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></li>
24 <li><a href="#callingconv">Calling Conventions</a></li>
25 <li><a href="#namedtypes">Named Types</a></li>
26 <li><a href="#globalvars">Global Variables</a></li>
27 <li><a href="#functionstructure">Functions</a></li>
28 <li><a href="#aliasstructure">Aliases</a></li>
29 <li><a href="#paramattrs">Parameter Attributes</a></li>
30 <li><a href="#fnattrs">Function Attributes</a></li>
31 <li><a href="#gc">Garbage Collector Names</a></li>
32 <li><a href="#moduleasm">Module-Level Inline Assembly</a></li>
33 <li><a href="#datalayout">Data Layout</a></li>
36 <li><a href="#typesystem">Type System</a>
38 <li><a href="#t_classifications">Type Classifications</a></li>
39 <li><a href="#t_primitive">Primitive Types</a>
41 <li><a href="#t_floating">Floating Point Types</a></li>
42 <li><a href="#t_void">Void Type</a></li>
43 <li><a href="#t_label">Label Type</a></li>
46 <li><a href="#t_derived">Derived Types</a>
48 <li><a href="#t_integer">Integer Type</a></li>
49 <li><a href="#t_array">Array Type</a></li>
50 <li><a href="#t_function">Function Type</a></li>
51 <li><a href="#t_pointer">Pointer Type</a></li>
52 <li><a href="#t_struct">Structure Type</a></li>
53 <li><a href="#t_pstruct">Packed Structure Type</a></li>
54 <li><a href="#t_vector">Vector Type</a></li>
55 <li><a href="#t_opaque">Opaque Type</a></li>
58 <li><a href="#t_uprefs">Type Up-references</a></li>
61 <li><a href="#constants">Constants</a>
63 <li><a href="#simpleconstants">Simple Constants</a></li>
64 <li><a href="#aggregateconstants">Aggregate Constants</a></li>
65 <li><a href="#globalconstants">Global Variable and Function Addresses</a></li>
66 <li><a href="#undefvalues">Undefined Values</a></li>
67 <li><a href="#constantexprs">Constant Expressions</a></li>
70 <li><a href="#othervalues">Other Values</a>
72 <li><a href="#inlineasm">Inline Assembler Expressions</a></li>
75 <li><a href="#instref">Instruction Reference</a>
77 <li><a href="#terminators">Terminator Instructions</a>
79 <li><a href="#i_ret">'<tt>ret</tt>' Instruction</a></li>
80 <li><a href="#i_br">'<tt>br</tt>' Instruction</a></li>
81 <li><a href="#i_switch">'<tt>switch</tt>' Instruction</a></li>
82 <li><a href="#i_invoke">'<tt>invoke</tt>' Instruction</a></li>
83 <li><a href="#i_unwind">'<tt>unwind</tt>' Instruction</a></li>
84 <li><a href="#i_unreachable">'<tt>unreachable</tt>' Instruction</a></li>
87 <li><a href="#binaryops">Binary Operations</a>
89 <li><a href="#i_add">'<tt>add</tt>' Instruction</a></li>
90 <li><a href="#i_sub">'<tt>sub</tt>' Instruction</a></li>
91 <li><a href="#i_mul">'<tt>mul</tt>' Instruction</a></li>
92 <li><a href="#i_udiv">'<tt>udiv</tt>' Instruction</a></li>
93 <li><a href="#i_sdiv">'<tt>sdiv</tt>' Instruction</a></li>
94 <li><a href="#i_fdiv">'<tt>fdiv</tt>' Instruction</a></li>
95 <li><a href="#i_urem">'<tt>urem</tt>' Instruction</a></li>
96 <li><a href="#i_srem">'<tt>srem</tt>' Instruction</a></li>
97 <li><a href="#i_frem">'<tt>frem</tt>' Instruction</a></li>
100 <li><a href="#bitwiseops">Bitwise Binary Operations</a>
102 <li><a href="#i_shl">'<tt>shl</tt>' Instruction</a></li>
103 <li><a href="#i_lshr">'<tt>lshr</tt>' Instruction</a></li>
104 <li><a href="#i_ashr">'<tt>ashr</tt>' Instruction</a></li>
105 <li><a href="#i_and">'<tt>and</tt>' Instruction</a></li>
106 <li><a href="#i_or">'<tt>or</tt>' Instruction</a></li>
107 <li><a href="#i_xor">'<tt>xor</tt>' Instruction</a></li>
110 <li><a href="#vectorops">Vector Operations</a>
112 <li><a href="#i_extractelement">'<tt>extractelement</tt>' Instruction</a></li>
113 <li><a href="#i_insertelement">'<tt>insertelement</tt>' Instruction</a></li>
114 <li><a href="#i_shufflevector">'<tt>shufflevector</tt>' Instruction</a></li>
117 <li><a href="#aggregateops">Aggregate Operations</a>
119 <li><a href="#i_extractvalue">'<tt>extractvalue</tt>' Instruction</a></li>
120 <li><a href="#i_insertvalue">'<tt>insertvalue</tt>' Instruction</a></li>
123 <li><a href="#memoryops">Memory Access and Addressing Operations</a>
125 <li><a href="#i_malloc">'<tt>malloc</tt>' Instruction</a></li>
126 <li><a href="#i_free">'<tt>free</tt>' Instruction</a></li>
127 <li><a href="#i_alloca">'<tt>alloca</tt>' Instruction</a></li>
128 <li><a href="#i_load">'<tt>load</tt>' Instruction</a></li>
129 <li><a href="#i_store">'<tt>store</tt>' Instruction</a></li>
130 <li><a href="#i_getelementptr">'<tt>getelementptr</tt>' Instruction</a></li>
133 <li><a href="#convertops">Conversion Operations</a>
135 <li><a href="#i_trunc">'<tt>trunc .. to</tt>' Instruction</a></li>
136 <li><a href="#i_zext">'<tt>zext .. to</tt>' Instruction</a></li>
137 <li><a href="#i_sext">'<tt>sext .. to</tt>' Instruction</a></li>
138 <li><a href="#i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a></li>
139 <li><a href="#i_fpext">'<tt>fpext .. to</tt>' Instruction</a></li>
140 <li><a href="#i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a></li>
141 <li><a href="#i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a></li>
142 <li><a href="#i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a></li>
143 <li><a href="#i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a></li>
144 <li><a href="#i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a></li>
145 <li><a href="#i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a></li>
146 <li><a href="#i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a></li>
149 <li><a href="#otherops">Other Operations</a>
151 <li><a href="#i_icmp">'<tt>icmp</tt>' Instruction</a></li>
152 <li><a href="#i_fcmp">'<tt>fcmp</tt>' Instruction</a></li>
153 <li><a href="#i_vicmp">'<tt>vicmp</tt>' Instruction</a></li>
154 <li><a href="#i_vfcmp">'<tt>vfcmp</tt>' Instruction</a></li>
155 <li><a href="#i_phi">'<tt>phi</tt>' Instruction</a></li>
156 <li><a href="#i_select">'<tt>select</tt>' Instruction</a></li>
157 <li><a href="#i_call">'<tt>call</tt>' Instruction</a></li>
158 <li><a href="#i_va_arg">'<tt>va_arg</tt>' Instruction</a></li>
163 <li><a href="#intrinsics">Intrinsic Functions</a>
165 <li><a href="#int_varargs">Variable Argument Handling Intrinsics</a>
167 <li><a href="#int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a></li>
168 <li><a href="#int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a></li>
169 <li><a href="#int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a></li>
172 <li><a href="#int_gc">Accurate Garbage Collection Intrinsics</a>
174 <li><a href="#int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a></li>
175 <li><a href="#int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a></li>
176 <li><a href="#int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a></li>
179 <li><a href="#int_codegen">Code Generator Intrinsics</a>
181 <li><a href="#int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a></li>
182 <li><a href="#int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a></li>
183 <li><a href="#int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a></li>
184 <li><a href="#int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a></li>
185 <li><a href="#int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a></li>
186 <li><a href="#int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a></li>
187 <li><a href="#int_readcyclecounter"><tt>llvm.readcyclecounter</tt>' Intrinsic</a></li>
190 <li><a href="#int_libc">Standard C Library Intrinsics</a>
192 <li><a href="#int_memcpy">'<tt>llvm.memcpy.*</tt>' Intrinsic</a></li>
193 <li><a href="#int_memmove">'<tt>llvm.memmove.*</tt>' Intrinsic</a></li>
194 <li><a href="#int_memset">'<tt>llvm.memset.*</tt>' Intrinsic</a></li>
195 <li><a href="#int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a></li>
196 <li><a href="#int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a></li>
197 <li><a href="#int_sin">'<tt>llvm.sin.*</tt>' Intrinsic</a></li>
198 <li><a href="#int_cos">'<tt>llvm.cos.*</tt>' Intrinsic</a></li>
199 <li><a href="#int_pow">'<tt>llvm.pow.*</tt>' Intrinsic</a></li>
202 <li><a href="#int_manip">Bit Manipulation Intrinsics</a>
204 <li><a href="#int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a></li>
205 <li><a href="#int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic </a></li>
206 <li><a href="#int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic </a></li>
207 <li><a href="#int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic </a></li>
208 <li><a href="#int_part_select">'<tt>llvm.part.select.*</tt>' Intrinsic </a></li>
209 <li><a href="#int_part_set">'<tt>llvm.part.set.*</tt>' Intrinsic </a></li>
212 <li><a href="#int_overflow">Arithmetic with Overflow Intrinsics</a>
214 <li><a href="#int_sadd_overflow">'<tt>llvm.sadd.with.overflow.*</tt> Intrinsics</a></li>
215 <li><a href="#int_uadd_overflow">'<tt>llvm.uadd.with.overflow.*</tt> Intrinsics</a></li>
216 <li><a href="#int_ssub_overflow">'<tt>llvm.ssub.with.overflow.*</tt> Intrinsics</a></li>
217 <li><a href="#int_usub_overflow">'<tt>llvm.usub.with.overflow.*</tt> Intrinsics</a></li>
218 <li><a href="#int_smul_overflow">'<tt>llvm.smul.with.overflow.*</tt> Intrinsics</a></li>
219 <li><a href="#int_umul_overflow">'<tt>llvm.umul.with.overflow.*</tt> Intrinsics</a></li>
222 <li><a href="#int_debugger">Debugger intrinsics</a></li>
223 <li><a href="#int_eh">Exception Handling intrinsics</a></li>
224 <li><a href="#int_trampoline">Trampoline Intrinsic</a>
226 <li><a href="#int_it">'<tt>llvm.init.trampoline</tt>' Intrinsic</a></li>
229 <li><a href="#int_atomics">Atomic intrinsics</a>
231 <li><a href="#int_memory_barrier"><tt>llvm.memory_barrier</tt></a></li>
232 <li><a href="#int_atomic_cmp_swap"><tt>llvm.atomic.cmp.swap</tt></a></li>
233 <li><a href="#int_atomic_swap"><tt>llvm.atomic.swap</tt></a></li>
234 <li><a href="#int_atomic_load_add"><tt>llvm.atomic.load.add</tt></a></li>
235 <li><a href="#int_atomic_load_sub"><tt>llvm.atomic.load.sub</tt></a></li>
236 <li><a href="#int_atomic_load_and"><tt>llvm.atomic.load.and</tt></a></li>
237 <li><a href="#int_atomic_load_nand"><tt>llvm.atomic.load.nand</tt></a></li>
238 <li><a href="#int_atomic_load_or"><tt>llvm.atomic.load.or</tt></a></li>
239 <li><a href="#int_atomic_load_xor"><tt>llvm.atomic.load.xor</tt></a></li>
240 <li><a href="#int_atomic_load_max"><tt>llvm.atomic.load.max</tt></a></li>
241 <li><a href="#int_atomic_load_min"><tt>llvm.atomic.load.min</tt></a></li>
242 <li><a href="#int_atomic_load_umax"><tt>llvm.atomic.load.umax</tt></a></li>
243 <li><a href="#int_atomic_load_umin"><tt>llvm.atomic.load.umin</tt></a></li>
246 <li><a href="#int_general">General intrinsics</a>
248 <li><a href="#int_var_annotation">
249 '<tt>llvm.var.annotation</tt>' Intrinsic</a></li>
250 <li><a href="#int_annotation">
251 '<tt>llvm.annotation.*</tt>' Intrinsic</a></li>
252 <li><a href="#int_trap">
253 '<tt>llvm.trap</tt>' Intrinsic</a></li>
254 <li><a href="#int_stackprotector">
255 '<tt>llvm.stackprotector</tt>' Intrinsic</a></li>
262 <div class="doc_author">
263 <p>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a>
264 and <a href="mailto:vadve@cs.uiuc.edu">Vikram Adve</a></p>
267 <!-- *********************************************************************** -->
268 <div class="doc_section"> <a name="abstract">Abstract </a></div>
269 <!-- *********************************************************************** -->
271 <div class="doc_text">
272 <p>This document is a reference manual for the LLVM assembly language.
273 LLVM is a Static Single Assignment (SSA) based representation that provides
274 type safety, low-level operations, flexibility, and the capability of
275 representing 'all' high-level languages cleanly. It is the common code
276 representation used throughout all phases of the LLVM compilation
280 <!-- *********************************************************************** -->
281 <div class="doc_section"> <a name="introduction">Introduction</a> </div>
282 <!-- *********************************************************************** -->
284 <div class="doc_text">
286 <p>The LLVM code representation is designed to be used in three
287 different forms: as an in-memory compiler IR, as an on-disk bitcode
288 representation (suitable for fast loading by a Just-In-Time compiler),
289 and as a human readable assembly language representation. This allows
290 LLVM to provide a powerful intermediate representation for efficient
291 compiler transformations and analysis, while providing a natural means
292 to debug and visualize the transformations. The three different forms
293 of LLVM are all equivalent. This document describes the human readable
294 representation and notation.</p>
296 <p>The LLVM representation aims to be light-weight and low-level
297 while being expressive, typed, and extensible at the same time. It
298 aims to be a "universal IR" of sorts, by being at a low enough level
299 that high-level ideas may be cleanly mapped to it (similar to how
300 microprocessors are "universal IR's", allowing many source languages to
301 be mapped to them). By providing type information, LLVM can be used as
302 the target of optimizations: for example, through pointer analysis, it
303 can be proven that a C automatic variable is never accessed outside of
304 the current function... allowing it to be promoted to a simple SSA
305 value instead of a memory location.</p>
309 <!-- _______________________________________________________________________ -->
310 <div class="doc_subsubsection"> <a name="wellformed">Well-Formedness</a> </div>
312 <div class="doc_text">
314 <p>It is important to note that this document describes 'well formed'
315 LLVM assembly language. There is a difference between what the parser
316 accepts and what is considered 'well formed'. For example, the
317 following instruction is syntactically okay, but not well formed:</p>
319 <div class="doc_code">
321 %x = <a href="#i_add">add</a> i32 1, %x
325 <p>...because the definition of <tt>%x</tt> does not dominate all of
326 its uses. The LLVM infrastructure provides a verification pass that may
327 be used to verify that an LLVM module is well formed. This pass is
328 automatically run by the parser after parsing input assembly and by
329 the optimizer before it outputs bitcode. The violations pointed out
330 by the verifier pass indicate bugs in transformation passes or input to
334 <!-- Describe the typesetting conventions here. -->
336 <!-- *********************************************************************** -->
337 <div class="doc_section"> <a name="identifiers">Identifiers</a> </div>
338 <!-- *********************************************************************** -->
340 <div class="doc_text">
342 <p>LLVM identifiers come in two basic types: global and local. Global
343 identifiers (functions, global variables) begin with the @ character. Local
344 identifiers (register names, types) begin with the % character. Additionally,
345 there are three different formats for identifiers, for different purposes:</p>
348 <li>Named values are represented as a string of characters with their prefix.
349 For example, %foo, @DivisionByZero, %a.really.long.identifier. The actual
350 regular expression used is '<tt>[%@][a-zA-Z$._][a-zA-Z$._0-9]*</tt>'.
351 Identifiers which require other characters in their names can be surrounded
352 with quotes. Special characters may be escaped using "\xx" where xx is the
353 ASCII code for the character in hexadecimal. In this way, any character can
354 be used in a name value, even quotes themselves.
356 <li>Unnamed values are represented as an unsigned numeric value with their
357 prefix. For example, %12, @2, %44.</li>
359 <li>Constants, which are described in a <a href="#constants">section about
360 constants</a>, below.</li>
363 <p>LLVM requires that values start with a prefix for two reasons: Compilers
364 don't need to worry about name clashes with reserved words, and the set of
365 reserved words may be expanded in the future without penalty. Additionally,
366 unnamed identifiers allow a compiler to quickly come up with a temporary
367 variable without having to avoid symbol table conflicts.</p>
369 <p>Reserved words in LLVM are very similar to reserved words in other
370 languages. There are keywords for different opcodes
371 ('<tt><a href="#i_add">add</a></tt>',
372 '<tt><a href="#i_bitcast">bitcast</a></tt>',
373 '<tt><a href="#i_ret">ret</a></tt>', etc...), for primitive type names ('<tt><a
374 href="#t_void">void</a></tt>', '<tt><a href="#t_primitive">i32</a></tt>', etc...),
375 and others. These reserved words cannot conflict with variable names, because
376 none of them start with a prefix character ('%' or '@').</p>
378 <p>Here is an example of LLVM code to multiply the integer variable
379 '<tt>%X</tt>' by 8:</p>
383 <div class="doc_code">
385 %result = <a href="#i_mul">mul</a> i32 %X, 8
389 <p>After strength reduction:</p>
391 <div class="doc_code">
393 %result = <a href="#i_shl">shl</a> i32 %X, i8 3
397 <p>And the hard way:</p>
399 <div class="doc_code">
401 <a href="#i_add">add</a> i32 %X, %X <i>; yields {i32}:%0</i>
402 <a href="#i_add">add</a> i32 %0, %0 <i>; yields {i32}:%1</i>
403 %result = <a href="#i_add">add</a> i32 %1, %1
407 <p>This last way of multiplying <tt>%X</tt> by 8 illustrates several
408 important lexical features of LLVM:</p>
412 <li>Comments are delimited with a '<tt>;</tt>' and go until the end of
415 <li>Unnamed temporaries are created when the result of a computation is not
416 assigned to a named value.</li>
418 <li>Unnamed temporaries are numbered sequentially</li>
422 <p>...and it also shows a convention that we follow in this document. When
423 demonstrating instructions, we will follow an instruction with a comment that
424 defines the type and name of value produced. Comments are shown in italic
429 <!-- *********************************************************************** -->
430 <div class="doc_section"> <a name="highlevel">High Level Structure</a> </div>
431 <!-- *********************************************************************** -->
433 <!-- ======================================================================= -->
434 <div class="doc_subsection"> <a name="modulestructure">Module Structure</a>
437 <div class="doc_text">
439 <p>LLVM programs are composed of "Module"s, each of which is a
440 translation unit of the input programs. Each module consists of
441 functions, global variables, and symbol table entries. Modules may be
442 combined together with the LLVM linker, which merges function (and
443 global variable) definitions, resolves forward declarations, and merges
444 symbol table entries. Here is an example of the "hello world" module:</p>
446 <div class="doc_code">
447 <pre><i>; Declare the string constant as a global constant...</i>
448 <a href="#identifiers">@.LC0</a> = <a href="#linkage_internal">internal</a> <a
449 href="#globalvars">constant</a> <a href="#t_array">[13 x i8]</a> c"hello world\0A\00" <i>; [13 x i8]*</i>
451 <i>; External declaration of the puts function</i>
452 <a href="#functionstructure">declare</a> i32 @puts(i8 *) <i>; i32(i8 *)* </i>
454 <i>; Definition of main function</i>
455 define i32 @main() { <i>; i32()* </i>
456 <i>; Convert [13 x i8]* to i8 *...</i>
458 href="#i_getelementptr">getelementptr</a> [13 x i8]* @.LC0, i64 0, i64 0 <i>; i8 *</i>
460 <i>; Call puts function to write out the string to stdout...</i>
462 href="#i_call">call</a> i32 @puts(i8 * %cast210) <i>; i32</i>
464 href="#i_ret">ret</a> i32 0<br>}<br>
468 <p>This example is made up of a <a href="#globalvars">global variable</a>
469 named "<tt>.LC0</tt>", an external declaration of the "<tt>puts</tt>"
470 function, and a <a href="#functionstructure">function definition</a>
471 for "<tt>main</tt>".</p>
473 <p>In general, a module is made up of a list of global values,
474 where both functions and global variables are global values. Global values are
475 represented by a pointer to a memory location (in this case, a pointer to an
476 array of char, and a pointer to a function), and have one of the following <a
477 href="#linkage">linkage types</a>.</p>
481 <!-- ======================================================================= -->
482 <div class="doc_subsection">
483 <a name="linkage">Linkage Types</a>
486 <div class="doc_text">
489 All Global Variables and Functions have one of the following types of linkage:
494 <dt><tt><b><a name="linkage_private">private</a></b></tt>: </dt>
496 <dd>Global values with private linkage are only directly accessible by
497 objects in the current module. In particular, linking code into a module with
498 an private global value may cause the private to be renamed as necessary to
499 avoid collisions. Because the symbol is private to the module, all
500 references can be updated. This doesn't show up in any symbol table in the
504 <dt><tt><b><a name="linkage_internal">internal</a></b></tt>: </dt>
506 <dd> Similar to private, but the value shows as a local symbol (STB_LOCAL in
507 the case of ELF) in the object file. This corresponds to the notion of the
508 '<tt>static</tt>' keyword in C.
511 <dt><tt><b><a name="linkage_linkonce">linkonce</a></b></tt>: </dt>
513 <dd>Globals with "<tt>linkonce</tt>" linkage are merged with other globals of
514 the same name when linkage occurs. This is typically used to implement
515 inline functions, templates, or other code which must be generated in each
516 translation unit that uses it. Unreferenced <tt>linkonce</tt> globals are
517 allowed to be discarded.
520 <dt><tt><b><a name="linkage_common">common</a></b></tt>: </dt>
522 <dd>"<tt>common</tt>" linkage is exactly the same as <tt>linkonce</tt>
523 linkage, except that unreferenced <tt>common</tt> globals may not be
524 discarded. This is used for globals that may be emitted in multiple
525 translation units, but that are not guaranteed to be emitted into every
526 translation unit that uses them. One example of this is tentative
527 definitions in C, such as "<tt>int X;</tt>" at global scope.
530 <dt><tt><b><a name="linkage_weak">weak</a></b></tt>: </dt>
532 <dd>"<tt>weak</tt>" linkage is the same as <tt>common</tt> linkage, except
533 that some targets may choose to emit different assembly sequences for them
534 for target-dependent reasons. This is used for globals that are declared
535 "weak" in C source code.
538 <dt><tt><b><a name="linkage_appending">appending</a></b></tt>: </dt>
540 <dd>"<tt>appending</tt>" linkage may only be applied to global variables of
541 pointer to array type. When two global variables with appending linkage are
542 linked together, the two global arrays are appended together. This is the
543 LLVM, typesafe, equivalent of having the system linker append together
544 "sections" with identical names when .o files are linked.
547 <dt><tt><b><a name="linkage_externweak">extern_weak</a></b></tt>: </dt>
548 <dd>The semantics of this linkage follow the ELF object file model: the
549 symbol is weak until linked, if not linked, the symbol becomes null instead
550 of being an undefined reference.
553 <dt><tt><b><a name="linkage_external">externally visible</a></b></tt>:</dt>
555 <dd>If none of the above identifiers are used, the global is externally
556 visible, meaning that it participates in linkage and can be used to resolve
557 external symbol references.
562 The next two types of linkage are targeted for Microsoft Windows platform
563 only. They are designed to support importing (exporting) symbols from (to)
564 DLLs (Dynamic Link Libraries).
568 <dt><tt><b><a name="linkage_dllimport">dllimport</a></b></tt>: </dt>
570 <dd>"<tt>dllimport</tt>" linkage causes the compiler to reference a function
571 or variable via a global pointer to a pointer that is set up by the DLL
572 exporting the symbol. On Microsoft Windows targets, the pointer name is
573 formed by combining <code>__imp_</code> and the function or variable name.
576 <dt><tt><b><a name="linkage_dllexport">dllexport</a></b></tt>: </dt>
578 <dd>"<tt>dllexport</tt>" linkage causes the compiler to provide a global
579 pointer to a pointer in a DLL, so that it can be referenced with the
580 <tt>dllimport</tt> attribute. On Microsoft Windows targets, the pointer
581 name is formed by combining <code>__imp_</code> and the function or variable
587 <p>For example, since the "<tt>.LC0</tt>"
588 variable is defined to be internal, if another module defined a "<tt>.LC0</tt>"
589 variable and was linked with this one, one of the two would be renamed,
590 preventing a collision. Since "<tt>main</tt>" and "<tt>puts</tt>" are
591 external (i.e., lacking any linkage declarations), they are accessible
592 outside of the current module.</p>
593 <p>It is illegal for a function <i>declaration</i>
594 to have any linkage type other than "externally visible", <tt>dllimport</tt>,
595 or <tt>extern_weak</tt>.</p>
596 <p>Aliases can have only <tt>external</tt>, <tt>internal</tt> and <tt>weak</tt>
600 <!-- ======================================================================= -->
601 <div class="doc_subsection">
602 <a name="callingconv">Calling Conventions</a>
605 <div class="doc_text">
607 <p>LLVM <a href="#functionstructure">functions</a>, <a href="#i_call">calls</a>
608 and <a href="#i_invoke">invokes</a> can all have an optional calling convention
609 specified for the call. The calling convention of any pair of dynamic
610 caller/callee must match, or the behavior of the program is undefined. The
611 following calling conventions are supported by LLVM, and more may be added in
615 <dt><b>"<tt>ccc</tt>" - The C calling convention</b>:</dt>
617 <dd>This calling convention (the default if no other calling convention is
618 specified) matches the target C calling conventions. This calling convention
619 supports varargs function calls and tolerates some mismatch in the declared
620 prototype and implemented declaration of the function (as does normal C).
623 <dt><b>"<tt>fastcc</tt>" - The fast calling convention</b>:</dt>
625 <dd>This calling convention attempts to make calls as fast as possible
626 (e.g. by passing things in registers). This calling convention allows the
627 target to use whatever tricks it wants to produce fast code for the target,
628 without having to conform to an externally specified ABI (Application Binary
629 Interface). Implementations of this convention should allow arbitrary
630 <a href="CodeGenerator.html#tailcallopt">tail call optimization</a> to be
631 supported. This calling convention does not support varargs and requires the
632 prototype of all callees to exactly match the prototype of the function
636 <dt><b>"<tt>coldcc</tt>" - The cold calling convention</b>:</dt>
638 <dd>This calling convention attempts to make code in the caller as efficient
639 as possible under the assumption that the call is not commonly executed. As
640 such, these calls often preserve all registers so that the call does not break
641 any live ranges in the caller side. This calling convention does not support
642 varargs and requires the prototype of all callees to exactly match the
643 prototype of the function definition.
646 <dt><b>"<tt>cc <<em>n</em>></tt>" - Numbered convention</b>:</dt>
648 <dd>Any calling convention may be specified by number, allowing
649 target-specific calling conventions to be used. Target specific calling
650 conventions start at 64.
654 <p>More calling conventions can be added/defined on an as-needed basis, to
655 support pascal conventions or any other well-known target-independent
660 <!-- ======================================================================= -->
661 <div class="doc_subsection">
662 <a name="visibility">Visibility Styles</a>
665 <div class="doc_text">
668 All Global Variables and Functions have one of the following visibility styles:
672 <dt><b>"<tt>default</tt>" - Default style</b>:</dt>
674 <dd>On targets that use the ELF object file format, default visibility means
675 that the declaration is visible to other
676 modules and, in shared libraries, means that the declared entity may be
677 overridden. On Darwin, default visibility means that the declaration is
678 visible to other modules. Default visibility corresponds to "external
679 linkage" in the language.
682 <dt><b>"<tt>hidden</tt>" - Hidden style</b>:</dt>
684 <dd>Two declarations of an object with hidden visibility refer to the same
685 object if they are in the same shared object. Usually, hidden visibility
686 indicates that the symbol will not be placed into the dynamic symbol table,
687 so no other module (executable or shared library) can reference it
691 <dt><b>"<tt>protected</tt>" - Protected style</b>:</dt>
693 <dd>On ELF, protected visibility indicates that the symbol will be placed in
694 the dynamic symbol table, but that references within the defining module will
695 bind to the local symbol. That is, the symbol cannot be overridden by another
702 <!-- ======================================================================= -->
703 <div class="doc_subsection">
704 <a name="namedtypes">Named Types</a>
707 <div class="doc_text">
709 <p>LLVM IR allows you to specify name aliases for certain types. This can make
710 it easier to read the IR and make the IR more condensed (particularly when
711 recursive types are involved). An example of a name specification is:
714 <div class="doc_code">
716 %mytype = type { %mytype*, i32 }
720 <p>You may give a name to any <a href="#typesystem">type</a> except "<a
721 href="t_void">void</a>". Type name aliases may be used anywhere a type is
722 expected with the syntax "%mytype".</p>
724 <p>Note that type names are aliases for the structural type that they indicate,
725 and that you can therefore specify multiple names for the same type. This often
726 leads to confusing behavior when dumping out a .ll file. Since LLVM IR uses
727 structural typing, the name is not part of the type. When printing out LLVM IR,
728 the printer will pick <em>one name</em> to render all types of a particular
729 shape. This means that if you have code where two different source types end up
730 having the same LLVM type, that the dumper will sometimes print the "wrong" or
731 unexpected type. This is an important design point and isn't going to
736 <!-- ======================================================================= -->
737 <div class="doc_subsection">
738 <a name="globalvars">Global Variables</a>
741 <div class="doc_text">
743 <p>Global variables define regions of memory allocated at compilation time
744 instead of run-time. Global variables may optionally be initialized, may have
745 an explicit section to be placed in, and may have an optional explicit alignment
746 specified. A variable may be defined as "thread_local", which means that it
747 will not be shared by threads (each thread will have a separated copy of the
748 variable). A variable may be defined as a global "constant," which indicates
749 that the contents of the variable will <b>never</b> be modified (enabling better
750 optimization, allowing the global data to be placed in the read-only section of
751 an executable, etc). Note that variables that need runtime initialization
752 cannot be marked "constant" as there is a store to the variable.</p>
755 LLVM explicitly allows <em>declarations</em> of global variables to be marked
756 constant, even if the final definition of the global is not. This capability
757 can be used to enable slightly better optimization of the program, but requires
758 the language definition to guarantee that optimizations based on the
759 'constantness' are valid for the translation units that do not include the
763 <p>As SSA values, global variables define pointer values that are in
764 scope (i.e. they dominate) all basic blocks in the program. Global
765 variables always define a pointer to their "content" type because they
766 describe a region of memory, and all memory objects in LLVM are
767 accessed through pointers.</p>
769 <p>A global variable may be declared to reside in a target-specifc numbered
770 address space. For targets that support them, address spaces may affect how
771 optimizations are performed and/or what target instructions are used to access
772 the variable. The default address space is zero. The address space qualifier
773 must precede any other attributes.</p>
775 <p>LLVM allows an explicit section to be specified for globals. If the target
776 supports it, it will emit globals to the section specified.</p>
778 <p>An explicit alignment may be specified for a global. If not present, or if
779 the alignment is set to zero, the alignment of the global is set by the target
780 to whatever it feels convenient. If an explicit alignment is specified, the
781 global is forced to have at least that much alignment. All alignments must be
784 <p>For example, the following defines a global in a numbered address space with
785 an initializer, section, and alignment:</p>
787 <div class="doc_code">
789 @G = addrspace(5) constant float 1.0, section "foo", align 4
796 <!-- ======================================================================= -->
797 <div class="doc_subsection">
798 <a name="functionstructure">Functions</a>
801 <div class="doc_text">
803 <p>LLVM function definitions consist of the "<tt>define</tt>" keyord,
804 an optional <a href="#linkage">linkage type</a>, an optional
805 <a href="#visibility">visibility style</a>, an optional
806 <a href="#callingconv">calling convention</a>, a return type, an optional
807 <a href="#paramattrs">parameter attribute</a> for the return type, a function
808 name, a (possibly empty) argument list (each with optional
809 <a href="#paramattrs">parameter attributes</a>), optional
810 <a href="#fnattrs">function attributes</a>, an optional section,
811 an optional alignment, an optional <a href="#gc">garbage collector name</a>,
812 an opening curly brace, a list of basic blocks, and a closing curly brace.
