided depend on the type of the first pointer argument. The
2003'<tt>getelementptr</tt>' instruction is used to index down through the type
John Criswellfc6b8952005-05-16 16:17:45 +00002004levels of a structure or to a specific index in an array. When indexing into a
2005structure, only <tt>uint</tt>
John Criswellc1f786c2005-05-13 22:25:59 +00002006integer constants are allowed. When indexing into an array or pointer,
Chris Lattnerf74d5c72004-04-05 01:30:49 +00002007<tt>int</tt> and <tt>long</tt> indexes are allowed of any sign.</p>
2008
Chris Lattner261efe92003-11-25 01:02:51 +00002009<p>For example, let's consider a C code fragment and how it gets
2010compiled to LLVM:</p>
Chris Lattnerf74d5c72004-04-05 01:30:49 +00002011
2012<pre>
2013 struct RT {
2014 char A;
2015 int B[10][20];
2016 char C;
2017 };
2018 struct ST {
2019 int X;
2020 double Y;
2021 struct RT Z;
2022 };
2023
2024 int *foo(struct ST *s) {
2025 return &amp;s[1].Z.B[5][13];
2026 }
2027</pre>
2028
Misha Brukman9d0919f2003-11-08 01:05:38 +00002029<p>The LLVM code generated by the GCC frontend is:</p>
Chris Lattnerf74d5c72004-04-05 01:30:49 +00002030
2031<pre>
2032 %RT = type { sbyte, [10 x [20 x int]], sbyte }
2033 %ST = type { int, double, %RT }
2034
Brian Gaeke7283e7c2004-07-02 21:08:14 +00002035 implementation
2036
2037 int* %foo(%ST* %s) {
2038 entry:
2039 %reg = getelementptr %ST* %s, int 1, uint 2, uint 1, int 5, int 13
Chris Lattnerf74d5c72004-04-05 01:30:49 +00002040 ret int* %reg
2041 }
2042</pre>
2043
Chris Lattner7faa8832002-04-14 06:13:44 +00002044<h5>Semantics:</h5>
Chris Lattnerf74d5c72004-04-05 01:30:49 +00002045
2046<p>The index types specified for the '<tt>getelementptr</tt>' instruction depend
John Criswellc1f786c2005-05-13 22:25:59 +00002047on the pointer type that is being indexed into. <a href="#t_pointer">Pointer</a>
Chris Lattnere53e5082004-06-03 22:57:15 +00002048and <a href="#t_array">array</a> types require <tt>uint</tt>, <tt>int</tt>,
2049<tt>ulong</tt>, or <tt>long</tt> values, and <a href="#t_struct">structure</a>
Chris Lattnerf74d5c72004-04-05 01:30:49 +00002050types require <tt>uint</tt> <b>constants</b>.</p>
2051
Misha Brukman9d0919f2003-11-08 01:05:38 +00002052<p>In the example above, the first index is indexing into the '<tt>%ST*</tt>'
Chris Lattnerf74d5c72004-04-05 01:30:49 +00002053type, which is a pointer, yielding a '<tt>%ST</tt>' = '<tt>{ int, double, %RT
2054}</tt>' type, a structure. The second index indexes into the third element of
2055the structure, yielding a '<tt>%RT</tt>' = '<tt>{ sbyte, [10 x [20 x int]],
2056sbyte }</tt>' type, another structure. The third index indexes into the second
2057element of the structure, yielding a '<tt>[10 x [20 x int]]</tt>' type, an
2058array. The two dimensions of the array are subscripted into, yielding an
John Criswellfc6b8952005-05-16 16:17:45 +00002059'<tt>int</tt>' type. The '<tt>getelementptr</tt>' instruction returns a pointer
Chris Lattnerf74d5c72004-04-05 01:30:49 +00002060to this element, thus computing a value of '<tt>int*</tt>' type.</p>
2061
Chris Lattner261efe92003-11-25 01:02:51 +00002062<p>Note that it is perfectly legal to index partially through a
2063structure, returning a pointer to an inner element. Because of this,
2064the LLVM code for the given testcase is equivalent to:</p>
Chris Lattnerf74d5c72004-04-05 01:30:49 +00002065
2066<pre>
Chris Lattnerd4f6b172005-03-07 22:13:59 +00002067 int* %foo(%ST* %s) {
Chris Lattnerf74d5c72004-04-05 01:30:49 +00002068 %t1 = getelementptr %ST* %s, int 1 <i>; yields %ST*:%t1</i>
2069 %t2 = getelementptr %ST* %t1, int 0, uint 2 <i>; yields %RT*:%t2</i>
2070 %t3 = getelementptr %RT* %t2, int 0, uint 1 <i>; yields [10 x [20 x int]]*:%t3</i>
2071 %t4 = getelementptr [10 x [20 x int]]* %t3, int 0, int 5 <i>; yields [20 x int]*:%t4</i>
2072 %t5 = getelementptr [20 x int]* %t4, int 0, int 13 <i>; yields int*:%t5</i>
2073 ret int* %t5
2074 }
Chris Lattner6536cfe2002-05-06 22:08:29 +00002075</pre>
Chris Lattnere67a9512005-06-24 17:22:57 +00002076
2077<p>Note that it is undefined to access an array out of bounds: array and
2078pointer indexes must always be within the defined bounds of the array type.
2079The one exception for this rules is zero length arrays. These arrays are
2080defined to be accessible as variable length arrays, which requires access
2081beyond the zero'th element.</p>
2082
Chris Lattner7faa8832002-04-14 06:13:44 +00002083<h5>Example:</h5>
Chris Lattnere67a9512005-06-24 17:22:57 +00002084
Chris Lattnerf74d5c72004-04-05 01:30:49 +00002085<pre>
2086 <i>; yields [12 x ubyte]*:aptr</i>
2087 %aptr = getelementptr {int, [12 x ubyte]}* %sptr, long 0, uint 1
2088</pre>
2089
2090</div>
Chris Lattner00950542001-06-06 20:29:01 +00002091<!-- ======================================================================= -->
Chris Lattner261efe92003-11-25 01:02:51 +00002092<div class="doc_subsection"> <a name="otherops">Other Operations</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +00002093<div class="doc_text">
John Criswell4457dc92004-04-09 16:48:45 +00002094<p>The instructions in this category are the "miscellaneous"
Chris Lattner261efe92003-11-25 01:02:51 +00002095instructions, which defy better classification.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +00002096</div>
Chris Lattner00950542001-06-06 20:29:01 +00002097<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +00002098<div class="doc_subsubsection"> <a name="i_phi">'<tt>phi</tt>'
2099Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +00002100<div class="doc_text">
Chris Lattner33ba0d92001-07-09 00:26:23 +00002101<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00002102<pre> &lt;result&gt; = phi &lt;ty&gt; [ &lt;val0&gt;, &lt;label0&gt;], ...<br></pre>
Chris Lattner33ba0d92001-07-09 00:26:23 +00002103<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00002104<p>The '<tt>phi</tt>' instruction is used to implement the &#966; node in
2105the SSA graph representing the function.</p>
Chris Lattner33ba0d92001-07-09 00:26:23 +00002106<h5>Arguments:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00002107<p>The type of the incoming values are specified with the first type
2108field. After this, the '<tt>phi</tt>' instruction takes a list of pairs
2109as arguments, with one pair for each predecessor basic block of the
2110current block. Only values of <a href="#t_firstclass">first class</a>
2111type may be used as the value arguments to the PHI node. Only labels
2112may be used as the label arguments.</p>
2113<p>There must be no non-phi instructions between the start of a basic
2114block and the PHI instructions: i.e. PHI instructions must be first in
2115a basic block.</p>
Chris Lattner33ba0d92001-07-09 00:26:23 +00002116<h5>Semantics:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00002117<p>At runtime, the '<tt>phi</tt>' instruction logically takes on the
2118value specified by the parameter, depending on which basic block we
2119came from in the last <a href="#terminators">terminator</a> instruction.</p>
Chris Lattner6536cfe2002-05-06 22:08:29 +00002120<h5>Example:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00002121<pre>Loop: ; Infinite loop that counts from 0 on up...<br> %indvar = phi uint [ 0, %LoopHeader ], [ %nextindvar, %Loop ]<br> %nextindvar = add uint %indvar, 1<br> br label %Loop<br></pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +00002122</div>
Chris Lattnercc37aae2004-03-12 05:50:16 +00002123
Chris Lattner6536cfe2002-05-06 22:08:29 +00002124<!-- _______________________________________________________________________ -->
Chris Lattnercc37aae2004-03-12 05:50:16 +00002125<div class="doc_subsubsection">
2126 <a name="i_cast">'<tt>cast .. to</tt>' Instruction</a>
2127</div>
2128
Misha Brukman9d0919f2003-11-08 01:05:38 +00002129<div class="doc_text">
Chris Lattnercc37aae2004-03-12 05:50:16 +00002130
Chris Lattner6536cfe2002-05-06 22:08:29 +00002131<h5>Syntax:</h5>
Chris Lattnercc37aae2004-03-12 05:50:16 +00002132
2133<pre>
2134 &lt;result&gt; = cast &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
Chris Lattner6536cfe2002-05-06 22:08:29 +00002135</pre>
Chris Lattnercc37aae2004-03-12 05:50:16 +00002136
Chris Lattner6536cfe2002-05-06 22:08:29 +00002137<h5>Overview:</h5>
Chris Lattnercc37aae2004-03-12 05:50:16 +00002138
2139<p>
2140The '<tt>cast</tt>' instruction is used as the primitive means to convert
2141integers to floating point, change data type sizes, and break type safety (by
2142casting pointers).
