1 <!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN"
2 "http://www.w3.org/TR/html4/strict.dtd">
5 <title>LLVM Assembly Language Reference Manual</title>
6 <meta http-equiv="Content-Type" content="text/html; charset=utf-8">
7 <meta name="author" content="Chris Lattner">
8 <meta name="description"
9 content="LLVM Assembly Language Reference Manual.">
10 <link rel="stylesheet" href="llvm.css" type="text/css">
15 <h1>LLVM Language Reference Manual</h1>
17 <li><a href="#abstract">Abstract</a></li>
18 <li><a href="#introduction">Introduction</a></li>
19 <li><a href="#identifiers">Identifiers</a></li>
20 <li><a href="#highlevel">High Level Structure</a>
22 <li><a href="#modulestructure">Module Structure</a></li>
23 <li><a href="#linkage">Linkage Types</a>
25 <li><a href="#linkage_private">'<tt>private</tt>' Linkage</a></li>
26 <li><a href="#linkage_linker_private">'<tt>linker_private</tt>' Linkage</a></li>
27 <li><a href="#linkage_linker_private_weak">'<tt>linker_private_weak</tt>' Linkage</a></li>
28 <li><a href="#linkage_linker_private_weak_def_auto">'<tt>linker_private_weak_def_auto</tt>' Linkage</a></li>
29 <li><a href="#linkage_internal">'<tt>internal</tt>' Linkage</a></li>
30 <li><a href="#linkage_available_externally">'<tt>available_externally</tt>' Linkage</a></li>
31 <li><a href="#linkage_linkonce">'<tt>linkonce</tt>' Linkage</a></li>
32 <li><a href="#linkage_common">'<tt>common</tt>' Linkage</a></li>
33 <li><a href="#linkage_weak">'<tt>weak</tt>' Linkage</a></li>
34 <li><a href="#linkage_appending">'<tt>appending</tt>' Linkage</a></li>
35 <li><a href="#linkage_externweak">'<tt>extern_weak</tt>' Linkage</a></li>
36 <li><a href="#linkage_linkonce_odr">'<tt>linkonce_odr</tt>' Linkage</a></li>
37 <li><a href="#linkage_weak">'<tt>weak_odr</tt>' Linkage</a></li>
38 <li><a href="#linkage_external">'<tt>external</tt>' Linkage</a></li>
39 <li><a href="#linkage_dllimport">'<tt>dllimport</tt>' Linkage</a></li>
40 <li><a href="#linkage_dllexport">'<tt>dllexport</tt>' Linkage</a></li>
43 <li><a href="#callingconv">Calling Conventions</a></li>
44 <li><a href="#namedtypes">Named Types</a></li>
45 <li><a href="#globalvars">Global Variables</a></li>
46 <li><a href="#functionstructure">Functions</a></li>
47 <li><a href="#aliasstructure">Aliases</a></li>
48 <li><a href="#namedmetadatastructure">Named Metadata</a></li>
49 <li><a href="#paramattrs">Parameter Attributes</a></li>
50 <li><a href="#fnattrs">Function Attributes</a></li>
51 <li><a href="#gc">Garbage Collector Names</a></li>
52 <li><a href="#moduleasm">Module-Level Inline Assembly</a></li>
53 <li><a href="#datalayout">Data Layout</a></li>
54 <li><a href="#pointeraliasing">Pointer Aliasing Rules</a></li>
55 <li><a href="#volatile">Volatile Memory Accesses</a></li>
56 <li><a href="#memmodel">Memory Model for Concurrent Operations</a></li>
57 <li><a href="#ordering">Atomic Memory Ordering Constraints</a></li>
60 <li><a href="#typesystem">Type System</a>
62 <li><a href="#t_classifications">Type Classifications</a></li>
63 <li><a href="#t_primitive">Primitive Types</a>
65 <li><a href="#t_integer">Integer Type</a></li>
66 <li><a href="#t_floating">Floating Point Types</a></li>
67 <li><a href="#t_x86mmx">X86mmx Type</a></li>
68 <li><a href="#t_void">Void Type</a></li>
69 <li><a href="#t_label">Label Type</a></li>
70 <li><a href="#t_metadata">Metadata Type</a></li>
73 <li><a href="#t_derived">Derived Types</a>
75 <li><a href="#t_aggregate">Aggregate Types</a>
77 <li><a href="#t_array">Array Type</a></li>
78 <li><a href="#t_struct">Structure Type</a></li>
79 <li><a href="#t_opaque">Opaque Structure Types</a></li>
80 <li><a href="#t_vector">Vector Type</a></li>
83 <li><a href="#t_function">Function Type</a></li>
84 <li><a href="#t_pointer">Pointer Type</a></li>
89 <li><a href="#constants">Constants</a>
91 <li><a href="#simpleconstants">Simple Constants</a></li>
92 <li><a href="#complexconstants">Complex Constants</a></li>
93 <li><a href="#globalconstants">Global Variable and Function Addresses</a></li>
94 <li><a href="#undefvalues">Undefined Values</a></li>
95 <li><a href="#trapvalues">Trap Values</a></li>
96 <li><a href="#blockaddress">Addresses of Basic Blocks</a></li>
97 <li><a href="#constantexprs">Constant Expressions</a></li>
100 <li><a href="#othervalues">Other Values</a>
102 <li><a href="#inlineasm">Inline Assembler Expressions</a></li>
103 <li><a href="#metadata">Metadata Nodes and Metadata Strings</a>
105 <li><a href="#tbaa">'<tt>tbaa</tt>' Metadata</a></li>
106 <li><a href="#fpaccuracy">'<tt>fpaccuracy</tt>' Metadata</a></li>
111 <li><a href="#intrinsic_globals">Intrinsic Global Variables</a>
113 <li><a href="#intg_used">The '<tt>llvm.used</tt>' Global Variable</a></li>
114 <li><a href="#intg_compiler_used">The '<tt>llvm.compiler.used</tt>'
115 Global Variable</a></li>
116 <li><a href="#intg_global_ctors">The '<tt>llvm.global_ctors</tt>'
117 Global Variable</a></li>
118 <li><a href="#intg_global_dtors">The '<tt>llvm.global_dtors</tt>'
119 Global Variable</a></li>
122 <li><a href="#instref">Instruction Reference</a>
124 <li><a href="#terminators">Terminator Instructions</a>
126 <li><a href="#i_ret">'<tt>ret</tt>' Instruction</a></li>
127 <li><a href="#i_br">'<tt>br</tt>' Instruction</a></li>
128 <li><a href="#i_switch">'<tt>switch</tt>' Instruction</a></li>
129 <li><a href="#i_indirectbr">'<tt>indirectbr</tt>' Instruction</a></li>
130 <li><a href="#i_invoke">'<tt>invoke</tt>' Instruction</a></li>
131 <li><a href="#i_unwind">'<tt>unwind</tt>' Instruction</a></li>
132 <li><a href="#i_resume">'<tt>resume</tt>' Instruction</a></li>
133 <li><a href="#i_unreachable">'<tt>unreachable</tt>' Instruction</a></li>
136 <li><a href="#binaryops">Binary Operations</a>
138 <li><a href="#i_add">'<tt>add</tt>' Instruction</a></li>
139 <li><a href="#i_fadd">'<tt>fadd</tt>' Instruction</a></li>
140 <li><a href="#i_sub">'<tt>sub</tt>' Instruction</a></li>
141 <li><a href="#i_fsub">'<tt>fsub</tt>' Instruction</a></li>
142 <li><a href="#i_mul">'<tt>mul</tt>' Instruction</a></li>
143 <li><a href="#i_fmul">'<tt>fmul</tt>' Instruction</a></li>
144 <li><a href="#i_udiv">'<tt>udiv</tt>' Instruction</a></li>
145 <li><a href="#i_sdiv">'<tt>sdiv</tt>' Instruction</a></li>
146 <li><a href="#i_fdiv">'<tt>fdiv</tt>' Instruction</a></li>
147 <li><a href="#i_urem">'<tt>urem</tt>' Instruction</a></li>
148 <li><a href="#i_srem">'<tt>srem</tt>' Instruction</a></li>
149 <li><a href="#i_frem">'<tt>frem</tt>' Instruction</a></li>
152 <li><a href="#bitwiseops">Bitwise Binary Operations</a>
154 <li><a href="#i_shl">'<tt>shl</tt>' Instruction</a></li>
155 <li><a href="#i_lshr">'<tt>lshr</tt>' Instruction</a></li>
156 <li><a href="#i_ashr">'<tt>ashr</tt>' Instruction</a></li>
157 <li><a href="#i_and">'<tt>and</tt>' Instruction</a></li>
158 <li><a href="#i_or">'<tt>or</tt>' Instruction</a></li>
159 <li><a href="#i_xor">'<tt>xor</tt>' Instruction</a></li>
162 <li><a href="#vectorops">Vector Operations</a>
164 <li><a href="#i_extractelement">'<tt>extractelement</tt>' Instruction</a></li>
165 <li><a href="#i_insertelement">'<tt>insertelement</tt>' Instruction</a></li>
166 <li><a href="#i_shufflevector">'<tt>shufflevector</tt>' Instruction</a></li>
169 <li><a href="#aggregateops">Aggregate Operations</a>
171 <li><a href="#i_extractvalue">'<tt>extractvalue</tt>' Instruction</a></li>
172 <li><a href="#i_insertvalue">'<tt>insertvalue</tt>' Instruction</a></li>
175 <li><a href="#memoryops">Memory Access and Addressing Operations</a>
177 <li><a href="#i_alloca">'<tt>alloca</tt>' Instruction</a></li>
178 <li><a href="#i_load">'<tt>load</tt>' Instruction</a></li>
179 <li><a href="#i_store">'<tt>store</tt>' Instruction</a></li>
180 <li><a href="#i_fence">'<tt>fence</tt>' Instruction</a></li>
181 <li><a href="#i_cmpxchg">'<tt>cmpxchg</tt>' Instruction</a></li>
182 <li><a href="#i_atomicrmw">'<tt>atomicrmw</tt>' Instruction</a></li>
183 <li><a href="#i_getelementptr">'<tt>getelementptr</tt>' Instruction</a></li>
186 <li><a href="#convertops">Conversion Operations</a>
188 <li><a href="#i_trunc">'<tt>trunc .. to</tt>' Instruction</a></li>
189 <li><a href="#i_zext">'<tt>zext .. to</tt>' Instruction</a></li>
190 <li><a href="#i_sext">'<tt>sext .. to</tt>' Instruction</a></li>
191 <li><a href="#i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a></li>
192 <li><a href="#i_fpext">'<tt>fpext .. to</tt>' Instruction</a></li>
193 <li><a href="#i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a></li>
194 <li><a href="#i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a></li>
195 <li><a href="#i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a></li>
196 <li><a href="#i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a></li>
197 <li><a href="#i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a></li>
198 <li><a href="#i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a></li>
199 <li><a href="#i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a></li>
202 <li><a href="#otherops">Other Operations</a>
204 <li><a href="#i_icmp">'<tt>icmp</tt>' Instruction</a></li>
205 <li><a href="#i_fcmp">'<tt>fcmp</tt>' Instruction</a></li>
206 <li><a href="#i_phi">'<tt>phi</tt>' Instruction</a></li>
207 <li><a href="#i_select">'<tt>select</tt>' Instruction</a></li>
208 <li><a href="#i_call">'<tt>call</tt>' Instruction</a></li>
209 <li><a href="#i_va_arg">'<tt>va_arg</tt>' Instruction</a></li>
210 <li><a href="#i_landingpad">'<tt>landingpad</tt>' Instruction</a></li>
215 <li><a href="#intrinsics">Intrinsic Functions</a>
217 <li><a href="#int_varargs">Variable Argument Handling Intrinsics</a>
219 <li><a href="#int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a></li>
220 <li><a href="#int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a></li>
221 <li><a href="#int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a></li>
224 <li><a href="#int_gc">Accurate Garbage Collection Intrinsics</a>
226 <li><a href="#int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a></li>
227 <li><a href="#int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a></li>
228 <li><a href="#int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a></li>
231 <li><a href="#int_codegen">Code Generator Intrinsics</a>
233 <li><a href="#int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a></li>
234 <li><a href="#int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a></li>
235 <li><a href="#int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a></li>
236 <li><a href="#int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a></li>
237 <li><a href="#int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a></li>
238 <li><a href="#int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a></li>
239 <li><a href="#int_readcyclecounter">'<tt>llvm.readcyclecounter</tt>' Intrinsic</a></li>
242 <li><a href="#int_libc">Standard C Library Intrinsics</a>
244 <li><a href="#int_memcpy">'<tt>llvm.memcpy.*</tt>' Intrinsic</a></li>
245 <li><a href="#int_memmove">'<tt>llvm.memmove.*</tt>' Intrinsic</a></li>
246 <li><a href="#int_memset">'<tt>llvm.memset.*</tt>' Intrinsic</a></li>
247 <li><a href="#int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a></li>
248 <li><a href="#int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a></li>
249 <li><a href="#int_sin">'<tt>llvm.sin.*</tt>' Intrinsic</a></li>
250 <li><a href="#int_cos">'<tt>llvm.cos.*</tt>' Intrinsic</a></li>
251 <li><a href="#int_pow">'<tt>llvm.pow.*</tt>' Intrinsic</a></li>
252 <li><a href="#int_exp">'<tt>llvm.exp.*</tt>' Intrinsic</a></li>
253 <li><a href="#int_log">'<tt>llvm.log.*</tt>' Intrinsic</a></li>
254 <li><a href="#int_fma">'<tt>llvm.fma.*</tt>' Intrinsic</a></li>
257 <li><a href="#int_manip">Bit Manipulation Intrinsics</a>
259 <li><a href="#int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a></li>
260 <li><a href="#int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic </a></li>
261 <li><a href="#int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic </a></li>
262 <li><a href="#int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic </a></li>
265 <li><a href="#int_overflow">Arithmetic with Overflow Intrinsics</a>
267 <li><a href="#int_sadd_overflow">'<tt>llvm.sadd.with.overflow.*</tt> Intrinsics</a></li>
268 <li><a href="#int_uadd_overflow">'<tt>llvm.uadd.with.overflow.*</tt> Intrinsics</a></li>
269 <li><a href="#int_ssub_overflow">'<tt>llvm.ssub.with.overflow.*</tt> Intrinsics</a></li>
270 <li><a href="#int_usub_overflow">'<tt>llvm.usub.with.overflow.*</tt> Intrinsics</a></li>
271 <li><a href="#int_smul_overflow">'<tt>llvm.smul.with.overflow.*</tt> Intrinsics</a></li>
272 <li><a href="#int_umul_overflow">'<tt>llvm.umul.with.overflow.*</tt> Intrinsics</a></li>
275 <li><a href="#int_fp16">Half Precision Floating Point Intrinsics</a>
277 <li><a href="#int_convert_to_fp16">'<tt>llvm.convert.to.fp16</tt>' Intrinsic</a></li>
278 <li><a href="#int_convert_from_fp16">'<tt>llvm.convert.from.fp16</tt>' Intrinsic</a></li>
281 <li><a href="#int_debugger">Debugger intrinsics</a></li>
282 <li><a href="#int_eh">Exception Handling intrinsics</a></li>
283 <li><a href="#int_trampoline">Trampoline Intrinsics</a>
285 <li><a href="#int_it">'<tt>llvm.init.trampoline</tt>' Intrinsic</a></li>
286 <li><a href="#int_at">'<tt>llvm.adjust.trampoline</tt>' Intrinsic</a></li>
289 <li><a href="#int_memorymarkers">Memory Use Markers</a>
291 <li><a href="#int_lifetime_start"><tt>llvm.lifetime.start</tt></a></li>
292 <li><a href="#int_lifetime_end"><tt>llvm.lifetime.end</tt></a></li>
293 <li><a href="#int_invariant_start"><tt>llvm.invariant.start</tt></a></li>
294 <li><a href="#int_invariant_end"><tt>llvm.invariant.end</tt></a></li>
297 <li><a href="#int_general">General intrinsics</a>
299 <li><a href="#int_var_annotation">
300 '<tt>llvm.var.annotation</tt>' Intrinsic</a></li>
301 <li><a href="#int_annotation">
302 '<tt>llvm.annotation.*</tt>' Intrinsic</a></li>
303 <li><a href="#int_trap">
304 '<tt>llvm.trap</tt>' Intrinsic</a></li>
305 <li><a href="#int_stackprotector">
306 '<tt>llvm.stackprotector</tt>' Intrinsic</a></li>
307 <li><a href="#int_objectsize">
308 '<tt>llvm.objectsize</tt>' Intrinsic</a></li>
315 <div class="doc_author">
316 <p>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a>
317 and <a href="mailto:vadve@cs.uiuc.edu">Vikram Adve</a></p>
320 <!-- *********************************************************************** -->
321 <h2><a name="abstract">Abstract</a></h2>
322 <!-- *********************************************************************** -->
326 <p>This document is a reference manual for the LLVM assembly language. LLVM is
327 a Static Single Assignment (SSA) based representation that provides type
328 safety, low-level operations, flexibility, and the capability of representing
329 'all' high-level languages cleanly. It is the common code representation
330 used throughout all phases of the LLVM compilation strategy.</p>
334 <!-- *********************************************************************** -->
335 <h2><a name="introduction">Introduction</a></h2>
336 <!-- *********************************************************************** -->
340 <p>The LLVM code representation is designed to be used in three different forms:
341 as an in-memory compiler IR, as an on-disk bitcode representation (suitable
342 for fast loading by a Just-In-Time compiler), and as a human readable
343 assembly language representation. This allows LLVM to provide a powerful
344 intermediate representation for efficient compiler transformations and
345 analysis, while providing a natural means to debug and visualize the
346 transformations. The three different forms of LLVM are all equivalent. This
347 document describes the human readable representation and notation.</p>
349 <p>The LLVM representation aims to be light-weight and low-level while being
350 expressive, typed, and extensible at the same time. It aims to be a
351 "universal IR" of sorts, by being at a low enough level that high-level ideas
352 may be cleanly mapped to it (similar to how microprocessors are "universal
353 IR's", allowing many source languages to be mapped to them). By providing
354 type information, LLVM can be used as the target of optimizations: for
355 example, through pointer analysis, it can be proven that a C automatic
356 variable is never accessed outside of the current function, allowing it to
357 be promoted to a simple SSA value instead of a memory location.</p>
359 <!-- _______________________________________________________________________ -->
361 <a name="wellformed">Well-Formedness</a>
366 <p>It is important to note that this document describes 'well formed' LLVM
367 assembly language. There is a difference between what the parser accepts and
368 what is considered 'well formed'. For example, the following instruction is
369 syntactically okay, but not well formed:</p>
371 <pre class="doc_code">
372 %x = <a href="#i_add">add</a> i32 1, %x
375 <p>because the definition of <tt>%x</tt> does not dominate all of its uses. The
376 LLVM infrastructure provides a verification pass that may be used to verify
377 that an LLVM module is well formed. This pass is automatically run by the
378 parser after parsing input assembly and by the optimizer before it outputs
379 bitcode. The violations pointed out by the verifier pass indicate bugs in
380 transformation passes or input to the parser.</p>
386 <!-- Describe the typesetting conventions here. -->
388 <!-- *********************************************************************** -->
389 <h2><a name="identifiers">Identifiers</a></h2>
390 <!-- *********************************************************************** -->
394 <p>LLVM identifiers come in two basic types: global and local. Global
395 identifiers (functions, global variables) begin with the <tt>'@'</tt>
396 character. Local identifiers (register names, types) begin with
397 the <tt>'%'</tt> character. Additionally, there are three different formats
398 for identifiers, for different purposes:</p>
401 <li>Named values are represented as a string of characters with their prefix.
402 For example, <tt>%foo</tt>, <tt>@DivisionByZero</tt>,
403 <tt>%a.really.long.identifier</tt>. The actual regular expression used is
404 '<tt>[%@][a-zA-Z$._][a-zA-Z$._0-9]*</tt>'. Identifiers which require
405 other characters in their names can be surrounded with quotes. Special
406 characters may be escaped using <tt>"\xx"</tt> where <tt>xx</tt> is the
407 ASCII code for the character in hexadecimal. In this way, any character
408 can be used in a name value, even quotes themselves.</li>
410 <li>Unnamed values are represented as an unsigned numeric value with their
411 prefix. For example, <tt>%12</tt>, <tt>@2</tt>, <tt>%44</tt>.</li>
413 <li>Constants, which are described in a <a href="#constants">section about
414 constants</a>, below.</li>
417 <p>LLVM requires that values start with a prefix for two reasons: Compilers
418 don't need to worry about name clashes with reserved words, and the set of
419 reserved words may be expanded in the future without penalty. Additionally,
420 unnamed identifiers allow a compiler to quickly come up with a temporary
421 variable without having to avoid symbol table conflicts.</p>
423 <p>Reserved words in LLVM are very similar to reserved words in other
424 languages. There are keywords for different opcodes
425 ('<tt><a href="#i_add">add</a></tt>',
426 '<tt><a href="#i_bitcast">bitcast</a></tt>',
427 '<tt><a href="#i_ret">ret</a></tt>', etc...), for primitive type names
428 ('<tt><a href="#t_void">void</a></tt>',
429 '<tt><a href="#t_primitive">i32</a></tt>', etc...), and others. These
430 reserved words cannot conflict with variable names, because none of them
431 start with a prefix character (<tt>'%'</tt> or <tt>'@'</tt>).</p>
433 <p>Here is an example of LLVM code to multiply the integer variable
434 '<tt>%X</tt>' by 8:</p>
438 <pre class="doc_code">
439 %result = <a href="#i_mul">mul</a> i32 %X, 8
442 <p>After strength reduction:</p>
444 <pre class="doc_code">
445 %result = <a href="#i_shl">shl</a> i32 %X, i8 3
448 <p>And the hard way:</p>
450 <pre class="doc_code">
451 %0 = <a href="#i_add">add</a> i32 %X, %X <i>; yields {i32}:%0</i>
452 %1 = <a href="#i_add">add</a> i32 %0, %0 <i>; yields {i32}:%1</i>
453 %result = <a href="#i_add">add</a> i32 %1, %1
456 <p>This last way of multiplying <tt>%X</tt> by 8 illustrates several important
457 lexical features of LLVM:</p>
460 <li>Comments are delimited with a '<tt>;</tt>' and go until the end of
463 <li>Unnamed temporaries are created when the result of a computation is not
464 assigned to a named value.</li>
466 <li>Unnamed temporaries are numbered sequentially</li>
469 <p>It also shows a convention that we follow in this document. When
470 demonstrating instructions, we will follow an instruction with a comment that
471 defines the type and name of value produced. Comments are shown in italic
476 <!-- *********************************************************************** -->
477 <h2><a name="highlevel">High Level Structure</a></h2>
478 <!-- *********************************************************************** -->
480 <!-- ======================================================================= -->
482 <a name="modulestructure">Module Structure</a>
487 <p>LLVM programs are composed of "Module"s, each of which is a translation unit
488 of the input programs. Each module consists of functions, global variables,
489 and symbol table entries. Modules may be combined together with the LLVM
490 linker, which merges function (and global variable) definitions, resolves
491 forward declarations, and merges symbol table entries. Here is an example of
492 the "hello world" module:</p>
494 <pre class="doc_code">
495 <i>; Declare the string constant as a global constant.</i>
496 <a href="#identifiers">@.LC0</a> = <a href="#linkage_internal">internal</a> <a href="#globalvars">constant</a> <a href="#t_array">[13 x i8]</a> c"hello world\0A\00" <i>; [13 x i8]*</i>
498 <i>; External declaration of the puts function</i>
499 <a href="#functionstructure">declare</a> i32 @puts(i8*) <i>; i32 (i8*)* </i>
501 <i>; Definition of main function</i>
502 define i32 @main() { <i>; i32()* </i>
503 <i>; Convert [13 x i8]* to i8 *...</i>
504 %cast210 = <a href="#i_getelementptr">getelementptr</a> [13 x i8]* @.LC0, i64 0, i64 0 <i>; i8*</i>
506 <i>; Call puts function to write out the string to stdout.</i>
507 <a href="#i_call">call</a> i32 @puts(i8* %cast210) <i>; i32</i>
508 <a href="#i_ret">ret</a> i32 0
511 <i>; Named metadata</i>
512 !1 = metadata !{i32 41}
516 <p>This example is made up of a <a href="#globalvars">global variable</a> named
517 "<tt>.LC0</tt>", an external declaration of the "<tt>puts</tt>" function,
518 a <a href="#functionstructure">function definition</a> for
519 "<tt>main</tt>" and <a href="#namedmetadatastructure">named metadata</a>
522 <p>In general, a module is made up of a list of global values, where both
523 functions and global variables are global values. Global values are
524 represented by a pointer to a memory location (in this case, a pointer to an
525 array of char, and a pointer to a function), and have one of the
526 following <a href="#linkage">linkage types</a>.</p>
530 <!-- ======================================================================= -->
532 <a name="linkage">Linkage Types</a>
537 <p>All Global Variables and Functions have one of the following types of
541 <dt><tt><b><a name="linkage_private">private</a></b></tt></dt>
542 <dd>Global values with "<tt>private</tt>" linkage are only directly accessible
543 by objects in the current module. In particular, linking code into a
544 module with an private global value may cause the private to be renamed as
545 necessary to avoid collisions. Because the symbol is private to the
546 module, all references can be updated. This doesn't show up in any symbol
547 table in the object file.</dd>
549 <dt><tt><b><a name="linkage_linker_private">linker_private</a></b></tt></dt>
550 <dd>Similar to <tt>private</tt>, but the symbol is passed through the
551 assembler and evaluated by the linker. Unlike normal strong symbols, they
552 are removed by the linker from the final linked image (executable or
553 dynamic library).</dd>
555 <dt><tt><b><a name="linkage_linker_private_weak">linker_private_weak</a></b></tt></dt>
556 <dd>Similar to "<tt>linker_private</tt>", but the symbol is weak. Note that
557 <tt>linker_private_weak</tt> symbols are subject to coalescing by the
558 linker. The symbols are removed by the linker from the final linked image
559 (executable or dynamic library).</dd>
561 <dt><tt><b><a name="linkage_linker_private_weak_def_auto">linker_private_weak_def_auto</a></b></tt></dt>
562 <dd>Similar to "<tt>linker_private_weak</tt>", but it's known that the address
563 of the object is not taken. For instance, functions that had an inline
564 definition, but the compiler decided not to inline it. Note,
565 unlike <tt>linker_private</tt> and <tt>linker_private_weak</tt>,
566 <tt>linker_private_weak_def_auto</tt> may have only <tt>default</tt>
567 visibility. The symbols are removed by the linker from the final linked
568 image (executable or dynamic library).</dd>
570 <dt><tt><b><a name="linkage_internal">internal</a></b></tt></dt>
571 <dd>Similar to private, but the value shows as a local symbol
572 (<tt>STB_LOCAL</tt> in the case of ELF) in the object file. This
573 corresponds to the notion of the '<tt>static</tt>' keyword in C.</dd>
575 <dt><tt><b><a name="linkage_available_externally">available_externally</a></b></tt></dt>
576 <dd>Globals with "<tt>available_externally</tt>" linkage are never emitted
577 into the object file corresponding to the LLVM module. They exist to
578 allow inlining and other optimizations to take place given knowledge of
579 the definition of the global, which is known to be somewhere outside the
580 module. Globals with <tt>available_externally</tt> linkage are allowed to
581 be discarded at will, and are otherwise the same as <tt>linkonce_odr</tt>.
582 This linkage type is only allowed on definitions, not declarations.</dd>
584 <dt><tt><b><a name="linkage_linkonce">linkonce</a></b></tt></dt>
585 <dd>Globals with "<tt>linkonce</tt>" linkage are merged with other globals of
586 the same name when linkage occurs. This can be used to implement
587 some forms of inline functions, templates, or other code which must be
588 generated in each translation unit that uses it, but where the body may
589 be overridden with a more definitive definition later. Unreferenced
590 <tt>linkonce</tt> globals are allowed to be discarded. Note that
591 <tt>linkonce</tt> linkage does not actually allow the optimizer to
592 inline the body of this function into callers because it doesn't know if
593 this definition of the function is the definitive definition within the
594 program or whether it will be overridden by a stronger definition.
595 To enable inlining and other optimizations, use "<tt>linkonce_odr</tt>"
598 <dt><tt><b><a name="linkage_weak">weak</a></b></tt></dt>
599 <dd>"<tt>weak</tt>" linkage has the same merging semantics as
600 <tt>linkonce</tt> linkage, except that unreferenced globals with
601 <tt>weak</tt> linkage may not be discarded. This is used for globals that
602 are declared "weak" in C source code.</dd>
604 <dt><tt><b><a name="linkage_common">common</a></b></tt></dt>
605 <dd>"<tt>common</tt>" linkage is most similar to "<tt>weak</tt>" linkage, but
606 they are used for tentative definitions in C, such as "<tt>int X;</tt>" at
608 Symbols with "<tt>common</tt>" linkage are merged in the same way as
609 <tt>weak symbols</tt>, and they may not be deleted if unreferenced.
610 <tt>common</tt> symbols may not have an explicit section,
611 must have a zero initializer, and may not be marked '<a
612 href="#globalvars"><tt>constant</tt></a>'. Functions and aliases may not
613 have common linkage.</dd>
616 <dt><tt><b><a name="linkage_appending">appending</a></b></tt></dt>
617 <dd>"<tt>appending</tt>" linkage may only be applied to global variables of
618 pointer to array type. When two global variables with appending linkage
619 are linked together, the two global arrays are appended together. This is
620 the LLVM, typesafe, equivalent of having the system linker append together
621 "sections" with identical names when .o files are linked.</dd>
623 <dt><tt><b><a name="linkage_externweak">extern_weak</a></b></tt></dt>
624 <dd>The semantics of this linkage follow the ELF object file model: the symbol
625 is weak until linked, if not linked, the symbol becomes null instead of
626 being an undefined reference.</dd>
628 <dt><tt><b><a name="linkage_linkonce_odr">linkonce_odr</a></b></tt></dt>
629 <dt><tt><b><a name="linkage_weak_odr">weak_odr</a></b></tt></dt>
630 <dd>Some languages allow differing globals to be merged, such as two functions
631 with different semantics. Other languages, such as <tt>C++</tt>, ensure
632 that only equivalent globals are ever merged (the "one definition rule"
633 — "ODR"). Such languages can use the <tt>linkonce_odr</tt>
634 and <tt>weak_odr</tt> linkage types to indicate that the global will only
635 be merged with equivalent globals. These linkage types are otherwise the
636 same as their non-<tt>odr</tt> versions.</dd>
638 <dt><tt><b><a name="linkage_external">external</a></b></tt></dt>
639 <dd>If none of the above identifiers are used, the global is externally
640 visible, meaning that it participates in linkage and can be used to
641 resolve external symbol references.</dd>
644 <p>The next two types of linkage are targeted for Microsoft Windows platform
645 only. They are designed to support importing (exporting) symbols from (to)
646 DLLs (Dynamic Link Libraries).</p>
649 <dt><tt><b><a name="linkage_dllimport">dllimport</a></b></tt></dt>
650 <dd>"<tt>dllimport</tt>" linkage causes the compiler to reference a function
651 or variable via a global pointer to a pointer that is set up by the DLL
652 exporting the symbol. On Microsoft Windows targets, the pointer name is
653 formed by combining <code>__imp_</code> and the function or variable
656 <dt><tt><b><a name="linkage_dllexport">dllexport</a></b></tt></dt>
657 <dd>"<tt>dllexport</tt>" linkage causes the compiler to provide a global
658 pointer to a pointer in a DLL, so that it can be referenced with the
659 <tt>dllimport</tt> attribute. On Microsoft Windows targets, the pointer
660 name is formed by combining <code>__imp_</code> and the function or
664 <p>For example, since the "<tt>.LC0</tt>" variable is defined to be internal, if
665 another module defined a "<tt>.LC0</tt>" variable and was linked with this
666 one, one of the two would be renamed, preventing a collision. Since
667 "<tt>main</tt>" and "<tt>puts</tt>" are external (i.e., lacking any linkage
668 declarations), they are accessible outside of the current module.</p>
670 <p>It is illegal for a function <i>declaration</i> to have any linkage type
671 other than <tt>external</tt>, <tt>dllimport</tt>
672 or <tt>extern_weak</tt>.</p>
674 <p>Aliases can have only <tt>external</tt>, <tt>internal</tt>, <tt>weak</tt>
675 or <tt>weak_odr</tt> linkages.</p>
679 <!-- ======================================================================= -->
681 <a name="callingconv">Calling Conventions</a>
686 <p>LLVM <a href="#functionstructure">functions</a>, <a href="#i_call">calls</a>
687 and <a href="#i_invoke">invokes</a> can all have an optional calling
688 convention specified for the call. The calling convention of any pair of
689 dynamic caller/callee must match, or the behavior of the program is
690 undefined. The following calling conventions are supported by LLVM, and more
691 may be added in the future:</p>
694 <dt><b>"<tt>ccc</tt>" - The C calling convention</b>:</dt>
695 <dd>This calling convention (the default if no other calling convention is
696 specified) matches the target C calling conventions. This calling
697 convention supports varargs function calls and tolerates some mismatch in
698 the declared prototype and implemented declaration of the function (as
701 <dt><b>"<tt>fastcc</tt>" - The fast calling convention</b>:</dt>
702 <dd>This calling convention attempts to make calls as fast as possible
703 (e.g. by passing things in registers). This calling convention allows the
704 target to use whatever tricks it wants to produce fast code for the
705 target, without having to conform to an externally specified ABI
706 (Application Binary Interface).
707 <a href="CodeGenerator.html#tailcallopt">Tail calls can only be optimized
708 when this or the GHC convention is used.</a> This calling convention
709 does not support varargs and requires the prototype of all callees to
710 exactly match the prototype of the function definition.</dd>
712 <dt><b>"<tt>coldcc</tt>" - The cold calling convention</b>:</dt>
713 <dd>This calling convention attempts to make code in the caller as efficient
714 as possible under the assumption that the call is not commonly executed.
715 As such, these calls often preserve all registers so that the call does
716 not break any live ranges in the caller side. This calling convention
717 does not support varargs and requires the prototype of all callees to
718 exactly match the prototype of the function definition.</dd>
720 <dt><b>"<tt>cc <em>10</em></tt>" - GHC convention</b>:</dt>
721 <dd>This calling convention has been implemented specifically for use by the
722 <a href="http://www.haskell.org/ghc">Glasgow Haskell Compiler (GHC)</a>.
723 It passes everything in registers, going to extremes to achieve this by
724 disabling callee save registers. This calling convention should not be
725 used lightly but only for specific situations such as an alternative to
726 the <em>register pinning</em> performance technique often used when
727 implementing functional programming languages.At the moment only X86
728 supports this convention and it has the following limitations:
730 <li>On <em>X86-32</em> only supports up to 4 bit type parameters. No
731 floating point types are supported.</li>
732 <li>On <em>X86-64</em> only supports up to 10 bit type parameters and
733 6 floating point parameters.</li>
735 This calling convention supports
736 <a href="CodeGenerator.html#tailcallopt">tail call optimization</a> but
737 requires both the caller and callee are using it.
