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="#poisonvalues">Poison 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="#module_flags">Module Flags Metadata</a>
115 <li><a href="#intrinsic_globals">Intrinsic Global Variables</a>
117 <li><a href="#intg_used">The '<tt>llvm.used</tt>' Global Variable</a></li>
118 <li><a href="#intg_compiler_used">The '<tt>llvm.compiler.used</tt>'
119 Global Variable</a></li>
120 <li><a href="#intg_global_ctors">The '<tt>llvm.global_ctors</tt>'
121 Global Variable</a></li>
122 <li><a href="#intg_global_dtors">The '<tt>llvm.global_dtors</tt>'
123 Global Variable</a></li>
126 <li><a href="#instref">Instruction Reference</a>
128 <li><a href="#terminators">Terminator Instructions</a>
130 <li><a href="#i_ret">'<tt>ret</tt>' Instruction</a></li>
131 <li><a href="#i_br">'<tt>br</tt>' Instruction</a></li>
132 <li><a href="#i_switch">'<tt>switch</tt>' Instruction</a></li>
133 <li><a href="#i_indirectbr">'<tt>indirectbr</tt>' Instruction</a></li>
134 <li><a href="#i_invoke">'<tt>invoke</tt>' Instruction</a></li>
135 <li><a href="#i_resume">'<tt>resume</tt>' Instruction</a></li>
136 <li><a href="#i_unreachable">'<tt>unreachable</tt>' Instruction</a></li>
139 <li><a href="#binaryops">Binary Operations</a>
141 <li><a href="#i_add">'<tt>add</tt>' Instruction</a></li>
142 <li><a href="#i_fadd">'<tt>fadd</tt>' Instruction</a></li>
143 <li><a href="#i_sub">'<tt>sub</tt>' Instruction</a></li>
144 <li><a href="#i_fsub">'<tt>fsub</tt>' Instruction</a></li>
145 <li><a href="#i_mul">'<tt>mul</tt>' Instruction</a></li>
146 <li><a href="#i_fmul">'<tt>fmul</tt>' Instruction</a></li>
147 <li><a href="#i_udiv">'<tt>udiv</tt>' Instruction</a></li>
148 <li><a href="#i_sdiv">'<tt>sdiv</tt>' Instruction</a></li>
149 <li><a href="#i_fdiv">'<tt>fdiv</tt>' Instruction</a></li>
150 <li><a href="#i_urem">'<tt>urem</tt>' Instruction</a></li>
151 <li><a href="#i_srem">'<tt>srem</tt>' Instruction</a></li>
152 <li><a href="#i_frem">'<tt>frem</tt>' Instruction</a></li>
155 <li><a href="#bitwiseops">Bitwise Binary Operations</a>
157 <li><a href="#i_shl">'<tt>shl</tt>' Instruction</a></li>
158 <li><a href="#i_lshr">'<tt>lshr</tt>' Instruction</a></li>
159 <li><a href="#i_ashr">'<tt>ashr</tt>' Instruction</a></li>
160 <li><a href="#i_and">'<tt>and</tt>' Instruction</a></li>
161 <li><a href="#i_or">'<tt>or</tt>' Instruction</a></li>
162 <li><a href="#i_xor">'<tt>xor</tt>' Instruction</a></li>
165 <li><a href="#vectorops">Vector Operations</a>
167 <li><a href="#i_extractelement">'<tt>extractelement</tt>' Instruction</a></li>
168 <li><a href="#i_insertelement">'<tt>insertelement</tt>' Instruction</a></li>
169 <li><a href="#i_shufflevector">'<tt>shufflevector</tt>' Instruction</a></li>
172 <li><a href="#aggregateops">Aggregate Operations</a>
174 <li><a href="#i_extractvalue">'<tt>extractvalue</tt>' Instruction</a></li>
175 <li><a href="#i_insertvalue">'<tt>insertvalue</tt>' Instruction</a></li>
178 <li><a href="#memoryops">Memory Access and Addressing Operations</a>
180 <li><a href="#i_alloca">'<tt>alloca</tt>' Instruction</a></li>
181 <li><a href="#i_load">'<tt>load</tt>' Instruction</a></li>
182 <li><a href="#i_store">'<tt>store</tt>' Instruction</a></li>
183 <li><a href="#i_fence">'<tt>fence</tt>' Instruction</a></li>
184 <li><a href="#i_cmpxchg">'<tt>cmpxchg</tt>' Instruction</a></li>
185 <li><a href="#i_atomicrmw">'<tt>atomicrmw</tt>' Instruction</a></li>
186 <li><a href="#i_getelementptr">'<tt>getelementptr</tt>' Instruction</a></li>
189 <li><a href="#convertops">Conversion Operations</a>
191 <li><a href="#i_trunc">'<tt>trunc .. to</tt>' Instruction</a></li>
192 <li><a href="#i_zext">'<tt>zext .. to</tt>' Instruction</a></li>
193 <li><a href="#i_sext">'<tt>sext .. to</tt>' Instruction</a></li>
194 <li><a href="#i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a></li>
195 <li><a href="#i_fpext">'<tt>fpext .. to</tt>' Instruction</a></li>
196 <li><a href="#i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a></li>
197 <li><a href="#i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a></li>
198 <li><a href="#i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a></li>
199 <li><a href="#i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a></li>
200 <li><a href="#i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a></li>
201 <li><a href="#i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a></li>
202 <li><a href="#i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a></li>
205 <li><a href="#otherops">Other Operations</a>
207 <li><a href="#i_icmp">'<tt>icmp</tt>' Instruction</a></li>
208 <li><a href="#i_fcmp">'<tt>fcmp</tt>' Instruction</a></li>
209 <li><a href="#i_phi">'<tt>phi</tt>' Instruction</a></li>
210 <li><a href="#i_select">'<tt>select</tt>' Instruction</a></li>
211 <li><a href="#i_call">'<tt>call</tt>' Instruction</a></li>
212 <li><a href="#i_va_arg">'<tt>va_arg</tt>' Instruction</a></li>
213 <li><a href="#i_landingpad">'<tt>landingpad</tt>' Instruction</a></li>
218 <li><a href="#intrinsics">Intrinsic Functions</a>
220 <li><a href="#int_varargs">Variable Argument Handling Intrinsics</a>
222 <li><a href="#int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a></li>
223 <li><a href="#int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a></li>
224 <li><a href="#int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a></li>
227 <li><a href="#int_gc">Accurate Garbage Collection Intrinsics</a>
229 <li><a href="#int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a></li>
230 <li><a href="#int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a></li>
231 <li><a href="#int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a></li>
234 <li><a href="#int_codegen">Code Generator Intrinsics</a>
236 <li><a href="#int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a></li>
237 <li><a href="#int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a></li>
238 <li><a href="#int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a></li>
239 <li><a href="#int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a></li>
240 <li><a href="#int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a></li>
241 <li><a href="#int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a></li>
242 <li><a href="#int_readcyclecounter">'<tt>llvm.readcyclecounter</tt>' Intrinsic</a></li>
245 <li><a href="#int_libc">Standard C Library Intrinsics</a>
247 <li><a href="#int_memcpy">'<tt>llvm.memcpy.*</tt>' Intrinsic</a></li>
248 <li><a href="#int_memmove">'<tt>llvm.memmove.*</tt>' Intrinsic</a></li>
249 <li><a href="#int_memset">'<tt>llvm.memset.*</tt>' Intrinsic</a></li>
250 <li><a href="#int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a></li>
251 <li><a href="#int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a></li>
252 <li><a href="#int_sin">'<tt>llvm.sin.*</tt>' Intrinsic</a></li>
253 <li><a href="#int_cos">'<tt>llvm.cos.*</tt>' Intrinsic</a></li>
254 <li><a href="#int_pow">'<tt>llvm.pow.*</tt>' Intrinsic</a></li>
255 <li><a href="#int_exp">'<tt>llvm.exp.*</tt>' Intrinsic</a></li>
256 <li><a href="#int_log">'<tt>llvm.log.*</tt>' Intrinsic</a></li>
257 <li><a href="#int_fma">'<tt>llvm.fma.*</tt>' Intrinsic</a></li>
260 <li><a href="#int_manip">Bit Manipulation Intrinsics</a>
262 <li><a href="#int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a></li>
263 <li><a href="#int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic </a></li>
264 <li><a href="#int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic </a></li>
265 <li><a href="#int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic </a></li>
268 <li><a href="#int_overflow">Arithmetic with Overflow Intrinsics</a>
270 <li><a href="#int_sadd_overflow">'<tt>llvm.sadd.with.overflow.*</tt> Intrinsics</a></li>
271 <li><a href="#int_uadd_overflow">'<tt>llvm.uadd.with.overflow.*</tt> Intrinsics</a></li>
272 <li><a href="#int_ssub_overflow">'<tt>llvm.ssub.with.overflow.*</tt> Intrinsics</a></li>
273 <li><a href="#int_usub_overflow">'<tt>llvm.usub.with.overflow.*</tt> Intrinsics</a></li>
274 <li><a href="#int_smul_overflow">'<tt>llvm.smul.with.overflow.*</tt> Intrinsics</a></li>
275 <li><a href="#int_umul_overflow">'<tt>llvm.umul.with.overflow.*</tt> Intrinsics</a></li>
278 <li><a href="#int_fp16">Half Precision Floating Point Intrinsics</a>
280 <li><a href="#int_convert_to_fp16">'<tt>llvm.convert.to.fp16</tt>' Intrinsic</a></li>
281 <li><a href="#int_convert_from_fp16">'<tt>llvm.convert.from.fp16</tt>' Intrinsic</a></li>
284 <li><a href="#int_debugger">Debugger intrinsics</a></li>
285 <li><a href="#int_eh">Exception Handling intrinsics</a></li>
286 <li><a href="#int_trampoline">Trampoline Intrinsics</a>
288 <li><a href="#int_it">'<tt>llvm.init.trampoline</tt>' Intrinsic</a></li>
289 <li><a href="#int_at">'<tt>llvm.adjust.trampoline</tt>' Intrinsic</a></li>
292 <li><a href="#int_memorymarkers">Memory Use Markers</a>
294 <li><a href="#int_lifetime_start">'<tt>llvm.lifetime.start</tt>' Intrinsic</a></li>
295 <li><a href="#int_lifetime_end">'<tt>llvm.lifetime.end</tt>' Intrinsic</a></li>
296 <li><a href="#int_invariant_start">'<tt>llvm.invariant.start</tt>' Intrinsic</a></li>
297 <li><a href="#int_invariant_end">'<tt>llvm.invariant.end</tt>' Intrinsic</a></li>
300 <li><a href="#int_general">General intrinsics</a>
302 <li><a href="#int_var_annotation">
303 '<tt>llvm.var.annotation</tt>' Intrinsic</a></li>
304 <li><a href="#int_annotation">
305 '<tt>llvm.annotation.*</tt>' Intrinsic</a></li>
306 <li><a href="#int_trap">
307 '<tt>llvm.trap</tt>' Intrinsic</a></li>
308 <li><a href="#int_stackprotector">
309 '<tt>llvm.stackprotector</tt>' Intrinsic</a></li>
310 <li><a href="#int_objectsize">
311 '<tt>llvm.objectsize</tt>' Intrinsic</a></li>
312 <li><a href="#int_expect">
313 '<tt>llvm.expect</tt>' Intrinsic</a></li>
320 <div class="doc_author">
321 <p>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a>
322 and <a href="mailto:vadve@cs.uiuc.edu">Vikram Adve</a></p>
325 <!-- *********************************************************************** -->
326 <h2><a name="abstract">Abstract</a></h2>
327 <!-- *********************************************************************** -->
331 <p>This document is a reference manual for the LLVM assembly language. LLVM is
332 a Static Single Assignment (SSA) based representation that provides type
333 safety, low-level operations, flexibility, and the capability of representing
334 'all' high-level languages cleanly. It is the common code representation
335 used throughout all phases of the LLVM compilation strategy.</p>
339 <!-- *********************************************************************** -->
340 <h2><a name="introduction">Introduction</a></h2>
341 <!-- *********************************************************************** -->
345 <p>The LLVM code representation is designed to be used in three different forms:
346 as an in-memory compiler IR, as an on-disk bitcode representation (suitable
347 for fast loading by a Just-In-Time compiler), and as a human readable
348 assembly language representation. This allows LLVM to provide a powerful
349 intermediate representation for efficient compiler transformations and
350 analysis, while providing a natural means to debug and visualize the
351 transformations. The three different forms of LLVM are all equivalent. This
352 document describes the human readable representation and notation.</p>
354 <p>The LLVM representation aims to be light-weight and low-level while being
355 expressive, typed, and extensible at the same time. It aims to be a
356 "universal IR" of sorts, by being at a low enough level that high-level ideas
357 may be cleanly mapped to it (similar to how microprocessors are "universal
358 IR's", allowing many source languages to be mapped to them). By providing
359 type information, LLVM can be used as the target of optimizations: for
360 example, through pointer analysis, it can be proven that a C automatic
361 variable is never accessed outside of the current function, allowing it to
362 be promoted to a simple SSA value instead of a memory location.</p>
364 <!-- _______________________________________________________________________ -->
366 <a name="wellformed">Well-Formedness</a>
371 <p>It is important to note that this document describes 'well formed' LLVM
372 assembly language. There is a difference between what the parser accepts and
373 what is considered 'well formed'. For example, the following instruction is
374 syntactically okay, but not well formed:</p>
376 <pre class="doc_code">
377 %x = <a href="#i_add">add</a> i32 1, %x
380 <p>because the definition of <tt>%x</tt> does not dominate all of its uses. The
381 LLVM infrastructure provides a verification pass that may be used to verify
382 that an LLVM module is well formed. This pass is automatically run by the
383 parser after parsing input assembly and by the optimizer before it outputs
384 bitcode. The violations pointed out by the verifier pass indicate bugs in
385 transformation passes or input to the parser.</p>
391 <!-- Describe the typesetting conventions here. -->
393 <!-- *********************************************************************** -->
394 <h2><a name="identifiers">Identifiers</a></h2>
395 <!-- *********************************************************************** -->
399 <p>LLVM identifiers come in two basic types: global and local. Global
400 identifiers (functions, global variables) begin with the <tt>'@'</tt>
401 character. Local identifiers (register names, types) begin with
402 the <tt>'%'</tt> character. Additionally, there are three different formats
403 for identifiers, for different purposes:</p>
406 <li>Named values are represented as a string of characters with their prefix.
407 For example, <tt>%foo</tt>, <tt>@DivisionByZero</tt>,
408 <tt>%a.really.long.identifier</tt>. The actual regular expression used is
409 '<tt>[%@][a-zA-Z$._][a-zA-Z$._0-9]*</tt>'. Identifiers which require
410 other characters in their names can be surrounded with quotes. Special
411 characters may be escaped using <tt>"\xx"</tt> where <tt>xx</tt> is the
412 ASCII code for the character in hexadecimal. In this way, any character
413 can be used in a name value, even quotes themselves.</li>
415 <li>Unnamed values are represented as an unsigned numeric value with their
416 prefix. For example, <tt>%12</tt>, <tt>@2</tt>, <tt>%44</tt>.</li>
418 <li>Constants, which are described in a <a href="#constants">section about
419 constants</a>, below.</li>
422 <p>LLVM requires that values start with a prefix for two reasons: Compilers
423 don't need to worry about name clashes with reserved words, and the set of
424 reserved words may be expanded in the future without penalty. Additionally,
425 unnamed identifiers allow a compiler to quickly come up with a temporary
426 variable without having to avoid symbol table conflicts.</p>
428 <p>Reserved words in LLVM are very similar to reserved words in other
429 languages. There are keywords for different opcodes
430 ('<tt><a href="#i_add">add</a></tt>',
431 '<tt><a href="#i_bitcast">bitcast</a></tt>',
432 '<tt><a href="#i_ret">ret</a></tt>', etc...), for primitive type names
433 ('<tt><a href="#t_void">void</a></tt>',
434 '<tt><a href="#t_primitive">i32</a></tt>', etc...), and others. These
435 reserved words cannot conflict with variable names, because none of them
436 start with a prefix character (<tt>'%'</tt> or <tt>'@'</tt>).</p>
438 <p>Here is an example of LLVM code to multiply the integer variable
439 '<tt>%X</tt>' by 8:</p>
443 <pre class="doc_code">
444 %result = <a href="#i_mul">mul</a> i32 %X, 8
447 <p>After strength reduction:</p>
449 <pre class="doc_code">
450 %result = <a href="#i_shl">shl</a> i32 %X, i8 3
453 <p>And the hard way:</p>
455 <pre class="doc_code">
456 %0 = <a href="#i_add">add</a> i32 %X, %X <i>; yields {i32}:%0</i>
457 %1 = <a href="#i_add">add</a> i32 %0, %0 <i>; yields {i32}:%1</i>
458 %result = <a href="#i_add">add</a> i32 %1, %1
461 <p>This last way of multiplying <tt>%X</tt> by 8 illustrates several important
462 lexical features of LLVM:</p>
465 <li>Comments are delimited with a '<tt>;</tt>' and go until the end of
468 <li>Unnamed temporaries are created when the result of a computation is not
469 assigned to a named value.</li>
471 <li>Unnamed temporaries are numbered sequentially</li>
474 <p>It also shows a convention that we follow in this document. When
475 demonstrating instructions, we will follow an instruction with a comment that
476 defines the type and name of value produced. Comments are shown in italic
481 <!-- *********************************************************************** -->
482 <h2><a name="highlevel">High Level Structure</a></h2>
483 <!-- *********************************************************************** -->
485 <!-- ======================================================================= -->
487 <a name="modulestructure">Module Structure</a>
492 <p>LLVM programs are composed of "Module"s, each of which is a translation unit
493 of the input programs. Each module consists of functions, global variables,
494 and symbol table entries. Modules may be combined together with the LLVM
495 linker, which merges function (and global variable) definitions, resolves
496 forward declarations, and merges symbol table entries. Here is an example of
497 the "hello world" module:</p>
499 <pre class="doc_code">
500 <i>; Declare the string constant as a global constant.</i>
501 <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>
503 <i>; External declaration of the puts function</i>
504 <a href="#functionstructure">declare</a> i32 @puts(i8*) <i>; i32 (i8*)* </i>
506 <i>; Definition of main function</i>
507 define i32 @main() { <i>; i32()* </i>
508 <i>; Convert [13 x i8]* to i8 *...</i>
509 %cast210 = <a href="#i_getelementptr">getelementptr</a> [13 x i8]* @.LC0, i64 0, i64 0 <i>; i8*</i>
511 <i>; Call puts function to write out the string to stdout.</i>
512 <a href="#i_call">call</a> i32 @puts(i8* %cast210) <i>; i32</i>
513 <a href="#i_ret">ret</a> i32 0
516 <i>; Named metadata</i>
517 !1 = metadata !{i32 41}
521 <p>This example is made up of a <a href="#globalvars">global variable</a> named
522 "<tt>.LC0</tt>", an external declaration of the "<tt>puts</tt>" function,
523 a <a href="#functionstructure">function definition</a> for
524 "<tt>main</tt>" and <a href="#namedmetadatastructure">named metadata</a>
527 <p>In general, a module is made up of a list of global values, where both
528 functions and global variables are global values. Global values are
529 represented by a pointer to a memory location (in this case, a pointer to an
530 array of char, and a pointer to a function), and have one of the
531 following <a href="#linkage">linkage types</a>.</p>
535 <!-- ======================================================================= -->
537 <a name="linkage">Linkage Types</a>
542 <p>All Global Variables and Functions have one of the following types of
546 <dt><tt><b><a name="linkage_private">private</a></b></tt></dt>
547 <dd>Global values with "<tt>private</tt>" linkage are only directly accessible
548 by objects in the current module. In particular, linking code into a
549 module with an private global value may cause the private to be renamed as
550 necessary to avoid collisions. Because the symbol is private to the
551 module, all references can be updated. This doesn't show up in any symbol
552 table in the object file.</dd>
554 <dt><tt><b><a name="linkage_linker_private">linker_private</a></b></tt></dt>
555 <dd>Similar to <tt>private</tt>, but the symbol is passed through the
556 assembler and evaluated by the linker. Unlike normal strong symbols, they
557 are removed by the linker from the final linked image (executable or
558 dynamic library).</dd>
560 <dt><tt><b><a name="linkage_linker_private_weak">linker_private_weak</a></b></tt></dt>
561 <dd>Similar to "<tt>linker_private</tt>", but the symbol is weak. Note that
562 <tt>linker_private_weak</tt> symbols are subject to coalescing by the
563 linker. The symbols are removed by the linker from the final linked image
564 (executable or dynamic library).</dd>
566 <dt><tt><b><a name="linkage_linker_private_weak_def_auto">linker_private_weak_def_auto</a></b></tt></dt>
567 <dd>Similar to "<tt>linker_private_weak</tt>", but it's known that the address
568 of the object is not taken. For instance, functions that had an inline
569 definition, but the compiler decided not to inline it. Note,
570 unlike <tt>linker_private</tt> and <tt>linker_private_weak</tt>,
571 <tt>linker_private_weak_def_auto</tt> may have only <tt>default</tt>
572 visibility. The symbols are removed by the linker from the final linked
573 image (executable or dynamic library).</dd>
575 <dt><tt><b><a name="linkage_internal">internal</a></b></tt></dt>
576 <dd>Similar to private, but the value shows as a local symbol
577 (<tt>STB_LOCAL</tt> in the case of ELF) in the object file. This
578 corresponds to the notion of the '<tt>static</tt>' keyword in C.</dd>
580 <dt><tt><b><a name="linkage_available_externally">available_externally</a></b></tt></dt>
581 <dd>Globals with "<tt>available_externally</tt>" linkage are never emitted
582 into the object file corresponding to the LLVM module. They exist to
583 allow inlining and other optimizations to take place given knowledge of
584 the definition of the global, which is known to be somewhere outside the
585 module. Globals with <tt>available_externally</tt> linkage are allowed to
586 be discarded at will, and are otherwise the same as <tt>linkonce_odr</tt>.
587 This linkage type is only allowed on definitions, not declarations.</dd>
589 <dt><tt><b><a name="linkage_linkonce">linkonce</a></b></tt></dt>
590 <dd>Globals with "<tt>linkonce</tt>" linkage are merged with other globals of
591 the same name when linkage occurs. This can be used to implement
592 some forms of inline functions, templates, or other code which must be
593 generated in each translation unit that uses it, but where the body may
594 be overridden with a more definitive definition later. Unreferenced
595 <tt>linkonce</tt> globals are allowed to be discarded. Note that
596 <tt>linkonce</tt> linkage does not actually allow the optimizer to
597 inline the body of this function into callers because it doesn't know if
598 this definition of the function is the definitive definition within the
599 program or whether it will be overridden by a stronger definition.
600 To enable inlining and other optimizations, use "<tt>linkonce_odr</tt>"
603 <dt><tt><b><a name="linkage_weak">weak</a></b></tt></dt>
604 <dd>"<tt>weak</tt>" linkage has the same merging semantics as
605 <tt>linkonce</tt> linkage, except that unreferenced globals with
606 <tt>weak</tt> linkage may not be discarded. This is used for globals that
607 are declared "weak" in C source code.</dd>
609 <dt><tt><b><a name="linkage_common">common</a></b></tt></dt>
610 <dd>"<tt>common</tt>" linkage is most similar to "<tt>weak</tt>" linkage, but
611 they are used for tentative definitions in C, such as "<tt>int X;</tt>" at
613 Symbols with "<tt>common</tt>" linkage are merged in the same way as
614 <tt>weak symbols</tt>, and they may not be deleted if unreferenced.
615 <tt>common</tt> symbols may not have an explicit section,
616 must have a zero initializer, and may not be marked '<a
617 href="#globalvars"><tt>constant</tt></a>'. Functions and aliases may not
618 have common linkage.</dd>
621 <dt><tt><b><a name="linkage_appending">appending</a></b></tt></dt>
622 <dd>"<tt>appending</tt>" linkage may only be applied to global variables of
623 pointer to array type. When two global variables with appending linkage
624 are linked together, the two global arrays are appended together. This is
625 the LLVM, typesafe, equivalent of having the system linker append together
626 "sections" with identical names when .o files are linked.</dd>
628 <dt><tt><b><a name="linkage_externweak">extern_weak</a></b></tt></dt>
629 <dd>The semantics of this linkage follow the ELF object file model: the symbol
630 is weak until linked, if not linked, the symbol becomes null instead of
631 being an undefined reference.</dd>
633 <dt><tt><b><a name="linkage_linkonce_odr">linkonce_odr</a></b></tt></dt>
634 <dt><tt><b><a name="linkage_weak_odr">weak_odr</a></b></tt></dt>
635 <dd>Some languages allow differing globals to be merged, such as two functions
636 with different semantics. Other languages, such as <tt>C++</tt>, ensure
637 that only equivalent globals are ever merged (the "one definition rule"
638 — "ODR"). Such languages can use the <tt>linkonce_odr</tt>
639 and <tt>weak_odr</tt> linkage types to indicate that the global will only
640 be merged with equivalent globals. These linkage types are otherwise the
641 same as their non-<tt>odr</tt> versions.</dd>
643 <dt><tt><b><a name="linkage_external">external</a></b></tt></dt>
644 <dd>If none of the above identifiers are used, the global is externally
645 visible, meaning that it participates in linkage and can be used to
646 resolve external symbol references.</dd>
649 <p>The next two types of linkage are targeted for Microsoft Windows platform
650 only. They are designed to support importing (exporting) symbols from (to)
651 DLLs (Dynamic Link Libraries).</p>
654 <dt><tt><b><a name="linkage_dllimport">dllimport</a></b></tt></dt>
655 <dd>"<tt>dllimport</tt>" linkage causes the compiler to reference a function
656 or variable via a global pointer to a pointer that is set up by the DLL
657 exporting the symbol. On Microsoft Windows targets, the pointer name is
658 formed by combining <code>__imp_</code> and the function or variable
661 <dt><tt><b><a name="linkage_dllexport">dllexport</a></b></tt></dt>
662 <dd>"<tt>dllexport</tt>" linkage causes the compiler to provide a global
663 pointer to a pointer in a DLL, so that it can be referenced with the
664 <tt>dllimport</tt> attribute. On Microsoft Windows targets, the pointer
665 name is formed by combining <code>__imp_</code> and the function or
669 <p>For example, since the "<tt>.LC0</tt>" variable is defined to be internal, if
670 another module defined a "<tt>.LC0</tt>" variable and was linked with this
671 one, one of the two would be renamed, preventing a collision. Since
672 "<tt>main</tt>" and "<tt>puts</tt>" are external (i.e., lacking any linkage
673 declarations), they are accessible outside of the current module.</p>
675 <p>It is illegal for a function <i>declaration</i> to have any linkage type
676 other than <tt>external</tt>, <tt>dllimport</tt>
677 or <tt>extern_weak</tt>.</p>
679 <p>Aliases can have only <tt>external</tt>, <tt>internal</tt>, <tt>weak</tt>
680 or <tt>weak_odr</tt> linkages.</p>
684 <!-- ======================================================================= -->
686 <a name="callingconv">Calling Conventions</a>
691 <p>LLVM <a href="#functionstructure">functions</a>, <a href="#i_call">calls</a>
692 and <a href="#i_invoke">invokes</a> can all have an optional calling
693 convention specified for the call. The calling convention of any pair of
694 dynamic caller/callee must match, or the behavior of the program is
695 undefined. The following calling conventions are supported by LLVM, and more
696 may be added in the future:</p>
699 <dt><b>"<tt>ccc</tt>" - The C calling convention</b>:</dt>
700 <dd>This calling convention (the default if no other calling convention is
701 specified) matches the target C calling conventions. This calling
702 convention supports varargs function calls and tolerates some mismatch in
703 the declared prototype and implemented declaration of the function (as
706 <dt><b>"<tt>fastcc</tt>" - The fast calling convention</b>:</dt>
707 <dd>This calling convention attempts to make calls as fast as possible
708 (e.g. by passing things in registers). This calling convention allows the
709 target to use whatever tricks it wants to produce fast code for the
710 target, without having to conform to an externally specified ABI
711 (Application Binary Interface).
712 <a href="CodeGenerator.html#tailcallopt">Tail calls can only be optimized
713 when this or the GHC convention is used.</a> This calling convention
714 does not support varargs and requires the prototype of all callees to
715 exactly match the prototype of the function definition.</dd>
717 <dt><b>"<tt>coldcc</tt>" - The cold calling convention</b>:</dt>
718 <dd>This calling convention attempts to make code in the caller as efficient
719 as possible under the assumption that the call is not commonly executed.
720 As such, these calls often preserve all registers so that the call does
721 not break any live ranges in the caller side. This calling convention
722 does not support varargs and requires the prototype of all callees to
723 exactly match the prototype of the function definition.</dd>
725 <dt><b>"<tt>cc <em>10</em></tt>" - GHC convention</b>:</dt>
726 <dd>This calling convention has been implemented specifically for use by the
727 <a href="http://www.haskell.org/ghc">Glasgow Haskell Compiler (GHC)</a>.
728 It passes everything in registers, going to extremes to achieve this by
729 disabling callee save registers. This calling convention should not be
730 used lightly but only for specific situations such as an alternative to
731 the <em>register pinning</em> performance technique often used when
732 implementing functional programming languages.At the moment only X86
733 supports this convention and it has the following limitations:
735 <li>On <em>X86-32</em> only supports up to 4 bit type parameters. No
736 floating point types are supported.</li>
737 <li>On <em>X86-64</em> only supports up to 10 bit type parameters and
738 6 floating point parameters.</li>
740 This calling convention supports
741 <a href="CodeGenerator.html#tailcallopt">tail call optimization</a> but
742 requires both the caller and callee are using it.
745 <dt><b>"<tt>cc <<em>n</em>></tt>" - Numbered convention</b>:</dt>
746 <dd>Any calling convention may be specified by number, allowing
747 target-specific calling conventions to be used. Target specific calling
748 conventions start at 64.</dd>
751 <p>More calling conventions can be added/defined on an as-needed basis, to
752 support Pascal conventions or any other well-known target-independent
757 <!-- ======================================================================= -->
759 <a name="visibility">Visibility Styles</a>
764 <p>All Global Variables and Functions have one of the following visibility
768 <dt><b>"<tt>default</tt>" - Default style</b>:</dt>
769 <dd>On targets that use the ELF object file format, default visibility means
770 that the declaration is visible to other modules and, in shared libraries,
771 means that the declared entity may be overridden. On Darwin, default
772 visibility means that the declaration is visible to other modules. Default
773 visibility corresponds to "external linkage" in the language.</dd>
775 <dt><b>"<tt>hidden</tt>" - Hidden style</b>:</dt>
776 <dd>Two declarations of an object with hidden visibility refer to the same
777 object if they are in the same shared object. Usually, hidden visibility
778 indicates that the symbol will not be placed into the dynamic symbol
779 table, so no other module (executable or shared library) can reference it
782 <dt><b>"<tt>protected</tt>" - Protected style</b>:</dt>
783 <dd>On ELF, protected visibility indicates that the symbol will be placed in
784 the dynamic symbol table, but that references within the defining module
785 will bind to the local symbol. That is, the symbol cannot be overridden by
791 <!-- ======================================================================= -->
793 <a name="namedtypes">Named Types</a>
798 <p>LLVM IR allows you to specify name aliases for certain types. This can make
799 it easier to read the IR and make the IR more condensed (particularly when
800 recursive types are involved). An example of a name specification is:</p>
802 <pre class="doc_code">
803 %mytype = type { %mytype*, i32 }
806 <p>You may give a name to any <a href="#typesystem">type</a> except
807 "<a href="#t_void">void</a>". Type name aliases may be used anywhere a type
808 is expected with the syntax "%mytype".</p>
810 <p>Note that type names are aliases for the structural type that they indicate,
811 and that you can therefore specify multiple names for the same type. This
812 often leads to confusing behavior when dumping out a .ll file. Since LLVM IR
813 uses structural typing, the name is not part of the type. When printing out
814 LLVM IR, the printer will pick <em>one name</em> to render all types of a
815 particular shape. This means that if you have code where two different
816 source types end up having the same LLVM type, that the dumper will sometimes
817 print the "wrong" or unexpected type. This is an important design point and
818 isn't going to change.</p>
822 <!-- ======================================================================= -->
824 <a name="globalvars">Global Variables</a>
829 <p>Global variables define regions of memory allocated at compilation time
830 instead of run-time. Global variables may optionally be initialized, may
831 have an explicit section to be placed in, and may have an optional explicit
832 alignment specified. A variable may be defined as "thread_local", which
833 means that it will not be shared by threads (each thread will have a
834 separated copy of the variable). A variable may be defined as a global
835 "constant," which indicates that the contents of the variable
836 will <b>never</b> be modified (enabling better optimization, allowing the
837 global data to be placed in the read-only section of an executable, etc).
838 Note that variables that need runtime initialization cannot be marked
839 "constant" as there is a store to the variable.</p>
841 <p>LLVM explicitly allows <em>declarations</em> of global variables to be marked
842 constant, even if the final definition of the global is not. This capability
843 can be used to enable slightly better optimization of the program, but
844 requires the language definition to guarantee that optimizations based on the
845 'constantness' are valid for the translation units that do not include the
848 <p>As SSA values, global variables define pointer values that are in scope
849 (i.e. they dominate) all basic blocks in the program. Global variables
850 always define a pointer to their "content" type because they describe a
851 region of memory, and all memory objects in LLVM are accessed through
854 <p>Global variables can be marked with <tt>unnamed_addr</tt> which indicates
855 that the address is not significant, only the content. Constants marked
856 like this can be merged with other constants if they have the same
857 initializer. Note that a constant with significant address <em>can</em>
858 be merged with a <tt>unnamed_addr</tt> constant, the result being a
859 constant whose address is significant.</p>
861 <p>A global variable may be declared to reside in a target-specific numbered
862 address space. For targets that support them, address spaces may affect how
863 optimizations are performed and/or what target instructions are used to
864 access the variable. The default address space is zero. The address space
865 qualifier must precede any other attributes.</p>
867 <p>LLVM allows an explicit section to be specified for globals. If the target
868 supports it, it will emit globals to the section specified.</p>
870 <p>An explicit alignment may be specified for a global, which must be a power
871 of 2. If not present, or if the alignment is set to zero, the alignment of
872 the global is set by the target to whatever it feels convenient. If an
873 explicit alignment is specified, the global is forced to have exactly that
874 alignment. Targets and optimizers are not allowed to over-align the global
875 if the global has an assigned section. In this case, the extra alignment
876 could be observable: for example, code could assume that the globals are
877 densely packed in their section and try to iterate over them as an array,
878 alignment padding would break this iteration.</p>
880 <p>For example, the following defines a global in a numbered address space with
881 an initializer, section, and alignment:</p>
883 <pre class="doc_code">
884 @G = addrspace(5) constant float 1.0, section "foo", align 4
890 <!-- ======================================================================= -->
892 <a name="functionstructure">Functions</a>
897 <p>LLVM function definitions consist of the "<tt>define</tt>" keyword, an
898 optional <a href="#linkage">linkage type</a>, an optional
899 <a href="#visibility">visibility style</a>, an optional
900 <a href="#callingconv">calling convention</a>,
901 an optional <tt>unnamed_addr</tt> attribute, a return type, an optional
902 <a href="#paramattrs">parameter attribute</a> for the return type, a function
903 name, a (possibly empty) argument list (each with optional
904 <a href="#paramattrs">parameter attributes</a>), optional
905 <a href="#fnattrs">function attributes</a>, an optional section, an optional
906 alignment, an optional <a href="#gc">garbage collector name</a>, an opening
907 curly brace, a list of basic blocks, and a closing curly brace.</p>
909 <p>LLVM function declarations consist of the "<tt>declare</tt>" keyword, an
910 optional <a href="#linkage">linkage type</a>, an optional
911 <a href="#visibility">visibility style</a>, an optional
912 <a href="#callingconv">calling convention</a>,
913 an optional <tt>unnamed_addr</tt> attribute, a return type, an optional
914 <a href="#paramattrs">parameter attribute</a> for the return type, a function
915 name, a possibly empty list of arguments, an optional alignment, and an
916 optional <a href="#gc">garbage collector name</a>.</p>
918 <p>A function definition contains a list of basic blocks, forming the CFG
919 (Control Flow Graph) for the function. Each basic block may optionally start
920 with a label (giving the basic block a symbol table entry), contains a list
921 of instructions, and ends with a <a href="#terminators">terminator</a>
922 instruction (such as a branch or function return).</p>
924 <p>The first basic block in a function is special in two ways: it is immediately
925 executed on entrance to the function, and it is not allowed to have
926 predecessor basic blocks (i.e. there can not be any branches to the entry
927 block of a function). Because the block can have no predecessors, it also
928 cannot have any <a href="#i_phi">PHI nodes</a>.</p>
930 <p>LLVM allows an explicit section to be specified for functions. If the target
931 supports it, it will emit functions to the section specified.</p>
933 <p>An explicit alignment may be specified for a function. If not present, or if
934 the alignment is set to zero, the alignment of the function is set by the
935 target to whatever it feels convenient. If an explicit alignment is
936 specified, the function is forced to have at least that much alignment. All
937 alignments must be a power of 2.</p>
939 <p>If the <tt>unnamed_addr</tt> attribute is given, the address is know to not
940 be significant and two identical functions can be merged.</p>
943 <pre class="doc_code">
944 define [<a href="#linkage">linkage</a>] [<a href="#visibility">visibility</a>]
945 [<a href="#callingconv">cconv</a>] [<a href="#paramattrs">ret attrs</a>]
946 <ResultType> @<FunctionName> ([argument list])
947 [<a href="#fnattrs">fn Attrs</a>] [section "name"] [align N]
948 [<a href="#gc">gc</a>] { ... }
953 <!-- ======================================================================= -->
955 <a name="aliasstructure">Aliases</a>
960 <p>Aliases act as "second name" for the aliasee value (which can be either
961 function, global variable, another alias or bitcast of global value). Aliases
962 may have an optional <a href="#linkage">linkage type</a>, and an
963 optional <a href="#visibility">visibility style</a>.</p>
966 <pre class="doc_code">
967 @<Name> = alias [Linkage] [Visibility] <AliaseeTy> @<Aliasee>
972 <!-- ======================================================================= -->
974 <a name="namedmetadatastructure">Named Metadata</a>
979 <p>Named metadata is a collection of metadata. <a href="#metadata">Metadata
980 nodes</a> (but not metadata strings) are the only valid operands for
981 a named metadata.</p>
984 <pre class="doc_code">
985 ; Some unnamed metadata nodes, which are referenced by the named metadata.
