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 <div class="doc_title"> LLVM Language Reference Manual </div>
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>externally visible</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>
58 <li><a href="#typesystem">Type System</a>
60 <li><a href="#t_classifications">Type Classifications</a></li>
61 <li><a href="#t_primitive">Primitive Types</a>
63 <li><a href="#t_integer">Integer Type</a></li>
64 <li><a href="#t_floating">Floating Point Types</a></li>
65 <li><a href="#t_x86mmx">X86mmx Type</a></li>
66 <li><a href="#t_void">Void Type</a></li>
67 <li><a href="#t_label">Label Type</a></li>
68 <li><a href="#t_metadata">Metadata Type</a></li>
71 <li><a href="#t_derived">Derived Types</a>
73 <li><a href="#t_aggregate">Aggregate Types</a>
75 <li><a href="#t_array">Array Type</a></li>
76 <li><a href="#t_struct">Structure Type</a></li>
77 <li><a href="#t_pstruct">Packed Structure Type</a></li>
78 <li><a href="#t_vector">Vector Type</a></li>
81 <li><a href="#t_function">Function Type</a></li>
82 <li><a href="#t_pointer">Pointer Type</a></li>
83 <li><a href="#t_opaque">Opaque Type</a></li>
86 <li><a href="#t_uprefs">Type Up-references</a></li>
89 <li><a href="#constants">Constants</a>
91 <li><a href="#simpleconstants">Simple Constants</a></li>
92 <li><a href="#complexconstants">Complex Constants</a></li>
93 <li><a href="#globalconstants">Global Variable and Function Addresses</a></li>
94 <li><a href="#undefvalues">Undefined Values</a></li>
95 <li><a href="#trapvalues">Trap Values</a></li>
96 <li><a href="#blockaddress">Addresses of Basic Blocks</a></li>
97 <li><a href="#constantexprs">Constant Expressions</a></li>
100 <li><a href="#othervalues">Other Values</a>
102 <li><a href="#inlineasm">Inline Assembler Expressions</a></li>
103 <li><a href="#metadata">Metadata Nodes and Metadata Strings</a></li>
106 <li><a href="#intrinsic_globals">Intrinsic Global Variables</a>
108 <li><a href="#intg_used">The '<tt>llvm.used</tt>' Global Variable</a></li>
109 <li><a href="#intg_compiler_used">The '<tt>llvm.compiler.used</tt>'
110 Global Variable</a></li>
111 <li><a href="#intg_global_ctors">The '<tt>llvm.global_ctors</tt>'
112 Global Variable</a></li>
113 <li><a href="#intg_global_dtors">The '<tt>llvm.global_dtors</tt>'
114 Global Variable</a></li>
117 <li><a href="#instref">Instruction Reference</a>
119 <li><a href="#terminators">Terminator Instructions</a>
121 <li><a href="#i_ret">'<tt>ret</tt>' Instruction</a></li>
122 <li><a href="#i_br">'<tt>br</tt>' Instruction</a></li>
123 <li><a href="#i_switch">'<tt>switch</tt>' Instruction</a></li>
124 <li><a href="#i_indirectbr">'<tt>indirectbr</tt>' Instruction</a></li>
125 <li><a href="#i_invoke">'<tt>invoke</tt>' Instruction</a></li>
126 <li><a href="#i_unwind">'<tt>unwind</tt>' Instruction</a></li>
127 <li><a href="#i_unreachable">'<tt>unreachable</tt>' Instruction</a></li>
130 <li><a href="#binaryops">Binary Operations</a>
132 <li><a href="#i_add">'<tt>add</tt>' Instruction</a></li>
133 <li><a href="#i_fadd">'<tt>fadd</tt>' Instruction</a></li>
134 <li><a href="#i_sub">'<tt>sub</tt>' Instruction</a></li>
135 <li><a href="#i_fsub">'<tt>fsub</tt>' Instruction</a></li>
136 <li><a href="#i_mul">'<tt>mul</tt>' Instruction</a></li>
137 <li><a href="#i_fmul">'<tt>fmul</tt>' Instruction</a></li>
138 <li><a href="#i_udiv">'<tt>udiv</tt>' Instruction</a></li>
139 <li><a href="#i_sdiv">'<tt>sdiv</tt>' Instruction</a></li>
140 <li><a href="#i_fdiv">'<tt>fdiv</tt>' Instruction</a></li>
141 <li><a href="#i_urem">'<tt>urem</tt>' Instruction</a></li>
142 <li><a href="#i_srem">'<tt>srem</tt>' Instruction</a></li>
143 <li><a href="#i_frem">'<tt>frem</tt>' Instruction</a></li>
146 <li><a href="#bitwiseops">Bitwise Binary Operations</a>
148 <li><a href="#i_shl">'<tt>shl</tt>' Instruction</a></li>
149 <li><a href="#i_lshr">'<tt>lshr</tt>' Instruction</a></li>
150 <li><a href="#i_ashr">'<tt>ashr</tt>' Instruction</a></li>
151 <li><a href="#i_and">'<tt>and</tt>' Instruction</a></li>
152 <li><a href="#i_or">'<tt>or</tt>' Instruction</a></li>
153 <li><a href="#i_xor">'<tt>xor</tt>' Instruction</a></li>
156 <li><a href="#vectorops">Vector Operations</a>
158 <li><a href="#i_extractelement">'<tt>extractelement</tt>' Instruction</a></li>
159 <li><a href="#i_insertelement">'<tt>insertelement</tt>' Instruction</a></li>
160 <li><a href="#i_shufflevector">'<tt>shufflevector</tt>' Instruction</a></li>
163 <li><a href="#aggregateops">Aggregate Operations</a>
165 <li><a href="#i_extractvalue">'<tt>extractvalue</tt>' Instruction</a></li>
166 <li><a href="#i_insertvalue">'<tt>insertvalue</tt>' Instruction</a></li>
169 <li><a href="#memoryops">Memory Access and Addressing Operations</a>
171 <li><a href="#i_alloca">'<tt>alloca</tt>' Instruction</a></li>
172 <li><a href="#i_load">'<tt>load</tt>' Instruction</a></li>
173 <li><a href="#i_store">'<tt>store</tt>' Instruction</a></li>
174 <li><a href="#i_getelementptr">'<tt>getelementptr</tt>' Instruction</a></li>
177 <li><a href="#convertops">Conversion Operations</a>
179 <li><a href="#i_trunc">'<tt>trunc .. to</tt>' Instruction</a></li>
180 <li><a href="#i_zext">'<tt>zext .. to</tt>' Instruction</a></li>
181 <li><a href="#i_sext">'<tt>sext .. to</tt>' Instruction</a></li>
182 <li><a href="#i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a></li>
183 <li><a href="#i_fpext">'<tt>fpext .. to</tt>' Instruction</a></li>
184 <li><a href="#i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a></li>
185 <li><a href="#i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a></li>
186 <li><a href="#i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a></li>
187 <li><a href="#i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a></li>
188 <li><a href="#i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a></li>
189 <li><a href="#i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a></li>
190 <li><a href="#i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a></li>
193 <li><a href="#otherops">Other Operations</a>
195 <li><a href="#i_icmp">'<tt>icmp</tt>' Instruction</a></li>
196 <li><a href="#i_fcmp">'<tt>fcmp</tt>' Instruction</a></li>
197 <li><a href="#i_phi">'<tt>phi</tt>' Instruction</a></li>
198 <li><a href="#i_select">'<tt>select</tt>' Instruction</a></li>
199 <li><a href="#i_call">'<tt>call</tt>' Instruction</a></li>
200 <li><a href="#i_va_arg">'<tt>va_arg</tt>' Instruction</a></li>
205 <li><a href="#intrinsics">Intrinsic Functions</a>
207 <li><a href="#int_varargs">Variable Argument Handling Intrinsics</a>
209 <li><a href="#int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a></li>
210 <li><a href="#int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a></li>
211 <li><a href="#int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a></li>
214 <li><a href="#int_gc">Accurate Garbage Collection Intrinsics</a>
216 <li><a href="#int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a></li>
217 <li><a href="#int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a></li>
218 <li><a href="#int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a></li>
221 <li><a href="#int_codegen">Code Generator Intrinsics</a>
223 <li><a href="#int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a></li>
224 <li><a href="#int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a></li>
225 <li><a href="#int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a></li>
226 <li><a href="#int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a></li>
227 <li><a href="#int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a></li>
228 <li><a href="#int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a></li>
229 <li><a href="#int_readcyclecounter">'<tt>llvm.readcyclecounter</tt>' Intrinsic</a></li>
232 <li><a href="#int_libc">Standard C Library Intrinsics</a>
234 <li><a href="#int_memcpy">'<tt>llvm.memcpy.*</tt>' Intrinsic</a></li>
235 <li><a href="#int_memmove">'<tt>llvm.memmove.*</tt>' Intrinsic</a></li>
236 <li><a href="#int_memset">'<tt>llvm.memset.*</tt>' Intrinsic</a></li>
237 <li><a href="#int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a></li>
238 <li><a href="#int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a></li>
239 <li><a href="#int_sin">'<tt>llvm.sin.*</tt>' Intrinsic</a></li>
240 <li><a href="#int_cos">'<tt>llvm.cos.*</tt>' Intrinsic</a></li>
241 <li><a href="#int_pow">'<tt>llvm.pow.*</tt>' Intrinsic</a></li>
244 <li><a href="#int_manip">Bit Manipulation Intrinsics</a>
246 <li><a href="#int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a></li>
247 <li><a href="#int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic </a></li>
248 <li><a href="#int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic </a></li>
249 <li><a href="#int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic </a></li>
252 <li><a href="#int_overflow">Arithmetic with Overflow Intrinsics</a>
254 <li><a href="#int_sadd_overflow">'<tt>llvm.sadd.with.overflow.*</tt> Intrinsics</a></li>
255 <li><a href="#int_uadd_overflow">'<tt>llvm.uadd.with.overflow.*</tt> Intrinsics</a></li>
256 <li><a href="#int_ssub_overflow">'<tt>llvm.ssub.with.overflow.*</tt> Intrinsics</a></li>
257 <li><a href="#int_usub_overflow">'<tt>llvm.usub.with.overflow.*</tt> Intrinsics</a></li>
258 <li><a href="#int_smul_overflow">'<tt>llvm.smul.with.overflow.*</tt> Intrinsics</a></li>
259 <li><a href="#int_umul_overflow">'<tt>llvm.umul.with.overflow.*</tt> Intrinsics</a></li>
262 <li><a href="#int_fp16">Half Precision Floating Point Intrinsics</a>
264 <li><a href="#int_convert_to_fp16">'<tt>llvm.convert.to.fp16</tt>' Intrinsic</a></li>
265 <li><a href="#int_convert_from_fp16">'<tt>llvm.convert.from.fp16</tt>' Intrinsic</a></li>
268 <li><a href="#int_debugger">Debugger intrinsics</a></li>
269 <li><a href="#int_eh">Exception Handling intrinsics</a></li>
270 <li><a href="#int_trampoline">Trampoline Intrinsic</a>
272 <li><a href="#int_it">'<tt>llvm.init.trampoline</tt>' Intrinsic</a></li>
275 <li><a href="#int_atomics">Atomic intrinsics</a>
277 <li><a href="#int_memory_barrier"><tt>llvm.memory_barrier</tt></a></li>
278 <li><a href="#int_atomic_cmp_swap"><tt>llvm.atomic.cmp.swap</tt></a></li>
279 <li><a href="#int_atomic_swap"><tt>llvm.atomic.swap</tt></a></li>
280 <li><a href="#int_atomic_load_add"><tt>llvm.atomic.load.add</tt></a></li>
281 <li><a href="#int_atomic_load_sub"><tt>llvm.atomic.load.sub</tt></a></li>
282 <li><a href="#int_atomic_load_and"><tt>llvm.atomic.load.and</tt></a></li>
283 <li><a href="#int_atomic_load_nand"><tt>llvm.atomic.load.nand</tt></a></li>
284 <li><a href="#int_atomic_load_or"><tt>llvm.atomic.load.or</tt></a></li>
285 <li><a href="#int_atomic_load_xor"><tt>llvm.atomic.load.xor</tt></a></li>
286 <li><a href="#int_atomic_load_max"><tt>llvm.atomic.load.max</tt></a></li>
287 <li><a href="#int_atomic_load_min"><tt>llvm.atomic.load.min</tt></a></li>
288 <li><a href="#int_atomic_load_umax"><tt>llvm.atomic.load.umax</tt></a></li>
289 <li><a href="#int_atomic_load_umin"><tt>llvm.atomic.load.umin</tt></a></li>
292 <li><a href="#int_memorymarkers">Memory Use Markers</a>
294 <li><a href="#int_lifetime_start"><tt>llvm.lifetime.start</tt></a></li>
295 <li><a href="#int_lifetime_end"><tt>llvm.lifetime.end</tt></a></li>
296 <li><a href="#int_invariant_start"><tt>llvm.invariant.start</tt></a></li>
297 <li><a href="#int_invariant_end"><tt>llvm.invariant.end</tt></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>
318 <div class="doc_author">
319 <p>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a>
320 and <a href="mailto:vadve@cs.uiuc.edu">Vikram Adve</a></p>
323 <!-- *********************************************************************** -->
324 <div class="doc_section"> <a name="abstract">Abstract </a></div>
325 <!-- *********************************************************************** -->
327 <div class="doc_text">
329 <p>This document is a reference manual for the LLVM assembly language. LLVM is
330 a Static Single Assignment (SSA) based representation that provides type
331 safety, low-level operations, flexibility, and the capability of representing
332 'all' high-level languages cleanly. It is the common code representation
333 used throughout all phases of the LLVM compilation strategy.</p>
337 <!-- *********************************************************************** -->
338 <div class="doc_section"> <a name="introduction">Introduction</a> </div>
339 <!-- *********************************************************************** -->
341 <div class="doc_text">
343 <p>The LLVM code representation is designed to be used in three different forms:
344 as an in-memory compiler IR, as an on-disk bitcode representation (suitable
345 for fast loading by a Just-In-Time compiler), and as a human readable
346 assembly language representation. This allows LLVM to provide a powerful
347 intermediate representation for efficient compiler transformations and
348 analysis, while providing a natural means to debug and visualize the
349 transformations. The three different forms of LLVM are all equivalent. This
350 document describes the human readable representation and notation.</p>
352 <p>The LLVM representation aims to be light-weight and low-level while being
353 expressive, typed, and extensible at the same time. It aims to be a
354 "universal IR" of sorts, by being at a low enough level that high-level ideas
355 may be cleanly mapped to it (similar to how microprocessors are "universal
356 IR's", allowing many source languages to be mapped to them). By providing
357 type information, LLVM can be used as the target of optimizations: for
358 example, through pointer analysis, it can be proven that a C automatic
359 variable is never accessed outside of the current function, allowing it to
360 be promoted to a simple SSA value instead of a memory location.</p>
364 <!-- _______________________________________________________________________ -->
365 <div class="doc_subsubsection"> <a name="wellformed">Well-Formedness</a> </div>
367 <div class="doc_text">
369 <p>It is important to note that this document describes 'well formed' LLVM
370 assembly language. There is a difference between what the parser accepts and
371 what is considered 'well formed'. For example, the following instruction is
372 syntactically okay, but not well formed:</p>
374 <pre class="doc_code">
375 %x = <a href="#i_add">add</a> i32 1, %x
378 <p>because the definition of <tt>%x</tt> does not dominate all of its uses. The
379 LLVM infrastructure provides a verification pass that may be used to verify
380 that an LLVM module is well formed. This pass is automatically run by the
381 parser after parsing input assembly and by the optimizer before it outputs
382 bitcode. The violations pointed out by the verifier pass indicate bugs in
383 transformation passes or input to the parser.</p>
387 <!-- Describe the typesetting conventions here. -->
389 <!-- *********************************************************************** -->
390 <div class="doc_section"> <a name="identifiers">Identifiers</a> </div>
391 <!-- *********************************************************************** -->
393 <div class="doc_text">
395 <p>LLVM identifiers come in two basic types: global and local. Global
396 identifiers (functions, global variables) begin with the <tt>'@'</tt>
397 character. Local identifiers (register names, types) begin with
398 the <tt>'%'</tt> character. Additionally, there are three different formats
399 for identifiers, for different purposes:</p>
402 <li>Named values are represented as a string of characters with their prefix.
403 For example, <tt>%foo</tt>, <tt>@DivisionByZero</tt>,
404 <tt>%a.really.long.identifier</tt>. The actual regular expression used is
405 '<tt>[%@][a-zA-Z$._][a-zA-Z$._0-9]*</tt>'. Identifiers which require
406 other characters in their names can be surrounded with quotes. Special
407 characters may be escaped using <tt>"\xx"</tt> where <tt>xx</tt> is the
408 ASCII code for the character in hexadecimal. In this way, any character
409 can be used in a name value, even quotes themselves.</li>
411 <li>Unnamed values are represented as an unsigned numeric value with their
412 prefix. For example, <tt>%12</tt>, <tt>@2</tt>, <tt>%44</tt>.</li>
414 <li>Constants, which are described in a <a href="#constants">section about
415 constants</a>, below.</li>
418 <p>LLVM requires that values start with a prefix for two reasons: Compilers
419 don't need to worry about name clashes with reserved words, and the set of
420 reserved words may be expanded in the future without penalty. Additionally,
421 unnamed identifiers allow a compiler to quickly come up with a temporary
422 variable without having to avoid symbol table conflicts.</p>
424 <p>Reserved words in LLVM are very similar to reserved words in other
425 languages. There are keywords for different opcodes
426 ('<tt><a href="#i_add">add</a></tt>',
427 '<tt><a href="#i_bitcast">bitcast</a></tt>',
428 '<tt><a href="#i_ret">ret</a></tt>', etc...), for primitive type names
429 ('<tt><a href="#t_void">void</a></tt>',
430 '<tt><a href="#t_primitive">i32</a></tt>', etc...), and others. These
431 reserved words cannot conflict with variable names, because none of them
432 start with a prefix character (<tt>'%'</tt> or <tt>'@'</tt>).</p>
434 <p>Here is an example of LLVM code to multiply the integer variable
435 '<tt>%X</tt>' by 8:</p>
439 <pre class="doc_code">
440 %result = <a href="#i_mul">mul</a> i32 %X, 8
443 <p>After strength reduction:</p>
445 <pre class="doc_code">
446 %result = <a href="#i_shl">shl</a> i32 %X, i8 3
449 <p>And the hard way:</p>
451 <pre class="doc_code">
452 %0 = <a href="#i_add">add</a> i32 %X, %X <i>; yields {i32}:%0</i>
453 %1 = <a href="#i_add">add</a> i32 %0, %0 <i>; yields {i32}:%1</i>
454 %result = <a href="#i_add">add</a> i32 %1, %1
457 <p>This last way of multiplying <tt>%X</tt> by 8 illustrates several important
458 lexical features of LLVM:</p>
461 <li>Comments are delimited with a '<tt>;</tt>' and go until the end of
464 <li>Unnamed temporaries are created when the result of a computation is not
465 assigned to a named value.</li>
467 <li>Unnamed temporaries are numbered sequentially</li>
470 <p>It also shows a convention that we follow in this document. When
471 demonstrating instructions, we will follow an instruction with a comment that
472 defines the type and name of value produced. Comments are shown in italic
477 <!-- *********************************************************************** -->
478 <div class="doc_section"> <a name="highlevel">High Level Structure</a> </div>
479 <!-- *********************************************************************** -->
481 <!-- ======================================================================= -->
482 <div class="doc_subsection"> <a name="modulestructure">Module Structure</a>
485 <div class="doc_text">
487 <p>LLVM programs are composed of "Module"s, each of which is a translation unit
488 of the input programs. Each module consists of functions, global variables,
489 and symbol table entries. Modules may be combined together with the LLVM
490 linker, which merges function (and global variable) definitions, resolves
491 forward declarations, and merges symbol table entries. Here is an example of
492 the "hello world" module:</p>
494 <pre class="doc_code">
495 <i>; Declare the string constant as a global constant.</i>
496 <a href="#identifiers">@.LC0</a> = <a href="#linkage_internal">internal</a> <a href="#globalvars">constant</a> <a href="#t_array">[13 x i8]</a> c"hello world\0A\00" <i>; [13 x i8]*</i>
498 <i>; External declaration of the puts function</i>
499 <a href="#functionstructure">declare</a> i32 @puts(i8*) <i>; i32 (i8*)* </i>
501 <i>; Definition of main function</i>
502 define i32 @main() { <i>; i32()* </i>
503 <i>; Convert [13 x i8]* to i8 *...</i>
504 %cast210 = <a href="#i_getelementptr">getelementptr</a> [13 x i8]* @.LC0, i64 0, i64 0 <i>; i8*</i>
506 <i>; Call puts function to write out the string to stdout.</i>
507 <a href="#i_call">call</a> i32 @puts(i8* %cast210) <i>; i32</i>
508 <a href="#i_ret">ret</a> i32 0
511 <i>; Named metadata</i>
512 !1 = metadata !{i32 41}
516 <p>This example is made up of a <a href="#globalvars">global variable</a> named
517 "<tt>.LC0</tt>", an external declaration of the "<tt>puts</tt>" function,
518 a <a href="#functionstructure">function definition</a> for
519 "<tt>main</tt>" and <a href="#namedmetadatastructure">named metadata</a>
522 <p>In general, a module is made up of a list of global values, where both
523 functions and global variables are global values. Global values are
524 represented by a pointer to a memory location (in this case, a pointer to an
525 array of char, and a pointer to a function), and have one of the
526 following <a href="#linkage">linkage types</a>.</p>
530 <!-- ======================================================================= -->
531 <div class="doc_subsection">
532 <a name="linkage">Linkage Types</a>
535 <div class="doc_text">
537 <p>All Global Variables and Functions have one of the following types of
541 <dt><tt><b><a name="linkage_private">private</a></b></tt></dt>
542 <dd>Global values with "<tt>private</tt>" linkage are only directly accessible
543 by objects in the current module. In particular, linking code into a
544 module with an private global value may cause the private to be renamed as
545 necessary to avoid collisions. Because the symbol is private to the
546 module, all references can be updated. This doesn't show up in any symbol
547 table in the object file.</dd>
549 <dt><tt><b><a name="linkage_linker_private">linker_private</a></b></tt></dt>
550 <dd>Similar to <tt>private</tt>, but the symbol is passed through the
551 assembler and evaluated by the linker. Unlike normal strong symbols, they
552 are removed by the linker from the final linked image (executable or
553 dynamic library).</dd>
555 <dt><tt><b><a name="linkage_linker_private_weak">linker_private_weak</a></b></tt></dt>
556 <dd>Similar to "<tt>linker_private</tt>", but the symbol is weak. Note that
557 <tt>linker_private_weak</tt> symbols are subject to coalescing by the
558 linker. The symbols are removed by the linker from the final linked image
559 (executable or dynamic library).</dd>
561 <dt><tt><b><a name="linkage_linker_private_weak_def_auto">linker_private_weak_def_auto</a></b></tt></dt>
562 <dd>Similar to "<tt>linker_private_weak</tt>", but it's known that the address
563 of the object is not taken. For instance, functions that had an inline
564 definition, but the compiler decided not to inline it. Note,
565 unlike <tt>linker_private</tt> and <tt>linker_private_weak</tt>,
566 <tt>linker_private_weak_def_auto</tt> may have only <tt>default</tt>
567 visibility. The symbols are removed by the linker from the final linked
568 image (executable or dynamic library).</dd>
570 <dt><tt><b><a name="linkage_internal">internal</a></b></tt></dt>
571 <dd>Similar to private, but the value shows as a local symbol
572 (<tt>STB_LOCAL</tt> in the case of ELF) in the object file. This
573 corresponds to the notion of the '<tt>static</tt>' keyword in C.</dd>
575 <dt><tt><b><a name="linkage_available_externally">available_externally</a></b></tt></dt>
576 <dd>Globals with "<tt>available_externally</tt>" linkage are never emitted
577 into the object file corresponding to the LLVM module. They exist to
578 allow inlining and other optimizations to take place given knowledge of
579 the definition of the global, which is known to be somewhere outside the
580 module. Globals with <tt>available_externally</tt> linkage are allowed to
581 be discarded at will, and are otherwise the same as <tt>linkonce_odr</tt>.
582 This linkage type is only allowed on definitions, not declarations.</dd>
584 <dt><tt><b><a name="linkage_linkonce">linkonce</a></b></tt></dt>
585 <dd>Globals with "<tt>linkonce</tt>" linkage are merged with other globals of
586 the same name when linkage occurs. This can be used to implement
587 some forms of inline functions, templates, or other code which must be
588 generated in each translation unit that uses it, but where the body may
589 be overridden with a more definitive definition later. Unreferenced
590 <tt>linkonce</tt> globals are allowed to be discarded. Note that
591 <tt>linkonce</tt> linkage does not actually allow the optimizer to
592 inline the body of this function into callers because it doesn't know if
593 this definition of the function is the definitive definition within the
594 program or whether it will be overridden by a stronger definition.
595 To enable inlining and other optimizations, use "<tt>linkonce_odr</tt>"
598 <dt><tt><b><a name="linkage_weak">weak</a></b></tt></dt>
599 <dd>"<tt>weak</tt>" linkage has the same merging semantics as
600 <tt>linkonce</tt> linkage, except that unreferenced globals with
601 <tt>weak</tt> linkage may not be discarded. This is used for globals that
602 are declared "weak" in C source code.</dd>
604 <dt><tt><b><a name="linkage_common">common</a></b></tt></dt>
605 <dd>"<tt>common</tt>" linkage is most similar to "<tt>weak</tt>" linkage, but
606 they are used for tentative definitions in C, such as "<tt>int X;</tt>" at
608 Symbols with "<tt>common</tt>" linkage are merged in the same way as
609 <tt>weak symbols</tt>, and they may not be deleted if unreferenced.
610 <tt>common</tt> symbols may not have an explicit section,
611 must have a zero initializer, and may not be marked '<a
612 href="#globalvars"><tt>constant</tt></a>'. Functions and aliases may not
613 have common linkage.</dd>
616 <dt><tt><b><a name="linkage_appending">appending</a></b></tt></dt>
617 <dd>"<tt>appending</tt>" linkage may only be applied to global variables of
618 pointer to array type. When two global variables with appending linkage
619 are linked together, the two global arrays are appended together. This is
620 the LLVM, typesafe, equivalent of having the system linker append together
621 "sections" with identical names when .o files are linked.</dd>
623 <dt><tt><b><a name="linkage_externweak">extern_weak</a></b></tt></dt>
624 <dd>The semantics of this linkage follow the ELF object file model: the symbol
625 is weak until linked, if not linked, the symbol becomes null instead of
626 being an undefined reference.</dd>
628 <dt><tt><b><a name="linkage_linkonce_odr">linkonce_odr</a></b></tt></dt>
629 <dt><tt><b><a name="linkage_weak_odr">weak_odr</a></b></tt></dt>
630 <dd>Some languages allow differing globals to be merged, such as two functions
631 with different semantics. Other languages, such as <tt>C++</tt>, ensure
632 that only equivalent globals are ever merged (the "one definition rule"
633 — "ODR"). Such languages can use the <tt>linkonce_odr</tt>
634 and <tt>weak_odr</tt> linkage types to indicate that the global will only
635 be merged with equivalent globals. These linkage types are otherwise the
636 same as their non-<tt>odr</tt> versions.</dd>
638 <dt><tt><b><a name="linkage_external">externally visible</a></b></tt>:</dt>
639 <dd>If none of the above identifiers are used, the global is externally
640 visible, meaning that it participates in linkage and can be used to
641 resolve external symbol references.</dd>
644 <p>The next two types of linkage are targeted for Microsoft Windows platform
645 only. They are designed to support importing (exporting) symbols from (to)
646 DLLs (Dynamic Link Libraries).</p>
649 <dt><tt><b><a name="linkage_dllimport">dllimport</a></b></tt></dt>
650 <dd>"<tt>dllimport</tt>" linkage causes the compiler to reference a function
651 or variable via a global pointer to a pointer that is set up by the DLL
652 exporting the symbol. On Microsoft Windows targets, the pointer name is
653 formed by combining <code>__imp_</code> and the function or variable
656 <dt><tt><b><a name="linkage_dllexport">dllexport</a></b></tt></dt>
657 <dd>"<tt>dllexport</tt>" linkage causes the compiler to provide a global
658 pointer to a pointer in a DLL, so that it can be referenced with the
659 <tt>dllimport</tt> attribute. On Microsoft Windows targets, the pointer
660 name is formed by combining <code>__imp_</code> and the function or
664 <p>For example, since the "<tt>.LC0</tt>" variable is defined to be internal, if
665 another module defined a "<tt>.LC0</tt>" variable and was linked with this
666 one, one of the two would be renamed, preventing a collision. Since
667 "<tt>main</tt>" and "<tt>puts</tt>" are external (i.e., lacking any linkage
668 declarations), they are accessible outside of the current module.</p>
670 <p>It is illegal for a function <i>declaration</i> to have any linkage type
671 other than "externally visible", <tt>dllimport</tt>
672 or <tt>extern_weak</tt>.</p>
674 <p>Aliases can have only <tt>external</tt>, <tt>internal</tt>, <tt>weak</tt>
675 or <tt>weak_odr</tt> linkages.</p>
679 <!-- ======================================================================= -->
680 <div class="doc_subsection">
681 <a name="callingconv">Calling Conventions</a>
684 <div class="doc_text">
686 <p>LLVM <a href="#functionstructure">functions</a>, <a href="#i_call">calls</a>
687 and <a href="#i_invoke">invokes</a> can all have an optional calling
688 convention specified for the call. The calling convention of any pair of
689 dynamic caller/callee must match, or the behavior of the program is
690 undefined. The following calling conventions are supported by LLVM, and more
691 may be added in the future:</p>
694 <dt><b>"<tt>ccc</tt>" - The C calling convention</b>:</dt>
695 <dd>This calling convention (the default if no other calling convention is
696 specified) matches the target C calling conventions. This calling
697 convention supports varargs function calls and tolerates some mismatch in
698 the declared prototype and implemented declaration of the function (as
701 <dt><b>"<tt>fastcc</tt>" - The fast calling convention</b>:</dt>
702 <dd>This calling convention attempts to make calls as fast as possible
703 (e.g. by passing things in registers). This calling convention allows the
704 target to use whatever tricks it wants to produce fast code for the
705 target, without having to conform to an externally specified ABI
706 (Application Binary Interface).
707 <a href="CodeGenerator.html#tailcallopt">Tail calls can only be optimized
708 when this or the GHC convention is used.</a> This calling convention
709 does not support varargs and requires the prototype of all callees to
710 exactly match the prototype of the function definition.</dd>
712 <dt><b>"<tt>coldcc</tt>" - The cold calling convention</b>:</dt>
713 <dd>This calling convention attempts to make code in the caller as efficient
714 as possible under the assumption that the call is not commonly executed.
715 As such, these calls often preserve all registers so that the call does
716 not break any live ranges in the caller side. This calling convention
717 does not support varargs and requires the prototype of all callees to
718 exactly match the prototype of the function definition.</dd>
720 <dt><b>"<tt>cc <em>10</em></tt>" - GHC convention</b>:</dt>
721 <dd>This calling convention has been implemented specifically for use by the
722 <a href="http://www.haskell.org/ghc">Glasgow Haskell Compiler (GHC)</a>.
723 It passes everything in registers, going to extremes to achieve this by
724 disabling callee save registers. This calling convention should not be
725 used lightly but only for specific situations such as an alternative to
726 the <em>register pinning</em> performance technique often used when
727 implementing functional programming languages.At the moment only X86
728 supports this convention and it has the following limitations:
730 <li>On <em>X86-32</em> only supports up to 4 bit type parameters. No
731 floating point types are supported.</li>
732 <li>On <em>X86-64</em> only supports up to 10 bit type parameters and
733 6 floating point parameters.</li>
735 This calling convention supports
736 <a href="CodeGenerator.html#tailcallopt">tail call optimization</a> but
737 requires both the caller and callee are using it.
740 <dt><b>"<tt>cc <<em>n</em>></tt>" - Numbered convention</b>:</dt>
741 <dd>Any calling convention may be specified by number, allowing
742 target-specific calling conventions to be used. Target specific calling
743 conventions start at 64.</dd>
746 <p>More calling conventions can be added/defined on an as-needed basis, to
747 support Pascal conventions or any other well-known target-independent
752 <!-- ======================================================================= -->
753 <div class="doc_subsection">
754 <a name="visibility">Visibility Styles</a>
757 <div class="doc_text">
759 <p>All Global Variables and Functions have one of the following visibility
763 <dt><b>"<tt>default</tt>" - Default style</b>:</dt>
764 <dd>On targets that use the ELF object file format, default visibility means
765 that the declaration is visible to other modules and, in shared libraries,
766 means that the declared entity may be overridden. On Darwin, default
767 visibility means that the declaration is visible to other modules. Default
768 visibility corresponds to "external linkage" in the language.</dd>
770 <dt><b>"<tt>hidden</tt>" - Hidden style</b>:</dt>
771 <dd>Two declarations of an object with hidden visibility refer to the same
772 object if they are in the same shared object. Usually, hidden visibility
773 indicates that the symbol will not be placed into the dynamic symbol
774 table, so no other module (executable or shared library) can reference it
777 <dt><b>"<tt>protected</tt>" - Protected style</b>:</dt>
778 <dd>On ELF, protected visibility indicates that the symbol will be placed in
779 the dynamic symbol table, but that references within the defining module
780 will bind to the local symbol. That is, the symbol cannot be overridden by
786 <!-- ======================================================================= -->
787 <div class="doc_subsection">
788 <a name="namedtypes">Named Types</a>
791 <div class="doc_text">
793 <p>LLVM IR allows you to specify name aliases for certain types. This can make
794 it easier to read the IR and make the IR more condensed (particularly when
795 recursive types are involved). An example of a name specification is:</p>
797 <pre class="doc_code">
798 %mytype = type { %mytype*, i32 }
801 <p>You may give a name to any <a href="#typesystem">type</a> except
802 "<a href="#t_void">void</a>". Type name aliases may be used anywhere a type
803 is expected with the syntax "%mytype".</p>
805 <p>Note that type names are aliases for the structural type that they indicate,
806 and that you can therefore specify multiple names for the same type. This
807 often leads to confusing behavior when dumping out a .ll file. Since LLVM IR
808 uses structural typing, the name is not part of the type. When printing out
809 LLVM IR, the printer will pick <em>one name</em> to render all types of a
810 particular shape. This means that if you have code where two different
811 source types end up having the same LLVM type, that the dumper will sometimes
812 print the "wrong" or unexpected type. This is an important design point and
813 isn't going to change.</p>
817 <!-- ======================================================================= -->
818 <div class="doc_subsection">
819 <a name="globalvars">Global Variables</a>
822 <div class="doc_text">
824 <p>Global variables define regions of memory allocated at compilation time
825 instead of run-time. Global variables may optionally be initialized, may
826 have an explicit section to be placed in, and may have an optional explicit
827 alignment specified. A variable may be defined as "thread_local", which
828 means that it will not be shared by threads (each thread will have a
829 separated copy of the variable). A variable may be defined as a global
830 "constant," which indicates that the contents of the variable
831 will <b>never</b> be modified (enabling better optimization, allowing the
832 global data to be placed in the read-only section of an executable, etc).
