Gitiles
Code Review Sign In
gerrit-public.fairphone.software / fp2-dev / platform / external / llvm / f17a0b7117af9fed21bd2232d50d4fc4c27b5944 / . / docs / LangRef.html
blob: 4954113d0d4eb09f8eb94906bf76fb21147ea082 [file] [log] [blame]
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000[diff] [blame]1<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN"
2 "http://www.w3.org/TR/html4/strict.dtd">
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3<html>
4<head>
5 <title>LLVM Assembly Language Reference Manual</title>
Reid Spencer3921c742004-08-26 20:44:00 +0000[diff] [blame]6 <meta http-equiv="Content-Type" content="text/html; charset=utf-8">
7 <meta name="author" content="Chris Lattner">
8 <meta name="description"
9 content="LLVM Assembly Language Reference Manual.">
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]10 <link rel="stylesheet" href="llvm.css" type="text/css">
11</head>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]12
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]13<body>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]14
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]15<div class="doc_title"> LLVM Language Reference Manual </div>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]16<ol>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]17 <li><a href="#abstract">Abstract</a></li>
18 <li><a href="#introduction">Introduction</a></li>
19 <li><a href="#identifiers">Identifiers</a></li>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]20 <li><a href="#highlevel">High Level Structure</a>
21 <ol>
22 <li><a href="#modulestructure">Module Structure</a></li>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]23 <li><a href="#linkage">Linkage Types</a></li>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]24 <li><a href="#callingconv">Calling Conventions</a></li>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]25 <li><a href="#globalvars">Global Variables</a></li>
Chris Lattner4e9aba72006-01-23 23:23:47 +0000[diff] [blame]26 <li><a href="#functionstructure">Functions</a></li>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]27 <li><a href="#paramattrs">Parameter Attributes</a></li>
Chris Lattner4e9aba72006-01-23 23:23:47 +0000[diff] [blame]28 <li><a href="#moduleasm">Module-Level Inline Assembly</a></li>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]29 </ol>
30 </li>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]31 <li><a href="#typesystem">Type System</a>
32 <ol>
Robert Bocchino7b81c752006-02-17 21:18:08 +0000[diff] [blame]33 <li><a href="#t_primitive">Primitive Types</a>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]34 <ol>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]35 <li><a href="#t_classifications">Type Classifications</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]36 </ol>
37 </li>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]38 <li><a href="#t_derived">Derived Types</a>
39 <ol>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]40 <li><a href="#t_array">Array Type</a></li>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]41 <li><a href="#t_function">Function Type</a></li>
42 <li><a href="#t_pointer">Pointer Type</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]43 <li><a href="#t_struct">Structure Type</a></li>
Andrew Lenharth75e10682006-12-08 17:13:00 +0000[diff] [blame]44 <li><a href="#t_pstruct">Packed Structure Type</a></li>
Chris Lattnera58561b2004-08-12 19:12:28 +0000[diff] [blame]45 <li><a href="#t_packed">Packed Type</a></li>
Chris Lattner69c11bb2005-04-25 17:34:15 +0000[diff] [blame]46 <li><a href="#t_opaque">Opaque Type</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]47 </ol>
48 </li>
49 </ol>
50 </li>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]51 <li><a href="#constants">Constants</a>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]52 <ol>
53 <li><a href="#simpleconstants">Simple Constants</a>
54 <li><a href="#aggregateconstants">Aggregate Constants</a>
55 <li><a href="#globalconstants">Global Variable and Function Addresses</a>
56 <li><a href="#undefvalues">Undefined Values</a>
57 <li><a href="#constantexprs">Constant Expressions</a>
58 </ol>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]59 </li>
Chris Lattnere87d6532006-01-25 23:47:57 +0000[diff] [blame]60 <li><a href="#othervalues">Other Values</a>
61 <ol>
62 <li><a href="#inlineasm">Inline Assembler Expressions</a>
63 </ol>
64 </li>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]65 <li><a href="#instref">Instruction Reference</a>
66 <ol>
67 <li><a href="#terminators">Terminator Instructions</a>
68 <ol>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]69 <li><a href="#i_ret">'<tt>ret</tt>' Instruction</a></li>
70 <li><a href="#i_br">'<tt>br</tt>' Instruction</a></li>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]71 <li><a href="#i_switch">'<tt>switch</tt>' Instruction</a></li>
72 <li><a href="#i_invoke">'<tt>invoke</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]73 <li><a href="#i_unwind">'<tt>unwind</tt>' Instruction</a></li>
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]74 <li><a href="#i_unreachable">'<tt>unreachable</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]75 </ol>
76 </li>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]77 <li><a href="#binaryops">Binary Operations</a>
78 <ol>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]79 <li><a href="#i_add">'<tt>add</tt>' Instruction</a></li>
80 <li><a href="#i_sub">'<tt>sub</tt>' Instruction</a></li>
81 <li><a href="#i_mul">'<tt>mul</tt>' Instruction</a></li>
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]82 <li><a href="#i_udiv">'<tt>udiv</tt>' Instruction</a></li>
83 <li><a href="#i_sdiv">'<tt>sdiv</tt>' Instruction</a></li>
84 <li><a href="#i_fdiv">'<tt>fdiv</tt>' Instruction</a></li>
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]85 <li><a href="#i_urem">'<tt>urem</tt>' Instruction</a></li>
86 <li><a href="#i_srem">'<tt>srem</tt>' Instruction</a></li>
87 <li><a href="#i_frem">'<tt>frem</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]88 </ol>
89 </li>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]90 <li><a href="#bitwiseops">Bitwise Binary Operations</a>
91 <ol>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]92 <li><a href="#i_and">'<tt>and</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]93 <li><a href="#i_or">'<tt>or</tt>' Instruction</a></li>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]94 <li><a href="#i_xor">'<tt>xor</tt>' Instruction</a></li>
95 <li><a href="#i_shl">'<tt>shl</tt>' Instruction</a></li>
Reid Spencer3822ff52006-11-08 06:47:33 +0000[diff] [blame]96 <li><a href="#i_lshr">'<tt>lshr</tt>' Instruction</a></li>
97 <li><a href="#i_ashr">'<tt>ashr</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]98 </ol>
99 </li>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]100 <li><a href="#vectorops">Vector Operations</a>
101 <ol>
102 <li><a href="#i_extractelement">'<tt>extractelement</tt>' Instruction</a></li>
103 <li><a href="#i_insertelement">'<tt>insertelement</tt>' Instruction</a></li>
104 <li><a href="#i_shufflevector">'<tt>shufflevector</tt>' Instruction</a></li>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]105 </ol>
106 </li>
Chris Lattner884a9702006-08-15 00:45:58 +0000[diff] [blame]107 <li><a href="#memoryops">Memory Access and Addressing Operations</a>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]108 <ol>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]109 <li><a href="#i_malloc">'<tt>malloc</tt>' Instruction</a></li>
110 <li><a href="#i_free">'<tt>free</tt>' Instruction</a></li>
111 <li><a href="#i_alloca">'<tt>alloca</tt>' Instruction</a></li>
Robert Bocchino7b81c752006-02-17 21:18:08 +0000[diff] [blame]112 <li><a href="#i_load">'<tt>load</tt>' Instruction</a></li>
113 <li><a href="#i_store">'<tt>store</tt>' Instruction</a></li>
114 <li><a href="#i_getelementptr">'<tt>getelementptr</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]115 </ol>
116 </li>
Reid Spencer2fd21e62006-11-08 01:18:52 +0000[diff] [blame]117 <li><a href="#convertops">Conversion Operations</a>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]118 <ol>
119 <li><a href="#i_trunc">'<tt>trunc .. to</tt>' Instruction</a></li>
120 <li><a href="#i_zext">'<tt>zext .. to</tt>' Instruction</a></li>
121 <li><a href="#i_sext">'<tt>sext .. to</tt>' Instruction</a></li>
122 <li><a href="#i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a></li>
123 <li><a href="#i_fpext">'<tt>fpext .. to</tt>' Instruction</a></li>
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]124 <li><a href="#i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a></li>
125 <li><a href="#i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a></li>
126 <li><a href="#i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a></li>
127 <li><a href="#i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a></li>
Reid Spencer72679252006-11-11 21:00:47 +0000[diff] [blame]128 <li><a href="#i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a></li>
129 <li><a href="#i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a></li>
Reid Spencer5c0ef472006-11-11 23:08:07 +0000[diff] [blame]130 <li><a href="#i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a></li>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]131 </ol>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]132 <li><a href="#otherops">Other Operations</a>
133 <ol>
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]134 <li><a href="#i_icmp">'<tt>icmp</tt>' Instruction</a></li>
135 <li><a href="#i_fcmp">'<tt>fcmp</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]136 <li><a href="#i_phi">'<tt>phi</tt>' Instruction</a></li>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]137 <li><a href="#i_select">'<tt>select</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]138 <li><a href="#i_call">'<tt>call</tt>' Instruction</a></li>
Chris Lattnerfb6977d2006-01-13 23:26:01 +0000[diff] [blame]139 <li><a href="#i_va_arg">'<tt>va_arg</tt>' Instruction</a></li>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]140 </ol>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]141 </li>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]142 </ol>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]143 </li>
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]144 <li><a href="#intrinsics">Intrinsic Functions</a>
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]145 <ol>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]146 <li><a href="#int_varargs">Variable Argument Handling Intrinsics</a>
147 <ol>
148 <li><a href="#i_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a></li>
149 <li><a href="#i_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a></li>
150 <li><a href="#i_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a></li>
151 </ol>
152 </li>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]153 <li><a href="#int_gc">Accurate Garbage Collection Intrinsics</a>
154 <ol>
155 <li><a href="#i_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a></li>
156 <li><a href="#i_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a></li>
157 <li><a href="#i_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a></li>
158 </ol>
159 </li>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]160 <li><a href="#int_codegen">Code Generator Intrinsics</a>
161 <ol>
162 <li><a href="#i_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a></li>
163 <li><a href="#i_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a></li>
Chris Lattner57e1f392006-01-13 02:03:13 +0000[diff] [blame]164 <li><a href="#i_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a></li>
165 <li><a href="#i_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a></li>
Chris Lattner9a9d7ac2005-02-28 19:24:19 +0000[diff] [blame]166 <li><a href="#i_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a></li>
Andrew Lenharth7f4ec3b2005-03-28 20:05:49 +0000[diff] [blame]167 <li><a href="#i_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a></li>
Andrew Lenharth51b8d542005-11-11 16:47:30 +0000[diff] [blame]168 <li><a href="#i_readcyclecounter"><tt>llvm.readcyclecounter</tt>' Intrinsic</a></li>
John Criswell7123e272004-04-09 16:43:20 +0000[diff] [blame]169 </ol>
170 </li>
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]171 <li><a href="#int_libc">Standard C Library Intrinsics</a>
172 <ol>
Chris Lattner5b310c32006-03-03 00:07:20 +0000[diff] [blame]173 <li><a href="#i_memcpy">'<tt>llvm.memcpy.*</tt>' Intrinsic</a></li>
174 <li><a href="#i_memmove">'<tt>llvm.memmove.*</tt>' Intrinsic</a></li>
175 <li><a href="#i_memset">'<tt>llvm.memset.*</tt>' Intrinsic</a></li>
Chris Lattnerec6cb612006-01-16 22:38:59 +0000[diff] [blame]176 <li><a href="#i_isunordered">'<tt>llvm.isunordered.*</tt>' Intrinsic</a></li>
177 <li><a href="#i_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a></li>
Chris Lattnerf4d252d2006-09-08 06:34:02 +0000[diff] [blame]178 <li><a href="#i_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a></li>
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]179 </ol>
180 </li>
Nate Begeman7e36c472006-01-13 23:26:38 +0000[diff] [blame]181 <li><a href="#int_manip">Bit Manipulation Intrinsics</a>
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]182 <ol>
Nate Begeman7e36c472006-01-13 23:26:38 +0000[diff] [blame]183 <li><a href="#i_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a></li>
Chris Lattner8a886be2006-01-16 22:34:14 +0000[diff] [blame]184 <li><a href="#int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic </a></li>
185 <li><a href="#int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic </a></li>
186 <li><a href="#int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic </a></li>
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]187 </ol>
188 </li>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]189 <li><a href="#int_debugger">Debugger intrinsics</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]190 </ol>
191 </li>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]192</ol>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]193
194<div class="doc_author">
195 <p>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a>
196 and <a href="mailto:vadve@cs.uiuc.edu">Vikram Adve</a></p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]197</div>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]198
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]199<!-- *********************************************************************** -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]200<div class="doc_section"> <a name="abstract">Abstract </a></div>
201<!-- *********************************************************************** -->
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]202
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]203<div class="doc_text">
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]204<p>This document is a reference manual for the LLVM assembly language.
205LLVM is an SSA based representation that provides type safety,
206low-level operations, flexibility, and the capability of representing
207'all' high-level languages cleanly. It is the common code
208representation used throughout all phases of the LLVM compilation
209strategy.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]210</div>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]211
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]212<!-- *********************************************************************** -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]213<div class="doc_section"> <a name="introduction">Introduction</a> </div>
214<!-- *********************************************************************** -->
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]215
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]216<div class="doc_text">
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]217
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]218<p>The LLVM code representation is designed to be used in three
219different forms: as an in-memory compiler IR, as an on-disk bytecode
220representation (suitable for fast loading by a Just-In-Time compiler),
221and as a human readable assembly language representation. This allows
222LLVM to provide a powerful intermediate representation for efficient
223compiler transformations and analysis, while providing a natural means
224to debug and visualize the transformations. The three different forms
225of LLVM are all equivalent. This document describes the human readable
226representation and notation.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]227
John Criswellc1f786c2005-05-13 22:25:59 +0000[diff] [blame]228<p>The LLVM representation aims to be light-weight and low-level
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]229while being expressive, typed, and extensible at the same time. It
230aims to be a "universal IR" of sorts, by being at a low enough level
231that high-level ideas may be cleanly mapped to it (similar to how
232microprocessors are "universal IR's", allowing many source languages to
233be mapped to them). By providing type information, LLVM can be used as
234the target of optimizations: for example, through pointer analysis, it
235can be proven that a C automatic variable is never accessed outside of
236the current function... allowing it to be promoted to a simple SSA
237value instead of a memory location.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]238
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]239</div>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]240
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]241<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]242<div class="doc_subsubsection"> <a name="wellformed">Well-Formedness</a> </div>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]243
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]244<div class="doc_text">
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]245
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]246<p>It is important to note that this document describes 'well formed'
247LLVM assembly language. There is a difference between what the parser
248accepts and what is considered 'well formed'. For example, the
249following instruction is syntactically okay, but not well formed:</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]250
251<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]252 %x = <a href="#i_add">add</a> i32 1, %x
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]253</pre>
254
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]255<p>...because the definition of <tt>%x</tt> does not dominate all of
256its uses. The LLVM infrastructure provides a verification pass that may
257be used to verify that an LLVM module is well formed. This pass is
John Criswellc1f786c2005-05-13 22:25:59 +0000[diff] [blame]258automatically run by the parser after parsing input assembly and by
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]259the optimizer before it outputs bytecode. The violations pointed out
260by the verifier pass indicate bugs in transformation passes or input to
261the parser.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]262
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]263<!-- Describe the typesetting conventions here. --> </div>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]264
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]265<!-- *********************************************************************** -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]266<div class="doc_section"> <a name="identifiers">Identifiers</a> </div>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]267<!-- *********************************************************************** -->
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]268
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]269<div class="doc_text">
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]270
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]271<p>LLVM uses three different forms of identifiers, for different
272purposes:</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]273
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]274<ol>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]275 <li>Named values are represented as a string of characters with a '%' prefix.
276 For example, %foo, %DivisionByZero, %a.really.long.identifier. The actual
277 regular expression used is '<tt>%[a-zA-Z$._][a-zA-Z$._0-9]*</tt>'.
278 Identifiers which require other characters in their names can be surrounded
279 with quotes. In this way, anything except a <tt>"</tt> character can be used
280 in a name.</li>
281
282 <li>Unnamed values are represented as an unsigned numeric value with a '%'
283 prefix. For example, %12, %2, %44.</li>
284
Reid Spencercc16dc32004-12-09 18:02:53 +0000[diff] [blame]285 <li>Constants, which are described in a <a href="#constants">section about
286 constants</a>, below.</li>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]287</ol>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]288
289<p>LLVM requires that values start with a '%' sign for two reasons: Compilers
290don't need to worry about name clashes with reserved words, and the set of
291reserved words may be expanded in the future without penalty. Additionally,
292unnamed identifiers allow a compiler to quickly come up with a temporary
293variable without having to avoid symbol table conflicts.</p>
294
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]295<p>Reserved words in LLVM are very similar to reserved words in other
Reid Spencer5c0ef472006-11-11 23:08:07 +0000[diff] [blame]296languages. There are keywords for different opcodes
297('<tt><a href="#i_add">add</a></tt>',
298 '<tt><a href="#i_bitcast">bitcast</a></tt>',
299 '<tt><a href="#i_ret">ret</a></tt>', etc...), for primitive type names ('<tt><a
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]300href="#t_void">void</a></tt>', '<tt><a href="#t_primitive">i32</a></tt>', etc...),
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]301and others. These reserved words cannot conflict with variable names, because
302none of them start with a '%' character.</p>
303
304<p>Here is an example of LLVM code to multiply the integer variable
305'<tt>%X</tt>' by 8:</p>
306
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]307<p>The easy way:</p>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]308
309<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]310 %result = <a href="#i_mul">mul</a> i32 %X, 8
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]311</pre>
312
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]313<p>After strength reduction:</p>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]314
315<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]316 %result = <a href="#i_shl">shl</a> i32 %X, i8 3
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]317</pre>
318
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]319<p>And the hard way:</p>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]320
321<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]322 <a href="#i_add">add</a> i32 %X, %X <i>; yields {i32}:%0</i>
323 <a href="#i_add">add</a> i32 %0, %0 <i>; yields {i32}:%1</i>
324 %result = <a href="#i_add">add</a> i32 %1, %1
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]325</pre>
326
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]327<p>This last way of multiplying <tt>%X</tt> by 8 illustrates several
328important lexical features of LLVM:</p>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]329
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]330<ol>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]331
332 <li>Comments are delimited with a '<tt>;</tt>' and go until the end of
333 line.</li>
334
335 <li>Unnamed temporaries are created when the result of a computation is not
336 assigned to a named value.</li>
337
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]338 <li>Unnamed temporaries are numbered sequentially</li>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]339
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]340</ol>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]341
John Criswelle4c57cc2005-05-12 16:52:32 +0000[diff] [blame]342<p>...and it also shows a convention that we follow in this document. When
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]343demonstrating instructions, we will follow an instruction with a comment that
344defines the type and name of value produced. Comments are shown in italic
345text.</p>
346
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]347</div>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]348
349<!-- *********************************************************************** -->
350<div class="doc_section"> <a name="highlevel">High Level Structure</a> </div>
351<!-- *********************************************************************** -->
352
353<!-- ======================================================================= -->
354<div class="doc_subsection"> <a name="modulestructure">Module Structure</a>
355</div>
356
357<div class="doc_text">
358
359<p>LLVM programs are composed of "Module"s, each of which is a
360translation unit of the input programs. Each module consists of
361functions, global variables, and symbol table entries. Modules may be
362combined together with the LLVM linker, which merges function (and
363global variable) definitions, resolves forward declarations, and merges
364symbol table entries. Here is an example of the "hello world" module:</p>
365
366<pre><i>; Declare the string constant as a global constant...</i>
367<a href="#identifiers">%.LC0</a> = <a href="#linkage_internal">internal</a> <a
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]368 href="#globalvars">constant</a> <a href="#t_array">[13 x i8 ]</a> c"hello world\0A\00" <i>; [13 x i8 ]*</i>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]369
370<i>; External declaration of the puts function</i>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]371<a href="#functionstructure">declare</a> i32 %puts(i8 *) <i>; i32(i8 *)* </i>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]372
Chris Lattner81c01f02006-06-13 03:05:47 +0000[diff] [blame]373<i>; Global variable / Function body section separator</i>
374implementation
375
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]376<i>; Definition of main function</i>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]377define i32 %main() { <i>; i32()* </i>
378 <i>; Convert [13x i8 ]* to i8 *...</i>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]379 %cast210 = <a
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]380 href="#i_getelementptr">getelementptr</a> [13 x i8 ]* %.LC0, i64 0, i64 0 <i>; i8 *</i>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]381
382 <i>; Call puts function to write out the string to stdout...</i>
383 <a
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]384 href="#i_call">call</a> i32 %puts(i8 * %cast210) <i>; i32</i>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]385 <a
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]386 href="#i_ret">ret</a> i32 0<br>}<br></pre>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]387
388<p>This example is made up of a <a href="#globalvars">global variable</a>
389named "<tt>.LC0</tt>", an external declaration of the "<tt>puts</tt>"
390function, and a <a href="#functionstructure">function definition</a>
391for "<tt>main</tt>".</p>
392
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]393<p>In general, a module is made up of a list of global values,
394where both functions and global variables are global values. Global values are
395represented by a pointer to a memory location (in this case, a pointer to an
396array of char, and a pointer to a function), and have one of the following <a
397href="#linkage">linkage types</a>.</p>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]398
Chris Lattner81c01f02006-06-13 03:05:47 +0000[diff] [blame]399<p>Due to a limitation in the current LLVM assembly parser (it is limited by
400one-token lookahead), modules are split into two pieces by the "implementation"
401keyword. Global variable prototypes and definitions must occur before the
402keyword, and function definitions must occur after it. Function prototypes may
403occur either before or after it. In the future, the implementation keyword may
404become a noop, if the parser gets smarter.</p>
405
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]406</div>
407
408<!-- ======================================================================= -->
409<div class="doc_subsection">
410 <a name="linkage">Linkage Types</a>
411</div>
412
413<div class="doc_text">
414
415<p>
416All Global Variables and Functions have one of the following types of linkage:
417</p>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]418
419<dl>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]420
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]421 <dt><tt><b><a name="linkage_internal">internal</a></b></tt> </dt>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]422
423 <dd>Global values with internal linkage are only directly accessible by
424 objects in the current module. In particular, linking code into a module with
425 an internal global value may cause the internal to be renamed as necessary to
426 avoid collisions. Because the symbol is internal to the module, all
427 references can be updated. This corresponds to the notion of the
428 '<tt>static</tt>' keyword in C, or the idea of "anonymous namespaces" in C++.
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]429 </dd>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]430
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]431 <dt><tt><b><a name="linkage_linkonce">linkonce</a></b></tt>: </dt>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]432
433 <dd>"<tt>linkonce</tt>" linkage is similar to <tt>internal</tt> linkage, with
434 the twist that linking together two modules defining the same
435 <tt>linkonce</tt> globals will cause one of the globals to be discarded. This
436 is typically used to implement inline functions. Unreferenced
437 <tt>linkonce</tt> globals are allowed to be discarded.
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]438 </dd>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]439
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]440 <dt><tt><b><a name="linkage_weak">weak</a></b></tt>: </dt>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]441
442 <dd>"<tt>weak</tt>" linkage is exactly the same as <tt>linkonce</tt> linkage,
443 except that unreferenced <tt>weak</tt> globals may not be discarded. This is
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]444 used to implement constructs in C such as "<tt>i32 X;</tt>" at global scope.
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]445 </dd>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]446
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]447 <dt><tt><b><a name="linkage_appending">appending</a></b></tt>: </dt>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]448
449 <dd>"<tt>appending</tt>" linkage may only be applied to global variables of
450 pointer to array type. When two global variables with appending linkage are
451 linked together, the two global arrays are appended together. This is the
452 LLVM, typesafe, equivalent of having the system linker append together
453 "sections" with identical names when .o files are linked.
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]454 </dd>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]455
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]456 <dt><tt><b><a name="linkage_external">externally visible</a></b></tt>:</dt>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]457
458 <dd>If none of the above identifiers are used, the global is externally
459 visible, meaning that it participates in linkage and can be used to resolve
460 external symbol references.
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]461 </dd>
Anton Korobeynikovb74ed072006-09-14 18:23:27 +0000[diff] [blame]462
463 <dt><tt><b><a name="linkage_externweak">extern_weak</a></b></tt>: </dt>
464
465 <dd>"<tt>extern_weak</tt>" TBD
466 </dd>
467
468 <p>
469 The next two types of linkage are targeted for Microsoft Windows platform
470 only. They are designed to support importing (exporting) symbols from (to)
471 DLLs.
472 </p>
473
474 <dt><tt><b><a name="linkage_dllimport">dllimport</a></b></tt>: </dt>
475
476 <dd>"<tt>dllimport</tt>" linkage causes the compiler to reference a function
477 or variable via a global pointer to a pointer that is set up by the DLL
478 exporting the symbol. On Microsoft Windows targets, the pointer name is
479 formed by combining <code>_imp__</code> and the function or variable name.
480 </dd>
481
482 <dt><tt><b><a name="linkage_dllexport">dllexport</a></b></tt>: </dt>
483
484 <dd>"<tt>dllexport</tt>" linkage causes the compiler to provide a global
485 pointer to a pointer in a DLL, so that it can be referenced with the
486 <tt>dllimport</tt> attribute. On Microsoft Windows targets, the pointer
487 name is formed by combining <code>_imp__</code> and the function or variable
488 name.
489 </dd>
490
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]491</dl>
492
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]493<p><a name="linkage_external">For example, since the "<tt>.LC0</tt>"
494variable is defined to be internal, if another module defined a "<tt>.LC0</tt>"
495variable and was linked with this one, one of the two would be renamed,
496preventing a collision. Since "<tt>main</tt>" and "<tt>puts</tt>" are
497external (i.e., lacking any linkage declarations), they are accessible
498outside of the current module. It is illegal for a function <i>declaration</i>
499to have any linkage type other than "externally visible".</a></p>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]500
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]501</div>
502
503<!-- ======================================================================= -->
504<div class="doc_subsection">
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]505 <a name="callingconv">Calling Conventions</a>
506</div>
507
508<div class="doc_text">
509
510<p>LLVM <a href="#functionstructure">functions</a>, <a href="#i_call">calls</a>
511and <a href="#i_invoke">invokes</a> can all have an optional calling convention
512specified for the call. The calling convention of any pair of dynamic
513caller/callee must match, or the behavior of the program is undefined. The
514following calling conventions are supported by LLVM, and more may be added in
515the future:</p>
516
517<dl>
518 <dt><b>"<tt>ccc</tt>" - The C calling convention</b>:</dt>
519
520 <dd>This calling convention (the default if no other calling convention is
521 specified) matches the target C calling conventions. This calling convention
John Criswelle4c57cc2005-05-12 16:52:32 +0000[diff] [blame]522 supports varargs function calls and tolerates some mismatch in the declared
Reid Spencer6f85b9d2006-12-28 16:55:55 +0000[diff] [blame]523 prototype and implemented declaration of the function (as does normal C). For
524 integer arguments less than 32-bits, the value will be sign-extended to
525 32-bits before the call is made. If zero-extension is required, use the
526 <tt>cextcc</tt> calling convention.