986 !0 = metadata !{metadata !"zero"}
987 !1 = metadata !{metadata !"one"}
988 !2 = metadata !{metadata !"two"}
990 !name = !{!0, !1, !2}
995 <!-- ======================================================================= -->
997 <a name="paramattrs">Parameter Attributes</a>
1002 <p>The return type and each parameter of a function type may have a set of
1003 <i>parameter attributes</i> associated with them. Parameter attributes are
1004 used to communicate additional information about the result or parameters of
1005 a function. Parameter attributes are considered to be part of the function,
1006 not of the function type, so functions with different parameter attributes
1007 can have the same function type.</p>
1009 <p>Parameter attributes are simple keywords that follow the type specified. If
1010 multiple parameter attributes are needed, they are space separated. For
1013 <pre class="doc_code">
1014 declare i32 @printf(i8* noalias nocapture, ...)
1015 declare i32 @atoi(i8 zeroext)
1016 declare signext i8 @returns_signed_char()
1019 <p>Note that any attributes for the function result (<tt>nounwind</tt>,
1020 <tt>readonly</tt>) come immediately after the argument list.</p>
1022 <p>Currently, only the following parameter attributes are defined:</p>
1025 <dt><tt><b>zeroext</b></tt></dt>
1026 <dd>This indicates to the code generator that the parameter or return value
1027 should be zero-extended to the extent required by the target's ABI (which
1028 is usually 32-bits, but is 8-bits for a i1 on x86-64) by the caller (for a
1029 parameter) or the callee (for a return value).</dd>
1031 <dt><tt><b>signext</b></tt></dt>
1032 <dd>This indicates to the code generator that the parameter or return value
1033 should be sign-extended to the extent required by the target's ABI (which
1034 is usually 32-bits) by the caller (for a parameter) or the callee (for a
1037 <dt><tt><b>inreg</b></tt></dt>
1038 <dd>This indicates that this parameter or return value should be treated in a
1039 special target-dependent fashion during while emitting code for a function
1040 call or return (usually, by putting it in a register as opposed to memory,
1041 though some targets use it to distinguish between two different kinds of
1042 registers). Use of this attribute is target-specific.</dd>
1044 <dt><tt><b><a name="byval">byval</a></b></tt></dt>
1045 <dd><p>This indicates that the pointer parameter should really be passed by
1046 value to the function. The attribute implies that a hidden copy of the
1048 is made between the caller and the callee, so the callee is unable to
1049 modify the value in the callee. This attribute is only valid on LLVM
1050 pointer arguments. It is generally used to pass structs and arrays by
1051 value, but is also valid on pointers to scalars. The copy is considered
1052 to belong to the caller not the callee (for example,
1053 <tt><a href="#readonly">readonly</a></tt> functions should not write to
1054 <tt>byval</tt> parameters). This is not a valid attribute for return
1057 <p>The byval attribute also supports specifying an alignment with
1058 the align attribute. It indicates the alignment of the stack slot to
1059 form and the known alignment of the pointer specified to the call site. If
1060 the alignment is not specified, then the code generator makes a
1061 target-specific assumption.</p></dd>
1063 <dt><tt><b><a name="sret">sret</a></b></tt></dt>
1064 <dd>This indicates that the pointer parameter specifies the address of a
1065 structure that is the return value of the function in the source program.
1066 This pointer must be guaranteed by the caller to be valid: loads and
1067 stores to the structure may be assumed by the callee to not to trap. This
1068 may only be applied to the first parameter. This is not a valid attribute
1069 for return values. </dd>
1071 <dt><tt><b><a name="noalias">noalias</a></b></tt></dt>
1072 <dd>This indicates that pointer values
1073 <a href="#pointeraliasing"><i>based</i></a> on the argument or return
1074 value do not alias pointer values which are not <i>based</i> on it,
1075 ignoring certain "irrelevant" dependencies.
1076 For a call to the parent function, dependencies between memory
1077 references from before or after the call and from those during the call
1078 are "irrelevant" to the <tt>noalias</tt> keyword for the arguments and
1079 return value used in that call.
1080 The caller shares the responsibility with the callee for ensuring that
1081 these requirements are met.
1082 For further details, please see the discussion of the NoAlias response in
1083 <a href="AliasAnalysis.html#MustMayNo">alias analysis</a>.<br>
1085 Note that this definition of <tt>noalias</tt> is intentionally
1086 similar to the definition of <tt>restrict</tt> in C99 for function
1087 arguments, though it is slightly weaker.
1089 For function return values, C99's <tt>restrict</tt> is not meaningful,
1090 while LLVM's <tt>noalias</tt> is.
1093 <dt><tt><b><a name="nocapture">nocapture</a></b></tt></dt>
1094 <dd>This indicates that the callee does not make any copies of the pointer
1095 that outlive the callee itself. This is not a valid attribute for return
1098 <dt><tt><b><a name="nest">nest</a></b></tt></dt>
1099 <dd>This indicates that the pointer parameter can be excised using the
1100 <a href="#int_trampoline">trampoline intrinsics</a>. This is not a valid
1101 attribute for return values.</dd>
1106 <!-- ======================================================================= -->
1108 <a name="gc">Garbage Collector Names</a>
1113 <p>Each function may specify a garbage collector name, which is simply a
1116 <pre class="doc_code">
1117 define void @f() gc "name" { ... }
1120 <p>The compiler declares the supported values of <i>name</i>. Specifying a
1121 collector which will cause the compiler to alter its output in order to
1122 support the named garbage collection algorithm.</p>
1126 <!-- ======================================================================= -->
1128 <a name="fnattrs">Function Attributes</a>
1133 <p>Function attributes are set to communicate additional information about a
1134 function. Function attributes are considered to be part of the function, not
1135 of the function type, so functions with different parameter attributes can
1136 have the same function type.</p>
1138 <p>Function attributes are simple keywords that follow the type specified. If
1139 multiple attributes are needed, they are space separated. For example:</p>
1141 <pre class="doc_code">
1142 define void @f() noinline { ... }
1143 define void @f() alwaysinline { ... }
1144 define void @f() alwaysinline optsize { ... }
1145 define void @f() optsize { ... }
1149 <dt><tt><b>address_safety</b></tt></dt>
1150 <dd>This attribute indicates that the address safety analysis
1151 is enabled for this function. </dd>
1153 <dt><tt><b>alignstack(<<em>n</em>>)</b></tt></dt>
1154 <dd>This attribute indicates that, when emitting the prologue and epilogue,
1155 the backend should forcibly align the stack pointer. Specify the
1156 desired alignment, which must be a power of two, in parentheses.
1158 <dt><tt><b>alwaysinline</b></tt></dt>
1159 <dd>This attribute indicates that the inliner should attempt to inline this
1160 function into callers whenever possible, ignoring any active inlining size
1161 threshold for this caller.</dd>
1163 <dt><tt><b>nonlazybind</b></tt></dt>
1164 <dd>This attribute suppresses lazy symbol binding for the function. This
1165 may make calls to the function faster, at the cost of extra program
1166 startup time if the function is not called during program startup.</dd>
1168 <dt><tt><b>inlinehint</b></tt></dt>
1169 <dd>This attribute indicates that the source code contained a hint that inlining
1170 this function is desirable (such as the "inline" keyword in C/C++). It
1171 is just a hint; it imposes no requirements on the inliner.</dd>
1173 <dt><tt><b>naked</b></tt></dt>
1174 <dd>This attribute disables prologue / epilogue emission for the function.
1175 This can have very system-specific consequences.</dd>
1177 <dt><tt><b>noimplicitfloat</b></tt></dt>
1178 <dd>This attributes disables implicit floating point instructions.</dd>
1180 <dt><tt><b>noinline</b></tt></dt>
1181 <dd>This attribute indicates that the inliner should never inline this
1182 function in any situation. This attribute may not be used together with
1183 the <tt>alwaysinline</tt> attribute.</dd>
1185 <dt><tt><b>noredzone</b></tt></dt>
1186 <dd>This attribute indicates that the code generator should not use a red
1187 zone, even if the target-specific ABI normally permits it.</dd>
1189 <dt><tt><b>noreturn</b></tt></dt>
1190 <dd>This function attribute indicates that the function never returns
1191 normally. This produces undefined behavior at runtime if the function
1192 ever does dynamically return.</dd>
1194 <dt><tt><b>nounwind</b></tt></dt>
1195 <dd>This function attribute indicates that the function never returns with an
1196 unwind or exceptional control flow. If the function does unwind, its
1197 runtime behavior is undefined.</dd>
1199 <dt><tt><b>optsize</b></tt></dt>
1200 <dd>This attribute suggests that optimization passes and code generator passes
1201 make choices that keep the code size of this function low, and otherwise
1202 do optimizations specifically to reduce code size.</dd>
1204 <dt><tt><b>readnone</b></tt></dt>
1205 <dd>This attribute indicates that the function computes its result (or decides
1206 to unwind an exception) based strictly on its arguments, without
1207 dereferencing any pointer arguments or otherwise accessing any mutable
1208 state (e.g. memory, control registers, etc) visible to caller functions.
1209 It does not write through any pointer arguments
1210 (including <tt><a href="#byval">byval</a></tt> arguments) and never
1211 changes any state visible to callers. This means that it cannot unwind
1212 exceptions by calling the <tt>C++</tt> exception throwing methods.</dd>
1214 <dt><tt><b><a name="readonly">readonly</a></b></tt></dt>
1215 <dd>This attribute indicates that the function does not write through any
1216 pointer arguments (including <tt><a href="#byval">byval</a></tt>
1217 arguments) or otherwise modify any state (e.g. memory, control registers,
1218 etc) visible to caller functions. It may dereference pointer arguments
1219 and read state that may be set in the caller. A readonly function always
1220 returns the same value (or unwinds an exception identically) when called
1221 with the same set of arguments and global state. It cannot unwind an
1222 exception by calling the <tt>C++</tt> exception throwing methods.</dd>
1224 <dt><tt><b><a name="returns_twice">returns_twice</a></b></tt></dt>
1225 <dd>This attribute indicates that this function can return twice. The
1226 C <code>setjmp</code> is an example of such a function. The compiler
1227 disables some optimizations (like tail calls) in the caller of these
1230 <dt><tt><b><a name="ssp">ssp</a></b></tt></dt>
1231 <dd>This attribute indicates that the function should emit a stack smashing
1232 protector. It is in the form of a "canary"—a random value placed on
1233 the stack before the local variables that's checked upon return from the
1234 function to see if it has been overwritten. A heuristic is used to
1235 determine if a function needs stack protectors or not.<br>
1237 If a function that has an <tt>ssp</tt> attribute is inlined into a
1238 function that doesn't have an <tt>ssp</tt> attribute, then the resulting
1239 function will have an <tt>ssp</tt> attribute.</dd>
1241 <dt><tt><b>sspreq</b></tt></dt>
1242 <dd>This attribute indicates that the function should <em>always</em> emit a
1243 stack smashing protector. This overrides
1244 the <tt><a href="#ssp">ssp</a></tt> function attribute.<br>
1246 If a function that has an <tt>sspreq</tt> attribute is inlined into a
1247 function that doesn't have an <tt>sspreq</tt> attribute or which has
1248 an <tt>ssp</tt> attribute, then the resulting function will have
1249 an <tt>sspreq</tt> attribute.</dd>
1251 <dt><tt><b><a name="uwtable">uwtable</a></b></tt></dt>
1252 <dd>This attribute indicates that the ABI being targeted requires that
1253 an unwind table entry be produce for this function even if we can
1254 show that no exceptions passes by it. This is normally the case for
1255 the ELF x86-64 abi, but it can be disabled for some compilation
1261 <!-- ======================================================================= -->
1263 <a name="moduleasm">Module-Level Inline Assembly</a>
1268 <p>Modules may contain "module-level inline asm" blocks, which corresponds to
1269 the GCC "file scope inline asm" blocks. These blocks are internally
1270 concatenated by LLVM and treated as a single unit, but may be separated in
1271 the <tt>.ll</tt> file if desired. The syntax is very simple:</p>
1273 <pre class="doc_code">
1274 module asm "inline asm code goes here"
1275 module asm "more can go here"
1278 <p>The strings can contain any character by escaping non-printable characters.
1279 The escape sequence used is simply "\xx" where "xx" is the two digit hex code
1282 <p>The inline asm code is simply printed to the machine code .s file when
1283 assembly code is generated.</p>
1287 <!-- ======================================================================= -->
1289 <a name="datalayout">Data Layout</a>
1294 <p>A module may specify a target specific data layout string that specifies how
1295 data is to be laid out in memory. The syntax for the data layout is
1298 <pre class="doc_code">
1299 target datalayout = "<i>layout specification</i>"
1302 <p>The <i>layout specification</i> consists of a list of specifications
1303 separated by the minus sign character ('-'). Each specification starts with
1304 a letter and may include other information after the letter to define some
1305 aspect of the data layout. The specifications accepted are as follows:</p>
1309 <dd>Specifies that the target lays out data in big-endian form. That is, the
1310 bits with the most significance have the lowest address location.</dd>
1313 <dd>Specifies that the target lays out data in little-endian form. That is,
1314 the bits with the least significance have the lowest address
1317 <dt><tt>S<i>size</i></tt></dt>
1318 <dd>Specifies the natural alignment of the stack in bits. Alignment promotion
1319 of stack variables is limited to the natural stack alignment to avoid
1320 dynamic stack realignment. The stack alignment must be a multiple of
1321 8-bits. If omitted, the natural stack alignment defaults to "unspecified",
1322 which does not prevent any alignment promotions.</dd>
1324 <dt><tt>p:<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1325 <dd>This specifies the <i>size</i> of a pointer and its <i>abi</i> and
1326 <i>preferred</i> alignments. All sizes are in bits. Specifying
1327 the <i>pref</i> alignment is optional. If omitted, the
1328 preceding <tt>:</tt> should be omitted too.</dd>
1330 <dt><tt>i<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1331 <dd>This specifies the alignment for an integer type of a given bit
1332 <i>size</i>. The value of <i>size</i> must be in the range [1,2^23).</dd>
1334 <dt><tt>v<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1335 <dd>This specifies the alignment for a vector type of a given bit
1338 <dt><tt>f<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1339 <dd>This specifies the alignment for a floating point type of a given bit
1340 <i>size</i>. Only values of <i>size</i> that are supported by the target
1341 will work. 32 (float) and 64 (double) are supported on all targets;
1342 80 or 128 (different flavors of long double) are also supported on some
1345 <dt><tt>a<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1346 <dd>This specifies the alignment for an aggregate type of a given bit
1349 <dt><tt>s<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1350 <dd>This specifies the alignment for a stack object of a given bit
1353 <dt><tt>n<i>size1</i>:<i>size2</i>:<i>size3</i>...</tt></dt>
1354 <dd>This specifies a set of native integer widths for the target CPU
1355 in bits. For example, it might contain "n32" for 32-bit PowerPC,
1356 "n32:64" for PowerPC 64, or "n8:16:32:64" for X86-64. Elements of
1357 this set are considered to support most general arithmetic
1358 operations efficiently.</dd>
1361 <p>When constructing the data layout for a given target, LLVM starts with a
1362 default set of specifications which are then (possibly) overridden by the
1363 specifications in the <tt>datalayout</tt> keyword. The default specifications
1364 are given in this list:</p>
1367 <li><tt>E</tt> - big endian</li>
1368 <li><tt>p:64:64:64</tt> - 64-bit pointers with 64-bit alignment</li>
1369 <li><tt>i1:8:8</tt> - i1 is 8-bit (byte) aligned</li>
1370 <li><tt>i8:8:8</tt> - i8 is 8-bit (byte) aligned</li>
1371 <li><tt>i16:16:16</tt> - i16 is 16-bit aligned</li>
1372 <li><tt>i32:32:32</tt> - i32 is 32-bit aligned</li>
1373 <li><tt>i64:32:64</tt> - i64 has ABI alignment of 32-bits but preferred
1374 alignment of 64-bits</li>
1375 <li><tt>f32:32:32</tt> - float is 32-bit aligned</li>
1376 <li><tt>f64:64:64</tt> - double is 64-bit aligned</li>
1377 <li><tt>v64:64:64</tt> - 64-bit vector is 64-bit aligned</li>
1378 <li><tt>v128:128:128</tt> - 128-bit vector is 128-bit aligned</li>
1379 <li><tt>a0:0:1</tt> - aggregates are 8-bit aligned</li>
1380 <li><tt>s0:64:64</tt> - stack objects are 64-bit aligned</li>
1383 <p>When LLVM is determining the alignment for a given type, it uses the
1384 following rules:</p>
1387 <li>If the type sought is an exact match for one of the specifications, that
1388 specification is used.</li>
1390 <li>If no match is found, and the type sought is an integer type, then the
1391 smallest integer type that is larger than the bitwidth of the sought type
1392 is used. If none of the specifications are larger than the bitwidth then
1393 the the largest integer type is used. For example, given the default
1394 specifications above, the i7 type will use the alignment of i8 (next
1395 largest) while both i65 and i256 will use the alignment of i64 (largest
1398 <li>If no match is found, and the type sought is a vector type, then the
1399 largest vector type that is smaller than the sought vector type will be
1400 used as a fall back. This happens because <128 x double> can be
1401 implemented in terms of 64 <2 x double>, for example.</li>
1404 <p>The function of the data layout string may not be what you expect. Notably,
1405 this is not a specification from the frontend of what alignment the code
1406 generator should use.</p>
1408 <p>Instead, if specified, the target data layout is required to match what the
1409 ultimate <em>code generator</em> expects. This string is used by the
1410 mid-level optimizers to
1411 improve code, and this only works if it matches what the ultimate code
1412 generator uses. If you would like to generate IR that does not embed this
1413 target-specific detail into the IR, then you don't have to specify the
1414 string. This will disable some optimizations that require precise layout
1415 information, but this also prevents those optimizations from introducing
1416 target specificity into the IR.</p>
1422 <!-- ======================================================================= -->
1424 <a name="pointeraliasing">Pointer Aliasing Rules</a>
1429 <p>Any memory access must be done through a pointer value associated
1430 with an address range of the memory access, otherwise the behavior
1431 is undefined. Pointer values are associated with address ranges
1432 according to the following rules:</p>
1435 <li>A pointer value is associated with the addresses associated with
1436 any value it is <i>based</i> on.
1437 <li>An address of a global variable is associated with the address
1438 range of the variable's storage.</li>
1439 <li>The result value of an allocation instruction is associated with
1440 the address range of the allocated storage.</li>
1441 <li>A null pointer in the default address-space is associated with
1443 <li>An integer constant other than zero or a pointer value returned
1444 from a function not defined within LLVM may be associated with address
1445 ranges allocated through mechanisms other than those provided by
1446 LLVM. Such ranges shall not overlap with any ranges of addresses
1447 allocated by mechanisms provided by LLVM.</li>
1450 <p>A pointer value is <i>based</i> on another pointer value according
1451 to the following rules:</p>
1454 <li>A pointer value formed from a
1455 <tt><a href="#i_getelementptr">getelementptr</a></tt> operation
1456 is <i>based</i> on the first operand of the <tt>getelementptr</tt>.</li>
1457 <li>The result value of a
1458 <tt><a href="#i_bitcast">bitcast</a></tt> is <i>based</i> on the operand
1459 of the <tt>bitcast</tt>.</li>
1460 <li>A pointer value formed by an
1461 <tt><a href="#i_inttoptr">inttoptr</a></tt> is <i>based</i> on all
1462 pointer values that contribute (directly or indirectly) to the
1463 computation of the pointer's value.</li>
1464 <li>The "<i>based</i> on" relationship is transitive.</li>
1467 <p>Note that this definition of <i>"based"</i> is intentionally
1468 similar to the definition of <i>"based"</i> in C99, though it is
1469 slightly weaker.</p>
1471 <p>LLVM IR does not associate types with memory. The result type of a
1472 <tt><a href="#i_load">load</a></tt> merely indicates the size and
1473 alignment of the memory from which to load, as well as the
1474 interpretation of the value. The first operand type of a
1475 <tt><a href="#i_store">store</a></tt> similarly only indicates the size
1476 and alignment of the store.</p>
1478 <p>Consequently, type-based alias analysis, aka TBAA, aka
1479 <tt>-fstrict-aliasing</tt>, is not applicable to general unadorned
1480 LLVM IR. <a href="#metadata">Metadata</a> may be used to encode
1481 additional information which specialized optimization passes may use
1482 to implement type-based alias analysis.</p>
1486 <!-- ======================================================================= -->
1488 <a name="volatile">Volatile Memory Accesses</a>
1493 <p>Certain memory accesses, such as <a href="#i_load"><tt>load</tt></a>s, <a
1494 href="#i_store"><tt>store</tt></a>s, and <a
1495 href="#int_memcpy"><tt>llvm.memcpy</tt></a>s may be marked <tt>volatile</tt>.
1496 The optimizers must not change the number of volatile operations or change their
1497 order of execution relative to other volatile operations. The optimizers
1498 <i>may</i> change the order of volatile operations relative to non-volatile
1499 operations. This is not Java's "volatile" and has no cross-thread
1500 synchronization behavior.</p>
1504 <!-- ======================================================================= -->
1506 <a name="memmodel">Memory Model for Concurrent Operations</a>
1511 <p>The LLVM IR does not define any way to start parallel threads of execution
1512 or to register signal handlers. Nonetheless, there are platform-specific
1513 ways to create them, and we define LLVM IR's behavior in their presence. This
1514 model is inspired by the C++0x memory model.</p>
1516 <p>For a more informal introduction to this model, see the
1517 <a href="Atomics.html">LLVM Atomic Instructions and Concurrency Guide</a>.
1519 <p>We define a <i>happens-before</i> partial order as the least partial order
1522 <li>Is a superset of single-thread program order, and</li>
1523 <li>When a <i>synchronizes-with</i> <tt>b</tt>, includes an edge from
1524 <tt>a</tt> to <tt>b</tt>. <i>Synchronizes-with</i> pairs are introduced
1525 by platform-specific techniques, like pthread locks, thread
1526 creation, thread joining, etc., and by atomic instructions.
1527 (See also <a href="#ordering">Atomic Memory Ordering Constraints</a>).
1531 <p>Note that program order does not introduce <i>happens-before</i> edges
1532 between a thread and signals executing inside that thread.</p>
1534 <p>Every (defined) read operation (load instructions, memcpy, atomic
1535 loads/read-modify-writes, etc.) <var>R</var> reads a series of bytes written by
1536 (defined) write operations (store instructions, atomic
1537 stores/read-modify-writes, memcpy, etc.). For the purposes of this section,
1538 initialized globals are considered to have a write of the initializer which is
1539 atomic and happens before any other read or write of the memory in question.
1540 For each byte of a read <var>R</var>, <var>R<sub>byte</sub></var> may see
1541 any write to the same byte, except:</p>
1544 <li>If <var>write<sub>1</sub></var> happens before
1545 <var>write<sub>2</sub></var>, and <var>write<sub>2</sub></var> happens
1546 before <var>R<sub>byte</sub></var>, then <var>R<sub>byte</sub></var>
1547 does not see <var>write<sub>1</sub></var>.
1548 <li>If <var>R<sub>byte</sub></var> happens before
1549 <var>write<sub>3</sub></var>, then <var>R<sub>byte</sub></var> does not
1550 see <var>write<sub>3</sub></var>.
1553 <p>Given that definition, <var>R<sub>byte</sub></var> is defined as follows:
1555 <li>If <var>R</var> is volatile, the result is target-dependent. (Volatile
1556 is supposed to give guarantees which can support
1557 <code>sig_atomic_t</code> in C/C++, and may be used for accesses to
1558 addresses which do not behave like normal memory. It does not generally
1559 provide cross-thread synchronization.)
1560 <li>Otherwise, if there is no write to the same byte that happens before
1561 <var>R<sub>byte</sub></var>, <var>R<sub>byte</sub></var> returns
1562 <tt>undef</tt> for that byte.
1563 <li>Otherwise, if <var>R<sub>byte</sub></var> may see exactly one write,
1564 <var>R<sub>byte</sub></var> returns the value written by that
1566 <li>Otherwise, if <var>R</var> is atomic, and all the writes
1567 <var>R<sub>byte</sub></var> may see are atomic, it chooses one of the
1568 values written. See the <a href="#ordering">Atomic Memory Ordering
1569 Constraints</a> section for additional constraints on how the choice
1571 <li>Otherwise <var>R<sub>byte</sub></var> returns <tt>undef</tt>.</li>
1574 <p><var>R</var> returns the value composed of the series of bytes it read.
1575 This implies that some bytes within the value may be <tt>undef</tt>
1576 <b>without</b> the entire value being <tt>undef</tt>. Note that this only
1577 defines the semantics of the operation; it doesn't mean that targets will
1578 emit more than one instruction to read the series of bytes.</p>
1580 <p>Note that in cases where none of the atomic intrinsics are used, this model
1581 places only one restriction on IR transformations on top of what is required
1582 for single-threaded execution: introducing a store to a byte which might not
1583 otherwise be stored is not allowed in general. (Specifically, in the case
1584 where another thread might write to and read from an address, introducing a
1585 store can change a load that may see exactly one write into a load that may
1586 see multiple writes.)</p>
1588 <!-- FIXME: This model assumes all targets where concurrency is relevant have
1589 a byte-size store which doesn't affect adjacent bytes. As far as I can tell,
1590 none of the backends currently in the tree fall into this category; however,
1591 there might be targets which care. If there are, we want a paragraph
1594 Targets may specify that stores narrower than a certain width are not
1595 available; on such a target, for the purposes of this model, treat any
1596 non-atomic write with an alignment or width less than the minimum width
1597 as if it writes to the relevant surrounding bytes.
1602 <!-- ======================================================================= -->
1604 <a name="ordering">Atomic Memory Ordering Constraints</a>
1609 <p>Atomic instructions (<a href="#i_cmpxchg"><code>cmpxchg</code></a>,
1610 <a href="#i_atomicrmw"><code>atomicrmw</code></a>,
1611 <a href="#i_fence"><code>fence</code></a>,
1612 <a href="#i_load"><code>atomic load</code></a>, and
1613 <a href="#i_store"><code>atomic store</code></a>) take an ordering parameter
1614 that determines which other atomic instructions on the same address they
1615 <i>synchronize with</i>. These semantics are borrowed from Java and C++0x,
1616 but are somewhat more colloquial. If these descriptions aren't precise enough,
1617 check those specs (see spec references in the
1618 <a href="Atomics.html#introduction">atomics guide</a>).
1619 <a href="#i_fence"><code>fence</code></a> instructions
1620 treat these orderings somewhat differently since they don't take an address.
1621 See that instruction's documentation for details.</p>
1623 <p>For a simpler introduction to the ordering constraints, see the
1624 <a href="Atomics.html">LLVM Atomic Instructions and Concurrency Guide</a>.</p>
1627 <dt><code>unordered</code></dt>
1628 <dd>The set of values that can be read is governed by the happens-before
1629 partial order. A value cannot be read unless some operation wrote it.
1630 This is intended to provide a guarantee strong enough to model Java's
1631 non-volatile shared variables. This ordering cannot be specified for
1632 read-modify-write operations; it is not strong enough to make them atomic
1633 in any interesting way.</dd>
1634 <dt><code>monotonic</code></dt>
1635 <dd>In addition to the guarantees of <code>unordered</code>, there is a single
1636 total order for modifications by <code>monotonic</code> operations on each
1637 address. All modification orders must be compatible with the happens-before
1638 order. There is no guarantee that the modification orders can be combined to
1639 a global total order for the whole program (and this often will not be
1640 possible). The read in an atomic read-modify-write operation
1641 (<a href="#i_cmpxchg"><code>cmpxchg</code></a> and
1642 <a href="#i_atomicrmw"><code>atomicrmw</code></a>)
1643 reads the value in the modification order immediately before the value it
1644 writes. If one atomic read happens before another atomic read of the same
1645 address, the later read must see the same value or a later value in the
1646 address's modification order. This disallows reordering of
1647 <code>monotonic</code> (or stronger) operations on the same address. If an
1648 address is written <code>monotonic</code>ally by one thread, and other threads
1649 <code>monotonic</code>ally read that address repeatedly, the other threads must
1650 eventually see the write. This corresponds to the C++0x/C1x
1651 <code>memory_order_relaxed</code>.</dd>
1652 <dt><code>acquire</code></dt>
1653 <dd>In addition to the guarantees of <code>monotonic</code>,
1654 a <i>synchronizes-with</i> edge may be formed with a <code>release</code>
1655 operation. This is intended to model C++'s <code>memory_order_acquire</code>.</dd>
1656 <dt><code>release</code></dt>
1657 <dd>In addition to the guarantees of <code>monotonic</code>, if this operation
1658 writes a value which is subsequently read by an <code>acquire</code> operation,
1659 it <i>synchronizes-with</i> that operation. (This isn't a complete
1660 description; see the C++0x definition of a release sequence.) This corresponds
1661 to the C++0x/C1x <code>memory_order_release</code>.</dd>
1662 <dt><code>acq_rel</code> (acquire+release)</dt><dd>Acts as both an
1663 <code>acquire</code> and <code>release</code> operation on its address.
1664 This corresponds to the C++0x/C1x <code>memory_order_acq_rel</code>.</dd>
1665 <dt><code>seq_cst</code> (sequentially consistent)</dt><dd>
1666 <dd>In addition to the guarantees of <code>acq_rel</code>
1667 (<code>acquire</code> for an operation which only reads, <code>release</code>
1668 for an operation which only writes), there is a global total order on all
1669 sequentially-consistent operations on all addresses, which is consistent with
1670 the <i>happens-before</i> partial order and with the modification orders of
1671 all the affected addresses. Each sequentially-consistent read sees the last
1672 preceding write to the same address in this global order. This corresponds
1673 to the C++0x/C1x <code>memory_order_seq_cst</code> and Java volatile.</dd>
1676 <p id="singlethread">If an atomic operation is marked <code>singlethread</code>,
1677 it only <i>synchronizes with</i> or participates in modification and seq_cst
1678 total orderings with other operations running in the same thread (for example,
1679 in signal handlers).</p>
1685 <!-- *********************************************************************** -->
1686 <h2><a name="typesystem">Type System</a></h2>
1687 <!-- *********************************************************************** -->
1691 <p>The LLVM type system is one of the most important features of the
1692 intermediate representation. Being typed enables a number of optimizations
1693 to be performed on the intermediate representation directly, without having
1694 to do extra analyses on the side before the transformation. A strong type
1695 system makes it easier to read the generated code and enables novel analyses
1696 and transformations that are not feasible to perform on normal three address
1697 code representations.</p>
1699 <!-- ======================================================================= -->
1701 <a name="t_classifications">Type Classifications</a>
1706 <p>The types fall into a few useful classifications:</p>
1708 <table border="1" cellspacing="0" cellpadding="4">
1710 <tr><th>Classification</th><th>Types</th></tr>
1712 <td><a href="#t_integer">integer</a></td>
1713 <td><tt>i1, i2, i3, ... i8, ... i16, ... i32, ... i64, ... </tt></td>
1716 <td><a href="#t_floating">floating point</a></td>
1717 <td><tt>half, float, double, x86_fp80, fp128, ppc_fp128</tt></td>
1720 <td><a name="t_firstclass">first class</a></td>
1721 <td><a href="#t_integer">integer</a>,
1722 <a href="#t_floating">floating point</a>,
1723 <a href="#t_pointer">pointer</a>,
1724 <a href="#t_vector">vector</a>,
1725 <a href="#t_struct">structure</a>,
1726 <a href="#t_array">array</a>,
1727 <a href="#t_label">label</a>,
1728 <a href="#t_metadata">metadata</a>.
1732 <td><a href="#t_primitive">primitive</a></td>
1733 <td><a href="#t_label">label</a>,
1734 <a href="#t_void">void</a>,
1735 <a href="#t_integer">integer</a>,
1736 <a href="#t_floating">floating point</a>,
1737 <a href="#t_x86mmx">x86mmx</a>,
1738 <a href="#t_metadata">metadata</a>.</td>
1741 <td><a href="#t_derived">derived</a></td>
1742 <td><a href="#t_array">array</a>,
1743 <a href="#t_function">function</a>,
1744 <a href="#t_pointer">pointer</a>,
1745 <a href="#t_struct">structure</a>,
1746 <a href="#t_vector">vector</a>,
1747 <a href="#t_opaque">opaque</a>.
