blob: 47fc33ebbb7f59e08c2af983c7c4a044158143db [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>
6 <link rel="stylesheet" href="llvm.css" type="text/css">
7</head>
Chris Lattnerd7923912004-05-23 21:06:01 +00008
Misha Brukman9d0919f2003-11-08 01:05:38 +00009<body>
Chris Lattnerd7923912004-05-23 21:06:01 +000010
Chris Lattner261efe92003-11-25 01:02:51 +000011<div class="doc_title"> LLVM Language Reference Manual </div>
Chris Lattner00950542001-06-06 20:29:01 +000012<ol>
Misha Brukman9d0919f2003-11-08 01:05:38 +000013 <li><a href="#abstract">Abstract</a></li>
14 <li><a href="#introduction">Introduction</a></li>
15 <li><a href="#identifiers">Identifiers</a></li>
Chris Lattner00950542001-06-06 20:29:01 +000016 <li><a href="#typesystem">Type System</a>
17 <ol>
Chris Lattner261efe92003-11-25 01:02:51 +000018 <li><a href="#t_primitive">Primitive Types</a>
19 <ol>
Misha Brukman9d0919f2003-11-08 01:05:38 +000020 <li><a href="#t_classifications">Type Classifications</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +000021 </ol>
22 </li>
Chris Lattner00950542001-06-06 20:29:01 +000023 <li><a href="#t_derived">Derived Types</a>
24 <ol>
Chris Lattner261efe92003-11-25 01:02:51 +000025 <li><a href="#t_array">Array Type</a></li>
Misha Brukman9d0919f2003-11-08 01:05:38 +000026 <li><a href="#t_function">Function Type</a></li>
27 <li><a href="#t_pointer">Pointer Type</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +000028 <li><a href="#t_struct">Structure Type</a></li>
29<!-- <li><a href="#t_packed" >Packed Type</a> -->
30 </ol>
31 </li>
32 </ol>
33 </li>
Chris Lattner00950542001-06-06 20:29:01 +000034 <li><a href="#highlevel">High Level Structure</a>
35 <ol>
Misha Brukman9d0919f2003-11-08 01:05:38 +000036 <li><a href="#modulestructure">Module Structure</a></li>
37 <li><a href="#globalvars">Global Variables</a></li>
38 <li><a href="#functionstructure">Function Structure</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +000039 </ol>
40 </li>
Chris Lattner00950542001-06-06 20:29:01 +000041 <li><a href="#instref">Instruction Reference</a>
42 <ol>
43 <li><a href="#terminators">Terminator Instructions</a>
44 <ol>
Chris Lattner261efe92003-11-25 01:02:51 +000045 <li><a href="#i_ret">'<tt>ret</tt>' Instruction</a></li>
46 <li><a href="#i_br">'<tt>br</tt>' Instruction</a></li>
Misha Brukman9d0919f2003-11-08 01:05:38 +000047 <li><a href="#i_switch">'<tt>switch</tt>' Instruction</a></li>
48 <li><a href="#i_invoke">'<tt>invoke</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +000049 <li><a href="#i_unwind">'<tt>unwind</tt>' Instruction</a></li>
50 </ol>
51 </li>
Chris Lattner00950542001-06-06 20:29:01 +000052 <li><a href="#binaryops">Binary Operations</a>
53 <ol>
Chris Lattner261efe92003-11-25 01:02:51 +000054 <li><a href="#i_add">'<tt>add</tt>' Instruction</a></li>
55 <li><a href="#i_sub">'<tt>sub</tt>' Instruction</a></li>
56 <li><a href="#i_mul">'<tt>mul</tt>' Instruction</a></li>
57 <li><a href="#i_div">'<tt>div</tt>' Instruction</a></li>
58 <li><a href="#i_rem">'<tt>rem</tt>' Instruction</a></li>
Misha Brukman9d0919f2003-11-08 01:05:38 +000059 <li><a href="#i_setcc">'<tt>set<i>cc</i></tt>' Instructions</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +000060 </ol>
61 </li>
Chris Lattner00950542001-06-06 20:29:01 +000062 <li><a href="#bitwiseops">Bitwise Binary Operations</a>
63 <ol>
Misha Brukman9d0919f2003-11-08 01:05:38 +000064 <li><a href="#i_and">'<tt>and</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +000065 <li><a href="#i_or">'<tt>or</tt>' Instruction</a></li>
Misha Brukman9d0919f2003-11-08 01:05:38 +000066 <li><a href="#i_xor">'<tt>xor</tt>' Instruction</a></li>
67 <li><a href="#i_shl">'<tt>shl</tt>' Instruction</a></li>
68 <li><a href="#i_shr">'<tt>shr</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +000069 </ol>
70 </li>
Chris Lattner00950542001-06-06 20:29:01 +000071 <li><a href="#memoryops">Memory Access Operations</a>
72 <ol>
Chris Lattner261efe92003-11-25 01:02:51 +000073 <li><a href="#i_malloc">'<tt>malloc</tt>' Instruction</a></li>
74 <li><a href="#i_free">'<tt>free</tt>' Instruction</a></li>
75 <li><a href="#i_alloca">'<tt>alloca</tt>' Instruction</a></li>
76 <li><a href="#i_load">'<tt>load</tt>' Instruction</a></li>
77 <li><a href="#i_store">'<tt>store</tt>' Instruction</a></li>
78 <li><a href="#i_getelementptr">'<tt>getelementptr</tt>' Instruction</a></li>
79 </ol>
80 </li>
Chris Lattner00950542001-06-06 20:29:01 +000081 <li><a href="#otherops">Other Operations</a>
82 <ol>
Chris Lattner261efe92003-11-25 01:02:51 +000083 <li><a href="#i_phi">'<tt>phi</tt>' Instruction</a></li>
Misha Brukman9d0919f2003-11-08 01:05:38 +000084 <li><a href="#i_cast">'<tt>cast .. to</tt>' Instruction</a></li>
Chris Lattnercc37aae2004-03-12 05:50:16 +000085 <li><a href="#i_select">'<tt>select</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +000086 <li><a href="#i_call">'<tt>call</tt>' Instruction</a></li>
Misha Brukman9d0919f2003-11-08 01:05:38 +000087 <li><a href="#i_vanext">'<tt>vanext</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +000088 <li><a href="#i_vaarg">'<tt>vaarg</tt>' Instruction</a></li>
Chris Lattner00950542001-06-06 20:29:01 +000089 </ol>
Chris Lattner261efe92003-11-25 01:02:51 +000090 </li>
Chris Lattner00950542001-06-06 20:29:01 +000091 </ol>
Chris Lattner261efe92003-11-25 01:02:51 +000092 </li>
Chris Lattnerd9ad5b32003-05-08 04:57:36 +000093 <li><a href="#intrinsics">Intrinsic Functions</a>
Chris Lattnerd9ad5b32003-05-08 04:57:36 +000094 <ol>
Chris Lattner261efe92003-11-25 01:02:51 +000095 <li><a href="#int_varargs">Variable Argument Handling Intrinsics</a>
96 <ol>
97 <li><a href="#i_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a></li>
98 <li><a href="#i_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a></li>
99 <li><a href="#i_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a></li>
100 </ol>
101 </li>
Chris Lattnerd7923912004-05-23 21:06:01 +0000102 <li><a href="#int_gc">Accurate Garbage Collection Intrinsics</a>
103 <ol>
104 <li><a href="#i_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a></li>
105 <li><a href="#i_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a></li>
106 <li><a href="#i_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a></li>
107 </ol>
108 </li>
Chris Lattner10610642004-02-14 04:08:35 +0000109 <li><a href="#int_codegen">Code Generator Intrinsics</a>
110 <ol>
111 <li><a href="#i_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a></li>
112 <li><a href="#i_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a></li>
John Criswell7123e272004-04-09 16:43:20 +0000113 </ol>
114 </li>
115 <li><a href="#int_os">Operating System Intrinsics</a>
116 <ol>
Chris Lattner32006282004-06-11 02:28:03 +0000117 <li><a href="#i_readport">'<tt>llvm.readport</tt>' Intrinsic</a></li>
118 <li><a href="#i_writeport">'<tt>llvm.writeport</tt>' Intrinsic</a></li>
John Criswell183402a2004-04-12 15:02:16 +0000119 <li><a href="#i_readio">'<tt>llvm.readio</tt>' Intrinsic</a></li>
120 <li><a href="#i_writeio">'<tt>llvm.writeio</tt>' Intrinsic</a></li>
Chris Lattner10610642004-02-14 04:08:35 +0000121 </ol>
Chris Lattner33aec9e2004-02-12 17:01:32 +0000122 <li><a href="#int_libc">Standard C Library Intrinsics</a>
123 <ol>
124 <li><a href="#i_memcpy">'<tt>llvm.memcpy</tt>' Intrinsic</a></li>
Chris Lattner0eb51b42004-02-12 18:10:10 +0000125 <li><a href="#i_memmove">'<tt>llvm.memmove</tt>' Intrinsic</a></li>
Chris Lattner10610642004-02-14 04:08:35 +0000126 <li><a href="#i_memset">'<tt>llvm.memset</tt>' Intrinsic</a></li>
Alkis Evlogimenos96853722004-06-12 19:19:14 +0000127 <li><a href="#i_isunordered">'<tt>llvm.isunordered</tt>' Intrinsic</a></li>
Chris Lattner33aec9e2004-02-12 17:01:32 +0000128 </ol>
129 </li>
Chris Lattnerd7923912004-05-23 21:06:01 +0000130 <li><a href="#int_debugger">Debugger intrinsics</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000131 </ol>
132 </li>
Chris Lattner00950542001-06-06 20:29:01 +0000133</ol>
Chris Lattnerd7923912004-05-23 21:06:01 +0000134
135<div class="doc_author">
136 <p>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a>
137 and <a href="mailto:vadve@cs.uiuc.edu">Vikram Adve</a></p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000138</div>
Chris Lattnerd7923912004-05-23 21:06:01 +0000139
Chris Lattner00950542001-06-06 20:29:01 +0000140<!-- *********************************************************************** -->
Chris Lattner261efe92003-11-25 01:02:51 +0000141<div class="doc_section"> <a name="abstract">Abstract </a></div>
142<!-- *********************************************************************** -->
Chris Lattnerd7923912004-05-23 21:06:01 +0000143
Misha Brukman9d0919f2003-11-08 01:05:38 +0000144<div class="doc_text">
Chris Lattner261efe92003-11-25 01:02:51 +0000145<p>This document is a reference manual for the LLVM assembly language.
146LLVM is an SSA based representation that provides type safety,
147low-level operations, flexibility, and the capability of representing
148'all' high-level languages cleanly. It is the common code
149representation used throughout all phases of the LLVM compilation
150strategy.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000151</div>
Chris Lattnerd7923912004-05-23 21:06:01 +0000152
Chris Lattner00950542001-06-06 20:29:01 +0000153<!-- *********************************************************************** -->
Chris Lattner261efe92003-11-25 01:02:51 +0000154<div class="doc_section"> <a name="introduction">Introduction</a> </div>
155<!-- *********************************************************************** -->
Chris Lattnerd7923912004-05-23 21:06:01 +0000156
Misha Brukman9d0919f2003-11-08 01:05:38 +0000157<div class="doc_text">
Chris Lattnerd7923912004-05-23 21:06:01 +0000158
Chris Lattner261efe92003-11-25 01:02:51 +0000159<p>The LLVM code representation is designed to be used in three
160different forms: as an in-memory compiler IR, as an on-disk bytecode
161representation (suitable for fast loading by a Just-In-Time compiler),
162and as a human readable assembly language representation. This allows
163LLVM to provide a powerful intermediate representation for efficient
164compiler transformations and analysis, while providing a natural means
165to debug and visualize the transformations. The three different forms
166of LLVM are all equivalent. This document describes the human readable
167representation and notation.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000168
Chris Lattner261efe92003-11-25 01:02:51 +0000169<p>The LLVM representation aims to be a light-weight and low-level
170while being expressive, typed, and extensible at the same time. It
171aims to be a "universal IR" of sorts, by being at a low enough level
172that high-level ideas may be cleanly mapped to it (similar to how
173microprocessors are "universal IR's", allowing many source languages to
174be mapped to them). By providing type information, LLVM can be used as
175the target of optimizations: for example, through pointer analysis, it
176can be proven that a C automatic variable is never accessed outside of
177the current function... allowing it to be promoted to a simple SSA
178value instead of a memory location.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000179
Misha Brukman9d0919f2003-11-08 01:05:38 +0000180</div>
Chris Lattnerd7923912004-05-23 21:06:01 +0000181
Chris Lattner00950542001-06-06 20:29:01 +0000182<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000183<div class="doc_subsubsection"> <a name="wellformed">Well-Formedness</a> </div>
Chris Lattnerd7923912004-05-23 21:06:01 +0000184
Misha Brukman9d0919f2003-11-08 01:05:38 +0000185<div class="doc_text">
Chris Lattnerd7923912004-05-23 21:06:01 +0000186
Chris Lattner261efe92003-11-25 01:02:51 +0000187<p>It is important to note that this document describes 'well formed'
188LLVM assembly language. There is a difference between what the parser
189accepts and what is considered 'well formed'. For example, the
190following instruction is syntactically okay, but not well formed:</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000191
192<pre>
193 %x = <a href="#i_add">add</a> int 1, %x
194</pre>
195
Chris Lattner261efe92003-11-25 01:02:51 +0000196<p>...because the definition of <tt>%x</tt> does not dominate all of
197its uses. The LLVM infrastructure provides a verification pass that may
198be used to verify that an LLVM module is well formed. This pass is
199automatically run by the parser after parsing input assembly, and by
200the optimizer before it outputs bytecode. The violations pointed out
201by the verifier pass indicate bugs in transformation passes or input to
202the parser.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000203
Chris Lattner261efe92003-11-25 01:02:51 +0000204<!-- Describe the typesetting conventions here. --> </div>
Chris Lattnerd7923912004-05-23 21:06:01 +0000205
Chris Lattner00950542001-06-06 20:29:01 +0000206<!-- *********************************************************************** -->
Chris Lattner261efe92003-11-25 01:02:51 +0000207<div class="doc_section"> <a name="identifiers">Identifiers</a> </div>
Chris Lattner00950542001-06-06 20:29:01 +0000208<!-- *********************************************************************** -->
Chris Lattnerd7923912004-05-23 21:06:01 +0000209
Misha Brukman9d0919f2003-11-08 01:05:38 +0000210<div class="doc_text">
Chris Lattnerd7923912004-05-23 21:06:01 +0000211
Chris Lattner261efe92003-11-25 01:02:51 +0000212<p>LLVM uses three different forms of identifiers, for different
213purposes:</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000214
Chris Lattner00950542001-06-06 20:29:01 +0000215<ol>
Chris Lattner261efe92003-11-25 01:02:51 +0000216 <li>Numeric constants are represented as you would expect: 12, -3
John Criswell4457dc92004-04-09 16:48:45 +0000217123.421, etc. Floating point constants have an optional hexadecimal
Chris Lattner261efe92003-11-25 01:02:51 +0000218notation.</li>
219 <li>Named values are represented as a string of characters with a '%'
220prefix. For example, %foo, %DivisionByZero,
221%a.really.long.identifier. The actual regular expression used is '<tt>%[a-zA-Z$._][a-zA-Z$._0-9]*</tt>'.
222Identifiers which require other characters in their names can be
223surrounded with quotes. In this way, anything except a <tt>"</tt>
224character can be used in a name.</li>
225 <li>Unnamed values are represented as an unsigned numeric value with
226a '%' prefix. For example, %12, %2, %44.</li>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000227</ol>
John Criswell6794d922004-03-12 21:19:06 +0000228<p>LLVM requires that values start with a '%' sign for two reasons:
Chris Lattner261efe92003-11-25 01:02:51 +0000229Compilers don't need to worry about name clashes with reserved words,
230and the set of reserved words may be expanded in the future without
231penalty. Additionally, unnamed identifiers allow a compiler to quickly
232come up with a temporary variable without having to avoid symbol table
233conflicts.</p>
234<p>Reserved words in LLVM are very similar to reserved words in other
235languages. There are keywords for different opcodes ('<tt><a
236 href="#i_add">add</a></tt>', '<tt><a href="#i_cast">cast</a></tt>', '<tt><a
237 href="#i_ret">ret</a></tt>', etc...), for primitive type names ('<tt><a
238 href="#t_void">void</a></tt>', '<tt><a href="#t_uint">uint</a></tt>',
239etc...), and others. These reserved words cannot conflict with
240variable names, because none of them start with a '%' character.</p>
241<p>Here is an example of LLVM code to multiply the integer variable '<tt>%X</tt>'
242by 8:</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000243<p>The easy way:</p>
Chris Lattner261efe92003-11-25 01:02:51 +0000244<pre> %result = <a href="#i_mul">mul</a> uint %X, 8<br></pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000245<p>After strength reduction:</p>
Chris Lattner261efe92003-11-25 01:02:51 +0000246<pre> %result = <a href="#i_shl">shl</a> uint %X, ubyte 3<br></pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000247<p>And the hard way:</p>
Chris Lattner261efe92003-11-25 01:02:51 +0000248<pre> <a href="#i_add">add</a> uint %X, %X <i>; yields {uint}:%0</i>
249 <a
250 href="#i_add">add</a> uint %0, %0 <i>; yields {uint}:%1</i>
251 %result = <a
252 href="#i_add">add</a> uint %1, %1<br></pre>
253<p>This last way of multiplying <tt>%X</tt> by 8 illustrates several
254important lexical features of LLVM:</p>
Chris Lattner00950542001-06-06 20:29:01 +0000255<ol>
Chris Lattner261efe92003-11-25 01:02:51 +0000256 <li>Comments are delimited with a '<tt>;</tt>' and go until the end
257of line.</li>
258 <li>Unnamed temporaries are created when the result of a computation
259is not assigned to a named value.</li>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000260 <li>Unnamed temporaries are numbered sequentially</li>
261</ol>
Chris Lattner261efe92003-11-25 01:02:51 +0000262<p>...and it also show a convention that we follow in this document.