2143</p>
2144
2145
Chris Lattner6536cfe2002-05-06 22:08:29 +00002146<h5>Arguments:</h5>
Chris Lattnercc37aae2004-03-12 05:50:16 +00002147
2148<p>
2149The '<tt>cast</tt>' instruction takes a value to cast, which must be a first
2150class value, and a type to cast it to, which must also be a <a
2151href="#t_firstclass">first class</a> type.
2152</p>
2153
Chris Lattner6536cfe2002-05-06 22:08:29 +00002154<h5>Semantics:</h5>
Chris Lattnercc37aae2004-03-12 05:50:16 +00002155
2156<p>
2157This instruction follows the C rules for explicit casts when determining how the
2158data being cast must change to fit in its new container.
2159</p>
2160
2161<p>
2162When casting to bool, any value that would be considered true in the context of
2163a C '<tt>if</tt>' condition is converted to the boolean '<tt>true</tt>' values,
2164all else are '<tt>false</tt>'.
2165</p>
2166
2167<p>
2168When extending an integral value from a type of one signness to another (for
2169example '<tt>sbyte</tt>' to '<tt>ulong</tt>'), the value is sign-extended if the
2170<b>source</b> value is signed, and zero-extended if the source value is
2171unsigned. <tt>bool</tt> values are always zero extended into either zero or
2172one.
2173</p>
2174
Chris Lattner33ba0d92001-07-09 00:26:23 +00002175<h5>Example:</h5>
Chris Lattnercc37aae2004-03-12 05:50:16 +00002176
2177<pre>
2178 %X = cast int 257 to ubyte <i>; yields ubyte:1</i>
Chris Lattner7bae3952002-06-25 18:03:17 +00002179 %Y = cast int 123 to bool <i>; yields bool:true</i>
Chris Lattner33ba0d92001-07-09 00:26:23 +00002180</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +00002181</div>
Chris Lattnercc37aae2004-03-12 05:50:16 +00002182
2183<!-- _______________________________________________________________________ -->
2184<div class="doc_subsubsection">
2185 <a name="i_select">'<tt>select</tt>' Instruction</a>
2186</div>
2187
2188<div class="doc_text">
2189
2190<h5>Syntax:</h5>
2191
2192<pre>
2193 &lt;result&gt; = select bool &lt;cond&gt;, &lt;ty&gt; &lt;val1&gt;, &lt;ty&gt; &lt;val2&gt; <i>; yields ty</i>
2194</pre>
2195
2196<h5>Overview:</h5>
2197
2198<p>
2199The '<tt>select</tt>' instruction is used to choose one value based on a
2200condition, without branching.
2201</p>
2202
2203
2204<h5>Arguments:</h5>
2205
2206<p>
2207The '<tt>select</tt>' instruction requires a boolean value indicating the condition, and two values of the same <a href="#t_firstclass">first class</a> type.
2208</p>
2209
2210<h5>Semantics:</h5>
2211
2212<p>
2213If the boolean condition evaluates to true, the instruction returns the first
John Criswellfc6b8952005-05-16 16:17:45 +00002214value argument; otherwise, it returns the second value argument.
Chris Lattnercc37aae2004-03-12 05:50:16 +00002215</p>
2216
2217<h5>Example:</h5>
2218
2219<pre>
2220 %X = select bool true, ubyte 17, ubyte 42 <i>; yields ubyte:17</i>
2221</pre>
2222</div>
2223
2224
2225
2226
2227
Chris Lattner33ba0d92001-07-09 00:26:23 +00002228<!-- _______________________________________________________________________ -->
Chris Lattner2bff5242005-05-06 05:47:36 +00002229<div class="doc_subsubsection">
2230 <a name="i_call">'<tt>call</tt>' Instruction</a>
2231</div>
2232
Misha Brukman9d0919f2003-11-08 01:05:38 +00002233<div class="doc_text">
Chris Lattner2bff5242005-05-06 05:47:36 +00002234
Chris Lattner00950542001-06-06 20:29:01 +00002235<h5>Syntax:</h5>
Chris Lattner2bff5242005-05-06 05:47:36 +00002236<pre>
Chris Lattnerbad10ee2005-05-06 22:57:40 +00002237 &lt;result&gt; = [tail] call [<a href="#callingconv">cconv</a>] &lt;ty&gt;* &lt;fnptrval&gt;(&lt;param list&gt;)
Chris Lattner2bff5242005-05-06 05:47:36 +00002238</pre>
2239
Chris Lattner00950542001-06-06 20:29:01 +00002240<h5>Overview:</h5>
Chris Lattner2bff5242005-05-06 05:47:36 +00002241
Misha Brukman9d0919f2003-11-08 01:05:38 +00002242<p>The '<tt>call</tt>' instruction represents a simple function call.</p>
Chris Lattner2bff5242005-05-06 05:47:36 +00002243
Chris Lattner00950542001-06-06 20:29:01 +00002244<h5>Arguments:</h5>
Chris Lattner2bff5242005-05-06 05:47:36 +00002245
Misha Brukman9d0919f2003-11-08 01:05:38 +00002246<p>This instruction requires several arguments:</p>
Chris Lattner2bff5242005-05-06 05:47:36 +00002247
Chris Lattner6536cfe2002-05-06 22:08:29 +00002248<ol>
Chris Lattner261efe92003-11-25 01:02:51 +00002249 <li>
Chris Lattnerbad10ee2005-05-06 22:57:40 +00002250 <p>The optional "tail" marker indicates whether the callee function accesses
2251 any allocas or varargs in the caller. If the "tail" marker is present, the
Chris Lattner2bff5242005-05-06 05:47:36 +00002252 function call is eligible for tail call optimization. Note that calls may
2253 be marked "tail" even if they do not occur before a <a
2254 href="#i_ret"><tt>ret</tt></a> instruction.
Chris Lattner261efe92003-11-25 01:02:51 +00002255 </li>
2256 <li>
Chris Lattnerbad10ee2005-05-06 22:57:40 +00002257 <p>The optional "cconv" marker indicates which <a href="callingconv">calling
2258 convention</a> the call should use. If none is specified, the call defaults
2259 to using C calling conventions.