1753 <p>The <a href="#t_firstclass">first class</a> types are perhaps the most
1754 important. Values of these types are the only ones which can be produced by
1759 <!-- ======================================================================= -->
1761 <a name="t_primitive">Primitive Types</a>
1766 <p>The primitive types are the fundamental building blocks of the LLVM
1769 <!-- _______________________________________________________________________ -->
1771 <a name="t_integer">Integer Type</a>
1777 <p>The integer type is a very simple type that simply specifies an arbitrary
1778 bit width for the integer type desired. Any bit width from 1 bit to
1779 2<sup>23</sup>-1 (about 8 million) can be specified.</p>
1786 <p>The number of bits the integer will occupy is specified by the <tt>N</tt>
1790 <table class="layout">
1792 <td class="left"><tt>i1</tt></td>
1793 <td class="left">a single-bit integer.</td>
1796 <td class="left"><tt>i32</tt></td>
1797 <td class="left">a 32-bit integer.</td>
1800 <td class="left"><tt>i1942652</tt></td>
1801 <td class="left">a really big integer of over 1 million bits.</td>
1807 <!-- _______________________________________________________________________ -->
1809 <a name="t_floating">Floating Point Types</a>
1816 <tr><th>Type</th><th>Description</th></tr>
1817 <tr><td><tt>half</tt></td><td>16-bit floating point value</td></tr>
1818 <tr><td><tt>float</tt></td><td>32-bit floating point value</td></tr>
1819 <tr><td><tt>double</tt></td><td>64-bit floating point value</td></tr>
1820 <tr><td><tt>fp128</tt></td><td>128-bit floating point value (112-bit mantissa)</td></tr>
1821 <tr><td><tt>x86_fp80</tt></td><td>80-bit floating point value (X87)</td></tr>
1822 <tr><td><tt>ppc_fp128</tt></td><td>128-bit floating point value (two 64-bits)</td></tr>
1828 <!-- _______________________________________________________________________ -->
1830 <a name="t_x86mmx">X86mmx Type</a>
1836 <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>
1845 <!-- _______________________________________________________________________ -->
1847 <a name="t_void">Void Type</a>
1853 <p>The void type does not represent any value and has no size.</p>
1862 <!-- _______________________________________________________________________ -->
1864 <a name="t_label">Label Type</a>
1870 <p>The label type represents code labels.</p>
1879 <!-- _______________________________________________________________________ -->
1881 <a name="t_metadata">Metadata Type</a>
1887 <p>The metadata type represents embedded metadata. No derived types may be
1888 created from metadata except for <a href="#t_function">function</a>
1900 <!-- ======================================================================= -->
1902 <a name="t_derived">Derived Types</a>
1907 <p>The real power in LLVM comes from the derived types in the system. This is
1908 what allows a programmer to represent arrays, functions, pointers, and other
1909 useful types. Each of these types contain one or more element types which
1910 may be a primitive type, or another derived type. For example, it is
1911 possible to have a two dimensional array, using an array as the element type
1912 of another array.</p>
1914 <!-- _______________________________________________________________________ -->
1916 <a name="t_aggregate">Aggregate Types</a>
1921 <p>Aggregate Types are a subset of derived types that can contain multiple
1922 member types. <a href="#t_array">Arrays</a> and
1923 <a href="#t_struct">structs</a> are aggregate types.
1924 <a href="#t_vector">Vectors</a> are not considered to be aggregate types.</p>
1928 <!-- _______________________________________________________________________ -->
1930 <a name="t_array">Array Type</a>
1936 <p>The array type is a very simple derived type that arranges elements
1937 sequentially in memory. The array type requires a size (number of elements)
1938 and an underlying data type.</p>
1942 [<# elements> x <elementtype>]
1945 <p>The number of elements is a constant integer value; <tt>elementtype</tt> may
1946 be any type with a size.</p>
1949 <table class="layout">
1951 <td class="left"><tt>[40 x i32]</tt></td>
1952 <td class="left">Array of 40 32-bit integer values.</td>
1955 <td class="left"><tt>[41 x i32]</tt></td>
1956 <td class="left">Array of 41 32-bit integer values.</td>
1959 <td class="left"><tt>[4 x i8]</tt></td>
1960 <td class="left">Array of 4 8-bit integer values.</td>
1963 <p>Here are some examples of multidimensional arrays:</p>
1964 <table class="layout">
1966 <td class="left"><tt>[3 x [4 x i32]]</tt></td>
1967 <td class="left">3x4 array of 32-bit integer values.</td>
1970 <td class="left"><tt>[12 x [10 x float]]</tt></td>
1971 <td class="left">12x10 array of single precision floating point values.</td>
1974 <td class="left"><tt>[2 x [3 x [4 x i16]]]</tt></td>
1975 <td class="left">2x3x4 array of 16-bit integer values.</td>
1979 <p>There is no restriction on indexing beyond the end of the array implied by
1980 a static type (though there are restrictions on indexing beyond the bounds
1981 of an allocated object in some cases). This means that single-dimension
1982 'variable sized array' addressing can be implemented in LLVM with a zero
1983 length array type. An implementation of 'pascal style arrays' in LLVM could
1984 use the type "<tt>{ i32, [0 x float]}</tt>", for example.</p>
1988 <!-- _______________________________________________________________________ -->
1990 <a name="t_function">Function Type</a>
1996 <p>The function type can be thought of as a function signature. It consists of
1997 a return type and a list of formal parameter types. The return type of a
1998 function type is a first class type or a void type.</p>
2002 <returntype> (<parameter list>)
2005 <p>...where '<tt><parameter list></tt>' is a comma-separated list of type
2006 specifiers. Optionally, the parameter list may include a type <tt>...</tt>,
2007 which indicates that the function takes a variable number of arguments.
2008 Variable argument functions can access their arguments with
2009 the <a href="#int_varargs">variable argument handling intrinsic</a>
2010 functions. '<tt><returntype></tt>' is any type except
2011 <a href="#t_label">label</a>.</p>
2014 <table class="layout">
2016 <td class="left"><tt>i32 (i32)</tt></td>
2017 <td class="left">function taking an <tt>i32</tt>, returning an <tt>i32</tt>
2019 </tr><tr class="layout">
2020 <td class="left"><tt>float (i16, i32 *) *
2022 <td class="left"><a href="#t_pointer">Pointer</a> to a function that takes
2023 an <tt>i16</tt> and a <a href="#t_pointer">pointer</a> to <tt>i32</tt>,
2024 returning <tt>float</tt>.
2026 </tr><tr class="layout">
2027 <td class="left"><tt>i32 (i8*, ...)</tt></td>
2028 <td class="left">A vararg function that takes at least one
2029 <a href="#t_pointer">pointer</a> to <tt>i8 </tt> (char in C),
2030 which returns an integer. This is the signature for <tt>printf</tt> in
2033 </tr><tr class="layout">
2034 <td class="left"><tt>{i32, i32} (i32)</tt></td>
2035 <td class="left">A function taking an <tt>i32</tt>, returning a
2036 <a href="#t_struct">structure</a> containing two <tt>i32</tt> values
2043 <!-- _______________________________________________________________________ -->
2045 <a name="t_struct">Structure Type</a>
2051 <p>The structure type is used to represent a collection of data members together
2052 in memory. The elements of a structure may be any type that has a size.</p>
2054 <p>Structures in memory are accessed using '<tt><a href="#i_load">load</a></tt>'
2055 and '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a field
2056 with the '<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.
2057 Structures in registers are accessed using the
2058 '<tt><a href="#i_extractvalue">extractvalue</a></tt>' and
2059 '<tt><a href="#i_insertvalue">insertvalue</a></tt>' instructions.</p>
2061 <p>Structures may optionally be "packed" structures, which indicate that the
2062 alignment of the struct is one byte, and that there is no padding between
2063 the elements. In non-packed structs, padding between field types is inserted
2064 as defined by the TargetData string in the module, which is required to match
2065 what the underlying code generator expects.</p>
2067 <p>Structures can either be "literal" or "identified". A literal structure is
2068 defined inline with other types (e.g. <tt>{i32, i32}*</tt>) whereas identified
2069 types are always defined at the top level with a name. Literal types are
2070 uniqued by their contents and can never be recursive or opaque since there is
2071 no way to write one. Identified types can be recursive, can be opaqued, and are
2077 %T1 = type { <type list> } <i>; Identified normal struct type</i>
2078 %T2 = type <{ <type list> }> <i>; Identified packed struct type</i>
2082 <table class="layout">
2084 <td class="left"><tt>{ i32, i32, i32 }</tt></td>
2085 <td class="left">A triple of three <tt>i32</tt> values</td>
2088 <td class="left"><tt>{ float, i32 (i32) * }</tt></td>
2089 <td class="left">A pair, where the first element is a <tt>float</tt> and the
2090 second element is a <a href="#t_pointer">pointer</a> to a
2091 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
2092 an <tt>i32</tt>.</td>
2095 <td class="left"><tt><{ i8, i32 }></tt></td>
2096 <td class="left">A packed struct known to be 5 bytes in size.</td>
2102 <!-- _______________________________________________________________________ -->
2104 <a name="t_opaque">Opaque Structure Types</a>
2110 <p>Opaque structure types are used to represent named structure types that do
2111 not have a body specified. This corresponds (for example) to the C notion of
2112 a forward declared structure.</p>
2121 <table class="layout">
2123 <td class="left"><tt>opaque</tt></td>
2124 <td class="left">An opaque type.</td>
2132 <!-- _______________________________________________________________________ -->
2134 <a name="t_pointer">Pointer Type</a>
2140 <p>The pointer type is used to specify memory locations.
2141 Pointers are commonly used to reference objects in memory.</p>
2143 <p>Pointer types may have an optional address space attribute defining the
2144 numbered address space where the pointed-to object resides. The default
2145 address space is number zero. The semantics of non-zero address
2146 spaces are target-specific.</p>
2148 <p>Note that LLVM does not permit pointers to void (<tt>void*</tt>) nor does it
2149 permit pointers to labels (<tt>label*</tt>). Use <tt>i8*</tt> instead.</p>
2157 <table class="layout">
2159 <td class="left"><tt>[4 x i32]*</tt></td>
2160 <td class="left">A <a href="#t_pointer">pointer</a> to <a
2161 href="#t_array">array</a> of four <tt>i32</tt> values.</td>
2164 <td class="left"><tt>i32 (i32*) *</tt></td>
2165 <td class="left"> A <a href="#t_pointer">pointer</a> to a <a
2166 href="#t_function">function</a> that takes an <tt>i32*</tt>, returning an
2170 <td class="left"><tt>i32 addrspace(5)*</tt></td>
2171 <td class="left">A <a href="#t_pointer">pointer</a> to an <tt>i32</tt> value
2172 that resides in address space #5.</td>
2178 <!-- _______________________________________________________________________ -->
2180 <a name="t_vector">Vector Type</a>
2186 <p>A vector type is a simple derived type that represents a vector of elements.
2187 Vector types are used when multiple primitive data are operated in parallel
2188 using a single instruction (SIMD). A vector type requires a size (number of
2189 elements) and an underlying primitive data type. Vector types are considered
2190 <a href="#t_firstclass">first class</a>.</p>
2194 < <# elements> x <elementtype> >
2197 <p>The number of elements is a constant integer value larger than 0; elementtype
2198 may be any integer or floating point type, or a pointer to these types.
2199 Vectors of size zero are not allowed. </p>
2202 <table class="layout">
2204 <td class="left"><tt><4 x i32></tt></td>
2205 <td class="left">Vector of 4 32-bit integer values.</td>
2208 <td class="left"><tt><8 x float></tt></td>
2209 <td class="left">Vector of 8 32-bit floating-point values.</td>
2212 <td class="left"><tt><2 x i64></tt></td>
2213 <td class="left">Vector of 2 64-bit integer values.</td>
2216 <td class="left"><tt><4 x i64*></tt></td>
2217 <td class="left">Vector of 4 pointers to 64-bit integer values.</td>
2227 <!-- *********************************************************************** -->
2228 <h2><a name="constants">Constants</a></h2>
2229 <!-- *********************************************************************** -->
2233 <p>LLVM has several different basic types of constants. This section describes
2234 them all and their syntax.</p>
2236 <!-- ======================================================================= -->
2238 <a name="simpleconstants">Simple Constants</a>
2244 <dt><b>Boolean constants</b></dt>
2245 <dd>The two strings '<tt>true</tt>' and '<tt>false</tt>' are both valid
2246 constants of the <tt><a href="#t_integer">i1</a></tt> type.</dd>
2248 <dt><b>Integer constants</b></dt>
2249 <dd>Standard integers (such as '4') are constants of
2250 the <a href="#t_integer">integer</a> type. Negative numbers may be used
2251 with integer types.</dd>
2253 <dt><b>Floating point constants</b></dt>
2254 <dd>Floating point constants use standard decimal notation (e.g. 123.421),
2255 exponential notation (e.g. 1.23421e+2), or a more precise hexadecimal
2256 notation (see below). The assembler requires the exact decimal value of a
2257 floating-point constant. For example, the assembler accepts 1.25 but
2258 rejects 1.3 because 1.3 is a repeating decimal in binary. Floating point
2259 constants must have a <a href="#t_floating">floating point</a> type. </dd>
2261 <dt><b>Null pointer constants</b></dt>
2262 <dd>The identifier '<tt>null</tt>' is recognized as a null pointer constant
2263 and must be of <a href="#t_pointer">pointer type</a>.</dd>
2266 <p>The one non-intuitive notation for constants is the hexadecimal form of
2267 floating point constants. For example, the form '<tt>double
2268 0x432ff973cafa8000</tt>' is equivalent to (but harder to read than)
2269 '<tt>double 4.5e+15</tt>'. The only time hexadecimal floating point
2270 constants are required (and the only time that they are generated by the
2271 disassembler) is when a floating point constant must be emitted but it cannot
2272 be represented as a decimal floating point number in a reasonable number of
2273 digits. For example, NaN's, infinities, and other special values are
2274 represented in their IEEE hexadecimal format so that assembly and disassembly
2275 do not cause any bits to change in the constants.</p>
2277 <p>When using the hexadecimal form, constants of types half, float, and double are
2278 represented using the 16-digit form shown above (which matches the IEEE754
2279 representation for double); half and float values must, however, be exactly
2280 representable as IEE754 half and single precision, respectively.
2281 Hexadecimal format is always used
2282 for long double, and there are three forms of long double. The 80-bit format
2283 used by x86 is represented as <tt>0xK</tt> followed by 20 hexadecimal digits.
2284 The 128-bit format used by PowerPC (two adjacent doubles) is represented
2285 by <tt>0xM</tt> followed by 32 hexadecimal digits. The IEEE 128-bit format
2286 is represented by <tt>0xL</tt> followed by 32 hexadecimal digits; no
2287 currently supported target uses this format. Long doubles will only work if
2288 they match the long double format on your target. All hexadecimal formats
2289 are big-endian (sign bit at the left).</p>
2291 <p>There are no constants of type x86mmx.</p>
2294 <!-- ======================================================================= -->
2296 <a name="aggregateconstants"></a> <!-- old anchor -->
2297 <a name="complexconstants">Complex Constants</a>
2302 <p>Complex constants are a (potentially recursive) combination of simple
2303 constants and smaller complex constants.</p>
2306 <dt><b>Structure constants</b></dt>
2307 <dd>Structure constants are represented with notation similar to structure
2308 type definitions (a comma separated list of elements, surrounded by braces
2309 (<tt>{}</tt>)). For example: "<tt>{ i32 4, float 17.0, i32* @G }</tt>",
2310 where "<tt>@G</tt>" is declared as "<tt>@G = external global i32</tt>".
2311 Structure constants must have <a href="#t_struct">structure type</a>, and
2312 the number and types of elements must match those specified by the
2315 <dt><b>Array constants</b></dt>
2316 <dd>Array constants are represented with notation similar to array type
2317 definitions (a comma separated list of elements, surrounded by square
2318 brackets (<tt>[]</tt>)). For example: "<tt>[ i32 42, i32 11, i32 74
2319 ]</tt>". Array constants must have <a href="#t_array">array type</a>, and
2320 the number and types of elements must match those specified by the
2323 <dt><b>Vector constants</b></dt>
2324 <dd>Vector constants are represented with notation similar to vector type
2325 definitions (a comma separated list of elements, surrounded by
2326 less-than/greater-than's (<tt><></tt>)). For example: "<tt>< i32
2327 42, i32 11, i32 74, i32 100 ></tt>". Vector constants must
2328 have <a href="#t_vector">vector type</a>, and the number and types of
2329 elements must match those specified by the type.</dd>
2331 <dt><b>Zero initialization</b></dt>
2332 <dd>The string '<tt>zeroinitializer</tt>' can be used to zero initialize a
2333 value to zero of <em>any</em> type, including scalar and
2334 <a href="#t_aggregate">aggregate</a> types.
2335 This is often used to avoid having to print large zero initializers
2336 (e.g. for large arrays) and is always exactly equivalent to using explicit
2337 zero initializers.</dd>
2339 <dt><b>Metadata node</b></dt>
2340 <dd>A metadata node is a structure-like constant with
2341 <a href="#t_metadata">metadata type</a>. For example: "<tt>metadata !{
2342 i32 0, metadata !"test" }</tt>". Unlike other constants that are meant to
2343 be interpreted as part of the instruction stream, metadata is a place to
2344 attach additional information such as debug info.</dd>
2349 <!-- ======================================================================= -->
2351 <a name="globalconstants">Global Variable and Function Addresses</a>
2356 <p>The addresses of <a href="#globalvars">global variables</a>
2357 and <a href="#functionstructure">functions</a> are always implicitly valid
2358 (link-time) constants. These constants are explicitly referenced when
2359 the <a href="#identifiers">identifier for the global</a> is used and always
2360 have <a href="#t_pointer">pointer</a> type. For example, the following is a
2361 legal LLVM file:</p>
2363 <pre class="doc_code">
2366 @Z = global [2 x i32*] [ i32* @X, i32* @Y ]
2371 <!-- ======================================================================= -->
2373 <a name="undefvalues">Undefined Values</a>
2378 <p>The string '<tt>undef</tt>' can be used anywhere a constant is expected, and
2379 indicates that the user of the value may receive an unspecified bit-pattern.
2380 Undefined values may be of any type (other than '<tt>label</tt>'
2381 or '<tt>void</tt>') and be used anywhere a constant is permitted.</p>
2383 <p>Undefined values are useful because they indicate to the compiler that the
2384 program is well defined no matter what value is used. This gives the
2385 compiler more freedom to optimize. Here are some examples of (potentially
2386 surprising) transformations that are valid (in pseudo IR):</p>
2389 <pre class="doc_code">
2399 <p>This is safe because all of the output bits are affected by the undef bits.
2400 Any output bit can have a zero or one depending on the input bits.</p>
2402 <pre class="doc_code">
2413 <p>These logical operations have bits that are not always affected by the input.
2414 For example, if <tt>%X</tt> has a zero bit, then the output of the
2415 '<tt>and</tt>' operation will always be a zero for that bit, no matter what
2416 the corresponding bit from the '<tt>undef</tt>' is. As such, it is unsafe to
2417 optimize or assume that the result of the '<tt>and</tt>' is '<tt>undef</tt>'.
2418 However, it is safe to assume that all bits of the '<tt>undef</tt>' could be
2419 0, and optimize the '<tt>and</tt>' to 0. Likewise, it is safe to assume that
2420 all the bits of the '<tt>undef</tt>' operand to the '<tt>or</tt>' could be
2421 set, allowing the '<tt>or</tt>' to be folded to -1.</p>
2423 <pre class="doc_code">
2424 %A = select undef, %X, %Y
2425 %B = select undef, 42, %Y
2426 %C = select %X, %Y, undef
2437 <p>This set of examples shows that undefined '<tt>select</tt>' (and conditional
2438 branch) conditions can go <em>either way</em>, but they have to come from one
2439 of the two operands. In the <tt>%A</tt> example, if <tt>%X</tt> and
2440 <tt>%Y</tt> were both known to have a clear low bit, then <tt>%A</tt> would
2441 have to have a cleared low bit. However, in the <tt>%C</tt> example, the
2442 optimizer is allowed to assume that the '<tt>undef</tt>' operand could be the
2443 same as <tt>%Y</tt>, allowing the whole '<tt>select</tt>' to be
2446 <pre class="doc_code">
2447 %A = xor undef, undef
2465 <p>This example points out that two '<tt>undef</tt>' operands are not
2466 necessarily the same. This can be surprising to people (and also matches C
2467 semantics) where they assume that "<tt>X^X</tt>" is always zero, even
2468 if <tt>X</tt> is undefined. This isn't true for a number of reasons, but the
2469 short answer is that an '<tt>undef</tt>' "variable" can arbitrarily change
2470 its value over its "live range". This is true because the variable doesn't
2471 actually <em>have a live range</em>. Instead, the value is logically read
2472 from arbitrary registers that happen to be around when needed, so the value
2473 is not necessarily consistent over time. In fact, <tt>%A</tt> and <tt>%C</tt>
2474 need to have the same semantics or the core LLVM "replace all uses with"
2475 concept would not hold.</p>
2477 <pre class="doc_code">
2485 <p>These examples show the crucial difference between an <em>undefined
2486 value</em> and <em>undefined behavior</em>. An undefined value (like
2487 '<tt>undef</tt>') is allowed to have an arbitrary bit-pattern. This means that
2488 the <tt>%A</tt> operation can be constant folded to '<tt>undef</tt>', because
2489 the '<tt>undef</tt>' could be an SNaN, and <tt>fdiv</tt> is not (currently)
2490 defined on SNaN's. However, in the second example, we can make a more
2491 aggressive assumption: because the <tt>undef</tt> is allowed to be an
2492 arbitrary value, we are allowed to assume that it could be zero. Since a
2493 divide by zero has <em>undefined behavior</em>, we are allowed to assume that
2494 the operation does not execute at all. This allows us to delete the divide and
2495 all code after it. Because the undefined operation "can't happen", the
2496 optimizer can assume that it occurs in dead code.</p>
2498 <pre class="doc_code">
2499 a: store undef -> %X
2500 b: store %X -> undef
2506 <p>These examples reiterate the <tt>fdiv</tt> example: a store <em>of</em> an
2507 undefined value can be assumed to not have any effect; we can assume that the
2508 value is overwritten with bits that happen to match what was already there.
2509 However, a store <em>to</em> an undefined location could clobber arbitrary
2510 memory, therefore, it has undefined behavior.</p>
2514 <!-- ======================================================================= -->
2516 <a name="poisonvalues">Poison Values</a>
2521 <p>Poison values are similar to <a href="#undefvalues">undef values</a>, however
2522 they also represent the fact that an instruction or constant expression which
2523 cannot evoke side effects has nevertheless detected a condition which results
2524 in undefined behavior.</p>
2526 <p>There is currently no way of representing a poison value in the IR; they
2527 only exist when produced by operations such as
2528 <a href="#i_add"><tt>add</tt></a> with the <tt>nsw</tt> flag.</p>
2530 <p>Poison value behavior is defined in terms of value <i>dependence</i>:</p>
2533 <li>Values other than <a href="#i_phi"><tt>phi</tt></a> nodes depend on
2534 their operands.</li>
2536 <li><a href="#i_phi"><tt>Phi</tt></a> nodes depend on the operand corresponding
2537 to their dynamic predecessor basic block.</li>
2539 <li>Function arguments depend on the corresponding actual argument values in
2540 the dynamic callers of their functions.</li>
2542 <li><a href="#i_call"><tt>Call</tt></a> instructions depend on the
2543 <a href="#i_ret"><tt>ret</tt></a> instructions that dynamically transfer
2544 control back to them.</li>
2546 <li><a href="#i_invoke"><tt>Invoke</tt></a> instructions depend on the
2547 <a href="#i_ret"><tt>ret</tt></a>, <a href="#i_resume"><tt>resume</tt></a>,
2548 or exception-throwing call instructions that dynamically transfer control
2551 <li>Non-volatile loads and stores depend on the most recent stores to all of the
2552 referenced memory addresses, following the order in the IR
2553 (including loads and stores implied by intrinsics such as
2554 <a href="#int_memcpy"><tt>@llvm.memcpy</tt></a>.)</li>
2556 <!-- TODO: In the case of multiple threads, this only applies if the store
2557 "happens-before" the load or store. -->
2559 <!-- TODO: floating-point exception state -->
2561 <li>An instruction with externally visible side effects depends on the most
2562 recent preceding instruction with externally visible side effects, following
2563 the order in the IR. (This includes
2564 <a href="#volatile">volatile operations</a>.)</li>
2566 <li>An instruction <i>control-depends</i> on a
2567 <a href="#terminators">terminator instruction</a>
2568 if the terminator instruction has multiple successors and the instruction
2569 is always executed when control transfers to one of the successors, and
2570 may not be executed when control is transferred to another.</li>
2572 <li>Additionally, an instruction also <i>control-depends</i> on a terminator
2573 instruction if the set of instructions it otherwise depends on would be
2574 different if the terminator had transferred control to a different
2577 <li>Dependence is transitive.</li>
2581 <p>Poison Values have the same behavior as <a href="#undefvalues">undef values</a>,
2582 with the additional affect that any instruction which has a <i>dependence</i>
2583 on a poison value has undefined behavior.</p>
2585 <p>Here are some examples:</p>
2587 <pre class="doc_code">
2589 %poison = sub nuw i32 0, 1 ; Results in a poison value.
2590 %still_poison = and i32 %poison, 0 ; 0, but also poison.
2591 %poison_yet_again = getelementptr i32* @h, i32 %still_poison
2592 store i32 0, i32* %poison_yet_again ; memory at @h[0] is poisoned
2594 store i32 %poison, i32* @g ; Poison value stored to memory.
2595 %poison2 = load i32* @g ; Poison value loaded back from memory.
2597 store volatile i32 %poison, i32* @g ; External observation; undefined behavior.
2599 %narrowaddr = bitcast i32* @g to i16*
2600 %wideaddr = bitcast i32* @g to i64*
2601 %poison3 = load i16* %narrowaddr ; Returns a poison value.
2602 %poison4 = load i64* %wideaddr ; Returns a poison value.
2604 %cmp = icmp slt i32 %poison, 0 ; Returns a poison value.
2605 br i1 %cmp, label %true, label %end ; Branch to either destination.
2608 store volatile i32 0, i32* @g ; This is control-dependent on %cmp, so
2609 ; it has undefined behavior.
2613 %p = phi i32 [ 0, %entry ], [ 1, %true ]
2614 ; Both edges into this PHI are
2615 ; control-dependent on %cmp, so this
2616 ; always results in a poison value.
2618 store volatile i32 0, i32* @g ; This would depend on the store in %true
2619 ; if %cmp is true, or the store in %entry
2620 ; otherwise, so this is undefined behavior.
2622 br i1 %cmp, label %second_true, label %second_end
2623 ; The same branch again, but this time the
2624 ; true block doesn't have side effects.
2631 store volatile i32 0, i32* @g ; This time, the instruction always depends
2632 ; on the store in %end. Also, it is
2633 ; control-equivalent to %end, so this is
2634 ; well-defined (ignoring earlier undefined
2635 ; behavior in this example).
2640 <!-- ======================================================================= -->
2642 <a name="blockaddress">Addresses of Basic Blocks</a>
2647 <p><b><tt>blockaddress(@function, %block)</tt></b></p>
2649 <p>The '<tt>blockaddress</tt>' constant computes the address of the specified
2650 basic block in the specified function, and always has an i8* type. Taking
2651 the address of the entry block is illegal.</p>
2653 <p>This value only has defined behavior when used as an operand to the
2654 '<a href="#i_indirectbr"><tt>indirectbr</tt></a>' instruction, or for
2655 comparisons against null. Pointer equality tests between labels addresses
2656 results in undefined behavior — though, again, comparison against null
2657 is ok, and no label is equal to the null pointer. This may be passed around
2658 as an opaque pointer sized value as long as the bits are not inspected. This
2659 allows <tt>ptrtoint</tt> and arithmetic to be performed on these values so
2660 long as the original value is reconstituted before the <tt>indirectbr</tt>
2663 <p>Finally, some targets may provide defined semantics when using the value as
2664 the operand to an inline assembly, but that is target specific.</p>
2669 <!-- ======================================================================= -->
2671 <a name="constantexprs">Constant Expressions</a>
2676 <p>Constant expressions are used to allow expressions involving other constants
2677 to be used as constants. Constant expressions may be of
2678 any <a href="#t_firstclass">first class</a> type and may involve any LLVM
2679 operation that does not have side effects (e.g. load and call are not
2680 supported). The following is the syntax for constant expressions:</p>
2683 <dt><b><tt>trunc (CST to TYPE)</tt></b></dt>
2684 <dd>Truncate a constant to another type. The bit size of CST must be larger
2685 than the bit size of TYPE. Both types must be integers.</dd>
2687 <dt><b><tt>zext (CST to TYPE)</tt></b></dt>
2688 <dd>Zero extend a constant to another type. The bit size of CST must be
2689 smaller than the bit size of TYPE. Both types must be integers.</dd>
2691 <dt><b><tt>sext (CST to TYPE)</tt></b></dt>
2692 <dd>Sign extend a constant to another type. The bit size of CST must be
2693 smaller than the bit size of TYPE. Both types must be integers.</dd>
2695 <dt><b><tt>fptrunc (CST to TYPE)</tt></b></dt>
2696 <dd>Truncate a floating point constant to another floating point type. The
2697 size of CST must be larger than the size of TYPE. Both types must be
2698 floating point.</dd>
2700 <dt><b><tt>fpext (CST to TYPE)</tt></b></dt>
2701 <dd>Floating point extend a constant to another type. The size of CST must be
2702 smaller or equal to the size of TYPE. Both types must be floating
2705 <dt><b><tt>fptoui (CST to TYPE)</tt></b></dt>
2706 <dd>Convert a floating point constant to the corresponding unsigned integer
2707 constant. TYPE must be a scalar or vector integer type. CST must be of
2708 scalar or vector floating point type. Both CST and TYPE must be scalars,
2709 or vectors of the same number of elements. If the value won't fit in the
2710 integer type, the results are undefined.</dd>
2712 <dt><b><tt>fptosi (CST to TYPE)</tt></b></dt>
2713 <dd>Convert a floating point constant to the corresponding signed integer
2714 constant. TYPE must be a scalar or vector integer type. CST must be of
2715 scalar or vector floating point type. Both CST and TYPE must be scalars,
2716 or vectors of the same number of elements. If the value won't fit in the
2717 integer type, the results are undefined.</dd>
2719 <dt><b><tt>uitofp (CST to TYPE)</tt></b></dt>
2720 <dd>Convert an unsigned integer constant to the corresponding floating point
2721 constant. TYPE must be a scalar or vector floating point type. CST must be
2722 of scalar or vector integer type. Both CST and TYPE must be scalars, or
2723 vectors of the same number of elements. If the value won't fit in the
2724 floating point type, the results are undefined.</dd>
2726 <dt><b><tt>sitofp (CST to TYPE)</tt></b></dt>
2727 <dd>Convert a signed integer constant to the corresponding floating point
2728 constant. TYPE must be a scalar or vector floating point type. CST must be
2729 of scalar or vector integer type. Both CST and TYPE must be scalars, or
2730 vectors of the same number of elements. If the value won't fit in the
2731 floating point type, the results are undefined.</dd>
2733 <dt><b><tt>ptrtoint (CST to TYPE)</tt></b></dt>
2734 <dd>Convert a pointer typed constant to the corresponding integer constant
2735 <tt>TYPE</tt> must be an integer type. <tt>CST</tt> must be of pointer
2736 type. The <tt>CST</tt> value is zero extended, truncated, or unchanged to
2737 make it fit in <tt>TYPE</tt>.</dd>
2739 <dt><b><tt>inttoptr (CST to TYPE)</tt></b></dt>
2740 <dd>Convert a integer constant to a pointer constant. TYPE must be a pointer
2741 type. CST must be of integer type. The CST value is zero extended,
2742 truncated, or unchanged to make it fit in a pointer size. This one is
2743 <i>really</i> dangerous!</dd>
2745 <dt><b><tt>bitcast (CST to TYPE)</tt></b></dt>
2746 <dd>Convert a constant, CST, to another TYPE. The constraints of the operands
2747 are the same as those for the <a href="#i_bitcast">bitcast
2748 instruction</a>.</dd>
2750 <dt><b><tt>getelementptr (CSTPTR, IDX0, IDX1, ...)</tt></b></dt>
2751 <dt><b><tt>getelementptr inbounds (CSTPTR, IDX0, IDX1, ...)</tt></b></dt>
2752 <dd>Perform the <a href="#i_getelementptr">getelementptr operation</a> on
2753 constants. As with the <a href="#i_getelementptr">getelementptr</a>
2754 instruction, the index list may have zero or more indexes, which are
2755 required to make sense for the type of "CSTPTR".</dd>
2757 <dt><b><tt>select (COND, VAL1, VAL2)</tt></b></dt>
2758 <dd>Perform the <a href="#i_select">select operation</a> on constants.</dd>
2760 <dt><b><tt>icmp COND (VAL1, VAL2)</tt></b></dt>
2761 <dd>Performs the <a href="#i_icmp">icmp operation</a> on constants.</dd>
2763 <dt><b><tt>fcmp COND (VAL1, VAL2)</tt></b></dt>
2764 <dd>Performs the <a href="#i_fcmp">fcmp operation</a> on constants.</dd>
2766 <dt><b><tt>extractelement (VAL, IDX)</tt></b></dt>
2767 <dd>Perform the <a href="#i_extractelement">extractelement operation</a> on
2770 <dt><b><tt>insertelement (VAL, ELT, IDX)</tt></b></dt>
2771 <dd>Perform the <a href="#i_insertelement">insertelement operation</a> on
2774 <dt><b><tt>shufflevector (VEC1, VEC2, IDXMASK)</tt></b></dt>
2775 <dd>Perform the <a href="#i_shufflevector">shufflevector operation</a> on
2778 <dt><b><tt>extractvalue (VAL, IDX0, IDX1, ...)</tt></b></dt>
2779 <dd>Perform the <a href="#i_extractvalue">extractvalue operation</a> on
2780 constants. The index list is interpreted in a similar manner as indices in
2781 a '<a href="#i_getelementptr">getelementptr</a>' operation. At least one
2782 index value must be specified.</dd>
2784 <dt><b><tt>insertvalue (VAL, ELT, IDX0, IDX1, ...)</tt></b></dt>
2785 <dd>Perform the <a href="#i_insertvalue">insertvalue operation</a> on
2786 constants. The index list is interpreted in a similar manner as indices in
2787 a '<a href="#i_getelementptr">getelementptr</a>' operation. At least one
2788 index value must be specified.</dd>
2790 <dt><b><tt>OPCODE (LHS, RHS)</tt></b></dt>
2791 <dd>Perform the specified operation of the LHS and RHS constants. OPCODE may
2792 be any of the <a href="#binaryops">binary</a>
2793 or <a href="#bitwiseops">bitwise binary</a> operations. The constraints
2794 on operands are the same as those for the corresponding instruction
2795 (e.g. no bitwise operations on floating point values are allowed).</dd>
2802 <!-- *********************************************************************** -->
2803 <h2><a name="othervalues">Other Values</a></h2>
2804 <!-- *********************************************************************** -->
2806 <!-- ======================================================================= -->
2808 <a name="inlineasm">Inline Assembler Expressions</a>
2813 <p>LLVM supports inline assembler expressions (as opposed
2814 to <a href="#moduleasm">Module-Level Inline Assembly</a>) through the use of
2815 a special value. This value represents the inline assembler as a string
2816 (containing the instructions to emit), a list of operand constraints (stored
2817 as a string), a flag that indicates whether or not the inline asm
2818 expression has side effects, and a flag indicating whether the function
2819 containing the asm needs to align its stack conservatively. An example
2820 inline assembler expression is:</p>
2822 <pre class="doc_code">
2823 i32 (i32) asm "bswap $0", "=r,r"
2826 <p>Inline assembler expressions may <b>only</b> be used as the callee operand of
2827 a <a href="#i_call"><tt>call</tt> instruction</a>. Thus, typically we
2830 <pre class="doc_code">
2831 %X = call i32 asm "<a href="#int_bswap">bswap</a> $0", "=r,r"(i32 %Y)
2834 <p>Inline asms with side effects not visible in the constraint list must be
2835 marked as having side effects. This is done through the use of the
2836 '<tt>sideeffect</tt>' keyword, like so:</p>
2838 <pre class="doc_code">
2839 call void asm sideeffect "eieio", ""()
2842 <p>In some cases inline asms will contain code that will not work unless the
2843 stack is aligned in some way, such as calls or SSE instructions on x86,
2844 yet will not contain code that does that alignment within the asm.
2845 The compiler should make conservative assumptions about what the asm might
2846 contain and should generate its usual stack alignment code in the prologue
2847 if the '<tt>alignstack</tt>' keyword is present:</p>
2849 <pre class="doc_code">
2850 call void asm alignstack "eieio", ""()
2853 <p>If both keywords appear the '<tt>sideeffect</tt>' keyword must come
2857 <p>TODO: The format of the asm and constraints string still need to be
2858 documented here. Constraints on what can be done (e.g. duplication, moving,
2859 etc need to be documented). This is probably best done by reference to
2860 another document that covers inline asm from a holistic perspective.</p>
2863 <!-- _______________________________________________________________________ -->
2865 <a name="inlineasm_md">Inline Asm Metadata</a>
2870 <p>The call instructions that wrap inline asm nodes may have a
2871 "<tt>!srcloc</tt>" MDNode attached to it that contains a list of constant
2872 integers. If present, the code generator will use the integer as the
2873 location cookie value when report errors through the <tt>LLVMContext</tt>
2874 error reporting mechanisms. This allows a front-end to correlate backend
2875 errors that occur with inline asm back to the source code that produced it.
