blob: 7b03c45511f059f31ec3046593f81b529d24b6f1 [file] [log] [blame]
Chris Lattner261efe92003-11-25 01:02:51 +00001<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN" "http://www.w3.org/TR/html4/strict.dtd">
Misha Brukman9d0919f2003-11-08 01:05:38 +00002<html>
3<head>
4 <title>LLVM Assembly Language Reference Manual</title>
5 <link rel="stylesheet" href="llvm.css" type="text/css">
6</head>
7<body>
Chris Lattner261efe92003-11-25 01:02:51 +00008<div class="doc_title"> LLVM Language Reference Manual </div>
Chris Lattner00950542001-06-06 20:29:01 +00009<ol>
Misha Brukman9d0919f2003-11-08 01:05:38 +000010 <li><a href="#abstract">Abstract</a></li>
11 <li><a href="#introduction">Introduction</a></li>
12 <li><a href="#identifiers">Identifiers</a></li>
Chris Lattner00950542001-06-06 20:29:01 +000013 <li><a href="#typesystem">Type System</a>
14 <ol>
Chris Lattner261efe92003-11-25 01:02:51 +000015 <li><a href="#t_primitive">Primitive Types</a>
16 <ol>
Misha Brukman9d0919f2003-11-08 01:05:38 +000017 <li><a href="#t_classifications">Type Classifications</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +000018 </ol>
19 </li>
Chris Lattner00950542001-06-06 20:29:01 +000020 <li><a href="#t_derived">Derived Types</a>
21 <ol>
Chris Lattner261efe92003-11-25 01:02:51 +000022 <li><a href="#t_array">Array Type</a></li>
Misha Brukman9d0919f2003-11-08 01:05:38 +000023 <li><a href="#t_function">Function Type</a></li>
24 <li><a href="#t_pointer">Pointer Type</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +000025 <li><a href="#t_struct">Structure Type</a></li>
26<!-- <li><a href="#t_packed" >Packed Type</a> -->
27 </ol>
28 </li>
29 </ol>
30 </li>
Chris Lattner00950542001-06-06 20:29:01 +000031 <li><a href="#highlevel">High Level Structure</a>
32 <ol>
Misha Brukman9d0919f2003-11-08 01:05:38 +000033 <li><a href="#modulestructure">Module Structure</a></li>
34 <li><a href="#globalvars">Global Variables</a></li>
35 <li><a href="#functionstructure">Function Structure</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +000036 </ol>
37 </li>
Chris Lattner00950542001-06-06 20:29:01 +000038 <li><a href="#instref">Instruction Reference</a>
39 <ol>
40 <li><a href="#terminators">Terminator Instructions</a>
41 <ol>
Chris Lattner261efe92003-11-25 01:02:51 +000042 <li><a href="#i_ret">'<tt>ret</tt>' Instruction</a></li>
43 <li><a href="#i_br">'<tt>br</tt>' Instruction</a></li>
Misha Brukman9d0919f2003-11-08 01:05:38 +000044 <li><a href="#i_switch">'<tt>switch</tt>' Instruction</a></li>
45 <li><a href="#i_invoke">'<tt>invoke</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +000046 <li><a href="#i_unwind">'<tt>unwind</tt>' Instruction</a></li>
47 </ol>
48 </li>
Chris Lattner00950542001-06-06 20:29:01 +000049 <li><a href="#binaryops">Binary Operations</a>
50 <ol>
Chris Lattner261efe92003-11-25 01:02:51 +000051 <li><a href="#i_add">'<tt>add</tt>' Instruction</a></li>
52 <li><a href="#i_sub">'<tt>sub</tt>' Instruction</a></li>
53 <li><a href="#i_mul">'<tt>mul</tt>' Instruction</a></li>
54 <li><a href="#i_div">'<tt>div</tt>' Instruction</a></li>
55 <li><a href="#i_rem">'<tt>rem</tt>' Instruction</a></li>
Misha Brukman9d0919f2003-11-08 01:05:38 +000056 <li><a href="#i_setcc">'<tt>set<i>cc</i></tt>' Instructions</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +000057 </ol>
58 </li>
Chris Lattner00950542001-06-06 20:29:01 +000059 <li><a href="#bitwiseops">Bitwise Binary Operations</a>
60 <ol>
Misha Brukman9d0919f2003-11-08 01:05:38 +000061 <li><a href="#i_and">'<tt>and</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +000062 <li><a href="#i_or">'<tt>or</tt>' Instruction</a></li>
Misha Brukman9d0919f2003-11-08 01:05:38 +000063 <li><a href="#i_xor">'<tt>xor</tt>' Instruction</a></li>
64 <li><a href="#i_shl">'<tt>shl</tt>' Instruction</a></li>
65 <li><a href="#i_shr">'<tt>shr</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +000066 </ol>
67 </li>
Chris Lattner00950542001-06-06 20:29:01 +000068 <li><a href="#memoryops">Memory Access Operations</a>
69 <ol>
Chris Lattner261efe92003-11-25 01:02:51 +000070 <li><a href="#i_malloc">'<tt>malloc</tt>' Instruction</a></li>
71 <li><a href="#i_free">'<tt>free</tt>' Instruction</a></li>
72 <li><a href="#i_alloca">'<tt>alloca</tt>' Instruction</a></li>
73 <li><a href="#i_load">'<tt>load</tt>' Instruction</a></li>
74 <li><a href="#i_store">'<tt>store</tt>' Instruction</a></li>
75 <li><a href="#i_getelementptr">'<tt>getelementptr</tt>' Instruction</a></li>
76 </ol>
77 </li>
Chris Lattner00950542001-06-06 20:29:01 +000078 <li><a href="#otherops">Other Operations</a>
79 <ol>
Chris Lattner261efe92003-11-25 01:02:51 +000080 <li><a href="#i_phi">'<tt>phi</tt>' Instruction</a></li>
Misha Brukman9d0919f2003-11-08 01:05:38 +000081 <li><a href="#i_cast">'<tt>cast .. to</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +000082 <li><a href="#i_call">'<tt>call</tt>' Instruction</a></li>
Misha Brukman9d0919f2003-11-08 01:05:38 +000083 <li><a href="#i_vanext">'<tt>vanext</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +000084 <li><a href="#i_vaarg">'<tt>vaarg</tt>' Instruction</a></li>
Chris Lattner00950542001-06-06 20:29:01 +000085 </ol>
Chris Lattner261efe92003-11-25 01:02:51 +000086 </li>
Chris Lattner00950542001-06-06 20:29:01 +000087 </ol>
Chris Lattner261efe92003-11-25 01:02:51 +000088 </li>
Chris Lattnerd9ad5b32003-05-08 04:57:36 +000089 <li><a href="#intrinsics">Intrinsic Functions</a>
Chris Lattnerd9ad5b32003-05-08 04:57:36 +000090 <ol>
Chris Lattner261efe92003-11-25 01:02:51 +000091 <li><a href="#int_varargs">Variable Argument Handling Intrinsics</a>
92 <ol>
93 <li><a href="#i_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a></li>
94 <li><a href="#i_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a></li>
95 <li><a href="#i_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a></li>
96 </ol>
97 </li>
Chris Lattner33aec9e2004-02-12 17:01:32 +000098 <li><a href="#int_libc">Standard C Library Intrinsics</a>
99 <ol>
100 <li><a href="#i_memcpy">'<tt>llvm.memcpy</tt>' Intrinsic</a></li>
101 </ol>
102 </li>
Chris Lattner8ff75902004-01-06 05:31:32 +0000103 <li><a href="#int_debugger">Debugger intrinsics</a>
Chris Lattner261efe92003-11-25 01:02:51 +0000104 </ol>
105 </li>
Chris Lattner00950542001-06-06 20:29:01 +0000106</ol>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000107<div class="doc_text">
Chris Lattner261efe92003-11-25 01:02:51 +0000108<p><b>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a>
109and <a href="mailto:vadve@cs.uiuc.edu">Vikram Adve</a></b></p>
110<p> </p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000111</div>
Chris Lattner00950542001-06-06 20:29:01 +0000112<!-- *********************************************************************** -->
Chris Lattner261efe92003-11-25 01:02:51 +0000113<div class="doc_section"> <a name="abstract">Abstract </a></div>
114<!-- *********************************************************************** -->
Misha Brukman9d0919f2003-11-08 01:05:38 +0000115<div class="doc_text">
Chris Lattner261efe92003-11-25 01:02:51 +0000116<p>This document is a reference manual for the LLVM assembly language.
117LLVM is an SSA based representation that provides type safety,
118low-level operations, flexibility, and the capability of representing
119'all' high-level languages cleanly. It is the common code
120representation used throughout all phases of the LLVM compilation
121strategy.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000122</div>
Chris Lattner00950542001-06-06 20:29:01 +0000123<!-- *********************************************************************** -->
Chris Lattner261efe92003-11-25 01:02:51 +0000124<div class="doc_section"> <a name="introduction">Introduction</a> </div>
125<!-- *********************************************************************** -->
Misha Brukman9d0919f2003-11-08 01:05:38 +0000126<div class="doc_text">
Chris Lattner261efe92003-11-25 01:02:51 +0000127<p>The LLVM code representation is designed to be used in three
128different forms: as an in-memory compiler IR, as an on-disk bytecode
129representation (suitable for fast loading by a Just-In-Time compiler),
130and as a human readable assembly language representation. This allows
131LLVM to provide a powerful intermediate representation for efficient
132compiler transformations and analysis, while providing a natural means
133to debug and visualize the transformations. The three different forms
134of LLVM are all equivalent. This document describes the human readable
135representation and notation.</p>
136<p>The LLVM representation aims to be a light-weight and low-level
137while being expressive, typed, and extensible at the same time. It
138aims to be a "universal IR" of sorts, by being at a low enough level
139that high-level ideas may be cleanly mapped to it (similar to how
140microprocessors are "universal IR's", allowing many source languages to
141be mapped to them). By providing type information, LLVM can be used as
142the target of optimizations: for example, through pointer analysis, it
143can be proven that a C automatic variable is never accessed outside of
144the current function... allowing it to be promoted to a simple SSA
145value instead of a memory location.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000146</div>
Chris Lattner00950542001-06-06 20:29:01 +0000147<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000148<div class="doc_subsubsection"> <a name="wellformed">Well-Formedness</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000149<div class="doc_text">
Chris Lattner261efe92003-11-25 01:02:51 +0000150<p>It is important to note that this document describes 'well formed'
151LLVM assembly language. There is a difference between what the parser
152accepts and what is considered 'well formed'. For example, the
153following instruction is syntactically okay, but not well formed:</p>
154<pre> %x = <a href="#i_add">add</a> int 1, %x<br></pre>
155<p>...because the definition of <tt>%x</tt> does not dominate all of
156its uses. The LLVM infrastructure provides a verification pass that may
157be used to verify that an LLVM module is well formed. This pass is
158automatically run by the parser after parsing input assembly, and by
159the optimizer before it outputs bytecode. The violations pointed out
160by the verifier pass indicate bugs in transformation passes or input to
161the parser.</p>
162<!-- Describe the typesetting conventions here. --> </div>
Chris Lattner00950542001-06-06 20:29:01 +0000163<!-- *********************************************************************** -->
Chris Lattner261efe92003-11-25 01:02:51 +0000164<div class="doc_section"> <a name="identifiers">Identifiers</a> </div>
Chris Lattner00950542001-06-06 20:29:01 +0000165<!-- *********************************************************************** -->
Misha Brukman9d0919f2003-11-08 01:05:38 +0000166<div class="doc_text">
Chris Lattner261efe92003-11-25 01:02:51 +0000167<p>LLVM uses three different forms of identifiers, for different
168purposes:</p>
Chris Lattner00950542001-06-06 20:29:01 +0000169<ol>
Chris Lattner261efe92003-11-25 01:02:51 +0000170 <li>Numeric constants are represented as you would expect: 12, -3
171123.421, etc. Floating point constants have an optional hexidecimal
172notation.</li>
173 <li>Named values are represented as a string of characters with a '%'
174prefix. For example, %foo, %DivisionByZero,
175%a.really.long.identifier. The actual regular expression used is '<tt>%[a-zA-Z$._][a-zA-Z$._0-9]*</tt>'.