814 LLVM function declarations consist of the "<tt>declare</tt>" keyword, an
815 optional <a href="#linkage">linkage type</a>, an optional
816 <a href="#visibility">visibility style</a>, an optional
817 <a href="#callingconv">calling convention</a>, a return type, an optional
818 <a href="#paramattrs">parameter attribute</a> for the return type, a function
819 name, a possibly empty list of arguments, an optional alignment, and an optional
820 <a href="#gc">garbage collector name</a>.</p>
822 <p>A function definition contains a list of basic blocks, forming the CFG
823 (Control Flow Graph) for
824 the function. Each basic block may optionally start with a label (giving the
825 basic block a symbol table entry), contains a list of instructions, and ends
826 with a <a href="#terminators">terminator</a> instruction (such as a branch or
827 function return).</p>
829 <p>The first basic block in a function is special in two ways: it is immediately
830 executed on entrance to the function, and it is not allowed to have predecessor
831 basic blocks (i.e. there can not be any branches to the entry block of a
832 function). Because the block can have no predecessors, it also cannot have any
833 <a href="#i_phi">PHI nodes</a>.</p>
835 <p>LLVM allows an explicit section to be specified for functions. If the target
836 supports it, it will emit functions to the section specified.</p>
838 <p>An explicit alignment may be specified for a function. If not present, or if
839 the alignment is set to zero, the alignment of the function is set by the target
840 to whatever it feels convenient. If an explicit alignment is specified, the
841 function is forced to have at least that much alignment. All alignments must be
846 <div class="doc_code">
848 define [<a href="#linkage">linkage</a>] [<a href="#visibility">visibility</a>]
849 [<a href="#callingconv">cconv</a>] [<a href="#paramattrs">ret attrs</a>]
850 <ResultType> @<FunctionName> ([argument list])
851 [<a href="#fnattrs">fn Attrs</a>] [section "name"] [align N]
852 [<a href="#gc">gc</a>] { ... }
859 <!-- ======================================================================= -->
860 <div class="doc_subsection">
861 <a name="aliasstructure">Aliases</a>
863 <div class="doc_text">
864 <p>Aliases act as "second name" for the aliasee value (which can be either
865 function, global variable, another alias or bitcast of global value). Aliases
866 may have an optional <a href="#linkage">linkage type</a>, and an
867 optional <a href="#visibility">visibility style</a>.</p>
871 <div class="doc_code">
873 @<Name> = alias [Linkage] [Visibility] <AliaseeTy> @<Aliasee>
881 <!-- ======================================================================= -->
882 <div class="doc_subsection"><a name="paramattrs">Parameter Attributes</a></div>
883 <div class="doc_text">
884 <p>The return type and each parameter of a function type may have a set of
885 <i>parameter attributes</i> associated with them. Parameter attributes are
886 used to communicate additional information about the result or parameters of
887 a function. Parameter attributes are considered to be part of the function,
888 not of the function type, so functions with different parameter attributes
889 can have the same function type.</p>
891 <p>Parameter attributes are simple keywords that follow the type specified. If
892 multiple parameter attributes are needed, they are space separated. For
895 <div class="doc_code">
897 declare i32 @printf(i8* noalias nocapture, ...)
898 declare i32 @atoi(i8 zeroext)
899 declare signext i8 @returns_signed_char()
903 <p>Note that any attributes for the function result (<tt>nounwind</tt>,
904 <tt>readonly</tt>) come immediately after the argument list.</p>
906 <p>Currently, only the following parameter attributes are defined:</p>
908 <dt><tt>zeroext</tt></dt>
909 <dd>This indicates to the code generator that the parameter or return value
910 should be zero-extended to a 32-bit value by the caller (for a parameter)
911 or the callee (for a return value).</dd>
913 <dt><tt>signext</tt></dt>
914 <dd>This indicates to the code generator that the parameter or return value
915 should be sign-extended to a 32-bit value by the caller (for a parameter)
916 or the callee (for a return value).</dd>
918 <dt><tt>inreg</tt></dt>
919 <dd>This indicates that this parameter or return value should be treated
920 in a special target-dependent fashion during while emitting code for a
921 function call or return (usually, by putting it in a register as opposed
922 to memory, though some targets use it to distinguish between two different
923 kinds of registers). Use of this attribute is target-specific.</dd>
925 <dt><tt><a name="byval">byval</a></tt></dt>
926 <dd>This indicates that the pointer parameter should really be passed by
927 value to the function. The attribute implies that a hidden copy of the
928 pointee is made between the caller and the callee, so the callee is unable
929 to modify the value in the callee. This attribute is only valid on LLVM
930 pointer arguments. It is generally used to pass structs and arrays by
931 value, but is also valid on pointers to scalars. The copy is considered to
932 belong to the caller not the callee (for example,
933 <tt><a href="#readonly">readonly</a></tt> functions should not write to
934 <tt>byval</tt> parameters). This is not a valid attribute for return
935 values. The byval attribute also supports specifying an alignment with the
936 align attribute. This has a target-specific effect on the code generator
937 that usually indicates a desired alignment for the synthesized stack
940 <dt><tt>sret</tt></dt>
941 <dd>This indicates that the pointer parameter specifies the address of a
942 structure that is the return value of the function in the source program.
943 This pointer must be guaranteed by the caller to be valid: loads and stores
944 to the structure may be assumed by the callee to not to trap. This may only
945 be applied to the first parameter. This is not a valid attribute for
948 <dt><tt>noalias</tt></dt>
949 <dd>This indicates that the pointer does not alias any global or any other
950 parameter. The caller is responsible for ensuring that this is the
951 case. On a function return value, <tt>noalias</tt> additionally indicates
952 that the pointer does not alias any other pointers visible to the
953 caller. For further details, please see the discussion of the NoAlias
955 <a href="http://llvm.org/docs/AliasAnalysis.html#MustMayNo">alias
958 <dt><tt>nocapture</tt></dt>
959 <dd>This indicates that the callee does not make any copies of the pointer
960 that outlive the callee itself. This is not a valid attribute for return
963 <dt><tt>nest</tt></dt>
964 <dd>This indicates that the pointer parameter can be excised using the
965 <a href="#int_trampoline">trampoline intrinsics</a>. This is not a valid
966 attribute for return values.</dd>
971 <!-- ======================================================================= -->
972 <div class="doc_subsection">
973 <a name="gc">Garbage Collector Names</a>
976 <div class="doc_text">
977 <p>Each function may specify a garbage collector name, which is simply a
980 <div class="doc_code"><pre
981 >define void @f() gc "name" { ...</pre></div>
983 <p>The compiler declares the supported values of <i>name</i>. Specifying a
984 collector which will cause the compiler to alter its output in order to support
985 the named garbage collection algorithm.</p>
988 <!-- ======================================================================= -->
989 <div class="doc_subsection">
990 <a name="fnattrs">Function Attributes</a>
993 <div class="doc_text">
995 <p>Function attributes are set to communicate additional information about
996 a function. Function attributes are considered to be part of the function,
997 not of the function type, so functions with different parameter attributes
998 can have the same function type.</p>
1000 <p>Function attributes are simple keywords that follow the type specified. If
1001 multiple attributes are needed, they are space separated. For
1004 <div class="doc_code">
1006 define void @f() noinline { ... }
1007 define void @f() alwaysinline { ... }
1008 define void @f() alwaysinline optsize { ... }
1009 define void @f() optsize
1014 <dt><tt>alwaysinline</tt></dt>
1015 <dd>This attribute indicates that the inliner should attempt to inline this
1016 function into callers whenever possible, ignoring any active inlining size
1017 threshold for this caller.</dd>
1019 <dt><tt>noinline</tt></dt>
1020 <dd>This attribute indicates that the inliner should never inline this function
1021 in any situation. This attribute may not be used together with the
1022 <tt>alwaysinline</tt> attribute.</dd>
1024 <dt><tt>optsize</tt></dt>
1025 <dd>This attribute suggests that optimization passes and code generator passes
1026 make choices that keep the code size of this function low, and otherwise do
1027 optimizations specifically to reduce code size.</dd>
1029 <dt><tt>noreturn</tt></dt>
1030 <dd>This function attribute indicates that the function never returns normally.
1031 This produces undefined behavior at runtime if the function ever does
1032 dynamically return.</dd>
1034 <dt><tt>nounwind</tt></dt>
1035 <dd>This function attribute indicates that the function never returns with an
1036 unwind or exceptional control flow. If the function does unwind, its runtime
1037 behavior is undefined.</dd>
1039 <dt><tt>readnone</tt></dt>
1040 <dd>This attribute indicates that the function computes its result (or the
1041 exception it throws) based strictly on its arguments, without dereferencing any
1042 pointer arguments or otherwise accessing any mutable state (e.g. memory, control
1043 registers, etc) visible to caller functions. It does not write through any
1044 pointer arguments (including <tt><a href="#byval">byval</a></tt> arguments) and
1045 never changes any state visible to callers.</dd>
1047 <dt><tt><a name="readonly">readonly</a></tt></dt>
1048 <dd>This attribute indicates that the function does not write through any
1049 pointer arguments (including <tt><a href="#byval">byval</a></tt> arguments)
1050 or otherwise modify any state (e.g. memory, control registers, etc) visible to
1051 caller functions. It may dereference pointer arguments and read state that may
1052 be set in the caller. A readonly function always returns the same value (or
1053 throws the same exception) when called with the same set of arguments and global
1056 <dt><tt><a name="ssp">ssp</a></tt></dt>
1057 <dd>This attribute indicates that the function should emit a stack smashing
1058 protector. It is in the form of a "canary"—a random value placed on the
1059 stack before the local variables that's checked upon return from the function to
1060 see if it has been overwritten. A heuristic is used to determine if a function
1061 needs stack protectors or not.
1063 <p>If a function that has an <tt>ssp</tt> attribute is inlined into a function
1064 that doesn't have an <tt>ssp</tt> attribute, then the resulting function will
1065 have an <tt>ssp</tt> attribute.</p></dd>
1067 <dt><tt>sspreq</tt></dt>
1068 <dd>This attribute indicates that the function should <em>always</em> emit a
1069 stack smashing protector. This overrides the <tt><a href="#ssp">ssp</a></tt>
1072 <p>If a function that has an <tt>sspreq</tt> attribute is inlined into a
1073 function that doesn't have an <tt>sspreq</tt> attribute or which has
1074 an <tt>ssp</tt> attribute, then the resulting function will have
1075 an <tt>sspreq</tt> attribute.</p></dd>
1080 <!-- ======================================================================= -->
1081 <div class="doc_subsection">
1082 <a name="moduleasm">Module-Level Inline Assembly</a>
1085 <div class="doc_text">
1087 Modules may contain "module-level inline asm" blocks, which corresponds to the
1088 GCC "file scope inline asm" blocks. These blocks are internally concatenated by
1089 LLVM and treated as a single unit, but may be separated in the .ll file if
1090 desired. The syntax is very simple:
1093 <div class="doc_code">
1095 module asm "inline asm code goes here"
1096 module asm "more can go here"
1100 <p>The strings can contain any character by escaping non-printable characters.
1101 The escape sequence used is simply "\xx" where "xx" is the two digit hex code
1106 The inline asm code is simply printed to the machine code .s file when
1107 assembly code is generated.
1111 <!-- ======================================================================= -->
1112 <div class="doc_subsection">
1113 <a name="datalayout">Data Layout</a>
1116 <div class="doc_text">
1117 <p>A module may specify a target specific data layout string that specifies how
1118 data is to be laid out in memory. The syntax for the data layout is simply:</p>
1119 <pre> target datalayout = "<i>layout specification</i>"</pre>
1120 <p>The <i>layout specification</i> consists of a list of specifications
1121 separated by the minus sign character ('-'). Each specification starts with a
1122 letter and may include other information after the letter to define some
1123 aspect of the data layout. The specifications accepted are as follows: </p>
1126 <dd>Specifies that the target lays out data in big-endian form. That is, the
1127 bits with the most significance have the lowest address location.</dd>
1129 <dd>Specifies that the target lays out data in little-endian form. That is,
1130 the bits with the least significance have the lowest address location.</dd>
1131 <dt><tt>p:<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1132 <dd>This specifies the <i>size</i> of a pointer and its <i>abi</i> and
1133 <i>preferred</i> alignments. All sizes are in bits. Specifying the <i>pref</i>
1134 alignment is optional. If omitted, the preceding <tt>:</tt> should be omitted
1136 <dt><tt>i<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1137 <dd>This specifies the alignment for an integer type of a given bit
1138 <i>size</i>. The value of <i>size</i> must be in the range [1,2^23).</dd>
1139 <dt><tt>v<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1140 <dd>This specifies the alignment for a vector type of a given bit
1142 <dt><tt>f<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1143 <dd>This specifies the alignment for a floating point type of a given bit
1144 <i>size</i>. The value of <i>size</i> must be either 32 (float) or 64
1146 <dt><tt>a<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1147 <dd>This specifies the alignment for an aggregate type of a given bit
1150 <p>When constructing the data layout for a given target, LLVM starts with a
1151 default set of specifications which are then (possibly) overriden by the
1152 specifications in the <tt>datalayout</tt> keyword. The default specifications
1153 are given in this list:</p>
1155 <li><tt>E</tt> - big endian</li>
1156 <li><tt>p:32:64:64</tt> - 32-bit pointers with 64-bit alignment</li>
1157 <li><tt>i1:8:8</tt> - i1 is 8-bit (byte) aligned</li>
1158 <li><tt>i8:8:8</tt> - i8 is 8-bit (byte) aligned</li>
1159 <li><tt>i16:16:16</tt> - i16 is 16-bit aligned</li>
1160 <li><tt>i32:32:32</tt> - i32 is 32-bit aligned</li>
1161 <li><tt>i64:32:64</tt> - i64 has ABI alignment of 32-bits but preferred
1162 alignment of 64-bits</li>
1163 <li><tt>f32:32:32</tt> - float is 32-bit aligned</li>
1164 <li><tt>f64:64:64</tt> - double is 64-bit aligned</li>
1165 <li><tt>v64:64:64</tt> - 64-bit vector is 64-bit aligned</li>
1166 <li><tt>v128:128:128</tt> - 128-bit vector is 128-bit aligned</li>
1167 <li><tt>a0:0:1</tt> - aggregates are 8-bit aligned</li>
1169 <p>When LLVM is determining the alignment for a given type, it uses the
1170 following rules:</p>
1172 <li>If the type sought is an exact match for one of the specifications, that
1173 specification is used.</li>
1174 <li>If no match is found, and the type sought is an integer type, then the
1175 smallest integer type that is larger than the bitwidth of the sought type is
1176 used. If none of the specifications are larger than the bitwidth then the the
1177 largest integer type is used. For example, given the default specifications
1178 above, the i7 type will use the alignment of i8 (next largest) while both
1179 i65 and i256 will use the alignment of i64 (largest specified).</li>
1180 <li>If no match is found, and the type sought is a vector type, then the
1181 largest vector type that is smaller than the sought vector type will be used
1182 as a fall back. This happens because <128 x double> can be implemented
1183 in terms of 64 <2 x double>, for example.</li>
1187 <!-- *********************************************************************** -->
1188 <div class="doc_section"> <a name="typesystem">Type System</a> </div>
1189 <!-- *********************************************************************** -->
1191 <div class="doc_text">
1193 <p>The LLVM type system is one of the most important features of the
1194 intermediate representation. Being typed enables a number of
1195 optimizations to be performed on the intermediate representation directly,
1196 without having to do
1197 extra analyses on the side before the transformation. A strong type
1198 system makes it easier to read the generated code and enables novel
1199 analyses and transformations that are not feasible to perform on normal
1200 three address code representations.</p>
1204 <!-- ======================================================================= -->
1205 <div class="doc_subsection"> <a name="t_classifications">Type
1206 Classifications</a> </div>
1207 <div class="doc_text">
1208 <p>The types fall into a few useful
1209 classifications:</p>
1211 <table border="1" cellspacing="0" cellpadding="4">
1213 <tr><th>Classification</th><th>Types</th></tr>
1215 <td><a href="#t_integer">integer</a></td>
1216 <td><tt>i1, i2, i3, ... i8, ... i16, ... i32, ... i64, ... </tt></td>
1219 <td><a href="#t_floating">floating point</a></td>
1220 <td><tt>float, double, x86_fp80, fp128, ppc_fp128</tt></td>
1223 <td><a name="t_firstclass">first class</a></td>
1224 <td><a href="#t_integer">integer</a>,
1225 <a href="#t_floating">floating point</a>,
1226 <a href="#t_pointer">pointer</a>,
1227 <a href="#t_vector">vector</a>,
1228 <a href="#t_struct">structure</a>,
1229 <a href="#t_array">array</a>,
1230 <a href="#t_label">label</a>.
1234 <td><a href="#t_primitive">primitive</a></td>
1235 <td><a href="#t_label">label</a>,
1236 <a href="#t_void">void</a>,
1237 <a href="#t_floating">floating point</a>.</td>
1240 <td><a href="#t_derived">derived</a></td>
1241 <td><a href="#t_integer">integer</a>,
1242 <a href="#t_array">array</a>,
1243 <a href="#t_function">function</a>,
1244 <a href="#t_pointer">pointer</a>,
1245 <a href="#t_struct">structure</a>,
1246 <a href="#t_pstruct">packed structure</a>,
1247 <a href="#t_vector">vector</a>,
1248 <a href="#t_opaque">opaque</a>.
1254 <p>The <a href="#t_firstclass">first class</a> types are perhaps the
1255 most important. Values of these types are the only ones which can be
1256 produced by instructions, passed as arguments, or used as operands to
1260 <!-- ======================================================================= -->
1261 <div class="doc_subsection"> <a name="t_primitive">Primitive Types</a> </div>
1263 <div class="doc_text">
1264 <p>The primitive types are the fundamental building blocks of the LLVM
1269 <!-- _______________________________________________________________________ -->
1270 <div class="doc_subsubsection"> <a name="t_floating">Floating Point Types</a> </div>
1272 <div class="doc_text">
1275 <tr><th>Type</th><th>Description</th></tr>
1276 <tr><td><tt>float</tt></td><td>32-bit floating point value</td></tr>
1277 <tr><td><tt>double</tt></td><td>64-bit floating point value</td></tr>
1278 <tr><td><tt>fp128</tt></td><td>128-bit floating point value (112-bit mantissa)</td></tr>
1279 <tr><td><tt>x86_fp80</tt></td><td>80-bit floating point value (X87)</td></tr>
1280 <tr><td><tt>ppc_fp128</tt></td><td>128-bit floating point value (two 64-bits)</td></tr>
1285 <!-- _______________________________________________________________________ -->
1286 <div class="doc_subsubsection"> <a name="t_void">Void Type</a> </div>
1288 <div class="doc_text">
1290 <p>The void type does not represent any value and has no size.</p>
1299 <!-- _______________________________________________________________________ -->
1300 <div class="doc_subsubsection"> <a name="t_label">Label Type</a> </div>
1302 <div class="doc_text">
1304 <p>The label type represents code labels.</p>
1314 <!-- ======================================================================= -->
1315 <div class="doc_subsection"> <a name="t_derived">Derived Types</a> </div>
1317 <div class="doc_text">
1319 <p>The real power in LLVM comes from the derived types in the system.
1320 This is what allows a programmer to represent arrays, functions,
1321 pointers, and other useful types. Note that these derived types may be
1322 recursive: For example, it is possible to have a two dimensional array.</p>
1326 <!-- _______________________________________________________________________ -->
1327 <div class="doc_subsubsection"> <a name="t_integer">Integer Type</a> </div>
1329 <div class="doc_text">
1332 <p>The integer type is a very simple derived type that simply specifies an
1333 arbitrary bit width for the integer type desired. Any bit width from 1 bit to
1334 2^23-1 (about 8 million) can be specified.</p>
1342 <p>The number of bits the integer will occupy is specified by the <tt>N</tt>
1346 <table class="layout">
1349 <td><tt>i1</tt></td>
1350 <td>a single-bit integer.</td>
1352 <td><tt>i32</tt></td>
1353 <td>a 32-bit integer.</td>
1355 <td><tt>i1942652</tt></td>
1356 <td>a really big integer of over 1 million bits.</td>
1361 <p>Note that the code generator does not yet support large integer types
1362 to be used as function return types. The specific limit on how large a
1363 return type the code generator can currently handle is target-dependent;
1364 currently it's often 64 bits for 32-bit targets and 128 bits for 64-bit
1369 <!-- _______________________________________________________________________ -->
1370 <div class="doc_subsubsection"> <a name="t_array">Array Type</a> </div>
1372 <div class="doc_text">
1376 <p>The array type is a very simple derived type that arranges elements
1377 sequentially in memory. The array type requires a size (number of
1378 elements) and an underlying data type.</p>
1383 [<# elements> x <elementtype>]
1386 <p>The number of elements is a constant integer value; elementtype may
1387 be any type with a size.</p>
1390 <table class="layout">
1392 <td class="left"><tt>[40 x i32]</tt></td>
1393 <td class="left">Array of 40 32-bit integer values.</td>
1396 <td class="left"><tt>[41 x i32]</tt></td>
1397 <td class="left">Array of 41 32-bit integer values.</td>
1400 <td class="left"><tt>[4 x i8]</tt></td>
1401 <td class="left">Array of 4 8-bit integer values.</td>
1404 <p>Here are some examples of multidimensional arrays:</p>
1405 <table class="layout">
1407 <td class="left"><tt>[3 x [4 x i32]]</tt></td>
1408 <td class="left">3x4 array of 32-bit integer values.</td>
1411 <td class="left"><tt>[12 x [10 x float]]</tt></td>
1412 <td class="left">12x10 array of single precision floating point values.</td>
1415 <td class="left"><tt>[2 x [3 x [4 x i16]]]</tt></td>
1416 <td class="left">2x3x4 array of 16-bit integer values.</td>
1420 <p>Note that 'variable sized arrays' can be implemented in LLVM with a zero
1421 length array. Normally, accesses past the end of an array are undefined in
1422 LLVM (e.g. it is illegal to access the 5th element of a 3 element array).
1423 As a special case, however, zero length arrays are recognized to be variable
1424 length. This allows implementation of 'pascal style arrays' with the LLVM
1425 type "{ i32, [0 x float]}", for example.</p>
1427 <p>Note that the code generator does not yet support large aggregate types
1428 to be used as function return types. The specific limit on how large an
1429 aggregate return type the code generator can currently handle is
1430 target-dependent, and also dependent on the aggregate element types.</p>
1434 <!-- _______________________________________________________________________ -->
1435 <div class="doc_subsubsection"> <a name="t_function">Function Type</a> </div>
1436 <div class="doc_text">
1440 <p>The function type can be thought of as a function signature. It
1441 consists of a return type and a list of formal parameter types. The
1442 return type of a function type is a scalar type, a void type, or a struct type.
1443 If the return type is a struct type then all struct elements must be of first
1444 class types, and the struct must have at least one element.</p>
1449 <returntype list> (<parameter list>)
1452 <p>...where '<tt><parameter list></tt>' is a comma-separated list of type
1453 specifiers. Optionally, the parameter list may include a type <tt>...</tt>,
1454 which indicates that the function takes a variable number of arguments.
1455 Variable argument functions can access their arguments with the <a
1456 href="#int_varargs">variable argument handling intrinsic</a> functions.
1457 '<tt><returntype list></tt>' is a comma-separated list of
1458 <a href="#t_firstclass">first class</a> type specifiers.</p>
1461 <table class="layout">
1463 <td class="left"><tt>i32 (i32)</tt></td>
1464 <td class="left">function taking an <tt>i32</tt>, returning an <tt>i32</tt>
1466 </tr><tr class="layout">
1467 <td class="left"><tt>float (i16 signext, i32 *) *
1469 <td class="left"><a href="#t_pointer">Pointer</a> to a function that takes
1470 an <tt>i16</tt> that should be sign extended and a
1471 <a href="#t_pointer">pointer</a> to <tt>i32</tt>, returning
1474 </tr><tr class="layout">
1475 <td class="left"><tt>i32 (i8*, ...)</tt></td>
1476 <td class="left">A vararg function that takes at least one
1477 <a href="#t_pointer">pointer</a> to <tt>i8 </tt> (char in C),
1478 which returns an integer. This is the signature for <tt>printf</tt> in
1481 </tr><tr class="layout">
1482 <td class="left"><tt>{i32, i32} (i32)</tt></td>
1483 <td class="left">A function taking an <tt>i32</tt>, returning two
1484 <tt>i32</tt> values as an aggregate of type <tt>{ i32, i32 }</tt>
1490 <!-- _______________________________________________________________________ -->
1491 <div class="doc_subsubsection"> <a name="t_struct">Structure Type</a> </div>
1492 <div class="doc_text">
1494 <p>The structure type is used to represent a collection of data members
1495 together in memory. The packing of the field types is defined to match
1496 the ABI of the underlying processor. The elements of a structure may
1497 be any type that has a size.</p>
1498 <p>Structures are accessed using '<tt><a href="#i_load">load</a></tt>
1499 and '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a
1500 field with the '<tt><a href="#i_getelementptr">getelementptr</a></tt>'
1503 <pre> { <type list> }<br></pre>
1505 <table class="layout">
1507 <td class="left"><tt>{ i32, i32, i32 }</tt></td>
1508 <td class="left">A triple of three <tt>i32</tt> values</td>
1509 </tr><tr class="layout">
1510 <td class="left"><tt>{ float, i32 (i32) * }</tt></td>
1511 <td class="left">A pair, where the first element is a <tt>float</tt> and the
1512 second element is a <a href="#t_pointer">pointer</a> to a
1513 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
1514 an <tt>i32</tt>.</td>
1518 <p>Note that the code generator does not yet support large aggregate types
1519 to be used as function return types. The specific limit on how large an
1520 aggregate return type the code generator can currently handle is
1521 target-dependent, and also dependent on the aggregate element types.</p>
1525 <!-- _______________________________________________________________________ -->
1526 <div class="doc_subsubsection"> <a name="t_pstruct">Packed Structure Type</a>
1528 <div class="doc_text">
1530 <p>The packed structure type is used to represent a collection of data members
1531 together in memory. There is no padding between fields. Further, the alignment
1532 of a packed structure is 1 byte. The elements of a packed structure may
1533 be any type that has a size.</p>
1534 <p>Structures are accessed using '<tt><a href="#i_load">load</a></tt>
1535 and '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a
1536 field with the '<tt><a href="#i_getelementptr">getelementptr</a></tt>'
1539 <pre> < { <type list> } > <br></pre>
1541 <table class="layout">
1543 <td class="left"><tt>< { i32, i32, i32 } ></tt></td>
1544 <td class="left">A triple of three <tt>i32</tt> values</td>
1545 </tr><tr class="layout">
1547 <tt>< { float, i32 (i32)* } ></tt></td>
1548 <td class="left">A pair, where the first element is a <tt>float</tt> and the
1549 second element is a <a href="#t_pointer">pointer</a> to a
1550 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
1551 an <tt>i32</tt>.</td>
1556 <!-- _______________________________________________________________________ -->
1557 <div class="doc_subsubsection"> <a name="t_pointer">Pointer Type</a> </div>
1558 <div class="doc_text">
1560 <p>As in many languages, the pointer type represents a pointer or
1561 reference to another object, which must live in memory. Pointer types may have
1562 an optional address space attribute defining the target-specific numbered
1563 address space where the pointed-to object resides. The default address space is
1566 <p>Note that LLVM does not permit pointers to void (<tt>void*</tt>) nor does
1567 it permit pointers to labels (<tt>label*</tt>). Use <tt>i8*</tt> instead.</p>
1570 <pre> <type> *<br></pre>
1572 <table class="layout">
1574 <td class="left"><tt>[4 x i32]*</tt></td>
1575 <td class="left">A <a href="#t_pointer">pointer</a> to <a
1576 href="#t_array">array</a> of four <tt>i32</tt> values.</td>
1579 <td class="left"><tt>i32 (i32 *) *</tt></td>
1580 <td class="left"> A <a href="#t_pointer">pointer</a> to a <a
1581 href="#t_function">function</a> that takes an <tt>i32*</tt>, returning an
1585 <td class="left"><tt>i32 addrspace(5)*</tt></td>
1586 <td class="left">A <a href="#t_pointer">pointer</a> to an <tt>i32</tt> value
1587 that resides in address space #5.</td>
1592 <!-- _______________________________________________________________________ -->
1593 <div class="doc_subsubsection"> <a name="t_vector">Vector Type</a> </div>
1594 <div class="doc_text">
1598 <p>A vector type is a simple derived type that represents a vector
1599 of elements. Vector types are used when multiple primitive data
1600 are operated in parallel using a single instruction (SIMD).
1601 A vector type requires a size (number of
1602 elements) and an underlying primitive data type. Vectors must have a power
1603 of two length (1, 2, 4, 8, 16 ...). Vector types are
1604 considered <a href="#t_firstclass">first class</a>.</p>
1609 < <# elements> x <elementtype> >
1612 <p>The number of elements is a constant integer value; elementtype may
1613 be any integer or floating point type.</p>
1617 <table class="layout">
1619 <td class="left"><tt><4 x i32></tt></td>
1620 <td class="left">Vector of 4 32-bit integer values.</td>
1623 <td class="left"><tt><8 x float></tt></td>
1624 <td class="left">Vector of 8 32-bit floating-point values.</td>
1627 <td class="left"><tt><2 x i64></tt></td>
1628 <td class="left">Vector of 2 64-bit integer values.</td>
1632 <p>Note that the code generator does not yet support large vector types
1633 to be used as function return types. The specific limit on how large a
1634 vector return type codegen can currently handle is target-dependent;
1635 currently it's often a few times longer than a hardware vector register.</p>
1639 <!-- _______________________________________________________________________ -->
1640 <div class="doc_subsubsection"> <a name="t_opaque">Opaque Type</a> </div>
1641 <div class="doc_text">
1645 <p>Opaque types are used to represent unknown types in the system. This
1646 corresponds (for example) to the C notion of a forward declared structure type.