833 Note that variables that need runtime initialization cannot be marked
834 "constant" as there is a store to the variable.</p>
836 <p>LLVM explicitly allows <em>declarations</em> of global variables to be marked
837 constant, even if the final definition of the global is not. This capability
838 can be used to enable slightly better optimization of the program, but
839 requires the language definition to guarantee that optimizations based on the
840 'constantness' are valid for the translation units that do not include the
843 <p>As SSA values, global variables define pointer values that are in scope
844 (i.e. they dominate) all basic blocks in the program. Global variables
845 always define a pointer to their "content" type because they describe a
846 region of memory, and all memory objects in LLVM are accessed through
849 <p>A global variable may be declared to reside in a target-specific numbered
850 address space. For targets that support them, address spaces may affect how
851 optimizations are performed and/or what target instructions are used to
852 access the variable. The default address space is zero. The address space
853 qualifier must precede any other attributes.</p>
855 <p>LLVM allows an explicit section to be specified for globals. If the target
856 supports it, it will emit globals to the section specified.</p>
858 <p>An explicit alignment may be specified for a global, which must be a power
859 of 2. If not present, or if the alignment is set to zero, the alignment of
860 the global is set by the target to whatever it feels convenient. If an
861 explicit alignment is specified, the global is forced to have exactly that
862 alignment. Targets and optimizers are not allowed to over-align the global
863 if the global has an assigned section. In this case, the extra alignment
864 could be observable: for example, code could assume that the globals are
865 densely packed in their section and try to iterate over them as an array,
866 alignment padding would break this iteration.</p>
868 <p>For example, the following defines a global in a numbered address space with
869 an initializer, section, and alignment:</p>
871 <pre class="doc_code">
872 @G = addrspace(5) constant float 1.0, section "foo", align 4
878 <!-- ======================================================================= -->
879 <div class="doc_subsection">
880 <a name="functionstructure">Functions</a>
883 <div class="doc_text">
885 <p>LLVM function definitions consist of the "<tt>define</tt>" keyword, an
886 optional <a href="#linkage">linkage type</a>, an optional
887 <a href="#visibility">visibility style</a>, an optional
888 <a href="#callingconv">calling convention</a>, a return type, an optional
889 <a href="#paramattrs">parameter attribute</a> for the return type, a function
890 name, a (possibly empty) argument list (each with optional
891 <a href="#paramattrs">parameter attributes</a>), optional
892 <a href="#fnattrs">function attributes</a>, an optional section, an optional
893 alignment, an optional <a href="#gc">garbage collector name</a>, an opening
894 curly brace, a list of basic blocks, and a closing curly brace.</p>
896 <p>LLVM function declarations consist of the "<tt>declare</tt>" keyword, an
897 optional <a href="#linkage">linkage type</a>, an optional
898 <a href="#visibility">visibility style</a>, an optional
899 <a href="#callingconv">calling convention</a>, a return type, an optional
900 <a href="#paramattrs">parameter attribute</a> for the return type, a function
901 name, a possibly empty list of arguments, an optional alignment, and an
902 optional <a href="#gc">garbage collector name</a>.</p>
904 <p>A function definition contains a list of basic blocks, forming the CFG
905 (Control Flow Graph) for the function. Each basic block may optionally start
906 with a label (giving the basic block a symbol table entry), contains a list
907 of instructions, and ends with a <a href="#terminators">terminator</a>
908 instruction (such as a branch or function return).</p>
910 <p>The first basic block in a function is special in two ways: it is immediately
911 executed on entrance to the function, and it is not allowed to have
912 predecessor basic blocks (i.e. there can not be any branches to the entry
913 block of a function). Because the block can have no predecessors, it also
914 cannot have any <a href="#i_phi">PHI nodes</a>.</p>
916 <p>LLVM allows an explicit section to be specified for functions. If the target
917 supports it, it will emit functions to the section specified.</p>
919 <p>An explicit alignment may be specified for a function. If not present, or if
920 the alignment is set to zero, the alignment of the function is set by the
921 target to whatever it feels convenient. If an explicit alignment is
922 specified, the function is forced to have at least that much alignment. All
923 alignments must be a power of 2.</p>
926 <pre class="doc_code">
927 define [<a href="#linkage">linkage</a>] [<a href="#visibility">visibility</a>]
928 [<a href="#callingconv">cconv</a>] [<a href="#paramattrs">ret attrs</a>]
929 <ResultType> @<FunctionName> ([argument list])
930 [<a href="#fnattrs">fn Attrs</a>] [section "name"] [align N]
931 [<a href="#gc">gc</a>] { ... }
936 <!-- ======================================================================= -->
937 <div class="doc_subsection">
938 <a name="aliasstructure">Aliases</a>
941 <div class="doc_text">
943 <p>Aliases act as "second name" for the aliasee value (which can be either
944 function, global variable, another alias or bitcast of global value). Aliases
945 may have an optional <a href="#linkage">linkage type</a>, and an
946 optional <a href="#visibility">visibility style</a>.</p>
949 <pre class="doc_code">
950 @<Name> = alias [Linkage] [Visibility] <AliaseeTy> @<Aliasee>
955 <!-- ======================================================================= -->
956 <div class="doc_subsection">
957 <a name="namedmetadatastructure">Named Metadata</a>
960 <div class="doc_text">
962 <p>Named metadata is a collection of metadata. <a href="#metadata">Metadata
963 nodes</a> (but not metadata strings) are the only valid operands for
964 a named metadata.</p>
967 <pre class="doc_code">
968 ; Some unnamed metadata nodes, which are referenced by the named metadata.
969 !0 = metadata !{metadata !"zero"}
970 !1 = metadata !{metadata !"one"}
971 !2 = metadata !{metadata !"two"}
973 !name = !{!0, !1, !2}
978 <!-- ======================================================================= -->
979 <div class="doc_subsection"><a name="paramattrs">Parameter Attributes</a></div>
981 <div class="doc_text">
983 <p>The return type and each parameter of a function type may have a set of
984 <i>parameter attributes</i> associated with them. Parameter attributes are
985 used to communicate additional information about the result or parameters of
986 a function. Parameter attributes are considered to be part of the function,
987 not of the function type, so functions with different parameter attributes
988 can have the same function type.</p>
990 <p>Parameter attributes are simple keywords that follow the type specified. If
991 multiple parameter attributes are needed, they are space separated. For
994 <pre class="doc_code">
995 declare i32 @printf(i8* noalias nocapture, ...)
996 declare i32 @atoi(i8 zeroext)
997 declare signext i8 @returns_signed_char()
1000 <p>Note that any attributes for the function result (<tt>nounwind</tt>,
1001 <tt>readonly</tt>) come immediately after the argument list.</p>
1003 <p>Currently, only the following parameter attributes are defined:</p>
1006 <dt><tt><b>zeroext</b></tt></dt>
1007 <dd>This indicates to the code generator that the parameter or return value
1008 should be zero-extended to a 32-bit value by the caller (for a parameter)
1009 or the callee (for a return value).</dd>
1011 <dt><tt><b>signext</b></tt></dt>
1012 <dd>This indicates to the code generator that the parameter or return value
1013 should be sign-extended to a 32-bit value by the caller (for a parameter)
1014 or the callee (for a return value).</dd>
1016 <dt><tt><b>inreg</b></tt></dt>
1017 <dd>This indicates that this parameter or return value should be treated in a
1018 special target-dependent fashion during while emitting code for a function
1019 call or return (usually, by putting it in a register as opposed to memory,
1020 though some targets use it to distinguish between two different kinds of
1021 registers). Use of this attribute is target-specific.</dd>
1023 <dt><tt><b><a name="byval">byval</a></b></tt></dt>
1024 <dd>This indicates that the pointer parameter should really be passed by value
1025 to the function. The attribute implies that a hidden copy of the pointee
1026 is made between the caller and the callee, so the callee is unable to
1027 modify the value in the callee. This attribute is only valid on LLVM
1028 pointer arguments. It is generally used to pass structs and arrays by
1029 value, but is also valid on pointers to scalars. The copy is considered
1030 to belong to the caller not the callee (for example,
1031 <tt><a href="#readonly">readonly</a></tt> functions should not write to
1032 <tt>byval</tt> parameters). This is not a valid attribute for return
1033 values. The byval attribute also supports specifying an alignment with
1034 the align attribute. This has a target-specific effect on the code
1035 generator that usually indicates a desired alignment for the synthesized
1038 <dt><tt><b><a name="sret">sret</a></b></tt></dt>
1039 <dd>This indicates that the pointer parameter specifies the address of a
1040 structure that is the return value of the function in the source program.
1041 This pointer must be guaranteed by the caller to be valid: loads and
1042 stores to the structure may be assumed by the callee to not to trap. This
1043 may only be applied to the first parameter. This is not a valid attribute
1044 for return values. </dd>
1046 <dt><tt><b><a name="noalias">noalias</a></b></tt></dt>
1047 <dd>This indicates that pointer values
1048 <a href="#pointeraliasing"><i>based</i></a> on the argument or return
1049 value do not alias pointer values which are not <i>based</i> on it,
1050 ignoring certain "irrelevant" dependencies.
1051 For a call to the parent function, dependencies between memory
1052 references from before or after the call and from those during the call
1053 are "irrelevant" to the <tt>noalias</tt> keyword for the arguments and
1054 return value used in that call.
1055 The caller shares the responsibility with the callee for ensuring that
1056 these requirements are met.
1057 For further details, please see the discussion of the NoAlias response in
1058 <a href="AliasAnalysis.html#MustMayNo">alias analysis</a>.<br>
1060 Note that this definition of <tt>noalias</tt> is intentionally
1061 similar to the definition of <tt>restrict</tt> in C99 for function
1062 arguments, though it is slightly weaker.
1064 For function return values, C99's <tt>restrict</tt> is not meaningful,
1065 while LLVM's <tt>noalias</tt> is.
1068 <dt><tt><b><a name="nocapture">nocapture</a></b></tt></dt>
1069 <dd>This indicates that the callee does not make any copies of the pointer
1070 that outlive the callee itself. This is not a valid attribute for return
1073 <dt><tt><b><a name="nest">nest</a></b></tt></dt>
1074 <dd>This indicates that the pointer parameter can be excised using the
1075 <a href="#int_trampoline">trampoline intrinsics</a>. This is not a valid
1076 attribute for return values.</dd>
1081 <!-- ======================================================================= -->
1082 <div class="doc_subsection">
1083 <a name="gc">Garbage Collector Names</a>
1086 <div class="doc_text">
1088 <p>Each function may specify a garbage collector name, which is simply a
1091 <pre class="doc_code">
1092 define void @f() gc "name" { ... }
1095 <p>The compiler declares the supported values of <i>name</i>. Specifying a
1096 collector which will cause the compiler to alter its output in order to
1097 support the named garbage collection algorithm.</p>
1101 <!-- ======================================================================= -->
1102 <div class="doc_subsection">
1103 <a name="fnattrs">Function Attributes</a>
1106 <div class="doc_text">
1108 <p>Function attributes are set to communicate additional information about a
1109 function. Function attributes are considered to be part of the function, not
1110 of the function type, so functions with different parameter attributes can
1111 have the same function type.</p>
1113 <p>Function attributes are simple keywords that follow the type specified. If
1114 multiple attributes are needed, they are space separated. For example:</p>
1116 <pre class="doc_code">
1117 define void @f() noinline { ... }
1118 define void @f() alwaysinline { ... }
1119 define void @f() alwaysinline optsize { ... }
1120 define void @f() optsize { ... }
1124 <dt><tt><b>alignstack(<<em>n</em>>)</b></tt></dt>
1125 <dd>This attribute indicates that, when emitting the prologue and epilogue,
1126 the backend should forcibly align the stack pointer. Specify the
1127 desired alignment, which must be a power of two, in parentheses.
1129 <dt><tt><b>alwaysinline</b></tt></dt>
1130 <dd>This attribute indicates that the inliner should attempt to inline this
1131 function into callers whenever possible, ignoring any active inlining size
1132 threshold for this caller.</dd>
1134 <dt><tt><b>hotpatch</b></tt></dt>
1135 <dd>This attribute indicates that the prologue should contain a 'hotpatch'
1136 sequence at the beginning. This is the same sequence used in the
1137 system DLLs in Microsoft Windows XP Service Pack 2 and higher.</dd>
1139 <dt><tt><b>inlinehint</b></tt></dt>
1140 <dd>This attribute indicates that the source code contained a hint that inlining
1141 this function is desirable (such as the "inline" keyword in C/C++). It
1142 is just a hint; it imposes no requirements on the inliner.</dd>
1144 <dt><tt><b>naked</b></tt></dt>
1145 <dd>This attribute disables prologue / epilogue emission for the function.
1146 This can have very system-specific consequences.</dd>
1148 <dt><tt><b>noimplicitfloat</b></tt></dt>
1149 <dd>This attributes disables implicit floating point instructions.</dd>
1151 <dt><tt><b>noinline</b></tt></dt>
1152 <dd>This attribute indicates that the inliner should never inline this
1153 function in any situation. This attribute may not be used together with
1154 the <tt>alwaysinline</tt> attribute.</dd>
1156 <dt><tt><b>noredzone</b></tt></dt>
1157 <dd>This attribute indicates that the code generator should not use a red
1158 zone, even if the target-specific ABI normally permits it.</dd>
1160 <dt><tt><b>noreturn</b></tt></dt>
1161 <dd>This function attribute indicates that the function never returns
1162 normally. This produces undefined behavior at runtime if the function
1163 ever does dynamically return.</dd>
1165 <dt><tt><b>nounwind</b></tt></dt>
1166 <dd>This function attribute indicates that the function never returns with an
1167 unwind or exceptional control flow. If the function does unwind, its
1168 runtime behavior is undefined.</dd>
1170 <dt><tt><b>optsize</b></tt></dt>
1171 <dd>This attribute suggests that optimization passes and code generator passes
1172 make choices that keep the code size of this function low, and otherwise
1173 do optimizations specifically to reduce code size.</dd>
1175 <dt><tt><b>readnone</b></tt></dt>
1176 <dd>This attribute indicates that the function computes its result (or decides
1177 to unwind an exception) based strictly on its arguments, without
1178 dereferencing any pointer arguments or otherwise accessing any mutable
1179 state (e.g. memory, control registers, etc) visible to caller functions.
1180 It does not write through any pointer arguments
1181 (including <tt><a href="#byval">byval</a></tt> arguments) and never
1182 changes any state visible to callers. This means that it cannot unwind
1183 exceptions by calling the <tt>C++</tt> exception throwing methods, but
1184 could use the <tt>unwind</tt> instruction.</dd>
1186 <dt><tt><b><a name="readonly">readonly</a></b></tt></dt>
1187 <dd>This attribute indicates that the function does not write through any
1188 pointer arguments (including <tt><a href="#byval">byval</a></tt>
1189 arguments) or otherwise modify any state (e.g. memory, control registers,
1190 etc) visible to caller functions. It may dereference pointer arguments
1191 and read state that may be set in the caller. A readonly function always
1192 returns the same value (or unwinds an exception identically) when called
1193 with the same set of arguments and global state. It cannot unwind an
1194 exception by calling the <tt>C++</tt> exception throwing methods, but may
1195 use the <tt>unwind</tt> instruction.</dd>
1197 <dt><tt><b><a name="ssp">ssp</a></b></tt></dt>
1198 <dd>This attribute indicates that the function should emit a stack smashing
1199 protector. It is in the form of a "canary"—a random value placed on
1200 the stack before the local variables that's checked upon return from the
1201 function to see if it has been overwritten. A heuristic is used to
1202 determine if a function needs stack protectors or not.<br>
1204 If a function that has an <tt>ssp</tt> attribute is inlined into a
1205 function that doesn't have an <tt>ssp</tt> attribute, then the resulting
1206 function will have an <tt>ssp</tt> attribute.</dd>
1208 <dt><tt><b>sspreq</b></tt></dt>
1209 <dd>This attribute indicates that the function should <em>always</em> emit a
1210 stack smashing protector. This overrides
1211 the <tt><a href="#ssp">ssp</a></tt> function attribute.<br>
1213 If a function that has an <tt>sspreq</tt> attribute is inlined into a
1214 function that doesn't have an <tt>sspreq</tt> attribute or which has
1215 an <tt>ssp</tt> attribute, then the resulting function will have
1216 an <tt>sspreq</tt> attribute.</dd>
1221 <!-- ======================================================================= -->
1222 <div class="doc_subsection">
1223 <a name="moduleasm">Module-Level Inline Assembly</a>
1226 <div class="doc_text">
1228 <p>Modules may contain "module-level inline asm" blocks, which corresponds to
1229 the GCC "file scope inline asm" blocks. These blocks are internally
1230 concatenated by LLVM and treated as a single unit, but may be separated in
1231 the <tt>.ll</tt> file if desired. The syntax is very simple:</p>
1233 <pre class="doc_code">
1234 module asm "inline asm code goes here"
1235 module asm "more can go here"
1238 <p>The strings can contain any character by escaping non-printable characters.
1239 The escape sequence used is simply "\xx" where "xx" is the two digit hex code
1242 <p>The inline asm code is simply printed to the machine code .s file when
1243 assembly code is generated.</p>
1247 <!-- ======================================================================= -->
1248 <div class="doc_subsection">
1249 <a name="datalayout">Data Layout</a>
1252 <div class="doc_text">
1254 <p>A module may specify a target specific data layout string that specifies how
1255 data is to be laid out in memory. The syntax for the data layout is
1258 <pre class="doc_code">
1259 target datalayout = "<i>layout specification</i>"
1262 <p>The <i>layout specification</i> consists of a list of specifications
1263 separated by the minus sign character ('-'). Each specification starts with
1264 a letter and may include other information after the letter to define some
1265 aspect of the data layout. The specifications accepted are as follows:</p>
1269 <dd>Specifies that the target lays out data in big-endian form. That is, the
1270 bits with the most significance have the lowest address location.</dd>
1273 <dd>Specifies that the target lays out data in little-endian form. That is,
1274 the bits with the least significance have the lowest address
1277 <dt><tt>p:<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1278 <dd>This specifies the <i>size</i> of a pointer and its <i>abi</i> and
1279 <i>preferred</i> alignments. All sizes are in bits. Specifying
1280 the <i>pref</i> alignment is optional. If omitted, the
1281 preceding <tt>:</tt> should be omitted too.</dd>
1283 <dt><tt>i<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1284 <dd>This specifies the alignment for an integer type of a given bit
1285 <i>size</i>. The value of <i>size</i> must be in the range [1,2^23).</dd>
1287 <dt><tt>v<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1288 <dd>This specifies the alignment for a vector type of a given bit
1291 <dt><tt>f<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1292 <dd>This specifies the alignment for a floating point type of a given bit
1293 <i>size</i>. Only values of <i>size</i> that are supported by the target
1294 will work. 32 (float) and 64 (double) are supported on all targets;
1295 80 or 128 (different flavors of long double) are also supported on some
1298 <dt><tt>a<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1299 <dd>This specifies the alignment for an aggregate type of a given bit
1302 <dt><tt>s<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1303 <dd>This specifies the alignment for a stack object of a given bit
1306 <dt><tt>n<i>size1</i>:<i>size2</i>:<i>size3</i>...</tt></dt>
1307 <dd>This specifies a set of native integer widths for the target CPU
1308 in bits. For example, it might contain "n32" for 32-bit PowerPC,
1309 "n32:64" for PowerPC 64, or "n8:16:32:64" for X86-64. Elements of
1310 this set are considered to support most general arithmetic
1311 operations efficiently.</dd>
1314 <p>When constructing the data layout for a given target, LLVM starts with a
1315 default set of specifications which are then (possibly) overridden by the
1316 specifications in the <tt>datalayout</tt> keyword. The default specifications
1317 are given in this list:</p>
1320 <li><tt>E</tt> - big endian</li>
1321 <li><tt>p:64:64:64</tt> - 64-bit pointers with 64-bit alignment</li>
1322 <li><tt>i1:8:8</tt> - i1 is 8-bit (byte) aligned</li>
1323 <li><tt>i8:8:8</tt> - i8 is 8-bit (byte) aligned</li>
1324 <li><tt>i16:16:16</tt> - i16 is 16-bit aligned</li>
1325 <li><tt>i32:32:32</tt> - i32 is 32-bit aligned</li>
1326 <li><tt>i64:32:64</tt> - i64 has ABI alignment of 32-bits but preferred
1327 alignment of 64-bits</li>
1328 <li><tt>f32:32:32</tt> - float is 32-bit aligned</li>
1329 <li><tt>f64:64:64</tt> - double is 64-bit aligned</li>
1330 <li><tt>v64:64:64</tt> - 64-bit vector is 64-bit aligned</li>
1331 <li><tt>v128:128:128</tt> - 128-bit vector is 128-bit aligned</li>
1332 <li><tt>a0:0:1</tt> - aggregates are 8-bit aligned</li>
1333 <li><tt>s0:64:64</tt> - stack objects are 64-bit aligned</li>
1336 <p>When LLVM is determining the alignment for a given type, it uses the
1337 following rules:</p>
1340 <li>If the type sought is an exact match for one of the specifications, that
1341 specification is used.</li>
1343 <li>If no match is found, and the type sought is an integer type, then the
1344 smallest integer type that is larger than the bitwidth of the sought type
1345 is used. If none of the specifications are larger than the bitwidth then
1346 the the largest integer type is used. For example, given the default
1347 specifications above, the i7 type will use the alignment of i8 (next
1348 largest) while both i65 and i256 will use the alignment of i64 (largest
1351 <li>If no match is found, and the type sought is a vector type, then the
1352 largest vector type that is smaller than the sought vector type will be
1353 used as a fall back. This happens because <128 x double> can be
1354 implemented in terms of 64 <2 x double>, for example.</li>
1359 <!-- ======================================================================= -->
1360 <div class="doc_subsection">
1361 <a name="pointeraliasing">Pointer Aliasing Rules</a>
1364 <div class="doc_text">
1366 <p>Any memory access must be done through a pointer value associated
1367 with an address range of the memory access, otherwise the behavior
1368 is undefined. Pointer values are associated with address ranges
1369 according to the following rules:</p>
1372 <li>A pointer value is associated with the addresses associated with
1373 any value it is <i>based</i> on.
1374 <li>An address of a global variable is associated with the address
1375 range of the variable's storage.</li>
1376 <li>The result value of an allocation instruction is associated with
1377 the address range of the allocated storage.</li>
1378 <li>A null pointer in the default address-space is associated with
1380 <li>An integer constant other than zero or a pointer value returned
1381 from a function not defined within LLVM may be associated with address
1382 ranges allocated through mechanisms other than those provided by
1383 LLVM. Such ranges shall not overlap with any ranges of addresses
1384 allocated by mechanisms provided by LLVM.</li>
1387 <p>A pointer value is <i>based</i> on another pointer value according
1388 to the following rules:</p>
1391 <li>A pointer value formed from a
1392 <tt><a href="#i_getelementptr">getelementptr</a></tt> operation
1393 is <i>based</i> on the first operand of the <tt>getelementptr</tt>.</li>
1394 <li>The result value of a
1395 <tt><a href="#i_bitcast">bitcast</a></tt> is <i>based</i> on the operand
1396 of the <tt>bitcast</tt>.</li>
1397 <li>A pointer value formed by an
1398 <tt><a href="#i_inttoptr">inttoptr</a></tt> is <i>based</i> on all
1399 pointer values that contribute (directly or indirectly) to the
1400 computation of the pointer's value.</li>
1401 <li>The "<i>based</i> on" relationship is transitive.</li>
1404 <p>Note that this definition of <i>"based"</i> is intentionally
1405 similar to the definition of <i>"based"</i> in C99, though it is
1406 slightly weaker.</p>
1408 <p>LLVM IR does not associate types with memory. The result type of a
1409 <tt><a href="#i_load">load</a></tt> merely indicates the size and
1410 alignment of the memory from which to load, as well as the
1411 interpretation of the value. The first operand type of a
1412 <tt><a href="#i_store">store</a></tt> similarly only indicates the size
1413 and alignment of the store.</p>
1415 <p>Consequently, type-based alias analysis, aka TBAA, aka
1416 <tt>-fstrict-aliasing</tt>, is not applicable to general unadorned
1417 LLVM IR. <a href="#metadata">Metadata</a> may be used to encode
1418 additional information which specialized optimization passes may use
1419 to implement type-based alias analysis.</p>
1423 <!-- ======================================================================= -->
1424 <div class="doc_subsection">
1425 <a name="volatile">Volatile Memory Accesses</a>
1428 <div class="doc_text">
1430 <p>Certain memory accesses, such as <a href="#i_load"><tt>load</tt></a>s, <a
1431 href="#i_store"><tt>store</tt></a>s, and <a
1432 href="#int_memcpy"><tt>llvm.memcpy</tt></a>s may be marked <tt>volatile</tt>.
1433 The optimizers must not change the number of volatile operations or change their
1434 order of execution relative to other volatile operations. The optimizers
1435 <i>may</i> change the order of volatile operations relative to non-volatile
1436 operations. This is not Java's "volatile" and has no cross-thread
1437 synchronization behavior.</p>
1441 <!-- *********************************************************************** -->
1442 <div class="doc_section"> <a name="typesystem">Type System</a> </div>
1443 <!-- *********************************************************************** -->
1445 <div class="doc_text">
1447 <p>The LLVM type system is one of the most important features of the
1448 intermediate representation. Being typed enables a number of optimizations
1449 to be performed on the intermediate representation directly, without having
1450 to do extra analyses on the side before the transformation. A strong type
1451 system makes it easier to read the generated code and enables novel analyses
1452 and transformations that are not feasible to perform on normal three address
1453 code representations.</p>
1457 <!-- ======================================================================= -->
1458 <div class="doc_subsection"> <a name="t_classifications">Type
1459 Classifications</a> </div>
1461 <div class="doc_text">
1463 <p>The types fall into a few useful classifications:</p>
1465 <table border="1" cellspacing="0" cellpadding="4">
1467 <tr><th>Classification</th><th>Types</th></tr>
1469 <td><a href="#t_integer">integer</a></td>
1470 <td><tt>i1, i2, i3, ... i8, ... i16, ... i32, ... i64, ... </tt></td>
1473 <td><a href="#t_floating">floating point</a></td>
1474 <td><tt>float, double, x86_fp80, fp128, ppc_fp128</tt></td>
1477 <td><a name="t_firstclass">first class</a></td>
1478 <td><a href="#t_integer">integer</a>,
1479 <a href="#t_floating">floating point</a>,
1480 <a href="#t_pointer">pointer</a>,
1481 <a href="#t_vector">vector</a>,
1482 <a href="#t_struct">structure</a>,
1483 <a href="#t_array">array</a>,
1484 <a href="#t_label">label</a>,
1485 <a href="#t_metadata">metadata</a>.
1489 <td><a href="#t_primitive">primitive</a></td>
1490 <td><a href="#t_label">label</a>,
1491 <a href="#t_void">void</a>,
1492 <a href="#t_floating">floating point</a>,
1493 <a href="#t_x86mmx">x86mmx</a>,
1494 <a href="#t_metadata">metadata</a>.</td>
1497 <td><a href="#t_derived">derived</a></td>
1498 <td><a href="#t_array">array</a>,
1499 <a href="#t_function">function</a>,
1500 <a href="#t_pointer">pointer</a>,
1501 <a href="#t_struct">structure</a>,
1502 <a href="#t_pstruct">packed structure</a>,
1503 <a href="#t_vector">vector</a>,
1504 <a href="#t_opaque">opaque</a>.
1510 <p>The <a href="#t_firstclass">first class</a> types are perhaps the most
1511 important. Values of these types are the only ones which can be produced by
1516 <!-- ======================================================================= -->
1517 <div class="doc_subsection"> <a name="t_primitive">Primitive Types</a> </div>
1519 <div class="doc_text">
1521 <p>The primitive types are the fundamental building blocks of the LLVM
1526 <!-- _______________________________________________________________________ -->
1527 <div class="doc_subsubsection"> <a name="t_integer">Integer Type</a> </div>
1529 <div class="doc_text">
1532 <p>The integer type is a very simple type that simply specifies an arbitrary
1533 bit width for the integer type desired. Any bit width from 1 bit to
1534 2<sup>23</sup>-1 (about 8 million) can be specified.</p>
1541 <p>The number of bits the integer will occupy is specified by the <tt>N</tt>
1545 <table class="layout">
1547 <td class="left"><tt>i1</tt></td>
1548 <td class="left">a single-bit integer.</td>
1551 <td class="left"><tt>i32</tt></td>
1552 <td class="left">a 32-bit integer.</td>
1555 <td class="left"><tt>i1942652</tt></td>
1556 <td class="left">a really big integer of over 1 million bits.</td>
1562 <!-- _______________________________________________________________________ -->
1563 <div class="doc_subsubsection"> <a name="t_floating">Floating Point Types</a> </div>
1565 <div class="doc_text">
1569 <tr><th>Type</th><th>Description</th></tr>
1570 <tr><td><tt>float</tt></td><td>32-bit floating point value</td></tr>
1571 <tr><td><tt>double</tt></td><td>64-bit floating point value</td></tr>
1572 <tr><td><tt>fp128</tt></td><td>128-bit floating point value (112-bit mantissa)</td></tr>
1573 <tr><td><tt>x86_fp80</tt></td><td>80-bit floating point value (X87)</td></tr>
1574 <tr><td><tt>ppc_fp128</tt></td><td>128-bit floating point value (two 64-bits)</td></tr>
1580 <!-- _______________________________________________________________________ -->
1581 <div class="doc_subsubsection"> <a name="t_x86mmx">X86mmx Type</a> </div>
1583 <div class="doc_text">
1586 <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>
1595 <!-- _______________________________________________________________________ -->
1596 <div class="doc_subsubsection"> <a name="t_void">Void Type</a> </div>
1598 <div class="doc_text">
1601 <p>The void type does not represent any value and has no size.</p>
1610 <!-- _______________________________________________________________________ -->
1611 <div class="doc_subsubsection"> <a name="t_label">Label Type</a> </div>
1613 <div class="doc_text">
1616 <p>The label type represents code labels.</p>
1625 <!-- _______________________________________________________________________ -->
1626 <div class="doc_subsubsection"> <a name="t_metadata">Metadata Type</a> </div>
1628 <div class="doc_text">
1631 <p>The metadata type represents embedded metadata. No derived types may be
1632 created from metadata except for <a href="#t_function">function</a>
1643 <!-- ======================================================================= -->
1644 <div class="doc_subsection"> <a name="t_derived">Derived Types</a> </div>
1646 <div class="doc_text">
1648 <p>The real power in LLVM comes from the derived types in the system. This is
1649 what allows a programmer to represent arrays, functions, pointers, and other
1650 useful types. Each of these types contain one or more element types which
1651 may be a primitive type, or another derived type. For example, it is
1652 possible to have a two dimensional array, using an array as the element type
1653 of another array.</p>
1658 <!-- _______________________________________________________________________ -->
1659 <div class="doc_subsubsection"> <a name="t_aggregate">Aggregate Types</a> </div>
1661 <div class="doc_text">
1663 <p>Aggregate Types are a subset of derived types that can contain multiple
1664 member types. <a href="#t_array">Arrays</a>,
1665 <a href="#t_struct">structs</a>, and <a href="#t_vector">vectors</a> are
1666 aggregate types.</p>
1670 <!-- _______________________________________________________________________ -->
1671 <div class="doc_subsubsection"> <a name="t_array">Array Type</a> </div>
1673 <div class="doc_text">
1676 <p>The array type is a very simple derived type that arranges elements
1677 sequentially in memory. The array type requires a size (number of elements)
1678 and an underlying data type.</p>
1682 [<# elements> x <elementtype>]
1685 <p>The number of elements is a constant integer value; <tt>elementtype</tt> may
1686 be any type with a size.</p>
1689 <table class="layout">
1691 <td class="left"><tt>[40 x i32]</tt></td>
1692 <td class="left">Array of 40 32-bit integer values.</td>
1695 <td class="left"><tt>[41 x i32]</tt></td>
1696 <td class="left">Array of 41 32-bit integer values.</td>
1699 <td class="left"><tt>[4 x i8]</tt></td>
1700 <td class="left">Array of 4 8-bit integer values.</td>
1703 <p>Here are some examples of multidimensional arrays:</p>
1704 <table class="layout">
1706 <td class="left"><tt>[3 x [4 x i32]]</tt></td>
1707 <td class="left">3x4 array of 32-bit integer values.</td>
1710 <td class="left"><tt>[12 x [10 x float]]</tt></td>
1711 <td class="left">12x10 array of single precision floating point values.</td>
1714 <td class="left"><tt>[2 x [3 x [4 x i16]]]</tt></td>
1715 <td class="left">2x3x4 array of 16-bit integer values.</td>
1719 <p>There is no restriction on indexing beyond the end of the array implied by
1720 a static type (though there are restrictions on indexing beyond the bounds
1721 of an allocated object in some cases). This means that single-dimension
1722 'variable sized array' addressing can be implemented in LLVM with a zero
1723 length array type. An implementation of 'pascal style arrays' in LLVM could
1724 use the type "<tt>{ i32, [0 x float]}</tt>", for example.</p>
1728 <!-- _______________________________________________________________________ -->
1729 <div class="doc_subsubsection"> <a name="t_function">Function Type</a> </div>
1731 <div class="doc_text">
1734 <p>The function type can be thought of as a function signature. It consists of
1735 a return type and a list of formal parameter types. The return type of a
1736 function type is a first class type or a void type.</p>
1740 <returntype> (<parameter list>)
1743 <p>...where '<tt><parameter list></tt>' is a comma-separated list of type
1744 specifiers. Optionally, the parameter list may include a type <tt>...</tt>,
1745 which indicates that the function takes a variable number of arguments.
1746 Variable argument functions can access their arguments with
1747 the <a href="#int_varargs">variable argument handling intrinsic</a>
1748 functions. '<tt><returntype></tt>' is any type except
1749 <a href="#t_label">label</a>.</p>
1752 <table class="layout">
1754 <td class="left"><tt>i32 (i32)</tt></td>
1755 <td class="left">function taking an <tt>i32</tt>, returning an <tt>i32</tt>
1757 </tr><tr class="layout">
1758 <td class="left"><tt>float (i16, i32 *) *
1760 <td class="left"><a href="#t_pointer">Pointer</a> to a function that takes
1761 an <tt>i16</tt> and a <a href="#t_pointer">pointer</a> to <tt>i32</tt>,
1762 returning <tt>float</tt>.
1764 </tr><tr class="layout">
1765 <td class="left"><tt>i32 (i8*, ...)</tt></td>
1766 <td class="left">A vararg function that takes at least one
1767 <a href="#t_pointer">pointer</a> to <tt>i8 </tt> (char in C),
1768 which returns an integer. This is the signature for <tt>printf</tt> in
1771 </tr><tr class="layout">
1772 <td class="left"><tt>{i32, i32} (i32)</tt></td>
1773 <td class="left">A function taking an <tt>i32</tt>, returning a
1774 <a href="#t_struct">structure</a> containing two <tt>i32</tt> values
1781 <!-- _______________________________________________________________________ -->
1782 <div class="doc_subsubsection"> <a name="t_struct">Structure Type</a> </div>
1784 <div class="doc_text">
1787 <p>The structure type is used to represent a collection of data members together
1788 in memory. The packing of the field types is defined to match the ABI of the
1789 underlying processor. The elements of a structure may be any type that has a
1792 <p>Structures in memory are accessed using '<tt><a href="#i_load">load</a></tt>'
1793 and '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a field
1794 with the '<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.
