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>
60 <li><a href="#constants">Constants</a>
62 <li><a href="#simpleconstants">Simple Constants</a></li>
63 <li><a href="#aggregateconstants">Aggregate Constants</a></li>
64 <li><a href="#globalconstants">Global Variable and Function Addresses</a></li>
65 <li><a href="#undefvalues">Undefined Values</a></li>
66 <li><a href="#constantexprs">Constant Expressions</a></li>
69 <li><a href="#othervalues">Other Values</a>
71 <li><a href="#inlineasm">Inline Assembler Expressions</a></li>
74 <li><a href="#instref">Instruction Reference</a>
76 <li><a href="#terminators">Terminator Instructions</a>
78 <li><a href="#i_ret">'<tt>ret</tt>' Instruction</a></li>
79 <li><a href="#i_br">'<tt>br</tt>' Instruction</a></li>
80 <li><a href="#i_switch">'<tt>switch</tt>' Instruction</a></li>
81 <li><a href="#i_invoke">'<tt>invoke</tt>' Instruction</a></li>
82 <li><a href="#i_unwind">'<tt>unwind</tt>' Instruction</a></li>
83 <li><a href="#i_unreachable">'<tt>unreachable</tt>' Instruction</a></li>
86 <li><a href="#binaryops">Binary Operations</a>
88 <li><a href="#i_add">'<tt>add</tt>' Instruction</a></li>
89 <li><a href="#i_sub">'<tt>sub</tt>' Instruction</a></li>
90 <li><a href="#i_mul">'<tt>mul</tt>' Instruction</a></li>
91 <li><a href="#i_udiv">'<tt>udiv</tt>' Instruction</a></li>
92 <li><a href="#i_sdiv">'<tt>sdiv</tt>' Instruction</a></li>
93 <li><a href="#i_fdiv">'<tt>fdiv</tt>' Instruction</a></li>
94 <li><a href="#i_urem">'<tt>urem</tt>' Instruction</a></li>
95 <li><a href="#i_srem">'<tt>srem</tt>' Instruction</a></li>
96 <li><a href="#i_frem">'<tt>frem</tt>' Instruction</a></li>
99 <li><a href="#bitwiseops">Bitwise Binary Operations</a>
101 <li><a href="#i_shl">'<tt>shl</tt>' Instruction</a></li>
102 <li><a href="#i_lshr">'<tt>lshr</tt>' Instruction</a></li>
103 <li><a href="#i_ashr">'<tt>ashr</tt>' Instruction</a></li>
104 <li><a href="#i_and">'<tt>and</tt>' Instruction</a></li>
105 <li><a href="#i_or">'<tt>or</tt>' Instruction</a></li>
106 <li><a href="#i_xor">'<tt>xor</tt>' Instruction</a></li>
109 <li><a href="#vectorops">Vector Operations</a>
111 <li><a href="#i_extractelement">'<tt>extractelement</tt>' Instruction</a></li>
112 <li><a href="#i_insertelement">'<tt>insertelement</tt>' Instruction</a></li>
113 <li><a href="#i_shufflevector">'<tt>shufflevector</tt>' Instruction</a></li>
116 <li><a href="#aggregateops">Aggregate Operations</a>
118 <li><a href="#i_extractvalue">'<tt>extractvalue</tt>' Instruction</a></li>
119 <li><a href="#i_insertvalue">'<tt>insertvalue</tt>' Instruction</a></li>
122 <li><a href="#memoryops">Memory Access and Addressing Operations</a>
124 <li><a href="#i_malloc">'<tt>malloc</tt>' Instruction</a></li>
125 <li><a href="#i_free">'<tt>free</tt>' Instruction</a></li>
126 <li><a href="#i_alloca">'<tt>alloca</tt>' Instruction</a></li>
127 <li><a href="#i_load">'<tt>load</tt>' Instruction</a></li>
128 <li><a href="#i_store">'<tt>store</tt>' Instruction</a></li>
129 <li><a href="#i_getelementptr">'<tt>getelementptr</tt>' Instruction</a></li>
132 <li><a href="#convertops">Conversion Operations</a>
134 <li><a href="#i_trunc">'<tt>trunc .. to</tt>' Instruction</a></li>
135 <li><a href="#i_zext">'<tt>zext .. to</tt>' Instruction</a></li>
136 <li><a href="#i_sext">'<tt>sext .. to</tt>' Instruction</a></li>
137 <li><a href="#i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a></li>
138 <li><a href="#i_fpext">'<tt>fpext .. to</tt>' Instruction</a></li>
139 <li><a href="#i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a></li>
140 <li><a href="#i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a></li>
141 <li><a href="#i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a></li>
142 <li><a href="#i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a></li>
143 <li><a href="#i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a></li>
144 <li><a href="#i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a></li>
145 <li><a href="#i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a></li>
148 <li><a href="#otherops">Other Operations</a>
150 <li><a href="#i_icmp">'<tt>icmp</tt>' Instruction</a></li>
151 <li><a href="#i_fcmp">'<tt>fcmp</tt>' Instruction</a></li>
152 <li><a href="#i_vicmp">'<tt>vicmp</tt>' Instruction</a></li>
153 <li><a href="#i_vfcmp">'<tt>vfcmp</tt>' Instruction</a></li>
154 <li><a href="#i_phi">'<tt>phi</tt>' Instruction</a></li>
155 <li><a href="#i_select">'<tt>select</tt>' Instruction</a></li>
156 <li><a href="#i_call">'<tt>call</tt>' Instruction</a></li>
157 <li><a href="#i_va_arg">'<tt>va_arg</tt>' Instruction</a></li>
162 <li><a href="#intrinsics">Intrinsic Functions</a>
164 <li><a href="#int_varargs">Variable Argument Handling Intrinsics</a>
166 <li><a href="#int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a></li>
167 <li><a href="#int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a></li>
168 <li><a href="#int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a></li>
171 <li><a href="#int_gc">Accurate Garbage Collection Intrinsics</a>
173 <li><a href="#int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a></li>
174 <li><a href="#int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a></li>
175 <li><a href="#int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a></li>
178 <li><a href="#int_codegen">Code Generator Intrinsics</a>
180 <li><a href="#int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a></li>
181 <li><a href="#int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a></li>
182 <li><a href="#int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a></li>
183 <li><a href="#int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a></li>
184 <li><a href="#int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a></li>
185 <li><a href="#int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a></li>
186 <li><a href="#int_readcyclecounter"><tt>llvm.readcyclecounter</tt>' Intrinsic</a></li>
189 <li><a href="#int_libc">Standard C Library Intrinsics</a>
191 <li><a href="#int_memcpy">'<tt>llvm.memcpy.*</tt>' Intrinsic</a></li>
192 <li><a href="#int_memmove">'<tt>llvm.memmove.*</tt>' Intrinsic</a></li>
193 <li><a href="#int_memset">'<tt>llvm.memset.*</tt>' Intrinsic</a></li>
194 <li><a href="#int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a></li>
195 <li><a href="#int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a></li>
196 <li><a href="#int_sin">'<tt>llvm.sin.*</tt>' Intrinsic</a></li>
197 <li><a href="#int_cos">'<tt>llvm.cos.*</tt>' Intrinsic</a></li>
198 <li><a href="#int_pow">'<tt>llvm.pow.*</tt>' Intrinsic</a></li>
201 <li><a href="#int_manip">Bit Manipulation Intrinsics</a>
203 <li><a href="#int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a></li>
204 <li><a href="#int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic </a></li>
205 <li><a href="#int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic </a></li>
206 <li><a href="#int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic </a></li>
207 <li><a href="#int_part_select">'<tt>llvm.part.select.*</tt>' Intrinsic </a></li>
208 <li><a href="#int_part_set">'<tt>llvm.part.set.*</tt>' Intrinsic </a></li>
211 <li><a href="#int_debugger">Debugger intrinsics</a></li>
212 <li><a href="#int_eh">Exception Handling intrinsics</a></li>
213 <li><a href="#int_trampoline">Trampoline Intrinsic</a>
215 <li><a href="#int_it">'<tt>llvm.init.trampoline</tt>' Intrinsic</a></li>
218 <li><a href="#int_atomics">Atomic intrinsics</a>
220 <li><a href="#int_memory_barrier"><tt>llvm.memory_barrier</tt></a></li>
221 <li><a href="#int_atomic_cmp_swap"><tt>llvm.atomic.cmp.swap</tt></a></li>
222 <li><a href="#int_atomic_swap"><tt>llvm.atomic.swap</tt></a></li>
223 <li><a href="#int_atomic_load_add"><tt>llvm.atomic.load.add</tt></a></li>
224 <li><a href="#int_atomic_load_sub"><tt>llvm.atomic.load.sub</tt></a></li>
225 <li><a href="#int_atomic_load_and"><tt>llvm.atomic.load.and</tt></a></li>
226 <li><a href="#int_atomic_load_nand"><tt>llvm.atomic.load.nand</tt></a></li>
227 <li><a href="#int_atomic_load_or"><tt>llvm.atomic.load.or</tt></a></li>
228 <li><a href="#int_atomic_load_xor"><tt>llvm.atomic.load.xor</tt></a></li>
229 <li><a href="#int_atomic_load_max"><tt>llvm.atomic.load.max</tt></a></li>
230 <li><a href="#int_atomic_load_min"><tt>llvm.atomic.load.min</tt></a></li>
231 <li><a href="#int_atomic_load_umax"><tt>llvm.atomic.load.umax</tt></a></li>
232 <li><a href="#int_atomic_load_umin"><tt>llvm.atomic.load.umin</tt></a></li>
235 <li><a href="#int_general">General intrinsics</a>
237 <li><a href="#int_var_annotation">
238 '<tt>llvm.var.annotation</tt>' Intrinsic</a></li>
239 <li><a href="#int_annotation">
240 '<tt>llvm.annotation.*</tt>' Intrinsic</a></li>
241 <li><a href="#int_trap">
242 '<tt>llvm.trap</tt>' Intrinsic</a></li>
243 <li><a href="#int_stackprotector">
244 '<tt>llvm.stackprotector</tt>' Intrinsic</a></li>
251 <div class="doc_author">
252 <p>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a>
253 and <a href="mailto:vadve@cs.uiuc.edu">Vikram Adve</a></p>
256 <!-- *********************************************************************** -->
257 <div class="doc_section"> <a name="abstract">Abstract </a></div>
258 <!-- *********************************************************************** -->
260 <div class="doc_text">
261 <p>This document is a reference manual for the LLVM assembly language.
262 LLVM is a Static Single Assignment (SSA) based representation that provides
263 type safety, low-level operations, flexibility, and the capability of
264 representing 'all' high-level languages cleanly. It is the common code
265 representation used throughout all phases of the LLVM compilation
269 <!-- *********************************************************************** -->
270 <div class="doc_section"> <a name="introduction">Introduction</a> </div>
271 <!-- *********************************************************************** -->
273 <div class="doc_text">
275 <p>The LLVM code representation is designed to be used in three
276 different forms: as an in-memory compiler IR, as an on-disk bitcode
277 representation (suitable for fast loading by a Just-In-Time compiler),
278 and as a human readable assembly language representation. This allows
279 LLVM to provide a powerful intermediate representation for efficient
280 compiler transformations and analysis, while providing a natural means
281 to debug and visualize the transformations. The three different forms
282 of LLVM are all equivalent. This document describes the human readable
283 representation and notation.</p>
285 <p>The LLVM representation aims to be light-weight and low-level
286 while being expressive, typed, and extensible at the same time. It
287 aims to be a "universal IR" of sorts, by being at a low enough level
288 that high-level ideas may be cleanly mapped to it (similar to how
289 microprocessors are "universal IR's", allowing many source languages to
290 be mapped to them). By providing type information, LLVM can be used as
291 the target of optimizations: for example, through pointer analysis, it
292 can be proven that a C automatic variable is never accessed outside of
293 the current function... allowing it to be promoted to a simple SSA
294 value instead of a memory location.</p>
298 <!-- _______________________________________________________________________ -->
299 <div class="doc_subsubsection"> <a name="wellformed">Well-Formedness</a> </div>
301 <div class="doc_text">
303 <p>It is important to note that this document describes 'well formed'
304 LLVM assembly language. There is a difference between what the parser
305 accepts and what is considered 'well formed'. For example, the
306 following instruction is syntactically okay, but not well formed:</p>
308 <div class="doc_code">
310 %x = <a href="#i_add">add</a> i32 1, %x
314 <p>...because the definition of <tt>%x</tt> does not dominate all of
315 its uses. The LLVM infrastructure provides a verification pass that may
316 be used to verify that an LLVM module is well formed. This pass is
317 automatically run by the parser after parsing input assembly and by
318 the optimizer before it outputs bitcode. The violations pointed out
319 by the verifier pass indicate bugs in transformation passes or input to
323 <!-- Describe the typesetting conventions here. -->
325 <!-- *********************************************************************** -->
326 <div class="doc_section"> <a name="identifiers">Identifiers</a> </div>
327 <!-- *********************************************************************** -->
329 <div class="doc_text">
331 <p>LLVM identifiers come in two basic types: global and local. Global
332 identifiers (functions, global variables) begin with the @ character. Local
333 identifiers (register names, types) begin with the % character. Additionally,
334 there are three different formats for identifiers, for different purposes:</p>
337 <li>Named values are represented as a string of characters with their prefix.
338 For example, %foo, @DivisionByZero, %a.really.long.identifier. The actual
339 regular expression used is '<tt>[%@][a-zA-Z$._][a-zA-Z$._0-9]*</tt>'.
340 Identifiers which require other characters in their names can be surrounded
341 with quotes. Special characters may be escaped using "\xx" where xx is the
342 ASCII code for the character in hexadecimal. In this way, any character can
343 be used in a name value, even quotes themselves.
345 <li>Unnamed values are represented as an unsigned numeric value with their
346 prefix. For example, %12, @2, %44.</li>
348 <li>Constants, which are described in a <a href="#constants">section about
349 constants</a>, below.</li>
352 <p>LLVM requires that values start with a prefix for two reasons: Compilers
353 don't need to worry about name clashes with reserved words, and the set of
354 reserved words may be expanded in the future without penalty. Additionally,
355 unnamed identifiers allow a compiler to quickly come up with a temporary
356 variable without having to avoid symbol table conflicts.</p>
358 <p>Reserved words in LLVM are very similar to reserved words in other
359 languages. There are keywords for different opcodes
360 ('<tt><a href="#i_add">add</a></tt>',
361 '<tt><a href="#i_bitcast">bitcast</a></tt>',
362 '<tt><a href="#i_ret">ret</a></tt>', etc...), for primitive type names ('<tt><a
363 href="#t_void">void</a></tt>', '<tt><a href="#t_primitive">i32</a></tt>', etc...),
364 and others. These reserved words cannot conflict with variable names, because
365 none of them start with a prefix character ('%' or '@').</p>
367 <p>Here is an example of LLVM code to multiply the integer variable
368 '<tt>%X</tt>' by 8:</p>
372 <div class="doc_code">
374 %result = <a href="#i_mul">mul</a> i32 %X, 8
378 <p>After strength reduction:</p>
380 <div class="doc_code">
382 %result = <a href="#i_shl">shl</a> i32 %X, i8 3
386 <p>And the hard way:</p>
388 <div class="doc_code">
390 <a href="#i_add">add</a> i32 %X, %X <i>; yields {i32}:%0</i>
391 <a href="#i_add">add</a> i32 %0, %0 <i>; yields {i32}:%1</i>
392 %result = <a href="#i_add">add</a> i32 %1, %1
396 <p>This last way of multiplying <tt>%X</tt> by 8 illustrates several
397 important lexical features of LLVM:</p>
401 <li>Comments are delimited with a '<tt>;</tt>' and go until the end of
404 <li>Unnamed temporaries are created when the result of a computation is not
405 assigned to a named value.</li>
407 <li>Unnamed temporaries are numbered sequentially</li>
411 <p>...and it also shows a convention that we follow in this document. When
412 demonstrating instructions, we will follow an instruction with a comment that
413 defines the type and name of value produced. Comments are shown in italic
418 <!-- *********************************************************************** -->
419 <div class="doc_section"> <a name="highlevel">High Level Structure</a> </div>
420 <!-- *********************************************************************** -->
422 <!-- ======================================================================= -->
423 <div class="doc_subsection"> <a name="modulestructure">Module Structure</a>
426 <div class="doc_text">
428 <p>LLVM programs are composed of "Module"s, each of which is a
429 translation unit of the input programs. Each module consists of
430 functions, global variables, and symbol table entries. Modules may be
431 combined together with the LLVM linker, which merges function (and
432 global variable) definitions, resolves forward declarations, and merges
433 symbol table entries. Here is an example of the "hello world" module:</p>
435 <div class="doc_code">
436 <pre><i>; Declare the string constant as a global constant...</i>
437 <a href="#identifiers">@.LC0</a> = <a href="#linkage_internal">internal</a> <a
438 href="#globalvars">constant</a> <a href="#t_array">[13 x i8]</a> c"hello world\0A\00" <i>; [13 x i8]*</i>
440 <i>; External declaration of the puts function</i>
441 <a href="#functionstructure">declare</a> i32 @puts(i8 *) <i>; i32(i8 *)* </i>
443 <i>; Definition of main function</i>
444 define i32 @main() { <i>; i32()* </i>
445 <i>; Convert [13 x i8]* to i8 *...</i>
447 href="#i_getelementptr">getelementptr</a> [13 x i8]* @.LC0, i64 0, i64 0 <i>; i8 *</i>
449 <i>; Call puts function to write out the string to stdout...</i>
451 href="#i_call">call</a> i32 @puts(i8 * %cast210) <i>; i32</i>
453 href="#i_ret">ret</a> i32 0<br>}<br>
457 <p>This example is made up of a <a href="#globalvars">global variable</a>
458 named "<tt>.LC0</tt>", an external declaration of the "<tt>puts</tt>"
459 function, and a <a href="#functionstructure">function definition</a>
460 for "<tt>main</tt>".</p>
462 <p>In general, a module is made up of a list of global values,
463 where both functions and global variables are global values. Global values are
464 represented by a pointer to a memory location (in this case, a pointer to an
465 array of char, and a pointer to a function), and have one of the following <a
466 href="#linkage">linkage types</a>.</p>
470 <!-- ======================================================================= -->
471 <div class="doc_subsection">
472 <a name="linkage">Linkage Types</a>
475 <div class="doc_text">
478 All Global Variables and Functions have one of the following types of linkage:
483 <dt><tt><b><a name="linkage_private">private</a></b></tt>: </dt>
485 <dd>Global values with private linkage are only directly accessible by
486 objects in the current module. In particular, linking code into a module with
487 an private global value may cause the private to be renamed as necessary to
488 avoid collisions. Because the symbol is private to the module, all
489 references can be updated. This doesn't show up in any symbol table in the
493 <dt><tt><b><a name="linkage_internal">internal</a></b></tt>: </dt>
495 <dd> Similar to private, but the value show as a local symbol (STB_LOCAL in
496 the case of ELF) in the object file. This corresponds to the notion of the
497 '<tt>static</tt>' keyword in C.
500 <dt><tt><b><a name="linkage_linkonce">linkonce</a></b></tt>: </dt>
502 <dd>Globals with "<tt>linkonce</tt>" linkage are merged with other globals of
503 the same name when linkage occurs. This is typically used to implement
504 inline functions, templates, or other code which must be generated in each
505 translation unit that uses it. Unreferenced <tt>linkonce</tt> globals are
506 allowed to be discarded.
509 <dt><tt><b><a name="linkage_common">common</a></b></tt>: </dt>
511 <dd>"<tt>common</tt>" linkage is exactly the same as <tt>linkonce</tt>
512 linkage, except that unreferenced <tt>common</tt> globals may not be
513 discarded. This is used for globals that may be emitted in multiple
514 translation units, but that are not guaranteed to be emitted into every
515 translation unit that uses them. One example of this is tentative
516 definitions in C, such as "<tt>int X;</tt>" at global scope.
519 <dt><tt><b><a name="linkage_weak">weak</a></b></tt>: </dt>
521 <dd>"<tt>weak</tt>" linkage is the same as <tt>common</tt> linkage, except
522 that some targets may choose to emit different assembly sequences for them
523 for target-dependent reasons. This is used for globals that are declared
524 "weak" in C source code.
527 <dt><tt><b><a name="linkage_appending">appending</a></b></tt>: </dt>
529 <dd>"<tt>appending</tt>" linkage may only be applied to global variables of
530 pointer to array type. When two global variables with appending linkage are
531 linked together, the two global arrays are appended together. This is the
532 LLVM, typesafe, equivalent of having the system linker append together
533 "sections" with identical names when .o files are linked.
536 <dt><tt><b><a name="linkage_externweak">extern_weak</a></b></tt>: </dt>
537 <dd>The semantics of this linkage follow the ELF object file model: the
538 symbol is weak until linked, if not linked, the symbol becomes null instead
539 of being an undefined reference.
542 <dt><tt><b><a name="linkage_external">externally visible</a></b></tt>:</dt>
544 <dd>If none of the above identifiers are used, the global is externally
545 visible, meaning that it participates in linkage and can be used to resolve
546 external symbol references.
551 The next two types of linkage are targeted for Microsoft Windows platform
552 only. They are designed to support importing (exporting) symbols from (to)
553 DLLs (Dynamic Link Libraries).
557 <dt><tt><b><a name="linkage_dllimport">dllimport</a></b></tt>: </dt>
559 <dd>"<tt>dllimport</tt>" linkage causes the compiler to reference a function
560 or variable via a global pointer to a pointer that is set up by the DLL
561 exporting the symbol. On Microsoft Windows targets, the pointer name is
562 formed by combining <code>__imp_</code> and the function or variable name.
565 <dt><tt><b><a name="linkage_dllexport">dllexport</a></b></tt>: </dt>
567 <dd>"<tt>dllexport</tt>" linkage causes the compiler to provide a global
568 pointer to a pointer in a DLL, so that it can be referenced with the
569 <tt>dllimport</tt> attribute. On Microsoft Windows targets, the pointer
570 name is formed by combining <code>__imp_</code> and the function or variable
576 <p>For example, since the "<tt>.LC0</tt>"
577 variable is defined to be internal, if another module defined a "<tt>.LC0</tt>"
578 variable and was linked with this one, one of the two would be renamed,
579 preventing a collision. Since "<tt>main</tt>" and "<tt>puts</tt>" are
580 external (i.e., lacking any linkage declarations), they are accessible
581 outside of the current module.</p>
582 <p>It is illegal for a function <i>declaration</i>
583 to have any linkage type other than "externally visible", <tt>dllimport</tt>,
584 or <tt>extern_weak</tt>.</p>
585 <p>Aliases can have only <tt>external</tt>, <tt>internal</tt> and <tt>weak</tt>
589 <!-- ======================================================================= -->
590 <div class="doc_subsection">
591 <a name="callingconv">Calling Conventions</a>
594 <div class="doc_text">
596 <p>LLVM <a href="#functionstructure">functions</a>, <a href="#i_call">calls</a>
597 and <a href="#i_invoke">invokes</a> can all have an optional calling convention
598 specified for the call. The calling convention of any pair of dynamic
599 caller/callee must match, or the behavior of the program is undefined. The
600 following calling conventions are supported by LLVM, and more may be added in
604 <dt><b>"<tt>ccc</tt>" - The C calling convention</b>:</dt>
606 <dd>This calling convention (the default if no other calling convention is
607 specified) matches the target C calling conventions. This calling convention
608 supports varargs function calls and tolerates some mismatch in the declared
609 prototype and implemented declaration of the function (as does normal C).
612 <dt><b>"<tt>fastcc</tt>" - The fast calling convention</b>:</dt>
614 <dd>This calling convention attempts to make calls as fast as possible
615 (e.g. by passing things in registers). This calling convention allows the
616 target to use whatever tricks it wants to produce fast code for the target,
617 without having to conform to an externally specified ABI (Application Binary
618 Interface). Implementations of this convention should allow arbitrary
619 <a href="CodeGenerator.html#tailcallopt">tail call optimization</a> to be
620 supported. This calling convention does not support varargs and requires the
621 prototype of all callees to exactly match the prototype of the function
625 <dt><b>"<tt>coldcc</tt>" - The cold calling convention</b>:</dt>
627 <dd>This calling convention attempts to make code in the caller as efficient
628 as possible under the assumption that the call is not commonly executed. As
629 such, these calls often preserve all registers so that the call does not break
630 any live ranges in the caller side. This calling convention does not support
631 varargs and requires the prototype of all callees to exactly match the
632 prototype of the function definition.
635 <dt><b>"<tt>cc <<em>n</em>></tt>" - Numbered convention</b>:</dt>
637 <dd>Any calling convention may be specified by number, allowing
638 target-specific calling conventions to be used. Target specific calling
639 conventions start at 64.
643 <p>More calling conventions can be added/defined on an as-needed basis, to
644 support pascal conventions or any other well-known target-independent
649 <!-- ======================================================================= -->
650 <div class="doc_subsection">
651 <a name="visibility">Visibility Styles</a>
654 <div class="doc_text">
657 All Global Variables and Functions have one of the following visibility styles:
661 <dt><b>"<tt>default</tt>" - Default style</b>:</dt>
663 <dd>On targets that use the ELF object file format, default visibility means
664 that the declaration is visible to other
665 modules and, in shared libraries, means that the declared entity may be
666 overridden. On Darwin, default visibility means that the declaration is
667 visible to other modules. Default visibility corresponds to "external
668 linkage" in the language.
671 <dt><b>"<tt>hidden</tt>" - Hidden style</b>:</dt>
673 <dd>Two declarations of an object with hidden visibility refer to the same
674 object if they are in the same shared object. Usually, hidden visibility
675 indicates that the symbol will not be placed into the dynamic symbol table,
676 so no other module (executable or shared library) can reference it
680 <dt><b>"<tt>protected</tt>" - Protected style</b>:</dt>
682 <dd>On ELF, protected visibility indicates that the symbol will be placed in
683 the dynamic symbol table, but that references within the defining module will
684 bind to the local symbol. That is, the symbol cannot be overridden by another
691 <!-- ======================================================================= -->
692 <div class="doc_subsection">
693 <a name="namedtypes">Named Types</a>
696 <div class="doc_text">
698 <p>LLVM IR allows you to specify name aliases for certain types. This can make
699 it easier to read the IR and make the IR more condensed (particularly when
700 recursive types are involved). An example of a name specification is:
703 <div class="doc_code">
705 %mytype = type { %mytype*, i32 }
709 <p>You may give a name to any <a href="#typesystem">type</a> except "<a
710 href="t_void">void</a>". Type name aliases may be used anywhere a type is
711 expected with the syntax "%mytype".</p>
713 <p>Note that type names are aliases for the structural type that they indicate,
714 and that you can therefore specify multiple names for the same type. This often
715 leads to confusing behavior when dumping out a .ll file. Since LLVM IR uses
716 structural typing, the name is not part of the type. When printing out LLVM IR,
717 the printer will pick <em>one name</em> to render all types of a particular
718 shape. This means that if you have code where two different source types end up
719 having the same LLVM type, that the dumper will sometimes print the "wrong" or
720 unexpected type. This is an important design point and isn't going to
726 <!-- ======================================================================= -->
727 <div class="doc_subsection">
728 <a name="globalvars">Global Variables</a>
731 <div class="doc_text">
733 <p>Global variables define regions of memory allocated at compilation time
734 instead of run-time. Global variables may optionally be initialized, may have
735 an explicit section to be placed in, and may have an optional explicit alignment
736 specified. A variable may be defined as "thread_local", which means that it
737 will not be shared by threads (each thread will have a separated copy of the
738 variable). A variable may be defined as a global "constant," which indicates
739 that the contents of the variable will <b>never</b> be modified (enabling better
740 optimization, allowing the global data to be placed in the read-only section of
741 an executable, etc). Note that variables that need runtime initialization
742 cannot be marked "constant" as there is a store to the variable.</p>
745 LLVM explicitly allows <em>declarations</em> of global variables to be marked
746 constant, even if the final definition of the global is not. This capability
747 can be used to enable slightly better optimization of the program, but requires
748 the language definition to guarantee that optimizations based on the
749 'constantness' are valid for the translation units that do not include the
753 <p>As SSA values, global variables define pointer values that are in
754 scope (i.e. they dominate) all basic blocks in the program. Global
755 variables always define a pointer to their "content" type because they
756 describe a region of memory, and all memory objects in LLVM are
757 accessed through pointers.</p>
759 <p>A global variable may be declared to reside in a target-specifc numbered
760 address space. For targets that support them, address spaces may affect how
761 optimizations are performed and/or what target instructions are used to access
762 the variable. The default address space is zero. The address space qualifier
763 must precede any other attributes.</p>
765 <p>LLVM allows an explicit section to be specified for globals. If the target
766 supports it, it will emit globals to the section specified.</p>
768 <p>An explicit alignment may be specified for a global. If not present, or if
769 the alignment is set to zero, the alignment of the global is set by the target
770 to whatever it feels convenient. If an explicit alignment is specified, the
771 global is forced to have at least that much alignment. All alignments must be
774 <p>For example, the following defines a global in a numbered address space with
775 an initializer, section, and alignment:</p>
777 <div class="doc_code">
779 @G = addrspace(5) constant float 1.0, section "foo", align 4
786 <!-- ======================================================================= -->
787 <div class="doc_subsection">
788 <a name="functionstructure">Functions</a>
791 <div class="doc_text">
793 <p>LLVM function definitions consist of the "<tt>define</tt>" keyord,
794 an optional <a href="#linkage">linkage type</a>, an optional
795 <a href="#visibility">visibility style</a>, an optional
796 <a href="#callingconv">calling convention</a>, a return type, an optional
797 <a href="#paramattrs">parameter attribute</a> for the return type, a function
798 name, a (possibly empty) argument list (each with optional
799 <a href="#paramattrs">parameter attributes</a>), optional
800 <a href="#fnattrs">function attributes</a>, an optional section,
801 an optional alignment, an optional <a href="#gc">garbage collector name</a>,
802 an opening curly brace, a list of basic blocks, and a closing curly brace.