2260 </li>
2261 <li>
Chris Lattner2bff5242005-05-06 05:47:36 +00002262 <p>'<tt>ty</tt>': shall be the signature of the pointer to function value
2263 being invoked. The argument types must match the types implied by this
John Criswellfc6b8952005-05-16 16:17:45 +00002264 signature. This type can be omitted if the function is not varargs and
2265 if the function type does not return a pointer to a function.</p>
Chris Lattner2bff5242005-05-06 05:47:36 +00002266 </li>
2267 <li>
2268 <p>'<tt>fnptrval</tt>': An LLVM value containing a pointer to a function to
2269 be invoked. In most cases, this is a direct function invocation, but
2270 indirect <tt>call</tt>s are just as possible, calling an arbitrary pointer
John Criswellfc6b8952005-05-16 16:17:45 +00002271 to function value.</p>
Chris Lattner261efe92003-11-25 01:02:51 +00002272 </li>
2273 <li>
2274 <p>'<tt>function args</tt>': argument list whose types match the
Reid Spencera7e302a2005-05-01 22:22:57 +00002275 function signature argument types. All arguments must be of
2276 <a href="#t_firstclass">first class</a> type. If the function signature
2277 indicates the function accepts a variable number of arguments, the extra
2278 arguments can be specified.</p>
Chris Lattner261efe92003-11-25 01:02:51 +00002279 </li>
Chris Lattner6536cfe2002-05-06 22:08:29 +00002280</ol>
Chris Lattner2bff5242005-05-06 05:47:36 +00002281
Chris Lattner00950542001-06-06 20:29:01 +00002282<h5>Semantics:</h5>
Chris Lattner2bff5242005-05-06 05:47:36 +00002283
Chris Lattner261efe92003-11-25 01:02:51 +00002284<p>The '<tt>call</tt>' instruction is used to cause control flow to
2285transfer to a specified function, with its incoming arguments bound to
2286the specified values. Upon a '<tt><a href="#i_ret">ret</a></tt>'
2287instruction in the called function, control flow continues with the
2288instruction after the function call, and the return value of the
2289function is bound to the result argument. This is a simpler case of
2290the <a href="#i_invoke">invoke</a> instruction.</p>
Chris Lattner2bff5242005-05-06 05:47:36 +00002291
Chris Lattner00950542001-06-06 20:29:01 +00002292<h5>Example:</h5>
Chris Lattner2bff5242005-05-06 05:47:36 +00002293
2294<pre>
2295 %retval = call int %test(int %argc)
2296 call int(sbyte*, ...) *%printf(sbyte* %msg, int 12, sbyte 42);
2297 %X = tail call int %foo()
Chris Lattnerbad10ee2005-05-06 22:57:40 +00002298 %Y = tail call <a href="#callingconv">fastcc</a> int %foo()
Chris Lattner2bff5242005-05-06 05:47:36 +00002299</pre>
2300
Misha Brukman9d0919f2003-11-08 01:05:38 +00002301</div>
Chris Lattnere19d7a72004-09-27 21:51:25 +00002302
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00002303<!-- _______________________________________________________________________ -->
Chris Lattnere19d7a72004-09-27 21:51:25 +00002304<div class="doc_subsubsection">
Chris Lattnere19d7a72004-09-27 21:51:25 +00002305 <a name="i_vaarg">'<tt>vaarg</tt>' Instruction</a>
2306</div>
2307
Misha Brukman9d0919f2003-11-08 01:05:38 +00002308<div class="doc_text">
Chris Lattnere19d7a72004-09-27 21:51:25 +00002309
Chris Lattner8d1a81d2003-10-18 05:51:36 +00002310<h5>Syntax:</h5>
Chris Lattnere19d7a72004-09-27 21:51:25 +00002311
2312<pre>
Andrew Lenharth8bf607a2005-06-18 18:28:17 +00002313 &lt;resultval&gt; = va_arg &lt;va_list*&gt; &lt;arglist&gt;, &lt;argty&gt;
Chris Lattnere19d7a72004-09-27 21:51:25 +00002314</pre>
2315
Chris Lattner8d1a81d2003-10-18 05:51:36 +00002316<h5>Overview:</h5>
Chris Lattnere19d7a72004-09-27 21:51:25 +00002317
Andrew Lenharth8bf607a2005-06-18 18:28:17 +00002318<p>The '<tt>va_arg</tt>' instruction is used to access arguments passed through
Chris Lattnere19d7a72004-09-27 21:51:25 +00002319the "variable argument" area of a function call. It is used to implement the
2320<tt>va_arg</tt> macro in C.</p>
2321
Chris Lattner8d1a81d2003-10-18 05:51:36 +00002322<h5>Arguments:</h5>
Chris Lattnere19d7a72004-09-27 21:51:25 +00002323
Andrew Lenharth8bf607a2005-06-18 18:28:17 +00002324<p>This instruction takes a <tt>va_list*</tt> value and the type of
2325the argument. It returns a value of the specified argument type and
2326increments the <tt>va_list</tt> to poin to the next argument. Again, the
2327actual type of <tt>va_list</tt> is target specific.</p>
Chris Lattnere19d7a72004-09-27 21:51:25 +00002328
Chris Lattner8d1a81d2003-10-18 05:51:36 +00002329<h5>Semantics:</h5>
Chris Lattnere19d7a72004-09-27 21:51:25 +00002330
Andrew Lenharth8bf607a2005-06-18 18:28:17 +00002331<p>The '<tt>va_arg</tt>' instruction loads an argument of the specified
2332type from the specified <tt>va_list</tt> and causes the
2333<tt>va_list</tt> to point to the next argument. For more information,
2334see the variable argument handling <a href="#int_varargs">Intrinsic
2335Functions</a>.</p>
Chris Lattnere19d7a72004-09-27 21:51:25 +00002336
2337<p>It is legal for this instruction to be called in a function which does not
2338take a variable number of arguments, for example, the <tt>vfprintf</tt>
Misha Brukman9d0919f2003-11-08 01:05:38 +00002339function.</p>
Chris Lattnere19d7a72004-09-27 21:51:25 +00002340
Andrew Lenharth8bf607a2005-06-18 18:28:17 +00002341<p><tt>va_arg</tt> is an LLVM instruction instead of an <a
John Criswellfc6b8952005-05-16 16:17:45 +00002342href="#intrinsics">intrinsic function</a> because it takes a type as an
Chris Lattnere19d7a72004-09-27 21:51:25 +00002343argument.</p>
2344
Chris Lattner8d1a81d2003-10-18 05:51:36 +00002345<h5>Example:</h5>
Chris Lattnere19d7a72004-09-27 21:51:25 +00002346
2347<p>See the <a href="#int_varargs">variable argument processing</a> section.</p>
2348
Misha Brukman9d0919f2003-11-08 01:05:38 +00002349</div>
Chris Lattner8ff75902004-01-06 05:31:32 +00002350
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00002351<!-- *********************************************************************** -->
Chris Lattner261efe92003-11-25 01:02:51 +00002352<div class="doc_section"> <a name="intrinsics">Intrinsic Functions</a> </div>
2353<!-- *********************************************************************** -->
Chris Lattner8ff75902004-01-06 05:31:32 +00002354
Misha Brukman9d0919f2003-11-08 01:05:38 +00002355<div class="doc_text">
Chris Lattner33aec9e2004-02-12 17:01:32 +00002356
2357<p>LLVM supports the notion of an "intrinsic function". These functions have
John Criswellfc6b8952005-05-16 16:17:45 +00002358well known names and semantics and are required to follow certain
Chris Lattner33aec9e2004-02-12 17:01:32 +00002359restrictions. Overall, these instructions represent an extension mechanism for
2360the LLVM language that does not require changing all of the transformations in
2361LLVM to add to the language (or the bytecode reader/writer, the parser,
2362etc...).</p>
2363
John Criswellfc6b8952005-05-16 16:17:45 +00002364<p>Intrinsic function names must all start with an "<tt>llvm.</tt>" prefix. This
2365prefix is reserved in LLVM for intrinsic names; thus, functions may not be named
Chris Lattner33aec9e2004-02-12 17:01:32 +00002366this. Intrinsic functions must always be external functions: you cannot define
2367the body of intrinsic functions. Intrinsic functions may only be used in call
2368or invoke instructions: it is illegal to take the address of an intrinsic
2369function. Additionally, because intrinsic functions are part of the LLVM
2370language, it is required that they all be documented here if any are added.</p>
2371
2372
John Criswellfc6b8952005-05-16 16:17:45 +00002373<p>To learn how to add an intrinsic function, please see the <a
Chris Lattner590cff32005-05-11 03:35:57 +00002374href="ExtendingLLVM.html">Extending LLVM Guide</a>.
Chris Lattner33aec9e2004-02-12 17:01:32 +00002375</p>
2376
Misha Brukman9d0919f2003-11-08 01:05:38 +00002377</div>
Chris Lattner8ff75902004-01-06 05:31:32 +00002378
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00002379<!-- ======================================================================= -->
Chris Lattner8ff75902004-01-06 05:31:32 +00002380<div class="doc_subsection">
2381 <a name="int_varargs">Variable Argument Handling Intrinsics</a>
2382</div>
2383
Misha Brukman9d0919f2003-11-08 01:05:38 +00002384<div class="doc_text">
Chris Lattnerd7923912004-05-23 21:06:01 +00002385
Misha Brukman9d0919f2003-11-08 01:05:38 +00002386<p>Variable argument support is defined in LLVM with the <a
Chris Lattner261efe92003-11-25 01:02:51 +00002387 href="#i_vanext"><tt>vanext</tt></a> instruction and these three
2388intrinsic functions. These functions are related to the similarly
2389named macros defined in the <tt>&lt;stdarg.h&gt;</tt> header file.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +00002390
Chris Lattner261efe92003-11-25 01:02:51 +00002391<p>All of these functions operate on arguments that use a
2392target-specific value type "<tt>va_list</tt>". The LLVM assembly
2393language reference manual does not define what this type is, so all
2394transformations should be prepared to handle intrinsics with any type
2395used.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +00002396
Misha Brukman9d0919f2003-11-08 01:05:38 +00002397<p>This example shows how the <a href="#i_vanext"><tt>vanext</tt></a>
Chris Lattner261efe92003-11-25 01:02:51 +00002398instruction and the variable argument handling intrinsic functions are
2399used.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +00002400
Chris Lattner33aec9e2004-02-12 17:01:32 +00002401<pre>
2402int %test(int %X, ...) {
2403 ; Initialize variable argument processing
Andrew Lenharth8bf607a2005-06-18 18:28:17 +00002404 %ap = alloca sbyte*
2405 call void %<a href="#i_va_start">llvm.va_start</a>(sbyte** %ap)
Chris Lattner33aec9e2004-02-12 17:01:32 +00002406
2407 ; Read a single integer argument
Andrew Lenharth8bf607a2005-06-18 18:28:17 +00002408 %tmp = va_arg sbyte** %ap, int
Chris Lattner33aec9e2004-02-12 17:01:32 +00002409
2410 ; Demonstrate usage of llvm.va_copy and llvm.va_end
Andrew Lenharth8bf607a2005-06-18 18:28:17 +00002411 %aq = alloca sbyte*
Andrew Lenharthd0a4c622005-06-22 20:38:11 +00002412 call void %<a href="#i_va_copy">llvm.va_copy</a>(sbyte** %aq, sbyte** %ap)
Andrew Lenharth8bf607a2005-06-18 18:28:17 +00002413 call void %<a href="#i_va_end">llvm.va_end</a>(sbyte** %aq)
Chris Lattner33aec9e2004-02-12 17:01:32 +00002414
2415 ; Stop processing of arguments.