1795 Structures in registers are accessed using the
1796 '<tt><a href="#i_extractvalue">extractvalue</a></tt>' and
1797 '<tt><a href="#i_insertvalue">insertvalue</a></tt>' instructions.</p>
1800 { <type list> }
1804 <table class="layout">
1806 <td class="left"><tt>{ i32, i32, i32 }</tt></td>
1807 <td class="left">A triple of three <tt>i32</tt> values</td>
1808 </tr><tr class="layout">
1809 <td class="left"><tt>{ float, i32 (i32) * }</tt></td>
1810 <td class="left">A pair, where the first element is a <tt>float</tt> and the
1811 second element is a <a href="#t_pointer">pointer</a> to a
1812 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
1813 an <tt>i32</tt>.</td>
1819 <!-- _______________________________________________________________________ -->
1820 <div class="doc_subsubsection"> <a name="t_pstruct">Packed Structure Type</a>
1823 <div class="doc_text">
1826 <p>The packed structure type is used to represent a collection of data members
1827 together in memory. There is no padding between fields. Further, the
1828 alignment of a packed structure is 1 byte. The elements of a packed
1829 structure may be any type that has a size.</p>
1831 <p>Structures are accessed using '<tt><a href="#i_load">load</a></tt> and
1832 '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a field with
1833 the '<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.</p>
1837 < { <type list> } >
1841 <table class="layout">
1843 <td class="left"><tt>< { i32, i32, i32 } ></tt></td>
1844 <td class="left">A triple of three <tt>i32</tt> values</td>
1845 </tr><tr class="layout">
1847 <tt>< { float, i32 (i32)* } ></tt></td>
1848 <td class="left">A pair, where the first element is a <tt>float</tt> and the
1849 second element is a <a href="#t_pointer">pointer</a> to a
1850 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
1851 an <tt>i32</tt>.</td>
1857 <!-- _______________________________________________________________________ -->
1858 <div class="doc_subsubsection"> <a name="t_pointer">Pointer Type</a> </div>
1860 <div class="doc_text">
1863 <p>The pointer type is used to specify memory locations.
1864 Pointers are commonly used to reference objects in memory.</p>
1866 <p>Pointer types may have an optional address space attribute defining the
1867 numbered address space where the pointed-to object resides. The default
1868 address space is number zero. The semantics of non-zero address
1869 spaces are target-specific.</p>
1871 <p>Note that LLVM does not permit pointers to void (<tt>void*</tt>) nor does it
1872 permit pointers to labels (<tt>label*</tt>). Use <tt>i8*</tt> instead.</p>
1880 <table class="layout">
1882 <td class="left"><tt>[4 x i32]*</tt></td>
1883 <td class="left">A <a href="#t_pointer">pointer</a> to <a
1884 href="#t_array">array</a> of four <tt>i32</tt> values.</td>
1887 <td class="left"><tt>i32 (i32*) *</tt></td>
1888 <td class="left"> A <a href="#t_pointer">pointer</a> to a <a
1889 href="#t_function">function</a> that takes an <tt>i32*</tt>, returning an
1893 <td class="left"><tt>i32 addrspace(5)*</tt></td>
1894 <td class="left">A <a href="#t_pointer">pointer</a> to an <tt>i32</tt> value
1895 that resides in address space #5.</td>
1901 <!-- _______________________________________________________________________ -->
1902 <div class="doc_subsubsection"> <a name="t_vector">Vector Type</a> </div>
1904 <div class="doc_text">
1907 <p>A vector type is a simple derived type that represents a vector of elements.
1908 Vector types are used when multiple primitive data are operated in parallel
1909 using a single instruction (SIMD). A vector type requires a size (number of
1910 elements) and an underlying primitive data type. Vector types are considered
1911 <a href="#t_firstclass">first class</a>.</p>
1915 < <# elements> x <elementtype> >
1918 <p>The number of elements is a constant integer value larger than 0; elementtype
1919 may be any integer or floating point type. Vectors of size zero are not
1920 allowed, and pointers are not allowed as the element type.</p>
1923 <table class="layout">
1925 <td class="left"><tt><4 x i32></tt></td>
1926 <td class="left">Vector of 4 32-bit integer values.</td>
1929 <td class="left"><tt><8 x float></tt></td>
1930 <td class="left">Vector of 8 32-bit floating-point values.</td>
1933 <td class="left"><tt><2 x i64></tt></td>
1934 <td class="left">Vector of 2 64-bit integer values.</td>
1940 <!-- _______________________________________________________________________ -->
1941 <div class="doc_subsubsection"> <a name="t_opaque">Opaque Type</a> </div>
1942 <div class="doc_text">
1945 <p>Opaque types are used to represent unknown types in the system. This
1946 corresponds (for example) to the C notion of a forward declared structure
1947 type. In LLVM, opaque types can eventually be resolved to any type (not just
1948 a structure type).</p>
1956 <table class="layout">
1958 <td class="left"><tt>opaque</tt></td>
1959 <td class="left">An opaque type.</td>
1965 <!-- ======================================================================= -->
1966 <div class="doc_subsection">
1967 <a name="t_uprefs">Type Up-references</a>
1970 <div class="doc_text">
1973 <p>An "up reference" allows you to refer to a lexically enclosing type without
1974 requiring it to have a name. For instance, a structure declaration may
1975 contain a pointer to any of the types it is lexically a member of. Example
1976 of up references (with their equivalent as named type declarations)
1980 { \2 * } %x = type { %x* }
1981 { \2 }* %y = type { %y }*
1985 <p>An up reference is needed by the asmprinter for printing out cyclic types
1986 when there is no declared name for a type in the cycle. Because the
1987 asmprinter does not want to print out an infinite type string, it needs a
1988 syntax to handle recursive types that have no names (all names are optional
1996 <p>The level is the count of the lexical type that is being referred to.</p>
1999 <table class="layout">
2001 <td class="left"><tt>\1*</tt></td>
2002 <td class="left">Self-referential pointer.</td>
2005 <td class="left"><tt>{ { \3*, i8 }, i32 }</tt></td>
2006 <td class="left">Recursive structure where the upref refers to the out-most
2013 <!-- *********************************************************************** -->
2014 <div class="doc_section"> <a name="constants">Constants</a> </div>
2015 <!-- *********************************************************************** -->
2017 <div class="doc_text">
2019 <p>LLVM has several different basic types of constants. This section describes
2020 them all and their syntax.</p>
2024 <!-- ======================================================================= -->
2025 <div class="doc_subsection"><a name="simpleconstants">Simple Constants</a></div>
2027 <div class="doc_text">
2030 <dt><b>Boolean constants</b></dt>
2031 <dd>The two strings '<tt>true</tt>' and '<tt>false</tt>' are both valid
2032 constants of the <tt><a href="#t_integer">i1</a></tt> type.</dd>
2034 <dt><b>Integer constants</b></dt>
2035 <dd>Standard integers (such as '4') are constants of
2036 the <a href="#t_integer">integer</a> type. Negative numbers may be used
2037 with integer types.</dd>
2039 <dt><b>Floating point constants</b></dt>
2040 <dd>Floating point constants use standard decimal notation (e.g. 123.421),
2041 exponential notation (e.g. 1.23421e+2), or a more precise hexadecimal
2042 notation (see below). The assembler requires the exact decimal value of a
2043 floating-point constant. For example, the assembler accepts 1.25 but
2044 rejects 1.3 because 1.3 is a repeating decimal in binary. Floating point
2045 constants must have a <a href="#t_floating">floating point</a> type. </dd>
2047 <dt><b>Null pointer constants</b></dt>
2048 <dd>The identifier '<tt>null</tt>' is recognized as a null pointer constant
2049 and must be of <a href="#t_pointer">pointer type</a>.</dd>
2052 <p>The one non-intuitive notation for constants is the hexadecimal form of
2053 floating point constants. For example, the form '<tt>double
2054 0x432ff973cafa8000</tt>' is equivalent to (but harder to read than)
2055 '<tt>double 4.5e+15</tt>'. The only time hexadecimal floating point
2056 constants are required (and the only time that they are generated by the
2057 disassembler) is when a floating point constant must be emitted but it cannot
2058 be represented as a decimal floating point number in a reasonable number of
2059 digits. For example, NaN's, infinities, and other special values are
2060 represented in their IEEE hexadecimal format so that assembly and disassembly
2061 do not cause any bits to change in the constants.</p>
2063 <p>When using the hexadecimal form, constants of types float and double are
2064 represented using the 16-digit form shown above (which matches the IEEE754
2065 representation for double); float values must, however, be exactly
2066 representable as IEE754 single precision. Hexadecimal format is always used
2067 for long double, and there are three forms of long double. The 80-bit format
2068 used by x86 is represented as <tt>0xK</tt> followed by 20 hexadecimal digits.
2069 The 128-bit format used by PowerPC (two adjacent doubles) is represented
2070 by <tt>0xM</tt> followed by 32 hexadecimal digits. The IEEE 128-bit format
2071 is represented by <tt>0xL</tt> followed by 32 hexadecimal digits; no
2072 currently supported target uses this format. Long doubles will only work if
2073 they match the long double format on your target. All hexadecimal formats
2074 are big-endian (sign bit at the left).</p>
2076 <p>There are no constants of type x86mmx.</p>
2079 <!-- ======================================================================= -->
2080 <div class="doc_subsection">
2081 <a name="aggregateconstants"></a> <!-- old anchor -->
2082 <a name="complexconstants">Complex Constants</a>
2085 <div class="doc_text">
2087 <p>Complex constants are a (potentially recursive) combination of simple
2088 constants and smaller complex constants.</p>
2091 <dt><b>Structure constants</b></dt>
2092 <dd>Structure constants are represented with notation similar to structure
2093 type definitions (a comma separated list of elements, surrounded by braces
2094 (<tt>{}</tt>)). For example: "<tt>{ i32 4, float 17.0, i32* @G }</tt>",
2095 where "<tt>@G</tt>" is declared as "<tt>@G = external global i32</tt>".
2096 Structure constants must have <a href="#t_struct">structure type</a>, and
2097 the number and types of elements must match those specified by the
2100 <dt><b>Array constants</b></dt>
2101 <dd>Array constants are represented with notation similar to array type
2102 definitions (a comma separated list of elements, surrounded by square
2103 brackets (<tt>[]</tt>)). For example: "<tt>[ i32 42, i32 11, i32 74
2104 ]</tt>". Array constants must have <a href="#t_array">array type</a>, and
2105 the number and types of elements must match those specified by the
2108 <dt><b>Vector constants</b></dt>
2109 <dd>Vector constants are represented with notation similar to vector type
2110 definitions (a comma separated list of elements, surrounded by
2111 less-than/greater-than's (<tt><></tt>)). For example: "<tt>< i32
2112 42, i32 11, i32 74, i32 100 ></tt>". Vector constants must
2113 have <a href="#t_vector">vector type</a>, and the number and types of
2114 elements must match those specified by the type.</dd>
2116 <dt><b>Zero initialization</b></dt>
2117 <dd>The string '<tt>zeroinitializer</tt>' can be used to zero initialize a
2118 value to zero of <em>any</em> type, including scalar and
2119 <a href="#t_aggregate">aggregate</a> types.
2120 This is often used to avoid having to print large zero initializers
2121 (e.g. for large arrays) and is always exactly equivalent to using explicit
2122 zero initializers.</dd>
2124 <dt><b>Metadata node</b></dt>
2125 <dd>A metadata node is a structure-like constant with
2126 <a href="#t_metadata">metadata type</a>. For example: "<tt>metadata !{
2127 i32 0, metadata !"test" }</tt>". Unlike other constants that are meant to
2128 be interpreted as part of the instruction stream, metadata is a place to
2129 attach additional information such as debug info.</dd>
2134 <!-- ======================================================================= -->
2135 <div class="doc_subsection">
2136 <a name="globalconstants">Global Variable and Function Addresses</a>
2139 <div class="doc_text">
2141 <p>The addresses of <a href="#globalvars">global variables</a>
2142 and <a href="#functionstructure">functions</a> are always implicitly valid
2143 (link-time) constants. These constants are explicitly referenced when
2144 the <a href="#identifiers">identifier for the global</a> is used and always
2145 have <a href="#t_pointer">pointer</a> type. For example, the following is a
2146 legal LLVM file:</p>
2148 <pre class="doc_code">
2151 @Z = global [2 x i32*] [ i32* @X, i32* @Y ]
2156 <!-- ======================================================================= -->
2157 <div class="doc_subsection"><a name="undefvalues">Undefined Values</a></div>
2158 <div class="doc_text">
2160 <p>The string '<tt>undef</tt>' can be used anywhere a constant is expected, and
2161 indicates that the user of the value may receive an unspecified bit-pattern.
2162 Undefined values may be of any type (other than label or void) and be used
2163 anywhere a constant is permitted.</p>
2165 <p>Undefined values are useful because they indicate to the compiler that the
2166 program is well defined no matter what value is used. This gives the
2167 compiler more freedom to optimize. Here are some examples of (potentially
2168 surprising) transformations that are valid (in pseudo IR):</p>
2171 <pre class="doc_code">
2181 <p>This is safe because all of the output bits are affected by the undef bits.
2182 Any output bit can have a zero or one depending on the input bits.</p>
2184 <pre class="doc_code">
2195 <p>These logical operations have bits that are not always affected by the input.
2196 For example, if "%X" has a zero bit, then the output of the 'and' operation will
2197 always be a zero, no matter what the corresponding bit from the undef is. As
2198 such, it is unsafe to optimize or assume that the result of the and is undef.
2199 However, it is safe to assume that all bits of the undef could be 0, and
2200 optimize the and to 0. Likewise, it is safe to assume that all the bits of
2201 the undef operand to the or could be set, allowing the or to be folded to
2204 <pre class="doc_code">
2205 %A = select undef, %X, %Y
2206 %B = select undef, 42, %Y
2207 %C = select %X, %Y, undef
2218 <p>This set of examples show that undefined select (and conditional branch)
2219 conditions can go "either way" but they have to come from one of the two
2220 operands. In the %A example, if %X and %Y were both known to have a clear low
2221 bit, then %A would have to have a cleared low bit. However, in the %C example,
2222 the optimizer is allowed to assume that the undef operand could be the same as
2223 %Y, allowing the whole select to be eliminated.</p>
2226 <pre class="doc_code">
2227 %A = xor undef, undef
2245 <p>This example points out that two undef operands are not necessarily the same.
2246 This can be surprising to people (and also matches C semantics) where they
2247 assume that "X^X" is always zero, even if X is undef. This isn't true for a
2248 number of reasons, but the short answer is that an undef "variable" can
2249 arbitrarily change its value over its "live range". This is true because the
2250 "variable" doesn't actually <em>have a live range</em>. Instead, the value is
2251 logically read from arbitrary registers that happen to be around when needed,
2252 so the value is not necessarily consistent over time. In fact, %A and %C need
2253 to have the same semantics or the core LLVM "replace all uses with" concept
2256 <pre class="doc_code">
2264 <p>These examples show the crucial difference between an <em>undefined
2265 value</em> and <em>undefined behavior</em>. An undefined value (like undef) is
2266 allowed to have an arbitrary bit-pattern. This means that the %A operation
2267 can be constant folded to undef because the undef could be an SNaN, and fdiv is
2268 not (currently) defined on SNaN's. However, in the second example, we can make
2269 a more aggressive assumption: because the undef is allowed to be an arbitrary
2270 value, we are allowed to assume that it could be zero. Since a divide by zero
2271 has <em>undefined behavior</em>, we are allowed to assume that the operation
2272 does not execute at all. This allows us to delete the divide and all code after
2273 it: since the undefined operation "can't happen", the optimizer can assume that
2274 it occurs in dead code.
2277 <pre class="doc_code">
2278 a: store undef -> %X
2279 b: store %X -> undef
2285 <p>These examples reiterate the fdiv example: a store "of" an undefined value
2286 can be assumed to not have any effect: we can assume that the value is
2287 overwritten with bits that happen to match what was already there. However, a
2288 store "to" an undefined location could clobber arbitrary memory, therefore, it
2289 has undefined behavior.</p>
2293 <!-- ======================================================================= -->
2294 <div class="doc_subsection"><a name="trapvalues">Trap Values</a></div>
2295 <div class="doc_text">
2297 <p>Trap values are similar to <a href="#undefvalues">undef values</a>, however
2298 instead of representing an unspecified bit pattern, they represent the
2299 fact that an instruction or constant expression which cannot evoke side
2300 effects has nevertheless detected a condition which results in undefined
2303 <p>There is currently no way of representing a trap value in the IR; they
2304 only exist when produced by operations such as
2305 <a href="#i_add"><tt>add</tt></a> with the <tt>nsw</tt> flag.</p>
2307 <p>Trap value behavior is defined in terms of value <i>dependence</i>:</p>
2310 <li>Values other than <a href="#i_phi"><tt>phi</tt></a> nodes depend on
2311 their operands.</li>
2313 <li><a href="#i_phi"><tt>Phi</tt></a> nodes depend on the operand corresponding
2314 to their dynamic predecessor basic block.</li>
2316 <li>Function arguments depend on the corresponding actual argument values in
2317 the dynamic callers of their functions.</li>
2319 <li><a href="#i_call"><tt>Call</tt></a> instructions depend on the
2320 <a href="#i_ret"><tt>ret</tt></a> instructions that dynamically transfer
2321 control back to them.</li>
2323 <li><a href="#i_invoke"><tt>Invoke</tt></a> instructions depend on the
2324 <a href="#i_ret"><tt>ret</tt></a>, <a href="#i_unwind"><tt>unwind</tt></a>,
2325 or exception-throwing call instructions that dynamically transfer control
2328 <li>Non-volatile loads and stores depend on the most recent stores to all of the
2329 referenced memory addresses, following the order in the IR
2330 (including loads and stores implied by intrinsics such as
2331 <a href="#int_memcpy"><tt>@llvm.memcpy</tt></a>.)</li>
2333 <!-- TODO: In the case of multiple threads, this only applies if the store
2334 "happens-before" the load or store. -->
2336 <!-- TODO: floating-point exception state -->
2338 <li>An instruction with externally visible side effects depends on the most
2339 recent preceding instruction with externally visible side effects, following
2340 the order in the IR. (This includes
2341 <a href="#volatile">volatile operations</a>.)</li>
2343 <li>An instruction <i>control-depends</i> on a
2344 <a href="#terminators">terminator instruction</a>
2345 if the terminator instruction has multiple successors and the instruction
2346 is always executed when control transfers to one of the successors, and
2347 may not be executed when control is transfered to another.</li>
2349 <li>Dependence is transitive.</li>
2353 <p>Whenever a trap value is generated, all values which depend on it evaluate
2354 to trap. If they have side effects, the evoke their side effects as if each
2355 operand with a trap value were undef. If they have externally-visible side
2356 effects, the behavior is undefined.</p>
2358 <p>Here are some examples:</p>
2360 <pre class="doc_code">
2362 %trap = sub nuw i32 0, 1 ; Results in a trap value.
2363 %still_trap = and i32 %trap, 0 ; Whereas (and i32 undef, 0) would return 0.
2364 %trap_yet_again = getelementptr i32* @h, i32 %still_trap
2365 store i32 0, i32* %trap_yet_again ; undefined behavior
2367 store i32 %trap, i32* @g ; Trap value conceptually stored to memory.
2368 %trap2 = load i32* @g ; Returns a trap value, not just undef.
2370 volatile store i32 %trap, i32* @g ; External observation; undefined behavior.
2372 %narrowaddr = bitcast i32* @g to i16*
2373 %wideaddr = bitcast i32* @g to i64*
2374 %trap3 = load 16* %narrowaddr ; Returns a trap value.
2375 %trap4 = load i64* %widaddr ; Returns a trap value.
2377 %cmp = icmp i32 slt %trap, 0 ; Returns a trap value.
2378 %br i1 %cmp, %true, %end ; Branch to either destination.
2381 volatile store i32 0, i32* @g ; This is control-dependent on %cmp, so
2382 ; it has undefined behavior.
2386 %p = phi i32 [ 0, %entry ], [ 1, %true ]
2387 ; Both edges into this PHI are
2388 ; control-dependent on %cmp, so this
2389 ; always results in a trap value.
2391 volatile store i32 0, i32* @g ; %end is control-equivalent to %entry
2392 ; so this is defined (ignoring earlier
2393 ; undefined behavior in this example).
2398 <!-- ======================================================================= -->
2399 <div class="doc_subsection"><a name="blockaddress">Addresses of Basic
2401 <div class="doc_text">
2403 <p><b><tt>blockaddress(@function, %block)</tt></b></p>
2405 <p>The '<tt>blockaddress</tt>' constant computes the address of the specified
2406 basic block in the specified function, and always has an i8* type. Taking
2407 the address of the entry block is illegal.</p>
2409 <p>This value only has defined behavior when used as an operand to the
2410 '<a href="#i_indirectbr"><tt>indirectbr</tt></a>' instruction or for comparisons
2411 against null. Pointer equality tests between labels addresses is undefined
2412 behavior - though, again, comparison against null is ok, and no label is
2413 equal to the null pointer. This may also be passed around as an opaque
2414 pointer sized value as long as the bits are not inspected. This allows
2415 <tt>ptrtoint</tt> and arithmetic to be performed on these values so long as
2416 the original value is reconstituted before the <tt>indirectbr</tt>.</p>
2418 <p>Finally, some targets may provide defined semantics when
2419 using the value as the operand to an inline assembly, but that is target
2426 <!-- ======================================================================= -->
2427 <div class="doc_subsection"><a name="constantexprs">Constant Expressions</a>
2430 <div class="doc_text">
2432 <p>Constant expressions are used to allow expressions involving other constants
2433 to be used as constants. Constant expressions may be of
2434 any <a href="#t_firstclass">first class</a> type and may involve any LLVM
2435 operation that does not have side effects (e.g. load and call are not
2436 supported). The following is the syntax for constant expressions:</p>
2439 <dt><b><tt>trunc (CST to TYPE)</tt></b></dt>
2440 <dd>Truncate a constant to another type. The bit size of CST must be larger
2441 than the bit size of TYPE. Both types must be integers.</dd>
2443 <dt><b><tt>zext (CST to TYPE)</tt></b></dt>
2444 <dd>Zero extend a constant to another type. The bit size of CST must be
2445 smaller than the bit size of TYPE. Both types must be integers.</dd>
2447 <dt><b><tt>sext (CST to TYPE)</tt></b></dt>
2448 <dd>Sign extend a constant to another type. The bit size of CST must be
2449 smaller than the bit size of TYPE. Both types must be integers.</dd>
2451 <dt><b><tt>fptrunc (CST to TYPE)</tt></b></dt>
2452 <dd>Truncate a floating point constant to another floating point type. The
2453 size of CST must be larger than the size of TYPE. Both types must be
2454 floating point.</dd>
2456 <dt><b><tt>fpext (CST to TYPE)</tt></b></dt>
2457 <dd>Floating point extend a constant to another type. The size of CST must be
2458 smaller or equal to the size of TYPE. Both types must be floating
2461 <dt><b><tt>fptoui (CST to TYPE)</tt></b></dt>
2462 <dd>Convert a floating point constant to the corresponding unsigned integer
2463 constant. TYPE must be a scalar or vector integer type. CST must be of
2464 scalar or vector floating point type. Both CST and TYPE must be scalars,
2465 or vectors of the same number of elements. If the value won't fit in the
2466 integer type, the results are undefined.</dd>
2468 <dt><b><tt>fptosi (CST to TYPE)</tt></b></dt>
2469 <dd>Convert a floating point constant to the corresponding signed integer
2470 constant. TYPE must be a scalar or vector integer type. CST must be of
2471 scalar or vector floating point type. Both CST and TYPE must be scalars,
2472 or vectors of the same number of elements. If the value won't fit in the
2473 integer type, the results are undefined.</dd>
2475 <dt><b><tt>uitofp (CST to TYPE)</tt></b></dt>
2476 <dd>Convert an unsigned integer constant to the corresponding floating point
2477 constant. TYPE must be a scalar or vector floating point type. CST must be
2478 of scalar or vector integer type. Both CST and TYPE must be scalars, or
2479 vectors of the same number of elements. If the value won't fit in the
2480 floating point type, the results are undefined.</dd>
2482 <dt><b><tt>sitofp (CST to TYPE)</tt></b></dt>
2483 <dd>Convert a signed integer constant to the corresponding floating point
2484 constant. TYPE must be a scalar or vector floating point type. CST must be
2485 of scalar or vector integer type. Both CST and TYPE must be scalars, or
2486 vectors of the same number of elements. If the value won't fit in the
2487 floating point type, the results are undefined.</dd>
2489 <dt><b><tt>ptrtoint (CST to TYPE)</tt></b></dt>
2490 <dd>Convert a pointer typed constant to the corresponding integer constant
2491 <tt>TYPE</tt> must be an integer type. <tt>CST</tt> must be of pointer
2492 type. The <tt>CST</tt> value is zero extended, truncated, or unchanged to
2493 make it fit in <tt>TYPE</tt>.</dd>
2495 <dt><b><tt>inttoptr (CST to TYPE)</tt></b></dt>
2496 <dd>Convert a integer constant to a pointer constant. TYPE must be a pointer
2497 type. CST must be of integer type. The CST value is zero extended,
2498 truncated, or unchanged to make it fit in a pointer size. This one is
2499 <i>really</i> dangerous!</dd>
2501 <dt><b><tt>bitcast (CST to TYPE)</tt></b></dt>
2502 <dd>Convert a constant, CST, to another TYPE. The constraints of the operands
2503 are the same as those for the <a href="#i_bitcast">bitcast
2504 instruction</a>.</dd>
2506 <dt><b><tt>getelementptr (CSTPTR, IDX0, IDX1, ...)</tt></b></dt>
2507 <dt><b><tt>getelementptr inbounds (CSTPTR, IDX0, IDX1, ...)</tt></b></dt>
2508 <dd>Perform the <a href="#i_getelementptr">getelementptr operation</a> on
2509 constants. As with the <a href="#i_getelementptr">getelementptr</a>
2510 instruction, the index list may have zero or more indexes, which are
2511 required to make sense for the type of "CSTPTR".</dd>
2513 <dt><b><tt>select (COND, VAL1, VAL2)</tt></b></dt>
2514 <dd>Perform the <a href="#i_select">select operation</a> on constants.</dd>
2516 <dt><b><tt>icmp COND (VAL1, VAL2)</tt></b></dt>
2517 <dd>Performs the <a href="#i_icmp">icmp operation</a> on constants.</dd>
2519 <dt><b><tt>fcmp COND (VAL1, VAL2)</tt></b></dt>
2520 <dd>Performs the <a href="#i_fcmp">fcmp operation</a> on constants.</dd>
2522 <dt><b><tt>extractelement (VAL, IDX)</tt></b></dt>
2523 <dd>Perform the <a href="#i_extractelement">extractelement operation</a> on
2526 <dt><b><tt>insertelement (VAL, ELT, IDX)</tt></b></dt>
2527 <dd>Perform the <a href="#i_insertelement">insertelement operation</a> on
2530 <dt><b><tt>shufflevector (VEC1, VEC2, IDXMASK)</tt></b></dt>
2531 <dd>Perform the <a href="#i_shufflevector">shufflevector operation</a> on
2534 <dt><b><tt>extractvalue (VAL, IDX0, IDX1, ...)</tt></b></dt>
2535 <dd>Perform the <a href="#i_extractvalue">extractvalue operation</a> on
2536 constants. The index list is interpreted in a similar manner as indices in
2537 a '<a href="#i_getelementptr">getelementptr</a>' operation. At least one
2538 index value must be specified.</dd>
2540 <dt><b><tt>insertvalue (VAL, ELT, IDX0, IDX1, ...)</tt></b></dt>
2541 <dd>Perform the <a href="#i_insertvalue">insertvalue operation</a> on
2542 constants. The index list is interpreted in a similar manner as indices in
2543 a '<a href="#i_getelementptr">getelementptr</a>' operation. At least one
2544 index value must be specified.</dd>
2546 <dt><b><tt>OPCODE (LHS, RHS)</tt></b></dt>
2547 <dd>Perform the specified operation of the LHS and RHS constants. OPCODE may
2548 be any of the <a href="#binaryops">binary</a>
2549 or <a href="#bitwiseops">bitwise binary</a> operations. The constraints
2550 on operands are the same as those for the corresponding instruction
2551 (e.g. no bitwise operations on floating point values are allowed).</dd>
2556 <!-- *********************************************************************** -->
2557 <div class="doc_section"> <a name="othervalues">Other Values</a> </div>
2558 <!-- *********************************************************************** -->
2560 <!-- ======================================================================= -->
2561 <div class="doc_subsection">
2562 <a name="inlineasm">Inline Assembler Expressions</a>
2565 <div class="doc_text">
2567 <p>LLVM supports inline assembler expressions (as opposed
2568 to <a href="#moduleasm"> Module-Level Inline Assembly</a>) through the use of
2569 a special value. This value represents the inline assembler as a string
2570 (containing the instructions to emit), a list of operand constraints (stored
2571 as a string), a flag that indicates whether or not the inline asm
2572 expression has side effects, and a flag indicating whether the function
2573 containing the asm needs to align its stack conservatively. An example
2574 inline assembler expression is:</p>
2576 <pre class="doc_code">
2577 i32 (i32) asm "bswap $0", "=r,r"
2580 <p>Inline assembler expressions may <b>only</b> be used as the callee operand of
2581 a <a href="#i_call"><tt>call</tt> instruction</a>. Thus, typically we
2584 <pre class="doc_code">
2585 %X = call i32 asm "<a href="#int_bswap">bswap</a> $0", "=r,r"(i32 %Y)
2588 <p>Inline asms with side effects not visible in the constraint list must be
2589 marked as having side effects. This is done through the use of the
2590 '<tt>sideeffect</tt>' keyword, like so:</p>
2592 <pre class="doc_code">
2593 call void asm sideeffect "eieio", ""()
2596 <p>In some cases inline asms will contain code that will not work unless the
2597 stack is aligned in some way, such as calls or SSE instructions on x86,
2598 yet will not contain code that does that alignment within the asm.
2599 The compiler should make conservative assumptions about what the asm might
2600 contain and should generate its usual stack alignment code in the prologue
2601 if the '<tt>alignstack</tt>' keyword is present:</p>
2603 <pre class="doc_code">
2604 call void asm alignstack "eieio", ""()
2607 <p>If both keywords appear the '<tt>sideeffect</tt>' keyword must come
2610 <p>TODO: The format of the asm and constraints string still need to be
2611 documented here. Constraints on what can be done (e.g. duplication, moving,
2612 etc need to be documented). This is probably best done by reference to
2613 another document that covers inline asm from a holistic perspective.</p>
2616 <div class="doc_subsubsection">
2617 <a name="inlineasm_md">Inline Asm Metadata</a>
2620 <div class="doc_text">
2622 <p>The call instructions that wrap inline asm nodes may have a "!srcloc" MDNode
2623 attached to it that contains a constant integer. If present, the code
2624 generator will use the integer as the location cookie value when report
2625 errors through the LLVMContext error reporting mechanisms. This allows a
2626 front-end to correlate backend errors that occur with inline asm back to the
2627 source code that produced it. For example:</p>
2629 <pre class="doc_code">
2630 call void asm sideeffect "something bad", ""()<b>, !srcloc !42</b>
2632 !42 = !{ i32 1234567 }
2635 <p>It is up to the front-end to make sense of the magic numbers it places in the
2640 <!-- ======================================================================= -->
2641 <div class="doc_subsection"><a name="metadata">Metadata Nodes and Metadata
2645 <div class="doc_text">
2647 <p>LLVM IR allows metadata to be attached to instructions in the program that
2648 can convey extra information about the code to the optimizers and code
2649 generator. One example application of metadata is source-level debug
2650 information. There are two metadata primitives: strings and nodes. All
2651 metadata has the <tt>metadata</tt> type and is identified in syntax by a
2652 preceding exclamation point ('<tt>!</tt>').</p>
2654 <p>A metadata string is a string surrounded by double quotes. It can contain
2655 any character by escaping non-printable characters with "\xx" where "xx" is
2656 the two digit hex code. For example: "<tt>!"test\00"</tt>".</p>
2658 <p>Metadata nodes are represented with notation similar to structure constants
2659 (a comma separated list of elements, surrounded by braces and preceded by an
2660 exclamation point). For example: "<tt>!{ metadata !"test\00", i32
2661 10}</tt>". Metadata nodes can have any values as their operand.</p>
2663 <p>A <a href="#namedmetadatastructure">named metadata</a> is a collection of
2664 metadata nodes, which can be looked up in the module symbol table. For
2665 example: "<tt>!foo = metadata !{!4, !3}</tt>".
2667 <p>Metadata can be used as function arguments. Here <tt>llvm.dbg.value</tt>
2668 function is using two metadata arguments.</p>
2670 <pre class="doc_code">
2671 call void @llvm.dbg.value(metadata !24, i64 0, metadata !25)
2674 <p>Metadata can be attached with an instruction. Here metadata <tt>!21</tt> is
2675 attached with <tt>add</tt> instruction using <tt>!dbg</tt> identifier.</p>
2677 <pre class="doc_code">
2678 %indvar.next = add i64 %indvar, 1, !dbg !21
2683 <!-- *********************************************************************** -->
2684 <div class="doc_section">
2685 <a name="intrinsic_globals">Intrinsic Global Variables</a>
2687 <!-- *********************************************************************** -->
2689 <p>LLVM has a number of "magic" global variables that contain data that affect
2690 code generation or other IR semantics. These are documented here. All globals
2691 of this sort should have a section specified as "<tt>llvm.metadata</tt>". This
2692 section and all globals that start with "<tt>llvm.</tt>" are reserved for use
2695 <!-- ======================================================================= -->
2696 <div class="doc_subsection">
2697 <a name="intg_used">The '<tt>llvm.used</tt>' Global Variable</a>
2700 <div class="doc_text">
2702 <p>The <tt>@llvm.used</tt> global is an array with i8* element type which has <a
2703 href="#linkage_appending">appending linkage</a>. This array contains a list of
2704 pointers to global variables and functions which may optionally have a pointer
2705 cast formed of bitcast or getelementptr. For example, a legal use of it is:</p>
2711 @llvm.used = appending global [2 x i8*] [
2713 i8* bitcast (i32* @Y to i8*)
2714 ], section "llvm.metadata"
2717 <p>If a global variable appears in the <tt>@llvm.used</tt> list, then the
2718 compiler, assembler, and linker are required to treat the symbol as if there is
2719 a reference to the global that it cannot see. For example, if a variable has
2720 internal linkage and no references other than that from the <tt>@llvm.used</tt>
2721 list, it cannot be deleted. This is commonly used to represent references from
2722 inline asms and other things the compiler cannot "see", and corresponds to
2723 "attribute((used))" in GNU C.</p>
2725 <p>On some targets, the code generator must emit a directive to the assembler or
2726 object file to prevent the assembler and linker from molesting the symbol.</p>
2730 <!-- ======================================================================= -->
2731 <div class="doc_subsection">
2732 <a name="intg_compiler_used">The '<tt>llvm.compiler.used</tt>' Global Variable</a>
2735 <div class="doc_text">
2737 <p>The <tt>@llvm.compiler.used</tt> directive is the same as the
2738 <tt>@llvm.used</tt> directive, except that it only prevents the compiler from
2739 touching the symbol. On targets that support it, this allows an intelligent
2740 linker to optimize references to the symbol without being impeded as it would be
2741 by <tt>@llvm.used</tt>.</p>
2743 <p>This is a rare construct that should only be used in rare circumstances, and
2744 should not be exposed to source languages.</p>
2748 <!-- ======================================================================= -->
2749 <div class="doc_subsection">
2750 <a name="intg_global_ctors">The '<tt>llvm.global_ctors</tt>' Global Variable</a>
2753 <div class="doc_text">
2755 %0 = type { i32, void ()* }
2756 @llvm.global_ctors = appending global [1 x %0] [%0 { i32 65535, void ()* @ctor }]
2758 <p>The <tt>@llvm.global_ctors</tt> array contains a list of constructor functions and associated priorities. The functions referenced by this array will be called in ascending order of priority (i.e. lowest first) when the module is loaded. The order of functions with the same priority is not defined.