804 LLVM function declarations consist of the "<tt>declare</tt>" keyword, an
805 optional <a href="#linkage">linkage type</a>, an optional
806 <a href="#visibility">visibility style</a>, an optional
807 <a href="#callingconv">calling convention</a>, a return type, an optional
808 <a href="#paramattrs">parameter attribute</a> for the return type, a function
809 name, a possibly empty list of arguments, an optional alignment, and an optional
810 <a href="#gc">garbage collector name</a>.</p>
812 <p>A function definition contains a list of basic blocks, forming the CFG
813 (Control Flow Graph) for
814 the function. Each basic block may optionally start with a label (giving the
815 basic block a symbol table entry), contains a list of instructions, and ends
816 with a <a href="#terminators">terminator</a> instruction (such as a branch or
817 function return).</p>
819 <p>The first basic block in a function is special in two ways: it is immediately
820 executed on entrance to the function, and it is not allowed to have predecessor
821 basic blocks (i.e. there can not be any branches to the entry block of a
822 function). Because the block can have no predecessors, it also cannot have any
823 <a href="#i_phi">PHI nodes</a>.</p>
825 <p>LLVM allows an explicit section to be specified for functions. If the target
826 supports it, it will emit functions to the section specified.</p>
828 <p>An explicit alignment may be specified for a function. If not present, or if
829 the alignment is set to zero, the alignment of the function is set by the target
830 to whatever it feels convenient. If an explicit alignment is specified, the
831 function is forced to have at least that much alignment. All alignments must be
836 <div class="doc_code">
838 define [<a href="#linkage">linkage</a>] [<a href="#visibility">visibility</a>]
839 [<a href="#callingconv">cconv</a>] [<a href="#paramattrs">ret attrs</a>]
840 <ResultType> @<FunctionName> ([argument list])
841 [<a href="#fnattrs">fn Attrs</a>] [section "name"] [align N]
842 [<a href="#gc">gc</a>] { ... }
849 <!-- ======================================================================= -->
850 <div class="doc_subsection">
851 <a name="aliasstructure">Aliases</a>
853 <div class="doc_text">
854 <p>Aliases act as "second name" for the aliasee value (which can be either
855 function, global variable, another alias or bitcast of global value). Aliases
856 may have an optional <a href="#linkage">linkage type</a>, and an
857 optional <a href="#visibility">visibility style</a>.</p>
861 <div class="doc_code">
863 @<Name> = alias [Linkage] [Visibility] <AliaseeTy> @<Aliasee>
871 <!-- ======================================================================= -->
872 <div class="doc_subsection"><a name="paramattrs">Parameter Attributes</a></div>
873 <div class="doc_text">
874 <p>The return type and each parameter of a function type may have a set of
875 <i>parameter attributes</i> associated with them. Parameter attributes are
876 used to communicate additional information about the result or parameters of
877 a function. Parameter attributes are considered to be part of the function,
878 not of the function type, so functions with different parameter attributes
879 can have the same function type.</p>
881 <p>Parameter attributes are simple keywords that follow the type specified. If
882 multiple parameter attributes are needed, they are space separated. For
885 <div class="doc_code">
887 declare i32 @printf(i8* noalias , ...)
888 declare i32 @atoi(i8 zeroext)
889 declare signext i8 @returns_signed_char()
893 <p>Note that any attributes for the function result (<tt>nounwind</tt>,
894 <tt>readonly</tt>) come immediately after the argument list.</p>
896 <p>Currently, only the following parameter attributes are defined:</p>
898 <dt><tt>zeroext</tt></dt>
899 <dd>This indicates to the code generator that the parameter or return value
900 should be zero-extended to a 32-bit value by the caller (for a parameter)
901 or the callee (for a return value).</dd>
903 <dt><tt>signext</tt></dt>
904 <dd>This indicates to the code generator that the parameter or return value
905 should be sign-extended to a 32-bit value by the caller (for a parameter)
906 or the callee (for a return value).</dd>
908 <dt><tt>inreg</tt></dt>
909 <dd>This indicates that this parameter or return value should be treated
910 in a special target-dependent fashion during while emitting code for a
911 function call or return (usually, by putting it in a register as opposed
912 to memory, though some targets use it to distinguish between two different
913 kinds of registers). Use of this attribute is target-specific.</dd>
915 <dt><tt><a name="byval">byval</a></tt></dt>
916 <dd>This indicates that the pointer parameter should really be passed by
917 value to the function. The attribute implies that a hidden copy of the
918 pointee is made between the caller and the callee, so the callee is unable
919 to modify the value in the callee. This attribute is only valid on LLVM
920 pointer arguments. It is generally used to pass structs and arrays by
921 value, but is also valid on pointers to scalars. The copy is considered to
922 belong to the caller not the callee (for example,
923 <tt><a href="#readonly">readonly</a></tt> functions should not write to
924 <tt>byval</tt> parameters). This is not a valid attribute for return
927 <dt><tt>sret</tt></dt>
928 <dd>This indicates that the pointer parameter specifies the address of a
929 structure that is the return value of the function in the source program.
930 This pointer must be guaranteed by the caller to be valid: loads and stores
931 to the structure may be assumed by the callee to not to trap. This may only
932 be applied to the first parameter. This is not a valid attribute for
935 <dt><tt>noalias</tt></dt>
936 <dd>This indicates that the pointer does not alias any global or any other
937 parameter. The caller is responsible for ensuring that this is the
938 case. On a function return value, <tt>noalias</tt> additionally indicates
939 that the pointer does not alias any other pointers visible to the
940 caller. For further details, please see the discussion of the NoAlias
942 <a href="http://llvm.org/docs/AliasAnalysis.html#MustMayNo">alias
945 <dt><tt>nocapture</tt></dt>
946 <dd>This indicates that the callee does not make any copies of the pointer
947 that outlive the callee itself. This is not a valid attribute for return
950 <dt><tt>nest</tt></dt>
951 <dd>This indicates that the pointer parameter can be excised using the
952 <a href="#int_trampoline">trampoline intrinsics</a>. This is not a valid
953 attribute for return values.</dd>
958 <!-- ======================================================================= -->
959 <div class="doc_subsection">
960 <a name="gc">Garbage Collector Names</a>
963 <div class="doc_text">
964 <p>Each function may specify a garbage collector name, which is simply a
967 <div class="doc_code"><pre
968 >define void @f() gc "name" { ...</pre></div>
970 <p>The compiler declares the supported values of <i>name</i>. Specifying a
971 collector which will cause the compiler to alter its output in order to support
972 the named garbage collection algorithm.</p>
975 <!-- ======================================================================= -->
976 <div class="doc_subsection">
977 <a name="fnattrs">Function Attributes</a>
980 <div class="doc_text">
982 <p>Function attributes are set to communicate additional information about
983 a function. Function attributes are considered to be part of the function,
984 not of the function type, so functions with different parameter attributes
985 can have the same function type.</p>
987 <p>Function attributes are simple keywords that follow the type specified. If
988 multiple attributes are needed, they are space separated. For
991 <div class="doc_code">
993 define void @f() noinline { ... }
994 define void @f() alwaysinline { ... }
995 define void @f() alwaysinline optsize { ... }
996 define void @f() optsize
1001 <dt><tt>alwaysinline</tt></dt>
1002 <dd>This attribute indicates that the inliner should attempt to inline this
1003 function into callers whenever possible, ignoring any active inlining size
1004 threshold for this caller.</dd>
1006 <dt><tt>noinline</tt></dt>
1007 <dd>This attribute indicates that the inliner should never inline this function
1008 in any situation. This attribute may not be used together with the
1009 <tt>alwaysinline</tt> attribute.</dd>
1011 <dt><tt>optsize</tt></dt>
1012 <dd>This attribute suggests that optimization passes and code generator passes
1013 make choices that keep the code size of this function low, and otherwise do
1014 optimizations specifically to reduce code size.</dd>
1016 <dt><tt>noreturn</tt></dt>
1017 <dd>This function attribute indicates that the function never returns normally.
1018 This produces undefined behavior at runtime if the function ever does
1019 dynamically return.</dd>
1021 <dt><tt>nounwind</tt></dt>
1022 <dd>This function attribute indicates that the function never returns with an
1023 unwind or exceptional control flow. If the function does unwind, its runtime
1024 behavior is undefined.</dd>
1026 <dt><tt>readnone</tt></dt>
1027 <dd>This attribute indicates that the function computes its result (or the
1028 exception it throws) based strictly on its arguments, without dereferencing any
1029 pointer arguments or otherwise accessing any mutable state (e.g. memory, control
1030 registers, etc) visible to caller functions. It does not write through any
1031 pointer arguments (including <tt><a href="#byval">byval</a></tt> arguments) and
1032 never changes any state visible to callers.</dd>
1034 <dt><tt><a name="readonly">readonly</a></tt></dt>
1035 <dd>This attribute indicates that the function does not write through any
1036 pointer arguments (including <tt><a href="#byval">byval</a></tt> arguments)
1037 or otherwise modify any state (e.g. memory, control registers, etc) visible to
1038 caller functions. It may dereference pointer arguments and read state that may
1039 be set in the caller. A readonly function always returns the same value (or
1040 throws the same exception) when called with the same set of arguments and global
1043 <dt><tt><a name="ssp">ssp</a></tt></dt>
1044 <dd>This attribute indicates that the function should emit a stack smashing
1045 protector. It is in the form of a "canary"—a random value placed on the
1046 stack before the local variables that's checked upon return from the function to
1047 see if it has been overwritten. A heuristic is used to determine if a function
1048 needs stack protectors or not.
1050 <p>If a function that has an <tt>ssp</tt> attribute is inlined into a function
1051 that doesn't have an <tt>ssp</tt> attribute, then the resulting function will
1052 have an <tt>ssp</tt> attribute.</p></dd>
1054 <dt><tt>sspreq</tt></dt>
1055 <dd>This attribute indicates that the function should <em>always</em> emit a
1056 stack smashing protector. This overrides the <tt><a href="#ssp">ssp</a></tt>
1059 <p>If a function that has an <tt>sspreq</tt> attribute is inlined into a
1060 function that doesn't have an <tt>sspreq</tt> attribute or which has
1061 an <tt>ssp</tt> attribute, then the resulting function will have
1062 an <tt>sspreq</tt> attribute.</p></dd>
1067 <!-- ======================================================================= -->
1068 <div class="doc_subsection">
1069 <a name="moduleasm">Module-Level Inline Assembly</a>
1072 <div class="doc_text">
1074 Modules may contain "module-level inline asm" blocks, which corresponds to the
1075 GCC "file scope inline asm" blocks. These blocks are internally concatenated by
1076 LLVM and treated as a single unit, but may be separated in the .ll file if
1077 desired. The syntax is very simple:
1080 <div class="doc_code">
1082 module asm "inline asm code goes here"
1083 module asm "more can go here"
1087 <p>The strings can contain any character by escaping non-printable characters.
1088 The escape sequence used is simply "\xx" where "xx" is the two digit hex code
1093 The inline asm code is simply printed to the machine code .s file when
1094 assembly code is generated.
1098 <!-- ======================================================================= -->
1099 <div class="doc_subsection">
1100 <a name="datalayout">Data Layout</a>
1103 <div class="doc_text">
1104 <p>A module may specify a target specific data layout string that specifies how
1105 data is to be laid out in memory. The syntax for the data layout is simply:</p>
1106 <pre> target datalayout = "<i>layout specification</i>"</pre>
1107 <p>The <i>layout specification</i> consists of a list of specifications
1108 separated by the minus sign character ('-'). Each specification starts with a
1109 letter and may include other information after the letter to define some
1110 aspect of the data layout. The specifications accepted are as follows: </p>
1113 <dd>Specifies that the target lays out data in big-endian form. That is, the
1114 bits with the most significance have the lowest address location.</dd>
1116 <dd>Specifies that the target lays out data in little-endian form. That is,
1117 the bits with the least significance have the lowest address location.</dd>
1118 <dt><tt>p:<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1119 <dd>This specifies the <i>size</i> of a pointer and its <i>abi</i> and
1120 <i>preferred</i> alignments. All sizes are in bits. Specifying the <i>pref</i>
1121 alignment is optional. If omitted, the preceding <tt>:</tt> should be omitted
1123 <dt><tt>i<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1124 <dd>This specifies the alignment for an integer type of a given bit
1125 <i>size</i>. The value of <i>size</i> must be in the range [1,2^23).</dd>
1126 <dt><tt>v<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1127 <dd>This specifies the alignment for a vector type of a given bit
1129 <dt><tt>f<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1130 <dd>This specifies the alignment for a floating point type of a given bit
1131 <i>size</i>. The value of <i>size</i> must be either 32 (float) or 64
1133 <dt><tt>a<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1134 <dd>This specifies the alignment for an aggregate type of a given bit
1137 <p>When constructing the data layout for a given target, LLVM starts with a
1138 default set of specifications which are then (possibly) overriden by the
1139 specifications in the <tt>datalayout</tt> keyword. The default specifications
1140 are given in this list:</p>
1142 <li><tt>E</tt> - big endian</li>
1143 <li><tt>p:32:64:64</tt> - 32-bit pointers with 64-bit alignment</li>
1144 <li><tt>i1:8:8</tt> - i1 is 8-bit (byte) aligned</li>
1145 <li><tt>i8:8:8</tt> - i8 is 8-bit (byte) aligned</li>
1146 <li><tt>i16:16:16</tt> - i16 is 16-bit aligned</li>
1147 <li><tt>i32:32:32</tt> - i32 is 32-bit aligned</li>
1148 <li><tt>i64:32:64</tt> - i64 has ABI alignment of 32-bits but preferred
1149 alignment of 64-bits</li>
1150 <li><tt>f32:32:32</tt> - float is 32-bit aligned</li>
1151 <li><tt>f64:64:64</tt> - double is 64-bit aligned</li>
1152 <li><tt>v64:64:64</tt> - 64-bit vector is 64-bit aligned</li>
1153 <li><tt>v128:128:128</tt> - 128-bit vector is 128-bit aligned</li>
1154 <li><tt>a0:0:1</tt> - aggregates are 8-bit aligned</li>
1156 <p>When LLVM is determining the alignment for a given type, it uses the
1157 following rules:</p>
1159 <li>If the type sought is an exact match for one of the specifications, that
1160 specification is used.</li>
1161 <li>If no match is found, and the type sought is an integer type, then the
1162 smallest integer type that is larger than the bitwidth of the sought type is
1163 used. If none of the specifications are larger than the bitwidth then the the
1164 largest integer type is used. For example, given the default specifications
1165 above, the i7 type will use the alignment of i8 (next largest) while both
1166 i65 and i256 will use the alignment of i64 (largest specified).</li>
1167 <li>If no match is found, and the type sought is a vector type, then the
1168 largest vector type that is smaller than the sought vector type will be used
1169 as a fall back. This happens because <128 x double> can be implemented
1170 in terms of 64 <2 x double>, for example.</li>
1174 <!-- *********************************************************************** -->
1175 <div class="doc_section"> <a name="typesystem">Type System</a> </div>
1176 <!-- *********************************************************************** -->
1178 <div class="doc_text">
1180 <p>The LLVM type system is one of the most important features of the
1181 intermediate representation. Being typed enables a number of
1182 optimizations to be performed on the intermediate representation directly,
1183 without having to do
1184 extra analyses on the side before the transformation. A strong type
1185 system makes it easier to read the generated code and enables novel
1186 analyses and transformations that are not feasible to perform on normal
1187 three address code representations.</p>
1191 <!-- ======================================================================= -->
1192 <div class="doc_subsection"> <a name="t_classifications">Type
1193 Classifications</a> </div>
1194 <div class="doc_text">
1195 <p>The types fall into a few useful
1196 classifications:</p>
1198 <table border="1" cellspacing="0" cellpadding="4">
1200 <tr><th>Classification</th><th>Types</th></tr>
1202 <td><a href="#t_integer">integer</a></td>
1203 <td><tt>i1, i2, i3, ... i8, ... i16, ... i32, ... i64, ... </tt></td>
1206 <td><a href="#t_floating">floating point</a></td>
1207 <td><tt>float, double, x86_fp80, fp128, ppc_fp128</tt></td>
1210 <td><a name="t_firstclass">first class</a></td>
1211 <td><a href="#t_integer">integer</a>,
1212 <a href="#t_floating">floating point</a>,
1213 <a href="#t_pointer">pointer</a>,
1214 <a href="#t_vector">vector</a>,
1215 <a href="#t_struct">structure</a>,
1216 <a href="#t_array">array</a>,
1217 <a href="#t_label">label</a>.
1221 <td><a href="#t_primitive">primitive</a></td>
1222 <td><a href="#t_label">label</a>,
1223 <a href="#t_void">void</a>,
1224 <a href="#t_floating">floating point</a>.</td>
1227 <td><a href="#t_derived">derived</a></td>
1228 <td><a href="#t_integer">integer</a>,
1229 <a href="#t_array">array</a>,
1230 <a href="#t_function">function</a>,
1231 <a href="#t_pointer">pointer</a>,
1232 <a href="#t_struct">structure</a>,
1233 <a href="#t_pstruct">packed structure</a>,
1234 <a href="#t_vector">vector</a>,
1235 <a href="#t_opaque">opaque</a>.
1241 <p>The <a href="#t_firstclass">first class</a> types are perhaps the
1242 most important. Values of these types are the only ones which can be
1243 produced by instructions, passed as arguments, or used as operands to
1247 <!-- ======================================================================= -->
1248 <div class="doc_subsection"> <a name="t_primitive">Primitive Types</a> </div>
1250 <div class="doc_text">
1251 <p>The primitive types are the fundamental building blocks of the LLVM
1256 <!-- _______________________________________________________________________ -->
1257 <div class="doc_subsubsection"> <a name="t_floating">Floating Point Types</a> </div>
1259 <div class="doc_text">
1262 <tr><th>Type</th><th>Description</th></tr>
1263 <tr><td><tt>float</tt></td><td>32-bit floating point value</td></tr>
1264 <tr><td><tt>double</tt></td><td>64-bit floating point value</td></tr>
1265 <tr><td><tt>fp128</tt></td><td>128-bit floating point value (112-bit mantissa)</td></tr>
1266 <tr><td><tt>x86_fp80</tt></td><td>80-bit floating point value (X87)</td></tr>
1267 <tr><td><tt>ppc_fp128</tt></td><td>128-bit floating point value (two 64-bits)</td></tr>
1272 <!-- _______________________________________________________________________ -->
1273 <div class="doc_subsubsection"> <a name="t_void">Void Type</a> </div>
1275 <div class="doc_text">
1277 <p>The void type does not represent any value and has no size.</p>
1286 <!-- _______________________________________________________________________ -->
1287 <div class="doc_subsubsection"> <a name="t_label">Label Type</a> </div>
1289 <div class="doc_text">
1291 <p>The label type represents code labels.</p>
1301 <!-- ======================================================================= -->
1302 <div class="doc_subsection"> <a name="t_derived">Derived Types</a> </div>
1304 <div class="doc_text">
1306 <p>The real power in LLVM comes from the derived types in the system.
1307 This is what allows a programmer to represent arrays, functions,
1308 pointers, and other useful types. Note that these derived types may be
1309 recursive: For example, it is possible to have a two dimensional array.</p>
1313 <!-- _______________________________________________________________________ -->
1314 <div class="doc_subsubsection"> <a name="t_integer">Integer Type</a> </div>
1316 <div class="doc_text">
1319 <p>The integer type is a very simple derived type that simply specifies an
1320 arbitrary bit width for the integer type desired. Any bit width from 1 bit to
1321 2^23-1 (about 8 million) can be specified.</p>
1329 <p>The number of bits the integer will occupy is specified by the <tt>N</tt>
1333 <table class="layout">
1336 <td><tt>i1</tt></td>
1337 <td>a single-bit integer.</td>
1339 <td><tt>i32</tt></td>
1340 <td>a 32-bit integer.</td>
1342 <td><tt>i1942652</tt></td>
1343 <td>a really big integer of over 1 million bits.</td>
1349 <!-- _______________________________________________________________________ -->
1350 <div class="doc_subsubsection"> <a name="t_array">Array Type</a> </div>
1352 <div class="doc_text">
1356 <p>The array type is a very simple derived type that arranges elements
1357 sequentially in memory. The array type requires a size (number of
1358 elements) and an underlying data type.</p>
1363 [<# elements> x <elementtype>]
1366 <p>The number of elements is a constant integer value; elementtype may
1367 be any type with a size.</p>
1370 <table class="layout">
1372 <td class="left"><tt>[40 x i32]</tt></td>
1373 <td class="left">Array of 40 32-bit integer values.</td>
1376 <td class="left"><tt>[41 x i32]</tt></td>
1377 <td class="left">Array of 41 32-bit integer values.</td>
1380 <td class="left"><tt>[4 x i8]</tt></td>
1381 <td class="left">Array of 4 8-bit integer values.</td>
1384 <p>Here are some examples of multidimensional arrays:</p>
1385 <table class="layout">
1387 <td class="left"><tt>[3 x [4 x i32]]</tt></td>
1388 <td class="left">3x4 array of 32-bit integer values.</td>
1391 <td class="left"><tt>[12 x [10 x float]]</tt></td>
1392 <td class="left">12x10 array of single precision floating point values.</td>
1395 <td class="left"><tt>[2 x [3 x [4 x i16]]]</tt></td>
1396 <td class="left">2x3x4 array of 16-bit integer values.</td>
1400 <p>Note that 'variable sized arrays' can be implemented in LLVM with a zero
1401 length array. Normally, accesses past the end of an array are undefined in
1402 LLVM (e.g. it is illegal to access the 5th element of a 3 element array).
1403 As a special case, however, zero length arrays are recognized to be variable
1404 length. This allows implementation of 'pascal style arrays' with the LLVM
1405 type "{ i32, [0 x float]}", for example.</p>
1409 <!-- _______________________________________________________________________ -->
1410 <div class="doc_subsubsection"> <a name="t_function">Function Type</a> </div>
1411 <div class="doc_text">
1415 <p>The function type can be thought of as a function signature. It
1416 consists of a return type and a list of formal parameter types. The
1417 return type of a function type is a scalar type, a void type, or a struct type.
1418 If the return type is a struct type then all struct elements must be of first
1419 class types, and the struct must have at least one element.</p>
1424 <returntype list> (<parameter list>)
1427 <p>...where '<tt><parameter list></tt>' is a comma-separated list of type
1428 specifiers. Optionally, the parameter list may include a type <tt>...</tt>,
1429 which indicates that the function takes a variable number of arguments.
1430 Variable argument functions can access their arguments with the <a
1431 href="#int_varargs">variable argument handling intrinsic</a> functions.
1432 '<tt><returntype list></tt>' is a comma-separated list of
1433 <a href="#t_firstclass">first class</a> type specifiers.</p>
1436 <table class="layout">
1438 <td class="left"><tt>i32 (i32)</tt></td>
1439 <td class="left">function taking an <tt>i32</tt>, returning an <tt>i32</tt>
1441 </tr><tr class="layout">
1442 <td class="left"><tt>float (i16 signext, i32 *) *
1444 <td class="left"><a href="#t_pointer">Pointer</a> to a function that takes
1445 an <tt>i16</tt> that should be sign extended and a
1446 <a href="#t_pointer">pointer</a> to <tt>i32</tt>, returning
1449 </tr><tr class="layout">
1450 <td class="left"><tt>i32 (i8*, ...)</tt></td>
1451 <td class="left">A vararg function that takes at least one
1452 <a href="#t_pointer">pointer</a> to <tt>i8 </tt> (char in C),
1453 which returns an integer. This is the signature for <tt>printf</tt> in
1456 </tr><tr class="layout">
1457 <td class="left"><tt>{i32, i32} (i32)</tt></td>
1458 <td class="left">A function taking an <tt>i32</tt>, returning two
1459 <tt>i32</tt> values as an aggregate of type <tt>{ i32, i32 }</tt>
1465 <!-- _______________________________________________________________________ -->
1466 <div class="doc_subsubsection"> <a name="t_struct">Structure Type</a> </div>
1467 <div class="doc_text">
1469 <p>The structure type is used to represent a collection of data members
1470 together in memory. The packing of the field types is defined to match
1471 the ABI of the underlying processor. The elements of a structure may
1472 be any type that has a size.</p>
1473 <p>Structures are accessed using '<tt><a href="#i_load">load</a></tt>
1474 and '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a
1475 field with the '<tt><a href="#i_getelementptr">getelementptr</a></tt>'
1478 <pre> { <type list> }<br></pre>
1480 <table class="layout">
1482 <td class="left"><tt>{ i32, i32, i32 }</tt></td>
1483 <td class="left">A triple of three <tt>i32</tt> values</td>
1484 </tr><tr class="layout">
1485 <td class="left"><tt>{ float, i32 (i32) * }</tt></td>
1486 <td class="left">A pair, where the first element is a <tt>float</tt> and the
1487 second element is a <a href="#t_pointer">pointer</a> to a
1488 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
1489 an <tt>i32</tt>.</td>
1494 <!-- _______________________________________________________________________ -->
1495 <div class="doc_subsubsection"> <a name="t_pstruct">Packed Structure Type</a>
1497 <div class="doc_text">
1499 <p>The packed structure type is used to represent a collection of data members
1500 together in memory. There is no padding between fields. Further, the alignment
1501 of a packed structure is 1 byte. The elements of a packed structure may
1502 be any type that has a size.</p>
1503 <p>Structures are accessed using '<tt><a href="#i_load">load</a></tt>
1504 and '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a
1505 field with the '<tt><a href="#i_getelementptr">getelementptr</a></tt>'
1508 <pre> < { <type list> } > <br></pre>
1510 <table class="layout">
1512 <td class="left"><tt>< { i32, i32, i32 } ></tt></td>
1513 <td class="left">A triple of three <tt>i32</tt> values</td>
1514 </tr><tr class="layout">
1516 <tt>< { float, i32 (i32)* } ></tt></td>
1517 <td class="left">A pair, where the first element is a <tt>float</tt> and the
1518 second element is a <a href="#t_pointer">pointer</a> to a
1519 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
1520 an <tt>i32</tt>.</td>
1525 <!-- _______________________________________________________________________ -->
1526 <div class="doc_subsubsection"> <a name="t_pointer">Pointer Type</a> </div>
1527 <div class="doc_text">
1529 <p>As in many languages, the pointer type represents a pointer or
1530 reference to another object, which must live in memory. Pointer types may have
1531 an optional address space attribute defining the target-specific numbered
1532 address space where the pointed-to object resides. The default address space is
1535 <pre> <type> *<br></pre>
1537 <table class="layout">
1539 <td class="left"><tt>[4 x i32]*</tt></td>
1540 <td class="left">A <a href="#t_pointer">pointer</a> to <a
1541 href="#t_array">array</a> of four <tt>i32</tt> values.</td>
1544 <td class="left"><tt>i32 (i32 *) *</tt></td>
1545 <td class="left"> A <a href="#t_pointer">pointer</a> to a <a
1546 href="#t_function">function</a> that takes an <tt>i32*</tt>, returning an
1550 <td class="left"><tt>i32 addrspace(5)*</tt></td>
1551 <td class="left">A <a href="#t_pointer">pointer</a> to an <tt>i32</tt> value
1552 that resides in address space #5.</td>
1557 <!-- _______________________________________________________________________ -->
1558 <div class="doc_subsubsection"> <a name="t_vector">Vector Type</a> </div>
1559 <div class="doc_text">
1563 <p>A vector type is a simple derived type that represents a vector
1564 of elements. Vector types are used when multiple primitive data
1565 are operated in parallel using a single instruction (SIMD).