176Identifiers which require other characters in their names can be
177surrounded with quotes. In this way, anything except a <tt>"</tt>
178character can be used in a name.</li>
179 <li>Unnamed values are represented as an unsigned numeric value with
180a '%' prefix. For example, %12, %2, %44.</li>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000181</ol>
Chris Lattner261efe92003-11-25 01:02:51 +0000182<p>LLVM requires the values start with a '%' sign for two reasons:
183Compilers don't need to worry about name clashes with reserved words,
184and the set of reserved words may be expanded in the future without
185penalty. Additionally, unnamed identifiers allow a compiler to quickly
186come up with a temporary variable without having to avoid symbol table
187conflicts.</p>
188<p>Reserved words in LLVM are very similar to reserved words in other
189languages. There are keywords for different opcodes ('<tt><a
190 href="#i_add">add</a></tt>', '<tt><a href="#i_cast">cast</a></tt>', '<tt><a
191 href="#i_ret">ret</a></tt>', etc...), for primitive type names ('<tt><a
192 href="#t_void">void</a></tt>', '<tt><a href="#t_uint">uint</a></tt>',
193etc...), and others. These reserved words cannot conflict with
194variable names, because none of them start with a '%' character.</p>
195<p>Here is an example of LLVM code to multiply the integer variable '<tt>%X</tt>'
196by 8:</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000197<p>The easy way:</p>
Chris Lattner261efe92003-11-25 01:02:51 +0000198<pre> %result = <a href="#i_mul">mul</a> uint %X, 8<br></pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000199<p>After strength reduction:</p>
Chris Lattner261efe92003-11-25 01:02:51 +0000200<pre> %result = <a href="#i_shl">shl</a> uint %X, ubyte 3<br></pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000201<p>And the hard way:</p>
Chris Lattner261efe92003-11-25 01:02:51 +0000202<pre> <a href="#i_add">add</a> uint %X, %X <i>; yields {uint}:%0</i>
203 <a
204 href="#i_add">add</a> uint %0, %0 <i>; yields {uint}:%1</i>
205 %result = <a
206 href="#i_add">add</a> uint %1, %1<br></pre>
207<p>This last way of multiplying <tt>%X</tt> by 8 illustrates several
208important lexical features of LLVM:</p>
Chris Lattner00950542001-06-06 20:29:01 +0000209<ol>
Chris Lattner261efe92003-11-25 01:02:51 +0000210 <li>Comments are delimited with a '<tt>;</tt>' and go until the end
211of line.</li>
212 <li>Unnamed temporaries are created when the result of a computation
213is not assigned to a named value.</li>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000214 <li>Unnamed temporaries are numbered sequentially</li>
215</ol>
Chris Lattner261efe92003-11-25 01:02:51 +0000216<p>...and it also show a convention that we follow in this document.
217When demonstrating instructions, we will follow an instruction with a
218comment that defines the type and name of value produced. Comments are
219shown in italic text.</p>
220<p>The one non-intuitive notation for constants is the optional
221hexidecimal form of floating point constants. For example, the form '<tt>double
Chris Lattner2b7d3202002-05-06 03:03:22 +00002220x432ff973cafa8000</tt>' is equivalent to (but harder to read than) '<tt>double
Chris Lattner261efe92003-11-25 01:02:51 +00002234.5e+15</tt>' which is also supported by the parser. The only time
224hexadecimal floating point constants are useful (and the only time that
225they are generated by the disassembler) is when an FP constant has to
226be emitted that is not representable as a decimal floating point number
227exactly. For example, NaN's, infinities, and other special cases are
228represented in their IEEE hexadecimal format so that assembly and
229disassembly do not cause any bits to change in the constants.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000230</div>
Chris Lattner00950542001-06-06 20:29:01 +0000231<!-- *********************************************************************** -->
Chris Lattner261efe92003-11-25 01:02:51 +0000232<div class="doc_section"> <a name="typesystem">Type System</a> </div>
233<!-- *********************************************************************** -->
Misha Brukman9d0919f2003-11-08 01:05:38 +0000234<div class="doc_text">
Misha Brukman9d0919f2003-11-08 01:05:38 +0000235<p>The LLVM type system is one of the most important features of the
Chris Lattner261efe92003-11-25 01:02:51 +0000236intermediate representation. Being typed enables a number of
237optimizations to be performed on the IR directly, without having to do
238extra analyses on the side before the transformation. A strong type
239system makes it easier to read the generated code and enables novel
240analyses and transformations that are not feasible to perform on normal
241three address code representations.</p>
Chris Lattner7bae3952002-06-25 18:03:17 +0000242<!-- The written form for the type system was heavily influenced by the
243syntactic problems with types in the C language<sup><a
Chris Lattner261efe92003-11-25 01:02:51 +0000244href="#rw_stroustrup">1</a></sup>.<p> --> </div>
Chris Lattner00950542001-06-06 20:29:01 +0000245<!-- ======================================================================= -->
Chris Lattner261efe92003-11-25 01:02:51 +0000246<div class="doc_subsection"> <a name="t_primitive">Primitive Types</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000247<div class="doc_text">
Chris Lattner261efe92003-11-25 01:02:51 +0000248<p>The primitive types are the fundemental building blocks of the LLVM
249system. The current set of primitive types are as follows:</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000250<p>
251<table border="0" align="center">
Chris Lattner261efe92003-11-25 01:02:51 +0000252 <tbody>
253 <tr>
254 <td>
255 <table border="1" cellspacing="0" cellpadding="4" align="center">
256 <tbody>
257 <tr>
258 <td><tt>void</tt></td>
259 <td>No value</td>
260 </tr>
261 <tr>
262 <td><tt>ubyte</tt></td>
263 <td>Unsigned 8 bit value</td>
264 </tr>
265 <tr>
266 <td><tt>ushort</tt></td>
267 <td>Unsigned 16 bit value</td>
268 </tr>
269 <tr>
270 <td><tt>uint</tt></td>
271 <td>Unsigned 32 bit value</td>
272 </tr>
273 <tr>
274 <td><tt>ulong</tt></td>
275 <td>Unsigned 64 bit value</td>
276 </tr>
277 <tr>
278 <td><tt>float</tt></td>
279 <td>32 bit floating point value</td>
280 </tr>
281 <tr>
282 <td><tt>label</tt></td>
283 <td>Branch destination</td>
284 </tr>
285 </tbody>
286 </table>
287 </td>
288 <td valign="top">
289 <table border="1" cellspacing="0" cellpadding="4" align="center&quot;">
290 <tbody>
291 <tr>
292 <td><tt>bool</tt></td>
293 <td>True or False value</td>
294 </tr>
295 <tr>
296 <td><tt>sbyte</tt></td>
297 <td>Signed 8 bit value</td>
298 </tr>
299 <tr>
300 <td><tt>short</tt></td>
301 <td>Signed 16 bit value</td>
302 </tr>
303 <tr>
304 <td><tt>int</tt></td>
305 <td>Signed 32 bit value</td>
306 </tr>
307 <tr>
308 <td><tt>long</tt></td>
309 <td>Signed 64 bit value</td>
310 </tr>
311 <tr>
312 <td><tt>double</tt></td>
313 <td>64 bit floating point value</td>
314 </tr>
315 </tbody>
316 </table>
317 </td>
318 </tr>
319 </tbody>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000320</table>
321</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000322</div>
Chris Lattner00950542001-06-06 20:29:01 +0000323<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000324<div class="doc_subsubsection"> <a name="t_classifications">Type
325Classifications</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000326<div class="doc_text">
Chris Lattner261efe92003-11-25 01:02:51 +0000327<p>These different primitive types fall into a few useful
328classifications:</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000329<p>
330<table border="1" cellspacing="0" cellpadding="4" align="center">
Chris Lattner261efe92003-11-25 01:02:51 +0000331 <tbody>
332 <tr>
333 <td><a name="t_signed">signed</a></td>
334 <td><tt>sbyte, short, int, long, float, double</tt></td>
335 </tr>
336 <tr>
337 <td><a name="t_unsigned">unsigned</a></td>
338 <td><tt>ubyte, ushort, uint, ulong</tt></td>
339 </tr>
340 <tr>
341 <td><a name="t_integer">integer</a></td>
342 <td><tt>ubyte, sbyte, ushort, short, uint, int, ulong, long</tt></td>
343 </tr>
344 <tr>
345 <td><a name="t_integral">integral</a></td>
346 <td><tt>bool, ubyte, sbyte, ushort, short, uint, int, ulong, long</tt></td>
347 </tr>
348 <tr>
349 <td><a name="t_floating">floating point</a></td>
350 <td><tt>float, double</tt></td>
351 </tr>
352 <tr>
353 <td><a name="t_firstclass">first class</a></td>
354 <td><tt>bool, ubyte, sbyte, ushort, short,<br>
355uint, int, ulong, long, float, double, <a href="#t_pointer">pointer</a></tt></td>
356 </tr>
357 </tbody>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000358</table>
359</p>
Chris Lattner261efe92003-11-25 01:02:51 +0000360<p>The <a href="#t_firstclass">first class</a> types are perhaps the
361most important. Values of these types are the only ones which can be
362produced by instructions, passed as arguments, or used as operands to
363instructions. This means that all structures and arrays must be
364manipulated either by pointer or by component.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000365</div>
Chris Lattner00950542001-06-06 20:29:01 +0000366<!-- ======================================================================= -->
Chris Lattner261efe92003-11-25 01:02:51 +0000367<div class="doc_subsection"> <a name="t_derived">Derived Types</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000368<div class="doc_text">
Chris Lattner261efe92003-11-25 01:02:51 +0000369<p>The real power in LLVM comes from the derived types in the system.
370This is what allows a programmer to represent arrays, functions,
371pointers, and other useful types. Note that these derived types may be
372recursive: For example, it is possible to have a two dimensional array.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000373</div>
Chris Lattner00950542001-06-06 20:29:01 +0000374<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000375<div class="doc_subsubsection"> <a name="t_array">Array Type</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000376<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +0000377<h5>Overview:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000378<p>The array type is a very simple derived type that arranges elements
Chris Lattner261efe92003-11-25 01:02:51 +0000379sequentially in memory. The array type requires a size (number of
380elements) and an underlying data type.</p>
Chris Lattner7faa8832002-04-14 06:13:44 +0000381<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000382<pre> [&lt;# elements&gt; x &lt;elementtype&gt;]<br></pre>
383<p>The number of elements is a constant integer value, elementtype may
384be any type with a size.</p>
Chris Lattner7faa8832002-04-14 06:13:44 +0000385<h5>Examples:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000386<p> <tt>[40 x int ]</tt>: Array of 40 integer values.<br>
387<tt>[41 x int ]</tt>: Array of 41 integer values.<br>
388<tt>[40 x uint]</tt>: Array of 40 unsigned integer values.</p>
389<p> </p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000390<p>Here are some examples of multidimensional arrays:</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000391<p>
392<table border="0" cellpadding="0" cellspacing="0">
Chris Lattner261efe92003-11-25 01:02:51 +0000393 <tbody>
394 <tr>
395 <td><tt>[3 x [4 x int]]</tt></td>
396 <td>: 3x4 array integer values.</td>
397 </tr>
398 <tr>
399 <td><tt>[12 x [10 x float]]</tt></td>
400 <td>: 12x10 array of single precision floating point values.</td>
401 </tr>
402 <tr>
403 <td><tt>[2 x [3 x [4 x uint]]]</tt></td>
404 <td>: 2x3x4 array of unsigned integer values.</td>
405 </tr>
406 </tbody>
Chris Lattner00950542001-06-06 20:29:01 +0000407</table>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000408</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000409</div>
Chris Lattner00950542001-06-06 20:29:01 +0000410<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000411<div class="doc_subsubsection"> <a name="t_function">Function Type</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000412<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +0000413<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000414<p>The function type can be thought of as a function signature. It
415consists of a return type and a list of formal parameter types.
John Criswell009900b2003-11-25 21:45:46 +0000416Function types are usually used to build virtual function tables
Chris Lattner261efe92003-11-25 01:02:51 +0000417(which are structures of pointers to functions), for indirect function
418calls, and when defining a function.</p>
John Criswell009900b2003-11-25 21:45:46 +0000419<p>
420The return type of a function type cannot be an aggregate type.
421</p>
Chris Lattner00950542001-06-06 20:29:01 +0000422<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000423<pre> &lt;returntype&gt; (&lt;parameter list&gt;)<br></pre>
424<p>Where '<tt>&lt;parameter list&gt;</tt>' is a comma-separated list of
425type specifiers. Optionally, the parameter list may include a type <tt>...</tt>,
Chris Lattner27f71f22003-09-03 00:41:47 +0000426which indicates that the function takes a variable number of arguments.