2763 <!-- ======================================================================= -->
2764 <div class="doc_subsection">
2765 <a name="intg_global_dtors">The '<tt>llvm.global_dtors</tt>' Global Variable</a>
2768 <div class="doc_text">
2770 %0 = type { i32, void ()* }
2771 @llvm.global_dtors = appending global [1 x %0] [%0 { i32 65535, void ()* @dtor }]
2774 <p>The <tt>@llvm.global_dtors</tt> array contains a list of destructor functions and associated priorities. The functions referenced by this array will be called in descending order of priority (i.e. highest first) when the module is loaded. The order of functions with the same priority is not defined.
2780 <!-- *********************************************************************** -->
2781 <div class="doc_section"> <a name="instref">Instruction Reference</a> </div>
2782 <!-- *********************************************************************** -->
2784 <div class="doc_text">
2786 <p>The LLVM instruction set consists of several different classifications of
2787 instructions: <a href="#terminators">terminator
2788 instructions</a>, <a href="#binaryops">binary instructions</a>,
2789 <a href="#bitwiseops">bitwise binary instructions</a>,
2790 <a href="#memoryops">memory instructions</a>, and
2791 <a href="#otherops">other instructions</a>.</p>
2795 <!-- ======================================================================= -->
2796 <div class="doc_subsection"> <a name="terminators">Terminator
2797 Instructions</a> </div>
2799 <div class="doc_text">
2801 <p>As mentioned <a href="#functionstructure">previously</a>, every basic block
2802 in a program ends with a "Terminator" instruction, which indicates which
2803 block should be executed after the current block is finished. These
2804 terminator instructions typically yield a '<tt>void</tt>' value: they produce
2805 control flow, not values (the one exception being the
2806 '<a href="#i_invoke"><tt>invoke</tt></a>' instruction).</p>
2808 <p>There are seven different terminator instructions: the
2809 '<a href="#i_ret"><tt>ret</tt></a>' instruction, the
2810 '<a href="#i_br"><tt>br</tt></a>' instruction, the
2811 '<a href="#i_switch"><tt>switch</tt></a>' instruction, the
2812 '<a href="#i_indirectbr">'<tt>indirectbr</tt></a>' Instruction, the
2813 '<a href="#i_invoke"><tt>invoke</tt></a>' instruction, the
2814 '<a href="#i_unwind"><tt>unwind</tt></a>' instruction, and the
2815 '<a href="#i_unreachable"><tt>unreachable</tt></a>' instruction.</p>
2819 <!-- _______________________________________________________________________ -->
2820 <div class="doc_subsubsection"> <a name="i_ret">'<tt>ret</tt>'
2821 Instruction</a> </div>
2823 <div class="doc_text">
2827 ret <type> <value> <i>; Return a value from a non-void function</i>
2828 ret void <i>; Return from void function</i>
2832 <p>The '<tt>ret</tt>' instruction is used to return control flow (and optionally
2833 a value) from a function back to the caller.</p>
2835 <p>There are two forms of the '<tt>ret</tt>' instruction: one that returns a
2836 value and then causes control flow, and one that just causes control flow to
2840 <p>The '<tt>ret</tt>' instruction optionally accepts a single argument, the
2841 return value. The type of the return value must be a
2842 '<a href="#t_firstclass">first class</a>' type.</p>
2844 <p>A function is not <a href="#wellformed">well formed</a> if it it has a
2845 non-void return type and contains a '<tt>ret</tt>' instruction with no return
2846 value or a return value with a type that does not match its type, or if it
2847 has a void return type and contains a '<tt>ret</tt>' instruction with a
2851 <p>When the '<tt>ret</tt>' instruction is executed, control flow returns back to
2852 the calling function's context. If the caller is a
2853 "<a href="#i_call"><tt>call</tt></a>" instruction, execution continues at the
2854 instruction after the call. If the caller was an
2855 "<a href="#i_invoke"><tt>invoke</tt></a>" instruction, execution continues at
2856 the beginning of the "normal" destination block. If the instruction returns
2857 a value, that value shall set the call or invoke instruction's return
2862 ret i32 5 <i>; Return an integer value of 5</i>
2863 ret void <i>; Return from a void function</i>
2864 ret { i32, i8 } { i32 4, i8 2 } <i>; Return a struct of values 4 and 2</i>
2868 <!-- _______________________________________________________________________ -->
2869 <div class="doc_subsubsection"> <a name="i_br">'<tt>br</tt>' Instruction</a> </div>
2871 <div class="doc_text">
2875 br i1 <cond>, label <iftrue>, label <iffalse><br> br label <dest> <i>; Unconditional branch</i>
2879 <p>The '<tt>br</tt>' instruction is used to cause control flow to transfer to a
2880 different basic block in the current function. There are two forms of this
2881 instruction, corresponding to a conditional branch and an unconditional
2885 <p>The conditional branch form of the '<tt>br</tt>' instruction takes a single
2886 '<tt>i1</tt>' value and two '<tt>label</tt>' values. The unconditional form
2887 of the '<tt>br</tt>' instruction takes a single '<tt>label</tt>' value as a
2891 <p>Upon execution of a conditional '<tt>br</tt>' instruction, the '<tt>i1</tt>'
2892 argument is evaluated. If the value is <tt>true</tt>, control flows to the
2893 '<tt>iftrue</tt>' <tt>label</tt> argument. If "cond" is <tt>false</tt>,
2894 control flows to the '<tt>iffalse</tt>' <tt>label</tt> argument.</p>
2899 %cond = <a href="#i_icmp">icmp</a> eq i32 %a, %b
2900 br i1 %cond, label %IfEqual, label %IfUnequal
2902 <a href="#i_ret">ret</a> i32 1
2904 <a href="#i_ret">ret</a> i32 0
2909 <!-- _______________________________________________________________________ -->
2910 <div class="doc_subsubsection">
2911 <a name="i_switch">'<tt>switch</tt>' Instruction</a>
2914 <div class="doc_text">
2918 switch <intty> <value>, label <defaultdest> [ <intty> <val>, label <dest> ... ]
2922 <p>The '<tt>switch</tt>' instruction is used to transfer control flow to one of
2923 several different places. It is a generalization of the '<tt>br</tt>'
2924 instruction, allowing a branch to occur to one of many possible
2928 <p>The '<tt>switch</tt>' instruction uses three parameters: an integer
2929 comparison value '<tt>value</tt>', a default '<tt>label</tt>' destination,
2930 and an array of pairs of comparison value constants and '<tt>label</tt>'s.
2931 The table is not allowed to contain duplicate constant entries.</p>
2934 <p>The <tt>switch</tt> instruction specifies a table of values and
2935 destinations. When the '<tt>switch</tt>' instruction is executed, this table
2936 is searched for the given value. If the value is found, control flow is
2937 transferred to the corresponding destination; otherwise, control flow is
2938 transferred to the default destination.</p>
2940 <h5>Implementation:</h5>
2941 <p>Depending on properties of the target machine and the particular
2942 <tt>switch</tt> instruction, this instruction may be code generated in
2943 different ways. For example, it could be generated as a series of chained
2944 conditional branches or with a lookup table.</p>
2948 <i>; Emulate a conditional br instruction</i>
2949 %Val = <a href="#i_zext">zext</a> i1 %value to i32
2950 switch i32 %Val, label %truedest [ i32 0, label %falsedest ]
2952 <i>; Emulate an unconditional br instruction</i>
2953 switch i32 0, label %dest [ ]
2955 <i>; Implement a jump table:</i>
2956 switch i32 %val, label %otherwise [ i32 0, label %onzero
2958 i32 2, label %ontwo ]
2964 <!-- _______________________________________________________________________ -->
2965 <div class="doc_subsubsection">
2966 <a name="i_indirectbr">'<tt>indirectbr</tt>' Instruction</a>
2969 <div class="doc_text">
2973 indirectbr <somety>* <address>, [ label <dest1>, label <dest2>, ... ]
2978 <p>The '<tt>indirectbr</tt>' instruction implements an indirect branch to a label
2979 within the current function, whose address is specified by
2980 "<tt>address</tt>". Address must be derived from a <a
2981 href="#blockaddress">blockaddress</a> constant.</p>
2985 <p>The '<tt>address</tt>' argument is the address of the label to jump to. The
2986 rest of the arguments indicate the full set of possible destinations that the
2987 address may point to. Blocks are allowed to occur multiple times in the
2988 destination list, though this isn't particularly useful.</p>
2990 <p>This destination list is required so that dataflow analysis has an accurate
2991 understanding of the CFG.</p>
2995 <p>Control transfers to the block specified in the address argument. All
2996 possible destination blocks must be listed in the label list, otherwise this
2997 instruction has undefined behavior. This implies that jumps to labels
2998 defined in other functions have undefined behavior as well.</p>
3000 <h5>Implementation:</h5>
3002 <p>This is typically implemented with a jump through a register.</p>
3006 indirectbr i8* %Addr, [ label %bb1, label %bb2, label %bb3 ]
3012 <!-- _______________________________________________________________________ -->
3013 <div class="doc_subsubsection">
3014 <a name="i_invoke">'<tt>invoke</tt>' Instruction</a>
3017 <div class="doc_text">
3021 <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>]
3022 to label <normal label> unwind label <exception label>
3026 <p>The '<tt>invoke</tt>' instruction causes control to transfer to a specified
3027 function, with the possibility of control flow transfer to either the
3028 '<tt>normal</tt>' label or the '<tt>exception</tt>' label. If the callee
3029 function returns with the "<tt><a href="#i_ret">ret</a></tt>" instruction,
3030 control flow will return to the "normal" label. If the callee (or any
3031 indirect callees) returns with the "<a href="#i_unwind"><tt>unwind</tt></a>"
3032 instruction, control is interrupted and continued at the dynamically nearest
3033 "exception" label.</p>
3036 <p>This instruction requires several arguments:</p>
3039 <li>The optional "cconv" marker indicates which <a href="#callingconv">calling
3040 convention</a> the call should use. If none is specified, the call
3041 defaults to using C calling conventions.</li>
3043 <li>The optional <a href="#paramattrs">Parameter Attributes</a> list for
3044 return values. Only '<tt>zeroext</tt>', '<tt>signext</tt>', and
3045 '<tt>inreg</tt>' attributes are valid here.</li>
3047 <li>'<tt>ptr to function ty</tt>': shall be the signature of the pointer to
3048 function value being invoked. In most cases, this is a direct function
3049 invocation, but indirect <tt>invoke</tt>s are just as possible, branching
3050 off an arbitrary pointer to function value.</li>
3052 <li>'<tt>function ptr val</tt>': An LLVM value containing a pointer to a
3053 function to be invoked. </li>
3055 <li>'<tt>function args</tt>': argument list whose types match the function
3056 signature argument types and parameter attributes. All arguments must be
3057 of <a href="#t_firstclass">first class</a> type. If the function
3058 signature indicates the function accepts a variable number of arguments,
3059 the extra arguments can be specified.</li>
3061 <li>'<tt>normal label</tt>': the label reached when the called function
3062 executes a '<tt><a href="#i_ret">ret</a></tt>' instruction. </li>
3064 <li>'<tt>exception label</tt>': the label reached when a callee returns with
3065 the <a href="#i_unwind"><tt>unwind</tt></a> instruction. </li>
3067 <li>The optional <a href="#fnattrs">function attributes</a> list. Only
3068 '<tt>noreturn</tt>', '<tt>nounwind</tt>', '<tt>readonly</tt>' and
3069 '<tt>readnone</tt>' attributes are valid here.</li>
3073 <p>This instruction is designed to operate as a standard
3074 '<tt><a href="#i_call">call</a></tt>' instruction in most regards. The
3075 primary difference is that it establishes an association with a label, which
3076 is used by the runtime library to unwind the stack.</p>
3078 <p>This instruction is used in languages with destructors to ensure that proper
3079 cleanup is performed in the case of either a <tt>longjmp</tt> or a thrown
3080 exception. Additionally, this is important for implementation of
3081 '<tt>catch</tt>' clauses in high-level languages that support them.</p>
3083 <p>For the purposes of the SSA form, the definition of the value returned by the
3084 '<tt>invoke</tt>' instruction is deemed to occur on the edge from the current
3085 block to the "normal" label. If the callee unwinds then no return value is
3088 <p>Note that the code generator does not yet completely support unwind, and
3089 that the invoke/unwind semantics are likely to change in future versions.</p>
3093 %retval = invoke i32 @Test(i32 15) to label %Continue
3094 unwind label %TestCleanup <i>; {i32}:retval set</i>
3095 %retval = invoke <a href="#callingconv">coldcc</a> i32 %Testfnptr(i32 15) to label %Continue
3096 unwind label %TestCleanup <i>; {i32}:retval set</i>
3101 <!-- _______________________________________________________________________ -->
3103 <div class="doc_subsubsection"> <a name="i_unwind">'<tt>unwind</tt>'
3104 Instruction</a> </div>
3106 <div class="doc_text">
3114 <p>The '<tt>unwind</tt>' instruction unwinds the stack, continuing control flow
3115 at the first callee in the dynamic call stack which used
3116 an <a href="#i_invoke"><tt>invoke</tt></a> instruction to perform the call.
3117 This is primarily used to implement exception handling.</p>
3120 <p>The '<tt>unwind</tt>' instruction causes execution of the current function to
3121 immediately halt. The dynamic call stack is then searched for the
3122 first <a href="#i_invoke"><tt>invoke</tt></a> instruction on the call stack.
3123 Once found, execution continues at the "exceptional" destination block
3124 specified by the <tt>invoke</tt> instruction. If there is no <tt>invoke</tt>
3125 instruction in the dynamic call chain, undefined behavior results.</p>
3127 <p>Note that the code generator does not yet completely support unwind, and
3128 that the invoke/unwind semantics are likely to change in future versions.</p>
3132 <!-- _______________________________________________________________________ -->
3134 <div class="doc_subsubsection"> <a name="i_unreachable">'<tt>unreachable</tt>'
3135 Instruction</a> </div>
3137 <div class="doc_text">
3145 <p>The '<tt>unreachable</tt>' instruction has no defined semantics. This
3146 instruction is used to inform the optimizer that a particular portion of the
3147 code is not reachable. This can be used to indicate that the code after a
3148 no-return function cannot be reached, and other facts.</p>
3151 <p>The '<tt>unreachable</tt>' instruction has no defined semantics.</p>
3155 <!-- ======================================================================= -->
3156 <div class="doc_subsection"> <a name="binaryops">Binary Operations</a> </div>
3158 <div class="doc_text">
3160 <p>Binary operators are used to do most of the computation in a program. They
3161 require two operands of the same type, execute an operation on them, and
3162 produce a single value. The operands might represent multiple data, as is
3163 the case with the <a href="#t_vector">vector</a> data type. The result value
3164 has the same type as its operands.</p>
3166 <p>There are several different binary operators:</p>
3170 <!-- _______________________________________________________________________ -->
3171 <div class="doc_subsubsection">
3172 <a name="i_add">'<tt>add</tt>' Instruction</a>
3175 <div class="doc_text">
3179 <result> = add <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3180 <result> = add nuw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3181 <result> = add nsw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3182 <result> = add nuw nsw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3186 <p>The '<tt>add</tt>' instruction returns the sum of its two operands.</p>
3189 <p>The two arguments to the '<tt>add</tt>' instruction must
3190 be <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
3191 integer values. Both arguments must have identical types.</p>
3194 <p>The value produced is the integer sum of the two operands.</p>
3196 <p>If the sum has unsigned overflow, the result returned is the mathematical
3197 result modulo 2<sup>n</sup>, where n is the bit width of the result.</p>
3199 <p>Because LLVM integers use a two's complement representation, this instruction
3200 is appropriate for both signed and unsigned integers.</p>
3202 <p><tt>nuw</tt> and <tt>nsw</tt> stand for "No Unsigned Wrap"
3203 and "No Signed Wrap", respectively. If the <tt>nuw</tt> and/or
3204 <tt>nsw</tt> keywords are present, the result value of the <tt>add</tt>
3205 is a <a href="#trapvalues">trap value</a> if unsigned and/or signed overflow,
3206 respectively, occurs.</p>
3210 <result> = add i32 4, %var <i>; yields {i32}:result = 4 + %var</i>
3215 <!-- _______________________________________________________________________ -->
3216 <div class="doc_subsubsection">
3217 <a name="i_fadd">'<tt>fadd</tt>' Instruction</a>
3220 <div class="doc_text">
3224 <result> = fadd <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3228 <p>The '<tt>fadd</tt>' instruction returns the sum of its two operands.</p>
3231 <p>The two arguments to the '<tt>fadd</tt>' instruction must be
3232 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3233 floating point values. Both arguments must have identical types.</p>
3236 <p>The value produced is the floating point sum of the two operands.</p>
3240 <result> = fadd float 4.0, %var <i>; yields {float}:result = 4.0 + %var</i>
3245 <!-- _______________________________________________________________________ -->
3246 <div class="doc_subsubsection">
3247 <a name="i_sub">'<tt>sub</tt>' Instruction</a>
3250 <div class="doc_text">
3254 <result> = sub <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3255 <result> = sub nuw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3256 <result> = sub nsw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3257 <result> = sub nuw nsw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3261 <p>The '<tt>sub</tt>' instruction returns the difference of its two
3264 <p>Note that the '<tt>sub</tt>' instruction is used to represent the
3265 '<tt>neg</tt>' instruction present in most other intermediate
3266 representations.</p>
3269 <p>The two arguments to the '<tt>sub</tt>' instruction must
3270 be <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
3271 integer values. Both arguments must have identical types.</p>
3274 <p>The value produced is the integer difference of the two operands.</p>
3276 <p>If the difference has unsigned overflow, the result returned is the
3277 mathematical result modulo 2<sup>n</sup>, where n is the bit width of the
3280 <p>Because LLVM integers use a two's complement representation, this instruction
3281 is appropriate for both signed and unsigned integers.</p>
3283 <p><tt>nuw</tt> and <tt>nsw</tt> stand for "No Unsigned Wrap"
3284 and "No Signed Wrap", respectively. If the <tt>nuw</tt> and/or
3285 <tt>nsw</tt> keywords are present, the result value of the <tt>sub</tt>
3286 is a <a href="#trapvalues">trap value</a> if unsigned and/or signed overflow,
3287 respectively, occurs.</p>
3291 <result> = sub i32 4, %var <i>; yields {i32}:result = 4 - %var</i>
3292 <result> = sub i32 0, %val <i>; yields {i32}:result = -%var</i>
3297 <!-- _______________________________________________________________________ -->
3298 <div class="doc_subsubsection">
3299 <a name="i_fsub">'<tt>fsub</tt>' Instruction</a>
3302 <div class="doc_text">
3306 <result> = fsub <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3310 <p>The '<tt>fsub</tt>' instruction returns the difference of its two
3313 <p>Note that the '<tt>fsub</tt>' instruction is used to represent the
3314 '<tt>fneg</tt>' instruction present in most other intermediate
3315 representations.</p>
3318 <p>The two arguments to the '<tt>fsub</tt>' instruction must be
3319 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3320 floating point values. Both arguments must have identical types.</p>
3323 <p>The value produced is the floating point difference of the two operands.</p>
3327 <result> = fsub float 4.0, %var <i>; yields {float}:result = 4.0 - %var</i>
3328 <result> = fsub float -0.0, %val <i>; yields {float}:result = -%var</i>
3333 <!-- _______________________________________________________________________ -->
3334 <div class="doc_subsubsection">
3335 <a name="i_mul">'<tt>mul</tt>' Instruction</a>
3338 <div class="doc_text">
3342 <result> = mul <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3343 <result> = mul nuw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3344 <result> = mul nsw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3345 <result> = mul nuw nsw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3349 <p>The '<tt>mul</tt>' instruction returns the product of its two operands.</p>
3352 <p>The two arguments to the '<tt>mul</tt>' instruction must
3353 be <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
3354 integer values. Both arguments must have identical types.</p>
3357 <p>The value produced is the integer product of the two operands.</p>
3359 <p>If the result of the multiplication has unsigned overflow, the result
3360 returned is the mathematical result modulo 2<sup>n</sup>, where n is the bit
3361 width of the result.</p>
3363 <p>Because LLVM integers use a two's complement representation, and the result
3364 is the same width as the operands, this instruction returns the correct
3365 result for both signed and unsigned integers. If a full product
3366 (e.g. <tt>i32</tt>x<tt>i32</tt>-><tt>i64</tt>) is needed, the operands should
3367 be sign-extended or zero-extended as appropriate to the width of the full
3370 <p><tt>nuw</tt> and <tt>nsw</tt> stand for "No Unsigned Wrap"
3371 and "No Signed Wrap", respectively. If the <tt>nuw</tt> and/or
3372 <tt>nsw</tt> keywords are present, the result value of the <tt>mul</tt>
3373 is a <a href="#trapvalues">trap value</a> if unsigned and/or signed overflow,
3374 respectively, occurs.</p>
3378 <result> = mul i32 4, %var <i>; yields {i32}:result = 4 * %var</i>
3383 <!-- _______________________________________________________________________ -->
3384 <div class="doc_subsubsection">
3385 <a name="i_fmul">'<tt>fmul</tt>' Instruction</a>
3388 <div class="doc_text">
3392 <result> = fmul <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3396 <p>The '<tt>fmul</tt>' instruction returns the product of its two operands.</p>
3399 <p>The two arguments to the '<tt>fmul</tt>' instruction must be
3400 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3401 floating point values. Both arguments must have identical types.</p>
3404 <p>The value produced is the floating point product of the two operands.</p>
3408 <result> = fmul float 4.0, %var <i>; yields {float}:result = 4.0 * %var</i>
3413 <!-- _______________________________________________________________________ -->
3414 <div class="doc_subsubsection"> <a name="i_udiv">'<tt>udiv</tt>' Instruction
3417 <div class="doc_text">
3421 <result> = udiv <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3425 <p>The '<tt>udiv</tt>' instruction returns the quotient of its two operands.</p>
3428 <p>The two arguments to the '<tt>udiv</tt>' instruction must be
3429 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3430 values. Both arguments must have identical types.</p>
3433 <p>The value produced is the unsigned integer quotient of the two operands.</p>
3435 <p>Note that unsigned integer division and signed integer division are distinct
3436 operations; for signed integer division, use '<tt>sdiv</tt>'.</p>
3438 <p>Division by zero leads to undefined behavior.</p>
3442 <result> = udiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
3447 <!-- _______________________________________________________________________ -->
3448 <div class="doc_subsubsection"> <a name="i_sdiv">'<tt>sdiv</tt>' Instruction
3451 <div class="doc_text">
3455 <result> = sdiv <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3456 <result> = sdiv exact <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3460 <p>The '<tt>sdiv</tt>' instruction returns the quotient of its two operands.</p>
3463 <p>The two arguments to the '<tt>sdiv</tt>' instruction must be
3464 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3465 values. Both arguments must have identical types.</p>
3468 <p>The value produced is the signed integer quotient of the two operands rounded
3471 <p>Note that signed integer division and unsigned integer division are distinct
3472 operations; for unsigned integer division, use '<tt>udiv</tt>'.</p>
3474 <p>Division by zero leads to undefined behavior. Overflow also leads to
3475 undefined behavior; this is a rare case, but can occur, for example, by doing
3476 a 32-bit division of -2147483648 by -1.</p>
3478 <p>If the <tt>exact</tt> keyword is present, the result value of the
3479 <tt>sdiv</tt> is a <a href="#trapvalues">trap value</a> if the result would
3484 <result> = sdiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
3489 <!-- _______________________________________________________________________ -->
3490 <div class="doc_subsubsection"> <a name="i_fdiv">'<tt>fdiv</tt>'
3491 Instruction</a> </div>
3493 <div class="doc_text">
3497 <result> = fdiv <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3501 <p>The '<tt>fdiv</tt>' instruction returns the quotient of its two operands.</p>
3504 <p>The two arguments to the '<tt>fdiv</tt>' instruction must be
3505 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3506 floating point values. Both arguments must have identical types.</p>
3509 <p>The value produced is the floating point quotient of the two operands.</p>
3513 <result> = fdiv float 4.0, %var <i>; yields {float}:result = 4.0 / %var</i>
3518 <!-- _______________________________________________________________________ -->
3519 <div class="doc_subsubsection"> <a name="i_urem">'<tt>urem</tt>' Instruction</a>
3522 <div class="doc_text">
3526 <result> = urem <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3530 <p>The '<tt>urem</tt>' instruction returns the remainder from the unsigned
3531 division of its two arguments.</p>
3534 <p>The two arguments to the '<tt>urem</tt>' instruction must be
3535 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3536 values. Both arguments must have identical types.</p>
3539 <p>This instruction returns the unsigned integer <i>remainder</i> of a division.
3540 This instruction always performs an unsigned division to get the
3543 <p>Note that unsigned integer remainder and signed integer remainder are
3544 distinct operations; for signed integer remainder, use '<tt>srem</tt>'.</p>
3546 <p>Taking the remainder of a division by zero leads to undefined behavior.</p>
3550 <result> = urem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
3555 <!-- _______________________________________________________________________ -->
3556 <div class="doc_subsubsection">
3557 <a name="i_srem">'<tt>srem</tt>' Instruction</a>
3560 <div class="doc_text">
3564 <result> = srem <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3568 <p>The '<tt>srem</tt>' instruction returns the remainder from the signed
3569 division of its two operands. This instruction can also take
3570 <a href="#t_vector">vector</a> versions of the values in which case the
3571 elements must be integers.</p>
3574 <p>The two arguments to the '<tt>srem</tt>' instruction must be
3575 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3576 values. Both arguments must have identical types.</p>
3579 <p>This instruction returns the <i>remainder</i> of a division (where the result
3580 has the same sign as the dividend, <tt>op1</tt>), not the <i>modulo</i>
3581 operator (where the result has the same sign as the divisor, <tt>op2</tt>) of
3582 a value. For more information about the difference,
3583 see <a href="http://mathforum.org/dr.math/problems/anne.4.28.99.html">The
3584 Math Forum</a>. For a table of how this is implemented in various languages,
3585 please see <a href="http://en.wikipedia.org/wiki/Modulo_operation">
3586 Wikipedia: modulo operation</a>.</p>
3588 <p>Note that signed integer remainder and unsigned integer remainder are
3589 distinct operations; for unsigned integer remainder, use '<tt>urem</tt>'.</p>
3591 <p>Taking the remainder of a division by zero leads to undefined behavior.
3592 Overflow also leads to undefined behavior; this is a rare case, but can
3593 occur, for example, by taking the remainder of a 32-bit division of
3594 -2147483648 by -1. (The remainder doesn't actually overflow, but this rule
3595 lets srem be implemented using instructions that return both the result of
3596 the division and the remainder.)</p>
3600 <result> = srem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
3605 <!-- _______________________________________________________________________ -->
3606 <div class="doc_subsubsection">
3607 <a name="i_frem">'<tt>frem</tt>' Instruction</a> </div>
3609 <div class="doc_text">
3613 <result> = frem <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3617 <p>The '<tt>frem</tt>' instruction returns the remainder from the division of
3618 its two operands.</p>
3621 <p>The two arguments to the '<tt>frem</tt>' instruction must be
3622 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3623 floating point values. Both arguments must have identical types.</p>
3626 <p>This instruction returns the <i>remainder</i> of a division. The remainder
3627 has the same sign as the dividend.</p>
3631 <result> = frem float 4.0, %var <i>; yields {float}:result = 4.0 % %var</i>
3636 <!-- ======================================================================= -->
3637 <div class="doc_subsection"> <a name="bitwiseops">Bitwise Binary
3638 Operations</a> </div>
3640 <div class="doc_text">
3642 <p>Bitwise binary operators are used to do various forms of bit-twiddling in a
3643 program. They are generally very efficient instructions and can commonly be
3644 strength reduced from other instructions. They require two operands of the
3645 same type, execute an operation on them, and produce a single value. The
3646 resulting value is the same type as its operands.</p>
3650 <!-- _______________________________________________________________________ -->
3651 <div class="doc_subsubsection"> <a name="i_shl">'<tt>shl</tt>'
3652 Instruction</a> </div>
3654 <div class="doc_text">
3658 <result> = shl <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3662 <p>The '<tt>shl</tt>' instruction returns the first operand shifted to the left
3663 a specified number of bits.</p>
3666 <p>Both arguments to the '<tt>shl</tt>' instruction must be the
3667 same <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
3668 integer type. '<tt>op2</tt>' is treated as an unsigned value.</p>
3671 <p>The value produced is <tt>op1</tt> * 2<sup><tt>op2</tt></sup> mod
3672 2<sup>n</sup>, where <tt>n</tt> is the width of the result. If <tt>op2</tt>
3673 is (statically or dynamically) negative or equal to or larger than the number
3674 of bits in <tt>op1</tt>, the result is undefined. If the arguments are
3675 vectors, each vector element of <tt>op1</tt> is shifted by the corresponding
3676 shift amount in <tt>op2</tt>.</p>
3680 <result> = shl i32 4, %var <i>; yields {i32}: 4 << %var</i>
3681 <result> = shl i32 4, 2 <i>; yields {i32}: 16</i>
3682 <result> = shl i32 1, 10 <i>; yields {i32}: 1024</i>
3683 <result> = shl i32 1, 32 <i>; undefined</i>
3684 <result> = shl <2 x i32> < i32 1, i32 1>, < i32 1, i32 2> <i>; yields: result=<2 x i32> < i32 2, i32 4></i>
3689 <!-- _______________________________________________________________________ -->
3690 <div class="doc_subsubsection"> <a name="i_lshr">'<tt>lshr</tt>'
3691 Instruction</a> </div>
3693 <div class="doc_text">
3697 <result> = lshr <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3701 <p>The '<tt>lshr</tt>' instruction (logical shift right) returns the first
3702 operand shifted to the right a specified number of bits with zero fill.</p>
3705 <p>Both arguments to the '<tt>lshr</tt>' instruction must be the same
3706 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3707 type. '<tt>op2</tt>' is treated as an unsigned value.</p>
3710 <p>This instruction always performs a logical shift right operation. The most
3711 significant bits of the result will be filled with zero bits after the shift.
3712 If <tt>op2</tt> is (statically or dynamically) equal to or larger than the
3713 number of bits in <tt>op1</tt>, the result is undefined. If the arguments are
3714 vectors, each vector element of <tt>op1</tt> is shifted by the corresponding
3715 shift amount in <tt>op2</tt>.</p>
3719 <result> = lshr i32 4, 1 <i>; yields {i32}:result = 2</i>
3720 <result> = lshr i32 4, 2 <i>; yields {i32}:result = 1</i>
3721 <result> = lshr i8 4, 3 <i>; yields {i8}:result = 0</i>
3722 <result> = lshr i8 -2, 1 <i>; yields {i8}:result = 0x7FFFFFFF </i>
3723 <result> = lshr i32 1, 32 <i>; undefined</i>
3724 <result> = lshr <2 x i32> < i32 -2, i32 4>, < i32 1, i32 2> <i>; yields: result=<2 x i32> < i32 0x7FFFFFFF, i32 1></i>
3729 <!-- _______________________________________________________________________ -->
3730 <div class="doc_subsubsection"> <a name="i_ashr">'<tt>ashr</tt>'
3731 Instruction</a> </div>
3732 <div class="doc_text">
3736 <result> = ashr <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3740 <p>The '<tt>ashr</tt>' instruction (arithmetic shift right) returns the first
3741 operand shifted to the right a specified number of bits with sign
3745 <p>Both arguments to the '<tt>ashr</tt>' instruction must be the same
3746 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3747 type. '<tt>op2</tt>' is treated as an unsigned value.</p>
3750 <p>This instruction always performs an arithmetic shift right operation, The
3751 most significant bits of the result will be filled with the sign bit
3752 of <tt>op1</tt>. If <tt>op2</tt> is (statically or dynamically) equal to or
3753 larger than the number of bits in <tt>op1</tt>, the result is undefined. If
3754 the arguments are vectors, each vector element of <tt>op1</tt> is shifted by
3755 the corresponding shift amount in <tt>op2</tt>.</p>
3759 <result> = ashr i32 4, 1 <i>; yields {i32}:result = 2</i>
3760 <result> = ashr i32 4, 2 <i>; yields {i32}:result = 1</i>
3761 <result> = ashr i8 4, 3 <i>; yields {i8}:result = 0</i>
3762 <result> = ashr i8 -2, 1 <i>; yields {i8}:result = -1</i>
3763 <result> = ashr i32 1, 32 <i>; undefined</i>
3764 <result> = ashr <2 x i32> < i32 -2, i32 4>, < i32 1, i32 3> <i>; yields: result=<2 x i32> < i32 -1, i32 0></i>
3769 <!-- _______________________________________________________________________ -->
3770 <div class="doc_subsubsection"> <a name="i_and">'<tt>and</tt>'
3771 Instruction</a> </div>
3773 <div class="doc_text">
3777 <result> = and <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3781 <p>The '<tt>and</tt>' instruction returns the bitwise logical and of its two
3785 <p>The two arguments to the '<tt>and</tt>' instruction must be
3786 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3787 values. Both arguments must have identical types.</p>
3790 <p>The truth table used for the '<tt>and</tt>' instruction is:</p>
3792 <table border="1" cellspacing="0" cellpadding="4">
3824 <result> = and i32 4, %var <i>; yields {i32}:result = 4 & %var</i>
3825 <result> = and i32 15, 40 <i>; yields {i32}:result = 8</i>
3826 <result> = and i32 4, 8 <i>; yields {i32}:result = 0</i>
3829 <!-- _______________________________________________________________________ -->
3830 <div class="doc_subsubsection"> <a name="i_or">'<tt>or</tt>' Instruction</a> </div>
3832 <div class="doc_text">
3836 <result> = or <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3840 <p>The '<tt>or</tt>' instruction returns the bitwise logical inclusive or of its
3844 <p>The two arguments to the '<tt>or</tt>' instruction must be
3845 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3846 values. Both arguments must have identical types.</p>
3849 <p>The truth table used for the '<tt>or</tt>' instruction is:</p>
3851 <table border="1" cellspacing="0" cellpadding="4">
3883 <result> = or i32 4, %var <i>; yields {i32}:result = 4 | %var</i>
3884 <result> = or i32 15, 40 <i>; yields {i32}:result = 47</i>
3885 <result> = or i32 4, 8 <i>; yields {i32}:result = 12</i>
3890 <!-- _______________________________________________________________________ -->
3891 <div class="doc_subsubsection"> <a name="i_xor">'<tt>xor</tt>'
3892 Instruction</a> </div>
3894 <div class="doc_text">
3898 <result> = xor <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3902 <p>The '<tt>xor</tt>' instruction returns the bitwise logical exclusive or of
3903 its two operands. The <tt>xor</tt> is used to implement the "one's
3904 complement" operation, which is the "~" operator in C.</p>
3907 <p>The two arguments to the '<tt>xor</tt>' instruction must be
3908 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3909 values. Both arguments must have identical types.</p>
3912 <p>The truth table used for the '<tt>xor</tt>' instruction is:</p>
3914 <table border="1" cellspacing="0" cellpadding="4">
3946 <result> = xor i32 4, %var <i>; yields {i32}:result = 4 ^ %var</i>
3947 <result> = xor i32 15, 40 <i>; yields {i32}:result = 39</i>
3948 <result> = xor i32 4, 8 <i>; yields {i32}:result = 12</i>
3949 <result> = xor i32 %V, -1 <i>; yields {i32}:result = ~%V</i>
3954 <!-- ======================================================================= -->
3955 <div class="doc_subsection">
3956 <a name="vectorops">Vector Operations</a>
3959 <div class="doc_text">
3961 <p>LLVM supports several instructions to represent vector operations in a
3962 target-independent manner. These instructions cover the element-access and
3963 vector-specific operations needed to process vectors effectively. While LLVM
3964 does directly support these vector operations, many sophisticated algorithms
3965 will want to use target-specific intrinsics to take full advantage of a
3966 specific target.</p>
3970 <!-- _______________________________________________________________________ -->
3971 <div class="doc_subsubsection">
3972 <a name="i_extractelement">'<tt>extractelement</tt>' Instruction</a>
3975 <div class="doc_text">
3979 <result> = extractelement <n x <ty>> <val>, i32 <idx> <i>; yields <ty></i>
3983 <p>The '<tt>extractelement</tt>' instruction extracts a single scalar element
3984 from a vector at a specified index.</p>
3988 <p>The first operand of an '<tt>extractelement</tt>' instruction is a value
3989 of <a href="#t_vector">vector</a> type. The second operand is an index
3990 indicating the position from which to extract the element. The index may be
3994 <p>The result is a scalar of the same type as the element type of
3995 <tt>val</tt>. Its value is the value at position <tt>idx</tt> of
3996 <tt>val</tt>. If <tt>idx</tt> exceeds the length of <tt>val</tt>, the
3997 results are undefined.</p>
4001 <result> = extractelement <4 x i32> %vec, i32 0 <i>; yields i32</i>
4006 <!-- _______________________________________________________________________ -->
4007 <div class="doc_subsubsection">
4008 <a name="i_insertelement">'<tt>insertelement</tt>' Instruction</a>
4011 <div class="doc_text">
4015 <result> = insertelement <n x <ty>> <val>, <ty> <elt>, i32 <idx> <i>; yields <n x <ty>></i>
4019 <p>The '<tt>insertelement</tt>' instruction inserts a scalar element into a
4020 vector at a specified index.</p>
4023 <p>The first operand of an '<tt>insertelement</tt>' instruction is a value
4024 of <a href="#t_vector">vector</a> type. The second operand is a scalar value
4025 whose type must equal the element type of the first operand. The third
4026 operand is an index indicating the position at which to insert the value.