1566 A vector type requires a size (number of
1567 elements) and an underlying primitive data type. Vectors must have a power
1568 of two length (1, 2, 4, 8, 16 ...). Vector types are
1569 considered <a href="#t_firstclass">first class</a>.</p>
1574 < <# elements> x <elementtype> >
1577 <p>The number of elements is a constant integer value; elementtype may
1578 be any integer or floating point type.</p>
1582 <table class="layout">
1584 <td class="left"><tt><4 x i32></tt></td>
1585 <td class="left">Vector of 4 32-bit integer values.</td>
1588 <td class="left"><tt><8 x float></tt></td>
1589 <td class="left">Vector of 8 32-bit floating-point values.</td>
1592 <td class="left"><tt><2 x i64></tt></td>
1593 <td class="left">Vector of 2 64-bit integer values.</td>
1598 <!-- _______________________________________________________________________ -->
1599 <div class="doc_subsubsection"> <a name="t_opaque">Opaque Type</a> </div>
1600 <div class="doc_text">
1604 <p>Opaque types are used to represent unknown types in the system. This
1605 corresponds (for example) to the C notion of a forward declared structure type.
1606 In LLVM, opaque types can eventually be resolved to any type (not just a
1607 structure type).</p>
1617 <table class="layout">
1619 <td class="left"><tt>opaque</tt></td>
1620 <td class="left">An opaque type.</td>
1626 <!-- *********************************************************************** -->
1627 <div class="doc_section"> <a name="constants">Constants</a> </div>
1628 <!-- *********************************************************************** -->
1630 <div class="doc_text">
1632 <p>LLVM has several different basic types of constants. This section describes
1633 them all and their syntax.</p>
1637 <!-- ======================================================================= -->
1638 <div class="doc_subsection"><a name="simpleconstants">Simple Constants</a></div>
1640 <div class="doc_text">
1643 <dt><b>Boolean constants</b></dt>
1645 <dd>The two strings '<tt>true</tt>' and '<tt>false</tt>' are both valid
1646 constants of the <tt><a href="#t_primitive">i1</a></tt> type.
1649 <dt><b>Integer constants</b></dt>
1651 <dd>Standard integers (such as '4') are constants of the <a
1652 href="#t_integer">integer</a> type. Negative numbers may be used with
1656 <dt><b>Floating point constants</b></dt>
1658 <dd>Floating point constants use standard decimal notation (e.g. 123.421),
1659 exponential notation (e.g. 1.23421e+2), or a more precise hexadecimal
1660 notation (see below). The assembler requires the exact decimal value of
1661 a floating-point constant. For example, the assembler accepts 1.25 but
1662 rejects 1.3 because 1.3 is a repeating decimal in binary. Floating point
1663 constants must have a <a href="#t_floating">floating point</a> type. </dd>
1665 <dt><b>Null pointer constants</b></dt>
1667 <dd>The identifier '<tt>null</tt>' is recognized as a null pointer constant
1668 and must be of <a href="#t_pointer">pointer type</a>.</dd>
1672 <p>The one non-intuitive notation for constants is the optional hexadecimal form
1673 of floating point constants. For example, the form '<tt>double
1674 0x432ff973cafa8000</tt>' is equivalent to (but harder to read than) '<tt>double
1675 4.5e+15</tt>'. The only time hexadecimal floating point constants are required
1676 (and the only time that they are generated by the disassembler) is when a
1677 floating point constant must be emitted but it cannot be represented as a
1678 decimal floating point number. For example, NaN's, infinities, and other
1679 special values are represented in their IEEE hexadecimal format so that
1680 assembly and disassembly do not cause any bits to change in the constants.</p>
1684 <!-- ======================================================================= -->
1685 <div class="doc_subsection"><a name="aggregateconstants">Aggregate Constants</a>
1688 <div class="doc_text">
1689 <p>Aggregate constants arise from aggregation of simple constants
1690 and smaller aggregate constants.</p>
1693 <dt><b>Structure constants</b></dt>
1695 <dd>Structure constants are represented with notation similar to structure
1696 type definitions (a comma separated list of elements, surrounded by braces
1697 (<tt>{}</tt>)). For example: "<tt>{ i32 4, float 17.0, i32* @G }</tt>",
1698 where "<tt>@G</tt>" is declared as "<tt>@G = external global i32</tt>". Structure constants
1699 must have <a href="#t_struct">structure type</a>, and the number and
1700 types of elements must match those specified by the type.
1703 <dt><b>Array constants</b></dt>
1705 <dd>Array constants are represented with notation similar to array type
1706 definitions (a comma separated list of elements, surrounded by square brackets
1707 (<tt>[]</tt>)). For example: "<tt>[ i32 42, i32 11, i32 74 ]</tt>". Array
1708 constants must have <a href="#t_array">array type</a>, and the number and
1709 types of elements must match those specified by the type.
1712 <dt><b>Vector constants</b></dt>
1714 <dd>Vector constants are represented with notation similar to vector type
1715 definitions (a comma separated list of elements, surrounded by
1716 less-than/greater-than's (<tt><></tt>)). For example: "<tt>< i32 42,
1717 i32 11, i32 74, i32 100 ></tt>". Vector constants must have <a
1718 href="#t_vector">vector type</a>, and the number and types of elements must
1719 match those specified by the type.
1722 <dt><b>Zero initialization</b></dt>
1724 <dd>The string '<tt>zeroinitializer</tt>' can be used to zero initialize a
1725 value to zero of <em>any</em> type, including scalar and aggregate types.
1726 This is often used to avoid having to print large zero initializers (e.g. for
1727 large arrays) and is always exactly equivalent to using explicit zero
1734 <!-- ======================================================================= -->
1735 <div class="doc_subsection">
1736 <a name="globalconstants">Global Variable and Function Addresses</a>
1739 <div class="doc_text">
1741 <p>The addresses of <a href="#globalvars">global variables</a> and <a
1742 href="#functionstructure">functions</a> are always implicitly valid (link-time)
1743 constants. These constants are explicitly referenced when the <a
1744 href="#identifiers">identifier for the global</a> is used and always have <a
1745 href="#t_pointer">pointer</a> type. For example, the following is a legal LLVM
1748 <div class="doc_code">
1752 @Z = global [2 x i32*] [ i32* @X, i32* @Y ]
1758 <!-- ======================================================================= -->
1759 <div class="doc_subsection"><a name="undefvalues">Undefined Values</a></div>
1760 <div class="doc_text">
1761 <p>The string '<tt>undef</tt>' is recognized as a type-less constant that has
1762 no specific value. Undefined values may be of any type and be used anywhere
1763 a constant is permitted.</p>
1765 <p>Undefined values indicate to the compiler that the program is well defined
1766 no matter what value is used, giving the compiler more freedom to optimize.
1770 <!-- ======================================================================= -->
1771 <div class="doc_subsection"><a name="constantexprs">Constant Expressions</a>
1774 <div class="doc_text">
1776 <p>Constant expressions are used to allow expressions involving other constants
1777 to be used as constants. Constant expressions may be of any <a
1778 href="#t_firstclass">first class</a> type and may involve any LLVM operation
1779 that does not have side effects (e.g. load and call are not supported). The
1780 following is the syntax for constant expressions:</p>
1783 <dt><b><tt>trunc ( CST to TYPE )</tt></b></dt>
1784 <dd>Truncate a constant to another type. The bit size of CST must be larger
1785 than the bit size of TYPE. Both types must be integers.</dd>
1787 <dt><b><tt>zext ( CST to TYPE )</tt></b></dt>
1788 <dd>Zero extend a constant to another type. The bit size of CST must be
1789 smaller or equal to the bit size of TYPE. Both types must be integers.</dd>
1791 <dt><b><tt>sext ( CST to TYPE )</tt></b></dt>
1792 <dd>Sign extend a constant to another type. The bit size of CST must be
1793 smaller or equal to the bit size of TYPE. Both types must be integers.</dd>
1795 <dt><b><tt>fptrunc ( CST to TYPE )</tt></b></dt>
1796 <dd>Truncate a floating point constant to another floating point type. The
1797 size of CST must be larger than the size of TYPE. Both types must be
1798 floating point.</dd>
1800 <dt><b><tt>fpext ( CST to TYPE )</tt></b></dt>
1801 <dd>Floating point extend a constant to another type. The size of CST must be
1802 smaller or equal to the size of TYPE. Both types must be floating point.</dd>
1804 <dt><b><tt>fptoui ( CST to TYPE )</tt></b></dt>
1805 <dd>Convert a floating point constant to the corresponding unsigned integer
1806 constant. TYPE must be a scalar or vector integer type. CST must be of scalar
1807 or vector floating point type. Both CST and TYPE must be scalars, or vectors
1808 of the same number of elements. If the value won't fit in the integer type,
1809 the results are undefined.</dd>
1811 <dt><b><tt>fptosi ( CST to TYPE )</tt></b></dt>
1812 <dd>Convert a floating point constant to the corresponding signed integer
1813 constant. TYPE must be a scalar or vector integer type. CST must be of scalar
1814 or vector floating point type. Both CST and TYPE must be scalars, or vectors
1815 of the same number of elements. If the value won't fit in the integer type,
1816 the results are undefined.</dd>
1818 <dt><b><tt>uitofp ( CST to TYPE )</tt></b></dt>
1819 <dd>Convert an unsigned integer constant to the corresponding floating point
1820 constant. TYPE must be a scalar or vector floating point type. CST must be of
1821 scalar or vector integer type. Both CST and TYPE must be scalars, or vectors
1822 of the same number of elements. If the value won't fit in the floating point
1823 type, the results are undefined.</dd>
1825 <dt><b><tt>sitofp ( CST to TYPE )</tt></b></dt>
1826 <dd>Convert a signed integer constant to the corresponding floating point
1827 constant. TYPE must be a scalar or vector floating point type. CST must be of
1828 scalar or vector integer type. Both CST and TYPE must be scalars, or vectors
1829 of the same number of elements. If the value won't fit in the floating point
1830 type, the results are undefined.</dd>
1832 <dt><b><tt>ptrtoint ( CST to TYPE )</tt></b></dt>
1833 <dd>Convert a pointer typed constant to the corresponding integer constant
1834 TYPE must be an integer type. CST must be of pointer type. The CST value is
1835 zero extended, truncated, or unchanged to make it fit in TYPE.</dd>
1837 <dt><b><tt>inttoptr ( CST to TYPE )</tt></b></dt>
1838 <dd>Convert a integer constant to a pointer constant. TYPE must be a
1839 pointer type. CST must be of integer type. The CST value is zero extended,
1840 truncated, or unchanged to make it fit in a pointer size. This one is
1841 <i>really</i> dangerous!</dd>
1843 <dt><b><tt>bitcast ( CST to TYPE )</tt></b></dt>
1844 <dd>Convert a constant, CST, to another TYPE. The size of CST and TYPE must be
1845 identical (same number of bits). The conversion is done as if the CST value
1846 was stored to memory and read back as TYPE. In other words, no bits change
1847 with this operator, just the type. This can be used for conversion of
1848 vector types to any other type, as long as they have the same bit width. For
1849 pointers it is only valid to cast to another pointer type. It is not valid
1850 to bitcast to or from an aggregate type.
1853 <dt><b><tt>getelementptr ( CSTPTR, IDX0, IDX1, ... )</tt></b></dt>
1855 <dd>Perform the <a href="#i_getelementptr">getelementptr operation</a> on
1856 constants. As with the <a href="#i_getelementptr">getelementptr</a>
1857 instruction, the index list may have zero or more indexes, which are required
1858 to make sense for the type of "CSTPTR".</dd>
1860 <dt><b><tt>select ( COND, VAL1, VAL2 )</tt></b></dt>
1862 <dd>Perform the <a href="#i_select">select operation</a> on
1865 <dt><b><tt>icmp COND ( VAL1, VAL2 )</tt></b></dt>
1866 <dd>Performs the <a href="#i_icmp">icmp operation</a> on constants.</dd>
1868 <dt><b><tt>fcmp COND ( VAL1, VAL2 )</tt></b></dt>
1869 <dd>Performs the <a href="#i_fcmp">fcmp operation</a> on constants.</dd>
1871 <dt><b><tt>vicmp COND ( VAL1, VAL2 )</tt></b></dt>
1872 <dd>Performs the <a href="#i_vicmp">vicmp operation</a> on constants.</dd>
1874 <dt><b><tt>vfcmp COND ( VAL1, VAL2 )</tt></b></dt>
1875 <dd>Performs the <a href="#i_vfcmp">vfcmp operation</a> on constants.</dd>
1877 <dt><b><tt>extractelement ( VAL, IDX )</tt></b></dt>
1879 <dd>Perform the <a href="#i_extractelement">extractelement
1880 operation</a> on constants.</dd>
1882 <dt><b><tt>insertelement ( VAL, ELT, IDX )</tt></b></dt>
1884 <dd>Perform the <a href="#i_insertelement">insertelement
1885 operation</a> on constants.</dd>
1888 <dt><b><tt>shufflevector ( VEC1, VEC2, IDXMASK )</tt></b></dt>
1890 <dd>Perform the <a href="#i_shufflevector">shufflevector
1891 operation</a> on constants.</dd>
1893 <dt><b><tt>OPCODE ( LHS, RHS )</tt></b></dt>
1895 <dd>Perform the specified operation of the LHS and RHS constants. OPCODE may
1896 be any of the <a href="#binaryops">binary</a> or <a href="#bitwiseops">bitwise
1897 binary</a> operations. The constraints on operands are the same as those for
1898 the corresponding instruction (e.g. no bitwise operations on floating point
1899 values are allowed).</dd>
1903 <!-- *********************************************************************** -->
1904 <div class="doc_section"> <a name="othervalues">Other Values</a> </div>
1905 <!-- *********************************************************************** -->
1907 <!-- ======================================================================= -->
1908 <div class="doc_subsection">
1909 <a name="inlineasm">Inline Assembler Expressions</a>
1912 <div class="doc_text">
1915 LLVM supports inline assembler expressions (as opposed to <a href="#moduleasm">
1916 Module-Level Inline Assembly</a>) through the use of a special value. This
1917 value represents the inline assembler as a string (containing the instructions
1918 to emit), a list of operand constraints (stored as a string), and a flag that
1919 indicates whether or not the inline asm expression has side effects. An example
1920 inline assembler expression is:
1923 <div class="doc_code">
1925 i32 (i32) asm "bswap $0", "=r,r"
1930 Inline assembler expressions may <b>only</b> be used as the callee operand of
1931 a <a href="#i_call"><tt>call</tt> instruction</a>. Thus, typically we have:
1934 <div class="doc_code">
1936 %X = call i32 asm "<a href="#int_bswap">bswap</a> $0", "=r,r"(i32 %Y)
1941 Inline asms with side effects not visible in the constraint list must be marked
1942 as having side effects. This is done through the use of the
1943 '<tt>sideeffect</tt>' keyword, like so:
1946 <div class="doc_code">
1948 call void asm sideeffect "eieio", ""()
1952 <p>TODO: The format of the asm and constraints string still need to be
1953 documented here. Constraints on what can be done (e.g. duplication, moving, etc
1954 need to be documented). This is probably best done by reference to another
1955 document that covers inline asm from a holistic perspective.
1960 <!-- *********************************************************************** -->
1961 <div class="doc_section"> <a name="instref">Instruction Reference</a> </div>
1962 <!-- *********************************************************************** -->
1964 <div class="doc_text">
1966 <p>The LLVM instruction set consists of several different
1967 classifications of instructions: <a href="#terminators">terminator
1968 instructions</a>, <a href="#binaryops">binary instructions</a>,
1969 <a href="#bitwiseops">bitwise binary instructions</a>, <a
1970 href="#memoryops">memory instructions</a>, and <a href="#otherops">other
1971 instructions</a>.</p>
1975 <!-- ======================================================================= -->
1976 <div class="doc_subsection"> <a name="terminators">Terminator
1977 Instructions</a> </div>
1979 <div class="doc_text">
1981 <p>As mentioned <a href="#functionstructure">previously</a>, every
1982 basic block in a program ends with a "Terminator" instruction, which
1983 indicates which block should be executed after the current block is
1984 finished. These terminator instructions typically yield a '<tt>void</tt>'
1985 value: they produce control flow, not values (the one exception being
1986 the '<a href="#i_invoke"><tt>invoke</tt></a>' instruction).</p>
1987 <p>There are six different terminator instructions: the '<a
1988 href="#i_ret"><tt>ret</tt></a>' instruction, the '<a href="#i_br"><tt>br</tt></a>'
1989 instruction, the '<a href="#i_switch"><tt>switch</tt></a>' instruction,
1990 the '<a href="#i_invoke"><tt>invoke</tt></a>' instruction, the '<a
1991 href="#i_unwind"><tt>unwind</tt></a>' instruction, and the '<a
1992 href="#i_unreachable"><tt>unreachable</tt></a>' instruction.</p>
1996 <!-- _______________________________________________________________________ -->
1997 <div class="doc_subsubsection"> <a name="i_ret">'<tt>ret</tt>'
1998 Instruction</a> </div>
1999 <div class="doc_text">
2002 ret <type> <value> <i>; Return a value from a non-void function</i>
2003 ret void <i>; Return from void function</i>
2008 <p>The '<tt>ret</tt>' instruction is used to return control flow (and
2009 optionally a value) from a function back to the caller.</p>
2010 <p>There are two forms of the '<tt>ret</tt>' instruction: one that
2011 returns a value and then causes control flow, and one that just causes
2012 control flow to occur.</p>
2016 <p>The '<tt>ret</tt>' instruction optionally accepts a single argument,
2017 the return value. The type of the return value must be a
2018 '<a href="#t_firstclass">first class</a>' type.</p>
2020 <p>A function is not <a href="#wellformed">well formed</a> if
2021 it it has a non-void return type and contains a '<tt>ret</tt>'
2022 instruction with no return value or a return value with a type that
2023 does not match its type, or if it has a void return type and contains
2024 a '<tt>ret</tt>' instruction with a return value.</p>
2028 <p>When the '<tt>ret</tt>' instruction is executed, control flow
2029 returns back to the calling function's context. If the caller is a "<a
2030 href="#i_call"><tt>call</tt></a>" instruction, execution continues at
2031 the instruction after the call. If the caller was an "<a
2032 href="#i_invoke"><tt>invoke</tt></a>" instruction, execution continues
2033 at the beginning of the "normal" destination block. If the instruction
2034 returns a value, that value shall set the call or invoke instruction's
2040 ret i32 5 <i>; Return an integer value of 5</i>
2041 ret void <i>; Return from a void function</i>
2042 ret { i32, i8 } { i32 4, i8 2 } <i>; Return an aggregate of values 4 and 2</i>
2045 <p>Note that the code generator does not yet fully support larger
2046 aggregate return values.</p>
2049 <!-- _______________________________________________________________________ -->
2050 <div class="doc_subsubsection"> <a name="i_br">'<tt>br</tt>' Instruction</a> </div>
2051 <div class="doc_text">
2053 <pre> br i1 <cond>, label <iftrue>, label <iffalse><br> br label <dest> <i>; Unconditional branch</i>
2056 <p>The '<tt>br</tt>' instruction is used to cause control flow to
2057 transfer to a different basic block in the current function. There are
2058 two forms of this instruction, corresponding to a conditional branch
2059 and an unconditional branch.</p>
2061 <p>The conditional branch form of the '<tt>br</tt>' instruction takes a
2062 single '<tt>i1</tt>' value and two '<tt>label</tt>' values. The
2063 unconditional form of the '<tt>br</tt>' instruction takes a single
2064 '<tt>label</tt>' value as a target.</p>
2066 <p>Upon execution of a conditional '<tt>br</tt>' instruction, the '<tt>i1</tt>'
2067 argument is evaluated. If the value is <tt>true</tt>, control flows
2068 to the '<tt>iftrue</tt>' <tt>label</tt> argument. If "cond" is <tt>false</tt>,
2069 control flows to the '<tt>iffalse</tt>' <tt>label</tt> argument.</p>
2071 <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
2072 href="#i_ret">ret</a> i32 1<br>IfUnequal:<br> <a href="#i_ret">ret</a> i32 0<br></pre>
2074 <!-- _______________________________________________________________________ -->
2075 <div class="doc_subsubsection">
2076 <a name="i_switch">'<tt>switch</tt>' Instruction</a>
2079 <div class="doc_text">
2083 switch <intty> <value>, label <defaultdest> [ <intty> <val>, label <dest> ... ]
2088 <p>The '<tt>switch</tt>' instruction is used to transfer control flow to one of
2089 several different places. It is a generalization of the '<tt>br</tt>'
2090 instruction, allowing a branch to occur to one of many possible
2096 <p>The '<tt>switch</tt>' instruction uses three parameters: an integer
2097 comparison value '<tt>value</tt>', a default '<tt>label</tt>' destination, and
2098 an array of pairs of comparison value constants and '<tt>label</tt>'s. The
2099 table is not allowed to contain duplicate constant entries.</p>
2103 <p>The <tt>switch</tt> instruction specifies a table of values and
2104 destinations. When the '<tt>switch</tt>' instruction is executed, this
2105 table is searched for the given value. If the value is found, control flow is
2106 transfered to the corresponding destination; otherwise, control flow is
2107 transfered to the default destination.</p>
2109 <h5>Implementation:</h5>
2111 <p>Depending on properties of the target machine and the particular
2112 <tt>switch</tt> instruction, this instruction may be code generated in different
2113 ways. For example, it could be generated as a series of chained conditional
2114 branches or with a lookup table.</p>
2119 <i>; Emulate a conditional br instruction</i>
2120 %Val = <a href="#i_zext">zext</a> i1 %value to i32
2121 switch i32 %Val, label %truedest [ i32 0, label %falsedest ]
2123 <i>; Emulate an unconditional br instruction</i>
2124 switch i32 0, label %dest [ ]
2126 <i>; Implement a jump table:</i>
2127 switch i32 %val, label %otherwise [ i32 0, label %onzero
2129 i32 2, label %ontwo ]
2133 <!-- _______________________________________________________________________ -->
2134 <div class="doc_subsubsection">
2135 <a name="i_invoke">'<tt>invoke</tt>' Instruction</a>
2138 <div class="doc_text">
2143 <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>]
2144 to label <normal label> unwind label <exception label>
2149 <p>The '<tt>invoke</tt>' instruction causes control to transfer to a specified
2150 function, with the possibility of control flow transfer to either the
2151 '<tt>normal</tt>' label or the
2152 '<tt>exception</tt>' label. If the callee function returns with the
2153 "<tt><a href="#i_ret">ret</a></tt>" instruction, control flow will return to the
2154 "normal" label. If the callee (or any indirect callees) returns with the "<a
2155 href="#i_unwind"><tt>unwind</tt></a>" instruction, control is interrupted and
2156 continued at the dynamically nearest "exception" label.</p>
2160 <p>This instruction requires several arguments:</p>
2164 The optional "cconv" marker indicates which <a href="#callingconv">calling
2165 convention</a> the call should use. If none is specified, the call defaults
2166 to using C calling conventions.
2169 <li>The optional <a href="#paramattrs">Parameter Attributes</a> list for
2170 return values. Only '<tt>zeroext</tt>', '<tt>signext</tt>',
2171 and '<tt>inreg</tt>' attributes are valid here.</li>
2173 <li>'<tt>ptr to function ty</tt>': shall be the signature of the pointer to
2174 function value being invoked. In most cases, this is a direct function
2175 invocation, but indirect <tt>invoke</tt>s are just as possible, branching off
2176 an arbitrary pointer to function value.
2179 <li>'<tt>function ptr val</tt>': An LLVM value containing a pointer to a
2180 function to be invoked. </li>
2182 <li>'<tt>function args</tt>': argument list whose types match the function
2183 signature argument types. If the function signature indicates the function
2184 accepts a variable number of arguments, the extra arguments can be
2187 <li>'<tt>normal label</tt>': the label reached when the called function
2188 executes a '<tt><a href="#i_ret">ret</a></tt>' instruction. </li>
2190 <li>'<tt>exception label</tt>': the label reached when a callee returns with
2191 the <a href="#i_unwind"><tt>unwind</tt></a> instruction. </li>
2193 <li>The optional <a href="#fnattrs">function attributes</a> list. Only
2194 '<tt>noreturn</tt>', '<tt>nounwind</tt>', '<tt>readonly</tt>' and
2195 '<tt>readnone</tt>' attributes are valid here.</li>
2200 <p>This instruction is designed to operate as a standard '<tt><a
2201 href="#i_call">call</a></tt>' instruction in most regards. The primary
2202 difference is that it establishes an association with a label, which is used by
2203 the runtime library to unwind the stack.</p>
2205 <p>This instruction is used in languages with destructors to ensure that proper
2206 cleanup is performed in the case of either a <tt>longjmp</tt> or a thrown
2207 exception. Additionally, this is important for implementation of
2208 '<tt>catch</tt>' clauses in high-level languages that support them.</p>
2212 %retval = invoke i32 @Test(i32 15) to label %Continue
2213 unwind label %TestCleanup <i>; {i32}:retval set</i>
2214 %retval = invoke <a href="#callingconv">coldcc</a> i32 %Testfnptr(i32 15) to label %Continue
2215 unwind label %TestCleanup <i>; {i32}:retval set</i>
2220 <!-- _______________________________________________________________________ -->
2222 <div class="doc_subsubsection"> <a name="i_unwind">'<tt>unwind</tt>'
2223 Instruction</a> </div>
2225 <div class="doc_text">
2234 <p>The '<tt>unwind</tt>' instruction unwinds the stack, continuing control flow
2235 at the first callee in the dynamic call stack which used an <a
2236 href="#i_invoke"><tt>invoke</tt></a> instruction to perform the call. This is
2237 primarily used to implement exception handling.</p>
2241 <p>The '<tt>unwind</tt>' instruction causes execution of the current function to
2242 immediately halt. The dynamic call stack is then searched for the first <a
2243 href="#i_invoke"><tt>invoke</tt></a> instruction on the call stack. Once found,
2244 execution continues at the "exceptional" destination block specified by the
2245 <tt>invoke</tt> instruction. If there is no <tt>invoke</tt> instruction in the
2246 dynamic call chain, undefined behavior results.</p>
2249 <!-- _______________________________________________________________________ -->
2251 <div class="doc_subsubsection"> <a name="i_unreachable">'<tt>unreachable</tt>'
2252 Instruction</a> </div>
2254 <div class="doc_text">
2263 <p>The '<tt>unreachable</tt>' instruction has no defined semantics. This
2264 instruction is used to inform the optimizer that a particular portion of the
2265 code is not reachable. This can be used to indicate that the code after a
2266 no-return function cannot be reached, and other facts.</p>
2270 <p>The '<tt>unreachable</tt>' instruction has no defined semantics.</p>
2275 <!-- ======================================================================= -->
2276 <div class="doc_subsection"> <a name="binaryops">Binary Operations</a> </div>
2277 <div class="doc_text">
2278 <p>Binary operators are used to do most of the computation in a
2279 program. They require two operands of the same type, execute an operation on them, and
2280 produce a single value. The operands might represent
2281 multiple data, as is the case with the <a href="#t_vector">vector</a> data type.