Andrew Lenharth8bf607a2005-06-18 18:28:17 +00002416 call void %<a href="#i_va_end">llvm.va_end</a>(sbyte** %ap)
Chris Lattner33aec9e2004-02-12 17:01:32 +00002417 ret int %tmp
2418}
2419</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +00002420</div>
Chris Lattner8ff75902004-01-06 05:31:32 +00002421
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00002422<!-- _______________________________________________________________________ -->
Chris Lattner8ff75902004-01-06 05:31:32 +00002423<div class="doc_subsubsection">
2424 <a name="i_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a>
2425</div>
2426
2427
Misha Brukman9d0919f2003-11-08 01:05:38 +00002428<div class="doc_text">
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00002429<h5>Syntax:</h5>
Andrew Lenharth8bf607a2005-06-18 18:28:17 +00002430<pre> declare void %llvm.va_start(&lt;va_list&gt;* &lt;arglist&gt;)<br></pre>
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00002431<h5>Overview:</h5>
Andrew Lenharth8bf607a2005-06-18 18:28:17 +00002432<P>The '<tt>llvm.va_start</tt>' intrinsic initializes
2433<tt>*&lt;arglist&gt;</tt> for subsequent use by <tt><a
2434href="#i_va_arg">va_arg</a></tt>.</p>
2435
2436<h5>Arguments:</h5>
2437
2438<P>The argument is a pointer to a <tt>va_list</tt> element to initialize.</p>
2439
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00002440<h5>Semantics:</h5>
Andrew Lenharth8bf607a2005-06-18 18:28:17 +00002441
2442<P>The '<tt>llvm.va_start</tt>' intrinsic works just like the <tt>va_start</tt>
2443macro available in C. In a target-dependent way, it initializes the
2444<tt>va_list</tt> element the argument points to, so that the next call to
2445<tt>va_arg</tt> will produce the first variable argument passed to the function.
2446Unlike the C <tt>va_start</tt> macro, this intrinsic does not need to know the
2447last argument of the function, the compiler can figure that out.</p>
2448
Misha Brukman9d0919f2003-11-08 01:05:38 +00002449</div>
Chris Lattner8ff75902004-01-06 05:31:32 +00002450
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00002451<!-- _______________________________________________________________________ -->
Chris Lattner8ff75902004-01-06 05:31:32 +00002452<div class="doc_subsubsection">
2453 <a name="i_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a>
2454</div>
2455
Misha Brukman9d0919f2003-11-08 01:05:38 +00002456<div class="doc_text">
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00002457<h5>Syntax:</h5>
Andrew Lenharth8bf607a2005-06-18 18:28:17 +00002458<pre> declare void %llvm.va_end(&lt;va_list*&gt; &lt;arglist&gt;)<br></pre>
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00002459<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00002460<p>The '<tt>llvm.va_end</tt>' intrinsic destroys <tt>&lt;arglist&gt;</tt>
2461which has been initialized previously with <tt><a href="#i_va_start">llvm.va_start</a></tt>
2462or <tt><a href="#i_va_copy">llvm.va_copy</a></tt>.</p>
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00002463<h5>Arguments:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00002464<p>The argument is a <tt>va_list</tt> to destroy.</p>
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00002465<h5>Semantics:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00002466<p>The '<tt>llvm.va_end</tt>' intrinsic works just like the <tt>va_end</tt>
Chris Lattner261efe92003-11-25 01:02:51 +00002467macro available in C. In a target-dependent way, it destroys the <tt>va_list</tt>.
2468Calls to <a href="#i_va_start"><tt>llvm.va_start</tt></a> and <a
2469 href="#i_va_copy"><tt>llvm.va_copy</tt></a> must be matched exactly
2470with calls to <tt>llvm.va_end</tt>.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +00002471</div>
Chris Lattner8ff75902004-01-06 05:31:32 +00002472
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00002473<!-- _______________________________________________________________________ -->
Chris Lattner8ff75902004-01-06 05:31:32 +00002474<div class="doc_subsubsection">
2475 <a name="i_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a>
2476</div>
2477
Misha Brukman9d0919f2003-11-08 01:05:38 +00002478<div class="doc_text">
Chris Lattnerd7923912004-05-23 21:06:01 +00002479
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00002480<h5>Syntax:</h5>
Chris Lattnerd7923912004-05-23 21:06:01 +00002481
2482<pre>
Andrew Lenharth8bf607a2005-06-18 18:28:17 +00002483 declare void %llvm.va_copy(&lt;va_list&gt;* &lt;destarglist&gt;,
Andrew Lenharthd0a4c622005-06-22 20:38:11 +00002484 &lt;va_list&gt;* &lt;srcarglist&gt;)
Chris Lattnerd7923912004-05-23 21:06:01 +00002485</pre>
2486
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00002487<h5>Overview:</h5>
Chris Lattnerd7923912004-05-23 21:06:01 +00002488
Andrew Lenharth8bf607a2005-06-18 18:28:17 +00002489<p>The '<tt>llvm.va_copy</tt>' intrinsic copies the current argument position from
2490the source argument list to the destination argument list.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +00002491
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00002492<h5>Arguments:</h5>
Chris Lattnerd7923912004-05-23 21:06:01 +00002493
Andrew Lenharth8bf607a2005-06-18 18:28:17 +00002494<p>The first argument is a pointer to a <tt>va_list</tt> element to initialize.
Andrew Lenharthd0a4c622005-06-22 20:38:11 +00002495The second argument is a pointer to a <tt>va_list</tt> element to copy from.</p>
Andrew Lenharth8bf607a2005-06-18 18:28:17 +00002496
Chris Lattnerd7923912004-05-23 21:06:01 +00002497
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00002498<h5>Semantics:</h5>
Chris Lattnerd7923912004-05-23 21:06:01 +00002499
Andrew Lenharth8bf607a2005-06-18 18:28:17 +00002500<p>The '<tt>llvm.va_copy</tt>' intrinsic works just like the <tt>va_copy</tt> macro
2501available in C. In a target-dependent way, it copies the source
2502<tt>va_list</tt> element into the destination list. This intrinsic is necessary
2503because the <tt><a href="i_va_begin">llvm.va_begin</a></tt> intrinsic may be
Chris Lattnerd7923912004-05-23 21:06:01 +00002504arbitrarily complex and require memory allocation, for example.</p>
2505
Misha Brukman9d0919f2003-11-08 01:05:38 +00002506</div>
Chris Lattner8ff75902004-01-06 05:31:32 +00002507
Chris Lattner33aec9e2004-02-12 17:01:32 +00002508<!-- ======================================================================= -->
2509<div class="doc_subsection">
Chris Lattnerd7923912004-05-23 21:06:01 +00002510 <a name="int_gc">Accurate Garbage Collection Intrinsics</a>
2511</div>
2512
2513<div class="doc_text">
2514
2515<p>
2516LLVM support for <a href="GarbageCollection.html">Accurate Garbage
2517Collection</a> requires the implementation and generation of these intrinsics.
2518These intrinsics allow identification of <a href="#i_gcroot">GC roots on the
2519stack</a>, as well as garbage collector implementations that require <a
2520href="#i_gcread">read</a> and <a href="#i_gcwrite">write</a> barriers.
2521Front-ends for type-safe garbage collected languages should generate these
2522intrinsics to make use of the LLVM garbage collectors. For more details, see <a
2523href="GarbageCollection.html">Accurate Garbage Collection with LLVM</a>.