4027 The index may be a variable.</p>
4030 <p>The result is a vector of the same type as <tt>val</tt>. Its element values
4031 are those of <tt>val</tt> except at position <tt>idx</tt>, where it gets the
4032 value <tt>elt</tt>. If <tt>idx</tt> exceeds the length of <tt>val</tt>, the
4033 results are undefined.</p>
4037 <result> = insertelement <4 x i32> %vec, i32 1, i32 0 <i>; yields <4 x i32></i>
4042 <!-- _______________________________________________________________________ -->
4043 <div class="doc_subsubsection">
4044 <a name="i_shufflevector">'<tt>shufflevector</tt>' Instruction</a>
4047 <div class="doc_text">
4051 <result> = shufflevector <n x <ty>> <v1>, <n x <ty>> <v2>, <m x i32> <mask> <i>; yields <m x <ty>></i>
4055 <p>The '<tt>shufflevector</tt>' instruction constructs a permutation of elements
4056 from two input vectors, returning a vector with the same element type as the
4057 input and length that is the same as the shuffle mask.</p>
4060 <p>The first two operands of a '<tt>shufflevector</tt>' instruction are vectors
4061 with types that match each other. The third argument is a shuffle mask whose
4062 element type is always 'i32'. The result of the instruction is a vector
4063 whose length is the same as the shuffle mask and whose element type is the
4064 same as the element type of the first two operands.</p>
4066 <p>The shuffle mask operand is required to be a constant vector with either
4067 constant integer or undef values.</p>
4070 <p>The elements of the two input vectors are numbered from left to right across
4071 both of the vectors. The shuffle mask operand specifies, for each element of
4072 the result vector, which element of the two input vectors the result element
4073 gets. The element selector may be undef (meaning "don't care") and the
4074 second operand may be undef if performing a shuffle from only one vector.</p>
4078 <result> = shufflevector <4 x i32> %v1, <4 x i32> %v2,
4079 <4 x i32> <i32 0, i32 4, i32 1, i32 5> <i>; yields <4 x i32></i>
4080 <result> = shufflevector <4 x i32> %v1, <4 x i32> undef,
4081 <4 x i32> <i32 0, i32 1, i32 2, i32 3> <i>; yields <4 x i32></i> - Identity shuffle.
4082 <result> = shufflevector <8 x i32> %v1, <8 x i32> undef,
4083 <4 x i32> <i32 0, i32 1, i32 2, i32 3> <i>; yields <4 x i32></i>
4084 <result> = shufflevector <4 x i32> %v1, <4 x i32> %v2,
4085 <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>
4090 <!-- ======================================================================= -->
4091 <div class="doc_subsection">
4092 <a name="aggregateops">Aggregate Operations</a>
4095 <div class="doc_text">
4097 <p>LLVM supports several instructions for working with
4098 <a href="#t_aggregate">aggregate</a> values.</p>
4102 <!-- _______________________________________________________________________ -->
4103 <div class="doc_subsubsection">
4104 <a name="i_extractvalue">'<tt>extractvalue</tt>' Instruction</a>
4107 <div class="doc_text">
4111 <result> = extractvalue <aggregate type> <val>, <idx>{, <idx>}*
4115 <p>The '<tt>extractvalue</tt>' instruction extracts the value of a member field
4116 from an <a href="#t_aggregate">aggregate</a> value.</p>
4119 <p>The first operand of an '<tt>extractvalue</tt>' instruction is a value
4120 of <a href="#t_struct">struct</a> or
4121 <a href="#t_array">array</a> type. The operands are constant indices to
4122 specify which value to extract in a similar manner as indices in a
4123 '<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.</p>
4126 <p>The result is the value at the position in the aggregate specified by the
4131 <result> = extractvalue {i32, float} %agg, 0 <i>; yields i32</i>
4136 <!-- _______________________________________________________________________ -->
4137 <div class="doc_subsubsection">
4138 <a name="i_insertvalue">'<tt>insertvalue</tt>' Instruction</a>
4141 <div class="doc_text">
4145 <result> = insertvalue <aggregate type> <val>, <ty> <elt>, <idx> <i>; yields <aggregate type></i>
4149 <p>The '<tt>insertvalue</tt>' instruction inserts a value into a member field
4150 in an <a href="#t_aggregate">aggregate</a> value.</p>
4153 <p>The first operand of an '<tt>insertvalue</tt>' instruction is a value
4154 of <a href="#t_struct">struct</a> or
4155 <a href="#t_array">array</a> type. The second operand is a first-class
4156 value to insert. The following operands are constant indices indicating
4157 the position at which to insert the value in a similar manner as indices in a
4158 '<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction. The
4159 value to insert must have the same type as the value identified by the
4163 <p>The result is an aggregate of the same type as <tt>val</tt>. Its value is
4164 that of <tt>val</tt> except that the value at the position specified by the
4165 indices is that of <tt>elt</tt>.</p>
4169 %agg1 = insertvalue {i32, float} undef, i32 1, 0 <i>; yields {i32 1, float undef}</i>
4170 %agg2 = insertvalue {i32, float} %agg1, float %val, 1 <i>; yields {i32 1, float %val}</i>
4176 <!-- ======================================================================= -->
4177 <div class="doc_subsection">
4178 <a name="memoryops">Memory Access and Addressing Operations</a>
4181 <div class="doc_text">
4183 <p>A key design point of an SSA-based representation is how it represents
4184 memory. In LLVM, no memory locations are in SSA form, which makes things
4185 very simple. This section describes how to read, write, and allocate
4190 <!-- _______________________________________________________________________ -->
4191 <div class="doc_subsubsection">
4192 <a name="i_alloca">'<tt>alloca</tt>' Instruction</a>
4195 <div class="doc_text">
4199 <result> = alloca <type>[, <ty> <NumElements>][, align <alignment>] <i>; yields {type*}:result</i>
4203 <p>The '<tt>alloca</tt>' instruction allocates memory on the stack frame of the
4204 currently executing function, to be automatically released when this function
4205 returns to its caller. The object is always allocated in the generic address
4206 space (address space zero).</p>
4209 <p>The '<tt>alloca</tt>' instruction
4210 allocates <tt>sizeof(<type>)*NumElements</tt> bytes of memory on the
4211 runtime stack, returning a pointer of the appropriate type to the program.
4212 If "NumElements" is specified, it is the number of elements allocated,
4213 otherwise "NumElements" is defaulted to be one. If a constant alignment is
4214 specified, the value result of the allocation is guaranteed to be aligned to
4215 at least that boundary. If not specified, or if zero, the target can choose
4216 to align the allocation on any convenient boundary compatible with the
4219 <p>'<tt>type</tt>' may be any sized type.</p>
4222 <p>Memory is allocated; a pointer is returned. The operation is undefined if
4223 there is insufficient stack space for the allocation. '<tt>alloca</tt>'d
4224 memory is automatically released when the function returns. The
4225 '<tt>alloca</tt>' instruction is commonly used to represent automatic
4226 variables that must have an address available. When the function returns
4227 (either with the <tt><a href="#i_ret">ret</a></tt>
4228 or <tt><a href="#i_unwind">unwind</a></tt> instructions), the memory is
4229 reclaimed. Allocating zero bytes is legal, but the result is undefined.</p>
4233 %ptr = alloca i32 <i>; yields {i32*}:ptr</i>
4234 %ptr = alloca i32, i32 4 <i>; yields {i32*}:ptr</i>
4235 %ptr = alloca i32, i32 4, align 1024 <i>; yields {i32*}:ptr</i>
4236 %ptr = alloca i32, align 1024 <i>; yields {i32*}:ptr</i>
4241 <!-- _______________________________________________________________________ -->
4242 <div class="doc_subsubsection"> <a name="i_load">'<tt>load</tt>'
4243 Instruction</a> </div>
4245 <div class="doc_text">
4249 <result> = load <ty>* <pointer>[, align <alignment>][, !nontemporal !<index>]
4250 <result> = volatile load <ty>* <pointer>[, align <alignment>][, !nontemporal !<index>]
4251 !<index> = !{ i32 1 }
4255 <p>The '<tt>load</tt>' instruction is used to read from memory.</p>
4258 <p>The argument to the '<tt>load</tt>' instruction specifies the memory address
4259 from which to load. The pointer must point to
4260 a <a href="#t_firstclass">first class</a> type. If the <tt>load</tt> is
4261 marked as <tt>volatile</tt>, then the optimizer is not allowed to modify the
4262 number or order of execution of this <tt>load</tt> with other <a
4263 href="#volatile">volatile operations</a>.</p>
4265 <p>The optional constant <tt>align</tt> argument specifies the alignment of the
4266 operation (that is, the alignment of the memory address). A value of 0 or an
4267 omitted <tt>align</tt> argument means that the operation has the preferential
4268 alignment for the target. It is the responsibility of the code emitter to
4269 ensure that the alignment information is correct. Overestimating the
4270 alignment results in undefined behavior. Underestimating the alignment may
4271 produce less efficient code. An alignment of 1 is always safe.</p>
4273 <p>The optional <tt>!nontemporal</tt> metadata must reference a single
4274 metatadata name <index> corresponding to a metadata node with
4275 one <tt>i32</tt> entry of value 1. The existence of
4276 the <tt>!nontemporal</tt> metatadata on the instruction tells the optimizer
4277 and code generator that this load is not expected to be reused in the cache.
4278 The code generator may select special instructions to save cache bandwidth,
4279 such as the <tt>MOVNT</tt> instruction on x86.</p>
4282 <p>The location of memory pointed to is loaded. If the value being loaded is of
4283 scalar type then the number of bytes read does not exceed the minimum number
4284 of bytes needed to hold all bits of the type. For example, loading an
4285 <tt>i24</tt> reads at most three bytes. When loading a value of a type like
4286 <tt>i20</tt> with a size that is not an integral number of bytes, the result
4287 is undefined if the value was not originally written using a store of the
4292 %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
4293 <a href="#i_store">store</a> i32 3, i32* %ptr <i>; yields {void}</i>
4294 %val = load i32* %ptr <i>; yields {i32}:val = i32 3</i>
4299 <!-- _______________________________________________________________________ -->
4300 <div class="doc_subsubsection"> <a name="i_store">'<tt>store</tt>'
4301 Instruction</a> </div>
4303 <div class="doc_text">
4307 store <ty> <value>, <ty>* <pointer>[, align <alignment>][, !nontemporal !<index>] <i>; yields {void}</i>
4308 volatile store <ty> <value>, <ty>* <pointer>[, align <alignment>][, !nontemporal !<index>] <i>; yields {void}</i>
4312 <p>The '<tt>store</tt>' instruction is used to write to memory.</p>
4315 <p>There are two arguments to the '<tt>store</tt>' instruction: a value to store
4316 and an address at which to store it. The type of the
4317 '<tt><pointer></tt>' operand must be a pointer to
4318 the <a href="#t_firstclass">first class</a> type of the
4319 '<tt><value></tt>' operand. If the <tt>store</tt> is marked as
4320 <tt>volatile</tt>, then the optimizer is not allowed to modify the number or
4321 order of execution of this <tt>store</tt> with other <a
4322 href="#volatile">volatile operations</a>.</p>
4324 <p>The optional constant "align" argument specifies the alignment of the
4325 operation (that is, the alignment of the memory address). A value of 0 or an
4326 omitted "align" argument means that the operation has the preferential
4327 alignment for the target. It is the responsibility of the code emitter to
4328 ensure that the alignment information is correct. Overestimating the
4329 alignment results in an undefined behavior. Underestimating the alignment may
4330 produce less efficient code. An alignment of 1 is always safe.</p>
4332 <p>The optional !nontemporal metadata must reference a single metatadata
4333 name <index> corresponding to a metadata node with one i32 entry of
4334 value 1. The existence of the !nontemporal metatadata on the
4335 instruction tells the optimizer and code generator that this load is
4336 not expected to be reused in the cache. The code generator may
4337 select special instructions to save cache bandwidth, such as the
4338 MOVNT instruction on x86.</p>
4342 <p>The contents of memory are updated to contain '<tt><value></tt>' at the
4343 location specified by the '<tt><pointer></tt>' operand. If
4344 '<tt><value></tt>' is of scalar type then the number of bytes written
4345 does not exceed the minimum number of bytes needed to hold all bits of the
4346 type. For example, storing an <tt>i24</tt> writes at most three bytes. When
4347 writing a value of a type like <tt>i20</tt> with a size that is not an
4348 integral number of bytes, it is unspecified what happens to the extra bits
4349 that do not belong to the type, but they will typically be overwritten.</p>
4353 %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
4354 store i32 3, i32* %ptr <i>; yields {void}</i>
4355 %val = <a href="#i_load">load</a> i32* %ptr <i>; yields {i32}:val = i32 3</i>
4360 <!-- _______________________________________________________________________ -->
4361 <div class="doc_subsubsection">
4362 <a name="i_getelementptr">'<tt>getelementptr</tt>' Instruction</a>
4365 <div class="doc_text">
4369 <result> = getelementptr <pty>* <ptrval>{, <ty> <idx>}*
4370 <result> = getelementptr inbounds <pty>* <ptrval>{, <ty> <idx>}*
4374 <p>The '<tt>getelementptr</tt>' instruction is used to get the address of a
4375 subelement of an <a href="#t_aggregate">aggregate</a> data structure.
4376 It performs address calculation only and does not access memory.</p>
4379 <p>The first argument is always a pointer, and forms the basis of the
4380 calculation. The remaining arguments are indices that indicate which of the
4381 elements of the aggregate object are indexed. The interpretation of each
4382 index is dependent on the type being indexed into. The first index always
4383 indexes the pointer value given as the first argument, the second index
4384 indexes a value of the type pointed to (not necessarily the value directly
4385 pointed to, since the first index can be non-zero), etc. The first type
4386 indexed into must be a pointer value, subsequent types can be arrays,
4387 vectors, and structs. Note that subsequent types being indexed into
4388 can never be pointers, since that would require loading the pointer before
4389 continuing calculation.</p>
4391 <p>The type of each index argument depends on the type it is indexing into.
4392 When indexing into a (optionally packed) structure, only <tt>i32</tt>
4393 integer <b>constants</b> are allowed. When indexing into an array, pointer
4394 or vector, integers of any width are allowed, and they are not required to be
4397 <p>For example, let's consider a C code fragment and how it gets compiled to
4400 <pre class="doc_code">
4412 int *foo(struct ST *s) {
4413 return &s[1].Z.B[5][13];
4417 <p>The LLVM code generated by the GCC frontend is:</p>
4419 <pre class="doc_code">
4420 %RT = <a href="#namedtypes">type</a> { i8 , [10 x [20 x i32]], i8 }
4421 %ST = <a href="#namedtypes">type</a> { i32, double, %RT }
4423 define i32* @foo(%ST* %s) {
4425 %reg = getelementptr %ST* %s, i32 1, i32 2, i32 1, i32 5, i32 13
4431 <p>In the example above, the first index is indexing into the '<tt>%ST*</tt>'
4432 type, which is a pointer, yielding a '<tt>%ST</tt>' = '<tt>{ i32, double, %RT
4433 }</tt>' type, a structure. The second index indexes into the third element
4434 of the structure, yielding a '<tt>%RT</tt>' = '<tt>{ i8 , [10 x [20 x i32]],
4435 i8 }</tt>' type, another structure. The third index indexes into the second
4436 element of the structure, yielding a '<tt>[10 x [20 x i32]]</tt>' type, an
4437 array. The two dimensions of the array are subscripted into, yielding an
4438 '<tt>i32</tt>' type. The '<tt>getelementptr</tt>' instruction returns a
4439 pointer to this element, thus computing a value of '<tt>i32*</tt>' type.</p>
4441 <p>Note that it is perfectly legal to index partially through a structure,
4442 returning a pointer to an inner element. Because of this, the LLVM code for
4443 the given testcase is equivalent to:</p>
4446 define i32* @foo(%ST* %s) {
4447 %t1 = getelementptr %ST* %s, i32 1 <i>; yields %ST*:%t1</i>
4448 %t2 = getelementptr %ST* %t1, i32 0, i32 2 <i>; yields %RT*:%t2</i>
4449 %t3 = getelementptr %RT* %t2, i32 0, i32 1 <i>; yields [10 x [20 x i32]]*:%t3</i>
4450 %t4 = getelementptr [10 x [20 x i32]]* %t3, i32 0, i32 5 <i>; yields [20 x i32]*:%t4</i>
4451 %t5 = getelementptr [20 x i32]* %t4, i32 0, i32 13 <i>; yields i32*:%t5</i>
4456 <p>If the <tt>inbounds</tt> keyword is present, the result value of the
4457 <tt>getelementptr</tt> is a <a href="#trapvalues">trap value</a> if the
4458 base pointer is not an <i>in bounds</i> address of an allocated object,
4459 or if any of the addresses that would be formed by successive addition of
4460 the offsets implied by the indices to the base address with infinitely
4461 precise arithmetic are not an <i>in bounds</i> address of that allocated
4462 object. The <i>in bounds</i> addresses for an allocated object are all
4463 the addresses that point into the object, plus the address one byte past
4466 <p>If the <tt>inbounds</tt> keyword is not present, the offsets are added to
4467 the base address with silently-wrapping two's complement arithmetic, and
4468 the result value of the <tt>getelementptr</tt> may be outside the object
4469 pointed to by the base pointer. The result value may not necessarily be
4470 used to access memory though, even if it happens to point into allocated
4471 storage. See the <a href="#pointeraliasing">Pointer Aliasing Rules</a>
4472 section for more information.</p>
4474 <p>The getelementptr instruction is often confusing. For some more insight into
4475 how it works, see <a href="GetElementPtr.html">the getelementptr FAQ</a>.</p>
4479 <i>; yields [12 x i8]*:aptr</i>
4480 %aptr = getelementptr {i32, [12 x i8]}* %saptr, i64 0, i32 1
4481 <i>; yields i8*:vptr</i>
4482 %vptr = getelementptr {i32, <2 x i8>}* %svptr, i64 0, i32 1, i32 1
4483 <i>; yields i8*:eptr</i>
4484 %eptr = getelementptr [12 x i8]* %aptr, i64 0, i32 1
4485 <i>; yields i32*:iptr</i>
4486 %iptr = getelementptr [10 x i32]* @arr, i16 0, i16 0
4491 <!-- ======================================================================= -->
4492 <div class="doc_subsection"> <a name="convertops">Conversion Operations</a>
4495 <div class="doc_text">
4497 <p>The instructions in this category are the conversion instructions (casting)
4498 which all take a single operand and a type. They perform various bit
4499 conversions on the operand.</p>
4503 <!-- _______________________________________________________________________ -->
4504 <div class="doc_subsubsection">
4505 <a name="i_trunc">'<tt>trunc .. to</tt>' Instruction</a>
4507 <div class="doc_text">
4511 <result> = trunc <ty> <value> to <ty2> <i>; yields ty2</i>
4515 <p>The '<tt>trunc</tt>' instruction truncates its operand to the
4516 type <tt>ty2</tt>.</p>
4519 <p>The '<tt>trunc</tt>' instruction takes a <tt>value</tt> to trunc, which must
4520 be an <a href="#t_integer">integer</a> type, and a type that specifies the
4521 size and type of the result, which must be
4522 an <a href="#t_integer">integer</a> type. The bit size of <tt>value</tt> must
4523 be larger than the bit size of <tt>ty2</tt>. Equal sized types are not
4527 <p>The '<tt>trunc</tt>' instruction truncates the high order bits
4528 in <tt>value</tt> and converts the remaining bits to <tt>ty2</tt>. Since the
4529 source size must be larger than the destination size, <tt>trunc</tt> cannot
4530 be a <i>no-op cast</i>. It will always truncate bits.</p>
4534 %X = trunc i32 257 to i8 <i>; yields i8:1</i>
4535 %Y = trunc i32 123 to i1 <i>; yields i1:true</i>
4536 %Z = trunc i32 122 to i1 <i>; yields i1:false</i>
4541 <!-- _______________________________________________________________________ -->
4542 <div class="doc_subsubsection">
4543 <a name="i_zext">'<tt>zext .. to</tt>' Instruction</a>
4545 <div class="doc_text">
4549 <result> = zext <ty> <value> to <ty2> <i>; yields ty2</i>
4553 <p>The '<tt>zext</tt>' instruction zero extends its operand to type
4558 <p>The '<tt>zext</tt>' instruction takes a value to cast, which must be of
4559 <a href="#t_integer">integer</a> type, and a type to cast it to, which must
4560 also be of <a href="#t_integer">integer</a> type. The bit size of the
4561 <tt>value</tt> must be smaller than the bit size of the destination type,
4565 <p>The <tt>zext</tt> fills the high order bits of the <tt>value</tt> with zero
4566 bits until it reaches the size of the destination type, <tt>ty2</tt>.</p>
4568 <p>When zero extending from i1, the result will always be either 0 or 1.</p>
4572 %X = zext i32 257 to i64 <i>; yields i64:257</i>
4573 %Y = zext i1 true to i32 <i>; yields i32:1</i>
4578 <!-- _______________________________________________________________________ -->
4579 <div class="doc_subsubsection">
4580 <a name="i_sext">'<tt>sext .. to</tt>' Instruction</a>
4582 <div class="doc_text">
4586 <result> = sext <ty> <value> to <ty2> <i>; yields ty2</i>
4590 <p>The '<tt>sext</tt>' sign extends <tt>value</tt> to the type <tt>ty2</tt>.</p>
4593 <p>The '<tt>sext</tt>' instruction takes a value to cast, which must be of
4594 <a href="#t_integer">integer</a> type, and a type to cast it to, which must
4595 also be of <a href="#t_integer">integer</a> type. The bit size of the
4596 <tt>value</tt> must be smaller than the bit size of the destination type,
4600 <p>The '<tt>sext</tt>' instruction performs a sign extension by copying the sign
4601 bit (highest order bit) of the <tt>value</tt> until it reaches the bit size
4602 of the type <tt>ty2</tt>.</p>
4604 <p>When sign extending from i1, the extension always results in -1 or 0.</p>
4608 %X = sext i8 -1 to i16 <i>; yields i16 :65535</i>
4609 %Y = sext i1 true to i32 <i>; yields i32:-1</i>
4614 <!-- _______________________________________________________________________ -->
4615 <div class="doc_subsubsection">
4616 <a name="i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a>
4619 <div class="doc_text">
4623 <result> = fptrunc <ty> <value> to <ty2> <i>; yields ty2</i>
4627 <p>The '<tt>fptrunc</tt>' instruction truncates <tt>value</tt> to type
4631 <p>The '<tt>fptrunc</tt>' instruction takes a <a href="#t_floating">floating
4632 point</a> value to cast and a <a href="#t_floating">floating point</a> type
4633 to cast it to. The size of <tt>value</tt> must be larger than the size of
4634 <tt>ty2</tt>. This implies that <tt>fptrunc</tt> cannot be used to make a
4635 <i>no-op cast</i>.</p>
4638 <p>The '<tt>fptrunc</tt>' instruction truncates a <tt>value</tt> from a larger
4639 <a href="#t_floating">floating point</a> type to a smaller
4640 <a href="#t_floating">floating point</a> type. If the value cannot fit
4641 within the destination type, <tt>ty2</tt>, then the results are
4646 %X = fptrunc double 123.0 to float <i>; yields float:123.0</i>
4647 %Y = fptrunc double 1.0E+300 to float <i>; yields undefined</i>
4652 <!-- _______________________________________________________________________ -->
4653 <div class="doc_subsubsection">
4654 <a name="i_fpext">'<tt>fpext .. to</tt>' Instruction</a>
4656 <div class="doc_text">
4660 <result> = fpext <ty> <value> to <ty2> <i>; yields ty2</i>
4664 <p>The '<tt>fpext</tt>' extends a floating point <tt>value</tt> to a larger
4665 floating point value.</p>
4668 <p>The '<tt>fpext</tt>' instruction takes a
4669 <a href="#t_floating">floating point</a> <tt>value</tt> to cast, and
4670 a <a href="#t_floating">floating point</a> type to cast it to. The source
4671 type must be smaller than the destination type.</p>
4674 <p>The '<tt>fpext</tt>' instruction extends the <tt>value</tt> from a smaller
4675 <a href="#t_floating">floating point</a> type to a larger
4676 <a href="#t_floating">floating point</a> type. The <tt>fpext</tt> cannot be
4677 used to make a <i>no-op cast</i> because it always changes bits. Use
4678 <tt>bitcast</tt> to make a <i>no-op cast</i> for a floating point cast.</p>
4682 %X = fpext float 3.1415 to double <i>; yields double:3.1415</i>
4683 %Y = fpext float 1.0 to float <i>; yields float:1.0 (no-op)</i>
4688 <!-- _______________________________________________________________________ -->
4689 <div class="doc_subsubsection">
4690 <a name="i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a>
4692 <div class="doc_text">
4696 <result> = fptoui <ty> <value> to <ty2> <i>; yields ty2</i>
4700 <p>The '<tt>fptoui</tt>' converts a floating point <tt>value</tt> to its
4701 unsigned integer equivalent of type <tt>ty2</tt>.</p>
4704 <p>The '<tt>fptoui</tt>' instruction takes a value to cast, which must be a
4705 scalar or vector <a href="#t_floating">floating point</a> value, and a type
4706 to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
4707 type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
4708 vector integer type with the same number of elements as <tt>ty</tt></p>
4711 <p>The '<tt>fptoui</tt>' instruction converts its
4712 <a href="#t_floating">floating point</a> operand into the nearest (rounding
4713 towards zero) unsigned integer value. If the value cannot fit
4714 in <tt>ty2</tt>, the results are undefined.</p>
4718 %X = fptoui double 123.0 to i32 <i>; yields i32:123</i>
4719 %Y = fptoui float 1.0E+300 to i1 <i>; yields undefined:1</i>
4720 %Z = fptoui float 1.04E+17 to i8 <i>; yields undefined:1</i>
4725 <!-- _______________________________________________________________________ -->
4726 <div class="doc_subsubsection">
4727 <a name="i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a>
4729 <div class="doc_text">
4733 <result> = fptosi <ty> <value> to <ty2> <i>; yields ty2</i>
4737 <p>The '<tt>fptosi</tt>' instruction converts
4738 <a href="#t_floating">floating point</a> <tt>value</tt> to
4739 type <tt>ty2</tt>.</p>
4742 <p>The '<tt>fptosi</tt>' instruction takes a value to cast, which must be a
4743 scalar or vector <a href="#t_floating">floating point</a> value, and a type
4744 to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
4745 type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
4746 vector integer type with the same number of elements as <tt>ty</tt></p>
4749 <p>The '<tt>fptosi</tt>' instruction converts its
4750 <a href="#t_floating">floating point</a> operand into the nearest (rounding
4751 towards zero) signed integer value. If the value cannot fit in <tt>ty2</tt>,
4752 the results are undefined.</p>
4756 %X = fptosi double -123.0 to i32 <i>; yields i32:-123</i>
4757 %Y = fptosi float 1.0E-247 to i1 <i>; yields undefined:1</i>
4758 %Z = fptosi float 1.04E+17 to i8 <i>; yields undefined:1</i>
4763 <!-- _______________________________________________________________________ -->
4764 <div class="doc_subsubsection">
4765 <a name="i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a>
4767 <div class="doc_text">
4771 <result> = uitofp <ty> <value> to <ty2> <i>; yields ty2</i>
4775 <p>The '<tt>uitofp</tt>' instruction regards <tt>value</tt> as an unsigned
4776 integer and converts that value to the <tt>ty2</tt> type.</p>
4779 <p>The '<tt>uitofp</tt>' instruction takes a value to cast, which must be a
4780 scalar or vector <a href="#t_integer">integer</a> value, and a type to cast
4781 it to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
4782 type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
4783 floating point type with the same number of elements as <tt>ty</tt></p>
4786 <p>The '<tt>uitofp</tt>' instruction interprets its operand as an unsigned
4787 integer quantity and converts it to the corresponding floating point
4788 value. If the value cannot fit in the floating point value, the results are
4793 %X = uitofp i32 257 to float <i>; yields float:257.0</i>
4794 %Y = uitofp i8 -1 to double <i>; yields double:255.0</i>
4799 <!-- _______________________________________________________________________ -->
4800 <div class="doc_subsubsection">
4801 <a name="i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a>
4803 <div class="doc_text">
4807 <result> = sitofp <ty> <value> to <ty2> <i>; yields ty2</i>
4811 <p>The '<tt>sitofp</tt>' instruction regards <tt>value</tt> as a signed integer
4812 and converts that value to the <tt>ty2</tt> type.</p>
4815 <p>The '<tt>sitofp</tt>' instruction takes a value to cast, which must be a
4816 scalar or vector <a href="#t_integer">integer</a> value, and a type to cast
4817 it to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
4818 type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
4819 floating point type with the same number of elements as <tt>ty</tt></p>
4822 <p>The '<tt>sitofp</tt>' instruction interprets its operand as a signed integer
4823 quantity and converts it to the corresponding floating point value. If the
4824 value cannot fit in the floating point value, the results are undefined.</p>
4828 %X = sitofp i32 257 to float <i>; yields float:257.0</i>
4829 %Y = sitofp i8 -1 to double <i>; yields double:-1.0</i>
4834 <!-- _______________________________________________________________________ -->
4835 <div class="doc_subsubsection">
4836 <a name="i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a>
4838 <div class="doc_text">
4842 <result> = ptrtoint <ty> <value> to <ty2> <i>; yields ty2</i>
4846 <p>The '<tt>ptrtoint</tt>' instruction converts the pointer <tt>value</tt> to
4847 the integer type <tt>ty2</tt>.</p>
4850 <p>The '<tt>ptrtoint</tt>' instruction takes a <tt>value</tt> to cast, which
4851 must be a <a href="#t_pointer">pointer</a> value, and a type to cast it to
4852 <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a> type.</p>
4855 <p>The '<tt>ptrtoint</tt>' instruction converts <tt>value</tt> to integer type
4856 <tt>ty2</tt> by interpreting the pointer value as an integer and either
4857 truncating or zero extending that value to the size of the integer type. If
4858 <tt>value</tt> is smaller than <tt>ty2</tt> then a zero extension is done. If
4859 <tt>value</tt> is larger than <tt>ty2</tt> then a truncation is done. If they
4860 are the same size, then nothing is done (<i>no-op cast</i>) other than a type
4865 %X = ptrtoint i32* %X to i8 <i>; yields truncation on 32-bit architecture</i>
4866 %Y = ptrtoint i32* %x to i64 <i>; yields zero extension on 32-bit architecture</i>
4871 <!-- _______________________________________________________________________ -->
4872 <div class="doc_subsubsection">
4873 <a name="i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a>
4875 <div class="doc_text">
4879 <result> = inttoptr <ty> <value> to <ty2> <i>; yields ty2</i>
4883 <p>The '<tt>inttoptr</tt>' instruction converts an integer <tt>value</tt> to a
4884 pointer type, <tt>ty2</tt>.</p>
4887 <p>The '<tt>inttoptr</tt>' instruction takes an <a href="#t_integer">integer</a>
4888 value to cast, and a type to cast it to, which must be a
4889 <a href="#t_pointer">pointer</a> type.</p>
4892 <p>The '<tt>inttoptr</tt>' instruction converts <tt>value</tt> to type
4893 <tt>ty2</tt> by applying either a zero extension or a truncation depending on
4894 the size of the integer <tt>value</tt>. If <tt>value</tt> is larger than the
4895 size of a pointer then a truncation is done. If <tt>value</tt> is smaller
4896 than the size of a pointer then a zero extension is done. If they are the
4897 same size, nothing is done (<i>no-op cast</i>).</p>
4901 %X = inttoptr i32 255 to i32* <i>; yields zero extension on 64-bit architecture</i>
4902 %Y = inttoptr i32 255 to i32* <i>; yields no-op on 32-bit architecture</i>
4903 %Z = inttoptr i64 0 to i32* <i>; yields truncation on 32-bit architecture</i>
4908 <!-- _______________________________________________________________________ -->
4909 <div class="doc_subsubsection">
4910 <a name="i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a>
4912 <div class="doc_text">
4916 <result> = bitcast <ty> <value> to <ty2> <i>; yields ty2</i>
4920 <p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
4921 <tt>ty2</tt> without changing any bits.</p>
4924 <p>The '<tt>bitcast</tt>' instruction takes a value to cast, which must be a
4925 non-aggregate first class value, and a type to cast it to, which must also be
4926 a non-aggregate <a href="#t_firstclass">first class</a> type. The bit sizes
4927 of <tt>value</tt> and the destination type, <tt>ty2</tt>, must be
4928 identical. If the source type is a pointer, the destination type must also be
4929 a pointer. This instruction supports bitwise conversion of vectors to
4930 integers and to vectors of other types (as long as they have the same
4934 <p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
4935 <tt>ty2</tt>. It is always a <i>no-op cast</i> because no bits change with
4936 this conversion. The conversion is done as if the <tt>value</tt> had been
4937 stored to memory and read back as type <tt>ty2</tt>. Pointer types may only
4938 be converted to other pointer types with this instruction. To convert
4939 pointers to other types, use the <a href="#i_inttoptr">inttoptr</a> or
4940 <a href="#i_ptrtoint">ptrtoint</a> instructions first.</p>
4944 %X = bitcast i8 255 to i8 <i>; yields i8 :-1</i>
4945 %Y = bitcast i32* %x to sint* <i>; yields sint*:%x</i>
4946 %Z = bitcast <2 x int> %V to i64; <i>; yields i64: %V</i>
4951 <!-- ======================================================================= -->
4952 <div class="doc_subsection"> <a name="otherops">Other Operations</a> </div>
4954 <div class="doc_text">
4956 <p>The instructions in this category are the "miscellaneous" instructions, which