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]527 </dd>
528
Reid Spencer6f85b9d2006-12-28 16:55:55 +0000[diff] [blame]529 <dt><b>"<tt>cextcc(bitmask)</tt>" - The C with explicit extend calling
530 convention </b>:</dt>
531 <dd>This calling convention is exactly like the C calling convention except
532 that it is parameterized to provide a <tt>bitmask</tt> that indicates how
533 integer arguments of less than 32-bits should be extended. A zero bit
534 indicates zero-extension while a 1-bit indicates sign-extension. The least
535 significant bit always corresponds to the return type of the function. The
536 bits in the <tt>bitmask</tt> are assigned to the integer parameters of the
537 function that are smaller than 32-bits. For example, a bitmask of value
538 5 (0b0101) indicates that the return value is to be sign extended, the first
539 small integer argument is to be zero extended and the second small integer
540 argument is to be sign extended.</dd>
541
542
Chris Lattner5710ce92006-05-19 21:15:36 +0000[diff] [blame]543 <dt><b>"<tt>csretcc</tt>" - The C struct return calling convention</b>:</dt>
544
545 <dd>This calling convention matches the target C calling conventions, except
546 that functions with this convention are required to take a pointer as their
547 first argument, and the return type of the function must be void. This is
548 used for C functions that return aggregates by-value. In this case, the
549 function has been transformed to take a pointer to the struct as the first
550 argument to the function. For targets where the ABI specifies specific
551 behavior for structure-return calls, the calling convention can be used to
552 distinguish between struct return functions and other functions that take a
553 pointer to a struct as the first argument.
554 </dd>
555
Reid Spencer6f85b9d2006-12-28 16:55:55 +0000[diff] [blame]556 <dt><b>"<tt>csretextcc(bitmask)</tt>" - The C struct return with explicit
557 extend calling convention</b>:</dt>
558 <dd>This calling convention is exactly like the <tt>csret</tt> calling
559 convention except that it is parameterized to provide a <tt>bitmask</tt>
560 that indicates how integer arguments of less than 32-bits should be extended.
561 A zero bit indicates zero-extension while a 1-bit indicates sign-extension.
562 </dd>
563
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]564 <dt><b>"<tt>fastcc</tt>" - The fast calling convention</b>:</dt>
565
566 <dd>This calling convention attempts to make calls as fast as possible
567 (e.g. by passing things in registers). This calling convention allows the
568 target to use whatever tricks it wants to produce fast code for the target,
Chris Lattner8cdc5bc2005-05-06 23:08:23 +0000[diff] [blame]569 without having to conform to an externally specified ABI. Implementations of
570 this convention should allow arbitrary tail call optimization to be supported.
571 This calling convention does not support varargs and requires the prototype of
572 all callees to exactly match the prototype of the function definition.
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]573 </dd>
574
575 <dt><b>"<tt>coldcc</tt>" - The cold calling convention</b>:</dt>
576
577 <dd>This calling convention attempts to make code in the caller as efficient
578 as possible under the assumption that the call is not commonly executed. As
579 such, these calls often preserve all registers so that the call does not break
580 any live ranges in the caller side. This calling convention does not support
581 varargs and requires the prototype of all callees to exactly match the
582 prototype of the function definition.
583 </dd>
584
Chris Lattnercfe6b372005-05-07 01:46:40 +0000[diff] [blame]585 <dt><b>"<tt>cc &lt;<em>n</em>&gt;</tt>" - Numbered convention</b>:</dt>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]586
587 <dd>Any calling convention may be specified by number, allowing
588 target-specific calling conventions to be used. Target specific calling
589 conventions start at 64.
590 </dd>
Chris Lattnercfe6b372005-05-07 01:46:40 +0000[diff] [blame]591</dl>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]592
593<p>More calling conventions can be added/defined on an as-needed basis, to
594support pascal conventions or any other well-known target-independent
595convention.</p>
596
597</div>
598
599<!-- ======================================================================= -->
600<div class="doc_subsection">
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]601 <a name="globalvars">Global Variables</a>
602</div>
603
604<div class="doc_text">
605
Chris Lattner3689a342005-02-12 19:30:21 +0000[diff] [blame]606<p>Global variables define regions of memory allocated at compilation time
Chris Lattner88f6c462005-11-12 00:45:07 +0000[diff] [blame]607instead of run-time. Global variables may optionally be initialized, may have
608an explicit section to be placed in, and may
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]609have an optional explicit alignment specified. A
John Criswell0ec250c2005-10-24 16:17:18 +0000[diff] [blame]610variable may be defined as a global "constant," which indicates that the
Chris Lattner3689a342005-02-12 19:30:21 +0000[diff] [blame]611contents of the variable will <b>never</b> be modified (enabling better
612optimization, allowing the global data to be placed in the read-only section of
613an executable, etc). Note that variables that need runtime initialization
John Criswell0ec250c2005-10-24 16:17:18 +0000[diff] [blame]614cannot be marked "constant" as there is a store to the variable.</p>
Chris Lattner3689a342005-02-12 19:30:21 +0000[diff] [blame]615
616<p>
617LLVM explicitly allows <em>declarations</em> of global variables to be marked
618constant, even if the final definition of the global is not. This capability
619can be used to enable slightly better optimization of the program, but requires
620the language definition to guarantee that optimizations based on the
621'constantness' are valid for the translation units that do not include the
622definition.
623</p>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]624
625<p>As SSA values, global variables define pointer values that are in
626scope (i.e. they dominate) all basic blocks in the program. Global
627variables always define a pointer to their "content" type because they
628describe a region of memory, and all memory objects in LLVM are
629accessed through pointers.</p>
630
Chris Lattner88f6c462005-11-12 00:45:07 +0000[diff] [blame]631<p>LLVM allows an explicit section to be specified for globals. If the target
632supports it, it will emit globals to the section specified.</p>
633
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]634<p>An explicit alignment may be specified for a global. If not present, or if
635the alignment is set to zero, the alignment of the global is set by the target
636to whatever it feels convenient. If an explicit alignment is specified, the
637global is forced to have at least that much alignment. All alignments must be
638a power of 2.</p>
639
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]640</div>
641
642
643<!-- ======================================================================= -->
644<div class="doc_subsection">
645 <a name="functionstructure">Functions</a>
646</div>
647
648<div class="doc_text">
649
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]650<p>LLVM function definitions consist of the "<tt>define</tt>" keyord,
651an optional <a href="#linkage">linkage type</a>, an optional
652<a href="#callingconv">calling convention</a>, a return type, an optional
653<a href="#paramattrs">parameter attribute</a> for the return type, a function
654name, a (possibly empty) argument list (each with optional
Reid Spencer92f82302006-12-31 07:18:34 +0000[diff] [blame]655<a href="#paramattrs">parameter attributes</a>), an optional section, an
656optional alignment, an opening curly brace, a list of basic blocks, and a
657closing curly brace. LLVM function declarations
658consist of the "<tt>declare</tt>" keyword, an optional <a
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]659 href="#callingconv">calling convention</a>, a return type, an optional
660<a href="#paramattrs">parameter attribute</a> for the return type, a function
661name, a possibly empty list of arguments, and an optional alignment.</p>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]662
663<p>A function definition contains a list of basic blocks, forming the CFG for
664the function. Each basic block may optionally start with a label (giving the
665basic block a symbol table entry), contains a list of instructions, and ends
666with a <a href="#terminators">terminator</a> instruction (such as a branch or
667function return).</p>
668
John Criswelle4c57cc2005-05-12 16:52:32 +0000[diff] [blame]669<p>The first basic block in a program is special in two ways: it is immediately
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]670executed on entrance to the function, and it is not allowed to have predecessor
671basic blocks (i.e. there can not be any branches to the entry block of a
672function). Because the block can have no predecessors, it also cannot have any
673<a href="#i_phi">PHI nodes</a>.</p>
674
675<p>LLVM functions are identified by their name and type signature. Hence, two
676functions with the same name but different parameter lists or return values are
Chris Lattnerd4f6b172005-03-07 22:13:59 +0000[diff] [blame]677considered different functions, and LLVM will resolve references to each
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]678appropriately.</p>
679
Chris Lattner88f6c462005-11-12 00:45:07 +0000[diff] [blame]680<p>LLVM allows an explicit section to be specified for functions. If the target
681supports it, it will emit functions to the section specified.</p>
682
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]683<p>An explicit alignment may be specified for a function. If not present, or if
684the alignment is set to zero, the alignment of the function is set by the target
685to whatever it feels convenient. If an explicit alignment is specified, the
686function is forced to have at least that much alignment. All alignments must be
687a power of 2.</p>
688
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]689</div>
690
Chris Lattner4e9aba72006-01-23 23:23:47 +0000[diff] [blame]691<!-- ======================================================================= -->
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]692<div class="doc_subsection"><a name="paramattrs">Parameter Attributes</a></div>
693<div class="doc_text">
694 <p>The return type and each parameter of a function type may have a set of
695 <i>parameter attributes</i> associated with them. Parameter attributes are
696 used to communicate additional information about the result or parameters of
697 a function. Parameter attributes are considered to be part of the function
698 type so two functions types that differ only by the parameter attributes
699 are different function types.</p>
700
Reid Spencer92f82302006-12-31 07:18:34 +0000[diff] [blame]701 <p>Parameter attributes consist of an at sign (@) followed by either a single
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]702 keyword or a comma separate list of keywords enclosed in parentheses. For
703 example:<pre>
704 %someFunc = i16 @zext (i8 @(sext) %someParam)
Reid Spencer92f82302006-12-31 07:18:34 +0000[diff] [blame]705 %someFunc = i16 @zext (i8 @zext %someParam)</pre>
706 Note that the two function types above are unique because the parameter
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]707 has a different attribute (@sext in the first one, @zext in the second).</p>
708
709 <p>Currently, only the following parameter attributes are defined:
710 <dl>
711 <dt><tt>@zext</tt></dt>
712 <dd>This indicates that the parameter should be zero extended just before
713 a call to this function.</dd>
714 <dt><tt>@sext</tt></dt>
715 <dd>This indicates that the parameter should be sign extended just before
716 a call to this function.</dd>
717 </dl></p>
718
719 <p>The current motivation for parameter attributes is to enable the sign and
720 zero extend information necessary for the C calling convention to be passed
721 from the front end to LLVM. The <tt>@zext</tt> and <tt>@sext</tt> attributes
722 are used by the code generator to perform the required extension. However,
723 parameter attributes are an orthogonal feature to calling conventions and
724 may be used for other purposes in the future.</p>
725</div>
726
727<!-- ======================================================================= -->
Chris Lattner4e9aba72006-01-23 23:23:47 +0000[diff] [blame]728<div class="doc_subsection">
Chris Lattner1eeeb0c2006-04-08 04:40:53 +0000[diff] [blame]729 <a name="moduleasm">Module-Level Inline Assembly</a>
Chris Lattner4e9aba72006-01-23 23:23:47 +0000[diff] [blame]730</div>
731
732<div class="doc_text">
733<p>
734Modules may contain "module-level inline asm" blocks, which corresponds to the
735GCC "file scope inline asm" blocks. These blocks are internally concatenated by
736LLVM and treated as a single unit, but may be separated in the .ll file if
737desired. The syntax is very simple:
738</p>
739
740<div class="doc_code"><pre>
Chris Lattner52599e12006-01-24 00:37:20 +0000[diff] [blame]741 module asm "inline asm code goes here"
742 module asm "more can go here"
Chris Lattner4e9aba72006-01-23 23:23:47 +0000[diff] [blame]743</pre></div>
744
745<p>The strings can contain any character by escaping non-printable characters.
746 The escape sequence used is simply "\xx" where "xx" is the two digit hex code
747 for the number.
748</p>
749
750<p>
751 The inline asm code is simply printed to the machine code .s file when
752 assembly code is generated.
753</p>
754</div>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]755
756
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]757<!-- *********************************************************************** -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]758<div class="doc_section"> <a name="typesystem">Type System</a> </div>
759<!-- *********************************************************************** -->
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]760
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]761<div class="doc_text">
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]762
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]763<p>The LLVM type system is one of the most important features of the
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]764intermediate representation. Being typed enables a number of
765optimizations to be performed on the IR directly, without having to do
766extra analyses on the side before the transformation. A strong type
767system makes it easier to read the generated code and enables novel
768analyses and transformations that are not feasible to perform on normal
769three address code representations.</p>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]770
771</div>
772
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]773<!-- ======================================================================= -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]774<div class="doc_subsection"> <a name="t_primitive">Primitive Types</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]775<div class="doc_text">
John Criswell4457dc92004-04-09 16:48:45 +0000[diff] [blame]776<p>The primitive types are the fundamental building blocks of the LLVM
Chris Lattnerd4f6b172005-03-07 22:13:59 +0000[diff] [blame]777system. The current set of primitive types is as follows:</p>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000[diff] [blame]778
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]779<table class="layout">
780 <tr class="layout">
781 <td class="left">
782 <table>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]783 <tbody>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]784 <tr><th>Type</th><th>Description</th></tr>
785 <tr><td><tt>void</tt></td><td>No value</td></tr>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]786 <tr><td><tt>i8</tt></td><td>Signless 8-bit value</td></tr>
787 <tr><td><tt>i32</tt></td><td>Signless 32-bit value</td></tr>
Misha Brukmancfa87bc2005-04-22 18:02:52 +0000[diff] [blame]788 <tr><td><tt>float</tt></td><td>32-bit floating point value</td></tr>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]789 <tr><td><tt>label</tt></td><td>Branch destination</td></tr>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]790 </tbody>
791 </table>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]792 </td>
793 <td class="right">
794 <table>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]795 <tbody>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]796 <tr><th>Type</th><th>Description</th></tr>
797 <tr><td><tt>bool</tt></td><td>True or False value</td></tr>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]798 <tr><td><tt>i16</tt></td><td>Signless 16-bit value</td></tr>
799 <tr><td><tt>i64</tt></td><td>Signless 64-bit value</td></tr>
800 <tr><td><tt>double</tt></td><td>64-bit floating point value</td></tr>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]801 </tbody>
802 </table>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]803 </td>
804 </tr>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]805</table>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]806</div>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]807
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]808<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]809<div class="doc_subsubsection"> <a name="t_classifications">Type
810Classifications</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]811<div class="doc_text">
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]812<p>These different primitive types fall into a few useful
813classifications:</p>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000[diff] [blame]814
815<table border="1" cellspacing="0" cellpadding="4">
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]816 <tbody>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]817 <tr><th>Classification</th><th>Types</th></tr>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]818 <tr>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]819 <td><a name="t_integer">integer</a></td>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]820 <td><tt>i8, i16, i32, i64</tt></td>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]821 </tr>
822 <tr>
823 <td><a name="t_integral">integral</a></td>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]824 <td><tt>bool, i8, i16, i32, i64</tt>
Misha Brukmanc24b7582004-08-12 20:16:08 +0000[diff] [blame]825 </td>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]826 </tr>
827 <tr>
828 <td><a name="t_floating">floating point</a></td>
829 <td><tt>float, double</tt></td>
830 </tr>
831 <tr>
832 <td><a name="t_firstclass">first class</a></td>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]833 <td><tt>bool, i8, i16, i32, i64, float, double, <br/>
834 <a href="#t_pointer">pointer</a>,<a href="#t_packed">packed</a></tt>
835 </td>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]836 </tr>
837 </tbody>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]838</table>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000[diff] [blame]839
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]840<p>The <a href="#t_firstclass">first class</a> types are perhaps the
841most important. Values of these types are the only ones which can be
842produced by instructions, passed as arguments, or used as operands to
843instructions. This means that all structures and arrays must be
844manipulated either by pointer or by component.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]845</div>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]846
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]847<!-- ======================================================================= -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]848<div class="doc_subsection"> <a name="t_derived">Derived Types</a> </div>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]849
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]850<div class="doc_text">
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]851
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]852<p>The real power in LLVM comes from the derived types in the system.
853This is what allows a programmer to represent arrays, functions,
854pointers, and other useful types. Note that these derived types may be
855recursive: For example, it is possible to have a two dimensional array.</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]856
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]857</div>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]858
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]859<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]860<div class="doc_subsubsection"> <a name="t_array">Array Type</a> </div>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]861
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]862<div class="doc_text">
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]863
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]864<h5>Overview:</h5>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]865
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]866<p>The array type is a very simple derived type that arranges elements
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]867sequentially in memory. The array type requires a size (number of
868elements) and an underlying data type.</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]869
Chris Lattner7faa8832002-04-14 06:13:44 +0000[diff] [blame]870<h5>Syntax:</h5>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]871
872<pre>
873 [&lt;# elements&gt; x &lt;elementtype&gt;]
874</pre>
875
John Criswelle4c57cc2005-05-12 16:52:32 +0000[diff] [blame]876<p>The number of elements is a constant integer value; elementtype may
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]877be any type with a size.</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]878
Chris Lattner7faa8832002-04-14 06:13:44 +0000[diff] [blame]879<h5>Examples:</h5>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]880<table class="layout">
881 <tr class="layout">
882 <td class="left">
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]883 <tt>[40 x i32 ]</tt><br/>
884 <tt>[41 x i32 ]</tt><br/>
885 <tt>[40 x i32]</tt><br/>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]886 </td>
887 <td class="left">
888 Array of 40 integer values.<br/>
889 Array of 41 integer values.<br/>
890 Array of 40 unsigned integer values.<br/>
891 </td>
892 </tr>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]893</table>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]894<p>Here are some examples of multidimensional arrays:</p>
895<table class="layout">
896 <tr class="layout">
897 <td class="left">
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]898 <tt>[3 x [4 x i32]]</tt><br/>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]899 <tt>[12 x [10 x float]]</tt><br/>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]900 <tt>[2 x [3 x [4 x i32]]]</tt><br/>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]901 </td>
902 <td class="left">
John Criswellc1f786c2005-05-13 22:25:59 +0000[diff] [blame]903 3x4 array of integer values.<br/>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]904 12x10 array of single precision floating point values.<br/>
905 2x3x4 array of unsigned integer values.<br/>
906 </td>
907 </tr>
908</table>
Chris Lattnere67a9512005-06-24 17:22:57 +0000[diff] [blame]909
John Criswell0ec250c2005-10-24 16:17:18 +0000[diff] [blame]910<p>Note that 'variable sized arrays' can be implemented in LLVM with a zero
911length array. Normally, accesses past the end of an array are undefined in
Chris Lattnere67a9512005-06-24 17:22:57 +0000[diff] [blame]912LLVM (e.g. it is illegal to access the 5th element of a 3 element array).
913As a special case, however, zero length arrays are recognized to be variable
914length. This allows implementation of 'pascal style arrays' with the LLVM
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]915type "{ i32, [0 x float]}", for example.</p>
Chris Lattnere67a9512005-06-24 17:22:57 +0000[diff] [blame]916
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]917</div>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]918
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]919<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]920<div class="doc_subsubsection"> <a name="t_function">Function Type</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]921<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]922<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]923<p>The function type can be thought of as a function signature. It
924consists of a return type and a list of formal parameter types.
John Criswell009900b2003-11-25 21:45:46 +0000[diff] [blame]925Function types are usually used to build virtual function tables
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]926(which are structures of pointers to functions), for indirect function
927calls, and when defining a function.</p>
John Criswell009900b2003-11-25 21:45:46 +0000[diff] [blame]928<p>
929The return type of a function type cannot be an aggregate type.
930</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]931<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]932<pre> &lt;returntype&gt; (&lt;parameter list&gt;)<br></pre>
John Criswell0ec250c2005-10-24 16:17:18 +0000[diff] [blame]933<p>...where '<tt>&lt;parameter list&gt;</tt>' is a comma-separated list of type
Misha Brukmanc24b7582004-08-12 20:16:08 +0000[diff] [blame]934specifiers. Optionally, the parameter list may include a type <tt>...</tt>,
Chris Lattner27f71f22003-09-03 00:41:47 +0000[diff] [blame]935which indicates that the function takes a variable number of arguments.
936Variable argument functions can access their arguments with the <a
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]937 href="#int_varargs">variable argument handling intrinsic</a> functions.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]938<h5>Examples:</h5>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]939<table class="layout">
940 <tr class="layout">
Reid Spencer92f82302006-12-31 07:18:34 +0000[diff] [blame]941 <td class="left"><tt>i32 (i32)</tt></td>
942 <td class="left">function taking an <tt>i32</tt>, returning an <tt>i32</tt>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]943 </td>
Reid Spencer92f82302006-12-31 07:18:34 +0000[diff] [blame]944 </tr><tr class="layout">
Reid Spencerf17a0b72006-12-31 07:20:23 +0000[diff] [blame^]945 <td class="left"><tt>float&nbsp;(i16&nbsp;@sext,&nbsp;i32&nbsp;*)&nbsp;*
946 </tt></td>
Reid Spencer92f82302006-12-31 07:18:34 +0000[diff] [blame]947 <td class="left"><a href="#t_pointer">Pointer</a> to a function that takes
948 an <tt>i16</tt> that should be sign extended and a
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]949 <a href="#t_pointer">pointer</a> to <tt>i32</tt>, returning
Reid Spencer92f82302006-12-31 07:18:34 +0000[diff] [blame]950 <tt>float</tt>.
951 </td>
952 </tr><tr class="layout">
953 <td class="left"><tt>i32 (i8*, ...)</tt></td>
954 <td class="left">A vararg function that takes at least one
955 <a href="#t_pointer">pointer</a> to <tt>i8 </tt> (signed char in C),
956 which returns an integer. This is the signature for <tt>printf</tt> in
957 LLVM.
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]958 </td>
959 </tr>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]960</table>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000[diff] [blame]961
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]962</div>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]963<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]964<div class="doc_subsubsection"> <a name="t_struct">Structure Type</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]965<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]966<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]967<p>The structure type is used to represent a collection of data members
968together in memory. The packing of the field types is defined to match
969the ABI of the underlying processor. The elements of a structure may
970be any type that has a size.</p>
971<p>Structures are accessed using '<tt><a href="#i_load">load</a></tt>
972and '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a
973field with the '<tt><a href="#i_getelementptr">getelementptr</a></tt>'
974instruction.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]975<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]976<pre> { &lt;type list&gt; }<br></pre>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]977<h5>Examples:</h5>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]978<table class="layout">
979 <tr class="layout">
980 <td class="left">
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]981 <tt>{ i32, i32, i32 }</tt><br/>
982 <tt>{ float, i32 (i32) * }</tt><br/>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]983 </td>
984 <td class="left">
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]985 a triple of three <tt>i32</tt> values<br/>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]986 A pair, where the first element is a <tt>float</tt> and the second element
987 is a <a href="#t_pointer">pointer</a> to a <a href="#t_function">function</a>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]988 that takes an <tt>i32</tt>, returning an <tt>i32</tt>.<br/>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]989 </td>
990 </tr>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]991</table>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]992</div>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]993
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]994<!-- _______________________________________________________________________ -->
Andrew Lenharth75e10682006-12-08 17:13:00 +0000[diff] [blame]995<div class="doc_subsubsection"> <a name="t_pstruct">Packed Structure Type</a>
996</div>
997<div class="doc_text">
998<h5>Overview:</h5>
999<p>The packed structure type is used to represent a collection of data members
1000together in memory. There is no padding between fields. Further, the alignment
1001of a packed structure is 1 byte. The elements of a packed structure may
1002be any type that has a size.</p>
1003<p>Structures are accessed using '<tt><a href="#i_load">load</a></tt>
1004and '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a
1005field with the '<tt><a href="#i_getelementptr">getelementptr</a></tt>'
1006instruction.</p>
1007<h5>Syntax:</h5>
1008<pre> &lt; { &lt;type list&gt; } &gt; <br></pre>
1009<h5>Examples:</h5>
1010<table class="layout">
1011 <tr class="layout">
1012 <td class="left">
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]1013 <tt> &lt; { i32, i32, i32 } &gt; </tt><br/>
1014 <tt> &lt; { float, i32 (i32) * } &gt; </tt><br/>
Andrew Lenharth75e10682006-12-08 17:13:00 +0000[diff] [blame]1015 </td>
1016 <td class="left">
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]1017 a triple of three <tt>i32</tt> values<br/>
Andrew Lenharth75e10682006-12-08 17:13:00 +0000[diff] [blame]1018 A pair, where the first element is a <tt>float</tt> and the second element
1019 is a <a href="#t_pointer">pointer</a> to a <a href="#t_function">function</a>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]1020 that takes an <tt>i32</tt>, returning an <tt>i32</tt>.<br/>
Andrew Lenharth75e10682006-12-08 17:13:00 +0000[diff] [blame]1021 </td>
1022 </tr>
1023</table>
1024</div>
1025
1026<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1027<div class="doc_subsubsection"> <a name="t_pointer">Pointer Type</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1028<div class="doc_text">
Chris Lattner7faa8832002-04-14 06:13:44 +0000[diff] [blame]1029<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1030<p>As in many languages, the pointer type represents a pointer or
1031reference to another object, which must live in memory.</p>
Chris Lattner7faa8832002-04-14 06:13:44 +0000[diff] [blame]1032<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1033<pre> &lt;type&gt; *<br></pre>
Chris Lattner7faa8832002-04-14 06:13:44 +0000[diff] [blame]1034<h5>Examples:</h5>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]1035<table class="layout">
1036 <tr class="layout">
1037 <td class="left">
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]1038 <tt>[4x i32]*</tt><br/>
1039 <tt>i32 (i32 *) *</tt><br/>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]1040 </td>
1041 <td class="left">
1042 A <a href="#t_pointer">pointer</a> to <a href="#t_array">array</a> of
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]1043 four <tt>i32</tt> values<br/>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]1044 A <a href="#t_pointer">pointer</a> to a <a
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]1045 href="#t_function">function</a> that takes an <tt>i32*</tt>, returning an
1046 <tt>i32</tt>.<br/>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]1047 </td>
1048 </tr>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1049</table>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1050</div>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]1051
Chris Lattnera58561b2004-08-12 19:12:28 +0000[diff] [blame]1052<!-- _______________________________________________________________________ -->
1053<div class="doc_subsubsection"> <a name="t_packed">Packed Type</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1054<div class="doc_text">
Chris Lattner69c11bb2005-04-25 17:34:15 +0000[diff] [blame]1055
Chris Lattnera58561b2004-08-12 19:12:28 +0000[diff] [blame]1056<h5>Overview:</h5>
Chris Lattner69c11bb2005-04-25 17:34:15 +0000[diff] [blame]1057
Chris Lattnera58561b2004-08-12 19:12:28 +0000[diff] [blame]1058<p>A packed type is a simple derived type that represents a vector
1059of elements. Packed types are used when multiple primitive data
1060are operated in parallel using a single instruction (SIMD).
1061A packed type requires a size (number of
Chris Lattnerb8d172f2005-11-10 01:44:22 +0000[diff] [blame]1062elements) and an underlying primitive data type. Vectors must have a power
1063of two length (1, 2, 4, 8, 16 ...). Packed types are
Chris Lattnera58561b2004-08-12 19:12:28 +0000[diff] [blame]1064considered <a href="#t_firstclass">first class</a>.</p>
Chris Lattner69c11bb2005-04-25 17:34:15 +0000[diff] [blame]1065
Chris Lattnera58561b2004-08-12 19:12:28 +0000[diff] [blame]1066<h5>Syntax:</h5>
Chris Lattner69c11bb2005-04-25 17:34:15 +0000[diff] [blame]1067
1068<pre>
1069 &lt; &lt;# elements&gt; x &lt;elementtype&gt; &gt;
1070</pre>
1071
John Criswellc1f786c2005-05-13 22:25:59 +0000[diff] [blame]1072<p>The number of elements is a constant integer value; elementtype may
Chris Lattnera58561b2004-08-12 19:12:28 +0000[diff] [blame]1073be any integral or floating point type.</p>
Chris Lattner69c11bb2005-04-25 17:34:15 +0000[diff] [blame]1074
Chris Lattnera58561b2004-08-12 19:12:28 +0000[diff] [blame]1075<h5>Examples:</h5>
Chris Lattner69c11bb2005-04-25 17:34:15 +0000[diff] [blame]1076
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]1077<table class="layout">
1078 <tr class="layout">
1079 <td class="left">
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]1080 <tt>&lt;4 x i32&gt;</tt><br/>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]1081 <tt>&lt;8 x float&gt;</tt><br/>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]1082 <tt>&lt;2 x i32&gt;</tt><br/>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]1083 </td>
1084 <td class="left">
1085 Packed vector of 4 integer values.<br/>
1086 Packed vector of 8 floating-point values.<br/>
1087 Packed vector of 2 unsigned integer values.<br/>
1088 </td>
1089 </tr>
1090</table>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1091</div>
1092
Chris Lattner69c11bb2005-04-25 17:34:15 +0000[diff] [blame]1093<!-- _______________________________________________________________________ -->
1094<div class="doc_subsubsection"> <a name="t_opaque">Opaque Type</a> </div>
1095<div class="doc_text">
1096
1097<h5>Overview:</h5>
1098
1099<p>Opaque types are used to represent unknown types in the system. This
1100corresponds (for example) to the C notion of a foward declared structure type.