2524</p>
2525</div>
2526
2527<!-- _______________________________________________________________________ -->
2528<div class="doc_subsubsection">
2529 <a name="i_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a>
2530</div>
2531
2532<div class="doc_text">
2533
2534<h5>Syntax:</h5>
2535
2536<pre>
Reid Spencera8d451e2005-04-26 20:50:44 +00002537 declare void %llvm.gcroot(&lt;ty&gt;** %ptrloc, &lt;ty2&gt;* %metadata)
Chris Lattnerd7923912004-05-23 21:06:01 +00002538</pre>
2539
2540<h5>Overview:</h5>
2541
John Criswell9e2485c2004-12-10 15:51:16 +00002542<p>The '<tt>llvm.gcroot</tt>' intrinsic declares the existence of a GC root to
Chris Lattnerd7923912004-05-23 21:06:01 +00002543the code generator, and allows some metadata to be associated with it.</p>
2544
2545<h5>Arguments:</h5>
2546
2547<p>The first argument specifies the address of a stack object that contains the
2548root pointer. The second pointer (which must be either a constant or a global
2549value address) contains the meta-data to be associated with the root.</p>
2550
2551<h5>Semantics:</h5>
2552
2553<p>At runtime, a call to this intrinsics stores a null pointer into the "ptrloc"
2554location. At compile-time, the code generator generates information to allow
2555the runtime to find the pointer at GC safe points.
2556</p>
2557
2558</div>
2559
2560
2561<!-- _______________________________________________________________________ -->
2562<div class="doc_subsubsection">
2563 <a name="i_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a>
2564</div>
2565
2566<div class="doc_text">
2567
2568<h5>Syntax:</h5>
2569
2570<pre>
Reid Spencera8d451e2005-04-26 20:50:44 +00002571 declare sbyte* %llvm.gcread(sbyte** %Ptr)
Chris Lattnerd7923912004-05-23 21:06:01 +00002572</pre>
2573
2574<h5>Overview:</h5>
2575
2576<p>The '<tt>llvm.gcread</tt>' intrinsic identifies reads of references from heap
2577locations, allowing garbage collector implementations that require read
2578barriers.</p>
2579
2580<h5>Arguments:</h5>
2581
2582<p>The argument is the address to read from, which should be an address
2583allocated from the garbage collector.</p>
2584
2585<h5>Semantics:</h5>
2586
2587<p>The '<tt>llvm.gcread</tt>' intrinsic has the same semantics as a load
2588instruction, but may be replaced with substantially more complex code by the
2589garbage collector runtime, as needed.</p>
2590
2591</div>
2592
2593
2594<!-- _______________________________________________________________________ -->
2595<div class="doc_subsubsection">
2596 <a name="i_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a>
2597</div>
2598
2599<div class="doc_text">
2600
2601<h5>Syntax:</h5>
2602
2603<pre>
Reid Spencera8d451e2005-04-26 20:50:44 +00002604 declare void %llvm.gcwrite(sbyte* %P1, sbyte** %P2)
Chris Lattnerd7923912004-05-23 21:06:01 +00002605</pre>
2606
2607<h5>Overview:</h5>
2608
2609<p>The '<tt>llvm.gcwrite</tt>' intrinsic identifies writes of references to heap
2610locations, allowing garbage collector implementations that require write
2611barriers (such as generational or reference counting collectors).</p>
2612
2613<h5>Arguments:</h5>
2614
2615<p>The first argument is the reference to store, and the second is the heap
2616location to store to.</p>
2617
2618<h5>Semantics:</h5>
2619
2620<p>The '<tt>llvm.gcwrite</tt>' intrinsic has the same semantics as a store
2621instruction, but may be replaced with substantially more complex code by the
2622garbage collector runtime, as needed.</p>
2623
2624</div>
2625
2626
2627
2628<!-- ======================================================================= -->
2629<div class="doc_subsection">
Chris Lattner10610642004-02-14 04:08:35 +00002630 <a name="int_codegen">Code Generator Intrinsics</a>
2631</div>
2632
2633<div class="doc_text">
2634<p>
2635These intrinsics are provided by LLVM to expose special features that may only
2636be implemented with code generator support.
2637</p>
2638
2639</div>
2640
2641<!-- _______________________________________________________________________ -->
2642<div class="doc_subsubsection">
2643 <a name="i_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a>
2644</div>
2645
2646<div class="doc_text">
2647
2648<h5>Syntax:</h5>
2649<pre>
Reid Spencera8d451e2005-04-26 20:50:44 +00002650 declare void* %llvm.returnaddress(uint &lt;level&gt;)
Chris Lattner10610642004-02-14 04:08:35 +00002651</pre>
2652
2653<h5>Overview:</h5>
2654
2655<p>
2656The '<tt>llvm.returnaddress</tt>' intrinsic returns a target-specific value
2657indicating the return address of the current function or one of its callers.
2658</p>
2659
2660<h5>Arguments:</h5>
2661
2662<p>
2663The argument to this intrinsic indicates which function to return the address
2664for. Zero indicates the calling function, one indicates its caller, etc. The
2665argument is <b>required</b> to be a constant integer value.
2666</p>
2667
2668<h5>Semantics:</h5>
2669
2670<p>
2671The '<tt>llvm.returnaddress</tt>' intrinsic either returns a pointer indicating
2672the return address of the specified call frame, or zero if it cannot be
2673identified. The value returned by this intrinsic is likely to be incorrect or 0
2674for arguments other than zero, so it should only be used for debugging purposes.
2675</p>
2676
2677<p>
2678Note that calling this intrinsic does not prevent function inlining or other
Chris Lattnerb40bb382005-03-07 20:30:51 +00002679aggressive transformations, so the value returned may not be that of the obvious
Chris Lattner10610642004-02-14 04:08:35 +00002680source-language caller.
2681</p>
2682</div>
2683
2684
2685<!-- _______________________________________________________________________ -->
2686<div class="doc_subsubsection">
2687 <a name="i_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a>
2688</div>
2689
2690<div class="doc_text">
2691
2692<h5>Syntax:</h5>
2693<pre>
Reid Spencera8d451e2005-04-26 20:50:44 +00002694 declare void* %llvm.frameaddress(uint &lt;level&gt;)
Chris Lattner10610642004-02-14 04:08:35 +00002695</pre>
2696
2697<h5>Overview:</h5>
2698
2699<p>
2700The '<tt>llvm.frameaddress</tt>' intrinsic returns the target-specific frame
2701pointer value for the specified stack frame.
2702</p>
2703
2704<h5>Arguments:</h5>
2705
2706<p>
2707The argument to this intrinsic indicates which function to return the frame
2708pointer for. Zero indicates the calling function, one indicates its caller,
2709etc. The argument is <b>required</b> to be a constant integer value.
2710</p>
2711
2712<h5>Semantics:</h5>
2713
2714<p>
2715The '<tt>llvm.frameaddress</tt>' intrinsic either returns a pointer indicating
2716the frame address of the specified call frame, or zero if it cannot be
2717identified. The value returned by this intrinsic is likely to be incorrect or 0
2718for arguments other than zero, so it should only be used for debugging purposes.
2719</p>
2720
2721<p>
2722Note that calling this intrinsic does not prevent function inlining or other
Chris Lattnerb40bb382005-03-07 20:30:51 +00002723aggressive transformations, so the value returned may not be that of the obvious
Chris Lattner10610642004-02-14 04:08:35 +00002724source-language caller.
2725</p>
2726</div>
2727
Chris Lattner9a9d7ac2005-02-28 19:24:19 +00002728<!-- _______________________________________________________________________ -->
2729<div class="doc_subsubsection">
2730 <a name="i_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a>
2731</div>
2732
2733<div class="doc_text">
2734
2735<h5>Syntax:</h5>
2736<pre>
Reid Spencera8d451e2005-04-26 20:50:44 +00002737 declare void %llvm.prefetch(sbyte * &lt;address&gt;,
2738 uint &lt;rw&gt;, uint &lt;locality&gt;)
Chris Lattner9a9d7ac2005-02-28 19:24:19 +00002739</pre>
2740
2741<h5>Overview:</h5>
2742
2743
2744<p>
2745The '<tt>llvm.prefetch</tt>' intrinsic is a hint to the code generator to insert
John Criswellfc6b8952005-05-16 16:17:45 +00002746a prefetch instruction if supported; otherwise, it is a noop. Prefetches have
2747no
2748effect on the behavior of the program but can change its performance
Chris Lattner2a615362005-02-28 19:47:14 +00002749characteristics.
Chris Lattner9a9d7ac2005-02-28 19:24:19 +00002750</p>
2751
2752<h5>Arguments:</h5>
2753
2754<p>
2755<tt>address</tt> is the address to be prefetched, <tt>rw</tt> is the specifier
2756determining if the fetch should be for a read (0) or write (1), and
2757<tt>locality</tt> is a temporal locality specifier ranging from (0) - no
Chris Lattneraeffb4a2005-03-07 20:31:38 +00002758locality, to (3) - extremely local keep in cache. The <tt>rw</tt> and
Chris Lattner9a9d7ac2005-02-28 19:24:19 +00002759<tt>locality</tt> arguments must be constant integers.
2760</p>
2761
2762<h5>Semantics:</h5>
2763
2764<p>
2765This intrinsic does not modify the behavior of the program. In particular,
2766prefetches cannot trap and do not produce a value. On targets that support this
2767intrinsic, the prefetch can provide hints to the processor cache for better
2768performance.