2878 <pre class="doc_code">
2879 call void asm sideeffect "something bad", ""()<b>, !srcloc !42</b>
2881 !42 = !{ i32 1234567 }
2884 <p>It is up to the front-end to make sense of the magic numbers it places in the
2885 IR. If the MDNode contains multiple constants, the code generator will use
2886 the one that corresponds to the line of the asm that the error occurs on.</p>
2892 <!-- ======================================================================= -->
2894 <a name="metadata">Metadata Nodes and Metadata Strings</a>
2899 <p>LLVM IR allows metadata to be attached to instructions in the program that
2900 can convey extra information about the code to the optimizers and code
2901 generator. One example application of metadata is source-level debug
2902 information. There are two metadata primitives: strings and nodes. All
2903 metadata has the <tt>metadata</tt> type and is identified in syntax by a
2904 preceding exclamation point ('<tt>!</tt>').</p>
2906 <p>A metadata string is a string surrounded by double quotes. It can contain
2907 any character by escaping non-printable characters with "<tt>\xx</tt>" where
2908 "<tt>xx</tt>" is the two digit hex code. For example:
2909 "<tt>!"test\00"</tt>".</p>
2911 <p>Metadata nodes are represented with notation similar to structure constants
2912 (a comma separated list of elements, surrounded by braces and preceded by an
2913 exclamation point). Metadata nodes can have any values as their operand. For
2916 <div class="doc_code">
2918 !{ metadata !"test\00", i32 10}
2922 <p>A <a href="#namedmetadatastructure">named metadata</a> is a collection of
2923 metadata nodes, which can be looked up in the module symbol table. For
2926 <div class="doc_code">
2928 !foo = metadata !{!4, !3}
2932 <p>Metadata can be used as function arguments. Here <tt>llvm.dbg.value</tt>
2933 function is using two metadata arguments:</p>
2935 <div class="doc_code">
2937 call void @llvm.dbg.value(metadata !24, i64 0, metadata !25)
2941 <p>Metadata can be attached with an instruction. Here metadata <tt>!21</tt> is
2942 attached to the <tt>add</tt> instruction using the <tt>!dbg</tt>
2945 <div class="doc_code">
2947 %indvar.next = add i64 %indvar, 1, !dbg !21
2951 <p>More information about specific metadata nodes recognized by the optimizers
2952 and code generator is found below.</p>
2954 <!-- _______________________________________________________________________ -->
2956 <a name="tbaa">'<tt>tbaa</tt>' Metadata</a>
2961 <p>In LLVM IR, memory does not have types, so LLVM's own type system is not
2962 suitable for doing TBAA. Instead, metadata is added to the IR to describe
2963 a type system of a higher level language. This can be used to implement
2964 typical C/C++ TBAA, but it can also be used to implement custom alias
2965 analysis behavior for other languages.</p>
2967 <p>The current metadata format is very simple. TBAA metadata nodes have up to
2968 three fields, e.g.:</p>
2970 <div class="doc_code">
2972 !0 = metadata !{ metadata !"an example type tree" }
2973 !1 = metadata !{ metadata !"int", metadata !0 }
2974 !2 = metadata !{ metadata !"float", metadata !0 }
2975 !3 = metadata !{ metadata !"const float", metadata !2, i64 1 }
2979 <p>The first field is an identity field. It can be any value, usually
2980 a metadata string, which uniquely identifies the type. The most important
2981 name in the tree is the name of the root node. Two trees with
2982 different root node names are entirely disjoint, even if they
2983 have leaves with common names.</p>
2985 <p>The second field identifies the type's parent node in the tree, or
2986 is null or omitted for a root node. A type is considered to alias
2987 all of its descendants and all of its ancestors in the tree. Also,
2988 a type is considered to alias all types in other trees, so that
2989 bitcode produced from multiple front-ends is handled conservatively.</p>
2991 <p>If the third field is present, it's an integer which if equal to 1
2992 indicates that the type is "constant" (meaning
2993 <tt>pointsToConstantMemory</tt> should return true; see
2994 <a href="AliasAnalysis.html#OtherItfs">other useful
2995 <tt>AliasAnalysis</tt> methods</a>).</p>
2999 <!-- _______________________________________________________________________ -->
3001 <a name="fpaccuracy">'<tt>fpaccuracy</tt>' Metadata</a>
3006 <p><tt>fpaccuracy</tt> metadata may be attached to any instruction of floating
3007 point type. It expresses the maximum relative error of the result of
3008 that instruction, in ULPs. ULP is defined as follows:</p>
3012 <p>If <tt>x</tt> is a real number that lies between two finite consecutive
3013 floating-point numbers <tt>a</tt> and <tt>b</tt>, without being equal to one
3014 of them, then <tt>ulp(x) = |b - a|</tt>, otherwise <tt>ulp(x)</tt> is the
3015 distance between the two non-equal finite floating-point numbers nearest
3016 <tt>x</tt>. Moreover, <tt>ulp(NaN)</tt> is <tt>NaN</tt>.</p>
3020 <p>The maximum relative error may be any rational number. The metadata node
3021 shall consist of a pair of unsigned integers respectively representing
3022 the numerator and denominator. For example, 2.5 ULP:</p>
3024 <div class="doc_code">
3026 !0 = metadata !{ i32 5, i32 2 }
3036 <!-- *********************************************************************** -->
3038 <a name="module_flags">Module Flags Metadata</a>
3040 <!-- *********************************************************************** -->
3044 <p>Information about the module as a whole is difficult to convey to LLVM's
3045 subsystems. The LLVM IR isn't sufficient to transmit this
3046 information. The <tt>llvm.module.flags</tt> named metadata exists in order to
3047 facilitate this. These flags are in the form of key / value pairs —
3048 much like a dictionary — making it easy for any subsystem who cares
3049 about a flag to look it up.</p>
3051 <p>The <tt>llvm.module.flags</tt> metadata contains a list of metadata
3052 triplets. Each triplet has the following form:</p>
3055 <li>The first element is a <i>behavior</i> flag, which specifies the behavior
3056 when two (or more) modules are merged together, and it encounters two (or
3057 more) metadata with the same ID. The supported behaviors are described
3060 <li>The second element is a metadata string that is a unique ID for the
3061 metadata. How each ID is interpreted is documented below.</li>
3063 <li>The third element is the value of the flag.</li>
3066 <p>When two (or more) modules are merged together, the resulting
3067 <tt>llvm.module.flags</tt> metadata is the union of the
3068 modules' <tt>llvm.module.flags</tt> metadata. The only exception being a flag
3069 with the <i>Override</i> behavior, which may override another flag's value
3072 <p>The following behaviors are supported:</p>
3074 <table border="1" cellspacing="0" cellpadding="4">
3083 <dt><b>Error</b></dt>
3084 <dd>Emits an error if two values disagree. It is an error to have an ID
3085 with both an Error and a Warning behavior.</dd>
3091 <dt><b>Warning</b></dt>
3092 <dd>Emits a warning if two values disagree.</dd>
3098 <dt><b>Require</b></dt>
3099 <dd>Emits an error when the specified value is not present or doesn't
3100 have the specified value. It is an error for two (or more)
3101 <tt>llvm.module.flags</tt> with the same ID to have the Require
3102 behavior but different values. There may be multiple Require flags
3109 <dt><b>Override</b></dt>
3110 <dd>Uses the specified value if the two values disagree. It is an error
3111 for two (or more) <tt>llvm.module.flags</tt> with the same ID to
3112 have the Override behavior but different values.</dd>
3118 <p>An example of module flags:</p>
3120 <pre class="doc_code">
3121 !0 = metadata !{ i32 1, metadata !"foo", i32 1 }
3122 !1 = metadata !{ i32 4, metadata !"bar", i32 37 }
3123 !2 = metadata !{ i32 2, metadata !"qux", i32 42 }
3124 !3 = metadata !{ i32 3, metadata !"qux",
3126 metadata !"foo", i32 1
3129 !llvm.module.flags = !{ !0, !1, !2, !3 }
3133 <li><p>Metadata <tt>!0</tt> has the ID <tt>!"foo"</tt> and the value '1'. The
3134 behavior if two or more <tt>!"foo"</tt> flags are seen is to emit an
3135 error if their values are not equal.</p></li>
3137 <li><p>Metadata <tt>!1</tt> has the ID <tt>!"bar"</tt> and the value '37'. The
3138 behavior if two or more <tt>!"bar"</tt> flags are seen is to use the
3139 value '37' if their values are not equal.</p></li>
3141 <li><p>Metadata <tt>!2</tt> has the ID <tt>!"qux"</tt> and the value '42'. The
3142 behavior if two or more <tt>!"qux"</tt> flags are seen is to emit a
3143 warning if their values are not equal.</p></li>
3145 <li><p>Metadata <tt>!3</tt> has the ID <tt>!"qux"</tt> and the value:</p>
3147 <pre class="doc_code">
3148 metadata !{ metadata !"foo", i32 1 }
3150 <p>The behavior is to emit an error if the <tt>llvm.module.flags</tt> does
3151 not contain a flag with the ID <tt>!"foo"</tt> that has the value
3152 '1'. If two or more <tt>!"qux"</tt> flags exist, then they must have
3153 the same value or an error will be issued.</p></li>
3158 <!-- *********************************************************************** -->
3160 <a name="intrinsic_globals">Intrinsic Global Variables</a>
3162 <!-- *********************************************************************** -->
3164 <p>LLVM has a number of "magic" global variables that contain data that affect
3165 code generation or other IR semantics. These are documented here. All globals
3166 of this sort should have a section specified as "<tt>llvm.metadata</tt>". This
3167 section and all globals that start with "<tt>llvm.</tt>" are reserved for use
3170 <!-- ======================================================================= -->
3172 <a name="intg_used">The '<tt>llvm.used</tt>' Global Variable</a>
3177 <p>The <tt>@llvm.used</tt> global is an array with i8* element type which has <a
3178 href="#linkage_appending">appending linkage</a>. This array contains a list of
3179 pointers to global variables and functions which may optionally have a pointer
3180 cast formed of bitcast or getelementptr. For example, a legal use of it is:</p>
3182 <div class="doc_code">
3187 @llvm.used = appending global [2 x i8*] [
3189 i8* bitcast (i32* @Y to i8*)
3190 ], section "llvm.metadata"
3194 <p>If a global variable appears in the <tt>@llvm.used</tt> list, then the
3195 compiler, assembler, and linker are required to treat the symbol as if there
3196 is a reference to the global that it cannot see. For example, if a variable
3197 has internal linkage and no references other than that from
3198 the <tt>@llvm.used</tt> list, it cannot be deleted. This is commonly used to
3199 represent references from inline asms and other things the compiler cannot
3200 "see", and corresponds to "<tt>attribute((used))</tt>" in GNU C.</p>
3202 <p>On some targets, the code generator must emit a directive to the assembler or
3203 object file to prevent the assembler and linker from molesting the
3208 <!-- ======================================================================= -->
3210 <a name="intg_compiler_used">
3211 The '<tt>llvm.compiler.used</tt>' Global Variable
3217 <p>The <tt>@llvm.compiler.used</tt> directive is the same as the
3218 <tt>@llvm.used</tt> directive, except that it only prevents the compiler from
3219 touching the symbol. On targets that support it, this allows an intelligent
3220 linker to optimize references to the symbol without being impeded as it would
3221 be by <tt>@llvm.used</tt>.</p>
3223 <p>This is a rare construct that should only be used in rare circumstances, and
3224 should not be exposed to source languages.</p>
3228 <!-- ======================================================================= -->
3230 <a name="intg_global_ctors">The '<tt>llvm.global_ctors</tt>' Global Variable</a>
3235 <div class="doc_code">
3237 %0 = type { i32, void ()* }
3238 @llvm.global_ctors = appending global [1 x %0] [%0 { i32 65535, void ()* @ctor }]
3242 <p>The <tt>@llvm.global_ctors</tt> array contains a list of constructor
3243 functions and associated priorities. The functions referenced by this array
3244 will be called in ascending order of priority (i.e. lowest first) when the
3245 module is loaded. The order of functions with the same priority is not
3250 <!-- ======================================================================= -->
3252 <a name="intg_global_dtors">The '<tt>llvm.global_dtors</tt>' Global Variable</a>
3257 <div class="doc_code">
3259 %0 = type { i32, void ()* }
3260 @llvm.global_dtors = appending global [1 x %0] [%0 { i32 65535, void ()* @dtor }]
3264 <p>The <tt>@llvm.global_dtors</tt> array contains a list of destructor functions
3265 and associated priorities. The functions referenced by this array will be
3266 called in descending order of priority (i.e. highest first) when the module
3267 is loaded. The order of functions with the same priority is not defined.</p>
3273 <!-- *********************************************************************** -->
3274 <h2><a name="instref">Instruction Reference</a></h2>
3275 <!-- *********************************************************************** -->
3279 <p>The LLVM instruction set consists of several different classifications of
3280 instructions: <a href="#terminators">terminator
3281 instructions</a>, <a href="#binaryops">binary instructions</a>,
3282 <a href="#bitwiseops">bitwise binary instructions</a>,
3283 <a href="#memoryops">memory instructions</a>, and
3284 <a href="#otherops">other instructions</a>.</p>
3286 <!-- ======================================================================= -->
3288 <a name="terminators">Terminator Instructions</a>
3293 <p>As mentioned <a href="#functionstructure">previously</a>, every basic block
3294 in a program ends with a "Terminator" instruction, which indicates which
3295 block should be executed after the current block is finished. These
3296 terminator instructions typically yield a '<tt>void</tt>' value: they produce
3297 control flow, not values (the one exception being the
3298 '<a href="#i_invoke"><tt>invoke</tt></a>' instruction).</p>
3300 <p>The terminator instructions are:
3301 '<a href="#i_ret"><tt>ret</tt></a>',
3302 '<a href="#i_br"><tt>br</tt></a>',
3303 '<a href="#i_switch"><tt>switch</tt></a>',
3304 '<a href="#i_indirectbr"><tt>indirectbr</tt></a>',
3305 '<a href="#i_invoke"><tt>invoke</tt></a>',
3306 '<a href="#i_resume"><tt>resume</tt></a>', and
3307 '<a href="#i_unreachable"><tt>unreachable</tt></a>'.</p>
3309 <!-- _______________________________________________________________________ -->
3311 <a name="i_ret">'<tt>ret</tt>' Instruction</a>
3318 ret <type> <value> <i>; Return a value from a non-void function</i>
3319 ret void <i>; Return from void function</i>
3323 <p>The '<tt>ret</tt>' instruction is used to return control flow (and optionally
3324 a value) from a function back to the caller.</p>
3326 <p>There are two forms of the '<tt>ret</tt>' instruction: one that returns a
3327 value and then causes control flow, and one that just causes control flow to
3331 <p>The '<tt>ret</tt>' instruction optionally accepts a single argument, the
3332 return value. The type of the return value must be a
3333 '<a href="#t_firstclass">first class</a>' type.</p>
3335 <p>A function is not <a href="#wellformed">well formed</a> if it it has a
3336 non-void return type and contains a '<tt>ret</tt>' instruction with no return
3337 value or a return value with a type that does not match its type, or if it
3338 has a void return type and contains a '<tt>ret</tt>' instruction with a
3342 <p>When the '<tt>ret</tt>' instruction is executed, control flow returns back to
3343 the calling function's context. If the caller is a
3344 "<a href="#i_call"><tt>call</tt></a>" instruction, execution continues at the
3345 instruction after the call. If the caller was an
3346 "<a href="#i_invoke"><tt>invoke</tt></a>" instruction, execution continues at
3347 the beginning of the "normal" destination block. If the instruction returns
3348 a value, that value shall set the call or invoke instruction's return
3353 ret i32 5 <i>; Return an integer value of 5</i>
3354 ret void <i>; Return from a void function</i>
3355 ret { i32, i8 } { i32 4, i8 2 } <i>; Return a struct of values 4 and 2</i>
3359 <!-- _______________________________________________________________________ -->
3361 <a name="i_br">'<tt>br</tt>' Instruction</a>
3368 br i1 <cond>, label <iftrue>, label <iffalse>
3369 br label <dest> <i>; Unconditional branch</i>
3373 <p>The '<tt>br</tt>' instruction is used to cause control flow to transfer to a
3374 different basic block in the current function. There are two forms of this
3375 instruction, corresponding to a conditional branch and an unconditional
3379 <p>The conditional branch form of the '<tt>br</tt>' instruction takes a single
3380 '<tt>i1</tt>' value and two '<tt>label</tt>' values. The unconditional form
3381 of the '<tt>br</tt>' instruction takes a single '<tt>label</tt>' value as a
3385 <p>Upon execution of a conditional '<tt>br</tt>' instruction, the '<tt>i1</tt>'
3386 argument is evaluated. If the value is <tt>true</tt>, control flows to the
3387 '<tt>iftrue</tt>' <tt>label</tt> argument. If "cond" is <tt>false</tt>,
3388 control flows to the '<tt>iffalse</tt>' <tt>label</tt> argument.</p>
3393 %cond = <a href="#i_icmp">icmp</a> eq i32 %a, %b
3394 br i1 %cond, label %IfEqual, label %IfUnequal
3396 <a href="#i_ret">ret</a> i32 1
3398 <a href="#i_ret">ret</a> i32 0
3403 <!-- _______________________________________________________________________ -->
3405 <a name="i_switch">'<tt>switch</tt>' Instruction</a>
3412 switch <intty> <value>, label <defaultdest> [ <intty> <val>, label <dest> ... ]
3416 <p>The '<tt>switch</tt>' instruction is used to transfer control flow to one of
3417 several different places. It is a generalization of the '<tt>br</tt>'
3418 instruction, allowing a branch to occur to one of many possible
3422 <p>The '<tt>switch</tt>' instruction uses three parameters: an integer
3423 comparison value '<tt>value</tt>', a default '<tt>label</tt>' destination,
3424 and an array of pairs of comparison value constants and '<tt>label</tt>'s.
3425 The table is not allowed to contain duplicate constant entries.</p>
3428 <p>The <tt>switch</tt> instruction specifies a table of values and
3429 destinations. When the '<tt>switch</tt>' instruction is executed, this table
3430 is searched for the given value. If the value is found, control flow is
3431 transferred to the corresponding destination; otherwise, control flow is
3432 transferred to the default destination.</p>
3434 <h5>Implementation:</h5>
3435 <p>Depending on properties of the target machine and the particular
3436 <tt>switch</tt> instruction, this instruction may be code generated in
3437 different ways. For example, it could be generated as a series of chained
3438 conditional branches or with a lookup table.</p>
3442 <i>; Emulate a conditional br instruction</i>
3443 %Val = <a href="#i_zext">zext</a> i1 %value to i32
3444 switch i32 %Val, label %truedest [ i32 0, label %falsedest ]
3446 <i>; Emulate an unconditional br instruction</i>
3447 switch i32 0, label %dest [ ]
3449 <i>; Implement a jump table:</i>
3450 switch i32 %val, label %otherwise [ i32 0, label %onzero
3452 i32 2, label %ontwo ]
3458 <!-- _______________________________________________________________________ -->
3460 <a name="i_indirectbr">'<tt>indirectbr</tt>' Instruction</a>
3467 indirectbr <somety>* <address>, [ label <dest1>, label <dest2>, ... ]
3472 <p>The '<tt>indirectbr</tt>' instruction implements an indirect branch to a label
3473 within the current function, whose address is specified by
3474 "<tt>address</tt>". Address must be derived from a <a
3475 href="#blockaddress">blockaddress</a> constant.</p>
3479 <p>The '<tt>address</tt>' argument is the address of the label to jump to. The
3480 rest of the arguments indicate the full set of possible destinations that the
3481 address may point to. Blocks are allowed to occur multiple times in the
3482 destination list, though this isn't particularly useful.</p>
3484 <p>This destination list is required so that dataflow analysis has an accurate
3485 understanding of the CFG.</p>
3489 <p>Control transfers to the block specified in the address argument. All
3490 possible destination blocks must be listed in the label list, otherwise this
3491 instruction has undefined behavior. This implies that jumps to labels
3492 defined in other functions have undefined behavior as well.</p>
3494 <h5>Implementation:</h5>
3496 <p>This is typically implemented with a jump through a register.</p>
3500 indirectbr i8* %Addr, [ label %bb1, label %bb2, label %bb3 ]
3506 <!-- _______________________________________________________________________ -->
3508 <a name="i_invoke">'<tt>invoke</tt>' Instruction</a>
3515 <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>]
3516 to label <normal label> unwind label <exception label>
3520 <p>The '<tt>invoke</tt>' instruction causes control to transfer to a specified
3521 function, with the possibility of control flow transfer to either the
3522 '<tt>normal</tt>' label or the '<tt>exception</tt>' label. If the callee
3523 function returns with the "<tt><a href="#i_ret">ret</a></tt>" instruction,
3524 control flow will return to the "normal" label. If the callee (or any
3525 indirect callees) returns via the "<a href="#i_resume"><tt>resume</tt></a>"
3526 instruction or other exception handling mechanism, control is interrupted and
3527 continued at the dynamically nearest "exception" label.</p>
3529 <p>The '<tt>exception</tt>' label is a
3530 <i><a href="ExceptionHandling.html#overview">landing pad</a></i> for the
3531 exception. As such, '<tt>exception</tt>' label is required to have the
3532 "<a href="#i_landingpad"><tt>landingpad</tt></a>" instruction, which contains
3533 the information about the behavior of the program after unwinding
3534 happens, as its first non-PHI instruction. The restrictions on the
3535 "<tt>landingpad</tt>" instruction's tightly couples it to the
3536 "<tt>invoke</tt>" instruction, so that the important information contained
3537 within the "<tt>landingpad</tt>" instruction can't be lost through normal
3541 <p>This instruction requires several arguments:</p>
3544 <li>The optional "cconv" marker indicates which <a href="#callingconv">calling
3545 convention</a> the call should use. If none is specified, the call
3546 defaults to using C calling conventions.</li>
3548 <li>The optional <a href="#paramattrs">Parameter Attributes</a> list for
3549 return values. Only '<tt>zeroext</tt>', '<tt>signext</tt>', and
3550 '<tt>inreg</tt>' attributes are valid here.</li>
3552 <li>'<tt>ptr to function ty</tt>': shall be the signature of the pointer to
3553 function value being invoked. In most cases, this is a direct function
3554 invocation, but indirect <tt>invoke</tt>s are just as possible, branching
3555 off an arbitrary pointer to function value.</li>
3557 <li>'<tt>function ptr val</tt>': An LLVM value containing a pointer to a
3558 function to be invoked. </li>
3560 <li>'<tt>function args</tt>': argument list whose types match the function
3561 signature argument types and parameter attributes. All arguments must be
3562 of <a href="#t_firstclass">first class</a> type. If the function
3563 signature indicates the function accepts a variable number of arguments,
3564 the extra arguments can be specified.</li>
3566 <li>'<tt>normal label</tt>': the label reached when the called function
3567 executes a '<tt><a href="#i_ret">ret</a></tt>' instruction. </li>
3569 <li>'<tt>exception label</tt>': the label reached when a callee returns via
3570 the <a href="#i_resume"><tt>resume</tt></a> instruction or other exception
3571 handling mechanism.</li>
3573 <li>The optional <a href="#fnattrs">function attributes</a> list. Only
3574 '<tt>noreturn</tt>', '<tt>nounwind</tt>', '<tt>readonly</tt>' and
3575 '<tt>readnone</tt>' attributes are valid here.</li>
3579 <p>This instruction is designed to operate as a standard
3580 '<tt><a href="#i_call">call</a></tt>' instruction in most regards. The
3581 primary difference is that it establishes an association with a label, which
3582 is used by the runtime library to unwind the stack.</p>
3584 <p>This instruction is used in languages with destructors to ensure that proper
3585 cleanup is performed in the case of either a <tt>longjmp</tt> or a thrown
3586 exception. Additionally, this is important for implementation of
3587 '<tt>catch</tt>' clauses in high-level languages that support them.</p>
3589 <p>For the purposes of the SSA form, the definition of the value returned by the
3590 '<tt>invoke</tt>' instruction is deemed to occur on the edge from the current
3591 block to the "normal" label. If the callee unwinds then no return value is
3596 %retval = invoke i32 @Test(i32 15) to label %Continue
3597 unwind label %TestCleanup <i>; {i32}:retval set</i>
3598 %retval = invoke <a href="#callingconv">coldcc</a> i32 %Testfnptr(i32 15) to label %Continue
3599 unwind label %TestCleanup <i>; {i32}:retval set</i>
3604 <!-- _______________________________________________________________________ -->
3607 <a name="i_resume">'<tt>resume</tt>' Instruction</a>
3614 resume <type> <value>
3618 <p>The '<tt>resume</tt>' instruction is a terminator instruction that has no
3622 <p>The '<tt>resume</tt>' instruction requires one argument, which must have the
3623 same type as the result of any '<tt>landingpad</tt>' instruction in the same
3627 <p>The '<tt>resume</tt>' instruction resumes propagation of an existing
3628 (in-flight) exception whose unwinding was interrupted with
3629 a <a href="#i_landingpad"><tt>landingpad</tt></a> instruction.</p>
3633 resume { i8*, i32 } %exn
3638 <!-- _______________________________________________________________________ -->
3641 <a name="i_unreachable">'<tt>unreachable</tt>' Instruction</a>
3652 <p>The '<tt>unreachable</tt>' instruction has no defined semantics. This
3653 instruction is used to inform the optimizer that a particular portion of the
3654 code is not reachable. This can be used to indicate that the code after a
3655 no-return function cannot be reached, and other facts.</p>
3658 <p>The '<tt>unreachable</tt>' instruction has no defined semantics.</p>
3664 <!-- ======================================================================= -->
3666 <a name="binaryops">Binary Operations</a>
3671 <p>Binary operators are used to do most of the computation in a program. They
3672 require two operands of the same type, execute an operation on them, and
3673 produce a single value. The operands might represent multiple data, as is
3674 the case with the <a href="#t_vector">vector</a> data type. The result value
3675 has the same type as its operands.</p>
3677 <p>There are several different binary operators:</p>
3679 <!-- _______________________________________________________________________ -->
3681 <a name="i_add">'<tt>add</tt>' Instruction</a>
3688 <result> = add <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3689 <result> = add nuw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3690 <result> = add nsw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3691 <result> = add nuw nsw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3695 <p>The '<tt>add</tt>' instruction returns the sum of its two operands.</p>
3698 <p>The two arguments to the '<tt>add</tt>' instruction must
3699 be <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
3700 integer values. Both arguments must have identical types.</p>
3703 <p>The value produced is the integer sum of the two operands.</p>
3705 <p>If the sum has unsigned overflow, the result returned is the mathematical
3706 result modulo 2<sup>n</sup>, where n is the bit width of the result.</p>
3708 <p>Because LLVM integers use a two's complement representation, this instruction
3709 is appropriate for both signed and unsigned integers.</p>
3711 <p><tt>nuw</tt> and <tt>nsw</tt> stand for "No Unsigned Wrap"
3712 and "No Signed Wrap", respectively. If the <tt>nuw</tt> and/or
3713 <tt>nsw</tt> keywords are present, the result value of the <tt>add</tt>
3714 is a <a href="#poisonvalues">poison value</a> if unsigned and/or signed overflow,
3715 respectively, occurs.</p>
3719 <result> = add i32 4, %var <i>; yields {i32}:result = 4 + %var</i>
3724 <!-- _______________________________________________________________________ -->
3726 <a name="i_fadd">'<tt>fadd</tt>' Instruction</a>
3733 <result> = fadd <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3737 <p>The '<tt>fadd</tt>' instruction returns the sum of its two operands.</p>
3740 <p>The two arguments to the '<tt>fadd</tt>' instruction must be
3741 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3742 floating point values. Both arguments must have identical types.</p>
3745 <p>The value produced is the floating point sum of the two operands.</p>
3749 <result> = fadd float 4.0, %var <i>; yields {float}:result = 4.0 + %var</i>
3754 <!-- _______________________________________________________________________ -->
3756 <a name="i_sub">'<tt>sub</tt>' Instruction</a>
3763 <result> = sub <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3764 <result> = sub nuw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3765 <result> = sub nsw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3766 <result> = sub nuw nsw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3770 <p>The '<tt>sub</tt>' instruction returns the difference of its two
3773 <p>Note that the '<tt>sub</tt>' instruction is used to represent the
3774 '<tt>neg</tt>' instruction present in most other intermediate
3775 representations.</p>
3778 <p>The two arguments to the '<tt>sub</tt>' instruction must
3779 be <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
3780 integer values. Both arguments must have identical types.</p>
3783 <p>The value produced is the integer difference of the two operands.</p>
3785 <p>If the difference has unsigned overflow, the result returned is the
3786 mathematical result modulo 2<sup>n</sup>, where n is the bit width of the
3789 <p>Because LLVM integers use a two's complement representation, this instruction
3790 is appropriate for both signed and unsigned integers.</p>
3792 <p><tt>nuw</tt> and <tt>nsw</tt> stand for "No Unsigned Wrap"
3793 and "No Signed Wrap", respectively. If the <tt>nuw</tt> and/or
3794 <tt>nsw</tt> keywords are present, the result value of the <tt>sub</tt>
3795 is a <a href="#poisonvalues">poison value</a> if unsigned and/or signed overflow,
3796 respectively, occurs.</p>
3800 <result> = sub i32 4, %var <i>; yields {i32}:result = 4 - %var</i>
3801 <result> = sub i32 0, %val <i>; yields {i32}:result = -%var</i>
3806 <!-- _______________________________________________________________________ -->
3808 <a name="i_fsub">'<tt>fsub</tt>' Instruction</a>
3815 <result> = fsub <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3819 <p>The '<tt>fsub</tt>' instruction returns the difference of its two
3822 <p>Note that the '<tt>fsub</tt>' instruction is used to represent the
3823 '<tt>fneg</tt>' instruction present in most other intermediate
3824 representations.</p>
3827 <p>The two arguments to the '<tt>fsub</tt>' instruction must be
3828 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3829 floating point values. Both arguments must have identical types.</p>
3832 <p>The value produced is the floating point difference of the two operands.</p>
3836 <result> = fsub float 4.0, %var <i>; yields {float}:result = 4.0 - %var</i>
3837 <result> = fsub float -0.0, %val <i>; yields {float}:result = -%var</i>
3842 <!-- _______________________________________________________________________ -->
3844 <a name="i_mul">'<tt>mul</tt>' Instruction</a>
3851 <result> = mul <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3852 <result> = mul nuw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3853 <result> = mul nsw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3854 <result> = mul nuw nsw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3858 <p>The '<tt>mul</tt>' instruction returns the product of its two operands.</p>
3861 <p>The two arguments to the '<tt>mul</tt>' instruction must
3862 be <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
3863 integer values. Both arguments must have identical types.</p>
3866 <p>The value produced is the integer product of the two operands.</p>
3868 <p>If the result of the multiplication has unsigned overflow, the result
3869 returned is the mathematical result modulo 2<sup>n</sup>, where n is the bit
3870 width of the result.</p>
3872 <p>Because LLVM integers use a two's complement representation, and the result
3873 is the same width as the operands, this instruction returns the correct
3874 result for both signed and unsigned integers. If a full product
3875 (e.g. <tt>i32</tt>x<tt>i32</tt>-><tt>i64</tt>) is needed, the operands should
3876 be sign-extended or zero-extended as appropriate to the width of the full
3879 <p><tt>nuw</tt> and <tt>nsw</tt> stand for "No Unsigned Wrap"
3880 and "No Signed Wrap", respectively. If the <tt>nuw</tt> and/or
3881 <tt>nsw</tt> keywords are present, the result value of the <tt>mul</tt>
3882 is a <a href="#poisonvalues">poison value</a> if unsigned and/or signed overflow,
3883 respectively, occurs.</p>
3887 <result> = mul i32 4, %var <i>; yields {i32}:result = 4 * %var</i>
3892 <!-- _______________________________________________________________________ -->
3894 <a name="i_fmul">'<tt>fmul</tt>' Instruction</a>
3901 <result> = fmul <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3905 <p>The '<tt>fmul</tt>' instruction returns the product of its two operands.</p>
3908 <p>The two arguments to the '<tt>fmul</tt>' instruction must be
3909 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3910 floating point values. Both arguments must have identical types.</p>
3913 <p>The value produced is the floating point product of the two operands.</p>
3917 <result> = fmul float 4.0, %var <i>; yields {float}:result = 4.0 * %var</i>
3922 <!-- _______________________________________________________________________ -->
3924 <a name="i_udiv">'<tt>udiv</tt>' Instruction</a>
3931 <result> = udiv <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3932 <result> = udiv exact <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3936 <p>The '<tt>udiv</tt>' instruction returns the quotient of its two operands.</p>
3939 <p>The two arguments to the '<tt>udiv</tt>' instruction must be
3940 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3941 values. Both arguments must have identical types.</p>
3944 <p>The value produced is the unsigned integer quotient of the two operands.</p>
3946 <p>Note that unsigned integer division and signed integer division are distinct
3947 operations; for signed integer division, use '<tt>sdiv</tt>'.</p>
3949 <p>Division by zero leads to undefined behavior.</p>
3951 <p>If the <tt>exact</tt> keyword is present, the result value of the
3952 <tt>udiv</tt> is a <a href="#poisonvalues">poison value</a> if %op1 is not a
3953 multiple of %op2 (as such, "((a udiv exact b) mul b) == a").</p>
3958 <result> = udiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
3963 <!-- _______________________________________________________________________ -->
3965 <a name="i_sdiv">'<tt>sdiv</tt>' Instruction</a>
3972 <result> = sdiv <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3973 <result> = sdiv exact <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3977 <p>The '<tt>sdiv</tt>' instruction returns the quotient of its two operands.</p>
3980 <p>The two arguments to the '<tt>sdiv</tt>' instruction must be
3981 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3982 values. Both arguments must have identical types.</p>
3985 <p>The value produced is the signed integer quotient of the two operands rounded
3988 <p>Note that signed integer division and unsigned integer division are distinct
3989 operations; for unsigned integer division, use '<tt>udiv</tt>'.</p>
3991 <p>Division by zero leads to undefined behavior. Overflow also leads to
3992 undefined behavior; this is a rare case, but can occur, for example, by doing
3993 a 32-bit division of -2147483648 by -1.</p>
3995 <p>If the <tt>exact</tt> keyword is present, the result value of the
3996 <tt>sdiv</tt> is a <a href="#poisonvalues">poison value</a> if the result would
4001 <result> = sdiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
4006 <!-- _______________________________________________________________________ -->
4008 <a name="i_fdiv">'<tt>fdiv</tt>' Instruction</a>
4015 <result> = fdiv <ty> <op1>, <op2> <i>; yields {ty}:result</i>
4019 <p>The '<tt>fdiv</tt>' instruction returns the quotient of its two operands.</p>
4022 <p>The two arguments to the '<tt>fdiv</tt>' instruction must be
4023 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
4024 floating point values. Both arguments must have identical types.</p>
4027 <p>The value produced is the floating point quotient of the two operands.</p>
4031 <result> = fdiv float 4.0, %var <i>; yields {float}:result = 4.0 / %var</i>
4036 <!-- _______________________________________________________________________ -->
4038 <a name="i_urem">'<tt>urem</tt>' Instruction</a>
4045 <result> = urem <ty> <op1>, <op2> <i>; yields {ty}:result</i>
4049 <p>The '<tt>urem</tt>' instruction returns the remainder from the unsigned
4050 division of its two arguments.</p>
4053 <p>The two arguments to the '<tt>urem</tt>' instruction must be
4054 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
4055 values. Both arguments must have identical types.</p>
4058 <p>This instruction returns the unsigned integer <i>remainder</i> of a division.
4059 This instruction always performs an unsigned division to get the
4062 <p>Note that unsigned integer remainder and signed integer remainder are
4063 distinct operations; for signed integer remainder, use '<tt>srem</tt>'.</p>
4065 <p>Taking the remainder of a division by zero leads to undefined behavior.</p>
4069 <result> = urem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
4074 <!-- _______________________________________________________________________ -->
4076 <a name="i_srem">'<tt>srem</tt>' Instruction</a>
4083 <result> = srem <ty> <op1>, <op2> <i>; yields {ty}:result</i>
4087 <p>The '<tt>srem</tt>' instruction returns the remainder from the signed
4088 division of its two operands. This instruction can also take
4089 <a href="#t_vector">vector</a> versions of the values in which case the
4090 elements must be integers.</p>
4093 <p>The two arguments to the '<tt>srem</tt>' instruction must be
4094 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
4095 values. Both arguments must have identical types.</p>
4098 <p>This instruction returns the <i>remainder</i> of a division (where the result
4099 is either zero or has the same sign as the dividend, <tt>op1</tt>), not the
4100 <i>modulo</i> operator (where the result is either zero or has the same sign
4101 as the divisor, <tt>op2</tt>) of a value.