1101In LLVM, opaque types can eventually be resolved to any type (not just a
1102structure type).</p>
1103
1104<h5>Syntax:</h5>
1105
1106<pre>
1107 opaque
1108</pre>
1109
1110<h5>Examples:</h5>
1111
1112<table class="layout">
1113 <tr class="layout">
1114 <td class="left">
1115 <tt>opaque</tt>
1116 </td>
1117 <td class="left">
1118 An opaque type.<br/>
1119 </td>
1120 </tr>
1121</table>
1122</div>
1123
1124
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1125<!-- *********************************************************************** -->
1126<div class="doc_section"> <a name="constants">Constants</a> </div>
1127<!-- *********************************************************************** -->
1128
1129<div class="doc_text">
1130
1131<p>LLVM has several different basic types of constants. This section describes
1132them all and their syntax.</p>
1133
1134</div>
1135
1136<!-- ======================================================================= -->
Reid Spencercc16dc32004-12-09 18:02:53 +0000[diff] [blame]1137<div class="doc_subsection"><a name="simpleconstants">Simple Constants</a></div>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1138
1139<div class="doc_text">
1140
1141<dl>
1142 <dt><b>Boolean constants</b></dt>
1143
1144 <dd>The two strings '<tt>true</tt>' and '<tt>false</tt>' are both valid
1145 constants of the <tt><a href="#t_primitive">bool</a></tt> type.
1146 </dd>
1147
1148 <dt><b>Integer constants</b></dt>
1149
Reid Spencercc16dc32004-12-09 18:02:53 +0000[diff] [blame]1150 <dd>Standard integers (such as '4') are constants of the <a
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1151 href="#t_integer">integer</a> type. Negative numbers may be used with signed
1152 integer types.
1153 </dd>
1154
1155 <dt><b>Floating point constants</b></dt>
1156
1157 <dd>Floating point constants use standard decimal notation (e.g. 123.421),
1158 exponential notation (e.g. 1.23421e+2), or a more precise hexadecimal
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1159 notation (see below). Floating point constants must have a <a
1160 href="#t_floating">floating point</a> type. </dd>
1161
1162 <dt><b>Null pointer constants</b></dt>
1163
John Criswell9e2485c2004-12-10 15:51:16 +0000[diff] [blame]1164 <dd>The identifier '<tt>null</tt>' is recognized as a null pointer constant
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1165 and must be of <a href="#t_pointer">pointer type</a>.</dd>
1166
1167</dl>
1168
John Criswell9e2485c2004-12-10 15:51:16 +0000[diff] [blame]1169<p>The one non-intuitive notation for constants is the optional hexadecimal form
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1170of floating point constants. For example, the form '<tt>double
11710x432ff973cafa8000</tt>' is equivalent to (but harder to read than) '<tt>double
11724.5e+15</tt>'. The only time hexadecimal floating point constants are required
Reid Spencercc16dc32004-12-09 18:02:53 +0000[diff] [blame]1173(and the only time that they are generated by the disassembler) is when a
1174floating point constant must be emitted but it cannot be represented as a
1175decimal floating point number. For example, NaN's, infinities, and other
1176special values are represented in their IEEE hexadecimal format so that
1177assembly and disassembly do not cause any bits to change in the constants.</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1178
1179</div>
1180
1181<!-- ======================================================================= -->
1182<div class="doc_subsection"><a name="aggregateconstants">Aggregate Constants</a>
1183</div>
1184
1185<div class="doc_text">
Chris Lattnerd4f6b172005-03-07 22:13:59 +0000[diff] [blame]1186<p>Aggregate constants arise from aggregation of simple constants
1187and smaller aggregate constants.</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1188
1189<dl>
1190 <dt><b>Structure constants</b></dt>
1191
1192 <dd>Structure constants are represented with notation similar to structure
1193 type definitions (a comma separated list of elements, surrounded by braces
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]1194 (<tt>{}</tt>)). For example: "<tt>{ i32 4, float 17.0, i32* %G }</tt>",
1195 where "<tt>%G</tt>" is declared as "<tt>%G = external global i32</tt>". Structure constants
Chris Lattnerd4f6b172005-03-07 22:13:59 +0000[diff] [blame]1196 must have <a href="#t_struct">structure type</a>, and the number and
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1197 types of elements must match those specified by the type.
1198 </dd>
1199
1200 <dt><b>Array constants</b></dt>
1201
1202 <dd>Array constants are represented with notation similar to array type
1203 definitions (a comma separated list of elements, surrounded by square brackets
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]1204 (<tt>[]</tt>)). For example: "<tt>[ i32 42, i32 11, i32 74 ]</tt>". Array
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1205 constants must have <a href="#t_array">array type</a>, and the number and
1206 types of elements must match those specified by the type.
1207 </dd>
1208
1209 <dt><b>Packed constants</b></dt>
1210
1211 <dd>Packed constants are represented with notation similar to packed type
1212 definitions (a comma separated list of elements, surrounded by
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]1213 less-than/greater-than's (<tt>&lt;&gt;</tt>)). For example: "<tt>&lt; i32 42,
1214 i32 11, i32 74, i32 100 &gt;</tt>". Packed constants must have <a
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1215 href="#t_packed">packed type</a>, and the number and types of elements must
1216 match those specified by the type.
1217 </dd>
1218
1219 <dt><b>Zero initialization</b></dt>
1220
1221 <dd>The string '<tt>zeroinitializer</tt>' can be used to zero initialize a
1222 value to zero of <em>any</em> type, including scalar and aggregate types.
1223 This is often used to avoid having to print large zero initializers (e.g. for
John Criswell0ec250c2005-10-24 16:17:18 +0000[diff] [blame]1224 large arrays) and is always exactly equivalent to using explicit zero
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1225 initializers.
1226 </dd>
1227</dl>
1228
1229</div>
1230
1231<!-- ======================================================================= -->
1232<div class="doc_subsection">
1233 <a name="globalconstants">Global Variable and Function Addresses</a>
1234</div>
1235
1236<div class="doc_text">
1237
1238<p>The addresses of <a href="#globalvars">global variables</a> and <a
1239href="#functionstructure">functions</a> are always implicitly valid (link-time)
John Criswell9e2485c2004-12-10 15:51:16 +0000[diff] [blame]1240constants. These constants are explicitly referenced when the <a
1241href="#identifiers">identifier for the global</a> is used and always have <a
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1242href="#t_pointer">pointer</a> type. For example, the following is a legal LLVM
1243file:</p>
1244
1245<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]1246 %X = global i32 17
1247 %Y = global i32 42
1248 %Z = global [2 x i32*] [ i32* %X, i32* %Y ]
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1249</pre>
1250
1251</div>
1252
1253<!-- ======================================================================= -->
Reid Spencer2dc45b82004-12-09 18:13:12 +0000[diff] [blame]1254<div class="doc_subsection"><a name="undefvalues">Undefined Values</a></div>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1255<div class="doc_text">
Reid Spencer2dc45b82004-12-09 18:13:12 +0000[diff] [blame]1256 <p>The string '<tt>undef</tt>' is recognized as a type-less constant that has
John Criswellc1f786c2005-05-13 22:25:59 +0000[diff] [blame]1257 no specific value. Undefined values may be of any type and be used anywhere
Reid Spencer2dc45b82004-12-09 18:13:12 +0000[diff] [blame]1258 a constant is permitted.</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1259
Reid Spencer2dc45b82004-12-09 18:13:12 +0000[diff] [blame]1260 <p>Undefined values indicate to the compiler that the program is well defined
1261 no matter what value is used, giving the compiler more freedom to optimize.
1262 </p>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1263</div>
1264
1265<!-- ======================================================================= -->
1266<div class="doc_subsection"><a name="constantexprs">Constant Expressions</a>
1267</div>
1268
1269<div class="doc_text">
1270
1271<p>Constant expressions are used to allow expressions involving other constants
1272to be used as constants. Constant expressions may be of any <a
John Criswellc1f786c2005-05-13 22:25:59 +0000[diff] [blame]1273href="#t_firstclass">first class</a> type and may involve any LLVM operation
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1274that does not have side effects (e.g. load and call are not supported). The
1275following is the syntax for constant expressions:</p>
1276
1277<dl>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]1278 <dt><b><tt>trunc ( CST to TYPE )</tt></b></dt>
1279 <dd>Truncate a constant to another type. The bit size of CST must be larger
1280 than the bit size of TYPE. Both types must be integral.</dd>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1281
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]1282 <dt><b><tt>zext ( CST to TYPE )</tt></b></dt>
1283 <dd>Zero extend a constant to another type. The bit size of CST must be
1284 smaller or equal to the bit size of TYPE. Both types must be integral.</dd>
1285
1286 <dt><b><tt>sext ( CST to TYPE )</tt></b></dt>
1287 <dd>Sign extend a constant to another type. The bit size of CST must be
1288 smaller or equal to the bit size of TYPE. Both types must be integral.</dd>
1289
1290 <dt><b><tt>fptrunc ( CST to TYPE )</tt></b></dt>
1291 <dd>Truncate a floating point constant to another floating point type. The
1292 size of CST must be larger than the size of TYPE. Both types must be
1293 floating point.</dd>
1294
1295 <dt><b><tt>fpext ( CST to TYPE )</tt></b></dt>
1296 <dd>Floating point extend a constant to another type. The size of CST must be
1297 smaller or equal to the size of TYPE. Both types must be floating point.</dd>
1298
1299 <dt><b><tt>fp2uint ( CST to TYPE )</tt></b></dt>
1300 <dd>Convert a floating point constant to the corresponding unsigned integer
1301 constant. TYPE must be an integer type. CST must be floating point. If the
1302 value won't fit in the integer type, the results are undefined.</dd>
1303
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]1304 <dt><b><tt>fptosi ( CST to TYPE )</tt></b></dt>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]1305 <dd>Convert a floating point constant to the corresponding signed integer
1306 constant. TYPE must be an integer type. CST must be floating point. If the
1307 value won't fit in the integer type, the results are undefined.</dd>
1308
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]1309 <dt><b><tt>uitofp ( CST to TYPE )</tt></b></dt>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]1310 <dd>Convert an unsigned integer constant to the corresponding floating point
1311 constant. TYPE must be floating point. CST must be of integer type. If the
1312 value won't fit in the floating point type, the results are undefined.</dd>
1313
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]1314 <dt><b><tt>sitofp ( CST to TYPE )</tt></b></dt>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]1315 <dd>Convert a signed integer constant to the corresponding floating point
1316 constant. TYPE must be floating point. CST must be of integer type. If the
1317 value won't fit in the floating point type, the results are undefined.</dd>
1318
Reid Spencer5c0ef472006-11-11 23:08:07 +0000[diff] [blame]1319 <dt><b><tt>ptrtoint ( CST to TYPE )</tt></b></dt>
1320 <dd>Convert a pointer typed constant to the corresponding integer constant
1321 TYPE must be an integer type. CST must be of pointer type. The CST value is
1322 zero extended, truncated, or unchanged to make it fit in TYPE.</dd>
1323
1324 <dt><b><tt>inttoptr ( CST to TYPE )</tt></b></dt>
1325 <dd>Convert a integer constant to a pointer constant. TYPE must be a
1326 pointer type. CST must be of integer type. The CST value is zero extended,
1327 truncated, or unchanged to make it fit in a pointer size. This one is
1328 <i>really</i> dangerous!</dd>
1329
1330 <dt><b><tt>bitcast ( CST to TYPE )</tt></b></dt>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]1331 <dd>Convert a constant, CST, to another TYPE. The size of CST and TYPE must be
1332 identical (same number of bits). The conversion is done as if the CST value
1333 was stored to memory and read back as TYPE. In other words, no bits change
Reid Spencer5c0ef472006-11-11 23:08:07 +0000[diff] [blame]1334 with this operator, just the type. This can be used for conversion of
1335 packed types to any other type, as long as they have the same bit width. For
1336 pointers it is only valid to cast to another pointer type.
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]1337 </dd>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1338
1339 <dt><b><tt>getelementptr ( CSTPTR, IDX0, IDX1, ... )</tt></b></dt>
1340
1341 <dd>Perform the <a href="#i_getelementptr">getelementptr operation</a> on
1342 constants. As with the <a href="#i_getelementptr">getelementptr</a>
1343 instruction, the index list may have zero or more indexes, which are required
1344 to make sense for the type of "CSTPTR".</dd>
1345
Robert Bocchino9fbe1452006-01-10 19:31:34 +0000[diff] [blame]1346 <dt><b><tt>select ( COND, VAL1, VAL2 )</tt></b></dt>
1347
1348 <dd>Perform the <a href="#i_select">select operation</a> on
Reid Spencer01c42592006-12-04 19:23:19 +0000[diff] [blame]1349 constants.</dd>
1350
1351 <dt><b><tt>icmp COND ( VAL1, VAL2 )</tt></b></dt>
1352 <dd>Performs the <a href="#i_icmp">icmp operation</a> on constants.</dd>
1353
1354 <dt><b><tt>fcmp COND ( VAL1, VAL2 )</tt></b></dt>
1355 <dd>Performs the <a href="#i_fcmp">fcmp operation</a> on constants.</dd>
Robert Bocchino9fbe1452006-01-10 19:31:34 +0000[diff] [blame]1356
1357 <dt><b><tt>extractelement ( VAL, IDX )</tt></b></dt>
1358
1359 <dd>Perform the <a href="#i_extractelement">extractelement
1360 operation</a> on constants.
1361
Robert Bocchino05ccd702006-01-15 20:48:27 +0000[diff] [blame]1362 <dt><b><tt>insertelement ( VAL, ELT, IDX )</tt></b></dt>
1363
1364 <dd>Perform the <a href="#i_insertelement">insertelement
Reid Spencer01c42592006-12-04 19:23:19 +0000[diff] [blame]1365 operation</a> on constants.</dd>
Robert Bocchino05ccd702006-01-15 20:48:27 +0000[diff] [blame]1366
Chris Lattnerc1989542006-04-08 00:13:41 +0000[diff] [blame]1367
1368 <dt><b><tt>shufflevector ( VEC1, VEC2, IDXMASK )</tt></b></dt>
1369
1370 <dd>Perform the <a href="#i_shufflevector">shufflevector
Reid Spencer01c42592006-12-04 19:23:19 +0000[diff] [blame]1371 operation</a> on constants.</dd>
Chris Lattnerc1989542006-04-08 00:13:41 +0000[diff] [blame]1372
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1373 <dt><b><tt>OPCODE ( LHS, RHS )</tt></b></dt>
1374
Reid Spencer2dc45b82004-12-09 18:13:12 +0000[diff] [blame]1375 <dd>Perform the specified operation of the LHS and RHS constants. OPCODE may
1376 be any of the <a href="#binaryops">binary</a> or <a href="#bitwiseops">bitwise
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1377 binary</a> operations. The constraints on operands are the same as those for
1378 the corresponding instruction (e.g. no bitwise operations on floating point
John Criswelle4c57cc2005-05-12 16:52:32 +0000[diff] [blame]1379 values are allowed).</dd>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1380</dl>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1381</div>
Chris Lattner9ee5d222004-03-08 16:49:10 +0000[diff] [blame]1382
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1383<!-- *********************************************************************** -->
Chris Lattnere87d6532006-01-25 23:47:57 +0000[diff] [blame]1384<div class="doc_section"> <a name="othervalues">Other Values</a> </div>
1385<!-- *********************************************************************** -->
1386
1387<!-- ======================================================================= -->
1388<div class="doc_subsection">
1389<a name="inlineasm">Inline Assembler Expressions</a>
1390</div>
1391
1392<div class="doc_text">
1393
1394<p>
1395LLVM supports inline assembler expressions (as opposed to <a href="#moduleasm">
1396Module-Level Inline Assembly</a>) through the use of a special value. This
1397value represents the inline assembler as a string (containing the instructions
1398to emit), a list of operand constraints (stored as a string), and a flag that
1399indicates whether or not the inline asm expression has side effects. An example
1400inline assembler expression is:
1401</p>
1402
1403<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]1404 i32 (i32) asm "bswap $0", "=r,r"
Chris Lattnere87d6532006-01-25 23:47:57 +0000[diff] [blame]1405</pre>
1406
1407<p>
1408Inline assembler expressions may <b>only</b> be used as the callee operand of
1409a <a href="#i_call"><tt>call</tt> instruction</a>. Thus, typically we have:
1410</p>
1411
1412<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]1413 %X = call i32 asm "<a href="#i_bswap">bswap</a> $0", "=r,r"(i32 %Y)
Chris Lattnere87d6532006-01-25 23:47:57 +0000[diff] [blame]1414</pre>
1415
1416<p>
1417Inline asms with side effects not visible in the constraint list must be marked
1418as having side effects. This is done through the use of the
1419'<tt>sideeffect</tt>' keyword, like so:
1420</p>
1421
1422<pre>
1423 call void asm sideeffect "eieio", ""()
1424</pre>
1425
1426<p>TODO: The format of the asm and constraints string still need to be
1427documented here. Constraints on what can be done (e.g. duplication, moving, etc
1428need to be documented).
1429</p>
1430
1431</div>
1432
1433<!-- *********************************************************************** -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1434<div class="doc_section"> <a name="instref">Instruction Reference</a> </div>
1435<!-- *********************************************************************** -->
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1436
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1437<div class="doc_text">
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1438
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1439<p>The LLVM instruction set consists of several different
1440classifications of instructions: <a href="#terminators">terminator
John Criswellc1f786c2005-05-13 22:25:59 +0000[diff] [blame]1441instructions</a>, <a href="#binaryops">binary instructions</a>,
1442<a href="#bitwiseops">bitwise binary instructions</a>, <a
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1443 href="#memoryops">memory instructions</a>, and <a href="#otherops">other
1444instructions</a>.</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1445
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1446</div>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1447
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1448<!-- ======================================================================= -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1449<div class="doc_subsection"> <a name="terminators">Terminator
1450Instructions</a> </div>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1451
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1452<div class="doc_text">
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1453
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1454<p>As mentioned <a href="#functionstructure">previously</a>, every
1455basic block in a program ends with a "Terminator" instruction, which
1456indicates which block should be executed after the current block is
1457finished. These terminator instructions typically yield a '<tt>void</tt>'
1458value: they produce control flow, not values (the one exception being
1459the '<a href="#i_invoke"><tt>invoke</tt></a>' instruction).</p>
John Criswell9e2485c2004-12-10 15:51:16 +0000[diff] [blame]1460<p>There are six different terminator instructions: the '<a
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1461 href="#i_ret"><tt>ret</tt></a>' instruction, the '<a href="#i_br"><tt>br</tt></a>'
1462instruction, the '<a href="#i_switch"><tt>switch</tt></a>' instruction,
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]1463the '<a href="#i_invoke"><tt>invoke</tt></a>' instruction, the '<a
1464 href="#i_unwind"><tt>unwind</tt></a>' instruction, and the '<a
1465 href="#i_unreachable"><tt>unreachable</tt></a>' instruction.</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1466
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1467</div>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1468
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1469<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1470<div class="doc_subsubsection"> <a name="i_ret">'<tt>ret</tt>'
1471Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1472<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1473<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1474<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 +0000[diff] [blame]1475 ret void <i>; Return from void function</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1476</pre>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1477<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1478<p>The '<tt>ret</tt>' instruction is used to return control flow (and a
John Criswellc1f786c2005-05-13 22:25:59 +0000[diff] [blame]1479value) from a function back to the caller.</p>
John Criswell4457dc92004-04-09 16:48:45 +0000[diff] [blame]1480<p>There are two forms of the '<tt>ret</tt>' instruction: one that
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1481returns a value and then causes control flow, and one that just causes
1482control flow to occur.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1483<h5>Arguments:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1484<p>The '<tt>ret</tt>' instruction may return any '<a
1485 href="#t_firstclass">first class</a>' type. Notice that a function is
1486not <a href="#wellformed">well formed</a> if there exists a '<tt>ret</tt>'
1487instruction inside of the function that returns a value that does not
1488match the return type of the function.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1489<h5>Semantics:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1490<p>When the '<tt>ret</tt>' instruction is executed, control flow
1491returns back to the calling function's context. If the caller is a "<a
John Criswellfa081872004-06-25 15:16:57 +0000[diff] [blame]1492 href="#i_call"><tt>call</tt></a>" instruction, execution continues at
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1493the instruction after the call. If the caller was an "<a
1494 href="#i_invoke"><tt>invoke</tt></a>" instruction, execution continues
John Criswelle4c57cc2005-05-12 16:52:32 +0000[diff] [blame]1495at the beginning of the "normal" destination block. If the instruction
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1496returns a value, that value shall set the call or invoke instruction's
1497return value.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1498<h5>Example:</h5>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]1499<pre> ret i32 5 <i>; Return an integer value of 5</i>
Chris Lattner7faa8832002-04-14 06:13:44 +0000[diff] [blame]1500 ret void <i>; Return from a void function</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1501</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1502</div>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1503<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1504<div class="doc_subsubsection"> <a name="i_br">'<tt>br</tt>' Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1505<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1506<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1507<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 +0000[diff] [blame]1508</pre>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1509<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1510<p>The '<tt>br</tt>' instruction is used to cause control flow to
1511transfer to a different basic block in the current function. There are
1512two forms of this instruction, corresponding to a conditional branch
1513and an unconditional branch.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1514<h5>Arguments:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1515<p>The conditional branch form of the '<tt>br</tt>' instruction takes a
1516single '<tt>bool</tt>' value and two '<tt>label</tt>' values. The
1517unconditional form of the '<tt>br</tt>' instruction takes a single '<tt>label</tt>'
1518value as a target.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1519<h5>Semantics:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1520<p>Upon execution of a conditional '<tt>br</tt>' instruction, the '<tt>bool</tt>'
1521argument is evaluated. If the value is <tt>true</tt>, control flows
1522to the '<tt>iftrue</tt>' <tt>label</tt> argument. If "cond" is <tt>false</tt>,
1523control flows to the '<tt>iffalse</tt>' <tt>label</tt> argument.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1524<h5>Example:</h5>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]1525<pre>Test:<br> %cond = <a href="#i_icmp">icmp</a> eq, i32 %a, %b<br> br bool %cond, label %IfEqual, label %IfUnequal<br>IfEqual:<br> <a
1526 href="#i_ret">ret</a> i32 1<br>IfUnequal:<br> <a href="#i_ret">ret</a> i32 0<br></pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1527</div>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1528<!-- _______________________________________________________________________ -->
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000[diff] [blame]1529<div class="doc_subsubsection">
1530 <a name="i_switch">'<tt>switch</tt>' Instruction</a>
1531</div>
1532
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1533<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1534<h5>Syntax:</h5>
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000[diff] [blame]1535
1536<pre>
1537 switch &lt;intty&gt; &lt;value&gt;, label &lt;defaultdest&gt; [ &lt;intty&gt; &lt;val&gt;, label &lt;dest&gt; ... ]
1538</pre>
1539
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1540<h5>Overview:</h5>
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000[diff] [blame]1541
1542<p>The '<tt>switch</tt>' instruction is used to transfer control flow to one of
1543several different places. It is a generalization of the '<tt>br</tt>'
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1544instruction, allowing a branch to occur to one of many possible
1545destinations.</p>
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000[diff] [blame]1546
1547
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1548<h5>Arguments:</h5>
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000[diff] [blame]1549
1550<p>The '<tt>switch</tt>' instruction uses three parameters: an integer
1551comparison value '<tt>value</tt>', a default '<tt>label</tt>' destination, and
1552an array of pairs of comparison value constants and '<tt>label</tt>'s. The
1553table is not allowed to contain duplicate constant entries.</p>
1554
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1555<h5>Semantics:</h5>
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000[diff] [blame]1556
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1557<p>The <tt>switch</tt> instruction specifies a table of values and
1558destinations. When the '<tt>switch</tt>' instruction is executed, this
John Criswell84114752004-06-25 16:05:06 +0000[diff] [blame]1559table is searched for the given value. If the value is found, control flow is
1560transfered to the corresponding destination; otherwise, control flow is
1561transfered to the default destination.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1562
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000[diff] [blame]1563<h5>Implementation:</h5>
1564
1565<p>Depending on properties of the target machine and the particular
1566<tt>switch</tt> instruction, this instruction may be code generated in different
John Criswell84114752004-06-25 16:05:06 +0000[diff] [blame]1567ways. For example, it could be generated as a series of chained conditional
1568branches or with a lookup table.</p>
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000[diff] [blame]1569
1570<h5>Example:</h5>
1571
1572<pre>
1573 <i>; Emulate a conditional br instruction</i>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]1574 %Val = <a href="#i_zext">zext</a> bool %value to i32
1575 switch i32 %Val, label %truedest [i32 0, label %falsedest ]
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000[diff] [blame]1576
1577 <i>; Emulate an unconditional br instruction</i>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]1578 switch i32 0, label %dest [ ]
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000[diff] [blame]1579
1580 <i>; Implement a jump table:</i>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]1581 switch i32 %val, label %otherwise [ i32 0, label %onzero
1582 i32 1, label %onone
1583 i32 2, label %ontwo ]
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1584</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1585</div>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]1586
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1587<!-- _______________________________________________________________________ -->
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]1588<div class="doc_subsubsection">
1589 <a name="i_invoke">'<tt>invoke</tt>' Instruction</a>
1590</div>
1591
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1592<div class="doc_text">
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]1593
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1594<h5>Syntax:</h5>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]1595
1596<pre>
1597 &lt;result&gt; = invoke [<a href="#callingconv">cconv</a>] &lt;ptr to function ty&gt; %&lt;function ptr val&gt;(&lt;function args&gt;)
Chris Lattner76b8a332006-05-14 18:23:06 +0000[diff] [blame]1598 to label &lt;normal label&gt; unwind label &lt;exception label&gt;
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]1599</pre>
1600
Chris Lattner6536cfe2002-05-06 22:08:29 +0000[diff] [blame]1601<h5>Overview:</h5>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]1602
1603<p>The '<tt>invoke</tt>' instruction causes control to transfer to a specified
1604function, with the possibility of control flow transfer to either the
John Criswelle4c57cc2005-05-12 16:52:32 +0000[diff] [blame]1605'<tt>normal</tt>' label or the
1606'<tt>exception</tt>' label. If the callee function returns with the
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]1607"<tt><a href="#i_ret">ret</a></tt>" instruction, control flow will return to the
1608"normal" label. If the callee (or any indirect callees) returns with the "<a
John Criswelle4c57cc2005-05-12 16:52:32 +0000[diff] [blame]1609href="#i_unwind"><tt>unwind</tt></a>" instruction, control is interrupted and
1610continued at the dynamically nearest "exception" label.</p>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]1611
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1612<h5>Arguments:</h5>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]1613
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1614<p>This instruction requires several arguments:</p>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]1615
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1616<ol>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]1617 <li>
John Criswellc1f786c2005-05-13 22:25:59 +0000[diff] [blame]1618 The optional "cconv" marker indicates which <a href="callingconv">calling
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]1619 convention</a> the call should use. If none is specified, the call defaults
1620 to using C calling conventions.