1647 In LLVM, opaque types can eventually be resolved to any type (not just a
1648 structure type).</p>
1658 <table class="layout">
1660 <td class="left"><tt>opaque</tt></td>
1661 <td class="left">An opaque type.</td>
1666 <!-- ======================================================================= -->
1667 <div class="doc_subsection">
1668 <a name="t_uprefs">Type Up-references</a>
1671 <div class="doc_text">
1674 An "up reference" allows you to refer to a lexically enclosing type without
1675 requiring it to have a name. For instance, a structure declaration may contain a
1676 pointer to any of the types it is lexically a member of. Example of up
1677 references (with their equivalent as named type declarations) include:</p>
1680 { \2 * } %x = type { %x* }
1681 { \2 }* %y = type { %y }*
1686 An up reference is needed by the asmprinter for printing out cyclic types when
1687 there is no declared name for a type in the cycle. Because the asmprinter does
1688 not want to print out an infinite type string, it needs a syntax to handle
1689 recursive types that have no names (all names are optional in llvm IR).
1698 The level is the count of the lexical type that is being referred to.
1703 <table class="layout">
1705 <td class="left"><tt>\1*</tt></td>
1706 <td class="left">Self-referential pointer.</td>
1709 <td class="left"><tt>{ { \3*, i8 }, i32 }</tt></td>
1710 <td class="left">Recursive structure where the upref refers to the out-most
1717 <!-- *********************************************************************** -->
1718 <div class="doc_section"> <a name="constants">Constants</a> </div>
1719 <!-- *********************************************************************** -->
1721 <div class="doc_text">
1723 <p>LLVM has several different basic types of constants. This section describes
1724 them all and their syntax.</p>
1728 <!-- ======================================================================= -->
1729 <div class="doc_subsection"><a name="simpleconstants">Simple Constants</a></div>
1731 <div class="doc_text">
1734 <dt><b>Boolean constants</b></dt>
1736 <dd>The two strings '<tt>true</tt>' and '<tt>false</tt>' are both valid
1737 constants of the <tt><a href="#t_primitive">i1</a></tt> type.
1740 <dt><b>Integer constants</b></dt>
1742 <dd>Standard integers (such as '4') are constants of the <a
1743 href="#t_integer">integer</a> type. Negative numbers may be used with
1747 <dt><b>Floating point constants</b></dt>
1749 <dd>Floating point constants use standard decimal notation (e.g. 123.421),
1750 exponential notation (e.g. 1.23421e+2), or a more precise hexadecimal
1751 notation (see below). The assembler requires the exact decimal value of
1752 a floating-point constant. For example, the assembler accepts 1.25 but
1753 rejects 1.3 because 1.3 is a repeating decimal in binary. Floating point
1754 constants must have a <a href="#t_floating">floating point</a> type. </dd>
1756 <dt><b>Null pointer constants</b></dt>
1758 <dd>The identifier '<tt>null</tt>' is recognized as a null pointer constant
1759 and must be of <a href="#t_pointer">pointer type</a>.</dd>
1763 <p>The one non-intuitive notation for constants is the hexadecimal form
1764 of floating point constants. For example, the form '<tt>double
1765 0x432ff973cafa8000</tt>' is equivalent to (but harder to read than) '<tt>double
1766 4.5e+15</tt>'. The only time hexadecimal floating point constants are required
1767 (and the only time that they are generated by the disassembler) is when a
1768 floating point constant must be emitted but it cannot be represented as a
1769 decimal floating point number in a reasonable number of digits. For example,
1770 NaN's, infinities, and other
1771 special values are represented in their IEEE hexadecimal format so that
1772 assembly and disassembly do not cause any bits to change in the constants.</p>
1773 <p>When using the hexadecimal form, constants of types float and double are
1774 represented using the 16-digit form shown above (which matches the IEEE754
1775 representation for double); float values must, however, be exactly representable
1776 as IEE754 single precision.
1777 Hexadecimal format is always used for long
1778 double, and there are three forms of long double. The 80-bit
1779 format used by x86 is represented as <tt>0xK</tt>
1780 followed by 20 hexadecimal digits.
1781 The 128-bit format used by PowerPC (two adjacent doubles) is represented
1782 by <tt>0xM</tt> followed by 32 hexadecimal digits. The IEEE 128-bit
1783 format is represented
1784 by <tt>0xL</tt> followed by 32 hexadecimal digits; no currently supported
1785 target uses this format. Long doubles will only work if they match
1786 the long double format on your target. All hexadecimal formats are big-endian
1787 (sign bit at the left).</p>
1790 <!-- ======================================================================= -->
1791 <div class="doc_subsection"><a name="aggregateconstants">Aggregate Constants</a>
1794 <div class="doc_text">
1795 <p>Aggregate constants arise from aggregation of simple constants
1796 and smaller aggregate constants.</p>
1799 <dt><b>Structure constants</b></dt>
1801 <dd>Structure constants are represented with notation similar to structure
1802 type definitions (a comma separated list of elements, surrounded by braces
1803 (<tt>{}</tt>)). For example: "<tt>{ i32 4, float 17.0, i32* @G }</tt>",
1804 where "<tt>@G</tt>" is declared as "<tt>@G = external global i32</tt>". Structure constants
1805 must have <a href="#t_struct">structure type</a>, and the number and
1806 types of elements must match those specified by the type.
1809 <dt><b>Array constants</b></dt>
1811 <dd>Array constants are represented with notation similar to array type
1812 definitions (a comma separated list of elements, surrounded by square brackets
1813 (<tt>[]</tt>)). For example: "<tt>[ i32 42, i32 11, i32 74 ]</tt>". Array
1814 constants must have <a href="#t_array">array type</a>, and the number and
1815 types of elements must match those specified by the type.
1818 <dt><b>Vector constants</b></dt>
1820 <dd>Vector constants are represented with notation similar to vector type
1821 definitions (a comma separated list of elements, surrounded by
1822 less-than/greater-than's (<tt><></tt>)). For example: "<tt>< i32 42,
1823 i32 11, i32 74, i32 100 ></tt>". Vector constants must have <a
1824 href="#t_vector">vector type</a>, and the number and types of elements must
1825 match those specified by the type.
1828 <dt><b>Zero initialization</b></dt>
1830 <dd>The string '<tt>zeroinitializer</tt>' can be used to zero initialize a
1831 value to zero of <em>any</em> type, including scalar and aggregate types.
1832 This is often used to avoid having to print large zero initializers (e.g. for
1833 large arrays) and is always exactly equivalent to using explicit zero
1840 <!-- ======================================================================= -->
1841 <div class="doc_subsection">
1842 <a name="globalconstants">Global Variable and Function Addresses</a>
1845 <div class="doc_text">
1847 <p>The addresses of <a href="#globalvars">global variables</a> and <a
1848 href="#functionstructure">functions</a> are always implicitly valid (link-time)
1849 constants. These constants are explicitly referenced when the <a
1850 href="#identifiers">identifier for the global</a> is used and always have <a
1851 href="#t_pointer">pointer</a> type. For example, the following is a legal LLVM
1854 <div class="doc_code">
1858 @Z = global [2 x i32*] [ i32* @X, i32* @Y ]
1864 <!-- ======================================================================= -->
1865 <div class="doc_subsection"><a name="undefvalues">Undefined Values</a></div>
1866 <div class="doc_text">
1867 <p>The string '<tt>undef</tt>' is recognized as a type-less constant that has
1868 no specific value. Undefined values may be of any type and be used anywhere
1869 a constant is permitted.</p>
1871 <p>Undefined values indicate to the compiler that the program is well defined
1872 no matter what value is used, giving the compiler more freedom to optimize.
1876 <!-- ======================================================================= -->
1877 <div class="doc_subsection"><a name="constantexprs">Constant Expressions</a>
1880 <div class="doc_text">
1882 <p>Constant expressions are used to allow expressions involving other constants
1883 to be used as constants. Constant expressions may be of any <a
1884 href="#t_firstclass">first class</a> type and may involve any LLVM operation
1885 that does not have side effects (e.g. load and call are not supported). The
1886 following is the syntax for constant expressions:</p>
1889 <dt><b><tt>trunc ( CST to TYPE )</tt></b></dt>
1890 <dd>Truncate a constant to another type. The bit size of CST must be larger
1891 than the bit size of TYPE. Both types must be integers.</dd>
1893 <dt><b><tt>zext ( CST to TYPE )</tt></b></dt>
1894 <dd>Zero extend a constant to another type. The bit size of CST must be
1895 smaller or equal to the bit size of TYPE. Both types must be integers.</dd>
1897 <dt><b><tt>sext ( CST to TYPE )</tt></b></dt>
1898 <dd>Sign extend a constant to another type. The bit size of CST must be
1899 smaller or equal to the bit size of TYPE. Both types must be integers.</dd>
1901 <dt><b><tt>fptrunc ( CST to TYPE )</tt></b></dt>
1902 <dd>Truncate a floating point constant to another floating point type. The
1903 size of CST must be larger than the size of TYPE. Both types must be
1904 floating point.</dd>
1906 <dt><b><tt>fpext ( CST to TYPE )</tt></b></dt>
1907 <dd>Floating point extend a constant to another type. The size of CST must be
1908 smaller or equal to the size of TYPE. Both types must be floating point.</dd>
1910 <dt><b><tt>fptoui ( CST to TYPE )</tt></b></dt>
1911 <dd>Convert a floating point constant to the corresponding unsigned integer
1912 constant. TYPE must be a scalar or vector integer type. CST must be of scalar
1913 or vector floating point type. Both CST and TYPE must be scalars, or vectors
1914 of the same number of elements. If the value won't fit in the integer type,
1915 the results are undefined.</dd>
1917 <dt><b><tt>fptosi ( CST to TYPE )</tt></b></dt>
1918 <dd>Convert a floating point constant to the corresponding signed integer
1919 constant. TYPE must be a scalar or vector integer type. CST must be of scalar
1920 or vector floating point type. Both CST and TYPE must be scalars, or vectors
1921 of the same number of elements. If the value won't fit in the integer type,
1922 the results are undefined.</dd>
1924 <dt><b><tt>uitofp ( CST to TYPE )</tt></b></dt>
1925 <dd>Convert an unsigned integer constant to the corresponding floating point
1926 constant. TYPE must be a scalar or vector floating point type. CST must be of
1927 scalar or vector integer type. Both CST and TYPE must be scalars, or vectors
1928 of the same number of elements. If the value won't fit in the floating point
1929 type, the results are undefined.</dd>
1931 <dt><b><tt>sitofp ( CST to TYPE )</tt></b></dt>
1932 <dd>Convert a signed integer constant to the corresponding floating point
1933 constant. TYPE must be a scalar or vector floating point type. CST must be of
1934 scalar or vector integer type. Both CST and TYPE must be scalars, or vectors
1935 of the same number of elements. If the value won't fit in the floating point
1936 type, the results are undefined.</dd>
1938 <dt><b><tt>ptrtoint ( CST to TYPE )</tt></b></dt>
1939 <dd>Convert a pointer typed constant to the corresponding integer constant
1940 TYPE must be an integer type. CST must be of pointer type. The CST value is
1941 zero extended, truncated, or unchanged to make it fit in TYPE.</dd>
1943 <dt><b><tt>inttoptr ( CST to TYPE )</tt></b></dt>
1944 <dd>Convert a integer constant to a pointer constant. TYPE must be a
1945 pointer type. CST must be of integer type. The CST value is zero extended,
1946 truncated, or unchanged to make it fit in a pointer size. This one is
1947 <i>really</i> dangerous!</dd>
1949 <dt><b><tt>bitcast ( CST to TYPE )</tt></b></dt>
1950 <dd>Convert a constant, CST, to another TYPE. The constraints of the operands
1951 are the same as those for the <a href="#i_bitcast">bitcast
1952 instruction</a>.</dd>
1954 <dt><b><tt>getelementptr ( CSTPTR, IDX0, IDX1, ... )</tt></b></dt>
1956 <dd>Perform the <a href="#i_getelementptr">getelementptr operation</a> on
1957 constants. As with the <a href="#i_getelementptr">getelementptr</a>
1958 instruction, the index list may have zero or more indexes, which are required
1959 to make sense for the type of "CSTPTR".</dd>
1961 <dt><b><tt>select ( COND, VAL1, VAL2 )</tt></b></dt>
1963 <dd>Perform the <a href="#i_select">select operation</a> on
1966 <dt><b><tt>icmp COND ( VAL1, VAL2 )</tt></b></dt>
1967 <dd>Performs the <a href="#i_icmp">icmp operation</a> on constants.</dd>
1969 <dt><b><tt>fcmp COND ( VAL1, VAL2 )</tt></b></dt>
1970 <dd>Performs the <a href="#i_fcmp">fcmp operation</a> on constants.</dd>
1972 <dt><b><tt>vicmp COND ( VAL1, VAL2 )</tt></b></dt>
1973 <dd>Performs the <a href="#i_vicmp">vicmp operation</a> on constants.</dd>
1975 <dt><b><tt>vfcmp COND ( VAL1, VAL2 )</tt></b></dt>
1976 <dd>Performs the <a href="#i_vfcmp">vfcmp operation</a> on constants.</dd>
1978 <dt><b><tt>extractelement ( VAL, IDX )</tt></b></dt>
1980 <dd>Perform the <a href="#i_extractelement">extractelement
1981 operation</a> on constants.</dd>
1983 <dt><b><tt>insertelement ( VAL, ELT, IDX )</tt></b></dt>
1985 <dd>Perform the <a href="#i_insertelement">insertelement
1986 operation</a> on constants.</dd>
1989 <dt><b><tt>shufflevector ( VEC1, VEC2, IDXMASK )</tt></b></dt>
1991 <dd>Perform the <a href="#i_shufflevector">shufflevector
1992 operation</a> on constants.</dd>
1994 <dt><b><tt>OPCODE ( LHS, RHS )</tt></b></dt>
1996 <dd>Perform the specified operation of the LHS and RHS constants. OPCODE may
1997 be any of the <a href="#binaryops">binary</a> or <a href="#bitwiseops">bitwise
1998 binary</a> operations. The constraints on operands are the same as those for
1999 the corresponding instruction (e.g. no bitwise operations on floating point
2000 values are allowed).</dd>
2004 <!-- *********************************************************************** -->
2005 <div class="doc_section"> <a name="othervalues">Other Values</a> </div>
2006 <!-- *********************************************************************** -->
2008 <!-- ======================================================================= -->
2009 <div class="doc_subsection">
2010 <a name="inlineasm">Inline Assembler Expressions</a>
2013 <div class="doc_text">
2016 LLVM supports inline assembler expressions (as opposed to <a href="#moduleasm">
2017 Module-Level Inline Assembly</a>) through the use of a special value. This
2018 value represents the inline assembler as a string (containing the instructions
2019 to emit), a list of operand constraints (stored as a string), and a flag that
2020 indicates whether or not the inline asm expression has side effects. An example
2021 inline assembler expression is:
2024 <div class="doc_code">
2026 i32 (i32) asm "bswap $0", "=r,r"
2031 Inline assembler expressions may <b>only</b> be used as the callee operand of
2032 a <a href="#i_call"><tt>call</tt> instruction</a>. Thus, typically we have:
2035 <div class="doc_code">
2037 %X = call i32 asm "<a href="#int_bswap">bswap</a> $0", "=r,r"(i32 %Y)
2042 Inline asms with side effects not visible in the constraint list must be marked
2043 as having side effects. This is done through the use of the
2044 '<tt>sideeffect</tt>' keyword, like so:
2047 <div class="doc_code">
2049 call void asm sideeffect "eieio", ""()
2053 <p>TODO: The format of the asm and constraints string still need to be
2054 documented here. Constraints on what can be done (e.g. duplication, moving, etc
2055 need to be documented). This is probably best done by reference to another
2056 document that covers inline asm from a holistic perspective.
2061 <!-- *********************************************************************** -->
2062 <div class="doc_section"> <a name="instref">Instruction Reference</a> </div>
2063 <!-- *********************************************************************** -->
2065 <div class="doc_text">
2067 <p>The LLVM instruction set consists of several different
2068 classifications of instructions: <a href="#terminators">terminator
2069 instructions</a>, <a href="#binaryops">binary instructions</a>,
2070 <a href="#bitwiseops">bitwise binary instructions</a>, <a
2071 href="#memoryops">memory instructions</a>, and <a href="#otherops">other
2072 instructions</a>.</p>
2076 <!-- ======================================================================= -->
2077 <div class="doc_subsection"> <a name="terminators">Terminator
2078 Instructions</a> </div>
2080 <div class="doc_text">
2082 <p>As mentioned <a href="#functionstructure">previously</a>, every
2083 basic block in a program ends with a "Terminator" instruction, which
2084 indicates which block should be executed after the current block is
2085 finished. These terminator instructions typically yield a '<tt>void</tt>'
2086 value: they produce control flow, not values (the one exception being
2087 the '<a href="#i_invoke"><tt>invoke</tt></a>' instruction).</p>
2088 <p>There are six different terminator instructions: the '<a
2089 href="#i_ret"><tt>ret</tt></a>' instruction, the '<a href="#i_br"><tt>br</tt></a>'
2090 instruction, the '<a href="#i_switch"><tt>switch</tt></a>' instruction,
2091 the '<a href="#i_invoke"><tt>invoke</tt></a>' instruction, the '<a
2092 href="#i_unwind"><tt>unwind</tt></a>' instruction, and the '<a
2093 href="#i_unreachable"><tt>unreachable</tt></a>' instruction.</p>
2097 <!-- _______________________________________________________________________ -->
2098 <div class="doc_subsubsection"> <a name="i_ret">'<tt>ret</tt>'
2099 Instruction</a> </div>
2100 <div class="doc_text">
2103 ret <type> <value> <i>; Return a value from a non-void function</i>
2104 ret void <i>; Return from void function</i>
2109 <p>The '<tt>ret</tt>' instruction is used to return control flow (and
2110 optionally a value) from a function back to the caller.</p>
2111 <p>There are two forms of the '<tt>ret</tt>' instruction: one that
2112 returns a value and then causes control flow, and one that just causes
2113 control flow to occur.</p>
2117 <p>The '<tt>ret</tt>' instruction optionally accepts a single argument,
2118 the return value. The type of the return value must be a
2119 '<a href="#t_firstclass">first class</a>' type.</p>
2121 <p>A function is not <a href="#wellformed">well formed</a> if
2122 it it has a non-void return type and contains a '<tt>ret</tt>'
2123 instruction with no return value or a return value with a type that
2124 does not match its type, or if it has a void return type and contains
2125 a '<tt>ret</tt>' instruction with a return value.</p>
2129 <p>When the '<tt>ret</tt>' instruction is executed, control flow
2130 returns back to the calling function's context. If the caller is a "<a
2131 href="#i_call"><tt>call</tt></a>" instruction, execution continues at
2132 the instruction after the call. If the caller was an "<a
2133 href="#i_invoke"><tt>invoke</tt></a>" instruction, execution continues
2134 at the beginning of the "normal" destination block. If the instruction
2135 returns a value, that value shall set the call or invoke instruction's
2141 ret i32 5 <i>; Return an integer value of 5</i>
2142 ret void <i>; Return from a void function</i>
2143 ret { i32, i8 } { i32 4, i8 2 } <i>; Return an aggregate of values 4 and 2</i>
2146 <p>Note that the code generator does not yet fully support large
2147 return values. The specific sizes that are currently supported are
2148 dependent on the target. For integers, on 32-bit targets the limit
2149 is often 64 bits, and on 64-bit targets the limit is often 128 bits.
2150 For aggregate types, the current limits are dependent on the element
2151 types; for example targets are often limited to 2 total integer
2152 elements and 2 total floating-point elements.</p>
2155 <!-- _______________________________________________________________________ -->
2156 <div class="doc_subsubsection"> <a name="i_br">'<tt>br</tt>' Instruction</a> </div>
2157 <div class="doc_text">
2159 <pre> br i1 <cond>, label <iftrue>, label <iffalse><br> br label <dest> <i>; Unconditional branch</i>
2162 <p>The '<tt>br</tt>' instruction is used to cause control flow to
2163 transfer to a different basic block in the current function. There are
2164 two forms of this instruction, corresponding to a conditional branch
2165 and an unconditional branch.</p>
2167 <p>The conditional branch form of the '<tt>br</tt>' instruction takes a
2168 single '<tt>i1</tt>' value and two '<tt>label</tt>' values. The
2169 unconditional form of the '<tt>br</tt>' instruction takes a single
2170 '<tt>label</tt>' value as a target.</p>
2172 <p>Upon execution of a conditional '<tt>br</tt>' instruction, the '<tt>i1</tt>'
2173 argument is evaluated. If the value is <tt>true</tt>, control flows
2174 to the '<tt>iftrue</tt>' <tt>label</tt> argument. If "cond" is <tt>false</tt>,
2175 control flows to the '<tt>iffalse</tt>' <tt>label</tt> argument.</p>
2177 <pre>Test:<br> %cond = <a href="#i_icmp">icmp</a> eq, i32 %a, %b<br> br i1 %cond, label %IfEqual, label %IfUnequal<br>IfEqual:<br> <a
2178 href="#i_ret">ret</a> i32 1<br>IfUnequal:<br> <a href="#i_ret">ret</a> i32 0<br></pre>
2180 <!-- _______________________________________________________________________ -->
2181 <div class="doc_subsubsection">
2182 <a name="i_switch">'<tt>switch</tt>' Instruction</a>
2185 <div class="doc_text">
2189 switch <intty> <value>, label <defaultdest> [ <intty> <val>, label <dest> ... ]
2194 <p>The '<tt>switch</tt>' instruction is used to transfer control flow to one of
2195 several different places. It is a generalization of the '<tt>br</tt>'
2196 instruction, allowing a branch to occur to one of many possible
2202 <p>The '<tt>switch</tt>' instruction uses three parameters: an integer
2203 comparison value '<tt>value</tt>', a default '<tt>label</tt>' destination, and
2204 an array of pairs of comparison value constants and '<tt>label</tt>'s. The
2205 table is not allowed to contain duplicate constant entries.</p>
2209 <p>The <tt>switch</tt> instruction specifies a table of values and
2210 destinations. When the '<tt>switch</tt>' instruction is executed, this
2211 table is searched for the given value. If the value is found, control flow is
2212 transfered to the corresponding destination; otherwise, control flow is
2213 transfered to the default destination.</p>
2215 <h5>Implementation:</h5>
2217 <p>Depending on properties of the target machine and the particular
2218 <tt>switch</tt> instruction, this instruction may be code generated in different
2219 ways. For example, it could be generated as a series of chained conditional
2220 branches or with a lookup table.</p>
2225 <i>; Emulate a conditional br instruction</i>
2226 %Val = <a href="#i_zext">zext</a> i1 %value to i32
2227 switch i32 %Val, label %truedest [ i32 0, label %falsedest ]
2229 <i>; Emulate an unconditional br instruction</i>
2230 switch i32 0, label %dest [ ]
2232 <i>; Implement a jump table:</i>
2233 switch i32 %val, label %otherwise [ i32 0, label %onzero
2235 i32 2, label %ontwo ]
2239 <!-- _______________________________________________________________________ -->
2240 <div class="doc_subsubsection">
2241 <a name="i_invoke">'<tt>invoke</tt>' Instruction</a>
2244 <div class="doc_text">
2249 <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>]
2250 to label <normal label> unwind label <exception label>
2255 <p>The '<tt>invoke</tt>' instruction causes control to transfer to a specified
2256 function, with the possibility of control flow transfer to either the
2257 '<tt>normal</tt>' label or the
2258 '<tt>exception</tt>' label. If the callee function returns with the
2259 "<tt><a href="#i_ret">ret</a></tt>" instruction, control flow will return to the
2260 "normal" label. If the callee (or any indirect callees) returns with the "<a
2261 href="#i_unwind"><tt>unwind</tt></a>" instruction, control is interrupted and
2262 continued at the dynamically nearest "exception" label.</p>
2266 <p>This instruction requires several arguments:</p>
2270 The optional "cconv" marker indicates which <a href="#callingconv">calling
2271 convention</a> the call should use. If none is specified, the call defaults
2272 to using C calling conventions.
2275 <li>The optional <a href="#paramattrs">Parameter Attributes</a> list for
2276 return values. Only '<tt>zeroext</tt>', '<tt>signext</tt>',
2277 and '<tt>inreg</tt>' attributes are valid here.</li>
2279 <li>'<tt>ptr to function ty</tt>': shall be the signature of the pointer to
2280 function value being invoked. In most cases, this is a direct function
2281 invocation, but indirect <tt>invoke</tt>s are just as possible, branching off
2282 an arbitrary pointer to function value.
2285 <li>'<tt>function ptr val</tt>': An LLVM value containing a pointer to a
2286 function to be invoked. </li>
2288 <li>'<tt>function args</tt>': argument list whose types match the function
2289 signature argument types. If the function signature indicates the function
2290 accepts a variable number of arguments, the extra arguments can be
2293 <li>'<tt>normal label</tt>': the label reached when the called function
2294 executes a '<tt><a href="#i_ret">ret</a></tt>' instruction. </li>
2296 <li>'<tt>exception label</tt>': the label reached when a callee returns with
2297 the <a href="#i_unwind"><tt>unwind</tt></a> instruction. </li>
2299 <li>The optional <a href="#fnattrs">function attributes</a> list. Only
2300 '<tt>noreturn</tt>', '<tt>nounwind</tt>', '<tt>readonly</tt>' and
2301 '<tt>readnone</tt>' attributes are valid here.</li>
2306 <p>This instruction is designed to operate as a standard '<tt><a
2307 href="#i_call">call</a></tt>' instruction in most regards. The primary
2308 difference is that it establishes an association with a label, which is used by
2309 the runtime library to unwind the stack.</p>
2311 <p>This instruction is used in languages with destructors to ensure that proper
2312 cleanup is performed in the case of either a <tt>longjmp</tt> or a thrown
2313 exception. Additionally, this is important for implementation of
2314 '<tt>catch</tt>' clauses in high-level languages that support them.</p>
2318 %retval = invoke i32 @Test(i32 15) to label %Continue
2319 unwind label %TestCleanup <i>; {i32}:retval set</i>
2320 %retval = invoke <a href="#callingconv">coldcc</a> i32 %Testfnptr(i32 15) to label %Continue
2321 unwind label %TestCleanup <i>; {i32}:retval set</i>
2326 <!-- _______________________________________________________________________ -->
2328 <div class="doc_subsubsection"> <a name="i_unwind">'<tt>unwind</tt>'
2329 Instruction</a> </div>
2331 <div class="doc_text">
2340 <p>The '<tt>unwind</tt>' instruction unwinds the stack, continuing control flow
2341 at the first callee in the dynamic call stack which used an <a
2342 href="#i_invoke"><tt>invoke</tt></a> instruction to perform the call. This is
2343 primarily used to implement exception handling.</p>
2347 <p>The '<tt>unwind</tt>' instruction causes execution of the current function to
2348 immediately halt. The dynamic call stack is then searched for the first <a
2349 href="#i_invoke"><tt>invoke</tt></a> instruction on the call stack. Once found,
2350 execution continues at the "exceptional" destination block specified by the
2351 <tt>invoke</tt> instruction. If there is no <tt>invoke</tt> instruction in the
2352 dynamic call chain, undefined behavior results.</p>
2355 <!-- _______________________________________________________________________ -->
2357 <div class="doc_subsubsection"> <a name="i_unreachable">'<tt>unreachable</tt>'
2358 Instruction</a> </div>
2360 <div class="doc_text">
2369 <p>The '<tt>unreachable</tt>' instruction has no defined semantics. This
2370 instruction is used to inform the optimizer that a particular portion of the
2371 code is not reachable. This can be used to indicate that the code after a
2372 no-return function cannot be reached, and other facts.</p>
2376 <p>The '<tt>unreachable</tt>' instruction has no defined semantics.</p>
2381 <!-- ======================================================================= -->
2382 <div class="doc_subsection"> <a name="binaryops">Binary Operations</a> </div>
2383 <div class="doc_text">
2384 <p>Binary operators are used to do most of the computation in a
2385 program. They require two operands of the same type, execute an operation on them, and
2386 produce a single value. The operands might represent
2387 multiple data, as is the case with the <a href="#t_vector">vector</a> data type.