4957 defy better classification.</p>
4961 <!-- _______________________________________________________________________ -->
4962 <div class="doc_subsubsection"><a name="i_icmp">'<tt>icmp</tt>' Instruction</a>
4965 <div class="doc_text">
4969 <result> = icmp <cond> <ty> <op1>, <op2> <i>; yields {i1} or {<N x i1>}:result</i>
4973 <p>The '<tt>icmp</tt>' instruction returns a boolean value or a vector of
4974 boolean values based on comparison of its two integer, integer vector, or
4975 pointer operands.</p>
4978 <p>The '<tt>icmp</tt>' instruction takes three operands. The first operand is
4979 the condition code indicating the kind of comparison to perform. It is not a
4980 value, just a keyword. The possible condition code are:</p>
4983 <li><tt>eq</tt>: equal</li>
4984 <li><tt>ne</tt>: not equal </li>
4985 <li><tt>ugt</tt>: unsigned greater than</li>
4986 <li><tt>uge</tt>: unsigned greater or equal</li>
4987 <li><tt>ult</tt>: unsigned less than</li>
4988 <li><tt>ule</tt>: unsigned less or equal</li>
4989 <li><tt>sgt</tt>: signed greater than</li>
4990 <li><tt>sge</tt>: signed greater or equal</li>
4991 <li><tt>slt</tt>: signed less than</li>
4992 <li><tt>sle</tt>: signed less or equal</li>
4995 <p>The remaining two arguments must be <a href="#t_integer">integer</a> or
4996 <a href="#t_pointer">pointer</a> or integer <a href="#t_vector">vector</a>
4997 typed. They must also be identical types.</p>
5000 <p>The '<tt>icmp</tt>' compares <tt>op1</tt> and <tt>op2</tt> according to the
5001 condition code given as <tt>cond</tt>. The comparison performed always yields
5002 either an <a href="#t_integer"><tt>i1</tt></a> or vector of <tt>i1</tt>
5003 result, as follows:</p>
5006 <li><tt>eq</tt>: yields <tt>true</tt> if the operands are equal,
5007 <tt>false</tt> otherwise. No sign interpretation is necessary or
5010 <li><tt>ne</tt>: yields <tt>true</tt> if the operands are unequal,
5011 <tt>false</tt> otherwise. No sign interpretation is necessary or
5014 <li><tt>ugt</tt>: interprets the operands as unsigned values and yields
5015 <tt>true</tt> if <tt>op1</tt> is greater than <tt>op2</tt>.</li>
5017 <li><tt>uge</tt>: interprets the operands as unsigned values and yields
5018 <tt>true</tt> if <tt>op1</tt> is greater than or equal
5019 to <tt>op2</tt>.</li>
5021 <li><tt>ult</tt>: interprets the operands as unsigned values and yields
5022 <tt>true</tt> if <tt>op1</tt> is less than <tt>op2</tt>.</li>
5024 <li><tt>ule</tt>: interprets the operands as unsigned values and yields
5025 <tt>true</tt> if <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
5027 <li><tt>sgt</tt>: interprets the operands as signed values and yields
5028 <tt>true</tt> if <tt>op1</tt> is greater than <tt>op2</tt>.</li>
5030 <li><tt>sge</tt>: interprets the operands as signed values and yields
5031 <tt>true</tt> if <tt>op1</tt> is greater than or equal
5032 to <tt>op2</tt>.</li>
5034 <li><tt>slt</tt>: interprets the operands as signed values and yields
5035 <tt>true</tt> if <tt>op1</tt> is less than <tt>op2</tt>.</li>
5037 <li><tt>sle</tt>: interprets the operands as signed values and yields
5038 <tt>true</tt> if <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
5041 <p>If the operands are <a href="#t_pointer">pointer</a> typed, the pointer
5042 values are compared as if they were integers.</p>
5044 <p>If the operands are integer vectors, then they are compared element by
5045 element. The result is an <tt>i1</tt> vector with the same number of elements
5046 as the values being compared. Otherwise, the result is an <tt>i1</tt>.</p>
5050 <result> = icmp eq i32 4, 5 <i>; yields: result=false</i>
5051 <result> = icmp ne float* %X, %X <i>; yields: result=false</i>
5052 <result> = icmp ult i16 4, 5 <i>; yields: result=true</i>
5053 <result> = icmp sgt i16 4, 5 <i>; yields: result=false</i>
5054 <result> = icmp ule i16 -4, 5 <i>; yields: result=false</i>
5055 <result> = icmp sge i16 4, 5 <i>; yields: result=false</i>
5058 <p>Note that the code generator does not yet support vector types with
5059 the <tt>icmp</tt> instruction.</p>
5063 <!-- _______________________________________________________________________ -->
5064 <div class="doc_subsubsection"><a name="i_fcmp">'<tt>fcmp</tt>' Instruction</a>
5067 <div class="doc_text">
5071 <result> = fcmp <cond> <ty> <op1>, <op2> <i>; yields {i1} or {<N x i1>}:result</i>
5075 <p>The '<tt>fcmp</tt>' instruction returns a boolean value or vector of boolean
5076 values based on comparison of its operands.</p>
5078 <p>If the operands are floating point scalars, then the result type is a boolean
5079 (<a href="#t_integer"><tt>i1</tt></a>).</p>
5081 <p>If the operands are floating point vectors, then the result type is a vector
5082 of boolean with the same number of elements as the operands being
5086 <p>The '<tt>fcmp</tt>' instruction takes three operands. The first operand is
5087 the condition code indicating the kind of comparison to perform. It is not a
5088 value, just a keyword. The possible condition code are:</p>
5091 <li><tt>false</tt>: no comparison, always returns false</li>
5092 <li><tt>oeq</tt>: ordered and equal</li>
5093 <li><tt>ogt</tt>: ordered and greater than </li>
5094 <li><tt>oge</tt>: ordered and greater than or equal</li>
5095 <li><tt>olt</tt>: ordered and less than </li>
5096 <li><tt>ole</tt>: ordered and less than or equal</li>
5097 <li><tt>one</tt>: ordered and not equal</li>
5098 <li><tt>ord</tt>: ordered (no nans)</li>
5099 <li><tt>ueq</tt>: unordered or equal</li>
5100 <li><tt>ugt</tt>: unordered or greater than </li>
5101 <li><tt>uge</tt>: unordered or greater than or equal</li>
5102 <li><tt>ult</tt>: unordered or less than </li>
5103 <li><tt>ule</tt>: unordered or less than or equal</li>
5104 <li><tt>une</tt>: unordered or not equal</li>
5105 <li><tt>uno</tt>: unordered (either nans)</li>
5106 <li><tt>true</tt>: no comparison, always returns true</li>
5109 <p><i>Ordered</i> means that neither operand is a QNAN while
5110 <i>unordered</i> means that either operand may be a QNAN.</p>
5112 <p>Each of <tt>val1</tt> and <tt>val2</tt> arguments must be either
5113 a <a href="#t_floating">floating point</a> type or
5114 a <a href="#t_vector">vector</a> of floating point type. They must have
5115 identical types.</p>
5118 <p>The '<tt>fcmp</tt>' instruction compares <tt>op1</tt> and <tt>op2</tt>
5119 according to the condition code given as <tt>cond</tt>. If the operands are
5120 vectors, then the vectors are compared element by element. Each comparison
5121 performed always yields an <a href="#t_integer">i1</a> result, as
5125 <li><tt>false</tt>: always yields <tt>false</tt>, regardless of operands.</li>
5127 <li><tt>oeq</tt>: yields <tt>true</tt> if both operands are not a QNAN and
5128 <tt>op1</tt> is equal to <tt>op2</tt>.</li>
5130 <li><tt>ogt</tt>: yields <tt>true</tt> if both operands are not a QNAN and
5131 <tt>op1</tt> is greater than <tt>op2</tt>.</li>
5133 <li><tt>oge</tt>: yields <tt>true</tt> if both operands are not a QNAN and
5134 <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
5136 <li><tt>olt</tt>: yields <tt>true</tt> if both operands are not a QNAN and
5137 <tt>op1</tt> is less than <tt>op2</tt>.</li>
5139 <li><tt>ole</tt>: yields <tt>true</tt> if both operands are not a QNAN and
5140 <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
5142 <li><tt>one</tt>: yields <tt>true</tt> if both operands are not a QNAN and
5143 <tt>op1</tt> is not equal to <tt>op2</tt>.</li>
5145 <li><tt>ord</tt>: yields <tt>true</tt> if both operands are not a QNAN.</li>
5147 <li><tt>ueq</tt>: yields <tt>true</tt> if either operand is a QNAN or
5148 <tt>op1</tt> is equal to <tt>op2</tt>.</li>
5150 <li><tt>ugt</tt>: yields <tt>true</tt> if either operand is a QNAN or
5151 <tt>op1</tt> is greater than <tt>op2</tt>.</li>
5153 <li><tt>uge</tt>: yields <tt>true</tt> if either operand is a QNAN or
5154 <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
5156 <li><tt>ult</tt>: yields <tt>true</tt> if either operand is a QNAN or
5157 <tt>op1</tt> is less than <tt>op2</tt>.</li>
5159 <li><tt>ule</tt>: yields <tt>true</tt> if either operand is a QNAN or
5160 <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
5162 <li><tt>une</tt>: yields <tt>true</tt> if either operand is a QNAN or
5163 <tt>op1</tt> is not equal to <tt>op2</tt>.</li>
5165 <li><tt>uno</tt>: yields <tt>true</tt> if either operand is a QNAN.</li>
5167 <li><tt>true</tt>: always yields <tt>true</tt>, regardless of operands.</li>
5172 <result> = fcmp oeq float 4.0, 5.0 <i>; yields: result=false</i>
5173 <result> = fcmp one float 4.0, 5.0 <i>; yields: result=true</i>
5174 <result> = fcmp olt float 4.0, 5.0 <i>; yields: result=true</i>
5175 <result> = fcmp ueq double 1.0, 2.0 <i>; yields: result=false</i>
5178 <p>Note that the code generator does not yet support vector types with
5179 the <tt>fcmp</tt> instruction.</p>
5183 <!-- _______________________________________________________________________ -->
5184 <div class="doc_subsubsection">
5185 <a name="i_phi">'<tt>phi</tt>' Instruction</a>
5188 <div class="doc_text">
5192 <result> = phi <ty> [ <val0>, <label0>], ...
5196 <p>The '<tt>phi</tt>' instruction is used to implement the φ node in the
5197 SSA graph representing the function.</p>
5200 <p>The type of the incoming values is specified with the first type field. After
5201 this, the '<tt>phi</tt>' instruction takes a list of pairs as arguments, with
5202 one pair for each predecessor basic block of the current block. Only values
5203 of <a href="#t_firstclass">first class</a> type may be used as the value
5204 arguments to the PHI node. Only labels may be used as the label
5207 <p>There must be no non-phi instructions between the start of a basic block and
5208 the PHI instructions: i.e. PHI instructions must be first in a basic
5211 <p>For the purposes of the SSA form, the use of each incoming value is deemed to
5212 occur on the edge from the corresponding predecessor block to the current
5213 block (but after any definition of an '<tt>invoke</tt>' instruction's return
5214 value on the same edge).</p>
5217 <p>At runtime, the '<tt>phi</tt>' instruction logically takes on the value
5218 specified by the pair corresponding to the predecessor basic block that
5219 executed just prior to the current block.</p>
5223 Loop: ; Infinite loop that counts from 0 on up...
5224 %indvar = phi i32 [ 0, %LoopHeader ], [ %nextindvar, %Loop ]
5225 %nextindvar = add i32 %indvar, 1
5231 <!-- _______________________________________________________________________ -->
5232 <div class="doc_subsubsection">
5233 <a name="i_select">'<tt>select</tt>' Instruction</a>
5236 <div class="doc_text">
5240 <result> = select <i>selty</i> <cond>, <ty> <val1>, <ty> <val2> <i>; yields ty</i>
5242 <i>selty</i> is either i1 or {<N x i1>}
5246 <p>The '<tt>select</tt>' instruction is used to choose one value based on a
5247 condition, without branching.</p>
5251 <p>The '<tt>select</tt>' instruction requires an 'i1' value or a vector of 'i1'
5252 values indicating the condition, and two values of the
5253 same <a href="#t_firstclass">first class</a> type. If the val1/val2 are
5254 vectors and the condition is a scalar, then entire vectors are selected, not
5255 individual elements.</p>
5258 <p>If the condition is an i1 and it evaluates to 1, the instruction returns the
5259 first value argument; otherwise, it returns the second value argument.</p>
5261 <p>If the condition is a vector of i1, then the value arguments must be vectors
5262 of the same size, and the selection is done element by element.</p>
5266 %X = select i1 true, i8 17, i8 42 <i>; yields i8:17</i>
5269 <p>Note that the code generator does not yet support conditions
5270 with vector type.</p>
5274 <!-- _______________________________________________________________________ -->
5275 <div class="doc_subsubsection">
5276 <a name="i_call">'<tt>call</tt>' Instruction</a>
5279 <div class="doc_text">
5283 <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>]
5287 <p>The '<tt>call</tt>' instruction represents a simple function call.</p>
5290 <p>This instruction requires several arguments:</p>
5293 <li>The optional "tail" marker indicates that the callee function does not
5294 access any allocas or varargs in the caller. Note that calls may be
5295 marked "tail" even if they do not occur before
5296 a <a href="#i_ret"><tt>ret</tt></a> instruction. If the "tail" marker is
5297 present, the function call is eligible for tail call optimization,
5298 but <a href="CodeGenerator.html#tailcallopt">might not in fact be
5299 optimized into a jump</a>. The code generator may optimize calls marked
5300 "tail" with either 1) automatic <a href="CodeGenerator.html#sibcallopt">
5301 sibling call optimization</a> when the caller and callee have
5302 matching signatures, or 2) forced tail call optimization when the
5303 following extra requirements are met:
5305 <li>Caller and callee both have the calling
5306 convention <tt>fastcc</tt>.</li>
5307 <li>The call is in tail position (ret immediately follows call and ret
5308 uses value of call or is void).</li>
5309 <li>Option <tt>-tailcallopt</tt> is enabled,
5310 or <code>llvm::GuaranteedTailCallOpt</code> is <code>true</code>.</li>
5311 <li><a href="CodeGenerator.html#tailcallopt">Platform specific
5312 constraints are met.</a></li>
5316 <li>The optional "cconv" marker indicates which <a href="#callingconv">calling
5317 convention</a> the call should use. If none is specified, the call
5318 defaults to using C calling conventions. The calling convention of the
5319 call must match the calling convention of the target function, or else the
5320 behavior is undefined.</li>
5322 <li>The optional <a href="#paramattrs">Parameter Attributes</a> list for
5323 return values. Only '<tt>zeroext</tt>', '<tt>signext</tt>', and
5324 '<tt>inreg</tt>' attributes are valid here.</li>
5326 <li>'<tt>ty</tt>': the type of the call instruction itself which is also the
5327 type of the return value. Functions that return no value are marked
5328 <tt><a href="#t_void">void</a></tt>.</li>
5330 <li>'<tt>fnty</tt>': shall be the signature of the pointer to function value
5331 being invoked. The argument types must match the types implied by this
5332 signature. This type can be omitted if the function is not varargs and if
5333 the function type does not return a pointer to a function.</li>
5335 <li>'<tt>fnptrval</tt>': An LLVM value containing a pointer to a function to
5336 be invoked. In most cases, this is a direct function invocation, but
5337 indirect <tt>call</tt>s are just as possible, calling an arbitrary pointer
5338 to function value.</li>
5340 <li>'<tt>function args</tt>': argument list whose types match the function
5341 signature argument types and parameter attributes. All arguments must be
5342 of <a href="#t_firstclass">first class</a> type. If the function
5343 signature indicates the function accepts a variable number of arguments,
5344 the extra arguments can be specified.</li>
5346 <li>The optional <a href="#fnattrs">function attributes</a> list. Only
5347 '<tt>noreturn</tt>', '<tt>nounwind</tt>', '<tt>readonly</tt>' and
5348 '<tt>readnone</tt>' attributes are valid here.</li>
5352 <p>The '<tt>call</tt>' instruction is used to cause control flow to transfer to
5353 a specified function, with its incoming arguments bound to the specified
5354 values. Upon a '<tt><a href="#i_ret">ret</a></tt>' instruction in the called
5355 function, control flow continues with the instruction after the function
5356 call, and the return value of the function is bound to the result
5361 %retval = call i32 @test(i32 %argc)
5362 call i32 (i8*, ...)* @printf(i8* %msg, i32 12, i8 42) <i>; yields i32</i>
5363 %X = tail call i32 @foo() <i>; yields i32</i>
5364 %Y = tail call <a href="#callingconv">fastcc</a> i32 @foo() <i>; yields i32</i>
5365 call void %foo(i8 97 signext)
5367 %struct.A = type { i32, i8 }
5368 %r = call %struct.A @foo() <i>; yields { 32, i8 }</i>
5369 %gr = extractvalue %struct.A %r, 0 <i>; yields i32</i>
5370 %gr1 = extractvalue %struct.A %r, 1 <i>; yields i8</i>
5371 %Z = call void @foo() noreturn <i>; indicates that %foo never returns normally</i>
5372 %ZZ = call zeroext i32 @bar() <i>; Return value is %zero extended</i>
5375 <p>llvm treats calls to some functions with names and arguments that match the
5376 standard C99 library as being the C99 library functions, and may perform
5377 optimizations or generate code for them under that assumption. This is
5378 something we'd like to change in the future to provide better support for
5379 freestanding environments and non-C-based languages.</p>
5383 <!-- _______________________________________________________________________ -->
5384 <div class="doc_subsubsection">
5385 <a name="i_va_arg">'<tt>va_arg</tt>' Instruction</a>
5388 <div class="doc_text">
5392 <resultval> = va_arg <va_list*> <arglist>, <argty>
5396 <p>The '<tt>va_arg</tt>' instruction is used to access arguments passed through
5397 the "variable argument" area of a function call. It is used to implement the
5398 <tt>va_arg</tt> macro in C.</p>
5401 <p>This instruction takes a <tt>va_list*</tt> value and the type of the
5402 argument. It returns a value of the specified argument type and increments
5403 the <tt>va_list</tt> to point to the next argument. The actual type
5404 of <tt>va_list</tt> is target specific.</p>
5407 <p>The '<tt>va_arg</tt>' instruction loads an argument of the specified type
5408 from the specified <tt>va_list</tt> and causes the <tt>va_list</tt> to point
5409 to the next argument. For more information, see the variable argument
5410 handling <a href="#int_varargs">Intrinsic Functions</a>.</p>
5412 <p>It is legal for this instruction to be called in a function which does not
5413 take a variable number of arguments, for example, the <tt>vfprintf</tt>
5416 <p><tt>va_arg</tt> is an LLVM instruction instead of
5417 an <a href="#intrinsics">intrinsic function</a> because it takes a type as an
5421 <p>See the <a href="#int_varargs">variable argument processing</a> section.</p>
5423 <p>Note that the code generator does not yet fully support va_arg on many
5424 targets. Also, it does not currently support va_arg with aggregate types on
5429 <!-- *********************************************************************** -->
5430 <div class="doc_section"> <a name="intrinsics">Intrinsic Functions</a> </div>
5431 <!-- *********************************************************************** -->
5433 <div class="doc_text">
5435 <p>LLVM supports the notion of an "intrinsic function". These functions have
5436 well known names and semantics and are required to follow certain
5437 restrictions. Overall, these intrinsics represent an extension mechanism for
5438 the LLVM language that does not require changing all of the transformations
5439 in LLVM when adding to the language (or the bitcode reader/writer, the
5440 parser, etc...).</p>
5442 <p>Intrinsic function names must all start with an "<tt>llvm.</tt>" prefix. This
5443 prefix is reserved in LLVM for intrinsic names; thus, function names may not
5444 begin with this prefix. Intrinsic functions must always be external
5445 functions: you cannot define the body of intrinsic functions. Intrinsic
5446 functions may only be used in call or invoke instructions: it is illegal to
5447 take the address of an intrinsic function. Additionally, because intrinsic
5448 functions are part of the LLVM language, it is required if any are added that
5449 they be documented here.</p>
5451 <p>Some intrinsic functions can be overloaded, i.e., the intrinsic represents a
5452 family of functions that perform the same operation but on different data
5453 types. Because LLVM can represent over 8 million different integer types,
5454 overloading is used commonly to allow an intrinsic function to operate on any
5455 integer type. One or more of the argument types or the result type can be
5456 overloaded to accept any integer type. Argument types may also be defined as
5457 exactly matching a previous argument's type or the result type. This allows
5458 an intrinsic function which accepts multiple arguments, but needs all of them
5459 to be of the same type, to only be overloaded with respect to a single
5460 argument or the result.</p>
5462 <p>Overloaded intrinsics will have the names of its overloaded argument types
5463 encoded into its function name, each preceded by a period. Only those types
5464 which are overloaded result in a name suffix. Arguments whose type is matched
5465 against another type do not. For example, the <tt>llvm.ctpop</tt> function
5466 can take an integer of any width and returns an integer of exactly the same
5467 integer width. This leads to a family of functions such as
5468 <tt>i8 @llvm.ctpop.i8(i8 %val)</tt> and <tt>i29 @llvm.ctpop.i29(i29
5469 %val)</tt>. Only one type, the return type, is overloaded, and only one type
5470 suffix is required. Because the argument's type is matched against the return
5471 type, it does not require its own name suffix.</p>
5473 <p>To learn how to add an intrinsic function, please see the
5474 <a href="ExtendingLLVM.html">Extending LLVM Guide</a>.</p>
5478 <!-- ======================================================================= -->
5479 <div class="doc_subsection">
5480 <a name="int_varargs">Variable Argument Handling Intrinsics</a>
5483 <div class="doc_text">
5485 <p>Variable argument support is defined in LLVM with
5486 the <a href="#i_va_arg"><tt>va_arg</tt></a> instruction and these three
5487 intrinsic functions. These functions are related to the similarly named
5488 macros defined in the <tt><stdarg.h></tt> header file.</p>
5490 <p>All of these functions operate on arguments that use a target-specific value
5491 type "<tt>va_list</tt>". The LLVM assembly language reference manual does
5492 not define what this type is, so all transformations should be prepared to
5493 handle these functions regardless of the type used.</p>
5495 <p>This example shows how the <a href="#i_va_arg"><tt>va_arg</tt></a>
5496 instruction and the variable argument handling intrinsic functions are
5499 <pre class="doc_code">
5500 define i32 @test(i32 %X, ...) {
5501 ; Initialize variable argument processing
5503 %ap2 = bitcast i8** %ap to i8*
5504 call void @llvm.va_start(i8* %ap2)
5506 ; Read a single integer argument
5507 %tmp = va_arg i8** %ap, i32
5509 ; Demonstrate usage of llvm.va_copy and llvm.va_end
5511 %aq2 = bitcast i8** %aq to i8*
5512 call void @llvm.va_copy(i8* %aq2, i8* %ap2)
5513 call void @llvm.va_end(i8* %aq2)
5515 ; Stop processing of arguments.
5516 call void @llvm.va_end(i8* %ap2)
5520 declare void @llvm.va_start(i8*)
5521 declare void @llvm.va_copy(i8*, i8*)
5522 declare void @llvm.va_end(i8*)
5527 <!-- _______________________________________________________________________ -->
5528 <div class="doc_subsubsection">
5529 <a name="int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a>
5533 <div class="doc_text">
5537 declare void %llvm.va_start(i8* <arglist>)
5541 <p>The '<tt>llvm.va_start</tt>' intrinsic initializes <tt>*<arglist></tt>
5542 for subsequent use by <tt><a href="#i_va_arg">va_arg</a></tt>.</p>
5545 <p>The argument is a pointer to a <tt>va_list</tt> element to initialize.</p>
5548 <p>The '<tt>llvm.va_start</tt>' intrinsic works just like the <tt>va_start</tt>
5549 macro available in C. In a target-dependent way, it initializes
5550 the <tt>va_list</tt> element to which the argument points, so that the next
5551 call to <tt>va_arg</tt> will produce the first variable argument passed to
5552 the function. Unlike the C <tt>va_start</tt> macro, this intrinsic does not
5553 need to know the last argument of the function as the compiler can figure
5558 <!-- _______________________________________________________________________ -->
5559 <div class="doc_subsubsection">
5560 <a name="int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a>
5563 <div class="doc_text">
5567 declare void @llvm.va_end(i8* <arglist>)
5571 <p>The '<tt>llvm.va_end</tt>' intrinsic destroys <tt>*<arglist></tt>,
5572 which has been initialized previously
5573 with <tt><a href="#int_va_start">llvm.va_start</a></tt>
5574 or <tt><a href="#i_va_copy">llvm.va_copy</a></tt>.</p>
5577 <p>The argument is a pointer to a <tt>va_list</tt> to destroy.</p>
5580 <p>The '<tt>llvm.va_end</tt>' intrinsic works just like the <tt>va_end</tt>
5581 macro available in C. In a target-dependent way, it destroys
5582 the <tt>va_list</tt> element to which the argument points. Calls
5583 to <a href="#int_va_start"><tt>llvm.va_start</tt></a>
5584 and <a href="#int_va_copy"> <tt>llvm.va_copy</tt></a> must be matched exactly
5585 with calls to <tt>llvm.va_end</tt>.</p>
5589 <!-- _______________________________________________________________________ -->
5590 <div class="doc_subsubsection">
5591 <a name="int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a>
5594 <div class="doc_text">
5598 declare void @llvm.va_copy(i8* <destarglist>, i8* <srcarglist>)
5602 <p>The '<tt>llvm.va_copy</tt>' intrinsic copies the current argument position
5603 from the source argument list to the destination argument list.</p>
5606 <p>The first argument is a pointer to a <tt>va_list</tt> element to initialize.
5607 The second argument is a pointer to a <tt>va_list</tt> element to copy
5611 <p>The '<tt>llvm.va_copy</tt>' intrinsic works just like the <tt>va_copy</tt>
5612 macro available in C. In a target-dependent way, it copies the
5613 source <tt>va_list</tt> element into the destination <tt>va_list</tt>
5614 element. This intrinsic is necessary because
5615 the <tt><a href="#int_va_start"> llvm.va_start</a></tt> intrinsic may be
5616 arbitrarily complex and require, for example, memory allocation.</p>
5620 <!-- ======================================================================= -->
5621 <div class="doc_subsection">
5622 <a name="int_gc">Accurate Garbage Collection Intrinsics</a>
5625 <div class="doc_text">
5627 <p>LLVM support for <a href="GarbageCollection.html">Accurate Garbage
5628 Collection</a> (GC) requires the implementation and generation of these
5629 intrinsics. These intrinsics allow identification of <a href="#int_gcroot">GC
5630 roots on the stack</a>, as well as garbage collector implementations that
5631 require <a href="#int_gcread">read</a> and <a href="#int_gcwrite">write</a>
5632 barriers. Front-ends for type-safe garbage collected languages should generate
5633 these intrinsics to make use of the LLVM garbage collectors. For more details,
5634 see <a href="GarbageCollection.html">Accurate Garbage Collection with
5637 <p>The garbage collection intrinsics only operate on objects in the generic
5638 address space (address space zero).</p>
5642 <!-- _______________________________________________________________________ -->
5643 <div class="doc_subsubsection">
5644 <a name="int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a>
5647 <div class="doc_text">
5651 declare void @llvm.gcroot(i8** %ptrloc, i8* %metadata)
5655 <p>The '<tt>llvm.gcroot</tt>' intrinsic declares the existence of a GC root to
5656 the code generator, and allows some metadata to be associated with it.</p>
5659 <p>The first argument specifies the address of a stack object that contains the
5660 root pointer. The second pointer (which must be either a constant or a
5661 global value address) contains the meta-data to be associated with the
5665 <p>At runtime, a call to this intrinsic stores a null pointer into the "ptrloc"
5666 location. At compile-time, the code generator generates information to allow
5667 the runtime to find the pointer at GC safe points. The '<tt>llvm.gcroot</tt>'
5668 intrinsic may only be used in a function which <a href="#gc">specifies a GC
5673 <!-- _______________________________________________________________________ -->
5674 <div class="doc_subsubsection">
5675 <a name="int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a>
5678 <div class="doc_text">
5682 declare i8* @llvm.gcread(i8* %ObjPtr, i8** %Ptr)
5686 <p>The '<tt>llvm.gcread</tt>' intrinsic identifies reads of references from heap
5687 locations, allowing garbage collector implementations that require read
5691 <p>The second argument is the address to read from, which should be an address
5692 allocated from the garbage collector. The first object is a pointer to the
5693 start of the referenced object, if needed by the language runtime (otherwise
5697 <p>The '<tt>llvm.gcread</tt>' intrinsic has the same semantics as a load
5698 instruction, but may be replaced with substantially more complex code by the
5699 garbage collector runtime, as needed. The '<tt>llvm.gcread</tt>' intrinsic
5700 may only be used in a function which <a href="#gc">specifies a GC
5705 <!-- _______________________________________________________________________ -->
5706 <div class="doc_subsubsection">
5707 <a name="int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a>
5710 <div class="doc_text">
5714 declare void @llvm.gcwrite(i8* %P1, i8* %Obj, i8** %P2)
5718 <p>The '<tt>llvm.gcwrite</tt>' intrinsic identifies writes of references to heap
5719 locations, allowing garbage collector implementations that require write
5720 barriers (such as generational or reference counting collectors).</p>
5723 <p>The first argument is the reference to store, the second is the start of the
5724 object to store it to, and the third is the address of the field of Obj to
5725 store to. If the runtime does not require a pointer to the object, Obj may
5729 <p>The '<tt>llvm.gcwrite</tt>' intrinsic has the same semantics as a store
5730 instruction, but may be replaced with substantially more complex code by the
5731 garbage collector runtime, as needed. The '<tt>llvm.gcwrite</tt>' intrinsic
5732 may only be used in a function which <a href="#gc">specifies a GC
5737 <!-- ======================================================================= -->
5738 <div class="doc_subsection">
5739 <a name="int_codegen">Code Generator Intrinsics</a>
5742 <div class="doc_text">
5744 <p>These intrinsics are provided by LLVM to expose special features that may
5745 only be implemented with code generator support.</p>
5749 <!-- _______________________________________________________________________ -->
5750 <div class="doc_subsubsection">
5751 <a name="int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a>
5754 <div class="doc_text">
5758 declare i8 *@llvm.returnaddress(i32 <level>)
5762 <p>The '<tt>llvm.returnaddress</tt>' intrinsic attempts to compute a
5763 target-specific value indicating the return address of the current function
5764 or one of its callers.</p>
5767 <p>The argument to this intrinsic indicates which function to return the address
5768 for. Zero indicates the calling function, one indicates its caller, etc.
5769 The argument is <b>required</b> to be a constant integer value.</p>
5772 <p>The '<tt>llvm.returnaddress</tt>' intrinsic either returns a pointer
5773 indicating the return address of the specified call frame, or zero if it
5774 cannot be identified. The value returned by this intrinsic is likely to be
5775 incorrect or 0 for arguments other than zero, so it should only be used for
5776 debugging purposes.</p>
5778 <p>Note that calling this intrinsic does not prevent function inlining or other
5779 aggressive transformations, so the value returned may not be that of the
5780 obvious source-language caller.</p>
5784 <!-- _______________________________________________________________________ -->
5785 <div class="doc_subsubsection">
5786 <a name="int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a>
5789 <div class="doc_text">
5793 declare i8* @llvm.frameaddress(i32 <level>)
5797 <p>The '<tt>llvm.frameaddress</tt>' intrinsic attempts to return the
5798 target-specific frame pointer value for the specified stack frame.</p>
5801 <p>The argument to this intrinsic indicates which function to return the frame
5802 pointer for. Zero indicates the calling function, one indicates its caller,
5803 etc. The argument is <b>required</b> to be a constant integer value.</p>
5806 <p>The '<tt>llvm.frameaddress</tt>' intrinsic either returns a pointer
5807 indicating the frame address of the specified call frame, or zero if it
5808 cannot be identified. The value returned by this intrinsic is likely to be
5809 incorrect or 0 for arguments other than zero, so it should only be used for
5810 debugging purposes.</p>
5812 <p>Note that calling this intrinsic does not prevent function inlining or other
5813 aggressive transformations, so the value returned may not be that of the
5814 obvious source-language caller.</p>
5818 <!-- _______________________________________________________________________ -->
5819 <div class="doc_subsubsection">
5820 <a name="int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a>
5823 <div class="doc_text">
5827 declare i8* @llvm.stacksave()
5831 <p>The '<tt>llvm.stacksave</tt>' intrinsic is used to remember the current state
5832 of the function stack, for use
5833 with <a href="#int_stackrestore"> <tt>llvm.stackrestore</tt></a>. This is
5834 useful for implementing language features like scoped automatic variable
5835 sized arrays in C99.</p>
5838 <p>This intrinsic returns a opaque pointer value that can be passed
5839 to <a href="#int_stackrestore"><tt>llvm.stackrestore</tt></a>. When
5840 an <tt>llvm.stackrestore</tt> intrinsic is executed with a value saved
5841 from <tt>llvm.stacksave</tt>, it effectively restores the state of the stack
5842 to the state it was in when the <tt>llvm.stacksave</tt> intrinsic executed.