427Variable argument functions can access their arguments with the <a
Chris Lattner261efe92003-11-25 01:02:51 +0000428 href="#int_varargs">variable argument handling intrinsic</a> functions.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000429<h5>Examples:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000430<p>
431<table border="0" cellpadding="0" cellspacing="0">
Chris Lattner261efe92003-11-25 01:02:51 +0000432 <tbody>
433 <tr>
434 <td><tt>int (int)</tt></td>
435 <td>: function taking an <tt>int</tt>, returning an <tt>int</tt></td>
436 </tr>
437 <tr>
438 <td><tt>float (int, int *) *</tt></td>
439 <td>: <a href="#t_pointer">Pointer</a> to a function that takes
440an <tt>int</tt> and a <a href="#t_pointer">pointer</a> to <tt>int</tt>,
441returning <tt>float</tt>.</td>
442 </tr>
443 <tr>
444 <td><tt>int (sbyte *, ...)</tt></td>
445 <td>: A vararg function that takes at least one <a
446 href="#t_pointer">pointer</a> to <tt>sbyte</tt> (signed char in C),
447which returns an integer. This is the signature for <tt>printf</tt>
448in LLVM.</td>
449 </tr>
450 </tbody>
Chris Lattner00950542001-06-06 20:29:01 +0000451</table>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000452</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000453</div>
Chris Lattner00950542001-06-06 20:29:01 +0000454<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000455<div class="doc_subsubsection"> <a name="t_struct">Structure Type</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000456<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +0000457<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000458<p>The structure type is used to represent a collection of data members
459together in memory. The packing of the field types is defined to match
460the ABI of the underlying processor. The elements of a structure may
461be any type that has a size.</p>
462<p>Structures are accessed using '<tt><a href="#i_load">load</a></tt>
463and '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a
464field with the '<tt><a href="#i_getelementptr">getelementptr</a></tt>'
465instruction.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000466<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000467<pre> { &lt;type list&gt; }<br></pre>
Chris Lattner00950542001-06-06 20:29:01 +0000468<h5>Examples:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000469<p>
470<table border="0" cellpadding="0" cellspacing="0">
Chris Lattner261efe92003-11-25 01:02:51 +0000471 <tbody>
472 <tr>
473 <td><tt>{ int, int, int }</tt></td>
474 <td>: a triple of three <tt>int</tt> values</td>
475 </tr>
476 <tr>
477 <td><tt>{ float, int (int) * }</tt></td>
478 <td>: A pair, where the first element is a <tt>float</tt> and the
479second element is a <a href="#t_pointer">pointer</a> to a <a
480 href="t_function">function</a> that takes an <tt>int</tt>, returning
481an <tt>int</tt>.</td>
482 </tr>
483 </tbody>
Chris Lattner00950542001-06-06 20:29:01 +0000484</table>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000485</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000486</div>
Chris Lattner00950542001-06-06 20:29:01 +0000487<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000488<div class="doc_subsubsection"> <a name="t_pointer">Pointer Type</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000489<div class="doc_text">
Chris Lattner7faa8832002-04-14 06:13:44 +0000490<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000491<p>As in many languages, the pointer type represents a pointer or
492reference to another object, which must live in memory.</p>
Chris Lattner7faa8832002-04-14 06:13:44 +0000493<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000494<pre> &lt;type&gt; *<br></pre>
Chris Lattner7faa8832002-04-14 06:13:44 +0000495<h5>Examples:</h5>
Chris Lattner7faa8832002-04-14 06:13:44 +0000496<p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000497<table border="0" cellpadding="0" cellspacing="0">
Chris Lattner261efe92003-11-25 01:02:51 +0000498 <tbody>
499 <tr>
500 <td><tt>[4x int]*</tt></td>
501 <td>: <a href="#t_pointer">pointer</a> to <a href="#t_array">array</a>
502of four <tt>int</tt> values</td>
503 </tr>
504 <tr>
505 <td><tt>int (int *) *</tt></td>
506 <td>: A <a href="#t_pointer">pointer</a> to a <a
507 href="t_function">function</a> that takes an <tt>int</tt>, returning
508an <tt>int</tt>.</td>
509 </tr>
510 </tbody>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000511</table>
512</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000513</div>
Chris Lattner261efe92003-11-25 01:02:51 +0000514<!-- _______________________________________________________________________ --><!--
Misha Brukman9d0919f2003-11-08 01:05:38 +0000515<div class="doc_subsubsection">
516 <a name="t_packed">Packed Type</a>
517</div>
518
519<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +0000520
521Mention/decide that packed types work with saturation or not. Maybe have a packed+saturated type in addition to just a packed type.<p>
522
523Packed types should be 'nonsaturated' because standard data types are not saturated. Maybe have a saturated packed type?<p>
524
Misha Brukman9d0919f2003-11-08 01:05:38 +0000525</div>
526
Chris Lattner261efe92003-11-25 01:02:51 +0000527--><!-- *********************************************************************** -->
528<div class="doc_section"> <a name="highlevel">High Level Structure</a> </div>
529<!-- *********************************************************************** --><!-- ======================================================================= -->
530<div class="doc_subsection"> <a name="modulestructure">Module Structure</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000531<div class="doc_text">
Chris Lattner261efe92003-11-25 01:02:51 +0000532<p>LLVM programs are composed of "Module"s, each of which is a
533translation unit of the input programs. Each module consists of
534functions, global variables, and symbol table entries. Modules may be
535combined together with the LLVM linker, which merges function (and
536global variable) definitions, resolves forward declarations, and merges
537symbol table entries. Here is an example of the "hello world" module:</p>
538<pre><i>; Declare the string constant as a global constant...</i>
539<a href="#identifiers">%.LC0</a> = <a href="#linkage_internal">internal</a> <a
540 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 +0000541
Chris Lattner27f71f22003-09-03 00:41:47 +0000542<i>; External declaration of the puts function</i>
543<a href="#functionstructure">declare</a> int %puts(sbyte*) <i>; int(sbyte*)* </i>
Chris Lattner2b7d3202002-05-06 03:03:22 +0000544
545<i>; Definition of main function</i>
Chris Lattner27f71f22003-09-03 00:41:47 +0000546int %main() { <i>; int()* </i>
Chris Lattner2b7d3202002-05-06 03:03:22 +0000547 <i>; Convert [13x sbyte]* to sbyte *...</i>
Chris Lattner261efe92003-11-25 01:02:51 +0000548 %cast210 = <a
549 href="#i_getelementptr">getelementptr</a> [13 x sbyte]* %.LC0, long 0, long 0 <i>; sbyte*</i>
Chris Lattner2b7d3202002-05-06 03:03:22 +0000550
551 <i>; Call puts function to write out the string to stdout...</i>
Chris Lattner261efe92003-11-25 01:02:51 +0000552 <a
553 href="#i_call">call</a> int %puts(sbyte* %cast210) <i>; int</i>
554 <a
555 href="#i_ret">ret</a> int 0<br>}<br></pre>
556<p>This example is made up of a <a href="#globalvars">global variable</a>
557named "<tt>.LC0</tt>", an external declaration of the "<tt>puts</tt>"
558function, and a <a href="#functionstructure">function definition</a>
559for "<tt>main</tt>".</p>
560<a name="linkage"> In general, a module is made up of a list of global
561values, where both functions and global variables are global values.
562Global values are represented by a pointer to a memory location (in
563this case, a pointer to an array of char, and a pointer to a function),
564and have one of the following linkage types:</a>
565<p> </p>
Chris Lattner27f71f22003-09-03 00:41:47 +0000566<dl>
Chris Lattner261efe92003-11-25 01:02:51 +0000567 <a name="linkage_internal"> <dt><tt><b>internal</b></tt> </dt>
568 <dd>Global values with internal linkage are only directly accessible
569by objects in the current module. In particular, linking code into a
570module with an internal global value may cause the internal to be
571renamed as necessary to avoid collisions. Because the symbol is
572internal to the module, all references can be updated. This
573corresponds to the notion of the '<tt>static</tt>' keyword in C, or the
574idea of "anonymous namespaces" in C++.
575 <p> </p>
576 </dd>
577 </a><a name="linkage_linkonce"> <dt><tt><b>linkonce</b></tt>: </dt>
578 <dd>"<tt>linkonce</tt>" linkage is similar to <tt>internal</tt>
579linkage, with the twist that linking together two modules defining the
580same <tt>linkonce</tt> globals will cause one of the globals to be
581discarded. This is typically used to implement inline functions.
582Unreferenced <tt>linkonce</tt> globals are allowed to be discarded.
583 <p> </p>
584 </dd>
585 </a><a name="linkage_weak"> <dt><tt><b>weak</b></tt>: </dt>
586 <dd>"<tt>weak</tt>" linkage is exactly the same as <tt>linkonce</tt>
587linkage, except that unreferenced <tt>weak</tt> globals may not be
588discarded. This is used to implement constructs in C such as "<tt>int
589X;</tt>" at global scope.
590 <p> </p>
591 </dd>
592 </a><a name="linkage_appending"> <dt><tt><b>appending</b></tt>: </dt>
593 <dd>"<tt>appending</tt>" linkage may only be applied to global
594variables of pointer to array type. When two global variables with
595appending linkage are linked together, the two global arrays are
596appended together. This is the LLVM, typesafe, equivalent of having
597the system linker append together "sections" with identical names when
598.o files are linked.
599 <p> </p>
600 </dd>
601 </a><a name="linkage_external"> <dt><tt><b>externally visible</b></tt>:</dt>
602 <dd>If none of the above identifiers are used, the global is
603externally visible, meaning that it participates in linkage and can be
604used to resolve external symbol references.
605 <p> </p>
606 </dd>
607 </a>
608</dl>
609<p> </p>
610<p><a name="linkage_external">For example, since the "<tt>.LC0</tt>"
611variable is defined to be internal, if another module defined a "<tt>.LC0</tt>"
612variable and was linked with this one, one of the two would be renamed,
613preventing a collision. Since "<tt>main</tt>" and "<tt>puts</tt>" are
614external (i.e., lacking any linkage declarations), they are accessible
615outside of the current module. It is illegal for a function <i>declaration</i>
616to have any linkage type other than "externally visible".</a></p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000617</div>
Chris Lattner00950542001-06-06 20:29:01 +0000618<!-- ======================================================================= -->
Chris Lattner261efe92003-11-25 01:02:51 +0000619<div class="doc_subsection"> <a name="globalvars">Global Variables</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000620<div class="doc_text">
Chris Lattner261efe92003-11-25 01:02:51 +0000621<p>Global variables define regions of memory allocated at compilation
622time instead of run-time. Global variables may optionally be
623initialized. A variable may be defined as a global "constant", which
624indicates that the contents of the variable will never be modified
625(opening options for optimization). Constants must always have an
626initial value.</p>
627<p>As SSA values, global variables define pointer values that are in
628scope (i.e. they dominate) for all basic blocks in the program. Global
629variables always define a pointer to their "content" type because they
630describe a region of memory, and all memory objects in LLVM are
631accessed through pointers.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000632</div>
Chris Lattner2b7d3202002-05-06 03:03:22 +0000633<!-- ======================================================================= -->
Chris Lattner261efe92003-11-25 01:02:51 +0000634<div class="doc_subsection"> <a name="functionstructure">Functions</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000635<div class="doc_text">
Chris Lattner261efe92003-11-25 01:02:51 +0000636<p>LLVM function definitions are composed of a (possibly empty)
637argument list, an opening curly brace, a list of basic blocks, and a
638closing curly brace. LLVM function declarations are defined with the "<tt>declare</tt>"
639keyword, a function name, and a function signature.</p>
640<p>A function definition contains a list of basic blocks, forming the
641CFG for the function. Each basic block may optionally start with a
642label (giving the basic block a symbol table entry), contains a list of
643instructions, and ends with a <a href="#terminators">terminator</a>
644instruction (such as a branch or function return).</p>
645<p>The first basic block in program is special in two ways: it is
646immediately executed on entrance to the function, and it is not allowed
647to have predecessor basic blocks (i.e. there can not be any branches to
648the entry block of a function). Because the block can have no
649predecessors, it also cannot have any <a href="#i_phi">PHI nodes</a>.</p>
John Criswell009900b2003-11-25 21:45:46 +0000650<p>
651LLVM functions are identified by their name and type signature. Hence, two
652functions with the same name but different parameter lists or return values
653are considered different functions, and LLVM will resolves references to each
654appropriately.