263When demonstrating instructions, we will follow an instruction with a
264comment that defines the type and name of value produced. Comments are
265shown in italic text.</p>
266<p>The one non-intuitive notation for constants is the optional
267hexidecimal form of floating point constants. For example, the form '<tt>double
Chris Lattner2b7d3202002-05-06 03:03:22 +00002680x432ff973cafa8000</tt>' is equivalent to (but harder to read than) '<tt>double
Chris Lattner261efe92003-11-25 01:02:51 +00002694.5e+15</tt>' which is also supported by the parser. The only time
270hexadecimal floating point constants are useful (and the only time that
271they are generated by the disassembler) is when an FP constant has to
272be emitted that is not representable as a decimal floating point number
273exactly. For example, NaN's, infinities, and other special cases are
274represented in their IEEE hexadecimal format so that assembly and
275disassembly do not cause any bits to change in the constants.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000276</div>
Chris Lattner00950542001-06-06 20:29:01 +0000277<!-- *********************************************************************** -->
Chris Lattner261efe92003-11-25 01:02:51 +0000278<div class="doc_section"> <a name="typesystem">Type System</a> </div>
279<!-- *********************************************************************** -->
Misha Brukman9d0919f2003-11-08 01:05:38 +0000280<div class="doc_text">
Misha Brukman9d0919f2003-11-08 01:05:38 +0000281<p>The LLVM type system is one of the most important features of the
Chris Lattner261efe92003-11-25 01:02:51 +0000282intermediate representation. Being typed enables a number of
283optimizations to be performed on the IR directly, without having to do
284extra analyses on the side before the transformation. A strong type
285system makes it easier to read the generated code and enables novel
286analyses and transformations that are not feasible to perform on normal
287three address code representations.</p>
Chris Lattner7bae3952002-06-25 18:03:17 +0000288<!-- The written form for the type system was heavily influenced by the
289syntactic problems with types in the C language<sup><a
Chris Lattner261efe92003-11-25 01:02:51 +0000290href="#rw_stroustrup">1</a></sup>.<p> --> </div>
Chris Lattner00950542001-06-06 20:29:01 +0000291<!-- ======================================================================= -->
Chris Lattner261efe92003-11-25 01:02:51 +0000292<div class="doc_subsection"> <a name="t_primitive">Primitive Types</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000293<div class="doc_text">
John Criswell4457dc92004-04-09 16:48:45 +0000294<p>The primitive types are the fundamental building blocks of the LLVM
Chris Lattner261efe92003-11-25 01:02:51 +0000295system. The current set of primitive types are as follows:</p>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000296
297<table border="0" style="align: center">
Chris Lattner261efe92003-11-25 01:02:51 +0000298 <tbody>
299 <tr>
300 <td>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000301 <table border="1" cellspacing="0" cellpadding="4" style="align: center">
Chris Lattner261efe92003-11-25 01:02:51 +0000302 <tbody>
303 <tr>
304 <td><tt>void</tt></td>
305 <td>No value</td>
306 </tr>
307 <tr>
308 <td><tt>ubyte</tt></td>
309 <td>Unsigned 8 bit value</td>
310 </tr>
311 <tr>
312 <td><tt>ushort</tt></td>
313 <td>Unsigned 16 bit value</td>
314 </tr>
315 <tr>
316 <td><tt>uint</tt></td>
317 <td>Unsigned 32 bit value</td>
318 </tr>
319 <tr>
320 <td><tt>ulong</tt></td>
321 <td>Unsigned 64 bit value</td>
322 </tr>
323 <tr>
324 <td><tt>float</tt></td>
325 <td>32 bit floating point value</td>
326 </tr>
327 <tr>
328 <td><tt>label</tt></td>
329 <td>Branch destination</td>
330 </tr>
331 </tbody>
332 </table>
333 </td>
334 <td valign="top">
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000335 <table border="1" cellspacing="0" cellpadding="4">
Chris Lattner261efe92003-11-25 01:02:51 +0000336 <tbody>
337 <tr>
338 <td><tt>bool</tt></td>
339 <td>True or False value</td>
340 </tr>
341 <tr>
342 <td><tt>sbyte</tt></td>
343 <td>Signed 8 bit value</td>
344 </tr>
345 <tr>
346 <td><tt>short</tt></td>
347 <td>Signed 16 bit value</td>
348 </tr>
349 <tr>
350 <td><tt>int</tt></td>
351 <td>Signed 32 bit value</td>
352 </tr>
353 <tr>
354 <td><tt>long</tt></td>
355 <td>Signed 64 bit value</td>
356 </tr>
357 <tr>
358 <td><tt>double</tt></td>
359 <td>64 bit floating point value</td>
360 </tr>
361 </tbody>
362 </table>
363 </td>
364 </tr>
365 </tbody>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000366</table>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000367
Misha Brukman9d0919f2003-11-08 01:05:38 +0000368</div>
Chris Lattner00950542001-06-06 20:29:01 +0000369<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000370<div class="doc_subsubsection"> <a name="t_classifications">Type
371Classifications</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000372<div class="doc_text">
Chris Lattner261efe92003-11-25 01:02:51 +0000373<p>These different primitive types fall into a few useful
374classifications:</p>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000375
376<table border="1" cellspacing="0" cellpadding="4">
Chris Lattner261efe92003-11-25 01:02:51 +0000377 <tbody>
378 <tr>
379 <td><a name="t_signed">signed</a></td>
380 <td><tt>sbyte, short, int, long, float, double</tt></td>
381 </tr>
382 <tr>
383 <td><a name="t_unsigned">unsigned</a></td>
384 <td><tt>ubyte, ushort, uint, ulong</tt></td>
385 </tr>
386 <tr>
387 <td><a name="t_integer">integer</a></td>
388 <td><tt>ubyte, sbyte, ushort, short, uint, int, ulong, long</tt></td>
389 </tr>
390 <tr>
391 <td><a name="t_integral">integral</a></td>
392 <td><tt>bool, ubyte, sbyte, ushort, short, uint, int, ulong, long</tt></td>
393 </tr>
394 <tr>
395 <td><a name="t_floating">floating point</a></td>
396 <td><tt>float, double</tt></td>
397 </tr>
398 <tr>
399 <td><a name="t_firstclass">first class</a></td>
400 <td><tt>bool, ubyte, sbyte, ushort, short,<br>
401uint, int, ulong, long, float, double, <a href="#t_pointer">pointer</a></tt></td>
402 </tr>
403 </tbody>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000404</table>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000405
Chris Lattner261efe92003-11-25 01:02:51 +0000406<p>The <a href="#t_firstclass">first class</a> types are perhaps the
407most important. Values of these types are the only ones which can be
408produced by instructions, passed as arguments, or used as operands to
409instructions. This means that all structures and arrays must be
410manipulated either by pointer or by component.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000411</div>
Chris Lattner00950542001-06-06 20:29:01 +0000412<!-- ======================================================================= -->
Chris Lattner261efe92003-11-25 01:02:51 +0000413<div class="doc_subsection"> <a name="t_derived">Derived Types</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000414<div class="doc_text">
Chris Lattner261efe92003-11-25 01:02:51 +0000415<p>The real power in LLVM comes from the derived types in the system.
416This is what allows a programmer to represent arrays, functions,
417pointers, and other useful types. Note that these derived types may be
418recursive: For example, it is possible to have a two dimensional array.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000419</div>
Chris Lattner00950542001-06-06 20:29:01 +0000420<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000421<div class="doc_subsubsection"> <a name="t_array">Array Type</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000422<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +0000423<h5>Overview:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000424<p>The array type is a very simple derived type that arranges elements
Chris Lattner261efe92003-11-25 01:02:51 +0000425sequentially in memory. The array type requires a size (number of
426elements) and an underlying data type.</p>
Chris Lattner7faa8832002-04-14 06:13:44 +0000427<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000428<pre> [&lt;# elements&gt; x &lt;elementtype&gt;]<br></pre>
429<p>The number of elements is a constant integer value, elementtype may
430be any type with a size.</p>
Chris Lattner7faa8832002-04-14 06:13:44 +0000431<h5>Examples:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000432<p> <tt>[40 x int ]</tt>: Array of 40 integer values.<br>
433<tt>[41 x int ]</tt>: Array of 41 integer values.<br>
434<tt>[40 x uint]</tt>: Array of 40 unsigned integer values.</p>
435<p> </p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000436<p>Here are some examples of multidimensional arrays:</p>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000437
Misha Brukman9d0919f2003-11-08 01:05:38 +0000438<table border="0" cellpadding="0" cellspacing="0">
Chris Lattner261efe92003-11-25 01:02:51 +0000439 <tbody>
440 <tr>
441 <td><tt>[3 x [4 x int]]</tt></td>
442 <td>: 3x4 array integer values.</td>
443 </tr>
444 <tr>
445 <td><tt>[12 x [10 x float]]</tt></td>
446 <td>: 12x10 array of single precision floating point values.</td>
447 </tr>
448 <tr>
449 <td><tt>[2 x [3 x [4 x uint]]]</tt></td>
450 <td>: 2x3x4 array of unsigned integer values.</td>
451 </tr>
452 </tbody>
Chris Lattner00950542001-06-06 20:29:01 +0000453</table>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000454
Misha Brukman9d0919f2003-11-08 01:05:38 +0000455</div>
Chris Lattner00950542001-06-06 20:29:01 +0000456<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000457<div class="doc_subsubsection"> <a name="t_function">Function Type</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000458<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +0000459<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000460<p>The function type can be thought of as a function signature. It
461consists of a return type and a list of formal parameter types.
John Criswell009900b2003-11-25 21:45:46 +0000462Function types are usually used to build virtual function tables
Chris Lattner261efe92003-11-25 01:02:51 +0000463(which are structures of pointers to functions), for indirect function
464calls, and when defining a function.</p>
John Criswell009900b2003-11-25 21:45:46 +0000465<p>
466The return type of a function type cannot be an aggregate type.
467</p>
Chris Lattner00950542001-06-06 20:29:01 +0000468<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000469<pre> &lt;returntype&gt; (&lt;parameter list&gt;)<br></pre>
470<p>Where '<tt>&lt;parameter list&gt;</tt>' is a comma-separated list of
471type specifiers. Optionally, the parameter list may include a type <tt>...</tt>,
Chris Lattner27f71f22003-09-03 00:41:47 +0000472which indicates that the function takes a variable number of arguments.
473Variable argument functions can access their arguments with the <a
Chris Lattner261efe92003-11-25 01:02:51 +0000474 href="#int_varargs">variable argument handling intrinsic</a> functions.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000475<h5>Examples:</h5>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000476
Misha Brukman9d0919f2003-11-08 01:05:38 +0000477<table border="0" cellpadding="0" cellspacing="0">
Chris Lattner261efe92003-11-25 01:02:51 +0000478 <tbody>
479 <tr>
480 <td><tt>int (int)</tt></td>
481 <td>: function taking an <tt>int</tt>, returning an <tt>int</tt></td>
482 </tr>
483 <tr>
484 <td><tt>float (int, int *) *</tt></td>
485 <td>: <a href="#t_pointer">Pointer</a> to a function that takes
486an <tt>int</tt> and a <a href="#t_pointer">pointer</a> to <tt>int</tt>,
487returning <tt>float</tt>.</td>
488 </tr>
489 <tr>
490 <td><tt>int (sbyte *, ...)</tt></td>
491 <td>: A vararg function that takes at least one <a
492 href="#t_pointer">pointer</a> to <tt>sbyte</tt> (signed char in C),
493which returns an integer. This is the signature for <tt>printf</tt>
494in LLVM.</td>
495 </tr>
496 </tbody>
Chris Lattner00950542001-06-06 20:29:01 +0000497</table>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000498
Misha Brukman9d0919f2003-11-08 01:05:38 +0000499</div>
Chris Lattner00950542001-06-06 20:29:01 +0000500<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000501<div class="doc_subsubsection"> <a name="t_struct">Structure Type</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000502<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +0000503<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000504<p>The structure type is used to represent a collection of data members
505together in memory. The packing of the field types is defined to match
506the ABI of the underlying processor. The elements of a structure may
507be any type that has a size.</p>
508<p>Structures are accessed using '<tt><a href="#i_load">load</a></tt>
509and '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a
510field with the '<tt><a href="#i_getelementptr">getelementptr</a></tt>'
511instruction.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000512<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000513<pre> { &lt;type list&gt; }<br></pre>
Chris Lattner00950542001-06-06 20:29:01 +0000514<h5>Examples:</h5>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000515
Misha Brukman9d0919f2003-11-08 01:05:38 +0000516<table border="0" cellpadding="0" cellspacing="0">
Chris Lattner261efe92003-11-25 01:02:51 +0000517 <tbody>
518 <tr>
519 <td><tt>{ int, int, int }</tt></td>
520 <td>: a triple of three <tt>int</tt> values</td>
521 </tr>
522 <tr>
523 <td><tt>{ float, int (int) * }</tt></td>
524 <td>: A pair, where the first element is a <tt>float</tt> and the
525second element is a <a href="#t_pointer">pointer</a> to a <a
Chris Lattnere53e5082004-06-03 22:57:15 +0000526 href="#t_function">function</a> that takes an <tt>int</tt>, returning
Chris Lattner261efe92003-11-25 01:02:51 +0000527an <tt>int</tt>.</td>
528 </tr>
529 </tbody>
Chris Lattner00950542001-06-06 20:29:01 +0000530</table>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000531
Misha Brukman9d0919f2003-11-08 01:05:38 +0000532</div>
Chris Lattner00950542001-06-06 20:29:01 +0000533<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000534<div class="doc_subsubsection"> <a name="t_pointer">Pointer Type</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000535<div class="doc_text">
Chris Lattner7faa8832002-04-14 06:13:44 +0000536<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000537<p>As in many languages, the pointer type represents a pointer or
538reference to another object, which must live in memory.</p>
Chris Lattner7faa8832002-04-14 06:13:44 +0000539<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000540<pre> &lt;type&gt; *<br></pre>
Chris Lattner7faa8832002-04-14 06:13:44 +0000541<h5>Examples:</h5>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000542
Misha Brukman9d0919f2003-11-08 01:05:38 +0000543<table border="0" cellpadding="0" cellspacing="0">
Chris Lattner261efe92003-11-25 01:02:51 +0000544 <tbody>
545 <tr>
546 <td><tt>[4x int]*</tt></td>
547 <td>: <a href="#t_pointer">pointer</a> to <a href="#t_array">array</a>
548of four <tt>int</tt> values</td>
549 </tr>
550 <tr>
551 <td><tt>int (int *) *</tt></td>
552 <td>: A <a href="#t_pointer">pointer</a> to a <a
Chris Lattnere53e5082004-06-03 22:57:15 +0000553 href="#t_function">function</a> that takes an <tt>int</tt>, returning
Chris Lattner261efe92003-11-25 01:02:51 +0000554an <tt>int</tt>.</td>
555 </tr>
556 </tbody>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000557</table>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000558
Misha Brukman9d0919f2003-11-08 01:05:38 +0000559</div>
Chris Lattner261efe92003-11-25 01:02:51 +0000560<!-- _______________________________________________________________________ --><!--
Misha Brukman9d0919f2003-11-08 01:05:38 +0000561<div class="doc_subsubsection">
562 <a name="t_packed">Packed Type</a>
563</div>
564
565<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +0000566
567Mention/decide that packed types work with saturation or not. Maybe have a packed+saturated type in addition to just a packed type.<p>
568
569Packed types should be 'nonsaturated' because standard data types are not saturated. Maybe have a saturated packed type?<p>
570
Misha Brukman9d0919f2003-11-08 01:05:38 +0000571</div>
572
Chris Lattner261efe92003-11-25 01:02:51 +0000573--><!-- *********************************************************************** -->
574<div class="doc_section"> <a name="highlevel">High Level Structure</a> </div>
575<!-- *********************************************************************** --><!-- ======================================================================= -->
576<div class="doc_subsection"> <a name="modulestructure">Module Structure</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000577<div class="doc_text">
Chris Lattner261efe92003-11-25 01:02:51 +0000578<p>LLVM programs are composed of "Module"s, each of which is a
579translation unit of the input programs. Each module consists of
580functions, global variables, and symbol table entries. Modules may be
581combined together with the LLVM linker, which merges function (and
582global variable) definitions, resolves forward declarations, and merges
583symbol table entries. Here is an example of the "hello world" module:</p>
584<pre><i>; Declare the string constant as a global constant...</i>
585<a href="#identifiers">%.LC0</a> = <a href="#linkage_internal">internal</a> <a
586 href="#globalvars">constant</a> <a href="#t_array">[13 x sbyte]</a> c"hello world\0A\00" <i>; [13 x sbyte]*</i>
Chris Lattner2b7d3202002-05-06 03:03:22 +0000587
Chris Lattner27f71f22003-09-03 00:41:47 +0000588<i>; External declaration of the puts function</i>
589<a href="#functionstructure">declare</a> int %puts(sbyte*) <i>; int(sbyte*)* </i>
Chris Lattner2b7d3202002-05-06 03:03:22 +0000590
591<i>; Definition of main function</i>
Chris Lattner27f71f22003-09-03 00:41:47 +0000592int %main() { <i>; int()* </i>
Chris Lattner2b7d3202002-05-06 03:03:22 +0000593 <i>; Convert [13x sbyte]* to sbyte *...</i>
Chris Lattner261efe92003-11-25 01:02:51 +0000594 %cast210 = <a
595 href="#i_getelementptr">getelementptr</a> [13 x sbyte]* %.LC0, long 0, long 0 <i>; sbyte*</i>
Chris Lattner2b7d3202002-05-06 03:03:22 +0000596
597 <i>; Call puts function to write out the string to stdout...</i>
Chris Lattner261efe92003-11-25 01:02:51 +0000598 <a
599 href="#i_call">call</a> int %puts(sbyte* %cast210) <i>; int</i>
600 <a
601 href="#i_ret">ret</a> int 0<br>}<br></pre>
602<p>This example is made up of a <a href="#globalvars">global variable</a>
603named "<tt>.LC0</tt>", an external declaration of the "<tt>puts</tt>"
604function, and a <a href="#functionstructure">function definition</a>
605for "<tt>main</tt>".</p>
606<a name="linkage"> In general, a module is made up of a list of global
607values, where both functions and global variables are global values.