2388 The result value has the same type as its operands.</p>
2389 <p>There are several different binary operators:</p>
2391 <!-- _______________________________________________________________________ -->
2392 <div class="doc_subsubsection">
2393 <a name="i_add">'<tt>add</tt>' Instruction</a>
2396 <div class="doc_text">
2401 <result> = add <ty> <op1>, <op2> <i>; yields {ty}:result</i>
2406 <p>The '<tt>add</tt>' instruction returns the sum of its two operands.</p>
2410 <p>The two arguments to the '<tt>add</tt>' instruction must be <a
2411 href="#t_integer">integer</a>, <a href="#t_floating">floating point</a>, or
2412 <a href="#t_vector">vector</a> values. Both arguments must have identical
2417 <p>The value produced is the integer or floating point sum of the two
2420 <p>If an integer sum has unsigned overflow, the result returned is the
2421 mathematical result modulo 2<sup>n</sup>, where n is the bit width of
2424 <p>Because LLVM integers use a two's complement representation, this
2425 instruction is appropriate for both signed and unsigned integers.</p>
2430 <result> = add i32 4, %var <i>; yields {i32}:result = 4 + %var</i>
2433 <!-- _______________________________________________________________________ -->
2434 <div class="doc_subsubsection">
2435 <a name="i_sub">'<tt>sub</tt>' Instruction</a>
2438 <div class="doc_text">
2443 <result> = sub <ty> <op1>, <op2> <i>; yields {ty}:result</i>
2448 <p>The '<tt>sub</tt>' instruction returns the difference of its two
2451 <p>Note that the '<tt>sub</tt>' instruction is used to represent the
2452 '<tt>neg</tt>' instruction present in most other intermediate
2453 representations.</p>
2457 <p>The two arguments to the '<tt>sub</tt>' instruction must be <a
2458 href="#t_integer">integer</a>, <a href="#t_floating">floating point</a>,
2459 or <a href="#t_vector">vector</a> values. Both arguments must have identical
2464 <p>The value produced is the integer or floating point difference of
2465 the two operands.</p>
2467 <p>If an integer difference has unsigned overflow, the result returned is the
2468 mathematical result modulo 2<sup>n</sup>, where n is the bit width of
2471 <p>Because LLVM integers use a two's complement representation, this
2472 instruction is appropriate for both signed and unsigned integers.</p>
2476 <result> = sub i32 4, %var <i>; yields {i32}:result = 4 - %var</i>
2477 <result> = sub i32 0, %val <i>; yields {i32}:result = -%var</i>
2481 <!-- _______________________________________________________________________ -->
2482 <div class="doc_subsubsection">
2483 <a name="i_mul">'<tt>mul</tt>' Instruction</a>
2486 <div class="doc_text">
2489 <pre> <result> = mul <ty> <op1>, <op2> <i>; yields {ty}:result</i>
2492 <p>The '<tt>mul</tt>' instruction returns the product of its two
2497 <p>The two arguments to the '<tt>mul</tt>' instruction must be <a
2498 href="#t_integer">integer</a>, <a href="#t_floating">floating point</a>,
2499 or <a href="#t_vector">vector</a> values. Both arguments must have identical
2504 <p>The value produced is the integer or floating point product of the
2507 <p>If the result of an integer multiplication has unsigned overflow,
2508 the result returned is the mathematical result modulo
2509 2<sup>n</sup>, where n is the bit width of the result.</p>
2510 <p>Because LLVM integers use a two's complement representation, and the
2511 result is the same width as the operands, this instruction returns the
2512 correct result for both signed and unsigned integers. If a full product
2513 (e.g. <tt>i32</tt>x<tt>i32</tt>-><tt>i64</tt>) is needed, the operands
2514 should be sign-extended or zero-extended as appropriate to the
2515 width of the full product.</p>
2517 <pre> <result> = mul i32 4, %var <i>; yields {i32}:result = 4 * %var</i>
2521 <!-- _______________________________________________________________________ -->
2522 <div class="doc_subsubsection"> <a name="i_udiv">'<tt>udiv</tt>' Instruction
2524 <div class="doc_text">
2526 <pre> <result> = udiv <ty> <op1>, <op2> <i>; yields {ty}:result</i>
2529 <p>The '<tt>udiv</tt>' instruction returns the quotient of its two
2534 <p>The two arguments to the '<tt>udiv</tt>' instruction must be
2535 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2536 values. Both arguments must have identical types.</p>
2540 <p>The value produced is the unsigned integer quotient of the two operands.</p>
2541 <p>Note that unsigned integer division and signed integer division are distinct
2542 operations; for signed integer division, use '<tt>sdiv</tt>'.</p>
2543 <p>Division by zero leads to undefined behavior.</p>
2545 <pre> <result> = udiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
2548 <!-- _______________________________________________________________________ -->
2549 <div class="doc_subsubsection"> <a name="i_sdiv">'<tt>sdiv</tt>' Instruction
2551 <div class="doc_text">
2554 <result> = sdiv <ty> <op1>, <op2> <i>; yields {ty}:result</i>
2559 <p>The '<tt>sdiv</tt>' instruction returns the quotient of its two
2564 <p>The two arguments to the '<tt>sdiv</tt>' instruction must be
2565 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2566 values. Both arguments must have identical types.</p>
2569 <p>The value produced is the signed integer quotient of the two operands rounded towards zero.</p>
2570 <p>Note that signed integer division and unsigned integer division are distinct
2571 operations; for unsigned integer division, use '<tt>udiv</tt>'.</p>
2572 <p>Division by zero leads to undefined behavior. Overflow also leads to
2573 undefined behavior; this is a rare case, but can occur, for example,
2574 by doing a 32-bit division of -2147483648 by -1.</p>
2576 <pre> <result> = sdiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
2579 <!-- _______________________________________________________________________ -->
2580 <div class="doc_subsubsection"> <a name="i_fdiv">'<tt>fdiv</tt>'
2581 Instruction</a> </div>
2582 <div class="doc_text">
2585 <result> = fdiv <ty> <op1>, <op2> <i>; yields {ty}:result</i>
2589 <p>The '<tt>fdiv</tt>' instruction returns the quotient of its two
2594 <p>The two arguments to the '<tt>fdiv</tt>' instruction must be
2595 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a>
2596 of floating point values. Both arguments must have identical types.</p>
2600 <p>The value produced is the floating point quotient of the two operands.</p>
2605 <result> = fdiv float 4.0, %var <i>; yields {float}:result = 4.0 / %var</i>
2609 <!-- _______________________________________________________________________ -->
2610 <div class="doc_subsubsection"> <a name="i_urem">'<tt>urem</tt>' Instruction</a>
2612 <div class="doc_text">
2614 <pre> <result> = urem <ty> <op1>, <op2> <i>; yields {ty}:result</i>
2617 <p>The '<tt>urem</tt>' instruction returns the remainder from the
2618 unsigned division of its two arguments.</p>
2620 <p>The two arguments to the '<tt>urem</tt>' instruction must be
2621 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2622 values. Both arguments must have identical types.</p>
2624 <p>This instruction returns the unsigned integer <i>remainder</i> of a division.
2625 This instruction always performs an unsigned division to get the remainder.</p>
2626 <p>Note that unsigned integer remainder and signed integer remainder are
2627 distinct operations; for signed integer remainder, use '<tt>srem</tt>'.</p>
2628 <p>Taking the remainder of a division by zero leads to undefined behavior.</p>
2630 <pre> <result> = urem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
2634 <!-- _______________________________________________________________________ -->
2635 <div class="doc_subsubsection">
2636 <a name="i_srem">'<tt>srem</tt>' Instruction</a>
2639 <div class="doc_text">
2644 <result> = srem <ty> <op1>, <op2> <i>; yields {ty}:result</i>
2649 <p>The '<tt>srem</tt>' instruction returns the remainder from the
2650 signed division of its two operands. This instruction can also take
2651 <a href="#t_vector">vector</a> versions of the values in which case
2652 the elements must be integers.</p>
2656 <p>The two arguments to the '<tt>srem</tt>' instruction must be
2657 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2658 values. Both arguments must have identical types.</p>
2662 <p>This instruction returns the <i>remainder</i> of a division (where the result
2663 has the same sign as the dividend, <tt>op1</tt>), not the <i>modulo</i>
2664 operator (where the result has the same sign as the divisor, <tt>op2</tt>) of
2665 a value. For more information about the difference, see <a
2666 href="http://mathforum.org/dr.math/problems/anne.4.28.99.html">The
2667 Math Forum</a>. For a table of how this is implemented in various languages,
2668 please see <a href="http://en.wikipedia.org/wiki/Modulo_operation">
2669 Wikipedia: modulo operation</a>.</p>
2670 <p>Note that signed integer remainder and unsigned integer remainder are
2671 distinct operations; for unsigned integer remainder, use '<tt>urem</tt>'.</p>
2672 <p>Taking the remainder of a division by zero leads to undefined behavior.
2673 Overflow also leads to undefined behavior; this is a rare case, but can occur,
2674 for example, by taking the remainder of a 32-bit division of -2147483648 by -1.
2675 (The remainder doesn't actually overflow, but this rule lets srem be
2676 implemented using instructions that return both the result of the division
2677 and the remainder.)</p>
2679 <pre> <result> = srem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
2683 <!-- _______________________________________________________________________ -->
2684 <div class="doc_subsubsection">
2685 <a name="i_frem">'<tt>frem</tt>' Instruction</a> </div>
2687 <div class="doc_text">
2690 <pre> <result> = frem <ty> <op1>, <op2> <i>; yields {ty}:result</i>
2693 <p>The '<tt>frem</tt>' instruction returns the remainder from the
2694 division of its two operands.</p>
2696 <p>The two arguments to the '<tt>frem</tt>' instruction must be
2697 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a>
2698 of floating point values. Both arguments must have identical types.</p>
2702 <p>This instruction returns the <i>remainder</i> of a division.
2703 The remainder has the same sign as the dividend.</p>
2708 <result> = frem float 4.0, %var <i>; yields {float}:result = 4.0 % %var</i>
2712 <!-- ======================================================================= -->
2713 <div class="doc_subsection"> <a name="bitwiseops">Bitwise Binary
2714 Operations</a> </div>
2715 <div class="doc_text">
2716 <p>Bitwise binary operators are used to do various forms of
2717 bit-twiddling in a program. They are generally very efficient
2718 instructions and can commonly be strength reduced from other
2719 instructions. They require two operands of the same type, execute an operation on them,
2720 and produce a single value. The resulting value is the same type as its operands.</p>
2723 <!-- _______________________________________________________________________ -->
2724 <div class="doc_subsubsection"> <a name="i_shl">'<tt>shl</tt>'
2725 Instruction</a> </div>
2726 <div class="doc_text">
2728 <pre> <result> = shl <ty> <op1>, <op2> <i>; yields {ty}:result</i>
2733 <p>The '<tt>shl</tt>' instruction returns the first operand shifted to
2734 the left a specified number of bits.</p>
2738 <p>Both arguments to the '<tt>shl</tt>' instruction must be the same <a
2739 href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2740 type. '<tt>op2</tt>' is treated as an unsigned value.</p>
2744 <p>The value produced is <tt>op1</tt> * 2<sup><tt>op2</tt></sup> mod 2<sup>n</sup>,
2745 where n is the width of the result. If <tt>op2</tt> is (statically or dynamically) negative or
2746 equal to or larger than the number of bits in <tt>op1</tt>, the result is undefined.
2747 If the arguments are vectors, each vector element of <tt>op1</tt> is shifted by the
2748 corresponding shift amount in <tt>op2</tt>.</p>
2750 <h5>Example:</h5><pre>
2751 <result> = shl i32 4, %var <i>; yields {i32}: 4 << %var</i>
2752 <result> = shl i32 4, 2 <i>; yields {i32}: 16</i>
2753 <result> = shl i32 1, 10 <i>; yields {i32}: 1024</i>
2754 <result> = shl i32 1, 32 <i>; undefined</i>
2755 <result> = shl <2 x i32> < i32 1, i32 1>, < i32 1, i32 2> <i>; yields: result=<2 x i32> < i32 2, i32 4></i>
2758 <!-- _______________________________________________________________________ -->
2759 <div class="doc_subsubsection"> <a name="i_lshr">'<tt>lshr</tt>'
2760 Instruction</a> </div>
2761 <div class="doc_text">
2763 <pre> <result> = lshr <ty> <op1>, <op2> <i>; yields {ty}:result</i>
2767 <p>The '<tt>lshr</tt>' instruction (logical shift right) returns the first
2768 operand shifted to the right a specified number of bits with zero fill.</p>
2771 <p>Both arguments to the '<tt>lshr</tt>' instruction must be the same
2772 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2773 type. '<tt>op2</tt>' is treated as an unsigned value.</p>
2777 <p>This instruction always performs a logical shift right operation. The most
2778 significant bits of the result will be filled with zero bits after the
2779 shift. If <tt>op2</tt> is (statically or dynamically) equal to or larger than
2780 the number of bits in <tt>op1</tt>, the result is undefined. If the arguments are
2781 vectors, each vector element of <tt>op1</tt> is shifted by the corresponding shift
2782 amount in <tt>op2</tt>.</p>
2786 <result> = lshr i32 4, 1 <i>; yields {i32}:result = 2</i>
2787 <result> = lshr i32 4, 2 <i>; yields {i32}:result = 1</i>
2788 <result> = lshr i8 4, 3 <i>; yields {i8}:result = 0</i>
2789 <result> = lshr i8 -2, 1 <i>; yields {i8}:result = 0x7FFFFFFF </i>
2790 <result> = lshr i32 1, 32 <i>; undefined</i>
2791 <result> = lshr <2 x i32> < i32 -2, i32 4>, < i32 1, i32 2> <i>; yields: result=<2 x i32> < i32 0x7FFFFFFF, i32 1></i>
2795 <!-- _______________________________________________________________________ -->
2796 <div class="doc_subsubsection"> <a name="i_ashr">'<tt>ashr</tt>'
2797 Instruction</a> </div>
2798 <div class="doc_text">
2801 <pre> <result> = ashr <ty> <op1>, <op2> <i>; yields {ty}:result</i>
2805 <p>The '<tt>ashr</tt>' instruction (arithmetic shift right) returns the first
2806 operand shifted to the right a specified number of bits with sign extension.</p>
2809 <p>Both arguments to the '<tt>ashr</tt>' instruction must be the same
2810 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2811 type. '<tt>op2</tt>' is treated as an unsigned value.</p>
2814 <p>This instruction always performs an arithmetic shift right operation,
2815 The most significant bits of the result will be filled with the sign bit
2816 of <tt>op1</tt>. If <tt>op2</tt> is (statically or dynamically) equal to or
2817 larger than the number of bits in <tt>op1</tt>, the result is undefined. If the
2818 arguments are vectors, each vector element of <tt>op1</tt> is shifted by the
2819 corresponding shift amount in <tt>op2</tt>.</p>
2823 <result> = ashr i32 4, 1 <i>; yields {i32}:result = 2</i>
2824 <result> = ashr i32 4, 2 <i>; yields {i32}:result = 1</i>
2825 <result> = ashr i8 4, 3 <i>; yields {i8}:result = 0</i>
2826 <result> = ashr i8 -2, 1 <i>; yields {i8}:result = -1</i>
2827 <result> = ashr i32 1, 32 <i>; undefined</i>
2828 <result> = ashr <2 x i32> < i32 -2, i32 4>, < i32 1, i32 3> <i>; yields: result=<2 x i32> < i32 -1, i32 0></i>
2832 <!-- _______________________________________________________________________ -->
2833 <div class="doc_subsubsection"> <a name="i_and">'<tt>and</tt>'
2834 Instruction</a> </div>
2836 <div class="doc_text">
2841 <result> = and <ty> <op1>, <op2> <i>; yields {ty}:result</i>
2846 <p>The '<tt>and</tt>' instruction returns the bitwise logical and of
2847 its two operands.</p>
2851 <p>The two arguments to the '<tt>and</tt>' instruction must be
2852 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2853 values. Both arguments must have identical types.</p>
2856 <p>The truth table used for the '<tt>and</tt>' instruction is:</p>
2859 <table border="1" cellspacing="0" cellpadding="4">
2891 <result> = and i32 4, %var <i>; yields {i32}:result = 4 & %var</i>
2892 <result> = and i32 15, 40 <i>; yields {i32}:result = 8</i>
2893 <result> = and i32 4, 8 <i>; yields {i32}:result = 0</i>
2896 <!-- _______________________________________________________________________ -->
2897 <div class="doc_subsubsection"> <a name="i_or">'<tt>or</tt>' Instruction</a> </div>
2898 <div class="doc_text">
2900 <pre> <result> = or <ty> <op1>, <op2> <i>; yields {ty}:result</i>
2903 <p>The '<tt>or</tt>' instruction returns the bitwise logical inclusive
2904 or of its two operands.</p>
2907 <p>The two arguments to the '<tt>or</tt>' instruction must be
2908 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2909 values. Both arguments must have identical types.</p>
2911 <p>The truth table used for the '<tt>or</tt>' instruction is:</p>
2914 <table border="1" cellspacing="0" cellpadding="4">
2945 <pre> <result> = or i32 4, %var <i>; yields {i32}:result = 4 | %var</i>
2946 <result> = or i32 15, 40 <i>; yields {i32}:result = 47</i>
2947 <result> = or i32 4, 8 <i>; yields {i32}:result = 12</i>
2950 <!-- _______________________________________________________________________ -->
2951 <div class="doc_subsubsection"> <a name="i_xor">'<tt>xor</tt>'
2952 Instruction</a> </div>
2953 <div class="doc_text">
2955 <pre> <result> = xor <ty> <op1>, <op2> <i>; yields {ty}:result</i>
2958 <p>The '<tt>xor</tt>' instruction returns the bitwise logical exclusive
2959 or of its two operands. The <tt>xor</tt> is used to implement the
2960 "one's complement" operation, which is the "~" operator in C.</p>
2962 <p>The two arguments to the '<tt>xor</tt>' instruction must be
2963 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2964 values. Both arguments must have identical types.</p>
2968 <p>The truth table used for the '<tt>xor</tt>' instruction is:</p>
2971 <table border="1" cellspacing="0" cellpadding="4">
3003 <pre> <result> = xor i32 4, %var <i>; yields {i32}:result = 4 ^ %var</i>
3004 <result> = xor i32 15, 40 <i>; yields {i32}:result = 39</i>
3005 <result> = xor i32 4, 8 <i>; yields {i32}:result = 12</i>
3006 <result> = xor i32 %V, -1 <i>; yields {i32}:result = ~%V</i>
3010 <!-- ======================================================================= -->
3011 <div class="doc_subsection">
3012 <a name="vectorops">Vector Operations</a>
3015 <div class="doc_text">
3017 <p>LLVM supports several instructions to represent vector operations in a
3018 target-independent manner. These instructions cover the element-access and
3019 vector-specific operations needed to process vectors effectively. While LLVM
3020 does directly support these vector operations, many sophisticated algorithms
3021 will want to use target-specific intrinsics to take full advantage of a specific
3026 <!-- _______________________________________________________________________ -->
3027 <div class="doc_subsubsection">
3028 <a name="i_extractelement">'<tt>extractelement</tt>' Instruction</a>
3031 <div class="doc_text">
3036 <result> = extractelement <n x <ty>> <val>, i32 <idx> <i>; yields <ty></i>
3042 The '<tt>extractelement</tt>' instruction extracts a single scalar
3043 element from a vector at a specified index.
3050 The first operand of an '<tt>extractelement</tt>' instruction is a
3051 value of <a href="#t_vector">vector</a> type. The second operand is
3052 an index indicating the position from which to extract the element.
3053 The index may be a variable.</p>
3058 The result is a scalar of the same type as the element type of
3059 <tt>val</tt>. Its value is the value at position <tt>idx</tt> of
3060 <tt>val</tt>. If <tt>idx</tt> exceeds the length of <tt>val</tt>, the
3061 results are undefined.
3067 %result = extractelement <4 x i32> %vec, i32 0 <i>; yields i32</i>
3072 <!-- _______________________________________________________________________ -->
3073 <div class="doc_subsubsection">
3074 <a name="i_insertelement">'<tt>insertelement</tt>' Instruction</a>
3077 <div class="doc_text">
3082 <result> = insertelement <n x <ty>> <val>, <ty> <elt>, i32 <idx> <i>; yields <n x <ty>></i>
3088 The '<tt>insertelement</tt>' instruction inserts a scalar
3089 element into a vector at a specified index.
3096 The first operand of an '<tt>insertelement</tt>' instruction is a
3097 value of <a href="#t_vector">vector</a> type. The second operand is a
3098 scalar value whose type must equal the element type of the first
3099 operand. The third operand is an index indicating the position at
3100 which to insert the value. The index may be a variable.</p>
3105 The result is a vector of the same type as <tt>val</tt>. Its
3106 element values are those of <tt>val</tt> except at position
3107 <tt>idx</tt>, where it gets the value <tt>elt</tt>. If <tt>idx</tt>
3108 exceeds the length of <tt>val</tt>, the results are undefined.
3114 %result = insertelement <4 x i32> %vec, i32 1, i32 0 <i>; yields <4 x i32></i>
3118 <!-- _______________________________________________________________________ -->
3119 <div class="doc_subsubsection">
3120 <a name="i_shufflevector">'<tt>shufflevector</tt>' Instruction</a>
3123 <div class="doc_text">
3128 <result> = shufflevector <n x <ty>> <v1>, <n x <ty>> <v2>, <m x i32> <mask> <i>; yields <m x <ty>></i>
3134 The '<tt>shufflevector</tt>' instruction constructs a permutation of elements
3135 from two input vectors, returning a vector with the same element type as
3136 the input and length that is the same as the shuffle mask.
3142 The first two operands of a '<tt>shufflevector</tt>' instruction are vectors
3143 with types that match each other. The third argument is a shuffle mask whose
3144 element type is always 'i32'. The result of the instruction is a vector whose
3145 length is the same as the shuffle mask and whose element type is the same as
3146 the element type of the first two operands.
3150 The shuffle mask operand is required to be a constant vector with either
3151 constant integer or undef values.
3157 The elements of the two input vectors are numbered from left to right across
3158 both of the vectors. The shuffle mask operand specifies, for each element of
3159 the result vector, which element of the two input vectors the result element
3160 gets. The element selector may be undef (meaning "don't care") and the second
3161 operand may be undef if performing a shuffle from only one vector.
3167 %result = shufflevector <4 x i32> %v1, <4 x i32> %v2,
3168 <4 x i32> <i32 0, i32 4, i32 1, i32 5> <i>; yields <4 x i32></i>
3169 %result = shufflevector <4 x i32> %v1, <4 x i32> undef,
3170 <4 x i32> <i32 0, i32 1, i32 2, i32 3> <i>; yields <4 x i32></i> - Identity shuffle.
3171 %result = shufflevector <8 x i32> %v1, <8 x i32> undef,
3172 <4 x i32> <i32 0, i32 1, i32 2, i32 3> <i>; yields <4 x i32></i>
3173 %result = shufflevector <4 x i32> %v1, <4 x i32> %v2,
3174 <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>
3179 <!-- ======================================================================= -->
3180 <div class="doc_subsection">
3181 <a name="aggregateops">Aggregate Operations</a>
3184 <div class="doc_text">
3186 <p>LLVM supports several instructions for working with aggregate values.
3191 <!-- _______________________________________________________________________ -->
3192 <div class="doc_subsubsection">
3193 <a name="i_extractvalue">'<tt>extractvalue</tt>' Instruction</a>
3196 <div class="doc_text">
3201 <result> = extractvalue <aggregate type> <val>, <idx>{, <idx>}*
3207 The '<tt>extractvalue</tt>' instruction extracts the value of a struct field
3208 or array element from an aggregate value.
3215 The first operand of an '<tt>extractvalue</tt>' instruction is a
3216 value of <a href="#t_struct">struct</a> or <a href="#t_array">array</a>
3217 type. The operands are constant indices to specify which value to extract
3218 in a similar manner as indices in a
3219 '<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.
3225 The result is the value at the position in the aggregate specified by
3232 %result = extractvalue {i32, float} %agg, 0 <i>; yields i32</i>
3237 <!-- _______________________________________________________________________ -->
3238 <div class="doc_subsubsection">
3239 <a name="i_insertvalue">'<tt>insertvalue</tt>' Instruction</a>
3242 <div class="doc_text">
3247 <result> = insertvalue <aggregate type> <val>, <ty> <val>, <idx> <i>; yields <n x <ty>></i>
3253 The '<tt>insertvalue</tt>' instruction inserts a value
3254 into a struct field or array element in an aggregate.
3261 The first operand of an '<tt>insertvalue</tt>' instruction is a
3262 value of <a href="#t_struct">struct</a> or <a href="#t_array">array</a> type.
3263 The second operand is a first-class value to insert.
3264 The following operands are constant indices
3265 indicating the position at which to insert the value in a similar manner as
3267 '<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.
3268 The value to insert must have the same type as the value identified
3275 The result is an aggregate of the same type as <tt>val</tt>. Its
3276 value is that of <tt>val</tt> except that the value at the position
3277 specified by the indices is that of <tt>elt</tt>.
3283 %result = insertvalue {i32, float} %agg, i32 1, 0 <i>; yields {i32, float}</i>
3288 <!-- ======================================================================= -->
3289 <div class="doc_subsection">
3290 <a name="memoryops">Memory Access and Addressing Operations</a>
3293 <div class="doc_text">
3295 <p>A key design point of an SSA-based representation is how it
3296 represents memory. In LLVM, no memory locations are in SSA form, which
3297 makes things very simple. This section describes how to read, write,
3298 allocate, and free memory in LLVM.</p>
3302 <!-- _______________________________________________________________________ -->
3303 <div class="doc_subsubsection">
3304 <a name="i_malloc">'<tt>malloc</tt>' Instruction</a>
3307 <div class="doc_text">
3312 <result> = malloc <type>[, i32 <NumElements>][, align <alignment>] <i>; yields {type*}:result</i>
3317 <p>The '<tt>malloc</tt>' instruction allocates memory from the system
3318 heap and returns a pointer to it. The object is always allocated in the generic
3319 address space (address space zero).</p>
3323 <p>The '<tt>malloc</tt>' instruction allocates
3324 <tt>sizeof(<type>)*NumElements</tt>
3325 bytes of memory from the operating system and returns a pointer of the
3326 appropriate type to the program. If "NumElements" is specified, it is the
3327 number of elements allocated, otherwise "NumElements" is defaulted to be one.
3328 If a constant alignment is specified, the value result of the allocation is guaranteed to
3329 be aligned to at least that boundary. If not specified, or if zero, the target can
3330 choose to align the allocation on any convenient boundary.</p>
3332 <p>'<tt>type</tt>' must be a sized type.</p>
3336 <p>Memory is allocated using the system "<tt>malloc</tt>" function, and
3337 a pointer is returned. The result of a zero byte allocation is undefined. The
3338 result is null if there is insufficient memory available.</p>
3343 %array = malloc [4 x i8] <i>; yields {[%4 x i8]*}:array</i>
3345 %size = <a href="#i_add">add</a> i32 2, 2 <i>; yields {i32}:size = i32 4</i>
3346 %array1 = malloc i8, i32 4 <i>; yields {i8*}:array1</i>
3347 %array2 = malloc [12 x i8], i32 %size <i>; yields {[12 x i8]*}:array2</i>
3348 %array3 = malloc i32, i32 4, align 1024 <i>; yields {i32*}:array3</i>
3349 %array4 = malloc i32, align 1024 <i>; yields {i32*}:array4</i>
3352 <p>Note that the code generator does not yet respect the
3353 alignment value.</p>
3357 <!-- _______________________________________________________________________ -->
3358 <div class="doc_subsubsection">
3359 <a name="i_free">'<tt>free</tt>' Instruction</a>
3362 <div class="doc_text">
3367 free <type> <value> <i>; yields {void}</i>
3372 <p>The '<tt>free</tt>' instruction returns memory back to the unused
3373 memory heap to be reallocated in the future.</p>
3377 <p>'<tt>value</tt>' shall be a pointer value that points to a value
3378 that was allocated with the '<tt><a href="#i_malloc">malloc</a></tt>'
3383 <p>Access to the memory pointed to by the pointer is no longer defined
3384 after this instruction executes. If the pointer is null, the operation
3390 %array = <a href="#i_malloc">malloc</a> [4 x i8] <i>; yields {[4 x i8]*}:array</i>
3391 free [4 x i8]* %array
3395 <!-- _______________________________________________________________________ -->
3396 <div class="doc_subsubsection">
3397 <a name="i_alloca">'<tt>alloca</tt>' Instruction</a>
3400 <div class="doc_text">
3405 <result> = alloca <type>[, i32 <NumElements>][, align <alignment>] <i>; yields {type*}:result</i>
3410 <p>The '<tt>alloca</tt>' instruction allocates memory on the stack frame of the
3411 currently executing function, to be automatically released when this function
3412 returns to its caller. The object is always allocated in the generic address
3413 space (address space zero).</p>
3417 <p>The '<tt>alloca</tt>' instruction allocates <tt>sizeof(<type>)*NumElements</tt>
3418 bytes of memory on the runtime stack, returning a pointer of the
3419 appropriate type to the program. If "NumElements" is specified, it is the
3420 number of elements allocated, otherwise "NumElements" is defaulted to be one.
3421 If a constant alignment is specified, the value result of the allocation is guaranteed
3422 to be aligned to at least that boundary. If not specified, or if zero, the target
3423 can choose to align the allocation on any convenient boundary.</p>
3425 <p>'<tt>type</tt>' may be any sized type.</p>
3429 <p>Memory is allocated; a pointer is returned. The operation is undefiend if
3430 there is insufficient stack space for the allocation. '<tt>alloca</tt>'d
3431 memory is automatically released when the function returns. The '<tt>alloca</tt>'
3432 instruction is commonly used to represent automatic variables that must
3433 have an address available. When the function returns (either with the <tt><a
3434 href="#i_ret">ret</a></tt> or <tt><a href="#i_unwind">unwind</a></tt>
3435 instructions), the memory is reclaimed. Allocating zero bytes
3436 is legal, but the result is undefined.</p>
3441 %ptr = alloca i32 <i>; yields {i32*}:ptr</i>
3442 %ptr = alloca i32, i32 4 <i>; yields {i32*}:ptr</i>
3443 %ptr = alloca i32, i32 4, align 1024 <i>; yields {i32*}:ptr</i>
3444 %ptr = alloca i32, align 1024 <i>; yields {i32*}:ptr</i>
3448 <!-- _______________________________________________________________________ -->
3449 <div class="doc_subsubsection"> <a name="i_load">'<tt>load</tt>'
3450 Instruction</a> </div>
3451 <div class="doc_text">
3453 <pre> <result> = load <ty>* <pointer>[, align <alignment>]<br> <result> = volatile load <ty>* <pointer>[, align <alignment>]<br></pre>
3455 <p>The '<tt>load</tt>' instruction is used to read from memory.</p>
3457 <p>The argument to the '<tt>load</tt>' instruction specifies the memory
3458 address from which to load. The pointer must point to a <a
3459 href="#t_firstclass">first class</a> type. If the <tt>load</tt> is
3460 marked as <tt>volatile</tt>, then the optimizer is not allowed to modify
3461 the number or order of execution of this <tt>load</tt> with other
3462 volatile <tt>load</tt> and <tt><a href="#i_store">store</a></tt>
3465 The optional constant "align" argument specifies the alignment of the operation
3466 (that is, the alignment of the memory address). A value of 0 or an
3467 omitted "align" argument means that the operation has the preferential
3468 alignment for the target. It is the responsibility of the code emitter
3469 to ensure that the alignment information is correct. Overestimating
3470 the alignment results in an undefined behavior. Underestimating the
3471 alignment may produce less efficient code. An alignment of 1 is always
3475 <p>The location of memory pointed to is loaded.</p>
3477 <pre> %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
3479 href="#i_store">store</a> i32 3, i32* %ptr <i>; yields {void}</i>
3480 %val = load i32* %ptr <i>; yields {i32}:val = i32 3</i>
3483 <!-- _______________________________________________________________________ -->
3484 <div class="doc_subsubsection"> <a name="i_store">'<tt>store</tt>'
3485 Instruction</a> </div>
3486 <div class="doc_text">
3488 <pre> store <ty> <value>, <ty>* <pointer>[, align <alignment>] <i>; yields {void}</i>
3489 volatile store <ty> <value>, <ty>* <pointer>[, align <alignment>] <i>; yields {void}</i>
3492 <p>The '<tt>store</tt>' instruction is used to write to memory.</p>
3494 <p>There are two arguments to the '<tt>store</tt>' instruction: a value
3495 to store and an address at which to store it. The type of the '<tt><pointer></tt>'
3496 operand must be a pointer to the <a href="#t_firstclass">first class</a> type
3497 of the '<tt><value></tt>'
3498 operand. If the <tt>store</tt> is marked as <tt>volatile</tt>, then the
3499 optimizer is not allowed to modify the number or order of execution of
3500 this <tt>store</tt> with other volatile <tt>load</tt> and <tt><a
3501 href="#i_store">store</a></tt> instructions.</p>
3503 The optional constant "align" argument specifies the alignment of the operation
3504 (that is, the alignment of the memory address). A value of 0 or an
3505 omitted "align" argument means that the operation has the preferential
3506 alignment for the target. It is the responsibility of the code emitter
3507 to ensure that the alignment information is correct. Overestimating
3508 the alignment results in an undefined behavior. Underestimating the
3509 alignment may produce less efficient code. An alignment of 1 is always
3513 <p>The contents of memory are updated to contain '<tt><value></tt>'
3514 at the location specified by the '<tt><pointer></tt>' operand.</p>
3516 <pre> %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
3517 store i32 3, i32* %ptr <i>; yields {void}</i>
3518 %val = <a href="#i_load">load</a> i32* %ptr <i>; yields {i32}:val = i32 3</i>
3522 <!-- _______________________________________________________________________ -->
3523 <div class="doc_subsubsection">
3524 <a name="i_getelementptr">'<tt>getelementptr</tt>' Instruction</a>
3527 <div class="doc_text">
3530 <result> = getelementptr <pty>* <ptrval>{, <ty> <idx>}*
3536 The '<tt>getelementptr</tt>' instruction is used to get the address of a
3537 subelement of an aggregate data structure. It performs address calculation only
3538 and does not access memory.</p>
3542 <p>The first argument is always a pointer, and forms the basis of the
3543 calculation. The remaining arguments are indices, that indicate which of the
3544 elements of the aggregate object are indexed. The interpretation of each index
3545 is dependent on the type being indexed into. The first index always indexes the
3546 pointer value given as the first argument, the second index indexes a value of
3547 the type pointed to (not necessarily the value directly pointed to, since the
3548 first index can be non-zero), etc. The first type indexed into must be a pointer
3549 value, subsequent types can be arrays, vectors and structs. Note that subsequent
3550 types being indexed into can never be pointers, since that would require loading
3551 the pointer before continuing calculation.</p>
3553 <p>The type of each index argument depends on the type it is indexing into.