1621 </li>
1622 <li>'<tt>ptr to function ty</tt>': shall be the signature of the pointer to
1623 function value being invoked. In most cases, this is a direct function
1624 invocation, but indirect <tt>invoke</tt>s are just as possible, branching off
1625 an arbitrary pointer to function value.
1626 </li>
1627
1628 <li>'<tt>function ptr val</tt>': An LLVM value containing a pointer to a
1629 function to be invoked. </li>
1630
1631 <li>'<tt>function args</tt>': argument list whose types match the function
1632 signature argument types. If the function signature indicates the function
1633 accepts a variable number of arguments, the extra arguments can be
1634 specified. </li>
1635
1636 <li>'<tt>normal label</tt>': the label reached when the called function
1637 executes a '<tt><a href="#i_ret">ret</a></tt>' instruction. </li>
1638
1639 <li>'<tt>exception label</tt>': the label reached when a callee returns with
1640 the <a href="#i_unwind"><tt>unwind</tt></a> instruction. </li>
1641
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1642</ol>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]1643
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1644<h5>Semantics:</h5>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]1645
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1646<p>This instruction is designed to operate as a standard '<tt><a
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]1647href="#i_call">call</a></tt>' instruction in most regards. The primary
1648difference is that it establishes an association with a label, which is used by
1649the runtime library to unwind the stack.</p>
1650
1651<p>This instruction is used in languages with destructors to ensure that proper
1652cleanup is performed in the case of either a <tt>longjmp</tt> or a thrown
1653exception. Additionally, this is important for implementation of
1654'<tt>catch</tt>' clauses in high-level languages that support them.</p>
1655
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1656<h5>Example:</h5>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]1657<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]1658 %retval = invoke i32 %Test(i32 15) to label %Continue
1659 unwind label %TestCleanup <i>; {i32}:retval set</i>
1660 %retval = invoke <a href="#callingconv">coldcc</a> i32 %Test(i32 15) to label %Continue
1661 unwind label %TestCleanup <i>; {i32}:retval set</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1662</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1663</div>
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]1664
1665
Chris Lattner27f71f22003-09-03 00:41:47 +0000[diff] [blame]1666<!-- _______________________________________________________________________ -->
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]1667
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1668<div class="doc_subsubsection"> <a name="i_unwind">'<tt>unwind</tt>'
1669Instruction</a> </div>
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]1670
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1671<div class="doc_text">
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]1672
Chris Lattner27f71f22003-09-03 00:41:47 +0000[diff] [blame]1673<h5>Syntax:</h5>
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]1674<pre>
1675 unwind
1676</pre>
1677
Chris Lattner27f71f22003-09-03 00:41:47 +0000[diff] [blame]1678<h5>Overview:</h5>
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]1679
1680<p>The '<tt>unwind</tt>' instruction unwinds the stack, continuing control flow
1681at the first callee in the dynamic call stack which used an <a
1682href="#i_invoke"><tt>invoke</tt></a> instruction to perform the call. This is
1683primarily used to implement exception handling.</p>
1684
Chris Lattner27f71f22003-09-03 00:41:47 +0000[diff] [blame]1685<h5>Semantics:</h5>
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]1686
1687<p>The '<tt>unwind</tt>' intrinsic causes execution of the current function to
1688immediately halt. The dynamic call stack is then searched for the first <a
1689href="#i_invoke"><tt>invoke</tt></a> instruction on the call stack. Once found,
1690execution continues at the "exceptional" destination block specified by the
1691<tt>invoke</tt> instruction. If there is no <tt>invoke</tt> instruction in the
1692dynamic call chain, undefined behavior results.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1693</div>
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]1694
1695<!-- _______________________________________________________________________ -->
1696
1697<div class="doc_subsubsection"> <a name="i_unreachable">'<tt>unreachable</tt>'
1698Instruction</a> </div>
1699
1700<div class="doc_text">
1701
1702<h5>Syntax:</h5>
1703<pre>
1704 unreachable
1705</pre>
1706
1707<h5>Overview:</h5>
1708
1709<p>The '<tt>unreachable</tt>' instruction has no defined semantics. This
1710instruction is used to inform the optimizer that a particular portion of the
1711code is not reachable. This can be used to indicate that the code after a
1712no-return function cannot be reached, and other facts.</p>
1713
1714<h5>Semantics:</h5>
1715
1716<p>The '<tt>unreachable</tt>' instruction has no defined semantics.</p>
1717</div>
1718
1719
1720
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1721<!-- ======================================================================= -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1722<div class="doc_subsection"> <a name="binaryops">Binary Operations</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1723<div class="doc_text">
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1724<p>Binary operators are used to do most of the computation in a
1725program. They require two operands, execute an operation on them, and
John Criswell9e2485c2004-12-10 15:51:16 +0000[diff] [blame]1726produce a single value. The operands might represent
Chris Lattnera58561b2004-08-12 19:12:28 +0000[diff] [blame]1727multiple data, as is the case with the <a href="#t_packed">packed</a> data type.
1728The result value of a binary operator is not
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1729necessarily the same type as its operands.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1730<p>There are several different binary operators:</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1731</div>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1732<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1733<div class="doc_subsubsection"> <a name="i_add">'<tt>add</tt>'
1734Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1735<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1736<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1737<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 +0000[diff] [blame]1738</pre>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1739<h5>Overview:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1740<p>The '<tt>add</tt>' instruction returns the sum of its two operands.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1741<h5>Arguments:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1742<p>The two arguments to the '<tt>add</tt>' instruction must be either <a
Chris Lattnera58561b2004-08-12 19:12:28 +0000[diff] [blame]1743 href="#t_integer">integer</a> or <a href="#t_floating">floating point</a> values.
1744 This instruction can also take <a href="#t_packed">packed</a> versions of the values.
1745Both arguments must have identical types.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1746<h5>Semantics:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1747<p>The value produced is the integer or floating point sum of the two
1748operands.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1749<h5>Example:</h5>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]1750<pre> &lt;result&gt; = add i32 4, %var <i>; yields {i32}:result = 4 + %var</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1751</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1752</div>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1753<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1754<div class="doc_subsubsection"> <a name="i_sub">'<tt>sub</tt>'
1755Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1756<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1757<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1758<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 +0000[diff] [blame]1759</pre>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1760<h5>Overview:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1761<p>The '<tt>sub</tt>' instruction returns the difference of its two
1762operands.</p>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1763<p>Note that the '<tt>sub</tt>' instruction is used to represent the '<tt>neg</tt>'
1764instruction present in most other intermediate representations.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1765<h5>Arguments:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1766<p>The two arguments to the '<tt>sub</tt>' instruction must be either <a
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1767 href="#t_integer">integer</a> or <a href="#t_floating">floating point</a>
Chris Lattnera58561b2004-08-12 19:12:28 +0000[diff] [blame]1768values.
1769This instruction can also take <a href="#t_packed">packed</a> versions of the values.
1770Both arguments must have identical types.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1771<h5>Semantics:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1772<p>The value produced is the integer or floating point difference of
1773the two operands.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1774<h5>Example:</h5>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]1775<pre> &lt;result&gt; = sub i32 4, %var <i>; yields {i32}:result = 4 - %var</i>
1776 &lt;result&gt; = sub i32 0, %val <i>; yields {i32}:result = -%var</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1777</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1778</div>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1779<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1780<div class="doc_subsubsection"> <a name="i_mul">'<tt>mul</tt>'
1781Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1782<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1783<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1784<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 +0000[diff] [blame]1785</pre>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1786<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1787<p>The '<tt>mul</tt>' instruction returns the product of its two
1788operands.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1789<h5>Arguments:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1790<p>The two arguments to the '<tt>mul</tt>' instruction must be either <a
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1791 href="#t_integer">integer</a> or <a href="#t_floating">floating point</a>
Chris Lattnera58561b2004-08-12 19:12:28 +0000[diff] [blame]1792values.
1793This instruction can also take <a href="#t_packed">packed</a> versions of the values.
1794Both arguments must have identical types.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1795<h5>Semantics:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1796<p>The value produced is the integer or floating point product of the
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1797two operands.</p>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1798<p>There is no signed vs unsigned multiplication. The appropriate
1799action is taken based on the type of the operand.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1800<h5>Example:</h5>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]1801<pre> &lt;result&gt; = mul i32 4, %var <i>; yields {i32}:result = 4 * %var</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1802</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1803</div>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1804<!-- _______________________________________________________________________ -->
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]1805<div class="doc_subsubsection"> <a name="i_udiv">'<tt>udiv</tt>' Instruction
1806</a></div>
1807<div class="doc_text">
1808<h5>Syntax:</h5>
1809<pre> &lt;result&gt; = udiv &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
1810</pre>
1811<h5>Overview:</h5>
1812<p>The '<tt>udiv</tt>' instruction returns the quotient of its two
1813operands.</p>
1814<h5>Arguments:</h5>
1815<p>The two arguments to the '<tt>udiv</tt>' instruction must be
1816<a href="#t_integer">integer</a> values. Both arguments must have identical
1817types. This instruction can also take <a href="#t_packed">packed</a> versions
1818of the values in which case the elements must be integers.</p>
1819<h5>Semantics:</h5>
1820<p>The value produced is the unsigned integer quotient of the two operands. This
1821instruction always performs an unsigned division operation, regardless of
1822whether the arguments are unsigned or not.</p>
1823<h5>Example:</h5>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]1824<pre> &lt;result&gt; = udiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]1825</pre>
1826</div>
1827<!-- _______________________________________________________________________ -->
1828<div class="doc_subsubsection"> <a name="i_sdiv">'<tt>sdiv</tt>' Instruction
1829</a> </div>
1830<div class="doc_text">
1831<h5>Syntax:</h5>
1832<pre> &lt;result&gt; = sdiv &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
1833</pre>
1834<h5>Overview:</h5>
1835<p>The '<tt>sdiv</tt>' instruction returns the quotient of its two
1836operands.</p>
1837<h5>Arguments:</h5>
1838<p>The two arguments to the '<tt>sdiv</tt>' instruction must be
1839<a href="#t_integer">integer</a> values. Both arguments must have identical
1840types. This instruction can also take <a href="#t_packed">packed</a> versions
1841of the values in which case the elements must be integers.</p>
1842<h5>Semantics:</h5>
1843<p>The value produced is the signed integer quotient of the two operands. This
1844instruction always performs a signed division operation, regardless of whether
1845the arguments are signed or not.</p>
1846<h5>Example:</h5>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]1847<pre> &lt;result&gt; = sdiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]1848</pre>
1849</div>
1850<!-- _______________________________________________________________________ -->
1851<div class="doc_subsubsection"> <a name="i_fdiv">'<tt>fdiv</tt>'
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1852Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1853<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1854<h5>Syntax:</h5>
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]1855<pre> &lt;result&gt; = fdiv &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1856</pre>
1857<h5>Overview:</h5>
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]1858<p>The '<tt>fdiv</tt>' instruction returns the quotient of its two
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1859operands.</p>
1860<h5>Arguments:</h5>
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]1861<p>The two arguments to the '<tt>div</tt>' instruction must be
1862<a href="#t_floating">floating point</a> values. Both arguments must have
1863identical types. This instruction can also take <a href="#t_packed">packed</a>
1864versions of the values in which case the elements must be floating point.</p>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1865<h5>Semantics:</h5>
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]1866<p>The value produced is the floating point quotient of the two operands.</p>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1867<h5>Example:</h5>
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]1868<pre> &lt;result&gt; = fdiv float 4.0, %var <i>; yields {float}:result = 4.0 / %var</i>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1869</pre>
1870</div>
1871<!-- _______________________________________________________________________ -->
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]1872<div class="doc_subsubsection"> <a name="i_urem">'<tt>urem</tt>' Instruction</a>
1873</div>
1874<div class="doc_text">
1875<h5>Syntax:</h5>
1876<pre> &lt;result&gt; = urem &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
1877</pre>
1878<h5>Overview:</h5>
1879<p>The '<tt>urem</tt>' instruction returns the remainder from the
1880unsigned division of its two arguments.</p>
1881<h5>Arguments:</h5>
1882<p>The two arguments to the '<tt>urem</tt>' instruction must be
1883<a href="#t_integer">integer</a> values. Both arguments must have identical
1884types.</p>
1885<h5>Semantics:</h5>
1886<p>This instruction returns the unsigned integer <i>remainder</i> of a division.
1887This instruction always performs an unsigned division to get the remainder,
1888regardless of whether the arguments are unsigned or not.</p>
1889<h5>Example:</h5>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]1890<pre> &lt;result&gt; = urem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]1891</pre>
1892
1893</div>
1894<!-- _______________________________________________________________________ -->
1895<div class="doc_subsubsection"> <a name="i_srem">'<tt>srem</tt>'
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1896Instruction</a> </div>
1897<div class="doc_text">
1898<h5>Syntax:</h5>
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]1899<pre> &lt;result&gt; = srem &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1900</pre>
1901<h5>Overview:</h5>
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]1902<p>The '<tt>srem</tt>' instruction returns the remainder from the
1903signed division of its two operands.</p>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1904<h5>Arguments:</h5>
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]1905<p>The two arguments to the '<tt>srem</tt>' instruction must be
1906<a href="#t_integer">integer</a> values. Both arguments must have identical
1907types.</p>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1908<h5>Semantics:</h5>
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]1909<p>This instruction returns the <i>remainder</i> of a division (where the result
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1910has the same sign as the divisor), not the <i>modulus</i> (where the
1911result has the same sign as the dividend) of a value. For more
John Criswell0ec250c2005-10-24 16:17:18 +0000[diff] [blame]1912information about the difference, see <a
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1913 href="http://mathforum.org/dr.math/problems/anne.4.28.99.html">The
1914Math Forum</a>.</p>
1915<h5>Example:</h5>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]1916<pre> &lt;result&gt; = srem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]1917</pre>
1918
1919</div>
1920<!-- _______________________________________________________________________ -->
1921<div class="doc_subsubsection"> <a name="i_frem">'<tt>frem</tt>'
1922Instruction</a> </div>
1923<div class="doc_text">
1924<h5>Syntax:</h5>
1925<pre> &lt;result&gt; = frem &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
1926</pre>
1927<h5>Overview:</h5>
1928<p>The '<tt>frem</tt>' instruction returns the remainder from the
1929division of its two operands.</p>
1930<h5>Arguments:</h5>
1931<p>The two arguments to the '<tt>frem</tt>' instruction must be
1932<a href="#t_floating">floating point</a> values. Both arguments must have
1933identical types.</p>
1934<h5>Semantics:</h5>
1935<p>This instruction returns the <i>remainder</i> of a division.</p>
1936<h5>Example:</h5>
1937<pre> &lt;result&gt; = frem float 4.0, %var <i>; yields {float}:result = 4.0 % %var</i>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1938</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1939</div>
Robert Bocchino7b81c752006-02-17 21:18:08 +0000[diff] [blame]1940
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1941<!-- ======================================================================= -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1942<div class="doc_subsection"> <a name="bitwiseops">Bitwise Binary
1943Operations</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1944<div class="doc_text">
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1945<p>Bitwise binary operators are used to do various forms of
1946bit-twiddling in a program. They are generally very efficient
John Criswell9e2485c2004-12-10 15:51:16 +0000[diff] [blame]1947instructions and can commonly be strength reduced from other
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1948instructions. They require two operands, execute an operation on them,
1949and produce a single value. The resulting value of the bitwise binary
1950operators is always the same type as its first operand.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1951</div>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1952<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1953<div class="doc_subsubsection"> <a name="i_and">'<tt>and</tt>'
1954Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1955<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1956<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1957<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 +0000[diff] [blame]1958</pre>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1959<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1960<p>The '<tt>and</tt>' instruction returns the bitwise logical and of
1961its two operands.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1962<h5>Arguments:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1963<p>The two arguments to the '<tt>and</tt>' instruction must be <a
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1964 href="#t_integral">integral</a> values. Both arguments must have
1965identical types.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1966<h5>Semantics:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1967<p>The truth table used for the '<tt>and</tt>' instruction is:</p>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1968<p> </p>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000[diff] [blame]1969<div style="align: center">
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1970<table border="1" cellspacing="0" cellpadding="4">
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1971 <tbody>
1972 <tr>
1973 <td>In0</td>
1974 <td>In1</td>
1975 <td>Out</td>
1976 </tr>
1977 <tr>
1978 <td>0</td>
1979 <td>0</td>
1980 <td>0</td>
1981 </tr>
1982 <tr>
1983 <td>0</td>
1984 <td>1</td>
1985 <td>0</td>
1986 </tr>
1987 <tr>
1988 <td>1</td>
1989 <td>0</td>
1990 <td>0</td>
1991 </tr>
1992 <tr>
1993 <td>1</td>
1994 <td>1</td>
1995 <td>1</td>
1996 </tr>
1997 </tbody>
1998</table>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000[diff] [blame]1999</div>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2000<h5>Example:</h5>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2001<pre> &lt;result&gt; = and i32 4, %var <i>; yields {i32}:result = 4 &amp; %var</i>
2002 &lt;result&gt; = and i32 15, 40 <i>; yields {i32}:result = 8</i>
2003 &lt;result&gt; = and i32 4, 8 <i>; yields {i32}:result = 0</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2004</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2005</div>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2006<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2007<div class="doc_subsubsection"> <a name="i_or">'<tt>or</tt>' Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2008<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2009<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2010<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 +0000[diff] [blame]2011</pre>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2012<h5>Overview:</h5>
2013<p>The '<tt>or</tt>' instruction returns the bitwise logical inclusive
2014or of its two operands.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2015<h5>Arguments:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2016<p>The two arguments to the '<tt>or</tt>' instruction must be <a
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2017 href="#t_integral">integral</a> values. Both arguments must have
2018identical types.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2019<h5>Semantics:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2020<p>The truth table used for the '<tt>or</tt>' instruction is:</p>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2021<p> </p>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000[diff] [blame]2022<div style="align: center">
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2023<table border="1" cellspacing="0" cellpadding="4">
2024 <tbody>
2025 <tr>
2026 <td>In0</td>
2027 <td>In1</td>
2028 <td>Out</td>
2029 </tr>
2030 <tr>
2031 <td>0</td>
2032 <td>0</td>
2033 <td>0</td>
2034 </tr>
2035 <tr>
2036 <td>0</td>
2037 <td>1</td>
2038 <td>1</td>
2039 </tr>
2040 <tr>
2041 <td>1</td>
2042 <td>0</td>
2043 <td>1</td>
2044 </tr>
2045 <tr>
2046 <td>1</td>
2047 <td>1</td>
2048 <td>1</td>
2049 </tr>
2050 </tbody>
2051</table>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000[diff] [blame]2052</div>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2053<h5>Example:</h5>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2054<pre> &lt;result&gt; = or i32 4, %var <i>; yields {i32}:result = 4 | %var</i>
2055 &lt;result&gt; = or i32 15, 40 <i>; yields {i32}:result = 47</i>
2056 &lt;result&gt; = or i32 4, 8 <i>; yields {i32}:result = 12</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2057</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2058</div>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2059<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2060<div class="doc_subsubsection"> <a name="i_xor">'<tt>xor</tt>'
2061Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2062<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2063<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2064<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 +0000[diff] [blame]2065</pre>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2066<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2067<p>The '<tt>xor</tt>' instruction returns the bitwise logical exclusive
2068or of its two operands. The <tt>xor</tt> is used to implement the
2069"one's complement" operation, which is the "~" operator in C.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2070<h5>Arguments:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2071<p>The two arguments to the '<tt>xor</tt>' instruction must be <a
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2072 href="#t_integral">integral</a> values. Both arguments must have
2073identical types.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2074<h5>Semantics:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2075<p>The truth table used for the '<tt>xor</tt>' instruction is:</p>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2076<p> </p>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000[diff] [blame]2077<div style="align: center">
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2078<table border="1" cellspacing="0" cellpadding="4">
2079 <tbody>
2080 <tr>
2081 <td>In0</td>
2082 <td>In1</td>
2083 <td>Out</td>
2084 </tr>
2085 <tr>
2086 <td>0</td>
2087 <td>0</td>
2088 <td>0</td>
2089 </tr>
2090 <tr>
2091 <td>0</td>
2092 <td>1</td>
2093 <td>1</td>
2094 </tr>
2095 <tr>
2096 <td>1</td>
2097 <td>0</td>
2098 <td>1</td>
2099 </tr>
2100 <tr>
2101 <td>1</td>
2102 <td>1</td>
2103 <td>0</td>
2104 </tr>
2105 </tbody>
2106</table>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000[diff] [blame]2107</div>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2108<p> </p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2109<h5>Example:</h5>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2110<pre> &lt;result&gt; = xor i32 4, %var <i>; yields {i32}:result = 4 ^ %var</i>
2111 &lt;result&gt; = xor i32 15, 40 <i>; yields {i32}:result = 39</i>
2112 &lt;result&gt; = xor i32 4, 8 <i>; yields {i32}:result = 12</i>
2113 &lt;result&gt; = xor i32 %V, -1 <i>; yields {i32}:result = ~%V</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2114</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2115</div>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2116<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2117<div class="doc_subsubsection"> <a name="i_shl">'<tt>shl</tt>'
2118Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2119<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2120<h5>Syntax:</h5>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2121<pre> &lt;result&gt; = shl &lt;ty&gt; &lt;var1&gt;, i8 &lt;var2&gt; <i>; yields {ty}:result</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2122</pre>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2123<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2124<p>The '<tt>shl</tt>' instruction returns the first operand shifted to
2125the left a specified number of bits.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2126<h5>Arguments:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2127<p>The first argument to the '<tt>shl</tt>' instruction must be an <a
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2128 href="#t_integer">integer</a> type. The second argument must be an '<tt>i8</tt>'
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2129type.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2130<h5>Semantics:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2131<p>The value produced is <tt>var1</tt> * 2<sup><tt>var2</tt></sup>.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2132<h5>Example:</h5>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2133<pre> &lt;result&gt; = shl i32 4, i8 %var <i>; yields {i32}:result = 4 &lt;&lt; %var</i>
2134 &lt;result&gt; = shl i32 4, i8 2 <i>; yields {i32}:result = 16</i>
2135 &lt;result&gt; = shl i32 1, i8 10 <i>; yields {i32}:result = 1024</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2136</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2137</div>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2138<!-- _______________________________________________________________________ -->
Reid Spencer3822ff52006-11-08 06:47:33 +0000[diff] [blame]2139<div class="doc_subsubsection"> <a name="i_lshr">'<tt>lshr</tt>'
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2140Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2141<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2142<h5>Syntax:</h5>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2143<pre> &lt;result&gt; = lshr &lt;ty&gt; &lt;var1&gt;, i8 &lt;var2&gt; <i>; yields {ty}:result</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2144</pre>
Reid Spencer3822ff52006-11-08 06:47:33 +0000[diff] [blame]2145
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2146<h5>Overview:</h5>
Reid Spencer3822ff52006-11-08 06:47:33 +0000[diff] [blame]2147<p>The '<tt>lshr</tt>' instruction (logical shift right) returns the first
2148operand shifted to the right a specified number of bits.</p>
2149
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2150<h5>Arguments:</h5>
Reid Spencer3822ff52006-11-08 06:47:33 +0000[diff] [blame]2151<p>The first argument to the '<tt>lshr</tt>' instruction must be an <a
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2152 href="#t_integer">integer</a> type. The second argument must be an '<tt>i8</tt>' type.</p>
Reid Spencer3822ff52006-11-08 06:47:33 +0000[diff] [blame]2153
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2154<h5>Semantics:</h5>
Reid Spencer3822ff52006-11-08 06:47:33 +0000[diff] [blame]2155<p>This instruction always performs a logical shift right operation, regardless
2156of whether the arguments are unsigned or not. The <tt>var2</tt> most significant
2157bits will be filled with zero bits after the shift.</p>
2158
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2159<h5>Example:</h5>
Reid Spencer3822ff52006-11-08 06:47:33 +0000[diff] [blame]2160<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2161 &lt;result&gt; = lshr i32 4, i8 1 <i>; yields {i32}:result = 2</i>
2162 &lt;result&gt; = lshr i32 4, i8 2 <i>; yields {i32}:result = 1</i>
2163 &lt;result&gt; = lshr i8 4, i8 3 <i>; yields {i8 }:result = 0</i>
2164 &lt;result&gt; = lshr i8 -2, i8 1 <i>; yields {i8 }:result = 0x7FFFFFFF </i>
Reid Spencer3822ff52006-11-08 06:47:33 +0000[diff] [blame]2165</pre>
2166</div>
2167
2168<!-- ======================================================================= -->
2169<div class="doc_subsubsection"> <a name="i_ashr">'<tt>ashr</tt>'
2170Instruction</a> </div>
2171<div class="doc_text">
2172
2173<h5>Syntax:</h5>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2174<pre> &lt;result&gt; = ashr &lt;ty&gt; &lt;var1&gt;, i8 &lt;var2&gt; <i>; yields {ty}:result</i>
Reid Spencer3822ff52006-11-08 06:47:33 +0000[diff] [blame]2175</pre>
2176
2177<h5>Overview:</h5>
2178<p>The '<tt>ashr</tt>' instruction (arithmetic shift right) returns the first
2179operand shifted to the right a specified number of bits.</p>
2180
2181<h5>Arguments:</h5>
2182<p>The first argument to the '<tt>ashr</tt>' instruction must be an
2183<a href="#t_integer">integer</a> type. The second argument must be an
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2184'<tt>i8</tt>' type.</p>
Reid Spencer3822ff52006-11-08 06:47:33 +0000[diff] [blame]2185
2186<h5>Semantics:</h5>
2187<p>This instruction always performs an arithmetic shift right operation,
2188regardless of whether the arguments are signed or not. The <tt>var2</tt> most
2189significant bits will be filled with the sign bit of <tt>var1</tt>.</p>
2190
2191<h5>Example:</h5>
2192<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2193 &lt;result&gt; = ashr i32 4, i8 1 <i>; yields {i32}:result = 2</i>
2194 &lt;result&gt; = ashr i32 4, i8 2 <i>; yields {i32}:result = 1</i>
2195 &lt;result&gt; = ashr i8 4, i8 3 <i>; yields {i8}:result = 0</i>
2196 &lt;result&gt; = ashr i8 -2, i8 1 <i>; yields {i8 }:result = -1</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2197</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2198</div>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2199
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2200<!-- ======================================================================= -->
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2201<div class="doc_subsection">
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]2202 <a name="vectorops">Vector Operations</a>
2203</div>
2204
2205<div class="doc_text">
2206
2207<p>LLVM supports several instructions to represent vector operations in a
2208target-independent manner. This instructions cover the element-access and
2209vector-specific operations needed to process vectors effectively. While LLVM
2210does directly support these vector operations, many sophisticated algorithms
2211will want to use target-specific intrinsics to take full advantage of a specific
2212target.</p>
2213
2214</div>
2215
2216<!-- _______________________________________________________________________ -->
2217<div class="doc_subsubsection">
2218 <a name="i_extractelement">'<tt>extractelement</tt>' Instruction</a>
2219</div>
2220
2221<div class="doc_text">
2222
2223<h5>Syntax:</h5>
2224
2225<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2226 &lt;result&gt; = extractelement &lt;n x &lt;ty&gt;&gt; &lt;val&gt;, i32 &lt;idx&gt; <i>; yields &lt;ty&gt;</i>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]2227</pre>
2228
2229<h5>Overview:</h5>
2230
2231<p>
2232The '<tt>extractelement</tt>' instruction extracts a single scalar
2233element from a packed vector at a specified index.
2234</p>
2235
2236
2237<h5>Arguments:</h5>
2238
2239<p>
2240The first operand of an '<tt>extractelement</tt>' instruction is a
2241value of <a href="#t_packed">packed</a> type. The second operand is
2242an index indicating the position from which to extract the element.