608Global values are represented by a pointer to a memory location (in
609this case, a pointer to an array of char, and a pointer to a function),
610and have one of the following linkage types:</a>
611<p> </p>
Chris Lattner27f71f22003-09-03 00:41:47 +0000612<dl>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000613 <dt><tt><b><a name="linkage_internal">internal</a></b></tt> </dt>
Chris Lattner261efe92003-11-25 01:02:51 +0000614 <dd>Global values with internal linkage are only directly accessible
615by objects in the current module. In particular, linking code into a
616module with an internal global value may cause the internal to be
617renamed as necessary to avoid collisions. Because the symbol is
618internal to the module, all references can be updated. This
619corresponds to the notion of the '<tt>static</tt>' keyword in C, or the
620idea of "anonymous namespaces" in C++.
621 <p> </p>
622 </dd>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000623 <dt><tt><b><a name="linkage_linkonce">linkonce</a></b></tt>: </dt>
Chris Lattner261efe92003-11-25 01:02:51 +0000624 <dd>"<tt>linkonce</tt>" linkage is similar to <tt>internal</tt>
625linkage, with the twist that linking together two modules defining the
626same <tt>linkonce</tt> globals will cause one of the globals to be
627discarded. This is typically used to implement inline functions.
628Unreferenced <tt>linkonce</tt> globals are allowed to be discarded.
629 <p> </p>
630 </dd>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000631 <dt><tt><b><a name="linkage_weak">weak</a></b></tt>: </dt>
Chris Lattner261efe92003-11-25 01:02:51 +0000632 <dd>"<tt>weak</tt>" linkage is exactly the same as <tt>linkonce</tt>
633linkage, except that unreferenced <tt>weak</tt> globals may not be
634discarded. This is used to implement constructs in C such as "<tt>int
635X;</tt>" at global scope.
636 <p> </p>
637 </dd>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000638 <dt><tt><b><a name="linkage_appending">appending</a></b></tt>: </dt>
Chris Lattner261efe92003-11-25 01:02:51 +0000639 <dd>"<tt>appending</tt>" linkage may only be applied to global
640variables of pointer to array type. When two global variables with
641appending linkage are linked together, the two global arrays are
642appended together. This is the LLVM, typesafe, equivalent of having
643the system linker append together "sections" with identical names when
644.o files are linked.
645 <p> </p>
646 </dd>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000647 <dt><tt><b><a name="linkage_external">externally visible</a></b></tt>:</dt>
Chris Lattner261efe92003-11-25 01:02:51 +0000648 <dd>If none of the above identifiers are used, the global is
649externally visible, meaning that it participates in linkage and can be
650used to resolve external symbol references.
651 <p> </p>
652 </dd>
Chris Lattner261efe92003-11-25 01:02:51 +0000653</dl>
654<p> </p>
655<p><a name="linkage_external">For example, since the "<tt>.LC0</tt>"
656variable is defined to be internal, if another module defined a "<tt>.LC0</tt>"
657variable and was linked with this one, one of the two would be renamed,
658preventing a collision. Since "<tt>main</tt>" and "<tt>puts</tt>" are
659external (i.e., lacking any linkage declarations), they are accessible
660outside of the current module. It is illegal for a function <i>declaration</i>
661to have any linkage type other than "externally visible".</a></p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000662</div>
Chris Lattner9ee5d222004-03-08 16:49:10 +0000663
Chris Lattner00950542001-06-06 20:29:01 +0000664<!-- ======================================================================= -->
Chris Lattner9ee5d222004-03-08 16:49:10 +0000665<div class="doc_subsection">
666 <a name="globalvars">Global Variables</a>
667</div>
668
Misha Brukman9d0919f2003-11-08 01:05:38 +0000669<div class="doc_text">
Chris Lattner9ee5d222004-03-08 16:49:10 +0000670
Chris Lattner261efe92003-11-25 01:02:51 +0000671<p>Global variables define regions of memory allocated at compilation
672time instead of run-time. Global variables may optionally be
673initialized. A variable may be defined as a global "constant", which
674indicates that the contents of the variable will never be modified
Chris Lattner9ee5d222004-03-08 16:49:10 +0000675(opening options for optimization).</p>
676
Chris Lattner261efe92003-11-25 01:02:51 +0000677<p>As SSA values, global variables define pointer values that are in
678scope (i.e. they dominate) for all basic blocks in the program. Global
679variables always define a pointer to their "content" type because they
680describe a region of memory, and all memory objects in LLVM are
681accessed through pointers.</p>
Chris Lattner9ee5d222004-03-08 16:49:10 +0000682
Misha Brukman9d0919f2003-11-08 01:05:38 +0000683</div>
Chris Lattner9ee5d222004-03-08 16:49:10 +0000684
685
Chris Lattner2b7d3202002-05-06 03:03:22 +0000686<!-- ======================================================================= -->
Chris Lattner9ee5d222004-03-08 16:49:10 +0000687<div class="doc_subsection">
688 <a name="functionstructure">Functions</a>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000689</div>
Chris Lattner9ee5d222004-03-08 16:49:10 +0000690
691<div class="doc_text">
692
693<p>LLVM function definitions are composed of a (possibly empty) argument list,
694an opening curly brace, a list of basic blocks, and a closing curly brace. LLVM
695function declarations are defined with the "<tt>declare</tt>" keyword, a
696function name, and a function signature.</p>
697
698<p>A function definition contains a list of basic blocks, forming the CFG for
699the function. Each basic block may optionally start with a label (giving the
700basic block a symbol table entry), contains a list of instructions, and ends
701with a <a href="#terminators">terminator</a> instruction (such as a branch or
702function return).</p>
703
704<p>The first basic block in program is special in two ways: it is immediately
705executed on entrance to the function, and it is not allowed to have predecessor
706basic blocks (i.e. there can not be any branches to the entry block of a
707function). Because the block can have no predecessors, it also cannot have any
708<a href="#i_phi">PHI nodes</a>.</p>
709
710<p>LLVM functions are identified by their name and type signature. Hence, two
711functions with the same name but different parameter lists or return values are
712considered different functions, and LLVM will resolves references to each
713appropriately.</p>
714
715</div>
716
717
Chris Lattner00950542001-06-06 20:29:01 +0000718<!-- *********************************************************************** -->
Chris Lattner261efe92003-11-25 01:02:51 +0000719<div class="doc_section"> <a name="instref">Instruction Reference</a> </div>
720<!-- *********************************************************************** -->
Misha Brukman9d0919f2003-11-08 01:05:38 +0000721<div class="doc_text">
Chris Lattner261efe92003-11-25 01:02:51 +0000722<p>The LLVM instruction set consists of several different
723classifications of instructions: <a href="#terminators">terminator
724instructions</a>, <a href="#binaryops">binary instructions</a>, <a
725 href="#memoryops">memory instructions</a>, and <a href="#otherops">other
726instructions</a>.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000727</div>
Chris Lattner00950542001-06-06 20:29:01 +0000728<!-- ======================================================================= -->
Chris Lattner261efe92003-11-25 01:02:51 +0000729<div class="doc_subsection"> <a name="terminators">Terminator
730Instructions</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000731<div class="doc_text">
Chris Lattner261efe92003-11-25 01:02:51 +0000732<p>As mentioned <a href="#functionstructure">previously</a>, every
733basic block in a program ends with a "Terminator" instruction, which
734indicates which block should be executed after the current block is
735finished. These terminator instructions typically yield a '<tt>void</tt>'
736value: they produce control flow, not values (the one exception being
737the '<a href="#i_invoke"><tt>invoke</tt></a>' instruction).</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000738<p>There are five different terminator instructions: the '<a
Chris Lattner261efe92003-11-25 01:02:51 +0000739 href="#i_ret"><tt>ret</tt></a>' instruction, the '<a href="#i_br"><tt>br</tt></a>'
740instruction, the '<a href="#i_switch"><tt>switch</tt></a>' instruction,
741the '<a href="#i_invoke"><tt>invoke</tt></a>' instruction, and the '<a
742 href="#i_unwind"><tt>unwind</tt></a>' instruction.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000743</div>
Chris Lattner00950542001-06-06 20:29:01 +0000744<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000745<div class="doc_subsubsection"> <a name="i_ret">'<tt>ret</tt>'
746Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000747<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +0000748<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000749<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 +0000750 ret void <i>; Return from void function</i>
Chris Lattner00950542001-06-06 20:29:01 +0000751</pre>
Chris Lattner00950542001-06-06 20:29:01 +0000752<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000753<p>The '<tt>ret</tt>' instruction is used to return control flow (and a
754value) from a function, back to the caller.</p>
John Criswell4457dc92004-04-09 16:48:45 +0000755<p>There are two forms of the '<tt>ret</tt>' instruction: one that
Chris Lattner261efe92003-11-25 01:02:51 +0000756returns a value and then causes control flow, and one that just causes
757control flow to occur.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000758<h5>Arguments:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000759<p>The '<tt>ret</tt>' instruction may return any '<a
760 href="#t_firstclass">first class</a>' type. Notice that a function is
761not <a href="#wellformed">well formed</a> if there exists a '<tt>ret</tt>'
762instruction inside of the function that returns a value that does not
763match the return type of the function.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000764<h5>Semantics:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000765<p>When the '<tt>ret</tt>' instruction is executed, control flow
766returns back to the calling function's context. If the caller is a "<a
John Criswellfa081872004-06-25 15:16:57 +0000767 href="#i_call"><tt>call</tt></a>" instruction, execution continues at
Chris Lattner261efe92003-11-25 01:02:51 +0000768the instruction after the call. If the caller was an "<a
769 href="#i_invoke"><tt>invoke</tt></a>" instruction, execution continues
770at the beginning "normal" of the destination block. If the instruction
771returns a value, that value shall set the call or invoke instruction's
772return value.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000773<h5>Example:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000774<pre> ret int 5 <i>; Return an integer value of 5</i>
Chris Lattner7faa8832002-04-14 06:13:44 +0000775 ret void <i>; Return from a void function</i>
Chris Lattner00950542001-06-06 20:29:01 +0000776</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000777</div>
Chris Lattner00950542001-06-06 20:29:01 +0000778<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000779<div class="doc_subsubsection"> <a name="i_br">'<tt>br</tt>' Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000780<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +0000781<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000782<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 +0000783</pre>
Chris Lattner00950542001-06-06 20:29:01 +0000784<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000785<p>The '<tt>br</tt>' instruction is used to cause control flow to
786transfer to a different basic block in the current function. There are
787two forms of this instruction, corresponding to a conditional branch
788and an unconditional branch.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000789<h5>Arguments:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000790<p>The conditional branch form of the '<tt>br</tt>' instruction takes a
791single '<tt>bool</tt>' value and two '<tt>label</tt>' values. The
792unconditional form of the '<tt>br</tt>' instruction takes a single '<tt>label</tt>'
793value as a target.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000794<h5>Semantics:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000795<p>Upon execution of a conditional '<tt>br</tt>' instruction, the '<tt>bool</tt>'
796argument is evaluated. If the value is <tt>true</tt>, control flows
797to the '<tt>iftrue</tt>' <tt>label</tt> argument. If "cond" is <tt>false</tt>,
798control flows to the '<tt>iffalse</tt>' <tt>label</tt> argument.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000799<h5>Example:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000800<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
801 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 +0000802</div>
Chris Lattner00950542001-06-06 20:29:01 +0000803<!-- _______________________________________________________________________ -->
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000804<div class="doc_subsubsection">
805 <a name="i_switch">'<tt>switch</tt>' Instruction</a>
806</div>
807
Misha Brukman9d0919f2003-11-08 01:05:38 +0000808<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +0000809<h5>Syntax:</h5>
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000810
811<pre>
812 switch &lt;intty&gt; &lt;value&gt;, label &lt;defaultdest&gt; [ &lt;intty&gt; &lt;val&gt;, label &lt;dest&gt; ... ]
813</pre>
814
Chris Lattner00950542001-06-06 20:29:01 +0000815<h5>Overview:</h5>
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000816
817<p>The '<tt>switch</tt>' instruction is used to transfer control flow to one of
818several different places. It is a generalization of the '<tt>br</tt>'
Misha Brukman9d0919f2003-11-08 01:05:38 +0000819instruction, allowing a branch to occur to one of many possible
820destinations.</p>
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000821
822
Chris Lattner00950542001-06-06 20:29:01 +0000823<h5>Arguments:</h5>
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000824
825<p>The '<tt>switch</tt>' instruction uses three parameters: an integer
826comparison value '<tt>value</tt>', a default '<tt>label</tt>' destination, and
827an array of pairs of comparison value constants and '<tt>label</tt>'s. The
828table is not allowed to contain duplicate constant entries.</p>
829
Chris Lattner00950542001-06-06 20:29:01 +0000830<h5>Semantics:</h5>
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000831
Chris Lattner261efe92003-11-25 01:02:51 +0000832<p>The <tt>switch</tt> instruction specifies a table of values and
833destinations. When the '<tt>switch</tt>' instruction is executed, this
John Criswell84114752004-06-25 16:05:06 +0000834table is searched for the given value. If the value is found, control flow is
835transfered to the corresponding destination; otherwise, control flow is
836transfered to the default destination.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000837
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000838<h5>Implementation:</h5>
839
840<p>Depending on properties of the target machine and the particular
841<tt>switch</tt> instruction, this instruction may be code generated in different
John Criswell84114752004-06-25 16:05:06 +0000842ways. For example, it could be generated as a series of chained conditional
843branches or with a lookup table.</p>
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000844
845<h5>Example:</h5>
846
847<pre>
848 <i>; Emulate a conditional br instruction</i>
849 %Val = <a href="#i_cast">cast</a> bool %value to int
850 switch int %Val, label %truedest [int 0, label %falsedest ]
851
852 <i>; Emulate an unconditional br instruction</i>
853 switch uint 0, label %dest [ ]
854
855 <i>; Implement a jump table:</i>
856 switch uint %val, label %otherwise [ uint 0, label %onzero
857 uint 1, label %onone
858 uint 2, label %ontwo ]
Chris Lattner00950542001-06-06 20:29:01 +0000859</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000860</div>
Chris Lattner00950542001-06-06 20:29:01 +0000861<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000862<div class="doc_subsubsection"> <a name="i_invoke">'<tt>invoke</tt>'
863Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000864<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +0000865<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000866<pre> &lt;result&gt; = invoke &lt;ptr to function ty&gt; %&lt;function ptr val&gt;(&lt;function args&gt;)<br> to label &lt;normal label&gt; except label &lt;exception label&gt;<br></pre>
Chris Lattner6536cfe2002-05-06 22:08:29 +0000867<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000868<p>The '<tt>invoke</tt>' instruction causes control to transfer to a
869specified function, with the possibility of control flow transfer to
870either the '<tt>normal</tt>' <tt>label</tt> label or the '<tt>exception</tt>'<tt>label</tt>.