655</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000656</div>
Chris Lattner00950542001-06-06 20:29:01 +0000657<!-- *********************************************************************** -->
Chris Lattner261efe92003-11-25 01:02:51 +0000658<div class="doc_section"> <a name="instref">Instruction Reference</a> </div>
659<!-- *********************************************************************** -->
Misha Brukman9d0919f2003-11-08 01:05:38 +0000660<div class="doc_text">
Chris Lattner261efe92003-11-25 01:02:51 +0000661<p>The LLVM instruction set consists of several different
662classifications of instructions: <a href="#terminators">terminator
663instructions</a>, <a href="#binaryops">binary instructions</a>, <a
664 href="#memoryops">memory instructions</a>, and <a href="#otherops">other
665instructions</a>.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000666</div>
Chris Lattner00950542001-06-06 20:29:01 +0000667<!-- ======================================================================= -->
Chris Lattner261efe92003-11-25 01:02:51 +0000668<div class="doc_subsection"> <a name="terminators">Terminator
669Instructions</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000670<div class="doc_text">
Chris Lattner261efe92003-11-25 01:02:51 +0000671<p>As mentioned <a href="#functionstructure">previously</a>, every
672basic block in a program ends with a "Terminator" instruction, which
673indicates which block should be executed after the current block is
674finished. These terminator instructions typically yield a '<tt>void</tt>'
675value: they produce control flow, not values (the one exception being
676the '<a href="#i_invoke"><tt>invoke</tt></a>' instruction).</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000677<p>There are five different terminator instructions: the '<a
Chris Lattner261efe92003-11-25 01:02:51 +0000678 href="#i_ret"><tt>ret</tt></a>' instruction, the '<a href="#i_br"><tt>br</tt></a>'
679instruction, the '<a href="#i_switch"><tt>switch</tt></a>' instruction,
680the '<a href="#i_invoke"><tt>invoke</tt></a>' instruction, and the '<a
681 href="#i_unwind"><tt>unwind</tt></a>' instruction.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000682</div>
Chris Lattner00950542001-06-06 20:29:01 +0000683<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000684<div class="doc_subsubsection"> <a name="i_ret">'<tt>ret</tt>'
685Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000686<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +0000687<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000688<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 +0000689 ret void <i>; Return from void function</i>
Chris Lattner00950542001-06-06 20:29:01 +0000690</pre>
Chris Lattner00950542001-06-06 20:29:01 +0000691<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000692<p>The '<tt>ret</tt>' instruction is used to return control flow (and a
693value) from a function, back to the caller.</p>
694<p>There are two forms of the '<tt>ret</tt>' instructruction: one that
695returns a value and then causes control flow, and one that just causes
696control flow to occur.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000697<h5>Arguments:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000698<p>The '<tt>ret</tt>' instruction may return any '<a
699 href="#t_firstclass">first class</a>' type. Notice that a function is
700not <a href="#wellformed">well formed</a> if there exists a '<tt>ret</tt>'
701instruction inside of the function that returns a value that does not
702match the return type of the function.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000703<h5>Semantics:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000704<p>When the '<tt>ret</tt>' instruction is executed, control flow
705returns back to the calling function's context. If the caller is a "<a
706 href="#i_call"><tt>call</tt></a> instruction, execution continues at
707the instruction after the call. If the caller was an "<a
708 href="#i_invoke"><tt>invoke</tt></a>" instruction, execution continues
709at the beginning "normal" of the destination block. If the instruction
710returns a value, that value shall set the call or invoke instruction's
711return value.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000712<h5>Example:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000713<pre> ret int 5 <i>; Return an integer value of 5</i>
Chris Lattner7faa8832002-04-14 06:13:44 +0000714 ret void <i>; Return from a void function</i>
Chris Lattner00950542001-06-06 20:29:01 +0000715</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000716</div>
Chris Lattner00950542001-06-06 20:29:01 +0000717<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000718<div class="doc_subsubsection"> <a name="i_br">'<tt>br</tt>' Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000719<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +0000720<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000721<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 +0000722</pre>
Chris Lattner00950542001-06-06 20:29:01 +0000723<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000724<p>The '<tt>br</tt>' instruction is used to cause control flow to
725transfer to a different basic block in the current function. There are
726two forms of this instruction, corresponding to a conditional branch
727and an unconditional branch.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000728<h5>Arguments:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000729<p>The conditional branch form of the '<tt>br</tt>' instruction takes a
730single '<tt>bool</tt>' value and two '<tt>label</tt>' values. The
731unconditional form of the '<tt>br</tt>' instruction takes a single '<tt>label</tt>'
732value as a target.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000733<h5>Semantics:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000734<p>Upon execution of a conditional '<tt>br</tt>' instruction, the '<tt>bool</tt>'
735argument is evaluated. If the value is <tt>true</tt>, control flows
736to the '<tt>iftrue</tt>' <tt>label</tt> argument. If "cond" is <tt>false</tt>,
737control flows to the '<tt>iffalse</tt>' <tt>label</tt> argument.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000738<h5>Example:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000739<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
740 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 +0000741</div>
Chris Lattner00950542001-06-06 20:29:01 +0000742<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000743<div class="doc_subsubsection"> <a name="i_switch">'<tt>switch</tt>'
744Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000745<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +0000746<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000747<pre> switch uint &lt;value&gt;, label &lt;defaultdest&gt; [ int &lt;val&gt;, label &amp;dest&gt;, ... ]<br></pre>
Chris Lattner00950542001-06-06 20:29:01 +0000748<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000749<p>The '<tt>switch</tt>' instruction is used to transfer control flow
750to one of several different places. It is a generalization of the '<tt>br</tt>'
Misha Brukman9d0919f2003-11-08 01:05:38 +0000751instruction, allowing a branch to occur to one of many possible
752destinations.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000753<h5>Arguments:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000754<p>The '<tt>switch</tt>' instruction uses three parameters: a '<tt>uint</tt>'
Chris Lattner261efe92003-11-25 01:02:51 +0000755comparison value '<tt>value</tt>', a default '<tt>label</tt>'
756destination, and an array of pairs of comparison value constants and '<tt>label</tt>'s.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000757<h5>Semantics:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000758<p>The <tt>switch</tt> instruction specifies a table of values and
759destinations. When the '<tt>switch</tt>' instruction is executed, this
760table is searched for the given value. If the value is found, the
761corresponding destination is branched to, otherwise the default value
762it transfered to.</p>
Chris Lattnerc29b1252003-05-08 05:08:48 +0000763<h5>Implementation:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000764<p>Depending on properties of the target machine and the particular <tt>switch</tt>
765instruction, this instruction may be code generated as a series of
766chained conditional branches, or with a lookup table.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000767<h5>Example:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000768<pre> <i>; Emulate a conditional br instruction</i>
769 %Val = <a
770 href="#i_cast">cast</a> bool %value to uint<br> switch uint %Val, label %truedest [int 0, label %falsedest ]<br><br> <i>; Emulate an unconditional br instruction</i>
Chris Lattner27f71f22003-09-03 00:41:47 +0000771 switch uint 0, label %dest [ ]
Chris Lattner00950542001-06-06 20:29:01 +0000772
Chris Lattner2b7d3202002-05-06 03:03:22 +0000773 <i>; Implement a jump table:</i>
Chris Lattner27f71f22003-09-03 00:41:47 +0000774 switch uint %val, label %otherwise [ int 0, label %onzero,
775 int 1, label %onone,
776 int 2, label %ontwo ]
Chris Lattner00950542001-06-06 20:29:01 +0000777</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000778</div>
Chris Lattner00950542001-06-06 20:29:01 +0000779<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000780<div class="doc_subsubsection"> <a name="i_invoke">'<tt>invoke</tt>'
781Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000782<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +0000783<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000784<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 +0000785<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000786<p>The '<tt>invoke</tt>' instruction causes control to transfer to a
787specified function, with the possibility of control flow transfer to
788either the '<tt>normal</tt>' <tt>label</tt> label or the '<tt>exception</tt>'<tt>label</tt>.
789If the callee function returns with the "<tt><a href="#i_ret">ret</a></tt>"
790instruction, control flow will return to the "normal" label. If the
791callee (or any indirect callees) returns with the "<a href="#i_unwind"><tt>unwind</tt></a>"
792instruction, control is interrupted, and continued at the dynamically
793nearest "except" label.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000794<h5>Arguments:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000795<p>This instruction requires several arguments:</p>
Chris Lattner00950542001-06-06 20:29:01 +0000796<ol>
Chris Lattner261efe92003-11-25 01:02:51 +0000797 <li>'<tt>ptr to function ty</tt>': shall be the signature of the
798pointer to function value being invoked. In most cases, this is a
799direct function invocation, but indirect <tt>invoke</tt>s are just as
800possible, branching off an arbitrary pointer to function value. </li>
801 <li>'<tt>function ptr val</tt>': An LLVM value containing a pointer
802to a function to be invoked. </li>
803 <li>'<tt>function args</tt>': argument list whose types match the
804function signature argument types. If the function signature indicates
805the function accepts a variable number of arguments, the extra
806arguments can be specified. </li>
807 <li>'<tt>normal label</tt>': the label reached when the called
808function executes a '<tt><a href="#i_ret">ret</a></tt>' instruction. </li>
809 <li>'<tt>exception label</tt>': the label reached when a callee
810returns with the <a href="#i_unwind"><tt>unwind</tt></a> instruction. </li>
Chris Lattner00950542001-06-06 20:29:01 +0000811</ol>
Chris Lattner00950542001-06-06 20:29:01 +0000812<h5>Semantics:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000813<p>This instruction is designed to operate as a standard '<tt><a
Chris Lattner261efe92003-11-25 01:02:51 +0000814 href="#i_call">call</a></tt>' instruction in most regards. The
815primary difference is that it establishes an association with a label,
816which is used by the runtime library to unwind the stack.</p>
817<p>This instruction is used in languages with destructors to ensure
818that proper cleanup is performed in the case of either a <tt>longjmp</tt>
819or a thrown exception. Additionally, this is important for
820implementation of '<tt>catch</tt>' clauses in high-level languages that
821support them.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000822<h5>Example:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000823<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 +0000824</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000825</div>
Chris Lattner27f71f22003-09-03 00:41:47 +0000826<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000827<div class="doc_subsubsection"> <a name="i_unwind">'<tt>unwind</tt>'
828Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000829<div class="doc_text">
Chris Lattner27f71f22003-09-03 00:41:47 +0000830<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000831<pre> unwind<br></pre>
Chris Lattner27f71f22003-09-03 00:41:47 +0000832<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000833<p>The '<tt>unwind</tt>' instruction unwinds the stack, continuing
834control flow at the first callee in the dynamic call stack which used
835an <a href="#i_invoke"><tt>invoke</tt></a> instruction to perform the
836call. This is primarily used to implement exception handling.</p>
Chris Lattner27f71f22003-09-03 00:41:47 +0000837<h5>Semantics:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000838<p>The '<tt>unwind</tt>' intrinsic causes execution of the current
839function to immediately halt. The dynamic call stack is then searched
840for the first <a href="#i_invoke"><tt>invoke</tt></a> instruction on
841the call stack. Once found, execution continues at the "exceptional"
842destination block specified by the <tt>invoke</tt> instruction. If
843there is no <tt>invoke</tt> instruction in the dynamic call chain,
844undefined behavior results.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000845</div>
Chris Lattner00950542001-06-06 20:29:01 +0000846<!-- ======================================================================= -->
Chris Lattner261efe92003-11-25 01:02:51 +0000847<div class="doc_subsection"> <a name="binaryops">Binary Operations</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000848<div class="doc_text">
Chris Lattner261efe92003-11-25 01:02:51 +0000849<p>Binary operators are used to do most of the computation in a
850program. They require two operands, execute an operation on them, and
851produce a single value. The result value of a binary operator is not
852necessarily the same type as its operands.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000853<p>There are several different binary operators:</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000854</div>
Chris Lattner00950542001-06-06 20:29:01 +0000855<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000856<div class="doc_subsubsection"> <a name="i_add">'<tt>add</tt>'
857Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000858<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +0000859<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000860<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 +0000861</pre>
Chris Lattner00950542001-06-06 20:29:01 +0000862<h5>Overview:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000863<p>The '<tt>add</tt>' instruction returns the sum of its two operands.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000864<h5>Arguments:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000865<p>The two arguments to the '<tt>add</tt>' instruction must be either <a
Chris Lattner261efe92003-11-25 01:02:51 +0000866 href="#t_integer">integer</a> or <a href="#t_floating">floating point</a>
867values. Both arguments must have identical types.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000868<h5>Semantics:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000869<p>The value produced is the integer or floating point sum of the two
870operands.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000871<h5>Example:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000872<pre> &lt;result&gt; = add int 4, %var <i>; yields {int}:result = 4 + %var</i>
Chris Lattner00950542001-06-06 20:29:01 +0000873</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000874</div>
Chris Lattner00950542001-06-06 20:29:01 +0000875<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000876<div class="doc_subsubsection"> <a name="i_sub">'<tt>sub</tt>'
877Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000878<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +0000879<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000880<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 +0000881</pre>
Chris Lattner00950542001-06-06 20:29:01 +0000882<h5>Overview:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000883<p>The '<tt>sub</tt>' instruction returns the difference of its two
884operands.</p>
Chris Lattner261efe92003-11-25 01:02:51 +0000885<p>Note that the '<tt>sub</tt>' instruction is used to represent the '<tt>neg</tt>'
886instruction present in most other intermediate representations.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000887<h5>Arguments:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000888<p>The two arguments to the '<tt>sub</tt>' instruction must be either <a
Chris Lattner261efe92003-11-25 01:02:51 +0000889 href="#t_integer">integer</a> or <a href="#t_floating">floating point</a>
890values. Both arguments must have identical types.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000891<h5>Semantics:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000892<p>The value produced is the integer or floating point difference of