4102 For more information about the difference,
4103 see <a href="http://mathforum.org/dr.math/problems/anne.4.28.99.html">The
4104 Math Forum</a>. For a table of how this is implemented in various languages,
4105 please see <a href="http://en.wikipedia.org/wiki/Modulo_operation">
4106 Wikipedia: modulo operation</a>.</p>
4108 <p>Note that signed integer remainder and unsigned integer remainder are
4109 distinct operations; for unsigned integer remainder, use '<tt>urem</tt>'.</p>
4111 <p>Taking the remainder of a division by zero leads to undefined behavior.
4112 Overflow also leads to undefined behavior; this is a rare case, but can
4113 occur, for example, by taking the remainder of a 32-bit division of
4114 -2147483648 by -1. (The remainder doesn't actually overflow, but this rule
4115 lets srem be implemented using instructions that return both the result of
4116 the division and the remainder.)</p>
4120 <result> = srem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
4125 <!-- _______________________________________________________________________ -->
4127 <a name="i_frem">'<tt>frem</tt>' Instruction</a>
4134 <result> = frem <ty> <op1>, <op2> <i>; yields {ty}:result</i>
4138 <p>The '<tt>frem</tt>' instruction returns the remainder from the division of
4139 its two operands.</p>
4142 <p>The two arguments to the '<tt>frem</tt>' instruction must be
4143 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
4144 floating point values. Both arguments must have identical types.</p>
4147 <p>This instruction returns the <i>remainder</i> of a division. The remainder
4148 has the same sign as the dividend.</p>
4152 <result> = frem float 4.0, %var <i>; yields {float}:result = 4.0 % %var</i>
4159 <!-- ======================================================================= -->
4161 <a name="bitwiseops">Bitwise Binary Operations</a>
4166 <p>Bitwise binary operators are used to do various forms of bit-twiddling in a
4167 program. They are generally very efficient instructions and can commonly be
4168 strength reduced from other instructions. They require two operands of the
4169 same type, execute an operation on them, and produce a single value. The
4170 resulting value is the same type as its operands.</p>
4172 <!-- _______________________________________________________________________ -->
4174 <a name="i_shl">'<tt>shl</tt>' Instruction</a>
4181 <result> = shl <ty> <op1>, <op2> <i>; yields {ty}:result</i>
4182 <result> = shl nuw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
4183 <result> = shl nsw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
4184 <result> = shl nuw nsw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
4188 <p>The '<tt>shl</tt>' instruction returns the first operand shifted to the left
4189 a specified number of bits.</p>
4192 <p>Both arguments to the '<tt>shl</tt>' instruction must be the
4193 same <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
4194 integer type. '<tt>op2</tt>' is treated as an unsigned value.</p>
4197 <p>The value produced is <tt>op1</tt> * 2<sup><tt>op2</tt></sup> mod
4198 2<sup>n</sup>, where <tt>n</tt> is the width of the result. If <tt>op2</tt>
4199 is (statically or dynamically) negative or equal to or larger than the number
4200 of bits in <tt>op1</tt>, the result is undefined. If the arguments are
4201 vectors, each vector element of <tt>op1</tt> is shifted by the corresponding
4202 shift amount in <tt>op2</tt>.</p>
4204 <p>If the <tt>nuw</tt> keyword is present, then the shift produces a
4205 <a href="#poisonvalues">poison value</a> if it shifts out any non-zero bits. If
4206 the <tt>nsw</tt> keyword is present, then the shift produces a
4207 <a href="#poisonvalues">poison value</a> if it shifts out any bits that disagree
4208 with the resultant sign bit. As such, NUW/NSW have the same semantics as
4209 they would if the shift were expressed as a mul instruction with the same
4210 nsw/nuw bits in (mul %op1, (shl 1, %op2)).</p>
4214 <result> = shl i32 4, %var <i>; yields {i32}: 4 << %var</i>
4215 <result> = shl i32 4, 2 <i>; yields {i32}: 16</i>
4216 <result> = shl i32 1, 10 <i>; yields {i32}: 1024</i>
4217 <result> = shl i32 1, 32 <i>; undefined</i>
4218 <result> = shl <2 x i32> < i32 1, i32 1>, < i32 1, i32 2> <i>; yields: result=<2 x i32> < i32 2, i32 4></i>
4223 <!-- _______________________________________________________________________ -->
4225 <a name="i_lshr">'<tt>lshr</tt>' Instruction</a>
4232 <result> = lshr <ty> <op1>, <op2> <i>; yields {ty}:result</i>
4233 <result> = lshr exact <ty> <op1>, <op2> <i>; yields {ty}:result</i>
4237 <p>The '<tt>lshr</tt>' instruction (logical shift right) returns the first
4238 operand shifted to the right a specified number of bits with zero fill.</p>
4241 <p>Both arguments to the '<tt>lshr</tt>' instruction must be the same
4242 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
4243 type. '<tt>op2</tt>' is treated as an unsigned value.</p>
4246 <p>This instruction always performs a logical shift right operation. The most
4247 significant bits of the result will be filled with zero bits after the shift.
4248 If <tt>op2</tt> is (statically or dynamically) equal to or larger than the
4249 number of bits in <tt>op1</tt>, the result is undefined. If the arguments are
4250 vectors, each vector element of <tt>op1</tt> is shifted by the corresponding
4251 shift amount in <tt>op2</tt>.</p>
4253 <p>If the <tt>exact</tt> keyword is present, the result value of the
4254 <tt>lshr</tt> is a <a href="#poisonvalues">poison value</a> if any of the bits
4255 shifted out are non-zero.</p>
4260 <result> = lshr i32 4, 1 <i>; yields {i32}:result = 2</i>
4261 <result> = lshr i32 4, 2 <i>; yields {i32}:result = 1</i>
4262 <result> = lshr i8 4, 3 <i>; yields {i8}:result = 0</i>
4263 <result> = lshr i8 -2, 1 <i>; yields {i8}:result = 0x7FFFFFFF </i>
4264 <result> = lshr i32 1, 32 <i>; undefined</i>
4265 <result> = lshr <2 x i32> < i32 -2, i32 4>, < i32 1, i32 2> <i>; yields: result=<2 x i32> < i32 0x7FFFFFFF, i32 1></i>
4270 <!-- _______________________________________________________________________ -->
4272 <a name="i_ashr">'<tt>ashr</tt>' Instruction</a>
4279 <result> = ashr <ty> <op1>, <op2> <i>; yields {ty}:result</i>
4280 <result> = ashr exact <ty> <op1>, <op2> <i>; yields {ty}:result</i>
4284 <p>The '<tt>ashr</tt>' instruction (arithmetic shift right) returns the first
4285 operand shifted to the right a specified number of bits with sign
4289 <p>Both arguments to the '<tt>ashr</tt>' instruction must be the same
4290 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
4291 type. '<tt>op2</tt>' is treated as an unsigned value.</p>
4294 <p>This instruction always performs an arithmetic shift right operation, The
4295 most significant bits of the result will be filled with the sign bit
4296 of <tt>op1</tt>. If <tt>op2</tt> is (statically or dynamically) equal to or
4297 larger than the number of bits in <tt>op1</tt>, the result is undefined. If
4298 the arguments are vectors, each vector element of <tt>op1</tt> is shifted by
4299 the corresponding shift amount in <tt>op2</tt>.</p>
4301 <p>If the <tt>exact</tt> keyword is present, the result value of the
4302 <tt>ashr</tt> is a <a href="#poisonvalues">poison value</a> if any of the bits
4303 shifted out are non-zero.</p>
4307 <result> = ashr i32 4, 1 <i>; yields {i32}:result = 2</i>
4308 <result> = ashr i32 4, 2 <i>; yields {i32}:result = 1</i>
4309 <result> = ashr i8 4, 3 <i>; yields {i8}:result = 0</i>
4310 <result> = ashr i8 -2, 1 <i>; yields {i8}:result = -1</i>
4311 <result> = ashr i32 1, 32 <i>; undefined</i>
4312 <result> = ashr <2 x i32> < i32 -2, i32 4>, < i32 1, i32 3> <i>; yields: result=<2 x i32> < i32 -1, i32 0></i>
4317 <!-- _______________________________________________________________________ -->
4319 <a name="i_and">'<tt>and</tt>' Instruction</a>
4326 <result> = and <ty> <op1>, <op2> <i>; yields {ty}:result</i>
4330 <p>The '<tt>and</tt>' instruction returns the bitwise logical and of its two
4334 <p>The two arguments to the '<tt>and</tt>' instruction must be
4335 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
4336 values. Both arguments must have identical types.</p>
4339 <p>The truth table used for the '<tt>and</tt>' instruction is:</p>
4341 <table border="1" cellspacing="0" cellpadding="4">
4373 <result> = and i32 4, %var <i>; yields {i32}:result = 4 & %var</i>
4374 <result> = and i32 15, 40 <i>; yields {i32}:result = 8</i>
4375 <result> = and i32 4, 8 <i>; yields {i32}:result = 0</i>
4378 <!-- _______________________________________________________________________ -->
4380 <a name="i_or">'<tt>or</tt>' Instruction</a>
4387 <result> = or <ty> <op1>, <op2> <i>; yields {ty}:result</i>
4391 <p>The '<tt>or</tt>' instruction returns the bitwise logical inclusive or of its
4395 <p>The two arguments to the '<tt>or</tt>' instruction must be
4396 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
4397 values. Both arguments must have identical types.</p>
4400 <p>The truth table used for the '<tt>or</tt>' instruction is:</p>
4402 <table border="1" cellspacing="0" cellpadding="4">
4434 <result> = or i32 4, %var <i>; yields {i32}:result = 4 | %var</i>
4435 <result> = or i32 15, 40 <i>; yields {i32}:result = 47</i>
4436 <result> = or i32 4, 8 <i>; yields {i32}:result = 12</i>
4441 <!-- _______________________________________________________________________ -->
4443 <a name="i_xor">'<tt>xor</tt>' Instruction</a>
4450 <result> = xor <ty> <op1>, <op2> <i>; yields {ty}:result</i>
4454 <p>The '<tt>xor</tt>' instruction returns the bitwise logical exclusive or of
4455 its two operands. The <tt>xor</tt> is used to implement the "one's
4456 complement" operation, which is the "~" operator in C.</p>
4459 <p>The two arguments to the '<tt>xor</tt>' instruction must be
4460 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
4461 values. Both arguments must have identical types.</p>
4464 <p>The truth table used for the '<tt>xor</tt>' instruction is:</p>
4466 <table border="1" cellspacing="0" cellpadding="4">
4498 <result> = xor i32 4, %var <i>; yields {i32}:result = 4 ^ %var</i>
4499 <result> = xor i32 15, 40 <i>; yields {i32}:result = 39</i>
4500 <result> = xor i32 4, 8 <i>; yields {i32}:result = 12</i>
4501 <result> = xor i32 %V, -1 <i>; yields {i32}:result = ~%V</i>
4508 <!-- ======================================================================= -->
4510 <a name="vectorops">Vector Operations</a>
4515 <p>LLVM supports several instructions to represent vector operations in a
4516 target-independent manner. These instructions cover the element-access and
4517 vector-specific operations needed to process vectors effectively. While LLVM
4518 does directly support these vector operations, many sophisticated algorithms
4519 will want to use target-specific intrinsics to take full advantage of a
4520 specific target.</p>
4522 <!-- _______________________________________________________________________ -->
4524 <a name="i_extractelement">'<tt>extractelement</tt>' Instruction</a>
4531 <result> = extractelement <n x <ty>> <val>, i32 <idx> <i>; yields <ty></i>
4535 <p>The '<tt>extractelement</tt>' instruction extracts a single scalar element
4536 from a vector at a specified index.</p>
4540 <p>The first operand of an '<tt>extractelement</tt>' instruction is a value
4541 of <a href="#t_vector">vector</a> type. The second operand is an index
4542 indicating the position from which to extract the element. The index may be
4546 <p>The result is a scalar of the same type as the element type of
4547 <tt>val</tt>. Its value is the value at position <tt>idx</tt> of
4548 <tt>val</tt>. If <tt>idx</tt> exceeds the length of <tt>val</tt>, the
4549 results are undefined.</p>
4553 <result> = extractelement <4 x i32> %vec, i32 0 <i>; yields i32</i>
4558 <!-- _______________________________________________________________________ -->
4560 <a name="i_insertelement">'<tt>insertelement</tt>' Instruction</a>
4567 <result> = insertelement <n x <ty>> <val>, <ty> <elt>, i32 <idx> <i>; yields <n x <ty>></i>
4571 <p>The '<tt>insertelement</tt>' instruction inserts a scalar element into a
4572 vector at a specified index.</p>
4575 <p>The first operand of an '<tt>insertelement</tt>' instruction is a value
4576 of <a href="#t_vector">vector</a> type. The second operand is a scalar value
4577 whose type must equal the element type of the first operand. The third
4578 operand is an index indicating the position at which to insert the value.
4579 The index may be a variable.</p>
4582 <p>The result is a vector of the same type as <tt>val</tt>. Its element values
4583 are those of <tt>val</tt> except at position <tt>idx</tt>, where it gets the
4584 value <tt>elt</tt>. If <tt>idx</tt> exceeds the length of <tt>val</tt>, the
4585 results are undefined.</p>
4589 <result> = insertelement <4 x i32> %vec, i32 1, i32 0 <i>; yields <4 x i32></i>
4594 <!-- _______________________________________________________________________ -->
4596 <a name="i_shufflevector">'<tt>shufflevector</tt>' Instruction</a>
4603 <result> = shufflevector <n x <ty>> <v1>, <n x <ty>> <v2>, <m x i32> <mask> <i>; yields <m x <ty>></i>
4607 <p>The '<tt>shufflevector</tt>' instruction constructs a permutation of elements
4608 from two input vectors, returning a vector with the same element type as the
4609 input and length that is the same as the shuffle mask.</p>
4612 <p>The first two operands of a '<tt>shufflevector</tt>' instruction are vectors
4613 with types that match each other. The third argument is a shuffle mask whose
4614 element type is always 'i32'. The result of the instruction is a vector
4615 whose length is the same as the shuffle mask and whose element type is the
4616 same as the element type of the first two operands.</p>
4618 <p>The shuffle mask operand is required to be a constant vector with either
4619 constant integer or undef values.</p>
4622 <p>The elements of the two input vectors are numbered from left to right across
4623 both of the vectors. The shuffle mask operand specifies, for each element of
4624 the result vector, which element of the two input vectors the result element
4625 gets. The element selector may be undef (meaning "don't care") and the
4626 second operand may be undef if performing a shuffle from only one vector.</p>
4630 <result> = shufflevector <4 x i32> %v1, <4 x i32> %v2,
4631 <4 x i32> <i32 0, i32 4, i32 1, i32 5> <i>; yields <4 x i32></i>
4632 <result> = shufflevector <4 x i32> %v1, <4 x i32> undef,
4633 <4 x i32> <i32 0, i32 1, i32 2, i32 3> <i>; yields <4 x i32></i> - Identity shuffle.
4634 <result> = shufflevector <8 x i32> %v1, <8 x i32> undef,
4635 <4 x i32> <i32 0, i32 1, i32 2, i32 3> <i>; yields <4 x i32></i>
4636 <result> = shufflevector <4 x i32> %v1, <4 x i32> %v2,
4637 <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>
4644 <!-- ======================================================================= -->
4646 <a name="aggregateops">Aggregate Operations</a>
4651 <p>LLVM supports several instructions for working with
4652 <a href="#t_aggregate">aggregate</a> values.</p>
4654 <!-- _______________________________________________________________________ -->
4656 <a name="i_extractvalue">'<tt>extractvalue</tt>' Instruction</a>
4663 <result> = extractvalue <aggregate type> <val>, <idx>{, <idx>}*
4667 <p>The '<tt>extractvalue</tt>' instruction extracts the value of a member field
4668 from an <a href="#t_aggregate">aggregate</a> value.</p>
4671 <p>The first operand of an '<tt>extractvalue</tt>' instruction is a value
4672 of <a href="#t_struct">struct</a> or
4673 <a href="#t_array">array</a> type. The operands are constant indices to
4674 specify which value to extract in a similar manner as indices in a
4675 '<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.</p>
4676 <p>The major differences to <tt>getelementptr</tt> indexing are:</p>
4678 <li>Since the value being indexed is not a pointer, the first index is
4679 omitted and assumed to be zero.</li>
4680 <li>At least one index must be specified.</li>
4681 <li>Not only struct indices but also array indices must be in
4686 <p>The result is the value at the position in the aggregate specified by the
4691 <result> = extractvalue {i32, float} %agg, 0 <i>; yields i32</i>
4696 <!-- _______________________________________________________________________ -->
4698 <a name="i_insertvalue">'<tt>insertvalue</tt>' Instruction</a>
4705 <result> = insertvalue <aggregate type> <val>, <ty> <elt>, <idx>{, <idx>}* <i>; yields <aggregate type></i>
4709 <p>The '<tt>insertvalue</tt>' instruction inserts a value into a member field
4710 in an <a href="#t_aggregate">aggregate</a> value.</p>
4713 <p>The first operand of an '<tt>insertvalue</tt>' instruction is a value
4714 of <a href="#t_struct">struct</a> or
4715 <a href="#t_array">array</a> type. The second operand is a first-class
4716 value to insert. The following operands are constant indices indicating
4717 the position at which to insert the value in a similar manner as indices in a
4718 '<tt><a href="#i_extractvalue">extractvalue</a></tt>' instruction. The
4719 value to insert must have the same type as the value identified by the
4723 <p>The result is an aggregate of the same type as <tt>val</tt>. Its value is
4724 that of <tt>val</tt> except that the value at the position specified by the
4725 indices is that of <tt>elt</tt>.</p>
4729 %agg1 = insertvalue {i32, float} undef, i32 1, 0 <i>; yields {i32 1, float undef}</i>
4730 %agg2 = insertvalue {i32, float} %agg1, float %val, 1 <i>; yields {i32 1, float %val}</i>
4731 %agg3 = insertvalue {i32, {float}} %agg1, float %val, 1, 0 <i>; yields {i32 1, float %val}</i>
4738 <!-- ======================================================================= -->
4740 <a name="memoryops">Memory Access and Addressing Operations</a>
4745 <p>A key design point of an SSA-based representation is how it represents
4746 memory. In LLVM, no memory locations are in SSA form, which makes things
4747 very simple. This section describes how to read, write, and allocate
4750 <!-- _______________________________________________________________________ -->
4752 <a name="i_alloca">'<tt>alloca</tt>' Instruction</a>
4759 <result> = alloca <type>[, <ty> <NumElements>][, align <alignment>] <i>; yields {type*}:result</i>
4763 <p>The '<tt>alloca</tt>' instruction allocates memory on the stack frame of the
4764 currently executing function, to be automatically released when this function
4765 returns to its caller. The object is always allocated in the generic address
4766 space (address space zero).</p>
4769 <p>The '<tt>alloca</tt>' instruction
4770 allocates <tt>sizeof(<type>)*NumElements</tt> bytes of memory on the
4771 runtime stack, returning a pointer of the appropriate type to the program.
4772 If "NumElements" is specified, it is the number of elements allocated,
4773 otherwise "NumElements" is defaulted to be one. If a constant alignment is
4774 specified, the value result of the allocation is guaranteed to be aligned to
4775 at least that boundary. If not specified, or if zero, the target can choose
4776 to align the allocation on any convenient boundary compatible with the
4779 <p>'<tt>type</tt>' may be any sized type.</p>
4782 <p>Memory is allocated; a pointer is returned. The operation is undefined if
4783 there is insufficient stack space for the allocation. '<tt>alloca</tt>'d
4784 memory is automatically released when the function returns. The
4785 '<tt>alloca</tt>' instruction is commonly used to represent automatic
4786 variables that must have an address available. When the function returns
4787 (either with the <tt><a href="#i_ret">ret</a></tt>
4788 or <tt><a href="#i_resume">resume</a></tt> instructions), the memory is
4789 reclaimed. Allocating zero bytes is legal, but the result is undefined.</p>
4793 %ptr = alloca i32 <i>; yields {i32*}:ptr</i>
4794 %ptr = alloca i32, i32 4 <i>; yields {i32*}:ptr</i>
4795 %ptr = alloca i32, i32 4, align 1024 <i>; yields {i32*}:ptr</i>
4796 %ptr = alloca i32, align 1024 <i>; yields {i32*}:ptr</i>
4801 <!-- _______________________________________________________________________ -->
4803 <a name="i_load">'<tt>load</tt>' Instruction</a>
4810 <result> = load [volatile] <ty>* <pointer>[, align <alignment>][, !nontemporal !<index>][, !invariant.load !<index>]
4811 <result> = load atomic [volatile] <ty>* <pointer> [singlethread] <ordering>, align <alignment>
4812 !<index> = !{ i32 1 }
4816 <p>The '<tt>load</tt>' instruction is used to read from memory.</p>
4819 <p>The argument to the '<tt>load</tt>' instruction specifies the memory address
4820 from which to load. The pointer must point to
4821 a <a href="#t_firstclass">first class</a> type. If the <tt>load</tt> is
4822 marked as <tt>volatile</tt>, then the optimizer is not allowed to modify the
4823 number or order of execution of this <tt>load</tt> with other <a
4824 href="#volatile">volatile operations</a>.</p>
4826 <p>If the <code>load</code> is marked as <code>atomic</code>, it takes an extra
4827 <a href="#ordering">ordering</a> and optional <code>singlethread</code>
4828 argument. The <code>release</code> and <code>acq_rel</code> orderings are
4829 not valid on <code>load</code> instructions. Atomic loads produce <a
4830 href="#memorymodel">defined</a> results when they may see multiple atomic
4831 stores. The type of the pointee must be an integer type whose bit width
4832 is a power of two greater than or equal to eight and less than or equal
4833 to a target-specific size limit. <code>align</code> must be explicitly
4834 specified on atomic loads, and the load has undefined behavior if the
4835 alignment is not set to a value which is at least the size in bytes of
4836 the pointee. <code>!nontemporal</code> does not have any defined semantics
4837 for atomic loads.</p>
4839 <p>The optional constant <tt>align</tt> argument specifies the alignment of the
4840 operation (that is, the alignment of the memory address). A value of 0 or an
4841 omitted <tt>align</tt> argument means that the operation has the preferential
4842 alignment for the target. It is the responsibility of the code emitter to
4843 ensure that the alignment information is correct. Overestimating the
4844 alignment results in undefined behavior. Underestimating the alignment may
4845 produce less efficient code. An alignment of 1 is always safe.</p>
4847 <p>The optional <tt>!nontemporal</tt> metadata must reference a single
4848 metatadata name <index> corresponding to a metadata node with
4849 one <tt>i32</tt> entry of value 1. The existence of
4850 the <tt>!nontemporal</tt> metatadata on the instruction tells the optimizer
4851 and code generator that this load is not expected to be reused in the cache.
4852 The code generator may select special instructions to save cache bandwidth,
4853 such as the <tt>MOVNT</tt> instruction on x86.</p>
4855 <p>The optional <tt>!invariant.load</tt> metadata must reference a single
4856 metatadata name <index> corresponding to a metadata node with no
4857 entries. The existence of the <tt>!invariant.load</tt> metatadata on the
4858 instruction tells the optimizer and code generator that this load address
4859 points to memory which does not change value during program execution.
4860 The optimizer may then move this load around, for example, by hoisting it
4861 out of loops using loop invariant code motion.</p>
4864 <p>The location of memory pointed to is loaded. If the value being loaded is of
4865 scalar type then the number of bytes read does not exceed the minimum number
4866 of bytes needed to hold all bits of the type. For example, loading an
4867 <tt>i24</tt> reads at most three bytes. When loading a value of a type like
4868 <tt>i20</tt> with a size that is not an integral number of bytes, the result
4869 is undefined if the value was not originally written using a store of the
4874 %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
4875 <a href="#i_store">store</a> i32 3, i32* %ptr <i>; yields {void}</i>
4876 %val = load i32* %ptr <i>; yields {i32}:val = i32 3</i>
4881 <!-- _______________________________________________________________________ -->
4883 <a name="i_store">'<tt>store</tt>' Instruction</a>
4890 store [volatile] <ty> <value>, <ty>* <pointer>[, align <alignment>][, !nontemporal !<index>] <i>; yields {void}</i>
4891 store atomic [volatile] <ty> <value>, <ty>* <pointer> [singlethread] <ordering>, align <alignment> <i>; yields {void}</i>
4895 <p>The '<tt>store</tt>' instruction is used to write to memory.</p>
4898 <p>There are two arguments to the '<tt>store</tt>' instruction: a value to store
4899 and an address at which to store it. The type of the
4900 '<tt><pointer></tt>' operand must be a pointer to
4901 the <a href="#t_firstclass">first class</a> type of the
4902 '<tt><value></tt>' operand. If the <tt>store</tt> is marked as
4903 <tt>volatile</tt>, then the optimizer is not allowed to modify the number or
4904 order of execution of this <tt>store</tt> with other <a
4905 href="#volatile">volatile operations</a>.</p>
4907 <p>If the <code>store</code> is marked as <code>atomic</code>, it takes an extra
4908 <a href="#ordering">ordering</a> and optional <code>singlethread</code>
4909 argument. The <code>acquire</code> and <code>acq_rel</code> orderings aren't
4910 valid on <code>store</code> instructions. Atomic loads produce <a
4911 href="#memorymodel">defined</a> results when they may see multiple atomic
4912 stores. The type of the pointee must be an integer type whose bit width
4913 is a power of two greater than or equal to eight and less than or equal
4914 to a target-specific size limit. <code>align</code> must be explicitly
4915 specified on atomic stores, and the store has undefined behavior if the
4916 alignment is not set to a value which is at least the size in bytes of
4917 the pointee. <code>!nontemporal</code> does not have any defined semantics
4918 for atomic stores.</p>
4920 <p>The optional constant "align" argument specifies the alignment of the
4921 operation (that is, the alignment of the memory address). A value of 0 or an
4922 omitted "align" argument means that the operation has the preferential
4923 alignment for the target. It is the responsibility of the code emitter to
4924 ensure that the alignment information is correct. Overestimating the
4925 alignment results in an undefined behavior. Underestimating the alignment may
4926 produce less efficient code. An alignment of 1 is always safe.</p>
4928 <p>The optional !nontemporal metadata must reference a single metatadata
4929 name <index> corresponding to a metadata node with one i32 entry of
4930 value 1. The existence of the !nontemporal metatadata on the
4931 instruction tells the optimizer and code generator that this load is
4932 not expected to be reused in the cache. The code generator may
4933 select special instructions to save cache bandwidth, such as the
4934 MOVNT instruction on x86.</p>
4938 <p>The contents of memory are updated to contain '<tt><value></tt>' at the
4939 location specified by the '<tt><pointer></tt>' operand. If
4940 '<tt><value></tt>' is of scalar type then the number of bytes written
4941 does not exceed the minimum number of bytes needed to hold all bits of the
4942 type. For example, storing an <tt>i24</tt> writes at most three bytes. When
4943 writing a value of a type like <tt>i20</tt> with a size that is not an
4944 integral number of bytes, it is unspecified what happens to the extra bits
4945 that do not belong to the type, but they will typically be overwritten.</p>
4949 %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
4950 store i32 3, i32* %ptr <i>; yields {void}</i>
4951 %val = <a href="#i_load">load</a> i32* %ptr <i>; yields {i32}:val = i32 3</i>
4956 <!-- _______________________________________________________________________ -->
4958 <a name="i_fence">'<tt>fence</tt>' Instruction</a>
4965 fence [singlethread] <ordering> <i>; yields {void}</i>
4969 <p>The '<tt>fence</tt>' instruction is used to introduce happens-before edges
4970 between operations.</p>
4972 <h5>Arguments:</h5> <p>'<code>fence</code>' instructions take an <a
4973 href="#ordering">ordering</a> argument which defines what
4974 <i>synchronizes-with</i> edges they add. They can only be given
4975 <code>acquire</code>, <code>release</code>, <code>acq_rel</code>, and
4976 <code>seq_cst</code> orderings.</p>
4979 <p>A fence <var>A</var> which has (at least) <code>release</code> ordering
4980 semantics <i>synchronizes with</i> a fence <var>B</var> with (at least)
4981 <code>acquire</code> ordering semantics if and only if there exist atomic
4982 operations <var>X</var> and <var>Y</var>, both operating on some atomic object
4983 <var>M</var>, such that <var>A</var> is sequenced before <var>X</var>,
4984 <var>X</var> modifies <var>M</var> (either directly or through some side effect
4985 of a sequence headed by <var>X</var>), <var>Y</var> is sequenced before
4986 <var>B</var>, and <var>Y</var> observes <var>M</var>. This provides a
4987 <i>happens-before</i> dependency between <var>A</var> and <var>B</var>. Rather
4988 than an explicit <code>fence</code>, one (but not both) of the atomic operations
4989 <var>X</var> or <var>Y</var> might provide a <code>release</code> or
4990 <code>acquire</code> (resp.) ordering constraint and still
4991 <i>synchronize-with</i> the explicit <code>fence</code> and establish the
4992 <i>happens-before</i> edge.</p>
4994 <p>A <code>fence</code> which has <code>seq_cst</code> ordering, in addition to
4995 having both <code>acquire</code> and <code>release</code> semantics specified
4996 above, participates in the global program order of other <code>seq_cst</code>
4997 operations and/or fences.</p>
4999 <p>The optional "<a href="#singlethread"><code>singlethread</code></a>" argument
5000 specifies that the fence only synchronizes with other fences in the same
5001 thread. (This is useful for interacting with signal handlers.)</p>
5005 fence acquire <i>; yields {void}</i>
5006 fence singlethread seq_cst <i>; yields {void}</i>
5011 <!-- _______________________________________________________________________ -->
5013 <a name="i_cmpxchg">'<tt>cmpxchg</tt>' Instruction</a>
5020 cmpxchg [volatile] <ty>* <pointer>, <ty> <cmp>, <ty> <new> [singlethread] <ordering> <i>; yields {ty}</i>
5024 <p>The '<tt>cmpxchg</tt>' instruction is used to atomically modify memory.
5025 It loads a value in memory and compares it to a given value. If they are
5026 equal, it stores a new value into the memory.</p>
5029 <p>There are three arguments to the '<code>cmpxchg</code>' instruction: an
5030 address to operate on, a value to compare to the value currently be at that
5031 address, and a new value to place at that address if the compared values are
5032 equal. The type of '<var><cmp></var>' must be an integer type whose
5033 bit width is a power of two greater than or equal to eight and less than
5034 or equal to a target-specific size limit. '<var><cmp></var>' and
5035 '<var><new></var>' must have the same type, and the type of
5036 '<var><pointer></var>' must be a pointer to that type. If the
5037 <code>cmpxchg</code> is marked as <code>volatile</code>, then the
5038 optimizer is not allowed to modify the number or order of execution
5039 of this <code>cmpxchg</code> with other <a href="#volatile">volatile
5042 <!-- FIXME: Extend allowed types. -->
5044 <p>The <a href="#ordering"><var>ordering</var></a> argument specifies how this
5045 <code>cmpxchg</code> synchronizes with other atomic operations.</p>
5047 <p>The optional "<code>singlethread</code>" argument declares that the
5048 <code>cmpxchg</code> is only atomic with respect to code (usually signal
5049 handlers) running in the same thread as the <code>cmpxchg</code>. Otherwise the
5050 cmpxchg is atomic with respect to all other code in the system.</p>
5052 <p>The pointer passed into cmpxchg must have alignment greater than or equal to
5053 the size in memory of the operand.
5056 <p>The contents of memory at the location specified by the
5057 '<tt><pointer></tt>' operand is read and compared to
5058 '<tt><cmp></tt>'; if the read value is the equal,
5059 '<tt><new></tt>' is written. The original value at the location
5062 <p>A successful <code>cmpxchg</code> is a read-modify-write instruction for the
5063 purpose of identifying <a href="#release_sequence">release sequences</a>. A
5064 failed <code>cmpxchg</code> is equivalent to an atomic load with an ordering
5065 parameter determined by dropping any <code>release</code> part of the
5066 <code>cmpxchg</code>'s ordering.</p>
5069 FIXME: Is compare_exchange_weak() necessary? (Consider after we've done
5070 optimization work on ARM.)
5072 FIXME: Is a weaker ordering constraint on failure helpful in practice?
5078 %orig = atomic <a href="#i_load">load</a> i32* %ptr unordered <i>; yields {i32}</i>
5079 <a href="#i_br">br</a> label %loop
5082 %cmp = <a href="#i_phi">phi</a> i32 [ %orig, %entry ], [%old, %loop]
5083 %squared = <a href="#i_mul">mul</a> i32 %cmp, %cmp
5084 %old = cmpxchg i32* %ptr, i32 %cmp, i32 %squared <i>; yields {i32}</i>
5085 %success = <a href="#i_icmp">icmp</a> eq i32 %cmp, %old
5086 <a href="#i_br">br</a> i1 %success, label %done, label %loop
5094 <!-- _______________________________________________________________________ -->
5096 <a name="i_atomicrmw">'<tt>atomicrmw</tt>' Instruction</a>
5103 atomicrmw [volatile] <operation> <ty>* <pointer>, <ty> <value> [singlethread] <ordering> <i>; yields {ty}</i>
5107 <p>The '<tt>atomicrmw</tt>' instruction is used to atomically modify memory.</p>
5110 <p>There are three arguments to the '<code>atomicrmw</code>' instruction: an
5111 operation to apply, an address whose value to modify, an argument to the
5112 operation. The operation must be one of the following keywords:</p>
5127 <p>The type of '<var><value></var>' must be an integer type whose
5128 bit width is a power of two greater than or equal to eight and less than
5129 or equal to a target-specific size limit. The type of the
5130 '<code><pointer></code>' operand must be a pointer to that type.
5131 If the <code>atomicrmw</code> is marked as <code>volatile</code>, then the
5132 optimizer is not allowed to modify the number or order of execution of this
5133 <code>atomicrmw</code> with other <a href="#volatile">volatile
5136 <!-- FIXME: Extend allowed types. -->
5139 <p>The contents of memory at the location specified by the
5140 '<tt><pointer></tt>' operand are atomically read, modified, and written
5141 back. The original value at the location is returned. The modification is
5142 specified by the <var>operation</var> argument:</p>
5145 <li>xchg: <code>*ptr = val</code></li>
5146 <li>add: <code>*ptr = *ptr + val</code></li>
5147 <li>sub: <code>*ptr = *ptr - val</code></li>
5148 <li>and: <code>*ptr = *ptr & val</code></li>
5149 <li>nand: <code>*ptr = ~(*ptr & val)</code></li>
5150 <li>or: <code>*ptr = *ptr | val</code></li>
5151 <li>xor: <code>*ptr = *ptr ^ val</code></li>
5152 <li>max: <code>*ptr = *ptr > val ? *ptr : val</code> (using a signed comparison)</li>
5153 <li>min: <code>*ptr = *ptr < val ? *ptr : val</code> (using a signed comparison)</li>
5154 <li>umax: <code>*ptr = *ptr > val ? *ptr : val</code> (using an unsigned comparison)</li>
5155 <li>umin: <code>*ptr = *ptr < val ? *ptr : val</code> (using an unsigned comparison)</li>
5160 %old = atomicrmw add i32* %ptr, i32 1 acquire <i>; yields {i32}</i>
5165 <!-- _______________________________________________________________________ -->
5167 <a name="i_getelementptr">'<tt>getelementptr</tt>' Instruction</a>
5174 <result> = getelementptr <pty>* <ptrval>{, <ty> <idx>}*
5175 <result> = getelementptr inbounds <pty>* <ptrval>{, <ty> <idx>}*
5176 <result> = getelementptr <ptr vector> ptrval, <vector index type> idx
5180 <p>The '<tt>getelementptr</tt>' instruction is used to get the address of a
5181 subelement of an <a href="#t_aggregate">aggregate</a> data structure.