2282 The result value has the same type as its operands.</p>
2283 <p>There are several different binary operators:</p>
2285 <!-- _______________________________________________________________________ -->
2286 <div class="doc_subsubsection">
2287 <a name="i_add">'<tt>add</tt>' Instruction</a>
2290 <div class="doc_text">
2295 <result> = add <ty> <op1>, <op2> <i>; yields {ty}:result</i>
2300 <p>The '<tt>add</tt>' instruction returns the sum of its two operands.</p>
2304 <p>The two arguments to the '<tt>add</tt>' instruction must be <a
2305 href="#t_integer">integer</a>, <a href="#t_floating">floating point</a>, or
2306 <a href="#t_vector">vector</a> values. Both arguments must have identical
2311 <p>The value produced is the integer or floating point sum of the two
2314 <p>If an integer sum has unsigned overflow, the result returned is the
2315 mathematical result modulo 2<sup>n</sup>, where n is the bit width of
2318 <p>Because LLVM integers use a two's complement representation, this
2319 instruction is appropriate for both signed and unsigned integers.</p>
2324 <result> = add i32 4, %var <i>; yields {i32}:result = 4 + %var</i>
2327 <!-- _______________________________________________________________________ -->
2328 <div class="doc_subsubsection">
2329 <a name="i_sub">'<tt>sub</tt>' Instruction</a>
2332 <div class="doc_text">
2337 <result> = sub <ty> <op1>, <op2> <i>; yields {ty}:result</i>
2342 <p>The '<tt>sub</tt>' instruction returns the difference of its two
2345 <p>Note that the '<tt>sub</tt>' instruction is used to represent the
2346 '<tt>neg</tt>' instruction present in most other intermediate
2347 representations.</p>
2351 <p>The two arguments to the '<tt>sub</tt>' instruction must be <a
2352 href="#t_integer">integer</a>, <a href="#t_floating">floating point</a>,
2353 or <a href="#t_vector">vector</a> values. Both arguments must have identical
2358 <p>The value produced is the integer or floating point difference of
2359 the two operands.</p>
2361 <p>If an integer difference has unsigned overflow, the result returned is the
2362 mathematical result modulo 2<sup>n</sup>, where n is the bit width of
2365 <p>Because LLVM integers use a two's complement representation, this
2366 instruction is appropriate for both signed and unsigned integers.</p>
2370 <result> = sub i32 4, %var <i>; yields {i32}:result = 4 - %var</i>
2371 <result> = sub i32 0, %val <i>; yields {i32}:result = -%var</i>
2375 <!-- _______________________________________________________________________ -->
2376 <div class="doc_subsubsection">
2377 <a name="i_mul">'<tt>mul</tt>' Instruction</a>
2380 <div class="doc_text">
2383 <pre> <result> = mul <ty> <op1>, <op2> <i>; yields {ty}:result</i>
2386 <p>The '<tt>mul</tt>' instruction returns the product of its two
2391 <p>The two arguments to the '<tt>mul</tt>' instruction must be <a
2392 href="#t_integer">integer</a>, <a href="#t_floating">floating point</a>,
2393 or <a href="#t_vector">vector</a> values. Both arguments must have identical
2398 <p>The value produced is the integer or floating point product of the
2401 <p>If the result of an integer multiplication has unsigned overflow,
2402 the result returned is the mathematical result modulo
2403 2<sup>n</sup>, where n is the bit width of the result.</p>
2404 <p>Because LLVM integers use a two's complement representation, and the
2405 result is the same width as the operands, this instruction returns the
2406 correct result for both signed and unsigned integers. If a full product
2407 (e.g. <tt>i32</tt>x<tt>i32</tt>-><tt>i64</tt>) is needed, the operands
2408 should be sign-extended or zero-extended as appropriate to the
2409 width of the full product.</p>
2411 <pre> <result> = mul i32 4, %var <i>; yields {i32}:result = 4 * %var</i>
2415 <!-- _______________________________________________________________________ -->
2416 <div class="doc_subsubsection"> <a name="i_udiv">'<tt>udiv</tt>' Instruction
2418 <div class="doc_text">
2420 <pre> <result> = udiv <ty> <op1>, <op2> <i>; yields {ty}:result</i>
2423 <p>The '<tt>udiv</tt>' instruction returns the quotient of its two
2428 <p>The two arguments to the '<tt>udiv</tt>' instruction must be
2429 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2430 values. Both arguments must have identical types.</p>
2434 <p>The value produced is the unsigned integer quotient of the two operands.</p>
2435 <p>Note that unsigned integer division and signed integer division are distinct
2436 operations; for signed integer division, use '<tt>sdiv</tt>'.</p>
2437 <p>Division by zero leads to undefined behavior.</p>
2439 <pre> <result> = udiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
2442 <!-- _______________________________________________________________________ -->
2443 <div class="doc_subsubsection"> <a name="i_sdiv">'<tt>sdiv</tt>' Instruction
2445 <div class="doc_text">
2448 <result> = sdiv <ty> <op1>, <op2> <i>; yields {ty}:result</i>
2453 <p>The '<tt>sdiv</tt>' instruction returns the quotient of its two
2458 <p>The two arguments to the '<tt>sdiv</tt>' instruction must be
2459 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2460 values. Both arguments must have identical types.</p>
2463 <p>The value produced is the signed integer quotient of the two operands rounded towards zero.</p>
2464 <p>Note that signed integer division and unsigned integer division are distinct
2465 operations; for unsigned integer division, use '<tt>udiv</tt>'.</p>
2466 <p>Division by zero leads to undefined behavior. Overflow also leads to
2467 undefined behavior; this is a rare case, but can occur, for example,
2468 by doing a 32-bit division of -2147483648 by -1.</p>
2470 <pre> <result> = sdiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
2473 <!-- _______________________________________________________________________ -->
2474 <div class="doc_subsubsection"> <a name="i_fdiv">'<tt>fdiv</tt>'
2475 Instruction</a> </div>
2476 <div class="doc_text">
2479 <result> = fdiv <ty> <op1>, <op2> <i>; yields {ty}:result</i>
2483 <p>The '<tt>fdiv</tt>' instruction returns the quotient of its two
2488 <p>The two arguments to the '<tt>fdiv</tt>' instruction must be
2489 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a>
2490 of floating point values. Both arguments must have identical types.</p>
2494 <p>The value produced is the floating point quotient of the two operands.</p>
2499 <result> = fdiv float 4.0, %var <i>; yields {float}:result = 4.0 / %var</i>
2503 <!-- _______________________________________________________________________ -->
2504 <div class="doc_subsubsection"> <a name="i_urem">'<tt>urem</tt>' Instruction</a>
2506 <div class="doc_text">
2508 <pre> <result> = urem <ty> <op1>, <op2> <i>; yields {ty}:result</i>
2511 <p>The '<tt>urem</tt>' instruction returns the remainder from the
2512 unsigned division of its two arguments.</p>
2514 <p>The two arguments to the '<tt>urem</tt>' instruction must be
2515 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2516 values. Both arguments must have identical types.</p>
2518 <p>This instruction returns the unsigned integer <i>remainder</i> of a division.
2519 This instruction always performs an unsigned division to get the remainder.</p>
2520 <p>Note that unsigned integer remainder and signed integer remainder are
2521 distinct operations; for signed integer remainder, use '<tt>srem</tt>'.</p>
2522 <p>Taking the remainder of a division by zero leads to undefined behavior.</p>
2524 <pre> <result> = urem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
2528 <!-- _______________________________________________________________________ -->
2529 <div class="doc_subsubsection">
2530 <a name="i_srem">'<tt>srem</tt>' Instruction</a>
2533 <div class="doc_text">
2538 <result> = srem <ty> <op1>, <op2> <i>; yields {ty}:result</i>
2543 <p>The '<tt>srem</tt>' instruction returns the remainder from the
2544 signed division of its two operands. This instruction can also take
2545 <a href="#t_vector">vector</a> versions of the values in which case
2546 the elements must be integers.</p>
2550 <p>The two arguments to the '<tt>srem</tt>' instruction must be
2551 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2552 values. Both arguments must have identical types.</p>
2556 <p>This instruction returns the <i>remainder</i> of a division (where the result
2557 has the same sign as the dividend, <tt>op1</tt>), not the <i>modulo</i>
2558 operator (where the result has the same sign as the divisor, <tt>op2</tt>) of
2559 a value. For more information about the difference, see <a
2560 href="http://mathforum.org/dr.math/problems/anne.4.28.99.html">The
2561 Math Forum</a>. For a table of how this is implemented in various languages,
2562 please see <a href="http://en.wikipedia.org/wiki/Modulo_operation">
2563 Wikipedia: modulo operation</a>.</p>
2564 <p>Note that signed integer remainder and unsigned integer remainder are
2565 distinct operations; for unsigned integer remainder, use '<tt>urem</tt>'.</p>
2566 <p>Taking the remainder of a division by zero leads to undefined behavior.
2567 Overflow also leads to undefined behavior; this is a rare case, but can occur,
2568 for example, by taking the remainder of a 32-bit division of -2147483648 by -1.
2569 (The remainder doesn't actually overflow, but this rule lets srem be
2570 implemented using instructions that return both the result of the division
2571 and the remainder.)</p>
2573 <pre> <result> = srem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
2577 <!-- _______________________________________________________________________ -->
2578 <div class="doc_subsubsection">
2579 <a name="i_frem">'<tt>frem</tt>' Instruction</a> </div>
2581 <div class="doc_text">
2584 <pre> <result> = frem <ty> <op1>, <op2> <i>; yields {ty}:result</i>
2587 <p>The '<tt>frem</tt>' instruction returns the remainder from the
2588 division of its two operands.</p>
2590 <p>The two arguments to the '<tt>frem</tt>' instruction must be
2591 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a>
2592 of floating point values. Both arguments must have identical types.</p>
2596 <p>This instruction returns the <i>remainder</i> of a division.
2597 The remainder has the same sign as the dividend.</p>
2602 <result> = frem float 4.0, %var <i>; yields {float}:result = 4.0 % %var</i>
2606 <!-- ======================================================================= -->
2607 <div class="doc_subsection"> <a name="bitwiseops">Bitwise Binary
2608 Operations</a> </div>
2609 <div class="doc_text">
2610 <p>Bitwise binary operators are used to do various forms of
2611 bit-twiddling in a program. They are generally very efficient
2612 instructions and can commonly be strength reduced from other
2613 instructions. They require two operands of the same type, execute an operation on them,
2614 and produce a single value. The resulting value is the same type as its operands.</p>
2617 <!-- _______________________________________________________________________ -->
2618 <div class="doc_subsubsection"> <a name="i_shl">'<tt>shl</tt>'
2619 Instruction</a> </div>
2620 <div class="doc_text">
2622 <pre> <result> = shl <ty> <op1>, <op2> <i>; yields {ty}:result</i>
2627 <p>The '<tt>shl</tt>' instruction returns the first operand shifted to
2628 the left a specified number of bits.</p>
2632 <p>Both arguments to the '<tt>shl</tt>' instruction must be the same <a
2633 href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2634 type. '<tt>op2</tt>' is treated as an unsigned value.</p>
2638 <p>The value produced is <tt>op1</tt> * 2<sup><tt>op2</tt></sup> mod 2<sup>n</sup>,
2639 where n is the width of the result. If <tt>op2</tt> is (statically or dynamically) negative or
2640 equal to or larger than the number of bits in <tt>op1</tt>, the result is undefined.
2641 If the arguments are vectors, each vector element of <tt>op1</tt> is shifted by the
2642 corresponding shift amount in <tt>op2</tt>.</p>
2644 <h5>Example:</h5><pre>
2645 <result> = shl i32 4, %var <i>; yields {i32}: 4 << %var</i>
2646 <result> = shl i32 4, 2 <i>; yields {i32}: 16</i>
2647 <result> = shl i32 1, 10 <i>; yields {i32}: 1024</i>
2648 <result> = shl i32 1, 32 <i>; undefined</i>
2649 <result> = shl <2 x i32> < i32 1, i32 1>, < i32 1, i32 2> <i>; yields: result=<2 x i32> < i32 2, i32 4></i>
2652 <!-- _______________________________________________________________________ -->
2653 <div class="doc_subsubsection"> <a name="i_lshr">'<tt>lshr</tt>'
2654 Instruction</a> </div>
2655 <div class="doc_text">
2657 <pre> <result> = lshr <ty> <op1>, <op2> <i>; yields {ty}:result</i>
2661 <p>The '<tt>lshr</tt>' instruction (logical shift right) returns the first
2662 operand shifted to the right a specified number of bits with zero fill.</p>
2665 <p>Both arguments to the '<tt>lshr</tt>' instruction must be the same
2666 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2667 type. '<tt>op2</tt>' is treated as an unsigned value.</p>
2671 <p>This instruction always performs a logical shift right operation. The most
2672 significant bits of the result will be filled with zero bits after the
2673 shift. If <tt>op2</tt> is (statically or dynamically) equal to or larger than
2674 the number of bits in <tt>op1</tt>, the result is undefined. If the arguments are
2675 vectors, each vector element of <tt>op1</tt> is shifted by the corresponding shift
2676 amount in <tt>op2</tt>.</p>
2680 <result> = lshr i32 4, 1 <i>; yields {i32}:result = 2</i>
2681 <result> = lshr i32 4, 2 <i>; yields {i32}:result = 1</i>
2682 <result> = lshr i8 4, 3 <i>; yields {i8}:result = 0</i>
2683 <result> = lshr i8 -2, 1 <i>; yields {i8}:result = 0x7FFFFFFF </i>
2684 <result> = lshr i32 1, 32 <i>; undefined</i>
2685 <result> = lshr <2 x i32> < i32 -2, i32 4>, < i32 1, i32 2> <i>; yields: result=<2 x i32> < i32 0x7FFFFFFF, i32 1></i>
2689 <!-- _______________________________________________________________________ -->
2690 <div class="doc_subsubsection"> <a name="i_ashr">'<tt>ashr</tt>'
2691 Instruction</a> </div>
2692 <div class="doc_text">
2695 <pre> <result> = ashr <ty> <op1>, <op2> <i>; yields {ty}:result</i>
2699 <p>The '<tt>ashr</tt>' instruction (arithmetic shift right) returns the first
2700 operand shifted to the right a specified number of bits with sign extension.</p>
2703 <p>Both arguments to the '<tt>ashr</tt>' instruction must be the same
2704 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2705 type. '<tt>op2</tt>' is treated as an unsigned value.</p>
2708 <p>This instruction always performs an arithmetic shift right operation,
2709 The most significant bits of the result will be filled with the sign bit
2710 of <tt>op1</tt>. If <tt>op2</tt> is (statically or dynamically) equal to or
2711 larger than the number of bits in <tt>op1</tt>, the result is undefined. If the
2712 arguments are vectors, each vector element of <tt>op1</tt> is shifted by the
2713 corresponding shift amount in <tt>op2</tt>.</p>
2717 <result> = ashr i32 4, 1 <i>; yields {i32}:result = 2</i>
2718 <result> = ashr i32 4, 2 <i>; yields {i32}:result = 1</i>
2719 <result> = ashr i8 4, 3 <i>; yields {i8}:result = 0</i>
2720 <result> = ashr i8 -2, 1 <i>; yields {i8}:result = -1</i>
2721 <result> = ashr i32 1, 32 <i>; undefined</i>
2722 <result> = ashr <2 x i32> < i32 -2, i32 4>, < i32 1, i32 3> <i>; yields: result=<2 x i32> < i32 -1, i32 0></i>
2726 <!-- _______________________________________________________________________ -->
2727 <div class="doc_subsubsection"> <a name="i_and">'<tt>and</tt>'
2728 Instruction</a> </div>
2730 <div class="doc_text">
2735 <result> = and <ty> <op1>, <op2> <i>; yields {ty}:result</i>
2740 <p>The '<tt>and</tt>' instruction returns the bitwise logical and of
2741 its two operands.</p>
2745 <p>The two arguments to the '<tt>and</tt>' instruction must be
2746 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2747 values. Both arguments must have identical types.</p>
2750 <p>The truth table used for the '<tt>and</tt>' instruction is:</p>
2753 <table border="1" cellspacing="0" cellpadding="4">
2785 <result> = and i32 4, %var <i>; yields {i32}:result = 4 & %var</i>
2786 <result> = and i32 15, 40 <i>; yields {i32}:result = 8</i>
2787 <result> = and i32 4, 8 <i>; yields {i32}:result = 0</i>
2790 <!-- _______________________________________________________________________ -->
2791 <div class="doc_subsubsection"> <a name="i_or">'<tt>or</tt>' Instruction</a> </div>
2792 <div class="doc_text">
2794 <pre> <result> = or <ty> <op1>, <op2> <i>; yields {ty}:result</i>
2797 <p>The '<tt>or</tt>' instruction returns the bitwise logical inclusive
2798 or of its two operands.</p>
2801 <p>The two arguments to the '<tt>or</tt>' instruction must be
2802 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2803 values. Both arguments must have identical types.</p>
2805 <p>The truth table used for the '<tt>or</tt>' instruction is:</p>
2808 <table border="1" cellspacing="0" cellpadding="4">
2839 <pre> <result> = or i32 4, %var <i>; yields {i32}:result = 4 | %var</i>
2840 <result> = or i32 15, 40 <i>; yields {i32}:result = 47</i>
2841 <result> = or i32 4, 8 <i>; yields {i32}:result = 12</i>
2844 <!-- _______________________________________________________________________ -->
2845 <div class="doc_subsubsection"> <a name="i_xor">'<tt>xor</tt>'
2846 Instruction</a> </div>
2847 <div class="doc_text">
2849 <pre> <result> = xor <ty> <op1>, <op2> <i>; yields {ty}:result</i>
2852 <p>The '<tt>xor</tt>' instruction returns the bitwise logical exclusive
2853 or of its two operands. The <tt>xor</tt> is used to implement the
2854 "one's complement" operation, which is the "~" operator in C.</p>
2856 <p>The two arguments to the '<tt>xor</tt>' instruction must be
2857 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2858 values. Both arguments must have identical types.</p>
2862 <p>The truth table used for the '<tt>xor</tt>' instruction is:</p>
2865 <table border="1" cellspacing="0" cellpadding="4">
2897 <pre> <result> = xor i32 4, %var <i>; yields {i32}:result = 4 ^ %var</i>
2898 <result> = xor i32 15, 40 <i>; yields {i32}:result = 39</i>
2899 <result> = xor i32 4, 8 <i>; yields {i32}:result = 12</i>
2900 <result> = xor i32 %V, -1 <i>; yields {i32}:result = ~%V</i>
2904 <!-- ======================================================================= -->
2905 <div class="doc_subsection">
2906 <a name="vectorops">Vector Operations</a>
2909 <div class="doc_text">
2911 <p>LLVM supports several instructions to represent vector operations in a
2912 target-independent manner. These instructions cover the element-access and
2913 vector-specific operations needed to process vectors effectively. While LLVM
2914 does directly support these vector operations, many sophisticated algorithms
2915 will want to use target-specific intrinsics to take full advantage of a specific
2920 <!-- _______________________________________________________________________ -->
2921 <div class="doc_subsubsection">
2922 <a name="i_extractelement">'<tt>extractelement</tt>' Instruction</a>
2925 <div class="doc_text">
2930 <result> = extractelement <n x <ty>> <val>, i32 <idx> <i>; yields <ty></i>
2936 The '<tt>extractelement</tt>' instruction extracts a single scalar
2937 element from a vector at a specified index.
2944 The first operand of an '<tt>extractelement</tt>' instruction is a
2945 value of <a href="#t_vector">vector</a> type. The second operand is
2946 an index indicating the position from which to extract the element.
2947 The index may be a variable.</p>
2952 The result is a scalar of the same type as the element type of
2953 <tt>val</tt>. Its value is the value at position <tt>idx</tt> of
2954 <tt>val</tt>. If <tt>idx</tt> exceeds the length of <tt>val</tt>, the
2955 results are undefined.
2961 %result = extractelement <4 x i32> %vec, i32 0 <i>; yields i32</i>
2966 <!-- _______________________________________________________________________ -->
2967 <div class="doc_subsubsection">
2968 <a name="i_insertelement">'<tt>insertelement</tt>' Instruction</a>
2971 <div class="doc_text">
2976 <result> = insertelement <n x <ty>> <val>, <ty> <elt>, i32 <idx> <i>; yields <n x <ty>></i>
2982 The '<tt>insertelement</tt>' instruction inserts a scalar
2983 element into a vector at a specified index.
2990 The first operand of an '<tt>insertelement</tt>' instruction is a
2991 value of <a href="#t_vector">vector</a> type. The second operand is a
2992 scalar value whose type must equal the element type of the first
2993 operand. The third operand is an index indicating the position at
2994 which to insert the value. The index may be a variable.</p>
2999 The result is a vector of the same type as <tt>val</tt>. Its
3000 element values are those of <tt>val</tt> except at position
3001 <tt>idx</tt>, where it gets the value <tt>elt</tt>. If <tt>idx</tt>
3002 exceeds the length of <tt>val</tt>, the results are undefined.
3008 %result = insertelement <4 x i32> %vec, i32 1, i32 0 <i>; yields <4 x i32></i>
3012 <!-- _______________________________________________________________________ -->
3013 <div class="doc_subsubsection">
3014 <a name="i_shufflevector">'<tt>shufflevector</tt>' Instruction</a>
3017 <div class="doc_text">
3022 <result> = shufflevector <n x <ty>> <v1>, <n x <ty>> <v2>, <m x i32> <mask> <i>; yields <m x <ty>></i>
3028 The '<tt>shufflevector</tt>' instruction constructs a permutation of elements
3029 from two input vectors, returning a vector with the same element type as
3030 the input and length that is the same as the shuffle mask.
3036 The first two operands of a '<tt>shufflevector</tt>' instruction are vectors
3037 with types that match each other. The third argument is a shuffle mask whose
3038 element type is always 'i32'. The result of the instruction is a vector whose
3039 length is the same as the shuffle mask and whose element type is the same as
3040 the element type of the first two operands.
3044 The shuffle mask operand is required to be a constant vector with either
3045 constant integer or undef values.
3051 The elements of the two input vectors are numbered from left to right across
3052 both of the vectors. The shuffle mask operand specifies, for each element of
3053 the result vector, which element of the two input vectors the result element
3054 gets. The element selector may be undef (meaning "don't care") and the second
3055 operand may be undef if performing a shuffle from only one vector.
3061 %result = shufflevector <4 x i32> %v1, <4 x i32> %v2,
3062 <4 x i32> <i32 0, i32 4, i32 1, i32 5> <i>; yields <4 x i32></i>
3063 %result = shufflevector <4 x i32> %v1, <4 x i32> undef,
3064 <4 x i32> <i32 0, i32 1, i32 2, i32 3> <i>; yields <4 x i32></i> - Identity shuffle.
3065 %result = shufflevector <8 x i32> %v1, <8 x i32> undef,
3066 <4 x i32> <i32 0, i32 1, i32 2, i32 3> <i>; yields <4 x i32></i>
3067 %result = shufflevector <4 x i32> %v1, <4 x i32> %v2,
3068 <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>
3073 <!-- ======================================================================= -->
3074 <div class="doc_subsection">
3075 <a name="aggregateops">Aggregate Operations</a>
3078 <div class="doc_text">
3080 <p>LLVM supports several instructions for working with aggregate values.
3085 <!-- _______________________________________________________________________ -->
3086 <div class="doc_subsubsection">
3087 <a name="i_extractvalue">'<tt>extractvalue</tt>' Instruction</a>
3090 <div class="doc_text">
3095 <result> = extractvalue <aggregate type> <val>, <idx>{, <idx>}*
3101 The '<tt>extractvalue</tt>' instruction extracts the value of a struct field
3102 or array element from an aggregate value.
3109 The first operand of an '<tt>extractvalue</tt>' instruction is a
3110 value of <a href="#t_struct">struct</a> or <a href="#t_array">array</a>
3111 type. The operands are constant indices to specify which value to extract
3112 in a similar manner as indices in a
3113 '<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.
3119 The result is the value at the position in the aggregate specified by
3126 %result = extractvalue {i32, float} %agg, 0 <i>; yields i32</i>
3131 <!-- _______________________________________________________________________ -->
3132 <div class="doc_subsubsection">
3133 <a name="i_insertvalue">'<tt>insertvalue</tt>' Instruction</a>
3136 <div class="doc_text">
3141 <result> = insertvalue <aggregate type> <val>, <ty> <val>, <idx> <i>; yields <n x <ty>></i>
3147 The '<tt>insertvalue</tt>' instruction inserts a value
3148 into a struct field or array element in an aggregate.
3155 The first operand of an '<tt>insertvalue</tt>' instruction is a
3156 value of <a href="#t_struct">struct</a> or <a href="#t_array">array</a> type.
3157 The second operand is a first-class value to insert.
3158 The following operands are constant indices
3159 indicating the position at which to insert the value in a similar manner as
3161 '<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.
3162 The value to insert must have the same type as the value identified
3169 The result is an aggregate of the same type as <tt>val</tt>. Its
3170 value is that of <tt>val</tt> except that the value at the position
3171 specified by the indices is that of <tt>elt</tt>.
3177 %result = insertvalue {i32, float} %agg, i32 1, 0 <i>; yields {i32, float}</i>
3182 <!-- ======================================================================= -->
3183 <div class="doc_subsection">
3184 <a name="memoryops">Memory Access and Addressing Operations</a>
3187 <div class="doc_text">
3189 <p>A key design point of an SSA-based representation is how it
3190 represents memory. In LLVM, no memory locations are in SSA form, which
3191 makes things very simple. This section describes how to read, write,
3192 allocate, and free memory in LLVM.</p>
3196 <!-- _______________________________________________________________________ -->
3197 <div class="doc_subsubsection">
3198 <a name="i_malloc">'<tt>malloc</tt>' Instruction</a>
3201 <div class="doc_text">
3206 <result> = malloc <type>[, i32 <NumElements>][, align <alignment>] <i>; yields {type*}:result</i>
3211 <p>The '<tt>malloc</tt>' instruction allocates memory from the system
3212 heap and returns a pointer to it. The object is always allocated in the generic
3213 address space (address space zero).</p>
3217 <p>The '<tt>malloc</tt>' instruction allocates
3218 <tt>sizeof(<type>)*NumElements</tt>
3219 bytes of memory from the operating system and returns a pointer of the
3220 appropriate type to the program. If "NumElements" is specified, it is the
3221 number of elements allocated, otherwise "NumElements" is defaulted to be one.