740 <dt><b>"<tt>cc <<em>n</em>></tt>" - Numbered convention</b>:</dt>
741 <dd>Any calling convention may be specified by number, allowing
742 target-specific calling conventions to be used. Target specific calling
743 conventions start at 64.</dd>
746 <p>More calling conventions can be added/defined on an as-needed basis, to
747 support Pascal conventions or any other well-known target-independent
752 <!-- ======================================================================= -->
754 <a name="visibility">Visibility Styles</a>
759 <p>All Global Variables and Functions have one of the following visibility
763 <dt><b>"<tt>default</tt>" - Default style</b>:</dt>
764 <dd>On targets that use the ELF object file format, default visibility means
765 that the declaration is visible to other modules and, in shared libraries,
766 means that the declared entity may be overridden. On Darwin, default
767 visibility means that the declaration is visible to other modules. Default
768 visibility corresponds to "external linkage" in the language.</dd>
770 <dt><b>"<tt>hidden</tt>" - Hidden style</b>:</dt>
771 <dd>Two declarations of an object with hidden visibility refer to the same
772 object if they are in the same shared object. Usually, hidden visibility
773 indicates that the symbol will not be placed into the dynamic symbol
774 table, so no other module (executable or shared library) can reference it
777 <dt><b>"<tt>protected</tt>" - Protected style</b>:</dt>
778 <dd>On ELF, protected visibility indicates that the symbol will be placed in
779 the dynamic symbol table, but that references within the defining module
780 will bind to the local symbol. That is, the symbol cannot be overridden by
786 <!-- ======================================================================= -->
788 <a name="namedtypes">Named Types</a>
793 <p>LLVM IR allows you to specify name aliases for certain types. This can make
794 it easier to read the IR and make the IR more condensed (particularly when
795 recursive types are involved). An example of a name specification is:</p>
797 <pre class="doc_code">
798 %mytype = type { %mytype*, i32 }
801 <p>You may give a name to any <a href="#typesystem">type</a> except
802 "<a href="#t_void">void</a>". Type name aliases may be used anywhere a type
803 is expected with the syntax "%mytype".</p>
805 <p>Note that type names are aliases for the structural type that they indicate,
806 and that you can therefore specify multiple names for the same type. This
807 often leads to confusing behavior when dumping out a .ll file. Since LLVM IR
808 uses structural typing, the name is not part of the type. When printing out
809 LLVM IR, the printer will pick <em>one name</em> to render all types of a
810 particular shape. This means that if you have code where two different
811 source types end up having the same LLVM type, that the dumper will sometimes
812 print the "wrong" or unexpected type. This is an important design point and
813 isn't going to change.</p>
817 <!-- ======================================================================= -->
819 <a name="globalvars">Global Variables</a>
824 <p>Global variables define regions of memory allocated at compilation time
825 instead of run-time. Global variables may optionally be initialized, may
826 have an explicit section to be placed in, and may have an optional explicit
827 alignment specified. A variable may be defined as "thread_local", which
828 means that it will not be shared by threads (each thread will have a
829 separated copy of the variable). A variable may be defined as a global
830 "constant," which indicates that the contents of the variable
831 will <b>never</b> be modified (enabling better optimization, allowing the
832 global data to be placed in the read-only section of an executable, etc).
833 Note that variables that need runtime initialization cannot be marked
834 "constant" as there is a store to the variable.</p>
836 <p>LLVM explicitly allows <em>declarations</em> of global variables to be marked
837 constant, even if the final definition of the global is not. This capability
838 can be used to enable slightly better optimization of the program, but
839 requires the language definition to guarantee that optimizations based on the
840 'constantness' are valid for the translation units that do not include the
843 <p>As SSA values, global variables define pointer values that are in scope
844 (i.e. they dominate) all basic blocks in the program. Global variables
845 always define a pointer to their "content" type because they describe a
846 region of memory, and all memory objects in LLVM are accessed through
849 <p>Global variables can be marked with <tt>unnamed_addr</tt> which indicates
850 that the address is not significant, only the content. Constants marked
851 like this can be merged with other constants if they have the same
852 initializer. Note that a constant with significant address <em>can</em>
853 be merged with a <tt>unnamed_addr</tt> constant, the result being a
854 constant whose address is significant.</p>
856 <p>A global variable may be declared to reside in a target-specific numbered
857 address space. For targets that support them, address spaces may affect how
858 optimizations are performed and/or what target instructions are used to
859 access the variable. The default address space is zero. The address space
860 qualifier must precede any other attributes.</p>
862 <p>LLVM allows an explicit section to be specified for globals. If the target
863 supports it, it will emit globals to the section specified.</p>
865 <p>An explicit alignment may be specified for a global, which must be a power
866 of 2. If not present, or if the alignment is set to zero, the alignment of
867 the global is set by the target to whatever it feels convenient. If an
868 explicit alignment is specified, the global is forced to have exactly that
869 alignment. Targets and optimizers are not allowed to over-align the global
870 if the global has an assigned section. In this case, the extra alignment
871 could be observable: for example, code could assume that the globals are
872 densely packed in their section and try to iterate over them as an array,
873 alignment padding would break this iteration.</p>
875 <p>For example, the following defines a global in a numbered address space with
876 an initializer, section, and alignment:</p>
878 <pre class="doc_code">
879 @G = addrspace(5) constant float 1.0, section "foo", align 4
885 <!-- ======================================================================= -->
887 <a name="functionstructure">Functions</a>
892 <p>LLVM function definitions consist of the "<tt>define</tt>" keyword, an
893 optional <a href="#linkage">linkage type</a>, an optional
894 <a href="#visibility">visibility style</a>, an optional
895 <a href="#callingconv">calling convention</a>,
896 an optional <tt>unnamed_addr</tt> attribute, a return type, an optional
897 <a href="#paramattrs">parameter attribute</a> for the return type, a function
898 name, a (possibly empty) argument list (each with optional
899 <a href="#paramattrs">parameter attributes</a>), optional
900 <a href="#fnattrs">function attributes</a>, an optional section, an optional
901 alignment, an optional <a href="#gc">garbage collector name</a>, an opening
902 curly brace, a list of basic blocks, and a closing curly brace.</p>
904 <p>LLVM function declarations consist of the "<tt>declare</tt>" keyword, an
905 optional <a href="#linkage">linkage type</a>, an optional
906 <a href="#visibility">visibility style</a>, an optional
907 <a href="#callingconv">calling convention</a>,
908 an optional <tt>unnamed_addr</tt> attribute, a return type, an optional
909 <a href="#paramattrs">parameter attribute</a> for the return type, a function
910 name, a possibly empty list of arguments, an optional alignment, and an
911 optional <a href="#gc">garbage collector name</a>.</p>
913 <p>A function definition contains a list of basic blocks, forming the CFG
914 (Control Flow Graph) for the function. Each basic block may optionally start
915 with a label (giving the basic block a symbol table entry), contains a list
916 of instructions, and ends with a <a href="#terminators">terminator</a>
917 instruction (such as a branch or function return).</p>
919 <p>The first basic block in a function is special in two ways: it is immediately
920 executed on entrance to the function, and it is not allowed to have
921 predecessor basic blocks (i.e. there can not be any branches to the entry
922 block of a function). Because the block can have no predecessors, it also
923 cannot have any <a href="#i_phi">PHI nodes</a>.</p>
925 <p>LLVM allows an explicit section to be specified for functions. If the target
926 supports it, it will emit functions to the section specified.</p>
928 <p>An explicit alignment may be specified for a function. If not present, or if
929 the alignment is set to zero, the alignment of the function is set by the
930 target to whatever it feels convenient. If an explicit alignment is
931 specified, the function is forced to have at least that much alignment. All
932 alignments must be a power of 2.</p>
934 <p>If the <tt>unnamed_addr</tt> attribute is given, the address is know to not
935 be significant and two identical functions can be merged.</p>
938 <pre class="doc_code">
939 define [<a href="#linkage">linkage</a>] [<a href="#visibility">visibility</a>]
940 [<a href="#callingconv">cconv</a>] [<a href="#paramattrs">ret attrs</a>]
941 <ResultType> @<FunctionName> ([argument list])
942 [<a href="#fnattrs">fn Attrs</a>] [section "name"] [align N]
943 [<a href="#gc">gc</a>] { ... }
948 <!-- ======================================================================= -->
950 <a name="aliasstructure">Aliases</a>
955 <p>Aliases act as "second name" for the aliasee value (which can be either
956 function, global variable, another alias or bitcast of global value). Aliases
957 may have an optional <a href="#linkage">linkage type</a>, and an
958 optional <a href="#visibility">visibility style</a>.</p>
961 <pre class="doc_code">
962 @<Name> = alias [Linkage] [Visibility] <AliaseeTy> @<Aliasee>
967 <!-- ======================================================================= -->
969 <a name="namedmetadatastructure">Named Metadata</a>
974 <p>Named metadata is a collection of metadata. <a href="#metadata">Metadata
975 nodes</a> (but not metadata strings) are the only valid operands for
976 a named metadata.</p>
979 <pre class="doc_code">
980 ; Some unnamed metadata nodes, which are referenced by the named metadata.
981 !0 = metadata !{metadata !"zero"}
982 !1 = metadata !{metadata !"one"}
983 !2 = metadata !{metadata !"two"}
985 !name = !{!0, !1, !2}
990 <!-- ======================================================================= -->
992 <a name="paramattrs">Parameter Attributes</a>
997 <p>The return type and each parameter of a function type may have a set of
998 <i>parameter attributes</i> associated with them. Parameter attributes are
999 used to communicate additional information about the result or parameters of
1000 a function. Parameter attributes are considered to be part of the function,
1001 not of the function type, so functions with different parameter attributes
1002 can have the same function type.</p>
1004 <p>Parameter attributes are simple keywords that follow the type specified. If
1005 multiple parameter attributes are needed, they are space separated. For
1008 <pre class="doc_code">
1009 declare i32 @printf(i8* noalias nocapture, ...)
1010 declare i32 @atoi(i8 zeroext)
1011 declare signext i8 @returns_signed_char()
1014 <p>Note that any attributes for the function result (<tt>nounwind</tt>,
1015 <tt>readonly</tt>) come immediately after the argument list.</p>
1017 <p>Currently, only the following parameter attributes are defined:</p>
1020 <dt><tt><b>zeroext</b></tt></dt>
1021 <dd>This indicates to the code generator that the parameter or return value
1022 should be zero-extended to the extent required by the target's ABI (which
1023 is usually 32-bits, but is 8-bits for a i1 on x86-64) by the caller (for a
1024 parameter) or the callee (for a return value).</dd>
1026 <dt><tt><b>signext</b></tt></dt>
1027 <dd>This indicates to the code generator that the parameter or return value
1028 should be sign-extended to the extent required by the target's ABI (which
1029 is usually 32-bits) by the caller (for a parameter) or the callee (for a
1032 <dt><tt><b>inreg</b></tt></dt>
1033 <dd>This indicates that this parameter or return value should be treated in a
1034 special target-dependent fashion during while emitting code for a function
1035 call or return (usually, by putting it in a register as opposed to memory,
1036 though some targets use it to distinguish between two different kinds of
1037 registers). Use of this attribute is target-specific.</dd>
1039 <dt><tt><b><a name="byval">byval</a></b></tt></dt>
1040 <dd><p>This indicates that the pointer parameter should really be passed by
1041 value to the function. The attribute implies that a hidden copy of the
1043 is made between the caller and the callee, so the callee is unable to
1044 modify the value in the callee. This attribute is only valid on LLVM
1045 pointer arguments. It is generally used to pass structs and arrays by
1046 value, but is also valid on pointers to scalars. The copy is considered
1047 to belong to the caller not the callee (for example,
1048 <tt><a href="#readonly">readonly</a></tt> functions should not write to
1049 <tt>byval</tt> parameters). This is not a valid attribute for return
1052 <p>The byval attribute also supports specifying an alignment with
1053 the align attribute. It indicates the alignment of the stack slot to
1054 form and the known alignment of the pointer specified to the call site. If
1055 the alignment is not specified, then the code generator makes a
1056 target-specific assumption.</p></dd>
1058 <dt><tt><b><a name="sret">sret</a></b></tt></dt>
1059 <dd>This indicates that the pointer parameter specifies the address of a
1060 structure that is the return value of the function in the source program.
1061 This pointer must be guaranteed by the caller to be valid: loads and
1062 stores to the structure may be assumed by the callee to not to trap. This
1063 may only be applied to the first parameter. This is not a valid attribute
1064 for return values. </dd>
1066 <dt><tt><b><a name="noalias">noalias</a></b></tt></dt>
1067 <dd>This indicates that pointer values
1068 <a href="#pointeraliasing"><i>based</i></a> on the argument or return
1069 value do not alias pointer values which are not <i>based</i> on it,
1070 ignoring certain "irrelevant" dependencies.
1071 For a call to the parent function, dependencies between memory
1072 references from before or after the call and from those during the call
1073 are "irrelevant" to the <tt>noalias</tt> keyword for the arguments and
1074 return value used in that call.
1075 The caller shares the responsibility with the callee for ensuring that
1076 these requirements are met.
1077 For further details, please see the discussion of the NoAlias response in
1078 <a href="AliasAnalysis.html#MustMayNo">alias analysis</a>.<br>
1080 Note that this definition of <tt>noalias</tt> is intentionally
1081 similar to the definition of <tt>restrict</tt> in C99 for function
1082 arguments, though it is slightly weaker.
1084 For function return values, C99's <tt>restrict</tt> is not meaningful,
1085 while LLVM's <tt>noalias</tt> is.
1088 <dt><tt><b><a name="nocapture">nocapture</a></b></tt></dt>
1089 <dd>This indicates that the callee does not make any copies of the pointer
1090 that outlive the callee itself. This is not a valid attribute for return
1093 <dt><tt><b><a name="nest">nest</a></b></tt></dt>
1094 <dd>This indicates that the pointer parameter can be excised using the
1095 <a href="#int_trampoline">trampoline intrinsics</a>. This is not a valid
1096 attribute for return values.</dd>
1101 <!-- ======================================================================= -->
1103 <a name="gc">Garbage Collector Names</a>
1108 <p>Each function may specify a garbage collector name, which is simply a
1111 <pre class="doc_code">
1112 define void @f() gc "name" { ... }
1115 <p>The compiler declares the supported values of <i>name</i>. Specifying a
1116 collector which will cause the compiler to alter its output in order to
1117 support the named garbage collection algorithm.</p>
1121 <!-- ======================================================================= -->
1123 <a name="fnattrs">Function Attributes</a>
1128 <p>Function attributes are set to communicate additional information about a
1129 function. Function attributes are considered to be part of the function, not
1130 of the function type, so functions with different parameter attributes can
1131 have the same function type.</p>
1133 <p>Function attributes are simple keywords that follow the type specified. If
1134 multiple attributes are needed, they are space separated. For example:</p>
1136 <pre class="doc_code">
1137 define void @f() noinline { ... }
1138 define void @f() alwaysinline { ... }
1139 define void @f() alwaysinline optsize { ... }
1140 define void @f() optsize { ... }
1144 <dt><tt><b>alignstack(<<em>n</em>>)</b></tt></dt>
1145 <dd>This attribute indicates that, when emitting the prologue and epilogue,
1146 the backend should forcibly align the stack pointer. Specify the
1147 desired alignment, which must be a power of two, in parentheses.
1149 <dt><tt><b>alwaysinline</b></tt></dt>
1150 <dd>This attribute indicates that the inliner should attempt to inline this
1151 function into callers whenever possible, ignoring any active inlining size
1152 threshold for this caller.</dd>
1154 <dt><tt><b>nonlazybind</b></tt></dt>
1155 <dd>This attribute suppresses lazy symbol binding for the function. This
1156 may make calls to the function faster, at the cost of extra program
1157 startup time if the function is not called during program startup.</dd>
1159 <dt><tt><b>inlinehint</b></tt></dt>
1160 <dd>This attribute indicates that the source code contained a hint that inlining
1161 this function is desirable (such as the "inline" keyword in C/C++). It
1162 is just a hint; it imposes no requirements on the inliner.</dd>
1164 <dt><tt><b>naked</b></tt></dt>
1165 <dd>This attribute disables prologue / epilogue emission for the function.
1166 This can have very system-specific consequences.</dd>
1168 <dt><tt><b>noimplicitfloat</b></tt></dt>
1169 <dd>This attributes disables implicit floating point instructions.</dd>
1171 <dt><tt><b>noinline</b></tt></dt>
1172 <dd>This attribute indicates that the inliner should never inline this
1173 function in any situation. This attribute may not be used together with
1174 the <tt>alwaysinline</tt> attribute.</dd>
1176 <dt><tt><b>noredzone</b></tt></dt>
1177 <dd>This attribute indicates that the code generator should not use a red
1178 zone, even if the target-specific ABI normally permits it.</dd>
1180 <dt><tt><b>noreturn</b></tt></dt>
1181 <dd>This function attribute indicates that the function never returns
1182 normally. This produces undefined behavior at runtime if the function
1183 ever does dynamically return.</dd>
1185 <dt><tt><b>nounwind</b></tt></dt>
1186 <dd>This function attribute indicates that the function never returns with an
1187 unwind or exceptional control flow. If the function does unwind, its
1188 runtime behavior is undefined.</dd>
1190 <dt><tt><b>optsize</b></tt></dt>
1191 <dd>This attribute suggests that optimization passes and code generator passes
1192 make choices that keep the code size of this function low, and otherwise
1193 do optimizations specifically to reduce code size.</dd>
1195 <dt><tt><b>readnone</b></tt></dt>
1196 <dd>This attribute indicates that the function computes its result (or decides
1197 to unwind an exception) based strictly on its arguments, without
1198 dereferencing any pointer arguments or otherwise accessing any mutable
1199 state (e.g. memory, control registers, etc) visible to caller functions.
1200 It does not write through any pointer arguments
1201 (including <tt><a href="#byval">byval</a></tt> arguments) and never
1202 changes any state visible to callers. This means that it cannot unwind
1203 exceptions by calling the <tt>C++</tt> exception throwing methods, but
1204 could use the <tt>unwind</tt> instruction.</dd>
1206 <dt><tt><b><a name="readonly">readonly</a></b></tt></dt>
1207 <dd>This attribute indicates that the function does not write through any
1208 pointer arguments (including <tt><a href="#byval">byval</a></tt>
1209 arguments) or otherwise modify any state (e.g. memory, control registers,
1210 etc) visible to caller functions. It may dereference pointer arguments
1211 and read state that may be set in the caller. A readonly function always
1212 returns the same value (or unwinds an exception identically) when called
1213 with the same set of arguments and global state. It cannot unwind an
1214 exception by calling the <tt>C++</tt> exception throwing methods, but may
1215 use the <tt>unwind</tt> instruction.</dd>
1217 <dt><tt><b><a name="ssp">ssp</a></b></tt></dt>
1218 <dd>This attribute indicates that the function should emit a stack smashing
1219 protector. It is in the form of a "canary"—a random value placed on
1220 the stack before the local variables that's checked upon return from the
1221 function to see if it has been overwritten. A heuristic is used to
1222 determine if a function needs stack protectors or not.<br>
1224 If a function that has an <tt>ssp</tt> attribute is inlined into a
1225 function that doesn't have an <tt>ssp</tt> attribute, then the resulting
1226 function will have an <tt>ssp</tt> attribute.</dd>
1228 <dt><tt><b>sspreq</b></tt></dt>
1229 <dd>This attribute indicates that the function should <em>always</em> emit a
1230 stack smashing protector. This overrides
1231 the <tt><a href="#ssp">ssp</a></tt> function attribute.<br>
1233 If a function that has an <tt>sspreq</tt> attribute is inlined into a
1234 function that doesn't have an <tt>sspreq</tt> attribute or which has
1235 an <tt>ssp</tt> attribute, then the resulting function will have
1236 an <tt>sspreq</tt> attribute.</dd>
1238 <dt><tt><b><a name="uwtable">uwtable</a></b></tt></dt>
1239 <dd>This attribute indicates that the ABI being targeted requires that
1240 an unwind table entry be produce for this function even if we can
1241 show that no exceptions passes by it. This is normally the case for
1242 the ELF x86-64 abi, but it can be disabled for some compilation
1245 <dt><tt><b><a name="returns_twice">returns_twice</a></b></tt></dt>
1246 <dd>This attribute indicates that this function can return
1247 twice. The C <code>setjmp</code> is an example of such a function.
1248 The compiler disables some optimizations (like tail calls) in the caller of
1249 these functions.</dd>
1254 <!-- ======================================================================= -->
1256 <a name="moduleasm">Module-Level Inline Assembly</a>
1261 <p>Modules may contain "module-level inline asm" blocks, which corresponds to
1262 the GCC "file scope inline asm" blocks. These blocks are internally
1263 concatenated by LLVM and treated as a single unit, but may be separated in
1264 the <tt>.ll</tt> file if desired. The syntax is very simple:</p>
1266 <pre class="doc_code">
1267 module asm "inline asm code goes here"
1268 module asm "more can go here"
1271 <p>The strings can contain any character by escaping non-printable characters.
1272 The escape sequence used is simply "\xx" where "xx" is the two digit hex code
1275 <p>The inline asm code is simply printed to the machine code .s file when
1276 assembly code is generated.</p>
1280 <!-- ======================================================================= -->
1282 <a name="datalayout">Data Layout</a>
1287 <p>A module may specify a target specific data layout string that specifies how
1288 data is to be laid out in memory. The syntax for the data layout is
1291 <pre class="doc_code">
1292 target datalayout = "<i>layout specification</i>"
1295 <p>The <i>layout specification</i> consists of a list of specifications
1296 separated by the minus sign character ('-'). Each specification starts with
1297 a letter and may include other information after the letter to define some
1298 aspect of the data layout. The specifications accepted are as follows:</p>
1302 <dd>Specifies that the target lays out data in big-endian form. That is, the
1303 bits with the most significance have the lowest address location.</dd>
1306 <dd>Specifies that the target lays out data in little-endian form. That is,
1307 the bits with the least significance have the lowest address
1310 <dt><tt>S<i>size</i></tt></dt>
1311 <dd>Specifies the natural alignment of the stack in bits. Alignment promotion
1312 of stack variables is limited to the natural stack alignment to avoid
1313 dynamic stack realignment. The stack alignment must be a multiple of
1314 8-bits. If omitted, the natural stack alignment defaults to "unspecified",
1315 which does not prevent any alignment promotions.</dd>
1317 <dt><tt>p:<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1318 <dd>This specifies the <i>size</i> of a pointer and its <i>abi</i> and
1319 <i>preferred</i> alignments. All sizes are in bits. Specifying
1320 the <i>pref</i> alignment is optional. If omitted, the
1321 preceding <tt>:</tt> should be omitted too.</dd>
1323 <dt><tt>i<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1324 <dd>This specifies the alignment for an integer type of a given bit
1325 <i>size</i>. The value of <i>size</i> must be in the range [1,2^23).</dd>
1327 <dt><tt>v<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1328 <dd>This specifies the alignment for a vector type of a given bit
1331 <dt><tt>f<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1332 <dd>This specifies the alignment for a floating point type of a given bit
1333 <i>size</i>. Only values of <i>size</i> that are supported by the target
1334 will work. 32 (float) and 64 (double) are supported on all targets;
1335 80 or 128 (different flavors of long double) are also supported on some
1338 <dt><tt>a<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1339 <dd>This specifies the alignment for an aggregate type of a given bit
1342 <dt><tt>s<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1343 <dd>This specifies the alignment for a stack object of a given bit
1346 <dt><tt>n<i>size1</i>:<i>size2</i>:<i>size3</i>...</tt></dt>
1347 <dd>This specifies a set of native integer widths for the target CPU
1348 in bits. For example, it might contain "n32" for 32-bit PowerPC,
1349 "n32:64" for PowerPC 64, or "n8:16:32:64" for X86-64. Elements of
1350 this set are considered to support most general arithmetic
1351 operations efficiently.</dd>
1354 <p>When constructing the data layout for a given target, LLVM starts with a
1355 default set of specifications which are then (possibly) overridden by the
1356 specifications in the <tt>datalayout</tt> keyword. The default specifications
1357 are given in this list:</p>
1360 <li><tt>E</tt> - big endian</li>
1361 <li><tt>p:64:64:64</tt> - 64-bit pointers with 64-bit alignment</li>
1362 <li><tt>i1:8:8</tt> - i1 is 8-bit (byte) aligned</li>
1363 <li><tt>i8:8:8</tt> - i8 is 8-bit (byte) aligned</li>
1364 <li><tt>i16:16:16</tt> - i16 is 16-bit aligned</li>
1365 <li><tt>i32:32:32</tt> - i32 is 32-bit aligned</li>
1366 <li><tt>i64:32:64</tt> - i64 has ABI alignment of 32-bits but preferred
1367 alignment of 64-bits</li>
1368 <li><tt>f32:32:32</tt> - float is 32-bit aligned</li>
1369 <li><tt>f64:64:64</tt> - double is 64-bit aligned</li>
1370 <li><tt>v64:64:64</tt> - 64-bit vector is 64-bit aligned</li>
1371 <li><tt>v128:128:128</tt> - 128-bit vector is 128-bit aligned</li>
1372 <li><tt>a0:0:1</tt> - aggregates are 8-bit aligned</li>
1373 <li><tt>s0:64:64</tt> - stack objects are 64-bit aligned</li>
1376 <p>When LLVM is determining the alignment for a given type, it uses the
1377 following rules:</p>
1380 <li>If the type sought is an exact match for one of the specifications, that
1381 specification is used.</li>
1383 <li>If no match is found, and the type sought is an integer type, then the
1384 smallest integer type that is larger than the bitwidth of the sought type
1385 is used. If none of the specifications are larger than the bitwidth then
1386 the the largest integer type is used. For example, given the default
1387 specifications above, the i7 type will use the alignment of i8 (next
1388 largest) while both i65 and i256 will use the alignment of i64 (largest
1391 <li>If no match is found, and the type sought is a vector type, then the
1392 largest vector type that is smaller than the sought vector type will be
1393 used as a fall back. This happens because <128 x double> can be
1394 implemented in terms of 64 <2 x double>, for example.</li>
1397 <p>The function of the data layout string may not be what you expect. Notably,
1398 this is not a specification from the frontend of what alignment the code
1399 generator should use.</p>
1401 <p>Instead, if specified, the target data layout is required to match what the
1402 ultimate <em>code generator</em> expects. This string is used by the
1403 mid-level optimizers to
1404 improve code, and this only works if it matches what the ultimate code
1405 generator uses. If you would like to generate IR that does not embed this
1406 target-specific detail into the IR, then you don't have to specify the
1407 string. This will disable some optimizations that require precise layout
1408 information, but this also prevents those optimizations from introducing
1409 target specificity into the IR.</p>
1415 <!-- ======================================================================= -->
1417 <a name="pointeraliasing">Pointer Aliasing Rules</a>
1422 <p>Any memory access must be done through a pointer value associated
1423 with an address range of the memory access, otherwise the behavior
1424 is undefined. Pointer values are associated with address ranges
1425 according to the following rules:</p>
1428 <li>A pointer value is associated with the addresses associated with
1429 any value it is <i>based</i> on.
1430 <li>An address of a global variable is associated with the address
1431 range of the variable's storage.</li>
1432 <li>The result value of an allocation instruction is associated with
1433 the address range of the allocated storage.</li>
1434 <li>A null pointer in the default address-space is associated with
1436 <li>An integer constant other than zero or a pointer value returned
1437 from a function not defined within LLVM may be associated with address
1438 ranges allocated through mechanisms other than those provided by
1439 LLVM. Such ranges shall not overlap with any ranges of addresses
1440 allocated by mechanisms provided by LLVM.</li>
1443 <p>A pointer value is <i>based</i> on another pointer value according
1444 to the following rules:</p>
1447 <li>A pointer value formed from a
1448 <tt><a href="#i_getelementptr">getelementptr</a></tt> operation
1449 is <i>based</i> on the first operand of the <tt>getelementptr</tt>.</li>
1450 <li>The result value of a
1451 <tt><a href="#i_bitcast">bitcast</a></tt> is <i>based</i> on the operand
1452 of the <tt>bitcast</tt>.</li>
1453 <li>A pointer value formed by an
1454 <tt><a href="#i_inttoptr">inttoptr</a></tt> is <i>based</i> on all
1455 pointer values that contribute (directly or indirectly) to the
1456 computation of the pointer's value.</li>
1457 <li>The "<i>based</i> on" relationship is transitive.</li>
1460 <p>Note that this definition of <i>"based"</i> is intentionally
1461 similar to the definition of <i>"based"</i> in C99, though it is
1462 slightly weaker.</p>
1464 <p>LLVM IR does not associate types with memory. The result type of a
1465 <tt><a href="#i_load">load</a></tt> merely indicates the size and
1466 alignment of the memory from which to load, as well as the
1467 interpretation of the value. The first operand type of a
1468 <tt><a href="#i_store">store</a></tt> similarly only indicates the size
1469 and alignment of the store.</p>
1471 <p>Consequently, type-based alias analysis, aka TBAA, aka
1472 <tt>-fstrict-aliasing</tt>, is not applicable to general unadorned
1473 LLVM IR. <a href="#metadata">Metadata</a> may be used to encode
1474 additional information which specialized optimization passes may use
1475 to implement type-based alias analysis.</p>
1479 <!-- ======================================================================= -->
1481 <a name="volatile">Volatile Memory Accesses</a>
1486 <p>Certain memory accesses, such as <a href="#i_load"><tt>load</tt></a>s, <a
1487 href="#i_store"><tt>store</tt></a>s, and <a
1488 href="#int_memcpy"><tt>llvm.memcpy</tt></a>s may be marked <tt>volatile</tt>.
1489 The optimizers must not change the number of volatile operations or change their
1490 order of execution relative to other volatile operations. The optimizers
1491 <i>may</i> change the order of volatile operations relative to non-volatile
1492 operations. This is not Java's "volatile" and has no cross-thread
1493 synchronization behavior.</p>
1497 <!-- ======================================================================= -->
1499 <a name="memmodel">Memory Model for Concurrent Operations</a>
1504 <p>The LLVM IR does not define any way to start parallel threads of execution
1505 or to register signal handlers. Nonetheless, there are platform-specific
1506 ways to create them, and we define LLVM IR's behavior in their presence. This
1507 model is inspired by the C++0x memory model.</p>
1509 <p>For a more informal introduction to this model, see the
1510 <a href="Atomics.html">LLVM Atomic Instructions and Concurrency Guide</a>.
1512 <p>We define a <i>happens-before</i> partial order as the least partial order
1515 <li>Is a superset of single-thread program order, and</li>
1516 <li>When a <i>synchronizes-with</i> <tt>b</tt>, includes an edge from
1517 <tt>a</tt> to <tt>b</tt>. <i>Synchronizes-with</i> pairs are introduced
1518 by platform-specific techniques, like pthread locks, thread
1519 creation, thread joining, etc., and by atomic instructions.
1520 (See also <a href="#ordering">Atomic Memory Ordering Constraints</a>).
1524 <p>Note that program order does not introduce <i>happens-before</i> edges
1525 between a thread and signals executing inside that thread.</p>
1527 <p>Every (defined) read operation (load instructions, memcpy, atomic
1528 loads/read-modify-writes, etc.) <var>R</var> reads a series of bytes written by
1529 (defined) write operations (store instructions, atomic
1530 stores/read-modify-writes, memcpy, etc.). For the purposes of this section,
1531 initialized globals are considered to have a write of the initializer which is
1532 atomic and happens before any other read or write of the memory in question.
1533 For each byte of a read <var>R</var>, <var>R<sub>byte</sub></var> may see
1534 any write to the same byte, except:</p>
1537 <li>If <var>write<sub>1</sub></var> happens before
1538 <var>write<sub>2</sub></var>, and <var>write<sub>2</sub></var> happens
1539 before <var>R<sub>byte</sub></var>, then <var>R<sub>byte</sub></var>
1540 does not see <var>write<sub>1</sub></var>.
1541 <li>If <var>R<sub>byte</sub></var> happens before
1542 <var>write<sub>3</sub></var>, then <var>R<sub>byte</sub></var> does not
1543 see <var>write<sub>3</sub></var>.
1546 <p>Given that definition, <var>R<sub>byte</sub></var> is defined as follows:
1548 <li>If <var>R</var> is volatile, the result is target-dependent. (Volatile
1549 is supposed to give guarantees which can support
1550 <code>sig_atomic_t</code> in C/C++, and may be used for accesses to
1551 addresses which do not behave like normal memory. It does not generally
1552 provide cross-thread synchronization.)
1553 <li>Otherwise, if there is no write to the same byte that happens before
1554 <var>R<sub>byte</sub></var>, <var>R<sub>byte</sub></var> returns
1555 <tt>undef</tt> for that byte.
1556 <li>Otherwise, if <var>R<sub>byte</sub></var> may see exactly one write,
1557 <var>R<sub>byte</sub></var> returns the value written by that
1559 <li>Otherwise, if <var>R</var> is atomic, and all the writes
1560 <var>R<sub>byte</sub></var> may see are atomic, it chooses one of the
1561 values written. See the <a href="#ordering">Atomic Memory Ordering
1562 Constraints</a> section for additional constraints on how the choice
1564 <li>Otherwise <var>R<sub>byte</sub></var> returns <tt>undef</tt>.</li>
1567 <p><var>R</var> returns the value composed of the series of bytes it read.
1568 This implies that some bytes within the value may be <tt>undef</tt>
1569 <b>without</b> the entire value being <tt>undef</tt>. Note that this only
1570 defines the semantics of the operation; it doesn't mean that targets will
1571 emit more than one instruction to read the series of bytes.</p>
1573 <p>Note that in cases where none of the atomic intrinsics are used, this model
1574 places only one restriction on IR transformations on top of what is required
1575 for single-threaded execution: introducing a store to a byte which might not
1576 otherwise be stored is not allowed in general. (Specifically, in the case
1577 where another thread might write to and read from an address, introducing a
1578 store can change a load that may see exactly one write into a load that may
1579 see multiple writes.)</p>
1581 <!-- FIXME: This model assumes all targets where concurrency is relevant have
1582 a byte-size store which doesn't affect adjacent bytes. As far as I can tell,
1583 none of the backends currently in the tree fall into this category; however,
1584 there might be targets which care. If there are, we want a paragraph
1587 Targets may specify that stores narrower than a certain width are not
1588 available; on such a target, for the purposes of this model, treat any
1589 non-atomic write with an alignment or width less than the minimum width
1590 as if it writes to the relevant surrounding bytes.
1595 <!-- ======================================================================= -->
1597 <a name="ordering">Atomic Memory Ordering Constraints</a>
1602 <p>Atomic instructions (<a href="#i_cmpxchg"><code>cmpxchg</code></a>,
1603 <a href="#i_atomicrmw"><code>atomicrmw</code></a>,
1604 <a href="#i_fence"><code>fence</code></a>,
1605 <a href="#i_load"><code>atomic load</code></a>, and
1606 <a href="#i_store"><code>atomic store</code></a>) take an ordering parameter
1607 that determines which other atomic instructions on the same address they
1608 <i>synchronize with</i>. These semantics are borrowed from Java and C++0x,
1609 but are somewhat more colloquial. If these descriptions aren't precise enough,
1610 check those specs (see spec references in the
1611 <a href="Atomic.html#introduction">atomics guide</a>).
1612 <a href="#i_fence"><code>fence</code></a> instructions
1613 treat these orderings somewhat differently since they don't take an address.
1614 See that instruction's documentation for details.</p>
1616 <p>For a simpler introduction to the ordering constraints, see the
1617 <a href="Atomics.html">LLVM Atomic Instructions and Concurrency Guide</a>.</p>
1620 <dt><code>unordered</code></dt>
1621 <dd>The set of values that can be read is governed by the happens-before
1622 partial order. A value cannot be read unless some operation wrote it.
1623 This is intended to provide a guarantee strong enough to model Java's
1624 non-volatile shared variables. This ordering cannot be specified for
1625 read-modify-write operations; it is not strong enough to make them atomic
1626 in any interesting way.</dd>
1627 <dt><code>monotonic</code></dt>
1628 <dd>In addition to the guarantees of <code>unordered</code>, there is a single
1629 total order for modifications by <code>monotonic</code> operations on each
1630 address. All modification orders must be compatible with the happens-before
1631 order. There is no guarantee that the modification orders can be combined to
1632 a global total order for the whole program (and this often will not be
1633 possible). The read in an atomic read-modify-write operation
1634 (<a href="#i_cmpxchg"><code>cmpxchg</code></a> and
1635 <a href="#i_atomicrmw"><code>atomicrmw</code></a>)
1636 reads the value in the modification order immediately before the value it
1637 writes. If one atomic read happens before another atomic read of the same
1638 address, the later read must see the same value or a later value in the
1639 address's modification order. This disallows reordering of
1640 <code>monotonic</code> (or stronger) operations on the same address. If an
1641 address is written <code>monotonic</code>ally by one thread, and other threads
1642 <code>monotonic</code>ally read that address repeatedly, the other threads must
1643 eventually see the write. This corresponds to the C++0x/C1x
1644 <code>memory_order_relaxed</code>.</dd>
1645 <dt><code>acquire</code></dt>
1646 <dd>In addition to the guarantees of <code>monotonic</code>,
1647 a <i>synchronizes-with</i> edge may be formed with a <code>release</code>
1648 operation. This is intended to model C++'s <code>memory_order_acquire</code>.</dd>
1649 <dt><code>release</code></dt>
1650 <dd>In addition to the guarantees of <code>monotonic</code>, if this operation
1651 writes a value which is subsequently read by an <code>acquire</code> operation,
1652 it <i>synchronizes-with</i> that operation. (This isn't a complete
1653 description; see the C++0x definition of a release sequence.) This corresponds
1654 to the C++0x/C1x <code>memory_order_release</code>.</dd>
1655 <dt><code>acq_rel</code> (acquire+release)</dt><dd>Acts as both an
1656 <code>acquire</code> and <code>release</code> operation on its address.