3554 When indexing into a (packed) structure, only <tt>i32</tt> integer
3555 <b>constants</b> are allowed. When indexing into an array, pointer or vector,
3556 only integers of 32 or 64 bits are allowed (also non-constants). 32-bit values
3557 will be sign extended to 64-bits if required.</p>
3559 <p>For example, let's consider a C code fragment and how it gets
3560 compiled to LLVM:</p>
3562 <div class="doc_code">
3575 int *foo(struct ST *s) {
3576 return &s[1].Z.B[5][13];
3581 <p>The LLVM code generated by the GCC frontend is:</p>
3583 <div class="doc_code">
3585 %RT = <a href="#namedtypes">type</a> { i8 , [10 x [20 x i32]], i8 }
3586 %ST = <a href="#namedtypes">type</a> { i32, double, %RT }
3588 define i32* %foo(%ST* %s) {
3590 %reg = getelementptr %ST* %s, i32 1, i32 2, i32 1, i32 5, i32 13
3598 <p>In the example above, the first index is indexing into the '<tt>%ST*</tt>'
3599 type, which is a pointer, yielding a '<tt>%ST</tt>' = '<tt>{ i32, double, %RT
3600 }</tt>' type, a structure. The second index indexes into the third element of
3601 the structure, yielding a '<tt>%RT</tt>' = '<tt>{ i8 , [10 x [20 x i32]],
3602 i8 }</tt>' type, another structure. The third index indexes into the second
3603 element of the structure, yielding a '<tt>[10 x [20 x i32]]</tt>' type, an
3604 array. The two dimensions of the array are subscripted into, yielding an
3605 '<tt>i32</tt>' type. The '<tt>getelementptr</tt>' instruction returns a pointer
3606 to this element, thus computing a value of '<tt>i32*</tt>' type.</p>
3608 <p>Note that it is perfectly legal to index partially through a
3609 structure, returning a pointer to an inner element. Because of this,
3610 the LLVM code for the given testcase is equivalent to:</p>
3613 define i32* %foo(%ST* %s) {
3614 %t1 = getelementptr %ST* %s, i32 1 <i>; yields %ST*:%t1</i>
3615 %t2 = getelementptr %ST* %t1, i32 0, i32 2 <i>; yields %RT*:%t2</i>
3616 %t3 = getelementptr %RT* %t2, i32 0, i32 1 <i>; yields [10 x [20 x i32]]*:%t3</i>
3617 %t4 = getelementptr [10 x [20 x i32]]* %t3, i32 0, i32 5 <i>; yields [20 x i32]*:%t4</i>
3618 %t5 = getelementptr [20 x i32]* %t4, i32 0, i32 13 <i>; yields i32*:%t5</i>
3623 <p>Note that it is undefined to access an array out of bounds: array and
3624 pointer indexes must always be within the defined bounds of the array type.
3625 The one exception for this rule is zero length arrays. These arrays are
3626 defined to be accessible as variable length arrays, which requires access
3627 beyond the zero'th element.</p>
3629 <p>The getelementptr instruction is often confusing. For some more insight
3630 into how it works, see <a href="GetElementPtr.html">the getelementptr
3636 <i>; yields [12 x i8]*:aptr</i>
3637 %aptr = getelementptr {i32, [12 x i8]}* %saptr, i64 0, i32 1
3638 <i>; yields i8*:vptr</i>
3639 %vptr = getelementptr {i32, <2 x i8>}* %svptr, i64 0, i32 1, i32 1
3640 <i>; yields i8*:eptr</i>
3641 %eptr = getelementptr [12 x i8]* %aptr, i64 0, i32 1
3645 <!-- ======================================================================= -->
3646 <div class="doc_subsection"> <a name="convertops">Conversion Operations</a>
3648 <div class="doc_text">
3649 <p>The instructions in this category are the conversion instructions (casting)
3650 which all take a single operand and a type. They perform various bit conversions
3654 <!-- _______________________________________________________________________ -->
3655 <div class="doc_subsubsection">
3656 <a name="i_trunc">'<tt>trunc .. to</tt>' Instruction</a>
3658 <div class="doc_text">
3662 <result> = trunc <ty> <value> to <ty2> <i>; yields ty2</i>
3667 The '<tt>trunc</tt>' instruction truncates its operand to the type <tt>ty2</tt>.
3672 The '<tt>trunc</tt>' instruction takes a <tt>value</tt> to trunc, which must
3673 be an <a href="#t_integer">integer</a> type, and a type that specifies the size
3674 and type of the result, which must be an <a href="#t_integer">integer</a>
3675 type. The bit size of <tt>value</tt> must be larger than the bit size of
3676 <tt>ty2</tt>. Equal sized types are not allowed.</p>
3680 The '<tt>trunc</tt>' instruction truncates the high order bits in <tt>value</tt>
3681 and converts the remaining bits to <tt>ty2</tt>. Since the source size must be
3682 larger than the destination size, <tt>trunc</tt> cannot be a <i>no-op cast</i>.
3683 It will always truncate bits.</p>
3687 %X = trunc i32 257 to i8 <i>; yields i8:1</i>
3688 %Y = trunc i32 123 to i1 <i>; yields i1:true</i>
3689 %Y = trunc i32 122 to i1 <i>; yields i1:false</i>
3693 <!-- _______________________________________________________________________ -->
3694 <div class="doc_subsubsection">
3695 <a name="i_zext">'<tt>zext .. to</tt>' Instruction</a>
3697 <div class="doc_text">
3701 <result> = zext <ty> <value> to <ty2> <i>; yields ty2</i>
3705 <p>The '<tt>zext</tt>' instruction zero extends its operand to type
3710 <p>The '<tt>zext</tt>' instruction takes a value to cast, which must be of
3711 <a href="#t_integer">integer</a> type, and a type to cast it to, which must
3712 also be of <a href="#t_integer">integer</a> type. The bit size of the
3713 <tt>value</tt> must be smaller than the bit size of the destination type,
3717 <p>The <tt>zext</tt> fills the high order bits of the <tt>value</tt> with zero
3718 bits until it reaches the size of the destination type, <tt>ty2</tt>.</p>
3720 <p>When zero extending from i1, the result will always be either 0 or 1.</p>
3724 %X = zext i32 257 to i64 <i>; yields i64:257</i>
3725 %Y = zext i1 true to i32 <i>; yields i32:1</i>
3729 <!-- _______________________________________________________________________ -->
3730 <div class="doc_subsubsection">
3731 <a name="i_sext">'<tt>sext .. to</tt>' Instruction</a>
3733 <div class="doc_text">
3737 <result> = sext <ty> <value> to <ty2> <i>; yields ty2</i>
3741 <p>The '<tt>sext</tt>' sign extends <tt>value</tt> to the type <tt>ty2</tt>.</p>
3745 The '<tt>sext</tt>' instruction takes a value to cast, which must be of
3746 <a href="#t_integer">integer</a> type, and a type to cast it to, which must
3747 also be of <a href="#t_integer">integer</a> type. The bit size of the
3748 <tt>value</tt> must be smaller than the bit size of the destination type,
3753 The '<tt>sext</tt>' instruction performs a sign extension by copying the sign
3754 bit (highest order bit) of the <tt>value</tt> until it reaches the bit size of
3755 the type <tt>ty2</tt>.</p>
3757 <p>When sign extending from i1, the extension always results in -1 or 0.</p>
3761 %X = sext i8 -1 to i16 <i>; yields i16 :65535</i>
3762 %Y = sext i1 true to i32 <i>; yields i32:-1</i>
3766 <!-- _______________________________________________________________________ -->
3767 <div class="doc_subsubsection">
3768 <a name="i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a>
3771 <div class="doc_text">
3776 <result> = fptrunc <ty> <value> to <ty2> <i>; yields ty2</i>
3780 <p>The '<tt>fptrunc</tt>' instruction truncates <tt>value</tt> to type
3785 <p>The '<tt>fptrunc</tt>' instruction takes a <a href="#t_floating">floating
3786 point</a> value to cast and a <a href="#t_floating">floating point</a> type to
3787 cast it to. The size of <tt>value</tt> must be larger than the size of
3788 <tt>ty2</tt>. This implies that <tt>fptrunc</tt> cannot be used to make a
3789 <i>no-op cast</i>.</p>
3792 <p> The '<tt>fptrunc</tt>' instruction truncates a <tt>value</tt> from a larger
3793 <a href="#t_floating">floating point</a> type to a smaller
3794 <a href="#t_floating">floating point</a> type. If the value cannot fit within
3795 the destination type, <tt>ty2</tt>, then the results are undefined.</p>
3799 %X = fptrunc double 123.0 to float <i>; yields float:123.0</i>
3800 %Y = fptrunc double 1.0E+300 to float <i>; yields undefined</i>
3804 <!-- _______________________________________________________________________ -->
3805 <div class="doc_subsubsection">
3806 <a name="i_fpext">'<tt>fpext .. to</tt>' Instruction</a>
3808 <div class="doc_text">
3812 <result> = fpext <ty> <value> to <ty2> <i>; yields ty2</i>
3816 <p>The '<tt>fpext</tt>' extends a floating point <tt>value</tt> to a larger
3817 floating point value.</p>
3820 <p>The '<tt>fpext</tt>' instruction takes a
3821 <a href="#t_floating">floating point</a> <tt>value</tt> to cast,
3822 and a <a href="#t_floating">floating point</a> type to cast it to. The source
3823 type must be smaller than the destination type.</p>
3826 <p>The '<tt>fpext</tt>' instruction extends the <tt>value</tt> from a smaller
3827 <a href="#t_floating">floating point</a> type to a larger
3828 <a href="#t_floating">floating point</a> type. The <tt>fpext</tt> cannot be
3829 used to make a <i>no-op cast</i> because it always changes bits. Use
3830 <tt>bitcast</tt> to make a <i>no-op cast</i> for a floating point cast.</p>
3834 %X = fpext float 3.1415 to double <i>; yields double:3.1415</i>
3835 %Y = fpext float 1.0 to float <i>; yields float:1.0 (no-op)</i>
3839 <!-- _______________________________________________________________________ -->
3840 <div class="doc_subsubsection">
3841 <a name="i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a>
3843 <div class="doc_text">
3847 <result> = fptoui <ty> <value> to <ty2> <i>; yields ty2</i>
3851 <p>The '<tt>fptoui</tt>' converts a floating point <tt>value</tt> to its
3852 unsigned integer equivalent of type <tt>ty2</tt>.
3856 <p>The '<tt>fptoui</tt>' instruction takes a value to cast, which must be a
3857 scalar or vector <a href="#t_floating">floating point</a> value, and a type
3858 to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
3859 type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
3860 vector integer type with the same number of elements as <tt>ty</tt></p>
3863 <p> The '<tt>fptoui</tt>' instruction converts its
3864 <a href="#t_floating">floating point</a> operand into the nearest (rounding
3865 towards zero) unsigned integer value. If the value cannot fit in <tt>ty2</tt>,
3866 the results are undefined.</p>
3870 %X = fptoui double 123.0 to i32 <i>; yields i32:123</i>
3871 %Y = fptoui float 1.0E+300 to i1 <i>; yields undefined:1</i>
3872 %X = fptoui float 1.04E+17 to i8 <i>; yields undefined:1</i>
3876 <!-- _______________________________________________________________________ -->
3877 <div class="doc_subsubsection">
3878 <a name="i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a>
3880 <div class="doc_text">
3884 <result> = fptosi <ty> <value> to <ty2> <i>; yields ty2</i>
3888 <p>The '<tt>fptosi</tt>' instruction converts
3889 <a href="#t_floating">floating point</a> <tt>value</tt> to type <tt>ty2</tt>.
3893 <p> The '<tt>fptosi</tt>' instruction takes a value to cast, which must be a
3894 scalar or vector <a href="#t_floating">floating point</a> value, and a type
3895 to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
3896 type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
3897 vector integer type with the same number of elements as <tt>ty</tt></p>
3900 <p>The '<tt>fptosi</tt>' instruction converts its
3901 <a href="#t_floating">floating point</a> operand into the nearest (rounding
3902 towards zero) signed integer value. If the value cannot fit in <tt>ty2</tt>,
3903 the results are undefined.</p>
3907 %X = fptosi double -123.0 to i32 <i>; yields i32:-123</i>
3908 %Y = fptosi float 1.0E-247 to i1 <i>; yields undefined:1</i>
3909 %X = fptosi float 1.04E+17 to i8 <i>; yields undefined:1</i>
3913 <!-- _______________________________________________________________________ -->
3914 <div class="doc_subsubsection">
3915 <a name="i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a>
3917 <div class="doc_text">
3921 <result> = uitofp <ty> <value> to <ty2> <i>; yields ty2</i>
3925 <p>The '<tt>uitofp</tt>' instruction regards <tt>value</tt> as an unsigned
3926 integer and converts that value to the <tt>ty2</tt> type.</p>
3929 <p>The '<tt>uitofp</tt>' instruction takes a value to cast, which must be a
3930 scalar or vector <a href="#t_integer">integer</a> value, and a type to cast it
3931 to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
3932 type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
3933 floating point type with the same number of elements as <tt>ty</tt></p>
3936 <p>The '<tt>uitofp</tt>' instruction interprets its operand as an unsigned
3937 integer quantity and converts it to the corresponding floating point value. If
3938 the value cannot fit in the floating point value, the results are undefined.</p>
3942 %X = uitofp i32 257 to float <i>; yields float:257.0</i>
3943 %Y = uitofp i8 -1 to double <i>; yields double:255.0</i>
3947 <!-- _______________________________________________________________________ -->
3948 <div class="doc_subsubsection">
3949 <a name="i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a>
3951 <div class="doc_text">
3955 <result> = sitofp <ty> <value> to <ty2> <i>; yields ty2</i>
3959 <p>The '<tt>sitofp</tt>' instruction regards <tt>value</tt> as a signed
3960 integer and converts that value to the <tt>ty2</tt> type.</p>
3963 <p>The '<tt>sitofp</tt>' instruction takes a value to cast, which must be a
3964 scalar or vector <a href="#t_integer">integer</a> value, and a type to cast it
3965 to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
3966 type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
3967 floating point type with the same number of elements as <tt>ty</tt></p>
3970 <p>The '<tt>sitofp</tt>' instruction interprets its operand as a signed
3971 integer quantity and converts it to the corresponding floating point value. If
3972 the value cannot fit in the floating point value, the results are undefined.</p>
3976 %X = sitofp i32 257 to float <i>; yields float:257.0</i>
3977 %Y = sitofp i8 -1 to double <i>; yields double:-1.0</i>
3981 <!-- _______________________________________________________________________ -->
3982 <div class="doc_subsubsection">
3983 <a name="i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a>
3985 <div class="doc_text">
3989 <result> = ptrtoint <ty> <value> to <ty2> <i>; yields ty2</i>
3993 <p>The '<tt>ptrtoint</tt>' instruction converts the pointer <tt>value</tt> to
3994 the integer type <tt>ty2</tt>.</p>
3997 <p>The '<tt>ptrtoint</tt>' instruction takes a <tt>value</tt> to cast, which
3998 must be a <a href="#t_pointer">pointer</a> value, and a type to cast it to
3999 <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a> type.</p>
4002 <p>The '<tt>ptrtoint</tt>' instruction converts <tt>value</tt> to integer type
4003 <tt>ty2</tt> by interpreting the pointer value as an integer and either
4004 truncating or zero extending that value to the size of the integer type. If
4005 <tt>value</tt> is smaller than <tt>ty2</tt> then a zero extension is done. If
4006 <tt>value</tt> is larger than <tt>ty2</tt> then a truncation is done. If they
4007 are the same size, then nothing is done (<i>no-op cast</i>) other than a type
4012 %X = ptrtoint i32* %X to i8 <i>; yields truncation on 32-bit architecture</i>
4013 %Y = ptrtoint i32* %x to i64 <i>; yields zero extension on 32-bit architecture</i>
4017 <!-- _______________________________________________________________________ -->
4018 <div class="doc_subsubsection">
4019 <a name="i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a>
4021 <div class="doc_text">
4025 <result> = inttoptr <ty> <value> to <ty2> <i>; yields ty2</i>
4029 <p>The '<tt>inttoptr</tt>' instruction converts an integer <tt>value</tt> to
4030 a pointer type, <tt>ty2</tt>.</p>
4033 <p>The '<tt>inttoptr</tt>' instruction takes an <a href="#t_integer">integer</a>
4034 value to cast, and a type to cast it to, which must be a
4035 <a href="#t_pointer">pointer</a> type.</p>
4038 <p>The '<tt>inttoptr</tt>' instruction converts <tt>value</tt> to type
4039 <tt>ty2</tt> by applying either a zero extension or a truncation depending on
4040 the size of the integer <tt>value</tt>. If <tt>value</tt> is larger than the
4041 size of a pointer then a truncation is done. If <tt>value</tt> is smaller than
4042 the size of a pointer then a zero extension is done. If they are the same size,
4043 nothing is done (<i>no-op cast</i>).</p>
4047 %X = inttoptr i32 255 to i32* <i>; yields zero extension on 64-bit architecture</i>
4048 %X = inttoptr i32 255 to i32* <i>; yields no-op on 32-bit architecture</i>
4049 %Y = inttoptr i64 0 to i32* <i>; yields truncation on 32-bit architecture</i>
4053 <!-- _______________________________________________________________________ -->
4054 <div class="doc_subsubsection">
4055 <a name="i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a>
4057 <div class="doc_text">
4061 <result> = bitcast <ty> <value> to <ty2> <i>; yields ty2</i>
4066 <p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
4067 <tt>ty2</tt> without changing any bits.</p>
4071 <p>The '<tt>bitcast</tt>' instruction takes a value to cast, which must be
4072 a non-aggregate first class value, and a type to cast it to, which must also be
4073 a non-aggregate <a href="#t_firstclass">first class</a> type. The bit sizes of
4075 and the destination type, <tt>ty2</tt>, must be identical. If the source
4076 type is a pointer, the destination type must also be a pointer. This
4077 instruction supports bitwise conversion of vectors to integers and to vectors
4078 of other types (as long as they have the same size).</p>
4081 <p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
4082 <tt>ty2</tt>. It is always a <i>no-op cast</i> because no bits change with
4083 this conversion. The conversion is done as if the <tt>value</tt> had been
4084 stored to memory and read back as type <tt>ty2</tt>. Pointer types may only be
4085 converted to other pointer types with this instruction. To convert pointers to
4086 other types, use the <a href="#i_inttoptr">inttoptr</a> or
4087 <a href="#i_ptrtoint">ptrtoint</a> instructions first.</p>
4091 %X = bitcast i8 255 to i8 <i>; yields i8 :-1</i>
4092 %Y = bitcast i32* %x to sint* <i>; yields sint*:%x</i>
4093 %Z = bitcast <2 x int> %V to i64; <i>; yields i64: %V</i>
4097 <!-- ======================================================================= -->
4098 <div class="doc_subsection"> <a name="otherops">Other Operations</a> </div>
4099 <div class="doc_text">
4100 <p>The instructions in this category are the "miscellaneous"
4101 instructions, which defy better classification.</p>
4104 <!-- _______________________________________________________________________ -->
4105 <div class="doc_subsubsection"><a name="i_icmp">'<tt>icmp</tt>' Instruction</a>
4107 <div class="doc_text">
4109 <pre> <result> = icmp <cond> <ty> <op1>, <op2> <i>; yields {i1} or {<N x i1>}:result</i>
4112 <p>The '<tt>icmp</tt>' instruction returns a boolean value or
4113 a vector of boolean values based on comparison
4114 of its two integer, integer vector, or pointer operands.</p>
4116 <p>The '<tt>icmp</tt>' instruction takes three operands. The first operand is
4117 the condition code indicating the kind of comparison to perform. It is not
4118 a value, just a keyword. The possible condition code are:
4121 <li><tt>eq</tt>: equal</li>
4122 <li><tt>ne</tt>: not equal </li>
4123 <li><tt>ugt</tt>: unsigned greater than</li>
4124 <li><tt>uge</tt>: unsigned greater or equal</li>
4125 <li><tt>ult</tt>: unsigned less than</li>
4126 <li><tt>ule</tt>: unsigned less or equal</li>
4127 <li><tt>sgt</tt>: signed greater than</li>
4128 <li><tt>sge</tt>: signed greater or equal</li>
4129 <li><tt>slt</tt>: signed less than</li>
4130 <li><tt>sle</tt>: signed less or equal</li>
4132 <p>The remaining two arguments must be <a href="#t_integer">integer</a> or
4133 <a href="#t_pointer">pointer</a>
4134 or integer <a href="#t_vector">vector</a> typed.
4135 They must also be identical types.</p>
4137 <p>The '<tt>icmp</tt>' compares <tt>op1</tt> and <tt>op2</tt> according to
4138 the condition code given as <tt>cond</tt>. The comparison performed always
4139 yields either an <a href="#t_primitive"><tt>i1</tt></a> or vector of <tt>i1</tt> result, as follows:
4142 <li><tt>eq</tt>: yields <tt>true</tt> if the operands are equal,
4143 <tt>false</tt> otherwise. No sign interpretation is necessary or performed.
4145 <li><tt>ne</tt>: yields <tt>true</tt> if the operands are unequal,
4146 <tt>false</tt> otherwise. No sign interpretation is necessary or performed.</li>
4147 <li><tt>ugt</tt>: interprets the operands as unsigned values and yields
4148 <tt>true</tt> if <tt>op1</tt> is greater than <tt>op2</tt>.</li>
4149 <li><tt>uge</tt>: interprets the operands as unsigned values and yields
4150 <tt>true</tt> if <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
4151 <li><tt>ult</tt>: interprets the operands as unsigned values and yields
4152 <tt>true</tt> if <tt>op1</tt> is less than <tt>op2</tt>.</li>
4153 <li><tt>ule</tt>: interprets the operands as unsigned values and yields
4154 <tt>true</tt> if <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
4155 <li><tt>sgt</tt>: interprets the operands as signed values and yields
4156 <tt>true</tt> if <tt>op1</tt> is greater than <tt>op2</tt>.</li>
4157 <li><tt>sge</tt>: interprets the operands as signed values and yields
4158 <tt>true</tt> if <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
4159 <li><tt>slt</tt>: interprets the operands as signed values and yields
4160 <tt>true</tt> if <tt>op1</tt> is less than <tt>op2</tt>.</li>
4161 <li><tt>sle</tt>: interprets the operands as signed values and yields
4162 <tt>true</tt> if <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
4164 <p>If the operands are <a href="#t_pointer">pointer</a> typed, the pointer
4165 values are compared as if they were integers.</p>
4166 <p>If the operands are integer vectors, then they are compared
4167 element by element. The result is an <tt>i1</tt> vector with
4168 the same number of elements as the values being compared.
4169 Otherwise, the result is an <tt>i1</tt>.
4173 <pre> <result> = icmp eq i32 4, 5 <i>; yields: result=false</i>
4174 <result> = icmp ne float* %X, %X <i>; yields: result=false</i>
4175 <result> = icmp ult i16 4, 5 <i>; yields: result=true</i>
4176 <result> = icmp sgt i16 4, 5 <i>; yields: result=false</i>
4177 <result> = icmp ule i16 -4, 5 <i>; yields: result=false</i>
4178 <result> = icmp sge i16 4, 5 <i>; yields: result=false</i>
4181 <p>Note that the code generator does not yet support vector types with
4182 the <tt>icmp</tt> instruction.</p>
4186 <!-- _______________________________________________________________________ -->
4187 <div class="doc_subsubsection"><a name="i_fcmp">'<tt>fcmp</tt>' Instruction</a>
4189 <div class="doc_text">
4191 <pre> <result> = fcmp <cond> <ty> <op1>, <op2> <i>; yields {i1} or {<N x i1>}:result</i>
4194 <p>The '<tt>fcmp</tt>' instruction returns a boolean value
4195 or vector of boolean values based on comparison
4196 of its operands.</p>
4198 If the operands are floating point scalars, then the result
4199 type is a boolean (<a href="#t_primitive"><tt>i1</tt></a>).
4201 <p>If the operands are floating point vectors, then the result type
4202 is a vector of boolean with the same number of elements as the
4203 operands being compared.</p>
4205 <p>The '<tt>fcmp</tt>' instruction takes three operands. The first operand is
4206 the condition code indicating the kind of comparison to perform. It is not
4207 a value, just a keyword. The possible condition code are:</p>
4209 <li><tt>false</tt>: no comparison, always returns false</li>
4210 <li><tt>oeq</tt>: ordered and equal</li>
4211 <li><tt>ogt</tt>: ordered and greater than </li>
4212 <li><tt>oge</tt>: ordered and greater than or equal</li>
4213 <li><tt>olt</tt>: ordered and less than </li>
4214 <li><tt>ole</tt>: ordered and less than or equal</li>
4215 <li><tt>one</tt>: ordered and not equal</li>
4216 <li><tt>ord</tt>: ordered (no nans)</li>
4217 <li><tt>ueq</tt>: unordered or equal</li>
4218 <li><tt>ugt</tt>: unordered or greater than </li>
4219 <li><tt>uge</tt>: unordered or greater than or equal</li>
4220 <li><tt>ult</tt>: unordered or less than </li>
4221 <li><tt>ule</tt>: unordered or less than or equal</li>
4222 <li><tt>une</tt>: unordered or not equal</li>
4223 <li><tt>uno</tt>: unordered (either nans)</li>
4224 <li><tt>true</tt>: no comparison, always returns true</li>
4226 <p><i>Ordered</i> means that neither operand is a QNAN while
4227 <i>unordered</i> means that either operand may be a QNAN.</p>
4228 <p>Each of <tt>val1</tt> and <tt>val2</tt> arguments must be
4229 either a <a href="#t_floating">floating point</a> type
4230 or a <a href="#t_vector">vector</a> of floating point type.
4231 They must have identical types.</p>
4233 <p>The '<tt>fcmp</tt>' instruction compares <tt>op1</tt> and <tt>op2</tt>
4234 according to the condition code given as <tt>cond</tt>.