5182 It performs address calculation only and does not access memory.</p>
5185 <p>The first argument is always a pointer or a vector of pointers,
5186 and forms the basis of the
5187 calculation. The remaining arguments are indices that indicate which of the
5188 elements of the aggregate object are indexed. The interpretation of each
5189 index is dependent on the type being indexed into. The first index always
5190 indexes the pointer value given as the first argument, the second index
5191 indexes a value of the type pointed to (not necessarily the value directly
5192 pointed to, since the first index can be non-zero), etc. The first type
5193 indexed into must be a pointer value, subsequent types can be arrays,
5194 vectors, and structs. Note that subsequent types being indexed into
5195 can never be pointers, since that would require loading the pointer before
5196 continuing calculation.</p>
5198 <p>The type of each index argument depends on the type it is indexing into.
5199 When indexing into a (optionally packed) structure, only <tt>i32</tt>
5200 integer <b>constants</b> are allowed. When indexing into an array, pointer
5201 or vector, integers of any width are allowed, and they are not required to be
5202 constant. These integers are treated as signed values where relevant.</p>
5204 <p>For example, let's consider a C code fragment and how it gets compiled to
5207 <pre class="doc_code">
5219 int *foo(struct ST *s) {
5220 return &s[1].Z.B[5][13];
5224 <p>The LLVM code generated by Clang is:</p>
5226 <pre class="doc_code">
5227 %struct.RT = <a href="#namedtypes">type</a> { i8, [10 x [20 x i32]], i8 }
5228 %struct.ST = <a href="#namedtypes">type</a> { i32, double, %struct.RT }
5230 define i32* @foo(%struct.ST* %s) nounwind uwtable readnone optsize ssp {
5232 %arrayidx = getelementptr inbounds %struct.ST* %s, i64 1, i32 2, i32 1, i64 5, i64 13
5238 <p>In the example above, the first index is indexing into the
5239 '<tt>%struct.ST*</tt>' type, which is a pointer, yielding a
5240 '<tt>%struct.ST</tt>' = '<tt>{ i32, double, %struct.RT }</tt>' type, a
5241 structure. The second index indexes into the third element of the structure,
5242 yielding a '<tt>%struct.RT</tt>' = '<tt>{ i8 , [10 x [20 x i32]], i8 }</tt>'
5243 type, another structure. The third index indexes into the second element of
5244 the structure, yielding a '<tt>[10 x [20 x i32]]</tt>' type, an array. The
5245 two dimensions of the array are subscripted into, yielding an '<tt>i32</tt>'
5246 type. The '<tt>getelementptr</tt>' instruction returns a pointer to this
5247 element, thus computing a value of '<tt>i32*</tt>' type.</p>
5249 <p>Note that it is perfectly legal to index partially through a structure,
5250 returning a pointer to an inner element. Because of this, the LLVM code for
5251 the given testcase is equivalent to:</p>
5253 <pre class="doc_code">
5254 define i32* @foo(%struct.ST* %s) {
5255 %t1 = getelementptr %struct.ST* %s, i32 1 <i>; yields %struct.ST*:%t1</i>
5256 %t2 = getelementptr %struct.ST* %t1, i32 0, i32 2 <i>; yields %struct.RT*:%t2</i>
5257 %t3 = getelementptr %struct.RT* %t2, i32 0, i32 1 <i>; yields [10 x [20 x i32]]*:%t3</i>
5258 %t4 = getelementptr [10 x [20 x i32]]* %t3, i32 0, i32 5 <i>; yields [20 x i32]*:%t4</i>
5259 %t5 = getelementptr [20 x i32]* %t4, i32 0, i32 13 <i>; yields i32*:%t5</i>
5264 <p>If the <tt>inbounds</tt> keyword is present, the result value of the
5265 <tt>getelementptr</tt> is a <a href="#poisonvalues">poison value</a> if the
5266 base pointer is not an <i>in bounds</i> address of an allocated object,
5267 or if any of the addresses that would be formed by successive addition of
5268 the offsets implied by the indices to the base address with infinitely
5269 precise signed arithmetic are not an <i>in bounds</i> address of that
5270 allocated object. The <i>in bounds</i> addresses for an allocated object
5271 are all the addresses that point into the object, plus the address one
5273 In cases where the base is a vector of pointers the <tt>inbounds</tt> keyword
5274 applies to each of the computations element-wise. </p>
5276 <p>If the <tt>inbounds</tt> keyword is not present, the offsets are added to
5277 the base address with silently-wrapping two's complement arithmetic. If the
5278 offsets have a different width from the pointer, they are sign-extended or
5279 truncated to the width of the pointer. The result value of the
5280 <tt>getelementptr</tt> may be outside the object pointed to by the base
5281 pointer. The result value may not necessarily be used to access memory
5282 though, even if it happens to point into allocated storage. See the
5283 <a href="#pointeraliasing">Pointer Aliasing Rules</a> section for more
5286 <p>The getelementptr instruction is often confusing. For some more insight into
5287 how it works, see <a href="GetElementPtr.html">the getelementptr FAQ</a>.</p>
5291 <i>; yields [12 x i8]*:aptr</i>
5292 %aptr = getelementptr {i32, [12 x i8]}* %saptr, i64 0, i32 1
5293 <i>; yields i8*:vptr</i>
5294 %vptr = getelementptr {i32, <2 x i8>}* %svptr, i64 0, i32 1, i32 1
5295 <i>; yields i8*:eptr</i>
5296 %eptr = getelementptr [12 x i8]* %aptr, i64 0, i32 1
5297 <i>; yields i32*:iptr</i>
5298 %iptr = getelementptr [10 x i32]* @arr, i16 0, i16 0
5301 <p>In cases where the pointer argument is a vector of pointers, only a
5302 single index may be used, and the number of vector elements has to be
5303 the same. For example: </p>
5304 <pre class="doc_code">
5305 %A = getelementptr <4 x i8*> %ptrs, <4 x i64> %offsets,
5312 <!-- ======================================================================= -->
5314 <a name="convertops">Conversion Operations</a>
5319 <p>The instructions in this category are the conversion instructions (casting)
5320 which all take a single operand and a type. They perform various bit
5321 conversions on the operand.</p>
5323 <!-- _______________________________________________________________________ -->
5325 <a name="i_trunc">'<tt>trunc .. to</tt>' Instruction</a>
5332 <result> = trunc <ty> <value> to <ty2> <i>; yields ty2</i>
5336 <p>The '<tt>trunc</tt>' instruction truncates its operand to the
5337 type <tt>ty2</tt>.</p>
5340 <p>The '<tt>trunc</tt>' instruction takes a value to trunc, and a type to trunc it to.
5341 Both types must be of <a href="#t_integer">integer</a> types, or vectors
5342 of the same number of integers.
5343 The bit size of the <tt>value</tt> must be larger than
5344 the bit size of the destination type, <tt>ty2</tt>.
5345 Equal sized types are not allowed.</p>
5348 <p>The '<tt>trunc</tt>' instruction truncates the high order bits
5349 in <tt>value</tt> and converts the remaining bits to <tt>ty2</tt>. Since the
5350 source size must be larger than the destination size, <tt>trunc</tt> cannot
5351 be a <i>no-op cast</i>. It will always truncate bits.</p>
5355 %X = trunc i32 257 to i8 <i>; yields i8:1</i>
5356 %Y = trunc i32 123 to i1 <i>; yields i1:true</i>
5357 %Z = trunc i32 122 to i1 <i>; yields i1:false</i>
5358 %W = trunc <2 x i16> <i16 8, i16 7> to <2 x i8> <i>; yields <i8 8, i8 7></i>
5363 <!-- _______________________________________________________________________ -->
5365 <a name="i_zext">'<tt>zext .. to</tt>' Instruction</a>
5372 <result> = zext <ty> <value> to <ty2> <i>; yields ty2</i>
5376 <p>The '<tt>zext</tt>' instruction zero extends its operand to type
5381 <p>The '<tt>zext</tt>' instruction takes a value to cast, and a type to cast it to.
5382 Both types must be of <a href="#t_integer">integer</a> types, or vectors
5383 of the same number of integers.
5384 The bit size of the <tt>value</tt> must be smaller than
5385 the bit size of the destination type,
5389 <p>The <tt>zext</tt> fills the high order bits of the <tt>value</tt> with zero
5390 bits until it reaches the size of the destination type, <tt>ty2</tt>.</p>
5392 <p>When zero extending from i1, the result will always be either 0 or 1.</p>
5396 %X = zext i32 257 to i64 <i>; yields i64:257</i>
5397 %Y = zext i1 true to i32 <i>; yields i32:1</i>
5398 %Z = zext <2 x i16> <i16 8, i16 7> to <2 x i32> <i>; yields <i32 8, i32 7></i>
5403 <!-- _______________________________________________________________________ -->
5405 <a name="i_sext">'<tt>sext .. to</tt>' Instruction</a>
5412 <result> = sext <ty> <value> to <ty2> <i>; yields ty2</i>
5416 <p>The '<tt>sext</tt>' sign extends <tt>value</tt> to the type <tt>ty2</tt>.</p>
5419 <p>The '<tt>sext</tt>' instruction takes a value to cast, and a type to cast it to.
5420 Both types must be of <a href="#t_integer">integer</a> types, or vectors
5421 of the same number of integers.
5422 The bit size of the <tt>value</tt> must be smaller than
5423 the bit size of the destination type,
5427 <p>The '<tt>sext</tt>' instruction performs a sign extension by copying the sign
5428 bit (highest order bit) of the <tt>value</tt> until it reaches the bit size
5429 of the type <tt>ty2</tt>.</p>
5431 <p>When sign extending from i1, the extension always results in -1 or 0.</p>
5435 %X = sext i8 -1 to i16 <i>; yields i16 :65535</i>
5436 %Y = sext i1 true to i32 <i>; yields i32:-1</i>
5437 %Z = sext <2 x i16> <i16 8, i16 7> to <2 x i32> <i>; yields <i32 8, i32 7></i>
5442 <!-- _______________________________________________________________________ -->
5444 <a name="i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a>
5451 <result> = fptrunc <ty> <value> to <ty2> <i>; yields ty2</i>
5455 <p>The '<tt>fptrunc</tt>' instruction truncates <tt>value</tt> to type
5459 <p>The '<tt>fptrunc</tt>' instruction takes a <a href="#t_floating">floating
5460 point</a> value to cast and a <a href="#t_floating">floating point</a> type
5461 to cast it to. The size of <tt>value</tt> must be larger than the size of
5462 <tt>ty2</tt>. This implies that <tt>fptrunc</tt> cannot be used to make a
5463 <i>no-op cast</i>.</p>
5466 <p>The '<tt>fptrunc</tt>' instruction truncates a <tt>value</tt> from a larger
5467 <a href="#t_floating">floating point</a> type to a smaller
5468 <a href="#t_floating">floating point</a> type. If the value cannot fit
5469 within the destination type, <tt>ty2</tt>, then the results are
5474 %X = fptrunc double 123.0 to float <i>; yields float:123.0</i>
5475 %Y = fptrunc double 1.0E+300 to float <i>; yields undefined</i>
5480 <!-- _______________________________________________________________________ -->
5482 <a name="i_fpext">'<tt>fpext .. to</tt>' Instruction</a>
5489 <result> = fpext <ty> <value> to <ty2> <i>; yields ty2</i>
5493 <p>The '<tt>fpext</tt>' extends a floating point <tt>value</tt> to a larger
5494 floating point value.</p>
5497 <p>The '<tt>fpext</tt>' instruction takes a
5498 <a href="#t_floating">floating point</a> <tt>value</tt> to cast, and
5499 a <a href="#t_floating">floating point</a> type to cast it to. The source
5500 type must be smaller than the destination type.</p>
5503 <p>The '<tt>fpext</tt>' instruction extends the <tt>value</tt> from a smaller
5504 <a href="#t_floating">floating point</a> type to a larger
5505 <a href="#t_floating">floating point</a> type. The <tt>fpext</tt> cannot be
5506 used to make a <i>no-op cast</i> because it always changes bits. Use
5507 <tt>bitcast</tt> to make a <i>no-op cast</i> for a floating point cast.</p>
5511 %X = fpext float 3.125 to double <i>; yields double:3.125000e+00</i>
5512 %Y = fpext double %X to fp128 <i>; yields fp128:0xL00000000000000004000900000000000</i>
5517 <!-- _______________________________________________________________________ -->
5519 <a name="i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a>
5526 <result> = fptoui <ty> <value> to <ty2> <i>; yields ty2</i>
5530 <p>The '<tt>fptoui</tt>' converts a floating point <tt>value</tt> to its
5531 unsigned integer equivalent of type <tt>ty2</tt>.</p>
5534 <p>The '<tt>fptoui</tt>' instruction takes a value to cast, which must be a
5535 scalar or vector <a href="#t_floating">floating point</a> value, and a type
5536 to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
5537 type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
5538 vector integer type with the same number of elements as <tt>ty</tt></p>
5541 <p>The '<tt>fptoui</tt>' instruction converts its
5542 <a href="#t_floating">floating point</a> operand into the nearest (rounding
5543 towards zero) unsigned integer value. If the value cannot fit
5544 in <tt>ty2</tt>, the results are undefined.</p>
5548 %X = fptoui double 123.0 to i32 <i>; yields i32:123</i>
5549 %Y = fptoui float 1.0E+300 to i1 <i>; yields undefined:1</i>
5550 %Z = fptoui float 1.04E+17 to i8 <i>; yields undefined:1</i>
5555 <!-- _______________________________________________________________________ -->
5557 <a name="i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a>
5564 <result> = fptosi <ty> <value> to <ty2> <i>; yields ty2</i>
5568 <p>The '<tt>fptosi</tt>' instruction converts
5569 <a href="#t_floating">floating point</a> <tt>value</tt> to
5570 type <tt>ty2</tt>.</p>
5573 <p>The '<tt>fptosi</tt>' instruction takes a value to cast, which must be a
5574 scalar or vector <a href="#t_floating">floating point</a> value, and a type
5575 to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
5576 type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
5577 vector integer type with the same number of elements as <tt>ty</tt></p>
5580 <p>The '<tt>fptosi</tt>' instruction converts its
5581 <a href="#t_floating">floating point</a> operand into the nearest (rounding
5582 towards zero) signed integer value. If the value cannot fit in <tt>ty2</tt>,
5583 the results are undefined.</p>
5587 %X = fptosi double -123.0 to i32 <i>; yields i32:-123</i>
5588 %Y = fptosi float 1.0E-247 to i1 <i>; yields undefined:1</i>
5589 %Z = fptosi float 1.04E+17 to i8 <i>; yields undefined:1</i>
5594 <!-- _______________________________________________________________________ -->
5596 <a name="i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a>
5603 <result> = uitofp <ty> <value> to <ty2> <i>; yields ty2</i>
5607 <p>The '<tt>uitofp</tt>' instruction regards <tt>value</tt> as an unsigned
5608 integer and converts that value to the <tt>ty2</tt> type.</p>
5611 <p>The '<tt>uitofp</tt>' instruction takes a value to cast, which must be a
5612 scalar or vector <a href="#t_integer">integer</a> value, and a type to cast
5613 it to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
5614 type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
5615 floating point type with the same number of elements as <tt>ty</tt></p>
5618 <p>The '<tt>uitofp</tt>' instruction interprets its operand as an unsigned
5619 integer quantity and converts it to the corresponding floating point
5620 value. If the value cannot fit in the floating point value, the results are
5625 %X = uitofp i32 257 to float <i>; yields float:257.0</i>
5626 %Y = uitofp i8 -1 to double <i>; yields double:255.0</i>
5631 <!-- _______________________________________________________________________ -->
5633 <a name="i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a>
5640 <result> = sitofp <ty> <value> to <ty2> <i>; yields ty2</i>
5644 <p>The '<tt>sitofp</tt>' instruction regards <tt>value</tt> as a signed integer
5645 and converts that value to the <tt>ty2</tt> type.</p>
5648 <p>The '<tt>sitofp</tt>' instruction takes a value to cast, which must be a
5649 scalar or vector <a href="#t_integer">integer</a> value, and a type to cast
5650 it to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
5651 type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
5652 floating point type with the same number of elements as <tt>ty</tt></p>
5655 <p>The '<tt>sitofp</tt>' instruction interprets its operand as a signed integer
5656 quantity and converts it to the corresponding floating point value. If the
5657 value cannot fit in the floating point value, the results are undefined.</p>
5661 %X = sitofp i32 257 to float <i>; yields float:257.0</i>
5662 %Y = sitofp i8 -1 to double <i>; yields double:-1.0</i>
5667 <!-- _______________________________________________________________________ -->
5669 <a name="i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a>
5676 <result> = ptrtoint <ty> <value> to <ty2> <i>; yields ty2</i>
5680 <p>The '<tt>ptrtoint</tt>' instruction converts the pointer or a vector of
5681 pointers <tt>value</tt> to
5682 the integer (or vector of integers) type <tt>ty2</tt>.</p>
5685 <p>The '<tt>ptrtoint</tt>' instruction takes a <tt>value</tt> to cast, which
5686 must be a a value of type <a href="#t_pointer">pointer</a> or a vector of
5687 pointers, and a type to cast it to
5688 <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a> or a vector
5689 of integers type.</p>
5692 <p>The '<tt>ptrtoint</tt>' instruction converts <tt>value</tt> to integer type
5693 <tt>ty2</tt> by interpreting the pointer value as an integer and either
5694 truncating or zero extending that value to the size of the integer type. If
5695 <tt>value</tt> is smaller than <tt>ty2</tt> then a zero extension is done. If
5696 <tt>value</tt> is larger than <tt>ty2</tt> then a truncation is done. If they
5697 are the same size, then nothing is done (<i>no-op cast</i>) other than a type
5702 %X = ptrtoint i32* %P to i8 <i>; yields truncation on 32-bit architecture</i>
5703 %Y = ptrtoint i32* %P to i64 <i>; yields zero extension on 32-bit architecture</i>
5704 %Z = ptrtoint <4 x i32*> %P to <4 x i64><i>; yields vector zero extension for a vector of addresses on 32-bit architecture</i>
5709 <!-- _______________________________________________________________________ -->
5711 <a name="i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a>
5718 <result> = inttoptr <ty> <value> to <ty2> <i>; yields ty2</i>
5722 <p>The '<tt>inttoptr</tt>' instruction converts an integer <tt>value</tt> to a
5723 pointer type, <tt>ty2</tt>.</p>
5726 <p>The '<tt>inttoptr</tt>' instruction takes an <a href="#t_integer">integer</a>
5727 value to cast, and a type to cast it to, which must be a
5728 <a href="#t_pointer">pointer</a> type.</p>
5731 <p>The '<tt>inttoptr</tt>' instruction converts <tt>value</tt> to type
5732 <tt>ty2</tt> by applying either a zero extension or a truncation depending on
5733 the size of the integer <tt>value</tt>. If <tt>value</tt> is larger than the
5734 size of a pointer then a truncation is done. If <tt>value</tt> is smaller
5735 than the size of a pointer then a zero extension is done. If they are the
5736 same size, nothing is done (<i>no-op cast</i>).</p>
5740 %X = inttoptr i32 255 to i32* <i>; yields zero extension on 64-bit architecture</i>
5741 %Y = inttoptr i32 255 to i32* <i>; yields no-op on 32-bit architecture</i>
5742 %Z = inttoptr i64 0 to i32* <i>; yields truncation on 32-bit architecture</i>
5743 %Z = inttoptr <4 x i32> %G to <4 x i8*><i>; yields truncation of vector G to four pointers</i>
5748 <!-- _______________________________________________________________________ -->
5750 <a name="i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a>
5757 <result> = bitcast <ty> <value> to <ty2> <i>; yields ty2</i>
5761 <p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
5762 <tt>ty2</tt> without changing any bits.</p>
5765 <p>The '<tt>bitcast</tt>' instruction takes a value to cast, which must be a
5766 non-aggregate first class value, and a type to cast it to, which must also be
5767 a non-aggregate <a href="#t_firstclass">first class</a> type. The bit sizes
5768 of <tt>value</tt> and the destination type, <tt>ty2</tt>, must be
5769 identical. If the source type is a pointer, the destination type must also be
5770 a pointer. This instruction supports bitwise conversion of vectors to
5771 integers and to vectors of other types (as long as they have the same
5775 <p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
5776 <tt>ty2</tt>. It is always a <i>no-op cast</i> because no bits change with
5777 this conversion. The conversion is done as if the <tt>value</tt> had been
5778 stored to memory and read back as type <tt>ty2</tt>.
5779 Pointer (or vector of pointers) types may only be converted to other pointer
5780 (or vector of pointers) types with this instruction. To convert
5781 pointers to other types, use the <a href="#i_inttoptr">inttoptr</a> or
5782 <a href="#i_ptrtoint">ptrtoint</a> instructions first.</p>
5786 %X = bitcast i8 255 to i8 <i>; yields i8 :-1</i>
5787 %Y = bitcast i32* %x to sint* <i>; yields sint*:%x</i>
5788 %Z = bitcast <2 x int> %V to i64; <i>; yields i64: %V</i>
5789 %Z = bitcast <2 x i32*> %V to <2 x i64*> <i>; yields <2 x i64*></i>
5796 <!-- ======================================================================= -->
5798 <a name="otherops">Other Operations</a>
5803 <p>The instructions in this category are the "miscellaneous" instructions, which
5804 defy better classification.</p>
5806 <!-- _______________________________________________________________________ -->
5808 <a name="i_icmp">'<tt>icmp</tt>' Instruction</a>
5815 <result> = icmp <cond> <ty> <op1>, <op2> <i>; yields {i1} or {<N x i1>}:result</i>
5819 <p>The '<tt>icmp</tt>' instruction returns a boolean value or a vector of
5820 boolean values based on comparison of its two integer, integer vector,
5821 pointer, or pointer vector operands.</p>
5824 <p>The '<tt>icmp</tt>' instruction takes three operands. The first operand is
5825 the condition code indicating the kind of comparison to perform. It is not a
5826 value, just a keyword. The possible condition code are:</p>
5829 <li><tt>eq</tt>: equal</li>
5830 <li><tt>ne</tt>: not equal </li>
5831 <li><tt>ugt</tt>: unsigned greater than</li>
5832 <li><tt>uge</tt>: unsigned greater or equal</li>
5833 <li><tt>ult</tt>: unsigned less than</li>
5834 <li><tt>ule</tt>: unsigned less or equal</li>
5835 <li><tt>sgt</tt>: signed greater than</li>
5836 <li><tt>sge</tt>: signed greater or equal</li>
5837 <li><tt>slt</tt>: signed less than</li>
5838 <li><tt>sle</tt>: signed less or equal</li>
5841 <p>The remaining two arguments must be <a href="#t_integer">integer</a> or
5842 <a href="#t_pointer">pointer</a> or integer <a href="#t_vector">vector</a>
5843 typed. They must also be identical types.</p>
5846 <p>The '<tt>icmp</tt>' compares <tt>op1</tt> and <tt>op2</tt> according to the
5847 condition code given as <tt>cond</tt>. The comparison performed always yields
5848 either an <a href="#t_integer"><tt>i1</tt></a> or vector of <tt>i1</tt>
5849 result, as follows:</p>
5852 <li><tt>eq</tt>: yields <tt>true</tt> if the operands are equal,
5853 <tt>false</tt> otherwise. No sign interpretation is necessary or
5856 <li><tt>ne</tt>: yields <tt>true</tt> if the operands are unequal,
5857 <tt>false</tt> otherwise. No sign interpretation is necessary or
5860 <li><tt>ugt</tt>: interprets the operands as unsigned values and yields
5861 <tt>true</tt> if <tt>op1</tt> is greater than <tt>op2</tt>.</li>
5863 <li><tt>uge</tt>: interprets the operands as unsigned values and yields
5864 <tt>true</tt> if <tt>op1</tt> is greater than or equal
5865 to <tt>op2</tt>.</li>
5867 <li><tt>ult</tt>: interprets the operands as unsigned values and yields
5868 <tt>true</tt> if <tt>op1</tt> is less than <tt>op2</tt>.</li>
5870 <li><tt>ule</tt>: interprets the operands as unsigned values and yields
5871 <tt>true</tt> if <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
5873 <li><tt>sgt</tt>: interprets the operands as signed values and yields
5874 <tt>true</tt> if <tt>op1</tt> is greater than <tt>op2</tt>.</li>
5876 <li><tt>sge</tt>: interprets the operands as signed values and yields
5877 <tt>true</tt> if <tt>op1</tt> is greater than or equal
5878 to <tt>op2</tt>.</li>
5880 <li><tt>slt</tt>: interprets the operands as signed values and yields
5881 <tt>true</tt> if <tt>op1</tt> is less than <tt>op2</tt>.</li>
5883 <li><tt>sle</tt>: interprets the operands as signed values and yields
5884 <tt>true</tt> if <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
5887 <p>If the operands are <a href="#t_pointer">pointer</a> typed, the pointer
5888 values are compared as if they were integers.</p>
5890 <p>If the operands are integer vectors, then they are compared element by
5891 element. The result is an <tt>i1</tt> vector with the same number of elements
5892 as the values being compared. Otherwise, the result is an <tt>i1</tt>.</p>
5896 <result> = icmp eq i32 4, 5 <i>; yields: result=false</i>
5897 <result> = icmp ne float* %X, %X <i>; yields: result=false</i>
5898 <result> = icmp ult i16 4, 5 <i>; yields: result=true</i>
5899 <result> = icmp sgt i16 4, 5 <i>; yields: result=false</i>
5900 <result> = icmp ule i16 -4, 5 <i>; yields: result=false</i>
5901 <result> = icmp sge i16 4, 5 <i>; yields: result=false</i>
5904 <p>Note that the code generator does not yet support vector types with
5905 the <tt>icmp</tt> instruction.</p>
5909 <!-- _______________________________________________________________________ -->
5911 <a name="i_fcmp">'<tt>fcmp</tt>' Instruction</a>
5918 <result> = fcmp <cond> <ty> <op1>, <op2> <i>; yields {i1} or {<N x i1>}:result</i>
5922 <p>The '<tt>fcmp</tt>' instruction returns a boolean value or vector of boolean
5923 values based on comparison of its operands.</p>
5925 <p>If the operands are floating point scalars, then the result type is a boolean
5926 (<a href="#t_integer"><tt>i1</tt></a>).</p>
5928 <p>If the operands are floating point vectors, then the result type is a vector
5929 of boolean with the same number of elements as the operands being
5933 <p>The '<tt>fcmp</tt>' instruction takes three operands. The first operand is
5934 the condition code indicating the kind of comparison to perform. It is not a
5935 value, just a keyword. The possible condition code are:</p>
5938 <li><tt>false</tt>: no comparison, always returns false</li>
5939 <li><tt>oeq</tt>: ordered and equal</li>
5940 <li><tt>ogt</tt>: ordered and greater than </li>
5941 <li><tt>oge</tt>: ordered and greater than or equal</li>
5942 <li><tt>olt</tt>: ordered and less than </li>
5943 <li><tt>ole</tt>: ordered and less than or equal</li>
5944 <li><tt>one</tt>: ordered and not equal</li>
5945 <li><tt>ord</tt>: ordered (no nans)</li>
5946 <li><tt>ueq</tt>: unordered or equal</li>
5947 <li><tt>ugt</tt>: unordered or greater than </li>
5948 <li><tt>uge</tt>: unordered or greater than or equal</li>
5949 <li><tt>ult</tt>: unordered or less than </li>
5950 <li><tt>ule</tt>: unordered or less than or equal</li>
5951 <li><tt>une</tt>: unordered or not equal</li>
5952 <li><tt>uno</tt>: unordered (either nans)</li>
5953 <li><tt>true</tt>: no comparison, always returns true</li>
5956 <p><i>Ordered</i> means that neither operand is a QNAN while
5957 <i>unordered</i> means that either operand may be a QNAN.</p>
5959 <p>Each of <tt>val1</tt> and <tt>val2</tt> arguments must be either
5960 a <a href="#t_floating">floating point</a> type or
5961 a <a href="#t_vector">vector</a> of floating point type. They must have
5962 identical types.</p>
5965 <p>The '<tt>fcmp</tt>' instruction compares <tt>op1</tt> and <tt>op2</tt>
5966 according to the condition code given as <tt>cond</tt>. If the operands are
5967 vectors, then the vectors are compared element by element. Each comparison
5968 performed always yields an <a href="#t_integer">i1</a> result, as
5972 <li><tt>false</tt>: always yields <tt>false</tt>, regardless of operands.</li>
5974 <li><tt>oeq</tt>: yields <tt>true</tt> if both operands are not a QNAN and
5975 <tt>op1</tt> is equal to <tt>op2</tt>.</li>
5977 <li><tt>ogt</tt>: yields <tt>true</tt> if both operands are not a QNAN and
5978 <tt>op1</tt> is greater than <tt>op2</tt>.</li>
5980 <li><tt>oge</tt>: yields <tt>true</tt> if both operands are not a QNAN and
5981 <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
5983 <li><tt>olt</tt>: yields <tt>true</tt> if both operands are not a QNAN and
5984 <tt>op1</tt> is less than <tt>op2</tt>.</li>
5986 <li><tt>ole</tt>: yields <tt>true</tt> if both operands are not a QNAN and
5987 <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
5989 <li><tt>one</tt>: yields <tt>true</tt> if both operands are not a QNAN and
5990 <tt>op1</tt> is not equal to <tt>op2</tt>.</li>
5992 <li><tt>ord</tt>: yields <tt>true</tt> if both operands are not a QNAN.</li>
5994 <li><tt>ueq</tt>: yields <tt>true</tt> if either operand is a QNAN or
5995 <tt>op1</tt> is equal to <tt>op2</tt>.</li>
5997 <li><tt>ugt</tt>: yields <tt>true</tt> if either operand is a QNAN or
5998 <tt>op1</tt> is greater than <tt>op2</tt>.</li>
6000 <li><tt>uge</tt>: yields <tt>true</tt> if either operand is a QNAN or
6001 <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
6003 <li><tt>ult</tt>: yields <tt>true</tt> if either operand is a QNAN or
6004 <tt>op1</tt> is less than <tt>op2</tt>.</li>
6006 <li><tt>ule</tt>: yields <tt>true</tt> if either operand is a QNAN or
6007 <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
6009 <li><tt>une</tt>: yields <tt>true</tt> if either operand is a QNAN or
6010 <tt>op1</tt> is not equal to <tt>op2</tt>.</li>
6012 <li><tt>uno</tt>: yields <tt>true</tt> if either operand is a QNAN.</li>
6014 <li><tt>true</tt>: always yields <tt>true</tt>, regardless of operands.</li>
6019 <result> = fcmp oeq float 4.0, 5.0 <i>; yields: result=false</i>
6020 <result> = fcmp one float 4.0, 5.0 <i>; yields: result=true</i>
6021 <result> = fcmp olt float 4.0, 5.0 <i>; yields: result=true</i>
6022 <result> = fcmp ueq double 1.0, 2.0 <i>; yields: result=false</i>
6025 <p>Note that the code generator does not yet support vector types with
6026 the <tt>fcmp</tt> instruction.</p>
6030 <!-- _______________________________________________________________________ -->
6032 <a name="i_phi">'<tt>phi</tt>' Instruction</a>
6039 <result> = phi <ty> [ <val0>, <label0>], ...