3222 If a constant alignment is specified, the value result of the allocation is guaranteed to
3223 be aligned to at least that boundary. If not specified, or if zero, the target can
3224 choose to align the allocation on any convenient boundary.</p>
3226 <p>'<tt>type</tt>' must be a sized type.</p>
3230 <p>Memory is allocated using the system "<tt>malloc</tt>" function, and
3231 a pointer is returned. The result of a zero byte allocation is undefined. The
3232 result is null if there is insufficient memory available.</p>
3237 %array = malloc [4 x i8] <i>; yields {[%4 x i8]*}:array</i>
3239 %size = <a href="#i_add">add</a> i32 2, 2 <i>; yields {i32}:size = i32 4</i>
3240 %array1 = malloc i8, i32 4 <i>; yields {i8*}:array1</i>
3241 %array2 = malloc [12 x i8], i32 %size <i>; yields {[12 x i8]*}:array2</i>
3242 %array3 = malloc i32, i32 4, align 1024 <i>; yields {i32*}:array3</i>
3243 %array4 = malloc i32, align 1024 <i>; yields {i32*}:array4</i>
3246 <p>Note that the code generator does not yet respect the
3247 alignment value.</p>
3251 <!-- _______________________________________________________________________ -->
3252 <div class="doc_subsubsection">
3253 <a name="i_free">'<tt>free</tt>' Instruction</a>
3256 <div class="doc_text">
3261 free <type> <value> <i>; yields {void}</i>
3266 <p>The '<tt>free</tt>' instruction returns memory back to the unused
3267 memory heap to be reallocated in the future.</p>
3271 <p>'<tt>value</tt>' shall be a pointer value that points to a value
3272 that was allocated with the '<tt><a href="#i_malloc">malloc</a></tt>'
3277 <p>Access to the memory pointed to by the pointer is no longer defined
3278 after this instruction executes. If the pointer is null, the operation
3284 %array = <a href="#i_malloc">malloc</a> [4 x i8] <i>; yields {[4 x i8]*}:array</i>
3285 free [4 x i8]* %array
3289 <!-- _______________________________________________________________________ -->
3290 <div class="doc_subsubsection">
3291 <a name="i_alloca">'<tt>alloca</tt>' Instruction</a>
3294 <div class="doc_text">
3299 <result> = alloca <type>[, i32 <NumElements>][, align <alignment>] <i>; yields {type*}:result</i>
3304 <p>The '<tt>alloca</tt>' instruction allocates memory on the stack frame of the
3305 currently executing function, to be automatically released when this function
3306 returns to its caller. The object is always allocated in the generic address
3307 space (address space zero).</p>
3311 <p>The '<tt>alloca</tt>' instruction allocates <tt>sizeof(<type>)*NumElements</tt>
3312 bytes of memory on the runtime stack, returning a pointer of the
3313 appropriate type to the program. If "NumElements" is specified, it is the
3314 number of elements allocated, otherwise "NumElements" is defaulted to be one.
3315 If a constant alignment is specified, the value result of the allocation is guaranteed
3316 to be aligned to at least that boundary. If not specified, or if zero, the target
3317 can choose to align the allocation on any convenient boundary.</p>
3319 <p>'<tt>type</tt>' may be any sized type.</p>
3323 <p>Memory is allocated; a pointer is returned. The operation is undefiend if
3324 there is insufficient stack space for the allocation. '<tt>alloca</tt>'d
3325 memory is automatically released when the function returns. The '<tt>alloca</tt>'
3326 instruction is commonly used to represent automatic variables that must
3327 have an address available. When the function returns (either with the <tt><a
3328 href="#i_ret">ret</a></tt> or <tt><a href="#i_unwind">unwind</a></tt>
3329 instructions), the memory is reclaimed. Allocating zero bytes
3330 is legal, but the result is undefined.</p>
3335 %ptr = alloca i32 <i>; yields {i32*}:ptr</i>
3336 %ptr = alloca i32, i32 4 <i>; yields {i32*}:ptr</i>
3337 %ptr = alloca i32, i32 4, align 1024 <i>; yields {i32*}:ptr</i>
3338 %ptr = alloca i32, align 1024 <i>; yields {i32*}:ptr</i>
3342 <!-- _______________________________________________________________________ -->
3343 <div class="doc_subsubsection"> <a name="i_load">'<tt>load</tt>'
3344 Instruction</a> </div>
3345 <div class="doc_text">
3347 <pre> <result> = load <ty>* <pointer>[, align <alignment>]<br> <result> = volatile load <ty>* <pointer>[, align <alignment>]<br></pre>
3349 <p>The '<tt>load</tt>' instruction is used to read from memory.</p>
3351 <p>The argument to the '<tt>load</tt>' instruction specifies the memory
3352 address from which to load. The pointer must point to a <a
3353 href="#t_firstclass">first class</a> type. If the <tt>load</tt> is
3354 marked as <tt>volatile</tt>, then the optimizer is not allowed to modify
3355 the number or order of execution of this <tt>load</tt> with other
3356 volatile <tt>load</tt> and <tt><a href="#i_store">store</a></tt>
3359 The optional constant "align" argument specifies the alignment of the operation
3360 (that is, the alignment of the memory address). A value of 0 or an
3361 omitted "align" argument means that the operation has the preferential
3362 alignment for the target. It is the responsibility of the code emitter
3363 to ensure that the alignment information is correct. Overestimating
3364 the alignment results in an undefined behavior. Underestimating the
3365 alignment may produce less efficient code. An alignment of 1 is always
3369 <p>The location of memory pointed to is loaded.</p>
3371 <pre> %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
3373 href="#i_store">store</a> i32 3, i32* %ptr <i>; yields {void}</i>
3374 %val = load i32* %ptr <i>; yields {i32}:val = i32 3</i>
3377 <!-- _______________________________________________________________________ -->
3378 <div class="doc_subsubsection"> <a name="i_store">'<tt>store</tt>'
3379 Instruction</a> </div>
3380 <div class="doc_text">
3382 <pre> store <ty> <value>, <ty>* <pointer>[, align <alignment>] <i>; yields {void}</i>
3383 volatile store <ty> <value>, <ty>* <pointer>[, align <alignment>] <i>; yields {void}</i>
3386 <p>The '<tt>store</tt>' instruction is used to write to memory.</p>
3388 <p>There are two arguments to the '<tt>store</tt>' instruction: a value
3389 to store and an address at which to store it. The type of the '<tt><pointer></tt>'
3390 operand must be a pointer to the <a href="#t_firstclass">first class</a> type
3391 of the '<tt><value></tt>'
3392 operand. If the <tt>store</tt> is marked as <tt>volatile</tt>, then the
3393 optimizer is not allowed to modify the number or order of execution of
3394 this <tt>store</tt> with other volatile <tt>load</tt> and <tt><a
3395 href="#i_store">store</a></tt> instructions.</p>
3397 The optional constant "align" argument specifies the alignment of the operation
3398 (that is, the alignment of the memory address). A value of 0 or an
3399 omitted "align" argument means that the operation has the preferential
3400 alignment for the target. It is the responsibility of the code emitter
3401 to ensure that the alignment information is correct. Overestimating
3402 the alignment results in an undefined behavior. Underestimating the
3403 alignment may produce less efficient code. An alignment of 1 is always
3407 <p>The contents of memory are updated to contain '<tt><value></tt>'
3408 at the location specified by the '<tt><pointer></tt>' operand.</p>
3410 <pre> %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
3411 store i32 3, i32* %ptr <i>; yields {void}</i>
3412 %val = <a href="#i_load">load</a> i32* %ptr <i>; yields {i32}:val = i32 3</i>
3416 <!-- _______________________________________________________________________ -->
3417 <div class="doc_subsubsection">
3418 <a name="i_getelementptr">'<tt>getelementptr</tt>' Instruction</a>
3421 <div class="doc_text">
3424 <result> = getelementptr <pty>* <ptrval>{, <ty> <idx>}*
3430 The '<tt>getelementptr</tt>' instruction is used to get the address of a
3431 subelement of an aggregate data structure. It performs address calculation only
3432 and does not access memory.</p>
3436 <p>The first argument is always a pointer, and forms the basis of the
3437 calculation. The remaining arguments are indices, that indicate which of the
3438 elements of the aggregate object are indexed. The interpretation of each index
3439 is dependent on the type being indexed into. The first index always indexes the
3440 pointer value given as the first argument, the second index indexes a value of
3441 the type pointed to (not necessarily the value directly pointed to, since the
3442 first index can be non-zero), etc. The first type indexed into must be a pointer
3443 value, subsequent types can be arrays, vectors and structs. Note that subsequent
3444 types being indexed into can never be pointers, since that would require loading
3445 the pointer before continuing calculation.</p>
3447 <p>The type of each index argument depends on the type it is indexing into.
3448 When indexing into a (packed) structure, only <tt>i32</tt> integer
3449 <b>constants</b> are allowed. When indexing into an array, pointer or vector,
3450 only integers of 32 or 64 bits are allowed (also non-constants). 32-bit values
3451 will be sign extended to 64-bits if required.</p>
3453 <p>For example, let's consider a C code fragment and how it gets
3454 compiled to LLVM:</p>
3456 <div class="doc_code">
3469 int *foo(struct ST *s) {
3470 return &s[1].Z.B[5][13];
3475 <p>The LLVM code generated by the GCC frontend is:</p>
3477 <div class="doc_code">
3479 %RT = <a href="#namedtypes">type</a> { i8 , [10 x [20 x i32]], i8 }
3480 %ST = <a href="#namedtypes">type</a> { i32, double, %RT }
3482 define i32* %foo(%ST* %s) {
3484 %reg = getelementptr %ST* %s, i32 1, i32 2, i32 1, i32 5, i32 13
3492 <p>In the example above, the first index is indexing into the '<tt>%ST*</tt>'
3493 type, which is a pointer, yielding a '<tt>%ST</tt>' = '<tt>{ i32, double, %RT
3494 }</tt>' type, a structure. The second index indexes into the third element of
3495 the structure, yielding a '<tt>%RT</tt>' = '<tt>{ i8 , [10 x [20 x i32]],
3496 i8 }</tt>' type, another structure. The third index indexes into the second
3497 element of the structure, yielding a '<tt>[10 x [20 x i32]]</tt>' type, an
3498 array. The two dimensions of the array are subscripted into, yielding an
3499 '<tt>i32</tt>' type. The '<tt>getelementptr</tt>' instruction returns a pointer
3500 to this element, thus computing a value of '<tt>i32*</tt>' type.</p>
3502 <p>Note that it is perfectly legal to index partially through a
3503 structure, returning a pointer to an inner element. Because of this,
3504 the LLVM code for the given testcase is equivalent to:</p>
3507 define i32* %foo(%ST* %s) {
3508 %t1 = getelementptr %ST* %s, i32 1 <i>; yields %ST*:%t1</i>
3509 %t2 = getelementptr %ST* %t1, i32 0, i32 2 <i>; yields %RT*:%t2</i>
3510 %t3 = getelementptr %RT* %t2, i32 0, i32 1 <i>; yields [10 x [20 x i32]]*:%t3</i>
3511 %t4 = getelementptr [10 x [20 x i32]]* %t3, i32 0, i32 5 <i>; yields [20 x i32]*:%t4</i>
3512 %t5 = getelementptr [20 x i32]* %t4, i32 0, i32 13 <i>; yields i32*:%t5</i>
3517 <p>Note that it is undefined to access an array out of bounds: array and
3518 pointer indexes must always be within the defined bounds of the array type.
3519 The one exception for this rule is zero length arrays. These arrays are
3520 defined to be accessible as variable length arrays, which requires access
3521 beyond the zero'th element.</p>
3523 <p>The getelementptr instruction is often confusing. For some more insight
3524 into how it works, see <a href="GetElementPtr.html">the getelementptr
3530 <i>; yields [12 x i8]*:aptr</i>
3531 %aptr = getelementptr {i32, [12 x i8]}* %saptr, i64 0, i32 1
3532 <i>; yields i8*:vptr</i>
3533 %vptr = getelementptr {i32, <2 x i8>}* %svptr, i64 0, i32 1, i32 1
3534 <i>; yields i8*:eptr</i>
3535 %eptr = getelementptr [12 x i8]* %aptr, i64 0, i32 1
3539 <!-- ======================================================================= -->
3540 <div class="doc_subsection"> <a name="convertops">Conversion Operations</a>
3542 <div class="doc_text">
3543 <p>The instructions in this category are the conversion instructions (casting)
3544 which all take a single operand and a type. They perform various bit conversions
3548 <!-- _______________________________________________________________________ -->
3549 <div class="doc_subsubsection">
3550 <a name="i_trunc">'<tt>trunc .. to</tt>' Instruction</a>
3552 <div class="doc_text">
3556 <result> = trunc <ty> <value> to <ty2> <i>; yields ty2</i>
3561 The '<tt>trunc</tt>' instruction truncates its operand to the type <tt>ty2</tt>.
3566 The '<tt>trunc</tt>' instruction takes a <tt>value</tt> to trunc, which must
3567 be an <a href="#t_integer">integer</a> type, and a type that specifies the size
3568 and type of the result, which must be an <a href="#t_integer">integer</a>
3569 type. The bit size of <tt>value</tt> must be larger than the bit size of
3570 <tt>ty2</tt>. Equal sized types are not allowed.</p>
3574 The '<tt>trunc</tt>' instruction truncates the high order bits in <tt>value</tt>
3575 and converts the remaining bits to <tt>ty2</tt>. Since the source size must be
3576 larger than the destination size, <tt>trunc</tt> cannot be a <i>no-op cast</i>.
3577 It will always truncate bits.</p>
3581 %X = trunc i32 257 to i8 <i>; yields i8:1</i>
3582 %Y = trunc i32 123 to i1 <i>; yields i1:true</i>
3583 %Y = trunc i32 122 to i1 <i>; yields i1:false</i>
3587 <!-- _______________________________________________________________________ -->
3588 <div class="doc_subsubsection">
3589 <a name="i_zext">'<tt>zext .. to</tt>' Instruction</a>
3591 <div class="doc_text">
3595 <result> = zext <ty> <value> to <ty2> <i>; yields ty2</i>
3599 <p>The '<tt>zext</tt>' instruction zero extends its operand to type
3604 <p>The '<tt>zext</tt>' instruction takes a value to cast, which must be of
3605 <a href="#t_integer">integer</a> type, and a type to cast it to, which must
3606 also be of <a href="#t_integer">integer</a> type. The bit size of the
3607 <tt>value</tt> must be smaller than the bit size of the destination type,
3611 <p>The <tt>zext</tt> fills the high order bits of the <tt>value</tt> with zero
3612 bits until it reaches the size of the destination type, <tt>ty2</tt>.</p>
3614 <p>When zero extending from i1, the result will always be either 0 or 1.</p>
3618 %X = zext i32 257 to i64 <i>; yields i64:257</i>
3619 %Y = zext i1 true to i32 <i>; yields i32:1</i>
3623 <!-- _______________________________________________________________________ -->
3624 <div class="doc_subsubsection">
3625 <a name="i_sext">'<tt>sext .. to</tt>' Instruction</a>
3627 <div class="doc_text">
3631 <result> = sext <ty> <value> to <ty2> <i>; yields ty2</i>
3635 <p>The '<tt>sext</tt>' sign extends <tt>value</tt> to the type <tt>ty2</tt>.</p>
3639 The '<tt>sext</tt>' instruction takes a value to cast, which must be of
3640 <a href="#t_integer">integer</a> type, and a type to cast it to, which must
3641 also be of <a href="#t_integer">integer</a> type. The bit size of the
3642 <tt>value</tt> must be smaller than the bit size of the destination type,
3647 The '<tt>sext</tt>' instruction performs a sign extension by copying the sign
3648 bit (highest order bit) of the <tt>value</tt> until it reaches the bit size of
3649 the type <tt>ty2</tt>.</p>
3651 <p>When sign extending from i1, the extension always results in -1 or 0.</p>
3655 %X = sext i8 -1 to i16 <i>; yields i16 :65535</i>
3656 %Y = sext i1 true to i32 <i>; yields i32:-1</i>
3660 <!-- _______________________________________________________________________ -->
3661 <div class="doc_subsubsection">
3662 <a name="i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a>
3665 <div class="doc_text">
3670 <result> = fptrunc <ty> <value> to <ty2> <i>; yields ty2</i>
3674 <p>The '<tt>fptrunc</tt>' instruction truncates <tt>value</tt> to type
3679 <p>The '<tt>fptrunc</tt>' instruction takes a <a href="#t_floating">floating
3680 point</a> value to cast and a <a href="#t_floating">floating point</a> type to
3681 cast it to. The size of <tt>value</tt> must be larger than the size of
3682 <tt>ty2</tt>. This implies that <tt>fptrunc</tt> cannot be used to make a
3683 <i>no-op cast</i>.</p>
3686 <p> The '<tt>fptrunc</tt>' instruction truncates a <tt>value</tt> from a larger
3687 <a href="#t_floating">floating point</a> type to a smaller
3688 <a href="#t_floating">floating point</a> type. If the value cannot fit within
3689 the destination type, <tt>ty2</tt>, then the results are undefined.</p>
3693 %X = fptrunc double 123.0 to float <i>; yields float:123.0</i>
3694 %Y = fptrunc double 1.0E+300 to float <i>; yields undefined</i>
3698 <!-- _______________________________________________________________________ -->
3699 <div class="doc_subsubsection">
3700 <a name="i_fpext">'<tt>fpext .. to</tt>' Instruction</a>
3702 <div class="doc_text">
3706 <result> = fpext <ty> <value> to <ty2> <i>; yields ty2</i>
3710 <p>The '<tt>fpext</tt>' extends a floating point <tt>value</tt> to a larger
3711 floating point value.</p>
3714 <p>The '<tt>fpext</tt>' instruction takes a
3715 <a href="#t_floating">floating point</a> <tt>value</tt> to cast,
3716 and a <a href="#t_floating">floating point</a> type to cast it to. The source
3717 type must be smaller than the destination type.</p>
3720 <p>The '<tt>fpext</tt>' instruction extends the <tt>value</tt> from a smaller
3721 <a href="#t_floating">floating point</a> type to a larger
3722 <a href="#t_floating">floating point</a> type. The <tt>fpext</tt> cannot be
3723 used to make a <i>no-op cast</i> because it always changes bits. Use
3724 <tt>bitcast</tt> to make a <i>no-op cast</i> for a floating point cast.</p>
3728 %X = fpext float 3.1415 to double <i>; yields double:3.1415</i>
3729 %Y = fpext float 1.0 to float <i>; yields float:1.0 (no-op)</i>
3733 <!-- _______________________________________________________________________ -->
3734 <div class="doc_subsubsection">
3735 <a name="i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a>
3737 <div class="doc_text">
3741 <result> = fptoui <ty> <value> to <ty2> <i>; yields ty2</i>
3745 <p>The '<tt>fptoui</tt>' converts a floating point <tt>value</tt> to its
3746 unsigned integer equivalent of type <tt>ty2</tt>.
3750 <p>The '<tt>fptoui</tt>' instruction takes a value to cast, which must be a
3751 scalar or vector <a href="#t_floating">floating point</a> value, and a type
3752 to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
3753 type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
3754 vector integer type with the same number of elements as <tt>ty</tt></p>
3757 <p> The '<tt>fptoui</tt>' instruction converts its
3758 <a href="#t_floating">floating point</a> operand into the nearest (rounding
3759 towards zero) unsigned integer value. If the value cannot fit in <tt>ty2</tt>,
3760 the results are undefined.</p>
3764 %X = fptoui double 123.0 to i32 <i>; yields i32:123</i>
3765 %Y = fptoui float 1.0E+300 to i1 <i>; yields undefined:1</i>
3766 %X = fptoui float 1.04E+17 to i8 <i>; yields undefined:1</i>
3770 <!-- _______________________________________________________________________ -->
3771 <div class="doc_subsubsection">
3772 <a name="i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a>
3774 <div class="doc_text">
3778 <result> = fptosi <ty> <value> to <ty2> <i>; yields ty2</i>
3782 <p>The '<tt>fptosi</tt>' instruction converts
3783 <a href="#t_floating">floating point</a> <tt>value</tt> to type <tt>ty2</tt>.
3787 <p> The '<tt>fptosi</tt>' instruction takes a value to cast, which must be a
3788 scalar or vector <a href="#t_floating">floating point</a> value, and a type
3789 to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
3790 type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
3791 vector integer type with the same number of elements as <tt>ty</tt></p>
3794 <p>The '<tt>fptosi</tt>' instruction converts its
3795 <a href="#t_floating">floating point</a> operand into the nearest (rounding
3796 towards zero) signed integer value. If the value cannot fit in <tt>ty2</tt>,
3797 the results are undefined.</p>
3801 %X = fptosi double -123.0 to i32 <i>; yields i32:-123</i>
3802 %Y = fptosi float 1.0E-247 to i1 <i>; yields undefined:1</i>
3803 %X = fptosi float 1.04E+17 to i8 <i>; yields undefined:1</i>
3807 <!-- _______________________________________________________________________ -->
3808 <div class="doc_subsubsection">
3809 <a name="i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a>
3811 <div class="doc_text">
3815 <result> = uitofp <ty> <value> to <ty2> <i>; yields ty2</i>
3819 <p>The '<tt>uitofp</tt>' instruction regards <tt>value</tt> as an unsigned
3820 integer and converts that value to the <tt>ty2</tt> type.</p>
3823 <p>The '<tt>uitofp</tt>' instruction takes a value to cast, which must be a
3824 scalar or vector <a href="#t_integer">integer</a> value, and a type to cast it
3825 to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
3826 type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
3827 floating point type with the same number of elements as <tt>ty</tt></p>
3830 <p>The '<tt>uitofp</tt>' instruction interprets its operand as an unsigned
3831 integer quantity and converts it to the corresponding floating point value. If
3832 the value cannot fit in the floating point value, the results are undefined.</p>
3836 %X = uitofp i32 257 to float <i>; yields float:257.0</i>
3837 %Y = uitofp i8 -1 to double <i>; yields double:255.0</i>
3841 <!-- _______________________________________________________________________ -->
3842 <div class="doc_subsubsection">
3843 <a name="i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a>
3845 <div class="doc_text">
3849 <result> = sitofp <ty> <value> to <ty2> <i>; yields ty2</i>
3853 <p>The '<tt>sitofp</tt>' instruction regards <tt>value</tt> as a signed
3854 integer and converts that value to the <tt>ty2</tt> type.</p>
3857 <p>The '<tt>sitofp</tt>' instruction takes a value to cast, which must be a
3858 scalar or vector <a href="#t_integer">integer</a> value, and a type to cast it
3859 to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
3860 type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
3861 floating point type with the same number of elements as <tt>ty</tt></p>
3864 <p>The '<tt>sitofp</tt>' instruction interprets its operand as a signed
3865 integer quantity and converts it to the corresponding floating point value. If
3866 the value cannot fit in the floating point value, the results are undefined.</p>
3870 %X = sitofp i32 257 to float <i>; yields float:257.0</i>
3871 %Y = sitofp i8 -1 to double <i>; yields double:-1.0</i>
3875 <!-- _______________________________________________________________________ -->
3876 <div class="doc_subsubsection">
3877 <a name="i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a>
3879 <div class="doc_text">
3883 <result> = ptrtoint <ty> <value> to <ty2> <i>; yields ty2</i>
3887 <p>The '<tt>ptrtoint</tt>' instruction converts the pointer <tt>value</tt> to
3888 the integer type <tt>ty2</tt>.</p>
3891 <p>The '<tt>ptrtoint</tt>' instruction takes a <tt>value</tt> to cast, which
3892 must be a <a href="#t_pointer">pointer</a> value, and a type to cast it to
3893 <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a> type.</p>
3896 <p>The '<tt>ptrtoint</tt>' instruction converts <tt>value</tt> to integer type
3897 <tt>ty2</tt> by interpreting the pointer value as an integer and either
3898 truncating or zero extending that value to the size of the integer type. If
3899 <tt>value</tt> is smaller than <tt>ty2</tt> then a zero extension is done. If
3900 <tt>value</tt> is larger than <tt>ty2</tt> then a truncation is done. If they
3901 are the same size, then nothing is done (<i>no-op cast</i>) other than a type
3906 %X = ptrtoint i32* %X to i8 <i>; yields truncation on 32-bit architecture</i>
3907 %Y = ptrtoint i32* %x to i64 <i>; yields zero extension on 32-bit architecture</i>
3911 <!-- _______________________________________________________________________ -->
3912 <div class="doc_subsubsection">
3913 <a name="i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a>
3915 <div class="doc_text">
3919 <result> = inttoptr <ty> <value> to <ty2> <i>; yields ty2</i>
3923 <p>The '<tt>inttoptr</tt>' instruction converts an integer <tt>value</tt> to
3924 a pointer type, <tt>ty2</tt>.</p>
3927 <p>The '<tt>inttoptr</tt>' instruction takes an <a href="#t_integer">integer</a>
3928 value to cast, and a type to cast it to, which must be a
3929 <a href="#t_pointer">pointer</a> type.</p>
3932 <p>The '<tt>inttoptr</tt>' instruction converts <tt>value</tt> to type
3933 <tt>ty2</tt> by applying either a zero extension or a truncation depending on
3934 the size of the integer <tt>value</tt>. If <tt>value</tt> is larger than the
3935 size of a pointer then a truncation is done. If <tt>value</tt> is smaller than
3936 the size of a pointer then a zero extension is done. If they are the same size,
3937 nothing is done (<i>no-op cast</i>).</p>
3941 %X = inttoptr i32 255 to i32* <i>; yields zero extension on 64-bit architecture</i>
3942 %X = inttoptr i32 255 to i32* <i>; yields no-op on 32-bit architecture</i>
3943 %Y = inttoptr i64 0 to i32* <i>; yields truncation on 32-bit architecture</i>
3947 <!-- _______________________________________________________________________ -->
3948 <div class="doc_subsubsection">
3949 <a name="i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a>
3951 <div class="doc_text">
3955 <result> = bitcast <ty> <value> to <ty2> <i>; yields ty2</i>
3960 <p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
3961 <tt>ty2</tt> without changing any bits.</p>
3965 <p>The '<tt>bitcast</tt>' instruction takes a value to cast, which must be
3966 a non-aggregate first class value, and a type to cast it to, which must also be
3967 a non-aggregate <a href="#t_firstclass">first class</a> type. The bit sizes of
3969 and the destination type, <tt>ty2</tt>, must be identical. If the source
3970 type is a pointer, the destination type must also be a pointer. This
3971 instruction supports bitwise conversion of vectors to integers and to vectors
3972 of other types (as long as they have the same size).</p>
3975 <p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
3976 <tt>ty2</tt>. It is always a <i>no-op cast</i> because no bits change with
3977 this conversion. The conversion is done as if the <tt>value</tt> had been
3978 stored to memory and read back as type <tt>ty2</tt>. Pointer types may only be
3979 converted to other pointer types with this instruction. To convert pointers to
3980 other types, use the <a href="#i_inttoptr">inttoptr</a> or
3981 <a href="#i_ptrtoint">ptrtoint</a> instructions first.</p>
3985 %X = bitcast i8 255 to i8 <i>; yields i8 :-1</i>
3986 %Y = bitcast i32* %x to sint* <i>; yields sint*:%x</i>
3987 %Z = bitcast <2 x int> %V to i64; <i>; yields i64: %V</i>
3991 <!-- ======================================================================= -->
3992 <div class="doc_subsection"> <a name="otherops">Other Operations</a> </div>
3993 <div class="doc_text">
3994 <p>The instructions in this category are the "miscellaneous"
3995 instructions, which defy better classification.</p>
3998 <!-- _______________________________________________________________________ -->
3999 <div class="doc_subsubsection"><a name="i_icmp">'<tt>icmp</tt>' Instruction</a>
4001 <div class="doc_text">
4003 <pre> <result> = icmp <cond> <ty> <op1>, <op2> <i>; yields {i1} or {<N x i1>}:result</i>
4006 <p>The '<tt>icmp</tt>' instruction returns a boolean value or
4007 a vector of boolean values based on comparison
4008 of its two integer, integer vector, or pointer operands.</p>
4010 <p>The '<tt>icmp</tt>' instruction takes three operands. The first operand is
4011 the condition code indicating the kind of comparison to perform. It is not
4012 a value, just a keyword. The possible condition code are:
4015 <li><tt>eq</tt>: equal</li>
4016 <li><tt>ne</tt>: not equal </li>
4017 <li><tt>ugt</tt>: unsigned greater than</li>
4018 <li><tt>uge</tt>: unsigned greater or equal</li>
4019 <li><tt>ult</tt>: unsigned less than</li>
4020 <li><tt>ule</tt>: unsigned less or equal</li>
4021 <li><tt>sgt</tt>: signed greater than</li>
4022 <li><tt>sge</tt>: signed greater or equal</li>
4023 <li><tt>slt</tt>: signed less than</li>
4024 <li><tt>sle</tt>: signed less or equal</li>
4026 <p>The remaining two arguments must be <a href="#t_integer">integer</a> or
4027 <a href="#t_pointer">pointer</a>
4028 or integer <a href="#t_vector">vector</a> typed.
4029 They must also be identical types.</p>
4031 <p>The '<tt>icmp</tt>' compares <tt>op1</tt> and <tt>op2</tt> according to
4032 the condition code given as <tt>cond</tt>. The comparison performed always
4033 yields either an <a href="#t_primitive"><tt>i1</tt></a> or vector of <tt>i1</tt> result, as follows:
4036 <li><tt>eq</tt>: yields <tt>true</tt> if the operands are equal,
4037 <tt>false</tt> otherwise. No sign interpretation is necessary or performed.