2243The index may be a variable.</p>
2244
2245<h5>Semantics:</h5>
2246
2247<p>
2248The result is a scalar of the same type as the element type of
2249<tt>val</tt>. Its value is the value at position <tt>idx</tt> of
2250<tt>val</tt>. If <tt>idx</tt> exceeds the length of <tt>val</tt>, the
2251results are undefined.
2252</p>
2253
2254<h5>Example:</h5>
2255
2256<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2257 %result = extractelement &lt;4 x i32&gt; %vec, i32 0 <i>; yields i32</i>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]2258</pre>
2259</div>
2260
2261
2262<!-- _______________________________________________________________________ -->
2263<div class="doc_subsubsection">
2264 <a name="i_insertelement">'<tt>insertelement</tt>' Instruction</a>
2265</div>
2266
2267<div class="doc_text">
2268
2269<h5>Syntax:</h5>
2270
2271<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2272 &lt;result&gt; = insertelement &lt;n x &lt;ty&gt;&gt; &lt;val&gt;, &lt;ty&gt; &lt;elt&gt, i32 &lt;idx&gt; <i>; yields &lt;n x &lt;ty&gt;&gt;</i>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]2273</pre>
2274
2275<h5>Overview:</h5>
2276
2277<p>
2278The '<tt>insertelement</tt>' instruction inserts a scalar
2279element into a packed vector at a specified index.
2280</p>
2281
2282
2283<h5>Arguments:</h5>
2284
2285<p>
2286The first operand of an '<tt>insertelement</tt>' instruction is a
2287value of <a href="#t_packed">packed</a> type. The second operand is a
2288scalar value whose type must equal the element type of the first
2289operand. The third operand is an index indicating the position at
2290which to insert the value. The index may be a variable.</p>
2291
2292<h5>Semantics:</h5>
2293
2294<p>
2295The result is a packed vector of the same type as <tt>val</tt>. Its
2296element values are those of <tt>val</tt> except at position
2297<tt>idx</tt>, where it gets the value <tt>elt</tt>. If <tt>idx</tt>
2298exceeds the length of <tt>val</tt>, the results are undefined.
2299</p>
2300
2301<h5>Example:</h5>
2302
2303<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2304 %result = insertelement &lt;4 x i32&gt; %vec, i32 1, i32 0 <i>; yields &lt;4 x i32&gt;</i>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]2305</pre>
2306</div>
2307
2308<!-- _______________________________________________________________________ -->
2309<div class="doc_subsubsection">
2310 <a name="i_shufflevector">'<tt>shufflevector</tt>' Instruction</a>
2311</div>
2312
2313<div class="doc_text">
2314
2315<h5>Syntax:</h5>
2316
2317<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2318 &lt;result&gt; = shufflevector &lt;n x &lt;ty&gt;&gt; &lt;v1&gt;, &lt;n x &lt;ty&gt;&gt; &lt;v2&gt;, &lt;n x i32&gt; &lt;mask&gt; <i>; yields &lt;n x &lt;ty&gt;&gt;</i>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]2319</pre>
2320
2321<h5>Overview:</h5>
2322
2323<p>
2324The '<tt>shufflevector</tt>' instruction constructs a permutation of elements
2325from two input vectors, returning a vector of the same type.
2326</p>
2327
2328<h5>Arguments:</h5>
2329
2330<p>
2331The first two operands of a '<tt>shufflevector</tt>' instruction are vectors
2332with types that match each other and types that match the result of the
2333instruction. The third argument is a shuffle mask, which has the same number
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2334of elements as the other vector type, but whose element type is always 'i32'.
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]2335</p>
2336
2337<p>
2338The shuffle mask operand is required to be a constant vector with either
2339constant integer or undef values.
2340</p>
2341
2342<h5>Semantics:</h5>
2343
2344<p>
2345The elements of the two input vectors are numbered from left to right across
2346both of the vectors. The shuffle mask operand specifies, for each element of
2347the result vector, which element of the two input registers the result element
2348gets. The element selector may be undef (meaning "don't care") and the second
2349operand may be undef if performing a shuffle from only one vector.
2350</p>
2351
2352<h5>Example:</h5>
2353
2354<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2355 %result = shufflevector &lt;4 x i32&gt; %v1, &lt;4 x i32&gt; %v2,
2356 &lt;4 x i32&gt; &lt;i32 0, i32 4, i32 1, i32 5&gt; <i>; yields &lt;4 x i32&gt;</i>
2357 %result = shufflevector &lt;4 x i32&gt; %v1, &lt;4 x i32&gt; undef,
2358 &lt;4 x i32&gt; &lt;i32 0, i32 1, i32 2, i32 3&gt; <i>; yields &lt;4 x i32&gt;</i> - Identity shuffle.
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]2359</pre>
2360</div>
2361
Tanya Lattner09474292006-04-14 19:24:33 +0000[diff] [blame]2362
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]2363<!-- ======================================================================= -->
2364<div class="doc_subsection">
Chris Lattner884a9702006-08-15 00:45:58 +0000[diff] [blame]2365 <a name="memoryops">Memory Access and Addressing Operations</a>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2366</div>
2367
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2368<div class="doc_text">
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2369
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2370<p>A key design point of an SSA-based representation is how it
2371represents memory. In LLVM, no memory locations are in SSA form, which
2372makes things very simple. This section describes how to read, write,
John Criswell9e2485c2004-12-10 15:51:16 +0000[diff] [blame]2373allocate, and free memory in LLVM.</p>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2374
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2375</div>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2376
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2377<!-- _______________________________________________________________________ -->
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2378<div class="doc_subsubsection">
2379 <a name="i_malloc">'<tt>malloc</tt>' Instruction</a>
2380</div>
2381
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2382<div class="doc_text">
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2383
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2384<h5>Syntax:</h5>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2385
2386<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2387 &lt;result&gt; = malloc &lt;type&gt;[, i32 &lt;NumElements&gt;][, align &lt;alignment&gt;] <i>; yields {type*}:result</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2388</pre>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2389
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2390<h5>Overview:</h5>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2391
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2392<p>The '<tt>malloc</tt>' instruction allocates memory from the system
2393heap and returns a pointer to it.</p>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2394
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2395<h5>Arguments:</h5>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2396
2397<p>The '<tt>malloc</tt>' instruction allocates
2398<tt>sizeof(&lt;type&gt;)*NumElements</tt>
John Criswell6e4ca612004-02-24 16:13:56 +0000[diff] [blame]2399bytes of memory from the operating system and returns a pointer of the
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2400appropriate type to the program. If "NumElements" is specified, it is the
2401number of elements allocated. If an alignment is specified, the value result
2402of the allocation is guaranteed to be aligned to at least that boundary. If
2403not specified, or if zero, the target can choose to align the allocation on any
2404convenient boundary.</p>
2405
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2406<p>'<tt>type</tt>' must be a sized type.</p>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2407
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2408<h5>Semantics:</h5>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2409
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2410<p>Memory is allocated using the system "<tt>malloc</tt>" function, and
2411a pointer is returned.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2412
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2413<h5>Example:</h5>
2414
2415<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2416 %array = malloc [4 x i8 ] <i>; yields {[%4 x i8]*}:array</i>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2417
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2418 %size = <a href="#i_add">add</a> i32 2, 2 <i>; yields {i32}:size = i32 4</i>
2419 %array1 = malloc i8, i32 4 <i>; yields {i8*}:array1</i>
2420 %array2 = malloc [12 x i8], i32 %size <i>; yields {[12 x i8]*}:array2</i>
2421 %array3 = malloc i32, i32 4, align 1024 <i>; yields {i32*}:array3</i>
2422 %array4 = malloc i32, align 1024 <i>; yields {i32*}:array4</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2423</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2424</div>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2425
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2426<!-- _______________________________________________________________________ -->
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2427<div class="doc_subsubsection">
2428 <a name="i_free">'<tt>free</tt>' Instruction</a>
2429</div>
2430
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2431<div class="doc_text">
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2432
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2433<h5>Syntax:</h5>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2434
2435<pre>
2436 free &lt;type&gt; &lt;value&gt; <i>; yields {void}</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2437</pre>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2438
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2439<h5>Overview:</h5>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2440
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2441<p>The '<tt>free</tt>' instruction returns memory back to the unused
John Criswellc1f786c2005-05-13 22:25:59 +0000[diff] [blame]2442memory heap to be reallocated in the future.</p>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2443
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2444<h5>Arguments:</h5>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2445
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2446<p>'<tt>value</tt>' shall be a pointer value that points to a value
2447that was allocated with the '<tt><a href="#i_malloc">malloc</a></tt>'
2448instruction.</p>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2449
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2450<h5>Semantics:</h5>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2451
John Criswell9e2485c2004-12-10 15:51:16 +0000[diff] [blame]2452<p>Access to the memory pointed to by the pointer is no longer defined
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2453after this instruction executes.</p>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2454
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2455<h5>Example:</h5>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2456
2457<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2458 %array = <a href="#i_malloc">malloc</a> [4 x i8] <i>; yields {[4 x i8]*}:array</i>
2459 free [4 x i8]* %array
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2460</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2461</div>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2462
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2463<!-- _______________________________________________________________________ -->
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2464<div class="doc_subsubsection">
2465 <a name="i_alloca">'<tt>alloca</tt>' Instruction</a>
2466</div>
2467
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2468<div class="doc_text">
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2469
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2470<h5>Syntax:</h5>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2471
2472<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2473 &lt;result&gt; = alloca &lt;type&gt;[, i32 &lt;NumElements&gt;][, align &lt;alignment&gt;] <i>; yields {type*}:result</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2474</pre>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2475
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2476<h5>Overview:</h5>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2477
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2478<p>The '<tt>alloca</tt>' instruction allocates memory on the current
2479stack frame of the procedure that is live until the current function
2480returns to its caller.</p>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2481
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2482<h5>Arguments:</h5>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2483
John Criswell9e2485c2004-12-10 15:51:16 +0000[diff] [blame]2484<p>The '<tt>alloca</tt>' instruction allocates <tt>sizeof(&lt;type&gt;)*NumElements</tt>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2485bytes of memory on the runtime stack, returning a pointer of the
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2486appropriate type to the program. If "NumElements" is specified, it is the
2487number of elements allocated. If an alignment is specified, the value result
2488of the allocation is guaranteed to be aligned to at least that boundary. If
2489not specified, or if zero, the target can choose to align the allocation on any
2490convenient boundary.</p>
2491
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2492<p>'<tt>type</tt>' may be any sized type.</p>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2493
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2494<h5>Semantics:</h5>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2495
John Criswellc1f786c2005-05-13 22:25:59 +0000[diff] [blame]2496<p>Memory is allocated; a pointer is returned. '<tt>alloca</tt>'d
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2497memory is automatically released when the function returns. The '<tt>alloca</tt>'
2498instruction is commonly used to represent automatic variables that must
2499have an address available. When the function returns (either with the <tt><a
John Criswelldae2e932005-05-12 16:55:34 +0000[diff] [blame]2500 href="#i_ret">ret</a></tt> or <tt><a href="#i_unwind">unwind</a></tt>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2501instructions), the memory is reclaimed.</p>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2502
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2503<h5>Example:</h5>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2504
2505<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2506 %ptr = alloca i32 <i>; yields {i32*}:ptr</i>
2507 %ptr = alloca i32, i32 4 <i>; yields {i32*}:ptr</i>
2508 %ptr = alloca i32, i32 4, align 1024 <i>; yields {i32*}:ptr</i>
2509 %ptr = alloca i32, align 1024 <i>; yields {i32*}:ptr</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2510</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2511</div>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2512
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2513<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2514<div class="doc_subsubsection"> <a name="i_load">'<tt>load</tt>'
2515Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2516<div class="doc_text">
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]2517<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2518<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 +0000[diff] [blame]2519<h5>Overview:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2520<p>The '<tt>load</tt>' instruction is used to read from memory.</p>
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]2521<h5>Arguments:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2522<p>The argument to the '<tt>load</tt>' instruction specifies the memory
John Criswell0ec250c2005-10-24 16:17:18 +0000[diff] [blame]2523address from which to load. The pointer must point to a <a
Chris Lattnere53e5082004-06-03 22:57:15 +0000[diff] [blame]2524 href="#t_firstclass">first class</a> type. If the <tt>load</tt> is
John Criswell0ec250c2005-10-24 16:17:18 +0000[diff] [blame]2525marked as <tt>volatile</tt>, then the optimizer is not allowed to modify
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2526the number or order of execution of this <tt>load</tt> with other
2527volatile <tt>load</tt> and <tt><a href="#i_store">store</a></tt>
2528instructions. </p>
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]2529<h5>Semantics:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2530<p>The location of memory pointed to is loaded.</p>
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]2531<h5>Examples:</h5>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2532<pre> %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2533 <a
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2534 href="#i_store">store</a> i32 3, i32* %ptr <i>; yields {void}</i>
2535 %val = load i32* %ptr <i>; yields {i32}:val = i32 3</i>
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]2536</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2537</div>
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]2538<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2539<div class="doc_subsubsection"> <a name="i_store">'<tt>store</tt>'
2540Instruction</a> </div>
Reid Spencer035ab572006-11-09 21:18:01 +0000[diff] [blame]2541<div class="doc_text">
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]2542<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2543<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 +0000[diff] [blame]2544 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 +0000[diff] [blame]2545</pre>
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]2546<h5>Overview:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2547<p>The '<tt>store</tt>' instruction is used to write to memory.</p>
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]2548<h5>Arguments:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2549<p>There are two arguments to the '<tt>store</tt>' instruction: a value
John Criswell0ec250c2005-10-24 16:17:18 +0000[diff] [blame]2550to store and an address in which to store it. The type of the '<tt>&lt;pointer&gt;</tt>'
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2551operand must be a pointer to the type of the '<tt>&lt;value&gt;</tt>'
John Criswellc1f786c2005-05-13 22:25:59 +0000[diff] [blame]2552operand. If the <tt>store</tt> is marked as <tt>volatile</tt>, then the
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2553optimizer is not allowed to modify the number or order of execution of
2554this <tt>store</tt> with other volatile <tt>load</tt> and <tt><a
2555 href="#i_store">store</a></tt> instructions.</p>
2556<h5>Semantics:</h5>
2557<p>The contents of memory are updated to contain '<tt>&lt;value&gt;</tt>'
2558at the location specified by the '<tt>&lt;pointer&gt;</tt>' operand.</p>
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]2559<h5>Example:</h5>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2560<pre> %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2561 <a
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2562 href="#i_store">store</a> i32 3, i32* %ptr <i>; yields {void}</i>
2563 %val = load i32* %ptr <i>; yields {i32}:val = i32 3</i>
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]2564</pre>
Reid Spencer47ce1792006-11-09 21:15:49 +0000[diff] [blame]2565</div>
2566
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]2567<!-- _______________________________________________________________________ -->
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]2568<div class="doc_subsubsection">
2569 <a name="i_getelementptr">'<tt>getelementptr</tt>' Instruction</a>
2570</div>
2571
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2572<div class="doc_text">
Chris Lattner7faa8832002-04-14 06:13:44 +0000[diff] [blame]2573<h5>Syntax:</h5>
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]2574<pre>
2575 &lt;result&gt; = getelementptr &lt;ty&gt;* &lt;ptrval&gt;{, &lt;ty&gt; &lt;idx&gt;}*
2576</pre>
2577
Chris Lattner7faa8832002-04-14 06:13:44 +0000[diff] [blame]2578<h5>Overview:</h5>
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]2579
2580<p>
2581The '<tt>getelementptr</tt>' instruction is used to get the address of a
2582subelement of an aggregate data structure.</p>
2583
Chris Lattner7faa8832002-04-14 06:13:44 +0000[diff] [blame]2584<h5>Arguments:</h5>
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]2585
Reid Spencer85f5b5b2006-12-04 21:29:24 +0000[diff] [blame]2586<p>This instruction takes a list of integer operands that indicate what
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]2587elements of the aggregate object to index to. The actual types of the arguments
2588provided depend on the type of the first pointer argument. The
2589'<tt>getelementptr</tt>' instruction is used to index down through the type
John Criswellfc6b8952005-05-16 16:17:45 +0000[diff] [blame]2590levels of a structure or to a specific index in an array. When indexing into a
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2591structure, only <tt>i32</tt> integer constants are allowed. When indexing
Reid Spencer85f5b5b2006-12-04 21:29:24 +0000[diff] [blame]2592into an array or pointer, only integers of 32 or 64 bits are allowed, and will
2593be sign extended to 64-bit values.</p>
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]2594
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2595<p>For example, let's consider a C code fragment and how it gets
2596compiled to LLVM:</p>
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]2597
2598<pre>
2599 struct RT {
2600 char A;
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2601 i32 B[10][20];
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]2602 char C;
2603 };
2604 struct ST {
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2605 i32 X;
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]2606 double Y;
2607 struct RT Z;
2608 };
2609
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2610 define i32 *foo(struct ST *s) {
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]2611 return &amp;s[1].Z.B[5][13];
2612 }
2613</pre>
2614
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2615<p>The LLVM code generated by the GCC frontend is:</p>
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]2616
2617<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2618 %RT = type { i8 , [10 x [20 x i32]], i8 }
2619 %ST = type { i32, double, %RT }
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]2620
Brian Gaeke7283e7c2004-07-02 21:08:14 +0000[diff] [blame]2621 implementation
2622
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2623 define i32* %foo(%ST* %s) {
Brian Gaeke7283e7c2004-07-02 21:08:14 +0000[diff] [blame]2624 entry:
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2625 %reg = getelementptr %ST* %s, i32 1, i32 2, i32 1, i32 5, i32 13
2626 ret i32* %reg
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]2627 }
2628</pre>
2629
Chris Lattner7faa8832002-04-14 06:13:44 +0000[diff] [blame]2630<h5>Semantics:</h5>
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]2631
2632<p>The index types specified for the '<tt>getelementptr</tt>' instruction depend
John Criswellc1f786c2005-05-13 22:25:59 +0000[diff] [blame]2633on the pointer type that is being indexed into. <a href="#t_pointer">Pointer</a>
Reid Spencer85f5b5b2006-12-04 21:29:24 +0000[diff] [blame]2634and <a href="#t_array">array</a> types can use a 32-bit or 64-bit
Reid Spencer42ddd842006-12-03 16:53:48 +0000[diff] [blame]2635<a href="#t_integer">integer</a> type but the value will always be sign extended
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2636to 64-bits. <a href="#t_struct">Structure</a> types, require <tt>i32</tt>
Reid Spencer42ddd842006-12-03 16:53:48 +0000[diff] [blame]2637<b>constants</b>.</p>
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]2638
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2639<p>In the example above, the first index is indexing into the '<tt>%ST*</tt>'
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2640type, which is a pointer, yielding a '<tt>%ST</tt>' = '<tt>{ i32, double, %RT
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]2641}</tt>' type, a structure. The second index indexes into the third element of
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2642the structure, yielding a '<tt>%RT</tt>' = '<tt>{ i8 , [10 x [20 x i32]],
2643i8 }</tt>' type, another structure. The third index indexes into the second
2644element of the structure, yielding a '<tt>[10 x [20 x i32]]</tt>' type, an
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]2645array. The two dimensions of the array are subscripted into, yielding an
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2646'<tt>i32</tt>' type. The '<tt>getelementptr</tt>' instruction returns a pointer
2647to this element, thus computing a value of '<tt>i32*</tt>' type.</p>
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]2648
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2649<p>Note that it is perfectly legal to index partially through a
2650structure, returning a pointer to an inner element. Because of this,
2651the LLVM code for the given testcase is equivalent to:</p>
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]2652
2653<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2654 define i32* %foo(%ST* %s) {
2655 %t1 = getelementptr %ST* %s, i32 1 <i>; yields %ST*:%t1</i>
2656 %t2 = getelementptr %ST* %t1, i32 0, i32 2 <i>; yields %RT*:%t2</i>
2657 %t3 = getelementptr %RT* %t2, i32 0, i32 1 <i>; yields [10 x [20 x i32]]*:%t3</i>
2658 %t4 = getelementptr [10 x [20 x i32]]* %t3, i32 0, i32 5 <i>; yields [20 x i32]*:%t4</i>
2659 %t5 = getelementptr [20 x i32]* %t4, i32 0, i32 13 <i>; yields i32*:%t5</i>
2660 ret i32* %t5
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]2661 }
Chris Lattner6536cfe2002-05-06 22:08:29 +0000[diff] [blame]2662</pre>
Chris Lattnere67a9512005-06-24 17:22:57 +0000[diff] [blame]2663
2664<p>Note that it is undefined to access an array out of bounds: array and
2665pointer indexes must always be within the defined bounds of the array type.
2666The one exception for this rules is zero length arrays. These arrays are
2667defined to be accessible as variable length arrays, which requires access
2668beyond the zero'th element.</p>
2669
Chris Lattner884a9702006-08-15 00:45:58 +0000[diff] [blame]2670<p>The getelementptr instruction is often confusing. For some more insight
2671into how it works, see <a href="GetElementPtr.html">the getelementptr
2672FAQ</a>.</p>
2673
Chris Lattner7faa8832002-04-14 06:13:44 +0000[diff] [blame]2674<h5>Example:</h5>
Chris Lattnere67a9512005-06-24 17:22:57 +0000[diff] [blame]2675
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]2676<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2677 <i>; yields [12 x i8]*:aptr</i>
2678 %aptr = getelementptr {i32, [12 x i8]}* %sptr, i64 0, i32 1
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]2679</pre>
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]2680</div>
Reid Spencer47ce1792006-11-09 21:15:49 +0000[diff] [blame]2681
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2682<!-- ======================================================================= -->
Reid Spencer2fd21e62006-11-08 01:18:52 +0000[diff] [blame]2683<div class="doc_subsection"> <a name="convertops">Conversion Operations</a>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2684</div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2685<div class="doc_text">
Reid Spencer2fd21e62006-11-08 01:18:52 +0000[diff] [blame]2686<p>The instructions in this category are the conversion instructions (casting)
2687which all take a single operand and a type. They perform various bit conversions
2688on the operand.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2689</div>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]2690
Chris Lattner6536cfe2002-05-06 22:08:29 +0000[diff] [blame]2691<!-- _______________________________________________________________________ -->
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]2692<div class="doc_subsubsection">
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2693 <a name="i_trunc">'<tt>trunc .. to</tt>' Instruction</a>
2694</div>
2695<div class="doc_text">
2696
2697<h5>Syntax:</h5>
2698<pre>
2699 &lt;result&gt; = trunc &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
2700</pre>
2701
2702<h5>Overview:</h5>
2703<p>
2704The '<tt>trunc</tt>' instruction truncates its operand to the type <tt>ty2</tt>.
2705</p>
2706
2707<h5>Arguments:</h5>
2708<p>
2709The '<tt>trunc</tt>' instruction takes a <tt>value</tt> to trunc, which must
2710be an <a href="#t_integer">integer</a> type, and a type that specifies the size
2711and type of the result, which must be an <a href="#t_integral">integral</a>
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]2712type. The bit size of <tt>value</tt> must be larger than the bit size of
2713<tt>ty2</tt>. Equal sized types are not allowed.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2714
2715<h5>Semantics:</h5>
2716<p>
2717The '<tt>trunc</tt>' instruction truncates the high order bits in <tt>value</tt>
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]2718and converts the remaining bits to <tt>ty2</tt>. Since the source size must be
2719larger than the destination size, <tt>trunc</tt> cannot be a <i>no-op cast</i>.
2720It will always truncate bits.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2721
2722<h5>Example:</h5>
2723<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2724 %X = trunc i32 257 to i8 <i>; yields i8:1</i>
2725 %Y = trunc i32 123 to bool <i>; yields bool:true</i>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2726</pre>
2727</div>
2728
2729<!-- _______________________________________________________________________ -->
2730<div class="doc_subsubsection">
2731 <a name="i_zext">'<tt>zext .. to</tt>' Instruction</a>
2732</div>
2733<div class="doc_text">
2734
2735<h5>Syntax:</h5>
2736<pre>
2737 &lt;result&gt; = zext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
2738</pre>
2739
2740<h5>Overview:</h5>
2741<p>The '<tt>zext</tt>' instruction zero extends its operand to type
2742<tt>ty2</tt>.</p>
2743
2744
2745<h5>Arguments:</h5>
2746<p>The '<tt>zext</tt>' instruction takes a value to cast, which must be of
2747<a href="#t_integral">integral</a> type, and a type to cast it to, which must
2748also be of <a href="#t_integral">integral</a> type. The bit size of the
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]2749<tt>value</tt> must be smaller than the bit size of the destination type,
2750<tt>ty2</tt>.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2751
2752<h5>Semantics:</h5>
2753<p>The <tt>zext</tt> fills the high order bits of the <tt>value</tt> with zero
2754bits until it reaches the size of the destination type, <tt>ty2</tt>. When the
2755the operand and the type are the same size, no bit filling is done and the
2756cast is considered a <i>no-op cast</i> because no bits change (only the type
2757changes).</p>
2758
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]2759<p>When zero extending from bool, the result will alwasy be either 0 or 1.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2760
2761<h5>Example:</h5>
2762<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2763 %X = zext i32 257 to i64 <i>; yields i64:257</i>
2764 %Y = zext bool true to i32 <i>; yields i32:1</i>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2765</pre>
2766</div>
2767
2768<!-- _______________________________________________________________________ -->
2769<div class="doc_subsubsection">
2770 <a name="i_sext">'<tt>sext .. to</tt>' Instruction</a>
2771</div>
2772<div class="doc_text">
2773
2774<h5>Syntax:</h5>
2775<pre>
2776 &lt;result&gt; = sext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
2777</pre>
2778
2779<h5>Overview:</h5>
2780<p>The '<tt>sext</tt>' sign extends <tt>value</tt> to the type <tt>ty2</tt>.</p>
2781
2782<h5>Arguments:</h5>
2783<p>
2784The '<tt>sext</tt>' instruction takes a value to cast, which must be of
2785<a href="#t_integral">integral</a> type, and a type to cast it to, which must
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]2786also be of <a href="#t_integral">integral</a> type. The bit size of the
2787<tt>value</tt> must be smaller than the bit size of the destination type,
2788<tt>ty2</tt>.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2789
2790<h5>Semantics:</h5>
2791<p>
2792The '<tt>sext</tt>' instruction performs a sign extension by copying the sign
2793bit (highest order bit) of the <tt>value</tt> until it reaches the bit size of
2794the type <tt>ty2</tt>. When the the operand and the type are the same size,
2795no bit filling is done and the cast is considered a <i>no-op cast</i> because
2796no bits change (only the type changes).</p>
2797
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]2798<p>When sign extending from bool, the extension always results in -1 or 0.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2799
2800<h5>Example:</h5>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2801<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2802 %X = sext i8 -1 to i16 <i>; yields i16 :65535</i>
2803 %Y = sext bool true to i32 <i>; yields i32:-1</i>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2804</pre>
2805</div>
2806
2807<!-- _______________________________________________________________________ -->
2808<div class="doc_subsubsection">
Reid Spencer3fa91b02006-11-09 21:48:10 +0000[diff] [blame]2809 <a name="i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a>
2810</div>
2811
2812<div class="doc_text">
2813
2814<h5>Syntax:</h5>
2815
2816<pre>
2817 &lt;result&gt; = fptrunc &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
2818</pre>
2819
2820<h5>Overview:</h5>
2821<p>The '<tt>fptrunc</tt>' instruction truncates <tt>value</tt> to type
2822<tt>ty2</tt>.</p>
2823
2824
2825<h5>Arguments:</h5>
2826<p>The '<tt>fptrunc</tt>' instruction takes a <a href="#t_floating">floating
2827 point</a> value to cast and a <a href="#t_floating">floating point</a> type to
2828cast it to. The size of <tt>value</tt> must be larger than the size of
2829<tt>ty2</tt>. This implies that <tt>fptrunc</tt> cannot be used to make a
2830<i>no-op cast</i>.</p>
2831
2832<h5>Semantics:</h5>
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]2833<p> The '<tt>fptrunc</tt>' instruction truncates a <tt>value</tt> from a larger
2834<a href="#t_floating">floating point</a> type to a smaller
2835<a href="#t_floating">floating point</a> type. If the value cannot fit within
2836the destination type, <tt>ty2</tt>, then the results are undefined.</p>
Reid Spencer3fa91b02006-11-09 21:48:10 +0000[diff] [blame]2837
2838<h5>Example:</h5>
2839<pre>
2840 %X = fptrunc double 123.0 to float <i>; yields float:123.0</i>
2841 %Y = fptrunc double 1.0E+300 to float <i>; yields undefined</i>
2842</pre>
2843</div>
2844
2845<!-- _______________________________________________________________________ -->
2846<div class="doc_subsubsection">
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2847 <a name="i_fpext">'<tt>fpext .. to</tt>' Instruction</a>
2848</div>
2849<div class="doc_text">
2850
2851<h5>Syntax:</h5>
2852<pre>
2853 &lt;result&gt; = fpext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
2854</pre>
2855
2856<h5>Overview:</h5>
2857<p>The '<tt>fpext</tt>' extends a floating point <tt>value</tt> to a larger
2858floating point value.</p>
2859
2860<h5>Arguments:</h5>
2861<p>The '<tt>fpext</tt>' instruction takes a
2862<a href="#t_floating">floating point</a> <tt>value</tt> to cast,
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]2863and a <a href="#t_floating">floating point</a> type to cast it to. The source
2864type must be smaller than the destination type.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2865
2866<h5>Semantics:</h5>
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]2867<p>The '<tt>fpext</tt>' instruction extends the <tt>value</tt> from a smaller
2868<a href="t_floating">floating point</a> type to a larger
2869<a href="t_floating">floating point</a> type. The <tt>fpext</tt> cannot be
2870used to make a <i>no-op cast</i> because it always changes bits. Use
Reid Spencer5c0ef472006-11-11 23:08:07 +0000[diff] [blame]2871<tt>bitcast</tt> to make a <i>no-op cast</i> for a floating point cast.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2872
2873<h5>Example:</h5>
2874<pre>
2875 %X = fpext float 3.1415 to double <i>; yields double:3.1415</i>
2876 %Y = fpext float 1.0 to float <i>; yields float:1.0 (no-op)</i>
2877</pre>
2878</div>
2879
2880<!-- _______________________________________________________________________ -->
2881<div class="doc_subsubsection">
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]2882 <a name="i_fp2uint">'<tt>fptoui .. to</tt>' Instruction</a>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2883</div>
2884<div class="doc_text">
2885
2886<h5>Syntax:</h5>
2887<pre>
2888 &lt;result&gt; = fp2uint &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
2889</pre>
2890
2891<h5>Overview:</h5>
2892<p>The '<tt>fp2uint</tt>' converts a floating point <tt>value</tt> to its
2893unsigned integer equivalent of type <tt>ty2</tt>.