871If the callee function returns with the "<tt><a href="#i_ret">ret</a></tt>"
872instruction, control flow will return to the "normal" label. If the
873callee (or any indirect callees) returns with the "<a href="#i_unwind"><tt>unwind</tt></a>"
874instruction, control is interrupted, and continued at the dynamically
875nearest "except" label.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000876<h5>Arguments:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000877<p>This instruction requires several arguments:</p>
Chris Lattner00950542001-06-06 20:29:01 +0000878<ol>
Chris Lattner261efe92003-11-25 01:02:51 +0000879 <li>'<tt>ptr to function ty</tt>': shall be the signature of the
880pointer to function value being invoked. In most cases, this is a
881direct function invocation, but indirect <tt>invoke</tt>s are just as
882possible, branching off an arbitrary pointer to function value. </li>
883 <li>'<tt>function ptr val</tt>': An LLVM value containing a pointer
884to a function to be invoked. </li>
885 <li>'<tt>function args</tt>': argument list whose types match the
886function signature argument types. If the function signature indicates
887the function accepts a variable number of arguments, the extra
888arguments can be specified. </li>
889 <li>'<tt>normal label</tt>': the label reached when the called
890function executes a '<tt><a href="#i_ret">ret</a></tt>' instruction. </li>
891 <li>'<tt>exception label</tt>': the label reached when a callee
892returns with the <a href="#i_unwind"><tt>unwind</tt></a> instruction. </li>
Chris Lattner00950542001-06-06 20:29:01 +0000893</ol>
Chris Lattner00950542001-06-06 20:29:01 +0000894<h5>Semantics:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000895<p>This instruction is designed to operate as a standard '<tt><a
Chris Lattner261efe92003-11-25 01:02:51 +0000896 href="#i_call">call</a></tt>' instruction in most regards. The
897primary difference is that it establishes an association with a label,
898which is used by the runtime library to unwind the stack.</p>
899<p>This instruction is used in languages with destructors to ensure
900that proper cleanup is performed in the case of either a <tt>longjmp</tt>
901or a thrown exception. Additionally, this is important for
902implementation of '<tt>catch</tt>' clauses in high-level languages that
903support them.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000904<h5>Example:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000905<pre> %retval = invoke int %Test(int 15)<br> to label %Continue<br> except label %TestCleanup <i>; {int}:retval set</i>
Chris Lattner00950542001-06-06 20:29:01 +0000906</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000907</div>
Chris Lattner27f71f22003-09-03 00:41:47 +0000908<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000909<div class="doc_subsubsection"> <a name="i_unwind">'<tt>unwind</tt>'
910Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000911<div class="doc_text">
Chris Lattner27f71f22003-09-03 00:41:47 +0000912<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000913<pre> unwind<br></pre>
Chris Lattner27f71f22003-09-03 00:41:47 +0000914<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000915<p>The '<tt>unwind</tt>' instruction unwinds the stack, continuing
916control flow at the first callee in the dynamic call stack which used
917an <a href="#i_invoke"><tt>invoke</tt></a> instruction to perform the
918call. This is primarily used to implement exception handling.</p>
Chris Lattner27f71f22003-09-03 00:41:47 +0000919<h5>Semantics:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000920<p>The '<tt>unwind</tt>' intrinsic causes execution of the current
921function to immediately halt. The dynamic call stack is then searched
922for the first <a href="#i_invoke"><tt>invoke</tt></a> instruction on
923the call stack. Once found, execution continues at the "exceptional"
924destination block specified by the <tt>invoke</tt> instruction. If
925there is no <tt>invoke</tt> instruction in the dynamic call chain,
926undefined behavior results.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000927</div>
Chris Lattner00950542001-06-06 20:29:01 +0000928<!-- ======================================================================= -->
Chris Lattner261efe92003-11-25 01:02:51 +0000929<div class="doc_subsection"> <a name="binaryops">Binary Operations</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000930<div class="doc_text">
Chris Lattner261efe92003-11-25 01:02:51 +0000931<p>Binary operators are used to do most of the computation in a
932program. They require two operands, execute an operation on them, and
933produce a single value. The result value of a binary operator is not
934necessarily the same type as its operands.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000935<p>There are several different binary operators:</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000936</div>
Chris Lattner00950542001-06-06 20:29:01 +0000937<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000938<div class="doc_subsubsection"> <a name="i_add">'<tt>add</tt>'
939Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000940<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +0000941<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000942<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 +0000943</pre>
Chris Lattner00950542001-06-06 20:29:01 +0000944<h5>Overview:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000945<p>The '<tt>add</tt>' instruction returns the sum of its two operands.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000946<h5>Arguments:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000947<p>The two arguments to the '<tt>add</tt>' instruction must be either <a
Chris Lattner261efe92003-11-25 01:02:51 +0000948 href="#t_integer">integer</a> or <a href="#t_floating">floating point</a>
949values. Both arguments must have identical types.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000950<h5>Semantics:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000951<p>The value produced is the integer or floating point sum of the two
952operands.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000953<h5>Example:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000954<pre> &lt;result&gt; = add int 4, %var <i>; yields {int}:result = 4 + %var</i>
Chris Lattner00950542001-06-06 20:29:01 +0000955</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000956</div>
Chris Lattner00950542001-06-06 20:29:01 +0000957<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000958<div class="doc_subsubsection"> <a name="i_sub">'<tt>sub</tt>'
959Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000960<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +0000961<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000962<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 +0000963</pre>
Chris Lattner00950542001-06-06 20:29:01 +0000964<h5>Overview:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000965<p>The '<tt>sub</tt>' instruction returns the difference of its two
966operands.</p>
Chris Lattner261efe92003-11-25 01:02:51 +0000967<p>Note that the '<tt>sub</tt>' instruction is used to represent the '<tt>neg</tt>'
968instruction present in most other intermediate representations.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000969<h5>Arguments:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000970<p>The two arguments to the '<tt>sub</tt>' instruction must be either <a
Chris Lattner261efe92003-11-25 01:02:51 +0000971 href="#t_integer">integer</a> or <a href="#t_floating">floating point</a>
972values. Both arguments must have identical types.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000973<h5>Semantics:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000974<p>The value produced is the integer or floating point difference of
975the two operands.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000976<h5>Example:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000977<pre> &lt;result&gt; = sub int 4, %var <i>; yields {int}:result = 4 - %var</i>
Chris Lattner00950542001-06-06 20:29:01 +0000978 &lt;result&gt; = sub int 0, %val <i>; yields {int}:result = -%var</i>
979</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000980</div>
Chris Lattner00950542001-06-06 20:29:01 +0000981<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000982<div class="doc_subsubsection"> <a name="i_mul">'<tt>mul</tt>'
983Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000984<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +0000985<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000986<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 +0000987</pre>
Chris Lattner00950542001-06-06 20:29:01 +0000988<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000989<p>The '<tt>mul</tt>' instruction returns the product of its two
990operands.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000991<h5>Arguments:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000992<p>The two arguments to the '<tt>mul</tt>' instruction must be either <a
Chris Lattner261efe92003-11-25 01:02:51 +0000993 href="#t_integer">integer</a> or <a href="#t_floating">floating point</a>
994values. Both arguments must have identical types.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000995<h5>Semantics:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000996<p>The value produced is the integer or floating point product of the
Misha Brukman9d0919f2003-11-08 01:05:38 +0000997two operands.</p>
Chris Lattner261efe92003-11-25 01:02:51 +0000998<p>There is no signed vs unsigned multiplication. The appropriate
999action is taken based on the type of the operand.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001000<h5>Example:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001001<pre> &lt;result&gt; = mul int 4, %var <i>; yields {int}:result = 4 * %var</i>
Chris Lattner00950542001-06-06 20:29:01 +00001002</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001003</div>
Chris Lattner00950542001-06-06 20:29:01 +00001004<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +00001005<div class="doc_subsubsection"> <a name="i_div">'<tt>div</tt>'
1006Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001007<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +00001008<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001009<pre> &lt;result&gt; = div &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
1010</pre>
1011<h5>Overview:</h5>
1012<p>The '<tt>div</tt>' instruction returns the quotient of its two
1013operands.</p>
1014<h5>Arguments:</h5>
1015<p>The two arguments to the '<tt>div</tt>' instruction must be either <a
1016 href="#t_integer">integer</a> or <a href="#t_floating">floating point</a>
1017values. Both arguments must have identical types.</p>
1018<h5>Semantics:</h5>
1019<p>The value produced is the integer or floating point quotient of the
1020two operands.</p>
1021<h5>Example:</h5>
1022<pre> &lt;result&gt; = div int 4, %var <i>; yields {int}:result = 4 / %var</i>
1023</pre>
1024</div>
1025<!-- _______________________________________________________________________ -->
1026<div class="doc_subsubsection"> <a name="i_rem">'<tt>rem</tt>'
1027Instruction</a> </div>
1028<div class="doc_text">
1029<h5>Syntax:</h5>
1030<pre> &lt;result&gt; = rem &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
1031</pre>
1032<h5>Overview:</h5>
1033<p>The '<tt>rem</tt>' instruction returns the remainder from the
1034division of its two operands.</p>
1035<h5>Arguments:</h5>
1036<p>The two arguments to the '<tt>rem</tt>' instruction must be either <a
1037 href="#t_integer">integer</a> or <a href="#t_floating">floating point</a>
1038values. Both arguments must have identical types.</p>
1039<h5>Semantics:</h5>
1040<p>This returns the <i>remainder</i> of a division (where the result
1041has the same sign as the divisor), not the <i>modulus</i> (where the
1042result has the same sign as the dividend) of a value. For more
1043information about the difference, see: <a
1044 href="http://mathforum.org/dr.math/problems/anne.4.28.99.html">The
1045Math Forum</a>.</p>
1046<h5>Example:</h5>
1047<pre> &lt;result&gt; = rem int 4, %var <i>; yields {int}:result = 4 % %var</i>
1048</pre>
1049</div>
1050<!-- _______________________________________________________________________ -->
1051<div class="doc_subsubsection"> <a name="i_setcc">'<tt>set<i>cc</i></tt>'
1052Instructions</a> </div>
1053<div class="doc_text">
1054<h5>Syntax:</h5>
1055<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 +00001056 &lt;result&gt; = setne &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {bool}:result</i>
1057 &lt;result&gt; = setlt &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {bool}:result</i>
1058 &lt;result&gt; = setgt &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {bool}:result</i>
1059 &lt;result&gt; = setle &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {bool}:result</i>
1060 &lt;result&gt; = setge &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {bool}:result</i>
1061</pre>
Chris Lattner261efe92003-11-25 01:02:51 +00001062<h5>Overview:</h5>
1063<p>The '<tt>set<i>cc</i></tt>' family of instructions returns a boolean
1064value based on a comparison of their two operands.</p>
1065<h5>Arguments:</h5>
1066<p>The two arguments to the '<tt>set<i>cc</i></tt>' instructions must
1067be of <a href="#t_firstclass">first class</a> type (it is not possible
1068to compare '<tt>label</tt>'s, '<tt>array</tt>'s, '<tt>structure</tt>'
1069or '<tt>void</tt>' values, etc...). Both arguments must have identical
1070types.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001071<h5>Semantics:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001072<p>The '<tt>seteq</tt>' instruction yields a <tt>true</tt> '<tt>bool</tt>'
1073value if both operands are equal.<br>
1074The '<tt>setne</tt>' instruction yields a <tt>true</tt> '<tt>bool</tt>'
1075value if both operands are unequal.<br>
1076The '<tt>setlt</tt>' instruction yields a <tt>true</tt> '<tt>bool</tt>'
1077value if the first operand is less than the second operand.<br>
1078The '<tt>setgt</tt>' instruction yields a <tt>true</tt> '<tt>bool</tt>'
1079value if the first operand is greater than the second operand.<br>
1080The '<tt>setle</tt>' instruction yields a <tt>true</tt> '<tt>bool</tt>'
1081value if the first operand is less than or equal to the second operand.<br>
1082The '<tt>setge</tt>' instruction yields a <tt>true</tt> '<tt>bool</tt>'
1083value if the first operand is greater than or equal to the second
1084operand.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001085<h5>Example:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001086<pre> &lt;result&gt; = seteq int 4, 5 <i>; yields {bool}:result = false</i>
Chris Lattner00950542001-06-06 20:29:01 +00001087 &lt;result&gt; = setne float 4, 5 <i>; yields {bool}:result = true</i>
1088 &lt;result&gt; = setlt uint 4, 5 <i>; yields {bool}:result = true</i>
1089 &lt;result&gt; = setgt sbyte 4, 5 <i>; yields {bool}:result = false</i>
1090 &lt;result&gt; = setle sbyte 4, 5 <i>; yields {bool}:result = true</i>
1091 &lt;result&gt; = setge sbyte 4, 5 <i>; yields {bool}:result = false</i>
1092</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001093</div>
Chris Lattner00950542001-06-06 20:29:01 +00001094<!-- ======================================================================= -->
Chris Lattner261efe92003-11-25 01:02:51 +00001095<div class="doc_subsection"> <a name="bitwiseops">Bitwise Binary
1096Operations</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001097<div class="doc_text">
Chris Lattner261efe92003-11-25 01:02:51 +00001098<p>Bitwise binary operators are used to do various forms of
1099bit-twiddling in a program. They are generally very efficient
1100instructions, and can commonly be strength reduced from other
1101instructions. They require two operands, execute an operation on them,
1102and produce a single value. The resulting value of the bitwise binary
1103operators is always the same type as its first operand.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001104</div>
Chris Lattner00950542001-06-06 20:29:01 +00001105<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +00001106<div class="doc_subsubsection"> <a name="i_and">'<tt>and</tt>'
1107Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001108<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +00001109<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001110<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 +00001111</pre>
Chris Lattner00950542001-06-06 20:29:01 +00001112<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001113<p>The '<tt>and</tt>' instruction returns the bitwise logical and of
1114its two operands.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001115<h5>Arguments:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001116<p>The two arguments to the '<tt>and</tt>' instruction must be <a
Chris Lattner261efe92003-11-25 01:02:51 +00001117 href="#t_integral">integral</a> values. Both arguments must have
1118identical types.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001119<h5>Semantics:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001120<p>The truth table used for the '<tt>and</tt>' instruction is:</p>
Chris Lattner261efe92003-11-25 01:02:51 +00001121<p> </p>
Misha Brukmandaa4cb02004-03-01 17:47:27 +00001122<div style="align: center">
Misha Brukman9d0919f2003-11-08 01:05:38 +00001123<table border="1" cellspacing="0" cellpadding="4">
Chris Lattner261efe92003-11-25 01:02:51 +00001124 <tbody>
1125 <tr>
1126 <td>In0</td>
1127 <td>In1</td>
1128 <td>Out</td>
1129 </tr>
1130 <tr>
1131 <td>0</td>
1132 <td>0</td>
1133 <td>0</td>
1134 </tr>
1135 <tr>
1136 <td>0</td>
1137 <td>1</td>
1138 <td>0</td>
1139 </tr>
1140 <tr>
1141 <td>1</td>
1142 <td>0</td>
1143 <td>0</td>
1144 </tr>
1145 <tr>
1146 <td>1</td>
1147 <td>1</td>
1148 <td>1</td>
1149 </tr>
1150 </tbody>
1151</table>
Misha Brukmandaa4cb02004-03-01 17:47:27 +00001152</div>
Chris Lattner00950542001-06-06 20:29:01 +00001153<h5>Example:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001154<pre> &lt;result&gt; = and int 4, %var <i>; yields {int}:result = 4 &amp; %var</i>
Chris Lattner00950542001-06-06 20:29:01 +00001155 &lt;result&gt; = and int 15, 40 <i>; yields {int}:result = 8</i>
1156 &lt;result&gt; = and int 4, 8 <i>; yields {int}:result = 0</i>
1157</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001158</div>
Chris Lattner00950542001-06-06 20:29:01 +00001159<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +00001160<div class="doc_subsubsection"> <a name="i_or">'<tt>or</tt>' Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001161<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +00001162<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001163<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 +00001164</pre>
Chris Lattner261efe92003-11-25 01:02:51 +00001165<h5>Overview:</h5>
1166<p>The '<tt>or</tt>' instruction returns the bitwise logical inclusive
1167or of its two operands.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001168<h5>Arguments:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001169<p>The two arguments to the '<tt>or</tt>' instruction must be <a
Chris Lattner261efe92003-11-25 01:02:51 +00001170 href="#t_integral">integral</a> values. Both arguments must have
1171identical types.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001172<h5>Semantics:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001173<p>The truth table used for the '<tt>or</tt>' instruction is:</p>
Chris Lattner261efe92003-11-25 01:02:51 +00001174<p> </p>
Misha Brukmandaa4cb02004-03-01 17:47:27 +00001175<div style="align: center">
Chris Lattner261efe92003-11-25 01:02:51 +00001176<table border="1" cellspacing="0" cellpadding="4">
1177 <tbody>
1178 <tr>
1179 <td>In0</td>
1180 <td>In1</td>
1181 <td>Out</td>
1182 </tr>
1183 <tr>
1184 <td>0</td>
1185 <td>0</td>
1186 <td>0</td>
1187 </tr>
1188 <tr>
1189 <td>0</td>
1190 <td>1</td>
1191 <td>1</td>
1192 </tr>
1193 <tr>
1194 <td>1</td>
1195 <td>0</td>
1196 <td>1</td>
1197 </tr>
1198 <tr>
1199 <td>1</td>
1200 <td>1</td>
1201 <td>1</td>
1202 </tr>
1203 </tbody>
1204</table>
Misha Brukmandaa4cb02004-03-01 17:47:27 +00001205</div>
Chris Lattner00950542001-06-06 20:29:01 +00001206<h5>Example:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001207<pre> &lt;result&gt; = or int 4, %var <i>; yields {int}:result = 4 | %var</i>
Chris Lattner00950542001-06-06 20:29:01 +00001208 &lt;result&gt; = or int 15, 40 <i>; yields {int}:result = 47</i>
1209 &lt;result&gt; = or int 4, 8 <i>; yields {int}:result = 12</i>
1210</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001211</div>
Chris Lattner00950542001-06-06 20:29:01 +00001212<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +00001213<div class="doc_subsubsection"> <a name="i_xor">'<tt>xor</tt>'
1214Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001215<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +00001216<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001217<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 +00001218</pre>
Chris Lattner00950542001-06-06 20:29:01 +00001219<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001220<p>The '<tt>xor</tt>' instruction returns the bitwise logical exclusive
1221or of its two operands. The <tt>xor</tt> is used to implement the
1222"one's complement" operation, which is the "~" operator in C.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001223<h5>Arguments:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001224<p>The two arguments to the '<tt>xor</tt>' instruction must be <a
Chris Lattner261efe92003-11-25 01:02:51 +00001225 href="#t_integral">integral</a> values. Both arguments must have
1226identical types.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001227<h5>Semantics:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001228<p>The truth table used for the '<tt>xor</tt>' instruction is:</p>
Chris Lattner261efe92003-11-25 01:02:51 +00001229<p> </p>
Misha Brukmandaa4cb02004-03-01 17:47:27 +00001230<div style="align: center">
Chris Lattner261efe92003-11-25 01:02:51 +00001231<table border="1" cellspacing="0" cellpadding="4">
1232 <tbody>
1233 <tr>
1234 <td>In0</td>
1235 <td>In1</td>
1236 <td>Out</td>
1237 </tr>
1238 <tr>
1239 <td>0</td>
1240 <td>0</td>
1241 <td>0</td>
1242 </tr>
1243 <tr>
1244 <td>0</td>
1245 <td>1</td>
1246 <td>1</td>
1247 </tr>
1248 <tr>
1249 <td>1</td>
1250 <td>0</td>
1251 <td>1</td>
1252 </tr>
1253 <tr>
1254 <td>1</td>
1255 <td>1</td>
1256 <td>0</td>
1257 </tr>
1258 </tbody>
1259</table>
Misha Brukmandaa4cb02004-03-01 17:47:27 +00001260</div>
Chris Lattner261efe92003-11-25 01:02:51 +00001261<p> </p>
Chris Lattner00950542001-06-06 20:29:01 +00001262<h5>Example:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001263<pre> &lt;result&gt; = xor int 4, %var <i>; yields {int}:result = 4 ^ %var</i>
Chris Lattner00950542001-06-06 20:29:01 +00001264 &lt;result&gt; = xor int 15, 40 <i>; yields {int}:result = 39</i>
1265 &lt;result&gt; = xor int 4, 8 <i>; yields {int}:result = 12</i>
Chris Lattner27f71f22003-09-03 00:41:47 +00001266 &lt;result&gt; = xor int %V, -1 <i>; yields {int}:result = ~%V</i>
Chris Lattner00950542001-06-06 20:29:01 +00001267</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001268</div>
Chris Lattner00950542001-06-06 20:29:01 +00001269<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +00001270<div class="doc_subsubsection"> <a name="i_shl">'<tt>shl</tt>'
1271Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001272<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +00001273<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001274<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 +00001275</pre>
Chris Lattner00950542001-06-06 20:29:01 +00001276<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001277<p>The '<tt>shl</tt>' instruction returns the first operand shifted to
1278the left a specified number of bits.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001279<h5>Arguments:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001280<p>The first argument to the '<tt>shl</tt>' instruction must be an <a
Chris Lattner261efe92003-11-25 01:02:51 +00001281 href="#t_integer">integer</a> type. The second argument must be an '<tt>ubyte</tt>'
1282type.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001283<h5>Semantics:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001284<p>The value produced is <tt>var1</tt> * 2<sup><tt>var2</tt></sup>.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001285<h5>Example:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001286<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 +00001287 &lt;result&gt; = shl int 4, ubyte 2 <i>; yields {int}:result = 16</i>
1288 &lt;result&gt; = shl int 1, ubyte 10 <i>; yields {int}:result = 1024</i>
1289</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001290</div>
Chris Lattner00950542001-06-06 20:29:01 +00001291<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +00001292<div class="doc_subsubsection"> <a name="i_shr">'<tt>shr</tt>'
1293Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001294<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +00001295<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001296<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 +00001297</pre>
Chris Lattner00950542001-06-06 20:29:01 +00001298<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001299<p>The '<tt>shr</tt>' instruction returns the first operand shifted to
1300the right a specified number of bits.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001301<h5>Arguments:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001302<p>The first argument to the '<tt>shr</tt>' instruction must be an <a
Chris Lattner261efe92003-11-25 01:02:51 +00001303 href="#t_integer">integer</a> type. The second argument must be an '<tt>ubyte</tt>'
1304type.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001305<h5>Semantics:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001306<p>If the first argument is a <a href="#t_signed">signed</a> type, the
1307most significant bit is duplicated in the newly free'd bit positions.