6043 <p>The '<tt>phi</tt>' instruction is used to implement the φ node in the
6044 SSA graph representing the function.</p>
6047 <p>The type of the incoming values is specified with the first type field. After
6048 this, the '<tt>phi</tt>' instruction takes a list of pairs as arguments, with
6049 one pair for each predecessor basic block of the current block. Only values
6050 of <a href="#t_firstclass">first class</a> type may be used as the value
6051 arguments to the PHI node. Only labels may be used as the label
6054 <p>There must be no non-phi instructions between the start of a basic block and
6055 the PHI instructions: i.e. PHI instructions must be first in a basic
6058 <p>For the purposes of the SSA form, the use of each incoming value is deemed to
6059 occur on the edge from the corresponding predecessor block to the current
6060 block (but after any definition of an '<tt>invoke</tt>' instruction's return
6061 value on the same edge).</p>
6064 <p>At runtime, the '<tt>phi</tt>' instruction logically takes on the value
6065 specified by the pair corresponding to the predecessor basic block that
6066 executed just prior to the current block.</p>
6070 Loop: ; Infinite loop that counts from 0 on up...
6071 %indvar = phi i32 [ 0, %LoopHeader ], [ %nextindvar, %Loop ]
6072 %nextindvar = add i32 %indvar, 1
6078 <!-- _______________________________________________________________________ -->
6080 <a name="i_select">'<tt>select</tt>' Instruction</a>
6087 <result> = select <i>selty</i> <cond>, <ty> <val1>, <ty> <val2> <i>; yields ty</i>
6089 <i>selty</i> is either i1 or {<N x i1>}
6093 <p>The '<tt>select</tt>' instruction is used to choose one value based on a
6094 condition, without branching.</p>
6098 <p>The '<tt>select</tt>' instruction requires an 'i1' value or a vector of 'i1'
6099 values indicating the condition, and two values of the
6100 same <a href="#t_firstclass">first class</a> type. If the val1/val2 are
6101 vectors and the condition is a scalar, then entire vectors are selected, not
6102 individual elements.</p>
6105 <p>If the condition is an i1 and it evaluates to 1, the instruction returns the
6106 first value argument; otherwise, it returns the second value argument.</p>
6108 <p>If the condition is a vector of i1, then the value arguments must be vectors
6109 of the same size, and the selection is done element by element.</p>
6113 %X = select i1 true, i8 17, i8 42 <i>; yields i8:17</i>
6118 <!-- _______________________________________________________________________ -->
6120 <a name="i_call">'<tt>call</tt>' Instruction</a>
6127 <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>]
6131 <p>The '<tt>call</tt>' instruction represents a simple function call.</p>
6134 <p>This instruction requires several arguments:</p>
6137 <li>The optional "tail" marker indicates that the callee function does not
6138 access any allocas or varargs in the caller. Note that calls may be
6139 marked "tail" even if they do not occur before
6140 a <a href="#i_ret"><tt>ret</tt></a> instruction. If the "tail" marker is
6141 present, the function call is eligible for tail call optimization,
6142 but <a href="CodeGenerator.html#tailcallopt">might not in fact be
6143 optimized into a jump</a>. The code generator may optimize calls marked
6144 "tail" with either 1) automatic <a href="CodeGenerator.html#sibcallopt">
6145 sibling call optimization</a> when the caller and callee have
6146 matching signatures, or 2) forced tail call optimization when the
6147 following extra requirements are met:
6149 <li>Caller and callee both have the calling
6150 convention <tt>fastcc</tt>.</li>
6151 <li>The call is in tail position (ret immediately follows call and ret
6152 uses value of call or is void).</li>
6153 <li>Option <tt>-tailcallopt</tt> is enabled,
6154 or <code>llvm::GuaranteedTailCallOpt</code> is <code>true</code>.</li>
6155 <li><a href="CodeGenerator.html#tailcallopt">Platform specific
6156 constraints are met.</a></li>
6160 <li>The optional "cconv" marker indicates which <a href="#callingconv">calling
6161 convention</a> the call should use. If none is specified, the call
6162 defaults to using C calling conventions. The calling convention of the
6163 call must match the calling convention of the target function, or else the
6164 behavior is undefined.</li>
6166 <li>The optional <a href="#paramattrs">Parameter Attributes</a> list for
6167 return values. Only '<tt>zeroext</tt>', '<tt>signext</tt>', and
6168 '<tt>inreg</tt>' attributes are valid here.</li>
6170 <li>'<tt>ty</tt>': the type of the call instruction itself which is also the
6171 type of the return value. Functions that return no value are marked
6172 <tt><a href="#t_void">void</a></tt>.</li>
6174 <li>'<tt>fnty</tt>': shall be the signature of the pointer to function value
6175 being invoked. The argument types must match the types implied by this
6176 signature. This type can be omitted if the function is not varargs and if
6177 the function type does not return a pointer to a function.</li>
6179 <li>'<tt>fnptrval</tt>': An LLVM value containing a pointer to a function to
6180 be invoked. In most cases, this is a direct function invocation, but
6181 indirect <tt>call</tt>s are just as possible, calling an arbitrary pointer
6182 to function value.</li>
6184 <li>'<tt>function args</tt>': argument list whose types match the function
6185 signature argument types and parameter attributes. All arguments must be
6186 of <a href="#t_firstclass">first class</a> type. If the function
6187 signature indicates the function accepts a variable number of arguments,
6188 the extra arguments can be specified.</li>
6190 <li>The optional <a href="#fnattrs">function attributes</a> list. Only
6191 '<tt>noreturn</tt>', '<tt>nounwind</tt>', '<tt>readonly</tt>' and
6192 '<tt>readnone</tt>' attributes are valid here.</li>
6196 <p>The '<tt>call</tt>' instruction is used to cause control flow to transfer to
6197 a specified function, with its incoming arguments bound to the specified
6198 values. Upon a '<tt><a href="#i_ret">ret</a></tt>' instruction in the called
6199 function, control flow continues with the instruction after the function
6200 call, and the return value of the function is bound to the result
6205 %retval = call i32 @test(i32 %argc)
6206 call i32 (i8*, ...)* @printf(i8* %msg, i32 12, i8 42) <i>; yields i32</i>
6207 %X = tail call i32 @foo() <i>; yields i32</i>
6208 %Y = tail call <a href="#callingconv">fastcc</a> i32 @foo() <i>; yields i32</i>
6209 call void %foo(i8 97 signext)
6211 %struct.A = type { i32, i8 }
6212 %r = call %struct.A @foo() <i>; yields { 32, i8 }</i>
6213 %gr = extractvalue %struct.A %r, 0 <i>; yields i32</i>
6214 %gr1 = extractvalue %struct.A %r, 1 <i>; yields i8</i>
6215 %Z = call void @foo() noreturn <i>; indicates that %foo never returns normally</i>
6216 %ZZ = call zeroext i32 @bar() <i>; Return value is %zero extended</i>
6219 <p>llvm treats calls to some functions with names and arguments that match the
6220 standard C99 library as being the C99 library functions, and may perform
6221 optimizations or generate code for them under that assumption. This is
6222 something we'd like to change in the future to provide better support for
6223 freestanding environments and non-C-based languages.</p>
6227 <!-- _______________________________________________________________________ -->
6229 <a name="i_va_arg">'<tt>va_arg</tt>' Instruction</a>
6236 <resultval> = va_arg <va_list*> <arglist>, <argty>
6240 <p>The '<tt>va_arg</tt>' instruction is used to access arguments passed through
6241 the "variable argument" area of a function call. It is used to implement the
6242 <tt>va_arg</tt> macro in C.</p>
6245 <p>This instruction takes a <tt>va_list*</tt> value and the type of the
6246 argument. It returns a value of the specified argument type and increments
6247 the <tt>va_list</tt> to point to the next argument. The actual type
6248 of <tt>va_list</tt> is target specific.</p>
6251 <p>The '<tt>va_arg</tt>' instruction loads an argument of the specified type
6252 from the specified <tt>va_list</tt> and causes the <tt>va_list</tt> to point
6253 to the next argument. For more information, see the variable argument
6254 handling <a href="#int_varargs">Intrinsic Functions</a>.</p>
6256 <p>It is legal for this instruction to be called in a function which does not
6257 take a variable number of arguments, for example, the <tt>vfprintf</tt>
6260 <p><tt>va_arg</tt> is an LLVM instruction instead of
6261 an <a href="#intrinsics">intrinsic function</a> because it takes a type as an
6265 <p>See the <a href="#int_varargs">variable argument processing</a> section.</p>
6267 <p>Note that the code generator does not yet fully support va_arg on many
6268 targets. Also, it does not currently support va_arg with aggregate types on
6273 <!-- _______________________________________________________________________ -->
6275 <a name="i_landingpad">'<tt>landingpad</tt>' Instruction</a>
6282 <resultval> = landingpad <resultty> personality <type> <pers_fn> <clause>+
6283 <resultval> = landingpad <resultty> personality <type> <pers_fn> cleanup <clause>*
6285 <clause> := catch <type> <value>
6286 <clause> := filter <array constant type> <array constant>
6290 <p>The '<tt>landingpad</tt>' instruction is used by
6291 <a href="ExceptionHandling.html#overview">LLVM's exception handling
6292 system</a> to specify that a basic block is a landing pad — one where
6293 the exception lands, and corresponds to the code found in the
6294 <i><tt>catch</tt></i> portion of a <i><tt>try/catch</tt></i> sequence. It
6295 defines values supplied by the personality function (<tt>pers_fn</tt>) upon
6296 re-entry to the function. The <tt>resultval</tt> has the
6297 type <tt>resultty</tt>.</p>
6300 <p>This instruction takes a <tt>pers_fn</tt> value. This is the personality
6301 function associated with the unwinding mechanism. The optional
6302 <tt>cleanup</tt> flag indicates that the landing pad block is a cleanup.</p>
6304 <p>A <tt>clause</tt> begins with the clause type — <tt>catch</tt>
6305 or <tt>filter</tt> — and contains the global variable representing the
6306 "type" that may be caught or filtered respectively. Unlike the
6307 <tt>catch</tt> clause, the <tt>filter</tt> clause takes an array constant as
6308 its argument. Use "<tt>[0 x i8**] undef</tt>" for a filter which cannot
6309 throw. The '<tt>landingpad</tt>' instruction must contain <em>at least</em>
6310 one <tt>clause</tt> or the <tt>cleanup</tt> flag.</p>
6313 <p>The '<tt>landingpad</tt>' instruction defines the values which are set by the
6314 personality function (<tt>pers_fn</tt>) upon re-entry to the function, and
6315 therefore the "result type" of the <tt>landingpad</tt> instruction. As with
6316 calling conventions, how the personality function results are represented in
6317 LLVM IR is target specific.</p>
6319 <p>The clauses are applied in order from top to bottom. If two
6320 <tt>landingpad</tt> instructions are merged together through inlining, the
6321 clauses from the calling function are appended to the list of clauses.
6322 When the call stack is being unwound due to an exception being thrown, the
6323 exception is compared against each <tt>clause</tt> in turn. If it doesn't
6324 match any of the clauses, and the <tt>cleanup</tt> flag is not set, then
6325 unwinding continues further up the call stack.</p>
6327 <p>The <tt>landingpad</tt> instruction has several restrictions:</p>
6330 <li>A landing pad block is a basic block which is the unwind destination of an
6331 '<tt>invoke</tt>' instruction.</li>
6332 <li>A landing pad block must have a '<tt>landingpad</tt>' instruction as its
6333 first non-PHI instruction.</li>
6334 <li>There can be only one '<tt>landingpad</tt>' instruction within the landing
6336 <li>A basic block that is not a landing pad block may not include a
6337 '<tt>landingpad</tt>' instruction.</li>
6338 <li>All '<tt>landingpad</tt>' instructions in a function must have the same
6339 personality function.</li>
6344 ;; A landing pad which can catch an integer.
6345 %res = landingpad { i8*, i32 } personality i32 (...)* @__gxx_personality_v0
6347 ;; A landing pad that is a cleanup.
6348 %res = landingpad { i8*, i32 } personality i32 (...)* @__gxx_personality_v0
6350 ;; A landing pad which can catch an integer and can only throw a double.
6351 %res = landingpad { i8*, i32 } personality i32 (...)* @__gxx_personality_v0
6353 filter [1 x i8**] [@_ZTId]
6362 <!-- *********************************************************************** -->
6363 <h2><a name="intrinsics">Intrinsic Functions</a></h2>
6364 <!-- *********************************************************************** -->
6368 <p>LLVM supports the notion of an "intrinsic function". These functions have
6369 well known names and semantics and are required to follow certain
6370 restrictions. Overall, these intrinsics represent an extension mechanism for
6371 the LLVM language that does not require changing all of the transformations
6372 in LLVM when adding to the language (or the bitcode reader/writer, the
6373 parser, etc...).</p>
6375 <p>Intrinsic function names must all start with an "<tt>llvm.</tt>" prefix. This
6376 prefix is reserved in LLVM for intrinsic names; thus, function names may not
6377 begin with this prefix. Intrinsic functions must always be external
6378 functions: you cannot define the body of intrinsic functions. Intrinsic
6379 functions may only be used in call or invoke instructions: it is illegal to
6380 take the address of an intrinsic function. Additionally, because intrinsic
6381 functions are part of the LLVM language, it is required if any are added that
6382 they be documented here.</p>
6384 <p>Some intrinsic functions can be overloaded, i.e., the intrinsic represents a
6385 family of functions that perform the same operation but on different data
6386 types. Because LLVM can represent over 8 million different integer types,
6387 overloading is used commonly to allow an intrinsic function to operate on any
6388 integer type. One or more of the argument types or the result type can be
6389 overloaded to accept any integer type. Argument types may also be defined as
6390 exactly matching a previous argument's type or the result type. This allows
6391 an intrinsic function which accepts multiple arguments, but needs all of them
6392 to be of the same type, to only be overloaded with respect to a single
6393 argument or the result.</p>
6395 <p>Overloaded intrinsics will have the names of its overloaded argument types
6396 encoded into its function name, each preceded by a period. Only those types
6397 which are overloaded result in a name suffix. Arguments whose type is matched
6398 against another type do not. For example, the <tt>llvm.ctpop</tt> function
6399 can take an integer of any width and returns an integer of exactly the same
6400 integer width. This leads to a family of functions such as
6401 <tt>i8 @llvm.ctpop.i8(i8 %val)</tt> and <tt>i29 @llvm.ctpop.i29(i29
6402 %val)</tt>. Only one type, the return type, is overloaded, and only one type
6403 suffix is required. Because the argument's type is matched against the return
6404 type, it does not require its own name suffix.</p>
6406 <p>To learn how to add an intrinsic function, please see the
6407 <a href="ExtendingLLVM.html">Extending LLVM Guide</a>.</p>
6409 <!-- ======================================================================= -->
6411 <a name="int_varargs">Variable Argument Handling Intrinsics</a>
6416 <p>Variable argument support is defined in LLVM with
6417 the <a href="#i_va_arg"><tt>va_arg</tt></a> instruction and these three
6418 intrinsic functions. These functions are related to the similarly named
6419 macros defined in the <tt><stdarg.h></tt> header file.</p>
6421 <p>All of these functions operate on arguments that use a target-specific value
6422 type "<tt>va_list</tt>". The LLVM assembly language reference manual does
6423 not define what this type is, so all transformations should be prepared to
6424 handle these functions regardless of the type used.</p>
6426 <p>This example shows how the <a href="#i_va_arg"><tt>va_arg</tt></a>
6427 instruction and the variable argument handling intrinsic functions are
6430 <pre class="doc_code">
6431 define i32 @test(i32 %X, ...) {
6432 ; Initialize variable argument processing
6434 %ap2 = bitcast i8** %ap to i8*
6435 call void @llvm.va_start(i8* %ap2)
6437 ; Read a single integer argument
6438 %tmp = va_arg i8** %ap, i32
6440 ; Demonstrate usage of llvm.va_copy and llvm.va_end
6442 %aq2 = bitcast i8** %aq to i8*
6443 call void @llvm.va_copy(i8* %aq2, i8* %ap2)
6444 call void @llvm.va_end(i8* %aq2)
6446 ; Stop processing of arguments.
6447 call void @llvm.va_end(i8* %ap2)
6451 declare void @llvm.va_start(i8*)
6452 declare void @llvm.va_copy(i8*, i8*)
6453 declare void @llvm.va_end(i8*)
6456 <!-- _______________________________________________________________________ -->
6458 <a name="int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a>
6466 declare void %llvm.va_start(i8* <arglist>)
6470 <p>The '<tt>llvm.va_start</tt>' intrinsic initializes <tt>*<arglist></tt>
6471 for subsequent use by <tt><a href="#i_va_arg">va_arg</a></tt>.</p>
6474 <p>The argument is a pointer to a <tt>va_list</tt> element to initialize.</p>
6477 <p>The '<tt>llvm.va_start</tt>' intrinsic works just like the <tt>va_start</tt>
6478 macro available in C. In a target-dependent way, it initializes
6479 the <tt>va_list</tt> element to which the argument points, so that the next
6480 call to <tt>va_arg</tt> will produce the first variable argument passed to
6481 the function. Unlike the C <tt>va_start</tt> macro, this intrinsic does not
6482 need to know the last argument of the function as the compiler can figure
6487 <!-- _______________________________________________________________________ -->
6489 <a name="int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a>
6496 declare void @llvm.va_end(i8* <arglist>)
6500 <p>The '<tt>llvm.va_end</tt>' intrinsic destroys <tt>*<arglist></tt>,
6501 which has been initialized previously
6502 with <tt><a href="#int_va_start">llvm.va_start</a></tt>
6503 or <tt><a href="#i_va_copy">llvm.va_copy</a></tt>.</p>
6506 <p>The argument is a pointer to a <tt>va_list</tt> to destroy.</p>
6509 <p>The '<tt>llvm.va_end</tt>' intrinsic works just like the <tt>va_end</tt>
6510 macro available in C. In a target-dependent way, it destroys
6511 the <tt>va_list</tt> element to which the argument points. Calls
6512 to <a href="#int_va_start"><tt>llvm.va_start</tt></a>
6513 and <a href="#int_va_copy"> <tt>llvm.va_copy</tt></a> must be matched exactly
6514 with calls to <tt>llvm.va_end</tt>.</p>
6518 <!-- _______________________________________________________________________ -->
6520 <a name="int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a>
6527 declare void @llvm.va_copy(i8* <destarglist>, i8* <srcarglist>)
6531 <p>The '<tt>llvm.va_copy</tt>' intrinsic copies the current argument position
6532 from the source argument list to the destination argument list.</p>
6535 <p>The first argument is a pointer to a <tt>va_list</tt> element to initialize.
6536 The second argument is a pointer to a <tt>va_list</tt> element to copy
6540 <p>The '<tt>llvm.va_copy</tt>' intrinsic works just like the <tt>va_copy</tt>
6541 macro available in C. In a target-dependent way, it copies the
6542 source <tt>va_list</tt> element into the destination <tt>va_list</tt>
6543 element. This intrinsic is necessary because
6544 the <tt><a href="#int_va_start"> llvm.va_start</a></tt> intrinsic may be
6545 arbitrarily complex and require, for example, memory allocation.</p>
6551 <!-- ======================================================================= -->
6553 <a name="int_gc">Accurate Garbage Collection Intrinsics</a>
6558 <p>LLVM support for <a href="GarbageCollection.html">Accurate Garbage
6559 Collection</a> (GC) requires the implementation and generation of these
6560 intrinsics. These intrinsics allow identification of <a href="#int_gcroot">GC
6561 roots on the stack</a>, as well as garbage collector implementations that
6562 require <a href="#int_gcread">read</a> and <a href="#int_gcwrite">write</a>
6563 barriers. Front-ends for type-safe garbage collected languages should generate
6564 these intrinsics to make use of the LLVM garbage collectors. For more details,
6565 see <a href="GarbageCollection.html">Accurate Garbage Collection with
6568 <p>The garbage collection intrinsics only operate on objects in the generic
6569 address space (address space zero).</p>
6571 <!-- _______________________________________________________________________ -->
6573 <a name="int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a>
6580 declare void @llvm.gcroot(i8** %ptrloc, i8* %metadata)
6584 <p>The '<tt>llvm.gcroot</tt>' intrinsic declares the existence of a GC root to
6585 the code generator, and allows some metadata to be associated with it.</p>
6588 <p>The first argument specifies the address of a stack object that contains the
6589 root pointer. The second pointer (which must be either a constant or a
6590 global value address) contains the meta-data to be associated with the
6594 <p>At runtime, a call to this intrinsic stores a null pointer into the "ptrloc"
6595 location. At compile-time, the code generator generates information to allow
6596 the runtime to find the pointer at GC safe points. The '<tt>llvm.gcroot</tt>'
6597 intrinsic may only be used in a function which <a href="#gc">specifies a GC
6602 <!-- _______________________________________________________________________ -->
6604 <a name="int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a>
6611 declare i8* @llvm.gcread(i8* %ObjPtr, i8** %Ptr)
6615 <p>The '<tt>llvm.gcread</tt>' intrinsic identifies reads of references from heap
6616 locations, allowing garbage collector implementations that require read
6620 <p>The second argument is the address to read from, which should be an address
6621 allocated from the garbage collector. The first object is a pointer to the
6622 start of the referenced object, if needed by the language runtime (otherwise
6626 <p>The '<tt>llvm.gcread</tt>' intrinsic has the same semantics as a load
6627 instruction, but may be replaced with substantially more complex code by the
6628 garbage collector runtime, as needed. The '<tt>llvm.gcread</tt>' intrinsic
6629 may only be used in a function which <a href="#gc">specifies a GC
6634 <!-- _______________________________________________________________________ -->
6636 <a name="int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a>
6643 declare void @llvm.gcwrite(i8* %P1, i8* %Obj, i8** %P2)
6647 <p>The '<tt>llvm.gcwrite</tt>' intrinsic identifies writes of references to heap
6648 locations, allowing garbage collector implementations that require write
6649 barriers (such as generational or reference counting collectors).</p>
6652 <p>The first argument is the reference to store, the second is the start of the
6653 object to store it to, and the third is the address of the field of Obj to
6654 store to. If the runtime does not require a pointer to the object, Obj may
6658 <p>The '<tt>llvm.gcwrite</tt>' intrinsic has the same semantics as a store
6659 instruction, but may be replaced with substantially more complex code by the
6660 garbage collector runtime, as needed. The '<tt>llvm.gcwrite</tt>' intrinsic
6661 may only be used in a function which <a href="#gc">specifies a GC
6668 <!-- ======================================================================= -->
6670 <a name="int_codegen">Code Generator Intrinsics</a>
6675 <p>These intrinsics are provided by LLVM to expose special features that may
6676 only be implemented with code generator support.</p>
6678 <!-- _______________________________________________________________________ -->
6680 <a name="int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a>
6687 declare i8 *@llvm.returnaddress(i32 <level>)
6691 <p>The '<tt>llvm.returnaddress</tt>' intrinsic attempts to compute a
6692 target-specific value indicating the return address of the current function
6693 or one of its callers.</p>
6696 <p>The argument to this intrinsic indicates which function to return the address
6697 for. Zero indicates the calling function, one indicates its caller, etc.
6698 The argument is <b>required</b> to be a constant integer value.</p>
6701 <p>The '<tt>llvm.returnaddress</tt>' intrinsic either returns a pointer
6702 indicating the return address of the specified call frame, or zero if it
6703 cannot be identified. The value returned by this intrinsic is likely to be
6704 incorrect or 0 for arguments other than zero, so it should only be used for
6705 debugging purposes.</p>
6707 <p>Note that calling this intrinsic does not prevent function inlining or other
6708 aggressive transformations, so the value returned may not be that of the
6709 obvious source-language caller.</p>
6713 <!-- _______________________________________________________________________ -->
6715 <a name="int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a>
6722 declare i8* @llvm.frameaddress(i32 <level>)
6726 <p>The '<tt>llvm.frameaddress</tt>' intrinsic attempts to return the
6727 target-specific frame pointer value for the specified stack frame.</p>
6730 <p>The argument to this intrinsic indicates which function to return the frame
6731 pointer for. Zero indicates the calling function, one indicates its caller,
6732 etc. The argument is <b>required</b> to be a constant integer value.</p>
6735 <p>The '<tt>llvm.frameaddress</tt>' intrinsic either returns a pointer
6736 indicating the frame address of the specified call frame, or zero if it
6737 cannot be identified. The value returned by this intrinsic is likely to be
6738 incorrect or 0 for arguments other than zero, so it should only be used for
6739 debugging purposes.</p>
6741 <p>Note that calling this intrinsic does not prevent function inlining or other
6742 aggressive transformations, so the value returned may not be that of the
6743 obvious source-language caller.</p>
6747 <!-- _______________________________________________________________________ -->
6749 <a name="int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a>
6756 declare i8* @llvm.stacksave()
6760 <p>The '<tt>llvm.stacksave</tt>' intrinsic is used to remember the current state
6761 of the function stack, for use
6762 with <a href="#int_stackrestore"> <tt>llvm.stackrestore</tt></a>. This is
6763 useful for implementing language features like scoped automatic variable
6764 sized arrays in C99.</p>
6767 <p>This intrinsic returns a opaque pointer value that can be passed
6768 to <a href="#int_stackrestore"><tt>llvm.stackrestore</tt></a>. When
6769 an <tt>llvm.stackrestore</tt> intrinsic is executed with a value saved
6770 from <tt>llvm.stacksave</tt>, it effectively restores the state of the stack
6771 to the state it was in when the <tt>llvm.stacksave</tt> intrinsic executed.
6772 In practice, this pops any <a href="#i_alloca">alloca</a> blocks from the
6773 stack that were allocated after the <tt>llvm.stacksave</tt> was executed.</p>
6777 <!-- _______________________________________________________________________ -->
6779 <a name="int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a>
6786 declare void @llvm.stackrestore(i8* %ptr)
6790 <p>The '<tt>llvm.stackrestore</tt>' intrinsic is used to restore the state of
6791 the function stack to the state it was in when the
6792 corresponding <a href="#int_stacksave"><tt>llvm.stacksave</tt></a> intrinsic
6793 executed. This is useful for implementing language features like scoped
6794 automatic variable sized arrays in C99.</p>
6797 <p>See the description
6798 for <a href="#int_stacksave"><tt>llvm.stacksave</tt></a>.</p>
6802 <!-- _______________________________________________________________________ -->
6804 <a name="int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a>
6811 declare void @llvm.prefetch(i8* <address>, i32 <rw>, i32 <locality>, i32 <cache type>)
6815 <p>The '<tt>llvm.prefetch</tt>' intrinsic is a hint to the code generator to
6816 insert a prefetch instruction if supported; otherwise, it is a noop.
6817 Prefetches have no effect on the behavior of the program but can change its
6818 performance characteristics.</p>
6821 <p><tt>address</tt> is the address to be prefetched, <tt>rw</tt> is the
6822 specifier determining if the fetch should be for a read (0) or write (1),
6823 and <tt>locality</tt> is a temporal locality specifier ranging from (0) - no
6824 locality, to (3) - extremely local keep in cache. The <tt>cache type</tt>
6825 specifies whether the prefetch is performed on the data (1) or instruction (0)
6826 cache. The <tt>rw</tt>, <tt>locality</tt> and <tt>cache type</tt> arguments
6827 must be constant integers.</p>
6830 <p>This intrinsic does not modify the behavior of the program. In particular,
6831 prefetches cannot trap and do not produce a value. On targets that support
6832 this intrinsic, the prefetch can provide hints to the processor cache for
6833 better performance.</p>
6837 <!-- _______________________________________________________________________ -->
6839 <a name="int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a>
6846 declare void @llvm.pcmarker(i32 <id>)
6850 <p>The '<tt>llvm.pcmarker</tt>' intrinsic is a method to export a Program
6851 Counter (PC) in a region of code to simulators and other tools. The method
6852 is target specific, but it is expected that the marker will use exported
6853 symbols to transmit the PC of the marker. The marker makes no guarantees
6854 that it will remain with any specific instruction after optimizations. It is
6855 possible that the presence of a marker will inhibit optimizations. The
6856 intended use is to be inserted after optimizations to allow correlations of
6857 simulation runs.</p>
6860 <p><tt>id</tt> is a numerical id identifying the marker.</p>
6863 <p>This intrinsic does not modify the behavior of the program. Backends that do
6864 not support this intrinsic may ignore it.</p>
6868 <!-- _______________________________________________________________________ -->
6870 <a name="int_readcyclecounter">'<tt>llvm.readcyclecounter</tt>' Intrinsic</a>
6877 declare i64 @llvm.readcyclecounter()
6881 <p>The '<tt>llvm.readcyclecounter</tt>' intrinsic provides access to the cycle
6882 counter register (or similar low latency, high accuracy clocks) on those
6883 targets that support it. On X86, it should map to RDTSC. On Alpha, it
6884 should map to RPCC. As the backing counters overflow quickly (on the order
6885 of 9 seconds on alpha), this should only be used for small timings.</p>
6888 <p>When directly supported, reading the cycle counter should not modify any
6889 memory. Implementations are allowed to either return a application specific
6890 value or a system wide value. On backends without support, this is lowered
6891 to a constant 0.</p>
6897 <!-- ======================================================================= -->
6899 <a name="int_libc">Standard C Library Intrinsics</a>
6904 <p>LLVM provides intrinsics for a few important standard C library functions.
6905 These intrinsics allow source-language front-ends to pass information about
6906 the alignment of the pointer arguments to the code generator, providing
6907 opportunity for more efficient code generation.</p>
6909 <!-- _______________________________________________________________________ -->
6911 <a name="int_memcpy">'<tt>llvm.memcpy</tt>' Intrinsic</a>
6917 <p>This is an overloaded intrinsic. You can use <tt>llvm.memcpy</tt> on any
6918 integer bit width and for different address spaces. Not all targets support
6919 all bit widths however.</p>
6922 declare void @llvm.memcpy.p0i8.p0i8.i32(i8* <dest>, i8* <src>,
6923 i32 <len>, i32 <align>, i1 <isvolatile>)
6924 declare void @llvm.memcpy.p0i8.p0i8.i64(i8* <dest>, i8* <src>,
6925 i64 <len>, i32 <align>, i1 <isvolatile>)
6929 <p>The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the
6930 source location to the destination location.</p>
6932 <p>Note that, unlike the standard libc function, the <tt>llvm.memcpy.*</tt>
6933 intrinsics do not return a value, takes extra alignment/isvolatile arguments
6934 and the pointers can be in specified address spaces.</p>
6938 <p>The first argument is a pointer to the destination, the second is a pointer
6939 to the source. The third argument is an integer argument specifying the
6940 number of bytes to copy, the fourth argument is the alignment of the
6941 source and destination locations, and the fifth is a boolean indicating a
6942 volatile access.</p>
6944 <p>If the call to this intrinsic has an alignment value that is not 0 or 1,
6945 then the caller guarantees that both the source and destination pointers are
6946 aligned to that boundary.</p>
6948 <p>If the <tt>isvolatile</tt> parameter is <tt>true</tt>, the
6949 <tt>llvm.memcpy</tt> call is a <a href="#volatile">volatile operation</a>.
6950 The detailed access behavior is not very cleanly specified and it is unwise
6951 to depend on it.</p>
6955 <p>The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the
6956 source location to the destination location, which are not allowed to
6957 overlap. It copies "len" bytes of memory over. If the argument is known to
6958 be aligned to some boundary, this can be specified as the fourth argument,
6959 otherwise it should be set to 0 or 1.</p>
6963 <!-- _______________________________________________________________________ -->
6965 <a name="int_memmove">'<tt>llvm.memmove</tt>' Intrinsic</a>
6971 <p>This is an overloaded intrinsic. You can use llvm.memmove on any integer bit
6972 width and for different address space. Not all targets support all bit
6976 declare void @llvm.memmove.p0i8.p0i8.i32(i8* <dest>, i8* <src>,
6977 i32 <len>, i32 <align>, i1 <isvolatile>)
6978 declare void @llvm.memmove.p0i8.p0i8.i64(i8* <dest>, i8* <src>,
6979 i64 <len>, i32 <align>, i1 <isvolatile>)
6983 <p>The '<tt>llvm.memmove.*</tt>' intrinsics move a block of memory from the
6984 source location to the destination location. It is similar to the
6985 '<tt>llvm.memcpy</tt>' intrinsic but allows the two memory locations to
6988 <p>Note that, unlike the standard libc function, the <tt>llvm.memmove.*</tt>
6989 intrinsics do not return a value, takes extra alignment/isvolatile arguments
6990 and the pointers can be in specified address spaces.</p>
6994 <p>The first argument is a pointer to the destination, the second is a pointer
6995 to the source. The third argument is an integer argument specifying the
6996 number of bytes to copy, the fourth argument is the alignment of the
6997 source and destination locations, and the fifth is a boolean indicating a
6998 volatile access.</p>
7000 <p>If the call to this intrinsic has an alignment value that is not 0 or 1,
7001 then the caller guarantees that the source and destination pointers are
7002 aligned to that boundary.</p>
7004 <p>If the <tt>isvolatile</tt> parameter is <tt>true</tt>, the
7005 <tt>llvm.memmove</tt> call is a <a href="#volatile">volatile operation</a>.
7006 The detailed access behavior is not very cleanly specified and it is unwise
7007 to depend on it.</p>
7011 <p>The '<tt>llvm.memmove.*</tt>' intrinsics copy a block of memory from the
7012 source location to the destination location, which may overlap. It copies
7013 "len" bytes of memory over. If the argument is known to be aligned to some
7014 boundary, this can be specified as the fourth argument, otherwise it should
7015 be set to 0 or 1.</p>
7019 <!-- _______________________________________________________________________ -->
7021 <a name="int_memset">'<tt>llvm.memset.*</tt>' Intrinsics</a>
7027 <p>This is an overloaded intrinsic. You can use llvm.memset on any integer bit
7028 width and for different address spaces. However, not all targets support all
7032 declare void @llvm.memset.p0i8.i32(i8* <dest>, i8 <val>,
7033 i32 <len>, i32 <align>, i1 <isvolatile>)
7034 declare void @llvm.memset.p0i8.i64(i8* <dest>, i8 <val>,
7035 i64 <len>, i32 <align>, i1 <isvolatile>)
7039 <p>The '<tt>llvm.memset.*</tt>' intrinsics fill a block of memory with a
7040 particular byte value.</p>
7042 <p>Note that, unlike the standard libc function, the <tt>llvm.memset</tt>
7043 intrinsic does not return a value and takes extra alignment/volatile
7044 arguments. Also, the destination can be in an arbitrary address space.</p>
7047 <p>The first argument is a pointer to the destination to fill, the second is the
7048 byte value with which to fill it, the third argument is an integer argument
7049 specifying the number of bytes to fill, and the fourth argument is the known
7050 alignment of the destination location.</p>
7052 <p>If the call to this intrinsic has an alignment value that is not 0 or 1,
7053 then the caller guarantees that the destination pointer is aligned to that
7056 <p>If the <tt>isvolatile</tt> parameter is <tt>true</tt>, the
7057 <tt>llvm.memset</tt> call is a <a href="#volatile">volatile operation</a>.
7058 The detailed access behavior is not very cleanly specified and it is unwise
7059 to depend on it.</p>
7062 <p>The '<tt>llvm.memset.*</tt>' intrinsics fill "len" bytes of memory starting
7063 at the destination location. If the argument is known to be aligned to some
7064 boundary, this can be specified as the fourth argument, otherwise it should
7065 be set to 0 or 1.</p>
7069 <!-- _______________________________________________________________________ -->
7071 <a name="int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a>
7077 <p>This is an overloaded intrinsic. You can use <tt>llvm.sqrt</tt> on any
7078 floating point or vector of floating point type. Not all targets support all
7082 declare float @llvm.sqrt.f32(float %Val)
7083 declare double @llvm.sqrt.f64(double %Val)
7084 declare x86_fp80 @llvm.sqrt.f80(x86_fp80 %Val)
7085 declare fp128 @llvm.sqrt.f128(fp128 %Val)
7086 declare ppc_fp128 @llvm.sqrt.ppcf128(ppc_fp128 %Val)
7090 <p>The '<tt>llvm.sqrt</tt>' intrinsics return the sqrt of the specified operand,
7091 returning the same value as the libm '<tt>sqrt</tt>' functions would.
7092 Unlike <tt>sqrt</tt> in libm, however, <tt>llvm.sqrt</tt> has undefined
7093 behavior for negative numbers other than -0.0 (which allows for better
7094 optimization, because there is no need to worry about errno being
7095 set). <tt>llvm.sqrt(-0.0)</tt> is defined to return -0.0 like IEEE sqrt.</p>
7098 <p>The argument and return value are floating point numbers of the same
7102 <p>This function returns the sqrt of the specified operand if it is a
7103 nonnegative floating point number.</p>
7107 <!-- _______________________________________________________________________ -->
7109 <a name="int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a>
7115 <p>This is an overloaded intrinsic. You can use <tt>llvm.powi</tt> on any
7116 floating point or vector of floating point type. Not all targets support all
7120 declare float @llvm.powi.f32(float %Val, i32 %power)
7121 declare double @llvm.powi.f64(double %Val, i32 %power)
7122 declare x86_fp80 @llvm.powi.f80(x86_fp80 %Val, i32 %power)
7123 declare fp128 @llvm.powi.f128(fp128 %Val, i32 %power)
7124 declare ppc_fp128 @llvm.powi.ppcf128(ppc_fp128 %Val, i32 %power)
7128 <p>The '<tt>llvm.powi.*</tt>' intrinsics return the first operand raised to the
7129 specified (positive or negative) power. The order of evaluation of
7130 multiplications is not defined. When a vector of floating point type is
7131 used, the second argument remains a scalar integer value.</p>
7134 <p>The second argument is an integer power, and the first is a value to raise to
7138 <p>This function returns the first value raised to the second power with an
7139 unspecified sequence of rounding operations.</p>
7143 <!-- _______________________________________________________________________ -->
7145 <a name="int_sin">'<tt>llvm.sin.*</tt>' Intrinsic</a>
7151 <p>This is an overloaded intrinsic. You can use <tt>llvm.sin</tt> on any
7152 floating point or vector of floating point type. Not all targets support all
7156 declare float @llvm.sin.f32(float %Val)
7157 declare double @llvm.sin.f64(double %Val)
7158 declare x86_fp80 @llvm.sin.f80(x86_fp80 %Val)
7159 declare fp128 @llvm.sin.f128(fp128 %Val)
7160 declare ppc_fp128 @llvm.sin.ppcf128(ppc_fp128 %Val)
7164 <p>The '<tt>llvm.sin.*</tt>' intrinsics return the sine of the operand.</p>
7167 <p>The argument and return value are floating point numbers of the same
7171 <p>This function returns the sine of the specified operand, returning the same
7172 values as the libm <tt>sin</tt> functions would, and handles error conditions
7173 in the same way.</p>
7177 <!-- _______________________________________________________________________ -->
7179 <a name="int_cos">'<tt>llvm.cos.*</tt>' Intrinsic</a>
7185 <p>This is an overloaded intrinsic. You can use <tt>llvm.cos</tt> on any
7186 floating point or vector of floating point type. Not all targets support all
7190 declare float @llvm.cos.f32(float %Val)
7191 declare double @llvm.cos.f64(double %Val)
7192 declare x86_fp80 @llvm.cos.f80(x86_fp80 %Val)
7193 declare fp128 @llvm.cos.f128(fp128 %Val)
7194 declare ppc_fp128 @llvm.cos.ppcf128(ppc_fp128 %Val)
7198 <p>The '<tt>llvm.cos.*</tt>' intrinsics return the cosine of the operand.</p>
7201 <p>The argument and return value are floating point numbers of the same
7205 <p>This function returns the cosine of the specified operand, returning the same
7206 values as the libm <tt>cos</tt> functions would, and handles error conditions
7207 in the same way.</p>
7211 <!-- _______________________________________________________________________ -->
7213 <a name="int_pow">'<tt>llvm.pow.*</tt>' Intrinsic</a>
7219 <p>This is an overloaded intrinsic. You can use <tt>llvm.pow</tt> on any
7220 floating point or vector of floating point type. Not all targets support all
7224 declare float @llvm.pow.f32(float %Val, float %Power)
7225 declare double @llvm.pow.f64(double %Val, double %Power)
7226 declare x86_fp80 @llvm.pow.f80(x86_fp80 %Val, x86_fp80 %Power)
7227 declare fp128 @llvm.pow.f128(fp128 %Val, fp128 %Power)
7228 declare ppc_fp128 @llvm.pow.ppcf128(ppc_fp128 %Val, ppc_fp128 Power)
7232 <p>The '<tt>llvm.pow.*</tt>' intrinsics return the first operand raised to the
7233 specified (positive or negative) power.</p>
7236 <p>The second argument is a floating point power, and the first is a value to
7237 raise to that power.</p>
7240 <p>This function returns the first value raised to the second power, returning
7241 the same values as the libm <tt>pow</tt> functions would, and handles error
7242 conditions in the same way.</p>
7246 <!-- _______________________________________________________________________ -->
7248 <a name="int_exp">'<tt>llvm.exp.*</tt>' Intrinsic</a>
7254 <p>This is an overloaded intrinsic. You can use <tt>llvm.exp</tt> on any
7255 floating point or vector of floating point type. Not all targets support all
7259 declare float @llvm.exp.f32(float %Val)
7260 declare double @llvm.exp.f64(double %Val)
7261 declare x86_fp80 @llvm.exp.f80(x86_fp80 %Val)
7262 declare fp128 @llvm.exp.f128(fp128 %Val)
7263 declare ppc_fp128 @llvm.exp.ppcf128(ppc_fp128 %Val)
7267 <p>The '<tt>llvm.exp.*</tt>' intrinsics perform the exp function.</p>
7270 <p>The argument and return value are floating point numbers of the same
7274 <p>This function returns the same values as the libm <tt>exp</tt> functions
7275 would, and handles error conditions in the same way.</p>
7279 <!-- _______________________________________________________________________ -->
7281 <a name="int_log">'<tt>llvm.log.*</tt>' Intrinsic</a>
7287 <p>This is an overloaded intrinsic. You can use <tt>llvm.log</tt> on any
7288 floating point or vector of floating point type. Not all targets support all
7292 declare float @llvm.log.f32(float %Val)
7293 declare double @llvm.log.f64(double %Val)
7294 declare x86_fp80 @llvm.log.f80(x86_fp80 %Val)
7295 declare fp128 @llvm.log.f128(fp128 %Val)
7296 declare ppc_fp128 @llvm.log.ppcf128(ppc_fp128 %Val)
7300 <p>The '<tt>llvm.log.*</tt>' intrinsics perform the log function.</p>
7303 <p>The argument and return value are floating point numbers of the same
7307 <p>This function returns the same values as the libm <tt>log</tt> functions
7308 would, and handles error conditions in the same way.</p>
7312 <!-- _______________________________________________________________________ -->
7314 <a name="int_fma">'<tt>llvm.fma.*</tt>' Intrinsic</a>
7320 <p>This is an overloaded intrinsic. You can use <tt>llvm.fma</tt> on any
7321 floating point or vector of floating point type. Not all targets support all
7325 declare float @llvm.fma.f32(float %a, float %b, float %c)
7326 declare double @llvm.fma.f64(double %a, double %b, double %c)
7327 declare x86_fp80 @llvm.fma.f80(x86_fp80 %a, x86_fp80 %b, x86_fp80 %c)
7328 declare fp128 @llvm.fma.f128(fp128 %a, fp128 %b, fp128 %c)
7329 declare ppc_fp128 @llvm.fma.ppcf128(ppc_fp128 %a, ppc_fp128 %b, ppc_fp128 %c)
7333 <p>The '<tt>llvm.fma.*</tt>' intrinsics perform the fused multiply-add
7337 <p>The argument and return value are floating point numbers of the same
7341 <p>This function returns the same values as the libm <tt>fma</tt> functions
7348 <!-- ======================================================================= -->
7350 <a name="int_manip">Bit Manipulation Intrinsics</a>
7355 <p>LLVM provides intrinsics for a few important bit manipulation operations.