2769</p>
2770
2771</div>
2772
Andrew Lenharth7f4ec3b2005-03-28 20:05:49 +00002773<!-- _______________________________________________________________________ -->
2774<div class="doc_subsubsection">
2775 <a name="i_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a>
2776</div>
2777
2778<div class="doc_text">
2779
2780<h5>Syntax:</h5>
2781<pre>
Reid Spencera8d451e2005-04-26 20:50:44 +00002782 declare void %llvm.pcmarker( uint &lt;id&gt; )
Andrew Lenharth7f4ec3b2005-03-28 20:05:49 +00002783</pre>
2784
2785<h5>Overview:</h5>
2786
2787
2788<p>
John Criswellfc6b8952005-05-16 16:17:45 +00002789The '<tt>llvm.pcmarker</tt>' intrinsic is a method to export a Program Counter
2790(PC) in a region of
Andrew Lenharth7f4ec3b2005-03-28 20:05:49 +00002791code to simulators and other tools. The method is target specific, but it is
2792expected that the marker will use exported symbols to transmit the PC of the marker.
2793The marker makes no guaranties that it will remain with any specific instruction
2794after optimizations. It is possible that the presense of a marker will inhibit
2795optimizations. The intended use is to be inserted after optmizations to allow
John Criswellfc6b8952005-05-16 16:17:45 +00002796correlations of simulation runs.
Andrew Lenharth7f4ec3b2005-03-28 20:05:49 +00002797</p>
2798
2799<h5>Arguments:</h5>
2800
2801<p>
2802<tt>id</tt> is a numerical id identifying the marker.
2803</p>
2804
2805<h5>Semantics:</h5>
2806
2807<p>
2808This intrinsic does not modify the behavior of the program. Backends that do not
2809support this intrinisic may ignore it.
2810</p>
2811
2812</div>
2813
Chris Lattner9a9d7ac2005-02-28 19:24:19 +00002814
John Criswell7123e272004-04-09 16:43:20 +00002815<!-- ======================================================================= -->
2816<div class="doc_subsection">
2817 <a name="int_os">Operating System Intrinsics</a>
2818</div>
2819
2820<div class="doc_text">
2821<p>
2822These intrinsics are provided by LLVM to support the implementation of
2823operating system level code.
2824</p>
2825
2826</div>
John Criswell183402a2004-04-12 15:02:16 +00002827
John Criswellcfd3bac2004-04-09 15:23:37 +00002828<!-- _______________________________________________________________________ -->
2829<div class="doc_subsubsection">
2830 <a name="i_readport">'<tt>llvm.readport</tt>' Intrinsic</a>
2831</div>
2832
2833<div class="doc_text">
2834
2835<h5>Syntax:</h5>
2836<pre>
Reid Spencera8d451e2005-04-26 20:50:44 +00002837 declare &lt;integer type&gt; %llvm.readport (&lt;integer type&gt; &lt;address&gt;)
John Criswellcfd3bac2004-04-09 15:23:37 +00002838</pre>
2839
2840<h5>Overview:</h5>
2841
2842<p>
John Criswell7123e272004-04-09 16:43:20 +00002843The '<tt>llvm.readport</tt>' intrinsic reads data from the specified hardware
2844I/O port.
John Criswellcfd3bac2004-04-09 15:23:37 +00002845</p>
2846
2847<h5>Arguments:</h5>
2848
2849<p>
John Criswell7123e272004-04-09 16:43:20 +00002850The argument to this intrinsic indicates the hardware I/O address from which
2851to read the data. The address is in the hardware I/O address namespace (as
2852opposed to being a memory location for memory mapped I/O).
John Criswellcfd3bac2004-04-09 15:23:37 +00002853</p>
2854
2855<h5>Semantics:</h5>
2856
2857<p>
John Criswell7123e272004-04-09 16:43:20 +00002858The '<tt>llvm.readport</tt>' intrinsic reads data from the hardware I/O port
2859specified by <i>address</i> and returns the value. The address and return
2860value must be integers, but the size is dependent upon the platform upon which
2861the program is code generated. For example, on x86, the address must be an
Misha Brukmancfa87bc2005-04-22 18:02:52 +00002862unsigned 16-bit value, and the return value must be 8, 16, or 32 bits.
John Criswellcfd3bac2004-04-09 15:23:37 +00002863</p>
2864
2865</div>
2866
2867<!-- _______________________________________________________________________ -->
2868<div class="doc_subsubsection">
2869 <a name="i_writeport">'<tt>llvm.writeport</tt>' Intrinsic</a>
2870</div>
2871
2872<div class="doc_text">
2873
2874<h5>Syntax:</h5>
2875<pre>
Chris Lattnerc3f59762004-12-09 17:30:23 +00002876 call void (&lt;integer type&gt;, &lt;integer type&gt;)*
2877 %llvm.writeport (&lt;integer type&gt; &lt;value&gt;,
2878 &lt;integer type&gt; &lt;address&gt;)
John Criswellcfd3bac2004-04-09 15:23:37 +00002879</pre>
2880
2881<h5>Overview:</h5>
2882
2883<p>
John Criswell7123e272004-04-09 16:43:20 +00002884The '<tt>llvm.writeport</tt>' intrinsic writes data to the specified hardware
2885I/O port.
John Criswellcfd3bac2004-04-09 15:23:37 +00002886</p>
2887
2888<h5>Arguments:</h5>
2889
2890<p>
John Criswell96db6fc2004-04-12 16:33:19 +00002891The first argument is the value to write to the I/O port.
John Criswellcfd3bac2004-04-09 15:23:37 +00002892</p>
2893
2894<p>
John Criswell96db6fc2004-04-12 16:33:19 +00002895The second argument indicates the hardware I/O address to which data should be
2896written. The address is in the hardware I/O address namespace (as opposed to
2897being a memory location for memory mapped I/O).
John Criswellcfd3bac2004-04-09 15:23:37 +00002898</p>
2899
2900<h5>Semantics:</h5>
2901
2902<p>
2903The '<tt>llvm.writeport</tt>' intrinsic writes <i>value</i> to the I/O port
2904specified by <i>address</i>. The address and value must be integers, but the
2905size is dependent upon the platform upon which the program is code generated.
Misha Brukmancfa87bc2005-04-22 18:02:52 +00002906For example, on x86, the address must be an unsigned 16-bit value, and the
John Criswell7123e272004-04-09 16:43:20 +00002907value written must be 8, 16, or 32 bits in length.
John Criswellcfd3bac2004-04-09 15:23:37 +00002908</p>
2909
2910</div>
Chris Lattner10610642004-02-14 04:08:35 +00002911
John Criswell183402a2004-04-12 15:02:16 +00002912<!-- _______________________________________________________________________ -->
2913<div class="doc_subsubsection">
2914 <a name="i_readio">'<tt>llvm.readio</tt>' Intrinsic</a>
2915</div>
2916
2917<div class="doc_text">
2918
2919<h5>Syntax:</h5>
2920<pre>
Reid Spencera8d451e2005-04-26 20:50:44 +00002921 declare &lt;result&gt; %llvm.readio (&lt;ty&gt; * &lt;pointer&gt;)
John Criswell183402a2004-04-12 15:02:16 +00002922</pre>
2923
2924<h5>Overview:</h5>
2925
2926<p>
2927The '<tt>llvm.readio</tt>' intrinsic reads data from a memory mapped I/O
2928address.
2929</p>
2930
2931<h5>Arguments:</h5>
2932
2933<p>
John Criswell96db6fc2004-04-12 16:33:19 +00002934The argument to this intrinsic is a pointer indicating the memory address from
2935which to read the data. The data must be a
2936<a href="#t_firstclass">first class</a> type.
John Criswell183402a2004-04-12 15:02:16 +00002937</p>
2938
2939<h5>Semantics:</h5>
2940
2941<p>
2942The '<tt>llvm.readio</tt>' intrinsic reads data from a memory mapped I/O
John Criswell96db6fc2004-04-12 16:33:19 +00002943location specified by <i>pointer</i> and returns the value. The argument must
2944be a pointer, and the return value must be a
2945<a href="#t_firstclass">first class</a> type. However, certain architectures
Misha Brukmancfa87bc2005-04-22 18:02:52 +00002946may not support I/O on all first class types. For example, 32-bit processors
John Criswell96db6fc2004-04-12 16:33:19 +00002947may only support I/O on data types that are 32 bits or less.
John Criswell183402a2004-04-12 15:02:16 +00002948</p>
2949
2950<p>
John Criswell96db6fc2004-04-12 16:33:19 +00002951This intrinsic enforces an in-order memory model for llvm.readio and
2952llvm.writeio calls on machines that use dynamic scheduling. Dynamically
2953scheduled processors may execute loads and stores out of order, re-ordering at
2954run time accesses to memory mapped I/O registers. Using these intrinsics
2955ensures that accesses to memory mapped I/O registers occur in program order.