893the two operands.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000894<h5>Example:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000895<pre> &lt;result&gt; = sub int 4, %var <i>; yields {int}:result = 4 - %var</i>
Chris Lattner00950542001-06-06 20:29:01 +0000896 &lt;result&gt; = sub int 0, %val <i>; yields {int}:result = -%var</i>
897</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000898</div>
Chris Lattner00950542001-06-06 20:29:01 +0000899<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000900<div class="doc_subsubsection"> <a name="i_mul">'<tt>mul</tt>'
901Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000902<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +0000903<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000904<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 +0000905</pre>
Chris Lattner00950542001-06-06 20:29:01 +0000906<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000907<p>The '<tt>mul</tt>' instruction returns the product of its two
908operands.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000909<h5>Arguments:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000910<p>The two arguments to the '<tt>mul</tt>' instruction must be either <a
Chris Lattner261efe92003-11-25 01:02:51 +0000911 href="#t_integer">integer</a> or <a href="#t_floating">floating point</a>
912values. Both arguments must have identical types.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000913<h5>Semantics:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000914<p>The value produced is the integer or floating point product of the
Misha Brukman9d0919f2003-11-08 01:05:38 +0000915two operands.</p>
Chris Lattner261efe92003-11-25 01:02:51 +0000916<p>There is no signed vs unsigned multiplication. The appropriate
917action is taken based on the type of the operand.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000918<h5>Example:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000919<pre> &lt;result&gt; = mul int 4, %var <i>; yields {int}:result = 4 * %var</i>
Chris Lattner00950542001-06-06 20:29:01 +0000920</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000921</div>
Chris Lattner00950542001-06-06 20:29:01 +0000922<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000923<div class="doc_subsubsection"> <a name="i_div">'<tt>div</tt>'
924Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000925<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +0000926<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000927<pre> &lt;result&gt; = div &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
928</pre>
929<h5>Overview:</h5>
930<p>The '<tt>div</tt>' instruction returns the quotient of its two
931operands.</p>
932<h5>Arguments:</h5>
933<p>The two arguments to the '<tt>div</tt>' instruction must be either <a
934 href="#t_integer">integer</a> or <a href="#t_floating">floating point</a>
935values. Both arguments must have identical types.</p>
936<h5>Semantics:</h5>
937<p>The value produced is the integer or floating point quotient of the
938two operands.</p>
939<h5>Example:</h5>
940<pre> &lt;result&gt; = div int 4, %var <i>; yields {int}:result = 4 / %var</i>
941</pre>
942</div>
943<!-- _______________________________________________________________________ -->
944<div class="doc_subsubsection"> <a name="i_rem">'<tt>rem</tt>'
945Instruction</a> </div>
946<div class="doc_text">
947<h5>Syntax:</h5>
948<pre> &lt;result&gt; = rem &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
949</pre>
950<h5>Overview:</h5>
951<p>The '<tt>rem</tt>' instruction returns the remainder from the
952division of its two operands.</p>
953<h5>Arguments:</h5>
954<p>The two arguments to the '<tt>rem</tt>' instruction must be either <a
955 href="#t_integer">integer</a> or <a href="#t_floating">floating point</a>
956values. Both arguments must have identical types.</p>
957<h5>Semantics:</h5>
958<p>This returns the <i>remainder</i> of a division (where the result
959has the same sign as the divisor), not the <i>modulus</i> (where the
960result has the same sign as the dividend) of a value. For more
961information about the difference, see: <a
962 href="http://mathforum.org/dr.math/problems/anne.4.28.99.html">The
963Math Forum</a>.</p>
964<h5>Example:</h5>
965<pre> &lt;result&gt; = rem int 4, %var <i>; yields {int}:result = 4 % %var</i>
966</pre>
967</div>
968<!-- _______________________________________________________________________ -->
969<div class="doc_subsubsection"> <a name="i_setcc">'<tt>set<i>cc</i></tt>'
970Instructions</a> </div>
971<div class="doc_text">
972<h5>Syntax:</h5>
973<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 +0000974 &lt;result&gt; = setne &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {bool}:result</i>
975 &lt;result&gt; = setlt &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {bool}:result</i>
976 &lt;result&gt; = setgt &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {bool}:result</i>
977 &lt;result&gt; = setle &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {bool}:result</i>
978 &lt;result&gt; = setge &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {bool}:result</i>
979</pre>
Chris Lattner261efe92003-11-25 01:02:51 +0000980<h5>Overview:</h5>
981<p>The '<tt>set<i>cc</i></tt>' family of instructions returns a boolean
982value based on a comparison of their two operands.</p>
983<h5>Arguments:</h5>
984<p>The two arguments to the '<tt>set<i>cc</i></tt>' instructions must
985be of <a href="#t_firstclass">first class</a> type (it is not possible
986to compare '<tt>label</tt>'s, '<tt>array</tt>'s, '<tt>structure</tt>'
987or '<tt>void</tt>' values, etc...). Both arguments must have identical
988types.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000989<h5>Semantics:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000990<p>The '<tt>seteq</tt>' instruction yields a <tt>true</tt> '<tt>bool</tt>'
991value if both operands are equal.<br>
992The '<tt>setne</tt>' instruction yields a <tt>true</tt> '<tt>bool</tt>'
993value if both operands are unequal.<br>
994The '<tt>setlt</tt>' instruction yields a <tt>true</tt> '<tt>bool</tt>'
995value if the first operand is less than the second operand.<br>
996The '<tt>setgt</tt>' instruction yields a <tt>true</tt> '<tt>bool</tt>'
997value if the first operand is greater than the second operand.<br>
998The '<tt>setle</tt>' instruction yields a <tt>true</tt> '<tt>bool</tt>'
999value if the first operand is less than or equal to the second operand.<br>
1000The '<tt>setge</tt>' instruction yields a <tt>true</tt> '<tt>bool</tt>'
1001value if the first operand is greater than or equal to the second
1002operand.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001003<h5>Example:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001004<pre> &lt;result&gt; = seteq int 4, 5 <i>; yields {bool}:result = false</i>
Chris Lattner00950542001-06-06 20:29:01 +00001005 &lt;result&gt; = setne float 4, 5 <i>; yields {bool}:result = true</i>
1006 &lt;result&gt; = setlt uint 4, 5 <i>; yields {bool}:result = true</i>
1007 &lt;result&gt; = setgt sbyte 4, 5 <i>; yields {bool}:result = false</i>
1008 &lt;result&gt; = setle sbyte 4, 5 <i>; yields {bool}:result = true</i>
1009 &lt;result&gt; = setge sbyte 4, 5 <i>; yields {bool}:result = false</i>
1010</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001011</div>
Chris Lattner00950542001-06-06 20:29:01 +00001012<!-- ======================================================================= -->
Chris Lattner261efe92003-11-25 01:02:51 +00001013<div class="doc_subsection"> <a name="bitwiseops">Bitwise Binary
1014Operations</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001015<div class="doc_text">
Chris Lattner261efe92003-11-25 01:02:51 +00001016<p>Bitwise binary operators are used to do various forms of
1017bit-twiddling in a program. They are generally very efficient
1018instructions, and can commonly be strength reduced from other
1019instructions. They require two operands, execute an operation on them,
1020and produce a single value. The resulting value of the bitwise binary
1021operators is always the same type as its first operand.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001022</div>
Chris Lattner00950542001-06-06 20:29:01 +00001023<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +00001024<div class="doc_subsubsection"> <a name="i_and">'<tt>and</tt>'
1025Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001026<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +00001027<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001028<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 +00001029</pre>
Chris Lattner00950542001-06-06 20:29:01 +00001030<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001031<p>The '<tt>and</tt>' instruction returns the bitwise logical and of
1032its two operands.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001033<h5>Arguments:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001034<p>The two arguments to the '<tt>and</tt>' instruction must be <a
Chris Lattner261efe92003-11-25 01:02:51 +00001035 href="#t_integral">integral</a> values. Both arguments must have
1036identical types.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001037<h5>Semantics:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001038<p>The truth table used for the '<tt>and</tt>' instruction is:</p>
Chris Lattner261efe92003-11-25 01:02:51 +00001039<p> </p>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001040<center>
1041<table border="1" cellspacing="0" cellpadding="4">
Chris Lattner261efe92003-11-25 01:02:51 +00001042 <tbody>
1043 <tr>
1044 <td>In0</td>
1045 <td>In1</td>
1046 <td>Out</td>
1047 </tr>
1048 <tr>
1049 <td>0</td>
1050 <td>0</td>
1051 <td>0</td>
1052 </tr>
1053 <tr>
1054 <td>0</td>
1055 <td>1</td>
1056 <td>0</td>
1057 </tr>
1058 <tr>
1059 <td>1</td>
1060 <td>0</td>
1061 <td>0</td>
1062 </tr>
1063 <tr>
1064 <td>1</td>
1065 <td>1</td>
1066 <td>1</td>
1067 </tr>
1068 </tbody>
1069</table>
1070</center>
Chris Lattner00950542001-06-06 20:29:01 +00001071<h5>Example:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001072<pre> &lt;result&gt; = and int 4, %var <i>; yields {int}:result = 4 &amp; %var</i>
Chris Lattner00950542001-06-06 20:29:01 +00001073 &lt;result&gt; = and int 15, 40 <i>; yields {int}:result = 8</i>
1074 &lt;result&gt; = and int 4, 8 <i>; yields {int}:result = 0</i>
1075</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001076</div>
Chris Lattner00950542001-06-06 20:29:01 +00001077<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +00001078<div class="doc_subsubsection"> <a name="i_or">'<tt>or</tt>' Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001079<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +00001080<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001081<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 +00001082</pre>
Chris Lattner261efe92003-11-25 01:02:51 +00001083<h5>Overview:</h5>
1084<p>The '<tt>or</tt>' instruction returns the bitwise logical inclusive
1085or of its two operands.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001086<h5>Arguments:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001087<p>The two arguments to the '<tt>or</tt>' instruction must be <a
Chris Lattner261efe92003-11-25 01:02:51 +00001088 href="#t_integral">integral</a> values. Both arguments must have
1089identical types.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001090<h5>Semantics:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001091<p>The truth table used for the '<tt>or</tt>' instruction is:</p>
Chris Lattner261efe92003-11-25 01:02:51 +00001092<p> </p>
1093<center>
1094<table border="1" cellspacing="0" cellpadding="4">
1095 <tbody>
1096 <tr>
1097 <td>In0</td>
1098 <td>In1</td>
1099 <td>Out</td>
1100 </tr>
1101 <tr>
1102 <td>0</td>
1103 <td>0</td>
1104 <td>0</td>
1105 </tr>
1106 <tr>
1107 <td>0</td>
1108 <td>1</td>
1109 <td>1</td>
1110 </tr>
1111 <tr>
1112 <td>1</td>
1113 <td>0</td>
1114 <td>1</td>
1115 </tr>
1116 <tr>
1117 <td>1</td>
1118 <td>1</td>
1119 <td>1</td>
1120 </tr>
1121 </tbody>
1122</table>
1123</center>
Chris Lattner00950542001-06-06 20:29:01 +00001124<h5>Example:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001125<pre> &lt;result&gt; = or int 4, %var <i>; yields {int}:result = 4 | %var</i>
Chris Lattner00950542001-06-06 20:29:01 +00001126 &lt;result&gt; = or int 15, 40 <i>; yields {int}:result = 47</i>
1127 &lt;result&gt; = or int 4, 8 <i>; yields {int}:result = 12</i>
1128</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001129</div>
Chris Lattner00950542001-06-06 20:29:01 +00001130<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +00001131<div class="doc_subsubsection"> <a name="i_xor">'<tt>xor</tt>'
1132Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001133<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +00001134<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001135<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 +00001136</pre>
Chris Lattner00950542001-06-06 20:29:01 +00001137<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001138<p>The '<tt>xor</tt>' instruction returns the bitwise logical exclusive
1139or of its two operands. The <tt>xor</tt> is used to implement the
1140"one's complement" operation, which is the "~" operator in C.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001141<h5>Arguments:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001142<p>The two arguments to the '<tt>xor</tt>' instruction must be <a
Chris Lattner261efe92003-11-25 01:02:51 +00001143 href="#t_integral">integral</a> values. Both arguments must have
1144identical types.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001145<h5>Semantics:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001146<p>The truth table used for the '<tt>xor</tt>' instruction is:</p>
Chris Lattner261efe92003-11-25 01:02:51 +00001147<p> </p>
1148<center>
1149<table border="1" cellspacing="0" cellpadding="4">
1150 <tbody>
1151 <tr>
1152 <td>In0</td>
1153 <td>In1</td>
1154 <td>Out</td>
1155 </tr>
1156 <tr>
1157 <td>0</td>
1158 <td>0</td>
1159 <td>0</td>
1160 </tr>
1161 <tr>
1162 <td>0</td>
1163 <td>1</td>
1164 <td>1</td>
1165 </tr>
1166 <tr>
1167 <td>1</td>
1168 <td>0</td>
1169 <td>1</td>
1170 </tr>
1171 <tr>
1172 <td>1</td>
1173 <td>1</td>
1174 <td>0</td>
1175 </tr>
1176 </tbody>
1177</table>
1178</center>
1179<p> </p>
Chris Lattner00950542001-06-06 20:29:01 +00001180<h5>Example:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001181<pre> &lt;result&gt; = xor int 4, %var <i>; yields {int}:result = 4 ^ %var</i>
Chris Lattner00950542001-06-06 20:29:01 +00001182 &lt;result&gt; = xor int 15, 40 <i>; yields {int}:result = 39</i>
1183 &lt;result&gt; = xor int 4, 8 <i>; yields {int}:result = 12</i>
Chris Lattner27f71f22003-09-03 00:41:47 +00001184 &lt;result&gt; = xor int %V, -1 <i>; yields {int}:result = ~%V</i>
Chris Lattner00950542001-06-06 20:29:01 +00001185</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001186</div>
Chris Lattner00950542001-06-06 20:29:01 +00001187<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +00001188<div class="doc_subsubsection"> <a name="i_shl">'<tt>shl</tt>'
1189Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001190<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +00001191<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001192<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 +00001193</pre>
Chris Lattner00950542001-06-06 20:29:01 +00001194<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001195<p>The '<tt>shl</tt>' instruction returns the first operand shifted to
1196the left a specified number of bits.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001197<h5>Arguments:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001198<p>The first argument to the '<tt>shl</tt>' instruction must be an <a
Chris Lattner261efe92003-11-25 01:02:51 +00001199 href="#t_integer">integer</a> type. The second argument must be an '<tt>ubyte</tt>'
1200type.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001201<h5>Semantics:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001202<p>The value produced is <tt>var1</tt> * 2<sup><tt>var2</tt></sup>.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001203<h5>Example:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001204<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 +00001205 &lt;result&gt; = shl int 4, ubyte 2 <i>; yields {int}:result = 16</i>
1206 &lt;result&gt; = shl int 1, ubyte 10 <i>; yields {int}:result = 1024</i>
1207</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001208</div>
Chris Lattner00950542001-06-06 20:29:01 +00001209<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +00001210<div class="doc_subsubsection"> <a name="i_shr">'<tt>shr</tt>'
1211Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001212<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +00001213<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001214<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 +00001215</pre>
Chris Lattner00950542001-06-06 20:29:01 +00001216<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001217<p>The '<tt>shr</tt>' instruction returns the first operand shifted to
1218the right a specified number of bits.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001219<h5>Arguments:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001220<p>The first argument to the '<tt>shr</tt>' instruction must be an <a
Chris Lattner261efe92003-11-25 01:02:51 +00001221 href="#t_integer">integer</a> type. The second argument must be an '<tt>ubyte</tt>'
1222type.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001223<h5>Semantics:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001224<p>If the first argument is a <a href="#t_signed">signed</a> type, the
1225most significant bit is duplicated in the newly free'd bit positions.