5843 In practice, this pops any <a href="#i_alloca">alloca</a> blocks from the
5844 stack that were allocated after the <tt>llvm.stacksave</tt> was executed.</p>
5848 <!-- _______________________________________________________________________ -->
5849 <div class="doc_subsubsection">
5850 <a name="int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a>
5853 <div class="doc_text">
5857 declare void @llvm.stackrestore(i8* %ptr)
5861 <p>The '<tt>llvm.stackrestore</tt>' intrinsic is used to restore the state of
5862 the function stack to the state it was in when the
5863 corresponding <a href="#int_stacksave"><tt>llvm.stacksave</tt></a> intrinsic
5864 executed. This is useful for implementing language features like scoped
5865 automatic variable sized arrays in C99.</p>
5868 <p>See the description
5869 for <a href="#int_stacksave"><tt>llvm.stacksave</tt></a>.</p>
5873 <!-- _______________________________________________________________________ -->
5874 <div class="doc_subsubsection">
5875 <a name="int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a>
5878 <div class="doc_text">
5882 declare void @llvm.prefetch(i8* <address>, i32 <rw>, i32 <locality>)
5886 <p>The '<tt>llvm.prefetch</tt>' intrinsic is a hint to the code generator to
5887 insert a prefetch instruction if supported; otherwise, it is a noop.
5888 Prefetches have no effect on the behavior of the program but can change its
5889 performance characteristics.</p>
5892 <p><tt>address</tt> is the address to be prefetched, <tt>rw</tt> is the
5893 specifier determining if the fetch should be for a read (0) or write (1),
5894 and <tt>locality</tt> is a temporal locality specifier ranging from (0) - no
5895 locality, to (3) - extremely local keep in cache. The <tt>rw</tt>
5896 and <tt>locality</tt> arguments must be constant integers.</p>
5899 <p>This intrinsic does not modify the behavior of the program. In particular,
5900 prefetches cannot trap and do not produce a value. On targets that support
5901 this intrinsic, the prefetch can provide hints to the processor cache for
5902 better performance.</p>
5906 <!-- _______________________________________________________________________ -->
5907 <div class="doc_subsubsection">
5908 <a name="int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a>
5911 <div class="doc_text">
5915 declare void @llvm.pcmarker(i32 <id>)
5919 <p>The '<tt>llvm.pcmarker</tt>' intrinsic is a method to export a Program
5920 Counter (PC) in a region of code to simulators and other tools. The method
5921 is target specific, but it is expected that the marker will use exported
5922 symbols to transmit the PC of the marker. The marker makes no guarantees
5923 that it will remain with any specific instruction after optimizations. It is
5924 possible that the presence of a marker will inhibit optimizations. The
5925 intended use is to be inserted after optimizations to allow correlations of
5926 simulation runs.</p>
5929 <p><tt>id</tt> is a numerical id identifying the marker.</p>
5932 <p>This intrinsic does not modify the behavior of the program. Backends that do
5933 not support this intrinsic may ignore it.</p>
5937 <!-- _______________________________________________________________________ -->
5938 <div class="doc_subsubsection">
5939 <a name="int_readcyclecounter">'<tt>llvm.readcyclecounter</tt>' Intrinsic</a>
5942 <div class="doc_text">
5946 declare i64 @llvm.readcyclecounter()
5950 <p>The '<tt>llvm.readcyclecounter</tt>' intrinsic provides access to the cycle
5951 counter register (or similar low latency, high accuracy clocks) on those
5952 targets that support it. On X86, it should map to RDTSC. On Alpha, it
5953 should map to RPCC. As the backing counters overflow quickly (on the order
5954 of 9 seconds on alpha), this should only be used for small timings.</p>
5957 <p>When directly supported, reading the cycle counter should not modify any
5958 memory. Implementations are allowed to either return a application specific
5959 value or a system wide value. On backends without support, this is lowered
5960 to a constant 0.</p>
5964 <!-- ======================================================================= -->
5965 <div class="doc_subsection">
5966 <a name="int_libc">Standard C Library Intrinsics</a>
5969 <div class="doc_text">
5971 <p>LLVM provides intrinsics for a few important standard C library functions.
5972 These intrinsics allow source-language front-ends to pass information about
5973 the alignment of the pointer arguments to the code generator, providing
5974 opportunity for more efficient code generation.</p>
5978 <!-- _______________________________________________________________________ -->
5979 <div class="doc_subsubsection">
5980 <a name="int_memcpy">'<tt>llvm.memcpy</tt>' Intrinsic</a>
5983 <div class="doc_text">
5986 <p>This is an overloaded intrinsic. You can use <tt>llvm.memcpy</tt> on any
5987 integer bit width and for different address spaces. Not all targets support
5988 all bit widths however.</p>
5991 declare void @llvm.memcpy.p0i8.p0i8.i32(i8* <dest>, i8* <src>,
5992 i32 <len>, i32 <align>, i1 <isvolatile>)
5993 declare void @llvm.memcpy.p0i8.p0i8.i64(i8* <dest>, i8* <src>,
5994 i64 <len>, i32 <align>, i1 <isvolatile>)
5998 <p>The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the
5999 source location to the destination location.</p>
6001 <p>Note that, unlike the standard libc function, the <tt>llvm.memcpy.*</tt>
6002 intrinsics do not return a value, takes extra alignment/isvolatile arguments
6003 and the pointers can be in specified address spaces.</p>
6007 <p>The first argument is a pointer to the destination, the second is a pointer
6008 to the source. The third argument is an integer argument specifying the
6009 number of bytes to copy, the fourth argument is the alignment of the
6010 source and destination locations, and the fifth is a boolean indicating a
6011 volatile access.</p>
6013 <p>If the call to this intrinsic has an alignment value that is not 0 or 1,
6014 then the caller guarantees that both the source and destination pointers are
6015 aligned to that boundary.</p>
6017 <p>If the <tt>isvolatile</tt> parameter is <tt>true</tt>, the
6018 <tt>llvm.memcpy</tt> call is a <a href="#volatile">volatile operation</a>.
6019 The detailed access behavior is not very cleanly specified and it is unwise
6020 to depend on it.</p>
6024 <p>The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the
6025 source location to the destination location, which are not allowed to
6026 overlap. It copies "len" bytes of memory over. If the argument is known to
6027 be aligned to some boundary, this can be specified as the fourth argument,
6028 otherwise it should be set to 0 or 1.</p>
6032 <!-- _______________________________________________________________________ -->
6033 <div class="doc_subsubsection">
6034 <a name="int_memmove">'<tt>llvm.memmove</tt>' Intrinsic</a>
6037 <div class="doc_text">
6040 <p>This is an overloaded intrinsic. You can use llvm.memmove on any integer bit
6041 width and for different address space. Not all targets support all bit
6045 declare void @llvm.memmove.p0i8.p0i8.i32(i8* <dest>, i8* <src>,
6046 i32 <len>, i32 <align>, i1 <isvolatile>)
6047 declare void @llvm.memmove.p0i8.p0i8.i64(i8* <dest>, i8* <src>,
6048 i64 <len>, i32 <align>, i1 <isvolatile>)
6052 <p>The '<tt>llvm.memmove.*</tt>' intrinsics move a block of memory from the
6053 source location to the destination location. It is similar to the
6054 '<tt>llvm.memcpy</tt>' intrinsic but allows the two memory locations to
6057 <p>Note that, unlike the standard libc function, the <tt>llvm.memmove.*</tt>
6058 intrinsics do not return a value, takes extra alignment/isvolatile arguments
6059 and the pointers can be in specified address spaces.</p>
6063 <p>The first argument is a pointer to the destination, the second is a pointer
6064 to the source. The third argument is an integer argument specifying the
6065 number of bytes to copy, the fourth argument is the alignment of the
6066 source and destination locations, and the fifth is a boolean indicating a
6067 volatile access.</p>
6069 <p>If the call to this intrinsic has an alignment value that is not 0 or 1,
6070 then the caller guarantees that the source and destination pointers are
6071 aligned to that boundary.</p>
6073 <p>If the <tt>isvolatile</tt> parameter is <tt>true</tt>, the
6074 <tt>llvm.memmove</tt> call is a <a href="#volatile">volatile operation</a>.
6075 The detailed access behavior is not very cleanly specified and it is unwise
6076 to depend on it.</p>
6080 <p>The '<tt>llvm.memmove.*</tt>' intrinsics copy a block of memory from the
6081 source location to the destination location, which may overlap. It copies
6082 "len" bytes of memory over. If the argument is known to be aligned to some
6083 boundary, this can be specified as the fourth argument, otherwise it should
6084 be set to 0 or 1.</p>
6088 <!-- _______________________________________________________________________ -->
6089 <div class="doc_subsubsection">
6090 <a name="int_memset">'<tt>llvm.memset.*</tt>' Intrinsics</a>
6093 <div class="doc_text">
6096 <p>This is an overloaded intrinsic. You can use llvm.memset on any integer bit
6097 width and for different address spaces. However, not all targets support all
6101 declare void @llvm.memset.p0i8.i32(i8* <dest>, i8 <val>,
6102 i32 <len>, i32 <align>, i1 <isvolatile>)
6103 declare void @llvm.memset.p0i8.i64(i8* <dest>, i8 <val>,
6104 i64 <len>, i32 <align>, i1 <isvolatile>)
6108 <p>The '<tt>llvm.memset.*</tt>' intrinsics fill a block of memory with a
6109 particular byte value.</p>
6111 <p>Note that, unlike the standard libc function, the <tt>llvm.memset</tt>
6112 intrinsic does not return a value and takes extra alignment/volatile
6113 arguments. Also, the destination can be in an arbitrary address space.</p>
6116 <p>The first argument is a pointer to the destination to fill, the second is the
6117 byte value with which to fill it, the third argument is an integer argument
6118 specifying the number of bytes to fill, and the fourth argument is the known
6119 alignment of the destination location.</p>
6121 <p>If the call to this intrinsic has an alignment value that is not 0 or 1,
6122 then the caller guarantees that the destination pointer is aligned to that
6125 <p>If the <tt>isvolatile</tt> parameter is <tt>true</tt>, the
6126 <tt>llvm.memset</tt> call is a <a href="#volatile">volatile operation</a>.
6127 The detailed access behavior is not very cleanly specified and it is unwise
6128 to depend on it.</p>
6131 <p>The '<tt>llvm.memset.*</tt>' intrinsics fill "len" bytes of memory starting
6132 at the destination location. If the argument is known to be aligned to some
6133 boundary, this can be specified as the fourth argument, otherwise it should
6134 be set to 0 or 1.</p>
6138 <!-- _______________________________________________________________________ -->
6139 <div class="doc_subsubsection">
6140 <a name="int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a>
6143 <div class="doc_text">
6146 <p>This is an overloaded intrinsic. You can use <tt>llvm.sqrt</tt> on any
6147 floating point or vector of floating point type. Not all targets support all
6151 declare float @llvm.sqrt.f32(float %Val)
6152 declare double @llvm.sqrt.f64(double %Val)
6153 declare x86_fp80 @llvm.sqrt.f80(x86_fp80 %Val)
6154 declare fp128 @llvm.sqrt.f128(fp128 %Val)
6155 declare ppc_fp128 @llvm.sqrt.ppcf128(ppc_fp128 %Val)
6159 <p>The '<tt>llvm.sqrt</tt>' intrinsics return the sqrt of the specified operand,
6160 returning the same value as the libm '<tt>sqrt</tt>' functions would.
6161 Unlike <tt>sqrt</tt> in libm, however, <tt>llvm.sqrt</tt> has undefined
6162 behavior for negative numbers other than -0.0 (which allows for better
6163 optimization, because there is no need to worry about errno being
6164 set). <tt>llvm.sqrt(-0.0)</tt> is defined to return -0.0 like IEEE sqrt.</p>
6167 <p>The argument and return value are floating point numbers of the same
6171 <p>This function returns the sqrt of the specified operand if it is a
6172 nonnegative floating point number.</p>
6176 <!-- _______________________________________________________________________ -->
6177 <div class="doc_subsubsection">
6178 <a name="int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a>
6181 <div class="doc_text">
6184 <p>This is an overloaded intrinsic. You can use <tt>llvm.powi</tt> on any
6185 floating point or vector of floating point type. Not all targets support all
6189 declare float @llvm.powi.f32(float %Val, i32 %power)
6190 declare double @llvm.powi.f64(double %Val, i32 %power)
6191 declare x86_fp80 @llvm.powi.f80(x86_fp80 %Val, i32 %power)
6192 declare fp128 @llvm.powi.f128(fp128 %Val, i32 %power)
6193 declare ppc_fp128 @llvm.powi.ppcf128(ppc_fp128 %Val, i32 %power)
6197 <p>The '<tt>llvm.powi.*</tt>' intrinsics return the first operand raised to the
6198 specified (positive or negative) power. The order of evaluation of
6199 multiplications is not defined. When a vector of floating point type is
6200 used, the second argument remains a scalar integer value.</p>
6203 <p>The second argument is an integer power, and the first is a value to raise to
6207 <p>This function returns the first value raised to the second power with an
6208 unspecified sequence of rounding operations.</p>
6212 <!-- _______________________________________________________________________ -->
6213 <div class="doc_subsubsection">
6214 <a name="int_sin">'<tt>llvm.sin.*</tt>' Intrinsic</a>
6217 <div class="doc_text">
6220 <p>This is an overloaded intrinsic. You can use <tt>llvm.sin</tt> on any
6221 floating point or vector of floating point type. Not all targets support all
6225 declare float @llvm.sin.f32(float %Val)
6226 declare double @llvm.sin.f64(double %Val)
6227 declare x86_fp80 @llvm.sin.f80(x86_fp80 %Val)
6228 declare fp128 @llvm.sin.f128(fp128 %Val)
6229 declare ppc_fp128 @llvm.sin.ppcf128(ppc_fp128 %Val)
6233 <p>The '<tt>llvm.sin.*</tt>' intrinsics return the sine of the operand.</p>
6236 <p>The argument and return value are floating point numbers of the same
6240 <p>This function returns the sine of the specified operand, returning the same
6241 values as the libm <tt>sin</tt> functions would, and handles error conditions
6242 in the same way.</p>
6246 <!-- _______________________________________________________________________ -->
6247 <div class="doc_subsubsection">
6248 <a name="int_cos">'<tt>llvm.cos.*</tt>' Intrinsic</a>
6251 <div class="doc_text">
6254 <p>This is an overloaded intrinsic. You can use <tt>llvm.cos</tt> on any
6255 floating point or vector of floating point type. Not all targets support all
6259 declare float @llvm.cos.f32(float %Val)
6260 declare double @llvm.cos.f64(double %Val)
6261 declare x86_fp80 @llvm.cos.f80(x86_fp80 %Val)
6262 declare fp128 @llvm.cos.f128(fp128 %Val)
6263 declare ppc_fp128 @llvm.cos.ppcf128(ppc_fp128 %Val)
6267 <p>The '<tt>llvm.cos.*</tt>' intrinsics return the cosine of the operand.</p>
6270 <p>The argument and return value are floating point numbers of the same
6274 <p>This function returns the cosine of the specified operand, returning the same
6275 values as the libm <tt>cos</tt> functions would, and handles error conditions
6276 in the same way.</p>
6280 <!-- _______________________________________________________________________ -->
6281 <div class="doc_subsubsection">
6282 <a name="int_pow">'<tt>llvm.pow.*</tt>' Intrinsic</a>
6285 <div class="doc_text">
6288 <p>This is an overloaded intrinsic. You can use <tt>llvm.pow</tt> on any
6289 floating point or vector of floating point type. Not all targets support all
6293 declare float @llvm.pow.f32(float %Val, float %Power)
6294 declare double @llvm.pow.f64(double %Val, double %Power)
6295 declare x86_fp80 @llvm.pow.f80(x86_fp80 %Val, x86_fp80 %Power)
6296 declare fp128 @llvm.pow.f128(fp128 %Val, fp128 %Power)
6297 declare ppc_fp128 @llvm.pow.ppcf128(ppc_fp128 %Val, ppc_fp128 Power)
6301 <p>The '<tt>llvm.pow.*</tt>' intrinsics return the first operand raised to the
6302 specified (positive or negative) power.</p>
6305 <p>The second argument is a floating point power, and the first is a value to
6306 raise to that power.</p>
6309 <p>This function returns the first value raised to the second power, returning
6310 the same values as the libm <tt>pow</tt> functions would, and handles error
6311 conditions in the same way.</p>
6315 <!-- ======================================================================= -->
6316 <div class="doc_subsection">
6317 <a name="int_manip">Bit Manipulation Intrinsics</a>
6320 <div class="doc_text">
6322 <p>LLVM provides intrinsics for a few important bit manipulation operations.
6323 These allow efficient code generation for some algorithms.</p>
6327 <!-- _______________________________________________________________________ -->
6328 <div class="doc_subsubsection">
6329 <a name="int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a>
6332 <div class="doc_text">
6335 <p>This is an overloaded intrinsic function. You can use bswap on any integer
6336 type that is an even number of bytes (i.e. BitWidth % 16 == 0).</p>
6339 declare i16 @llvm.bswap.i16(i16 <id>)
6340 declare i32 @llvm.bswap.i32(i32 <id>)
6341 declare i64 @llvm.bswap.i64(i64 <id>)
6345 <p>The '<tt>llvm.bswap</tt>' family of intrinsics is used to byte swap integer
6346 values with an even number of bytes (positive multiple of 16 bits). These
6347 are useful for performing operations on data that is not in the target's
6348 native byte order.</p>
6351 <p>The <tt>llvm.bswap.i16</tt> intrinsic returns an i16 value that has the high
6352 and low byte of the input i16 swapped. Similarly,
6353 the <tt>llvm.bswap.i32</tt> intrinsic returns an i32 value that has the four
6354 bytes of the input i32 swapped, so that if the input bytes are numbered 0, 1,
6355 2, 3 then the returned i32 will have its bytes in 3, 2, 1, 0 order.
6356 The <tt>llvm.bswap.i48</tt>, <tt>llvm.bswap.i64</tt> and other intrinsics
6357 extend this concept to additional even-byte lengths (6 bytes, 8 bytes and
6358 more, respectively).</p>
6362 <!-- _______________________________________________________________________ -->
6363 <div class="doc_subsubsection">
6364 <a name="int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic</a>
6367 <div class="doc_text">
6370 <p>This is an overloaded intrinsic. You can use llvm.ctpop on any integer bit
6371 width. Not all targets support all bit widths however.</p>
6374 declare i8 @llvm.ctpop.i8(i8 <src>)
6375 declare i16 @llvm.ctpop.i16(i16 <src>)
6376 declare i32 @llvm.ctpop.i32(i32 <src>)
6377 declare i64 @llvm.ctpop.i64(i64 <src>)
6378 declare i256 @llvm.ctpop.i256(i256 <src>)
6382 <p>The '<tt>llvm.ctpop</tt>' family of intrinsics counts the number of bits set
6386 <p>The only argument is the value to be counted. The argument may be of any
6387 integer type. The return type must match the argument type.</p>
6390 <p>The '<tt>llvm.ctpop</tt>' intrinsic counts the 1's in a variable.</p>
6394 <!-- _______________________________________________________________________ -->
6395 <div class="doc_subsubsection">
6396 <a name="int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic</a>
6399 <div class="doc_text">
6402 <p>This is an overloaded intrinsic. You can use <tt>llvm.ctlz</tt> on any
6403 integer bit width. Not all targets support all bit widths however.</p>
6406 declare i8 @llvm.ctlz.i8 (i8 <src>)
6407 declare i16 @llvm.ctlz.i16(i16 <src>)
6408 declare i32 @llvm.ctlz.i32(i32 <src>)
6409 declare i64 @llvm.ctlz.i64(i64 <src>)
6410 declare i256 @llvm.ctlz.i256(i256 <src>)
6414 <p>The '<tt>llvm.ctlz</tt>' family of intrinsic functions counts the number of
6415 leading zeros in a variable.</p>
6418 <p>The only argument is the value to be counted. The argument may be of any
6419 integer type. The return type must match the argument type.</p>
6422 <p>The '<tt>llvm.ctlz</tt>' intrinsic counts the leading (most significant)
6423 zeros in a variable. If the src == 0 then the result is the size in bits of
6424 the type of src. For example, <tt>llvm.ctlz(i32 2) = 30</tt>.</p>
6428 <!-- _______________________________________________________________________ -->
6429 <div class="doc_subsubsection">
6430 <a name="int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic</a>
6433 <div class="doc_text">
6436 <p>This is an overloaded intrinsic. You can use <tt>llvm.cttz</tt> on any
6437 integer bit width. Not all targets support all bit widths however.</p>
6440 declare i8 @llvm.cttz.i8 (i8 <src>)
6441 declare i16 @llvm.cttz.i16(i16 <src>)
6442 declare i32 @llvm.cttz.i32(i32 <src>)
6443 declare i64 @llvm.cttz.i64(i64 <src>)
6444 declare i256 @llvm.cttz.i256(i256 <src>)
6448 <p>The '<tt>llvm.cttz</tt>' family of intrinsic functions counts the number of
6452 <p>The only argument is the value to be counted. The argument may be of any
6453 integer type. The return type must match the argument type.</p>
6456 <p>The '<tt>llvm.cttz</tt>' intrinsic counts the trailing (least significant)
6457 zeros in a variable. If the src == 0 then the result is the size in bits of
6458 the type of src. For example, <tt>llvm.cttz(2) = 1</tt>.</p>
6462 <!-- ======================================================================= -->
6463 <div class="doc_subsection">
6464 <a name="int_overflow">Arithmetic with Overflow Intrinsics</a>
6467 <div class="doc_text">
6469 <p>LLVM provides intrinsics for some arithmetic with overflow operations.</p>
6473 <!-- _______________________________________________________________________ -->
6474 <div class="doc_subsubsection">
6475 <a name="int_sadd_overflow">'<tt>llvm.sadd.with.overflow.*</tt>' Intrinsics</a>
6478 <div class="doc_text">
6481 <p>This is an overloaded intrinsic. You can use <tt>llvm.sadd.with.overflow</tt>
6482 on any integer bit width.</p>
6485 declare {i16, i1} @llvm.sadd.with.overflow.i16(i16 %a, i16 %b)
6486 declare {i32, i1} @llvm.sadd.with.overflow.i32(i32 %a, i32 %b)
6487 declare {i64, i1} @llvm.sadd.with.overflow.i64(i64 %a, i64 %b)
6491 <p>The '<tt>llvm.sadd.with.overflow</tt>' family of intrinsic functions perform
6492 a signed addition of the two arguments, and indicate whether an overflow
6493 occurred during the signed summation.</p>
6496 <p>The arguments (%a and %b) and the first element of the result structure may
6497 be of integer types of any bit width, but they must have the same bit
6498 width. The second element of the result structure must be of
6499 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
6500 undergo signed addition.</p>
6503 <p>The '<tt>llvm.sadd.with.overflow</tt>' family of intrinsic functions perform
6504 a signed addition of the two variables. They return a structure — the
6505 first element of which is the signed summation, and the second element of
6506 which is a bit specifying if the signed summation resulted in an
6511 %res = call {i32, i1} @llvm.sadd.with.overflow.i32(i32 %a, i32 %b)
6512 %sum = extractvalue {i32, i1} %res, 0
6513 %obit = extractvalue {i32, i1} %res, 1
6514 br i1 %obit, label %overflow, label %normal
6519 <!-- _______________________________________________________________________ -->
6520 <div class="doc_subsubsection">
6521 <a name="int_uadd_overflow">'<tt>llvm.uadd.with.overflow.*</tt>' Intrinsics</a>
6524 <div class="doc_text">
6527 <p>This is an overloaded intrinsic. You can use <tt>llvm.uadd.with.overflow</tt>
6528 on any integer bit width.</p>
6531 declare {i16, i1} @llvm.uadd.with.overflow.i16(i16 %a, i16 %b)
6532 declare {i32, i1} @llvm.uadd.with.overflow.i32(i32 %a, i32 %b)
6533 declare {i64, i1} @llvm.uadd.with.overflow.i64(i64 %a, i64 %b)
6537 <p>The '<tt>llvm.uadd.with.overflow</tt>' family of intrinsic functions perform
6538 an unsigned addition of the two arguments, and indicate whether a carry
6539 occurred during the unsigned summation.</p>
6542 <p>The arguments (%a and %b) and the first element of the result structure may
6543 be of integer types of any bit width, but they must have the same bit
6544 width. The second element of the result structure must be of
6545 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
6546 undergo unsigned addition.</p>
6549 <p>The '<tt>llvm.uadd.with.overflow</tt>' family of intrinsic functions perform
6550 an unsigned addition of the two arguments. They return a structure —
6551 the first element of which is the sum, and the second element of which is a
6552 bit specifying if the unsigned summation resulted in a carry.</p>
6556 %res = call {i32, i1} @llvm.uadd.with.overflow.i32(i32 %a, i32 %b)
6557 %sum = extractvalue {i32, i1} %res, 0
6558 %obit = extractvalue {i32, i1} %res, 1
6559 br i1 %obit, label %carry, label %normal
6564 <!-- _______________________________________________________________________ -->
6565 <div class="doc_subsubsection">
6566 <a name="int_ssub_overflow">'<tt>llvm.ssub.with.overflow.*</tt>' Intrinsics</a>
6569 <div class="doc_text">
6572 <p>This is an overloaded intrinsic. You can use <tt>llvm.ssub.with.overflow</tt>
6573 on any integer bit width.</p>
6576 declare {i16, i1} @llvm.ssub.with.overflow.i16(i16 %a, i16 %b)
6577 declare {i32, i1} @llvm.ssub.with.overflow.i32(i32 %a, i32 %b)
6578 declare {i64, i1} @llvm.ssub.with.overflow.i64(i64 %a, i64 %b)
6582 <p>The '<tt>llvm.ssub.with.overflow</tt>' family of intrinsic functions perform
6583 a signed subtraction of the two arguments, and indicate whether an overflow
6584 occurred during the signed subtraction.</p>
6587 <p>The arguments (%a and %b) and the first element of the result structure may
6588 be of integer types of any bit width, but they must have the same bit
6589 width. The second element of the result structure must be of
6590 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
6591 undergo signed subtraction.</p>
6594 <p>The '<tt>llvm.ssub.with.overflow</tt>' family of intrinsic functions perform
6595 a signed subtraction of the two arguments. They return a structure —
6596 the first element of which is the subtraction, and the second element of
6597 which is a bit specifying if the signed subtraction resulted in an
6602 %res = call {i32, i1} @llvm.ssub.with.overflow.i32(i32 %a, i32 %b)
6603 %sum = extractvalue {i32, i1} %res, 0
6604 %obit = extractvalue {i32, i1} %res, 1
6605 br i1 %obit, label %overflow, label %normal
6610 <!-- _______________________________________________________________________ -->
6611 <div class="doc_subsubsection">
6612 <a name="int_usub_overflow">'<tt>llvm.usub.with.overflow.*</tt>' Intrinsics</a>
6615 <div class="doc_text">
6618 <p>This is an overloaded intrinsic. You can use <tt>llvm.usub.with.overflow</tt>
6619 on any integer bit width.</p>
6622 declare {i16, i1} @llvm.usub.with.overflow.i16(i16 %a, i16 %b)
6623 declare {i32, i1} @llvm.usub.with.overflow.i32(i32 %a, i32 %b)
6624 declare {i64, i1} @llvm.usub.with.overflow.i64(i64 %a, i64 %b)
6628 <p>The '<tt>llvm.usub.with.overflow</tt>' family of intrinsic functions perform
6629 an unsigned subtraction of the two arguments, and indicate whether an
6630 overflow occurred during the unsigned subtraction.</p>
6633 <p>The arguments (%a and %b) and the first element of the result structure may
6634 be of integer types of any bit width, but they must have the same bit
6635 width. The second element of the result structure must be of
6636 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
6637 undergo unsigned subtraction.</p>
6640 <p>The '<tt>llvm.usub.with.overflow</tt>' family of intrinsic functions perform
6641 an unsigned subtraction of the two arguments. They return a structure —
6642 the first element of which is the subtraction, and the second element of
6643 which is a bit specifying if the unsigned subtraction resulted in an
6648 %res = call {i32, i1} @llvm.usub.with.overflow.i32(i32 %a, i32 %b)
6649 %sum = extractvalue {i32, i1} %res, 0
6650 %obit = extractvalue {i32, i1} %res, 1
6651 br i1 %obit, label %overflow, label %normal
6656 <!-- _______________________________________________________________________ -->
6657 <div class="doc_subsubsection">
6658 <a name="int_smul_overflow">'<tt>llvm.smul.with.overflow.*</tt>' Intrinsics</a>
6661 <div class="doc_text">
6664 <p>This is an overloaded intrinsic. You can use <tt>llvm.smul.with.overflow</tt>
6665 on any integer bit width.</p>
6668 declare {i16, i1} @llvm.smul.with.overflow.i16(i16 %a, i16 %b)
6669 declare {i32, i1} @llvm.smul.with.overflow.i32(i32 %a, i32 %b)
6670 declare {i64, i1} @llvm.smul.with.overflow.i64(i64 %a, i64 %b)
6675 <p>The '<tt>llvm.smul.with.overflow</tt>' family of intrinsic functions perform
6676 a signed multiplication of the two arguments, and indicate whether an
6677 overflow occurred during the signed multiplication.</p>
6680 <p>The arguments (%a and %b) and the first element of the result structure may
6681 be of integer types of any bit width, but they must have the same bit
6682 width. The second element of the result structure must be of
6683 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
6684 undergo signed multiplication.</p>
6687 <p>The '<tt>llvm.smul.with.overflow</tt>' family of intrinsic functions perform
6688 a signed multiplication of the two arguments. They return a structure —
6689 the first element of which is the multiplication, and the second element of
6690 which is a bit specifying if the signed multiplication resulted in an
6695 %res = call {i32, i1} @llvm.smul.with.overflow.i32(i32 %a, i32 %b)
6696 %sum = extractvalue {i32, i1} %res, 0
6697 %obit = extractvalue {i32, i1} %res, 1
6698 br i1 %obit, label %overflow, label %normal
6703 <!-- _______________________________________________________________________ -->
6704 <div class="doc_subsubsection">
6705 <a name="int_umul_overflow">'<tt>llvm.umul.with.overflow.*</tt>' Intrinsics</a>
6708 <div class="doc_text">
6711 <p>This is an overloaded intrinsic. You can use <tt>llvm.umul.with.overflow</tt>
6712 on any integer bit width.</p>
6715 declare {i16, i1} @llvm.umul.with.overflow.i16(i16 %a, i16 %b)
6716 declare {i32, i1} @llvm.umul.with.overflow.i32(i32 %a, i32 %b)
6717 declare {i64, i1} @llvm.umul.with.overflow.i64(i64 %a, i64 %b)
6721 <p>The '<tt>llvm.umul.with.overflow</tt>' family of intrinsic functions perform
6722 a unsigned multiplication of the two arguments, and indicate whether an
6723 overflow occurred during the unsigned multiplication.</p>
6726 <p>The arguments (%a and %b) and the first element of the result structure may
6727 be of integer types of any bit width, but they must have the same bit
6728 width. The second element of the result structure must be of
6729 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
6730 undergo unsigned multiplication.</p>
6733 <p>The '<tt>llvm.umul.with.overflow</tt>' family of intrinsic functions perform
6734 an unsigned multiplication of the two arguments. They return a structure
6735 — the first element of which is the multiplication, and the second
6736 element of which is a bit specifying if the unsigned multiplication resulted
6741 %res = call {i32, i1} @llvm.umul.with.overflow.i32(i32 %a, i32 %b)
6742 %sum = extractvalue {i32, i1} %res, 0
6743 %obit = extractvalue {i32, i1} %res, 1
6744 br i1 %obit, label %overflow, label %normal
6749 <!-- ======================================================================= -->
6750 <div class="doc_subsection">
6751 <a name="int_fp16">Half Precision Floating Point Intrinsics</a>
6754 <div class="doc_text">
6756 <p>Half precision floating point is a storage-only format. This means that it is
6757 a dense encoding (in memory) but does not support computation in the
6760 <p>This means that code must first load the half-precision floating point
6761 value as an i16, then convert it to float with <a
6762 href="#int_convert_from_fp16"><tt>llvm.convert.from.fp16</tt></a>.
6763 Computation can then be performed on the float value (including extending to
6764 double etc). To store the value back to memory, it is first converted to
6765 float if needed, then converted to i16 with
6766 <a href="#int_convert_to_fp16"><tt>llvm.convert.to.fp16</tt></a>, then
6767 storing as an i16 value.</p>
6770 <!-- _______________________________________________________________________ -->
6771 <div class="doc_subsubsection">
6772 <a name="int_convert_to_fp16">'<tt>llvm.convert.to.fp16</tt>' Intrinsic</a>
6775 <div class="doc_text">
6779 declare i16 @llvm.convert.to.fp16(f32 %a)
6783 <p>The '<tt>llvm.convert.to.fp16</tt>' intrinsic function performs
6784 a conversion from single precision floating point format to half precision
6785 floating point format.</p>
6788 <p>The intrinsic function contains single argument - the value to be
6792 <p>The '<tt>llvm.convert.to.fp16</tt>' intrinsic function performs
6793 a conversion from single precision floating point format to half precision
6794 floating point format. The return value is an <tt>i16</tt> which
6795 contains the converted number.</p>
6799 %res = call i16 @llvm.convert.to.fp16(f32 %a)
6800 store i16 %res, i16* @x, align 2
6805 <!-- _______________________________________________________________________ -->
6806 <div class="doc_subsubsection">
6807 <a name="int_convert_from_fp16">'<tt>llvm.convert.from.fp16</tt>' Intrinsic</a>
6810 <div class="doc_text">
6814 declare f32 @llvm.convert.from.fp16(i16 %a)
6818 <p>The '<tt>llvm.convert.from.fp16</tt>' intrinsic function performs
6819 a conversion from half precision floating point format to single precision
6820 floating point format.</p>
6823 <p>The intrinsic function contains single argument - the value to be
6827 <p>The '<tt>llvm.convert.from.fp16</tt>' intrinsic function performs a
6828 conversion from half single precision floating point format to single
6829 precision floating point format. The input half-float value is represented by
6830 an <tt>i16</tt> value.</p>
6834 %a = load i16* @x, align 2
6835 %res = call f32 @llvm.convert.from.fp16(i16 %a)
6840 <!-- ======================================================================= -->
6841 <div class="doc_subsection">
6842 <a name="int_debugger">Debugger Intrinsics</a>
6845 <div class="doc_text">
6847 <p>The LLVM debugger intrinsics (which all start with <tt>llvm.dbg.</tt>
6848 prefix), are described in
6849 the <a href="SourceLevelDebugging.html#format_common_intrinsics">LLVM Source
6850 Level Debugging</a> document.</p>
6854 <!-- ======================================================================= -->
6855 <div class="doc_subsection">
6856 <a name="int_eh">Exception Handling Intrinsics</a>
6859 <div class="doc_text">
6861 <p>The LLVM exception handling intrinsics (which all start with
6862 <tt>llvm.eh.</tt> prefix), are described in
6863 the <a href="ExceptionHandling.html#format_common_intrinsics">LLVM Exception
6864 Handling</a> document.</p>
6868 <!-- ======================================================================= -->
6869 <div class="doc_subsection">
6870 <a name="int_trampoline">Trampoline Intrinsic</a>
6873 <div class="doc_text">
6875 <p>This intrinsic makes it possible to excise one parameter, marked with
6876 the <a href="#nest"><tt>nest</tt></a> attribute, from a function.