4039 <li><tt>ne</tt>: yields <tt>true</tt> if the operands are unequal,
4040 <tt>false</tt> otherwise. No sign interpretation is necessary or performed.</li>
4041 <li><tt>ugt</tt>: interprets the operands as unsigned values and yields
4042 <tt>true</tt> if <tt>op1</tt> is greater than <tt>op2</tt>.</li>
4043 <li><tt>uge</tt>: interprets the operands as unsigned values and yields
4044 <tt>true</tt> if <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
4045 <li><tt>ult</tt>: interprets the operands as unsigned values and yields
4046 <tt>true</tt> if <tt>op1</tt> is less than <tt>op2</tt>.</li>
4047 <li><tt>ule</tt>: interprets the operands as unsigned values and yields
4048 <tt>true</tt> if <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
4049 <li><tt>sgt</tt>: interprets the operands as signed values and yields
4050 <tt>true</tt> if <tt>op1</tt> is greater than <tt>op2</tt>.</li>
4051 <li><tt>sge</tt>: interprets the operands as signed values and yields
4052 <tt>true</tt> if <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
4053 <li><tt>slt</tt>: interprets the operands as signed values and yields
4054 <tt>true</tt> if <tt>op1</tt> is less than <tt>op2</tt>.</li>
4055 <li><tt>sle</tt>: interprets the operands as signed values and yields
4056 <tt>true</tt> if <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
4058 <p>If the operands are <a href="#t_pointer">pointer</a> typed, the pointer
4059 values are compared as if they were integers.</p>
4060 <p>If the operands are integer vectors, then they are compared
4061 element by element. The result is an <tt>i1</tt> vector with
4062 the same number of elements as the values being compared.
4063 Otherwise, the result is an <tt>i1</tt>.
4067 <pre> <result> = icmp eq i32 4, 5 <i>; yields: result=false</i>
4068 <result> = icmp ne float* %X, %X <i>; yields: result=false</i>
4069 <result> = icmp ult i16 4, 5 <i>; yields: result=true</i>
4070 <result> = icmp sgt i16 4, 5 <i>; yields: result=false</i>
4071 <result> = icmp ule i16 -4, 5 <i>; yields: result=false</i>
4072 <result> = icmp sge i16 4, 5 <i>; yields: result=false</i>
4076 <!-- _______________________________________________________________________ -->
4077 <div class="doc_subsubsection"><a name="i_fcmp">'<tt>fcmp</tt>' Instruction</a>
4079 <div class="doc_text">
4081 <pre> <result> = fcmp <cond> <ty> <op1>, <op2> <i>; yields {i1} or {<N x i1>}:result</i>
4084 <p>The '<tt>fcmp</tt>' instruction returns a boolean value
4085 or vector of boolean values based on comparison
4086 of its operands.</p>
4088 If the operands are floating point scalars, then the result
4089 type is a boolean (<a href="#t_primitive"><tt>i1</tt></a>).
4091 <p>If the operands are floating point vectors, then the result type
4092 is a vector of boolean with the same number of elements as the
4093 operands being compared.</p>
4095 <p>The '<tt>fcmp</tt>' instruction takes three operands. The first operand is
4096 the condition code indicating the kind of comparison to perform. It is not
4097 a value, just a keyword. The possible condition code are:</p>
4099 <li><tt>false</tt>: no comparison, always returns false</li>
4100 <li><tt>oeq</tt>: ordered and equal</li>
4101 <li><tt>ogt</tt>: ordered and greater than </li>
4102 <li><tt>oge</tt>: ordered and greater than or equal</li>
4103 <li><tt>olt</tt>: ordered and less than </li>
4104 <li><tt>ole</tt>: ordered and less than or equal</li>
4105 <li><tt>one</tt>: ordered and not equal</li>
4106 <li><tt>ord</tt>: ordered (no nans)</li>
4107 <li><tt>ueq</tt>: unordered or equal</li>
4108 <li><tt>ugt</tt>: unordered or greater than </li>
4109 <li><tt>uge</tt>: unordered or greater than or equal</li>
4110 <li><tt>ult</tt>: unordered or less than </li>
4111 <li><tt>ule</tt>: unordered or less than or equal</li>
4112 <li><tt>une</tt>: unordered or not equal</li>
4113 <li><tt>uno</tt>: unordered (either nans)</li>
4114 <li><tt>true</tt>: no comparison, always returns true</li>
4116 <p><i>Ordered</i> means that neither operand is a QNAN while
4117 <i>unordered</i> means that either operand may be a QNAN.</p>
4118 <p>Each of <tt>val1</tt> and <tt>val2</tt> arguments must be
4119 either a <a href="#t_floating">floating point</a> type
4120 or a <a href="#t_vector">vector</a> of floating point type.
4121 They must have identical types.</p>
4123 <p>The '<tt>fcmp</tt>' instruction compares <tt>op1</tt> and <tt>op2</tt>
4124 according to the condition code given as <tt>cond</tt>.
4125 If the operands are vectors, then the vectors are compared
4127 Each comparison performed
4128 always yields an <a href="#t_primitive">i1</a> result, as follows:</p>
4130 <li><tt>false</tt>: always yields <tt>false</tt>, regardless of operands.</li>
4131 <li><tt>oeq</tt>: yields <tt>true</tt> if both operands are not a QNAN and
4132 <tt>op1</tt> is equal to <tt>op2</tt>.</li>
4133 <li><tt>ogt</tt>: yields <tt>true</tt> if both operands are not a QNAN and
4134 <tt>op1</tt> is greather than <tt>op2</tt>.</li>
4135 <li><tt>oge</tt>: yields <tt>true</tt> if both operands are not a QNAN and
4136 <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
4137 <li><tt>olt</tt>: yields <tt>true</tt> if both operands are not a QNAN and
4138 <tt>op1</tt> is less than <tt>op2</tt>.</li>
4139 <li><tt>ole</tt>: yields <tt>true</tt> if both operands are not a QNAN and
4140 <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
4141 <li><tt>one</tt>: yields <tt>true</tt> if both operands are not a QNAN and
4142 <tt>op1</tt> is not equal to <tt>op2</tt>.</li>
4143 <li><tt>ord</tt>: yields <tt>true</tt> if both operands are not a QNAN.</li>
4144 <li><tt>ueq</tt>: yields <tt>true</tt> if either operand is a QNAN or
4145 <tt>op1</tt> is equal to <tt>op2</tt>.</li>
4146 <li><tt>ugt</tt>: yields <tt>true</tt> if either operand is a QNAN or
4147 <tt>op1</tt> is greater than <tt>op2</tt>.</li>
4148 <li><tt>uge</tt>: yields <tt>true</tt> if either operand is a QNAN or
4149 <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
4150 <li><tt>ult</tt>: yields <tt>true</tt> if either operand is a QNAN or
4151 <tt>op1</tt> is less than <tt>op2</tt>.</li>
4152 <li><tt>ule</tt>: yields <tt>true</tt> if either operand is a QNAN or
4153 <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
4154 <li><tt>une</tt>: yields <tt>true</tt> if either operand is a QNAN or
4155 <tt>op1</tt> is not equal to <tt>op2</tt>.</li>
4156 <li><tt>uno</tt>: yields <tt>true</tt> if either operand is a QNAN.</li>
4157 <li><tt>true</tt>: always yields <tt>true</tt>, regardless of operands.</li>
4161 <pre> <result> = fcmp oeq float 4.0, 5.0 <i>; yields: result=false</i>
4162 <result> = fcmp one float 4.0, 5.0 <i>; yields: result=true</i>
4163 <result> = fcmp olt float 4.0, 5.0 <i>; yields: result=true</i>
4164 <result> = fcmp ueq double 1.0, 2.0 <i>; yields: result=false</i>
4168 <!-- _______________________________________________________________________ -->
4169 <div class="doc_subsubsection">
4170 <a name="i_vicmp">'<tt>vicmp</tt>' Instruction</a>
4172 <div class="doc_text">
4174 <pre> <result> = vicmp <cond> <ty> <op1>, <op2> <i>; yields {ty}:result</i>
4177 <p>The '<tt>vicmp</tt>' instruction returns an integer vector value based on
4178 element-wise comparison of its two integer vector operands.</p>
4180 <p>The '<tt>vicmp</tt>' instruction takes three operands. The first operand is
4181 the condition code indicating the kind of comparison to perform. It is not
4182 a value, just a keyword. The possible condition code are:</p>
4184 <li><tt>eq</tt>: equal</li>
4185 <li><tt>ne</tt>: not equal </li>
4186 <li><tt>ugt</tt>: unsigned greater than</li>
4187 <li><tt>uge</tt>: unsigned greater or equal</li>
4188 <li><tt>ult</tt>: unsigned less than</li>
4189 <li><tt>ule</tt>: unsigned less or equal</li>
4190 <li><tt>sgt</tt>: signed greater than</li>
4191 <li><tt>sge</tt>: signed greater or equal</li>
4192 <li><tt>slt</tt>: signed less than</li>
4193 <li><tt>sle</tt>: signed less or equal</li>
4195 <p>The remaining two arguments must be <a href="#t_vector">vector</a> or
4196 <a href="#t_integer">integer</a> typed. They must also be identical types.</p>
4198 <p>The '<tt>vicmp</tt>' instruction compares <tt>op1</tt> and <tt>op2</tt>
4199 according to the condition code given as <tt>cond</tt>. The comparison yields a
4200 <a href="#t_vector">vector</a> of <a href="#t_integer">integer</a> result, of
4201 identical type as the values being compared. The most significant bit in each
4202 element is 1 if the element-wise comparison evaluates to true, and is 0
4203 otherwise. All other bits of the result are undefined. The condition codes
4204 are evaluated identically to the <a href="#i_icmp">'<tt>icmp</tt>'
4205 instruction</a>.</p>
4209 <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>
4210 <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>
4214 <!-- _______________________________________________________________________ -->
4215 <div class="doc_subsubsection">
4216 <a name="i_vfcmp">'<tt>vfcmp</tt>' Instruction</a>
4218 <div class="doc_text">
4220 <pre> <result> = vfcmp <cond> <ty> <op1>, <op2></pre>
4222 <p>The '<tt>vfcmp</tt>' instruction returns an integer vector value based on
4223 element-wise comparison of its two floating point vector operands. The output
4224 elements have the same width as the input elements.</p>
4226 <p>The '<tt>vfcmp</tt>' instruction takes three operands. The first operand is
4227 the condition code indicating the kind of comparison to perform. It is not
4228 a value, just a keyword. The possible condition code are:</p>
4230 <li><tt>false</tt>: no comparison, always returns false</li>
4231 <li><tt>oeq</tt>: ordered and equal</li>
4232 <li><tt>ogt</tt>: ordered and greater than </li>
4233 <li><tt>oge</tt>: ordered and greater than or equal</li>
4234 <li><tt>olt</tt>: ordered and less than </li>
4235 <li><tt>ole</tt>: ordered and less than or equal</li>
4236 <li><tt>one</tt>: ordered and not equal</li>
4237 <li><tt>ord</tt>: ordered (no nans)</li>
4238 <li><tt>ueq</tt>: unordered or equal</li>
4239 <li><tt>ugt</tt>: unordered or greater than </li>
4240 <li><tt>uge</tt>: unordered or greater than or equal</li>
4241 <li><tt>ult</tt>: unordered or less than </li>
4242 <li><tt>ule</tt>: unordered or less than or equal</li>
4243 <li><tt>une</tt>: unordered or not equal</li>
4244 <li><tt>uno</tt>: unordered (either nans)</li>
4245 <li><tt>true</tt>: no comparison, always returns true</li>
4247 <p>The remaining two arguments must be <a href="#t_vector">vector</a> of
4248 <a href="#t_floating">floating point</a> typed. They must also be identical
4251 <p>The '<tt>vfcmp</tt>' instruction compares <tt>op1</tt> and <tt>op2</tt>
4252 according to the condition code given as <tt>cond</tt>. The comparison yields a
4253 <a href="#t_vector">vector</a> of <a href="#t_integer">integer</a> result, with
4254 an identical number of elements as the values being compared, and each element
4255 having identical with to the width of the floating point elements. The most
4256 significant bit in each element is 1 if the element-wise comparison evaluates to
4257 true, and is 0 otherwise. All other bits of the result are undefined. The
4258 condition codes are evaluated identically to the
4259 <a href="#i_fcmp">'<tt>fcmp</tt>' instruction</a>.</p>
4263 <i>; yields: result=<2 x i32> < i32 0, i32 -1 ></i>
4264 <result> = vfcmp oeq <2 x float> < float 4, float 0 >, < float 5, float 0 >
4266 <i>; yields: result=<2 x i64> < i64 -1, i64 0 ></i>
4267 <result> = vfcmp ult <2 x double> < double 1, double 2 >, < double 2, double 2>
4271 <!-- _______________________________________________________________________ -->
4272 <div class="doc_subsubsection">
4273 <a name="i_phi">'<tt>phi</tt>' Instruction</a>
4276 <div class="doc_text">
4280 <pre> <result> = phi <ty> [ <val0>, <label0>], ...<br></pre>
4282 <p>The '<tt>phi</tt>' instruction is used to implement the φ node in
4283 the SSA graph representing the function.</p>
4286 <p>The type of the incoming values is specified with the first type
4287 field. After this, the '<tt>phi</tt>' instruction takes a list of pairs
4288 as arguments, with one pair for each predecessor basic block of the
4289 current block. Only values of <a href="#t_firstclass">first class</a>
4290 type may be used as the value arguments to the PHI node. Only labels
4291 may be used as the label arguments.</p>
4293 <p>There must be no non-phi instructions between the start of a basic
4294 block and the PHI instructions: i.e. PHI instructions must be first in
4299 <p>At runtime, the '<tt>phi</tt>' instruction logically takes on the value
4300 specified by the pair corresponding to the predecessor basic block that executed
4301 just prior to the current block.</p>
4305 Loop: ; Infinite loop that counts from 0 on up...
4306 %indvar = phi i32 [ 0, %LoopHeader ], [ %nextindvar, %Loop ]
4307 %nextindvar = add i32 %indvar, 1
4312 <!-- _______________________________________________________________________ -->
4313 <div class="doc_subsubsection">
4314 <a name="i_select">'<tt>select</tt>' Instruction</a>
4317 <div class="doc_text">
4322 <result> = select <i>selty</i> <cond>, <ty> <val1>, <ty> <val2> <i>; yields ty</i>
4324 <i>selty</i> is either i1 or {<N x i1>}
4330 The '<tt>select</tt>' instruction is used to choose one value based on a
4331 condition, without branching.
4338 The '<tt>select</tt>' instruction requires an 'i1' value or
4339 a vector of 'i1' values indicating the
4340 condition, and two values of the same <a href="#t_firstclass">first class</a>
4341 type. If the val1/val2 are vectors and
4342 the condition is a scalar, then entire vectors are selected, not
4343 individual elements.
4349 If the condition is an i1 and it evaluates to 1, the instruction returns the first
4350 value argument; otherwise, it returns the second value argument.
4353 If the condition is a vector of i1, then the value arguments must
4354 be vectors of the same size, and the selection is done element
4361 %X = select i1 true, i8 17, i8 42 <i>; yields i8:17</i>
4366 <!-- _______________________________________________________________________ -->
4367 <div class="doc_subsubsection">
4368 <a name="i_call">'<tt>call</tt>' Instruction</a>
4371 <div class="doc_text">
4375 <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>]
4380 <p>The '<tt>call</tt>' instruction represents a simple function call.</p>
4384 <p>This instruction requires several arguments:</p>
4388 <p>The optional "tail" marker indicates whether the callee function accesses
4389 any allocas or varargs in the caller. If the "tail" marker is present, the
4390 function call is eligible for tail call optimization. Note that calls may
4391 be marked "tail" even if they do not occur before a <a
4392 href="#i_ret"><tt>ret</tt></a> instruction.</p>
4395 <p>The optional "cconv" marker indicates which <a href="#callingconv">calling
4396 convention</a> the call should use. If none is specified, the call defaults
4397 to using C calling conventions.</p>
4401 <p>The optional <a href="#paramattrs">Parameter Attributes</a> list for
4402 return values. Only '<tt>zeroext</tt>', '<tt>signext</tt>',
4403 and '<tt>inreg</tt>' attributes are valid here.</p>
4407 <p>'<tt>ty</tt>': the type of the call instruction itself which is also
4408 the type of the return value. Functions that return no value are marked
4409 <tt><a href="#t_void">void</a></tt>.</p>
4412 <p>'<tt>fnty</tt>': shall be the signature of the pointer to function
4413 value being invoked. The argument types must match the types implied by
4414 this signature. This type can be omitted if the function is not varargs
4415 and if the function type does not return a pointer to a function.</p>
4418 <p>'<tt>fnptrval</tt>': An LLVM value containing a pointer to a function to
4419 be invoked. In most cases, this is a direct function invocation, but
4420 indirect <tt>call</tt>s are just as possible, calling an arbitrary pointer
4421 to function value.</p>
4424 <p>'<tt>function args</tt>': argument list whose types match the
4425 function signature argument types. All arguments must be of
4426 <a href="#t_firstclass">first class</a> type. If the function signature
4427 indicates the function accepts a variable number of arguments, the extra
4428 arguments can be specified.</p>
4431 <p>The optional <a href="#fnattrs">function attributes</a> list. Only
4432 '<tt>noreturn</tt>', '<tt>nounwind</tt>', '<tt>readonly</tt>' and
4433 '<tt>readnone</tt>' attributes are valid here.</p>
4439 <p>The '<tt>call</tt>' instruction is used to cause control flow to
4440 transfer to a specified function, with its incoming arguments bound to
4441 the specified values. Upon a '<tt><a href="#i_ret">ret</a></tt>'
4442 instruction in the called function, control flow continues with the
4443 instruction after the function call, and the return value of the
4444 function is bound to the result argument.</p>
4449 %retval = call i32 @test(i32 %argc)
4450 call i32 (i8 *, ...)* @printf(i8 * %msg, i32 12, i8 42) <i>; yields i32</i>
4451 %X = tail call i32 @foo() <i>; yields i32</i>
4452 %Y = tail call <a href="#callingconv">fastcc</a> i32 @foo() <i>; yields i32</i>
4453 call void %foo(i8 97 signext)
4455 %struct.A = type { i32, i8 }
4456 %r = call %struct.A @foo() <i>; yields { 32, i8 }</i>
4457 %gr = extractvalue %struct.A %r, 0 <i>; yields i32</i>
4458 %gr1 = extractvalue %struct.A %r, 1 <i>; yields i8</i>
4459 %Z = call void @foo() noreturn <i>; indicates that %foo never returns normally</i>
4460 %ZZ = call zeroext i32 @bar() <i>; Return value is %zero extended</i>
4465 <!-- _______________________________________________________________________ -->
4466 <div class="doc_subsubsection">
4467 <a name="i_va_arg">'<tt>va_arg</tt>' Instruction</a>
4470 <div class="doc_text">
4475 <resultval> = va_arg <va_list*> <arglist>, <argty>
4480 <p>The '<tt>va_arg</tt>' instruction is used to access arguments passed through
4481 the "variable argument" area of a function call. It is used to implement the
4482 <tt>va_arg</tt> macro in C.</p>
4486 <p>This instruction takes a <tt>va_list*</tt> value and the type of
4487 the argument. It returns a value of the specified argument type and
4488 increments the <tt>va_list</tt> to point to the next argument. The
4489 actual type of <tt>va_list</tt> is target specific.</p>
4493 <p>The '<tt>va_arg</tt>' instruction loads an argument of the specified
4494 type from the specified <tt>va_list</tt> and causes the
4495 <tt>va_list</tt> to point to the next argument. For more information,
4496 see the variable argument handling <a href="#int_varargs">Intrinsic
4499 <p>It is legal for this instruction to be called in a function which does not
4500 take a variable number of arguments, for example, the <tt>vfprintf</tt>
4503 <p><tt>va_arg</tt> is an LLVM instruction instead of an <a
4504 href="#intrinsics">intrinsic function</a> because it takes a type as an
4509 <p>See the <a href="#int_varargs">variable argument processing</a> section.</p>
4511 <p>Note that the code generator does not yet fully support va_arg
4512 on many targets. Also, it does not currently support va_arg with
4513 aggregate types on any target.</p>
4517 <!-- *********************************************************************** -->
4518 <div class="doc_section"> <a name="intrinsics">Intrinsic Functions</a> </div>
4519 <!-- *********************************************************************** -->
4521 <div class="doc_text">
4523 <p>LLVM supports the notion of an "intrinsic function". These functions have
4524 well known names and semantics and are required to follow certain restrictions.
4525 Overall, these intrinsics represent an extension mechanism for the LLVM
4526 language that does not require changing all of the transformations in LLVM when
4527 adding to the language (or the bitcode reader/writer, the parser, etc...).</p>
4529 <p>Intrinsic function names must all start with an "<tt>llvm.</tt>" prefix. This
4530 prefix is reserved in LLVM for intrinsic names; thus, function names may not
4531 begin with this prefix. Intrinsic functions must always be external functions:
4532 you cannot define the body of intrinsic functions. Intrinsic functions may
4533 only be used in call or invoke instructions: it is illegal to take the address
4534 of an intrinsic function. Additionally, because intrinsic functions are part
4535 of the LLVM language, it is required if any are added that they be documented
4538 <p>Some intrinsic functions can be overloaded, i.e., the intrinsic represents
4539 a family of functions that perform the same operation but on different data
4540 types. Because LLVM can represent over 8 million different integer types,
4541 overloading is used commonly to allow an intrinsic function to operate on any
4542 integer type. One or more of the argument types or the result type can be
4543 overloaded to accept any integer type. Argument types may also be defined as
4544 exactly matching a previous argument's type or the result type. This allows an
4545 intrinsic function which accepts multiple arguments, but needs all of them to
4546 be of the same type, to only be overloaded with respect to a single argument or
4549 <p>Overloaded intrinsics will have the names of its overloaded argument types
4550 encoded into its function name, each preceded by a period. Only those types
4551 which are overloaded result in a name suffix. Arguments whose type is matched
4552 against another type do not. For example, the <tt>llvm.ctpop</tt> function can
4553 take an integer of any width and returns an integer of exactly the same integer
4554 width. This leads to a family of functions such as
4555 <tt>i8 @llvm.ctpop.i8(i8 %val)</tt> and <tt>i29 @llvm.ctpop.i29(i29 %val)</tt>.
4556 Only one type, the return type, is overloaded, and only one type suffix is
4557 required. Because the argument's type is matched against the return type, it
4558 does not require its own name suffix.</p>
4560 <p>To learn how to add an intrinsic function, please see the
4561 <a href="ExtendingLLVM.html">Extending LLVM Guide</a>.
4566 <!-- ======================================================================= -->
4567 <div class="doc_subsection">
4568 <a name="int_varargs">Variable Argument Handling Intrinsics</a>
4571 <div class="doc_text">
4573 <p>Variable argument support is defined in LLVM with the <a
4574 href="#i_va_arg"><tt>va_arg</tt></a> instruction and these three
4575 intrinsic functions. These functions are related to the similarly
4576 named macros defined in the <tt><stdarg.h></tt> header file.</p>
4578 <p>All of these functions operate on arguments that use a
4579 target-specific value type "<tt>va_list</tt>". The LLVM assembly
4580 language reference manual does not define what this type is, so all
4581 transformations should be prepared to handle these functions regardless of
4584 <p>This example shows how the <a href="#i_va_arg"><tt>va_arg</tt></a>
4585 instruction and the variable argument handling intrinsic functions are
4588 <div class="doc_code">
4590 define i32 @test(i32 %X, ...) {
4591 ; Initialize variable argument processing
4593 %ap2 = bitcast i8** %ap to i8*
4594 call void @llvm.va_start(i8* %ap2)
4596 ; Read a single integer argument
4597 %tmp = va_arg i8** %ap, i32
4599 ; Demonstrate usage of llvm.va_copy and llvm.va_end
4601 %aq2 = bitcast i8** %aq to i8*
4602 call void @llvm.va_copy(i8* %aq2, i8* %ap2)
4603 call void @llvm.va_end(i8* %aq2)
4605 ; Stop processing of arguments.
4606 call void @llvm.va_end(i8* %ap2)
4610 declare void @llvm.va_start(i8*)
4611 declare void @llvm.va_copy(i8*, i8*)
4612 declare void @llvm.va_end(i8*)
4618 <!-- _______________________________________________________________________ -->
4619 <div class="doc_subsubsection">
4620 <a name="int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a>
4624 <div class="doc_text">
4626 <pre> declare void %llvm.va_start(i8* <arglist>)<br></pre>
4628 <p>The '<tt>llvm.va_start</tt>' intrinsic initializes
4629 <tt>*<arglist></tt> for subsequent use by <tt><a
4630 href="#i_va_arg">va_arg</a></tt>.</p>
4634 <p>The argument is a pointer to a <tt>va_list</tt> element to initialize.</p>
4638 <p>The '<tt>llvm.va_start</tt>' intrinsic works just like the <tt>va_start</tt>
4639 macro available in C. In a target-dependent way, it initializes the
4640 <tt>va_list</tt> element to which the argument points, so that the next call to
4641 <tt>va_arg</tt> will produce the first variable argument passed to the function.
4642 Unlike the C <tt>va_start</tt> macro, this intrinsic does not need to know the
4643 last argument of the function as the compiler can figure that out.</p>
4647 <!-- _______________________________________________________________________ -->
4648 <div class="doc_subsubsection">
4649 <a name="int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a>
4652 <div class="doc_text">
4654 <pre> declare void @llvm.va_end(i8* <arglist>)<br></pre>
4657 <p>The '<tt>llvm.va_end</tt>' intrinsic destroys <tt>*<arglist></tt>,
4658 which has been initialized previously with <tt><a href="#int_va_start">llvm.va_start</a></tt>
4659 or <tt><a href="#i_va_copy">llvm.va_copy</a></tt>.</p>
4663 <p>The argument is a pointer to a <tt>va_list</tt> to destroy.</p>
4667 <p>The '<tt>llvm.va_end</tt>' intrinsic works just like the <tt>va_end</tt>
4668 macro available in C. In a target-dependent way, it destroys the
4669 <tt>va_list</tt> element to which the argument points. Calls to <a
4670 href="#int_va_start"><tt>llvm.va_start</tt></a> and <a href="#int_va_copy">
4671 <tt>llvm.va_copy</tt></a> must be matched exactly with calls to
4672 <tt>llvm.va_end</tt>.</p>
4676 <!-- _______________________________________________________________________ -->
4677 <div class="doc_subsubsection">
4678 <a name="int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a>
4681 <div class="doc_text">
4686 declare void @llvm.va_copy(i8* <destarglist>, i8* <srcarglist>)
4691 <p>The '<tt>llvm.va_copy</tt>' intrinsic copies the current argument position
4692 from the source argument list to the destination argument list.</p>
4696 <p>The first argument is a pointer to a <tt>va_list</tt> element to initialize.
4697 The second argument is a pointer to a <tt>va_list</tt> element to copy from.</p>
4702 <p>The '<tt>llvm.va_copy</tt>' intrinsic works just like the <tt>va_copy</tt>
4703 macro available in C. In a target-dependent way, it copies the source
4704 <tt>va_list</tt> element into the destination <tt>va_list</tt> element. This
4705 intrinsic is necessary because the <tt><a href="#int_va_start">
4706 llvm.va_start</a></tt> intrinsic may be arbitrarily complex and require, for
4707 example, memory allocation.</p>
4711 <!-- ======================================================================= -->
4712 <div class="doc_subsection">
4713 <a name="int_gc">Accurate Garbage Collection Intrinsics</a>
4716 <div class="doc_text">
4719 LLVM support for <a href="GarbageCollection.html">Accurate Garbage
4720 Collection</a> (GC) requires the implementation and generation of these
4722 These intrinsics allow identification of <a href="#int_gcroot">GC roots on the
4723 stack</a>, as well as garbage collector implementations that require <a
4724 href="#int_gcread">read</a> and <a href="#int_gcwrite">write</a> barriers.