4235 If the operands are vectors, then the vectors are compared
4237 Each comparison performed
4238 always yields an <a href="#t_primitive">i1</a> result, as follows:</p>
4240 <li><tt>false</tt>: always yields <tt>false</tt>, regardless of operands.</li>
4241 <li><tt>oeq</tt>: yields <tt>true</tt> if both operands are not a QNAN and
4242 <tt>op1</tt> is equal to <tt>op2</tt>.</li>
4243 <li><tt>ogt</tt>: yields <tt>true</tt> if both operands are not a QNAN and
4244 <tt>op1</tt> is greather than <tt>op2</tt>.</li>
4245 <li><tt>oge</tt>: yields <tt>true</tt> if both operands are not a QNAN and
4246 <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
4247 <li><tt>olt</tt>: yields <tt>true</tt> if both operands are not a QNAN and
4248 <tt>op1</tt> is less than <tt>op2</tt>.</li>
4249 <li><tt>ole</tt>: yields <tt>true</tt> if both operands are not a QNAN and
4250 <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
4251 <li><tt>one</tt>: yields <tt>true</tt> if both operands are not a QNAN and
4252 <tt>op1</tt> is not equal to <tt>op2</tt>.</li>
4253 <li><tt>ord</tt>: yields <tt>true</tt> if both operands are not a QNAN.</li>
4254 <li><tt>ueq</tt>: yields <tt>true</tt> if either operand is a QNAN or
4255 <tt>op1</tt> is equal to <tt>op2</tt>.</li>
4256 <li><tt>ugt</tt>: yields <tt>true</tt> if either operand is a QNAN or
4257 <tt>op1</tt> is greater than <tt>op2</tt>.</li>
4258 <li><tt>uge</tt>: yields <tt>true</tt> if either operand is a QNAN or
4259 <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
4260 <li><tt>ult</tt>: yields <tt>true</tt> if either operand is a QNAN or
4261 <tt>op1</tt> is less than <tt>op2</tt>.</li>
4262 <li><tt>ule</tt>: yields <tt>true</tt> if either operand is a QNAN or
4263 <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
4264 <li><tt>une</tt>: yields <tt>true</tt> if either operand is a QNAN or
4265 <tt>op1</tt> is not equal to <tt>op2</tt>.</li>
4266 <li><tt>uno</tt>: yields <tt>true</tt> if either operand is a QNAN.</li>
4267 <li><tt>true</tt>: always yields <tt>true</tt>, regardless of operands.</li>
4271 <pre> <result> = fcmp oeq float 4.0, 5.0 <i>; yields: result=false</i>
4272 <result> = fcmp one float 4.0, 5.0 <i>; yields: result=true</i>
4273 <result> = fcmp olt float 4.0, 5.0 <i>; yields: result=true</i>
4274 <result> = fcmp ueq double 1.0, 2.0 <i>; yields: result=false</i>
4277 <p>Note that the code generator does not yet support vector types with
4278 the <tt>fcmp</tt> instruction.</p>
4282 <!-- _______________________________________________________________________ -->
4283 <div class="doc_subsubsection">
4284 <a name="i_vicmp">'<tt>vicmp</tt>' Instruction</a>
4286 <div class="doc_text">
4288 <pre> <result> = vicmp <cond> <ty> <op1>, <op2> <i>; yields {ty}:result</i>
4291 <p>The '<tt>vicmp</tt>' instruction returns an integer vector value based on
4292 element-wise comparison of its two integer vector operands.</p>
4294 <p>The '<tt>vicmp</tt>' instruction takes three operands. The first operand is
4295 the condition code indicating the kind of comparison to perform. It is not
4296 a value, just a keyword. The possible condition code are:</p>
4298 <li><tt>eq</tt>: equal</li>
4299 <li><tt>ne</tt>: not equal </li>
4300 <li><tt>ugt</tt>: unsigned greater than</li>
4301 <li><tt>uge</tt>: unsigned greater or equal</li>
4302 <li><tt>ult</tt>: unsigned less than</li>
4303 <li><tt>ule</tt>: unsigned less or equal</li>
4304 <li><tt>sgt</tt>: signed greater than</li>
4305 <li><tt>sge</tt>: signed greater or equal</li>
4306 <li><tt>slt</tt>: signed less than</li>
4307 <li><tt>sle</tt>: signed less or equal</li>
4309 <p>The remaining two arguments must be <a href="#t_vector">vector</a> or
4310 <a href="#t_integer">integer</a> typed. They must also be identical types.</p>
4312 <p>The '<tt>vicmp</tt>' instruction compares <tt>op1</tt> and <tt>op2</tt>
4313 according to the condition code given as <tt>cond</tt>. The comparison yields a
4314 <a href="#t_vector">vector</a> of <a href="#t_integer">integer</a> result, of
4315 identical type as the values being compared. The most significant bit in each
4316 element is 1 if the element-wise comparison evaluates to true, and is 0
4317 otherwise. All other bits of the result are undefined. The condition codes
4318 are evaluated identically to the <a href="#i_icmp">'<tt>icmp</tt>'
4319 instruction</a>.</p>
4323 <result> = vicmp eq <2 x i32> < i32 4, i32 0>, < i32 5, i32 0> <i>; yields: result=<2 x i32> < i32 0, i32 -1 ></i>
4324 <result> = vicmp ult <2 x i8 > < i8 1, i8 2>, < i8 2, i8 2 > <i>; yields: result=<2 x i8> < i8 -1, i8 0 ></i>
4328 <!-- _______________________________________________________________________ -->
4329 <div class="doc_subsubsection">
4330 <a name="i_vfcmp">'<tt>vfcmp</tt>' Instruction</a>
4332 <div class="doc_text">
4334 <pre> <result> = vfcmp <cond> <ty> <op1>, <op2></pre>
4336 <p>The '<tt>vfcmp</tt>' instruction returns an integer vector value based on
4337 element-wise comparison of its two floating point vector operands. The output
4338 elements have the same width as the input elements.</p>
4340 <p>The '<tt>vfcmp</tt>' instruction takes three operands. The first operand is
4341 the condition code indicating the kind of comparison to perform. It is not
4342 a value, just a keyword. The possible condition code are:</p>
4344 <li><tt>false</tt>: no comparison, always returns false</li>
4345 <li><tt>oeq</tt>: ordered and equal</li>
4346 <li><tt>ogt</tt>: ordered and greater than </li>
4347 <li><tt>oge</tt>: ordered and greater than or equal</li>
4348 <li><tt>olt</tt>: ordered and less than </li>
4349 <li><tt>ole</tt>: ordered and less than or equal</li>
4350 <li><tt>one</tt>: ordered and not equal</li>
4351 <li><tt>ord</tt>: ordered (no nans)</li>
4352 <li><tt>ueq</tt>: unordered or equal</li>
4353 <li><tt>ugt</tt>: unordered or greater than </li>
4354 <li><tt>uge</tt>: unordered or greater than or equal</li>
4355 <li><tt>ult</tt>: unordered or less than </li>
4356 <li><tt>ule</tt>: unordered or less than or equal</li>
4357 <li><tt>une</tt>: unordered or not equal</li>
4358 <li><tt>uno</tt>: unordered (either nans)</li>
4359 <li><tt>true</tt>: no comparison, always returns true</li>
4361 <p>The remaining two arguments must be <a href="#t_vector">vector</a> of
4362 <a href="#t_floating">floating point</a> typed. They must also be identical
4365 <p>The '<tt>vfcmp</tt>' instruction compares <tt>op1</tt> and <tt>op2</tt>
4366 according to the condition code given as <tt>cond</tt>. The comparison yields a
4367 <a href="#t_vector">vector</a> of <a href="#t_integer">integer</a> result, with
4368 an identical number of elements as the values being compared, and each element
4369 having identical with to the width of the floating point elements. The most
4370 significant bit in each element is 1 if the element-wise comparison evaluates to
4371 true, and is 0 otherwise. All other bits of the result are undefined. The
4372 condition codes are evaluated identically to the
4373 <a href="#i_fcmp">'<tt>fcmp</tt>' instruction</a>.</p>
4377 <i>; yields: result=<2 x i32> < i32 0, i32 -1 ></i>
4378 <result> = vfcmp oeq <2 x float> < float 4, float 0 >, < float 5, float 0 >
4380 <i>; yields: result=<2 x i64> < i64 -1, i64 0 ></i>
4381 <result> = vfcmp ult <2 x double> < double 1, double 2 >, < double 2, double 2>
4385 <!-- _______________________________________________________________________ -->
4386 <div class="doc_subsubsection">
4387 <a name="i_phi">'<tt>phi</tt>' Instruction</a>
4390 <div class="doc_text">
4394 <pre> <result> = phi <ty> [ <val0>, <label0>], ...<br></pre>
4396 <p>The '<tt>phi</tt>' instruction is used to implement the φ node in
4397 the SSA graph representing the function.</p>
4400 <p>The type of the incoming values is specified with the first type
4401 field. After this, the '<tt>phi</tt>' instruction takes a list of pairs
4402 as arguments, with one pair for each predecessor basic block of the
4403 current block. Only values of <a href="#t_firstclass">first class</a>
4404 type may be used as the value arguments to the PHI node. Only labels
4405 may be used as the label arguments.</p>
4407 <p>There must be no non-phi instructions between the start of a basic
4408 block and the PHI instructions: i.e. PHI instructions must be first in
4413 <p>At runtime, the '<tt>phi</tt>' instruction logically takes on the value
4414 specified by the pair corresponding to the predecessor basic block that executed
4415 just prior to the current block.</p>
4419 Loop: ; Infinite loop that counts from 0 on up...
4420 %indvar = phi i32 [ 0, %LoopHeader ], [ %nextindvar, %Loop ]
4421 %nextindvar = add i32 %indvar, 1
4426 <!-- _______________________________________________________________________ -->
4427 <div class="doc_subsubsection">
4428 <a name="i_select">'<tt>select</tt>' Instruction</a>
4431 <div class="doc_text">
4436 <result> = select <i>selty</i> <cond>, <ty> <val1>, <ty> <val2> <i>; yields ty</i>
4438 <i>selty</i> is either i1 or {<N x i1>}
4444 The '<tt>select</tt>' instruction is used to choose one value based on a
4445 condition, without branching.
4452 The '<tt>select</tt>' instruction requires an 'i1' value or
4453 a vector of 'i1' values indicating the
4454 condition, and two values of the same <a href="#t_firstclass">first class</a>
4455 type. If the val1/val2 are vectors and
4456 the condition is a scalar, then entire vectors are selected, not
4457 individual elements.
4463 If the condition is an i1 and it evaluates to 1, the instruction returns the first
4464 value argument; otherwise, it returns the second value argument.
4467 If the condition is a vector of i1, then the value arguments must
4468 be vectors of the same size, and the selection is done element
4475 %X = select i1 true, i8 17, i8 42 <i>; yields i8:17</i>
4478 <p>Note that the code generator does not yet support conditions
4479 with vector type.</p>
4484 <!-- _______________________________________________________________________ -->
4485 <div class="doc_subsubsection">
4486 <a name="i_call">'<tt>call</tt>' Instruction</a>
4489 <div class="doc_text">
4493 <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>]
4498 <p>The '<tt>call</tt>' instruction represents a simple function call.</p>
4502 <p>This instruction requires several arguments:</p>
4506 <p>The optional "tail" marker indicates whether the callee function accesses
4507 any allocas or varargs in the caller. If the "tail" marker is present, the
4508 function call is eligible for tail call optimization. Note that calls may
4509 be marked "tail" even if they do not occur before a <a
4510 href="#i_ret"><tt>ret</tt></a> instruction.</p>
4513 <p>The optional "cconv" marker indicates which <a href="#callingconv">calling
4514 convention</a> the call should use. If none is specified, the call defaults
4515 to using C calling conventions.</p>
4519 <p>The optional <a href="#paramattrs">Parameter Attributes</a> list for
4520 return values. Only '<tt>zeroext</tt>', '<tt>signext</tt>',
4521 and '<tt>inreg</tt>' attributes are valid here.</p>
4525 <p>'<tt>ty</tt>': the type of the call instruction itself which is also
4526 the type of the return value. Functions that return no value are marked
4527 <tt><a href="#t_void">void</a></tt>.</p>
4530 <p>'<tt>fnty</tt>': shall be the signature of the pointer to function
4531 value being invoked. The argument types must match the types implied by
4532 this signature. This type can be omitted if the function is not varargs
4533 and if the function type does not return a pointer to a function.</p>
4536 <p>'<tt>fnptrval</tt>': An LLVM value containing a pointer to a function to
4537 be invoked. In most cases, this is a direct function invocation, but
4538 indirect <tt>call</tt>s are just as possible, calling an arbitrary pointer
4539 to function value.</p>
4542 <p>'<tt>function args</tt>': argument list whose types match the
4543 function signature argument types. All arguments must be of
4544 <a href="#t_firstclass">first class</a> type. If the function signature
4545 indicates the function accepts a variable number of arguments, the extra
4546 arguments can be specified.</p>
4549 <p>The optional <a href="#fnattrs">function attributes</a> list. Only
4550 '<tt>noreturn</tt>', '<tt>nounwind</tt>', '<tt>readonly</tt>' and
4551 '<tt>readnone</tt>' attributes are valid here.</p>
4557 <p>The '<tt>call</tt>' instruction is used to cause control flow to
4558 transfer to a specified function, with its incoming arguments bound to
4559 the specified values. Upon a '<tt><a href="#i_ret">ret</a></tt>'
4560 instruction in the called function, control flow continues with the
4561 instruction after the function call, and the return value of the
4562 function is bound to the result argument.</p>
4567 %retval = call i32 @test(i32 %argc)
4568 call i32 (i8 *, ...)* @printf(i8 * %msg, i32 12, i8 42) <i>; yields i32</i>
4569 %X = tail call i32 @foo() <i>; yields i32</i>
4570 %Y = tail call <a href="#callingconv">fastcc</a> i32 @foo() <i>; yields i32</i>
4571 call void %foo(i8 97 signext)
4573 %struct.A = type { i32, i8 }
4574 %r = call %struct.A @foo() <i>; yields { 32, i8 }</i>
4575 %gr = extractvalue %struct.A %r, 0 <i>; yields i32</i>
4576 %gr1 = extractvalue %struct.A %r, 1 <i>; yields i8</i>
4577 %Z = call void @foo() noreturn <i>; indicates that %foo never returns normally</i>
4578 %ZZ = call zeroext i32 @bar() <i>; Return value is %zero extended</i>
4583 <!-- _______________________________________________________________________ -->
4584 <div class="doc_subsubsection">
4585 <a name="i_va_arg">'<tt>va_arg</tt>' Instruction</a>
4588 <div class="doc_text">
4593 <resultval> = va_arg <va_list*> <arglist>, <argty>
4598 <p>The '<tt>va_arg</tt>' instruction is used to access arguments passed through
4599 the "variable argument" area of a function call. It is used to implement the
4600 <tt>va_arg</tt> macro in C.</p>
4604 <p>This instruction takes a <tt>va_list*</tt> value and the type of
4605 the argument. It returns a value of the specified argument type and
4606 increments the <tt>va_list</tt> to point to the next argument. The
4607 actual type of <tt>va_list</tt> is target specific.</p>
4611 <p>The '<tt>va_arg</tt>' instruction loads an argument of the specified
4612 type from the specified <tt>va_list</tt> and causes the
4613 <tt>va_list</tt> to point to the next argument. For more information,
4614 see the variable argument handling <a href="#int_varargs">Intrinsic
4617 <p>It is legal for this instruction to be called in a function which does not
4618 take a variable number of arguments, for example, the <tt>vfprintf</tt>
4621 <p><tt>va_arg</tt> is an LLVM instruction instead of an <a
4622 href="#intrinsics">intrinsic function</a> because it takes a type as an
4627 <p>See the <a href="#int_varargs">variable argument processing</a> section.</p>
4629 <p>Note that the code generator does not yet fully support va_arg
4630 on many targets. Also, it does not currently support va_arg with
4631 aggregate types on any target.</p>
4635 <!-- *********************************************************************** -->
4636 <div class="doc_section"> <a name="intrinsics">Intrinsic Functions</a> </div>
4637 <!-- *********************************************************************** -->
4639 <div class="doc_text">
4641 <p>LLVM supports the notion of an "intrinsic function". These functions have
4642 well known names and semantics and are required to follow certain restrictions.
4643 Overall, these intrinsics represent an extension mechanism for the LLVM
4644 language that does not require changing all of the transformations in LLVM when
4645 adding to the language (or the bitcode reader/writer, the parser, etc...).</p>
4647 <p>Intrinsic function names must all start with an "<tt>llvm.</tt>" prefix. This
4648 prefix is reserved in LLVM for intrinsic names; thus, function names may not
4649 begin with this prefix. Intrinsic functions must always be external functions:
4650 you cannot define the body of intrinsic functions. Intrinsic functions may
4651 only be used in call or invoke instructions: it is illegal to take the address
4652 of an intrinsic function. Additionally, because intrinsic functions are part
4653 of the LLVM language, it is required if any are added that they be documented
4656 <p>Some intrinsic functions can be overloaded, i.e., the intrinsic represents
4657 a family of functions that perform the same operation but on different data
4658 types. Because LLVM can represent over 8 million different integer types,
4659 overloading is used commonly to allow an intrinsic function to operate on any
4660 integer type. One or more of the argument types or the result type can be
4661 overloaded to accept any integer type. Argument types may also be defined as
4662 exactly matching a previous argument's type or the result type. This allows an
4663 intrinsic function which accepts multiple arguments, but needs all of them to
4664 be of the same type, to only be overloaded with respect to a single argument or
4667 <p>Overloaded intrinsics will have the names of its overloaded argument types
4668 encoded into its function name, each preceded by a period. Only those types
4669 which are overloaded result in a name suffix. Arguments whose type is matched
4670 against another type do not. For example, the <tt>llvm.ctpop</tt> function can
4671 take an integer of any width and returns an integer of exactly the same integer
4672 width. This leads to a family of functions such as
4673 <tt>i8 @llvm.ctpop.i8(i8 %val)</tt> and <tt>i29 @llvm.ctpop.i29(i29 %val)</tt>.
4674 Only one type, the return type, is overloaded, and only one type suffix is
4675 required. Because the argument's type is matched against the return type, it
4676 does not require its own name suffix.</p>
4678 <p>To learn how to add an intrinsic function, please see the
4679 <a href="ExtendingLLVM.html">Extending LLVM Guide</a>.
4684 <!-- ======================================================================= -->
4685 <div class="doc_subsection">
4686 <a name="int_varargs">Variable Argument Handling Intrinsics</a>
4689 <div class="doc_text">
4691 <p>Variable argument support is defined in LLVM with the <a
4692 href="#i_va_arg"><tt>va_arg</tt></a> instruction and these three
4693 intrinsic functions. These functions are related to the similarly
4694 named macros defined in the <tt><stdarg.h></tt> header file.</p>
4696 <p>All of these functions operate on arguments that use a
4697 target-specific value type "<tt>va_list</tt>". The LLVM assembly
4698 language reference manual does not define what this type is, so all
4699 transformations should be prepared to handle these functions regardless of
4702 <p>This example shows how the <a href="#i_va_arg"><tt>va_arg</tt></a>
4703 instruction and the variable argument handling intrinsic functions are
4706 <div class="doc_code">
4708 define i32 @test(i32 %X, ...) {
4709 ; Initialize variable argument processing
4711 %ap2 = bitcast i8** %ap to i8*
4712 call void @llvm.va_start(i8* %ap2)
4714 ; Read a single integer argument
4715 %tmp = va_arg i8** %ap, i32
4717 ; Demonstrate usage of llvm.va_copy and llvm.va_end
4719 %aq2 = bitcast i8** %aq to i8*
4720 call void @llvm.va_copy(i8* %aq2, i8* %ap2)
4721 call void @llvm.va_end(i8* %aq2)
4723 ; Stop processing of arguments.
4724 call void @llvm.va_end(i8* %ap2)
4728 declare void @llvm.va_start(i8*)
4729 declare void @llvm.va_copy(i8*, i8*)
4730 declare void @llvm.va_end(i8*)
4736 <!-- _______________________________________________________________________ -->
4737 <div class="doc_subsubsection">
4738 <a name="int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a>
4742 <div class="doc_text">
4744 <pre> declare void %llvm.va_start(i8* <arglist>)<br></pre>
4746 <p>The '<tt>llvm.va_start</tt>' intrinsic initializes
4747 <tt>*<arglist></tt> for subsequent use by <tt><a
4748 href="#i_va_arg">va_arg</a></tt>.</p>
4752 <p>The argument is a pointer to a <tt>va_list</tt> element to initialize.</p>
4756 <p>The '<tt>llvm.va_start</tt>' intrinsic works just like the <tt>va_start</tt>
4757 macro available in C. In a target-dependent way, it initializes the
4758 <tt>va_list</tt> element to which the argument points, so that the next call to
4759 <tt>va_arg</tt> will produce the first variable argument passed to the function.
4760 Unlike the C <tt>va_start</tt> macro, this intrinsic does not need to know the
4761 last argument of the function as the compiler can figure that out.</p>
4765 <!-- _______________________________________________________________________ -->
4766 <div class="doc_subsubsection">
4767 <a name="int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a>
4770 <div class="doc_text">
4772 <pre> declare void @llvm.va_end(i8* <arglist>)<br></pre>
4775 <p>The '<tt>llvm.va_end</tt>' intrinsic destroys <tt>*<arglist></tt>,
4776 which has been initialized previously with <tt><a href="#int_va_start">llvm.va_start</a></tt>
4777 or <tt><a href="#i_va_copy">llvm.va_copy</a></tt>.</p>
4781 <p>The argument is a pointer to a <tt>va_list</tt> to destroy.</p>
4785 <p>The '<tt>llvm.va_end</tt>' intrinsic works just like the <tt>va_end</tt>
4786 macro available in C. In a target-dependent way, it destroys the
4787 <tt>va_list</tt> element to which the argument points. Calls to <a
4788 href="#int_va_start"><tt>llvm.va_start</tt></a> and <a href="#int_va_copy">
4789 <tt>llvm.va_copy</tt></a> must be matched exactly with calls to
4790 <tt>llvm.va_end</tt>.</p>
4794 <!-- _______________________________________________________________________ -->
4795 <div class="doc_subsubsection">
4796 <a name="int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a>
4799 <div class="doc_text">
4804 declare void @llvm.va_copy(i8* <destarglist>, i8* <srcarglist>)
4809 <p>The '<tt>llvm.va_copy</tt>' intrinsic copies the current argument position
4810 from the source argument list to the destination argument list.</p>
4814 <p>The first argument is a pointer to a <tt>va_list</tt> element to initialize.
4815 The second argument is a pointer to a <tt>va_list</tt> element to copy from.</p>
4820 <p>The '<tt>llvm.va_copy</tt>' intrinsic works just like the <tt>va_copy</tt>
4821 macro available in C. In a target-dependent way, it copies the source
4822 <tt>va_list</tt> element into the destination <tt>va_list</tt> element. This
4823 intrinsic is necessary because the <tt><a href="#int_va_start">
4824 llvm.va_start</a></tt> intrinsic may be arbitrarily complex and require, for
4825 example, memory allocation.</p>
4829 <!-- ======================================================================= -->
4830 <div class="doc_subsection">
4831 <a name="int_gc">Accurate Garbage Collection Intrinsics</a>
4834 <div class="doc_text">
4837 LLVM support for <a href="GarbageCollection.html">Accurate Garbage
4838 Collection</a> (GC) requires the implementation and generation of these
4840 These intrinsics allow identification of <a href="#int_gcroot">GC roots on the
4841 stack</a>, as well as garbage collector implementations that require <a
4842 href="#int_gcread">read</a> and <a href="#int_gcwrite">write</a> barriers.
4843 Front-ends for type-safe garbage collected languages should generate these
4844 intrinsics to make use of the LLVM garbage collectors. For more details, see <a
4845 href="GarbageCollection.html">Accurate Garbage Collection with LLVM</a>.
4848 <p>The garbage collection intrinsics only operate on objects in the generic
4849 address space (address space zero).</p>
4853 <!-- _______________________________________________________________________ -->
4854 <div class="doc_subsubsection">
4855 <a name="int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a>
4858 <div class="doc_text">
4863 declare void @llvm.gcroot(i8** %ptrloc, i8* %metadata)
4868 <p>The '<tt>llvm.gcroot</tt>' intrinsic declares the existence of a GC root to
4869 the code generator, and allows some metadata to be associated with it.</p>
4873 <p>The first argument specifies the address of a stack object that contains the
4874 root pointer. The second pointer (which must be either a constant or a global
4875 value address) contains the meta-data to be associated with the root.</p>
4879 <p>At runtime, a call to this intrinsic stores a null pointer into the "ptrloc"
4880 location. At compile-time, the code generator generates information to allow
4881 the runtime to find the pointer at GC safe points. The '<tt>llvm.gcroot</tt>'
4882 intrinsic may only be used in a function which <a href="#gc">specifies a GC
4888 <!-- _______________________________________________________________________ -->
4889 <div class="doc_subsubsection">
4890 <a name="int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a>
4893 <div class="doc_text">
4898 declare i8* @llvm.gcread(i8* %ObjPtr, i8** %Ptr)
4903 <p>The '<tt>llvm.gcread</tt>' intrinsic identifies reads of references from heap
4904 locations, allowing garbage collector implementations that require read
4909 <p>The second argument is the address to read from, which should be an address
4910 allocated from the garbage collector. The first object is a pointer to the
4911 start of the referenced object, if needed by the language runtime (otherwise
4916 <p>The '<tt>llvm.gcread</tt>' intrinsic has the same semantics as a load
4917 instruction, but may be replaced with substantially more complex code by the
4918 garbage collector runtime, as needed. The '<tt>llvm.gcread</tt>' intrinsic
4919 may only be used in a function which <a href="#gc">specifies a GC
4925 <!-- _______________________________________________________________________ -->
4926 <div class="doc_subsubsection">
4927 <a name="int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a>
4930 <div class="doc_text">
4935 declare void @llvm.gcwrite(i8* %P1, i8* %Obj, i8** %P2)
4940 <p>The '<tt>llvm.gcwrite</tt>' intrinsic identifies writes of references to heap
4941 locations, allowing garbage collector implementations that require write
4942 barriers (such as generational or reference counting collectors).</p>
4946 <p>The first argument is the reference to store, the second is the start of the
4947 object to store it to, and the third is the address of the field of Obj to
4948 store to. If the runtime does not require a pointer to the object, Obj may be
4953 <p>The '<tt>llvm.gcwrite</tt>' intrinsic has the same semantics as a store
4954 instruction, but may be replaced with substantially more complex code by the
4955 garbage collector runtime, as needed. The '<tt>llvm.gcwrite</tt>' intrinsic
4956 may only be used in a function which <a href="#gc">specifies a GC
4963 <!-- ======================================================================= -->
4964 <div class="doc_subsection">
4965 <a name="int_codegen">Code Generator Intrinsics</a>
4968 <div class="doc_text">
4970 These intrinsics are provided by LLVM to expose special features that may only
4971 be implemented with code generator support.
4976 <!-- _______________________________________________________________________ -->
4977 <div class="doc_subsubsection">
4978 <a name="int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a>
4981 <div class="doc_text">
4985 declare i8 *@llvm.returnaddress(i32 <level>)
4991 The '<tt>llvm.returnaddress</tt>' intrinsic attempts to compute a
4992 target-specific value indicating the return address of the current function
4993 or one of its callers.
4999 The argument to this intrinsic indicates which function to return the address
5000 for. Zero indicates the calling function, one indicates its caller, etc. The
5001 argument is <b>required</b> to be a constant integer value.
5007 The '<tt>llvm.returnaddress</tt>' intrinsic either returns a pointer indicating
5008 the return address of the specified call frame, or zero if it cannot be
5009 identified. The value returned by this intrinsic is likely to be incorrect or 0
5010 for arguments other than zero, so it should only be used for debugging purposes.
5014 Note that calling this intrinsic does not prevent function inlining or other
5015 aggressive transformations, so the value returned may not be that of the obvious
5016 source-language caller.
5021 <!-- _______________________________________________________________________ -->
5022 <div class="doc_subsubsection">
5023 <a name="int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a>
5026 <div class="doc_text">
5030 declare i8 *@llvm.frameaddress(i32 <level>)
5036 The '<tt>llvm.frameaddress</tt>' intrinsic attempts to return the
5037 target-specific frame pointer value for the specified stack frame.
5043 The argument to this intrinsic indicates which function to return the frame
5044 pointer for. Zero indicates the calling function, one indicates its caller,
5045 etc. The argument is <b>required</b> to be a constant integer value.
5051 The '<tt>llvm.frameaddress</tt>' intrinsic either returns a pointer indicating
5052 the frame address of the specified call frame, or zero if it cannot be
5053 identified. The value returned by this intrinsic is likely to be incorrect or 0
5054 for arguments other than zero, so it should only be used for debugging purposes.
5058 Note that calling this intrinsic does not prevent function inlining or other
5059 aggressive transformations, so the value returned may not be that of the obvious
5060 source-language caller.
5064 <!-- _______________________________________________________________________ -->
5065 <div class="doc_subsubsection">
5066 <a name="int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a>
5069 <div class="doc_text">
5073 declare i8 *@llvm.stacksave()
5079 The '<tt>llvm.stacksave</tt>' intrinsic is used to remember the current state of
5080 the function stack, for use with <a href="#int_stackrestore">
5081 <tt>llvm.stackrestore</tt></a>. This is useful for implementing language
5082 features like scoped automatic variable sized arrays in C99.
5088 This intrinsic returns a opaque pointer value that can be passed to <a
5089 href="#int_stackrestore"><tt>llvm.stackrestore</tt></a>. When an
5090 <tt>llvm.stackrestore</tt> intrinsic is executed with a value saved from
5091 <tt>llvm.stacksave</tt>, it effectively restores the state of the stack to the
5092 state it was in when the <tt>llvm.stacksave</tt> intrinsic executed. In
5093 practice, this pops any <a href="#i_alloca">alloca</a> blocks from the stack
5094 that were allocated after the <tt>llvm.stacksave</tt> was executed.
5099 <!-- _______________________________________________________________________ -->
5100 <div class="doc_subsubsection">
5101 <a name="int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a>
5104 <div class="doc_text">
5108 declare void @llvm.stackrestore(i8 * %ptr)
5114 The '<tt>llvm.stackrestore</tt>' intrinsic is used to restore the state of
5115 the function stack to the state it was in when the corresponding <a
5116 href="#int_stacksave"><tt>llvm.stacksave</tt></a> intrinsic executed. This is
5117 useful for implementing language features like scoped automatic variable sized
5124 See the description for <a href="#int_stacksave"><tt>llvm.stacksave</tt></a>.
5130 <!-- _______________________________________________________________________ -->
5131 <div class="doc_subsubsection">
5132 <a name="int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a>
5135 <div class="doc_text">
5139 declare void @llvm.prefetch(i8* <address>, i32 <rw>, i32 <locality>)
5146 The '<tt>llvm.prefetch</tt>' intrinsic is a hint to the code generator to insert
5147 a prefetch instruction if supported; otherwise, it is a noop. Prefetches have
5149 effect on the behavior of the program but can change its performance
5156 <tt>address</tt> is the address to be prefetched, <tt>rw</tt> is the specifier
5157 determining if the fetch should be for a read (0) or write (1), and
5158 <tt>locality</tt> is a temporal locality specifier ranging from (0) - no
5159 locality, to (3) - extremely local keep in cache. The <tt>rw</tt> and
5160 <tt>locality</tt> arguments must be constant integers.
5166 This intrinsic does not modify the behavior of the program. In particular,
5167 prefetches cannot trap and do not produce a value. On targets that support this
5168 intrinsic, the prefetch can provide hints to the processor cache for better
5174 <!-- _______________________________________________________________________ -->
5175 <div class="doc_subsubsection">
5176 <a name="int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a>
5179 <div class="doc_text">
5183 declare void @llvm.pcmarker(i32 <id>)
5190 The '<tt>llvm.pcmarker</tt>' intrinsic is a method to export a Program Counter
5192 code to simulators and other tools. The method is target specific, but it is
5193 expected that the marker will use exported symbols to transmit the PC of the
5195 The marker makes no guarantees that it will remain with any specific instruction
5196 after optimizations. It is possible that the presence of a marker will inhibit
5197 optimizations. The intended use is to be inserted after optimizations to allow
5198 correlations of simulation runs.
5204 <tt>id</tt> is a numerical id identifying the marker.
5210 This intrinsic does not modify the behavior of the program. Backends that do not
5211 support this intrinisic may ignore it.
5216 <!-- _______________________________________________________________________ -->
5217 <div class="doc_subsubsection">
5218 <a name="int_readcyclecounter">'<tt>llvm.readcyclecounter</tt>' Intrinsic</a>
5221 <div class="doc_text">
5225 declare i64 @llvm.readcyclecounter( )
5232 The '<tt>llvm.readcyclecounter</tt>' intrinsic provides access to the cycle
5233 counter register (or similar low latency, high accuracy clocks) on those targets
5234 that support it. On X86, it should map to RDTSC. On Alpha, it should map to RPCC.
5235 As the backing counters overflow quickly (on the order of 9 seconds on alpha), this
5236 should only be used for small timings.
5242 When directly supported, reading the cycle counter should not modify any memory.
5243 Implementations are allowed to either return a application specific value or a
5244 system wide value. On backends without support, this is lowered to a constant 0.
5249 <!-- ======================================================================= -->
5250 <div class="doc_subsection">
5251 <a name="int_libc">Standard C Library Intrinsics</a>
5254 <div class="doc_text">
5256 LLVM provides intrinsics for a few important standard C library functions.
5257 These intrinsics allow source-language front-ends to pass information about the
5258 alignment of the pointer arguments to the code generator, providing opportunity
5259 for more efficient code generation.