John Criswell183402a2004-04-12 15:02:16 +00002956</p>
2957
2958</div>
2959
2960<!-- _______________________________________________________________________ -->
2961<div class="doc_subsubsection">
2962 <a name="i_writeio">'<tt>llvm.writeio</tt>' Intrinsic</a>
2963</div>
2964
2965<div class="doc_text">
2966
2967<h5>Syntax:</h5>
2968<pre>
Reid Spencera8d451e2005-04-26 20:50:44 +00002969 declare void %llvm.writeio (&lt;ty1&gt; &lt;value&gt;, &lt;ty2&gt; * &lt;pointer&gt;)
John Criswell183402a2004-04-12 15:02:16 +00002970</pre>
2971
2972<h5>Overview:</h5>
2973
2974<p>
2975The '<tt>llvm.writeio</tt>' intrinsic writes data to the specified memory
2976mapped I/O address.
2977</p>
2978
2979<h5>Arguments:</h5>
2980
2981<p>
John Criswell96db6fc2004-04-12 16:33:19 +00002982The first argument is the value to write to the memory mapped I/O location.
2983The second argument is a pointer indicating the memory address to which the
2984data should be written.
John Criswell183402a2004-04-12 15:02:16 +00002985</p>
2986
2987<h5>Semantics:</h5>
2988
2989<p>
2990The '<tt>llvm.writeio</tt>' intrinsic writes <i>value</i> to the memory mapped
John Criswell96db6fc2004-04-12 16:33:19 +00002991I/O address specified by <i>pointer</i>. The value must be a
2992<a href="#t_firstclass">first class</a> type. However, certain architectures
Misha Brukmancfa87bc2005-04-22 18:02:52 +00002993may not support I/O on all first class types. For example, 32-bit processors
John Criswell96db6fc2004-04-12 16:33:19 +00002994may only support I/O on data types that are 32 bits or less.
John Criswell183402a2004-04-12 15:02:16 +00002995</p>
2996
2997<p>
John Criswell96db6fc2004-04-12 16:33:19 +00002998This intrinsic enforces an in-order memory model for llvm.readio and
2999llvm.writeio calls on machines that use dynamic scheduling. Dynamically
3000scheduled processors may execute loads and stores out of order, re-ordering at
3001run time accesses to memory mapped I/O registers. Using these intrinsics
3002ensures that accesses to memory mapped I/O registers occur in program order.
John Criswell183402a2004-04-12 15:02:16 +00003003</p>
3004
3005</div>
3006
Chris Lattner10610642004-02-14 04:08:35 +00003007<!-- ======================================================================= -->
3008<div class="doc_subsection">
Chris Lattner33aec9e2004-02-12 17:01:32 +00003009 <a name="int_libc">Standard C Library Intrinsics</a>
3010</div>
3011
3012<div class="doc_text">
3013<p>
Chris Lattner10610642004-02-14 04:08:35 +00003014LLVM provides intrinsics for a few important standard C library functions.
3015These intrinsics allow source-language front-ends to pass information about the
3016alignment of the pointer arguments to the code generator, providing opportunity
3017for more efficient code generation.
Chris Lattner33aec9e2004-02-12 17:01:32 +00003018</p>
3019
3020</div>
3021
3022<!-- _______________________________________________________________________ -->
3023<div class="doc_subsubsection">
3024 <a name="i_memcpy">'<tt>llvm.memcpy</tt>' Intrinsic</a>
3025</div>
3026
3027<div class="doc_text">
3028
3029<h5>Syntax:</h5>
3030<pre>
Reid Spencerd4622352005-04-26 20:41:16 +00003031 declare void %llvm.memcpy(sbyte* &lt;dest&gt;, sbyte* &lt;src&gt;,
3032 uint &lt;len&gt;, uint &lt;align&gt;)
Chris Lattner33aec9e2004-02-12 17:01:32 +00003033</pre>
3034
3035<h5>Overview:</h5>
3036
3037<p>
3038The '<tt>llvm.memcpy</tt>' intrinsic copies a block of memory from the source
3039location to the destination location.
3040</p>
3041
3042<p>
3043Note that, unlike the standard libc function, the <tt>llvm.memcpy</tt> intrinsic
3044does not return a value, and takes an extra alignment argument.
3045</p>
3046
3047<h5>Arguments:</h5>
3048
3049<p>
3050The first argument is a pointer to the destination, the second is a pointer to
3051the source. The third argument is an (arbitrarily sized) integer argument
3052specifying the number of bytes to copy, and the fourth argument is the alignment
3053of the source and destination locations.
3054</p>
3055
Chris Lattner3301ced2004-02-12 21:18:15 +00003056<p>
3057If the call to this intrinisic has an alignment value that is not 0 or 1, then
3058the caller guarantees that the size of the copy is a multiple of the alignment
3059and that both the source and destination pointers are aligned to that boundary.
3060</p>
3061
Chris Lattner33aec9e2004-02-12 17:01:32 +00003062<h5>Semantics:</h5>
3063
3064<p>
3065The '<tt>llvm.memcpy</tt>' intrinsic copies a block of memory from the source
3066location to the destination location, which are not allowed to overlap. It
3067copies "len" bytes of memory over. If the argument is known to be aligned to
3068some boundary, this can be specified as the fourth argument, otherwise it should
3069be set to 0 or 1.
3070</p>
3071</div>
3072
3073
Chris Lattner0eb51b42004-02-12 18:10:10 +00003074<!-- _______________________________________________________________________ -->
3075<div class="doc_subsubsection">
3076 <a name="i_memmove">'<tt>llvm.memmove</tt>' Intrinsic</a>
3077</div>
3078
3079<div class="doc_text">
3080
3081<h5>Syntax:</h5>
3082<pre>
Reid Spencerd4622352005-04-26 20:41:16 +00003083 declare void %llvm.memmove(sbyte* &lt;dest&gt;, sbyte* &lt;src&gt;,
3084 uint &lt;len&gt;, uint &lt;align&gt;)
Chris Lattner0eb51b42004-02-12 18:10:10 +00003085</pre>
3086
3087<h5>Overview:</h5>
3088
3089<p>
3090The '<tt>llvm.memmove</tt>' intrinsic moves a block of memory from the source
3091location to the destination location. It is similar to the '<tt>llvm.memcpy</tt>'
3092intrinsic but allows the two memory locations to overlap.
3093</p>
3094
3095<p>
3096Note that, unlike the standard libc function, the <tt>llvm.memmove</tt> intrinsic
3097does not return a value, and takes an extra alignment argument.
3098</p>
3099
3100<h5>Arguments:</h5>
3101
3102<p>
3103The first argument is a pointer to the destination, the second is a pointer to
3104the source. The third argument is an (arbitrarily sized) integer argument
3105specifying the number of bytes to copy, and the fourth argument is the alignment
3106of the source and destination locations.
3107</p>
3108
Chris Lattner3301ced2004-02-12 21:18:15 +00003109<p>
3110If the call to this intrinisic has an alignment value that is not 0 or 1, then
3111the caller guarantees that the size of the copy is a multiple of the alignment
3112and that both the source and destination pointers are aligned to that boundary.
3113</p>
3114
Chris Lattner0eb51b42004-02-12 18:10:10 +00003115<h5>Semantics:</h5>
3116
3117<p>
3118The '<tt>llvm.memmove</tt>' intrinsic copies a block of memory from the source
3119location to the destination location, which may overlap. It
3120copies "len" bytes of memory over. If the argument is known to be aligned to
3121some boundary, this can be specified as the fourth argument, otherwise it should
3122be set to 0 or 1.
3123</p>
3124</div>
3125
Chris Lattner8ff75902004-01-06 05:31:32 +00003126
Chris Lattner10610642004-02-14 04:08:35 +00003127<!-- _______________________________________________________________________ -->
3128<div class="doc_subsubsection">
3129 <a name="i_memset">'<tt>llvm.memset</tt>' Intrinsic</a>
3130</div>
3131
3132<div class="doc_text">
3133
3134<h5>Syntax:</h5>
3135<pre>
Reid Spencerd4622352005-04-26 20:41:16 +00003136 declare void %llvm.memset(sbyte* &lt;dest&gt;, ubyte &lt;val&gt;,
3137 uint &lt;len&gt;, uint &lt;align&gt;)
Chris Lattner10610642004-02-14 04:08:35 +00003138</pre>
3139
3140<h5>Overview:</h5>
3141
3142<p>
3143The '<tt>llvm.memset</tt>' intrinsic fills a block of memory with a particular
3144byte value.
3145</p>
3146
3147<p>
3148Note that, unlike the standard libc function, the <tt>llvm.memset</tt> intrinsic
3149does not return a value, and takes an extra alignment argument.
3150</p>
3151
3152<h5>Arguments:</h5>
3153
3154<p>
3155The first argument is a pointer to the destination to fill, the second is the
3156byte value to fill it with, the third argument is an (arbitrarily sized) integer
3157argument specifying the number of bytes to fill, and the fourth argument is the
3158known alignment of destination location.