4725 Front-ends for type-safe garbage collected languages should generate these
4726 intrinsics to make use of the LLVM garbage collectors. For more details, see <a
4727 href="GarbageCollection.html">Accurate Garbage Collection with LLVM</a>.
4730 <p>The garbage collection intrinsics only operate on objects in the generic
4731 address space (address space zero).</p>
4735 <!-- _______________________________________________________________________ -->
4736 <div class="doc_subsubsection">
4737 <a name="int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a>
4740 <div class="doc_text">
4745 declare void @llvm.gcroot(i8** %ptrloc, i8* %metadata)
4750 <p>The '<tt>llvm.gcroot</tt>' intrinsic declares the existence of a GC root to
4751 the code generator, and allows some metadata to be associated with it.</p>
4755 <p>The first argument specifies the address of a stack object that contains the
4756 root pointer. The second pointer (which must be either a constant or a global
4757 value address) contains the meta-data to be associated with the root.</p>
4761 <p>At runtime, a call to this intrinsic stores a null pointer into the "ptrloc"
4762 location. At compile-time, the code generator generates information to allow
4763 the runtime to find the pointer at GC safe points. The '<tt>llvm.gcroot</tt>'
4764 intrinsic may only be used in a function which <a href="#gc">specifies a GC
4770 <!-- _______________________________________________________________________ -->
4771 <div class="doc_subsubsection">
4772 <a name="int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a>
4775 <div class="doc_text">
4780 declare i8* @llvm.gcread(i8* %ObjPtr, i8** %Ptr)
4785 <p>The '<tt>llvm.gcread</tt>' intrinsic identifies reads of references from heap
4786 locations, allowing garbage collector implementations that require read
4791 <p>The second argument is the address to read from, which should be an address
4792 allocated from the garbage collector. The first object is a pointer to the
4793 start of the referenced object, if needed by the language runtime (otherwise
4798 <p>The '<tt>llvm.gcread</tt>' intrinsic has the same semantics as a load
4799 instruction, but may be replaced with substantially more complex code by the
4800 garbage collector runtime, as needed. The '<tt>llvm.gcread</tt>' intrinsic
4801 may only be used in a function which <a href="#gc">specifies a GC
4807 <!-- _______________________________________________________________________ -->
4808 <div class="doc_subsubsection">
4809 <a name="int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a>
4812 <div class="doc_text">
4817 declare void @llvm.gcwrite(i8* %P1, i8* %Obj, i8** %P2)
4822 <p>The '<tt>llvm.gcwrite</tt>' intrinsic identifies writes of references to heap
4823 locations, allowing garbage collector implementations that require write
4824 barriers (such as generational or reference counting collectors).</p>
4828 <p>The first argument is the reference to store, the second is the start of the
4829 object to store it to, and the third is the address of the field of Obj to
4830 store to. If the runtime does not require a pointer to the object, Obj may be
4835 <p>The '<tt>llvm.gcwrite</tt>' intrinsic has the same semantics as a store
4836 instruction, but may be replaced with substantially more complex code by the
4837 garbage collector runtime, as needed. The '<tt>llvm.gcwrite</tt>' intrinsic
4838 may only be used in a function which <a href="#gc">specifies a GC
4845 <!-- ======================================================================= -->
4846 <div class="doc_subsection">
4847 <a name="int_codegen">Code Generator Intrinsics</a>
4850 <div class="doc_text">
4852 These intrinsics are provided by LLVM to expose special features that may only
4853 be implemented with code generator support.
4858 <!-- _______________________________________________________________________ -->
4859 <div class="doc_subsubsection">
4860 <a name="int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a>
4863 <div class="doc_text">
4867 declare i8 *@llvm.returnaddress(i32 <level>)
4873 The '<tt>llvm.returnaddress</tt>' intrinsic attempts to compute a
4874 target-specific value indicating the return address of the current function
4875 or one of its callers.
4881 The argument to this intrinsic indicates which function to return the address
4882 for. Zero indicates the calling function, one indicates its caller, etc. The
4883 argument is <b>required</b> to be a constant integer value.
4889 The '<tt>llvm.returnaddress</tt>' intrinsic either returns a pointer indicating
4890 the return address of the specified call frame, or zero if it cannot be
4891 identified. The value returned by this intrinsic is likely to be incorrect or 0
4892 for arguments other than zero, so it should only be used for debugging purposes.
4896 Note that calling this intrinsic does not prevent function inlining or other
4897 aggressive transformations, so the value returned may not be that of the obvious
4898 source-language caller.
4903 <!-- _______________________________________________________________________ -->
4904 <div class="doc_subsubsection">
4905 <a name="int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a>
4908 <div class="doc_text">
4912 declare i8 *@llvm.frameaddress(i32 <level>)
4918 The '<tt>llvm.frameaddress</tt>' intrinsic attempts to return the
4919 target-specific frame pointer value for the specified stack frame.
4925 The argument to this intrinsic indicates which function to return the frame
4926 pointer for. Zero indicates the calling function, one indicates its caller,
4927 etc. The argument is <b>required</b> to be a constant integer value.
4933 The '<tt>llvm.frameaddress</tt>' intrinsic either returns a pointer indicating
4934 the frame address of the specified call frame, or zero if it cannot be
4935 identified. The value returned by this intrinsic is likely to be incorrect or 0
4936 for arguments other than zero, so it should only be used for debugging purposes.
4940 Note that calling this intrinsic does not prevent function inlining or other
4941 aggressive transformations, so the value returned may not be that of the obvious
4942 source-language caller.
4946 <!-- _______________________________________________________________________ -->
4947 <div class="doc_subsubsection">
4948 <a name="int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a>
4951 <div class="doc_text">
4955 declare i8 *@llvm.stacksave()
4961 The '<tt>llvm.stacksave</tt>' intrinsic is used to remember the current state of
4962 the function stack, for use with <a href="#int_stackrestore">
4963 <tt>llvm.stackrestore</tt></a>. This is useful for implementing language
4964 features like scoped automatic variable sized arrays in C99.
4970 This intrinsic returns a opaque pointer value that can be passed to <a
4971 href="#int_stackrestore"><tt>llvm.stackrestore</tt></a>. When an
4972 <tt>llvm.stackrestore</tt> intrinsic is executed with a value saved from
4973 <tt>llvm.stacksave</tt>, it effectively restores the state of the stack to the
4974 state it was in when the <tt>llvm.stacksave</tt> intrinsic executed. In
4975 practice, this pops any <a href="#i_alloca">alloca</a> blocks from the stack
4976 that were allocated after the <tt>llvm.stacksave</tt> was executed.
4981 <!-- _______________________________________________________________________ -->
4982 <div class="doc_subsubsection">
4983 <a name="int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a>
4986 <div class="doc_text">
4990 declare void @llvm.stackrestore(i8 * %ptr)
4996 The '<tt>llvm.stackrestore</tt>' intrinsic is used to restore the state of
4997 the function stack to the state it was in when the corresponding <a
4998 href="#int_stacksave"><tt>llvm.stacksave</tt></a> intrinsic executed. This is
4999 useful for implementing language features like scoped automatic variable sized
5006 See the description for <a href="#int_stacksave"><tt>llvm.stacksave</tt></a>.
5012 <!-- _______________________________________________________________________ -->
5013 <div class="doc_subsubsection">
5014 <a name="int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a>
5017 <div class="doc_text">
5021 declare void @llvm.prefetch(i8* <address>, i32 <rw>, i32 <locality>)
5028 The '<tt>llvm.prefetch</tt>' intrinsic is a hint to the code generator to insert
5029 a prefetch instruction if supported; otherwise, it is a noop. Prefetches have
5031 effect on the behavior of the program but can change its performance
5038 <tt>address</tt> is the address to be prefetched, <tt>rw</tt> is the specifier
5039 determining if the fetch should be for a read (0) or write (1), and
5040 <tt>locality</tt> is a temporal locality specifier ranging from (0) - no
5041 locality, to (3) - extremely local keep in cache. The <tt>rw</tt> and
5042 <tt>locality</tt> arguments must be constant integers.
5048 This intrinsic does not modify the behavior of the program. In particular,
5049 prefetches cannot trap and do not produce a value. On targets that support this
5050 intrinsic, the prefetch can provide hints to the processor cache for better
5056 <!-- _______________________________________________________________________ -->
5057 <div class="doc_subsubsection">
5058 <a name="int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a>
5061 <div class="doc_text">
5065 declare void @llvm.pcmarker(i32 <id>)
5072 The '<tt>llvm.pcmarker</tt>' intrinsic is a method to export a Program Counter
5074 code to simulators and other tools. The method is target specific, but it is
5075 expected that the marker will use exported symbols to transmit the PC of the
5077 The marker makes no guarantees that it will remain with any specific instruction
5078 after optimizations. It is possible that the presence of a marker will inhibit
5079 optimizations. The intended use is to be inserted after optimizations to allow
5080 correlations of simulation runs.
5086 <tt>id</tt> is a numerical id identifying the marker.
5092 This intrinsic does not modify the behavior of the program. Backends that do not
5093 support this intrinisic may ignore it.
5098 <!-- _______________________________________________________________________ -->
5099 <div class="doc_subsubsection">
5100 <a name="int_readcyclecounter">'<tt>llvm.readcyclecounter</tt>' Intrinsic</a>
5103 <div class="doc_text">
5107 declare i64 @llvm.readcyclecounter( )
5114 The '<tt>llvm.readcyclecounter</tt>' intrinsic provides access to the cycle
5115 counter register (or similar low latency, high accuracy clocks) on those targets
5116 that support it. On X86, it should map to RDTSC. On Alpha, it should map to RPCC.
5117 As the backing counters overflow quickly (on the order of 9 seconds on alpha), this
5118 should only be used for small timings.
5124 When directly supported, reading the cycle counter should not modify any memory.
5125 Implementations are allowed to either return a application specific value or a
5126 system wide value. On backends without support, this is lowered to a constant 0.
5131 <!-- ======================================================================= -->
5132 <div class="doc_subsection">
5133 <a name="int_libc">Standard C Library Intrinsics</a>
5136 <div class="doc_text">
5138 LLVM provides intrinsics for a few important standard C library functions.
5139 These intrinsics allow source-language front-ends to pass information about the
5140 alignment of the pointer arguments to the code generator, providing opportunity
5141 for more efficient code generation.
5146 <!-- _______________________________________________________________________ -->
5147 <div class="doc_subsubsection">
5148 <a name="int_memcpy">'<tt>llvm.memcpy</tt>' Intrinsic</a>
5151 <div class="doc_text">
5154 <p>This is an overloaded intrinsic. You can use llvm.memcpy on any integer bit
5155 width. Not all targets support all bit widths however.</p>
5157 declare void @llvm.memcpy.i8(i8 * <dest>, i8 * <src>,
5158 i8 <len>, i32 <align>)
5159 declare void @llvm.memcpy.i16(i8 * <dest>, i8 * <src>,
5160 i16 <len>, i32 <align>)
5161 declare void @llvm.memcpy.i32(i8 * <dest>, i8 * <src>,
5162 i32 <len>, i32 <align>)
5163 declare void @llvm.memcpy.i64(i8 * <dest>, i8 * <src>,
5164 i64 <len>, i32 <align>)
5170 The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the source
5171 location to the destination location.
5175 Note that, unlike the standard libc function, the <tt>llvm.memcpy.*</tt>
5176 intrinsics do not return a value, and takes an extra alignment argument.
5182 The first argument is a pointer to the destination, the second is a pointer to
5183 the source. The third argument is an integer argument
5184 specifying the number of bytes to copy, and the fourth argument is the alignment
5185 of the source and destination locations.
5189 If the call to this intrinisic has an alignment value that is not 0 or 1, then
5190 the caller guarantees that both the source and destination pointers are aligned
5197 The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the source
5198 location to the destination location, which are not allowed to overlap. It
5199 copies "len" bytes of memory over. If the argument is known to be aligned to
5200 some boundary, this can be specified as the fourth argument, otherwise it should
5206 <!-- _______________________________________________________________________ -->
5207 <div class="doc_subsubsection">
5208 <a name="int_memmove">'<tt>llvm.memmove</tt>' Intrinsic</a>
5211 <div class="doc_text">
5214 <p>This is an overloaded intrinsic. You can use llvm.memmove on any integer bit
5215 width. Not all targets support all bit widths however.</p>
5217 declare void @llvm.memmove.i8(i8 * <dest>, i8 * <src>,
5218 i8 <len>, i32 <align>)
5219 declare void @llvm.memmove.i16(i8 * <dest>, i8 * <src>,
5220 i16 <len>, i32 <align>)
5221 declare void @llvm.memmove.i32(i8 * <dest>, i8 * <src>,
5222 i32 <len>, i32 <align>)
5223 declare void @llvm.memmove.i64(i8 * <dest>, i8 * <src>,
5224 i64 <len>, i32 <align>)
5230 The '<tt>llvm.memmove.*</tt>' intrinsics move a block of memory from the source
5231 location to the destination location. It is similar to the
5232 '<tt>llvm.memcpy</tt>' intrinsic but allows the two memory locations to overlap.
5236 Note that, unlike the standard libc function, the <tt>llvm.memmove.*</tt>
5237 intrinsics do not return a value, and takes an extra alignment argument.
5243 The first argument is a pointer to the destination, the second is a pointer to
5244 the source. The third argument is an integer argument
5245 specifying the number of bytes to copy, and the fourth argument is the alignment
5246 of the source and destination locations.
5250 If the call to this intrinisic has an alignment value that is not 0 or 1, then
5251 the caller guarantees that the source and destination pointers are aligned to
5258 The '<tt>llvm.memmove.*</tt>' intrinsics copy a block of memory from the source
5259 location to the destination location, which may overlap. It
5260 copies "len" bytes of memory over. If the argument is known to be aligned to
5261 some boundary, this can be specified as the fourth argument, otherwise it should
5267 <!-- _______________________________________________________________________ -->
5268 <div class="doc_subsubsection">
5269 <a name="int_memset">'<tt>llvm.memset.*</tt>' Intrinsics</a>
5272 <div class="doc_text">
5275 <p>This is an overloaded intrinsic. You can use llvm.memset on any integer bit
5276 width. Not all targets support all bit widths however.</p>
5278 declare void @llvm.memset.i8(i8 * <dest>, i8 <val>,
5279 i8 <len>, i32 <align>)
5280 declare void @llvm.memset.i16(i8 * <dest>, i8 <val>,
5281 i16 <len>, i32 <align>)
5282 declare void @llvm.memset.i32(i8 * <dest>, i8 <val>,
5283 i32 <len>, i32 <align>)
5284 declare void @llvm.memset.i64(i8 * <dest>, i8 <val>,
5285 i64 <len>, i32 <align>)
5291 The '<tt>llvm.memset.*</tt>' intrinsics fill a block of memory with a particular
5296 Note that, unlike the standard libc function, the <tt>llvm.memset</tt> intrinsic
5297 does not return a value, and takes an extra alignment argument.
5303 The first argument is a pointer to the destination to fill, the second is the
5304 byte value to fill it with, the third argument is an integer
5305 argument specifying the number of bytes to fill, and the fourth argument is the
5306 known alignment of destination location.
5310 If the call to this intrinisic has an alignment value that is not 0 or 1, then
5311 the caller guarantees that the destination pointer is aligned to that boundary.
5317 The '<tt>llvm.memset.*</tt>' intrinsics fill "len" bytes of memory starting at
5319 destination location. If the argument is known to be aligned to some boundary,
5320 this can be specified as the fourth argument, otherwise it should be set to 0 or
5326 <!-- _______________________________________________________________________ -->
5327 <div class="doc_subsubsection">
5328 <a name="int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a>
5331 <div class="doc_text">
5334 <p>This is an overloaded intrinsic. You can use <tt>llvm.sqrt</tt> on any
5335 floating point or vector of floating point type. Not all targets support all
5338 declare float @llvm.sqrt.f32(float %Val)
5339 declare double @llvm.sqrt.f64(double %Val)
5340 declare x86_fp80 @llvm.sqrt.f80(x86_fp80 %Val)
5341 declare fp128 @llvm.sqrt.f128(fp128 %Val)
5342 declare ppc_fp128 @llvm.sqrt.ppcf128(ppc_fp128 %Val)
5348 The '<tt>llvm.sqrt</tt>' intrinsics return the sqrt of the specified operand,
5349 returning the same value as the libm '<tt>sqrt</tt>' functions would. Unlike
5350 <tt>sqrt</tt> in libm, however, <tt>llvm.sqrt</tt> has undefined behavior for
5351 negative numbers other than -0.0 (which allows for better optimization, because
5352 there is no need to worry about errno being set). <tt>llvm.sqrt(-0.0)</tt> is
5353 defined to return -0.0 like IEEE sqrt.
5359 The argument and return value are floating point numbers of the same type.
5365 This function returns the sqrt of the specified operand if it is a nonnegative
5366 floating point number.
5370 <!-- _______________________________________________________________________ -->
5371 <div class="doc_subsubsection">
5372 <a name="int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a>
5375 <div class="doc_text">
5378 <p>This is an overloaded intrinsic. You can use <tt>llvm.powi</tt> on any
5379 floating point or vector of floating point type. Not all targets support all
5382 declare float @llvm.powi.f32(float %Val, i32 %power)
5383 declare double @llvm.powi.f64(double %Val, i32 %power)
5384 declare x86_fp80 @llvm.powi.f80(x86_fp80 %Val, i32 %power)
5385 declare fp128 @llvm.powi.f128(fp128 %Val, i32 %power)
5386 declare ppc_fp128 @llvm.powi.ppcf128(ppc_fp128 %Val, i32 %power)
5392 The '<tt>llvm.powi.*</tt>' intrinsics return the first operand raised to the
5393 specified (positive or negative) power. The order of evaluation of
5394 multiplications is not defined. When a vector of floating point type is
5395 used, the second argument remains a scalar integer value.
5401 The second argument is an integer power, and the first is a value to raise to
5408 This function returns the first value raised to the second power with an
5409 unspecified sequence of rounding operations.</p>
5412 <!-- _______________________________________________________________________ -->
5413 <div class="doc_subsubsection">
5414 <a name="int_sin">'<tt>llvm.sin.*</tt>' Intrinsic</a>
5417 <div class="doc_text">
5420 <p>This is an overloaded intrinsic. You can use <tt>llvm.sin</tt> on any
5421 floating point or vector of floating point type. Not all targets support all
5424 declare float @llvm.sin.f32(float %Val)
5425 declare double @llvm.sin.f64(double %Val)
5426 declare x86_fp80 @llvm.sin.f80(x86_fp80 %Val)
5427 declare fp128 @llvm.sin.f128(fp128 %Val)
5428 declare ppc_fp128 @llvm.sin.ppcf128(ppc_fp128 %Val)
5434 The '<tt>llvm.sin.*</tt>' intrinsics return the sine of the operand.
5440 The argument and return value are floating point numbers of the same type.
5446 This function returns the sine of the specified operand, returning the
5447 same values as the libm <tt>sin</tt> functions would, and handles error
5448 conditions in the same way.</p>
5451 <!-- _______________________________________________________________________ -->
5452 <div class="doc_subsubsection">
5453 <a name="int_cos">'<tt>llvm.cos.*</tt>' Intrinsic</a>
5456 <div class="doc_text">
5459 <p>This is an overloaded intrinsic. You can use <tt>llvm.cos</tt> on any
5460 floating point or vector of floating point type. Not all targets support all
5463 declare float @llvm.cos.f32(float %Val)
5464 declare double @llvm.cos.f64(double %Val)
5465 declare x86_fp80 @llvm.cos.f80(x86_fp80 %Val)
5466 declare fp128 @llvm.cos.f128(fp128 %Val)
5467 declare ppc_fp128 @llvm.cos.ppcf128(ppc_fp128 %Val)
5473 The '<tt>llvm.cos.*</tt>' intrinsics return the cosine of the operand.
5479 The argument and return value are floating point numbers of the same type.
5485 This function returns the cosine of the specified operand, returning the
5486 same values as the libm <tt>cos</tt> functions would, and handles error
5487 conditions in the same way.</p>
5490 <!-- _______________________________________________________________________ -->
5491 <div class="doc_subsubsection">
5492 <a name="int_pow">'<tt>llvm.pow.*</tt>' Intrinsic</a>
5495 <div class="doc_text">
5498 <p>This is an overloaded intrinsic. You can use <tt>llvm.pow</tt> on any
5499 floating point or vector of floating point type. Not all targets support all
5502 declare float @llvm.pow.f32(float %Val, float %Power)
5503 declare double @llvm.pow.f64(double %Val, double %Power)
5504 declare x86_fp80 @llvm.pow.f80(x86_fp80 %Val, x86_fp80 %Power)
5505 declare fp128 @llvm.pow.f128(fp128 %Val, fp128 %Power)
5506 declare ppc_fp128 @llvm.pow.ppcf128(ppc_fp128 %Val, ppc_fp128 Power)
5512 The '<tt>llvm.pow.*</tt>' intrinsics return the first operand raised to the
5513 specified (positive or negative) power.
5519 The second argument is a floating point power, and the first is a value to
5520 raise to that power.
5526 This function returns the first value raised to the second power,
5528 same values as the libm <tt>pow</tt> functions would, and handles error
5529 conditions in the same way.</p>
5533 <!-- ======================================================================= -->
5534 <div class="doc_subsection">
5535 <a name="int_manip">Bit Manipulation Intrinsics</a>
5538 <div class="doc_text">
5540 LLVM provides intrinsics for a few important bit manipulation operations.
5541 These allow efficient code generation for some algorithms.
5546 <!-- _______________________________________________________________________ -->
5547 <div class="doc_subsubsection">
5548 <a name="int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a>
5551 <div class="doc_text">
5554 <p>This is an overloaded intrinsic function. You can use bswap on any integer
5555 type that is an even number of bytes (i.e. BitWidth % 16 == 0).</p>
5557 declare i16 @llvm.bswap.i16(i16 <id>)
5558 declare i32 @llvm.bswap.i32(i32 <id>)
5559 declare i64 @llvm.bswap.i64(i64 <id>)
5565 The '<tt>llvm.bswap</tt>' family of intrinsics is used to byte swap integer
5566 values with an even number of bytes (positive multiple of 16 bits). These are
5567 useful for performing operations on data that is not in the target's native
5574 The <tt>llvm.bswap.i16</tt> intrinsic returns an i16 value that has the high
5575 and low byte of the input i16 swapped. Similarly, the <tt>llvm.bswap.i32</tt>
5576 intrinsic returns an i32 value that has the four bytes of the input i32
5577 swapped, so that if the input bytes are numbered 0, 1, 2, 3 then the returned
5578 i32 will have its bytes in 3, 2, 1, 0 order. The <tt>llvm.bswap.i48</tt>,
5579 <tt>llvm.bswap.i64</tt> and other intrinsics extend this concept to
5580 additional even-byte lengths (6 bytes, 8 bytes and more, respectively).
5585 <!-- _______________________________________________________________________ -->
5586 <div class="doc_subsubsection">
5587 <a name="int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic</a>
5590 <div class="doc_text">
5593 <p>This is an overloaded intrinsic. You can use llvm.ctpop on any integer bit
5594 width. Not all targets support all bit widths however.</p>
5596 declare i8 @llvm.ctpop.i8 (i8 <src>)
5597 declare i16 @llvm.ctpop.i16(i16 <src>)
5598 declare i32 @llvm.ctpop.i32(i32 <src>)
5599 declare i64 @llvm.ctpop.i64(i64 <src>)
5600 declare i256 @llvm.ctpop.i256(i256 <src>)
5606 The '<tt>llvm.ctpop</tt>' family of intrinsics counts the number of bits set in a
5613 The only argument is the value to be counted. The argument may be of any
5614 integer type. The return type must match the argument type.
5620 The '<tt>llvm.ctpop</tt>' intrinsic counts the 1's in a variable.
5624 <!-- _______________________________________________________________________ -->
5625 <div class="doc_subsubsection">
5626 <a name="int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic</a>
5629 <div class="doc_text">
5632 <p>This is an overloaded intrinsic. You can use <tt>llvm.ctlz</tt> on any
5633 integer bit width. Not all targets support all bit widths however.</p>
5635 declare i8 @llvm.ctlz.i8 (i8 <src>)
5636 declare i16 @llvm.ctlz.i16(i16 <src>)
5637 declare i32 @llvm.ctlz.i32(i32 <src>)
5638 declare i64 @llvm.ctlz.i64(i64 <src>)
5639 declare i256 @llvm.ctlz.i256(i256 <src>)
5645 The '<tt>llvm.ctlz</tt>' family of intrinsic functions counts the number of
5646 leading zeros in a variable.
5652 The only argument is the value to be counted. The argument may be of any
5653 integer type. The return type must match the argument type.
5659 The '<tt>llvm.ctlz</tt>' intrinsic counts the leading (most significant) zeros
5660 in a variable. If the src == 0 then the result is the size in bits of the type
5661 of src. For example, <tt>llvm.ctlz(i32 2) = 30</tt>.
5667 <!-- _______________________________________________________________________ -->
5668 <div class="doc_subsubsection">
5669 <a name="int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic</a>
5672 <div class="doc_text">
5675 <p>This is an overloaded intrinsic. You can use <tt>llvm.cttz</tt> on any
5676 integer bit width. Not all targets support all bit widths however.</p>
5678 declare i8 @llvm.cttz.i8 (i8 <src>)
5679 declare i16 @llvm.cttz.i16(i16 <src>)
5680 declare i32 @llvm.cttz.i32(i32 <src>)
5681 declare i64 @llvm.cttz.i64(i64 <src>)
5682 declare i256 @llvm.cttz.i256(i256 <src>)
5688 The '<tt>llvm.cttz</tt>' family of intrinsic functions counts the number of
5695 The only argument is the value to be counted. The argument may be of any
5696 integer type. The return type must match the argument type.
5702 The '<tt>llvm.cttz</tt>' intrinsic counts the trailing (least significant) zeros
5703 in a variable. If the src == 0 then the result is the size in bits of the type
5704 of src. For example, <tt>llvm.cttz(2) = 1</tt>.