2894</p>
2895
2896<h5>Arguments:</h5>
2897<p>The '<tt>fp2uint</tt>' instruction takes a value to cast, which must be a
2898<a href="#t_floating">floating point</a> value, and a type to cast it to, which
2899must be an <a href="#t_integral">integral</a> type.</p>
2900
2901<h5>Semantics:</h5>
2902<p> The '<tt>fp2uint</tt>' instruction converts its
2903<a href="#t_floating">floating point</a> operand into the nearest (rounding
2904towards zero) unsigned integer value. If the value cannot fit in <tt>ty2</tt>,
2905the results are undefined.</p>
2906
2907<p>When converting to bool, the conversion is done as a comparison against
2908zero. If the <tt>value</tt> was zero, the bool result will be <tt>false</tt>.
2909If the <tt>value</tt> was non-zero, the bool result will be <tt>true</tt>.</p>
2910
2911<h5>Example:</h5>
2912<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2913 %X = fp2uint double 123.0 to i32 <i>; yields i32:123</i>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2914 %Y = fp2uint float 1.0E+300 to bool <i>; yields bool:true</i>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2915 %X = fp2uint float 1.04E+17 to i8 <i>; yields undefined:1</i>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2916</pre>
2917</div>
2918
2919<!-- _______________________________________________________________________ -->
2920<div class="doc_subsubsection">
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]2921 <a name="i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2922</div>
2923<div class="doc_text">
2924
2925<h5>Syntax:</h5>
2926<pre>
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]2927 &lt;result&gt; = fptosi &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2928</pre>
2929
2930<h5>Overview:</h5>
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]2931<p>The '<tt>fptosi</tt>' instruction converts
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2932<a href="#t_floating">floating point</a> <tt>value</tt> to type <tt>ty2</tt>.
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]2933</p>
2934
2935
Chris Lattner6536cfe2002-05-06 22:08:29 +0000[diff] [blame]2936<h5>Arguments:</h5>
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]2937<p> The '<tt>fptosi</tt>' instruction takes a value to cast, which must be a
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2938<a href="#t_floating">floating point</a> value, and a type to cast it to, which
2939must also be an <a href="#t_integral">integral</a> type.</p>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]2940
Chris Lattner6536cfe2002-05-06 22:08:29 +0000[diff] [blame]2941<h5>Semantics:</h5>
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]2942<p>The '<tt>fptosi</tt>' instruction converts its
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2943<a href="#t_floating">floating point</a> operand into the nearest (rounding
2944towards zero) signed integer value. If the value cannot fit in <tt>ty2</tt>,
2945the results are undefined.</p>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]2946
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2947<p>When converting to bool, the conversion is done as a comparison against
2948zero. If the <tt>value</tt> was zero, the bool result will be <tt>false</tt>.
2949If the <tt>value</tt> was non-zero, the bool result will be <tt>true</tt>.</p>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]2950
Chris Lattner33ba0d92001-07-09 00:26:23 +0000[diff] [blame]2951<h5>Example:</h5>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]2952<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2953 %X = fptosi double -123.0 to i32 <i>; yields i32:-123</i>
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]2954 %Y = fptosi float 1.0E-247 to bool <i>; yields bool:true</i>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2955 %X = fptosi float 1.04E+17 to i8 <i>; yields undefined:1</i>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2956</pre>
2957</div>
2958
2959<!-- _______________________________________________________________________ -->
2960<div class="doc_subsubsection">
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]2961 <a name="i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2962</div>
2963<div class="doc_text">
2964
2965<h5>Syntax:</h5>
2966<pre>
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]2967 &lt;result&gt; = uitofp &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2968</pre>
2969
2970<h5>Overview:</h5>
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]2971<p>The '<tt>uitofp</tt>' instruction regards <tt>value</tt> as an unsigned
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2972integer and converts that value to the <tt>ty2</tt> type.</p>
2973
2974
2975<h5>Arguments:</h5>
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]2976<p>The '<tt>uitofp</tt>' instruction takes a value to cast, which must be an
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2977<a href="#t_integral">integral</a> value, and a type to cast it to, which must
2978be a <a href="#t_floating">floating point</a> type.</p>
2979
2980<h5>Semantics:</h5>
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]2981<p>The '<tt>uitofp</tt>' instruction interprets its operand as an unsigned
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2982integer quantity and converts it to the corresponding floating point value. If
2983the value cannot fit in the floating point value, the results are undefined.</p>
2984
2985
2986<h5>Example:</h5>
2987<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2988 %X = uitofp i32 257 to float <i>; yields float:257.0</i>
2989 %Y = uitofp i8 -1 to double <i>; yields double:255.0</i>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2990</pre>
2991</div>
2992
2993<!-- _______________________________________________________________________ -->
2994<div class="doc_subsubsection">
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]2995 <a name="i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2996</div>
2997<div class="doc_text">
2998
2999<h5>Syntax:</h5>
3000<pre>
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]3001 &lt;result&gt; = sitofp &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]3002</pre>
3003
3004<h5>Overview:</h5>
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]3005<p>The '<tt>sitofp</tt>' instruction regards <tt>value</tt> as a signed
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]3006integer and converts that value to the <tt>ty2</tt> type.</p>
3007
3008<h5>Arguments:</h5>
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]3009<p>The '<tt>sitofp</tt>' instruction takes a value to cast, which must be an
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]3010<a href="#t_integral">integral</a> value, and a type to cast it to, which must be
3011a <a href="#t_floating">floating point</a> type.</p>
3012
3013<h5>Semantics:</h5>
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]3014<p>The '<tt>sitofp</tt>' instruction interprets its operand as a signed
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]3015integer quantity and converts it to the corresponding floating point value. If
3016the value cannot fit in the floating point value, the results are undefined.</p>
3017
3018<h5>Example:</h5>
3019<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]3020 %X = sitofp i32 257 to float <i>; yields float:257.0</i>
3021 %Y = sitofp i8 -1 to double <i>; yields double:-1.0</i>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]3022</pre>
3023</div>
3024
3025<!-- _______________________________________________________________________ -->
3026<div class="doc_subsubsection">
Reid Spencer72679252006-11-11 21:00:47 +0000[diff] [blame]3027 <a name="i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a>
3028</div>
3029<div class="doc_text">
3030
3031<h5>Syntax:</h5>
3032<pre>
3033 &lt;result&gt; = ptrtoint &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3034</pre>
3035
3036<h5>Overview:</h5>
3037<p>The '<tt>ptrtoint</tt>' instruction converts the pointer <tt>value</tt> to
3038the integer type <tt>ty2</tt>.</p>
3039
3040<h5>Arguments:</h5>
3041<p>The '<tt>ptrtoint</tt>' instruction takes a <tt>value</tt> to cast, which
3042must be a <a href="t_pointer">pointer</a> value, and a type to cast it to
3043<tt>ty2</tt>, which must be an <a href="#t_integer">integer</a> type.
3044
3045<h5>Semantics:</h5>
3046<p>The '<tt>ptrtoint</tt>' instruction converts <tt>value</tt> to integer type
3047<tt>ty2</tt> by interpreting the pointer value as an integer and either
3048truncating or zero extending that value to the size of the integer type. If
3049<tt>value</tt> is smaller than <tt>ty2</tt> then a zero extension is done. If
3050<tt>value</tt> is larger than <tt>ty2</tt> then a truncation is done. If they
3051are the same size, then nothing is done (<i>no-op cast</i>).</p>
3052
3053<h5>Example:</h5>
3054<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]3055 %X = ptrtoint i32* %X to i8 <i>; yields truncation on 32-bit</i>
3056 %Y = ptrtoint i32* %x to i64 <i>; yields zero extend on 32-bit</i>
Reid Spencer72679252006-11-11 21:00:47 +0000[diff] [blame]3057</pre>
3058</div>
3059
3060<!-- _______________________________________________________________________ -->
3061<div class="doc_subsubsection">
3062 <a name="i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a>
3063</div>
3064<div class="doc_text">
3065
3066<h5>Syntax:</h5>
3067<pre>
3068 &lt;result&gt; = inttoptr &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3069</pre>
3070
3071<h5>Overview:</h5>
3072<p>The '<tt>inttoptr</tt>' instruction converts an integer <tt>value</tt> to
3073a pointer type, <tt>ty2</tt>.</p>
3074
3075<h5>Arguments:</h5>
3076<p>The '<tt>inttoptr</tt>' instruction takes an <a href="i_integer">integer</a>
3077value to cast, and a type to cast it to, which must be a
3078<a href="#t_pointer">pointer</a> type. </tt>
3079
3080<h5>Semantics:</h5>
3081<p>The '<tt>inttoptr</tt>' instruction converts <tt>value</tt> to type
3082<tt>ty2</tt> by applying either a zero extension or a truncation depending on
3083the size of the integer <tt>value</tt>. If <tt>value</tt> is larger than the
3084size of a pointer then a truncation is done. If <tt>value</tt> is smaller than
3085the size of a pointer then a zero extension is done. If they are the same size,
3086nothing is done (<i>no-op cast</i>).</p>
3087
3088<h5>Example:</h5>
3089<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]3090 %X = inttoptr i32 255 to i32* <i>; yields zero extend on 64-bit</i>
3091 %X = inttoptr i32 255 to i32* <i>; yields no-op on 32-bit </i>
3092 %Y = inttoptr i16 0 to i32* <i>; yields zero extend on 32-bit</i>
Reid Spencer72679252006-11-11 21:00:47 +0000[diff] [blame]3093</pre>
3094</div>
3095
3096<!-- _______________________________________________________________________ -->
3097<div class="doc_subsubsection">
Reid Spencer5c0ef472006-11-11 23:08:07 +0000[diff] [blame]3098 <a name="i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]3099</div>
3100<div class="doc_text">
3101
3102<h5>Syntax:</h5>
3103<pre>
Reid Spencer5c0ef472006-11-11 23:08:07 +0000[diff] [blame]3104 &lt;result&gt; = bitcast &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]3105</pre>
3106
3107<h5>Overview:</h5>
Reid Spencer5c0ef472006-11-11 23:08:07 +0000[diff] [blame]3108<p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]3109<tt>ty2</tt> without changing any bits.</p>
3110
3111<h5>Arguments:</h5>
Reid Spencer5c0ef472006-11-11 23:08:07 +0000[diff] [blame]3112<p>The '<tt>bitcast</tt>' instruction takes a value to cast, which must be
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]3113a first class value, and a type to cast it to, which must also be a <a
3114 href="#t_firstclass">first class</a> type. The bit sizes of <tt>value</tt>
3115and the destination type, <tt>ty2</tt>, must be identical.</p>
3116
3117<h5>Semantics:</h5>
Reid Spencer5c0ef472006-11-11 23:08:07 +0000[diff] [blame]3118<p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
Reid Spencer72679252006-11-11 21:00:47 +0000[diff] [blame]3119<tt>ty2</tt>. It is always a <i>no-op cast</i> because no bits change with
3120this conversion. The conversion is done as if the <tt>value</tt> had been
3121stored to memory and read back as type <tt>ty2</tt>. Pointer types may only be
3122converted to other pointer types with this instruction. To convert pointers to
3123other types, use the <a href="#i_inttoptr">inttoptr</a> or
3124<a href="#i_ptrtoint">ptrtoint</a> instructions first.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]3125
3126<h5>Example:</h5>
3127<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]3128 %X = bitcast i8 255 to i8 <i>; yields i8 :-1</i>
3129 %Y = bitcast i32* %x to sint* <i>; yields sint*:%x</i>
3130 %Z = bitcast <2xint> %V to i64; <i>; yields i64: %V</i>
Chris Lattner33ba0d92001-07-09 00:26:23 +0000[diff] [blame]3131</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3132</div>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]3133
Reid Spencer2fd21e62006-11-08 01:18:52 +0000[diff] [blame]3134<!-- ======================================================================= -->
3135<div class="doc_subsection"> <a name="otherops">Other Operations</a> </div>
3136<div class="doc_text">
3137<p>The instructions in this category are the "miscellaneous"
3138instructions, which defy better classification.</p>
3139</div>
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]3140
3141<!-- _______________________________________________________________________ -->
3142<div class="doc_subsubsection"><a name="i_icmp">'<tt>icmp</tt>' Instruction</a>
3143</div>
3144<div class="doc_text">
3145<h5>Syntax:</h5>
3146<pre> &lt;result&gt; = icmp &lt;cond&gt; &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {bool}:result</i>
3147</pre>
3148<h5>Overview:</h5>
3149<p>The '<tt>icmp</tt>' instruction returns a boolean value based on comparison
3150of its two integer operands.</p>
3151<h5>Arguments:</h5>
3152<p>The '<tt>icmp</tt>' instruction takes three operands. The first operand is
3153the condition code which indicates the kind of comparison to perform. It is not
3154a value, just a keyword. The possibilities for the condition code are:
3155<ol>
3156 <li><tt>eq</tt>: equal</li>
3157 <li><tt>ne</tt>: not equal </li>
3158 <li><tt>ugt</tt>: unsigned greater than</li>
3159 <li><tt>uge</tt>: unsigned greater or equal</li>
3160 <li><tt>ult</tt>: unsigned less than</li>
3161 <li><tt>ule</tt>: unsigned less or equal</li>
3162 <li><tt>sgt</tt>: signed greater than</li>
3163 <li><tt>sge</tt>: signed greater or equal</li>
3164 <li><tt>slt</tt>: signed less than</li>
3165 <li><tt>sle</tt>: signed less or equal</li>
3166</ol>
3167<p>The remaining two arguments must be of <a href="#t_integral">integral</a>,
3168<a href="#t_pointer">pointer</a> or a <a href="#t_packed">packed</a> integral
3169type. They must have identical types.</p>
3170<h5>Semantics:</h5>
3171<p>The '<tt>icmp</tt>' compares <tt>var1</tt> and <tt>var2</tt> according to
3172the condition code given as <tt>cond</tt>. The comparison performed always
3173yields a <a href="#t_bool">bool</a> result, as follows:
3174<ol>
3175 <li><tt>eq</tt>: yields <tt>true</tt> if the operands are equal,
3176 <tt>false</tt> otherwise. No sign interpretation is necessary or performed.
3177 </li>
3178 <li><tt>ne</tt>: yields <tt>true</tt> if the operands are unequal,
3179 <tt>false</tt> otherwise. No sign interpretation is necessary or performed.
3180 <li><tt>ugt</tt>: interprets the operands as unsigned values and yields
3181 <tt>true</tt> if <tt>var1</tt> is greater than <tt>var2</tt>.</li>
3182 <li><tt>uge</tt>: interprets the operands as unsigned values and yields
3183 <tt>true</tt> if <tt>var1</tt> is greater than or equal to <tt>var2</tt>.</li>
3184 <li><tt>ult</tt>: interprets the operands as unsigned values and yields
3185 <tt>true</tt> if <tt>var1</tt> is less than <tt>var2</tt>.</li>
3186 <li><tt>ule</tt>: interprets the operands as unsigned values and yields
3187 <tt>true</tt> if <tt>var1</tt> is less than or equal to <tt>var2</tt>.</li>
3188 <li><tt>sgt</tt>: interprets the operands as signed values and yields
3189 <tt>true</tt> if <tt>var1</tt> is greater than <tt>var2</tt>.</li>
3190 <li><tt>sge</tt>: interprets the operands as signed values and yields
3191 <tt>true</tt> if <tt>var1</tt> is greater than or equal to <tt>var2</tt>.</li>
3192 <li><tt>slt</tt>: interprets the operands as signed values and yields
3193 <tt>true</tt> if <tt>var1</tt> is less than <tt>var2</tt>.</li>
3194 <li><tt>sle</tt>: interprets the operands as signed values and yields
3195 <tt>true</tt> if <tt>var1</tt> is less than or equal to <tt>var2</tt>.</li>
3196 </li>
3197</ol>
3198<p>If the operands are <a href="#t_pointer">pointer</a> typed, the pointer
3199values are treated as integers and then compared.</p>
3200<p>If the operands are <a href="#t_packed">packed</a> typed, the elements of
Reid Spencerb7f26282006-11-19 03:00:14 +0000[diff] [blame]3201the vector are compared in turn and the predicate must hold for all
3202elements.</p>
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]3203
3204<h5>Example:</h5>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]3205<pre> &lt;result&gt; = icmp eq i32 4, 5 <i>; yields: result=false</i>
3206 &lt;result&gt; = icmp ne float* %X, %X <i>; yields: result=false</i>
3207 &lt;result&gt; = icmp ult i16 4, 5 <i>; yields: result=true</i>
3208 &lt;result&gt; = icmp sgt i16 4, 5 <i>; yields: result=false</i>
3209 &lt;result&gt; = icmp ule i16 -4, 5 <i>; yields: result=false</i>
3210 &lt;result&gt; = icmp sge i16 4, 5 <i>; yields: result=false</i>
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]3211</pre>
3212</div>
3213
3214<!-- _______________________________________________________________________ -->
3215<div class="doc_subsubsection"><a name="i_fcmp">'<tt>fcmp</tt>' Instruction</a>
3216</div>
3217<div class="doc_text">
3218<h5>Syntax:</h5>
3219<pre> &lt;result&gt; = fcmp &lt;cond&gt; &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {bool}:result</i>
3220</pre>
3221<h5>Overview:</h5>
3222<p>The '<tt>fcmp</tt>' instruction returns a boolean value based on comparison
3223of its floating point operands.</p>
3224<h5>Arguments:</h5>
3225<p>The '<tt>fcmp</tt>' instruction takes three operands. The first operand is
3226the condition code which indicates the kind of comparison to perform. It is not
3227a value, just a keyword. The possibilities for the condition code are:
3228<ol>
Reid Spencerb7f26282006-11-19 03:00:14 +0000[diff] [blame]3229 <li><tt>false</tt>: no comparison, always returns false</li>
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]3230 <li><tt>oeq</tt>: ordered and equal</li>
3231 <li><tt>ogt</tt>: ordered and greater than </li>
3232 <li><tt>oge</tt>: ordered and greater than or equal</li>
3233 <li><tt>olt</tt>: ordered and less than </li>
3234 <li><tt>ole</tt>: ordered and less than or equal</li>
3235 <li><tt>one</tt>: ordered and not equal</li>
3236 <li><tt>ord</tt>: ordered (no nans)</li>
3237 <li><tt>ueq</tt>: unordered or equal</li>
3238 <li><tt>ugt</tt>: unordered or greater than </li>
3239 <li><tt>uge</tt>: unordered or greater than or equal</li>
3240 <li><tt>ult</tt>: unordered or less than </li>
3241 <li><tt>ule</tt>: unordered or less than or equal</li>
3242 <li><tt>une</tt>: unordered or not equal</li>
3243 <li><tt>uno</tt>: unordered (either nans)</li>
Reid Spencerb7f26282006-11-19 03:00:14 +0000[diff] [blame]3244 <li><tt>true</tt>: no comparison, always returns true</li>
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]3245</ol>
Reid Spencer93a49852006-12-06 07:08:07 +0000[diff] [blame]3246<p>In the preceding, <i>ordered</i> means that neither operand is a QNAN while
3247<i>unordered</i> means that either operand may be a QNAN.</p>
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]3248<p>The <tt>val1</tt> and <tt>val2</tt> arguments must be of
3249<a href="#t_floating">floating point</a>, or a <a href="#t_packed">packed</a>
3250floating point type. They must have identical types.</p>
Reid Spencerb7f26282006-11-19 03:00:14 +0000[diff] [blame]3251<p>In the foregoing, <i>ordered</i> means that neither operand is a QNAN and
3252<i>unordered</i> means that either operand is a QNAN.</p>
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]3253<h5>Semantics:</h5>
3254<p>The '<tt>fcmp</tt>' compares <tt>var1</tt> and <tt>var2</tt> according to
3255the condition code given as <tt>cond</tt>. The comparison performed always
3256yields a <a href="#t_bool">bool</a> result, as follows:
3257<ol>
3258 <li><tt>false</tt>: always yields <tt>false</tt>, regardless of operands.</li>
Reid Spencerb7f26282006-11-19 03:00:14 +0000[diff] [blame]3259 <li><tt>oeq</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]3260 <tt>var1</tt> is equal to <tt>var2</tt>.</li>
Reid Spencerb7f26282006-11-19 03:00:14 +0000[diff] [blame]3261 <li><tt>ogt</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]3262 <tt>var1</tt> is greather than <tt>var2</tt>.</li>
Reid Spencerb7f26282006-11-19 03:00:14 +0000[diff] [blame]3263 <li><tt>oge</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]3264 <tt>var1</tt> is greater than or equal to <tt>var2</tt>.</li>
Reid Spencerb7f26282006-11-19 03:00:14 +0000[diff] [blame]3265 <li><tt>olt</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]3266 <tt>var1</tt> is less than <tt>var2</tt>.</li>
Reid Spencerb7f26282006-11-19 03:00:14 +0000[diff] [blame]3267 <li><tt>ole</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]3268 <tt>var1</tt> is less than or equal to <tt>var2</tt>.</li>
Reid Spencerb7f26282006-11-19 03:00:14 +0000[diff] [blame]3269 <li><tt>one</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]3270 <tt>var1</tt> is not equal to <tt>var2</tt>.</li>
Reid Spencerb7f26282006-11-19 03:00:14 +0000[diff] [blame]3271 <li><tt>ord</tt>: yields <tt>true</tt> if both operands are not a QNAN.</li>
3272 <li><tt>ueq</tt>: yields <tt>true</tt> if either operand is a QNAN or
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]3273 <tt>var1</tt> is equal to <tt>var2</tt>.</li>
Reid Spencerb7f26282006-11-19 03:00:14 +0000[diff] [blame]3274 <li><tt>ugt</tt>: yields <tt>true</tt> if either operand is a QNAN or
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]3275 <tt>var1</tt> is greater than <tt>var2</tt>.</li>
Reid Spencerb7f26282006-11-19 03:00:14 +0000[diff] [blame]3276 <li><tt>uge</tt>: yields <tt>true</tt> if either operand is a QNAN or
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]3277 <tt>var1</tt> is greater than or equal to <tt>var2</tt>.</li>
Reid Spencerb7f26282006-11-19 03:00:14 +0000[diff] [blame]3278 <li><tt>ult</tt>: yields <tt>true</tt> if either operand is a QNAN or
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]3279 <tt>var1</tt> is less than <tt>var2</tt>.</li>
Reid Spencerb7f26282006-11-19 03:00:14 +0000[diff] [blame]3280 <li><tt>ule</tt>: yields <tt>true</tt> if either operand is a QNAN or
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]3281 <tt>var1</tt> is less than or equal to <tt>var2</tt>.</li>
Reid Spencerb7f26282006-11-19 03:00:14 +0000[diff] [blame]3282 <li><tt>une</tt>: yields <tt>true</tt> if either operand is a QNAN or
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]3283 <tt>var1</tt> is not equal to <tt>var2</tt>.</li>
Reid Spencerb7f26282006-11-19 03:00:14 +0000[diff] [blame]3284 <li><tt>uno</tt>: yields <tt>true</tt> if either operand is a QNAN.</li>
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]3285 <li><tt>true</tt>: always yields <tt>true</tt>, regardless of operands.</li>
3286</ol>
3287<p>If the operands are <a href="#t_packed">packed</a> typed, the elements of
3288the vector are compared in turn and the predicate must hold for all elements.