6877 The result is a callable
6878 function pointer lacking the nest parameter - the caller does not need to
6879 provide a value for it. Instead, the value to use is stored in advance in a
6880 "trampoline", a block of memory usually allocated on the stack, which also
6881 contains code to splice the nest value into the argument list. This is used
6882 to implement the GCC nested function address extension.</p>
6884 <p>For example, if the function is
6885 <tt>i32 f(i8* nest %c, i32 %x, i32 %y)</tt> then the resulting function
6886 pointer has signature <tt>i32 (i32, i32)*</tt>. It can be created as
6889 <pre class="doc_code">
6890 %tramp = alloca [10 x i8], align 4 ; size and alignment only correct for X86
6891 %tramp1 = getelementptr [10 x i8]* %tramp, i32 0, i32 0
6892 %p = call i8* @llvm.init.trampoline(i8* %tramp1, i8* bitcast (i32 (i8* nest , i32, i32)* @f to i8*), i8* %nval)
6893 %fp = bitcast i8* %p to i32 (i32, i32)*
6896 <p>The call <tt>%val = call i32 %fp(i32 %x, i32 %y)</tt> is then equivalent
6897 to <tt>%val = call i32 %f(i8* %nval, i32 %x, i32 %y)</tt>.</p>
6901 <!-- _______________________________________________________________________ -->
6902 <div class="doc_subsubsection">
6903 <a name="int_it">'<tt>llvm.init.trampoline</tt>' Intrinsic</a>
6906 <div class="doc_text">
6910 declare i8* @llvm.init.trampoline(i8* <tramp>, i8* <func>, i8* <nval>)
6914 <p>This fills the memory pointed to by <tt>tramp</tt> with code and returns a
6915 function pointer suitable for executing it.</p>
6918 <p>The <tt>llvm.init.trampoline</tt> intrinsic takes three arguments, all
6919 pointers. The <tt>tramp</tt> argument must point to a sufficiently large and
6920 sufficiently aligned block of memory; this memory is written to by the
6921 intrinsic. Note that the size and the alignment are target-specific - LLVM
6922 currently provides no portable way of determining them, so a front-end that
6923 generates this intrinsic needs to have some target-specific knowledge.
6924 The <tt>func</tt> argument must hold a function bitcast to
6925 an <tt>i8*</tt>.</p>
6928 <p>The block of memory pointed to by <tt>tramp</tt> is filled with target
6929 dependent code, turning it into a function. A pointer to this function is
6930 returned, but needs to be bitcast to an <a href="#int_trampoline">appropriate
6931 function pointer type</a> before being called. The new function's signature
6932 is the same as that of <tt>func</tt> with any arguments marked with
6933 the <tt>nest</tt> attribute removed. At most one such <tt>nest</tt> argument
6934 is allowed, and it must be of pointer type. Calling the new function is
6935 equivalent to calling <tt>func</tt> with the same argument list, but
6936 with <tt>nval</tt> used for the missing <tt>nest</tt> argument. If, after
6937 calling <tt>llvm.init.trampoline</tt>, the memory pointed to
6938 by <tt>tramp</tt> is modified, then the effect of any later call to the
6939 returned function pointer is undefined.</p>
6943 <!-- ======================================================================= -->
6944 <div class="doc_subsection">
6945 <a name="int_atomics">Atomic Operations and Synchronization Intrinsics</a>
6948 <div class="doc_text">
6950 <p>These intrinsic functions expand the "universal IR" of LLVM to represent
6951 hardware constructs for atomic operations and memory synchronization. This
6952 provides an interface to the hardware, not an interface to the programmer. It
6953 is aimed at a low enough level to allow any programming models or APIs
6954 (Application Programming Interfaces) which need atomic behaviors to map
6955 cleanly onto it. It is also modeled primarily on hardware behavior. Just as
6956 hardware provides a "universal IR" for source languages, it also provides a
6957 starting point for developing a "universal" atomic operation and
6958 synchronization IR.</p>
6960 <p>These do <em>not</em> form an API such as high-level threading libraries,
6961 software transaction memory systems, atomic primitives, and intrinsic
6962 functions as found in BSD, GNU libc, atomic_ops, APR, and other system and
6963 application libraries. The hardware interface provided by LLVM should allow
6964 a clean implementation of all of these APIs and parallel programming models.
6965 No one model or paradigm should be selected above others unless the hardware
6966 itself ubiquitously does so.</p>
6970 <!-- _______________________________________________________________________ -->
6971 <div class="doc_subsubsection">
6972 <a name="int_memory_barrier">'<tt>llvm.memory.barrier</tt>' Intrinsic</a>
6974 <div class="doc_text">
6977 declare void @llvm.memory.barrier(i1 <ll>, i1 <ls>, i1 <sl>, i1 <ss>, i1 <device>)
6981 <p>The <tt>llvm.memory.barrier</tt> intrinsic guarantees ordering between
6982 specific pairs of memory access types.</p>
6985 <p>The <tt>llvm.memory.barrier</tt> intrinsic requires five boolean arguments.
6986 The first four arguments enables a specific barrier as listed below. The
6987 fifth argument specifies that the barrier applies to io or device or uncached
6991 <li><tt>ll</tt>: load-load barrier</li>
6992 <li><tt>ls</tt>: load-store barrier</li>
6993 <li><tt>sl</tt>: store-load barrier</li>
6994 <li><tt>ss</tt>: store-store barrier</li>
6995 <li><tt>device</tt>: barrier applies to device and uncached memory also.</li>
6999 <p>This intrinsic causes the system to enforce some ordering constraints upon
7000 the loads and stores of the program. This barrier does not
7001 indicate <em>when</em> any events will occur, it only enforces
7002 an <em>order</em> in which they occur. For any of the specified pairs of load
7003 and store operations (f.ex. load-load, or store-load), all of the first
7004 operations preceding the barrier will complete before any of the second
7005 operations succeeding the barrier begin. Specifically the semantics for each
7006 pairing is as follows:</p>
7009 <li><tt>ll</tt>: All loads before the barrier must complete before any load
7010 after the barrier begins.</li>
7011 <li><tt>ls</tt>: All loads before the barrier must complete before any
7012 store after the barrier begins.</li>
7013 <li><tt>ss</tt>: All stores before the barrier must complete before any
7014 store after the barrier begins.</li>
7015 <li><tt>sl</tt>: All stores before the barrier must complete before any
7016 load after the barrier begins.</li>
7019 <p>These semantics are applied with a logical "and" behavior when more than one
7020 is enabled in a single memory barrier intrinsic.</p>
7022 <p>Backends may implement stronger barriers than those requested when they do
7023 not support as fine grained a barrier as requested. Some architectures do
7024 not need all types of barriers and on such architectures, these become
7029 %mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
7030 %ptr = bitcast i8* %mallocP to i32*
7033 %result1 = load i32* %ptr <i>; yields {i32}:result1 = 4</i>
7034 call void @llvm.memory.barrier(i1 false, i1 true, i1 false, i1 false)
7035 <i>; guarantee the above finishes</i>
7036 store i32 8, %ptr <i>; before this begins</i>
7041 <!-- _______________________________________________________________________ -->
7042 <div class="doc_subsubsection">
7043 <a name="int_atomic_cmp_swap">'<tt>llvm.atomic.cmp.swap.*</tt>' Intrinsic</a>
7046 <div class="doc_text">
7049 <p>This is an overloaded intrinsic. You can use <tt>llvm.atomic.cmp.swap</tt> on
7050 any integer bit width and for different address spaces. Not all targets
7051 support all bit widths however.</p>
7054 declare i8 @llvm.atomic.cmp.swap.i8.p0i8(i8* <ptr>, i8 <cmp>, i8 <val>)
7055 declare i16 @llvm.atomic.cmp.swap.i16.p0i16(i16* <ptr>, i16 <cmp>, i16 <val>)
7056 declare i32 @llvm.atomic.cmp.swap.i32.p0i32(i32* <ptr>, i32 <cmp>, i32 <val>)
7057 declare i64 @llvm.atomic.cmp.swap.i64.p0i64(i64* <ptr>, i64 <cmp>, i64 <val>)
7061 <p>This loads a value in memory and compares it to a given value. If they are
7062 equal, it stores a new value into the memory.</p>
7065 <p>The <tt>llvm.atomic.cmp.swap</tt> intrinsic takes three arguments. The result
7066 as well as both <tt>cmp</tt> and <tt>val</tt> must be integer values with the
7067 same bit width. The <tt>ptr</tt> argument must be a pointer to a value of
7068 this integer type. While any bit width integer may be used, targets may only
7069 lower representations they support in hardware.</p>
7072 <p>This entire intrinsic must be executed atomically. It first loads the value
7073 in memory pointed to by <tt>ptr</tt> and compares it with the
7074 value <tt>cmp</tt>. If they are equal, <tt>val</tt> is stored into the
7075 memory. The loaded value is yielded in all cases. This provides the
7076 equivalent of an atomic compare-and-swap operation within the SSA
7081 %mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
7082 %ptr = bitcast i8* %mallocP to i32*
7085 %val1 = add i32 4, 4
7086 %result1 = call i32 @llvm.atomic.cmp.swap.i32.p0i32(i32* %ptr, i32 4, %val1)
7087 <i>; yields {i32}:result1 = 4</i>
7088 %stored1 = icmp eq i32 %result1, 4 <i>; yields {i1}:stored1 = true</i>
7089 %memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 8</i>
7091 %val2 = add i32 1, 1
7092 %result2 = call i32 @llvm.atomic.cmp.swap.i32.p0i32(i32* %ptr, i32 5, %val2)
7093 <i>; yields {i32}:result2 = 8</i>
7094 %stored2 = icmp eq i32 %result2, 5 <i>; yields {i1}:stored2 = false</i>
7096 %memval2 = load i32* %ptr <i>; yields {i32}:memval2 = 8</i>
7101 <!-- _______________________________________________________________________ -->
7102 <div class="doc_subsubsection">
7103 <a name="int_atomic_swap">'<tt>llvm.atomic.swap.*</tt>' Intrinsic</a>
7105 <div class="doc_text">
7108 <p>This is an overloaded intrinsic. You can use <tt>llvm.atomic.swap</tt> on any
7109 integer bit width. Not all targets support all bit widths however.</p>
7112 declare i8 @llvm.atomic.swap.i8.p0i8(i8* <ptr>, i8 <val>)
7113 declare i16 @llvm.atomic.swap.i16.p0i16(i16* <ptr>, i16 <val>)
7114 declare i32 @llvm.atomic.swap.i32.p0i32(i32* <ptr>, i32 <val>)
7115 declare i64 @llvm.atomic.swap.i64.p0i64(i64* <ptr>, i64 <val>)
7119 <p>This intrinsic loads the value stored in memory at <tt>ptr</tt> and yields
7120 the value from memory. It then stores the value in <tt>val</tt> in the memory
7121 at <tt>ptr</tt>.</p>
7124 <p>The <tt>llvm.atomic.swap</tt> intrinsic takes two arguments. Both
7125 the <tt>val</tt> argument and the result must be integers of the same bit
7126 width. The first argument, <tt>ptr</tt>, must be a pointer to a value of this
7127 integer type. The targets may only lower integer representations they
7131 <p>This intrinsic loads the value pointed to by <tt>ptr</tt>, yields it, and
7132 stores <tt>val</tt> back into <tt>ptr</tt> atomically. This provides the
7133 equivalent of an atomic swap operation within the SSA framework.</p>
7137 %mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
7138 %ptr = bitcast i8* %mallocP to i32*
7141 %val1 = add i32 4, 4
7142 %result1 = call i32 @llvm.atomic.swap.i32.p0i32(i32* %ptr, i32 %val1)
7143 <i>; yields {i32}:result1 = 4</i>
7144 %stored1 = icmp eq i32 %result1, 4 <i>; yields {i1}:stored1 = true</i>
7145 %memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 8</i>
7147 %val2 = add i32 1, 1
7148 %result2 = call i32 @llvm.atomic.swap.i32.p0i32(i32* %ptr, i32 %val2)
7149 <i>; yields {i32}:result2 = 8</i>
7151 %stored2 = icmp eq i32 %result2, 8 <i>; yields {i1}:stored2 = true</i>
7152 %memval2 = load i32* %ptr <i>; yields {i32}:memval2 = 2</i>
7157 <!-- _______________________________________________________________________ -->
7158 <div class="doc_subsubsection">
7159 <a name="int_atomic_load_add">'<tt>llvm.atomic.load.add.*</tt>' Intrinsic</a>
7163 <div class="doc_text">
7166 <p>This is an overloaded intrinsic. You can use <tt>llvm.atomic.load.add</tt> on
7167 any integer bit width. Not all targets support all bit widths however.</p>
7170 declare i8 @llvm.atomic.load.add.i8.p0i8(i8* <ptr>, i8 <delta>)
7171 declare i16 @llvm.atomic.load.add.i16.p0i16(i16* <ptr>, i16 <delta>)
7172 declare i32 @llvm.atomic.load.add.i32.p0i32(i32* <ptr>, i32 <delta>)
7173 declare i64 @llvm.atomic.load.add.i64.p0i64(i64* <ptr>, i64 <delta>)
7177 <p>This intrinsic adds <tt>delta</tt> to the value stored in memory
7178 at <tt>ptr</tt>. It yields the original value at <tt>ptr</tt>.</p>
7181 <p>The intrinsic takes two arguments, the first a pointer to an integer value
7182 and the second an integer value. The result is also an integer value. These
7183 integer types can have any bit width, but they must all have the same bit
7184 width. The targets may only lower integer representations they support.</p>
7187 <p>This intrinsic does a series of operations atomically. It first loads the
7188 value stored at <tt>ptr</tt>. It then adds <tt>delta</tt>, stores the result
7189 to <tt>ptr</tt>. It yields the original value stored at <tt>ptr</tt>.</p>
7193 %mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
7194 %ptr = bitcast i8* %mallocP to i32*
7196 %result1 = call i32 @llvm.atomic.load.add.i32.p0i32(i32* %ptr, i32 4)
7197 <i>; yields {i32}:result1 = 4</i>
7198 %result2 = call i32 @llvm.atomic.load.add.i32.p0i32(i32* %ptr, i32 2)
7199 <i>; yields {i32}:result2 = 8</i>
7200 %result3 = call i32 @llvm.atomic.load.add.i32.p0i32(i32* %ptr, i32 5)
7201 <i>; yields {i32}:result3 = 10</i>
7202 %memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 15</i>
7207 <!-- _______________________________________________________________________ -->
7208 <div class="doc_subsubsection">
7209 <a name="int_atomic_load_sub">'<tt>llvm.atomic.load.sub.*</tt>' Intrinsic</a>
7213 <div class="doc_text">
7216 <p>This is an overloaded intrinsic. You can use <tt>llvm.atomic.load.sub</tt> on
7217 any integer bit width and for different address spaces. Not all targets
7218 support all bit widths however.</p>
7221 declare i8 @llvm.atomic.load.sub.i8.p0i32(i8* <ptr>, i8 <delta>)
7222 declare i16 @llvm.atomic.load.sub.i16.p0i32(i16* <ptr>, i16 <delta>)
7223 declare i32 @llvm.atomic.load.sub.i32.p0i32(i32* <ptr>, i32 <delta>)
7224 declare i64 @llvm.atomic.load.sub.i64.p0i32(i64* <ptr>, i64 <delta>)
7228 <p>This intrinsic subtracts <tt>delta</tt> to the value stored in memory at
7229 <tt>ptr</tt>. It yields the original value at <tt>ptr</tt>.</p>
7232 <p>The intrinsic takes two arguments, the first a pointer to an integer value
7233 and the second an integer value. The result is also an integer value. These
7234 integer types can have any bit width, but they must all have the same bit
7235 width. The targets may only lower integer representations they support.</p>
7238 <p>This intrinsic does a series of operations atomically. It first loads the
7239 value stored at <tt>ptr</tt>. It then subtracts <tt>delta</tt>, stores the
7240 result to <tt>ptr</tt>. It yields the original value stored
7241 at <tt>ptr</tt>.</p>
7245 %mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
7246 %ptr = bitcast i8* %mallocP to i32*
7248 %result1 = call i32 @llvm.atomic.load.sub.i32.p0i32(i32* %ptr, i32 4)
7249 <i>; yields {i32}:result1 = 8</i>
7250 %result2 = call i32 @llvm.atomic.load.sub.i32.p0i32(i32* %ptr, i32 2)
7251 <i>; yields {i32}:result2 = 4</i>
7252 %result3 = call i32 @llvm.atomic.load.sub.i32.p0i32(i32* %ptr, i32 5)
7253 <i>; yields {i32}:result3 = 2</i>
7254 %memval1 = load i32* %ptr <i>; yields {i32}:memval1 = -3</i>
7259 <!-- _______________________________________________________________________ -->
7260 <div class="doc_subsubsection">
7261 <a name="int_atomic_load_and">'<tt>llvm.atomic.load.and.*</tt>' Intrinsic</a><br>
7262 <a name="int_atomic_load_nand">'<tt>llvm.atomic.load.nand.*</tt>' Intrinsic</a><br>
7263 <a name="int_atomic_load_or">'<tt>llvm.atomic.load.or.*</tt>' Intrinsic</a><br>
7264 <a name="int_atomic_load_xor">'<tt>llvm.atomic.load.xor.*</tt>' Intrinsic</a><br>
7267 <div class="doc_text">
7270 <p>These are overloaded intrinsics. You can
7271 use <tt>llvm.atomic.load_and</tt>, <tt>llvm.atomic.load_nand</tt>,
7272 <tt>llvm.atomic.load_or</tt>, and <tt>llvm.atomic.load_xor</tt> on any integer
7273 bit width and for different address spaces. Not all targets support all bit
7277 declare i8 @llvm.atomic.load.and.i8.p0i8(i8* <ptr>, i8 <delta>)
7278 declare i16 @llvm.atomic.load.and.i16.p0i16(i16* <ptr>, i16 <delta>)
7279 declare i32 @llvm.atomic.load.and.i32.p0i32(i32* <ptr>, i32 <delta>)
7280 declare i64 @llvm.atomic.load.and.i64.p0i64(i64* <ptr>, i64 <delta>)
7284 declare i8 @llvm.atomic.load.or.i8.p0i8(i8* <ptr>, i8 <delta>)
7285 declare i16 @llvm.atomic.load.or.i16.p0i16(i16* <ptr>, i16 <delta>)
7286 declare i32 @llvm.atomic.load.or.i32.p0i32(i32* <ptr>, i32 <delta>)
7287 declare i64 @llvm.atomic.load.or.i64.p0i64(i64* <ptr>, i64 <delta>)
7291 declare i8 @llvm.atomic.load.nand.i8.p0i32(i8* <ptr>, i8 <delta>)
7292 declare i16 @llvm.atomic.load.nand.i16.p0i32(i16* <ptr>, i16 <delta>)
7293 declare i32 @llvm.atomic.load.nand.i32.p0i32(i32* <ptr>, i32 <delta>)
7294 declare i64 @llvm.atomic.load.nand.i64.p0i32(i64* <ptr>, i64 <delta>)
7298 declare i8 @llvm.atomic.load.xor.i8.p0i32(i8* <ptr>, i8 <delta>)
7299 declare i16 @llvm.atomic.load.xor.i16.p0i32(i16* <ptr>, i16 <delta>)
7300 declare i32 @llvm.atomic.load.xor.i32.p0i32(i32* <ptr>, i32 <delta>)
7301 declare i64 @llvm.atomic.load.xor.i64.p0i32(i64* <ptr>, i64 <delta>)
7305 <p>These intrinsics bitwise the operation (and, nand, or, xor) <tt>delta</tt> to
7306 the value stored in memory at <tt>ptr</tt>. It yields the original value
7307 at <tt>ptr</tt>.</p>
7310 <p>These intrinsics take two arguments, the first a pointer to an integer value
7311 and the second an integer value. The result is also an integer value. These
7312 integer types can have any bit width, but they must all have the same bit
7313 width. The targets may only lower integer representations they support.</p>
7316 <p>These intrinsics does a series of operations atomically. They first load the
7317 value stored at <tt>ptr</tt>. They then do the bitwise
7318 operation <tt>delta</tt>, store the result to <tt>ptr</tt>. They yield the
7319 original value stored at <tt>ptr</tt>.</p>
7323 %mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
7324 %ptr = bitcast i8* %mallocP to i32*
7325 store i32 0x0F0F, %ptr
7326 %result0 = call i32 @llvm.atomic.load.nand.i32.p0i32(i32* %ptr, i32 0xFF)
7327 <i>; yields {i32}:result0 = 0x0F0F</i>
7328 %result1 = call i32 @llvm.atomic.load.and.i32.p0i32(i32* %ptr, i32 0xFF)
7329 <i>; yields {i32}:result1 = 0xFFFFFFF0</i>
7330 %result2 = call i32 @llvm.atomic.load.or.i32.p0i32(i32* %ptr, i32 0F)
7331 <i>; yields {i32}:result2 = 0xF0</i>
7332 %result3 = call i32 @llvm.atomic.load.xor.i32.p0i32(i32* %ptr, i32 0F)
7333 <i>; yields {i32}:result3 = FF</i>
7334 %memval1 = load i32* %ptr <i>; yields {i32}:memval1 = F0</i>
7339 <!-- _______________________________________________________________________ -->
7340 <div class="doc_subsubsection">
7341 <a name="int_atomic_load_max">'<tt>llvm.atomic.load.max.*</tt>' Intrinsic</a><br>
7342 <a name="int_atomic_load_min">'<tt>llvm.atomic.load.min.*</tt>' Intrinsic</a><br>
7343 <a name="int_atomic_load_umax">'<tt>llvm.atomic.load.umax.*</tt>' Intrinsic</a><br>
7344 <a name="int_atomic_load_umin">'<tt>llvm.atomic.load.umin.*</tt>' Intrinsic</a><br>
7347 <div class="doc_text">
7350 <p>These are overloaded intrinsics. You can use <tt>llvm.atomic.load_max</tt>,
7351 <tt>llvm.atomic.load_min</tt>, <tt>llvm.atomic.load_umax</tt>, and
7352 <tt>llvm.atomic.load_umin</tt> on any integer bit width and for different
7353 address spaces. Not all targets support all bit widths however.</p>
7356 declare i8 @llvm.atomic.load.max.i8.p0i8(i8* <ptr>, i8 <delta>)
7357 declare i16 @llvm.atomic.load.max.i16.p0i16(i16* <ptr>, i16 <delta>)
7358 declare i32 @llvm.atomic.load.max.i32.p0i32(i32* <ptr>, i32 <delta>)
7359 declare i64 @llvm.atomic.load.max.i64.p0i64(i64* <ptr>, i64 <delta>)
7363 declare i8 @llvm.atomic.load.min.i8.p0i8(i8* <ptr>, i8 <delta>)
7364 declare i16 @llvm.atomic.load.min.i16.p0i16(i16* <ptr>, i16 <delta>)
7365 declare i32 @llvm.atomic.load.min.i32.p0i32(i32* <ptr>, i32 <delta>)
7366 declare i64 @llvm.atomic.load.min.i64.p0i64(i64* <ptr>, i64 <delta>)
7370 declare i8 @llvm.atomic.load.umax.i8.p0i8(i8* <ptr>, i8 <delta>)
7371 declare i16 @llvm.atomic.load.umax.i16.p0i16(i16* <ptr>, i16 <delta>)
7372 declare i32 @llvm.atomic.load.umax.i32.p0i32(i32* <ptr>, i32 <delta>)
7373 declare i64 @llvm.atomic.load.umax.i64.p0i64(i64* <ptr>, i64 <delta>)
7377 declare i8 @llvm.atomic.load.umin.i8.p0i8(i8* <ptr>, i8 <delta>)
7378 declare i16 @llvm.atomic.load.umin.i16.p0i16(i16* <ptr>, i16 <delta>)
7379 declare i32 @llvm.atomic.load.umin.i32.p0i32(i32* <ptr>, i32 <delta>)
7380 declare i64 @llvm.atomic.load.umin.i64.p0i64(i64* <ptr>, i64 <delta>)
7384 <p>These intrinsics takes the signed or unsigned minimum or maximum of
7385 <tt>delta</tt> and the value stored in memory at <tt>ptr</tt>. It yields the
7386 original value at <tt>ptr</tt>.</p>
7389 <p>These intrinsics take two arguments, the first a pointer to an integer value
7390 and the second an integer value. The result is also an integer value. These
7391 integer types can have any bit width, but they must all have the same bit
7392 width. The targets may only lower integer representations they support.</p>
7395 <p>These intrinsics does a series of operations atomically. They first load the
7396 value stored at <tt>ptr</tt>. They then do the signed or unsigned min or
7397 max <tt>delta</tt> and the value, store the result to <tt>ptr</tt>. They
7398 yield the original value stored at <tt>ptr</tt>.</p>
7402 %mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
7403 %ptr = bitcast i8* %mallocP to i32*
7405 %result0 = call i32 @llvm.atomic.load.min.i32.p0i32(i32* %ptr, i32 -2)
7406 <i>; yields {i32}:result0 = 7</i>
7407 %result1 = call i32 @llvm.atomic.load.max.i32.p0i32(i32* %ptr, i32 8)
7408 <i>; yields {i32}:result1 = -2</i>
7409 %result2 = call i32 @llvm.atomic.load.umin.i32.p0i32(i32* %ptr, i32 10)
7410 <i>; yields {i32}:result2 = 8</i>
7411 %result3 = call i32 @llvm.atomic.load.umax.i32.p0i32(i32* %ptr, i32 30)
7412 <i>; yields {i32}:result3 = 8</i>
7413 %memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 30</i>
7419 <!-- ======================================================================= -->
7420 <div class="doc_subsection">
7421 <a name="int_memorymarkers">Memory Use Markers</a>
7424 <div class="doc_text">
7426 <p>This class of intrinsics exists to information about the lifetime of memory
7427 objects and ranges where variables are immutable.</p>
7431 <!-- _______________________________________________________________________ -->
7432 <div class="doc_subsubsection">
7433 <a name="int_lifetime_start">'<tt>llvm.lifetime.start</tt>' Intrinsic</a>
7436 <div class="doc_text">
7440 declare void @llvm.lifetime.start(i64 <size>, i8* nocapture <ptr>)
7444 <p>The '<tt>llvm.lifetime.start</tt>' intrinsic specifies the start of a memory
7445 object's lifetime.</p>
7448 <p>The first argument is a constant integer representing the size of the
7449 object, or -1 if it is variable sized. The second argument is a pointer to
7453 <p>This intrinsic indicates that before this point in the code, the value of the
7454 memory pointed to by <tt>ptr</tt> is dead. This means that it is known to
7455 never be used and has an undefined value. A load from the pointer that
7456 precedes this intrinsic can be replaced with
7457 <tt>'<a href="#undefvalues">undef</a>'</tt>.</p>
7461 <!-- _______________________________________________________________________ -->
7462 <div class="doc_subsubsection">
7463 <a name="int_lifetime_end">'<tt>llvm.lifetime.end</tt>' Intrinsic</a>
7466 <div class="doc_text">
7470 declare void @llvm.lifetime.end(i64 <size>, i8* nocapture <ptr>)
7474 <p>The '<tt>llvm.lifetime.end</tt>' intrinsic specifies the end of a memory
7475 object's lifetime.</p>
7478 <p>The first argument is a constant integer representing the size of the
7479 object, or -1 if it is variable sized. The second argument is a pointer to
7483 <p>This intrinsic indicates that after this point in the code, the value of the
7484 memory pointed to by <tt>ptr</tt> is dead. This means that it is known to
7485 never be used and has an undefined value. Any stores into the memory object
7486 following this intrinsic may be removed as dead.
7490 <!-- _______________________________________________________________________ -->
7491 <div class="doc_subsubsection">
7492 <a name="int_invariant_start">'<tt>llvm.invariant.start</tt>' Intrinsic</a>
7495 <div class="doc_text">
7499 declare {}* @llvm.invariant.start(i64 <size>, i8* nocapture <ptr>) readonly
7503 <p>The '<tt>llvm.invariant.start</tt>' intrinsic specifies that the contents of
7504 a memory object will not change.</p>
7507 <p>The first argument is a constant integer representing the size of the
7508 object, or -1 if it is variable sized. The second argument is a pointer to
7512 <p>This intrinsic indicates that until an <tt>llvm.invariant.end</tt> that uses
7513 the return value, the referenced memory location is constant and
7518 <!-- _______________________________________________________________________ -->
7519 <div class="doc_subsubsection">
7520 <a name="int_invariant_end">'<tt>llvm.invariant.end</tt>' Intrinsic</a>
7523 <div class="doc_text">
7527 declare void @llvm.invariant.end({}* <start>, i64 <size>, i8* nocapture <ptr>)
7531 <p>The '<tt>llvm.invariant.end</tt>' intrinsic specifies that the contents of
7532 a memory object are mutable.</p>
7535 <p>The first argument is the matching <tt>llvm.invariant.start</tt> intrinsic.
7536 The second argument is a constant integer representing the size of the
7537 object, or -1 if it is variable sized and the third argument is a pointer
7541 <p>This intrinsic indicates that the memory is mutable again.</p>
7545 <!-- ======================================================================= -->
7546 <div class="doc_subsection">
7547 <a name="int_general">General Intrinsics</a>
7550 <div class="doc_text">
7552 <p>This class of intrinsics is designed to be generic and has no specific
7557 <!-- _______________________________________________________________________ -->
7558 <div class="doc_subsubsection">
7559 <a name="int_var_annotation">'<tt>llvm.var.annotation</tt>' Intrinsic</a>
7562 <div class="doc_text">
7566 declare void @llvm.var.annotation(i8* <val>, i8* <str>, i8* <str>, i32 <int>)
7570 <p>The '<tt>llvm.var.annotation</tt>' intrinsic.</p>
7573 <p>The first argument is a pointer to a value, the second is a pointer to a
7574 global string, the third is a pointer to a global string which is the source
7575 file name, and the last argument is the line number.</p>
7578 <p>This intrinsic allows annotation of local variables with arbitrary strings.
7579 This can be useful for special purpose optimizations that want to look for
7580 these annotations. These have no other defined use, they are ignored by code
7581 generation and optimization.</p>
7585 <!-- _______________________________________________________________________ -->
7586 <div class="doc_subsubsection">
7587 <a name="int_annotation">'<tt>llvm.annotation.*</tt>' Intrinsic</a>
7590 <div class="doc_text">
7593 <p>This is an overloaded intrinsic. You can use '<tt>llvm.annotation</tt>' on
7594 any integer bit width.</p>
7597 declare i8 @llvm.annotation.i8(i8 <val>, i8* <str>, i8* <str>, i32 <int>)
7598 declare i16 @llvm.annotation.i16(i16 <val>, i8* <str>, i8* <str>, i32 <int>)
7599 declare i32 @llvm.annotation.i32(i32 <val>, i8* <str>, i8* <str>, i32 <int>)
7600 declare i64 @llvm.annotation.i64(i64 <val>, i8* <str>, i8* <str>, i32 <int>)
7601 declare i256 @llvm.annotation.i256(i256 <val>, i8* <str>, i8* <str>, i32 <int>)
7605 <p>The '<tt>llvm.annotation</tt>' intrinsic.</p>
7608 <p>The first argument is an integer value (result of some expression), the
7609 second is a pointer to a global string, the third is a pointer to a global
7610 string which is the source file name, and the last argument is the line
7611 number. It returns the value of the first argument.</p>
7614 <p>This intrinsic allows annotations to be put on arbitrary expressions with
7615 arbitrary strings. This can be useful for special purpose optimizations that
7616 want to look for these annotations. These have no other defined use, they
7617 are ignored by code generation and optimization.</p>
7621 <!-- _______________________________________________________________________ -->
7622 <div class="doc_subsubsection">
7623 <a name="int_trap">'<tt>llvm.trap</tt>' Intrinsic</a>
7626 <div class="doc_text">
7630 declare void @llvm.trap()
7634 <p>The '<tt>llvm.trap</tt>' intrinsic.</p>
7640 <p>This intrinsics is lowered to the target dependent trap instruction. If the
7641 target does not have a trap instruction, this intrinsic will be lowered to
7642 the call of the <tt>abort()</tt> function.</p>
7646 <!-- _______________________________________________________________________ -->
7647 <div class="doc_subsubsection">
7648 <a name="int_stackprotector">'<tt>llvm.stackprotector</tt>' Intrinsic</a>
7651 <div class="doc_text">
7655 declare void @llvm.stackprotector(i8* <guard>, i8** <slot>)
7659 <p>The <tt>llvm.stackprotector</tt> intrinsic takes the <tt>guard</tt> and
7660 stores it onto the stack at <tt>slot</tt>. The stack slot is adjusted to
7661 ensure that it is placed on the stack before local variables.</p>
7664 <p>The <tt>llvm.stackprotector</tt> intrinsic requires two pointer
7665 arguments. The first argument is the value loaded from the stack
7666 guard <tt>@__stack_chk_guard</tt>. The second variable is an <tt>alloca</tt>
7667 that has enough space to hold the value of the guard.</p>
7670 <p>This intrinsic causes the prologue/epilogue inserter to force the position of
7671 the <tt>AllocaInst</tt> stack slot to be before local variables on the
7672 stack. This is to ensure that if a local variable on the stack is
7673 overwritten, it will destroy the value of the guard. When the function exits,
7674 the guard on the stack is checked against the original guard. If they're
7675 different, then the program aborts by calling the <tt>__stack_chk_fail()</tt>
7680 <!-- _______________________________________________________________________ -->
7681 <div class="doc_subsubsection">
7682 <a name="int_objectsize">'<tt>llvm.objectsize</tt>' Intrinsic</a>
7685 <div class="doc_text">
7689 declare i32 @llvm.objectsize.i32(i8* <object>, i1 <type>)
7690 declare i64 @llvm.objectsize.i64(i8* <object>, i1 <type>)
7694 <p>The <tt>llvm.objectsize</tt> intrinsic is designed to provide information
7695 to the optimizers to discover at compile time either a) when an
7696 operation like memcpy will either overflow a buffer that corresponds to
7697 an object, or b) to determine that a runtime check for overflow isn't
7698 necessary. An object in this context means an allocation of a
7699 specific class, structure, array, or other object.</p>
7702 <p>The <tt>llvm.objectsize</tt> intrinsic takes two arguments. The first
7703 argument is a pointer to or into the <tt>object</tt>. The second argument
7704 is a boolean 0 or 1. This argument determines whether you want the
7705 maximum (0) or minimum (1) bytes remaining. This needs to be a literal 0 or
7706 1, variables are not allowed.</p>
7709 <p>The <tt>llvm.objectsize</tt> intrinsic is lowered to either a constant
7710 representing the size of the object concerned or <tt>i32/i64 -1 or 0</tt>
7711 (depending on the <tt>type</tt> argument if the size cannot be determined
7712 at compile time.</p>
7716 <!-- *********************************************************************** -->
7719 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
7720 src="http://jigsaw.w3.org/css-validator/images/vcss-blue" alt="Valid CSS"></a>
7721 <a href="http://validator.w3.org/check/referer"><img
7722 src="http://www.w3.org/Icons/valid-html401-blue" alt="Valid HTML 4.01"></a>
7724 <a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
7725 <a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br>
7726 Last modified: $Date$