1657 This corresponds to the C++0x/C1x <code>memory_order_acq_rel</code>.</dd>
1658 <dt><code>seq_cst</code> (sequentially consistent)</dt><dd>
1659 <dd>In addition to the guarantees of <code>acq_rel</code>
1660 (<code>acquire</code> for an operation which only reads, <code>release</code>
1661 for an operation which only writes), there is a global total order on all
1662 sequentially-consistent operations on all addresses, which is consistent with
1663 the <i>happens-before</i> partial order and with the modification orders of
1664 all the affected addresses. Each sequentially-consistent read sees the last
1665 preceding write to the same address in this global order. This corresponds
1666 to the C++0x/C1x <code>memory_order_seq_cst</code> and Java volatile.</dd>
1669 <p id="singlethread">If an atomic operation is marked <code>singlethread</code>,
1670 it only <i>synchronizes with</i> or participates in modification and seq_cst
1671 total orderings with other operations running in the same thread (for example,
1672 in signal handlers).</p>
1678 <!-- *********************************************************************** -->
1679 <h2><a name="typesystem">Type System</a></h2>
1680 <!-- *********************************************************************** -->
1684 <p>The LLVM type system is one of the most important features of the
1685 intermediate representation. Being typed enables a number of optimizations
1686 to be performed on the intermediate representation directly, without having
1687 to do extra analyses on the side before the transformation. A strong type
1688 system makes it easier to read the generated code and enables novel analyses
1689 and transformations that are not feasible to perform on normal three address
1690 code representations.</p>
1692 <!-- ======================================================================= -->
1694 <a name="t_classifications">Type Classifications</a>
1699 <p>The types fall into a few useful classifications:</p>
1701 <table border="1" cellspacing="0" cellpadding="4">
1703 <tr><th>Classification</th><th>Types</th></tr>
1705 <td><a href="#t_integer">integer</a></td>
1706 <td><tt>i1, i2, i3, ... i8, ... i16, ... i32, ... i64, ... </tt></td>
1709 <td><a href="#t_floating">floating point</a></td>
1710 <td><tt>float, double, x86_fp80, fp128, ppc_fp128</tt></td>
1713 <td><a name="t_firstclass">first class</a></td>
1714 <td><a href="#t_integer">integer</a>,
1715 <a href="#t_floating">floating point</a>,
1716 <a href="#t_pointer">pointer</a>,
1717 <a href="#t_vector">vector</a>,
1718 <a href="#t_struct">structure</a>,
1719 <a href="#t_array">array</a>,
1720 <a href="#t_label">label</a>,
1721 <a href="#t_metadata">metadata</a>.
1725 <td><a href="#t_primitive">primitive</a></td>
1726 <td><a href="#t_label">label</a>,
1727 <a href="#t_void">void</a>,
1728 <a href="#t_integer">integer</a>,
1729 <a href="#t_floating">floating point</a>,
1730 <a href="#t_x86mmx">x86mmx</a>,
1731 <a href="#t_metadata">metadata</a>.</td>
1734 <td><a href="#t_derived">derived</a></td>
1735 <td><a href="#t_array">array</a>,
1736 <a href="#t_function">function</a>,
1737 <a href="#t_pointer">pointer</a>,
1738 <a href="#t_struct">structure</a>,
1739 <a href="#t_vector">vector</a>,
1740 <a href="#t_opaque">opaque</a>.
1746 <p>The <a href="#t_firstclass">first class</a> types are perhaps the most
1747 important. Values of these types are the only ones which can be produced by
1752 <!-- ======================================================================= -->
1754 <a name="t_primitive">Primitive Types</a>
1759 <p>The primitive types are the fundamental building blocks of the LLVM
1762 <!-- _______________________________________________________________________ -->
1764 <a name="t_integer">Integer Type</a>
1770 <p>The integer type is a very simple type that simply specifies an arbitrary
1771 bit width for the integer type desired. Any bit width from 1 bit to
1772 2<sup>23</sup>-1 (about 8 million) can be specified.</p>
1779 <p>The number of bits the integer will occupy is specified by the <tt>N</tt>
1783 <table class="layout">
1785 <td class="left"><tt>i1</tt></td>
1786 <td class="left">a single-bit integer.</td>
1789 <td class="left"><tt>i32</tt></td>
1790 <td class="left">a 32-bit integer.</td>
1793 <td class="left"><tt>i1942652</tt></td>
1794 <td class="left">a really big integer of over 1 million bits.</td>
1800 <!-- _______________________________________________________________________ -->
1802 <a name="t_floating">Floating Point Types</a>
1809 <tr><th>Type</th><th>Description</th></tr>
1810 <tr><td><tt>float</tt></td><td>32-bit floating point value</td></tr>
1811 <tr><td><tt>double</tt></td><td>64-bit floating point value</td></tr>
1812 <tr><td><tt>fp128</tt></td><td>128-bit floating point value (112-bit mantissa)</td></tr>
1813 <tr><td><tt>x86_fp80</tt></td><td>80-bit floating point value (X87)</td></tr>
1814 <tr><td><tt>ppc_fp128</tt></td><td>128-bit floating point value (two 64-bits)</td></tr>
1820 <!-- _______________________________________________________________________ -->
1822 <a name="t_x86mmx">X86mmx Type</a>
1828 <p>The x86mmx type represents a value held in an MMX register on an x86 machine. The operations allowed on it are quite limited: parameters and return values, load and store, and bitcast. User-specified MMX instructions are represented as intrinsic or asm calls with arguments and/or results of this type. There are no arrays, vectors or constants of this type.</p>
1837 <!-- _______________________________________________________________________ -->
1839 <a name="t_void">Void Type</a>
1845 <p>The void type does not represent any value and has no size.</p>
1854 <!-- _______________________________________________________________________ -->
1856 <a name="t_label">Label Type</a>
1862 <p>The label type represents code labels.</p>
1871 <!-- _______________________________________________________________________ -->
1873 <a name="t_metadata">Metadata Type</a>
1879 <p>The metadata type represents embedded metadata. No derived types may be
1880 created from metadata except for <a href="#t_function">function</a>
1892 <!-- ======================================================================= -->
1894 <a name="t_derived">Derived Types</a>
1899 <p>The real power in LLVM comes from the derived types in the system. This is
1900 what allows a programmer to represent arrays, functions, pointers, and other
1901 useful types. Each of these types contain one or more element types which
1902 may be a primitive type, or another derived type. For example, it is
1903 possible to have a two dimensional array, using an array as the element type
1904 of another array.</p>
1906 <!-- _______________________________________________________________________ -->
1908 <a name="t_aggregate">Aggregate Types</a>
1913 <p>Aggregate Types are a subset of derived types that can contain multiple
1914 member types. <a href="#t_array">Arrays</a>,
1915 <a href="#t_struct">structs</a>, and <a href="#t_vector">vectors</a> are
1916 aggregate types.</p>
1920 <!-- _______________________________________________________________________ -->
1922 <a name="t_array">Array Type</a>
1928 <p>The array type is a very simple derived type that arranges elements
1929 sequentially in memory. The array type requires a size (number of elements)
1930 and an underlying data type.</p>
1934 [<# elements> x <elementtype>]
1937 <p>The number of elements is a constant integer value; <tt>elementtype</tt> may
1938 be any type with a size.</p>
1941 <table class="layout">
1943 <td class="left"><tt>[40 x i32]</tt></td>
1944 <td class="left">Array of 40 32-bit integer values.</td>
1947 <td class="left"><tt>[41 x i32]</tt></td>
1948 <td class="left">Array of 41 32-bit integer values.</td>
1951 <td class="left"><tt>[4 x i8]</tt></td>
1952 <td class="left">Array of 4 8-bit integer values.</td>
1955 <p>Here are some examples of multidimensional arrays:</p>
1956 <table class="layout">
1958 <td class="left"><tt>[3 x [4 x i32]]</tt></td>
1959 <td class="left">3x4 array of 32-bit integer values.</td>
1962 <td class="left"><tt>[12 x [10 x float]]</tt></td>
1963 <td class="left">12x10 array of single precision floating point values.</td>
1966 <td class="left"><tt>[2 x [3 x [4 x i16]]]</tt></td>
1967 <td class="left">2x3x4 array of 16-bit integer values.</td>
1971 <p>There is no restriction on indexing beyond the end of the array implied by
1972 a static type (though there are restrictions on indexing beyond the bounds
1973 of an allocated object in some cases). This means that single-dimension
1974 'variable sized array' addressing can be implemented in LLVM with a zero
1975 length array type. An implementation of 'pascal style arrays' in LLVM could
1976 use the type "<tt>{ i32, [0 x float]}</tt>", for example.</p>
1980 <!-- _______________________________________________________________________ -->
1982 <a name="t_function">Function Type</a>
1988 <p>The function type can be thought of as a function signature. It consists of
1989 a return type and a list of formal parameter types. The return type of a
1990 function type is a first class type or a void type.</p>
1994 <returntype> (<parameter list>)
1997 <p>...where '<tt><parameter list></tt>' is a comma-separated list of type
1998 specifiers. Optionally, the parameter list may include a type <tt>...</tt>,
1999 which indicates that the function takes a variable number of arguments.
2000 Variable argument functions can access their arguments with
2001 the <a href="#int_varargs">variable argument handling intrinsic</a>
2002 functions. '<tt><returntype></tt>' is any type except
2003 <a href="#t_label">label</a>.</p>
2006 <table class="layout">
2008 <td class="left"><tt>i32 (i32)</tt></td>
2009 <td class="left">function taking an <tt>i32</tt>, returning an <tt>i32</tt>
2011 </tr><tr class="layout">
2012 <td class="left"><tt>float (i16, i32 *) *
2014 <td class="left"><a href="#t_pointer">Pointer</a> to a function that takes
2015 an <tt>i16</tt> and a <a href="#t_pointer">pointer</a> to <tt>i32</tt>,
2016 returning <tt>float</tt>.
2018 </tr><tr class="layout">
2019 <td class="left"><tt>i32 (i8*, ...)</tt></td>
2020 <td class="left">A vararg function that takes at least one
2021 <a href="#t_pointer">pointer</a> to <tt>i8 </tt> (char in C),
2022 which returns an integer. This is the signature for <tt>printf</tt> in
2025 </tr><tr class="layout">
2026 <td class="left"><tt>{i32, i32} (i32)</tt></td>
2027 <td class="left">A function taking an <tt>i32</tt>, returning a
2028 <a href="#t_struct">structure</a> containing two <tt>i32</tt> values
2035 <!-- _______________________________________________________________________ -->
2037 <a name="t_struct">Structure Type</a>
2043 <p>The structure type is used to represent a collection of data members together
2044 in memory. The elements of a structure may be any type that has a size.</p>
2046 <p>Structures in memory are accessed using '<tt><a href="#i_load">load</a></tt>'
2047 and '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a field
2048 with the '<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.
2049 Structures in registers are accessed using the
2050 '<tt><a href="#i_extractvalue">extractvalue</a></tt>' and
2051 '<tt><a href="#i_insertvalue">insertvalue</a></tt>' instructions.</p>
2053 <p>Structures may optionally be "packed" structures, which indicate that the
2054 alignment of the struct is one byte, and that there is no padding between
2055 the elements. In non-packed structs, padding between field types is inserted
2056 as defined by the TargetData string in the module, which is required to match
2057 what the underlying code generator expects.</p>
2059 <p>Structures can either be "literal" or "identified". A literal structure is
2060 defined inline with other types (e.g. <tt>{i32, i32}*</tt>) whereas identified
2061 types are always defined at the top level with a name. Literal types are
2062 uniqued by their contents and can never be recursive or opaque since there is
2063 no way to write one. Identified types can be recursive, can be opaqued, and are
2069 %T1 = type { <type list> } <i>; Identified normal struct type</i>
2070 %T2 = type <{ <type list> }> <i>; Identified packed struct type</i>
2074 <table class="layout">
2076 <td class="left"><tt>{ i32, i32, i32 }</tt></td>
2077 <td class="left">A triple of three <tt>i32</tt> values</td>
2080 <td class="left"><tt>{ float, i32 (i32) * }</tt></td>
2081 <td class="left">A pair, where the first element is a <tt>float</tt> and the
2082 second element is a <a href="#t_pointer">pointer</a> to a
2083 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
2084 an <tt>i32</tt>.</td>
2087 <td class="left"><tt><{ i8, i32 }></tt></td>
2088 <td class="left">A packed struct known to be 5 bytes in size.</td>
2094 <!-- _______________________________________________________________________ -->
2096 <a name="t_opaque">Opaque Structure Types</a>
2102 <p>Opaque structure types are used to represent named structure types that do
2103 not have a body specified. This corresponds (for example) to the C notion of
2104 a forward declared structure.</p>
2113 <table class="layout">
2115 <td class="left"><tt>opaque</tt></td>
2116 <td class="left">An opaque type.</td>
2124 <!-- _______________________________________________________________________ -->
2126 <a name="t_pointer">Pointer Type</a>
2132 <p>The pointer type is used to specify memory locations.
2133 Pointers are commonly used to reference objects in memory.</p>
2135 <p>Pointer types may have an optional address space attribute defining the
2136 numbered address space where the pointed-to object resides. The default
2137 address space is number zero. The semantics of non-zero address
2138 spaces are target-specific.</p>
2140 <p>Note that LLVM does not permit pointers to void (<tt>void*</tt>) nor does it
2141 permit pointers to labels (<tt>label*</tt>). Use <tt>i8*</tt> instead.</p>
2149 <table class="layout">
2151 <td class="left"><tt>[4 x i32]*</tt></td>
2152 <td class="left">A <a href="#t_pointer">pointer</a> to <a
2153 href="#t_array">array</a> of four <tt>i32</tt> values.</td>
2156 <td class="left"><tt>i32 (i32*) *</tt></td>
2157 <td class="left"> A <a href="#t_pointer">pointer</a> to a <a
2158 href="#t_function">function</a> that takes an <tt>i32*</tt>, returning an
2162 <td class="left"><tt>i32 addrspace(5)*</tt></td>
2163 <td class="left">A <a href="#t_pointer">pointer</a> to an <tt>i32</tt> value
2164 that resides in address space #5.</td>
2170 <!-- _______________________________________________________________________ -->
2172 <a name="t_vector">Vector Type</a>
2178 <p>A vector type is a simple derived type that represents a vector of elements.
2179 Vector types are used when multiple primitive data are operated in parallel
2180 using a single instruction (SIMD). A vector type requires a size (number of
2181 elements) and an underlying primitive data type. Vector types are considered
2182 <a href="#t_firstclass">first class</a>.</p>
2186 < <# elements> x <elementtype> >
2189 <p>The number of elements is a constant integer value larger than 0; elementtype
2190 may be any integer or floating point type. Vectors of size zero are not
2191 allowed, and pointers are not allowed as the element type.</p>
2194 <table class="layout">
2196 <td class="left"><tt><4 x i32></tt></td>
2197 <td class="left">Vector of 4 32-bit integer values.</td>
2200 <td class="left"><tt><8 x float></tt></td>
2201 <td class="left">Vector of 8 32-bit floating-point values.</td>
2204 <td class="left"><tt><2 x i64></tt></td>
2205 <td class="left">Vector of 2 64-bit integer values.</td>
2215 <!-- *********************************************************************** -->
2216 <h2><a name="constants">Constants</a></h2>
2217 <!-- *********************************************************************** -->
2221 <p>LLVM has several different basic types of constants. This section describes
2222 them all and their syntax.</p>
2224 <!-- ======================================================================= -->
2226 <a name="simpleconstants">Simple Constants</a>
2232 <dt><b>Boolean constants</b></dt>
2233 <dd>The two strings '<tt>true</tt>' and '<tt>false</tt>' are both valid
2234 constants of the <tt><a href="#t_integer">i1</a></tt> type.</dd>
2236 <dt><b>Integer constants</b></dt>
2237 <dd>Standard integers (such as '4') are constants of
2238 the <a href="#t_integer">integer</a> type. Negative numbers may be used
2239 with integer types.</dd>
2241 <dt><b>Floating point constants</b></dt>
2242 <dd>Floating point constants use standard decimal notation (e.g. 123.421),
2243 exponential notation (e.g. 1.23421e+2), or a more precise hexadecimal
2244 notation (see below). The assembler requires the exact decimal value of a
2245 floating-point constant. For example, the assembler accepts 1.25 but
2246 rejects 1.3 because 1.3 is a repeating decimal in binary. Floating point
2247 constants must have a <a href="#t_floating">floating point</a> type. </dd>
2249 <dt><b>Null pointer constants</b></dt>
2250 <dd>The identifier '<tt>null</tt>' is recognized as a null pointer constant
2251 and must be of <a href="#t_pointer">pointer type</a>.</dd>
2254 <p>The one non-intuitive notation for constants is the hexadecimal form of
2255 floating point constants. For example, the form '<tt>double
2256 0x432ff973cafa8000</tt>' is equivalent to (but harder to read than)
2257 '<tt>double 4.5e+15</tt>'. The only time hexadecimal floating point
2258 constants are required (and the only time that they are generated by the
2259 disassembler) is when a floating point constant must be emitted but it cannot
2260 be represented as a decimal floating point number in a reasonable number of
2261 digits. For example, NaN's, infinities, and other special values are
2262 represented in their IEEE hexadecimal format so that assembly and disassembly
2263 do not cause any bits to change in the constants.</p>
2265 <p>When using the hexadecimal form, constants of types float and double are
2266 represented using the 16-digit form shown above (which matches the IEEE754
2267 representation for double); float values must, however, be exactly
2268 representable as IEE754 single precision. Hexadecimal format is always used
2269 for long double, and there are three forms of long double. The 80-bit format
2270 used by x86 is represented as <tt>0xK</tt> followed by 20 hexadecimal digits.
2271 The 128-bit format used by PowerPC (two adjacent doubles) is represented
2272 by <tt>0xM</tt> followed by 32 hexadecimal digits. The IEEE 128-bit format
2273 is represented by <tt>0xL</tt> followed by 32 hexadecimal digits; no
2274 currently supported target uses this format. Long doubles will only work if
2275 they match the long double format on your target. All hexadecimal formats
2276 are big-endian (sign bit at the left).</p>
2278 <p>There are no constants of type x86mmx.</p>
2281 <!-- ======================================================================= -->
2283 <a name="aggregateconstants"></a> <!-- old anchor -->
2284 <a name="complexconstants">Complex Constants</a>
2289 <p>Complex constants are a (potentially recursive) combination of simple
2290 constants and smaller complex constants.</p>
2293 <dt><b>Structure constants</b></dt>
2294 <dd>Structure constants are represented with notation similar to structure
2295 type definitions (a comma separated list of elements, surrounded by braces
2296 (<tt>{}</tt>)). For example: "<tt>{ i32 4, float 17.0, i32* @G }</tt>",
2297 where "<tt>@G</tt>" is declared as "<tt>@G = external global i32</tt>".
2298 Structure constants must have <a href="#t_struct">structure type</a>, and
2299 the number and types of elements must match those specified by the
2302 <dt><b>Array constants</b></dt>
2303 <dd>Array constants are represented with notation similar to array type
2304 definitions (a comma separated list of elements, surrounded by square
2305 brackets (<tt>[]</tt>)). For example: "<tt>[ i32 42, i32 11, i32 74
2306 ]</tt>". Array constants must have <a href="#t_array">array type</a>, and
2307 the number and types of elements must match those specified by the
2310 <dt><b>Vector constants</b></dt>
2311 <dd>Vector constants are represented with notation similar to vector type
2312 definitions (a comma separated list of elements, surrounded by
2313 less-than/greater-than's (<tt><></tt>)). For example: "<tt>< i32
2314 42, i32 11, i32 74, i32 100 ></tt>". Vector constants must
2315 have <a href="#t_vector">vector type</a>, and the number and types of
2316 elements must match those specified by the type.</dd>
2318 <dt><b>Zero initialization</b></dt>
2319 <dd>The string '<tt>zeroinitializer</tt>' can be used to zero initialize a
2320 value to zero of <em>any</em> type, including scalar and
2321 <a href="#t_aggregate">aggregate</a> types.
2322 This is often used to avoid having to print large zero initializers
2323 (e.g. for large arrays) and is always exactly equivalent to using explicit
2324 zero initializers.</dd>
2326 <dt><b>Metadata node</b></dt>
2327 <dd>A metadata node is a structure-like constant with
2328 <a href="#t_metadata">metadata type</a>. For example: "<tt>metadata !{
2329 i32 0, metadata !"test" }</tt>". Unlike other constants that are meant to
2330 be interpreted as part of the instruction stream, metadata is a place to
2331 attach additional information such as debug info.</dd>
2336 <!-- ======================================================================= -->
2338 <a name="globalconstants">Global Variable and Function Addresses</a>
2343 <p>The addresses of <a href="#globalvars">global variables</a>
2344 and <a href="#functionstructure">functions</a> are always implicitly valid
2345 (link-time) constants. These constants are explicitly referenced when
2346 the <a href="#identifiers">identifier for the global</a> is used and always
2347 have <a href="#t_pointer">pointer</a> type. For example, the following is a
2348 legal LLVM file:</p>
2350 <pre class="doc_code">
2353 @Z = global [2 x i32*] [ i32* @X, i32* @Y ]
2358 <!-- ======================================================================= -->
2360 <a name="undefvalues">Undefined Values</a>
2365 <p>The string '<tt>undef</tt>' can be used anywhere a constant is expected, and
2366 indicates that the user of the value may receive an unspecified bit-pattern.
2367 Undefined values may be of any type (other than '<tt>label</tt>'
2368 or '<tt>void</tt>') and be used anywhere a constant is permitted.</p>
2370 <p>Undefined values are useful because they indicate to the compiler that the
2371 program is well defined no matter what value is used. This gives the
2372 compiler more freedom to optimize. Here are some examples of (potentially
2373 surprising) transformations that are valid (in pseudo IR):</p>
2376 <pre class="doc_code">
2386 <p>This is safe because all of the output bits are affected by the undef bits.
2387 Any output bit can have a zero or one depending on the input bits.</p>
2389 <pre class="doc_code">
2400 <p>These logical operations have bits that are not always affected by the input.
2401 For example, if <tt>%X</tt> has a zero bit, then the output of the
2402 '<tt>and</tt>' operation will always be a zero for that bit, no matter what
2403 the corresponding bit from the '<tt>undef</tt>' is. As such, it is unsafe to
2404 optimize or assume that the result of the '<tt>and</tt>' is '<tt>undef</tt>'.
2405 However, it is safe to assume that all bits of the '<tt>undef</tt>' could be
2406 0, and optimize the '<tt>and</tt>' to 0. Likewise, it is safe to assume that
2407 all the bits of the '<tt>undef</tt>' operand to the '<tt>or</tt>' could be
2408 set, allowing the '<tt>or</tt>' to be folded to -1.</p>
2410 <pre class="doc_code">
2411 %A = select undef, %X, %Y
2412 %B = select undef, 42, %Y
2413 %C = select %X, %Y, undef
2424 <p>This set of examples shows that undefined '<tt>select</tt>' (and conditional
2425 branch) conditions can go <em>either way</em>, but they have to come from one
2426 of the two operands. In the <tt>%A</tt> example, if <tt>%X</tt> and
2427 <tt>%Y</tt> were both known to have a clear low bit, then <tt>%A</tt> would
2428 have to have a cleared low bit. However, in the <tt>%C</tt> example, the
2429 optimizer is allowed to assume that the '<tt>undef</tt>' operand could be the
2430 same as <tt>%Y</tt>, allowing the whole '<tt>select</tt>' to be
2433 <pre class="doc_code">
2434 %A = xor undef, undef
2452 <p>This example points out that two '<tt>undef</tt>' operands are not
2453 necessarily the same. This can be surprising to people (and also matches C
2454 semantics) where they assume that "<tt>X^X</tt>" is always zero, even
2455 if <tt>X</tt> is undefined. This isn't true for a number of reasons, but the
2456 short answer is that an '<tt>undef</tt>' "variable" can arbitrarily change
2457 its value over its "live range". This is true because the variable doesn't
2458 actually <em>have a live range</em>. Instead, the value is logically read
2459 from arbitrary registers that happen to be around when needed, so the value
2460 is not necessarily consistent over time. In fact, <tt>%A</tt> and <tt>%C</tt>
2461 need to have the same semantics or the core LLVM "replace all uses with"
2462 concept would not hold.</p>
2464 <pre class="doc_code">
2472 <p>These examples show the crucial difference between an <em>undefined
2473 value</em> and <em>undefined behavior</em>. An undefined value (like
2474 '<tt>undef</tt>') is allowed to have an arbitrary bit-pattern. This means that
2475 the <tt>%A</tt> operation can be constant folded to '<tt>undef</tt>', because
2476 the '<tt>undef</tt>' could be an SNaN, and <tt>fdiv</tt> is not (currently)
2477 defined on SNaN's. However, in the second example, we can make a more
2478 aggressive assumption: because the <tt>undef</tt> is allowed to be an
2479 arbitrary value, we are allowed to assume that it could be zero. Since a
2480 divide by zero has <em>undefined behavior</em>, we are allowed to assume that
2481 the operation does not execute at all. This allows us to delete the divide and
2482 all code after it. Because the undefined operation "can't happen", the
2483 optimizer can assume that it occurs in dead code.</p>
2485 <pre class="doc_code">
2486 a: store undef -> %X
2487 b: store %X -> undef
2493 <p>These examples reiterate the <tt>fdiv</tt> example: a store <em>of</em> an
2494 undefined value can be assumed to not have any effect; we can assume that the
2495 value is overwritten with bits that happen to match what was already there.
2496 However, a store <em>to</em> an undefined location could clobber arbitrary
2497 memory, therefore, it has undefined behavior.</p>
2501 <!-- ======================================================================= -->
2503 <a name="trapvalues">Trap Values</a>
2508 <p>Trap values are similar to <a href="#undefvalues">undef values</a>, however
2509 instead of representing an unspecified bit pattern, they represent the
2510 fact that an instruction or constant expression which cannot evoke side
2511 effects has nevertheless detected a condition which results in undefined
2514 <p>There is currently no way of representing a trap value in the IR; they
2515 only exist when produced by operations such as
2516 <a href="#i_add"><tt>add</tt></a> with the <tt>nsw</tt> flag.</p>
2518 <p>Trap value behavior is defined in terms of value <i>dependence</i>:</p>
2521 <li>Values other than <a href="#i_phi"><tt>phi</tt></a> nodes depend on
2522 their operands.</li>
2524 <li><a href="#i_phi"><tt>Phi</tt></a> nodes depend on the operand corresponding
2525 to their dynamic predecessor basic block.</li>
2527 <li>Function arguments depend on the corresponding actual argument values in
2528 the dynamic callers of their functions.</li>
2530 <li><a href="#i_call"><tt>Call</tt></a> instructions depend on the
2531 <a href="#i_ret"><tt>ret</tt></a> instructions that dynamically transfer
2532 control back to them.</li>
2534 <li><a href="#i_invoke"><tt>Invoke</tt></a> instructions depend on the
2535 <a href="#i_ret"><tt>ret</tt></a>, <a href="#i_unwind"><tt>unwind</tt></a>,
2536 or exception-throwing call instructions that dynamically transfer control
2539 <li>Non-volatile loads and stores depend on the most recent stores to all of the
2540 referenced memory addresses, following the order in the IR
2541 (including loads and stores implied by intrinsics such as
2542 <a href="#int_memcpy"><tt>@llvm.memcpy</tt></a>.)</li>
2544 <!-- TODO: In the case of multiple threads, this only applies if the store
2545 "happens-before" the load or store. -->
2547 <!-- TODO: floating-point exception state -->
2549 <li>An instruction with externally visible side effects depends on the most
2550 recent preceding instruction with externally visible side effects, following
2551 the order in the IR. (This includes
2552 <a href="#volatile">volatile operations</a>.)</li>
2554 <li>An instruction <i>control-depends</i> on a
2555 <a href="#terminators">terminator instruction</a>
2556 if the terminator instruction has multiple successors and the instruction
2557 is always executed when control transfers to one of the successors, and
2558 may not be executed when control is transferred to another.</li>
2560 <li>Additionally, an instruction also <i>control-depends</i> on a terminator
2561 instruction if the set of instructions it otherwise depends on would be
2562 different if the terminator had transferred control to a different
2565 <li>Dependence is transitive.</li>
2569 <p>Whenever a trap value is generated, all values which depend on it evaluate
2570 to trap. If they have side effects, they evoke their side effects as if each
2571 operand with a trap value were undef. If they have externally-visible side
2572 effects, the behavior is undefined.</p>
2574 <p>Here are some examples:</p>
2576 <pre class="doc_code">
2578 %trap = sub nuw i32 0, 1 ; Results in a trap value.
2579 %still_trap = and i32 %trap, 0 ; Whereas (and i32 undef, 0) would return 0.
2580 %trap_yet_again = getelementptr i32* @h, i32 %still_trap
2581 store i32 0, i32* %trap_yet_again ; undefined behavior
2583 store i32 %trap, i32* @g ; Trap value conceptually stored to memory.
2584 %trap2 = load i32* @g ; Returns a trap value, not just undef.
2586 store volatile i32 %trap, i32* @g ; External observation; undefined behavior.
2588 %narrowaddr = bitcast i32* @g to i16*
2589 %wideaddr = bitcast i32* @g to i64*
2590 %trap3 = load i16* %narrowaddr ; Returns a trap value.
2591 %trap4 = load i64* %wideaddr ; Returns a trap value.
2593 %cmp = icmp slt i32 %trap, 0 ; Returns a trap value.
2594 br i1 %cmp, label %true, label %end ; Branch to either destination.
2597 store volatile i32 0, i32* @g ; This is control-dependent on %cmp, so
2598 ; it has undefined behavior.
2602 %p = phi i32 [ 0, %entry ], [ 1, %true ]
2603 ; Both edges into this PHI are
2604 ; control-dependent on %cmp, so this
2605 ; always results in a trap value.
2607 store volatile i32 0, i32* @g ; This would depend on the store in %true
2608 ; if %cmp is true, or the store in %entry
2609 ; otherwise, so this is undefined behavior.
2611 br i1 %cmp, label %second_true, label %second_end
2612 ; The same branch again, but this time the
2613 ; true block doesn't have side effects.
2620 store volatile i32 0, i32* @g ; This time, the instruction always depends
2621 ; on the store in %end. Also, it is
2622 ; control-equivalent to %end, so this is
2623 ; well-defined (again, ignoring earlier
2624 ; undefined behavior in this example).