1226If the first argument is unsigned, zero bits shall fill the empty
1227positions.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001228<h5>Example:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001229<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 +00001230 &lt;result&gt; = shr uint 4, ubyte 1 <i>; yields {uint}:result = 2</i>
Chris Lattner00950542001-06-06 20:29:01 +00001231 &lt;result&gt; = shr int 4, ubyte 2 <i>; yields {int}:result = 1</i>
Chris Lattner8c6bb902003-06-18 21:30:51 +00001232 &lt;result&gt; = shr sbyte 4, ubyte 3 <i>; yields {sbyte}:result = 0</i>
1233 &lt;result&gt; = shr sbyte -2, ubyte 1 <i>; yields {sbyte}:result = -1</i>
Chris Lattner00950542001-06-06 20:29:01 +00001234</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001235</div>
Chris Lattner00950542001-06-06 20:29:01 +00001236<!-- ======================================================================= -->
Chris Lattner261efe92003-11-25 01:02:51 +00001237<div class="doc_subsection"> <a name="memoryops">Memory Access
1238Operations</a></div>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001239<div class="doc_text">
Chris Lattner261efe92003-11-25 01:02:51 +00001240<p>A key design point of an SSA-based representation is how it
1241represents memory. In LLVM, no memory locations are in SSA form, which
1242makes things very simple. This section describes how to read, write,
1243allocate and free memory in LLVM.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001244</div>
Chris Lattner00950542001-06-06 20:29:01 +00001245<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +00001246<div class="doc_subsubsection"> <a name="i_malloc">'<tt>malloc</tt>'
1247Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001248<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +00001249<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001250<pre> &lt;result&gt; = malloc &lt;type&gt;, uint &lt;NumElements&gt; <i>; yields {type*}:result</i>
Chris Lattner7faa8832002-04-14 06:13:44 +00001251 &lt;result&gt; = malloc &lt;type&gt; <i>; yields {type*}:result</i>
Chris Lattner00950542001-06-06 20:29:01 +00001252</pre>
Chris Lattner00950542001-06-06 20:29:01 +00001253<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001254<p>The '<tt>malloc</tt>' instruction allocates memory from the system
1255heap and returns a pointer to it.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001256<h5>Arguments:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001257<p>The the '<tt>malloc</tt>' instruction allocates <tt>sizeof(&lt;type&gt;)*NumElements</tt>
1258bytes of memory from the operating system, and returns a pointer of the
1259appropriate type to the program. The second form of the instruction is
1260a shorter version of the first instruction that defaults to allocating
1261one element.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001262<p>'<tt>type</tt>' must be a sized type.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001263<h5>Semantics:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001264<p>Memory is allocated using the system "<tt>malloc</tt>" function, and
1265a pointer is returned.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001266<h5>Example:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001267<pre> %array = malloc [4 x ubyte ] <i>; yields {[%4 x ubyte]*}:array</i>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001268
Chris Lattner261efe92003-11-25 01:02:51 +00001269 %size = <a
1270 href="#i_add">add</a> uint 2, 2 <i>; yields {uint}:size = uint 4</i>
Chris Lattner7faa8832002-04-14 06:13:44 +00001271 %array1 = malloc ubyte, uint 4 <i>; yields {ubyte*}:array1</i>
1272 %array2 = malloc [12 x ubyte], uint %size <i>; yields {[12 x ubyte]*}:array2</i>
Chris Lattner00950542001-06-06 20:29:01 +00001273</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001274</div>
Chris Lattner00950542001-06-06 20:29:01 +00001275<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +00001276<div class="doc_subsubsection"> <a name="i_free">'<tt>free</tt>'
1277Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001278<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +00001279<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001280<pre> free &lt;type&gt; &lt;value&gt; <i>; yields {void}</i>
Chris Lattner00950542001-06-06 20:29:01 +00001281</pre>
Chris Lattner00950542001-06-06 20:29:01 +00001282<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001283<p>The '<tt>free</tt>' instruction returns memory back to the unused
1284memory heap, to be reallocated in the future.</p>
1285<p> </p>
Chris Lattner00950542001-06-06 20:29:01 +00001286<h5>Arguments:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001287<p>'<tt>value</tt>' shall be a pointer value that points to a value
1288that was allocated with the '<tt><a href="#i_malloc">malloc</a></tt>'
1289instruction.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001290<h5>Semantics:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001291<p>Access to the memory pointed to by the pointer is not longer defined
1292after this instruction executes.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001293<h5>Example:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001294<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 +00001295 free [4 x ubyte]* %array
1296</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001297</div>
Chris Lattner00950542001-06-06 20:29:01 +00001298<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +00001299<div class="doc_subsubsection"> <a name="i_alloca">'<tt>alloca</tt>'
1300Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001301<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +00001302<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001303<pre> &lt;result&gt; = alloca &lt;type&gt;, uint &lt;NumElements&gt; <i>; yields {type*}:result</i>
Chris Lattner7faa8832002-04-14 06:13:44 +00001304 &lt;result&gt; = alloca &lt;type&gt; <i>; yields {type*}:result</i>
Chris Lattner00950542001-06-06 20:29:01 +00001305</pre>
Chris Lattner00950542001-06-06 20:29:01 +00001306<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001307<p>The '<tt>alloca</tt>' instruction allocates memory on the current
1308stack frame of the procedure that is live until the current function
1309returns to its caller.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001310<h5>Arguments:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001311<p>The the '<tt>alloca</tt>' instruction allocates <tt>sizeof(&lt;type&gt;)*NumElements</tt>
1312bytes of memory on the runtime stack, returning a pointer of the
1313appropriate type to the program. The second form of the instruction is
1314a shorter version of the first that defaults to allocating one element.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001315<p>'<tt>type</tt>' may be any sized type.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001316<h5>Semantics:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001317<p>Memory is allocated, a pointer is returned. '<tt>alloca</tt>'d
1318memory is automatically released when the function returns. The '<tt>alloca</tt>'
1319instruction is commonly used to represent automatic variables that must
1320have an address available. When the function returns (either with the <tt><a
1321 href="#i_ret">ret</a></tt> or <tt><a href="#i_invoke">invoke</a></tt>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001322instructions), the memory is reclaimed.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001323<h5>Example:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001324<pre> %ptr = alloca int <i>; yields {int*}:ptr</i>
Chris Lattner7faa8832002-04-14 06:13:44 +00001325 %ptr = alloca int, uint 4 <i>; yields {int*}:ptr</i>
Chris Lattner00950542001-06-06 20:29:01 +00001326</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001327</div>
Chris Lattner00950542001-06-06 20:29:01 +00001328<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +00001329<div class="doc_subsubsection"> <a name="i_load">'<tt>load</tt>'
1330Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001331<div class="doc_text">
Chris Lattner2b7d3202002-05-06 03:03:22 +00001332<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001333<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 +00001334<h5>Overview:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001335<p>The '<tt>load</tt>' instruction is used to read from memory.</p>
Chris Lattner2b7d3202002-05-06 03:03:22 +00001336<h5>Arguments:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001337<p>The argument to the '<tt>load</tt>' instruction specifies the memory
1338address to load from. The pointer must point to a <a
1339 href="t_firstclass">first class</a> type. If the <tt>load</tt> is
1340marked as <tt>volatile</tt> then the optimizer is not allowed to modify
1341the number or order of execution of this <tt>load</tt> with other
1342volatile <tt>load</tt> and <tt><a href="#i_store">store</a></tt>
1343instructions. </p>
Chris Lattner2b7d3202002-05-06 03:03:22 +00001344<h5>Semantics:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001345<p>The location of memory pointed to is loaded.</p>
Chris Lattner2b7d3202002-05-06 03:03:22 +00001346<h5>Examples:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001347<pre> %ptr = <a href="#i_alloca">alloca</a> int <i>; yields {int*}:ptr</i>
1348 <a
1349 href="#i_store">store</a> int 3, int* %ptr <i>; yields {void}</i>
Chris Lattner2b7d3202002-05-06 03:03:22 +00001350 %val = load int* %ptr <i>; yields {int}:val = int 3</i>
1351</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001352</div>
Chris Lattner2b7d3202002-05-06 03:03:22 +00001353<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +00001354<div class="doc_subsubsection"> <a name="i_store">'<tt>store</tt>'
1355Instruction</a> </div>
Chris Lattner2b7d3202002-05-06 03:03:22 +00001356<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001357<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 +00001358 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 +00001359</pre>
Chris Lattner2b7d3202002-05-06 03:03:22 +00001360<h5>Overview:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001361<p>The '<tt>store</tt>' instruction is used to write to memory.</p>
Chris Lattner2b7d3202002-05-06 03:03:22 +00001362<h5>Arguments:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001363<p>There are two arguments to the '<tt>store</tt>' instruction: a value
1364to store and an address to store it into. The type of the '<tt>&lt;pointer&gt;</tt>'
1365operand must be a pointer to the type of the '<tt>&lt;value&gt;</tt>'
1366operand. If the <tt>store</tt> is marked as <tt>volatile</tt> then the
1367optimizer is not allowed to modify the number or order of execution of
1368this <tt>store</tt> with other volatile <tt>load</tt> and <tt><a
1369 href="#i_store">store</a></tt> instructions.</p>
1370<h5>Semantics:</h5>
1371<p>The contents of memory are updated to contain '<tt>&lt;value&gt;</tt>'
1372at the location specified by the '<tt>&lt;pointer&gt;</tt>' operand.</p>
Chris Lattner2b7d3202002-05-06 03:03:22 +00001373<h5>Example:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001374<pre> %ptr = <a href="#i_alloca">alloca</a> int <i>; yields {int*}:ptr</i>
1375 <a
1376 href="#i_store">store</a> int 3, int* %ptr <i>; yields {void}</i>
Chris Lattner2b7d3202002-05-06 03:03:22 +00001377 %val = load int* %ptr <i>; yields {int}:val = int 3</i>
1378</pre>
Chris Lattner2b7d3202002-05-06 03:03:22 +00001379<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +00001380<div class="doc_subsubsection"> <a name="i_getelementptr">'<tt>getelementptr</tt>'
1381Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001382<div class="doc_text">
Chris Lattner7faa8832002-04-14 06:13:44 +00001383<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001384<pre> &lt;result&gt; = getelementptr &lt;ty&gt;* &lt;ptrval&gt;{, long &lt;aidx&gt;|, ubyte &lt;sidx&gt;}*<br></pre>
Chris Lattner7faa8832002-04-14 06:13:44 +00001385<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001386<p>The '<tt>getelementptr</tt>' instruction is used to get the address
1387of a subelement of an aggregate data structure.</p>
Chris Lattner7faa8832002-04-14 06:13:44 +00001388<h5>Arguments:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001389<p>This instruction takes a list of <tt>long</tt> values and <tt>ubyte</tt>
Chris Lattner261efe92003-11-25 01:02:51 +00001390constants that indicate what form of addressing to perform. The actual
1391types of the arguments provided depend on the type of the first pointer
1392argument. The '<tt>getelementptr</tt>' instruction is used to index
1393down through the type levels of a structure.</p>
1394<p>For example, let's consider a C code fragment and how it gets
1395compiled to LLVM:</p>
1396<pre>struct RT {<br> char A;<br> int B[10][20];<br> char C;<br>};<br>struct ST {<br> int X;<br> double Y;<br> struct RT Z;<br>};<br><br>int *foo(struct ST *s) {<br> return &amp;s[1].Z.B[5][13];<br>}<br></pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001397<p>The LLVM code generated by the GCC frontend is:</p>
Chris Lattner261efe92003-11-25 01:02:51 +00001398<pre>%RT = type { sbyte, [10 x [20 x int]], sbyte }<br>%ST = type { int, double, %RT }<br><br>int* "foo"(%ST* %s) {<br> %reg = getelementptr %ST* %s, long 1, ubyte 2, ubyte 1, long 5, long 13<br> ret int* %reg<br>}<br></pre>
Chris Lattner7faa8832002-04-14 06:13:44 +00001399<h5>Semantics:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001400<p>The index types specified for the '<tt>getelementptr</tt>'
1401instruction depend on the pointer type that is being index into. <a
1402 href="t_pointer">Pointer</a> and <a href="t_array">array</a> types
1403require '<tt>long</tt>' values, and <a href="t_struct">structure</a>
1404types require '<tt>ubyte</tt>' <b>constants</b>.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001405<p>In the example above, the first index is indexing into the '<tt>%ST*</tt>'
Chris Lattner261efe92003-11-25 01:02:51 +00001406type, which is a pointer, yielding a '<tt>%ST</tt>' = '<tt>{ int,
1407double, %RT }</tt>' type, a structure. The second index indexes into
1408the third element of the structure, yielding a '<tt>%RT</tt>' = '<tt>{
1409sbyte, [10 x [20 x int]], sbyte }</tt>' type, another structure. The
1410third index indexes into the second element of the structure, yielding
1411a '<tt>[10 x [20 x int]]</tt>' type, an array. The two dimensions of
1412the array are subscripted into, yielding an '<tt>int</tt>' type. The '<tt>getelementptr</tt>'
1413instruction return a pointer to this element, thus yielding a '<tt>int*</tt>'
1414type.</p>
1415<p>Note that it is perfectly legal to index partially through a
1416structure, returning a pointer to an inner element. Because of this,
1417the LLVM code for the given testcase is equivalent to:</p>