Reid Spencerb7f26282006-11-19 03:00:14 +0000[diff] [blame]3289</p>
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]3290
3291<h5>Example:</h5>
3292<pre> &lt;result&gt; = fcmp oeq float 4.0, 5.0 <i>; yields: result=false</i>
3293 &lt;result&gt; = icmp one float 4.0, 5.0 <i>; yields: result=true</i>
3294 &lt;result&gt; = icmp olt float 4.0, 5.0 <i>; yields: result=true</i>
3295 &lt;result&gt; = icmp ueq double 1.0, 2.0 <i>; yields: result=false</i>
3296</pre>
3297</div>
3298
Reid Spencer2fd21e62006-11-08 01:18:52 +0000[diff] [blame]3299<!-- _______________________________________________________________________ -->
3300<div class="doc_subsubsection"> <a name="i_phi">'<tt>phi</tt>'
3301Instruction</a> </div>
3302<div class="doc_text">
3303<h5>Syntax:</h5>
3304<pre> &lt;result&gt; = phi &lt;ty&gt; [ &lt;val0&gt;, &lt;label0&gt;], ...<br></pre>
3305<h5>Overview:</h5>
3306<p>The '<tt>phi</tt>' instruction is used to implement the &#966; node in
3307the SSA graph representing the function.</p>
3308<h5>Arguments:</h5>
3309<p>The type of the incoming values are specified with the first type
3310field. After this, the '<tt>phi</tt>' instruction takes a list of pairs
3311as arguments, with one pair for each predecessor basic block of the
3312current block. Only values of <a href="#t_firstclass">first class</a>
3313type may be used as the value arguments to the PHI node. Only labels
3314may be used as the label arguments.</p>
3315<p>There must be no non-phi instructions between the start of a basic
3316block and the PHI instructions: i.e. PHI instructions must be first in
3317a basic block.</p>
3318<h5>Semantics:</h5>
3319<p>At runtime, the '<tt>phi</tt>' instruction logically takes on the
3320value specified by the parameter, depending on which basic block we
3321came from in the last <a href="#terminators">terminator</a> instruction.</p>
3322<h5>Example:</h5>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]3323<pre>Loop: ; Infinite loop that counts from 0 on up...<br> %indvar = phi i32 [ 0, %LoopHeader ], [ %nextindvar, %Loop ]<br> %nextindvar = add i32 %indvar, 1<br> br label %Loop<br></pre>
Reid Spencer2fd21e62006-11-08 01:18:52 +0000[diff] [blame]3324</div>
3325
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]3326<!-- _______________________________________________________________________ -->
3327<div class="doc_subsubsection">
3328 <a name="i_select">'<tt>select</tt>' Instruction</a>
3329</div>
3330
3331<div class="doc_text">
3332
3333<h5>Syntax:</h5>
3334
3335<pre>
3336 &lt;result&gt; = select bool &lt;cond&gt;, &lt;ty&gt; &lt;val1&gt;, &lt;ty&gt; &lt;val2&gt; <i>; yields ty</i>
3337</pre>
3338
3339<h5>Overview:</h5>
3340
3341<p>
3342The '<tt>select</tt>' instruction is used to choose one value based on a
3343condition, without branching.
3344</p>
3345
3346
3347<h5>Arguments:</h5>
3348
3349<p>
3350The '<tt>select</tt>' instruction requires a boolean value indicating the condition, and two values of the same <a href="#t_firstclass">first class</a> type.
3351</p>
3352
3353<h5>Semantics:</h5>
3354
3355<p>
3356If the boolean condition evaluates to true, the instruction returns the first
John Criswellfc6b8952005-05-16 16:17:45 +0000[diff] [blame]3357value argument; otherwise, it returns the second value argument.
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]3358</p>
3359
3360<h5>Example:</h5>
3361
3362<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]3363 %X = select bool true, i8 17, i8 42 <i>; yields i8:17</i>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]3364</pre>
3365</div>
3366
Robert Bocchino05ccd702006-01-15 20:48:27 +0000[diff] [blame]3367
3368<!-- _______________________________________________________________________ -->
3369<div class="doc_subsubsection">
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]3370 <a name="i_call">'<tt>call</tt>' Instruction</a>
3371</div>
3372
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3373<div class="doc_text">
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]3374
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3375<h5>Syntax:</h5>
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]3376<pre>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]3377 &lt;result&gt; = [tail] call [<a href="#callingconv">cconv</a>] &lt;ty&gt;* &lt;fnptrval&gt;(&lt;param list&gt;)
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]3378</pre>
3379
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3380<h5>Overview:</h5>
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]3381
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3382<p>The '<tt>call</tt>' instruction represents a simple function call.</p>
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]3383
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3384<h5>Arguments:</h5>
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]3385
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3386<p>This instruction requires several arguments:</p>
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]3387
Chris Lattner6536cfe2002-05-06 22:08:29 +0000[diff] [blame]3388<ol>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3389 <li>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]3390 <p>The optional "tail" marker indicates whether the callee function accesses
3391 any allocas or varargs in the caller. If the "tail" marker is present, the
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]3392 function call is eligible for tail call optimization. Note that calls may
3393 be marked "tail" even if they do not occur before a <a
3394 href="#i_ret"><tt>ret</tt></a> instruction.
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3395 </li>
3396 <li>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]3397 <p>The optional "cconv" marker indicates which <a href="callingconv">calling
3398 convention</a> the call should use. If none is specified, the call defaults
3399 to using C calling conventions.
3400 </li>
3401 <li>
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]3402 <p>'<tt>ty</tt>': shall be the signature of the pointer to function value
3403 being invoked. The argument types must match the types implied by this
John Criswellfc6b8952005-05-16 16:17:45 +0000[diff] [blame]3404 signature. This type can be omitted if the function is not varargs and
3405 if the function type does not return a pointer to a function.</p>
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]3406 </li>
3407 <li>
3408 <p>'<tt>fnptrval</tt>': An LLVM value containing a pointer to a function to
3409 be invoked. In most cases, this is a direct function invocation, but
3410 indirect <tt>call</tt>s are just as possible, calling an arbitrary pointer
John Criswellfc6b8952005-05-16 16:17:45 +0000[diff] [blame]3411 to function value.</p>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3412 </li>
3413 <li>
3414 <p>'<tt>function args</tt>': argument list whose types match the
Reid Spencera7e302a2005-05-01 22:22:57 +0000[diff] [blame]3415 function signature argument types. All arguments must be of
3416 <a href="#t_firstclass">first class</a> type. If the function signature
3417 indicates the function accepts a variable number of arguments, the extra
3418 arguments can be specified.</p>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3419 </li>
Chris Lattner6536cfe2002-05-06 22:08:29 +0000[diff] [blame]3420</ol>
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]3421
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3422<h5>Semantics:</h5>
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]3423
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3424<p>The '<tt>call</tt>' instruction is used to cause control flow to
3425transfer to a specified function, with its incoming arguments bound to
3426the specified values. Upon a '<tt><a href="#i_ret">ret</a></tt>'
3427instruction in the called function, control flow continues with the
3428instruction after the function call, and the return value of the
3429function is bound to the result argument. This is a simpler case of
3430the <a href="#i_invoke">invoke</a> instruction.</p>
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]3431
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3432<h5>Example:</h5>
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]3433
3434<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]3435 %retval = call i32 %test(i32 %argc)
3436 call i32(i8 *, ...) *%printf(i8 * %msg, i32 12, i8 42);
3437 %X = tail call i32 %foo()
3438 %Y = tail call <a href="#callingconv">fastcc</a> i32 %foo()
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]3439</pre>
3440
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3441</div>
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]3442
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]3443<!-- _______________________________________________________________________ -->
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]3444<div class="doc_subsubsection">
Chris Lattnerfb6977d2006-01-13 23:26:01 +0000[diff] [blame]3445 <a name="i_va_arg">'<tt>va_arg</tt>' Instruction</a>
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]3446</div>
3447
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3448<div class="doc_text">
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]3449
Chris Lattner8d1a81d2003-10-18 05:51:36 +0000[diff] [blame]3450<h5>Syntax:</h5>
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]3451
3452<pre>
Andrew Lenharth8bf607a2005-06-18 18:28:17 +0000[diff] [blame]3453 &lt;resultval&gt; = va_arg &lt;va_list*&gt; &lt;arglist&gt;, &lt;argty&gt;
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]3454</pre>
3455
Chris Lattner8d1a81d2003-10-18 05:51:36 +0000[diff] [blame]3456<h5>Overview:</h5>
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]3457
Andrew Lenharth8bf607a2005-06-18 18:28:17 +0000[diff] [blame]3458<p>The '<tt>va_arg</tt>' instruction is used to access arguments passed through
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]3459the "variable argument" area of a function call. It is used to implement the
3460<tt>va_arg</tt> macro in C.</p>
3461
Chris Lattner8d1a81d2003-10-18 05:51:36 +0000[diff] [blame]3462<h5>Arguments:</h5>
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]3463
Andrew Lenharth8bf607a2005-06-18 18:28:17 +0000[diff] [blame]3464<p>This instruction takes a <tt>va_list*</tt> value and the type of
3465the argument. It returns a value of the specified argument type and
Jeff Cohen25d4f7e2005-11-11 02:15:27 +0000[diff] [blame]3466increments the <tt>va_list</tt> to point to the next argument. Again, the
Andrew Lenharth8bf607a2005-06-18 18:28:17 +0000[diff] [blame]3467actual type of <tt>va_list</tt> is target specific.</p>
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]3468
Chris Lattner8d1a81d2003-10-18 05:51:36 +0000[diff] [blame]3469<h5>Semantics:</h5>
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]3470
Andrew Lenharth8bf607a2005-06-18 18:28:17 +0000[diff] [blame]3471<p>The '<tt>va_arg</tt>' instruction loads an argument of the specified
3472type from the specified <tt>va_list</tt> and causes the
3473<tt>va_list</tt> to point to the next argument. For more information,
3474see the variable argument handling <a href="#int_varargs">Intrinsic
3475Functions</a>.</p>
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]3476
3477<p>It is legal for this instruction to be called in a function which does not
3478take a variable number of arguments, for example, the <tt>vfprintf</tt>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3479function.</p>
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]3480
Andrew Lenharth8bf607a2005-06-18 18:28:17 +0000[diff] [blame]3481<p><tt>va_arg</tt> is an LLVM instruction instead of an <a
John Criswellfc6b8952005-05-16 16:17:45 +0000[diff] [blame]3482href="#intrinsics">intrinsic function</a> because it takes a type as an
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]3483argument.</p>
3484
Chris Lattner8d1a81d2003-10-18 05:51:36 +0000[diff] [blame]3485<h5>Example:</h5>
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]3486
3487<p>See the <a href="#int_varargs">variable argument processing</a> section.</p>
3488
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3489</div>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]3490
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]3491<!-- *********************************************************************** -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3492<div class="doc_section"> <a name="intrinsics">Intrinsic Functions</a> </div>
3493<!-- *********************************************************************** -->
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]3494
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3495<div class="doc_text">
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]3496
3497<p>LLVM supports the notion of an "intrinsic function". These functions have
John Criswellfc6b8952005-05-16 16:17:45 +0000[diff] [blame]3498well known names and semantics and are required to follow certain
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]3499restrictions. Overall, these instructions represent an extension mechanism for
3500the LLVM language that does not require changing all of the transformations in
3501LLVM to add to the language (or the bytecode reader/writer, the parser,
3502etc...).</p>
3503
John Criswellfc6b8952005-05-16 16:17:45 +0000[diff] [blame]3504<p>Intrinsic function names must all start with an "<tt>llvm.</tt>" prefix. This
3505prefix is reserved in LLVM for intrinsic names; thus, functions may not be named
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]3506this. Intrinsic functions must always be external functions: you cannot define
3507the body of intrinsic functions. Intrinsic functions may only be used in call
3508or invoke instructions: it is illegal to take the address of an intrinsic
3509function. Additionally, because intrinsic functions are part of the LLVM
3510language, it is required that they all be documented here if any are added.</p>
3511
3512
John Criswellfc6b8952005-05-16 16:17:45 +0000[diff] [blame]3513<p>To learn how to add an intrinsic function, please see the <a
Chris Lattner590cff32005-05-11 03:35:57 +0000[diff] [blame]3514href="ExtendingLLVM.html">Extending LLVM Guide</a>.
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]3515</p>
3516
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3517</div>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]3518
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]3519<!-- ======================================================================= -->
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]3520<div class="doc_subsection">
3521 <a name="int_varargs">Variable Argument Handling Intrinsics</a>
3522</div>
3523
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3524<div class="doc_text">
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]3525
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3526<p>Variable argument support is defined in LLVM with the <a
Chris Lattnerfb6977d2006-01-13 23:26:01 +0000[diff] [blame]3527 href="#i_va_arg"><tt>va_arg</tt></a> instruction and these three
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3528intrinsic functions. These functions are related to the similarly
3529named macros defined in the <tt>&lt;stdarg.h&gt;</tt> header file.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]3530
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3531<p>All of these functions operate on arguments that use a
3532target-specific value type "<tt>va_list</tt>". The LLVM assembly
3533language reference manual does not define what this type is, so all
3534transformations should be prepared to handle intrinsics with any type
3535used.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]3536
Chris Lattner374ab302006-05-15 17:26:46 +0000[diff] [blame]3537<p>This example shows how the <a href="#i_va_arg"><tt>va_arg</tt></a>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3538instruction and the variable argument handling intrinsic functions are
3539used.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]3540
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]3541<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]3542define i32 %test(i32 %X, ...) {
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]3543 ; Initialize variable argument processing
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]3544 %ap = alloca i8 *
3545 call void %<a href="#i_va_start">llvm.va_start</a>(i8 ** %ap)
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]3546
3547 ; Read a single integer argument
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]3548 %tmp = va_arg i8 ** %ap, i32
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]3549
3550 ; Demonstrate usage of llvm.va_copy and llvm.va_end
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]3551 %aq = alloca i8 *
3552 call void %<a href="#i_va_copy">llvm.va_copy</a>(i8 ** %aq, i8 ** %ap)
3553 call void %<a href="#i_va_end">llvm.va_end</a>(i8 ** %aq)
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]3554
3555 ; Stop processing of arguments.
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]3556 call void %<a href="#i_va_end">llvm.va_end</a>(i8 ** %ap)
3557 ret i32 %tmp
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]3558}
3559</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3560</div>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]3561
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]3562<!-- _______________________________________________________________________ -->
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]3563<div class="doc_subsubsection">
3564 <a name="i_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a>
3565</div>
3566
3567
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3568<div class="doc_text">
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]3569<h5>Syntax:</h5>
Andrew Lenharth8bf607a2005-06-18 18:28:17 +0000[diff] [blame]3570<pre> declare void %llvm.va_start(&lt;va_list&gt;* &lt;arglist&gt;)<br></pre>
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]3571<h5>Overview:</h5>
Andrew Lenharth8bf607a2005-06-18 18:28:17 +0000[diff] [blame]3572<P>The '<tt>llvm.va_start</tt>' intrinsic initializes
3573<tt>*&lt;arglist&gt;</tt> for subsequent use by <tt><a
3574href="#i_va_arg">va_arg</a></tt>.</p>
3575
3576<h5>Arguments:</h5>
3577
3578<P>The argument is a pointer to a <tt>va_list</tt> element to initialize.</p>
3579
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]3580<h5>Semantics:</h5>
Andrew Lenharth8bf607a2005-06-18 18:28:17 +0000[diff] [blame]3581
3582<P>The '<tt>llvm.va_start</tt>' intrinsic works just like the <tt>va_start</tt>
3583macro available in C. In a target-dependent way, it initializes the
3584<tt>va_list</tt> element the argument points to, so that the next call to
3585<tt>va_arg</tt> will produce the first variable argument passed to the function.
3586Unlike the C <tt>va_start</tt> macro, this intrinsic does not need to know the
3587last argument of the function, the compiler can figure that out.</p>
3588
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3589</div>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]3590
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]3591<!-- _______________________________________________________________________ -->
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]3592<div class="doc_subsubsection">
3593 <a name="i_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a>
3594</div>
3595
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3596<div class="doc_text">
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]3597<h5>Syntax:</h5>
Andrew Lenharth8bf607a2005-06-18 18:28:17 +0000[diff] [blame]3598<pre> declare void %llvm.va_end(&lt;va_list*&gt; &lt;arglist&gt;)<br></pre>
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]3599<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3600<p>The '<tt>llvm.va_end</tt>' intrinsic destroys <tt>&lt;arglist&gt;</tt>
3601which has been initialized previously with <tt><a href="#i_va_start">llvm.va_start</a></tt>
3602or <tt><a href="#i_va_copy">llvm.va_copy</a></tt>.</p>
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]3603<h5>Arguments:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3604<p>The argument is a <tt>va_list</tt> to destroy.</p>
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]3605<h5>Semantics:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3606<p>The '<tt>llvm.va_end</tt>' intrinsic works just like the <tt>va_end</tt>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3607macro available in C. In a target-dependent way, it destroys the <tt>va_list</tt>.
3608Calls to <a href="#i_va_start"><tt>llvm.va_start</tt></a> and <a
3609 href="#i_va_copy"><tt>llvm.va_copy</tt></a> must be matched exactly
3610with calls to <tt>llvm.va_end</tt>.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3611</div>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]3612
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]3613<!-- _______________________________________________________________________ -->
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]3614<div class="doc_subsubsection">
3615 <a name="i_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a>
3616</div>
3617
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3618<div class="doc_text">
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]3619
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]3620<h5>Syntax:</h5>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]3621
3622<pre>
Andrew Lenharth8bf607a2005-06-18 18:28:17 +0000[diff] [blame]3623 declare void %llvm.va_copy(&lt;va_list&gt;* &lt;destarglist&gt;,
Andrew Lenharthd0a4c622005-06-22 20:38:11 +0000[diff] [blame]3624 &lt;va_list&gt;* &lt;srcarglist&gt;)
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]3625</pre>
3626
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]3627<h5>Overview:</h5>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]3628
Andrew Lenharth8bf607a2005-06-18 18:28:17 +0000[diff] [blame]3629<p>The '<tt>llvm.va_copy</tt>' intrinsic copies the current argument position from
3630the source argument list to the destination argument list.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]3631
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]3632<h5>Arguments:</h5>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]3633
Andrew Lenharth8bf607a2005-06-18 18:28:17 +0000[diff] [blame]3634<p>The first argument is a pointer to a <tt>va_list</tt> element to initialize.
Andrew Lenharthd0a4c622005-06-22 20:38:11 +0000[diff] [blame]3635The second argument is a pointer to a <tt>va_list</tt> element to copy from.</p>
Andrew Lenharth8bf607a2005-06-18 18:28:17 +0000[diff] [blame]3636
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]3637
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]3638<h5>Semantics:</h5>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]3639
Andrew Lenharth8bf607a2005-06-18 18:28:17 +0000[diff] [blame]3640<p>The '<tt>llvm.va_copy</tt>' intrinsic works just like the <tt>va_copy</tt> macro
3641available in C. In a target-dependent way, it copies the source
3642<tt>va_list</tt> element into the destination list. This intrinsic is necessary
3643because the <tt><a href="i_va_begin">llvm.va_begin</a></tt> intrinsic may be
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]3644arbitrarily complex and require memory allocation, for example.</p>
3645
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3646</div>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]3647
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]3648<!-- ======================================================================= -->
3649<div class="doc_subsection">
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]3650 <a name="int_gc">Accurate Garbage Collection Intrinsics</a>
3651</div>
3652
3653<div class="doc_text">
3654
3655<p>
3656LLVM support for <a href="GarbageCollection.html">Accurate Garbage
3657Collection</a> requires the implementation and generation of these intrinsics.
3658These intrinsics allow identification of <a href="#i_gcroot">GC roots on the
3659stack</a>, as well as garbage collector implementations that require <a
3660href="#i_gcread">read</a> and <a href="#i_gcwrite">write</a> barriers.
3661Front-ends for type-safe garbage collected languages should generate these
3662intrinsics to make use of the LLVM garbage collectors. For more details, see <a
3663href="GarbageCollection.html">Accurate Garbage Collection with LLVM</a>.
3664</p>
3665</div>
3666
3667<!-- _______________________________________________________________________ -->
3668<div class="doc_subsubsection">
3669 <a name="i_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a>
3670</div>
3671
3672<div class="doc_text">
3673
3674<h5>Syntax:</h5>
3675
3676<pre>
Reid Spencera8d451e2005-04-26 20:50:44 +0000[diff] [blame]3677 declare void %llvm.gcroot(&lt;ty&gt;** %ptrloc, &lt;ty2&gt;* %metadata)
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]3678</pre>
3679
3680<h5>Overview:</h5>
3681
John Criswell9e2485c2004-12-10 15:51:16 +0000[diff] [blame]3682<p>The '<tt>llvm.gcroot</tt>' intrinsic declares the existence of a GC root to
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]3683the code generator, and allows some metadata to be associated with it.</p>
3684
3685<h5>Arguments:</h5>
3686
3687<p>The first argument specifies the address of a stack object that contains the
3688root pointer. The second pointer (which must be either a constant or a global
3689value address) contains the meta-data to be associated with the root.</p>
3690
3691<h5>Semantics:</h5>
3692
3693<p>At runtime, a call to this intrinsics stores a null pointer into the "ptrloc"
3694location. At compile-time, the code generator generates information to allow
3695the runtime to find the pointer at GC safe points.
3696</p>
3697
3698</div>
3699
3700
3701<!-- _______________________________________________________________________ -->
3702<div class="doc_subsubsection">
3703 <a name="i_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a>
3704</div>
3705
3706<div class="doc_text">
3707
3708<h5>Syntax:</h5>
3709
3710<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]3711 declare i8 * %llvm.gcread(i8 * %ObjPtr, i8 ** %Ptr)
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]3712</pre>
3713
3714<h5>Overview:</h5>
3715
3716<p>The '<tt>llvm.gcread</tt>' intrinsic identifies reads of references from heap
3717locations, allowing garbage collector implementations that require read
3718barriers.</p>
3719
3720<h5>Arguments:</h5>
3721
Chris Lattner80626e92006-03-14 20:02:51 +0000[diff] [blame]3722<p>The second argument is the address to read from, which should be an address
3723allocated from the garbage collector. The first object is a pointer to the
3724start of the referenced object, if needed by the language runtime (otherwise
3725null).</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]3726
3727<h5>Semantics:</h5>
3728
3729<p>The '<tt>llvm.gcread</tt>' intrinsic has the same semantics as a load
3730instruction, but may be replaced with substantially more complex code by the
3731garbage collector runtime, as needed.</p>
3732
3733</div>
3734
3735
3736<!-- _______________________________________________________________________ -->
3737<div class="doc_subsubsection">
3738 <a name="i_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a>
3739</div>
3740
3741<div class="doc_text">
3742
3743<h5>Syntax:</h5>
3744
3745<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]3746 declare void %llvm.gcwrite(i8 * %P1, i8 * %Obj, i8 ** %P2)
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]3747</pre>
3748
3749<h5>Overview:</h5>
3750
3751<p>The '<tt>llvm.gcwrite</tt>' intrinsic identifies writes of references to heap
3752locations, allowing garbage collector implementations that require write
3753barriers (such as generational or reference counting collectors).</p>
3754
3755<h5>Arguments:</h5>
3756
Chris Lattner80626e92006-03-14 20:02:51 +0000[diff] [blame]3757<p>The first argument is the reference to store, the second is the start of the
3758object to store it to, and the third is the address of the field of Obj to
3759store to. If the runtime does not require a pointer to the object, Obj may be
3760null.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]3761
3762<h5>Semantics:</h5>
3763
3764<p>The '<tt>llvm.gcwrite</tt>' intrinsic has the same semantics as a store
3765instruction, but may be replaced with substantially more complex code by the
3766garbage collector runtime, as needed.</p>
3767
3768</div>
3769
3770
3771
3772<!-- ======================================================================= -->
3773<div class="doc_subsection">
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]3774 <a name="int_codegen">Code Generator Intrinsics</a>
3775</div>
3776
3777<div class="doc_text">
3778<p>
3779These intrinsics are provided by LLVM to expose special features that may only
3780be implemented with code generator support.
3781</p>
3782
3783</div>
3784
3785<!-- _______________________________________________________________________ -->
3786<div class="doc_subsubsection">
3787 <a name="i_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a>
3788</div>
3789
3790<div class="doc_text">
3791
3792<h5>Syntax:</h5>
3793<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]3794 declare i8 *%llvm.returnaddress(i32 &lt;level&gt;)
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]3795</pre>
3796
3797<h5>Overview:</h5>
3798
3799<p>
Chris Lattner32b5d712006-10-15 20:05:59 +0000[diff] [blame]3800The '<tt>llvm.returnaddress</tt>' intrinsic attempts to compute a
3801target-specific value indicating the return address of the current function
3802or one of its callers.
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]3803</p>
3804
3805<h5>Arguments:</h5>
3806
3807<p>
3808The argument to this intrinsic indicates which function to return the address
3809for. Zero indicates the calling function, one indicates its caller, etc. The
3810argument is <b>required</b> to be a constant integer value.
3811</p>
3812
3813<h5>Semantics:</h5>
3814
3815<p>
3816The '<tt>llvm.returnaddress</tt>' intrinsic either returns a pointer indicating
3817the return address of the specified call frame, or zero if it cannot be
3818identified. The value returned by this intrinsic is likely to be incorrect or 0
3819for arguments other than zero, so it should only be used for debugging purposes.
3820</p>
3821
3822<p>
3823Note that calling this intrinsic does not prevent function inlining or other
Chris Lattnerb40bb382005-03-07 20:30:51 +0000[diff] [blame]3824aggressive transformations, so the value returned may not be that of the obvious
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]3825source-language caller.
3826</p>
3827</div>
3828
3829
3830<!-- _______________________________________________________________________ -->
3831<div class="doc_subsubsection">
3832 <a name="i_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a>
3833</div>
3834
3835<div class="doc_text">
3836
3837<h5>Syntax:</h5>
3838<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]3839 declare i8 *%llvm.frameaddress(i32 &lt;level&gt;)
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]3840</pre>
3841
3842<h5>Overview:</h5>
3843
3844<p>
Chris Lattner32b5d712006-10-15 20:05:59 +0000[diff] [blame]3845The '<tt>llvm.frameaddress</tt>' intrinsic attempts to return the
3846target-specific frame pointer value for the specified stack frame.
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]3847</p>
3848
3849<h5>Arguments:</h5>
3850
3851<p>
3852The argument to this intrinsic indicates which function to return the frame
3853pointer for. Zero indicates the calling function, one indicates its caller,
3854etc. The argument is <b>required</b> to be a constant integer value.
3855</p>
3856
3857<h5>Semantics:</h5>
3858
3859<p>
3860The '<tt>llvm.frameaddress</tt>' intrinsic either returns a pointer indicating
3861the frame address of the specified call frame, or zero if it cannot be
3862identified. The value returned by this intrinsic is likely to be incorrect or 0
3863for arguments other than zero, so it should only be used for debugging purposes.
3864</p>
3865
3866<p>
3867Note that calling this intrinsic does not prevent function inlining or other
Chris Lattnerb40bb382005-03-07 20:30:51 +0000[diff] [blame]3868aggressive transformations, so the value returned may not be that of the obvious
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]3869source-language caller.
3870</p>
3871</div>
3872
Chris Lattner9a9d7ac2005-02-28 19:24:19 +0000[diff] [blame]3873<!-- _______________________________________________________________________ -->
3874<div class="doc_subsubsection">
Chris Lattner57e1f392006-01-13 02:03:13 +0000[diff] [blame]3875 <a name="i_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a>
3876</div>
3877
3878<div class="doc_text">
3879
3880<h5>Syntax:</h5>
3881<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]3882 declare i8 *%llvm.stacksave()
Chris Lattner57e1f392006-01-13 02:03:13 +0000[diff] [blame]3883</pre>
3884
3885<h5>Overview:</h5>
3886
3887<p>
3888The '<tt>llvm.stacksave</tt>' intrinsic is used to remember the current state of
3889the function stack, for use with <a href="#i_stackrestore">
3890<tt>llvm.stackrestore</tt></a>. This is useful for implementing language
3891features like scoped automatic variable sized arrays in C99.
3892</p>
3893
3894<h5>Semantics:</h5>
3895
3896<p>
3897This intrinsic returns a opaque pointer value that can be passed to <a
3898href="#i_stackrestore"><tt>llvm.stackrestore</tt></a>. When an
3899<tt>llvm.stackrestore</tt> intrinsic is executed with a value saved from
3900<tt>llvm.stacksave</tt>, it effectively restores the state of the stack to the
3901state it was in when the <tt>llvm.stacksave</tt> intrinsic executed. In
3902practice, this pops any <a href="#i_alloca">alloca</a> blocks from the stack
3903that were allocated after the <tt>llvm.stacksave</tt> was executed.