5264 <!-- _______________________________________________________________________ -->
5265 <div class="doc_subsubsection">
5266 <a name="int_memcpy">'<tt>llvm.memcpy</tt>' Intrinsic</a>
5269 <div class="doc_text">
5272 <p>This is an overloaded intrinsic. You can use llvm.memcpy on any integer bit
5273 width. Not all targets support all bit widths however.</p>
5275 declare void @llvm.memcpy.i8(i8 * <dest>, i8 * <src>,
5276 i8 <len>, i32 <align>)
5277 declare void @llvm.memcpy.i16(i8 * <dest>, i8 * <src>,
5278 i16 <len>, i32 <align>)
5279 declare void @llvm.memcpy.i32(i8 * <dest>, i8 * <src>,
5280 i32 <len>, i32 <align>)
5281 declare void @llvm.memcpy.i64(i8 * <dest>, i8 * <src>,
5282 i64 <len>, i32 <align>)
5288 The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the source
5289 location to the destination location.
5293 Note that, unlike the standard libc function, the <tt>llvm.memcpy.*</tt>
5294 intrinsics do not return a value, and takes an extra alignment argument.
5300 The first argument is a pointer to the destination, the second is a pointer to
5301 the source. The third argument is an integer argument
5302 specifying the number of bytes to copy, and the fourth argument is the alignment
5303 of the source and destination locations.
5307 If the call to this intrinisic has an alignment value that is not 0 or 1, then
5308 the caller guarantees that both the source and destination pointers are aligned
5315 The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the source
5316 location to the destination location, which are not allowed to overlap. It
5317 copies "len" bytes of memory over. If the argument is known to be aligned to
5318 some boundary, this can be specified as the fourth argument, otherwise it should
5324 <!-- _______________________________________________________________________ -->
5325 <div class="doc_subsubsection">
5326 <a name="int_memmove">'<tt>llvm.memmove</tt>' Intrinsic</a>
5329 <div class="doc_text">
5332 <p>This is an overloaded intrinsic. You can use llvm.memmove on any integer bit
5333 width. Not all targets support all bit widths however.</p>
5335 declare void @llvm.memmove.i8(i8 * <dest>, i8 * <src>,
5336 i8 <len>, i32 <align>)
5337 declare void @llvm.memmove.i16(i8 * <dest>, i8 * <src>,
5338 i16 <len>, i32 <align>)
5339 declare void @llvm.memmove.i32(i8 * <dest>, i8 * <src>,
5340 i32 <len>, i32 <align>)
5341 declare void @llvm.memmove.i64(i8 * <dest>, i8 * <src>,
5342 i64 <len>, i32 <align>)
5348 The '<tt>llvm.memmove.*</tt>' intrinsics move a block of memory from the source
5349 location to the destination location. It is similar to the
5350 '<tt>llvm.memcpy</tt>' intrinsic but allows the two memory locations to overlap.
5354 Note that, unlike the standard libc function, the <tt>llvm.memmove.*</tt>
5355 intrinsics do not return a value, and takes an extra alignment argument.
5361 The first argument is a pointer to the destination, the second is a pointer to
5362 the source. The third argument is an integer argument
5363 specifying the number of bytes to copy, and the fourth argument is the alignment
5364 of the source and destination locations.
5368 If the call to this intrinisic has an alignment value that is not 0 or 1, then
5369 the caller guarantees that the source and destination pointers are aligned to
5376 The '<tt>llvm.memmove.*</tt>' intrinsics copy a block of memory from the source
5377 location to the destination location, which may overlap. It
5378 copies "len" bytes of memory over. If the argument is known to be aligned to
5379 some boundary, this can be specified as the fourth argument, otherwise it should
5385 <!-- _______________________________________________________________________ -->
5386 <div class="doc_subsubsection">
5387 <a name="int_memset">'<tt>llvm.memset.*</tt>' Intrinsics</a>
5390 <div class="doc_text">
5393 <p>This is an overloaded intrinsic. You can use llvm.memset on any integer bit
5394 width. Not all targets support all bit widths however.</p>
5396 declare void @llvm.memset.i8(i8 * <dest>, i8 <val>,
5397 i8 <len>, i32 <align>)
5398 declare void @llvm.memset.i16(i8 * <dest>, i8 <val>,
5399 i16 <len>, i32 <align>)
5400 declare void @llvm.memset.i32(i8 * <dest>, i8 <val>,
5401 i32 <len>, i32 <align>)
5402 declare void @llvm.memset.i64(i8 * <dest>, i8 <val>,
5403 i64 <len>, i32 <align>)
5409 The '<tt>llvm.memset.*</tt>' intrinsics fill a block of memory with a particular
5414 Note that, unlike the standard libc function, the <tt>llvm.memset</tt> intrinsic
5415 does not return a value, and takes an extra alignment argument.
5421 The first argument is a pointer to the destination to fill, the second is the
5422 byte value to fill it with, the third argument is an integer
5423 argument specifying the number of bytes to fill, and the fourth argument is the
5424 known alignment of destination location.
5428 If the call to this intrinisic has an alignment value that is not 0 or 1, then
5429 the caller guarantees that the destination pointer is aligned to that boundary.
5435 The '<tt>llvm.memset.*</tt>' intrinsics fill "len" bytes of memory starting at
5437 destination location. If the argument is known to be aligned to some boundary,
5438 this can be specified as the fourth argument, otherwise it should be set to 0 or
5444 <!-- _______________________________________________________________________ -->
5445 <div class="doc_subsubsection">
5446 <a name="int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a>
5449 <div class="doc_text">
5452 <p>This is an overloaded intrinsic. You can use <tt>llvm.sqrt</tt> on any
5453 floating point or vector of floating point type. Not all targets support all
5456 declare float @llvm.sqrt.f32(float %Val)
5457 declare double @llvm.sqrt.f64(double %Val)
5458 declare x86_fp80 @llvm.sqrt.f80(x86_fp80 %Val)
5459 declare fp128 @llvm.sqrt.f128(fp128 %Val)
5460 declare ppc_fp128 @llvm.sqrt.ppcf128(ppc_fp128 %Val)
5466 The '<tt>llvm.sqrt</tt>' intrinsics return the sqrt of the specified operand,
5467 returning the same value as the libm '<tt>sqrt</tt>' functions would. Unlike
5468 <tt>sqrt</tt> in libm, however, <tt>llvm.sqrt</tt> has undefined behavior for
5469 negative numbers other than -0.0 (which allows for better optimization, because
5470 there is no need to worry about errno being set). <tt>llvm.sqrt(-0.0)</tt> is
5471 defined to return -0.0 like IEEE sqrt.
5477 The argument and return value are floating point numbers of the same type.
5483 This function returns the sqrt of the specified operand if it is a nonnegative
5484 floating point number.
5488 <!-- _______________________________________________________________________ -->
5489 <div class="doc_subsubsection">
5490 <a name="int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a>
5493 <div class="doc_text">
5496 <p>This is an overloaded intrinsic. You can use <tt>llvm.powi</tt> on any
5497 floating point or vector of floating point type. Not all targets support all
5500 declare float @llvm.powi.f32(float %Val, i32 %power)
5501 declare double @llvm.powi.f64(double %Val, i32 %power)
5502 declare x86_fp80 @llvm.powi.f80(x86_fp80 %Val, i32 %power)
5503 declare fp128 @llvm.powi.f128(fp128 %Val, i32 %power)
5504 declare ppc_fp128 @llvm.powi.ppcf128(ppc_fp128 %Val, i32 %power)
5510 The '<tt>llvm.powi.*</tt>' intrinsics return the first operand raised to the
5511 specified (positive or negative) power. The order of evaluation of
5512 multiplications is not defined. When a vector of floating point type is
5513 used, the second argument remains a scalar integer value.
5519 The second argument is an integer power, and the first is a value to raise to
5526 This function returns the first value raised to the second power with an
5527 unspecified sequence of rounding operations.</p>
5530 <!-- _______________________________________________________________________ -->
5531 <div class="doc_subsubsection">
5532 <a name="int_sin">'<tt>llvm.sin.*</tt>' Intrinsic</a>
5535 <div class="doc_text">
5538 <p>This is an overloaded intrinsic. You can use <tt>llvm.sin</tt> on any
5539 floating point or vector of floating point type. Not all targets support all
5542 declare float @llvm.sin.f32(float %Val)
5543 declare double @llvm.sin.f64(double %Val)
5544 declare x86_fp80 @llvm.sin.f80(x86_fp80 %Val)
5545 declare fp128 @llvm.sin.f128(fp128 %Val)
5546 declare ppc_fp128 @llvm.sin.ppcf128(ppc_fp128 %Val)
5552 The '<tt>llvm.sin.*</tt>' intrinsics return the sine of the operand.
5558 The argument and return value are floating point numbers of the same type.
5564 This function returns the sine of the specified operand, returning the
5565 same values as the libm <tt>sin</tt> functions would, and handles error
5566 conditions in the same way.</p>
5569 <!-- _______________________________________________________________________ -->
5570 <div class="doc_subsubsection">
5571 <a name="int_cos">'<tt>llvm.cos.*</tt>' Intrinsic</a>
5574 <div class="doc_text">
5577 <p>This is an overloaded intrinsic. You can use <tt>llvm.cos</tt> on any
5578 floating point or vector of floating point type. Not all targets support all
5581 declare float @llvm.cos.f32(float %Val)
5582 declare double @llvm.cos.f64(double %Val)
5583 declare x86_fp80 @llvm.cos.f80(x86_fp80 %Val)
5584 declare fp128 @llvm.cos.f128(fp128 %Val)
5585 declare ppc_fp128 @llvm.cos.ppcf128(ppc_fp128 %Val)
5591 The '<tt>llvm.cos.*</tt>' intrinsics return the cosine of the operand.
5597 The argument and return value are floating point numbers of the same type.
5603 This function returns the cosine of the specified operand, returning the
5604 same values as the libm <tt>cos</tt> functions would, and handles error
5605 conditions in the same way.</p>
5608 <!-- _______________________________________________________________________ -->
5609 <div class="doc_subsubsection">
5610 <a name="int_pow">'<tt>llvm.pow.*</tt>' Intrinsic</a>
5613 <div class="doc_text">
5616 <p>This is an overloaded intrinsic. You can use <tt>llvm.pow</tt> on any
5617 floating point or vector of floating point type. Not all targets support all
5620 declare float @llvm.pow.f32(float %Val, float %Power)
5621 declare double @llvm.pow.f64(double %Val, double %Power)
5622 declare x86_fp80 @llvm.pow.f80(x86_fp80 %Val, x86_fp80 %Power)
5623 declare fp128 @llvm.pow.f128(fp128 %Val, fp128 %Power)
5624 declare ppc_fp128 @llvm.pow.ppcf128(ppc_fp128 %Val, ppc_fp128 Power)
5630 The '<tt>llvm.pow.*</tt>' intrinsics return the first operand raised to the
5631 specified (positive or negative) power.
5637 The second argument is a floating point power, and the first is a value to
5638 raise to that power.
5644 This function returns the first value raised to the second power,
5646 same values as the libm <tt>pow</tt> functions would, and handles error
5647 conditions in the same way.</p>
5651 <!-- ======================================================================= -->
5652 <div class="doc_subsection">
5653 <a name="int_manip">Bit Manipulation Intrinsics</a>
5656 <div class="doc_text">
5658 LLVM provides intrinsics for a few important bit manipulation operations.
5659 These allow efficient code generation for some algorithms.
5664 <!-- _______________________________________________________________________ -->
5665 <div class="doc_subsubsection">
5666 <a name="int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a>
5669 <div class="doc_text">
5672 <p>This is an overloaded intrinsic function. You can use bswap on any integer
5673 type that is an even number of bytes (i.e. BitWidth % 16 == 0).</p>
5675 declare i16 @llvm.bswap.i16(i16 <id>)
5676 declare i32 @llvm.bswap.i32(i32 <id>)
5677 declare i64 @llvm.bswap.i64(i64 <id>)
5683 The '<tt>llvm.bswap</tt>' family of intrinsics is used to byte swap integer
5684 values with an even number of bytes (positive multiple of 16 bits). These are
5685 useful for performing operations on data that is not in the target's native
5692 The <tt>llvm.bswap.i16</tt> intrinsic returns an i16 value that has the high
5693 and low byte of the input i16 swapped. Similarly, the <tt>llvm.bswap.i32</tt>
5694 intrinsic returns an i32 value that has the four bytes of the input i32
5695 swapped, so that if the input bytes are numbered 0, 1, 2, 3 then the returned
5696 i32 will have its bytes in 3, 2, 1, 0 order. The <tt>llvm.bswap.i48</tt>,
5697 <tt>llvm.bswap.i64</tt> and other intrinsics extend this concept to
5698 additional even-byte lengths (6 bytes, 8 bytes and more, respectively).
5703 <!-- _______________________________________________________________________ -->
5704 <div class="doc_subsubsection">
5705 <a name="int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic</a>
5708 <div class="doc_text">
5711 <p>This is an overloaded intrinsic. You can use llvm.ctpop on any integer bit
5712 width. Not all targets support all bit widths however.</p>
5714 declare i8 @llvm.ctpop.i8(i8 <src>)
5715 declare i16 @llvm.ctpop.i16(i16 <src>)
5716 declare i32 @llvm.ctpop.i32(i32 <src>)
5717 declare i64 @llvm.ctpop.i64(i64 <src>)
5718 declare i256 @llvm.ctpop.i256(i256 <src>)
5724 The '<tt>llvm.ctpop</tt>' family of intrinsics counts the number of bits set in a
5731 The only argument is the value to be counted. The argument may be of any
5732 integer type. The return type must match the argument type.
5738 The '<tt>llvm.ctpop</tt>' intrinsic counts the 1's in a variable.
5742 <!-- _______________________________________________________________________ -->
5743 <div class="doc_subsubsection">
5744 <a name="int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic</a>
5747 <div class="doc_text">
5750 <p>This is an overloaded intrinsic. You can use <tt>llvm.ctlz</tt> on any
5751 integer bit width. Not all targets support all bit widths however.</p>
5753 declare i8 @llvm.ctlz.i8 (i8 <src>)
5754 declare i16 @llvm.ctlz.i16(i16 <src>)
5755 declare i32 @llvm.ctlz.i32(i32 <src>)
5756 declare i64 @llvm.ctlz.i64(i64 <src>)
5757 declare i256 @llvm.ctlz.i256(i256 <src>)
5763 The '<tt>llvm.ctlz</tt>' family of intrinsic functions counts the number of
5764 leading zeros in a variable.
5770 The only argument is the value to be counted. The argument may be of any
5771 integer type. The return type must match the argument type.
5777 The '<tt>llvm.ctlz</tt>' intrinsic counts the leading (most significant) zeros
5778 in a variable. If the src == 0 then the result is the size in bits of the type
5779 of src. For example, <tt>llvm.ctlz(i32 2) = 30</tt>.
5785 <!-- _______________________________________________________________________ -->
5786 <div class="doc_subsubsection">
5787 <a name="int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic</a>
5790 <div class="doc_text">
5793 <p>This is an overloaded intrinsic. You can use <tt>llvm.cttz</tt> on any
5794 integer bit width. Not all targets support all bit widths however.</p>
5796 declare i8 @llvm.cttz.i8 (i8 <src>)
5797 declare i16 @llvm.cttz.i16(i16 <src>)
5798 declare i32 @llvm.cttz.i32(i32 <src>)
5799 declare i64 @llvm.cttz.i64(i64 <src>)
5800 declare i256 @llvm.cttz.i256(i256 <src>)
5806 The '<tt>llvm.cttz</tt>' family of intrinsic functions counts the number of
5813 The only argument is the value to be counted. The argument may be of any
5814 integer type. The return type must match the argument type.
5820 The '<tt>llvm.cttz</tt>' intrinsic counts the trailing (least significant) zeros
5821 in a variable. If the src == 0 then the result is the size in bits of the type
5822 of src. For example, <tt>llvm.cttz(2) = 1</tt>.
5826 <!-- _______________________________________________________________________ -->
5827 <div class="doc_subsubsection">
5828 <a name="int_part_select">'<tt>llvm.part.select.*</tt>' Intrinsic</a>
5831 <div class="doc_text">
5834 <p>This is an overloaded intrinsic. You can use <tt>llvm.part.select</tt>
5835 on any integer bit width.</p>
5837 declare i17 @llvm.part.select.i17 (i17 %val, i32 %loBit, i32 %hiBit)
5838 declare i29 @llvm.part.select.i29 (i29 %val, i32 %loBit, i32 %hiBit)
5842 <p>The '<tt>llvm.part.select</tt>' family of intrinsic functions selects a
5843 range of bits from an integer value and returns them in the same bit width as
5844 the original value.</p>
5847 <p>The first argument, <tt>%val</tt> and the result may be integer types of
5848 any bit width but they must have the same bit width. The second and third
5849 arguments must be <tt>i32</tt> type since they specify only a bit index.</p>
5852 <p>The operation of the '<tt>llvm.part.select</tt>' intrinsic has two modes
5853 of operation: forwards and reverse. If <tt>%loBit</tt> is greater than
5854 <tt>%hiBits</tt> then the intrinsic operates in reverse mode. Otherwise it
5855 operates in forward mode.</p>
5856 <p>In forward mode, this intrinsic is the equivalent of shifting <tt>%val</tt>
5857 right by <tt>%loBit</tt> bits and then ANDing it with a mask with
5858 only the <tt>%hiBit - %loBit</tt> bits set, as follows:</p>
5860 <li>The <tt>%val</tt> is shifted right (LSHR) by the number of bits specified
5861 by <tt>%loBits</tt>. This normalizes the value to the low order bits.</li>
5862 <li>The <tt>%loBits</tt> value is subtracted from the <tt>%hiBits</tt> value
5863 to determine the number of bits to retain.</li>
5864 <li>A mask of the retained bits is created by shifting a -1 value.</li>
5865 <li>The mask is ANDed with <tt>%val</tt> to produce the result.</li>
5867 <p>In reverse mode, a similar computation is made except that the bits are
5868 returned in the reverse order. So, for example, if <tt>X</tt> has the value
5869 <tt>i16 0x0ACF (101011001111)</tt> and we apply
5870 <tt>part.select(i16 X, 8, 3)</tt> to it, we get back the value
5871 <tt>i16 0x0026 (000000100110)</tt>.</p>
5874 <div class="doc_subsubsection">
5875 <a name="int_part_set">'<tt>llvm.part.set.*</tt>' Intrinsic</a>
5878 <div class="doc_text">
5881 <p>This is an overloaded intrinsic. You can use <tt>llvm.part.set</tt>
5882 on any integer bit width.</p>
5884 declare i17 @llvm.part.set.i17.i9 (i17 %val, i9 %repl, i32 %lo, i32 %hi)
5885 declare i29 @llvm.part.set.i29.i9 (i29 %val, i9 %repl, i32 %lo, i32 %hi)
5889 <p>The '<tt>llvm.part.set</tt>' family of intrinsic functions replaces a range
5890 of bits in an integer value with another integer value. It returns the integer
5891 with the replaced bits.</p>
5894 <p>The first argument, <tt>%val</tt>, and the result may be integer types of
5895 any bit width, but they must have the same bit width. <tt>%val</tt> is the value
5896 whose bits will be replaced. The second argument, <tt>%repl</tt> may be an
5897 integer of any bit width. The third and fourth arguments must be <tt>i32</tt>
5898 type since they specify only a bit index.</p>
5901 <p>The operation of the '<tt>llvm.part.set</tt>' intrinsic has two modes
5902 of operation: forwards and reverse. If <tt>%lo</tt> is greater than
5903 <tt>%hi</tt> then the intrinsic operates in reverse mode. Otherwise it
5904 operates in forward mode.</p>
5906 <p>For both modes, the <tt>%repl</tt> value is prepared for use by either
5907 truncating it down to the size of the replacement area or zero extending it
5908 up to that size.</p>
5910 <p>In forward mode, the bits between <tt>%lo</tt> and <tt>%hi</tt> (inclusive)
5911 are replaced with corresponding bits from <tt>%repl</tt>. That is the 0th bit
5912 in <tt>%repl</tt> replaces the <tt>%lo</tt>th bit in <tt>%val</tt> and etc. up
5913 to the <tt>%hi</tt>th bit.</p>
5915 <p>In reverse mode, a similar computation is made except that the bits are
5916 reversed. That is, the <tt>0</tt>th bit in <tt>%repl</tt> replaces the
5917 <tt>%hi</tt> bit in <tt>%val</tt> and etc. down to the <tt>%lo</tt>th bit.</p>
5922 llvm.part.set(0xFFFF, 0, 4, 7) -> 0xFF0F
5923 llvm.part.set(0xFFFF, 0, 7, 4) -> 0xFF0F
5924 llvm.part.set(0xFFFF, 1, 7, 4) -> 0xFF8F
5925 llvm.part.set(0xFFFF, F, 8, 3) -> 0xFFE7
5926 llvm.part.set(0xFFFF, 0, 3, 8) -> 0xFE07
5931 <!-- ======================================================================= -->
5932 <div class="doc_subsection">
5933 <a name="int_overflow">Arithmetic with Overflow Intrinsics</a>
5936 <div class="doc_text">
5938 LLVM provides intrinsics for some arithmetic with overflow operations.
5943 <!-- _______________________________________________________________________ -->
5944 <div class="doc_subsubsection">
5945 <a name="int_sadd_overflow">'<tt>llvm.sadd.with.overflow.*</tt>' Intrinsics</a>
5948 <div class="doc_text">
5952 <p>This is an overloaded intrinsic. You can use <tt>llvm.sadd.with.overflow</tt>
5953 on any integer bit width.</p>
5956 declare {i16, i1} @llvm.sadd.with.overflow.i16(i16 %a, i16 %b)
5957 declare {i32, i1} @llvm.sadd.with.overflow.i32(i32 %a, i32 %b)
5958 declare {i64, i1} @llvm.sadd.with.overflow.i64(i64 %a, i64 %b)
5963 <p>The '<tt>llvm.sadd.with.overflow</tt>' family of intrinsic functions perform
5964 a signed addition of the two arguments, and indicate whether an overflow
5965 occurred during the signed summation.</p>
5969 <p>The arguments (%a and %b) and the first element of the result structure may
5970 be of integer types of any bit width, but they must have the same bit width. The
5971 second element of the result structure must be of type <tt>i1</tt>. <tt>%a</tt>
5972 and <tt>%b</tt> are the two values that will undergo signed addition.</p>
5976 <p>The '<tt>llvm.sadd.with.overflow</tt>' family of intrinsic functions perform
5977 a signed addition of the two variables. They return a structure — the
5978 first element of which is the signed summation, and the second element of which
5979 is a bit specifying if the signed summation resulted in an overflow.</p>
5983 %res = call {i32, i1} @llvm.sadd.with.overflow.i32(i32 %a, i32 %b)
5984 %sum = extractvalue {i32, i1} %res, 0
5985 %obit = extractvalue {i32, i1} %res, 1
5986 br i1 %obit, label %overflow, label %normal
5991 <!-- _______________________________________________________________________ -->
5992 <div class="doc_subsubsection">
5993 <a name="int_uadd_overflow">'<tt>llvm.uadd.with.overflow.*</tt>' Intrinsics</a>
5996 <div class="doc_text">
6000 <p>This is an overloaded intrinsic. You can use <tt>llvm.uadd.with.overflow</tt>
6001 on any integer bit width.</p>
6004 declare {i16, i1} @llvm.uadd.with.overflow.i16(i16 %a, i16 %b)
6005 declare {i32, i1} @llvm.uadd.with.overflow.i32(i32 %a, i32 %b)
6006 declare {i64, i1} @llvm.uadd.with.overflow.i64(i64 %a, i64 %b)
6011 <p>The '<tt>llvm.uadd.with.overflow</tt>' family of intrinsic functions perform
6012 an unsigned addition of the two arguments, and indicate whether a carry occurred
6013 during the unsigned summation.</p>
6017 <p>The arguments (%a and %b) and the first element of the result structure may
6018 be of integer types of any bit width, but they must have the same bit width. The
6019 second element of the result structure must be of type <tt>i1</tt>. <tt>%a</tt>
6020 and <tt>%b</tt> are the two values that will undergo unsigned addition.</p>
6024 <p>The '<tt>llvm.uadd.with.overflow</tt>' family of intrinsic functions perform
6025 an unsigned addition of the two arguments. They return a structure — the
6026 first element of which is the sum, and the second element of which is a bit
6027 specifying if the unsigned summation resulted in a carry.</p>
6031 %res = call {i32, i1} @llvm.uadd.with.overflow.i32(i32 %a, i32 %b)
6032 %sum = extractvalue {i32, i1} %res, 0
6033 %obit = extractvalue {i32, i1} %res, 1
6034 br i1 %obit, label %carry, label %normal
6039 <!-- _______________________________________________________________________ -->
6040 <div class="doc_subsubsection">
6041 <a name="int_ssub_overflow">'<tt>llvm.ssub.with.overflow.*</tt>' Intrinsics</a>
6044 <div class="doc_text">
6048 <p>This is an overloaded intrinsic. You can use <tt>llvm.ssub.with.overflow</tt>
6049 on any integer bit width.</p>
6052 declare {i16, i1} @llvm.ssub.with.overflow.i16(i16 %a, i16 %b)
6053 declare {i32, i1} @llvm.ssub.with.overflow.i32(i32 %a, i32 %b)
6054 declare {i64, i1} @llvm.ssub.with.overflow.i64(i64 %a, i64 %b)
6059 <p>The '<tt>llvm.ssub.with.overflow</tt>' family of intrinsic functions perform
6060 a signed subtraction of the two arguments, and indicate whether an overflow
6061 occurred during the signed subtraction.</p>
6065 <p>The arguments (%a and %b) and the first element of the result structure may
6066 be of integer types of any bit width, but they must have the same bit width. The
6067 second element of the result structure must be of type <tt>i1</tt>. <tt>%a</tt>
6068 and <tt>%b</tt> are the two values that will undergo signed subtraction.</p>
6072 <p>The '<tt>llvm.ssub.with.overflow</tt>' family of intrinsic functions perform
6073 a signed subtraction of the two arguments. They return a structure — the
6074 first element of which is the subtraction, and the second element of which is a bit
6075 specifying if the signed subtraction resulted in an overflow.</p>
6079 %res = call {i32, i1} @llvm.ssub.with.overflow.i32(i32 %a, i32 %b)
6080 %sum = extractvalue {i32, i1} %res, 0
6081 %obit = extractvalue {i32, i1} %res, 1
6082 br i1 %obit, label %overflow, label %normal
6087 <!-- _______________________________________________________________________ -->
6088 <div class="doc_subsubsection">
6089 <a name="int_usub_overflow">'<tt>llvm.usub.with.overflow.*</tt>' Intrinsics</a>
6092 <div class="doc_text">
6096 <p>This is an overloaded intrinsic. You can use <tt>llvm.usub.with.overflow</tt>
6097 on any integer bit width.</p>
6100 declare {i16, i1} @llvm.usub.with.overflow.i16(i16 %a, i16 %b)
6101 declare {i32, i1} @llvm.usub.with.overflow.i32(i32 %a, i32 %b)
6102 declare {i64, i1} @llvm.usub.with.overflow.i64(i64 %a, i64 %b)
6107 <p>The '<tt>llvm.usub.with.overflow</tt>' family of intrinsic functions perform
6108 an unsigned subtraction of the two arguments, and indicate whether an overflow
6109 occurred during the unsigned subtraction.</p>
6113 <p>The arguments (%a and %b) and the first element of the result structure may
6114 be of integer types of any bit width, but they must have the same bit width. The
6115 second element of the result structure must be of type <tt>i1</tt>. <tt>%a</tt>
6116 and <tt>%b</tt> are the two values that will undergo unsigned subtraction.</p>
6120 <p>The '<tt>llvm.usub.with.overflow</tt>' family of intrinsic functions perform
6121 an unsigned subtraction of the two arguments. They return a structure — the
6122 first element of which is the subtraction, and the second element of which is a bit
6123 specifying if the unsigned subtraction resulted in an overflow.</p>
6127 %res = call {i32, i1} @llvm.usub.with.overflow.i32(i32 %a, i32 %b)
6128 %sum = extractvalue {i32, i1} %res, 0
6129 %obit = extractvalue {i32, i1} %res, 1
6130 br i1 %obit, label %overflow, label %normal
6135 <!-- _______________________________________________________________________ -->
6136 <div class="doc_subsubsection">
6137 <a name="int_smul_overflow">'<tt>llvm.smul.with.overflow.*</tt>' Intrinsics</a>
6140 <div class="doc_text">
6144 <p>This is an overloaded intrinsic. You can use <tt>llvm.smul.with.overflow</tt>
6145 on any integer bit width.</p>
6148 declare {i16, i1} @llvm.smul.with.overflow.i16(i16 %a, i16 %b)
6149 declare {i32, i1} @llvm.smul.with.overflow.i32(i32 %a, i32 %b)
6150 declare {i64, i1} @llvm.smul.with.overflow.i64(i64 %a, i64 %b)
6155 <p>The '<tt>llvm.smul.with.overflow</tt>' family of intrinsic functions perform
6156 a signed multiplication of the two arguments, and indicate whether an overflow
6157 occurred during the signed multiplication.</p>
6161 <p>The arguments (%a and %b) and the first element of the result structure may
6162 be of integer types of any bit width, but they must have the same bit width. The
6163 second element of the result structure must be of type <tt>i1</tt>. <tt>%a</tt>
6164 and <tt>%b</tt> are the two values that will undergo signed multiplication.</p>
6168 <p>The '<tt>llvm.smul.with.overflow</tt>' family of intrinsic functions perform
6169 a signed multiplication of the two arguments. They return a structure —
6170 the first element of which is the multiplication, and the second element of
6171 which is a bit specifying if the signed multiplication resulted in an
6176 %res = call {i32, i1} @llvm.smul.with.overflow.i32(i32 %a, i32 %b)
6177 %sum = extractvalue {i32, i1} %res, 0
6178 %obit = extractvalue {i32, i1} %res, 1
6179 br i1 %obit, label %overflow, label %normal
6184 <!-- _______________________________________________________________________ -->
6185 <div class="doc_subsubsection">
6186 <a name="int_umul_overflow">'<tt>llvm.umul.with.overflow.*</tt>' Intrinsics</a>
6189 <div class="doc_text">
6193 <p>This is an overloaded intrinsic. You can use <tt>llvm.umul.with.overflow</tt>
6194 on any integer bit width.</p>
6197 declare {i16, i1} @llvm.umul.with.overflow.i16(i16 %a, i16 %b)
6198 declare {i32, i1} @llvm.umul.with.overflow.i32(i32 %a, i32 %b)
6199 declare {i64, i1} @llvm.umul.with.overflow.i64(i64 %a, i64 %b)
6204 <p><i><b>Warning:</b> '<tt>llvm.umul.with.overflow</tt>' is badly broken. It is
6205 actively being fixed, but it should not currently be used!</i></p>
6207 <p>The '<tt>llvm.umul.with.overflow</tt>' family of intrinsic functions perform
6208 a unsigned multiplication of the two arguments, and indicate whether an overflow
6209 occurred during the unsigned multiplication.</p>
6213 <p>The arguments (%a and %b) and the first element of the result structure may
6214 be of integer types of any bit width, but they must have the same bit width. The
6215 second element of the result structure must be of type <tt>i1</tt>. <tt>%a</tt>
6216 and <tt>%b</tt> are the two values that will undergo unsigned
6221 <p>The '<tt>llvm.umul.with.overflow</tt>' family of intrinsic functions perform
6222 an unsigned multiplication of the two arguments. They return a structure —
6223 the first element of which is the multiplication, and the second element of
6224 which is a bit specifying if the unsigned multiplication resulted in an
6229 %res = call {i32, i1} @llvm.umul.with.overflow.i32(i32 %a, i32 %b)
6230 %sum = extractvalue {i32, i1} %res, 0
6231 %obit = extractvalue {i32, i1} %res, 1
6232 br i1 %obit, label %overflow, label %normal
6237 <!-- ======================================================================= -->
6238 <div class="doc_subsection">
6239 <a name="int_debugger">Debugger Intrinsics</a>
6242 <div class="doc_text">
6244 The LLVM debugger intrinsics (which all start with <tt>llvm.dbg.</tt> prefix),
6245 are described in the <a
6246 href="SourceLevelDebugging.html#format_common_intrinsics">LLVM Source Level
6247 Debugging</a> document.
6252 <!-- ======================================================================= -->
6253 <div class="doc_subsection">
6254 <a name="int_eh">Exception Handling Intrinsics</a>
6257 <div class="doc_text">
6258 <p> The LLVM exception handling intrinsics (which all start with
6259 <tt>llvm.eh.</tt> prefix), are described in the <a
6260 href="ExceptionHandling.html#format_common_intrinsics">LLVM Exception
6261 Handling</a> document. </p>
6264 <!-- ======================================================================= -->
6265 <div class="doc_subsection">
6266 <a name="int_trampoline">Trampoline Intrinsic</a>
6269 <div class="doc_text">
6271 This intrinsic makes it possible to excise one parameter, marked with
6272 the <tt>nest</tt> attribute, from a function. The result is a callable
6273 function pointer lacking the nest parameter - the caller does not need
6274 to provide a value for it. Instead, the value to use is stored in
6275 advance in a "trampoline", a block of memory usually allocated
6276 on the stack, which also contains code to splice the nest value into the
6277 argument list. This is used to implement the GCC nested function address
6281 For example, if the function is
6282 <tt>i32 f(i8* nest %c, i32 %x, i32 %y)</tt> then the resulting function
6283 pointer has signature <tt>i32 (i32, i32)*</tt>. It can be created as follows:</p>
6285 %tramp = alloca [10 x i8], align 4 ; size and alignment only correct for X86
6286 %tramp1 = getelementptr [10 x i8]* %tramp, i32 0, i32 0
6287 %p = call i8* @llvm.init.trampoline( i8* %tramp1, i8* bitcast (i32 (i8* nest , i32, i32)* @f to i8*), i8* %nval )
6288 %fp = bitcast i8* %p to i32 (i32, i32)*
6290 <p>The call <tt>%val = call i32 %fp( i32 %x, i32 %y )</tt> is then equivalent
6291 to <tt>%val = call i32 %f( i8* %nval, i32 %x, i32 %y )</tt>.</p>
6294 <!-- _______________________________________________________________________ -->
6295 <div class="doc_subsubsection">
6296 <a name="int_it">'<tt>llvm.init.trampoline</tt>' Intrinsic</a>
6298 <div class="doc_text">
6301 declare i8* @llvm.init.trampoline(i8* <tramp>, i8* <func>, i8* <nval>)
6305 This fills the memory pointed to by <tt>tramp</tt> with code
6306 and returns a function pointer suitable for executing it.