7356 These allow efficient code generation for some algorithms.</p>
7358 <!-- _______________________________________________________________________ -->
7360 <a name="int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a>
7366 <p>This is an overloaded intrinsic function. You can use bswap on any integer
7367 type that is an even number of bytes (i.e. BitWidth % 16 == 0).</p>
7370 declare i16 @llvm.bswap.i16(i16 <id>)
7371 declare i32 @llvm.bswap.i32(i32 <id>)
7372 declare i64 @llvm.bswap.i64(i64 <id>)
7376 <p>The '<tt>llvm.bswap</tt>' family of intrinsics is used to byte swap integer
7377 values with an even number of bytes (positive multiple of 16 bits). These
7378 are useful for performing operations on data that is not in the target's
7379 native byte order.</p>
7382 <p>The <tt>llvm.bswap.i16</tt> intrinsic returns an i16 value that has the high
7383 and low byte of the input i16 swapped. Similarly,
7384 the <tt>llvm.bswap.i32</tt> intrinsic returns an i32 value that has the four
7385 bytes of the input i32 swapped, so that if the input bytes are numbered 0, 1,
7386 2, 3 then the returned i32 will have its bytes in 3, 2, 1, 0 order.
7387 The <tt>llvm.bswap.i48</tt>, <tt>llvm.bswap.i64</tt> and other intrinsics
7388 extend this concept to additional even-byte lengths (6 bytes, 8 bytes and
7389 more, respectively).</p>
7393 <!-- _______________________________________________________________________ -->
7395 <a name="int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic</a>
7401 <p>This is an overloaded intrinsic. You can use llvm.ctpop on any integer bit
7402 width, or on any vector with integer elements. Not all targets support all
7403 bit widths or vector types, however.</p>
7406 declare i8 @llvm.ctpop.i8(i8 <src>)
7407 declare i16 @llvm.ctpop.i16(i16 <src>)
7408 declare i32 @llvm.ctpop.i32(i32 <src>)
7409 declare i64 @llvm.ctpop.i64(i64 <src>)
7410 declare i256 @llvm.ctpop.i256(i256 <src>)
7411 declare <2 x i32> @llvm.ctpop.v2i32(<2 x i32> <src>)
7415 <p>The '<tt>llvm.ctpop</tt>' family of intrinsics counts the number of bits set
7419 <p>The only argument is the value to be counted. The argument may be of any
7420 integer type, or a vector with integer elements.
7421 The return type must match the argument type.</p>
7424 <p>The '<tt>llvm.ctpop</tt>' intrinsic counts the 1's in a variable, or within each
7425 element of a vector.</p>
7429 <!-- _______________________________________________________________________ -->
7431 <a name="int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic</a>
7437 <p>This is an overloaded intrinsic. You can use <tt>llvm.ctlz</tt> on any
7438 integer bit width, or any vector whose elements are integers. Not all
7439 targets support all bit widths or vector types, however.</p>
7442 declare i8 @llvm.ctlz.i8 (i8 <src>, i1 <is_zero_undef>)
7443 declare i16 @llvm.ctlz.i16 (i16 <src>, i1 <is_zero_undef>)
7444 declare i32 @llvm.ctlz.i32 (i32 <src>, i1 <is_zero_undef>)
7445 declare i64 @llvm.ctlz.i64 (i64 <src>, i1 <is_zero_undef>)
7446 declare i256 @llvm.ctlz.i256(i256 <src>, i1 <is_zero_undef>)
7447 declase <2 x i32> @llvm.ctlz.v2i32(<2 x i32> <src>, i1 <is_zero_undef>)
7451 <p>The '<tt>llvm.ctlz</tt>' family of intrinsic functions counts the number of
7452 leading zeros in a variable.</p>
7455 <p>The first argument is the value to be counted. This argument may be of any
7456 integer type, or a vectory with integer element type. The return type
7457 must match the first argument type.</p>
7459 <p>The second argument must be a constant and is a flag to indicate whether the
7460 intrinsic should ensure that a zero as the first argument produces a defined
7461 result. Historically some architectures did not provide a defined result for
7462 zero values as efficiently, and many algorithms are now predicated on
7463 avoiding zero-value inputs.</p>
7466 <p>The '<tt>llvm.ctlz</tt>' intrinsic counts the leading (most significant)
7467 zeros in a variable, or within each element of the vector.
7468 If <tt>src == 0</tt> then the result is the size in bits of the type of
7469 <tt>src</tt> if <tt>is_zero_undef == 0</tt> and <tt>undef</tt> otherwise.
7470 For example, <tt>llvm.ctlz(i32 2) = 30</tt>.</p>
7474 <!-- _______________________________________________________________________ -->
7476 <a name="int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic</a>
7482 <p>This is an overloaded intrinsic. You can use <tt>llvm.cttz</tt> on any
7483 integer bit width, or any vector of integer elements. Not all targets
7484 support all bit widths or vector types, however.</p>
7487 declare i8 @llvm.cttz.i8 (i8 <src>, i1 <is_zero_undef>)
7488 declare i16 @llvm.cttz.i16 (i16 <src>, i1 <is_zero_undef>)
7489 declare i32 @llvm.cttz.i32 (i32 <src>, i1 <is_zero_undef>)
7490 declare i64 @llvm.cttz.i64 (i64 <src>, i1 <is_zero_undef>)
7491 declare i256 @llvm.cttz.i256(i256 <src>, i1 <is_zero_undef>)
7492 declase <2 x i32> @llvm.cttz.v2i32(<2 x i32> <src>, i1 <is_zero_undef>)
7496 <p>The '<tt>llvm.cttz</tt>' family of intrinsic functions counts the number of
7500 <p>The first argument is the value to be counted. This argument may be of any
7501 integer type, or a vectory with integer element type. The return type
7502 must match the first argument type.</p>
7504 <p>The second argument must be a constant and is a flag to indicate whether the
7505 intrinsic should ensure that a zero as the first argument produces a defined
7506 result. Historically some architectures did not provide a defined result for
7507 zero values as efficiently, and many algorithms are now predicated on
7508 avoiding zero-value inputs.</p>
7511 <p>The '<tt>llvm.cttz</tt>' intrinsic counts the trailing (least significant)
7512 zeros in a variable, or within each element of a vector.
7513 If <tt>src == 0</tt> then the result is the size in bits of the type of
7514 <tt>src</tt> if <tt>is_zero_undef == 0</tt> and <tt>undef</tt> otherwise.
7515 For example, <tt>llvm.cttz(2) = 1</tt>.</p>
7521 <!-- ======================================================================= -->
7523 <a name="int_overflow">Arithmetic with Overflow Intrinsics</a>
7528 <p>LLVM provides intrinsics for some arithmetic with overflow operations.</p>
7530 <!-- _______________________________________________________________________ -->
7532 <a name="int_sadd_overflow">
7533 '<tt>llvm.sadd.with.overflow.*</tt>' Intrinsics
7540 <p>This is an overloaded intrinsic. You can use <tt>llvm.sadd.with.overflow</tt>
7541 on any integer bit width.</p>
7544 declare {i16, i1} @llvm.sadd.with.overflow.i16(i16 %a, i16 %b)
7545 declare {i32, i1} @llvm.sadd.with.overflow.i32(i32 %a, i32 %b)
7546 declare {i64, i1} @llvm.sadd.with.overflow.i64(i64 %a, i64 %b)
7550 <p>The '<tt>llvm.sadd.with.overflow</tt>' family of intrinsic functions perform
7551 a signed addition of the two arguments, and indicate whether an overflow
7552 occurred during the signed summation.</p>
7555 <p>The arguments (%a and %b) and the first element of the result structure may
7556 be of integer types of any bit width, but they must have the same bit
7557 width. The second element of the result structure must be of
7558 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
7559 undergo signed addition.</p>
7562 <p>The '<tt>llvm.sadd.with.overflow</tt>' family of intrinsic functions perform
7563 a signed addition of the two variables. They return a structure — the
7564 first element of which is the signed summation, and the second element of
7565 which is a bit specifying if the signed summation resulted in an
7570 %res = call {i32, i1} @llvm.sadd.with.overflow.i32(i32 %a, i32 %b)
7571 %sum = extractvalue {i32, i1} %res, 0
7572 %obit = extractvalue {i32, i1} %res, 1
7573 br i1 %obit, label %overflow, label %normal
7578 <!-- _______________________________________________________________________ -->
7580 <a name="int_uadd_overflow">
7581 '<tt>llvm.uadd.with.overflow.*</tt>' Intrinsics
7588 <p>This is an overloaded intrinsic. You can use <tt>llvm.uadd.with.overflow</tt>
7589 on any integer bit width.</p>
7592 declare {i16, i1} @llvm.uadd.with.overflow.i16(i16 %a, i16 %b)
7593 declare {i32, i1} @llvm.uadd.with.overflow.i32(i32 %a, i32 %b)
7594 declare {i64, i1} @llvm.uadd.with.overflow.i64(i64 %a, i64 %b)
7598 <p>The '<tt>llvm.uadd.with.overflow</tt>' family of intrinsic functions perform
7599 an unsigned addition of the two arguments, and indicate whether a carry
7600 occurred during the unsigned summation.</p>
7603 <p>The arguments (%a and %b) and the first element of the result structure may
7604 be of integer types of any bit width, but they must have the same bit
7605 width. The second element of the result structure must be of
7606 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
7607 undergo unsigned addition.</p>
7610 <p>The '<tt>llvm.uadd.with.overflow</tt>' family of intrinsic functions perform
7611 an unsigned addition of the two arguments. They return a structure —
7612 the first element of which is the sum, and the second element of which is a
7613 bit specifying if the unsigned summation resulted in a carry.</p>
7617 %res = call {i32, i1} @llvm.uadd.with.overflow.i32(i32 %a, i32 %b)
7618 %sum = extractvalue {i32, i1} %res, 0
7619 %obit = extractvalue {i32, i1} %res, 1
7620 br i1 %obit, label %carry, label %normal
7625 <!-- _______________________________________________________________________ -->
7627 <a name="int_ssub_overflow">
7628 '<tt>llvm.ssub.with.overflow.*</tt>' Intrinsics
7635 <p>This is an overloaded intrinsic. You can use <tt>llvm.ssub.with.overflow</tt>
7636 on any integer bit width.</p>
7639 declare {i16, i1} @llvm.ssub.with.overflow.i16(i16 %a, i16 %b)
7640 declare {i32, i1} @llvm.ssub.with.overflow.i32(i32 %a, i32 %b)
7641 declare {i64, i1} @llvm.ssub.with.overflow.i64(i64 %a, i64 %b)
7645 <p>The '<tt>llvm.ssub.with.overflow</tt>' family of intrinsic functions perform
7646 a signed subtraction of the two arguments, and indicate whether an overflow
7647 occurred during the signed subtraction.</p>
7650 <p>The arguments (%a and %b) and the first element of the result structure may
7651 be of integer types of any bit width, but they must have the same bit
7652 width. The second element of the result structure must be of
7653 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
7654 undergo signed subtraction.</p>
7657 <p>The '<tt>llvm.ssub.with.overflow</tt>' family of intrinsic functions perform
7658 a signed subtraction of the two arguments. They return a structure —
7659 the first element of which is the subtraction, and the second element of
7660 which is a bit specifying if the signed subtraction resulted in an
7665 %res = call {i32, i1} @llvm.ssub.with.overflow.i32(i32 %a, i32 %b)
7666 %sum = extractvalue {i32, i1} %res, 0
7667 %obit = extractvalue {i32, i1} %res, 1
7668 br i1 %obit, label %overflow, label %normal
7673 <!-- _______________________________________________________________________ -->
7675 <a name="int_usub_overflow">
7676 '<tt>llvm.usub.with.overflow.*</tt>' Intrinsics
7683 <p>This is an overloaded intrinsic. You can use <tt>llvm.usub.with.overflow</tt>
7684 on any integer bit width.</p>
7687 declare {i16, i1} @llvm.usub.with.overflow.i16(i16 %a, i16 %b)
7688 declare {i32, i1} @llvm.usub.with.overflow.i32(i32 %a, i32 %b)
7689 declare {i64, i1} @llvm.usub.with.overflow.i64(i64 %a, i64 %b)
7693 <p>The '<tt>llvm.usub.with.overflow</tt>' family of intrinsic functions perform
7694 an unsigned subtraction of the two arguments, and indicate whether an
7695 overflow occurred during the unsigned subtraction.</p>
7698 <p>The arguments (%a and %b) and the first element of the result structure may
7699 be of integer types of any bit width, but they must have the same bit
7700 width. The second element of the result structure must be of
7701 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
7702 undergo unsigned subtraction.</p>
7705 <p>The '<tt>llvm.usub.with.overflow</tt>' family of intrinsic functions perform
7706 an unsigned subtraction of the two arguments. They return a structure —
7707 the first element of which is the subtraction, and the second element of
7708 which is a bit specifying if the unsigned subtraction resulted in an
7713 %res = call {i32, i1} @llvm.usub.with.overflow.i32(i32 %a, i32 %b)
7714 %sum = extractvalue {i32, i1} %res, 0
7715 %obit = extractvalue {i32, i1} %res, 1
7716 br i1 %obit, label %overflow, label %normal
7721 <!-- _______________________________________________________________________ -->
7723 <a name="int_smul_overflow">
7724 '<tt>llvm.smul.with.overflow.*</tt>' Intrinsics
7731 <p>This is an overloaded intrinsic. You can use <tt>llvm.smul.with.overflow</tt>
7732 on any integer bit width.</p>
7735 declare {i16, i1} @llvm.smul.with.overflow.i16(i16 %a, i16 %b)
7736 declare {i32, i1} @llvm.smul.with.overflow.i32(i32 %a, i32 %b)
7737 declare {i64, i1} @llvm.smul.with.overflow.i64(i64 %a, i64 %b)
7742 <p>The '<tt>llvm.smul.with.overflow</tt>' family of intrinsic functions perform
7743 a signed multiplication of the two arguments, and indicate whether an
7744 overflow occurred during the signed multiplication.</p>
7747 <p>The arguments (%a and %b) and the first element of the result structure may
7748 be of integer types of any bit width, but they must have the same bit
7749 width. The second element of the result structure must be of
7750 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
7751 undergo signed multiplication.</p>
7754 <p>The '<tt>llvm.smul.with.overflow</tt>' family of intrinsic functions perform
7755 a signed multiplication of the two arguments. They return a structure —
7756 the first element of which is the multiplication, and the second element of
7757 which is a bit specifying if the signed multiplication resulted in an
7762 %res = call {i32, i1} @llvm.smul.with.overflow.i32(i32 %a, i32 %b)
7763 %sum = extractvalue {i32, i1} %res, 0
7764 %obit = extractvalue {i32, i1} %res, 1
7765 br i1 %obit, label %overflow, label %normal
7770 <!-- _______________________________________________________________________ -->
7772 <a name="int_umul_overflow">
7773 '<tt>llvm.umul.with.overflow.*</tt>' Intrinsics
7780 <p>This is an overloaded intrinsic. You can use <tt>llvm.umul.with.overflow</tt>
7781 on any integer bit width.</p>
7784 declare {i16, i1} @llvm.umul.with.overflow.i16(i16 %a, i16 %b)
7785 declare {i32, i1} @llvm.umul.with.overflow.i32(i32 %a, i32 %b)
7786 declare {i64, i1} @llvm.umul.with.overflow.i64(i64 %a, i64 %b)
7790 <p>The '<tt>llvm.umul.with.overflow</tt>' family of intrinsic functions perform
7791 a unsigned multiplication of the two arguments, and indicate whether an
7792 overflow occurred during the unsigned multiplication.</p>
7795 <p>The arguments (%a and %b) and the first element of the result structure may
7796 be of integer types of any bit width, but they must have the same bit
7797 width. The second element of the result structure must be of
7798 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
7799 undergo unsigned multiplication.</p>
7802 <p>The '<tt>llvm.umul.with.overflow</tt>' family of intrinsic functions perform
7803 an unsigned multiplication of the two arguments. They return a structure
7804 — the first element of which is the multiplication, and the second
7805 element of which is a bit specifying if the unsigned multiplication resulted
7810 %res = call {i32, i1} @llvm.umul.with.overflow.i32(i32 %a, i32 %b)
7811 %sum = extractvalue {i32, i1} %res, 0
7812 %obit = extractvalue {i32, i1} %res, 1
7813 br i1 %obit, label %overflow, label %normal
7820 <!-- ======================================================================= -->
7822 <a name="int_fp16">Half Precision Floating Point Intrinsics</a>
7827 <p>Half precision floating point is a storage-only format. This means that it is
7828 a dense encoding (in memory) but does not support computation in the
7831 <p>This means that code must first load the half-precision floating point
7832 value as an i16, then convert it to float with <a
7833 href="#int_convert_from_fp16"><tt>llvm.convert.from.fp16</tt></a>.
7834 Computation can then be performed on the float value (including extending to
7835 double etc). To store the value back to memory, it is first converted to
7836 float if needed, then converted to i16 with
7837 <a href="#int_convert_to_fp16"><tt>llvm.convert.to.fp16</tt></a>, then
7838 storing as an i16 value.</p>
7840 <!-- _______________________________________________________________________ -->
7842 <a name="int_convert_to_fp16">
7843 '<tt>llvm.convert.to.fp16</tt>' Intrinsic
7851 declare i16 @llvm.convert.to.fp16(f32 %a)
7855 <p>The '<tt>llvm.convert.to.fp16</tt>' intrinsic function performs
7856 a conversion from single precision floating point format to half precision
7857 floating point format.</p>
7860 <p>The intrinsic function contains single argument - the value to be
7864 <p>The '<tt>llvm.convert.to.fp16</tt>' intrinsic function performs
7865 a conversion from single precision floating point format to half precision
7866 floating point format. The return value is an <tt>i16</tt> which
7867 contains the converted number.</p>
7871 %res = call i16 @llvm.convert.to.fp16(f32 %a)
7872 store i16 %res, i16* @x, align 2
7877 <!-- _______________________________________________________________________ -->
7879 <a name="int_convert_from_fp16">
7880 '<tt>llvm.convert.from.fp16</tt>' Intrinsic
7888 declare f32 @llvm.convert.from.fp16(i16 %a)
7892 <p>The '<tt>llvm.convert.from.fp16</tt>' intrinsic function performs
7893 a conversion from half precision floating point format to single precision
7894 floating point format.</p>
7897 <p>The intrinsic function contains single argument - the value to be
7901 <p>The '<tt>llvm.convert.from.fp16</tt>' intrinsic function performs a
7902 conversion from half single precision floating point format to single
7903 precision floating point format. The input half-float value is represented by
7904 an <tt>i16</tt> value.</p>
7908 %a = load i16* @x, align 2
7909 %res = call f32 @llvm.convert.from.fp16(i16 %a)
7916 <!-- ======================================================================= -->
7918 <a name="int_debugger">Debugger Intrinsics</a>
7923 <p>The LLVM debugger intrinsics (which all start with <tt>llvm.dbg.</tt>
7924 prefix), are described in
7925 the <a href="SourceLevelDebugging.html#format_common_intrinsics">LLVM Source
7926 Level Debugging</a> document.</p>
7930 <!-- ======================================================================= -->
7932 <a name="int_eh">Exception Handling Intrinsics</a>
7937 <p>The LLVM exception handling intrinsics (which all start with
7938 <tt>llvm.eh.</tt> prefix), are described in
7939 the <a href="ExceptionHandling.html#format_common_intrinsics">LLVM Exception
7940 Handling</a> document.</p>
7944 <!-- ======================================================================= -->
7946 <a name="int_trampoline">Trampoline Intrinsics</a>
7951 <p>These intrinsics make it possible to excise one parameter, marked with
7952 the <a href="#nest"><tt>nest</tt></a> attribute, from a function.
7953 The result is a callable
7954 function pointer lacking the nest parameter - the caller does not need to
7955 provide a value for it. Instead, the value to use is stored in advance in a
7956 "trampoline", a block of memory usually allocated on the stack, which also
7957 contains code to splice the nest value into the argument list. This is used
7958 to implement the GCC nested function address extension.</p>
7960 <p>For example, if the function is
7961 <tt>i32 f(i8* nest %c, i32 %x, i32 %y)</tt> then the resulting function
7962 pointer has signature <tt>i32 (i32, i32)*</tt>. It can be created as
7965 <pre class="doc_code">
7966 %tramp = alloca [10 x i8], align 4 ; size and alignment only correct for X86
7967 %tramp1 = getelementptr [10 x i8]* %tramp, i32 0, i32 0
7968 call i8* @llvm.init.trampoline(i8* %tramp1, i8* bitcast (i32 (i8*, i32, i32)* @f to i8*), i8* %nval)
7969 %p = call i8* @llvm.adjust.trampoline(i8* %tramp1)
7970 %fp = bitcast i8* %p to i32 (i32, i32)*
7973 <p>The call <tt>%val = call i32 %fp(i32 %x, i32 %y)</tt> is then equivalent
7974 to <tt>%val = call i32 %f(i8* %nval, i32 %x, i32 %y)</tt>.</p>
7976 <!-- _______________________________________________________________________ -->
7979 '<tt>llvm.init.trampoline</tt>' Intrinsic
7987 declare void @llvm.init.trampoline(i8* <tramp>, i8* <func>, i8* <nval>)
7991 <p>This fills the memory pointed to by <tt>tramp</tt> with executable code,
7992 turning it into a trampoline.</p>
7995 <p>The <tt>llvm.init.trampoline</tt> intrinsic takes three arguments, all
7996 pointers. The <tt>tramp</tt> argument must point to a sufficiently large and
7997 sufficiently aligned block of memory; this memory is written to by the
7998 intrinsic. Note that the size and the alignment are target-specific - LLVM
7999 currently provides no portable way of determining them, so a front-end that
8000 generates this intrinsic needs to have some target-specific knowledge.
8001 The <tt>func</tt> argument must hold a function bitcast to
8002 an <tt>i8*</tt>.</p>
8005 <p>The block of memory pointed to by <tt>tramp</tt> is filled with target
8006 dependent code, turning it into a function. Then <tt>tramp</tt> needs to be
8007 passed to <a href="#int_at">llvm.adjust.trampoline</a> to get a pointer
8008 which can be <a href="#int_trampoline">bitcast (to a new function) and
8009 called</a>. The new function's signature is the same as that of
8010 <tt>func</tt> with any arguments marked with the <tt>nest</tt> attribute
8011 removed. At most one such <tt>nest</tt> argument is allowed, and it must be of
8012 pointer type. Calling the new function is equivalent to calling <tt>func</tt>
8013 with the same argument list, but with <tt>nval</tt> used for the missing
8014 <tt>nest</tt> argument. If, after calling <tt>llvm.init.trampoline</tt>, the
8015 memory pointed to by <tt>tramp</tt> is modified, then the effect of any later call
8016 to the returned function pointer is undefined.</p>
8019 <!-- _______________________________________________________________________ -->
8022 '<tt>llvm.adjust.trampoline</tt>' Intrinsic
8030 declare i8* @llvm.adjust.trampoline(i8* <tramp>)
8034 <p>This performs any required machine-specific adjustment to the address of a
8035 trampoline (passed as <tt>tramp</tt>).</p>
8038 <p><tt>tramp</tt> must point to a block of memory which already has trampoline code
8039 filled in by a previous call to <a href="#int_it"><tt>llvm.init.trampoline</tt>
8043 <p>On some architectures the address of the code to be executed needs to be
8044 different to the address where the trampoline is actually stored. This
8045 intrinsic returns the executable address corresponding to <tt>tramp</tt>
8046 after performing the required machine specific adjustments.
8047 The pointer returned can then be <a href="#int_trampoline"> bitcast and
8055 <!-- ======================================================================= -->
8057 <a name="int_memorymarkers">Memory Use Markers</a>
8062 <p>This class of intrinsics exists to information about the lifetime of memory
8063 objects and ranges where variables are immutable.</p>
8065 <!-- _______________________________________________________________________ -->
8067 <a name="int_lifetime_start">'<tt>llvm.lifetime.start</tt>' Intrinsic</a>
8074 declare void @llvm.lifetime.start(i64 <size>, i8* nocapture <ptr>)
8078 <p>The '<tt>llvm.lifetime.start</tt>' intrinsic specifies the start of a memory
8079 object's lifetime.</p>
8082 <p>The first argument is a constant integer representing the size of the
8083 object, or -1 if it is variable sized. The second argument is a pointer to
8087 <p>This intrinsic indicates that before this point in the code, the value of the
8088 memory pointed to by <tt>ptr</tt> is dead. This means that it is known to
8089 never be used and has an undefined value. A load from the pointer that
8090 precedes this intrinsic can be replaced with
8091 <tt>'<a href="#undefvalues">undef</a>'</tt>.</p>
8095 <!-- _______________________________________________________________________ -->
8097 <a name="int_lifetime_end">'<tt>llvm.lifetime.end</tt>' Intrinsic</a>
8104 declare void @llvm.lifetime.end(i64 <size>, i8* nocapture <ptr>)
8108 <p>The '<tt>llvm.lifetime.end</tt>' intrinsic specifies the end of a memory
8109 object's lifetime.</p>
8112 <p>The first argument is a constant integer representing the size of the
8113 object, or -1 if it is variable sized. The second argument is a pointer to
8117 <p>This intrinsic indicates that after this point in the code, the value of the
8118 memory pointed to by <tt>ptr</tt> is dead. This means that it is known to
8119 never be used and has an undefined value. Any stores into the memory object
8120 following this intrinsic may be removed as dead.
8124 <!-- _______________________________________________________________________ -->
8126 <a name="int_invariant_start">'<tt>llvm.invariant.start</tt>' Intrinsic</a>
8133 declare {}* @llvm.invariant.start(i64 <size>, i8* nocapture <ptr>)
8137 <p>The '<tt>llvm.invariant.start</tt>' intrinsic specifies that the contents of
8138 a memory object will not change.</p>
8141 <p>The first argument is a constant integer representing the size of the
8142 object, or -1 if it is variable sized. The second argument is a pointer to
8146 <p>This intrinsic indicates that until an <tt>llvm.invariant.end</tt> that uses
8147 the return value, the referenced memory location is constant and
8152 <!-- _______________________________________________________________________ -->
8154 <a name="int_invariant_end">'<tt>llvm.invariant.end</tt>' Intrinsic</a>
8161 declare void @llvm.invariant.end({}* <start>, i64 <size>, i8* nocapture <ptr>)
8165 <p>The '<tt>llvm.invariant.end</tt>' intrinsic specifies that the contents of
8166 a memory object are mutable.</p>
8169 <p>The first argument is the matching <tt>llvm.invariant.start</tt> intrinsic.
8170 The second argument is a constant integer representing the size of the
8171 object, or -1 if it is variable sized and the third argument is a pointer
8175 <p>This intrinsic indicates that the memory is mutable again.</p>
8181 <!-- ======================================================================= -->
8183 <a name="int_general">General Intrinsics</a>
8188 <p>This class of intrinsics is designed to be generic and has no specific
8191 <!-- _______________________________________________________________________ -->
8193 <a name="int_var_annotation">'<tt>llvm.var.annotation</tt>' Intrinsic</a>
8200 declare void @llvm.var.annotation(i8* <val>, i8* <str>, i8* <str>, i32 <int>)
8204 <p>The '<tt>llvm.var.annotation</tt>' intrinsic.</p>
8207 <p>The first argument is a pointer to a value, the second is a pointer to a
8208 global string, the third is a pointer to a global string which is the source
8209 file name, and the last argument is the line number.</p>
8212 <p>This intrinsic allows annotation of local variables with arbitrary strings.
8213 This can be useful for special purpose optimizations that want to look for
8214 these annotations. These have no other defined use; they are ignored by code
8215 generation and optimization.</p>
8219 <!-- _______________________________________________________________________ -->
8221 <a name="int_annotation">'<tt>llvm.annotation.*</tt>' Intrinsic</a>
8227 <p>This is an overloaded intrinsic. You can use '<tt>llvm.annotation</tt>' on
8228 any integer bit width.</p>
8231 declare i8 @llvm.annotation.i8(i8 <val>, i8* <str>, i8* <str>, i32 <int>)
8232 declare i16 @llvm.annotation.i16(i16 <val>, i8* <str>, i8* <str>, i32 <int>)
8233 declare i32 @llvm.annotation.i32(i32 <val>, i8* <str>, i8* <str>, i32 <int>)
8234 declare i64 @llvm.annotation.i64(i64 <val>, i8* <str>, i8* <str>, i32 <int>)
8235 declare i256 @llvm.annotation.i256(i256 <val>, i8* <str>, i8* <str>, i32 <int>)
8239 <p>The '<tt>llvm.annotation</tt>' intrinsic.</p>
8242 <p>The first argument is an integer value (result of some expression), the
8243 second is a pointer to a global string, the third is a pointer to a global
8244 string which is the source file name, and the last argument is the line
8245 number. It returns the value of the first argument.</p>
8248 <p>This intrinsic allows annotations to be put on arbitrary expressions with
8249 arbitrary strings. This can be useful for special purpose optimizations that
8250 want to look for these annotations. These have no other defined use; they
8251 are ignored by code generation and optimization.</p>
8255 <!-- _______________________________________________________________________ -->
8257 <a name="int_trap">'<tt>llvm.trap</tt>' Intrinsic</a>
8264 declare void @llvm.trap()
8268 <p>The '<tt>llvm.trap</tt>' intrinsic.</p>
8274 <p>This intrinsics is lowered to the target dependent trap instruction. If the
8275 target does not have a trap instruction, this intrinsic will be lowered to
8276 the call of the <tt>abort()</tt> function.</p>
8280 <!-- _______________________________________________________________________ -->
8282 <a name="int_stackprotector">'<tt>llvm.stackprotector</tt>' Intrinsic</a>
8289 declare void @llvm.stackprotector(i8* <guard>, i8** <slot>)
8293 <p>The <tt>llvm.stackprotector</tt> intrinsic takes the <tt>guard</tt> and
8294 stores it onto the stack at <tt>slot</tt>. The stack slot is adjusted to
8295 ensure that it is placed on the stack before local variables.</p>
8298 <p>The <tt>llvm.stackprotector</tt> intrinsic requires two pointer
8299 arguments. The first argument is the value loaded from the stack
8300 guard <tt>@__stack_chk_guard</tt>. The second variable is an <tt>alloca</tt>
8301 that has enough space to hold the value of the guard.</p>
8304 <p>This intrinsic causes the prologue/epilogue inserter to force the position of
8305 the <tt>AllocaInst</tt> stack slot to be before local variables on the
8306 stack. This is to ensure that if a local variable on the stack is
8307 overwritten, it will destroy the value of the guard. When the function exits,
8308 the guard on the stack is checked against the original guard. If they are
8309 different, then the program aborts by calling the <tt>__stack_chk_fail()</tt>
8314 <!-- _______________________________________________________________________ -->
8316 <a name="int_objectsize">'<tt>llvm.objectsize</tt>' Intrinsic</a>
8323 declare i32 @llvm.objectsize.i32(i8* <object>, i1 <type>)
8324 declare i64 @llvm.objectsize.i64(i8* <object>, i1 <type>)
8328 <p>The <tt>llvm.objectsize</tt> intrinsic is designed to provide information to
8329 the optimizers to determine at compile time whether a) an operation (like
8330 memcpy) will overflow a buffer that corresponds to an object, or b) that a
8331 runtime check for overflow isn't necessary. An object in this context means
8332 an allocation of a specific class, structure, array, or other object.</p>
8335 <p>The <tt>llvm.objectsize</tt> intrinsic takes two arguments. The first
8336 argument is a pointer to or into the <tt>object</tt>. The second argument
8337 is a boolean 0 or 1. This argument determines whether you want the
8338 maximum (0) or minimum (1) bytes remaining. This needs to be a literal 0 or
8339 1, variables are not allowed.</p>
8342 <p>The <tt>llvm.objectsize</tt> intrinsic is lowered to either a constant
8343 representing the size of the object concerned, or <tt>i32/i64 -1 or 0</tt>,
8344 depending on the <tt>type</tt> argument, if the size cannot be determined at
8348 <!-- _______________________________________________________________________ -->
8350 <a name="int_expect">'<tt>llvm.expect</tt>' Intrinsic</a>
8357 declare i32 @llvm.expect.i32(i32 <val>, i32 <expected_val>)
8358 declare i64 @llvm.expect.i64(i64 <val>, i64 <expected_val>)
8362 <p>The <tt>llvm.expect</tt> intrinsic provides information about expected (the
8363 most probable) value of <tt>val</tt>, which can be used by optimizers.</p>
8366 <p>The <tt>llvm.expect</tt> intrinsic takes two arguments. The first
8367 argument is a value. The second argument is an expected value, this needs to
8368 be a constant value, variables are not allowed.</p>
8371 <p>This intrinsic is lowered to the <tt>val</tt>.</p>
8377 <!-- *********************************************************************** -->
8380 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
8381 src="http://jigsaw.w3.org/css-validator/images/vcss-blue" alt="Valid CSS"></a>
8382 <a href="http://validator.w3.org/check/referer"><img
8383 src="http://www.w3.org/Icons/valid-html401-blue" alt="Valid HTML 4.01"></a>
8385 <a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
8386 <a href="http://llvm.org/">The LLVM Compiler Infrastructure</a><br>
8387 Last modified: $Date$