2629 <!-- ======================================================================= -->
2631 <a name="blockaddress">Addresses of Basic Blocks</a>
2636 <p><b><tt>blockaddress(@function, %block)</tt></b></p>
2638 <p>The '<tt>blockaddress</tt>' constant computes the address of the specified
2639 basic block in the specified function, and always has an i8* type. Taking
2640 the address of the entry block is illegal.</p>
2642 <p>This value only has defined behavior when used as an operand to the
2643 '<a href="#i_indirectbr"><tt>indirectbr</tt></a>' instruction, or for
2644 comparisons against null. Pointer equality tests between labels addresses
2645 results in undefined behavior — though, again, comparison against null
2646 is ok, and no label is equal to the null pointer. This may be passed around
2647 as an opaque pointer sized value as long as the bits are not inspected. This
2648 allows <tt>ptrtoint</tt> and arithmetic to be performed on these values so
2649 long as the original value is reconstituted before the <tt>indirectbr</tt>
2652 <p>Finally, some targets may provide defined semantics when using the value as
2653 the operand to an inline assembly, but that is target specific.</p>
2658 <!-- ======================================================================= -->
2660 <a name="constantexprs">Constant Expressions</a>
2665 <p>Constant expressions are used to allow expressions involving other constants
2666 to be used as constants. Constant expressions may be of
2667 any <a href="#t_firstclass">first class</a> type and may involve any LLVM
2668 operation that does not have side effects (e.g. load and call are not
2669 supported). The following is the syntax for constant expressions:</p>
2672 <dt><b><tt>trunc (CST to TYPE)</tt></b></dt>
2673 <dd>Truncate a constant to another type. The bit size of CST must be larger
2674 than the bit size of TYPE. Both types must be integers.</dd>
2676 <dt><b><tt>zext (CST to TYPE)</tt></b></dt>
2677 <dd>Zero extend a constant to another type. The bit size of CST must be
2678 smaller than the bit size of TYPE. Both types must be integers.</dd>
2680 <dt><b><tt>sext (CST to TYPE)</tt></b></dt>
2681 <dd>Sign extend a constant to another type. The bit size of CST must be
2682 smaller than the bit size of TYPE. Both types must be integers.</dd>
2684 <dt><b><tt>fptrunc (CST to TYPE)</tt></b></dt>
2685 <dd>Truncate a floating point constant to another floating point type. The
2686 size of CST must be larger than the size of TYPE. Both types must be
2687 floating point.</dd>
2689 <dt><b><tt>fpext (CST to TYPE)</tt></b></dt>
2690 <dd>Floating point extend a constant to another type. The size of CST must be
2691 smaller or equal to the size of TYPE. Both types must be floating
2694 <dt><b><tt>fptoui (CST to TYPE)</tt></b></dt>
2695 <dd>Convert a floating point constant to the corresponding unsigned integer
2696 constant. TYPE must be a scalar or vector integer type. CST must be of
2697 scalar or vector floating point type. Both CST and TYPE must be scalars,
2698 or vectors of the same number of elements. If the value won't fit in the
2699 integer type, the results are undefined.</dd>
2701 <dt><b><tt>fptosi (CST to TYPE)</tt></b></dt>
2702 <dd>Convert a floating point constant to the corresponding signed integer
2703 constant. TYPE must be a scalar or vector integer type. CST must be of
2704 scalar or vector floating point type. Both CST and TYPE must be scalars,
2705 or vectors of the same number of elements. If the value won't fit in the
2706 integer type, the results are undefined.</dd>
2708 <dt><b><tt>uitofp (CST to TYPE)</tt></b></dt>
2709 <dd>Convert an unsigned integer constant to the corresponding floating point
2710 constant. TYPE must be a scalar or vector floating point type. CST must be
2711 of scalar or vector integer type. Both CST and TYPE must be scalars, or
2712 vectors of the same number of elements. If the value won't fit in the
2713 floating point type, the results are undefined.</dd>
2715 <dt><b><tt>sitofp (CST to TYPE)</tt></b></dt>
2716 <dd>Convert a signed integer constant to the corresponding floating point
2717 constant. TYPE must be a scalar or vector floating point type. CST must be
2718 of scalar or vector integer type. Both CST and TYPE must be scalars, or
2719 vectors of the same number of elements. If the value won't fit in the
2720 floating point type, the results are undefined.</dd>
2722 <dt><b><tt>ptrtoint (CST to TYPE)</tt></b></dt>
2723 <dd>Convert a pointer typed constant to the corresponding integer constant
2724 <tt>TYPE</tt> must be an integer type. <tt>CST</tt> must be of pointer
2725 type. The <tt>CST</tt> value is zero extended, truncated, or unchanged to
2726 make it fit in <tt>TYPE</tt>.</dd>
2728 <dt><b><tt>inttoptr (CST to TYPE)</tt></b></dt>
2729 <dd>Convert a integer constant to a pointer constant. TYPE must be a pointer
2730 type. CST must be of integer type. The CST value is zero extended,
2731 truncated, or unchanged to make it fit in a pointer size. This one is
2732 <i>really</i> dangerous!</dd>
2734 <dt><b><tt>bitcast (CST to TYPE)</tt></b></dt>
2735 <dd>Convert a constant, CST, to another TYPE. The constraints of the operands
2736 are the same as those for the <a href="#i_bitcast">bitcast
2737 instruction</a>.</dd>
2739 <dt><b><tt>getelementptr (CSTPTR, IDX0, IDX1, ...)</tt></b></dt>
2740 <dt><b><tt>getelementptr inbounds (CSTPTR, IDX0, IDX1, ...)</tt></b></dt>
2741 <dd>Perform the <a href="#i_getelementptr">getelementptr operation</a> on
2742 constants. As with the <a href="#i_getelementptr">getelementptr</a>
2743 instruction, the index list may have zero or more indexes, which are
2744 required to make sense for the type of "CSTPTR".</dd>
2746 <dt><b><tt>select (COND, VAL1, VAL2)</tt></b></dt>
2747 <dd>Perform the <a href="#i_select">select operation</a> on constants.</dd>
2749 <dt><b><tt>icmp COND (VAL1, VAL2)</tt></b></dt>
2750 <dd>Performs the <a href="#i_icmp">icmp operation</a> on constants.</dd>
2752 <dt><b><tt>fcmp COND (VAL1, VAL2)</tt></b></dt>
2753 <dd>Performs the <a href="#i_fcmp">fcmp operation</a> on constants.</dd>
2755 <dt><b><tt>extractelement (VAL, IDX)</tt></b></dt>
2756 <dd>Perform the <a href="#i_extractelement">extractelement operation</a> on
2759 <dt><b><tt>insertelement (VAL, ELT, IDX)</tt></b></dt>
2760 <dd>Perform the <a href="#i_insertelement">insertelement operation</a> on
2763 <dt><b><tt>shufflevector (VEC1, VEC2, IDXMASK)</tt></b></dt>
2764 <dd>Perform the <a href="#i_shufflevector">shufflevector operation</a> on
2767 <dt><b><tt>extractvalue (VAL, IDX0, IDX1, ...)</tt></b></dt>
2768 <dd>Perform the <a href="#i_extractvalue">extractvalue operation</a> on
2769 constants. The index list is interpreted in a similar manner as indices in
2770 a '<a href="#i_getelementptr">getelementptr</a>' operation. At least one
2771 index value must be specified.</dd>
2773 <dt><b><tt>insertvalue (VAL, ELT, IDX0, IDX1, ...)</tt></b></dt>
2774 <dd>Perform the <a href="#i_insertvalue">insertvalue operation</a> on
2775 constants. The index list is interpreted in a similar manner as indices in
2776 a '<a href="#i_getelementptr">getelementptr</a>' operation. At least one
2777 index value must be specified.</dd>
2779 <dt><b><tt>OPCODE (LHS, RHS)</tt></b></dt>
2780 <dd>Perform the specified operation of the LHS and RHS constants. OPCODE may
2781 be any of the <a href="#binaryops">binary</a>
2782 or <a href="#bitwiseops">bitwise binary</a> operations. The constraints
2783 on operands are the same as those for the corresponding instruction
2784 (e.g. no bitwise operations on floating point values are allowed).</dd>
2791 <!-- *********************************************************************** -->
2792 <h2><a name="othervalues">Other Values</a></h2>
2793 <!-- *********************************************************************** -->
2795 <!-- ======================================================================= -->
2797 <a name="inlineasm">Inline Assembler Expressions</a>
2802 <p>LLVM supports inline assembler expressions (as opposed
2803 to <a href="#moduleasm"> Module-Level Inline Assembly</a>) through the use of
2804 a special value. This value represents the inline assembler as a string
2805 (containing the instructions to emit), a list of operand constraints (stored
2806 as a string), a flag that indicates whether or not the inline asm
2807 expression has side effects, and a flag indicating whether the function
2808 containing the asm needs to align its stack conservatively. An example
2809 inline assembler expression is:</p>
2811 <pre class="doc_code">
2812 i32 (i32) asm "bswap $0", "=r,r"
2815 <p>Inline assembler expressions may <b>only</b> be used as the callee operand of
2816 a <a href="#i_call"><tt>call</tt> instruction</a>. Thus, typically we
2819 <pre class="doc_code">
2820 %X = call i32 asm "<a href="#int_bswap">bswap</a> $0", "=r,r"(i32 %Y)
2823 <p>Inline asms with side effects not visible in the constraint list must be
2824 marked as having side effects. This is done through the use of the
2825 '<tt>sideeffect</tt>' keyword, like so:</p>
2827 <pre class="doc_code">
2828 call void asm sideeffect "eieio", ""()
2831 <p>In some cases inline asms will contain code that will not work unless the
2832 stack is aligned in some way, such as calls or SSE instructions on x86,
2833 yet will not contain code that does that alignment within the asm.
2834 The compiler should make conservative assumptions about what the asm might
2835 contain and should generate its usual stack alignment code in the prologue
2836 if the '<tt>alignstack</tt>' keyword is present:</p>
2838 <pre class="doc_code">
2839 call void asm alignstack "eieio", ""()
2842 <p>If both keywords appear the '<tt>sideeffect</tt>' keyword must come
2845 <p>TODO: The format of the asm and constraints string still need to be
2846 documented here. Constraints on what can be done (e.g. duplication, moving,
2847 etc need to be documented). This is probably best done by reference to
2848 another document that covers inline asm from a holistic perspective.</p>
2851 <a name="inlineasm_md">Inline Asm Metadata</a>
2856 <p>The call instructions that wrap inline asm nodes may have a "!srcloc" MDNode
2857 attached to it that contains a list of constant integers. If present, the
2858 code generator will use the integer as the location cookie value when report
2859 errors through the LLVMContext error reporting mechanisms. This allows a
2860 front-end to correlate backend errors that occur with inline asm back to the
2861 source code that produced it. For example:</p>
2863 <pre class="doc_code">
2864 call void asm sideeffect "something bad", ""()<b>, !srcloc !42</b>
2866 !42 = !{ i32 1234567 }
2869 <p>It is up to the front-end to make sense of the magic numbers it places in the
2870 IR. If the MDNode contains multiple constants, the code generator will use
2871 the one that corresponds to the line of the asm that the error occurs on.</p>
2877 <!-- ======================================================================= -->
2879 <a name="metadata">Metadata Nodes and Metadata Strings</a>
2884 <p>LLVM IR allows metadata to be attached to instructions in the program that
2885 can convey extra information about the code to the optimizers and code
2886 generator. One example application of metadata is source-level debug
2887 information. There are two metadata primitives: strings and nodes. All
2888 metadata has the <tt>metadata</tt> type and is identified in syntax by a
2889 preceding exclamation point ('<tt>!</tt>').</p>
2891 <p>A metadata string is a string surrounded by double quotes. It can contain
2892 any character by escaping non-printable characters with "\xx" where "xx" is
2893 the two digit hex code. For example: "<tt>!"test\00"</tt>".</p>
2895 <p>Metadata nodes are represented with notation similar to structure constants
2896 (a comma separated list of elements, surrounded by braces and preceded by an
2897 exclamation point). For example: "<tt>!{ metadata !"test\00", i32
2898 10}</tt>". Metadata nodes can have any values as their operand.</p>
2900 <p>A <a href="#namedmetadatastructure">named metadata</a> is a collection of
2901 metadata nodes, which can be looked up in the module symbol table. For
2902 example: "<tt>!foo = metadata !{!4, !3}</tt>".
2904 <p>Metadata can be used as function arguments. Here <tt>llvm.dbg.value</tt>
2905 function is using two metadata arguments.</p>
2907 <div class="doc_code">
2909 call void @llvm.dbg.value(metadata !24, i64 0, metadata !25)
2913 <p>Metadata can be attached with an instruction. Here metadata <tt>!21</tt> is
2914 attached with <tt>add</tt> instruction using <tt>!dbg</tt> identifier.</p>
2916 <div class="doc_code">
2918 %indvar.next = add i64 %indvar, 1, !dbg !21
2922 <p>More information about specific metadata nodes recognized by the optimizers
2923 and code generator is found below.</p>
2926 <a name="tbaa">'<tt>tbaa</tt>' Metadata</a>
2931 <p>In LLVM IR, memory does not have types, so LLVM's own type system is not
2932 suitable for doing TBAA. Instead, metadata is added to the IR to describe
2933 a type system of a higher level language. This can be used to implement
2934 typical C/C++ TBAA, but it can also be used to implement custom alias
2935 analysis behavior for other languages.</p>
2937 <p>The current metadata format is very simple. TBAA metadata nodes have up to
2938 three fields, e.g.:</p>
2940 <div class="doc_code">
2942 !0 = metadata !{ metadata !"an example type tree" }
2943 !1 = metadata !{ metadata !"int", metadata !0 }
2944 !2 = metadata !{ metadata !"float", metadata !0 }
2945 !3 = metadata !{ metadata !"const float", metadata !2, i64 1 }
2949 <p>The first field is an identity field. It can be any value, usually
2950 a metadata string, which uniquely identifies the type. The most important
2951 name in the tree is the name of the root node. Two trees with
2952 different root node names are entirely disjoint, even if they
2953 have leaves with common names.</p>
2955 <p>The second field identifies the type's parent node in the tree, or
2956 is null or omitted for a root node. A type is considered to alias
2957 all of its descendants and all of its ancestors in the tree. Also,
2958 a type is considered to alias all types in other trees, so that
2959 bitcode produced from multiple front-ends is handled conservatively.</p>
2961 <p>If the third field is present, it's an integer which if equal to 1
2962 indicates that the type is "constant" (meaning
2963 <tt>pointsToConstantMemory</tt> should return true; see
2964 <a href="AliasAnalysis.html#OtherItfs">other useful
2965 <tt>AliasAnalysis</tt> methods</a>).</p>
2970 <a name="fpaccuracy">'<tt>fpaccuracy</tt>' Metadata</a>
2975 <p><tt>fpaccuracy</tt> metadata may be attached to any instruction of floating
2976 point type. It expresses the maximum relative error of the result of
2977 that instruction, in ULPs. ULP is defined as follows:</p>
2981 <p>If <tt>x</tt> is a real number that lies between two finite consecutive
2982 floating-point numbers <tt>a</tt> and <tt>b</tt>, without being equal to one
2983 of them, then <tt>ulp(x) = |b - a|</tt>, otherwise <tt>ulp(x)</tt> is the
2984 distance between the two non-equal finite floating-point numbers nearest
2985 <tt>x</tt>. Moreover, <tt>ulp(NaN)</tt> is <tt>NaN</tt>.</p>
2989 <p>The maximum relative error may be any rational number. The metadata node
2990 shall consist of a pair of unsigned integers respectively representing
2991 the numerator and denominator. For example, 2.5 ULP:</p>
2993 <div class="doc_code">
2995 !0 = metadata !{ i32 5, i32 2 }
3005 <!-- *********************************************************************** -->
3007 <a name="intrinsic_globals">Intrinsic Global Variables</a>
3009 <!-- *********************************************************************** -->
3011 <p>LLVM has a number of "magic" global variables that contain data that affect
3012 code generation or other IR semantics. These are documented here. All globals
3013 of this sort should have a section specified as "<tt>llvm.metadata</tt>". This
3014 section and all globals that start with "<tt>llvm.</tt>" are reserved for use
3017 <!-- ======================================================================= -->
3019 <a name="intg_used">The '<tt>llvm.used</tt>' Global Variable</a>
3024 <p>The <tt>@llvm.used</tt> global is an array with i8* element type which has <a
3025 href="#linkage_appending">appending linkage</a>. This array contains a list of
3026 pointers to global variables and functions which may optionally have a pointer
3027 cast formed of bitcast or getelementptr. For example, a legal use of it is:</p>
3029 <div class="doc_code">
3034 @llvm.used = appending global [2 x i8*] [
3036 i8* bitcast (i32* @Y to i8*)
3037 ], section "llvm.metadata"
3041 <p>If a global variable appears in the <tt>@llvm.used</tt> list, then the
3042 compiler, assembler, and linker are required to treat the symbol as if there
3043 is a reference to the global that it cannot see. For example, if a variable
3044 has internal linkage and no references other than that from
3045 the <tt>@llvm.used</tt> list, it cannot be deleted. This is commonly used to
3046 represent references from inline asms and other things the compiler cannot
3047 "see", and corresponds to "<tt>attribute((used))</tt>" in GNU C.</p>
3049 <p>On some targets, the code generator must emit a directive to the assembler or
3050 object file to prevent the assembler and linker from molesting the
3055 <!-- ======================================================================= -->
3057 <a name="intg_compiler_used">
3058 The '<tt>llvm.compiler.used</tt>' Global Variable
3064 <p>The <tt>@llvm.compiler.used</tt> directive is the same as the
3065 <tt>@llvm.used</tt> directive, except that it only prevents the compiler from
3066 touching the symbol. On targets that support it, this allows an intelligent
3067 linker to optimize references to the symbol without being impeded as it would
3068 be by <tt>@llvm.used</tt>.</p>
3070 <p>This is a rare construct that should only be used in rare circumstances, and
3071 should not be exposed to source languages.</p>
3075 <!-- ======================================================================= -->
3077 <a name="intg_global_ctors">The '<tt>llvm.global_ctors</tt>' Global Variable</a>
3082 <div class="doc_code">
3084 %0 = type { i32, void ()* }
3085 @llvm.global_ctors = appending global [1 x %0] [%0 { i32 65535, void ()* @ctor }]
3089 <p>The <tt>@llvm.global_ctors</tt> array contains a list of constructor
3090 functions and associated priorities. The functions referenced by this array
3091 will be called in ascending order of priority (i.e. lowest first) when the
3092 module is loaded. The order of functions with the same priority is not
3097 <!-- ======================================================================= -->
3099 <a name="intg_global_dtors">The '<tt>llvm.global_dtors</tt>' Global Variable</a>
3104 <div class="doc_code">
3106 %0 = type { i32, void ()* }
3107 @llvm.global_dtors = appending global [1 x %0] [%0 { i32 65535, void ()* @dtor }]
3111 <p>The <tt>@llvm.global_dtors</tt> array contains a list of destructor functions
3112 and associated priorities. The functions referenced by this array will be
3113 called in descending order of priority (i.e. highest first) when the module
3114 is loaded. The order of functions with the same priority is not defined.</p>
3120 <!-- *********************************************************************** -->
3121 <h2><a name="instref">Instruction Reference</a></h2>
3122 <!-- *********************************************************************** -->
3126 <p>The LLVM instruction set consists of several different classifications of
3127 instructions: <a href="#terminators">terminator
3128 instructions</a>, <a href="#binaryops">binary instructions</a>,
3129 <a href="#bitwiseops">bitwise binary instructions</a>,
3130 <a href="#memoryops">memory instructions</a>, and
3131 <a href="#otherops">other instructions</a>.</p>
3133 <!-- ======================================================================= -->
3135 <a name="terminators">Terminator Instructions</a>
3140 <p>As mentioned <a href="#functionstructure">previously</a>, every basic block
3141 in a program ends with a "Terminator" instruction, which indicates which
3142 block should be executed after the current block is finished. These
3143 terminator instructions typically yield a '<tt>void</tt>' value: they produce
3144 control flow, not values (the one exception being the
3145 '<a href="#i_invoke"><tt>invoke</tt></a>' instruction).</p>
3147 <p>The terminator instructions are:
3148 '<a href="#i_ret"><tt>ret</tt></a>',
3149 '<a href="#i_br"><tt>br</tt></a>',
3150 '<a href="#i_switch"><tt>switch</tt></a>',
3151 '<a href="#i_indirectbr"><tt>indirectbr</tt></a>',
3152 '<a href="#i_invoke"><tt>invoke</tt></a>',
3153 '<a href="#i_unwind"><tt>unwind</tt></a>',
3154 '<a href="#i_resume"><tt>resume</tt></a>', and
3155 '<a href="#i_unreachable"><tt>unreachable</tt></a>'.</p>
3157 <!-- _______________________________________________________________________ -->
3159 <a name="i_ret">'<tt>ret</tt>' Instruction</a>
3166 ret <type> <value> <i>; Return a value from a non-void function</i>
3167 ret void <i>; Return from void function</i>
3171 <p>The '<tt>ret</tt>' instruction is used to return control flow (and optionally
3172 a value) from a function back to the caller.</p>
3174 <p>There are two forms of the '<tt>ret</tt>' instruction: one that returns a
3175 value and then causes control flow, and one that just causes control flow to
3179 <p>The '<tt>ret</tt>' instruction optionally accepts a single argument, the
3180 return value. The type of the return value must be a
3181 '<a href="#t_firstclass">first class</a>' type.</p>
3183 <p>A function is not <a href="#wellformed">well formed</a> if it it has a
3184 non-void return type and contains a '<tt>ret</tt>' instruction with no return
3185 value or a return value with a type that does not match its type, or if it
3186 has a void return type and contains a '<tt>ret</tt>' instruction with a
3190 <p>When the '<tt>ret</tt>' instruction is executed, control flow returns back to
3191 the calling function's context. If the caller is a
3192 "<a href="#i_call"><tt>call</tt></a>" instruction, execution continues at the
3193 instruction after the call. If the caller was an
3194 "<a href="#i_invoke"><tt>invoke</tt></a>" instruction, execution continues at
3195 the beginning of the "normal" destination block. If the instruction returns
3196 a value, that value shall set the call or invoke instruction's return
3201 ret i32 5 <i>; Return an integer value of 5</i>
3202 ret void <i>; Return from a void function</i>
3203 ret { i32, i8 } { i32 4, i8 2 } <i>; Return a struct of values 4 and 2</i>
3207 <!-- _______________________________________________________________________ -->
3209 <a name="i_br">'<tt>br</tt>' Instruction</a>
3216 br i1 <cond>, label <iftrue>, label <iffalse>
3217 br label <dest> <i>; Unconditional branch</i>
3221 <p>The '<tt>br</tt>' instruction is used to cause control flow to transfer to a
3222 different basic block in the current function. There are two forms of this
3223 instruction, corresponding to a conditional branch and an unconditional
3227 <p>The conditional branch form of the '<tt>br</tt>' instruction takes a single
3228 '<tt>i1</tt>' value and two '<tt>label</tt>' values. The unconditional form
3229 of the '<tt>br</tt>' instruction takes a single '<tt>label</tt>' value as a
3233 <p>Upon execution of a conditional '<tt>br</tt>' instruction, the '<tt>i1</tt>'
3234 argument is evaluated. If the value is <tt>true</tt>, control flows to the
3235 '<tt>iftrue</tt>' <tt>label</tt> argument. If "cond" is <tt>false</tt>,
3236 control flows to the '<tt>iffalse</tt>' <tt>label</tt> argument.</p>
3241 %cond = <a href="#i_icmp">icmp</a> eq i32 %a, %b
3242 br i1 %cond, label %IfEqual, label %IfUnequal
3244 <a href="#i_ret">ret</a> i32 1
3246 <a href="#i_ret">ret</a> i32 0
3251 <!-- _______________________________________________________________________ -->
3253 <a name="i_switch">'<tt>switch</tt>' Instruction</a>
3260 switch <intty> <value>, label <defaultdest> [ <intty> <val>, label <dest> ... ]
3264 <p>The '<tt>switch</tt>' instruction is used to transfer control flow to one of
3265 several different places. It is a generalization of the '<tt>br</tt>'
3266 instruction, allowing a branch to occur to one of many possible
3270 <p>The '<tt>switch</tt>' instruction uses three parameters: an integer
3271 comparison value '<tt>value</tt>', a default '<tt>label</tt>' destination,
3272 and an array of pairs of comparison value constants and '<tt>label</tt>'s.
3273 The table is not allowed to contain duplicate constant entries.</p>
3276 <p>The <tt>switch</tt> instruction specifies a table of values and
3277 destinations. When the '<tt>switch</tt>' instruction is executed, this table
3278 is searched for the given value. If the value is found, control flow is
3279 transferred to the corresponding destination; otherwise, control flow is
3280 transferred to the default destination.</p>
3282 <h5>Implementation:</h5>
3283 <p>Depending on properties of the target machine and the particular
3284 <tt>switch</tt> instruction, this instruction may be code generated in
3285 different ways. For example, it could be generated as a series of chained
3286 conditional branches or with a lookup table.</p>
3290 <i>; Emulate a conditional br instruction</i>
3291 %Val = <a href="#i_zext">zext</a> i1 %value to i32
3292 switch i32 %Val, label %truedest [ i32 0, label %falsedest ]
3294 <i>; Emulate an unconditional br instruction</i>
3295 switch i32 0, label %dest [ ]
3297 <i>; Implement a jump table:</i>
3298 switch i32 %val, label %otherwise [ i32 0, label %onzero
3300 i32 2, label %ontwo ]
3306 <!-- _______________________________________________________________________ -->
3308 <a name="i_indirectbr">'<tt>indirectbr</tt>' Instruction</a>
3315 indirectbr <somety>* <address>, [ label <dest1>, label <dest2>, ... ]
3320 <p>The '<tt>indirectbr</tt>' instruction implements an indirect branch to a label
3321 within the current function, whose address is specified by
3322 "<tt>address</tt>". Address must be derived from a <a
3323 href="#blockaddress">blockaddress</a> constant.</p>
3327 <p>The '<tt>address</tt>' argument is the address of the label to jump to. The
3328 rest of the arguments indicate the full set of possible destinations that the
3329 address may point to. Blocks are allowed to occur multiple times in the
3330 destination list, though this isn't particularly useful.</p>
3332 <p>This destination list is required so that dataflow analysis has an accurate
3333 understanding of the CFG.</p>
3337 <p>Control transfers to the block specified in the address argument. All
3338 possible destination blocks must be listed in the label list, otherwise this
3339 instruction has undefined behavior. This implies that jumps to labels
3340 defined in other functions have undefined behavior as well.</p>
3342 <h5>Implementation:</h5>
3344 <p>This is typically implemented with a jump through a register.</p>
3348 indirectbr i8* %Addr, [ label %bb1, label %bb2, label %bb3 ]
3354 <!-- _______________________________________________________________________ -->
3356 <a name="i_invoke">'<tt>invoke</tt>' Instruction</a>
3363 <result> = invoke [<a href="#callingconv">cconv</a>] [<a href="#paramattrs">ret attrs</a>] <ptr to function ty> <function ptr val>(<function args>) [<a href="#fnattrs">fn attrs</a>]
3364 to label <normal label> unwind label <exception label>
3368 <p>The '<tt>invoke</tt>' instruction causes control to transfer to a specified
3369 function, with the possibility of control flow transfer to either the
3370 '<tt>normal</tt>' label or the '<tt>exception</tt>' label. If the callee
3371 function returns with the "<tt><a href="#i_ret">ret</a></tt>" instruction,
3372 control flow will return to the "normal" label. If the callee (or any
3373 indirect callees) returns with the "<a href="#i_unwind"><tt>unwind</tt></a>"
3374 instruction, control is interrupted and continued at the dynamically nearest
3375 "exception" label.</p>
3377 <p>The '<tt>exception</tt>' label is a
3378 <i><a href="ExceptionHandling.html#overview">landing pad</a></i> for the
3379 exception. As such, '<tt>exception</tt>' label is required to have the
3380 "<a href="#i_landingpad"><tt>landingpad</tt></a>" instruction, which contains
3381 the information about about the behavior of the program after unwinding
3382 happens, as its first non-PHI instruction. The restrictions on the
3383 "<tt>landingpad</tt>" instruction's tightly couples it to the
3384 "<tt>invoke</tt>" instruction, so that the important information contained
3385 within the "<tt>landingpad</tt>" instruction can't be lost through normal
3389 <p>This instruction requires several arguments:</p>
3392 <li>The optional "cconv" marker indicates which <a href="#callingconv">calling
3393 convention</a> the call should use. If none is specified, the call
3394 defaults to using C calling conventions.</li>
3396 <li>The optional <a href="#paramattrs">Parameter Attributes</a> list for
3397 return values. Only '<tt>zeroext</tt>', '<tt>signext</tt>', and
3398 '<tt>inreg</tt>' attributes are valid here.</li>
3400 <li>'<tt>ptr to function ty</tt>': shall be the signature of the pointer to
3401 function value being invoked. In most cases, this is a direct function
3402 invocation, but indirect <tt>invoke</tt>s are just as possible, branching
3403 off an arbitrary pointer to function value.</li>
3405 <li>'<tt>function ptr val</tt>': An LLVM value containing a pointer to a
3406 function to be invoked. </li>
3408 <li>'<tt>function args</tt>': argument list whose types match the function
3409 signature argument types and parameter attributes. All arguments must be
3410 of <a href="#t_firstclass">first class</a> type. If the function
3411 signature indicates the function accepts a variable number of arguments,
3412 the extra arguments can be specified.</li>
3414 <li>'<tt>normal label</tt>': the label reached when the called function
3415 executes a '<tt><a href="#i_ret">ret</a></tt>' instruction. </li>
3417 <li>'<tt>exception label</tt>': the label reached when a callee returns with
3418 the <a href="#i_unwind"><tt>unwind</tt></a> instruction. </li>
3420 <li>The optional <a href="#fnattrs">function attributes</a> list. Only
3421 '<tt>noreturn</tt>', '<tt>nounwind</tt>', '<tt>readonly</tt>' and
3422 '<tt>readnone</tt>' attributes are valid here.</li>
3426 <p>This instruction is designed to operate as a standard
3427 '<tt><a href="#i_call">call</a></tt>' instruction in most regards. The
3428 primary difference is that it establishes an association with a label, which
3429 is used by the runtime library to unwind the stack.</p>
3431 <p>This instruction is used in languages with destructors to ensure that proper
3432 cleanup is performed in the case of either a <tt>longjmp</tt> or a thrown
3433 exception. Additionally, this is important for implementation of
3434 '<tt>catch</tt>' clauses in high-level languages that support them.</p>
3436 <p>For the purposes of the SSA form, the definition of the value returned by the
3437 '<tt>invoke</tt>' instruction is deemed to occur on the edge from the current
3438 block to the "normal" label. If the callee unwinds then no return value is
3441 <p>Note that the code generator does not yet completely support unwind, and
3442 that the invoke/unwind semantics are likely to change in future versions.</p>
3446 %retval = invoke i32 @Test(i32 15) to label %Continue
3447 unwind label %TestCleanup <i>; {i32}:retval set</i>
3448 %retval = invoke <a href="#callingconv">coldcc</a> i32 %Testfnptr(i32 15) to label %Continue
3449 unwind label %TestCleanup <i>; {i32}:retval set</i>
3454 <!-- _______________________________________________________________________ -->
3457 <a name="i_unwind">'<tt>unwind</tt>' Instruction</a>
3468 <p>The '<tt>unwind</tt>' instruction unwinds the stack, continuing control flow
3469 at the first callee in the dynamic call stack which used
3470 an <a href="#i_invoke"><tt>invoke</tt></a> instruction to perform the call.
3471 This is primarily used to implement exception handling.</p>
3474 <p>The '<tt>unwind</tt>' instruction causes execution of the current function to
3475 immediately halt. The dynamic call stack is then searched for the
3476 first <a href="#i_invoke"><tt>invoke</tt></a> instruction on the call stack.
3477 Once found, execution continues at the "exceptional" destination block
3478 specified by the <tt>invoke</tt> instruction. If there is no <tt>invoke</tt>
3479 instruction in the dynamic call chain, undefined behavior results.</p>
3481 <p>Note that the code generator does not yet completely support unwind, and
3482 that the invoke/unwind semantics are likely to change in future versions.</p>
3486 <!-- _______________________________________________________________________ -->
3489 <a name="i_resume">'<tt>resume</tt>' Instruction</a>
3496 resume <type> <value>
3500 <p>The '<tt>resume</tt>' instruction is a terminator instruction that has no
3504 <p>The '<tt>resume</tt>' instruction requires one argument, which must have the
3505 same type as the result of any '<tt>landingpad</tt>' instruction in the same
3509 <p>The '<tt>resume</tt>' instruction resumes propagation of an existing
3510 (in-flight) exception whose unwinding was interrupted with
3511 a <a href="#i_landingpad"><tt>landingpad</tt></a> instruction.</p>
3515 resume { i8*, i32 } %exn
3520 <!-- _______________________________________________________________________ -->
3523 <a name="i_unreachable">'<tt>unreachable</tt>' Instruction</a>
3534 <p>The '<tt>unreachable</tt>' instruction has no defined semantics. This
3535 instruction is used to inform the optimizer that a particular portion of the
3536 code is not reachable. This can be used to indicate that the code after a
3537 no-return function cannot be reached, and other facts.</p>
3540 <p>The '<tt>unreachable</tt>' instruction has no defined semantics.</p>
3546 <!-- ======================================================================= -->
3548 <a name="binaryops">Binary Operations</a>
3553 <p>Binary operators are used to do most of the computation in a program. They
3554 require two operands of the same type, execute an operation on them, and
3555 produce a single value. The operands might represent multiple data, as is
3556 the case with the <a href="#t_vector">vector</a> data type. The result value
3557 has the same type as its operands.</p>
3559 <p>There are several different binary operators:</p>
3561 <!-- _______________________________________________________________________ -->
3563 <a name="i_add">'<tt>add</tt>' Instruction</a>
3570 <result> = add <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3571 <result> = add nuw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3572 <result> = add nsw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3573 <result> = add nuw nsw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3577 <p>The '<tt>add</tt>' instruction returns the sum of its two operands.</p>
3580 <p>The two arguments to the '<tt>add</tt>' instruction must
3581 be <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
3582 integer values. Both arguments must have identical types.</p>
3585 <p>The value produced is the integer sum of the two operands.</p>
3587 <p>If the sum has unsigned overflow, the result returned is the mathematical
3588 result modulo 2<sup>n</sup>, where n is the bit width of the result.</p>
3590 <p>Because LLVM integers use a two's complement representation, this instruction
3591 is appropriate for both signed and unsigned integers.</p>
3593 <p><tt>nuw</tt> and <tt>nsw</tt> stand for "No Unsigned Wrap"
3594 and "No Signed Wrap", respectively. If the <tt>nuw</tt> and/or
3595 <tt>nsw</tt> keywords are present, the result value of the <tt>add</tt>
3596 is a <a href="#trapvalues">trap value</a> if unsigned and/or signed overflow,
3597 respectively, occurs.</p>
3601 <result> = add i32 4, %var <i>; yields {i32}:result = 4 + %var</i>
3606 <!-- _______________________________________________________________________ -->
3608 <a name="i_fadd">'<tt>fadd</tt>' Instruction</a>
3615 <result> = fadd <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3619 <p>The '<tt>fadd</tt>' instruction returns the sum of its two operands.</p>
3622 <p>The two arguments to the '<tt>fadd</tt>' instruction must be
3623 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3624 floating point values. Both arguments must have identical types.</p>
3627 <p>The value produced is the floating point sum of the two operands.</p>
3631 <result> = fadd float 4.0, %var <i>; yields {float}:result = 4.0 + %var</i>
3636 <!-- _______________________________________________________________________ -->
3638 <a name="i_sub">'<tt>sub</tt>' Instruction</a>
3645 <result> = sub <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3646 <result> = sub nuw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3647 <result> = sub nsw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3648 <result> = sub nuw nsw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3652 <p>The '<tt>sub</tt>' instruction returns the difference of its two
3655 <p>Note that the '<tt>sub</tt>' instruction is used to represent the
3656 '<tt>neg</tt>' instruction present in most other intermediate
3657 representations.</p>
3660 <p>The two arguments to the '<tt>sub</tt>' instruction must
3661 be <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
3662 integer values. Both arguments must have identical types.</p>
3665 <p>The value produced is the integer difference of the two operands.</p>
3667 <p>If the difference has unsigned overflow, the result returned is the
3668 mathematical result modulo 2<sup>n</sup>, where n is the bit width of the
3671 <p>Because LLVM integers use a two's complement representation, this instruction
3672 is appropriate for both signed and unsigned integers.</p>
3674 <p><tt>nuw</tt> and <tt>nsw</tt> stand for "No Unsigned Wrap"
3675 and "No Signed Wrap", respectively. If the <tt>nuw</tt> and/or
3676 <tt>nsw</tt> keywords are present, the result value of the <tt>sub</tt>
3677 is a <a href="#trapvalues">trap value</a> if unsigned and/or signed overflow,
3678 respectively, occurs.</p>
3682 <result> = sub i32 4, %var <i>; yields {i32}:result = 4 - %var</i>
3683 <result> = sub i32 0, %val <i>; yields {i32}:result = -%var</i>
3688 <!-- _______________________________________________________________________ -->
3690 <a name="i_fsub">'<tt>fsub</tt>' Instruction</a>
3697 <result> = fsub <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3701 <p>The '<tt>fsub</tt>' instruction returns the difference of its two
3704 <p>Note that the '<tt>fsub</tt>' instruction is used to represent the
3705 '<tt>fneg</tt>' instruction present in most other intermediate
3706 representations.</p>
3709 <p>The two arguments to the '<tt>fsub</tt>' instruction must be
3710 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3711 floating point values. Both arguments must have identical types.</p>
3714 <p>The value produced is the floating point difference of the two operands.</p>
3718 <result> = fsub float 4.0, %var <i>; yields {float}:result = 4.0 - %var</i>
3719 <result> = fsub float -0.0, %val <i>; yields {float}:result = -%var</i>
3724 <!-- _______________________________________________________________________ -->
3726 <a name="i_mul">'<tt>mul</tt>' Instruction</a>
3733 <result> = mul <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3734 <result> = mul nuw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3735 <result> = mul nsw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3736 <result> = mul nuw nsw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3740 <p>The '<tt>mul</tt>' instruction returns the product of its two operands.</p>
3743 <p>The two arguments to the '<tt>mul</tt>' instruction must
3744 be <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
3745 integer values. Both arguments must have identical types.</p>
3748 <p>The value produced is the integer product of the two operands.</p>
3750 <p>If the result of the multiplication has unsigned overflow, the result
3751 returned is the mathematical result modulo 2<sup>n</sup>, where n is the bit
3752 width of the result.</p>
3754 <p>Because LLVM integers use a two's complement representation, and the result
3755 is the same width as the operands, this instruction returns the correct
3756 result for both signed and unsigned integers. If a full product
3757 (e.g. <tt>i32</tt>x<tt>i32</tt>-><tt>i64</tt>) is needed, the operands should
3758 be sign-extended or zero-extended as appropriate to the width of the full
3761 <p><tt>nuw</tt> and <tt>nsw</tt> stand for "No Unsigned Wrap"
3762 and "No Signed Wrap", respectively. If the <tt>nuw</tt> and/or
3763 <tt>nsw</tt> keywords are present, the result value of the <tt>mul</tt>
3764 is a <a href="#trapvalues">trap value</a> if unsigned and/or signed overflow,
3765 respectively, occurs.</p>
3769 <result> = mul i32 4, %var <i>; yields {i32}:result = 4 * %var</i>
3774 <!-- _______________________________________________________________________ -->
3776 <a name="i_fmul">'<tt>fmul</tt>' Instruction</a>
3783 <result> = fmul <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3787 <p>The '<tt>fmul</tt>' instruction returns the product of its two operands.</p>
3790 <p>The two arguments to the '<tt>fmul</tt>' instruction must be
3791 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3792 floating point values. Both arguments must have identical types.</p>
3795 <p>The value produced is the floating point product of the two operands.</p>
3799 <result> = fmul float 4.0, %var <i>; yields {float}:result = 4.0 * %var</i>
3804 <!-- _______________________________________________________________________ -->
3806 <a name="i_udiv">'<tt>udiv</tt>' Instruction</a>
3813 <result> = udiv <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3814 <result> = udiv exact <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3818 <p>The '<tt>udiv</tt>' instruction returns the quotient of its two operands.</p>
3821 <p>The two arguments to the '<tt>udiv</tt>' instruction must be
3822 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3823 values. Both arguments must have identical types.</p>
3826 <p>The value produced is the unsigned integer quotient of the two operands.</p>
3828 <p>Note that unsigned integer division and signed integer division are distinct
3829 operations; for signed integer division, use '<tt>sdiv</tt>'.</p>
3831 <p>Division by zero leads to undefined behavior.</p>
3833 <p>If the <tt>exact</tt> keyword is present, the result value of the
3834 <tt>udiv</tt> is a <a href="#trapvalues">trap value</a> if %op1 is not a
3835 multiple of %op2 (as such, "((a udiv exact b) mul b) == a").</p>
3840 <result> = udiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
3845 <!-- _______________________________________________________________________ -->
3847 <a name="i_sdiv">'<tt>sdiv</tt>' Instruction</a>
3854 <result> = sdiv <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3855 <result> = sdiv exact <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3859 <p>The '<tt>sdiv</tt>' instruction returns the quotient of its two operands.</p>
3862 <p>The two arguments to the '<tt>sdiv</tt>' instruction must be
3863 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3864 values. Both arguments must have identical types.</p>
3867 <p>The value produced is the signed integer quotient of the two operands rounded
3870 <p>Note that signed integer division and unsigned integer division are distinct
3871 operations; for unsigned integer division, use '<tt>udiv</tt>'.</p>
3873 <p>Division by zero leads to undefined behavior. Overflow also leads to
3874 undefined behavior; this is a rare case, but can occur, for example, by doing
3875 a 32-bit division of -2147483648 by -1.</p>
3877 <p>If the <tt>exact</tt> keyword is present, the result value of the
3878 <tt>sdiv</tt> is a <a href="#trapvalues">trap value</a> if the result would
3883 <result> = sdiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
3888 <!-- _______________________________________________________________________ -->
3890 <a name="i_fdiv">'<tt>fdiv</tt>' Instruction</a>
3897 <result> = fdiv <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3901 <p>The '<tt>fdiv</tt>' instruction returns the quotient of its two operands.</p>
3904 <p>The two arguments to the '<tt>fdiv</tt>' instruction must be
3905 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3906 floating point values. Both arguments must have identical types.</p>
3909 <p>The value produced is the floating point quotient of the two operands.</p>
3913 <result> = fdiv float 4.0, %var <i>; yields {float}:result = 4.0 / %var</i>
3918 <!-- _______________________________________________________________________ -->
3920 <a name="i_urem">'<tt>urem</tt>' Instruction</a>
3927 <result> = urem <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3931 <p>The '<tt>urem</tt>' instruction returns the remainder from the unsigned
3932 division of its two arguments.</p>
3935 <p>The two arguments to the '<tt>urem</tt>' instruction must be
3936 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3937 values. Both arguments must have identical types.</p>
3940 <p>This instruction returns the unsigned integer <i>remainder</i> of a division.