1308If the first argument is unsigned, zero bits shall fill the empty
1309positions.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001310<h5>Example:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001311<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 +00001312 &lt;result&gt; = shr uint 4, ubyte 1 <i>; yields {uint}:result = 2</i>
Chris Lattner00950542001-06-06 20:29:01 +00001313 &lt;result&gt; = shr int 4, ubyte 2 <i>; yields {int}:result = 1</i>
Chris Lattner8c6bb902003-06-18 21:30:51 +00001314 &lt;result&gt; = shr sbyte 4, ubyte 3 <i>; yields {sbyte}:result = 0</i>
1315 &lt;result&gt; = shr sbyte -2, ubyte 1 <i>; yields {sbyte}:result = -1</i>
Chris Lattner00950542001-06-06 20:29:01 +00001316</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001317</div>
Chris Lattner00950542001-06-06 20:29:01 +00001318<!-- ======================================================================= -->
Chris Lattner261efe92003-11-25 01:02:51 +00001319<div class="doc_subsection"> <a name="memoryops">Memory Access
1320Operations</a></div>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001321<div class="doc_text">
Chris Lattner261efe92003-11-25 01:02:51 +00001322<p>A key design point of an SSA-based representation is how it
1323represents memory. In LLVM, no memory locations are in SSA form, which
1324makes things very simple. This section describes how to read, write,
1325allocate and free memory in LLVM.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001326</div>
Chris Lattner00950542001-06-06 20:29:01 +00001327<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +00001328<div class="doc_subsubsection"> <a name="i_malloc">'<tt>malloc</tt>'
1329Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001330<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +00001331<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001332<pre> &lt;result&gt; = malloc &lt;type&gt;, uint &lt;NumElements&gt; <i>; yields {type*}:result</i>
Chris Lattner7faa8832002-04-14 06:13:44 +00001333 &lt;result&gt; = malloc &lt;type&gt; <i>; yields {type*}:result</i>
Chris Lattner00950542001-06-06 20:29:01 +00001334</pre>
Chris Lattner00950542001-06-06 20:29:01 +00001335<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001336<p>The '<tt>malloc</tt>' instruction allocates memory from the system
1337heap and returns a pointer to it.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001338<h5>Arguments:</h5>
John Criswell6e4ca612004-02-24 16:13:56 +00001339<p>The '<tt>malloc</tt>' instruction allocates <tt>sizeof(&lt;type&gt;)*NumElements</tt>
1340bytes of memory from the operating system and returns a pointer of the
Chris Lattner261efe92003-11-25 01:02:51 +00001341appropriate type to the program. The second form of the instruction is
1342a shorter version of the first instruction that defaults to allocating
1343one element.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001344<p>'<tt>type</tt>' must be a sized type.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001345<h5>Semantics:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001346<p>Memory is allocated using the system "<tt>malloc</tt>" function, and
1347a pointer is returned.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001348<h5>Example:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001349<pre> %array = malloc [4 x ubyte ] <i>; yields {[%4 x ubyte]*}:array</i>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001350
Chris Lattner261efe92003-11-25 01:02:51 +00001351 %size = <a
1352 href="#i_add">add</a> uint 2, 2 <i>; yields {uint}:size = uint 4</i>
Chris Lattner7faa8832002-04-14 06:13:44 +00001353 %array1 = malloc ubyte, uint 4 <i>; yields {ubyte*}:array1</i>
1354 %array2 = malloc [12 x ubyte], uint %size <i>; yields {[12 x ubyte]*}:array2</i>
Chris Lattner00950542001-06-06 20:29:01 +00001355</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001356</div>
Chris Lattner00950542001-06-06 20:29:01 +00001357<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +00001358<div class="doc_subsubsection"> <a name="i_free">'<tt>free</tt>'
1359Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001360<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +00001361<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001362<pre> free &lt;type&gt; &lt;value&gt; <i>; yields {void}</i>
Chris Lattner00950542001-06-06 20:29:01 +00001363</pre>
Chris Lattner00950542001-06-06 20:29:01 +00001364<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001365<p>The '<tt>free</tt>' instruction returns memory back to the unused
1366memory heap, to be reallocated in the future.</p>
1367<p> </p>
Chris Lattner00950542001-06-06 20:29:01 +00001368<h5>Arguments:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001369<p>'<tt>value</tt>' shall be a pointer value that points to a value
1370that was allocated with the '<tt><a href="#i_malloc">malloc</a></tt>'
1371instruction.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001372<h5>Semantics:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001373<p>Access to the memory pointed to by the pointer is not longer defined
1374after this instruction executes.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001375<h5>Example:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001376<pre> %array = <a href="#i_malloc">malloc</a> [4 x ubyte] <i>; yields {[4 x ubyte]*}:array</i>
Chris Lattner00950542001-06-06 20:29:01 +00001377 free [4 x ubyte]* %array
1378</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001379</div>
Chris Lattner00950542001-06-06 20:29:01 +00001380<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +00001381<div class="doc_subsubsection"> <a name="i_alloca">'<tt>alloca</tt>'
1382Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001383<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +00001384<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001385<pre> &lt;result&gt; = alloca &lt;type&gt;, uint &lt;NumElements&gt; <i>; yields {type*}:result</i>
Chris Lattner7faa8832002-04-14 06:13:44 +00001386 &lt;result&gt; = alloca &lt;type&gt; <i>; yields {type*}:result</i>
Chris Lattner00950542001-06-06 20:29:01 +00001387</pre>
Chris Lattner00950542001-06-06 20:29:01 +00001388<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001389<p>The '<tt>alloca</tt>' instruction allocates memory on the current
1390stack frame of the procedure that is live until the current function
1391returns to its caller.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001392<h5>Arguments:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001393<p>The the '<tt>alloca</tt>' instruction allocates <tt>sizeof(&lt;type&gt;)*NumElements</tt>
1394bytes of memory on the runtime stack, returning a pointer of the
1395appropriate type to the program. The second form of the instruction is
1396a shorter version of the first that defaults to allocating one element.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001397<p>'<tt>type</tt>' may be any sized type.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001398<h5>Semantics:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001399<p>Memory is allocated, a pointer is returned. '<tt>alloca</tt>'d
1400memory is automatically released when the function returns. The '<tt>alloca</tt>'
1401instruction is commonly used to represent automatic variables that must
1402have an address available. When the function returns (either with the <tt><a
1403 href="#i_ret">ret</a></tt> or <tt><a href="#i_invoke">invoke</a></tt>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001404instructions), the memory is reclaimed.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001405<h5>Example:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001406<pre> %ptr = alloca int <i>; yields {int*}:ptr</i>
Chris Lattner7faa8832002-04-14 06:13:44 +00001407 %ptr = alloca int, uint 4 <i>; yields {int*}:ptr</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_load">'<tt>load</tt>'
1412Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001413<div class="doc_text">
Chris Lattner2b7d3202002-05-06 03:03:22 +00001414<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001415<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 +00001416<h5>Overview:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001417<p>The '<tt>load</tt>' instruction is used to read from memory.</p>
Chris Lattner2b7d3202002-05-06 03:03:22 +00001418<h5>Arguments:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001419<p>The argument to the '<tt>load</tt>' instruction specifies the memory
1420address to load from. The pointer must point to a <a
Chris Lattnere53e5082004-06-03 22:57:15 +00001421 href="#t_firstclass">first class</a> type. If the <tt>load</tt> is
Chris Lattner261efe92003-11-25 01:02:51 +00001422marked as <tt>volatile</tt> then the optimizer is not allowed to modify
1423the number or order of execution of this <tt>load</tt> with other
1424volatile <tt>load</tt> and <tt><a href="#i_store">store</a></tt>
1425instructions. </p>
Chris Lattner2b7d3202002-05-06 03:03:22 +00001426<h5>Semantics:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001427<p>The location of memory pointed to is loaded.</p>
Chris Lattner2b7d3202002-05-06 03:03:22 +00001428<h5>Examples:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001429<pre> %ptr = <a href="#i_alloca">alloca</a> int <i>; yields {int*}:ptr</i>
1430 <a
1431 href="#i_store">store</a> int 3, int* %ptr <i>; yields {void}</i>
Chris Lattner2b7d3202002-05-06 03:03:22 +00001432 %val = load int* %ptr <i>; yields {int}:val = int 3</i>
1433</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001434</div>
Chris Lattner2b7d3202002-05-06 03:03:22 +00001435<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +00001436<div class="doc_subsubsection"> <a name="i_store">'<tt>store</tt>'
1437Instruction</a> </div>
Chris Lattner2b7d3202002-05-06 03:03:22 +00001438<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001439<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 +00001440 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 +00001441</pre>
Chris Lattner2b7d3202002-05-06 03:03:22 +00001442<h5>Overview:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001443<p>The '<tt>store</tt>' instruction is used to write to memory.</p>
Chris Lattner2b7d3202002-05-06 03:03:22 +00001444<h5>Arguments:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001445<p>There are two arguments to the '<tt>store</tt>' instruction: a value
1446to store and an address to store it into. The type of the '<tt>&lt;pointer&gt;</tt>'
1447operand must be a pointer to the type of the '<tt>&lt;value&gt;</tt>'
1448operand. If the <tt>store</tt> is marked as <tt>volatile</tt> then the
1449optimizer is not allowed to modify the number or order of execution of
1450this <tt>store</tt> with other volatile <tt>load</tt> and <tt><a
1451 href="#i_store">store</a></tt> instructions.</p>
1452<h5>Semantics:</h5>
1453<p>The contents of memory are updated to contain '<tt>&lt;value&gt;</tt>'
1454at the location specified by the '<tt>&lt;pointer&gt;</tt>' operand.</p>
Chris Lattner2b7d3202002-05-06 03:03:22 +00001455<h5>Example:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001456<pre> %ptr = <a href="#i_alloca">alloca</a> int <i>; yields {int*}:ptr</i>
1457 <a
1458 href="#i_store">store</a> int 3, int* %ptr <i>; yields {void}</i>
Chris Lattner2b7d3202002-05-06 03:03:22 +00001459 %val = load int* %ptr <i>; yields {int}:val = int 3</i>
1460</pre>
Chris Lattner2b7d3202002-05-06 03:03:22 +00001461<!-- _______________________________________________________________________ -->
Chris Lattnerf74d5c72004-04-05 01:30:49 +00001462<div class="doc_subsubsection">
1463 <a name="i_getelementptr">'<tt>getelementptr</tt>' Instruction</a>
1464</div>
1465
Misha Brukman9d0919f2003-11-08 01:05:38 +00001466<div class="doc_text">
Chris Lattner7faa8832002-04-14 06:13:44 +00001467<h5>Syntax:</h5>
Chris Lattnerf74d5c72004-04-05 01:30:49 +00001468<pre>
1469 &lt;result&gt; = getelementptr &lt;ty&gt;* &lt;ptrval&gt;{, &lt;ty&gt; &lt;idx&gt;}*
1470</pre>
1471
Chris Lattner7faa8832002-04-14 06:13:44 +00001472<h5>Overview:</h5>
Chris Lattnerf74d5c72004-04-05 01:30:49 +00001473
1474<p>
1475The '<tt>getelementptr</tt>' instruction is used to get the address of a
1476subelement of an aggregate data structure.</p>
1477
Chris Lattner7faa8832002-04-14 06:13:44 +00001478<h5>Arguments:</h5>
Chris Lattnerf74d5c72004-04-05 01:30:49 +00001479
1480<p>This instruction takes a list of integer constants that indicate what
1481elements of the aggregate object to index to. The actual types of the arguments
1482provided depend on the type of the first pointer argument. The
1483'<tt>getelementptr</tt>' instruction is used to index down through the type
1484levels of a structure. When indexing into a structure, only <tt>uint</tt>
1485integer constants are allowed. When indexing into an array or pointer
1486<tt>int</tt> and <tt>long</tt> indexes are allowed of any sign.</p>
1487
Chris Lattner261efe92003-11-25 01:02:51 +00001488<p>For example, let's consider a C code fragment and how it gets
1489compiled to LLVM:</p>
Chris Lattnerf74d5c72004-04-05 01:30:49 +00001490
1491<pre>
1492 struct RT {
1493 char A;
1494 int B[10][20];
1495 char C;
1496 };
1497 struct ST {
1498 int X;
1499 double Y;
1500 struct RT Z;
1501 };
1502
1503 int *foo(struct ST *s) {
1504 return &amp;s[1].Z.B[5][13];
1505 }
1506</pre>
1507
Misha Brukman9d0919f2003-11-08 01:05:38 +00001508<p>The LLVM code generated by the GCC frontend is:</p>
Chris Lattnerf74d5c72004-04-05 01:30:49 +00001509
1510<pre>
1511 %RT = type { sbyte, [10 x [20 x int]], sbyte }
1512 %ST = type { int, double, %RT }
1513
Brian Gaeke7283e7c2004-07-02 21:08:14 +00001514 implementation
1515
1516 int* %foo(%ST* %s) {
1517 entry:
1518 %reg = getelementptr %ST* %s, int 1, uint 2, uint 1, int 5, int 13
Chris Lattnerf74d5c72004-04-05 01:30:49 +00001519 ret int* %reg
1520 }
1521</pre>
1522
Chris Lattner7faa8832002-04-14 06:13:44 +00001523<h5>Semantics:</h5>
Chris Lattnerf74d5c72004-04-05 01:30:49 +00001524
1525<p>The index types specified for the '<tt>getelementptr</tt>' instruction depend
Chris Lattnere53e5082004-06-03 22:57:15 +00001526on the pointer type that is being index into. <a href="#t_pointer">Pointer</a>
1527and <a href="#t_array">array</a> types require <tt>uint</tt>, <tt>int</tt>,
1528<tt>ulong</tt>, or <tt>long</tt> values, and <a href="#t_struct">structure</a>
Chris Lattnerf74d5c72004-04-05 01:30:49 +00001529types require <tt>uint</tt> <b>constants</b>.</p>
1530
Misha Brukman9d0919f2003-11-08 01:05:38 +00001531<p>In the example above, the first index is indexing into the '<tt>%ST*</tt>'
Chris Lattnerf74d5c72004-04-05 01:30:49 +00001532type, which is a pointer, yielding a '<tt>%ST</tt>' = '<tt>{ int, double, %RT
1533}</tt>' type, a structure. The second index indexes into the third element of
1534the structure, yielding a '<tt>%RT</tt>' = '<tt>{ sbyte, [10 x [20 x int]],
1535sbyte }</tt>' type, another structure. The third index indexes into the second
1536element of the structure, yielding a '<tt>[10 x [20 x int]]</tt>' type, an
1537array. The two dimensions of the array are subscripted into, yielding an
1538'<tt>int</tt>' type. The '<tt>getelementptr</tt>' instruction return a pointer
1539to this element, thus computing a value of '<tt>int*</tt>' type.</p>
1540
Chris Lattner261efe92003-11-25 01:02:51 +00001541<p>Note that it is perfectly legal to index partially through a
1542structure, returning a pointer to an inner element. Because of this,
1543the LLVM code for the given testcase is equivalent to:</p>
Chris Lattnerf74d5c72004-04-05 01:30:49 +00001544
1545<pre>
1546 int* "foo"(%ST* %s) {
1547 %t1 = getelementptr %ST* %s, int 1 <i>; yields %ST*:%t1</i>
1548 %t2 = getelementptr %ST* %t1, int 0, uint 2 <i>; yields %RT*:%t2</i>
1549 %t3 = getelementptr %RT* %t2, int 0, uint 1 <i>; yields [10 x [20 x int]]*:%t3</i>