3904</p>
3905
3906</div>
3907
3908<!-- _______________________________________________________________________ -->
3909<div class="doc_subsubsection">
3910 <a name="i_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a>
3911</div>
3912
3913<div class="doc_text">
3914
3915<h5>Syntax:</h5>
3916<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]3917 declare void %llvm.stackrestore(i8 * %ptr)
Chris Lattner57e1f392006-01-13 02:03:13 +0000[diff] [blame]3918</pre>
3919
3920<h5>Overview:</h5>
3921
3922<p>
3923The '<tt>llvm.stackrestore</tt>' intrinsic is used to restore the state of
3924the function stack to the state it was in when the corresponding <a
3925href="#llvm.stacksave"><tt>llvm.stacksave</tt></a> intrinsic executed. This is
3926useful for implementing language features like scoped automatic variable sized
3927arrays in C99.
3928</p>
3929
3930<h5>Semantics:</h5>
3931
3932<p>
3933See the description for <a href="#i_stacksave"><tt>llvm.stacksave</tt></a>.
3934</p>
3935
3936</div>
3937
3938
3939<!-- _______________________________________________________________________ -->
3940<div class="doc_subsubsection">
Chris Lattner9a9d7ac2005-02-28 19:24:19 +0000[diff] [blame]3941 <a name="i_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a>
3942</div>
3943
3944<div class="doc_text">
3945
3946<h5>Syntax:</h5>
3947<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]3948 declare void %llvm.prefetch(i8 * &lt;address&gt;,
3949 i32 &lt;rw&gt;, i32 &lt;locality&gt;)
Chris Lattner9a9d7ac2005-02-28 19:24:19 +0000[diff] [blame]3950</pre>
3951
3952<h5>Overview:</h5>
3953
3954
3955<p>
3956The '<tt>llvm.prefetch</tt>' intrinsic is a hint to the code generator to insert
John Criswellfc6b8952005-05-16 16:17:45 +0000[diff] [blame]3957a prefetch instruction if supported; otherwise, it is a noop. Prefetches have
3958no
3959effect on the behavior of the program but can change its performance
Chris Lattner2a615362005-02-28 19:47:14 +0000[diff] [blame]3960characteristics.
Chris Lattner9a9d7ac2005-02-28 19:24:19 +0000[diff] [blame]3961</p>
3962
3963<h5>Arguments:</h5>
3964
3965<p>
3966<tt>address</tt> is the address to be prefetched, <tt>rw</tt> is the specifier
3967determining if the fetch should be for a read (0) or write (1), and
3968<tt>locality</tt> is a temporal locality specifier ranging from (0) - no
Chris Lattneraeffb4a2005-03-07 20:31:38 +0000[diff] [blame]3969locality, to (3) - extremely local keep in cache. The <tt>rw</tt> and
Chris Lattner9a9d7ac2005-02-28 19:24:19 +0000[diff] [blame]3970<tt>locality</tt> arguments must be constant integers.
3971</p>
3972
3973<h5>Semantics:</h5>
3974
3975<p>
3976This intrinsic does not modify the behavior of the program. In particular,
3977prefetches cannot trap and do not produce a value. On targets that support this
3978intrinsic, the prefetch can provide hints to the processor cache for better
3979performance.
3980</p>
3981
3982</div>
3983
Andrew Lenharth7f4ec3b2005-03-28 20:05:49 +0000[diff] [blame]3984<!-- _______________________________________________________________________ -->
3985<div class="doc_subsubsection">
3986 <a name="i_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a>
3987</div>
3988
3989<div class="doc_text">
3990
3991<h5>Syntax:</h5>
3992<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]3993 declare void %llvm.pcmarker( i32 &lt;id&gt; )
Andrew Lenharth7f4ec3b2005-03-28 20:05:49 +0000[diff] [blame]3994</pre>
3995
3996<h5>Overview:</h5>
3997
3998
3999<p>
John Criswellfc6b8952005-05-16 16:17:45 +0000[diff] [blame]4000The '<tt>llvm.pcmarker</tt>' intrinsic is a method to export a Program Counter
4001(PC) in a region of
Andrew Lenharth7f4ec3b2005-03-28 20:05:49 +0000[diff] [blame]4002code to simulators and other tools. The method is target specific, but it is
4003expected that the marker will use exported symbols to transmit the PC of the marker.
Jeff Cohen25d4f7e2005-11-11 02:15:27 +0000[diff] [blame]4004The marker makes no guarantees that it will remain with any specific instruction
Chris Lattnerd07c3f42005-11-15 06:07:55 +0000[diff] [blame]4005after optimizations. It is possible that the presence of a marker will inhibit
Chris Lattnerb3e7afd2006-03-24 07:16:10 +0000[diff] [blame]4006optimizations. The intended use is to be inserted after optimizations to allow
John Criswellfc6b8952005-05-16 16:17:45 +0000[diff] [blame]4007correlations of simulation runs.
Andrew Lenharth7f4ec3b2005-03-28 20:05:49 +0000[diff] [blame]4008</p>
4009
4010<h5>Arguments:</h5>
4011
4012<p>
4013<tt>id</tt> is a numerical id identifying the marker.
4014</p>
4015
4016<h5>Semantics:</h5>
4017
4018<p>
4019This intrinsic does not modify the behavior of the program. Backends that do not
4020support this intrinisic may ignore it.
4021</p>
4022
4023</div>
4024
Andrew Lenharth51b8d542005-11-11 16:47:30 +0000[diff] [blame]4025<!-- _______________________________________________________________________ -->
4026<div class="doc_subsubsection">
4027 <a name="i_readcyclecounter">'<tt>llvm.readcyclecounter</tt>' Intrinsic</a>
4028</div>
4029
4030<div class="doc_text">
4031
4032<h5>Syntax:</h5>
4033<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]4034 declare i64 %llvm.readcyclecounter( )
Andrew Lenharth51b8d542005-11-11 16:47:30 +0000[diff] [blame]4035</pre>
4036
4037<h5>Overview:</h5>
4038
4039
4040<p>
4041The '<tt>llvm.readcyclecounter</tt>' intrinsic provides access to the cycle
4042counter register (or similar low latency, high accuracy clocks) on those targets
4043that support it. On X86, it should map to RDTSC. On Alpha, it should map to RPCC.
4044As the backing counters overflow quickly (on the order of 9 seconds on alpha), this
4045should only be used for small timings.
4046</p>
4047
4048<h5>Semantics:</h5>
4049
4050<p>
4051When directly supported, reading the cycle counter should not modify any memory.
4052Implementations are allowed to either return a application specific value or a
4053system wide value. On backends without support, this is lowered to a constant 0.
4054</p>
4055
4056</div>
4057
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]4058<!-- ======================================================================= -->
4059<div class="doc_subsection">
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]4060 <a name="int_libc">Standard C Library Intrinsics</a>
4061</div>
4062
4063<div class="doc_text">
4064<p>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]4065LLVM provides intrinsics for a few important standard C library functions.
4066These intrinsics allow source-language front-ends to pass information about the
4067alignment of the pointer arguments to the code generator, providing opportunity
4068for more efficient code generation.
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]4069</p>
4070
4071</div>
4072
4073<!-- _______________________________________________________________________ -->
4074<div class="doc_subsubsection">
4075 <a name="i_memcpy">'<tt>llvm.memcpy</tt>' Intrinsic</a>
4076</div>
4077
4078<div class="doc_text">
4079
4080<h5>Syntax:</h5>
4081<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]4082 declare void %llvm.memcpy.i32(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
4083 i32 &lt;len&gt;, i32 &lt;align&gt;)
4084 declare void %llvm.memcpy.i64(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
4085 i64 &lt;len&gt;, i32 &lt;align&gt;)
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]4086</pre>
4087
4088<h5>Overview:</h5>
4089
4090<p>
Chris Lattner5b310c32006-03-03 00:07:20 +0000[diff] [blame]4091The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the source
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]4092location to the destination location.
4093</p>
4094
4095<p>
Chris Lattner5b310c32006-03-03 00:07:20 +0000[diff] [blame]4096Note that, unlike the standard libc function, the <tt>llvm.memcpy.*</tt>
4097intrinsics do not return a value, and takes an extra alignment argument.
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]4098</p>
4099
4100<h5>Arguments:</h5>
4101
4102<p>
4103The first argument is a pointer to the destination, the second is a pointer to
Chris Lattner5b310c32006-03-03 00:07:20 +0000[diff] [blame]4104the source. The third argument is an integer argument
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]4105specifying the number of bytes to copy, and the fourth argument is the alignment
4106of the source and destination locations.
4107</p>
4108
Chris Lattner3301ced2004-02-12 21:18:15 +0000[diff] [blame]4109<p>
4110If the call to this intrinisic has an alignment value that is not 0 or 1, then
Chris Lattnerf0afc2c2006-03-04 00:02:10 +0000[diff] [blame]4111the caller guarantees that both the source and destination pointers are aligned
4112to that boundary.
Chris Lattner3301ced2004-02-12 21:18:15 +0000[diff] [blame]4113</p>
4114
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]4115<h5>Semantics:</h5>
4116
4117<p>
Chris Lattner5b310c32006-03-03 00:07:20 +0000[diff] [blame]4118The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the source
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]4119location to the destination location, which are not allowed to overlap. It
4120copies "len" bytes of memory over. If the argument is known to be aligned to
4121some boundary, this can be specified as the fourth argument, otherwise it should
4122be set to 0 or 1.
4123</p>
4124</div>
4125
4126
Chris Lattner0eb51b42004-02-12 18:10:10 +0000[diff] [blame]4127<!-- _______________________________________________________________________ -->
4128<div class="doc_subsubsection">
4129 <a name="i_memmove">'<tt>llvm.memmove</tt>' Intrinsic</a>
4130</div>
4131
4132<div class="doc_text">
4133
4134<h5>Syntax:</h5>
4135<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]4136 declare void %llvm.memmove.i32(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
4137 i32 &lt;len&gt;, i32 &lt;align&gt;)
4138 declare void %llvm.memmove.i64(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
4139 i64 &lt;len&gt;, i32 &lt;align&gt;)
Chris Lattner0eb51b42004-02-12 18:10:10 +0000[diff] [blame]4140</pre>
4141
4142<h5>Overview:</h5>
4143
4144<p>
Chris Lattner5b310c32006-03-03 00:07:20 +0000[diff] [blame]4145The '<tt>llvm.memmove.*</tt>' intrinsics move a block of memory from the source
4146location to the destination location. It is similar to the
4147'<tt>llvm.memcmp</tt>' intrinsic but allows the two memory locations to overlap.
Chris Lattner0eb51b42004-02-12 18:10:10 +0000[diff] [blame]4148</p>
4149
4150<p>
Chris Lattner5b310c32006-03-03 00:07:20 +0000[diff] [blame]4151Note that, unlike the standard libc function, the <tt>llvm.memmove.*</tt>
4152intrinsics do not return a value, and takes an extra alignment argument.
Chris Lattner0eb51b42004-02-12 18:10:10 +0000[diff] [blame]4153</p>
4154
4155<h5>Arguments:</h5>
4156
4157<p>
4158The first argument is a pointer to the destination, the second is a pointer to
Chris Lattner5b310c32006-03-03 00:07:20 +0000[diff] [blame]4159the source. The third argument is an integer argument
Chris Lattner0eb51b42004-02-12 18:10:10 +0000[diff] [blame]4160specifying the number of bytes to copy, and the fourth argument is the alignment
4161of the source and destination locations.
4162</p>
4163
Chris Lattner3301ced2004-02-12 21:18:15 +0000[diff] [blame]4164<p>
4165If the call to this intrinisic has an alignment value that is not 0 or 1, then
Chris Lattnerf0afc2c2006-03-04 00:02:10 +0000[diff] [blame]4166the caller guarantees that the source and destination pointers are aligned to
4167that boundary.
Chris Lattner3301ced2004-02-12 21:18:15 +0000[diff] [blame]4168</p>
4169
Chris Lattner0eb51b42004-02-12 18:10:10 +0000[diff] [blame]4170<h5>Semantics:</h5>
4171
4172<p>
Chris Lattner5b310c32006-03-03 00:07:20 +0000[diff] [blame]4173The '<tt>llvm.memmove.*</tt>' intrinsics copy a block of memory from the source
Chris Lattner0eb51b42004-02-12 18:10:10 +0000[diff] [blame]4174location to the destination location, which may overlap. It
4175copies "len" bytes of memory over. If the argument is known to be aligned to
4176some boundary, this can be specified as the fourth argument, otherwise it should
4177be set to 0 or 1.
4178</p>
4179</div>
4180
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]4181
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]4182<!-- _______________________________________________________________________ -->
4183<div class="doc_subsubsection">
Chris Lattner5b310c32006-03-03 00:07:20 +0000[diff] [blame]4184 <a name="i_memset">'<tt>llvm.memset.*</tt>' Intrinsics</a>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]4185</div>
4186
4187<div class="doc_text">
4188
4189<h5>Syntax:</h5>
4190<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]4191 declare void %llvm.memset.i32(i8 * &lt;dest&gt;, i8 &lt;val&gt;,
4192 i32 &lt;len&gt;, i32 &lt;align&gt;)
4193 declare void %llvm.memset.i64(i8 * &lt;dest&gt;, i8 &lt;val&gt;,
4194 i64 &lt;len&gt;, i32 &lt;align&gt;)
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]4195</pre>
4196
4197<h5>Overview:</h5>
4198
4199<p>
Chris Lattner5b310c32006-03-03 00:07:20 +0000[diff] [blame]4200The '<tt>llvm.memset.*</tt>' intrinsics fill a block of memory with a particular
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]4201byte value.
4202</p>
4203
4204<p>
4205Note that, unlike the standard libc function, the <tt>llvm.memset</tt> intrinsic
4206does not return a value, and takes an extra alignment argument.
4207</p>
4208
4209<h5>Arguments:</h5>
4210
4211<p>
4212The first argument is a pointer to the destination to fill, the second is the
Chris Lattner5b310c32006-03-03 00:07:20 +0000[diff] [blame]4213byte value to fill it with, the third argument is an integer
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]4214argument specifying the number of bytes to fill, and the fourth argument is the
4215known alignment of destination location.
4216</p>
4217
4218<p>
4219If the call to this intrinisic has an alignment value that is not 0 or 1, then
Chris Lattnerf0afc2c2006-03-04 00:02:10 +0000[diff] [blame]4220the caller guarantees that the destination pointer is aligned to that boundary.
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]4221</p>
4222
4223<h5>Semantics:</h5>
4224
4225<p>
Chris Lattner5b310c32006-03-03 00:07:20 +0000[diff] [blame]4226The '<tt>llvm.memset.*</tt>' intrinsics fill "len" bytes of memory starting at
4227the
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]4228destination location. If the argument is known to be aligned to some boundary,
4229this can be specified as the fourth argument, otherwise it should be set to 0 or
42301.
4231</p>
4232</div>
4233
4234
Chris Lattner32006282004-06-11 02:28:03 +0000[diff] [blame]4235<!-- _______________________________________________________________________ -->
4236<div class="doc_subsubsection">
Reid Spencer0b118202006-01-16 21:12:35 +0000[diff] [blame]4237 <a name="i_isunordered">'<tt>llvm.isunordered.*</tt>' Intrinsic</a>
Alkis Evlogimenos26bbe932004-06-13 01:16:15 +0000[diff] [blame]4238</div>
4239
4240<div class="doc_text">
4241
4242<h5>Syntax:</h5>
4243<pre>
Reid Spencer0b118202006-01-16 21:12:35 +0000[diff] [blame]4244 declare bool %llvm.isunordered.f32(float Val1, float Val2)
4245 declare bool %llvm.isunordered.f64(double Val1, double Val2)
Alkis Evlogimenos26bbe932004-06-13 01:16:15 +0000[diff] [blame]4246</pre>
4247
4248<h5>Overview:</h5>
4249
4250<p>
Reid Spencer0b118202006-01-16 21:12:35 +0000[diff] [blame]4251The '<tt>llvm.isunordered</tt>' intrinsics return true if either or both of the
Alkis Evlogimenos26bbe932004-06-13 01:16:15 +0000[diff] [blame]4252specified floating point values is a NAN.
4253</p>
4254
4255<h5>Arguments:</h5>
4256
4257<p>
4258The arguments are floating point numbers of the same type.
4259</p>
4260
4261<h5>Semantics:</h5>
4262
4263<p>
4264If either or both of the arguments is a SNAN or QNAN, it returns true, otherwise
4265false.
4266</p>
4267</div>
4268
4269
Chris Lattnera4d74142005-07-21 01:29:16 +0000[diff] [blame]4270<!-- _______________________________________________________________________ -->
4271<div class="doc_subsubsection">
Chris Lattnerec6cb612006-01-16 22:38:59 +0000[diff] [blame]4272 <a name="i_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a>
Chris Lattnera4d74142005-07-21 01:29:16 +0000[diff] [blame]4273</div>
4274
4275<div class="doc_text">
4276
4277<h5>Syntax:</h5>
4278<pre>
Chris Lattnerf4d252d2006-09-08 06:34:02 +0000[diff] [blame]4279 declare float %llvm.sqrt.f32(float %Val)
4280 declare double %llvm.sqrt.f64(double %Val)
Chris Lattnera4d74142005-07-21 01:29:16 +0000[diff] [blame]4281</pre>
4282
4283<h5>Overview:</h5>
4284
4285<p>
Reid Spencer0b118202006-01-16 21:12:35 +0000[diff] [blame]4286The '<tt>llvm.sqrt</tt>' intrinsics return the sqrt of the specified operand,
Chris Lattnera4d74142005-07-21 01:29:16 +0000[diff] [blame]4287returning the same value as the libm '<tt>sqrt</tt>' function would. Unlike
4288<tt>sqrt</tt> in libm, however, <tt>llvm.sqrt</tt> has undefined behavior for
4289negative numbers (which allows for better optimization).
4290</p>
4291
4292<h5>Arguments:</h5>
4293
4294<p>
4295The argument and return value are floating point numbers of the same type.
4296</p>
4297
4298<h5>Semantics:</h5>
4299
4300<p>
4301This function returns the sqrt of the specified operand if it is a positive
4302floating point number.
4303</p>
4304</div>
4305
Chris Lattnerf4d252d2006-09-08 06:34:02 +0000[diff] [blame]4306<!-- _______________________________________________________________________ -->
4307<div class="doc_subsubsection">
4308 <a name="i_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a>
4309</div>
4310
4311<div class="doc_text">
4312
4313<h5>Syntax:</h5>
4314<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]4315 declare float %llvm.powi.f32(float %Val, i32 %power)
4316 declare double %llvm.powi.f64(double %Val, i32 %power)
Chris Lattnerf4d252d2006-09-08 06:34:02 +0000[diff] [blame]4317</pre>
4318
4319<h5>Overview:</h5>
4320
4321<p>
4322The '<tt>llvm.powi.*</tt>' intrinsics return the first operand raised to the
4323specified (positive or negative) power. The order of evaluation of
4324multiplications is not defined.
4325</p>
4326
4327<h5>Arguments:</h5>
4328
4329<p>
4330The second argument is an integer power, and the first is a value to raise to
4331that power.
4332</p>
4333
4334<h5>Semantics:</h5>
4335
4336<p>
4337This function returns the first value raised to the second power with an
4338unspecified sequence of rounding operations.</p>
4339</div>
4340
4341
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]4342<!-- ======================================================================= -->
4343<div class="doc_subsection">
Nate Begeman7e36c472006-01-13 23:26:38 +0000[diff] [blame]4344 <a name="int_manip">Bit Manipulation Intrinsics</a>
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]4345</div>
4346
4347<div class="doc_text">
4348<p>
Nate Begeman7e36c472006-01-13 23:26:38 +0000[diff] [blame]4349LLVM provides intrinsics for a few important bit manipulation operations.
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]4350These allow efficient code generation for some algorithms.
4351</p>
4352
4353</div>
4354
4355<!-- _______________________________________________________________________ -->
4356<div class="doc_subsubsection">
Nate Begeman7e36c472006-01-13 23:26:38 +0000[diff] [blame]4357 <a name="i_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a>
4358</div>
4359
4360<div class="doc_text">
4361
4362<h5>Syntax:</h5>
4363<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]4364 declare i16 %llvm.bswap.i16(i16 &lt;id&gt;)
4365 declare i32 %llvm.bswap.i32(i32 &lt;id&gt;)
4366 declare i64 %llvm.bswap.i64(i64 &lt;id&gt;)
Nate Begeman7e36c472006-01-13 23:26:38 +0000[diff] [blame]4367</pre>
4368
4369<h5>Overview:</h5>
4370
4371<p>
4372The '<tt>llvm.bwsap</tt>' family of intrinsics is used to byteswap a 16, 32 or
437364 bit quantity. These are useful for performing operations on data that is not
4374in the target's native byte order.
4375</p>
4376
4377<h5>Semantics:</h5>
4378
4379<p>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]4380The <tt>llvm.bswap.16</tt> intrinsic returns an i16 value that has the high
4381and low byte of the input i16 swapped. Similarly, the <tt>llvm.bswap.i32</tt>
4382intrinsic returns an i32 value that has the four bytes of the input i32
4383swapped, so that if the input bytes are numbered 0, 1, 2, 3 then the returned
4384i32 will have its bytes in 3, 2, 1, 0 order. The <tt>llvm.bswap.i64</tt>
4385intrinsic extends this concept to 64 bits.
Nate Begeman7e36c472006-01-13 23:26:38 +0000[diff] [blame]4386</p>
4387
4388</div>
4389
4390<!-- _______________________________________________________________________ -->
4391<div class="doc_subsubsection">
Reid Spencer0b118202006-01-16 21:12:35 +0000[diff] [blame]4392 <a name="int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic</a>
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]4393</div>
4394
4395<div class="doc_text">
4396
4397<h5>Syntax:</h5>
4398<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]4399 declare i8 %llvm.ctpop.i8 (i8 &lt;src&gt;)
4400 declare i16 %llvm.ctpop.i16(i16 &lt;src&gt;)
4401 declare i32 %llvm.ctpop.i32(i32 &lt;src&gt;)
4402 declare i64 %llvm.ctpop.i64(i64 &lt;src&gt;)
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]4403</pre>
4404
4405<h5>Overview:</h5>
4406
4407<p>
Chris Lattnerec6cb612006-01-16 22:38:59 +0000[diff] [blame]4408The '<tt>llvm.ctpop</tt>' family of intrinsics counts the number of bits set in a
4409value.
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]4410</p>
4411
4412<h5>Arguments:</h5>
4413
4414<p>
Chris Lattnercfe6b372005-05-07 01:46:40 +0000[diff] [blame]4415The only argument is the value to be counted. The argument may be of any
Chris Lattnerec6cb612006-01-16 22:38:59 +0000[diff] [blame]4416unsigned integer type. The return type must match the argument type.
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]4417</p>
4418
4419<h5>Semantics:</h5>
4420
4421<p>
4422The '<tt>llvm.ctpop</tt>' intrinsic counts the 1's in a variable.
4423</p>
4424</div>
4425
4426<!-- _______________________________________________________________________ -->
4427<div class="doc_subsubsection">
Chris Lattner8a886be2006-01-16 22:34:14 +0000[diff] [blame]4428 <a name="int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic</a>
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]4429</div>
4430
4431<div class="doc_text">
4432
4433<h5>Syntax:</h5>
4434<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]4435 declare i8 %llvm.ctlz.i8 (i8 &lt;src&gt;)
4436 declare i16 %llvm.ctlz.i16(i16 &lt;src&gt;)
4437 declare i32 %llvm.ctlz.i32(i32 &lt;src&gt;)
4438 declare i64 %llvm.ctlz.i64(i64 &lt;src&gt;)
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]4439</pre>
4440
4441<h5>Overview:</h5>
4442
4443<p>
Reid Spencer0b118202006-01-16 21:12:35 +0000[diff] [blame]4444The '<tt>llvm.ctlz</tt>' family of intrinsic functions counts the number of
4445leading zeros in a variable.
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]4446</p>
4447
4448<h5>Arguments:</h5>
4449
4450<p>
Chris Lattnercfe6b372005-05-07 01:46:40 +0000[diff] [blame]4451The only argument is the value to be counted. The argument may be of any
Chris Lattnerec6cb612006-01-16 22:38:59 +0000[diff] [blame]4452unsigned integer type. The return type must match the argument type.
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]4453</p>
4454
4455<h5>Semantics:</h5>
4456
4457<p>
Chris Lattnereff29ab2005-05-15 19:39:26 +0000[diff] [blame]4458The '<tt>llvm.ctlz</tt>' intrinsic counts the leading (most significant) zeros
4459in a variable. If the src == 0 then the result is the size in bits of the type
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]4460of src. For example, <tt>llvm.ctlz(i32 2) = 30</tt>.
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]4461</p>
4462</div>
Chris Lattner32006282004-06-11 02:28:03 +0000[diff] [blame]4463
4464
Chris Lattnereff29ab2005-05-15 19:39:26 +0000[diff] [blame]4465
4466<!-- _______________________________________________________________________ -->
4467<div class="doc_subsubsection">
Chris Lattner8a886be2006-01-16 22:34:14 +0000[diff] [blame]4468 <a name="int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic</a>
Chris Lattnereff29ab2005-05-15 19:39:26 +0000[diff] [blame]4469</div>
4470
4471<div class="doc_text">
4472
4473<h5>Syntax:</h5>
4474<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]4475 declare i8 %llvm.cttz.i8 (i8 &lt;src&gt;)
4476 declare i16 %llvm.cttz.i16(i16 &lt;src&gt;)
4477 declare i32 %llvm.cttz.i32(i32 &lt;src&gt;)
4478 declare i64 %llvm.cttz.i64(i64 &lt;src&gt;)
Chris Lattnereff29ab2005-05-15 19:39:26 +0000[diff] [blame]4479</pre>
4480
4481<h5>Overview:</h5>
4482
4483<p>
Reid Spencer0b118202006-01-16 21:12:35 +0000[diff] [blame]4484The '<tt>llvm.cttz</tt>' family of intrinsic functions counts the number of
4485trailing zeros.
Chris Lattnereff29ab2005-05-15 19:39:26 +0000[diff] [blame]4486</p>
4487
4488<h5>Arguments:</h5>
4489
4490<p>
4491The only argument is the value to be counted. The argument may be of any
Chris Lattnerec6cb612006-01-16 22:38:59 +0000[diff] [blame]4492unsigned integer type. The return type must match the argument type.
Chris Lattnereff29ab2005-05-15 19:39:26 +0000[diff] [blame]4493</p>
4494
4495<h5>Semantics:</h5>
4496
4497<p>
4498The '<tt>llvm.cttz</tt>' intrinsic counts the trailing (least significant) zeros
4499in a variable. If the src == 0 then the result is the size in bits of the type
4500of src. For example, <tt>llvm.cttz(2) = 1</tt>.
4501</p>
4502</div>
4503
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]4504<!-- ======================================================================= -->
4505<div class="doc_subsection">
4506 <a name="int_debugger">Debugger Intrinsics</a>
4507</div>
4508
4509<div class="doc_text">
4510<p>
4511The LLVM debugger intrinsics (which all start with <tt>llvm.dbg.</tt> prefix),
4512are described in the <a
4513href="SourceLevelDebugging.html#format_common_intrinsics">LLVM Source Level
4514Debugging</a> document.
4515</p>
4516</div>
4517
4518
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4519<!-- *********************************************************************** -->
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4520<hr>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000[diff] [blame]4521<address>
4522 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
4523 src="http://jigsaw.w3.org/css-validator/images/vcss" alt="Valid CSS!"></a>
4524 <a href="http://validator.w3.org/check/referer"><img
4525 src="http://www.w3.org/Icons/valid-html401" alt="Valid HTML 4.01!" /></a>
4526
4527 <a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
Reid Spencer05fe4b02006-03-14 05:39:39 +0000[diff] [blame]4528 <a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000[diff] [blame]4529 Last modified: $Date$
4530</address>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]4531</body>
4532</html>