3941 This instruction always performs an unsigned division to get the
3944 <p>Note that unsigned integer remainder and signed integer remainder are
3945 distinct operations; for signed integer remainder, use '<tt>srem</tt>'.</p>
3947 <p>Taking the remainder of a division by zero leads to undefined behavior.</p>
3951 <result> = urem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
3956 <!-- _______________________________________________________________________ -->
3958 <a name="i_srem">'<tt>srem</tt>' Instruction</a>
3965 <result> = srem <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3969 <p>The '<tt>srem</tt>' instruction returns the remainder from the signed
3970 division of its two operands. This instruction can also take
3971 <a href="#t_vector">vector</a> versions of the values in which case the
3972 elements must be integers.</p>
3975 <p>The two arguments to the '<tt>srem</tt>' instruction must be
3976 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3977 values. Both arguments must have identical types.</p>
3980 <p>This instruction returns the <i>remainder</i> of a division (where the result
3981 is either zero or has the same sign as the dividend, <tt>op1</tt>), not the
3982 <i>modulo</i> operator (where the result is either zero or has the same sign
3983 as the divisor, <tt>op2</tt>) of a value.
3984 For more information about the difference,
3985 see <a href="http://mathforum.org/dr.math/problems/anne.4.28.99.html">The
3986 Math Forum</a>. For a table of how this is implemented in various languages,
3987 please see <a href="http://en.wikipedia.org/wiki/Modulo_operation">
3988 Wikipedia: modulo operation</a>.</p>
3990 <p>Note that signed integer remainder and unsigned integer remainder are
3991 distinct operations; for unsigned integer remainder, use '<tt>urem</tt>'.</p>
3993 <p>Taking the remainder of a division by zero leads to undefined behavior.
3994 Overflow also leads to undefined behavior; this is a rare case, but can
3995 occur, for example, by taking the remainder of a 32-bit division of
3996 -2147483648 by -1. (The remainder doesn't actually overflow, but this rule
3997 lets srem be implemented using instructions that return both the result of
3998 the division and the remainder.)</p>
4002 <result> = srem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
4007 <!-- _______________________________________________________________________ -->
4009 <a name="i_frem">'<tt>frem</tt>' Instruction</a>
4016 <result> = frem <ty> <op1>, <op2> <i>; yields {ty}:result</i>
4020 <p>The '<tt>frem</tt>' instruction returns the remainder from the division of
4021 its two operands.</p>
4024 <p>The two arguments to the '<tt>frem</tt>' instruction must be
4025 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
4026 floating point values. Both arguments must have identical types.</p>
4029 <p>This instruction returns the <i>remainder</i> of a division. The remainder
4030 has the same sign as the dividend.</p>
4034 <result> = frem float 4.0, %var <i>; yields {float}:result = 4.0 % %var</i>
4041 <!-- ======================================================================= -->
4043 <a name="bitwiseops">Bitwise Binary Operations</a>
4048 <p>Bitwise binary operators are used to do various forms of bit-twiddling in a
4049 program. They are generally very efficient instructions and can commonly be
4050 strength reduced from other instructions. They require two operands of the
4051 same type, execute an operation on them, and produce a single value. The
4052 resulting value is the same type as its operands.</p>
4054 <!-- _______________________________________________________________________ -->
4056 <a name="i_shl">'<tt>shl</tt>' Instruction</a>
4063 <result> = shl <ty> <op1>, <op2> <i>; yields {ty}:result</i>
4064 <result> = shl nuw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
4065 <result> = shl nsw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
4066 <result> = shl nuw nsw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
4070 <p>The '<tt>shl</tt>' instruction returns the first operand shifted to the left
4071 a specified number of bits.</p>
4074 <p>Both arguments to the '<tt>shl</tt>' instruction must be the
4075 same <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
4076 integer type. '<tt>op2</tt>' is treated as an unsigned value.</p>
4079 <p>The value produced is <tt>op1</tt> * 2<sup><tt>op2</tt></sup> mod
4080 2<sup>n</sup>, where <tt>n</tt> is the width of the result. If <tt>op2</tt>
4081 is (statically or dynamically) negative or equal to or larger than the number
4082 of bits in <tt>op1</tt>, the result is undefined. If the arguments are
4083 vectors, each vector element of <tt>op1</tt> is shifted by the corresponding
4084 shift amount in <tt>op2</tt>.</p>
4086 <p>If the <tt>nuw</tt> keyword is present, then the shift produces a
4087 <a href="#trapvalues">trap value</a> if it shifts out any non-zero bits. If
4088 the <tt>nsw</tt> keyword is present, then the shift produces a
4089 <a href="#trapvalues">trap value</a> if it shifts out any bits that disagree
4090 with the resultant sign bit. As such, NUW/NSW have the same semantics as
4091 they would if the shift were expressed as a mul instruction with the same
4092 nsw/nuw bits in (mul %op1, (shl 1, %op2)).</p>
4096 <result> = shl i32 4, %var <i>; yields {i32}: 4 << %var</i>
4097 <result> = shl i32 4, 2 <i>; yields {i32}: 16</i>
4098 <result> = shl i32 1, 10 <i>; yields {i32}: 1024</i>
4099 <result> = shl i32 1, 32 <i>; undefined</i>
4100 <result> = shl <2 x i32> < i32 1, i32 1>, < i32 1, i32 2> <i>; yields: result=<2 x i32> < i32 2, i32 4></i>
4105 <!-- _______________________________________________________________________ -->
4107 <a name="i_lshr">'<tt>lshr</tt>' Instruction</a>
4114 <result> = lshr <ty> <op1>, <op2> <i>; yields {ty}:result</i>
4115 <result> = lshr exact <ty> <op1>, <op2> <i>; yields {ty}:result</i>
4119 <p>The '<tt>lshr</tt>' instruction (logical shift right) returns the first
4120 operand shifted to the right a specified number of bits with zero fill.</p>
4123 <p>Both arguments to the '<tt>lshr</tt>' instruction must be the same
4124 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
4125 type. '<tt>op2</tt>' is treated as an unsigned value.</p>
4128 <p>This instruction always performs a logical shift right operation. The most
4129 significant bits of the result will be filled with zero bits after the shift.
4130 If <tt>op2</tt> is (statically or dynamically) equal to or larger than the
4131 number of bits in <tt>op1</tt>, the result is undefined. If the arguments are
4132 vectors, each vector element of <tt>op1</tt> is shifted by the corresponding
4133 shift amount in <tt>op2</tt>.</p>
4135 <p>If the <tt>exact</tt> keyword is present, the result value of the
4136 <tt>lshr</tt> is a <a href="#trapvalues">trap value</a> if any of the bits
4137 shifted out are non-zero.</p>
4142 <result> = lshr i32 4, 1 <i>; yields {i32}:result = 2</i>
4143 <result> = lshr i32 4, 2 <i>; yields {i32}:result = 1</i>
4144 <result> = lshr i8 4, 3 <i>; yields {i8}:result = 0</i>
4145 <result> = lshr i8 -2, 1 <i>; yields {i8}:result = 0x7FFFFFFF </i>
4146 <result> = lshr i32 1, 32 <i>; undefined</i>
4147 <result> = lshr <2 x i32> < i32 -2, i32 4>, < i32 1, i32 2> <i>; yields: result=<2 x i32> < i32 0x7FFFFFFF, i32 1></i>
4152 <!-- _______________________________________________________________________ -->
4154 <a name="i_ashr">'<tt>ashr</tt>' Instruction</a>
4161 <result> = ashr <ty> <op1>, <op2> <i>; yields {ty}:result</i>
4162 <result> = ashr exact <ty> <op1>, <op2> <i>; yields {ty}:result</i>
4166 <p>The '<tt>ashr</tt>' instruction (arithmetic shift right) returns the first
4167 operand shifted to the right a specified number of bits with sign
4171 <p>Both arguments to the '<tt>ashr</tt>' instruction must be the same
4172 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
4173 type. '<tt>op2</tt>' is treated as an unsigned value.</p>
4176 <p>This instruction always performs an arithmetic shift right operation, The
4177 most significant bits of the result will be filled with the sign bit
4178 of <tt>op1</tt>. If <tt>op2</tt> is (statically or dynamically) equal to or
4179 larger than the number of bits in <tt>op1</tt>, the result is undefined. If
4180 the arguments are vectors, each vector element of <tt>op1</tt> is shifted by
4181 the corresponding shift amount in <tt>op2</tt>.</p>
4183 <p>If the <tt>exact</tt> keyword is present, the result value of the
4184 <tt>ashr</tt> is a <a href="#trapvalues">trap value</a> if any of the bits
4185 shifted out are non-zero.</p>
4189 <result> = ashr i32 4, 1 <i>; yields {i32}:result = 2</i>
4190 <result> = ashr i32 4, 2 <i>; yields {i32}:result = 1</i>
4191 <result> = ashr i8 4, 3 <i>; yields {i8}:result = 0</i>
4192 <result> = ashr i8 -2, 1 <i>; yields {i8}:result = -1</i>
4193 <result> = ashr i32 1, 32 <i>; undefined</i>
4194 <result> = ashr <2 x i32> < i32 -2, i32 4>, < i32 1, i32 3> <i>; yields: result=<2 x i32> < i32 -1, i32 0></i>
4199 <!-- _______________________________________________________________________ -->
4201 <a name="i_and">'<tt>and</tt>' Instruction</a>
4208 <result> = and <ty> <op1>, <op2> <i>; yields {ty}:result</i>
4212 <p>The '<tt>and</tt>' instruction returns the bitwise logical and of its two
4216 <p>The two arguments to the '<tt>and</tt>' instruction must be
4217 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
4218 values. Both arguments must have identical types.</p>
4221 <p>The truth table used for the '<tt>and</tt>' instruction is:</p>
4223 <table border="1" cellspacing="0" cellpadding="4">
4255 <result> = and i32 4, %var <i>; yields {i32}:result = 4 & %var</i>
4256 <result> = and i32 15, 40 <i>; yields {i32}:result = 8</i>
4257 <result> = and i32 4, 8 <i>; yields {i32}:result = 0</i>
4260 <!-- _______________________________________________________________________ -->
4262 <a name="i_or">'<tt>or</tt>' Instruction</a>
4269 <result> = or <ty> <op1>, <op2> <i>; yields {ty}:result</i>
4273 <p>The '<tt>or</tt>' instruction returns the bitwise logical inclusive or of its
4277 <p>The two arguments to the '<tt>or</tt>' instruction must be
4278 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
4279 values. Both arguments must have identical types.</p>
4282 <p>The truth table used for the '<tt>or</tt>' instruction is:</p>
4284 <table border="1" cellspacing="0" cellpadding="4">
4316 <result> = or i32 4, %var <i>; yields {i32}:result = 4 | %var</i>
4317 <result> = or i32 15, 40 <i>; yields {i32}:result = 47</i>
4318 <result> = or i32 4, 8 <i>; yields {i32}:result = 12</i>
4323 <!-- _______________________________________________________________________ -->
4325 <a name="i_xor">'<tt>xor</tt>' Instruction</a>
4332 <result> = xor <ty> <op1>, <op2> <i>; yields {ty}:result</i>
4336 <p>The '<tt>xor</tt>' instruction returns the bitwise logical exclusive or of
4337 its two operands. The <tt>xor</tt> is used to implement the "one's
4338 complement" operation, which is the "~" operator in C.</p>
4341 <p>The two arguments to the '<tt>xor</tt>' instruction must be
4342 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
4343 values. Both arguments must have identical types.</p>
4346 <p>The truth table used for the '<tt>xor</tt>' instruction is:</p>
4348 <table border="1" cellspacing="0" cellpadding="4">
4380 <result> = xor i32 4, %var <i>; yields {i32}:result = 4 ^ %var</i>
4381 <result> = xor i32 15, 40 <i>; yields {i32}:result = 39</i>
4382 <result> = xor i32 4, 8 <i>; yields {i32}:result = 12</i>
4383 <result> = xor i32 %V, -1 <i>; yields {i32}:result = ~%V</i>
4390 <!-- ======================================================================= -->
4392 <a name="vectorops">Vector Operations</a>
4397 <p>LLVM supports several instructions to represent vector operations in a
4398 target-independent manner. These instructions cover the element-access and
4399 vector-specific operations needed to process vectors effectively. While LLVM
4400 does directly support these vector operations, many sophisticated algorithms
4401 will want to use target-specific intrinsics to take full advantage of a
4402 specific target.</p>
4404 <!-- _______________________________________________________________________ -->
4406 <a name="i_extractelement">'<tt>extractelement</tt>' Instruction</a>
4413 <result> = extractelement <n x <ty>> <val>, i32 <idx> <i>; yields <ty></i>
4417 <p>The '<tt>extractelement</tt>' instruction extracts a single scalar element
4418 from a vector at a specified index.</p>
4422 <p>The first operand of an '<tt>extractelement</tt>' instruction is a value
4423 of <a href="#t_vector">vector</a> type. The second operand is an index
4424 indicating the position from which to extract the element. The index may be
4428 <p>The result is a scalar of the same type as the element type of
4429 <tt>val</tt>. Its value is the value at position <tt>idx</tt> of
4430 <tt>val</tt>. If <tt>idx</tt> exceeds the length of <tt>val</tt>, the
4431 results are undefined.</p>
4435 <result> = extractelement <4 x i32> %vec, i32 0 <i>; yields i32</i>
4440 <!-- _______________________________________________________________________ -->
4442 <a name="i_insertelement">'<tt>insertelement</tt>' Instruction</a>
4449 <result> = insertelement <n x <ty>> <val>, <ty> <elt>, i32 <idx> <i>; yields <n x <ty>></i>
4453 <p>The '<tt>insertelement</tt>' instruction inserts a scalar element into a
4454 vector at a specified index.</p>
4457 <p>The first operand of an '<tt>insertelement</tt>' instruction is a value
4458 of <a href="#t_vector">vector</a> type. The second operand is a scalar value
4459 whose type must equal the element type of the first operand. The third
4460 operand is an index indicating the position at which to insert the value.
4461 The index may be a variable.</p>
4464 <p>The result is a vector of the same type as <tt>val</tt>. Its element values
4465 are those of <tt>val</tt> except at position <tt>idx</tt>, where it gets the
4466 value <tt>elt</tt>. If <tt>idx</tt> exceeds the length of <tt>val</tt>, the
4467 results are undefined.</p>
4471 <result> = insertelement <4 x i32> %vec, i32 1, i32 0 <i>; yields <4 x i32></i>
4476 <!-- _______________________________________________________________________ -->
4478 <a name="i_shufflevector">'<tt>shufflevector</tt>' Instruction</a>
4485 <result> = shufflevector <n x <ty>> <v1>, <n x <ty>> <v2>, <m x i32> <mask> <i>; yields <m x <ty>></i>
4489 <p>The '<tt>shufflevector</tt>' instruction constructs a permutation of elements
4490 from two input vectors, returning a vector with the same element type as the
4491 input and length that is the same as the shuffle mask.</p>
4494 <p>The first two operands of a '<tt>shufflevector</tt>' instruction are vectors
4495 with types that match each other. The third argument is a shuffle mask whose
4496 element type is always 'i32'. The result of the instruction is a vector
4497 whose length is the same as the shuffle mask and whose element type is the
4498 same as the element type of the first two operands.</p>
4500 <p>The shuffle mask operand is required to be a constant vector with either
4501 constant integer or undef values.</p>
4504 <p>The elements of the two input vectors are numbered from left to right across
4505 both of the vectors. The shuffle mask operand specifies, for each element of
4506 the result vector, which element of the two input vectors the result element
4507 gets. The element selector may be undef (meaning "don't care") and the
4508 second operand may be undef if performing a shuffle from only one vector.</p>
4512 <result> = shufflevector <4 x i32> %v1, <4 x i32> %v2,
4513 <4 x i32> <i32 0, i32 4, i32 1, i32 5> <i>; yields <4 x i32></i>
4514 <result> = shufflevector <4 x i32> %v1, <4 x i32> undef,
4515 <4 x i32> <i32 0, i32 1, i32 2, i32 3> <i>; yields <4 x i32></i> - Identity shuffle.
4516 <result> = shufflevector <8 x i32> %v1, <8 x i32> undef,
4517 <4 x i32> <i32 0, i32 1, i32 2, i32 3> <i>; yields <4 x i32></i>
4518 <result> = shufflevector <4 x i32> %v1, <4 x i32> %v2,
4519 <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7 > <i>; yields <8 x i32></i>
4526 <!-- ======================================================================= -->
4528 <a name="aggregateops">Aggregate Operations</a>
4533 <p>LLVM supports several instructions for working with
4534 <a href="#t_aggregate">aggregate</a> values.</p>
4536 <!-- _______________________________________________________________________ -->
4538 <a name="i_extractvalue">'<tt>extractvalue</tt>' Instruction</a>
4545 <result> = extractvalue <aggregate type> <val>, <idx>{, <idx>}*
4549 <p>The '<tt>extractvalue</tt>' instruction extracts the value of a member field
4550 from an <a href="#t_aggregate">aggregate</a> value.</p>
4553 <p>The first operand of an '<tt>extractvalue</tt>' instruction is a value
4554 of <a href="#t_struct">struct</a> or
4555 <a href="#t_array">array</a> type. The operands are constant indices to
4556 specify which value to extract in a similar manner as indices in a
4557 '<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.</p>
4558 <p>The major differences to <tt>getelementptr</tt> indexing are:</p>
4560 <li>Since the value being indexed is not a pointer, the first index is
4561 omitted and assumed to be zero.</li>
4562 <li>At least one index must be specified.</li>
4563 <li>Not only struct indices but also array indices must be in
4568 <p>The result is the value at the position in the aggregate specified by the
4573 <result> = extractvalue {i32, float} %agg, 0 <i>; yields i32</i>
4578 <!-- _______________________________________________________________________ -->
4580 <a name="i_insertvalue">'<tt>insertvalue</tt>' Instruction</a>
4587 <result> = insertvalue <aggregate type> <val>, <ty> <elt>, <idx>{, <idx>}* <i>; yields <aggregate type></i>
4591 <p>The '<tt>insertvalue</tt>' instruction inserts a value into a member field
4592 in an <a href="#t_aggregate">aggregate</a> value.</p>
4595 <p>The first operand of an '<tt>insertvalue</tt>' instruction is a value
4596 of <a href="#t_struct">struct</a> or
4597 <a href="#t_array">array</a> type. The second operand is a first-class
4598 value to insert. The following operands are constant indices indicating
4599 the position at which to insert the value in a similar manner as indices in a
4600 '<tt><a href="#i_extractvalue">extractvalue</a></tt>' instruction. The
4601 value to insert must have the same type as the value identified by the
4605 <p>The result is an aggregate of the same type as <tt>val</tt>. Its value is
4606 that of <tt>val</tt> except that the value at the position specified by the
4607 indices is that of <tt>elt</tt>.</p>
4611 %agg1 = insertvalue {i32, float} undef, i32 1, 0 <i>; yields {i32 1, float undef}</i>
4612 %agg2 = insertvalue {i32, float} %agg1, float %val, 1 <i>; yields {i32 1, float %val}</i>
4613 %agg3 = insertvalue {i32, {float}} %agg1, float %val, 1, 0 <i>; yields {i32 1, float %val}</i>
4620 <!-- ======================================================================= -->
4622 <a name="memoryops">Memory Access and Addressing Operations</a>
4627 <p>A key design point of an SSA-based representation is how it represents
4628 memory. In LLVM, no memory locations are in SSA form, which makes things
4629 very simple. This section describes how to read, write, and allocate
4632 <!-- _______________________________________________________________________ -->
4634 <a name="i_alloca">'<tt>alloca</tt>' Instruction</a>
4641 <result> = alloca <type>[, <ty> <NumElements>][, align <alignment>] <i>; yields {type*}:result</i>
4645 <p>The '<tt>alloca</tt>' instruction allocates memory on the stack frame of the
4646 currently executing function, to be automatically released when this function
4647 returns to its caller. The object is always allocated in the generic address
4648 space (address space zero).</p>
4651 <p>The '<tt>alloca</tt>' instruction
4652 allocates <tt>sizeof(<type>)*NumElements</tt> bytes of memory on the
4653 runtime stack, returning a pointer of the appropriate type to the program.
4654 If "NumElements" is specified, it is the number of elements allocated,
4655 otherwise "NumElements" is defaulted to be one. If a constant alignment is
4656 specified, the value result of the allocation is guaranteed to be aligned to
4657 at least that boundary. If not specified, or if zero, the target can choose
4658 to align the allocation on any convenient boundary compatible with the
4661 <p>'<tt>type</tt>' may be any sized type.</p>
4664 <p>Memory is allocated; a pointer is returned. The operation is undefined if
4665 there is insufficient stack space for the allocation. '<tt>alloca</tt>'d
4666 memory is automatically released when the function returns. The
4667 '<tt>alloca</tt>' instruction is commonly used to represent automatic
4668 variables that must have an address available. When the function returns
4669 (either with the <tt><a href="#i_ret">ret</a></tt>
4670 or <tt><a href="#i_unwind">unwind</a></tt> instructions), the memory is
4671 reclaimed. Allocating zero bytes is legal, but the result is undefined.</p>
4675 %ptr = alloca i32 <i>; yields {i32*}:ptr</i>
4676 %ptr = alloca i32, i32 4 <i>; yields {i32*}:ptr</i>
4677 %ptr = alloca i32, i32 4, align 1024 <i>; yields {i32*}:ptr</i>
4678 %ptr = alloca i32, align 1024 <i>; yields {i32*}:ptr</i>
4683 <!-- _______________________________________________________________________ -->
4685 <a name="i_load">'<tt>load</tt>' Instruction</a>
4692 <result> = load [volatile] <ty>* <pointer>[, align <alignment>][, !nontemporal !<index>]
4693 <result> = load atomic [volatile] <ty>* <pointer> [singlethread] <ordering>, align <alignment>
4694 !<index> = !{ i32 1 }
4698 <p>The '<tt>load</tt>' instruction is used to read from memory.</p>
4701 <p>The argument to the '<tt>load</tt>' instruction specifies the memory address
4702 from which to load. The pointer must point to
4703 a <a href="#t_firstclass">first class</a> type. If the <tt>load</tt> is
4704 marked as <tt>volatile</tt>, then the optimizer is not allowed to modify the
4705 number or order of execution of this <tt>load</tt> with other <a
4706 href="#volatile">volatile operations</a>.</p>
4708 <p>If the <code>load</code> is marked as <code>atomic</code>, it takes an extra
4709 <a href="#ordering">ordering</a> and optional <code>singlethread</code>
4710 argument. The <code>release</code> and <code>acq_rel</code> orderings are
4711 not valid on <code>load</code> instructions. Atomic loads produce <a
4712 href="#memorymodel">defined</a> results when they may see multiple atomic
4713 stores. The type of the pointee must be an integer type whose bit width
4714 is a power of two greater than or equal to eight and less than or equal
4715 to a target-specific size limit. <code>align</code> must be explicitly
4716 specified on atomic loads, and the load has undefined behavior if the
4717 alignment is not set to a value which is at least the size in bytes of
4718 the pointee. <code>!nontemporal</code> does not have any defined semantics
4719 for atomic loads.</p>
4721 <p>The optional constant <tt>align</tt> argument specifies the alignment of the
4722 operation (that is, the alignment of the memory address). A value of 0 or an
4723 omitted <tt>align</tt> argument means that the operation has the preferential
4724 alignment for the target. It is the responsibility of the code emitter to
4725 ensure that the alignment information is correct. Overestimating the
4726 alignment results in undefined behavior. Underestimating the alignment may
4727 produce less efficient code. An alignment of 1 is always safe.</p>
4729 <p>The optional <tt>!nontemporal</tt> metadata must reference a single
4730 metatadata name <index> corresponding to a metadata node with
4731 one <tt>i32</tt> entry of value 1. The existence of
4732 the <tt>!nontemporal</tt> metatadata on the instruction tells the optimizer
4733 and code generator that this load is not expected to be reused in the cache.
4734 The code generator may select special instructions to save cache bandwidth,
4735 such as the <tt>MOVNT</tt> instruction on x86.</p>
4738 <p>The location of memory pointed to is loaded. If the value being loaded is of
4739 scalar type then the number of bytes read does not exceed the minimum number
4740 of bytes needed to hold all bits of the type. For example, loading an
4741 <tt>i24</tt> reads at most three bytes. When loading a value of a type like
4742 <tt>i20</tt> with a size that is not an integral number of bytes, the result
4743 is undefined if the value was not originally written using a store of the
4748 %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
4749 <a href="#i_store">store</a> i32 3, i32* %ptr <i>; yields {void}</i>
4750 %val = load i32* %ptr <i>; yields {i32}:val = i32 3</i>
4755 <!-- _______________________________________________________________________ -->
4757 <a name="i_store">'<tt>store</tt>' Instruction</a>
4764 store [volatile] <ty> <value>, <ty>* <pointer>[, align <alignment>][, !nontemporal !<index>] <i>; yields {void}</i>
4765 store atomic [volatile] <ty> <value>, <ty>* <pointer> [singlethread] <ordering>, align <alignment> <i>; yields {void}</i>
4769 <p>The '<tt>store</tt>' instruction is used to write to memory.</p>
4772 <p>There are two arguments to the '<tt>store</tt>' instruction: a value to store
4773 and an address at which to store it. The type of the
4774 '<tt><pointer></tt>' operand must be a pointer to
4775 the <a href="#t_firstclass">first class</a> type of the
4776 '<tt><value></tt>' operand. If the <tt>store</tt> is marked as
4777 <tt>volatile</tt>, then the optimizer is not allowed to modify the number or
4778 order of execution of this <tt>store</tt> with other <a
4779 href="#volatile">volatile operations</a>.</p>
4781 <p>If the <code>store</code> is marked as <code>atomic</code>, it takes an extra
4782 <a href="#ordering">ordering</a> and optional <code>singlethread</code>
4783 argument. The <code>acquire</code> and <code>acq_rel</code> orderings aren't
4784 valid on <code>store</code> instructions. Atomic loads produce <a
4785 href="#memorymodel">defined</a> results when they may see multiple atomic
4786 stores. The type of the pointee must be an integer type whose bit width
4787 is a power of two greater than or equal to eight and less than or equal
4788 to a target-specific size limit. <code>align</code> must be explicitly
4789 specified on atomic stores, and the store has undefined behavior if the
4790 alignment is not set to a value which is at least the size in bytes of
4791 the pointee. <code>!nontemporal</code> does not have any defined semantics
4792 for atomic stores.</p>
4794 <p>The optional constant "align" argument specifies the alignment of the
4795 operation (that is, the alignment of the memory address). A value of 0 or an
4796 omitted "align" argument means that the operation has the preferential
4797 alignment for the target. It is the responsibility of the code emitter to
4798 ensure that the alignment information is correct. Overestimating the
4799 alignment results in an undefined behavior. Underestimating the alignment may
4800 produce less efficient code. An alignment of 1 is always safe.</p>
4802 <p>The optional !nontemporal metadata must reference a single metatadata
4803 name <index> corresponding to a metadata node with one i32 entry of
4804 value 1. The existence of the !nontemporal metatadata on the
4805 instruction tells the optimizer and code generator that this load is
4806 not expected to be reused in the cache. The code generator may
4807 select special instructions to save cache bandwidth, such as the
4808 MOVNT instruction on x86.</p>
4812 <p>The contents of memory are updated to contain '<tt><value></tt>' at the
4813 location specified by the '<tt><pointer></tt>' operand. If
4814 '<tt><value></tt>' is of scalar type then the number of bytes written
4815 does not exceed the minimum number of bytes needed to hold all bits of the
4816 type. For example, storing an <tt>i24</tt> writes at most three bytes. When
4817 writing a value of a type like <tt>i20</tt> with a size that is not an
4818 integral number of bytes, it is unspecified what happens to the extra bits
4819 that do not belong to the type, but they will typically be overwritten.</p>
4823 %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
4824 store i32 3, i32* %ptr <i>; yields {void}</i>
4825 %val = <a href="#i_load">load</a> i32* %ptr <i>; yields {i32}:val = i32 3</i>
4830 <!-- _______________________________________________________________________ -->
4832 <a name="i_fence">'<tt>fence</tt>' Instruction</a>
4839 fence [singlethread] <ordering> <i>; yields {void}</i>
4843 <p>The '<tt>fence</tt>' instruction is used to introduce happens-before edges
4844 between operations.</p>
4846 <h5>Arguments:</h5> <p>'<code>fence</code>' instructions take an <a
4847 href="#ordering">ordering</a> argument which defines what
4848 <i>synchronizes-with</i> edges they add. They can only be given
4849 <code>acquire</code>, <code>release</code>, <code>acq_rel</code>, and
4850 <code>seq_cst</code> orderings.</p>
4853 <p>A fence <var>A</var> which has (at least) <code>release</code> ordering
4854 semantics <i>synchronizes with</i> a fence <var>B</var> with (at least)
4855 <code>acquire</code> ordering semantics if and only if there exist atomic
4856 operations <var>X</var> and <var>Y</var>, both operating on some atomic object
4857 <var>M</var>, such that <var>A</var> is sequenced before <var>X</var>,
4858 <var>X</var> modifies <var>M</var> (either directly or through some side effect
4859 of a sequence headed by <var>X</var>), <var>Y</var> is sequenced before
4860 <var>B</var>, and <var>Y</var> observes <var>M</var>. This provides a
4861 <i>happens-before</i> dependency between <var>A</var> and <var>B</var>. Rather
4862 than an explicit <code>fence</code>, one (but not both) of the atomic operations
4863 <var>X</var> or <var>Y</var> might provide a <code>release</code> or
4864 <code>acquire</code> (resp.) ordering constraint and still
4865 <i>synchronize-with</i> the explicit <code>fence</code> and establish the
4866 <i>happens-before</i> edge.</p>
4868 <p>A <code>fence</code> which has <code>seq_cst</code> ordering, in addition to
4869 having both <code>acquire</code> and <code>release</code> semantics specified
4870 above, participates in the global program order of other <code>seq_cst</code>
4871 operations and/or fences.</p>
4873 <p>The optional "<a href="#singlethread"><code>singlethread</code></a>" argument
4874 specifies that the fence only synchronizes with other fences in the same
4875 thread. (This is useful for interacting with signal handlers.)</p>
4879 fence acquire <i>; yields {void}</i>
4880 fence singlethread seq_cst <i>; yields {void}</i>
4885 <!-- _______________________________________________________________________ -->
4887 <a name="i_cmpxchg">'<tt>cmpxchg</tt>' Instruction</a>
4894 cmpxchg [volatile] <ty>* <pointer>, <ty> <cmp>, <ty> <new> [singlethread] <ordering> <i>; yields {ty}</i>
4898 <p>The '<tt>cmpxchg</tt>' instruction is used to atomically modify memory.
4899 It loads a value in memory and compares it to a given value. If they are
4900 equal, it stores a new value into the memory.</p>
4903 <p>There are three arguments to the '<code>cmpxchg</code>' instruction: an
4904 address to operate on, a value to compare to the value currently be at that
4905 address, and a new value to place at that address if the compared values are
4906 equal. The type of '<var><cmp></var>' must be an integer type whose
4907 bit width is a power of two greater than or equal to eight and less than
4908 or equal to a target-specific size limit. '<var><cmp></var>' and
4909 '<var><new></var>' must have the same type, and the type of
4910 '<var><pointer></var>' must be a pointer to that type. If the
4911 <code>cmpxchg</code> is marked as <code>volatile</code>, then the
4912 optimizer is not allowed to modify the number or order of execution
4913 of this <code>cmpxchg</code> with other <a href="#volatile">volatile
4916 <!-- FIXME: Extend allowed types. -->
4918 <p>The <a href="#ordering"><var>ordering</var></a> argument specifies how this
4919 <code>cmpxchg</code> synchronizes with other atomic operations.</p>
4921 <p>The optional "<code>singlethread</code>" argument declares that the
4922 <code>cmpxchg</code> is only atomic with respect to code (usually signal
4923 handlers) running in the same thread as the <code>cmpxchg</code>. Otherwise the
4924 cmpxchg is atomic with respect to all other code in the system.</p>
4926 <p>The pointer passed into cmpxchg must have alignment greater than or equal to
4927 the size in memory of the operand.
4930 <p>The contents of memory at the location specified by the
4931 '<tt><pointer></tt>' operand is read and compared to
4932 '<tt><cmp></tt>'; if the read value is the equal,
4933 '<tt><new></tt>' is written. The original value at the location
4936 <p>A successful <code>cmpxchg</code> is a read-modify-write instruction for the
4937 purpose of identifying <a href="#release_sequence">release sequences</a>. A
4938 failed <code>cmpxchg</code> is equivalent to an atomic load with an ordering
4939 parameter determined by dropping any <code>release</code> part of the
4940 <code>cmpxchg</code>'s ordering.</p>
4943 FIXME: Is compare_exchange_weak() necessary? (Consider after we've done
4944 optimization work on ARM.)