1550 %t4 = getelementptr [10 x [20 x int]]* %t3, int 0, int 5 <i>; yields [20 x int]*:%t4</i>
1551 %t5 = getelementptr [20 x int]* %t4, int 0, int 13 <i>; yields int*:%t5</i>
1552 ret int* %t5
1553 }
Chris Lattner6536cfe2002-05-06 22:08:29 +00001554</pre>
Chris Lattner7faa8832002-04-14 06:13:44 +00001555<h5>Example:</h5>
Chris Lattnerf74d5c72004-04-05 01:30:49 +00001556<pre>
1557 <i>; yields [12 x ubyte]*:aptr</i>
1558 %aptr = getelementptr {int, [12 x ubyte]}* %sptr, long 0, uint 1
1559</pre>
1560
1561</div>
Chris Lattner00950542001-06-06 20:29:01 +00001562<!-- ======================================================================= -->
Chris Lattner261efe92003-11-25 01:02:51 +00001563<div class="doc_subsection"> <a name="otherops">Other Operations</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001564<div class="doc_text">
John Criswell4457dc92004-04-09 16:48:45 +00001565<p>The instructions in this category are the "miscellaneous"
Chris Lattner261efe92003-11-25 01:02:51 +00001566instructions, which defy better classification.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001567</div>
Chris Lattner00950542001-06-06 20:29:01 +00001568<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +00001569<div class="doc_subsubsection"> <a name="i_phi">'<tt>phi</tt>'
1570Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001571<div class="doc_text">
Chris Lattner33ba0d92001-07-09 00:26:23 +00001572<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001573<pre> &lt;result&gt; = phi &lt;ty&gt; [ &lt;val0&gt;, &lt;label0&gt;], ...<br></pre>
Chris Lattner33ba0d92001-07-09 00:26:23 +00001574<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001575<p>The '<tt>phi</tt>' instruction is used to implement the &#966; node in
1576the SSA graph representing the function.</p>
Chris Lattner33ba0d92001-07-09 00:26:23 +00001577<h5>Arguments:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001578<p>The type of the incoming values are specified with the first type
1579field. After this, the '<tt>phi</tt>' instruction takes a list of pairs
1580as arguments, with one pair for each predecessor basic block of the
1581current block. Only values of <a href="#t_firstclass">first class</a>
1582type may be used as the value arguments to the PHI node. Only labels
1583may be used as the label arguments.</p>
1584<p>There must be no non-phi instructions between the start of a basic
1585block and the PHI instructions: i.e. PHI instructions must be first in
1586a basic block.</p>
Chris Lattner33ba0d92001-07-09 00:26:23 +00001587<h5>Semantics:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001588<p>At runtime, the '<tt>phi</tt>' instruction logically takes on the
1589value specified by the parameter, depending on which basic block we
1590came from in the last <a href="#terminators">terminator</a> instruction.</p>
Chris Lattner6536cfe2002-05-06 22:08:29 +00001591<h5>Example:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001592<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 +00001593</div>
Chris Lattnercc37aae2004-03-12 05:50:16 +00001594
Chris Lattner6536cfe2002-05-06 22:08:29 +00001595<!-- _______________________________________________________________________ -->
Chris Lattnercc37aae2004-03-12 05:50:16 +00001596<div class="doc_subsubsection">
1597 <a name="i_cast">'<tt>cast .. to</tt>' Instruction</a>
1598</div>
1599
Misha Brukman9d0919f2003-11-08 01:05:38 +00001600<div class="doc_text">
Chris Lattnercc37aae2004-03-12 05:50:16 +00001601
Chris Lattner6536cfe2002-05-06 22:08:29 +00001602<h5>Syntax:</h5>
Chris Lattnercc37aae2004-03-12 05:50:16 +00001603
1604<pre>
1605 &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 +00001606</pre>
Chris Lattnercc37aae2004-03-12 05:50:16 +00001607
Chris Lattner6536cfe2002-05-06 22:08:29 +00001608<h5>Overview:</h5>
Chris Lattnercc37aae2004-03-12 05:50:16 +00001609
1610<p>
1611The '<tt>cast</tt>' instruction is used as the primitive means to convert
1612integers to floating point, change data type sizes, and break type safety (by
1613casting pointers).
1614</p>
1615
1616
Chris Lattner6536cfe2002-05-06 22:08:29 +00001617<h5>Arguments:</h5>
Chris Lattnercc37aae2004-03-12 05:50:16 +00001618
1619<p>
1620The '<tt>cast</tt>' instruction takes a value to cast, which must be a first
1621class value, and a type to cast it to, which must also be a <a
1622href="#t_firstclass">first class</a> type.
1623</p>
1624
Chris Lattner6536cfe2002-05-06 22:08:29 +00001625<h5>Semantics:</h5>
Chris Lattnercc37aae2004-03-12 05:50:16 +00001626
1627<p>
1628This instruction follows the C rules for explicit casts when determining how the
1629data being cast must change to fit in its new container.
1630</p>
1631
1632<p>
1633When casting to bool, any value that would be considered true in the context of
1634a C '<tt>if</tt>' condition is converted to the boolean '<tt>true</tt>' values,
1635all else are '<tt>false</tt>'.
1636</p>
1637
1638<p>
1639When extending an integral value from a type of one signness to another (for
1640example '<tt>sbyte</tt>' to '<tt>ulong</tt>'), the value is sign-extended if the
1641<b>source</b> value is signed, and zero-extended if the source value is
1642unsigned. <tt>bool</tt> values are always zero extended into either zero or
1643one.
1644</p>
1645
Chris Lattner33ba0d92001-07-09 00:26:23 +00001646<h5>Example:</h5>
Chris Lattnercc37aae2004-03-12 05:50:16 +00001647
1648<pre>
1649 %X = cast int 257 to ubyte <i>; yields ubyte:1</i>
Chris Lattner7bae3952002-06-25 18:03:17 +00001650 %Y = cast int 123 to bool <i>; yields bool:true</i>
Chris Lattner33ba0d92001-07-09 00:26:23 +00001651</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001652</div>
Chris Lattnercc37aae2004-03-12 05:50:16 +00001653
1654<!-- _______________________________________________________________________ -->
1655<div class="doc_subsubsection">
1656 <a name="i_select">'<tt>select</tt>' Instruction</a>
1657</div>
1658
1659<div class="doc_text">
1660
1661<h5>Syntax:</h5>
1662
1663<pre>
1664 &lt;result&gt; = select bool &lt;cond&gt;, &lt;ty&gt; &lt;val1&gt;, &lt;ty&gt; &lt;val2&gt; <i>; yields ty</i>
1665</pre>
1666
1667<h5>Overview:</h5>
1668
1669<p>
1670The '<tt>select</tt>' instruction is used to choose one value based on a
1671condition, without branching.
1672</p>
1673
1674
1675<h5>Arguments:</h5>
1676
1677<p>
1678The '<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.
1679</p>
1680
1681<h5>Semantics:</h5>
1682
1683<p>
1684If the boolean condition evaluates to true, the instruction returns the first
1685value argument, otherwise it returns the second value argument.
1686</p>
1687
1688<h5>Example:</h5>
1689
1690<pre>
1691 %X = select bool true, ubyte 17, ubyte 42 <i>; yields ubyte:17</i>
1692</pre>
1693</div>
1694
1695
1696
1697
1698
Chris Lattner33ba0d92001-07-09 00:26:23 +00001699<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +00001700<div class="doc_subsubsection"> <a name="i_call">'<tt>call</tt>'
1701Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001702<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +00001703<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001704<pre> &lt;result&gt; = call &lt;ty&gt;* &lt;fnptrval&gt;(&lt;param list&gt;)<br></pre>
Chris Lattner00950542001-06-06 20:29:01 +00001705<h5>Overview:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001706<p>The '<tt>call</tt>' instruction represents a simple function call.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001707<h5>Arguments:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001708<p>This instruction requires several arguments:</p>
Chris Lattner6536cfe2002-05-06 22:08:29 +00001709<ol>
Chris Lattner261efe92003-11-25 01:02:51 +00001710 <li>
1711 <p>'<tt>ty</tt>': shall be the signature of the pointer to function
1712value being invoked. The argument types must match the types implied
1713by this signature.</p>
1714 </li>
1715 <li>
1716 <p>'<tt>fnptrval</tt>': An LLVM value containing a pointer to a
1717function to be invoked. In most cases, this is a direct function
1718invocation, but indirect <tt>call</tt>s are just as possible,
1719calling an arbitrary pointer to function values.</p>
1720 </li>
1721 <li>
1722 <p>'<tt>function args</tt>': argument list whose types match the
1723function signature argument types. If the function signature
1724indicates the function accepts a variable number of arguments, the
1725extra arguments can be specified.</p>
1726 </li>
Chris Lattner6536cfe2002-05-06 22:08:29 +00001727</ol>
Chris Lattner00950542001-06-06 20:29:01 +00001728<h5>Semantics:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001729<p>The '<tt>call</tt>' instruction is used to cause control flow to
1730transfer to a specified function, with its incoming arguments bound to
1731the specified values. Upon a '<tt><a href="#i_ret">ret</a></tt>'
1732instruction in the called function, control flow continues with the
1733instruction after the function call, and the return value of the
1734function is bound to the result argument. This is a simpler case of
1735the <a href="#i_invoke">invoke</a> instruction.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001736<h5>Example:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001737<pre> %retval = call int %test(int %argc)<br> call int(sbyte*, ...) *%printf(sbyte* %msg, int 12, sbyte 42);<br></pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001738</div>
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00001739<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +00001740<div class="doc_subsubsection"> <a name="i_vanext">'<tt>vanext</tt>'
1741Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001742<div class="doc_text">
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00001743<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001744<pre> &lt;resultarglist&gt; = vanext &lt;va_list&gt; &lt;arglist&gt;, &lt;argty&gt;<br></pre>
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00001745<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001746<p>The '<tt>vanext</tt>' instruction is used to access arguments passed
1747through the "variable argument" area of a function call. It is used to
1748implement the <tt>va_arg</tt> macro in C.</p>
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00001749<h5>Arguments:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001750<p>This instruction takes a <tt>valist</tt> value and the type of the
1751argument. It returns another <tt>valist</tt>.</p>
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00001752<h5>Semantics:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001753<p>The '<tt>vanext</tt>' instruction advances the specified <tt>valist</tt>
1754past an argument of the specified type. In conjunction with the <a
1755 href="#i_vaarg"><tt>vaarg</tt></a> instruction, it is used to implement
1756the <tt>va_arg</tt> macro available in C. For more information, see
1757the variable argument handling <a href="#int_varargs">Intrinsic
1758Functions</a>.</p>
1759<p>It is legal for this instruction to be called in a function which
1760does not take a variable number of arguments, for example, the <tt>vfprintf</tt>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001761function.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001762<p><tt>vanext</tt> is an LLVM instruction instead of an <a
Chris Lattner261efe92003-11-25 01:02:51 +00001763 href="#intrinsics">intrinsic function</a> because it takes an type as
1764an argument.</p>
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00001765<h5>Example:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001766<p>See the <a href="#int_varargs">variable argument processing</a>
1767section.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001768</div>
Chris Lattner8d1a81d2003-10-18 05:51:36 +00001769<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +00001770<div class="doc_subsubsection"> <a name="i_vaarg">'<tt>vaarg</tt>'
1771Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001772<div class="doc_text">
Chris Lattner8d1a81d2003-10-18 05:51:36 +00001773<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001774<pre> &lt;resultval&gt; = vaarg &lt;va_list&gt; &lt;arglist&gt;, &lt;argty&gt;<br></pre>
Chris Lattner8d1a81d2003-10-18 05:51:36 +00001775<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001776<p>The '<tt>vaarg</tt>' instruction is used to access arguments passed
1777through the "variable argument" area of a function call. It is used to
1778implement the <tt>va_arg</tt> macro in C.</p>
Chris Lattner8d1a81d2003-10-18 05:51:36 +00001779<h5>Arguments:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001780<p>This instruction takes a <tt>valist</tt> value and the type of the
1781argument. It returns a value of the specified argument type.</p>
Chris Lattner8d1a81d2003-10-18 05:51:36 +00001782<h5>Semantics:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001783<p>The '<tt>vaarg</tt>' instruction loads an argument of the specified
1784type from the specified <tt>va_list</tt>. In conjunction with the <a
1785 href="#i_vanext"><tt>vanext</tt></a> instruction, it is used to
1786implement the <tt>va_arg</tt> macro available in C. For more
1787information, see the variable argument handling <a href="#int_varargs">Intrinsic
1788Functions</a>.</p>
1789<p>It is legal for this instruction to be called in a function which
1790does not take a variable number of arguments, for example, the <tt>vfprintf</tt>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001791function.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001792<p><tt>vaarg</tt> is an LLVM instruction instead of an <a
Chris Lattner261efe92003-11-25 01:02:51 +00001793 href="#intrinsics">intrinsic function</a> because it takes an type as
1794an argument.</p>
Chris Lattner8d1a81d2003-10-18 05:51:36 +00001795<h5>Example:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001796<p>See the <a href="#int_varargs">variable argument processing</a>
1797section.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001798</div>
Chris Lattner8ff75902004-01-06 05:31:32 +00001799
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00001800<!-- *********************************************************************** -->
Chris Lattner261efe92003-11-25 01:02:51 +00001801<div class="doc_section"> <a name="intrinsics">Intrinsic Functions</a> </div>
1802<!-- *********************************************************************** -->
Chris Lattner8ff75902004-01-06 05:31:32 +00001803
Misha Brukman9d0919f2003-11-08 01:05:38 +00001804<div class="doc_text">
Chris Lattner33aec9e2004-02-12 17:01:32 +00001805
1806<p>LLVM supports the notion of an "intrinsic function". These functions have
1807well known names and semantics, and are required to follow certain
1808restrictions. Overall, these instructions represent an extension mechanism for
1809the LLVM language that does not require changing all of the transformations in
1810LLVM to add to the language (or the bytecode reader/writer, the parser,
1811etc...).</p>
1812
1813<p>Intrinsic function names must all start with an "<tt>llvm.</tt>" prefix, this
1814prefix is reserved in LLVM for intrinsic names, thus functions may not be named
1815this. Intrinsic functions must always be external functions: you cannot define
1816the body of intrinsic functions. Intrinsic functions may only be used in call
1817or invoke instructions: it is illegal to take the address of an intrinsic
1818function. Additionally, because intrinsic functions are part of the LLVM
1819language, it is required that they all be documented here if any are added.</p>
1820
1821
1822<p>
1823Adding an intrinsic to LLVM is straight-forward if it is possible to express the
1824concept in LLVM directly (ie, code generator support is not _required_). To do
1825this, extend the default implementation of the IntrinsicLowering class to handle
1826the intrinsic. Code generators use this class to lower intrinsics they do not
1827understand to raw LLVM instructions that they do.