3159</p>
3160
3161<p>
3162If the call to this intrinisic has an alignment value that is not 0 or 1, then
3163the caller guarantees that the size of the copy is a multiple of the alignment
3164and that the destination pointer is aligned to that boundary.
3165</p>
3166
3167<h5>Semantics:</h5>
3168
3169<p>
3170The '<tt>llvm.memset</tt>' intrinsic fills "len" bytes of memory starting at the
3171destination location. If the argument is known to be aligned to some boundary,
3172this can be specified as the fourth argument, otherwise it should be set to 0 or
31731.
3174</p>
3175</div>
3176
3177
Chris Lattner32006282004-06-11 02:28:03 +00003178<!-- _______________________________________________________________________ -->
3179<div class="doc_subsubsection">
Alkis Evlogimenos26bbe932004-06-13 01:16:15 +00003180 <a name="i_isunordered">'<tt>llvm.isunordered</tt>' Intrinsic</a>
3181</div>
3182
3183<div class="doc_text">
3184
3185<h5>Syntax:</h5>
3186<pre>
Reid Spencera8d451e2005-04-26 20:50:44 +00003187 declare bool %llvm.isunordered(&lt;float or double&gt; Val1, &lt;float or double&gt; Val2)
Alkis Evlogimenos26bbe932004-06-13 01:16:15 +00003188</pre>
3189
3190<h5>Overview:</h5>
3191
3192<p>
3193The '<tt>llvm.isunordered</tt>' intrinsic returns true if either or both of the
3194specified floating point values is a NAN.
3195</p>
3196
3197<h5>Arguments:</h5>
3198
3199<p>
3200The arguments are floating point numbers of the same type.
3201</p>
3202
3203<h5>Semantics:</h5>
3204
3205<p>
3206If either or both of the arguments is a SNAN or QNAN, it returns true, otherwise
3207false.
3208</p>
3209</div>
3210
3211
Chris Lattnera4d74142005-07-21 01:29:16 +00003212<!-- _______________________________________________________________________ -->
3213<div class="doc_subsubsection">
3214 <a name="i_sqrt">'<tt>llvm.sqrt</tt>' Intrinsic</a>
3215</div>
3216
3217<div class="doc_text">
3218
3219<h5>Syntax:</h5>
3220<pre>
3221 declare &lt;float or double&gt; %llvm.sqrt(&lt;float or double&gt; Val)
3222</pre>
3223
3224<h5>Overview:</h5>
3225
3226<p>
3227The '<tt>llvm.sqrt</tt>' intrinsic returns the sqrt of the specified operand,
3228returning the same value as the libm '<tt>sqrt</tt>' function would. Unlike
3229<tt>sqrt</tt> in libm, however, <tt>llvm.sqrt</tt> has undefined behavior for
3230negative numbers (which allows for better optimization).
3231</p>
3232
3233<h5>Arguments:</h5>
3234
3235<p>
3236The argument and return value are floating point numbers of the same type.
3237</p>
3238
3239<h5>Semantics:</h5>
3240
3241<p>
3242This function returns the sqrt of the specified operand if it is a positive
3243floating point number.
3244</p>
3245</div>
3246
Andrew Lenharthec370fd2005-05-03 18:01:48 +00003247<!-- ======================================================================= -->
3248<div class="doc_subsection">
3249 <a name="int_count">Bit Counting Intrinsics</a>
3250</div>
3251
3252<div class="doc_text">
3253<p>
3254LLVM provides intrinsics for a few important bit counting operations.
3255These allow efficient code generation for some algorithms.
3256</p>
3257
3258</div>
3259
3260<!-- _______________________________________________________________________ -->
3261<div class="doc_subsubsection">
3262 <a name="int_ctpop">'<tt>llvm.ctpop</tt>' Intrinsic</a>
3263</div>
3264
3265<div class="doc_text">
3266
3267<h5>Syntax:</h5>
3268<pre>
3269 declare int %llvm.ctpop(int &lt;src&gt;)
3270
3271</pre>
3272
3273<h5>Overview:</h5>
3274
3275<p>
3276The '<tt>llvm.ctpop</tt>' intrinsic counts the number of ones in a variable.
3277</p>
3278
3279<h5>Arguments:</h5>
3280
3281<p>
Chris Lattnercfe6b372005-05-07 01:46:40 +00003282The only argument is the value to be counted. The argument may be of any
3283integer type. The return type must match the argument type.
Andrew Lenharthec370fd2005-05-03 18:01:48 +00003284</p>
3285
3286<h5>Semantics:</h5>
3287
3288<p>
3289The '<tt>llvm.ctpop</tt>' intrinsic counts the 1's in a variable.
3290</p>
3291</div>
3292
3293<!-- _______________________________________________________________________ -->
3294<div class="doc_subsubsection">
Andrew Lenharthec370fd2005-05-03 18:01:48 +00003295 <a name="int_ctlz">'<tt>llvm.ctlz</tt>' Intrinsic</a>
3296</div>
3297
3298<div class="doc_text">
3299
3300<h5>Syntax:</h5>
3301<pre>
3302 declare int %llvm.ctlz(int &lt;src&gt;)
3303
3304</pre>
3305
3306<h5>Overview:</h5>
3307
3308<p>
Chris Lattnercfe6b372005-05-07 01:46:40 +00003309The '<tt>llvm.ctlz</tt>' intrinsic counts the number of leading zeros in a
3310variable.
Andrew Lenharthec370fd2005-05-03 18:01:48 +00003311</p>
3312
3313<h5>Arguments:</h5>
3314
3315<p>
Chris Lattnercfe6b372005-05-07 01:46:40 +00003316The only argument is the value to be counted. The argument may be of any
3317integer type. The return type must match the argument type.
Andrew Lenharthec370fd2005-05-03 18:01:48 +00003318</p>
3319
3320<h5>Semantics:</h5>
3321
3322<p>
Chris Lattnereff29ab2005-05-15 19:39:26 +00003323The '<tt>llvm.ctlz</tt>' intrinsic counts the leading (most significant) zeros
3324in a variable. If the src == 0 then the result is the size in bits of the type
3325of src. For example, <tt>llvm.cttz(int 2) = 30</tt>.
Andrew Lenharthec370fd2005-05-03 18:01:48 +00003326</p>
3327</div>
Chris Lattner32006282004-06-11 02:28:03 +00003328
3329
Chris Lattnereff29ab2005-05-15 19:39:26 +00003330
3331<!-- _______________________________________________________________________ -->
3332<div class="doc_subsubsection">
3333 <a name="int_cttz">'<tt>llvm.cttz</tt>' Intrinsic</a>
3334</div>
3335
3336<div class="doc_text">
3337
3338<h5>Syntax:</h5>
3339<pre>
3340 declare int %llvm.cttz(int &lt;src&gt;)
3341
3342</pre>
3343
3344<h5>Overview:</h5>
3345
3346<p>
3347The '<tt>llvm.cttz</tt>' intrinsic counts the number of trailing zeros.
3348</p>
3349
3350<h5>Arguments:</h5>
3351
3352<p>
3353The only argument is the value to be counted. The argument may be of any
3354integer type. The return type must match the argument type.
3355</p>
3356
3357<h5>Semantics:</h5>
3358
3359<p>
3360The '<tt>llvm.cttz</tt>' intrinsic counts the trailing (least significant) zeros
3361in a variable. If the src == 0 then the result is the size in bits of the type
3362of src. For example, <tt>llvm.cttz(2) = 1</tt>.
3363</p>
3364</div>
3365
Chris Lattner8ff75902004-01-06 05:31:32 +00003366<!-- ======================================================================= -->
3367<div class="doc_subsection">
3368 <a name="int_debugger">Debugger Intrinsics</a>
3369</div>
3370
3371<div class="doc_text">
3372<p>
3373The LLVM debugger intrinsics (which all start with <tt>llvm.dbg.</tt> prefix),
3374are described in the <a
3375href="SourceLevelDebugging.html#format_common_intrinsics">LLVM Source Level
3376Debugging</a> document.
3377</p>
3378</div>
3379
3380
Chris Lattner00950542001-06-06 20:29:01 +00003381<!-- *********************************************************************** -->
Chris Lattner00950542001-06-06 20:29:01 +00003382<hr>
Misha Brukmandaa4cb02004-03-01 17:47:27 +00003383<address>
3384 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
3385 src="http://jigsaw.w3.org/css-validator/images/vcss" alt="Valid CSS!"></a>
3386 <a href="http://validator.w3.org/check/referer"><img
3387 src="http://www.w3.org/Icons/valid-html401" alt="Valid HTML 4.01!" /></a>
3388
3389 <a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
3390 <a href="http://llvm.cs.uiuc.edu">The LLVM Compiler Infrastructure</a><br>
3391 Last modified: $Date$
3392</address>
Misha Brukman9d0919f2003-11-08 01:05:38 +00003393</body>
3394</html>