4946 FIXME: Is a weaker ordering constraint on failure helpful in practice?
4952 %orig = atomic <a href="#i_load">load</a> i32* %ptr unordered <i>; yields {i32}</i>
4953 <a href="#i_br">br</a> label %loop
4956 %cmp = <a href="#i_phi">phi</a> i32 [ %orig, %entry ], [%old, %loop]
4957 %squared = <a href="#i_mul">mul</a> i32 %cmp, %cmp
4958 %old = cmpxchg i32* %ptr, i32 %cmp, i32 %squared <i>; yields {i32}</i>
4959 %success = <a href="#i_icmp">icmp</a> eq i32 %cmp, %old
4960 <a href="#i_br">br</a> i1 %success, label %done, label %loop
4968 <!-- _______________________________________________________________________ -->
4970 <a name="i_atomicrmw">'<tt>atomicrmw</tt>' Instruction</a>
4977 atomicrmw [volatile] <operation> <ty>* <pointer>, <ty> <value> [singlethread] <ordering> <i>; yields {ty}</i>
4981 <p>The '<tt>atomicrmw</tt>' instruction is used to atomically modify memory.</p>
4984 <p>There are three arguments to the '<code>atomicrmw</code>' instruction: an
4985 operation to apply, an address whose value to modify, an argument to the
4986 operation. The operation must be one of the following keywords:</p>
5001 <p>The type of '<var><value></var>' must be an integer type whose
5002 bit width is a power of two greater than or equal to eight and less than
5003 or equal to a target-specific size limit. The type of the
5004 '<code><pointer></code>' operand must be a pointer to that type.
5005 If the <code>atomicrmw</code> is marked as <code>volatile</code>, then the
5006 optimizer is not allowed to modify the number or order of execution of this
5007 <code>atomicrmw</code> with other <a href="#volatile">volatile
5010 <!-- FIXME: Extend allowed types. -->
5013 <p>The contents of memory at the location specified by the
5014 '<tt><pointer></tt>' operand are atomically read, modified, and written
5015 back. The original value at the location is returned. The modification is
5016 specified by the <var>operation</var> argument:</p>
5019 <li>xchg: <code>*ptr = val</code></li>
5020 <li>add: <code>*ptr = *ptr + val</code></li>
5021 <li>sub: <code>*ptr = *ptr - val</code></li>
5022 <li>and: <code>*ptr = *ptr & val</code></li>
5023 <li>nand: <code>*ptr = ~(*ptr & val)</code></li>
5024 <li>or: <code>*ptr = *ptr | val</code></li>
5025 <li>xor: <code>*ptr = *ptr ^ val</code></li>
5026 <li>max: <code>*ptr = *ptr > val ? *ptr : val</code> (using a signed comparison)</li>
5027 <li>min: <code>*ptr = *ptr < val ? *ptr : val</code> (using a signed comparison)</li>
5028 <li>umax: <code>*ptr = *ptr > val ? *ptr : val</code> (using an unsigned comparison)</li>
5029 <li>umin: <code>*ptr = *ptr < val ? *ptr : val</code> (using an unsigned comparison)</li>
5034 %old = atomicrmw add i32* %ptr, i32 1 acquire <i>; yields {i32}</i>
5039 <!-- _______________________________________________________________________ -->
5041 <a name="i_getelementptr">'<tt>getelementptr</tt>' Instruction</a>
5048 <result> = getelementptr <pty>* <ptrval>{, <ty> <idx>}*
5049 <result> = getelementptr inbounds <pty>* <ptrval>{, <ty> <idx>}*
5053 <p>The '<tt>getelementptr</tt>' instruction is used to get the address of a
5054 subelement of an <a href="#t_aggregate">aggregate</a> data structure.
5055 It performs address calculation only and does not access memory.</p>
5058 <p>The first argument is always a pointer, and forms the basis of the
5059 calculation. The remaining arguments are indices that indicate which of the
5060 elements of the aggregate object are indexed. The interpretation of each
5061 index is dependent on the type being indexed into. The first index always
5062 indexes the pointer value given as the first argument, the second index
5063 indexes a value of the type pointed to (not necessarily the value directly
5064 pointed to, since the first index can be non-zero), etc. The first type
5065 indexed into must be a pointer value, subsequent types can be arrays,
5066 vectors, and structs. Note that subsequent types being indexed into
5067 can never be pointers, since that would require loading the pointer before
5068 continuing calculation.</p>
5070 <p>The type of each index argument depends on the type it is indexing into.
5071 When indexing into a (optionally packed) structure, only <tt>i32</tt>
5072 integer <b>constants</b> are allowed. When indexing into an array, pointer
5073 or vector, integers of any width are allowed, and they are not required to be
5074 constant. These integers are treated as signed values where relevant.</p>
5076 <p>For example, let's consider a C code fragment and how it gets compiled to
5079 <pre class="doc_code">
5091 int *foo(struct ST *s) {
5092 return &s[1].Z.B[5][13];
5096 <p>The LLVM code generated by the GCC frontend is:</p>
5098 <pre class="doc_code">
5099 %RT = <a href="#namedtypes">type</a> { i8 , [10 x [20 x i32]], i8 }
5100 %ST = <a href="#namedtypes">type</a> { i32, double, %RT }
5102 define i32* @foo(%ST* %s) {
5104 %reg = getelementptr %ST* %s, i32 1, i32 2, i32 1, i32 5, i32 13
5110 <p>In the example above, the first index is indexing into the '<tt>%ST*</tt>'
5111 type, which is a pointer, yielding a '<tt>%ST</tt>' = '<tt>{ i32, double, %RT
5112 }</tt>' type, a structure. The second index indexes into the third element
5113 of the structure, yielding a '<tt>%RT</tt>' = '<tt>{ i8 , [10 x [20 x i32]],
5114 i8 }</tt>' type, another structure. The third index indexes into the second
5115 element of the structure, yielding a '<tt>[10 x [20 x i32]]</tt>' type, an
5116 array. The two dimensions of the array are subscripted into, yielding an
5117 '<tt>i32</tt>' type. The '<tt>getelementptr</tt>' instruction returns a
5118 pointer to this element, thus computing a value of '<tt>i32*</tt>' type.</p>
5120 <p>Note that it is perfectly legal to index partially through a structure,
5121 returning a pointer to an inner element. Because of this, the LLVM code for
5122 the given testcase is equivalent to:</p>
5125 define i32* @foo(%ST* %s) {
5126 %t1 = getelementptr %ST* %s, i32 1 <i>; yields %ST*:%t1</i>
5127 %t2 = getelementptr %ST* %t1, i32 0, i32 2 <i>; yields %RT*:%t2</i>
5128 %t3 = getelementptr %RT* %t2, i32 0, i32 1 <i>; yields [10 x [20 x i32]]*:%t3</i>
5129 %t4 = getelementptr [10 x [20 x i32]]* %t3, i32 0, i32 5 <i>; yields [20 x i32]*:%t4</i>
5130 %t5 = getelementptr [20 x i32]* %t4, i32 0, i32 13 <i>; yields i32*:%t5</i>
5135 <p>If the <tt>inbounds</tt> keyword is present, the result value of the
5136 <tt>getelementptr</tt> is a <a href="#trapvalues">trap value</a> if the
5137 base pointer is not an <i>in bounds</i> address of an allocated object,
5138 or if any of the addresses that would be formed by successive addition of
5139 the offsets implied by the indices to the base address with infinitely
5140 precise signed arithmetic are not an <i>in bounds</i> address of that
5141 allocated object. The <i>in bounds</i> addresses for an allocated object
5142 are all the addresses that point into the object, plus the address one
5143 byte past the end.</p>
5145 <p>If the <tt>inbounds</tt> keyword is not present, the offsets are added to
5146 the base address with silently-wrapping two's complement arithmetic. If the
5147 offsets have a different width from the pointer, they are sign-extended or
5148 truncated to the width of the pointer. The result value of the
5149 <tt>getelementptr</tt> may be outside the object pointed to by the base
5150 pointer. The result value may not necessarily be used to access memory
5151 though, even if it happens to point into allocated storage. See the
5152 <a href="#pointeraliasing">Pointer Aliasing Rules</a> section for more
5155 <p>The getelementptr instruction is often confusing. For some more insight into
5156 how it works, see <a href="GetElementPtr.html">the getelementptr FAQ</a>.</p>
5160 <i>; yields [12 x i8]*:aptr</i>
5161 %aptr = getelementptr {i32, [12 x i8]}* %saptr, i64 0, i32 1
5162 <i>; yields i8*:vptr</i>
5163 %vptr = getelementptr {i32, <2 x i8>}* %svptr, i64 0, i32 1, i32 1
5164 <i>; yields i8*:eptr</i>
5165 %eptr = getelementptr [12 x i8]* %aptr, i64 0, i32 1
5166 <i>; yields i32*:iptr</i>
5167 %iptr = getelementptr [10 x i32]* @arr, i16 0, i16 0
5174 <!-- ======================================================================= -->
5176 <a name="convertops">Conversion Operations</a>
5181 <p>The instructions in this category are the conversion instructions (casting)
5182 which all take a single operand and a type. They perform various bit
5183 conversions on the operand.</p>
5185 <!-- _______________________________________________________________________ -->
5187 <a name="i_trunc">'<tt>trunc .. to</tt>' Instruction</a>
5194 <result> = trunc <ty> <value> to <ty2> <i>; yields ty2</i>
5198 <p>The '<tt>trunc</tt>' instruction truncates its operand to the
5199 type <tt>ty2</tt>.</p>
5202 <p>The '<tt>trunc</tt>' instruction takes a value to trunc, and a type to trunc it to.
5203 Both types must be of <a href="#t_integer">integer</a> types, or vectors
5204 of the same number of integers.
5205 The bit size of the <tt>value</tt> must be larger than
5206 the bit size of the destination type, <tt>ty2</tt>.
5207 Equal sized types are not allowed.</p>
5210 <p>The '<tt>trunc</tt>' instruction truncates the high order bits
5211 in <tt>value</tt> and converts the remaining bits to <tt>ty2</tt>. Since the
5212 source size must be larger than the destination size, <tt>trunc</tt> cannot
5213 be a <i>no-op cast</i>. It will always truncate bits.</p>
5217 %X = trunc i32 257 to i8 <i>; yields i8:1</i>
5218 %Y = trunc i32 123 to i1 <i>; yields i1:true</i>
5219 %Z = trunc i32 122 to i1 <i>; yields i1:false</i>
5220 %W = trunc <2 x i16> <i16 8, i16 7> to <2 x i8> <i>; yields <i8 8, i8 7></i>
5225 <!-- _______________________________________________________________________ -->
5227 <a name="i_zext">'<tt>zext .. to</tt>' Instruction</a>
5234 <result> = zext <ty> <value> to <ty2> <i>; yields ty2</i>
5238 <p>The '<tt>zext</tt>' instruction zero extends its operand to type
5243 <p>The '<tt>zext</tt>' instruction takes a value to cast, and a type to cast it to.
5244 Both types must be of <a href="#t_integer">integer</a> types, or vectors
5245 of the same number of integers.
5246 The bit size of the <tt>value</tt> must be smaller than
5247 the bit size of the destination type,
5251 <p>The <tt>zext</tt> fills the high order bits of the <tt>value</tt> with zero
5252 bits until it reaches the size of the destination type, <tt>ty2</tt>.</p>
5254 <p>When zero extending from i1, the result will always be either 0 or 1.</p>
5258 %X = zext i32 257 to i64 <i>; yields i64:257</i>
5259 %Y = zext i1 true to i32 <i>; yields i32:1</i>
5260 %Z = zext <2 x i16> <i16 8, i16 7> to <2 x i32> <i>; yields <i32 8, i32 7></i>
5265 <!-- _______________________________________________________________________ -->
5267 <a name="i_sext">'<tt>sext .. to</tt>' Instruction</a>
5274 <result> = sext <ty> <value> to <ty2> <i>; yields ty2</i>
5278 <p>The '<tt>sext</tt>' sign extends <tt>value</tt> to the type <tt>ty2</tt>.</p>
5281 <p>The '<tt>sext</tt>' instruction takes a value to cast, and a type to cast it to.
5282 Both types must be of <a href="#t_integer">integer</a> types, or vectors
5283 of the same number of integers.
5284 The bit size of the <tt>value</tt> must be smaller than
5285 the bit size of the destination type,
5289 <p>The '<tt>sext</tt>' instruction performs a sign extension by copying the sign
5290 bit (highest order bit) of the <tt>value</tt> until it reaches the bit size
5291 of the type <tt>ty2</tt>.</p>
5293 <p>When sign extending from i1, the extension always results in -1 or 0.</p>
5297 %X = sext i8 -1 to i16 <i>; yields i16 :65535</i>
5298 %Y = sext i1 true to i32 <i>; yields i32:-1</i>
5299 %Z = sext <2 x i16> <i16 8, i16 7> to <2 x i32> <i>; yields <i32 8, i32 7></i>
5304 <!-- _______________________________________________________________________ -->
5306 <a name="i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a>
5313 <result> = fptrunc <ty> <value> to <ty2> <i>; yields ty2</i>
5317 <p>The '<tt>fptrunc</tt>' instruction truncates <tt>value</tt> to type
5321 <p>The '<tt>fptrunc</tt>' instruction takes a <a href="#t_floating">floating
5322 point</a> value to cast and a <a href="#t_floating">floating point</a> type
5323 to cast it to. The size of <tt>value</tt> must be larger than the size of
5324 <tt>ty2</tt>. This implies that <tt>fptrunc</tt> cannot be used to make a
5325 <i>no-op cast</i>.</p>
5328 <p>The '<tt>fptrunc</tt>' instruction truncates a <tt>value</tt> from a larger
5329 <a href="#t_floating">floating point</a> type to a smaller
5330 <a href="#t_floating">floating point</a> type. If the value cannot fit
5331 within the destination type, <tt>ty2</tt>, then the results are
5336 %X = fptrunc double 123.0 to float <i>; yields float:123.0</i>
5337 %Y = fptrunc double 1.0E+300 to float <i>; yields undefined</i>
5342 <!-- _______________________________________________________________________ -->
5344 <a name="i_fpext">'<tt>fpext .. to</tt>' Instruction</a>
5351 <result> = fpext <ty> <value> to <ty2> <i>; yields ty2</i>
5355 <p>The '<tt>fpext</tt>' extends a floating point <tt>value</tt> to a larger
5356 floating point value.</p>
5359 <p>The '<tt>fpext</tt>' instruction takes a
5360 <a href="#t_floating">floating point</a> <tt>value</tt> to cast, and
5361 a <a href="#t_floating">floating point</a> type to cast it to. The source
5362 type must be smaller than the destination type.</p>
5365 <p>The '<tt>fpext</tt>' instruction extends the <tt>value</tt> from a smaller
5366 <a href="#t_floating">floating point</a> type to a larger
5367 <a href="#t_floating">floating point</a> type. The <tt>fpext</tt> cannot be
5368 used to make a <i>no-op cast</i> because it always changes bits. Use
5369 <tt>bitcast</tt> to make a <i>no-op cast</i> for a floating point cast.</p>
5373 %X = fpext float 3.125 to double <i>; yields double:3.125000e+00</i>
5374 %Y = fpext double %X to fp128 <i>; yields fp128:0xL00000000000000004000900000000000</i>
5379 <!-- _______________________________________________________________________ -->
5381 <a name="i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a>
5388 <result> = fptoui <ty> <value> to <ty2> <i>; yields ty2</i>
5392 <p>The '<tt>fptoui</tt>' converts a floating point <tt>value</tt> to its
5393 unsigned integer equivalent of type <tt>ty2</tt>.</p>
5396 <p>The '<tt>fptoui</tt>' instruction takes a value to cast, which must be a
5397 scalar or vector <a href="#t_floating">floating point</a> value, and a type
5398 to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
5399 type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
5400 vector integer type with the same number of elements as <tt>ty</tt></p>
5403 <p>The '<tt>fptoui</tt>' instruction converts its
5404 <a href="#t_floating">floating point</a> operand into the nearest (rounding
5405 towards zero) unsigned integer value. If the value cannot fit
5406 in <tt>ty2</tt>, the results are undefined.</p>
5410 %X = fptoui double 123.0 to i32 <i>; yields i32:123</i>
5411 %Y = fptoui float 1.0E+300 to i1 <i>; yields undefined:1</i>
5412 %Z = fptoui float 1.04E+17 to i8 <i>; yields undefined:1</i>
5417 <!-- _______________________________________________________________________ -->
5419 <a name="i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a>
5426 <result> = fptosi <ty> <value> to <ty2> <i>; yields ty2</i>
5430 <p>The '<tt>fptosi</tt>' instruction converts
5431 <a href="#t_floating">floating point</a> <tt>value</tt> to
5432 type <tt>ty2</tt>.</p>
5435 <p>The '<tt>fptosi</tt>' instruction takes a value to cast, which must be a
5436 scalar or vector <a href="#t_floating">floating point</a> value, and a type
5437 to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
5438 type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
5439 vector integer type with the same number of elements as <tt>ty</tt></p>
5442 <p>The '<tt>fptosi</tt>' instruction converts its
5443 <a href="#t_floating">floating point</a> operand into the nearest (rounding
5444 towards zero) signed integer value. If the value cannot fit in <tt>ty2</tt>,
5445 the results are undefined.</p>
5449 %X = fptosi double -123.0 to i32 <i>; yields i32:-123</i>
5450 %Y = fptosi float 1.0E-247 to i1 <i>; yields undefined:1</i>
5451 %Z = fptosi float 1.04E+17 to i8 <i>; yields undefined:1</i>
5456 <!-- _______________________________________________________________________ -->
5458 <a name="i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a>
5465 <result> = uitofp <ty> <value> to <ty2> <i>; yields ty2</i>
5469 <p>The '<tt>uitofp</tt>' instruction regards <tt>value</tt> as an unsigned
5470 integer and converts that value to the <tt>ty2</tt> type.</p>
5473 <p>The '<tt>uitofp</tt>' instruction takes a value to cast, which must be a
5474 scalar or vector <a href="#t_integer">integer</a> value, and a type to cast
5475 it to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
5476 type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
5477 floating point type with the same number of elements as <tt>ty</tt></p>
5480 <p>The '<tt>uitofp</tt>' instruction interprets its operand as an unsigned
5481 integer quantity and converts it to the corresponding floating point
5482 value. If the value cannot fit in the floating point value, the results are
5487 %X = uitofp i32 257 to float <i>; yields float:257.0</i>
5488 %Y = uitofp i8 -1 to double <i>; yields double:255.0</i>
5493 <!-- _______________________________________________________________________ -->
5495 <a name="i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a>
5502 <result> = sitofp <ty> <value> to <ty2> <i>; yields ty2</i>
5506 <p>The '<tt>sitofp</tt>' instruction regards <tt>value</tt> as a signed integer
5507 and converts that value to the <tt>ty2</tt> type.</p>
5510 <p>The '<tt>sitofp</tt>' instruction takes a value to cast, which must be a
5511 scalar or vector <a href="#t_integer">integer</a> value, and a type to cast
5512 it to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
5513 type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
5514 floating point type with the same number of elements as <tt>ty</tt></p>
5517 <p>The '<tt>sitofp</tt>' instruction interprets its operand as a signed integer
5518 quantity and converts it to the corresponding floating point value. If the
5519 value cannot fit in the floating point value, the results are undefined.</p>
5523 %X = sitofp i32 257 to float <i>; yields float:257.0</i>
5524 %Y = sitofp i8 -1 to double <i>; yields double:-1.0</i>
5529 <!-- _______________________________________________________________________ -->
5531 <a name="i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a>
5538 <result> = ptrtoint <ty> <value> to <ty2> <i>; yields ty2</i>
5542 <p>The '<tt>ptrtoint</tt>' instruction converts the pointer <tt>value</tt> to
5543 the integer type <tt>ty2</tt>.</p>
5546 <p>The '<tt>ptrtoint</tt>' instruction takes a <tt>value</tt> to cast, which
5547 must be a <a href="#t_pointer">pointer</a> value, and a type to cast it to
5548 <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a> type.</p>
5551 <p>The '<tt>ptrtoint</tt>' instruction converts <tt>value</tt> to integer type
5552 <tt>ty2</tt> by interpreting the pointer value as an integer and either
5553 truncating or zero extending that value to the size of the integer type. If
5554 <tt>value</tt> is smaller than <tt>ty2</tt> then a zero extension is done. If
5555 <tt>value</tt> is larger than <tt>ty2</tt> then a truncation is done. If they
5556 are the same size, then nothing is done (<i>no-op cast</i>) other than a type
5561 %X = ptrtoint i32* %X to i8 <i>; yields truncation on 32-bit architecture</i>
5562 %Y = ptrtoint i32* %x to i64 <i>; yields zero extension on 32-bit architecture</i>
5567 <!-- _______________________________________________________________________ -->
5569 <a name="i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a>
5576 <result> = inttoptr <ty> <value> to <ty2> <i>; yields ty2</i>
5580 <p>The '<tt>inttoptr</tt>' instruction converts an integer <tt>value</tt> to a
5581 pointer type, <tt>ty2</tt>.</p>
5584 <p>The '<tt>inttoptr</tt>' instruction takes an <a href="#t_integer">integer</a>
5585 value to cast, and a type to cast it to, which must be a
5586 <a href="#t_pointer">pointer</a> type.</p>
5589 <p>The '<tt>inttoptr</tt>' instruction converts <tt>value</tt> to type
5590 <tt>ty2</tt> by applying either a zero extension or a truncation depending on
5591 the size of the integer <tt>value</tt>. If <tt>value</tt> is larger than the
5592 size of a pointer then a truncation is done. If <tt>value</tt> is smaller
5593 than the size of a pointer then a zero extension is done. If they are the
5594 same size, nothing is done (<i>no-op cast</i>).</p>
5598 %X = inttoptr i32 255 to i32* <i>; yields zero extension on 64-bit architecture</i>
5599 %Y = inttoptr i32 255 to i32* <i>; yields no-op on 32-bit architecture</i>
5600 %Z = inttoptr i64 0 to i32* <i>; yields truncation on 32-bit architecture</i>
5605 <!-- _______________________________________________________________________ -->
5607 <a name="i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a>
5614 <result> = bitcast <ty> <value> to <ty2> <i>; yields ty2</i>
5618 <p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
5619 <tt>ty2</tt> without changing any bits.</p>
5622 <p>The '<tt>bitcast</tt>' instruction takes a value to cast, which must be a
5623 non-aggregate first class value, and a type to cast it to, which must also be
5624 a non-aggregate <a href="#t_firstclass">first class</a> type. The bit sizes
5625 of <tt>value</tt> and the destination type, <tt>ty2</tt>, must be
5626 identical. If the source type is a pointer, the destination type must also be
5627 a pointer. This instruction supports bitwise conversion of vectors to
5628 integers and to vectors of other types (as long as they have the same
5632 <p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
5633 <tt>ty2</tt>. It is always a <i>no-op cast</i> because no bits change with
5634 this conversion. The conversion is done as if the <tt>value</tt> had been
5635 stored to memory and read back as type <tt>ty2</tt>. Pointer types may only
5636 be converted to other pointer types with this instruction. To convert
5637 pointers to other types, use the <a href="#i_inttoptr">inttoptr</a> or
5638 <a href="#i_ptrtoint">ptrtoint</a> instructions first.</p>
5642 %X = bitcast i8 255 to i8 <i>; yields i8 :-1</i>
5643 %Y = bitcast i32* %x to sint* <i>; yields sint*:%x</i>
5644 %Z = bitcast <2 x int> %V to i64; <i>; yields i64: %V</i>
5651 <!-- ======================================================================= -->
5653 <a name="otherops">Other Operations</a>
5658 <p>The instructions in this category are the "miscellaneous" instructions, which
5659 defy better classification.</p>
5661 <!-- _______________________________________________________________________ -->
5663 <a name="i_icmp">'<tt>icmp</tt>' Instruction</a>
5670 <result> = icmp <cond> <ty> <op1>, <op2> <i>; yields {i1} or {<N x i1>}:result</i>
5674 <p>The '<tt>icmp</tt>' instruction returns a boolean value or a vector of
5675 boolean values based on comparison of its two integer, integer vector, or
5676 pointer operands.</p>
5679 <p>The '<tt>icmp</tt>' instruction takes three operands. The first operand is
5680 the condition code indicating the kind of comparison to perform. It is not a
5681 value, just a keyword. The possible condition code are:</p>
5684 <li><tt>eq</tt>: equal</li>
5685 <li><tt>ne</tt>: not equal </li>
5686 <li><tt>ugt</tt>: unsigned greater than</li>
5687 <li><tt>uge</tt>: unsigned greater or equal</li>
5688 <li><tt>ult</tt>: unsigned less than</li>
5689 <li><tt>ule</tt>: unsigned less or equal</li>
5690 <li><tt>sgt</tt>: signed greater than</li>
5691 <li><tt>sge</tt>: signed greater or equal</li>
5692 <li><tt>slt</tt>: signed less than</li>
5693 <li><tt>sle</tt>: signed less or equal</li>
5696 <p>The remaining two arguments must be <a href="#t_integer">integer</a> or
5697 <a href="#t_pointer">pointer</a> or integer <a href="#t_vector">vector</a>
5698 typed. They must also be identical types.</p>
5701 <p>The '<tt>icmp</tt>' compares <tt>op1</tt> and <tt>op2</tt> according to the
5702 condition code given as <tt>cond</tt>. The comparison performed always yields
5703 either an <a href="#t_integer"><tt>i1</tt></a> or vector of <tt>i1</tt>
5704 result, as follows:</p>
5707 <li><tt>eq</tt>: yields <tt>true</tt> if the operands are equal,
5708 <tt>false</tt> otherwise. No sign interpretation is necessary or
5711 <li><tt>ne</tt>: yields <tt>true</tt> if the operands are unequal,
5712 <tt>false</tt> otherwise. No sign interpretation is necessary or
5715 <li><tt>ugt</tt>: interprets the operands as unsigned values and yields
5716 <tt>true</tt> if <tt>op1</tt> is greater than <tt>op2</tt>.</li>
5718 <li><tt>uge</tt>: interprets the operands as unsigned values and yields
5719 <tt>true</tt> if <tt>op1</tt> is greater than or equal
5720 to <tt>op2</tt>.</li>
5722 <li><tt>ult</tt>: interprets the operands as unsigned values and yields
5723 <tt>true</tt> if <tt>op1</tt> is less than <tt>op2</tt>.</li>
5725 <li><tt>ule</tt>: interprets the operands as unsigned values and yields
5726 <tt>true</tt> if <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
5728 <li><tt>sgt</tt>: interprets the operands as signed values and yields
5729 <tt>true</tt> if <tt>op1</tt> is greater than <tt>op2</tt>.</li>
5731 <li><tt>sge</tt>: interprets the operands as signed values and yields
5732 <tt>true</tt> if <tt>op1</tt> is greater than or equal
5733 to <tt>op2</tt>.</li>
5735 <li><tt>slt</tt>: interprets the operands as signed values and yields
5736 <tt>true</tt> if <tt>op1</tt> is less than <tt>op2</tt>.</li>
5738 <li><tt>sle</tt>: interprets the operands as signed values and yields
5739 <tt>true</tt> if <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
5742 <p>If the operands are <a href="#t_pointer">pointer</a> typed, the pointer
5743 values are compared as if they were integers.</p>
5745 <p>If the operands are integer vectors, then they are compared element by
5746 element. The result is an <tt>i1</tt> vector with the same number of elements
5747 as the values being compared. Otherwise, the result is an <tt>i1</tt>.</p>
5751 <result> = icmp eq i32 4, 5 <i>; yields: result=false</i>
5752 <result> = icmp ne float* %X, %X <i>; yields: result=false</i>
5753 <result> = icmp ult i16 4, 5 <i>; yields: result=true</i>
5754 <result> = icmp sgt i16 4, 5 <i>; yields: result=false</i>
5755 <result> = icmp ule i16 -4, 5 <i>; yields: result=false</i>
5756 <result> = icmp sge i16 4, 5 <i>; yields: result=false</i>
5759 <p>Note that the code generator does not yet support vector types with
5760 the <tt>icmp</tt> instruction.</p>
5764 <!-- _______________________________________________________________________ -->
5766 <a name="i_fcmp">'<tt>fcmp</tt>' Instruction</a>
5773 <result> = fcmp <cond> <ty> <op1>, <op2> <i>; yields {i1} or {<N x i1>}:result</i>
5777 <p>The '<tt>fcmp</tt>' instruction returns a boolean value or vector of boolean
5778 values based on comparison of its operands.</p>
5780 <p>If the operands are floating point scalars, then the result type is a boolean
5781 (<a href="#t_integer"><tt>i1</tt></a>).</p>
5783 <p>If the operands are floating point vectors, then the result type is a vector
5784 of boolean with the same number of elements as the operands being
5788 <p>The '<tt>fcmp</tt>' instruction takes three operands. The first operand is
5789 the condition code indicating the kind of comparison to perform. It is not a
5790 value, just a keyword. The possible condition code are:</p>
5793 <li><tt>false</tt>: no comparison, always returns false</li>
5794 <li><tt>oeq</tt>: ordered and equal</li>
5795 <li><tt>ogt</tt>: ordered and greater than </li>
5796 <li><tt>oge</tt>: ordered and greater than or equal</li>
5797 <li><tt>olt</tt>: ordered and less than </li>
5798 <li><tt>ole</tt>: ordered and less than or equal</li>
5799 <li><tt>one</tt>: ordered and not equal</li>
5800 <li><tt>ord</tt>: ordered (no nans)</li>
5801 <li><tt>ueq</tt>: unordered or equal</li>
5802 <li><tt>ugt</tt>: unordered or greater than </li>
5803 <li><tt>uge</tt>: unordered or greater than or equal</li>
5804 <li><tt>ult</tt>: unordered or less than </li>
5805 <li><tt>ule</tt>: unordered or less than or equal</li>
5806 <li><tt>une</tt>: unordered or not equal</li>
5807 <li><tt>uno</tt>: unordered (either nans)</li>
5808 <li><tt>true</tt>: no comparison, always returns true</li>
5811 <p><i>Ordered</i> means that neither operand is a QNAN while
5812 <i>unordered</i> means that either operand may be a QNAN.</p>
5814 <p>Each of <tt>val1</tt> and <tt>val2</tt> arguments must be either
5815 a <a href="#t_floating">floating point</a> type or
5816 a <a href="#t_vector">vector</a> of floating point type. They must have
5817 identical types.</p>
5820 <p>The '<tt>fcmp</tt>' instruction compares <tt>op1</tt> and <tt>op2</tt>
5821 according to the condition code given as <tt>cond</tt>. If the operands are
5822 vectors, then the vectors are compared element by element. Each comparison
5823 performed always yields an <a href="#t_integer">i1</a> result, as
5827 <li><tt>false</tt>: always yields <tt>false</tt>, regardless of operands.</li>
5829 <li><tt>oeq</tt>: yields <tt>true</tt> if both operands are not a QNAN and
5830 <tt>op1</tt> is equal to <tt>op2</tt>.</li>
5832 <li><tt>ogt</tt>: yields <tt>true</tt> if both operands are not a QNAN and
5833 <tt>op1</tt> is greater than <tt>op2</tt>.</li>
5835 <li><tt>oge</tt>: yields <tt>true</tt> if both operands are not a QNAN and
5836 <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
5838 <li><tt>olt</tt>: yields <tt>true</tt> if both operands are not a QNAN and
5839 <tt>op1</tt> is less than <tt>op2</tt>.</li>
5841 <li><tt>ole</tt>: yields <tt>true</tt> if both operands are not a QNAN and
5842 <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
5844 <li><tt>one</tt>: yields <tt>true</tt> if both operands are not a QNAN and
5845 <tt>op1</tt> is not equal to <tt>op2</tt>.</li>
5847 <li><tt>ord</tt>: yields <tt>true</tt> if both operands are not a QNAN.</li>
5849 <li><tt>ueq</tt>: yields <tt>true</tt> if either operand is a QNAN or
5850 <tt>op1</tt> is equal to <tt>op2</tt>.</li>
5852 <li><tt>ugt</tt>: yields <tt>true</tt> if either operand is a QNAN or
5853 <tt>op1</tt> is greater than <tt>op2</tt>.</li>
5855 <li><tt>uge</tt>: yields <tt>true</tt> if either operand is a QNAN or
5856 <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
5858 <li><tt>ult</tt>: yields <tt>true</tt> if either operand is a QNAN or
5859 <tt>op1</tt> is less than <tt>op2</tt>.</li>
5861 <li><tt>ule</tt>: yields <tt>true</tt> if either operand is a QNAN or
5862 <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
5864 <li><tt>une</tt>: yields <tt>true</tt> if either operand is a QNAN or
5865 <tt>op1</tt> is not equal to <tt>op2</tt>.</li>
5867 <li><tt>uno</tt>: yields <tt>true</tt> if either operand is a QNAN.</li>
5869 <li><tt>true</tt>: always yields <tt>true</tt>, regardless of operands.</li>
5874 <result> = fcmp oeq float 4.0, 5.0 <i>; yields: result=false</i>
5875 <result> = fcmp one float 4.0, 5.0 <i>; yields: result=true</i>
5876 <result> = fcmp olt float 4.0, 5.0 <i>; yields: result=true</i>
5877 <result> = fcmp ueq double 1.0, 2.0 <i>; yields: result=false</i>
5880 <p>Note that the code generator does not yet support vector types with
5881 the <tt>fcmp</tt> instruction.</p>
5885 <!-- _______________________________________________________________________ -->
5887 <a name="i_phi">'<tt>phi</tt>' Instruction</a>
5894 <result> = phi <ty> [ <val0>, <label0>], ...