1828</p>
1829
Misha Brukman9d0919f2003-11-08 01:05:38 +00001830</div>
Chris Lattner8ff75902004-01-06 05:31:32 +00001831
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00001832<!-- ======================================================================= -->
Chris Lattner8ff75902004-01-06 05:31:32 +00001833<div class="doc_subsection">
1834 <a name="int_varargs">Variable Argument Handling Intrinsics</a>
1835</div>
1836
Misha Brukman9d0919f2003-11-08 01:05:38 +00001837<div class="doc_text">
Chris Lattnerd7923912004-05-23 21:06:01 +00001838
Misha Brukman9d0919f2003-11-08 01:05:38 +00001839<p>Variable argument support is defined in LLVM with the <a
Chris Lattner261efe92003-11-25 01:02:51 +00001840 href="#i_vanext"><tt>vanext</tt></a> instruction and these three
1841intrinsic functions. These functions are related to the similarly
1842named macros defined in the <tt>&lt;stdarg.h&gt;</tt> header file.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +00001843
Chris Lattner261efe92003-11-25 01:02:51 +00001844<p>All of these functions operate on arguments that use a
1845target-specific value type "<tt>va_list</tt>". The LLVM assembly
1846language reference manual does not define what this type is, so all
1847transformations should be prepared to handle intrinsics with any type
1848used.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +00001849
Misha Brukman9d0919f2003-11-08 01:05:38 +00001850<p>This example shows how the <a href="#i_vanext"><tt>vanext</tt></a>
Chris Lattner261efe92003-11-25 01:02:51 +00001851instruction and the variable argument handling intrinsic functions are
1852used.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +00001853
Chris Lattner33aec9e2004-02-12 17:01:32 +00001854<pre>
1855int %test(int %X, ...) {
1856 ; Initialize variable argument processing
1857 %ap = call sbyte* %<a href="#i_va_start">llvm.va_start</a>()
1858
1859 ; Read a single integer argument
1860 %tmp = vaarg sbyte* %ap, int
1861
1862 ; Advance to the next argument
1863 %ap2 = vanext sbyte* %ap, int
1864
1865 ; Demonstrate usage of llvm.va_copy and llvm.va_end
1866 %aq = call sbyte* %<a href="#i_va_copy">llvm.va_copy</a>(sbyte* %ap2)
1867 call void %<a href="#i_va_end">llvm.va_end</a>(sbyte* %aq)
1868
1869 ; Stop processing of arguments.
1870 call void %<a href="#i_va_end">llvm.va_end</a>(sbyte* %ap2)
1871 ret int %tmp
1872}
1873</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001874</div>
Chris Lattner8ff75902004-01-06 05:31:32 +00001875
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00001876<!-- _______________________________________________________________________ -->
Chris Lattner8ff75902004-01-06 05:31:32 +00001877<div class="doc_subsubsection">
1878 <a name="i_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a>
1879</div>
1880
1881
Misha Brukman9d0919f2003-11-08 01:05:38 +00001882<div class="doc_text">
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00001883<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001884<pre> call va_list ()* %llvm.va_start()<br></pre>
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00001885<h5>Overview:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001886<p>The '<tt>llvm.va_start</tt>' intrinsic returns a new <tt>&lt;arglist&gt;</tt>
1887for subsequent use by the variable argument intrinsics.</p>
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00001888<h5>Semantics:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001889<p>The '<tt>llvm.va_start</tt>' intrinsic works just like the <tt>va_start</tt>
Chris Lattner261efe92003-11-25 01:02:51 +00001890macro available in C. In a target-dependent way, it initializes and
1891returns a <tt>va_list</tt> element, so that the next <tt>vaarg</tt>
1892will produce the first variable argument passed to the function. Unlike
1893the C <tt>va_start</tt> macro, this intrinsic does not need to know the
1894last argument of the function, the compiler can figure that out.</p>
1895<p>Note that this intrinsic function is only legal to be called from
1896within the body of a variable argument function.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001897</div>
Chris Lattner8ff75902004-01-06 05:31:32 +00001898
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00001899<!-- _______________________________________________________________________ -->
Chris Lattner8ff75902004-01-06 05:31:32 +00001900<div class="doc_subsubsection">
1901 <a name="i_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a>
1902</div>
1903
Misha Brukman9d0919f2003-11-08 01:05:38 +00001904<div class="doc_text">
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00001905<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001906<pre> call void (va_list)* %llvm.va_end(va_list &lt;arglist&gt;)<br></pre>
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00001907<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001908<p>The '<tt>llvm.va_end</tt>' intrinsic destroys <tt>&lt;arglist&gt;</tt>
1909which has been initialized previously with <tt><a href="#i_va_start">llvm.va_start</a></tt>
1910or <tt><a href="#i_va_copy">llvm.va_copy</a></tt>.</p>
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00001911<h5>Arguments:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001912<p>The argument is a <tt>va_list</tt> to destroy.</p>
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00001913<h5>Semantics:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001914<p>The '<tt>llvm.va_end</tt>' intrinsic works just like the <tt>va_end</tt>
Chris Lattner261efe92003-11-25 01:02:51 +00001915macro available in C. In a target-dependent way, it destroys the <tt>va_list</tt>.
1916Calls to <a href="#i_va_start"><tt>llvm.va_start</tt></a> and <a
1917 href="#i_va_copy"><tt>llvm.va_copy</tt></a> must be matched exactly
1918with calls to <tt>llvm.va_end</tt>.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001919</div>
Chris Lattner8ff75902004-01-06 05:31:32 +00001920
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00001921<!-- _______________________________________________________________________ -->
Chris Lattner8ff75902004-01-06 05:31:32 +00001922<div class="doc_subsubsection">
1923 <a name="i_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a>
1924</div>
1925
Misha Brukman9d0919f2003-11-08 01:05:38 +00001926<div class="doc_text">
Chris Lattnerd7923912004-05-23 21:06:01 +00001927
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00001928<h5>Syntax:</h5>
Chris Lattnerd7923912004-05-23 21:06:01 +00001929
1930<pre>
1931 call va_list (va_list)* %llvm.va_copy(va_list &lt;destarglist&gt;)
1932</pre>
1933
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00001934<h5>Overview:</h5>
Chris Lattnerd7923912004-05-23 21:06:01 +00001935
1936<p>The '<tt>llvm.va_copy</tt>' intrinsic copies the current argument position
1937from the source argument list to the destination argument list.</p>
1938
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00001939<h5>Arguments:</h5>
Chris Lattnerd7923912004-05-23 21:06:01 +00001940
Misha Brukman9d0919f2003-11-08 01:05:38 +00001941<p>The argument is the <tt>va_list</tt> to copy.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +00001942
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00001943<h5>Semantics:</h5>
Chris Lattnerd7923912004-05-23 21:06:01 +00001944
Misha Brukman9d0919f2003-11-08 01:05:38 +00001945<p>The '<tt>llvm.va_copy</tt>' intrinsic works just like the <tt>va_copy</tt>
Chris Lattnerd7923912004-05-23 21:06:01 +00001946macro available in C. In a target-dependent way, it copies the source
1947<tt>va_list</tt> element into the returned list. This intrinsic is necessary
Chris Lattnerfcd37252004-06-21 22:52:48 +00001948because the <tt><a href="#i_va_start">llvm.va_start</a></tt> intrinsic may be
Chris Lattnerd7923912004-05-23 21:06:01 +00001949arbitrarily complex and require memory allocation, for example.</p>
1950
Misha Brukman9d0919f2003-11-08 01:05:38 +00001951</div>
Chris Lattner8ff75902004-01-06 05:31:32 +00001952
Chris Lattner33aec9e2004-02-12 17:01:32 +00001953<!-- ======================================================================= -->
1954<div class="doc_subsection">
Chris Lattnerd7923912004-05-23 21:06:01 +00001955 <a name="int_gc">Accurate Garbage Collection Intrinsics</a>
1956</div>
1957
1958<div class="doc_text">
1959
1960<p>
1961LLVM support for <a href="GarbageCollection.html">Accurate Garbage
1962Collection</a> requires the implementation and generation of these intrinsics.
1963These intrinsics allow identification of <a href="#i_gcroot">GC roots on the
1964stack</a>, as well as garbage collector implementations that require <a
1965href="#i_gcread">read</a> and <a href="#i_gcwrite">write</a> barriers.
1966Front-ends for type-safe garbage collected languages should generate these
1967intrinsics to make use of the LLVM garbage collectors. For more details, see <a
1968href="GarbageCollection.html">Accurate Garbage Collection with LLVM</a>.
1969</p>
1970</div>
1971
1972<!-- _______________________________________________________________________ -->
1973<div class="doc_subsubsection">
1974 <a name="i_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a>
1975</div>
1976
1977<div class="doc_text">
1978
1979<h5>Syntax:</h5>
1980
1981<pre>
1982 call void (&lt;ty&gt;**, &lt;ty2&gt;*)* %llvm.gcroot(&lt;ty&gt;** %ptrloc, &lt;ty2&gt;* %metadata)
1983</pre>
1984
1985<h5>Overview:</h5>
1986
1987<p>The '<tt>llvm.gcroot</tt>' intrinsic declares the existance of a GC root to
1988the code generator, and allows some metadata to be associated with it.</p>
1989
1990<h5>Arguments:</h5>
1991
1992<p>The first argument specifies the address of a stack object that contains the
1993root pointer. The second pointer (which must be either a constant or a global
1994value address) contains the meta-data to be associated with the root.</p>
1995
1996<h5>Semantics:</h5>
1997
1998<p>At runtime, a call to this intrinsics stores a null pointer into the "ptrloc"
1999location. At compile-time, the code generator generates information to allow
2000the runtime to find the pointer at GC safe points.
2001</p>
2002
2003</div>
2004
2005
2006<!-- _______________________________________________________________________ -->
2007<div class="doc_subsubsection">
2008 <a name="i_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a>
2009</div>
2010
2011<div class="doc_text">
2012
2013<h5>Syntax:</h5>
2014
2015<pre>
2016 call sbyte* (sbyte**)* %llvm.gcread(sbyte** %Ptr)
2017</pre>
2018
2019<h5>Overview:</h5>
2020
2021<p>The '<tt>llvm.gcread</tt>' intrinsic identifies reads of references from heap
2022locations, allowing garbage collector implementations that require read
2023barriers.</p>
2024
2025<h5>Arguments:</h5>
2026
2027<p>The argument is the address to read from, which should be an address
2028allocated from the garbage collector.</p>
2029
2030<h5>Semantics:</h5>
2031
2032<p>The '<tt>llvm.gcread</tt>' intrinsic has the same semantics as a load
2033instruction, but may be replaced with substantially more complex code by the
2034garbage collector runtime, as needed.</p>
2035
2036</div>
2037
2038
2039<!-- _______________________________________________________________________ -->
2040<div class="doc_subsubsection">
2041 <a name="i_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a>
2042</div>
2043
2044<div class="doc_text">
2045
2046<h5>Syntax:</h5>
2047
2048<pre>
2049 call void (sbyte*, sbyte**)* %llvm.gcwrite(sbyte* %P1, sbyte** %P2)
2050</pre>
2051
2052<h5>Overview:</h5>
2053
2054<p>The '<tt>llvm.gcwrite</tt>' intrinsic identifies writes of references to heap
2055locations, allowing garbage collector implementations that require write
2056barriers (such as generational or reference counting collectors).</p>
2057
2058<h5>Arguments:</h5>
2059
2060<p>The first argument is the reference to store, and the second is the heap
2061location to store to.</p>
2062
2063<h5>Semantics:</h5>
2064
2065<p>The '<tt>llvm.gcwrite</tt>' intrinsic has the same semantics as a store
2066instruction, but may be replaced with substantially more complex code by the
2067garbage collector runtime, as needed.</p>
2068
2069</div>
2070
2071
2072
2073<!-- ======================================================================= -->
2074<div class="doc_subsection">
Chris Lattner10610642004-02-14 04:08:35 +00002075 <a name="int_codegen">Code Generator Intrinsics</a>
2076</div>
2077
2078<div class="doc_text">
2079<p>
2080These intrinsics are provided by LLVM to expose special features that may only
2081be implemented with code generator support.
2082</p>
2083
2084</div>
2085
2086<!-- _______________________________________________________________________ -->
2087<div class="doc_subsubsection">
2088 <a name="i_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a>
2089</div>
2090
2091<div class="doc_text">
2092
2093<h5>Syntax:</h5>
2094<pre>
2095 call void* ()* %llvm.returnaddress(uint &lt;level&gt;)
2096</pre>
2097
2098<h5>Overview:</h5>
2099
2100<p>
2101The '<tt>llvm.returnaddress</tt>' intrinsic returns a target-specific value
2102indicating the return address of the current function or one of its callers.
2103</p>
2104
2105<h5>Arguments:</h5>
2106
2107<p>
2108The argument to this intrinsic indicates which function to return the address
2109for. Zero indicates the calling function, one indicates its caller, etc. The
2110argument is <b>required</b> to be a constant integer value.
2111</p>
2112
2113<h5>Semantics:</h5>
2114
2115<p>
2116The '<tt>llvm.returnaddress</tt>' intrinsic either returns a pointer indicating
2117the return address of the specified call frame, or zero if it cannot be
2118identified. The value returned by this intrinsic is likely to be incorrect or 0
2119for arguments other than zero, so it should only be used for debugging purposes.
2120</p>
2121
2122<p>
2123Note that calling this intrinsic does not prevent function inlining or other
2124aggressive transformations, so the value returned may not that of the obvious
2125source-language caller.
2126</p>
2127</div>
2128
2129
2130<!-- _______________________________________________________________________ -->
2131<div class="doc_subsubsection">
2132 <a name="i_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a>
2133</div>
2134
2135<div class="doc_text">
2136
2137<h5>Syntax:</h5>
2138<pre>
2139 call void* ()* %llvm.frameaddress(uint &lt;level&gt;)
2140</pre>
2141
2142<h5>Overview:</h5>
2143
2144<p>
2145The '<tt>llvm.frameaddress</tt>' intrinsic returns the target-specific frame
2146pointer value for the specified stack frame.
2147</p>
2148
2149<h5>Arguments:</h5>
2150
2151<p>
2152The argument to this intrinsic indicates which function to return the frame
2153pointer for. Zero indicates the calling function, one indicates its caller,
2154etc. The argument is <b>required</b> to be a constant integer value.
2155</p>
2156
2157<h5>Semantics:</h5>
2158
2159<p>
2160The '<tt>llvm.frameaddress</tt>' intrinsic either returns a pointer indicating
2161the frame address of the specified call frame, or zero if it cannot be
2162identified. The value returned by this intrinsic is likely to be incorrect or 0
2163for arguments other than zero, so it should only be used for debugging purposes.
2164</p>
2165
2166<p>
2167Note that calling this intrinsic does not prevent function inlining or other
2168aggressive transformations, so the value returned may not that of the obvious
2169source-language caller.
2170</p>
2171</div>
2172
John Criswell7123e272004-04-09 16:43:20 +00002173<!-- ======================================================================= -->
2174<div class="doc_subsection">
2175 <a name="int_os">Operating System Intrinsics</a>
2176</div>
2177
2178<div class="doc_text">
2179<p>
2180These intrinsics are provided by LLVM to support the implementation of
2181operating system level code.
2182</p>
2183
2184</div>
John Criswell183402a2004-04-12 15:02:16 +00002185
John Criswellcfd3bac2004-04-09 15:23:37 +00002186<!-- _______________________________________________________________________ -->
2187<div class="doc_subsubsection">
2188 <a name="i_readport">'<tt>llvm.readport</tt>' Intrinsic</a>
2189</div>
2190
2191<div class="doc_text">
2192
2193<h5>Syntax:</h5>
2194<pre>
John Criswell7123e272004-04-09 16:43:20 +00002195 call &lt;integer type&gt; (&lt;integer type&gt;)* %llvm.readport (&lt;integer type&gt; &lt;address&gt;)
John Criswellcfd3bac2004-04-09 15:23:37 +00002196</pre>
2197
2198<h5>Overview:</h5>
2199
2200<p>
John Criswell7123e272004-04-09 16:43:20 +00002201The '<tt>llvm.readport</tt>' intrinsic reads data from the specified hardware
2202I/O port.