1418<pre>int* "foo"(%ST* %s) {<br> %t1 = getelementptr %ST* %s , long 1 <i>; yields %ST*:%t1</i>
Chris Lattner3dfa10b2002-12-13 06:01:21 +00001419 %t2 = getelementptr %ST* %t1, long 0, ubyte 2 <i>; yields %RT*:%t2</i>
1420 %t3 = getelementptr %RT* %t2, long 0, ubyte 1 <i>; yields [10 x [20 x int]]*:%t3</i>
1421 %t4 = getelementptr [10 x [20 x int]]* %t3, long 0, long 5 <i>; yields [20 x int]*:%t4</i>
1422 %t5 = getelementptr [20 x int]* %t4, long 0, long 13 <i>; yields int*:%t5</i>
Chris Lattner6536cfe2002-05-06 22:08:29 +00001423 ret int* %t5
1424}
1425</pre>
Chris Lattner7faa8832002-04-14 06:13:44 +00001426<h5>Example:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001427<pre> <i>; yields [12 x ubyte]*:aptr</i>
1428 %aptr = getelementptr {int, [12 x ubyte]}* %sptr, long 0, ubyte 1<br></pre>
1429<h5>&nbsp;Note To The Novice:</h5>
1430When using indexing into global arrays with the '<tt>getelementptr</tt>'
1431instruction, you must remember that the&nbsp; </div>
Chris Lattner00950542001-06-06 20:29:01 +00001432<!-- ======================================================================= -->
Chris Lattner261efe92003-11-25 01:02:51 +00001433<div class="doc_subsection"> <a name="otherops">Other Operations</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001434<div class="doc_text">
Chris Lattner261efe92003-11-25 01:02:51 +00001435<p>The instructions in this catagory are the "miscellaneous"
1436instructions, which defy better classification.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001437</div>
Chris Lattner00950542001-06-06 20:29:01 +00001438<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +00001439<div class="doc_subsubsection"> <a name="i_phi">'<tt>phi</tt>'
1440Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001441<div class="doc_text">
Chris Lattner33ba0d92001-07-09 00:26:23 +00001442<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001443<pre> &lt;result&gt; = phi &lt;ty&gt; [ &lt;val0&gt;, &lt;label0&gt;], ...<br></pre>
Chris Lattner33ba0d92001-07-09 00:26:23 +00001444<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001445<p>The '<tt>phi</tt>' instruction is used to implement the &#966; node in
1446the SSA graph representing the function.</p>
Chris Lattner33ba0d92001-07-09 00:26:23 +00001447<h5>Arguments:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001448<p>The type of the incoming values are specified with the first type
1449field. After this, the '<tt>phi</tt>' instruction takes a list of pairs
1450as arguments, with one pair for each predecessor basic block of the
1451current block. Only values of <a href="#t_firstclass">first class</a>
1452type may be used as the value arguments to the PHI node. Only labels
1453may be used as the label arguments.</p>
1454<p>There must be no non-phi instructions between the start of a basic
1455block and the PHI instructions: i.e. PHI instructions must be first in
1456a basic block.</p>
Chris Lattner33ba0d92001-07-09 00:26:23 +00001457<h5>Semantics:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001458<p>At runtime, the '<tt>phi</tt>' instruction logically takes on the
1459value specified by the parameter, depending on which basic block we
1460came from in the last <a href="#terminators">terminator</a> instruction.</p>
Chris Lattner6536cfe2002-05-06 22:08:29 +00001461<h5>Example:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001462<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 +00001463</div>
Chris Lattner6536cfe2002-05-06 22:08:29 +00001464<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +00001465<div class="doc_subsubsection"> <a name="i_cast">'<tt>cast .. to</tt>'
1466Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001467<div class="doc_text">
Chris Lattner6536cfe2002-05-06 22:08:29 +00001468<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001469<pre> &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 +00001470</pre>
Chris Lattner6536cfe2002-05-06 22:08:29 +00001471<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001472<p>The '<tt>cast</tt>' instruction is used as the primitive means to
1473convert integers to floating point, change data type sizes, and break
1474type safety (by casting pointers).</p>
Chris Lattner6536cfe2002-05-06 22:08:29 +00001475<h5>Arguments:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001476<p>The '<tt>cast</tt>' instruction takes a value to cast, which must be
1477a first class value, and a type to cast it to, which must also be a <a
1478 href="#t_firstclass">first class</a> type.</p>
Chris Lattner6536cfe2002-05-06 22:08:29 +00001479<h5>Semantics:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001480<p>This instruction follows the C rules for explicit casts when
1481determining how the data being cast must change to fit in its new
1482container.</p>
1483<p>When casting to bool, any value that would be considered true in the
1484context of a C '<tt>if</tt>' condition is converted to the boolean '<tt>true</tt>'
Misha Brukman9d0919f2003-11-08 01:05:38 +00001485values, all else are '<tt>false</tt>'.</p>
Chris Lattner261efe92003-11-25 01:02:51 +00001486<p>When extending an integral value from a type of one signness to
1487another (for example '<tt>sbyte</tt>' to '<tt>ulong</tt>'), the value
1488is sign-extended if the <b>source</b> value is signed, and
1489zero-extended if the source value is unsigned. <tt>bool</tt> values
1490are always zero extended into either zero or one.</p>
Chris Lattner33ba0d92001-07-09 00:26:23 +00001491<h5>Example:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001492<pre> %X = cast int 257 to ubyte <i>; yields ubyte:1</i>
Chris Lattner7bae3952002-06-25 18:03:17 +00001493 %Y = cast int 123 to bool <i>; yields bool:true</i>
Chris Lattner33ba0d92001-07-09 00:26:23 +00001494</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001495</div>
Chris Lattner33ba0d92001-07-09 00:26:23 +00001496<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +00001497<div class="doc_subsubsection"> <a name="i_call">'<tt>call</tt>'
1498Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001499<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +00001500<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001501<pre> &lt;result&gt; = call &lt;ty&gt;* &lt;fnptrval&gt;(&lt;param list&gt;)<br></pre>
Chris Lattner00950542001-06-06 20:29:01 +00001502<h5>Overview:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001503<p>The '<tt>call</tt>' instruction represents a simple function call.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001504<h5>Arguments:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001505<p>This instruction requires several arguments:</p>
Chris Lattner6536cfe2002-05-06 22:08:29 +00001506<ol>
Chris Lattner261efe92003-11-25 01:02:51 +00001507 <li>
1508 <p>'<tt>ty</tt>': shall be the signature of the pointer to function
1509value being invoked. The argument types must match the types implied
1510by this signature.</p>
1511 </li>
1512 <li>
1513 <p>'<tt>fnptrval</tt>': An LLVM value containing a pointer to a
1514function to be invoked. In most cases, this is a direct function
1515invocation, but indirect <tt>call</tt>s are just as possible,
1516calling an arbitrary pointer to function values.</p>
1517 </li>
1518 <li>
1519 <p>'<tt>function args</tt>': argument list whose types match the
1520function signature argument types. If the function signature
1521indicates the function accepts a variable number of arguments, the
1522extra arguments can be specified.</p>
1523 </li>
Chris Lattner6536cfe2002-05-06 22:08:29 +00001524</ol>
Chris Lattner00950542001-06-06 20:29:01 +00001525<h5>Semantics:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001526<p>The '<tt>call</tt>' instruction is used to cause control flow to
1527transfer to a specified function, with its incoming arguments bound to
1528the specified values. Upon a '<tt><a href="#i_ret">ret</a></tt>'
1529instruction in the called function, control flow continues with the
1530instruction after the function call, and the return value of the
1531function is bound to the result argument. This is a simpler case of
1532the <a href="#i_invoke">invoke</a> instruction.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001533<h5>Example:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001534<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 +00001535</div>
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00001536<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +00001537<div class="doc_subsubsection"> <a name="i_vanext">'<tt>vanext</tt>'
1538Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001539<div class="doc_text">
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00001540<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001541<pre> &lt;resultarglist&gt; = vanext &lt;va_list&gt; &lt;arglist&gt;, &lt;argty&gt;<br></pre>
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00001542<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001543<p>The '<tt>vanext</tt>' instruction is used to access arguments passed
1544through the "variable argument" area of a function call. It is used to
1545implement the <tt>va_arg</tt> macro in C.</p>
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00001546<h5>Arguments:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001547<p>This instruction takes a <tt>valist</tt> value and the type of the
1548argument. It returns another <tt>valist</tt>.</p>
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00001549<h5>Semantics:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001550<p>The '<tt>vanext</tt>' instruction advances the specified <tt>valist</tt>
1551past an argument of the specified type. In conjunction with the <a
1552 href="#i_vaarg"><tt>vaarg</tt></a> instruction, it is used to implement
1553the <tt>va_arg</tt> macro available in C. For more information, see
1554the variable argument handling <a href="#int_varargs">Intrinsic
1555Functions</a>.</p>
1556<p>It is legal for this instruction to be called in a function which
1557does not take a variable number of arguments, for example, the <tt>vfprintf</tt>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001558function.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001559<p><tt>vanext</tt> is an LLVM instruction instead of an <a
Chris Lattner261efe92003-11-25 01:02:51 +00001560 href="#intrinsics">intrinsic function</a> because it takes an type as
1561an argument.</p>
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00001562<h5>Example:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001563<p>See the <a href="#int_varargs">variable argument processing</a>
1564section.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001565</div>
Chris Lattner8d1a81d2003-10-18 05:51:36 +00001566<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +00001567<div class="doc_subsubsection"> <a name="i_vaarg">'<tt>vaarg</tt>'
1568Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001569<div class="doc_text">
Chris Lattner8d1a81d2003-10-18 05:51:36 +00001570<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001571<pre> &lt;resultval&gt; = vaarg &lt;va_list&gt; &lt;arglist&gt;, &lt;argty&gt;<br></pre>
Chris Lattner8d1a81d2003-10-18 05:51:36 +00001572<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001573<p>The '<tt>vaarg</tt>' instruction is used to access arguments passed
1574through the "variable argument" area of a function call. It is used to
1575implement the <tt>va_arg</tt> macro in C.</p>
Chris Lattner8d1a81d2003-10-18 05:51:36 +00001576<h5>Arguments:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001577<p>This instruction takes a <tt>valist</tt> value and the type of the
1578argument. It returns a value of the specified argument type.</p>
Chris Lattner8d1a81d2003-10-18 05:51:36 +00001579<h5>Semantics:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001580<p>The '<tt>vaarg</tt>' instruction loads an argument of the specified
1581type from the specified <tt>va_list</tt>. In conjunction with the <a
1582 href="#i_vanext"><tt>vanext</tt></a> instruction, it is used to
1583implement the <tt>va_arg</tt> macro available in C. For more
1584information, see the variable argument handling <a href="#int_varargs">Intrinsic
1585Functions</a>.</p>
1586<p>It is legal for this instruction to be called in a function which
1587does not take a variable number of arguments, for example, the <tt>vfprintf</tt>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001588function.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001589<p><tt>vaarg</tt> is an LLVM instruction instead of an <a
Chris Lattner261efe92003-11-25 01:02:51 +00001590 href="#intrinsics">intrinsic function</a> because it takes an type as
1591an argument.</p>
Chris Lattner8d1a81d2003-10-18 05:51:36 +00001592<h5>Example:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001593<p>See the <a href="#int_varargs">variable argument processing</a>
1594section.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001595</div>
Chris Lattner8ff75902004-01-06 05:31:32 +00001596
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00001597<!-- *********************************************************************** -->
Chris Lattner261efe92003-11-25 01:02:51 +00001598<div class="doc_section"> <a name="intrinsics">Intrinsic Functions</a> </div>
1599<!-- *********************************************************************** -->
Chris Lattner8ff75902004-01-06 05:31:32 +00001600
Misha Brukman9d0919f2003-11-08 01:05:38 +00001601<div class="doc_text">
Chris Lattner33aec9e2004-02-12 17:01:32 +00001602
1603<p>LLVM supports the notion of an "intrinsic function". These functions have
1604well known names and semantics, and are required to follow certain
1605restrictions. Overall, these instructions represent an extension mechanism for
1606the LLVM language that does not require changing all of the transformations in
1607LLVM to add to the language (or the bytecode reader/writer, the parser,
1608etc...).</p>
1609
1610<p>Intrinsic function names must all start with an "<tt>llvm.</tt>" prefix, this
1611prefix is reserved in LLVM for intrinsic names, thus functions may not be named
1612this. Intrinsic functions must always be external functions: you cannot define
1613the body of intrinsic functions. Intrinsic functions may only be used in call
1614or invoke instructions: it is illegal to take the address of an intrinsic
1615function. Additionally, because intrinsic functions are part of the LLVM
1616language, it is required that they all be documented here if any are added.</p>
1617
1618
1619<p>
1620Adding an intrinsic to LLVM is straight-forward if it is possible to express the
1621concept in LLVM directly (ie, code generator support is not _required_). To do
1622this, extend the default implementation of the IntrinsicLowering class to handle
1623the intrinsic. Code generators use this class to lower intrinsics they do not
1624understand to raw LLVM instructions that they do.