5898 <p>The '<tt>phi</tt>' instruction is used to implement the φ node in the
5899 SSA graph representing the function.</p>
5902 <p>The type of the incoming values is specified with the first type field. After
5903 this, the '<tt>phi</tt>' instruction takes a list of pairs as arguments, with
5904 one pair for each predecessor basic block of the current block. Only values
5905 of <a href="#t_firstclass">first class</a> type may be used as the value
5906 arguments to the PHI node. Only labels may be used as the label
5909 <p>There must be no non-phi instructions between the start of a basic block and
5910 the PHI instructions: i.e. PHI instructions must be first in a basic
5913 <p>For the purposes of the SSA form, the use of each incoming value is deemed to
5914 occur on the edge from the corresponding predecessor block to the current
5915 block (but after any definition of an '<tt>invoke</tt>' instruction's return
5916 value on the same edge).</p>
5919 <p>At runtime, the '<tt>phi</tt>' instruction logically takes on the value
5920 specified by the pair corresponding to the predecessor basic block that
5921 executed just prior to the current block.</p>
5925 Loop: ; Infinite loop that counts from 0 on up...
5926 %indvar = phi i32 [ 0, %LoopHeader ], [ %nextindvar, %Loop ]
5927 %nextindvar = add i32 %indvar, 1
5933 <!-- _______________________________________________________________________ -->
5935 <a name="i_select">'<tt>select</tt>' Instruction</a>
5942 <result> = select <i>selty</i> <cond>, <ty> <val1>, <ty> <val2> <i>; yields ty</i>
5944 <i>selty</i> is either i1 or {<N x i1>}
5948 <p>The '<tt>select</tt>' instruction is used to choose one value based on a
5949 condition, without branching.</p>
5953 <p>The '<tt>select</tt>' instruction requires an 'i1' value or a vector of 'i1'
5954 values indicating the condition, and two values of the
5955 same <a href="#t_firstclass">first class</a> type. If the val1/val2 are
5956 vectors and the condition is a scalar, then entire vectors are selected, not
5957 individual elements.</p>
5960 <p>If the condition is an i1 and it evaluates to 1, the instruction returns the
5961 first value argument; otherwise, it returns the second value argument.</p>
5963 <p>If the condition is a vector of i1, then the value arguments must be vectors
5964 of the same size, and the selection is done element by element.</p>
5968 %X = select i1 true, i8 17, i8 42 <i>; yields i8:17</i>
5971 <p>Note that the code generator does not yet support conditions
5972 with vector type.</p>
5976 <!-- _______________________________________________________________________ -->
5978 <a name="i_call">'<tt>call</tt>' Instruction</a>
5985 <result> = [tail] call [<a href="#callingconv">cconv</a>] [<a href="#paramattrs">ret attrs</a>] <ty> [<fnty>*] <fnptrval>(<function args>) [<a href="#fnattrs">fn attrs</a>]
5989 <p>The '<tt>call</tt>' instruction represents a simple function call.</p>
5992 <p>This instruction requires several arguments:</p>
5995 <li>The optional "tail" marker indicates that the callee function does not
5996 access any allocas or varargs in the caller. Note that calls may be
5997 marked "tail" even if they do not occur before
5998 a <a href="#i_ret"><tt>ret</tt></a> instruction. If the "tail" marker is
5999 present, the function call is eligible for tail call optimization,
6000 but <a href="CodeGenerator.html#tailcallopt">might not in fact be
6001 optimized into a jump</a>. The code generator may optimize calls marked
6002 "tail" with either 1) automatic <a href="CodeGenerator.html#sibcallopt">
6003 sibling call optimization</a> when the caller and callee have
6004 matching signatures, or 2) forced tail call optimization when the
6005 following extra requirements are met:
6007 <li>Caller and callee both have the calling
6008 convention <tt>fastcc</tt>.</li>
6009 <li>The call is in tail position (ret immediately follows call and ret
6010 uses value of call or is void).</li>
6011 <li>Option <tt>-tailcallopt</tt> is enabled,
6012 or <code>llvm::GuaranteedTailCallOpt</code> is <code>true</code>.</li>
6013 <li><a href="CodeGenerator.html#tailcallopt">Platform specific
6014 constraints are met.</a></li>
6018 <li>The optional "cconv" marker indicates which <a href="#callingconv">calling
6019 convention</a> the call should use. If none is specified, the call
6020 defaults to using C calling conventions. The calling convention of the
6021 call must match the calling convention of the target function, or else the
6022 behavior is undefined.</li>
6024 <li>The optional <a href="#paramattrs">Parameter Attributes</a> list for
6025 return values. Only '<tt>zeroext</tt>', '<tt>signext</tt>', and
6026 '<tt>inreg</tt>' attributes are valid here.</li>
6028 <li>'<tt>ty</tt>': the type of the call instruction itself which is also the
6029 type of the return value. Functions that return no value are marked
6030 <tt><a href="#t_void">void</a></tt>.</li>
6032 <li>'<tt>fnty</tt>': shall be the signature of the pointer to function value
6033 being invoked. The argument types must match the types implied by this
6034 signature. This type can be omitted if the function is not varargs and if
6035 the function type does not return a pointer to a function.</li>
6037 <li>'<tt>fnptrval</tt>': An LLVM value containing a pointer to a function to
6038 be invoked. In most cases, this is a direct function invocation, but
6039 indirect <tt>call</tt>s are just as possible, calling an arbitrary pointer
6040 to function value.</li>
6042 <li>'<tt>function args</tt>': argument list whose types match the function
6043 signature argument types and parameter attributes. All arguments must be
6044 of <a href="#t_firstclass">first class</a> type. If the function
6045 signature indicates the function accepts a variable number of arguments,
6046 the extra arguments can be specified.</li>
6048 <li>The optional <a href="#fnattrs">function attributes</a> list. Only
6049 '<tt>noreturn</tt>', '<tt>nounwind</tt>', '<tt>readonly</tt>' and
6050 '<tt>readnone</tt>' attributes are valid here.</li>
6054 <p>The '<tt>call</tt>' instruction is used to cause control flow to transfer to
6055 a specified function, with its incoming arguments bound to the specified
6056 values. Upon a '<tt><a href="#i_ret">ret</a></tt>' instruction in the called
6057 function, control flow continues with the instruction after the function
6058 call, and the return value of the function is bound to the result
6063 %retval = call i32 @test(i32 %argc)
6064 call i32 (i8*, ...)* @printf(i8* %msg, i32 12, i8 42) <i>; yields i32</i>
6065 %X = tail call i32 @foo() <i>; yields i32</i>
6066 %Y = tail call <a href="#callingconv">fastcc</a> i32 @foo() <i>; yields i32</i>
6067 call void %foo(i8 97 signext)
6069 %struct.A = type { i32, i8 }
6070 %r = call %struct.A @foo() <i>; yields { 32, i8 }</i>
6071 %gr = extractvalue %struct.A %r, 0 <i>; yields i32</i>
6072 %gr1 = extractvalue %struct.A %r, 1 <i>; yields i8</i>
6073 %Z = call void @foo() noreturn <i>; indicates that %foo never returns normally</i>
6074 %ZZ = call zeroext i32 @bar() <i>; Return value is %zero extended</i>
6077 <p>llvm treats calls to some functions with names and arguments that match the
6078 standard C99 library as being the C99 library functions, and may perform
6079 optimizations or generate code for them under that assumption. This is
6080 something we'd like to change in the future to provide better support for
6081 freestanding environments and non-C-based languages.</p>
6085 <!-- _______________________________________________________________________ -->
6087 <a name="i_va_arg">'<tt>va_arg</tt>' Instruction</a>
6094 <resultval> = va_arg <va_list*> <arglist>, <argty>
6098 <p>The '<tt>va_arg</tt>' instruction is used to access arguments passed through
6099 the "variable argument" area of a function call. It is used to implement the
6100 <tt>va_arg</tt> macro in C.</p>
6103 <p>This instruction takes a <tt>va_list*</tt> value and the type of the
6104 argument. It returns a value of the specified argument type and increments
6105 the <tt>va_list</tt> to point to the next argument. The actual type
6106 of <tt>va_list</tt> is target specific.</p>
6109 <p>The '<tt>va_arg</tt>' instruction loads an argument of the specified type
6110 from the specified <tt>va_list</tt> and causes the <tt>va_list</tt> to point
6111 to the next argument. For more information, see the variable argument
6112 handling <a href="#int_varargs">Intrinsic Functions</a>.</p>
6114 <p>It is legal for this instruction to be called in a function which does not
6115 take a variable number of arguments, for example, the <tt>vfprintf</tt>
6118 <p><tt>va_arg</tt> is an LLVM instruction instead of
6119 an <a href="#intrinsics">intrinsic function</a> because it takes a type as an
6123 <p>See the <a href="#int_varargs">variable argument processing</a> section.</p>
6125 <p>Note that the code generator does not yet fully support va_arg on many
6126 targets. Also, it does not currently support va_arg with aggregate types on
6131 <!-- _______________________________________________________________________ -->
6133 <a name="i_landingpad">'<tt>landingpad</tt>' Instruction</a>
6140 <resultval> = landingpad <somety> personality <type> <pers_fn> <clause>+
6141 <resultval> = landingpad <somety> personality <type> <pers_fn> cleanup <clause>*
6143 <clause> := catch <type> <value>
6144 <clause> := filter <array constant type> <array constant>
6148 <p>The '<tt>landingpad</tt>' instruction is used by
6149 <a href="ExceptionHandling.html#overview">LLVM's exception handling
6150 system</a> to specify that a basic block is a landing pad — one where
6151 the exception lands, and corresponds to the code found in the
6152 <i><tt>catch</tt></i> portion of a <i><tt>try/catch</tt></i> sequence. It
6153 defines values supplied by the personality function (<tt>pers_fn</tt>) upon
6154 re-entry to the function. The <tt>resultval</tt> has the
6155 type <tt>somety</tt>.</p>
6158 <p>This instruction takes a <tt>pers_fn</tt> value. This is the personality
6159 function associated with the unwinding mechanism. The optional
6160 <tt>cleanup</tt> flag indicates that the landing pad block is a cleanup.</p>
6162 <p>A <tt>clause</tt> begins with the clause type — <tt>catch</tt>
6163 or <tt>filter</tt> — and contains the global variable representing the
6164 "type" that may be caught or filtered respectively. Unlike the
6165 <tt>catch</tt> clause, the <tt>filter</tt> clause takes an array constant as
6166 its argument. Use "<tt>[0 x i8**] undef</tt>" for a filter which cannot
6167 throw. The '<tt>landingpad</tt>' instruction must contain <em>at least</em>
6168 one <tt>clause</tt> or the <tt>cleanup</tt> flag.</p>
6171 <p>The '<tt>landingpad</tt>' instruction defines the values which are set by the
6172 personality function (<tt>pers_fn</tt>) upon re-entry to the function, and
6173 therefore the "result type" of the <tt>landingpad</tt> instruction. As with
6174 calling conventions, how the personality function results are represented in
6175 LLVM IR is target specific.</p>
6177 <p>The clauses are applied in order from top to bottom. If two
6178 <tt>landingpad</tt> instructions are merged together through inlining, the
6179 clauses from the calling function are appended to the list of clauses.</p>
6181 <p>The <tt>landingpad</tt> instruction has several restrictions:</p>
6184 <li>A landing pad block is a basic block which is the unwind destination of an
6185 '<tt>invoke</tt>' instruction.</li>
6186 <li>A landing pad block must have a '<tt>landingpad</tt>' instruction as its
6187 first non-PHI instruction.</li>
6188 <li>There can be only one '<tt>landingpad</tt>' instruction within the landing
6190 <li>A basic block that is not a landing pad block may not include a
6191 '<tt>landingpad</tt>' instruction.</li>
6192 <li>All '<tt>landingpad</tt>' instructions in a function must have the same
6193 personality function.</li>
6198 ;; A landing pad which can catch an integer.
6199 %res = landingpad { i8*, i32 } personality i32 (...)* @__gxx_personality_v0
6201 ;; A landing pad that is a cleanup.
6202 %res = landingpad { i8*, i32 } personality i32 (...)* @__gxx_personality_v0
6204 ;; A landing pad which can catch an integer and can only throw a double.
6205 %res = landingpad { i8*, i32 } personality i32 (...)* @__gxx_personality_v0
6207 filter [1 x i8**] [@_ZTId]
6216 <!-- *********************************************************************** -->
6217 <h2><a name="intrinsics">Intrinsic Functions</a></h2>
6218 <!-- *********************************************************************** -->
6222 <p>LLVM supports the notion of an "intrinsic function". These functions have
6223 well known names and semantics and are required to follow certain
6224 restrictions. Overall, these intrinsics represent an extension mechanism for
6225 the LLVM language that does not require changing all of the transformations
6226 in LLVM when adding to the language (or the bitcode reader/writer, the
6227 parser, etc...).</p>
6229 <p>Intrinsic function names must all start with an "<tt>llvm.</tt>" prefix. This
6230 prefix is reserved in LLVM for intrinsic names; thus, function names may not
6231 begin with this prefix. Intrinsic functions must always be external
6232 functions: you cannot define the body of intrinsic functions. Intrinsic
6233 functions may only be used in call or invoke instructions: it is illegal to
6234 take the address of an intrinsic function. Additionally, because intrinsic
6235 functions are part of the LLVM language, it is required if any are added that
6236 they be documented here.</p>
6238 <p>Some intrinsic functions can be overloaded, i.e., the intrinsic represents a
6239 family of functions that perform the same operation but on different data
6240 types. Because LLVM can represent over 8 million different integer types,
6241 overloading is used commonly to allow an intrinsic function to operate on any
6242 integer type. One or more of the argument types or the result type can be
6243 overloaded to accept any integer type. Argument types may also be defined as
6244 exactly matching a previous argument's type or the result type. This allows
6245 an intrinsic function which accepts multiple arguments, but needs all of them
6246 to be of the same type, to only be overloaded with respect to a single
6247 argument or the result.</p>
6249 <p>Overloaded intrinsics will have the names of its overloaded argument types
6250 encoded into its function name, each preceded by a period. Only those types
6251 which are overloaded result in a name suffix. Arguments whose type is matched
6252 against another type do not. For example, the <tt>llvm.ctpop</tt> function
6253 can take an integer of any width and returns an integer of exactly the same
6254 integer width. This leads to a family of functions such as
6255 <tt>i8 @llvm.ctpop.i8(i8 %val)</tt> and <tt>i29 @llvm.ctpop.i29(i29
6256 %val)</tt>. Only one type, the return type, is overloaded, and only one type
6257 suffix is required. Because the argument's type is matched against the return
6258 type, it does not require its own name suffix.</p>
6260 <p>To learn how to add an intrinsic function, please see the
6261 <a href="ExtendingLLVM.html">Extending LLVM Guide</a>.</p>
6263 <!-- ======================================================================= -->
6265 <a name="int_varargs">Variable Argument Handling Intrinsics</a>
6270 <p>Variable argument support is defined in LLVM with
6271 the <a href="#i_va_arg"><tt>va_arg</tt></a> instruction and these three
6272 intrinsic functions. These functions are related to the similarly named
6273 macros defined in the <tt><stdarg.h></tt> header file.</p>
6275 <p>All of these functions operate on arguments that use a target-specific value
6276 type "<tt>va_list</tt>". The LLVM assembly language reference manual does
6277 not define what this type is, so all transformations should be prepared to
6278 handle these functions regardless of the type used.</p>
6280 <p>This example shows how the <a href="#i_va_arg"><tt>va_arg</tt></a>
6281 instruction and the variable argument handling intrinsic functions are
6284 <pre class="doc_code">
6285 define i32 @test(i32 %X, ...) {
6286 ; Initialize variable argument processing
6288 %ap2 = bitcast i8** %ap to i8*
6289 call void @llvm.va_start(i8* %ap2)
6291 ; Read a single integer argument
6292 %tmp = va_arg i8** %ap, i32
6294 ; Demonstrate usage of llvm.va_copy and llvm.va_end
6296 %aq2 = bitcast i8** %aq to i8*
6297 call void @llvm.va_copy(i8* %aq2, i8* %ap2)
6298 call void @llvm.va_end(i8* %aq2)
6300 ; Stop processing of arguments.
6301 call void @llvm.va_end(i8* %ap2)
6305 declare void @llvm.va_start(i8*)
6306 declare void @llvm.va_copy(i8*, i8*)
6307 declare void @llvm.va_end(i8*)
6310 <!-- _______________________________________________________________________ -->
6312 <a name="int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a>
6320 declare void %llvm.va_start(i8* <arglist>)
6324 <p>The '<tt>llvm.va_start</tt>' intrinsic initializes <tt>*<arglist></tt>
6325 for subsequent use by <tt><a href="#i_va_arg">va_arg</a></tt>.</p>
6328 <p>The argument is a pointer to a <tt>va_list</tt> element to initialize.</p>
6331 <p>The '<tt>llvm.va_start</tt>' intrinsic works just like the <tt>va_start</tt>
6332 macro available in C. In a target-dependent way, it initializes
6333 the <tt>va_list</tt> element to which the argument points, so that the next
6334 call to <tt>va_arg</tt> will produce the first variable argument passed to
6335 the function. Unlike the C <tt>va_start</tt> macro, this intrinsic does not
6336 need to know the last argument of the function as the compiler can figure
6341 <!-- _______________________________________________________________________ -->
6343 <a name="int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a>
6350 declare void @llvm.va_end(i8* <arglist>)
6354 <p>The '<tt>llvm.va_end</tt>' intrinsic destroys <tt>*<arglist></tt>,
6355 which has been initialized previously
6356 with <tt><a href="#int_va_start">llvm.va_start</a></tt>
6357 or <tt><a href="#i_va_copy">llvm.va_copy</a></tt>.</p>
6360 <p>The argument is a pointer to a <tt>va_list</tt> to destroy.</p>
6363 <p>The '<tt>llvm.va_end</tt>' intrinsic works just like the <tt>va_end</tt>
6364 macro available in C. In a target-dependent way, it destroys
6365 the <tt>va_list</tt> element to which the argument points. Calls
6366 to <a href="#int_va_start"><tt>llvm.va_start</tt></a>
6367 and <a href="#int_va_copy"> <tt>llvm.va_copy</tt></a> must be matched exactly
6368 with calls to <tt>llvm.va_end</tt>.</p>
6372 <!-- _______________________________________________________________________ -->
6374 <a name="int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a>
6381 declare void @llvm.va_copy(i8* <destarglist>, i8* <srcarglist>)
6385 <p>The '<tt>llvm.va_copy</tt>' intrinsic copies the current argument position
6386 from the source argument list to the destination argument list.</p>
6389 <p>The first argument is a pointer to a <tt>va_list</tt> element to initialize.
6390 The second argument is a pointer to a <tt>va_list</tt> element to copy
6394 <p>The '<tt>llvm.va_copy</tt>' intrinsic works just like the <tt>va_copy</tt>
6395 macro available in C. In a target-dependent way, it copies the
6396 source <tt>va_list</tt> element into the destination <tt>va_list</tt>
6397 element. This intrinsic is necessary because
6398 the <tt><a href="#int_va_start"> llvm.va_start</a></tt> intrinsic may be
6399 arbitrarily complex and require, for example, memory allocation.</p>
6405 <!-- ======================================================================= -->
6407 <a name="int_gc">Accurate Garbage Collection Intrinsics</a>
6412 <p>LLVM support for <a href="GarbageCollection.html">Accurate Garbage
6413 Collection</a> (GC) requires the implementation and generation of these
6414 intrinsics. These intrinsics allow identification of <a href="#int_gcroot">GC
6415 roots on the stack</a>, as well as garbage collector implementations that
6416 require <a href="#int_gcread">read</a> and <a href="#int_gcwrite">write</a>
6417 barriers. Front-ends for type-safe garbage collected languages should generate
6418 these intrinsics to make use of the LLVM garbage collectors. For more details,
6419 see <a href="GarbageCollection.html">Accurate Garbage Collection with
6422 <p>The garbage collection intrinsics only operate on objects in the generic
6423 address space (address space zero).</p>
6425 <!-- _______________________________________________________________________ -->
6427 <a name="int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a>
6434 declare void @llvm.gcroot(i8** %ptrloc, i8* %metadata)
6438 <p>The '<tt>llvm.gcroot</tt>' intrinsic declares the existence of a GC root to
6439 the code generator, and allows some metadata to be associated with it.</p>
6442 <p>The first argument specifies the address of a stack object that contains the
6443 root pointer. The second pointer (which must be either a constant or a
6444 global value address) contains the meta-data to be associated with the
6448 <p>At runtime, a call to this intrinsic stores a null pointer into the "ptrloc"
6449 location. At compile-time, the code generator generates information to allow
6450 the runtime to find the pointer at GC safe points. The '<tt>llvm.gcroot</tt>'
6451 intrinsic may only be used in a function which <a href="#gc">specifies a GC
6456 <!-- _______________________________________________________________________ -->
6458 <a name="int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a>
6465 declare i8* @llvm.gcread(i8* %ObjPtr, i8** %Ptr)
6469 <p>The '<tt>llvm.gcread</tt>' intrinsic identifies reads of references from heap
6470 locations, allowing garbage collector implementations that require read
6474 <p>The second argument is the address to read from, which should be an address
6475 allocated from the garbage collector. The first object is a pointer to the
6476 start of the referenced object, if needed by the language runtime (otherwise
6480 <p>The '<tt>llvm.gcread</tt>' intrinsic has the same semantics as a load
6481 instruction, but may be replaced with substantially more complex code by the
6482 garbage collector runtime, as needed. The '<tt>llvm.gcread</tt>' intrinsic
6483 may only be used in a function which <a href="#gc">specifies a GC
6488 <!-- _______________________________________________________________________ -->
6490 <a name="int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a>
6497 declare void @llvm.gcwrite(i8* %P1, i8* %Obj, i8** %P2)
6501 <p>The '<tt>llvm.gcwrite</tt>' intrinsic identifies writes of references to heap
6502 locations, allowing garbage collector implementations that require write
6503 barriers (such as generational or reference counting collectors).</p>
6506 <p>The first argument is the reference to store, the second is the start of the
6507 object to store it to, and the third is the address of the field of Obj to
6508 store to. If the runtime does not require a pointer to the object, Obj may
6512 <p>The '<tt>llvm.gcwrite</tt>' intrinsic has the same semantics as a store
6513 instruction, but may be replaced with substantially more complex code by the
6514 garbage collector runtime, as needed. The '<tt>llvm.gcwrite</tt>' intrinsic
6515 may only be used in a function which <a href="#gc">specifies a GC
6522 <!-- ======================================================================= -->
6524 <a name="int_codegen">Code Generator Intrinsics</a>
6529 <p>These intrinsics are provided by LLVM to expose special features that may
6530 only be implemented with code generator support.</p>
6532 <!-- _______________________________________________________________________ -->
6534 <a name="int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a>
6541 declare i8 *@llvm.returnaddress(i32 <level>)
6545 <p>The '<tt>llvm.returnaddress</tt>' intrinsic attempts to compute a
6546 target-specific value indicating the return address of the current function
6547 or one of its callers.</p>
6550 <p>The argument to this intrinsic indicates which function to return the address
6551 for. Zero indicates the calling function, one indicates its caller, etc.
6552 The argument is <b>required</b> to be a constant integer value.</p>
6555 <p>The '<tt>llvm.returnaddress</tt>' intrinsic either returns a pointer
6556 indicating the return address of the specified call frame, or zero if it
6557 cannot be identified. The value returned by this intrinsic is likely to be
6558 incorrect or 0 for arguments other than zero, so it should only be used for
6559 debugging purposes.</p>
6561 <p>Note that calling this intrinsic does not prevent function inlining or other
6562 aggressive transformations, so the value returned may not be that of the
6563 obvious source-language caller.</p>
6567 <!-- _______________________________________________________________________ -->
6569 <a name="int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a>
6576 declare i8* @llvm.frameaddress(i32 <level>)
6580 <p>The '<tt>llvm.frameaddress</tt>' intrinsic attempts to return the
6581 target-specific frame pointer value for the specified stack frame.</p>
6584 <p>The argument to this intrinsic indicates which function to return the frame
6585 pointer for. Zero indicates the calling function, one indicates its caller,
6586 etc. The argument is <b>required</b> to be a constant integer value.</p>
6589 <p>The '<tt>llvm.frameaddress</tt>' intrinsic either returns a pointer
6590 indicating the frame address of the specified call frame, or zero if it
6591 cannot be identified. The value returned by this intrinsic is likely to be
6592 incorrect or 0 for arguments other than zero, so it should only be used for
6593 debugging purposes.</p>
6595 <p>Note that calling this intrinsic does not prevent function inlining or other
6596 aggressive transformations, so the value returned may not be that of the
6597 obvious source-language caller.</p>
6601 <!-- _______________________________________________________________________ -->
6603 <a name="int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a>
6610 declare i8* @llvm.stacksave()
6614 <p>The '<tt>llvm.stacksave</tt>' intrinsic is used to remember the current state
6615 of the function stack, for use
6616 with <a href="#int_stackrestore"> <tt>llvm.stackrestore</tt></a>. This is
6617 useful for implementing language features like scoped automatic variable
6618 sized arrays in C99.</p>
6621 <p>This intrinsic returns a opaque pointer value that can be passed
6622 to <a href="#int_stackrestore"><tt>llvm.stackrestore</tt></a>. When
6623 an <tt>llvm.stackrestore</tt> intrinsic is executed with a value saved
6624 from <tt>llvm.stacksave</tt>, it effectively restores the state of the stack
6625 to the state it was in when the <tt>llvm.stacksave</tt> intrinsic executed.
6626 In practice, this pops any <a href="#i_alloca">alloca</a> blocks from the
6627 stack that were allocated after the <tt>llvm.stacksave</tt> was executed.</p>
6631 <!-- _______________________________________________________________________ -->
6633 <a name="int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a>
6640 declare void @llvm.stackrestore(i8* %ptr)
6644 <p>The '<tt>llvm.stackrestore</tt>' intrinsic is used to restore the state of
6645 the function stack to the state it was in when the
6646 corresponding <a href="#int_stacksave"><tt>llvm.stacksave</tt></a> intrinsic
6647 executed. This is useful for implementing language features like scoped
6648 automatic variable sized arrays in C99.</p>
6651 <p>See the description
6652 for <a href="#int_stacksave"><tt>llvm.stacksave</tt></a>.</p>
6656 <!-- _______________________________________________________________________ -->
6658 <a name="int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a>
6665 declare void @llvm.prefetch(i8* <address>, i32 <rw>, i32 <locality>, i32 <cache type>)
6669 <p>The '<tt>llvm.prefetch</tt>' intrinsic is a hint to the code generator to
6670 insert a prefetch instruction if supported; otherwise, it is a noop.
6671 Prefetches have no effect on the behavior of the program but can change its
6672 performance characteristics.</p>
6675 <p><tt>address</tt> is the address to be prefetched, <tt>rw</tt> is the
6676 specifier determining if the fetch should be for a read (0) or write (1),
6677 and <tt>locality</tt> is a temporal locality specifier ranging from (0) - no
6678 locality, to (3) - extremely local keep in cache. The <tt>cache type</tt>
6679 specifies whether the prefetch is performed on the data (1) or instruction (0)
6680 cache. The <tt>rw</tt>, <tt>locality</tt> and <tt>cache type</tt> arguments
6681 must be constant integers.</p>
6684 <p>This intrinsic does not modify the behavior of the program. In particular,
6685 prefetches cannot trap and do not produce a value. On targets that support
6686 this intrinsic, the prefetch can provide hints to the processor cache for
6687 better performance.</p>
6691 <!-- _______________________________________________________________________ -->
6693 <a name="int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a>
6700 declare void @llvm.pcmarker(i32 <id>)
6704 <p>The '<tt>llvm.pcmarker</tt>' intrinsic is a method to export a Program
6705 Counter (PC) in a region of code to simulators and other tools. The method
6706 is target specific, but it is expected that the marker will use exported
6707 symbols to transmit the PC of the marker. The marker makes no guarantees
6708 that it will remain with any specific instruction after optimizations. It is
6709 possible that the presence of a marker will inhibit optimizations. The
6710 intended use is to be inserted after optimizations to allow correlations of
6711 simulation runs.</p>
6714 <p><tt>id</tt> is a numerical id identifying the marker.</p>
6717 <p>This intrinsic does not modify the behavior of the program. Backends that do
6718 not support this intrinsic may ignore it.</p>
6722 <!-- _______________________________________________________________________ -->
6724 <a name="int_readcyclecounter">'<tt>llvm.readcyclecounter</tt>' Intrinsic</a>
6731 declare i64 @llvm.readcyclecounter()
6735 <p>The '<tt>llvm.readcyclecounter</tt>' intrinsic provides access to the cycle
6736 counter register (or similar low latency, high accuracy clocks) on those
6737 targets that support it. On X86, it should map to RDTSC. On Alpha, it
6738 should map to RPCC. As the backing counters overflow quickly (on the order
6739 of 9 seconds on alpha), this should only be used for small timings.</p>
6742 <p>When directly supported, reading the cycle counter should not modify any
6743 memory. Implementations are allowed to either return a application specific
6744 value or a system wide value. On backends without support, this is lowered
6745 to a constant 0.</p>
6751 <!-- ======================================================================= -->
6753 <a name="int_libc">Standard C Library Intrinsics</a>
6758 <p>LLVM provides intrinsics for a few important standard C library functions.
6759 These intrinsics allow source-language front-ends to pass information about
6760 the alignment of the pointer arguments to the code generator, providing
6761 opportunity for more efficient code generation.</p>
6763 <!-- _______________________________________________________________________ -->
6765 <a name="int_memcpy">'<tt>llvm.memcpy</tt>' Intrinsic</a>
6771 <p>This is an overloaded intrinsic. You can use <tt>llvm.memcpy</tt> on any
6772 integer bit width and for different address spaces. Not all targets support
6773 all bit widths however.</p>
6776 declare void @llvm.memcpy.p0i8.p0i8.i32(i8* <dest>, i8* <src>,
6777 i32 <len>, i32 <align>, i1 <isvolatile>)
6778 declare void @llvm.memcpy.p0i8.p0i8.i64(i8* <dest>, i8* <src>,
6779 i64 <len>, i32 <align>, i1 <isvolatile>)
6783 <p>The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the
6784 source location to the destination location.</p>
6786 <p>Note that, unlike the standard libc function, the <tt>llvm.memcpy.*</tt>
6787 intrinsics do not return a value, takes extra alignment/isvolatile arguments
6788 and the pointers can be in specified address spaces.</p>
6792 <p>The first argument is a pointer to the destination, the second is a pointer
6793 to the source. The third argument is an integer argument specifying the
6794 number of bytes to copy, the fourth argument is the alignment of the
6795 source and destination locations, and the fifth is a boolean indicating a
6796 volatile access.</p>
6798 <p>If the call to this intrinsic has an alignment value that is not 0 or 1,
6799 then the caller guarantees that both the source and destination pointers are
6800 aligned to that boundary.</p>
6802 <p>If the <tt>isvolatile</tt> parameter is <tt>true</tt>, the
6803 <tt>llvm.memcpy</tt> call is a <a href="#volatile">volatile operation</a>.
6804 The detailed access behavior is not very cleanly specified and it is unwise
6805 to depend on it.</p>
6809 <p>The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the
6810 source location to the destination location, which are not allowed to
6811 overlap. It copies "len" bytes of memory over. If the argument is known to
6812 be aligned to some boundary, this can be specified as the fourth argument,
6813 otherwise it should be set to 0 or 1.</p>
6817 <!-- _______________________________________________________________________ -->
6819 <a name="int_memmove">'<tt>llvm.memmove</tt>' Intrinsic</a>
6825 <p>This is an overloaded intrinsic. You can use llvm.memmove on any integer bit
6826 width and for different address space. Not all targets support all bit
6830 declare void @llvm.memmove.p0i8.p0i8.i32(i8* <dest>, i8* <src>,
6831 i32 <len>, i32 <align>, i1 <isvolatile>)
6832 declare void @llvm.memmove.p0i8.p0i8.i64(i8* <dest>, i8* <src>,
6833 i64 <len>, i32 <align>, i1 <isvolatile>)
6837 <p>The '<tt>llvm.memmove.*</tt>' intrinsics move a block of memory from the
6838 source location to the destination location. It is similar to the
6839 '<tt>llvm.memcpy</tt>' intrinsic but allows the two memory locations to
6842 <p>Note that, unlike the standard libc function, the <tt>llvm.memmove.*</tt>
6843 intrinsics do not return a value, takes extra alignment/isvolatile arguments
6844 and the pointers can be in specified address spaces.</p>
6848 <p>The first argument is a pointer to the destination, the second is a pointer
6849 to the source. The third argument is an integer argument specifying the
6850 number of bytes to copy, the fourth argument is the alignment of the
6851 source and destination locations, and the fifth is a boolean indicating a
6852 volatile access.</p>
6854 <p>If the call to this intrinsic has an alignment value that is not 0 or 1,
6855 then the caller guarantees that the source and destination pointers are
6856 aligned to that boundary.</p>
6858 <p>If the <tt>isvolatile</tt> parameter is <tt>true</tt>, the
6859 <tt>llvm.memmove</tt> call is a <a href="#volatile">volatile operation</a>.
6860 The detailed access behavior is not very cleanly specified and it is unwise
6861 to depend on it.</p>
6865 <p>The '<tt>llvm.memmove.*</tt>' intrinsics copy a block of memory from the
6866 source location to the destination location, which may overlap. It copies
6867 "len" bytes of memory over. If the argument is known to be aligned to some
6868 boundary, this can be specified as the fourth argument, otherwise it should
6869 be set to 0 or 1.</p>
6873 <!-- _______________________________________________________________________ -->
6875 <a name="int_memset">'<tt>llvm.memset.*</tt>' Intrinsics</a>
6881 <p>This is an overloaded intrinsic. You can use llvm.memset on any integer bit
6882 width and for different address spaces. However, not all targets support all
6886 declare void @llvm.memset.p0i8.i32(i8* <dest>, i8 <val>,
6887 i32 <len>, i32 <align>, i1 <isvolatile>)
6888 declare void @llvm.memset.p0i8.i64(i8* <dest>, i8 <val>,
6889 i64 <len>, i32 <align>, i1 <isvolatile>)
6893 <p>The '<tt>llvm.memset.*</tt>' intrinsics fill a block of memory with a
6894 particular byte value.</p>
6896 <p>Note that, unlike the standard libc function, the <tt>llvm.memset</tt>
6897 intrinsic does not return a value and takes extra alignment/volatile
6898 arguments. Also, the destination can be in an arbitrary address space.</p>
6901 <p>The first argument is a pointer to the destination to fill, the second is the
6902 byte value with which to fill it, the third argument is an integer argument
6903 specifying the number of bytes to fill, and the fourth argument is the known
6904 alignment of the destination location.</p>
6906 <p>If the call to this intrinsic has an alignment value that is not 0 or 1,
6907 then the caller guarantees that the destination pointer is aligned to that
6910 <p>If the <tt>isvolatile</tt> parameter is <tt>true</tt>, the
6911 <tt>llvm.memset</tt> call is a <a href="#volatile">volatile operation</a>.
6912 The detailed access behavior is not very cleanly specified and it is unwise
6913 to depend on it.</p>
6916 <p>The '<tt>llvm.memset.*</tt>' intrinsics fill "len" bytes of memory starting
6917 at the destination location. If the argument is known to be aligned to some
6918 boundary, this can be specified as the fourth argument, otherwise it should
6919 be set to 0 or 1.</p>
6923 <!-- _______________________________________________________________________ -->
6925 <a name="int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a>
6931 <p>This is an overloaded intrinsic. You can use <tt>llvm.sqrt</tt> on any
6932 floating point or vector of floating point type. Not all targets support all
6936 declare float @llvm.sqrt.f32(float %Val)
6937 declare double @llvm.sqrt.f64(double %Val)
6938 declare x86_fp80 @llvm.sqrt.f80(x86_fp80 %Val)
6939 declare fp128 @llvm.sqrt.f128(fp128 %Val)
6940 declare ppc_fp128 @llvm.sqrt.ppcf128(ppc_fp128 %Val)
6944 <p>The '<tt>llvm.sqrt</tt>' intrinsics return the sqrt of the specified operand,
6945 returning the same value as the libm '<tt>sqrt</tt>' functions would.