John Criswellcfd3bac2004-04-09 15:23:37 +00002203</p>
2204
2205<h5>Arguments:</h5>
2206
2207<p>
John Criswell7123e272004-04-09 16:43:20 +00002208The argument to this intrinsic indicates the hardware I/O address from which
2209to read the data. The address is in the hardware I/O address namespace (as
2210opposed to being a memory location for memory mapped I/O).
John Criswellcfd3bac2004-04-09 15:23:37 +00002211</p>
2212
2213<h5>Semantics:</h5>
2214
2215<p>
John Criswell7123e272004-04-09 16:43:20 +00002216The '<tt>llvm.readport</tt>' intrinsic reads data from the hardware I/O port
2217specified by <i>address</i> and returns the value. The address and return
2218value must be integers, but the size is dependent upon the platform upon which
2219the program is code generated. For example, on x86, the address must be an
2220unsigned 16 bit value, and the return value must be 8, 16, or 32 bits.
John Criswellcfd3bac2004-04-09 15:23:37 +00002221</p>
2222
2223</div>
2224
2225<!-- _______________________________________________________________________ -->
2226<div class="doc_subsubsection">
2227 <a name="i_writeport">'<tt>llvm.writeport</tt>' Intrinsic</a>
2228</div>
2229
2230<div class="doc_text">
2231
2232<h5>Syntax:</h5>
2233<pre>
John Criswell7123e272004-04-09 16:43:20 +00002234 call void (&lt;integer type&gt;, &lt;integer type&gt;)* %llvm.writeport (&lt;integer type&gt; &lt;value&gt;, &lt;integer type&gt; &lt;address&gt;)
John Criswellcfd3bac2004-04-09 15:23:37 +00002235</pre>
2236
2237<h5>Overview:</h5>
2238
2239<p>
John Criswell7123e272004-04-09 16:43:20 +00002240The '<tt>llvm.writeport</tt>' intrinsic writes data to the specified hardware
2241I/O port.
John Criswellcfd3bac2004-04-09 15:23:37 +00002242</p>
2243
2244<h5>Arguments:</h5>
2245
2246<p>
John Criswell96db6fc2004-04-12 16:33:19 +00002247The first argument is the value to write to the I/O port.
John Criswellcfd3bac2004-04-09 15:23:37 +00002248</p>
2249
2250<p>
John Criswell96db6fc2004-04-12 16:33:19 +00002251The second argument indicates the hardware I/O address to which data should be
2252written. The address is in the hardware I/O address namespace (as opposed to
2253being a memory location for memory mapped I/O).
John Criswellcfd3bac2004-04-09 15:23:37 +00002254</p>
2255
2256<h5>Semantics:</h5>
2257
2258<p>
2259The '<tt>llvm.writeport</tt>' intrinsic writes <i>value</i> to the I/O port
2260specified by <i>address</i>. The address and value must be integers, but the
2261size is dependent upon the platform upon which the program is code generated.
John Criswell7123e272004-04-09 16:43:20 +00002262For example, on x86, the address must be an unsigned 16 bit value, and the
2263value written must be 8, 16, or 32 bits in length.
John Criswellcfd3bac2004-04-09 15:23:37 +00002264</p>
2265
2266</div>
Chris Lattner10610642004-02-14 04:08:35 +00002267
John Criswell183402a2004-04-12 15:02:16 +00002268<!-- _______________________________________________________________________ -->
2269<div class="doc_subsubsection">
2270 <a name="i_readio">'<tt>llvm.readio</tt>' Intrinsic</a>
2271</div>
2272
2273<div class="doc_text">
2274
2275<h5>Syntax:</h5>
2276<pre>
John Criswell96db6fc2004-04-12 16:33:19 +00002277 call &lt;result&gt; (&lt;ty&gt;*)* %llvm.readio (&lt;ty&gt; * &lt;pointer&gt;)
John Criswell183402a2004-04-12 15:02:16 +00002278</pre>
2279
2280<h5>Overview:</h5>
2281
2282<p>
2283The '<tt>llvm.readio</tt>' intrinsic reads data from a memory mapped I/O
2284address.
2285</p>
2286
2287<h5>Arguments:</h5>
2288
2289<p>
John Criswell96db6fc2004-04-12 16:33:19 +00002290The argument to this intrinsic is a pointer indicating the memory address from
2291which to read the data. The data must be a
2292<a href="#t_firstclass">first class</a> type.
John Criswell183402a2004-04-12 15:02:16 +00002293</p>
2294
2295<h5>Semantics:</h5>
2296
2297<p>
2298The '<tt>llvm.readio</tt>' intrinsic reads data from a memory mapped I/O
John Criswell96db6fc2004-04-12 16:33:19 +00002299location specified by <i>pointer</i> and returns the value. The argument must
2300be a pointer, and the return value must be a
2301<a href="#t_firstclass">first class</a> type. However, certain architectures
2302may not support I/O on all first class types. For example, 32 bit processors
2303may only support I/O on data types that are 32 bits or less.
John Criswell183402a2004-04-12 15:02:16 +00002304</p>
2305
2306<p>
John Criswell96db6fc2004-04-12 16:33:19 +00002307This intrinsic enforces an in-order memory model for llvm.readio and
2308llvm.writeio calls on machines that use dynamic scheduling. Dynamically
2309scheduled processors may execute loads and stores out of order, re-ordering at
2310run time accesses to memory mapped I/O registers. Using these intrinsics
2311ensures that accesses to memory mapped I/O registers occur in program order.
John Criswell183402a2004-04-12 15:02:16 +00002312</p>
2313
2314</div>
2315
2316<!-- _______________________________________________________________________ -->
2317<div class="doc_subsubsection">
2318 <a name="i_writeio">'<tt>llvm.writeio</tt>' Intrinsic</a>
2319</div>
2320
2321<div class="doc_text">
2322
2323<h5>Syntax:</h5>
2324<pre>
John Criswell96db6fc2004-04-12 16:33:19 +00002325 call void (&lt;ty1&gt;, &lt;ty2&gt;*)* %llvm.writeio (&lt;ty1&gt; &lt;value&gt;, &lt;ty2&gt; * &lt;pointer&gt;)
John Criswell183402a2004-04-12 15:02:16 +00002326</pre>
2327
2328<h5>Overview:</h5>
2329
2330<p>
2331The '<tt>llvm.writeio</tt>' intrinsic writes data to the specified memory
2332mapped I/O address.
2333</p>
2334
2335<h5>Arguments:</h5>
2336
2337<p>
John Criswell96db6fc2004-04-12 16:33:19 +00002338The first argument is the value to write to the memory mapped I/O location.
2339The second argument is a pointer indicating the memory address to which the
2340data should be written.
John Criswell183402a2004-04-12 15:02:16 +00002341</p>
2342
2343<h5>Semantics:</h5>
2344
2345<p>
2346The '<tt>llvm.writeio</tt>' intrinsic writes <i>value</i> to the memory mapped
John Criswell96db6fc2004-04-12 16:33:19 +00002347I/O address specified by <i>pointer</i>. The value must be a
2348<a href="#t_firstclass">first class</a> type. However, certain architectures
2349may not support I/O on all first class types. For example, 32 bit processors
2350may only support I/O on data types that are 32 bits or less.
John Criswell183402a2004-04-12 15:02:16 +00002351</p>
2352
2353<p>
John Criswell96db6fc2004-04-12 16:33:19 +00002354This intrinsic enforces an in-order memory model for llvm.readio and
2355llvm.writeio calls on machines that use dynamic scheduling. Dynamically
2356scheduled processors may execute loads and stores out of order, re-ordering at
2357run time accesses to memory mapped I/O registers. Using these intrinsics
2358ensures that accesses to memory mapped I/O registers occur in program order.
John Criswell183402a2004-04-12 15:02:16 +00002359</p>
2360
2361</div>
2362
Chris Lattner10610642004-02-14 04:08:35 +00002363<!-- ======================================================================= -->
2364<div class="doc_subsection">
Chris Lattner33aec9e2004-02-12 17:01:32 +00002365 <a name="int_libc">Standard C Library Intrinsics</a>
2366</div>
2367
2368<div class="doc_text">
2369<p>
Chris Lattner10610642004-02-14 04:08:35 +00002370LLVM provides intrinsics for a few important standard C library functions.
2371These intrinsics allow source-language front-ends to pass information about the
2372alignment of the pointer arguments to the code generator, providing opportunity
2373for more efficient code generation.
Chris Lattner33aec9e2004-02-12 17:01:32 +00002374</p>
2375
2376</div>
2377
2378<!-- _______________________________________________________________________ -->
2379<div class="doc_subsubsection">
2380 <a name="i_memcpy">'<tt>llvm.memcpy</tt>' Intrinsic</a>
2381</div>
2382
2383<div class="doc_text">
2384
2385<h5>Syntax:</h5>
2386<pre>
2387 call void (sbyte*, sbyte*, uint, uint)* %llvm.memcpy(sbyte* &lt;dest&gt;, sbyte* &lt;src&gt;,
2388 uint &lt;len&gt;, uint &lt;align&gt;)
2389</pre>
2390
2391<h5>Overview:</h5>
2392
2393<p>
2394The '<tt>llvm.memcpy</tt>' intrinsic copies a block of memory from the source
2395location to the destination location.
2396</p>
2397
2398<p>
2399Note that, unlike the standard libc function, the <tt>llvm.memcpy</tt> intrinsic
2400does not return a value, and takes an extra alignment argument.
2401</p>
2402
2403<h5>Arguments:</h5>
2404
2405<p>
2406The first argument is a pointer to the destination, the second is a pointer to
2407the source. The third argument is an (arbitrarily sized) integer argument
2408specifying the number of bytes to copy, and the fourth argument is the alignment
2409of the source and destination locations.
2410</p>
2411
Chris Lattner3301ced2004-02-12 21:18:15 +00002412<p>
2413If the call to this intrinisic has an alignment value that is not 0 or 1, then
2414the caller guarantees that the size of the copy is a multiple of the alignment
2415and that both the source and destination pointers are aligned to that boundary.
2416</p>
2417
Chris Lattner33aec9e2004-02-12 17:01:32 +00002418<h5>Semantics:</h5>
2419
2420<p>
2421The '<tt>llvm.memcpy</tt>' intrinsic copies a block of memory from the source
2422location to the destination location, which are not allowed to overlap. It
2423copies "len" bytes of memory over. If the argument is known to be aligned to
2424some boundary, this can be specified as the fourth argument, otherwise it should
2425be set to 0 or 1.
2426</p>
2427</div>
2428
2429
Chris Lattner0eb51b42004-02-12 18:10:10 +00002430<!-- _______________________________________________________________________ -->
2431<div class="doc_subsubsection">
2432 <a name="i_memmove">'<tt>llvm.memmove</tt>' Intrinsic</a>
2433</div>
2434
2435<div class="doc_text">
2436
2437<h5>Syntax:</h5>
2438<pre>
2439 call void (sbyte*, sbyte*, uint, uint)* %llvm.memmove(sbyte* &lt;dest&gt;, sbyte* &lt;src&gt;,
2440 uint &lt;len&gt;, uint &lt;align&gt;)
2441</pre>
2442
2443<h5>Overview:</h5>
2444
2445<p>
2446The '<tt>llvm.memmove</tt>' intrinsic moves a block of memory from the source
2447location to the destination location. It is similar to the '<tt>llvm.memcpy</tt>'
2448intrinsic but allows the two memory locations to overlap.
2449</p>
2450
2451<p>
2452Note that, unlike the standard libc function, the <tt>llvm.memmove</tt> intrinsic
2453does not return a value, and takes an extra alignment argument.
2454</p>
2455
2456<h5>Arguments:</h5>
2457
2458<p>
2459The first argument is a pointer to the destination, the second is a pointer to
2460the source. The third argument is an (arbitrarily sized) integer argument
2461specifying the number of bytes to copy, and the fourth argument is the alignment
2462of the source and destination locations.
2463</p>
2464
Chris Lattner3301ced2004-02-12 21:18:15 +00002465<p>
2466If the call to this intrinisic has an alignment value that is not 0 or 1, then
2467the caller guarantees that the size of the copy is a multiple of the alignment
2468and that both the source and destination pointers are aligned to that boundary.
2469</p>
2470
Chris Lattner0eb51b42004-02-12 18:10:10 +00002471<h5>Semantics:</h5>
2472
2473<p>
2474The '<tt>llvm.memmove</tt>' intrinsic copies a block of memory from the source
2475location to the destination location, which may overlap. It
2476copies "len" bytes of memory over. If the argument is known to be aligned to
2477some boundary, this can be specified as the fourth argument, otherwise it should
2478be set to 0 or 1.
2479</p>
2480</div>
2481
Chris Lattner8ff75902004-01-06 05:31:32 +00002482
Chris Lattner10610642004-02-14 04:08:35 +00002483<!-- _______________________________________________________________________ -->
2484<div class="doc_subsubsection">
2485 <a name="i_memset">'<tt>llvm.memset</tt>' Intrinsic</a>
2486</div>
2487
2488<div class="doc_text">
2489
2490<h5>Syntax:</h5>
2491<pre>
2492 call void (sbyte*, ubyte, uint, uint)* %llvm.memset(sbyte* &lt;dest&gt;, ubyte &lt;val&gt;,
2493 uint &lt;len&gt;, uint &lt;align&gt;)
2494</pre>
2495
2496<h5>Overview:</h5>
2497
2498<p>
2499The '<tt>llvm.memset</tt>' intrinsic fills a block of memory with a particular
2500byte value.
2501</p>
2502
2503<p>
2504Note that, unlike the standard libc function, the <tt>llvm.memset</tt> intrinsic
2505does not return a value, and takes an extra alignment argument.
2506</p>
2507
2508<h5>Arguments:</h5>
2509
2510<p>
2511The first argument is a pointer to the destination to fill, the second is the
2512byte value to fill it with, the third argument is an (arbitrarily sized) integer
2513argument specifying the number of bytes to fill, and the fourth argument is the
2514known alignment of destination location.
2515</p>
2516
2517<p>
2518If the call to this intrinisic has an alignment value that is not 0 or 1, then
2519the caller guarantees that the size of the copy is a multiple of the alignment
2520and that the destination pointer is aligned to that boundary.
2521</p>
2522
2523<h5>Semantics:</h5>
2524
2525<p>
2526The '<tt>llvm.memset</tt>' intrinsic fills "len" bytes of memory starting at the
2527destination location. If the argument is known to be aligned to some boundary,
2528this can be specified as the fourth argument, otherwise it should be set to 0 or
25291.
2530</p>
2531</div>
2532
2533
Chris Lattner32006282004-06-11 02:28:03 +00002534<!-- _______________________________________________________________________ -->
2535<div class="doc_subsubsection">
Alkis Evlogimenos26bbe932004-06-13 01:16:15 +00002536 <a name="i_isunordered">'<tt>llvm.isunordered</tt>' Intrinsic</a>
2537</div>
2538
2539<div class="doc_text">
2540
2541<h5>Syntax:</h5>
2542<pre>
2543 call bool (&lt;float or double&gt;, &lt;float or double&gt;)* %llvm.isunordered(&lt;float or double&gt; Val1,
2544 &lt;float or double&gt; Val2)
2545</pre>
2546
2547<h5>Overview:</h5>
2548
2549<p>
2550The '<tt>llvm.isunordered</tt>' intrinsic returns true if either or both of the
2551specified floating point values is a NAN.
2552</p>
2553
2554<h5>Arguments:</h5>
2555
2556<p>
2557The arguments are floating point numbers of the same type.
2558</p>
2559
2560<h5>Semantics:</h5>
2561
2562<p>
2563If either or both of the arguments is a SNAN or QNAN, it returns true, otherwise
2564false.
2565</p>
2566</div>
2567
2568
Chris Lattner32006282004-06-11 02:28:03 +00002569
2570
Chris Lattner8ff75902004-01-06 05:31:32 +00002571<!-- ======================================================================= -->
2572<div class="doc_subsection">
2573 <a name="int_debugger">Debugger Intrinsics</a>
2574</div>
2575
2576<div class="doc_text">
2577<p>
2578The LLVM debugger intrinsics (which all start with <tt>llvm.dbg.</tt> prefix),
2579are described in the <a
2580href="SourceLevelDebugging.html#format_common_intrinsics">LLVM Source Level
2581Debugging</a> document.
2582</p>
2583</div>
2584
2585
Chris Lattner00950542001-06-06 20:29:01 +00002586<!-- *********************************************************************** -->
Chris Lattner00950542001-06-06 20:29:01 +00002587<hr>
Misha Brukmandaa4cb02004-03-01 17:47:27 +00002588<address>
2589 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
2590 src="http://jigsaw.w3.org/css-validator/images/vcss" alt="Valid CSS!"></a>
2591 <a href="http://validator.w3.org/check/referer"><img
2592 src="http://www.w3.org/Icons/valid-html401" alt="Valid HTML 4.01!" /></a>
2593
2594 <a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
2595 <a href="http://llvm.cs.uiuc.edu">The LLVM Compiler Infrastructure</a><br>
2596 Last modified: $Date$
2597</address>
Misha Brukman9d0919f2003-11-08 01:05:38 +00002598</body>
2599</html>