6310 The <tt>llvm.init.trampoline</tt> intrinsic takes three arguments, all
6311 pointers. The <tt>tramp</tt> argument must point to a sufficiently large
6312 and sufficiently aligned block of memory; this memory is written to by the
6313 intrinsic. Note that the size and the alignment are target-specific - LLVM
6314 currently provides no portable way of determining them, so a front-end that
6315 generates this intrinsic needs to have some target-specific knowledge.
6316 The <tt>func</tt> argument must hold a function bitcast to an <tt>i8*</tt>.
6320 The block of memory pointed to by <tt>tramp</tt> is filled with target
6321 dependent code, turning it into a function. A pointer to this function is
6322 returned, but needs to be bitcast to an
6323 <a href="#int_trampoline">appropriate function pointer type</a>
6324 before being called. The new function's signature is the same as that of
6325 <tt>func</tt> with any arguments marked with the <tt>nest</tt> attribute
6326 removed. At most one such <tt>nest</tt> argument is allowed, and it must be
6327 of pointer type. Calling the new function is equivalent to calling
6328 <tt>func</tt> with the same argument list, but with <tt>nval</tt> used for the
6329 missing <tt>nest</tt> argument. If, after calling
6330 <tt>llvm.init.trampoline</tt>, the memory pointed to by <tt>tramp</tt> is
6331 modified, then the effect of any later call to the returned function pointer is
6336 <!-- ======================================================================= -->
6337 <div class="doc_subsection">
6338 <a name="int_atomics">Atomic Operations and Synchronization Intrinsics</a>
6341 <div class="doc_text">
6343 These intrinsic functions expand the "universal IR" of LLVM to represent
6344 hardware constructs for atomic operations and memory synchronization. This
6345 provides an interface to the hardware, not an interface to the programmer. It
6346 is aimed at a low enough level to allow any programming models or APIs
6347 (Application Programming Interfaces) which
6348 need atomic behaviors to map cleanly onto it. It is also modeled primarily on
6349 hardware behavior. Just as hardware provides a "universal IR" for source
6350 languages, it also provides a starting point for developing a "universal"
6351 atomic operation and synchronization IR.
6354 These do <em>not</em> form an API such as high-level threading libraries,
6355 software transaction memory systems, atomic primitives, and intrinsic
6356 functions as found in BSD, GNU libc, atomic_ops, APR, and other system and
6357 application libraries. The hardware interface provided by LLVM should allow
6358 a clean implementation of all of these APIs and parallel programming models.
6359 No one model or paradigm should be selected above others unless the hardware
6360 itself ubiquitously does so.
6365 <!-- _______________________________________________________________________ -->
6366 <div class="doc_subsubsection">
6367 <a name="int_memory_barrier">'<tt>llvm.memory.barrier</tt>' Intrinsic</a>
6369 <div class="doc_text">
6372 declare void @llvm.memory.barrier( i1 <ll>, i1 <ls>, i1 <sl>, i1 <ss>,
6378 The <tt>llvm.memory.barrier</tt> intrinsic guarantees ordering between
6379 specific pairs of memory access types.
6383 The <tt>llvm.memory.barrier</tt> intrinsic requires five boolean arguments.
6384 The first four arguments enables a specific barrier as listed below. The fith
6385 argument specifies that the barrier applies to io or device or uncached memory.
6389 <li><tt>ll</tt>: load-load barrier</li>
6390 <li><tt>ls</tt>: load-store barrier</li>
6391 <li><tt>sl</tt>: store-load barrier</li>
6392 <li><tt>ss</tt>: store-store barrier</li>
6393 <li><tt>device</tt>: barrier applies to device and uncached memory also.</li>
6397 This intrinsic causes the system to enforce some ordering constraints upon
6398 the loads and stores of the program. This barrier does not indicate
6399 <em>when</em> any events will occur, it only enforces an <em>order</em> in
6400 which they occur. For any of the specified pairs of load and store operations
6401 (f.ex. load-load, or store-load), all of the first operations preceding the
6402 barrier will complete before any of the second operations succeeding the
6403 barrier begin. Specifically the semantics for each pairing is as follows:
6406 <li><tt>ll</tt>: All loads before the barrier must complete before any load
6407 after the barrier begins.</li>
6409 <li><tt>ls</tt>: All loads before the barrier must complete before any
6410 store after the barrier begins.</li>
6411 <li><tt>ss</tt>: All stores before the barrier must complete before any
6412 store after the barrier begins.</li>
6413 <li><tt>sl</tt>: All stores before the barrier must complete before any
6414 load after the barrier begins.</li>
6417 These semantics are applied with a logical "and" behavior when more than one
6418 is enabled in a single memory barrier intrinsic.
6421 Backends may implement stronger barriers than those requested when they do not
6422 support as fine grained a barrier as requested. Some architectures do not
6423 need all types of barriers and on such architectures, these become noops.
6430 %result1 = load i32* %ptr <i>; yields {i32}:result1 = 4</i>
6431 call void @llvm.memory.barrier( i1 false, i1 true, i1 false, i1 false )
6432 <i>; guarantee the above finishes</i>
6433 store i32 8, %ptr <i>; before this begins</i>
6437 <!-- _______________________________________________________________________ -->
6438 <div class="doc_subsubsection">
6439 <a name="int_atomic_cmp_swap">'<tt>llvm.atomic.cmp.swap.*</tt>' Intrinsic</a>
6441 <div class="doc_text">
6444 This is an overloaded intrinsic. You can use <tt>llvm.atomic.cmp.swap</tt> on
6445 any integer bit width and for different address spaces. Not all targets
6446 support all bit widths however.</p>
6449 declare i8 @llvm.atomic.cmp.swap.i8.p0i8( i8* <ptr>, i8 <cmp>, i8 <val> )
6450 declare i16 @llvm.atomic.cmp.swap.i16.p0i16( i16* <ptr>, i16 <cmp>, i16 <val> )
6451 declare i32 @llvm.atomic.cmp.swap.i32.p0i32( i32* <ptr>, i32 <cmp>, i32 <val> )
6452 declare i64 @llvm.atomic.cmp.swap.i64.p0i64( i64* <ptr>, i64 <cmp>, i64 <val> )
6457 This loads a value in memory and compares it to a given value. If they are
6458 equal, it stores a new value into the memory.
6462 The <tt>llvm.atomic.cmp.swap</tt> intrinsic takes three arguments. The result as
6463 well as both <tt>cmp</tt> and <tt>val</tt> must be integer values with the
6464 same bit width. The <tt>ptr</tt> argument must be a pointer to a value of
6465 this integer type. While any bit width integer may be used, targets may only
6466 lower representations they support in hardware.
6471 This entire intrinsic must be executed atomically. It first loads the value
6472 in memory pointed to by <tt>ptr</tt> and compares it with the value
6473 <tt>cmp</tt>. If they are equal, <tt>val</tt> is stored into the memory. The
6474 loaded value is yielded in all cases. This provides the equivalent of an
6475 atomic compare-and-swap operation within the SSA framework.
6483 %val1 = add i32 4, 4
6484 %result1 = call i32 @llvm.atomic.cmp.swap.i32.p0i32( i32* %ptr, i32 4, %val1 )
6485 <i>; yields {i32}:result1 = 4</i>
6486 %stored1 = icmp eq i32 %result1, 4 <i>; yields {i1}:stored1 = true</i>
6487 %memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 8</i>
6489 %val2 = add i32 1, 1
6490 %result2 = call i32 @llvm.atomic.cmp.swap.i32.p0i32( i32* %ptr, i32 5, %val2 )
6491 <i>; yields {i32}:result2 = 8</i>
6492 %stored2 = icmp eq i32 %result2, 5 <i>; yields {i1}:stored2 = false</i>
6494 %memval2 = load i32* %ptr <i>; yields {i32}:memval2 = 8</i>
6498 <!-- _______________________________________________________________________ -->
6499 <div class="doc_subsubsection">
6500 <a name="int_atomic_swap">'<tt>llvm.atomic.swap.*</tt>' Intrinsic</a>
6502 <div class="doc_text">
6506 This is an overloaded intrinsic. You can use <tt>llvm.atomic.swap</tt> on any
6507 integer bit width. Not all targets support all bit widths however.</p>
6509 declare i8 @llvm.atomic.swap.i8.p0i8( i8* <ptr>, i8 <val> )
6510 declare i16 @llvm.atomic.swap.i16.p0i16( i16* <ptr>, i16 <val> )
6511 declare i32 @llvm.atomic.swap.i32.p0i32( i32* <ptr>, i32 <val> )
6512 declare i64 @llvm.atomic.swap.i64.p0i64( i64* <ptr>, i64 <val> )
6517 This intrinsic loads the value stored in memory at <tt>ptr</tt> and yields
6518 the value from memory. It then stores the value in <tt>val</tt> in the memory
6524 The <tt>llvm.atomic.swap</tt> intrinsic takes two arguments. Both the
6525 <tt>val</tt> argument and the result must be integers of the same bit width.
6526 The first argument, <tt>ptr</tt>, must be a pointer to a value of this
6527 integer type. The targets may only lower integer representations they
6532 This intrinsic loads the value pointed to by <tt>ptr</tt>, yields it, and
6533 stores <tt>val</tt> back into <tt>ptr</tt> atomically. This provides the
6534 equivalent of an atomic swap operation within the SSA framework.
6542 %val1 = add i32 4, 4
6543 %result1 = call i32 @llvm.atomic.swap.i32.p0i32( i32* %ptr, i32 %val1 )
6544 <i>; yields {i32}:result1 = 4</i>
6545 %stored1 = icmp eq i32 %result1, 4 <i>; yields {i1}:stored1 = true</i>
6546 %memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 8</i>
6548 %val2 = add i32 1, 1
6549 %result2 = call i32 @llvm.atomic.swap.i32.p0i32( i32* %ptr, i32 %val2 )
6550 <i>; yields {i32}:result2 = 8</i>
6552 %stored2 = icmp eq i32 %result2, 8 <i>; yields {i1}:stored2 = true</i>
6553 %memval2 = load i32* %ptr <i>; yields {i32}:memval2 = 2</i>
6557 <!-- _______________________________________________________________________ -->
6558 <div class="doc_subsubsection">
6559 <a name="int_atomic_load_add">'<tt>llvm.atomic.load.add.*</tt>' Intrinsic</a>
6562 <div class="doc_text">
6565 This is an overloaded intrinsic. You can use <tt>llvm.atomic.load.add</tt> on any
6566 integer bit width. Not all targets support all bit widths however.</p>
6568 declare i8 @llvm.atomic.load.add.i8..p0i8( i8* <ptr>, i8 <delta> )
6569 declare i16 @llvm.atomic.load.add.i16..p0i16( i16* <ptr>, i16 <delta> )
6570 declare i32 @llvm.atomic.load.add.i32..p0i32( i32* <ptr>, i32 <delta> )
6571 declare i64 @llvm.atomic.load.add.i64..p0i64( i64* <ptr>, i64 <delta> )
6576 This intrinsic adds <tt>delta</tt> to the value stored in memory at
6577 <tt>ptr</tt>. It yields the original value at <tt>ptr</tt>.
6582 The intrinsic takes two arguments, the first a pointer to an integer value
6583 and the second an integer value. The result is also an integer value. These
6584 integer types can have any bit width, but they must all have the same bit
6585 width. The targets may only lower integer representations they support.
6589 This intrinsic does a series of operations atomically. It first loads the
6590 value stored at <tt>ptr</tt>. It then adds <tt>delta</tt>, stores the result
6591 to <tt>ptr</tt>. It yields the original value stored at <tt>ptr</tt>.
6598 %result1 = call i32 @llvm.atomic.load.add.i32.p0i32( i32* %ptr, i32 4 )
6599 <i>; yields {i32}:result1 = 4</i>
6600 %result2 = call i32 @llvm.atomic.load.add.i32.p0i32( i32* %ptr, i32 2 )
6601 <i>; yields {i32}:result2 = 8</i>
6602 %result3 = call i32 @llvm.atomic.load.add.i32.p0i32( i32* %ptr, i32 5 )
6603 <i>; yields {i32}:result3 = 10</i>
6604 %memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 15</i>
6608 <!-- _______________________________________________________________________ -->
6609 <div class="doc_subsubsection">
6610 <a name="int_atomic_load_sub">'<tt>llvm.atomic.load.sub.*</tt>' Intrinsic</a>
6613 <div class="doc_text">
6616 This is an overloaded intrinsic. You can use <tt>llvm.atomic.load.sub</tt> on
6617 any integer bit width and for different address spaces. Not all targets
6618 support all bit widths however.</p>
6620 declare i8 @llvm.atomic.load.sub.i8.p0i32( i8* <ptr>, i8 <delta> )
6621 declare i16 @llvm.atomic.load.sub.i16.p0i32( i16* <ptr>, i16 <delta> )
6622 declare i32 @llvm.atomic.load.sub.i32.p0i32( i32* <ptr>, i32 <delta> )
6623 declare i64 @llvm.atomic.load.sub.i64.p0i32( i64* <ptr>, i64 <delta> )
6628 This intrinsic subtracts <tt>delta</tt> to the value stored in memory at
6629 <tt>ptr</tt>. It yields the original value at <tt>ptr</tt>.
6634 The intrinsic takes two arguments, the first a pointer to an integer value
6635 and the second an integer value. The result is also an integer value. These
6636 integer types can have any bit width, but they must all have the same bit
6637 width. The targets may only lower integer representations they support.
6641 This intrinsic does a series of operations atomically. It first loads the
6642 value stored at <tt>ptr</tt>. It then subtracts <tt>delta</tt>, stores the
6643 result to <tt>ptr</tt>. It yields the original value stored at <tt>ptr</tt>.
6650 %result1 = call i32 @llvm.atomic.load.sub.i32.p0i32( i32* %ptr, i32 4 )
6651 <i>; yields {i32}:result1 = 8</i>
6652 %result2 = call i32 @llvm.atomic.load.sub.i32.p0i32( i32* %ptr, i32 2 )
6653 <i>; yields {i32}:result2 = 4</i>
6654 %result3 = call i32 @llvm.atomic.load.sub.i32.p0i32( i32* %ptr, i32 5 )
6655 <i>; yields {i32}:result3 = 2</i>
6656 %memval1 = load i32* %ptr <i>; yields {i32}:memval1 = -3</i>
6660 <!-- _______________________________________________________________________ -->
6661 <div class="doc_subsubsection">
6662 <a name="int_atomic_load_and">'<tt>llvm.atomic.load.and.*</tt>' Intrinsic</a><br>
6663 <a name="int_atomic_load_nand">'<tt>llvm.atomic.load.nand.*</tt>' Intrinsic</a><br>
6664 <a name="int_atomic_load_or">'<tt>llvm.atomic.load.or.*</tt>' Intrinsic</a><br>
6665 <a name="int_atomic_load_xor">'<tt>llvm.atomic.load.xor.*</tt>' Intrinsic</a><br>
6668 <div class="doc_text">
6671 These are overloaded intrinsics. You can use <tt>llvm.atomic.load_and</tt>,
6672 <tt>llvm.atomic.load_nand</tt>, <tt>llvm.atomic.load_or</tt>, and
6673 <tt>llvm.atomic.load_xor</tt> on any integer bit width and for different
6674 address spaces. Not all targets support all bit widths however.</p>
6676 declare i8 @llvm.atomic.load.and.i8.p0i8( i8* <ptr>, i8 <delta> )
6677 declare i16 @llvm.atomic.load.and.i16.p0i16( i16* <ptr>, i16 <delta> )
6678 declare i32 @llvm.atomic.load.and.i32.p0i32( i32* <ptr>, i32 <delta> )
6679 declare i64 @llvm.atomic.load.and.i64.p0i64( i64* <ptr>, i64 <delta> )
6684 declare i8 @llvm.atomic.load.or.i8.p0i8( i8* <ptr>, i8 <delta> )
6685 declare i16 @llvm.atomic.load.or.i16.p0i16( i16* <ptr>, i16 <delta> )
6686 declare i32 @llvm.atomic.load.or.i32.p0i32( i32* <ptr>, i32 <delta> )
6687 declare i64 @llvm.atomic.load.or.i64.p0i64( i64* <ptr>, i64 <delta> )
6692 declare i8 @llvm.atomic.load.nand.i8.p0i32( i8* <ptr>, i8 <delta> )
6693 declare i16 @llvm.atomic.load.nand.i16.p0i32( i16* <ptr>, i16 <delta> )
6694 declare i32 @llvm.atomic.load.nand.i32.p0i32( i32* <ptr>, i32 <delta> )
6695 declare i64 @llvm.atomic.load.nand.i64.p0i32( i64* <ptr>, i64 <delta> )
6700 declare i8 @llvm.atomic.load.xor.i8.p0i32( i8* <ptr>, i8 <delta> )
6701 declare i16 @llvm.atomic.load.xor.i16.p0i32( i16* <ptr>, i16 <delta> )
6702 declare i32 @llvm.atomic.load.xor.i32.p0i32( i32* <ptr>, i32 <delta> )
6703 declare i64 @llvm.atomic.load.xor.i64.p0i32( i64* <ptr>, i64 <delta> )
6708 These intrinsics bitwise the operation (and, nand, or, xor) <tt>delta</tt> to
6709 the value stored in memory at <tt>ptr</tt>. It yields the original value
6715 These intrinsics take two arguments, the first a pointer to an integer value
6716 and the second an integer value. The result is also an integer value. These
6717 integer types can have any bit width, but they must all have the same bit
6718 width. The targets may only lower integer representations they support.
6722 These intrinsics does a series of operations atomically. They first load the
6723 value stored at <tt>ptr</tt>. They then do the bitwise operation
6724 <tt>delta</tt>, store the result to <tt>ptr</tt>. They yield the original
6725 value stored at <tt>ptr</tt>.
6731 store i32 0x0F0F, %ptr
6732 %result0 = call i32 @llvm.atomic.load.nand.i32.p0i32( i32* %ptr, i32 0xFF )
6733 <i>; yields {i32}:result0 = 0x0F0F</i>
6734 %result1 = call i32 @llvm.atomic.load.and.i32.p0i32( i32* %ptr, i32 0xFF )
6735 <i>; yields {i32}:result1 = 0xFFFFFFF0</i>
6736 %result2 = call i32 @llvm.atomic.load.or.i32.p0i32( i32* %ptr, i32 0F )
6737 <i>; yields {i32}:result2 = 0xF0</i>
6738 %result3 = call i32 @llvm.atomic.load.xor.i32.p0i32( i32* %ptr, i32 0F )
6739 <i>; yields {i32}:result3 = FF</i>
6740 %memval1 = load i32* %ptr <i>; yields {i32}:memval1 = F0</i>
6745 <!-- _______________________________________________________________________ -->
6746 <div class="doc_subsubsection">
6747 <a name="int_atomic_load_max">'<tt>llvm.atomic.load.max.*</tt>' Intrinsic</a><br>
6748 <a name="int_atomic_load_min">'<tt>llvm.atomic.load.min.*</tt>' Intrinsic</a><br>
6749 <a name="int_atomic_load_umax">'<tt>llvm.atomic.load.umax.*</tt>' Intrinsic</a><br>
6750 <a name="int_atomic_load_umin">'<tt>llvm.atomic.load.umin.*</tt>' Intrinsic</a><br>
6753 <div class="doc_text">
6756 These are overloaded intrinsics. You can use <tt>llvm.atomic.load_max</tt>,
6757 <tt>llvm.atomic.load_min</tt>, <tt>llvm.atomic.load_umax</tt>, and
6758 <tt>llvm.atomic.load_umin</tt> on any integer bit width and for different
6759 address spaces. Not all targets
6760 support all bit widths however.</p>
6762 declare i8 @llvm.atomic.load.max.i8.p0i8( i8* <ptr>, i8 <delta> )
6763 declare i16 @llvm.atomic.load.max.i16.p0i16( i16* <ptr>, i16 <delta> )
6764 declare i32 @llvm.atomic.load.max.i32.p0i32( i32* <ptr>, i32 <delta> )
6765 declare i64 @llvm.atomic.load.max.i64.p0i64( i64* <ptr>, i64 <delta> )
6770 declare i8 @llvm.atomic.load.min.i8.p0i8( i8* <ptr>, i8 <delta> )
6771 declare i16 @llvm.atomic.load.min.i16.p0i16( i16* <ptr>, i16 <delta> )
6772 declare i32 @llvm.atomic.load.min.i32..p0i32( i32* <ptr>, i32 <delta> )
6773 declare i64 @llvm.atomic.load.min.i64..p0i64( i64* <ptr>, i64 <delta> )
6778 declare i8 @llvm.atomic.load.umax.i8.p0i8( i8* <ptr>, i8 <delta> )
6779 declare i16 @llvm.atomic.load.umax.i16.p0i16( i16* <ptr>, i16 <delta> )
6780 declare i32 @llvm.atomic.load.umax.i32.p0i32( i32* <ptr>, i32 <delta> )
6781 declare i64 @llvm.atomic.load.umax.i64.p0i64( i64* <ptr>, i64 <delta> )
6786 declare i8 @llvm.atomic.load.umin.i8..p0i8( i8* <ptr>, i8 <delta> )
6787 declare i16 @llvm.atomic.load.umin.i16.p0i16( i16* <ptr>, i16 <delta> )
6788 declare i32 @llvm.atomic.load.umin.i32..p0i32( i32* <ptr>, i32 <delta> )
6789 declare i64 @llvm.atomic.load.umin.i64..p0i64( i64* <ptr>, i64 <delta> )
6794 These intrinsics takes the signed or unsigned minimum or maximum of
6795 <tt>delta</tt> and the value stored in memory at <tt>ptr</tt>. It yields the
6796 original value at <tt>ptr</tt>.
6801 These intrinsics take two arguments, the first a pointer to an integer value
6802 and the second an integer value. The result is also an integer value. These
6803 integer types can have any bit width, but they must all have the same bit
6804 width. The targets may only lower integer representations they support.
6808 These intrinsics does a series of operations atomically. They first load the
6809 value stored at <tt>ptr</tt>. They then do the signed or unsigned min or max
6810 <tt>delta</tt> and the value, store the result to <tt>ptr</tt>. They yield
6811 the original value stored at <tt>ptr</tt>.
6818 %result0 = call i32 @llvm.atomic.load.min.i32.p0i32( i32* %ptr, i32 -2 )
6819 <i>; yields {i32}:result0 = 7</i>
6820 %result1 = call i32 @llvm.atomic.load.max.i32.p0i32( i32* %ptr, i32 8 )
6821 <i>; yields {i32}:result1 = -2</i>
6822 %result2 = call i32 @llvm.atomic.load.umin.i32.p0i32( i32* %ptr, i32 10 )
6823 <i>; yields {i32}:result2 = 8</i>
6824 %result3 = call i32 @llvm.atomic.load.umax.i32.p0i32( i32* %ptr, i32 30 )
6825 <i>; yields {i32}:result3 = 8</i>
6826 %memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 30</i>
6830 <!-- ======================================================================= -->
6831 <div class="doc_subsection">
6832 <a name="int_general">General Intrinsics</a>
6835 <div class="doc_text">
6836 <p> This class of intrinsics is designed to be generic and has
6837 no specific purpose. </p>
6840 <!-- _______________________________________________________________________ -->
6841 <div class="doc_subsubsection">
6842 <a name="int_var_annotation">'<tt>llvm.var.annotation</tt>' Intrinsic</a>
6845 <div class="doc_text">
6849 declare void @llvm.var.annotation(i8* <val>, i8* <str>, i8* <str>, i32 <int> )
6855 The '<tt>llvm.var.annotation</tt>' intrinsic
6861 The first argument is a pointer to a value, the second is a pointer to a
6862 global string, the third is a pointer to a global string which is the source
6863 file name, and the last argument is the line number.
6869 This intrinsic allows annotation of local variables with arbitrary strings.
6870 This can be useful for special purpose optimizations that want to look for these
6871 annotations. These have no other defined use, they are ignored by code
6872 generation and optimization.
6876 <!-- _______________________________________________________________________ -->
6877 <div class="doc_subsubsection">
6878 <a name="int_annotation">'<tt>llvm.annotation.*</tt>' Intrinsic</a>
6881 <div class="doc_text">
6884 <p>This is an overloaded intrinsic. You can use '<tt>llvm.annotation</tt>' on
6885 any integer bit width.
6888 declare i8 @llvm.annotation.i8(i8 <val>, i8* <str>, i8* <str>, i32 <int> )
6889 declare i16 @llvm.annotation.i16(i16 <val>, i8* <str>, i8* <str>, i32 <int> )
6890 declare i32 @llvm.annotation.i32(i32 <val>, i8* <str>, i8* <str>, i32 <int> )
6891 declare i64 @llvm.annotation.i64(i64 <val>, i8* <str>, i8* <str>, i32 <int> )
6892 declare i256 @llvm.annotation.i256(i256 <val>, i8* <str>, i8* <str>, i32 <int> )
6898 The '<tt>llvm.annotation</tt>' intrinsic.
6904 The first argument is an integer value (result of some expression),
6905 the second is a pointer to a global string, the third is a pointer to a global
6906 string which is the source file name, and the last argument is the line number.
6907 It returns the value of the first argument.
6913 This intrinsic allows annotations to be put on arbitrary expressions
6914 with arbitrary strings. This can be useful for special purpose optimizations
6915 that want to look for these annotations. These have no other defined use, they
6916 are ignored by code generation and optimization.
6920 <!-- _______________________________________________________________________ -->
6921 <div class="doc_subsubsection">
6922 <a name="int_trap">'<tt>llvm.trap</tt>' Intrinsic</a>
6925 <div class="doc_text">
6929 declare void @llvm.trap()
6935 The '<tt>llvm.trap</tt>' intrinsic
6947 This intrinsics is lowered to the target dependent trap instruction. If the
6948 target does not have a trap instruction, this intrinsic will be lowered to the
6949 call of the abort() function.
6953 <!-- _______________________________________________________________________ -->
6954 <div class="doc_subsubsection">
6955 <a name="int_stackprotector">'<tt>llvm.stackprotector</tt>' Intrinsic</a>
6957 <div class="doc_text">
6960 declare void @llvm.stackprotector( i8* <guard>, i8** <slot> )
6965 The <tt>llvm.stackprotector</tt> intrinsic takes the <tt>guard</tt> and stores
6966 it onto the stack at <tt>slot</tt>. The stack slot is adjusted to ensure that
6967 it is placed on the stack before local variables.
6971 The <tt>llvm.stackprotector</tt> intrinsic requires two pointer arguments. The
6972 first argument is the value loaded from the stack guard
6973 <tt>@__stack_chk_guard</tt>. The second variable is an <tt>alloca</tt> that
6974 has enough space to hold the value of the guard.
6978 This intrinsic causes the prologue/epilogue inserter to force the position of
6979 the <tt>AllocaInst</tt> stack slot to be before local variables on the
6980 stack. This is to ensure that if a local variable on the stack is overwritten,
6981 it will destroy the value of the guard. When the function exits, the guard on
6982 the stack is checked against the original guard. If they're different, then
6983 the program aborts by calling the <tt>__stack_chk_fail()</tt> function.
6987 <!-- *********************************************************************** -->
6990 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
6991 src="http://jigsaw.w3.org/css-validator/images/vcss-blue" alt="Valid CSS"></a>
6992 <a href="http://validator.w3.org/check/referer"><img
6993 src="http://www.w3.org/Icons/valid-html401-blue" alt="Valid HTML 4.01"></a>
6995 <a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
6996 <a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br>
6997 Last modified: $Date$