6946 Unlike <tt>sqrt</tt> in libm, however, <tt>llvm.sqrt</tt> has undefined
6947 behavior for negative numbers other than -0.0 (which allows for better
6948 optimization, because there is no need to worry about errno being
6949 set). <tt>llvm.sqrt(-0.0)</tt> is defined to return -0.0 like IEEE sqrt.</p>
6952 <p>The argument and return value are floating point numbers of the same
6956 <p>This function returns the sqrt of the specified operand if it is a
6957 nonnegative floating point number.</p>
6961 <!-- _______________________________________________________________________ -->
6963 <a name="int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a>
6969 <p>This is an overloaded intrinsic. You can use <tt>llvm.powi</tt> on any
6970 floating point or vector of floating point type. Not all targets support all
6974 declare float @llvm.powi.f32(float %Val, i32 %power)
6975 declare double @llvm.powi.f64(double %Val, i32 %power)
6976 declare x86_fp80 @llvm.powi.f80(x86_fp80 %Val, i32 %power)
6977 declare fp128 @llvm.powi.f128(fp128 %Val, i32 %power)
6978 declare ppc_fp128 @llvm.powi.ppcf128(ppc_fp128 %Val, i32 %power)
6982 <p>The '<tt>llvm.powi.*</tt>' intrinsics return the first operand raised to the
6983 specified (positive or negative) power. The order of evaluation of
6984 multiplications is not defined. When a vector of floating point type is
6985 used, the second argument remains a scalar integer value.</p>
6988 <p>The second argument is an integer power, and the first is a value to raise to
6992 <p>This function returns the first value raised to the second power with an
6993 unspecified sequence of rounding operations.</p>
6997 <!-- _______________________________________________________________________ -->
6999 <a name="int_sin">'<tt>llvm.sin.*</tt>' Intrinsic</a>
7005 <p>This is an overloaded intrinsic. You can use <tt>llvm.sin</tt> on any
7006 floating point or vector of floating point type. Not all targets support all
7010 declare float @llvm.sin.f32(float %Val)
7011 declare double @llvm.sin.f64(double %Val)
7012 declare x86_fp80 @llvm.sin.f80(x86_fp80 %Val)
7013 declare fp128 @llvm.sin.f128(fp128 %Val)
7014 declare ppc_fp128 @llvm.sin.ppcf128(ppc_fp128 %Val)
7018 <p>The '<tt>llvm.sin.*</tt>' intrinsics return the sine of the operand.</p>
7021 <p>The argument and return value are floating point numbers of the same
7025 <p>This function returns the sine of the specified operand, returning the same
7026 values as the libm <tt>sin</tt> functions would, and handles error conditions
7027 in the same way.</p>
7031 <!-- _______________________________________________________________________ -->
7033 <a name="int_cos">'<tt>llvm.cos.*</tt>' Intrinsic</a>
7039 <p>This is an overloaded intrinsic. You can use <tt>llvm.cos</tt> on any
7040 floating point or vector of floating point type. Not all targets support all
7044 declare float @llvm.cos.f32(float %Val)
7045 declare double @llvm.cos.f64(double %Val)
7046 declare x86_fp80 @llvm.cos.f80(x86_fp80 %Val)
7047 declare fp128 @llvm.cos.f128(fp128 %Val)
7048 declare ppc_fp128 @llvm.cos.ppcf128(ppc_fp128 %Val)
7052 <p>The '<tt>llvm.cos.*</tt>' intrinsics return the cosine of the operand.</p>
7055 <p>The argument and return value are floating point numbers of the same
7059 <p>This function returns the cosine of the specified operand, returning the same
7060 values as the libm <tt>cos</tt> functions would, and handles error conditions
7061 in the same way.</p>
7065 <!-- _______________________________________________________________________ -->
7067 <a name="int_pow">'<tt>llvm.pow.*</tt>' Intrinsic</a>
7073 <p>This is an overloaded intrinsic. You can use <tt>llvm.pow</tt> on any
7074 floating point or vector of floating point type. Not all targets support all
7078 declare float @llvm.pow.f32(float %Val, float %Power)
7079 declare double @llvm.pow.f64(double %Val, double %Power)
7080 declare x86_fp80 @llvm.pow.f80(x86_fp80 %Val, x86_fp80 %Power)
7081 declare fp128 @llvm.pow.f128(fp128 %Val, fp128 %Power)
7082 declare ppc_fp128 @llvm.pow.ppcf128(ppc_fp128 %Val, ppc_fp128 Power)
7086 <p>The '<tt>llvm.pow.*</tt>' intrinsics return the first operand raised to the
7087 specified (positive or negative) power.</p>
7090 <p>The second argument is a floating point power, and the first is a value to
7091 raise to that power.</p>
7094 <p>This function returns the first value raised to the second power, returning
7095 the same values as the libm <tt>pow</tt> functions would, and handles error
7096 conditions in the same way.</p>
7100 <!-- _______________________________________________________________________ -->
7102 <a name="int_exp">'<tt>llvm.exp.*</tt>' Intrinsic</a>
7108 <p>This is an overloaded intrinsic. You can use <tt>llvm.exp</tt> on any
7109 floating point or vector of floating point type. Not all targets support all
7113 declare float @llvm.exp.f32(float %Val)
7114 declare double @llvm.exp.f64(double %Val)
7115 declare x86_fp80 @llvm.exp.f80(x86_fp80 %Val)
7116 declare fp128 @llvm.exp.f128(fp128 %Val)
7117 declare ppc_fp128 @llvm.exp.ppcf128(ppc_fp128 %Val)
7121 <p>The '<tt>llvm.exp.*</tt>' intrinsics perform the exp function.</p>
7124 <p>The argument and return value are floating point numbers of the same
7128 <p>This function returns the same values as the libm <tt>exp</tt> functions
7129 would, and handles error conditions in the same way.</p>
7133 <!-- _______________________________________________________________________ -->
7135 <a name="int_log">'<tt>llvm.log.*</tt>' Intrinsic</a>
7141 <p>This is an overloaded intrinsic. You can use <tt>llvm.log</tt> on any
7142 floating point or vector of floating point type. Not all targets support all
7146 declare float @llvm.log.f32(float %Val)
7147 declare double @llvm.log.f64(double %Val)
7148 declare x86_fp80 @llvm.log.f80(x86_fp80 %Val)
7149 declare fp128 @llvm.log.f128(fp128 %Val)
7150 declare ppc_fp128 @llvm.log.ppcf128(ppc_fp128 %Val)
7154 <p>The '<tt>llvm.log.*</tt>' intrinsics perform the log function.</p>
7157 <p>The argument and return value are floating point numbers of the same
7161 <p>This function returns the same values as the libm <tt>log</tt> functions
7162 would, and handles error conditions in the same way.</p>
7166 <!-- _______________________________________________________________________ -->
7168 <a name="int_fma">'<tt>llvm.fma.*</tt>' Intrinsic</a>
7174 <p>This is an overloaded intrinsic. You can use <tt>llvm.fma</tt> on any
7175 floating point or vector of floating point type. Not all targets support all
7179 declare float @llvm.fma.f32(float %a, float %b, float %c)
7180 declare double @llvm.fma.f64(double %a, double %b, double %c)
7181 declare x86_fp80 @llvm.fma.f80(x86_fp80 %a, x86_fp80 %b, x86_fp80 %c)
7182 declare fp128 @llvm.fma.f128(fp128 %a, fp128 %b, fp128 %c)
7183 declare ppc_fp128 @llvm.fma.ppcf128(ppc_fp128 %a, ppc_fp128 %b, ppc_fp128 %c)
7187 <p>The '<tt>llvm.fma.*</tt>' intrinsics perform the fused multiply-add
7191 <p>The argument and return value are floating point numbers of the same
7195 <p>This function returns the same values as the libm <tt>fma</tt> functions
7202 <!-- ======================================================================= -->
7204 <a name="int_manip">Bit Manipulation Intrinsics</a>
7209 <p>LLVM provides intrinsics for a few important bit manipulation operations.
7210 These allow efficient code generation for some algorithms.</p>
7212 <!-- _______________________________________________________________________ -->
7214 <a name="int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a>
7220 <p>This is an overloaded intrinsic function. You can use bswap on any integer
7221 type that is an even number of bytes (i.e. BitWidth % 16 == 0).</p>
7224 declare i16 @llvm.bswap.i16(i16 <id>)
7225 declare i32 @llvm.bswap.i32(i32 <id>)
7226 declare i64 @llvm.bswap.i64(i64 <id>)
7230 <p>The '<tt>llvm.bswap</tt>' family of intrinsics is used to byte swap integer
7231 values with an even number of bytes (positive multiple of 16 bits). These
7232 are useful for performing operations on data that is not in the target's
7233 native byte order.</p>
7236 <p>The <tt>llvm.bswap.i16</tt> intrinsic returns an i16 value that has the high
7237 and low byte of the input i16 swapped. Similarly,
7238 the <tt>llvm.bswap.i32</tt> intrinsic returns an i32 value that has the four
7239 bytes of the input i32 swapped, so that if the input bytes are numbered 0, 1,
7240 2, 3 then the returned i32 will have its bytes in 3, 2, 1, 0 order.
7241 The <tt>llvm.bswap.i48</tt>, <tt>llvm.bswap.i64</tt> and other intrinsics
7242 extend this concept to additional even-byte lengths (6 bytes, 8 bytes and
7243 more, respectively).</p>
7247 <!-- _______________________________________________________________________ -->
7249 <a name="int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic</a>
7255 <p>This is an overloaded intrinsic. You can use llvm.ctpop on any integer bit
7256 width, or on any vector with integer elements. Not all targets support all
7257 bit widths or vector types, however.</p>
7260 declare i8 @llvm.ctpop.i8(i8 <src>)
7261 declare i16 @llvm.ctpop.i16(i16 <src>)
7262 declare i32 @llvm.ctpop.i32(i32 <src>)
7263 declare i64 @llvm.ctpop.i64(i64 <src>)
7264 declare i256 @llvm.ctpop.i256(i256 <src>)
7265 declare <2 x i32> @llvm.ctpop.v2i32(<2 x i32> <src>)
7269 <p>The '<tt>llvm.ctpop</tt>' family of intrinsics counts the number of bits set
7273 <p>The only argument is the value to be counted. The argument may be of any
7274 integer type, or a vector with integer elements.
7275 The return type must match the argument type.</p>
7278 <p>The '<tt>llvm.ctpop</tt>' intrinsic counts the 1's in a variable, or within each
7279 element of a vector.</p>
7283 <!-- _______________________________________________________________________ -->
7285 <a name="int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic</a>
7291 <p>This is an overloaded intrinsic. You can use <tt>llvm.ctlz</tt> on any
7292 integer bit width, or any vector whose elements are integers. Not all
7293 targets support all bit widths or vector types, however.</p>
7296 declare i8 @llvm.ctlz.i8 (i8 <src>)
7297 declare i16 @llvm.ctlz.i16(i16 <src>)
7298 declare i32 @llvm.ctlz.i32(i32 <src>)
7299 declare i64 @llvm.ctlz.i64(i64 <src>)
7300 declare i256 @llvm.ctlz.i256(i256 <src>)
7301 declare <2 x i32> @llvm.ctlz.v2i32(<2 x i32> <src;gt)
7305 <p>The '<tt>llvm.ctlz</tt>' family of intrinsic functions counts the number of
7306 leading zeros in a variable.</p>
7309 <p>The only argument is the value to be counted. The argument may be of any
7310 integer type, or any vector type with integer element type.
7311 The return type must match the argument type.</p>
7314 <p>The '<tt>llvm.ctlz</tt>' intrinsic counts the leading (most significant)
7315 zeros in a variable, or within each element of the vector if the operation
7316 is of vector type. If the src == 0 then the result is the size in bits of
7317 the type of src. For example, <tt>llvm.ctlz(i32 2) = 30</tt>.</p>
7321 <!-- _______________________________________________________________________ -->
7323 <a name="int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic</a>
7329 <p>This is an overloaded intrinsic. You can use <tt>llvm.cttz</tt> on any
7330 integer bit width, or any vector of integer elements. Not all targets
7331 support all bit widths or vector types, however.</p>
7334 declare i8 @llvm.cttz.i8 (i8 <src>)
7335 declare i16 @llvm.cttz.i16(i16 <src>)
7336 declare i32 @llvm.cttz.i32(i32 <src>)
7337 declare i64 @llvm.cttz.i64(i64 <src>)
7338 declare i256 @llvm.cttz.i256(i256 <src>)
7339 declase <2 x i32> @llvm.cttz.v2i32(<2 x i32> <src>)
7343 <p>The '<tt>llvm.cttz</tt>' family of intrinsic functions counts the number of
7347 <p>The only argument is the value to be counted. The argument may be of any
7348 integer type, or a vectory with integer element type.. The return type
7349 must match the argument type.</p>
7352 <p>The '<tt>llvm.cttz</tt>' intrinsic counts the trailing (least significant)
7353 zeros in a variable, or within each element of a vector.
7354 If the src == 0 then the result is the size in bits of
7355 the type of src. For example, <tt>llvm.cttz(2) = 1</tt>.</p>
7361 <!-- ======================================================================= -->
7363 <a name="int_overflow">Arithmetic with Overflow Intrinsics</a>
7368 <p>LLVM provides intrinsics for some arithmetic with overflow operations.</p>
7370 <!-- _______________________________________________________________________ -->
7372 <a name="int_sadd_overflow">
7373 '<tt>llvm.sadd.with.overflow.*</tt>' Intrinsics
7380 <p>This is an overloaded intrinsic. You can use <tt>llvm.sadd.with.overflow</tt>
7381 on any integer bit width.</p>
7384 declare {i16, i1} @llvm.sadd.with.overflow.i16(i16 %a, i16 %b)
7385 declare {i32, i1} @llvm.sadd.with.overflow.i32(i32 %a, i32 %b)
7386 declare {i64, i1} @llvm.sadd.with.overflow.i64(i64 %a, i64 %b)
7390 <p>The '<tt>llvm.sadd.with.overflow</tt>' family of intrinsic functions perform
7391 a signed addition of the two arguments, and indicate whether an overflow
7392 occurred during the signed summation.</p>
7395 <p>The arguments (%a and %b) and the first element of the result structure may
7396 be of integer types of any bit width, but they must have the same bit
7397 width. The second element of the result structure must be of
7398 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
7399 undergo signed addition.</p>
7402 <p>The '<tt>llvm.sadd.with.overflow</tt>' family of intrinsic functions perform
7403 a signed addition of the two variables. They return a structure — the
7404 first element of which is the signed summation, and the second element of
7405 which is a bit specifying if the signed summation resulted in an
7410 %res = call {i32, i1} @llvm.sadd.with.overflow.i32(i32 %a, i32 %b)
7411 %sum = extractvalue {i32, i1} %res, 0
7412 %obit = extractvalue {i32, i1} %res, 1
7413 br i1 %obit, label %overflow, label %normal
7418 <!-- _______________________________________________________________________ -->
7420 <a name="int_uadd_overflow">
7421 '<tt>llvm.uadd.with.overflow.*</tt>' Intrinsics
7428 <p>This is an overloaded intrinsic. You can use <tt>llvm.uadd.with.overflow</tt>
7429 on any integer bit width.</p>
7432 declare {i16, i1} @llvm.uadd.with.overflow.i16(i16 %a, i16 %b)
7433 declare {i32, i1} @llvm.uadd.with.overflow.i32(i32 %a, i32 %b)
7434 declare {i64, i1} @llvm.uadd.with.overflow.i64(i64 %a, i64 %b)
7438 <p>The '<tt>llvm.uadd.with.overflow</tt>' family of intrinsic functions perform
7439 an unsigned addition of the two arguments, and indicate whether a carry
7440 occurred during the unsigned summation.</p>
7443 <p>The arguments (%a and %b) and the first element of the result structure may
7444 be of integer types of any bit width, but they must have the same bit
7445 width. The second element of the result structure must be of
7446 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
7447 undergo unsigned addition.</p>
7450 <p>The '<tt>llvm.uadd.with.overflow</tt>' family of intrinsic functions perform
7451 an unsigned addition of the two arguments. They return a structure —
7452 the first element of which is the sum, and the second element of which is a
7453 bit specifying if the unsigned summation resulted in a carry.</p>
7457 %res = call {i32, i1} @llvm.uadd.with.overflow.i32(i32 %a, i32 %b)
7458 %sum = extractvalue {i32, i1} %res, 0
7459 %obit = extractvalue {i32, i1} %res, 1
7460 br i1 %obit, label %carry, label %normal
7465 <!-- _______________________________________________________________________ -->
7467 <a name="int_ssub_overflow">
7468 '<tt>llvm.ssub.with.overflow.*</tt>' Intrinsics
7475 <p>This is an overloaded intrinsic. You can use <tt>llvm.ssub.with.overflow</tt>
7476 on any integer bit width.</p>
7479 declare {i16, i1} @llvm.ssub.with.overflow.i16(i16 %a, i16 %b)
7480 declare {i32, i1} @llvm.ssub.with.overflow.i32(i32 %a, i32 %b)
7481 declare {i64, i1} @llvm.ssub.with.overflow.i64(i64 %a, i64 %b)
7485 <p>The '<tt>llvm.ssub.with.overflow</tt>' family of intrinsic functions perform
7486 a signed subtraction of the two arguments, and indicate whether an overflow
7487 occurred during the signed subtraction.</p>
7490 <p>The arguments (%a and %b) and the first element of the result structure may
7491 be of integer types of any bit width, but they must have the same bit
7492 width. The second element of the result structure must be of
7493 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
7494 undergo signed subtraction.</p>
7497 <p>The '<tt>llvm.ssub.with.overflow</tt>' family of intrinsic functions perform
7498 a signed subtraction of the two arguments. They return a structure —
7499 the first element of which is the subtraction, and the second element of
7500 which is a bit specifying if the signed subtraction resulted in an
7505 %res = call {i32, i1} @llvm.ssub.with.overflow.i32(i32 %a, i32 %b)
7506 %sum = extractvalue {i32, i1} %res, 0
7507 %obit = extractvalue {i32, i1} %res, 1
7508 br i1 %obit, label %overflow, label %normal
7513 <!-- _______________________________________________________________________ -->
7515 <a name="int_usub_overflow">
7516 '<tt>llvm.usub.with.overflow.*</tt>' Intrinsics
7523 <p>This is an overloaded intrinsic. You can use <tt>llvm.usub.with.overflow</tt>
7524 on any integer bit width.</p>
7527 declare {i16, i1} @llvm.usub.with.overflow.i16(i16 %a, i16 %b)
7528 declare {i32, i1} @llvm.usub.with.overflow.i32(i32 %a, i32 %b)
7529 declare {i64, i1} @llvm.usub.with.overflow.i64(i64 %a, i64 %b)
7533 <p>The '<tt>llvm.usub.with.overflow</tt>' family of intrinsic functions perform
7534 an unsigned subtraction of the two arguments, and indicate whether an
7535 overflow occurred during the unsigned subtraction.</p>
7538 <p>The arguments (%a and %b) and the first element of the result structure may
7539 be of integer types of any bit width, but they must have the same bit
7540 width. The second element of the result structure must be of
7541 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
7542 undergo unsigned subtraction.</p>
7545 <p>The '<tt>llvm.usub.with.overflow</tt>' family of intrinsic functions perform
7546 an unsigned subtraction of the two arguments. They return a structure —
7547 the first element of which is the subtraction, and the second element of
7548 which is a bit specifying if the unsigned subtraction resulted in an
7553 %res = call {i32, i1} @llvm.usub.with.overflow.i32(i32 %a, i32 %b)
7554 %sum = extractvalue {i32, i1} %res, 0
7555 %obit = extractvalue {i32, i1} %res, 1
7556 br i1 %obit, label %overflow, label %normal
7561 <!-- _______________________________________________________________________ -->
7563 <a name="int_smul_overflow">
7564 '<tt>llvm.smul.with.overflow.*</tt>' Intrinsics
7571 <p>This is an overloaded intrinsic. You can use <tt>llvm.smul.with.overflow</tt>
7572 on any integer bit width.</p>
7575 declare {i16, i1} @llvm.smul.with.overflow.i16(i16 %a, i16 %b)
7576 declare {i32, i1} @llvm.smul.with.overflow.i32(i32 %a, i32 %b)
7577 declare {i64, i1} @llvm.smul.with.overflow.i64(i64 %a, i64 %b)
7582 <p>The '<tt>llvm.smul.with.overflow</tt>' family of intrinsic functions perform
7583 a signed multiplication of the two arguments, and indicate whether an
7584 overflow occurred during the signed multiplication.</p>
7587 <p>The arguments (%a and %b) and the first element of the result structure may
7588 be of integer types of any bit width, but they must have the same bit
7589 width. The second element of the result structure must be of
7590 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
7591 undergo signed multiplication.</p>
7594 <p>The '<tt>llvm.smul.with.overflow</tt>' family of intrinsic functions perform
7595 a signed multiplication of the two arguments. They return a structure —
7596 the first element of which is the multiplication, and the second element of
7597 which is a bit specifying if the signed multiplication resulted in an
7602 %res = call {i32, i1} @llvm.smul.with.overflow.i32(i32 %a, i32 %b)
7603 %sum = extractvalue {i32, i1} %res, 0
7604 %obit = extractvalue {i32, i1} %res, 1
7605 br i1 %obit, label %overflow, label %normal
7610 <!-- _______________________________________________________________________ -->
7612 <a name="int_umul_overflow">
7613 '<tt>llvm.umul.with.overflow.*</tt>' Intrinsics
7620 <p>This is an overloaded intrinsic. You can use <tt>llvm.umul.with.overflow</tt>
7621 on any integer bit width.</p>
7624 declare {i16, i1} @llvm.umul.with.overflow.i16(i16 %a, i16 %b)
7625 declare {i32, i1} @llvm.umul.with.overflow.i32(i32 %a, i32 %b)
7626 declare {i64, i1} @llvm.umul.with.overflow.i64(i64 %a, i64 %b)
7630 <p>The '<tt>llvm.umul.with.overflow</tt>' family of intrinsic functions perform
7631 a unsigned multiplication of the two arguments, and indicate whether an
7632 overflow occurred during the unsigned multiplication.</p>
7635 <p>The arguments (%a and %b) and the first element of the result structure may
7636 be of integer types of any bit width, but they must have the same bit
7637 width. The second element of the result structure must be of
7638 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
7639 undergo unsigned multiplication.</p>
7642 <p>The '<tt>llvm.umul.with.overflow</tt>' family of intrinsic functions perform
7643 an unsigned multiplication of the two arguments. They return a structure
7644 — the first element of which is the multiplication, and the second
7645 element of which is a bit specifying if the unsigned multiplication resulted
7650 %res = call {i32, i1} @llvm.umul.with.overflow.i32(i32 %a, i32 %b)
7651 %sum = extractvalue {i32, i1} %res, 0
7652 %obit = extractvalue {i32, i1} %res, 1
7653 br i1 %obit, label %overflow, label %normal
7660 <!-- ======================================================================= -->
7662 <a name="int_fp16">Half Precision Floating Point Intrinsics</a>
7667 <p>Half precision floating point is a storage-only format. This means that it is
7668 a dense encoding (in memory) but does not support computation in the
7671 <p>This means that code must first load the half-precision floating point
7672 value as an i16, then convert it to float with <a
7673 href="#int_convert_from_fp16"><tt>llvm.convert.from.fp16</tt></a>.
7674 Computation can then be performed on the float value (including extending to
7675 double etc). To store the value back to memory, it is first converted to
7676 float if needed, then converted to i16 with
7677 <a href="#int_convert_to_fp16"><tt>llvm.convert.to.fp16</tt></a>, then
7678 storing as an i16 value.</p>
7680 <!-- _______________________________________________________________________ -->
7682 <a name="int_convert_to_fp16">
7683 '<tt>llvm.convert.to.fp16</tt>' Intrinsic
7691 declare i16 @llvm.convert.to.fp16(f32 %a)
7695 <p>The '<tt>llvm.convert.to.fp16</tt>' intrinsic function performs
7696 a conversion from single precision floating point format to half precision
7697 floating point format.</p>
7700 <p>The intrinsic function contains single argument - the value to be
7704 <p>The '<tt>llvm.convert.to.fp16</tt>' intrinsic function performs
7705 a conversion from single precision floating point format to half precision
7706 floating point format. The return value is an <tt>i16</tt> which
7707 contains the converted number.</p>
7711 %res = call i16 @llvm.convert.to.fp16(f32 %a)
7712 store i16 %res, i16* @x, align 2
7717 <!-- _______________________________________________________________________ -->
7719 <a name="int_convert_from_fp16">
7720 '<tt>llvm.convert.from.fp16</tt>' Intrinsic
7728 declare f32 @llvm.convert.from.fp16(i16 %a)
7732 <p>The '<tt>llvm.convert.from.fp16</tt>' intrinsic function performs
7733 a conversion from half precision floating point format to single precision
7734 floating point format.</p>
7737 <p>The intrinsic function contains single argument - the value to be
7741 <p>The '<tt>llvm.convert.from.fp16</tt>' intrinsic function performs a
7742 conversion from half single precision floating point format to single
7743 precision floating point format. The input half-float value is represented by
7744 an <tt>i16</tt> value.</p>
7748 %a = load i16* @x, align 2
7749 %res = call f32 @llvm.convert.from.fp16(i16 %a)
7756 <!-- ======================================================================= -->
7758 <a name="int_debugger">Debugger Intrinsics</a>
7763 <p>The LLVM debugger intrinsics (which all start with <tt>llvm.dbg.</tt>
7764 prefix), are described in
7765 the <a href="SourceLevelDebugging.html#format_common_intrinsics">LLVM Source
7766 Level Debugging</a> document.</p>
7770 <!-- ======================================================================= -->
7772 <a name="int_eh">Exception Handling Intrinsics</a>
7777 <p>The LLVM exception handling intrinsics (which all start with
7778 <tt>llvm.eh.</tt> prefix), are described in
7779 the <a href="ExceptionHandling.html#format_common_intrinsics">LLVM Exception
7780 Handling</a> document.</p>
7784 <!-- ======================================================================= -->
7786 <a name="int_trampoline">Trampoline Intrinsics</a>
7791 <p>These intrinsics make it possible to excise one parameter, marked with
7792 the <a href="#nest"><tt>nest</tt></a> attribute, from a function.
7793 The result is a callable
7794 function pointer lacking the nest parameter - the caller does not need to
7795 provide a value for it. Instead, the value to use is stored in advance in a
7796 "trampoline", a block of memory usually allocated on the stack, which also
7797 contains code to splice the nest value into the argument list. This is used
7798 to implement the GCC nested function address extension.</p>
7800 <p>For example, if the function is
7801 <tt>i32 f(i8* nest %c, i32 %x, i32 %y)</tt> then the resulting function
7802 pointer has signature <tt>i32 (i32, i32)*</tt>. It can be created as
7805 <pre class="doc_code">
7806 %tramp = alloca [10 x i8], align 4 ; size and alignment only correct for X86
7807 %tramp1 = getelementptr [10 x i8]* %tramp, i32 0, i32 0
7808 call i8* @llvm.init.trampoline(i8* %tramp1, i8* bitcast (i32 (i8*, i32, i32)* @f to i8*), i8* %nval)
7809 %p = call i8* @llvm.adjust.trampoline(i8* %tramp1)
7810 %fp = bitcast i8* %p to i32 (i32, i32)*
7813 <p>The call <tt>%val = call i32 %fp(i32 %x, i32 %y)</tt> is then equivalent
7814 to <tt>%val = call i32 %f(i8* %nval, i32 %x, i32 %y)</tt>.</p>
7816 <!-- _______________________________________________________________________ -->
7819 '<tt>llvm.init.trampoline</tt>' Intrinsic
7827 declare void @llvm.init.trampoline(i8* <tramp>, i8* <func>, i8* <nval>)
7831 <p>This fills the memory pointed to by <tt>tramp</tt> with executable code,
7832 turning it into a trampoline.</p>
7835 <p>The <tt>llvm.init.trampoline</tt> intrinsic takes three arguments, all
7836 pointers. The <tt>tramp</tt> argument must point to a sufficiently large and
7837 sufficiently aligned block of memory; this memory is written to by the
7838 intrinsic. Note that the size and the alignment are target-specific - LLVM
7839 currently provides no portable way of determining them, so a front-end that
7840 generates this intrinsic needs to have some target-specific knowledge.
7841 The <tt>func</tt> argument must hold a function bitcast to
7842 an <tt>i8*</tt>.</p>
7845 <p>The block of memory pointed to by <tt>tramp</tt> is filled with target
7846 dependent code, turning it into a function. Then <tt>tramp</tt> needs to be
7847 passed to <a href="#int_at">llvm.adjust.trampoline</a> to get a pointer
7848 which can be <a href="#int_trampoline">bitcast (to a new function) and
7849 called</a>. The new function's signature is the same as that of
7850 <tt>func</tt> with any arguments marked with the <tt>nest</tt> attribute
7851 removed. At most one such <tt>nest</tt> argument is allowed, and it must be of
7852 pointer type. Calling the new function is equivalent to calling <tt>func</tt>
7853 with the same argument list, but with <tt>nval</tt> used for the missing
7854 <tt>nest</tt> argument. If, after calling <tt>llvm.init.trampoline</tt>, the
7855 memory pointed to by <tt>tramp</tt> is modified, then the effect of any later call
7856 to the returned function pointer is undefined.</p>
7859 <!-- _______________________________________________________________________ -->
7862 '<tt>llvm.adjust.trampoline</tt>' Intrinsic
7870 declare i8* @llvm.adjust.trampoline(i8* <tramp>)
7874 <p>This performs any required machine-specific adjustment to the address of a
7875 trampoline (passed as <tt>tramp</tt>).</p>
7878 <p><tt>tramp</tt> must point to a block of memory which already has trampoline code
7879 filled in by a previous call to <a href="#int_it"><tt>llvm.init.trampoline</tt>
7883 <p>On some architectures the address of the code to be executed needs to be
7884 different to the address where the trampoline is actually stored. This
7885 intrinsic returns the executable address corresponding to <tt>tramp</tt>
7886 after performing the required machine specific adjustments.
7887 The pointer returned can then be <a href="#int_trampoline"> bitcast and
7895 <!-- ======================================================================= -->
7897 <a name="int_memorymarkers">Memory Use Markers</a>
7902 <p>This class of intrinsics exists to information about the lifetime of memory
7903 objects and ranges where variables are immutable.</p>
7905 <!-- _______________________________________________________________________ -->
7907 <a name="int_lifetime_start">'<tt>llvm.lifetime.start</tt>' Intrinsic</a>
7914 declare void @llvm.lifetime.start(i64 <size>, i8* nocapture <ptr>)
7918 <p>The '<tt>llvm.lifetime.start</tt>' intrinsic specifies the start of a memory
7919 object's lifetime.</p>
7922 <p>The first argument is a constant integer representing the size of the
7923 object, or -1 if it is variable sized. The second argument is a pointer to
7927 <p>This intrinsic indicates that before this point in the code, the value of the
7928 memory pointed to by <tt>ptr</tt> is dead. This means that it is known to
7929 never be used and has an undefined value. A load from the pointer that
7930 precedes this intrinsic can be replaced with
7931 <tt>'<a href="#undefvalues">undef</a>'</tt>.</p>
7935 <!-- _______________________________________________________________________ -->
7937 <a name="int_lifetime_end">'<tt>llvm.lifetime.end</tt>' Intrinsic</a>
7944 declare void @llvm.lifetime.end(i64 <size>, i8* nocapture <ptr>)
7948 <p>The '<tt>llvm.lifetime.end</tt>' intrinsic specifies the end of a memory
7949 object's lifetime.</p>
7952 <p>The first argument is a constant integer representing the size of the
7953 object, or -1 if it is variable sized. The second argument is a pointer to
7957 <p>This intrinsic indicates that after this point in the code, the value of the
7958 memory pointed to by <tt>ptr</tt> is dead. This means that it is known to
7959 never be used and has an undefined value. Any stores into the memory object
7960 following this intrinsic may be removed as dead.
7964 <!-- _______________________________________________________________________ -->
7966 <a name="int_invariant_start">'<tt>llvm.invariant.start</tt>' Intrinsic</a>
7973 declare {}* @llvm.invariant.start(i64 <size>, i8* nocapture <ptr>)
7977 <p>The '<tt>llvm.invariant.start</tt>' intrinsic specifies that the contents of
7978 a memory object will not change.</p>
7981 <p>The first argument is a constant integer representing the size of the
7982 object, or -1 if it is variable sized. The second argument is a pointer to
7986 <p>This intrinsic indicates that until an <tt>llvm.invariant.end</tt> that uses
7987 the return value, the referenced memory location is constant and
7992 <!-- _______________________________________________________________________ -->
7994 <a name="int_invariant_end">'<tt>llvm.invariant.end</tt>' Intrinsic</a>
8001 declare void @llvm.invariant.end({}* <start>, i64 <size>, i8* nocapture <ptr>)
8005 <p>The '<tt>llvm.invariant.end</tt>' intrinsic specifies that the contents of
8006 a memory object are mutable.</p>
8009 <p>The first argument is the matching <tt>llvm.invariant.start</tt> intrinsic.
8010 The second argument is a constant integer representing the size of the
8011 object, or -1 if it is variable sized and the third argument is a pointer
8015 <p>This intrinsic indicates that the memory is mutable again.</p>
8021 <!-- ======================================================================= -->
8023 <a name="int_general">General Intrinsics</a>
8028 <p>This class of intrinsics is designed to be generic and has no specific
8031 <!-- _______________________________________________________________________ -->
8033 <a name="int_var_annotation">'<tt>llvm.var.annotation</tt>' Intrinsic</a>
8040 declare void @llvm.var.annotation(i8* <val>, i8* <str>, i8* <str>, i32 <int>)
8044 <p>The '<tt>llvm.var.annotation</tt>' intrinsic.</p>
8047 <p>The first argument is a pointer to a value, the second is a pointer to a
8048 global string, the third is a pointer to a global string which is the source
8049 file name, and the last argument is the line number.</p>
8052 <p>This intrinsic allows annotation of local variables with arbitrary strings.
8053 This can be useful for special purpose optimizations that want to look for
8054 these annotations. These have no other defined use; they are ignored by code
8055 generation and optimization.</p>
8059 <!-- _______________________________________________________________________ -->
8061 <a name="int_annotation">'<tt>llvm.annotation.*</tt>' Intrinsic</a>
8067 <p>This is an overloaded intrinsic. You can use '<tt>llvm.annotation</tt>' on
8068 any integer bit width.</p>
8071 declare i8 @llvm.annotation.i8(i8 <val>, i8* <str>, i8* <str>, i32 <int>)
8072 declare i16 @llvm.annotation.i16(i16 <val>, i8* <str>, i8* <str>, i32 <int>)
8073 declare i32 @llvm.annotation.i32(i32 <val>, i8* <str>, i8* <str>, i32 <int>)
8074 declare i64 @llvm.annotation.i64(i64 <val>, i8* <str>, i8* <str>, i32 <int>)
8075 declare i256 @llvm.annotation.i256(i256 <val>, i8* <str>, i8* <str>, i32 <int>)
8079 <p>The '<tt>llvm.annotation</tt>' intrinsic.</p>
8082 <p>The first argument is an integer value (result of some expression), the
8083 second is a pointer to a global string, the third is a pointer to a global
8084 string which is the source file name, and the last argument is the line
8085 number. It returns the value of the first argument.</p>
8088 <p>This intrinsic allows annotations to be put on arbitrary expressions with
8089 arbitrary strings. This can be useful for special purpose optimizations that
8090 want to look for these annotations. These have no other defined use; they
8091 are ignored by code generation and optimization.</p>
8095 <!-- _______________________________________________________________________ -->
8097 <a name="int_trap">'<tt>llvm.trap</tt>' Intrinsic</a>
8104 declare void @llvm.trap()
8108 <p>The '<tt>llvm.trap</tt>' intrinsic.</p>
8114 <p>This intrinsics is lowered to the target dependent trap instruction. If the
8115 target does not have a trap instruction, this intrinsic will be lowered to
8116 the call of the <tt>abort()</tt> function.</p>
8120 <!-- _______________________________________________________________________ -->
8122 <a name="int_stackprotector">'<tt>llvm.stackprotector</tt>' Intrinsic</a>
8129 declare void @llvm.stackprotector(i8* <guard>, i8** <slot>)
8133 <p>The <tt>llvm.stackprotector</tt> intrinsic takes the <tt>guard</tt> and
8134 stores it onto the stack at <tt>slot</tt>. The stack slot is adjusted to
8135 ensure that it is placed on the stack before local variables.</p>
8138 <p>The <tt>llvm.stackprotector</tt> intrinsic requires two pointer
8139 arguments. The first argument is the value loaded from the stack
8140 guard <tt>@__stack_chk_guard</tt>. The second variable is an <tt>alloca</tt>
8141 that has enough space to hold the value of the guard.</p>
8144 <p>This intrinsic causes the prologue/epilogue inserter to force the position of
8145 the <tt>AllocaInst</tt> stack slot to be before local variables on the
8146 stack. This is to ensure that if a local variable on the stack is
8147 overwritten, it will destroy the value of the guard. When the function exits,
8148 the guard on the stack is checked against the original guard. If they are
8149 different, then the program aborts by calling the <tt>__stack_chk_fail()</tt>
8154 <!-- _______________________________________________________________________ -->
8156 <a name="int_objectsize">'<tt>llvm.objectsize</tt>' Intrinsic</a>
8163 declare i32 @llvm.objectsize.i32(i8* <object>, i1 <type>)
8164 declare i64 @llvm.objectsize.i64(i8* <object>, i1 <type>)
8168 <p>The <tt>llvm.objectsize</tt> intrinsic is designed to provide information to
8169 the optimizers to determine at compile time whether a) an operation (like
8170 memcpy) will overflow a buffer that corresponds to an object, or b) that a
8171 runtime check for overflow isn't necessary. An object in this context means
8172 an allocation of a specific class, structure, array, or other object.</p>
8175 <p>The <tt>llvm.objectsize</tt> intrinsic takes two arguments. The first
8176 argument is a pointer to or into the <tt>object</tt>. The second argument
8177 is a boolean 0 or 1. This argument determines whether you want the
8178 maximum (0) or minimum (1) bytes remaining. This needs to be a literal 0 or
8179 1, variables are not allowed.</p>
8182 <p>The <tt>llvm.objectsize</tt> intrinsic is lowered to either a constant
8183 representing the size of the object concerned, or <tt>i32/i64 -1 or 0</tt>,
8184 depending on the <tt>type</tt> argument, if the size cannot be determined at
8193 <!-- *********************************************************************** -->
8196 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
8197 src="http://jigsaw.w3.org/css-validator/images/vcss-blue" alt="Valid CSS"></a>
8198 <a href="http://validator.w3.org/check/referer"><img
8199 src="http://www.w3.org/Icons/valid-html401-blue" alt="Valid HTML 4.01"></a>
8201 <a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
8202 <a href="http://llvm.org/">The LLVM Compiler Infrastructure</a><br>
8203 Last modified: $Date$