1625</p>
1626
Misha Brukman9d0919f2003-11-08 01:05:38 +00001627</div>
Chris Lattner8ff75902004-01-06 05:31:32 +00001628
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00001629<!-- ======================================================================= -->
Chris Lattner8ff75902004-01-06 05:31:32 +00001630<div class="doc_subsection">
1631 <a name="int_varargs">Variable Argument Handling Intrinsics</a>
1632</div>
1633
Misha Brukman9d0919f2003-11-08 01:05:38 +00001634<div class="doc_text">
Misha Brukman9d0919f2003-11-08 01:05:38 +00001635<p>Variable argument support is defined in LLVM with the <a
Chris Lattner261efe92003-11-25 01:02:51 +00001636 href="#i_vanext"><tt>vanext</tt></a> instruction and these three
1637intrinsic functions. These functions are related to the similarly
1638named macros defined in the <tt>&lt;stdarg.h&gt;</tt> header file.</p>
1639<p>All of these functions operate on arguments that use a
1640target-specific value type "<tt>va_list</tt>". The LLVM assembly
1641language reference manual does not define what this type is, so all
1642transformations should be prepared to handle intrinsics with any type
1643used.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001644<p>This example shows how the <a href="#i_vanext"><tt>vanext</tt></a>
Chris Lattner261efe92003-11-25 01:02:51 +00001645instruction and the variable argument handling intrinsic functions are
1646used.</p>
Chris Lattner33aec9e2004-02-12 17:01:32 +00001647<pre>
1648int %test(int %X, ...) {
1649 ; Initialize variable argument processing
1650 %ap = call sbyte* %<a href="#i_va_start">llvm.va_start</a>()
1651
1652 ; Read a single integer argument
1653 %tmp = vaarg sbyte* %ap, int
1654
1655 ; Advance to the next argument
1656 %ap2 = vanext sbyte* %ap, int
1657
1658 ; Demonstrate usage of llvm.va_copy and llvm.va_end
1659 %aq = call sbyte* %<a href="#i_va_copy">llvm.va_copy</a>(sbyte* %ap2)
1660 call void %<a href="#i_va_end">llvm.va_end</a>(sbyte* %aq)
1661
1662 ; Stop processing of arguments.
1663 call void %<a href="#i_va_end">llvm.va_end</a>(sbyte* %ap2)
1664 ret int %tmp
1665}
1666</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001667</div>
Chris Lattner8ff75902004-01-06 05:31:32 +00001668
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00001669<!-- _______________________________________________________________________ -->
Chris Lattner8ff75902004-01-06 05:31:32 +00001670<div class="doc_subsubsection">
1671 <a name="i_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a>
1672</div>
1673
1674
Misha Brukman9d0919f2003-11-08 01:05:38 +00001675<div class="doc_text">
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00001676<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001677<pre> call va_list ()* %llvm.va_start()<br></pre>
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00001678<h5>Overview:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001679<p>The '<tt>llvm.va_start</tt>' intrinsic returns a new <tt>&lt;arglist&gt;</tt>
1680for subsequent use by the variable argument intrinsics.</p>
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00001681<h5>Semantics:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001682<p>The '<tt>llvm.va_start</tt>' intrinsic works just like the <tt>va_start</tt>
Chris Lattner261efe92003-11-25 01:02:51 +00001683macro available in C. In a target-dependent way, it initializes and
1684returns a <tt>va_list</tt> element, so that the next <tt>vaarg</tt>
1685will produce the first variable argument passed to the function. Unlike
1686the C <tt>va_start</tt> macro, this intrinsic does not need to know the
1687last argument of the function, the compiler can figure that out.</p>
1688<p>Note that this intrinsic function is only legal to be called from
1689within the body of a variable argument function.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001690</div>
Chris Lattner8ff75902004-01-06 05:31:32 +00001691
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00001692<!-- _______________________________________________________________________ -->
Chris Lattner8ff75902004-01-06 05:31:32 +00001693<div class="doc_subsubsection">
1694 <a name="i_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a>
1695</div>
1696
Misha Brukman9d0919f2003-11-08 01:05:38 +00001697<div class="doc_text">
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00001698<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001699<pre> call void (va_list)* %llvm.va_end(va_list &lt;arglist&gt;)<br></pre>
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00001700<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001701<p>The '<tt>llvm.va_end</tt>' intrinsic destroys <tt>&lt;arglist&gt;</tt>
1702which has been initialized previously with <tt><a href="#i_va_start">llvm.va_start</a></tt>
1703or <tt><a href="#i_va_copy">llvm.va_copy</a></tt>.</p>
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00001704<h5>Arguments:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001705<p>The argument is a <tt>va_list</tt> to destroy.</p>
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00001706<h5>Semantics:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001707<p>The '<tt>llvm.va_end</tt>' intrinsic works just like the <tt>va_end</tt>
Chris Lattner261efe92003-11-25 01:02:51 +00001708macro available in C. In a target-dependent way, it destroys the <tt>va_list</tt>.
1709Calls to <a href="#i_va_start"><tt>llvm.va_start</tt></a> and <a
1710 href="#i_va_copy"><tt>llvm.va_copy</tt></a> must be matched exactly
1711with calls to <tt>llvm.va_end</tt>.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001712</div>
Chris Lattner8ff75902004-01-06 05:31:32 +00001713
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00001714<!-- _______________________________________________________________________ -->
Chris Lattner8ff75902004-01-06 05:31:32 +00001715<div class="doc_subsubsection">
1716 <a name="i_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a>
1717</div>
1718
Misha Brukman9d0919f2003-11-08 01:05:38 +00001719<div class="doc_text">
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00001720<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001721<pre> call va_list (va_list)* %llvm.va_copy(va_list &lt;destarglist&gt;)<br></pre>
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00001722<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00001723<p>The '<tt>llvm.va_copy</tt>' intrinsic copies the current argument
1724position from the source argument list to the destination argument list.</p>
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00001725<h5>Arguments:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001726<p>The argument is the <tt>va_list</tt> to copy.</p>
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00001727<h5>Semantics:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001728<p>The '<tt>llvm.va_copy</tt>' intrinsic works just like the <tt>va_copy</tt>
Chris Lattner261efe92003-11-25 01:02:51 +00001729macro available in C. In a target-dependent way, it copies the source <tt>va_list</tt>
1730element into the returned list. This intrinsic is necessary because the <tt><a
1731 href="i_va_start">llvm.va_start</a></tt> intrinsic may be arbitrarily
1732complex and require memory allocation, for example.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001733</div>
Chris Lattner8ff75902004-01-06 05:31:32 +00001734
Chris Lattner33aec9e2004-02-12 17:01:32 +00001735<!-- ======================================================================= -->
1736<div class="doc_subsection">
1737 <a name="int_libc">Standard C Library Intrinsics</a>
1738</div>
1739
1740<div class="doc_text">
1741<p>
1742
1743</p>
1744
1745</div>
1746
1747<!-- _______________________________________________________________________ -->
1748<div class="doc_subsubsection">
1749 <a name="i_memcpy">'<tt>llvm.memcpy</tt>' Intrinsic</a>
1750</div>
1751
1752<div class="doc_text">
1753
1754<h5>Syntax:</h5>
1755<pre>
1756 call void (sbyte*, sbyte*, uint, uint)* %llvm.memcpy(sbyte* &lt;dest&gt;, sbyte* &lt;src&gt;,
1757 uint &lt;len&gt;, uint &lt;align&gt;)
1758</pre>
1759
1760<h5>Overview:</h5>
1761
1762<p>
1763The '<tt>llvm.memcpy</tt>' intrinsic copies a block of memory from the source
1764location to the destination location.
1765</p>
1766
1767<p>
1768Note that, unlike the standard libc function, the <tt>llvm.memcpy</tt> intrinsic
1769does not return a value, and takes an extra alignment argument.
1770</p>
1771
1772<h5>Arguments:</h5>
1773
1774<p>
1775The first argument is a pointer to the destination, the second is a pointer to
1776the source. The third argument is an (arbitrarily sized) integer argument
1777specifying the number of bytes to copy, and the fourth argument is the alignment
1778of the source and destination locations.
1779</p>
1780
1781<h5>Semantics:</h5>
1782
1783<p>
1784The '<tt>llvm.memcpy</tt>' intrinsic copies a block of memory from the source
1785location to the destination location, which are not allowed to overlap. It
1786copies "len" bytes of memory over. If the argument is known to be aligned to
1787some boundary, this can be specified as the fourth argument, otherwise it should
1788be set to 0 or 1.
1789</p>
1790</div>
1791
1792
Chris Lattner8ff75902004-01-06 05:31:32 +00001793
1794<!-- ======================================================================= -->
1795<div class="doc_subsection">
1796 <a name="int_debugger">Debugger Intrinsics</a>
1797</div>
1798
1799<div class="doc_text">
1800<p>
1801The LLVM debugger intrinsics (which all start with <tt>llvm.dbg.</tt> prefix),
1802are described in the <a
1803href="SourceLevelDebugging.html#format_common_intrinsics">LLVM Source Level
1804Debugging</a> document.
1805</p>
1806</div>
1807
1808
Chris Lattner00950542001-06-06 20:29:01 +00001809<!-- *********************************************************************** -->
Chris Lattner00950542001-06-06 20:29:01 +00001810<hr>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001811<div class="doc_footer">
Chris Lattner261efe92003-11-25 01:02:51 +00001812<address><a href="mailto:sabre@nondot.org">Chris Lattner</a></address>
1813<a href="http://llvm.cs.uiuc.edu">The LLVM Compiler Infrastructure</a> <br>
1814Last modified: $Date$ </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001815</body>
1816</html>