5708 <!-- _______________________________________________________________________ -->
5709 <div class="doc_subsubsection">
5710 <a name="int_part_select">'<tt>llvm.part.select.*</tt>' Intrinsic</a>
5713 <div class="doc_text">
5716 <p>This is an overloaded intrinsic. You can use <tt>llvm.part.select</tt>
5717 on any integer bit width.</p>
5719 declare i17 @llvm.part.select.i17 (i17 %val, i32 %loBit, i32 %hiBit)
5720 declare i29 @llvm.part.select.i29 (i29 %val, i32 %loBit, i32 %hiBit)
5724 <p>The '<tt>llvm.part.select</tt>' family of intrinsic functions selects a
5725 range of bits from an integer value and returns them in the same bit width as
5726 the original value.</p>
5729 <p>The first argument, <tt>%val</tt> and the result may be integer types of
5730 any bit width but they must have the same bit width. The second and third
5731 arguments must be <tt>i32</tt> type since they specify only a bit index.</p>
5734 <p>The operation of the '<tt>llvm.part.select</tt>' intrinsic has two modes
5735 of operation: forwards and reverse. If <tt>%loBit</tt> is greater than
5736 <tt>%hiBits</tt> then the intrinsic operates in reverse mode. Otherwise it
5737 operates in forward mode.</p>
5738 <p>In forward mode, this intrinsic is the equivalent of shifting <tt>%val</tt>
5739 right by <tt>%loBit</tt> bits and then ANDing it with a mask with
5740 only the <tt>%hiBit - %loBit</tt> bits set, as follows:</p>
5742 <li>The <tt>%val</tt> is shifted right (LSHR) by the number of bits specified
5743 by <tt>%loBits</tt>. This normalizes the value to the low order bits.</li>
5744 <li>The <tt>%loBits</tt> value is subtracted from the <tt>%hiBits</tt> value
5745 to determine the number of bits to retain.</li>
5746 <li>A mask of the retained bits is created by shifting a -1 value.</li>
5747 <li>The mask is ANDed with <tt>%val</tt> to produce the result.</li>
5749 <p>In reverse mode, a similar computation is made except that the bits are
5750 returned in the reverse order. So, for example, if <tt>X</tt> has the value
5751 <tt>i16 0x0ACF (101011001111)</tt> and we apply
5752 <tt>part.select(i16 X, 8, 3)</tt> to it, we get back the value
5753 <tt>i16 0x0026 (000000100110)</tt>.</p>
5756 <div class="doc_subsubsection">
5757 <a name="int_part_set">'<tt>llvm.part.set.*</tt>' Intrinsic</a>
5760 <div class="doc_text">
5763 <p>This is an overloaded intrinsic. You can use <tt>llvm.part.set</tt>
5764 on any integer bit width.</p>
5766 declare i17 @llvm.part.set.i17.i9 (i17 %val, i9 %repl, i32 %lo, i32 %hi)
5767 declare i29 @llvm.part.set.i29.i9 (i29 %val, i9 %repl, i32 %lo, i32 %hi)
5771 <p>The '<tt>llvm.part.set</tt>' family of intrinsic functions replaces a range
5772 of bits in an integer value with another integer value. It returns the integer
5773 with the replaced bits.</p>
5776 <p>The first argument, <tt>%val</tt> and the result may be integer types of
5777 any bit width but they must have the same bit width. <tt>%val</tt> is the value
5778 whose bits will be replaced. The second argument, <tt>%repl</tt> may be an
5779 integer of any bit width. The third and fourth arguments must be <tt>i32</tt>
5780 type since they specify only a bit index.</p>
5783 <p>The operation of the '<tt>llvm.part.set</tt>' intrinsic has two modes
5784 of operation: forwards and reverse. If <tt>%lo</tt> is greater than
5785 <tt>%hi</tt> then the intrinsic operates in reverse mode. Otherwise it
5786 operates in forward mode.</p>
5787 <p>For both modes, the <tt>%repl</tt> value is prepared for use by either
5788 truncating it down to the size of the replacement area or zero extending it
5789 up to that size.</p>
5790 <p>In forward mode, the bits between <tt>%lo</tt> and <tt>%hi</tt> (inclusive)
5791 are replaced with corresponding bits from <tt>%repl</tt>. That is the 0th bit
5792 in <tt>%repl</tt> replaces the <tt>%lo</tt>th bit in <tt>%val</tt> and etc. up
5793 to the <tt>%hi</tt>th bit.</p>
5794 <p>In reverse mode, a similar computation is made except that the bits are
5795 reversed. That is, the <tt>0</tt>th bit in <tt>%repl</tt> replaces the
5796 <tt>%hi</tt> bit in <tt>%val</tt> and etc. down to the <tt>%lo</tt>th bit.</p>
5799 llvm.part.set(0xFFFF, 0, 4, 7) -> 0xFF0F
5800 llvm.part.set(0xFFFF, 0, 7, 4) -> 0xFF0F
5801 llvm.part.set(0xFFFF, 1, 7, 4) -> 0xFF8F
5802 llvm.part.set(0xFFFF, F, 8, 3) -> 0xFFE7
5803 llvm.part.set(0xFFFF, 0, 3, 8) -> 0xFE07
5807 <!-- ======================================================================= -->
5808 <div class="doc_subsection">
5809 <a name="int_debugger">Debugger Intrinsics</a>
5812 <div class="doc_text">
5814 The LLVM debugger intrinsics (which all start with <tt>llvm.dbg.</tt> prefix),
5815 are described in the <a
5816 href="SourceLevelDebugging.html#format_common_intrinsics">LLVM Source Level
5817 Debugging</a> document.
5822 <!-- ======================================================================= -->
5823 <div class="doc_subsection">
5824 <a name="int_eh">Exception Handling Intrinsics</a>
5827 <div class="doc_text">
5828 <p> The LLVM exception handling intrinsics (which all start with
5829 <tt>llvm.eh.</tt> prefix), are described in the <a
5830 href="ExceptionHandling.html#format_common_intrinsics">LLVM Exception
5831 Handling</a> document. </p>
5834 <!-- ======================================================================= -->
5835 <div class="doc_subsection">
5836 <a name="int_trampoline">Trampoline Intrinsic</a>
5839 <div class="doc_text">
5841 This intrinsic makes it possible to excise one parameter, marked with
5842 the <tt>nest</tt> attribute, from a function. The result is a callable
5843 function pointer lacking the nest parameter - the caller does not need
5844 to provide a value for it. Instead, the value to use is stored in
5845 advance in a "trampoline", a block of memory usually allocated
5846 on the stack, which also contains code to splice the nest value into the
5847 argument list. This is used to implement the GCC nested function address
5851 For example, if the function is
5852 <tt>i32 f(i8* nest %c, i32 %x, i32 %y)</tt> then the resulting function
5853 pointer has signature <tt>i32 (i32, i32)*</tt>. It can be created as follows:</p>
5855 %tramp = alloca [10 x i8], align 4 ; size and alignment only correct for X86
5856 %tramp1 = getelementptr [10 x i8]* %tramp, i32 0, i32 0
5857 %p = call i8* @llvm.init.trampoline( i8* %tramp1, i8* bitcast (i32 (i8* nest , i32, i32)* @f to i8*), i8* %nval )
5858 %fp = bitcast i8* %p to i32 (i32, i32)*
5860 <p>The call <tt>%val = call i32 %fp( i32 %x, i32 %y )</tt> is then equivalent
5861 to <tt>%val = call i32 %f( i8* %nval, i32 %x, i32 %y )</tt>.</p>
5864 <!-- _______________________________________________________________________ -->
5865 <div class="doc_subsubsection">
5866 <a name="int_it">'<tt>llvm.init.trampoline</tt>' Intrinsic</a>
5868 <div class="doc_text">
5871 declare i8* @llvm.init.trampoline(i8* <tramp>, i8* <func>, i8* <nval>)
5875 This fills the memory pointed to by <tt>tramp</tt> with code
5876 and returns a function pointer suitable for executing it.
5880 The <tt>llvm.init.trampoline</tt> intrinsic takes three arguments, all
5881 pointers. The <tt>tramp</tt> argument must point to a sufficiently large
5882 and sufficiently aligned block of memory; this memory is written to by the
5883 intrinsic. Note that the size and the alignment are target-specific - LLVM
5884 currently provides no portable way of determining them, so a front-end that
5885 generates this intrinsic needs to have some target-specific knowledge.
5886 The <tt>func</tt> argument must hold a function bitcast to an <tt>i8*</tt>.
5890 The block of memory pointed to by <tt>tramp</tt> is filled with target
5891 dependent code, turning it into a function. A pointer to this function is
5892 returned, but needs to be bitcast to an
5893 <a href="#int_trampoline">appropriate function pointer type</a>
5894 before being called. The new function's signature is the same as that of
5895 <tt>func</tt> with any arguments marked with the <tt>nest</tt> attribute
5896 removed. At most one such <tt>nest</tt> argument is allowed, and it must be
5897 of pointer type. Calling the new function is equivalent to calling
5898 <tt>func</tt> with the same argument list, but with <tt>nval</tt> used for the
5899 missing <tt>nest</tt> argument. If, after calling
5900 <tt>llvm.init.trampoline</tt>, the memory pointed to by <tt>tramp</tt> is
5901 modified, then the effect of any later call to the returned function pointer is
5906 <!-- ======================================================================= -->
5907 <div class="doc_subsection">
5908 <a name="int_atomics">Atomic Operations and Synchronization Intrinsics</a>
5911 <div class="doc_text">
5913 These intrinsic functions expand the "universal IR" of LLVM to represent
5914 hardware constructs for atomic operations and memory synchronization. This
5915 provides an interface to the hardware, not an interface to the programmer. It
5916 is aimed at a low enough level to allow any programming models or APIs
5917 (Application Programming Interfaces) which
5918 need atomic behaviors to map cleanly onto it. It is also modeled primarily on
5919 hardware behavior. Just as hardware provides a "universal IR" for source
5920 languages, it also provides a starting point for developing a "universal"
5921 atomic operation and synchronization IR.
5924 These do <em>not</em> form an API such as high-level threading libraries,
5925 software transaction memory systems, atomic primitives, and intrinsic
5926 functions as found in BSD, GNU libc, atomic_ops, APR, and other system and
5927 application libraries. The hardware interface provided by LLVM should allow
5928 a clean implementation of all of these APIs and parallel programming models.
5929 No one model or paradigm should be selected above others unless the hardware
5930 itself ubiquitously does so.
5935 <!-- _______________________________________________________________________ -->
5936 <div class="doc_subsubsection">
5937 <a name="int_memory_barrier">'<tt>llvm.memory.barrier</tt>' Intrinsic</a>
5939 <div class="doc_text">
5942 declare void @llvm.memory.barrier( i1 <ll>, i1 <ls>, i1 <sl>, i1 <ss>,
5948 The <tt>llvm.memory.barrier</tt> intrinsic guarantees ordering between
5949 specific pairs of memory access types.
5953 The <tt>llvm.memory.barrier</tt> intrinsic requires five boolean arguments.
5954 The first four arguments enables a specific barrier as listed below. The fith
5955 argument specifies that the barrier applies to io or device or uncached memory.
5959 <li><tt>ll</tt>: load-load barrier</li>
5960 <li><tt>ls</tt>: load-store barrier</li>
5961 <li><tt>sl</tt>: store-load barrier</li>
5962 <li><tt>ss</tt>: store-store barrier</li>
5963 <li><tt>device</tt>: barrier applies to device and uncached memory also.</li>
5967 This intrinsic causes the system to enforce some ordering constraints upon
5968 the loads and stores of the program. This barrier does not indicate
5969 <em>when</em> any events will occur, it only enforces an <em>order</em> in
5970 which they occur. For any of the specified pairs of load and store operations
5971 (f.ex. load-load, or store-load), all of the first operations preceding the
5972 barrier will complete before any of the second operations succeeding the
5973 barrier begin. Specifically the semantics for each pairing is as follows:
5976 <li><tt>ll</tt>: All loads before the barrier must complete before any load
5977 after the barrier begins.</li>
5979 <li><tt>ls</tt>: All loads before the barrier must complete before any
5980 store after the barrier begins.</li>
5981 <li><tt>ss</tt>: All stores before the barrier must complete before any
5982 store after the barrier begins.</li>
5983 <li><tt>sl</tt>: All stores before the barrier must complete before any
5984 load after the barrier begins.</li>
5987 These semantics are applied with a logical "and" behavior when more than one
5988 is enabled in a single memory barrier intrinsic.
5991 Backends may implement stronger barriers than those requested when they do not
5992 support as fine grained a barrier as requested. Some architectures do not
5993 need all types of barriers and on such architectures, these become noops.
6000 %result1 = load i32* %ptr <i>; yields {i32}:result1 = 4</i>
6001 call void @llvm.memory.barrier( i1 false, i1 true, i1 false, i1 false )
6002 <i>; guarantee the above finishes</i>
6003 store i32 8, %ptr <i>; before this begins</i>
6007 <!-- _______________________________________________________________________ -->
6008 <div class="doc_subsubsection">
6009 <a name="int_atomic_cmp_swap">'<tt>llvm.atomic.cmp.swap.*</tt>' Intrinsic</a>
6011 <div class="doc_text">
6014 This is an overloaded intrinsic. You can use <tt>llvm.atomic.cmp.swap</tt> on
6015 any integer bit width and for different address spaces. Not all targets
6016 support all bit widths however.</p>
6019 declare i8 @llvm.atomic.cmp.swap.i8.p0i8( i8* <ptr>, i8 <cmp>, i8 <val> )
6020 declare i16 @llvm.atomic.cmp.swap.i16.p0i16( i16* <ptr>, i16 <cmp>, i16 <val> )
6021 declare i32 @llvm.atomic.cmp.swap.i32.p0i32( i32* <ptr>, i32 <cmp>, i32 <val> )
6022 declare i64 @llvm.atomic.cmp.swap.i64.p0i64( i64* <ptr>, i64 <cmp>, i64 <val> )
6027 This loads a value in memory and compares it to a given value. If they are
6028 equal, it stores a new value into the memory.
6032 The <tt>llvm.atomic.cmp.swap</tt> intrinsic takes three arguments. The result as
6033 well as both <tt>cmp</tt> and <tt>val</tt> must be integer values with the
6034 same bit width. The <tt>ptr</tt> argument must be a pointer to a value of
6035 this integer type. While any bit width integer may be used, targets may only
6036 lower representations they support in hardware.
6041 This entire intrinsic must be executed atomically. It first loads the value
6042 in memory pointed to by <tt>ptr</tt> and compares it with the value
6043 <tt>cmp</tt>. If they are equal, <tt>val</tt> is stored into the memory. The
6044 loaded value is yielded in all cases. This provides the equivalent of an
6045 atomic compare-and-swap operation within the SSA framework.
6053 %val1 = add i32 4, 4
6054 %result1 = call i32 @llvm.atomic.cmp.swap.i32.p0i32( i32* %ptr, i32 4, %val1 )
6055 <i>; yields {i32}:result1 = 4</i>
6056 %stored1 = icmp eq i32 %result1, 4 <i>; yields {i1}:stored1 = true</i>
6057 %memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 8</i>
6059 %val2 = add i32 1, 1
6060 %result2 = call i32 @llvm.atomic.cmp.swap.i32.p0i32( i32* %ptr, i32 5, %val2 )
6061 <i>; yields {i32}:result2 = 8</i>
6062 %stored2 = icmp eq i32 %result2, 5 <i>; yields {i1}:stored2 = false</i>
6064 %memval2 = load i32* %ptr <i>; yields {i32}:memval2 = 8</i>
6068 <!-- _______________________________________________________________________ -->
6069 <div class="doc_subsubsection">
6070 <a name="int_atomic_swap">'<tt>llvm.atomic.swap.*</tt>' Intrinsic</a>
6072 <div class="doc_text">
6076 This is an overloaded intrinsic. You can use <tt>llvm.atomic.swap</tt> on any
6077 integer bit width. Not all targets support all bit widths however.</p>
6079 declare i8 @llvm.atomic.swap.i8.p0i8( i8* <ptr>, i8 <val> )
6080 declare i16 @llvm.atomic.swap.i16.p0i16( i16* <ptr>, i16 <val> )
6081 declare i32 @llvm.atomic.swap.i32.p0i32( i32* <ptr>, i32 <val> )
6082 declare i64 @llvm.atomic.swap.i64.p0i64( i64* <ptr>, i64 <val> )
6087 This intrinsic loads the value stored in memory at <tt>ptr</tt> and yields
6088 the value from memory. It then stores the value in <tt>val</tt> in the memory
6094 The <tt>llvm.atomic.swap</tt> intrinsic takes two arguments. Both the
6095 <tt>val</tt> argument and the result must be integers of the same bit width.
6096 The first argument, <tt>ptr</tt>, must be a pointer to a value of this
6097 integer type. The targets may only lower integer representations they
6102 This intrinsic loads the value pointed to by <tt>ptr</tt>, yields it, and
6103 stores <tt>val</tt> back into <tt>ptr</tt> atomically. This provides the
6104 equivalent of an atomic swap operation within the SSA framework.
6112 %val1 = add i32 4, 4
6113 %result1 = call i32 @llvm.atomic.swap.i32.p0i32( i32* %ptr, i32 %val1 )
6114 <i>; yields {i32}:result1 = 4</i>
6115 %stored1 = icmp eq i32 %result1, 4 <i>; yields {i1}:stored1 = true</i>
6116 %memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 8</i>
6118 %val2 = add i32 1, 1
6119 %result2 = call i32 @llvm.atomic.swap.i32.p0i32( i32* %ptr, i32 %val2 )
6120 <i>; yields {i32}:result2 = 8</i>
6122 %stored2 = icmp eq i32 %result2, 8 <i>; yields {i1}:stored2 = true</i>
6123 %memval2 = load i32* %ptr <i>; yields {i32}:memval2 = 2</i>
6127 <!-- _______________________________________________________________________ -->
6128 <div class="doc_subsubsection">
6129 <a name="int_atomic_load_add">'<tt>llvm.atomic.load.add.*</tt>' Intrinsic</a>
6132 <div class="doc_text">
6135 This is an overloaded intrinsic. You can use <tt>llvm.atomic.load.add</tt> on any
6136 integer bit width. Not all targets support all bit widths however.</p>
6138 declare i8 @llvm.atomic.load.add.i8..p0i8( i8* <ptr>, i8 <delta> )
6139 declare i16 @llvm.atomic.load.add.i16..p0i16( i16* <ptr>, i16 <delta> )
6140 declare i32 @llvm.atomic.load.add.i32..p0i32( i32* <ptr>, i32 <delta> )
6141 declare i64 @llvm.atomic.load.add.i64..p0i64( i64* <ptr>, i64 <delta> )
6146 This intrinsic adds <tt>delta</tt> to the value stored in memory at
6147 <tt>ptr</tt>. It yields the original value at <tt>ptr</tt>.
6152 The intrinsic takes two arguments, the first a pointer to an integer value
6153 and the second an integer value. The result is also an integer value. These
6154 integer types can have any bit width, but they must all have the same bit
6155 width. The targets may only lower integer representations they support.
6159 This intrinsic does a series of operations atomically. It first loads the
6160 value stored at <tt>ptr</tt>. It then adds <tt>delta</tt>, stores the result
6161 to <tt>ptr</tt>. It yields the original value stored at <tt>ptr</tt>.
6168 %result1 = call i32 @llvm.atomic.load.add.i32.p0i32( i32* %ptr, i32 4 )
6169 <i>; yields {i32}:result1 = 4</i>
6170 %result2 = call i32 @llvm.atomic.load.add.i32.p0i32( i32* %ptr, i32 2 )
6171 <i>; yields {i32}:result2 = 8</i>
6172 %result3 = call i32 @llvm.atomic.load.add.i32.p0i32( i32* %ptr, i32 5 )
6173 <i>; yields {i32}:result3 = 10</i>
6174 %memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 15</i>
6178 <!-- _______________________________________________________________________ -->
6179 <div class="doc_subsubsection">
6180 <a name="int_atomic_load_sub">'<tt>llvm.atomic.load.sub.*</tt>' Intrinsic</a>
6183 <div class="doc_text">
6186 This is an overloaded intrinsic. You can use <tt>llvm.atomic.load.sub</tt> on
6187 any integer bit width and for different address spaces. Not all targets
6188 support all bit widths however.</p>
6190 declare i8 @llvm.atomic.load.sub.i8.p0i32( i8* <ptr>, i8 <delta> )
6191 declare i16 @llvm.atomic.load.sub.i16.p0i32( i16* <ptr>, i16 <delta> )
6192 declare i32 @llvm.atomic.load.sub.i32.p0i32( i32* <ptr>, i32 <delta> )
6193 declare i64 @llvm.atomic.load.sub.i64.p0i32( i64* <ptr>, i64 <delta> )
6198 This intrinsic subtracts <tt>delta</tt> to the value stored in memory at
6199 <tt>ptr</tt>. It yields the original value at <tt>ptr</tt>.
6204 The intrinsic takes two arguments, the first a pointer to an integer value
6205 and the second an integer value. The result is also an integer value. These
6206 integer types can have any bit width, but they must all have the same bit
6207 width. The targets may only lower integer representations they support.
6211 This intrinsic does a series of operations atomically. It first loads the
6212 value stored at <tt>ptr</tt>. It then subtracts <tt>delta</tt>, stores the
6213 result to <tt>ptr</tt>. It yields the original value stored at <tt>ptr</tt>.
6220 %result1 = call i32 @llvm.atomic.load.sub.i32.p0i32( i32* %ptr, i32 4 )
6221 <i>; yields {i32}:result1 = 8</i>
6222 %result2 = call i32 @llvm.atomic.load.sub.i32.p0i32( i32* %ptr, i32 2 )
6223 <i>; yields {i32}:result2 = 4</i>
6224 %result3 = call i32 @llvm.atomic.load.sub.i32.p0i32( i32* %ptr, i32 5 )
6225 <i>; yields {i32}:result3 = 2</i>
6226 %memval1 = load i32* %ptr <i>; yields {i32}:memval1 = -3</i>
6230 <!-- _______________________________________________________________________ -->
6231 <div class="doc_subsubsection">
6232 <a name="int_atomic_load_and">'<tt>llvm.atomic.load.and.*</tt>' Intrinsic</a><br>
6233 <a name="int_atomic_load_nand">'<tt>llvm.atomic.load.nand.*</tt>' Intrinsic</a><br>
6234 <a name="int_atomic_load_or">'<tt>llvm.atomic.load.or.*</tt>' Intrinsic</a><br>
6235 <a name="int_atomic_load_xor">'<tt>llvm.atomic.load.xor.*</tt>' Intrinsic</a><br>
6238 <div class="doc_text">
6241 These are overloaded intrinsics. You can use <tt>llvm.atomic.load_and</tt>,
6242 <tt>llvm.atomic.load_nand</tt>, <tt>llvm.atomic.load_or</tt>, and
6243 <tt>llvm.atomic.load_xor</tt> on any integer bit width and for different
6244 address spaces. Not all targets support all bit widths however.</p>
6246 declare i8 @llvm.atomic.load.and.i8.p0i8( i8* <ptr>, i8 <delta> )
6247 declare i16 @llvm.atomic.load.and.i16.p0i16( i16* <ptr>, i16 <delta> )
6248 declare i32 @llvm.atomic.load.and.i32.p0i32( i32* <ptr>, i32 <delta> )
6249 declare i64 @llvm.atomic.load.and.i64.p0i64( i64* <ptr>, i64 <delta> )
6254 declare i8 @llvm.atomic.load.or.i8.p0i8( i8* <ptr>, i8 <delta> )
6255 declare i16 @llvm.atomic.load.or.i16.p0i16( i16* <ptr>, i16 <delta> )
6256 declare i32 @llvm.atomic.load.or.i32.p0i32( i32* <ptr>, i32 <delta> )
6257 declare i64 @llvm.atomic.load.or.i64.p0i64( i64* <ptr>, i64 <delta> )
6262 declare i8 @llvm.atomic.load.nand.i8.p0i32( i8* <ptr>, i8 <delta> )
6263 declare i16 @llvm.atomic.load.nand.i16.p0i32( i16* <ptr>, i16 <delta> )
6264 declare i32 @llvm.atomic.load.nand.i32.p0i32( i32* <ptr>, i32 <delta> )
6265 declare i64 @llvm.atomic.load.nand.i64.p0i32( i64* <ptr>, i64 <delta> )
6270 declare i8 @llvm.atomic.load.xor.i8.p0i32( i8* <ptr>, i8 <delta> )
6271 declare i16 @llvm.atomic.load.xor.i16.p0i32( i16* <ptr>, i16 <delta> )
6272 declare i32 @llvm.atomic.load.xor.i32.p0i32( i32* <ptr>, i32 <delta> )
6273 declare i64 @llvm.atomic.load.xor.i64.p0i32( i64* <ptr>, i64 <delta> )
6278 These intrinsics bitwise the operation (and, nand, or, xor) <tt>delta</tt> to
6279 the value stored in memory at <tt>ptr</tt>. It yields the original value
6285 These intrinsics take two arguments, the first a pointer to an integer value
6286 and the second an integer value. The result is also an integer value. These
6287 integer types can have any bit width, but they must all have the same bit
6288 width. The targets may only lower integer representations they support.
6292 These intrinsics does a series of operations atomically. They first load the
6293 value stored at <tt>ptr</tt>. They then do the bitwise operation
6294 <tt>delta</tt>, store the result to <tt>ptr</tt>. They yield the original
6295 value stored at <tt>ptr</tt>.
6301 store i32 0x0F0F, %ptr
6302 %result0 = call i32 @llvm.atomic.load.nand.i32.p0i32( i32* %ptr, i32 0xFF )
6303 <i>; yields {i32}:result0 = 0x0F0F</i>
6304 %result1 = call i32 @llvm.atomic.load.and.i32.p0i32( i32* %ptr, i32 0xFF )
6305 <i>; yields {i32}:result1 = 0xFFFFFFF0</i>
6306 %result2 = call i32 @llvm.atomic.load.or.i32.p0i32( i32* %ptr, i32 0F )
6307 <i>; yields {i32}:result2 = 0xF0</i>
6308 %result3 = call i32 @llvm.atomic.load.xor.i32.p0i32( i32* %ptr, i32 0F )
6309 <i>; yields {i32}:result3 = FF</i>
6310 %memval1 = load i32* %ptr <i>; yields {i32}:memval1 = F0</i>
6315 <!-- _______________________________________________________________________ -->
6316 <div class="doc_subsubsection">
6317 <a name="int_atomic_load_max">'<tt>llvm.atomic.load.max.*</tt>' Intrinsic</a><br>
6318 <a name="int_atomic_load_min">'<tt>llvm.atomic.load.min.*</tt>' Intrinsic</a><br>
6319 <a name="int_atomic_load_umax">'<tt>llvm.atomic.load.umax.*</tt>' Intrinsic</a><br>
6320 <a name="int_atomic_load_umin">'<tt>llvm.atomic.load.umin.*</tt>' Intrinsic</a><br>
6323 <div class="doc_text">
6326 These are overloaded intrinsics. You can use <tt>llvm.atomic.load_max</tt>,
6327 <tt>llvm.atomic.load_min</tt>, <tt>llvm.atomic.load_umax</tt>, and
6328 <tt>llvm.atomic.load_umin</tt> on any integer bit width and for different
6329 address spaces. Not all targets
6330 support all bit widths however.</p>
6332 declare i8 @llvm.atomic.load.max.i8.p0i8( i8* <ptr>, i8 <delta> )
6333 declare i16 @llvm.atomic.load.max.i16.p0i16( i16* <ptr>, i16 <delta> )
6334 declare i32 @llvm.atomic.load.max.i32.p0i32( i32* <ptr>, i32 <delta> )
6335 declare i64 @llvm.atomic.load.max.i64.p0i64( i64* <ptr>, i64 <delta> )
6340 declare i8 @llvm.atomic.load.min.i8.p0i8( i8* <ptr>, i8 <delta> )
6341 declare i16 @llvm.atomic.load.min.i16.p0i16( i16* <ptr>, i16 <delta> )
6342 declare i32 @llvm.atomic.load.min.i32..p0i32( i32* <ptr>, i32 <delta> )
6343 declare i64 @llvm.atomic.load.min.i64..p0i64( i64* <ptr>, i64 <delta> )
6348 declare i8 @llvm.atomic.load.umax.i8.p0i8( i8* <ptr>, i8 <delta> )
6349 declare i16 @llvm.atomic.load.umax.i16.p0i16( i16* <ptr>, i16 <delta> )
6350 declare i32 @llvm.atomic.load.umax.i32.p0i32( i32* <ptr>, i32 <delta> )
6351 declare i64 @llvm.atomic.load.umax.i64.p0i64( i64* <ptr>, i64 <delta> )
6356 declare i8 @llvm.atomic.load.umin.i8..p0i8( i8* <ptr>, i8 <delta> )
6357 declare i16 @llvm.atomic.load.umin.i16.p0i16( i16* <ptr>, i16 <delta> )
6358 declare i32 @llvm.atomic.load.umin.i32..p0i32( i32* <ptr>, i32 <delta> )
6359 declare i64 @llvm.atomic.load.umin.i64..p0i64( i64* <ptr>, i64 <delta> )
6364 These intrinsics takes the signed or unsigned minimum or maximum of
6365 <tt>delta</tt> and the value stored in memory at <tt>ptr</tt>. It yields the
6366 original value at <tt>ptr</tt>.
6371 These intrinsics take two arguments, the first a pointer to an integer value
6372 and the second an integer value. The result is also an integer value. These
6373 integer types can have any bit width, but they must all have the same bit
6374 width. The targets may only lower integer representations they support.
6378 These intrinsics does a series of operations atomically. They first load the
6379 value stored at <tt>ptr</tt>. They then do the signed or unsigned min or max
6380 <tt>delta</tt> and the value, store the result to <tt>ptr</tt>. They yield
6381 the original value stored at <tt>ptr</tt>.
6388 %result0 = call i32 @llvm.atomic.load.min.i32.p0i32( i32* %ptr, i32 -2 )
6389 <i>; yields {i32}:result0 = 7</i>
6390 %result1 = call i32 @llvm.atomic.load.max.i32.p0i32( i32* %ptr, i32 8 )
6391 <i>; yields {i32}:result1 = -2</i>
6392 %result2 = call i32 @llvm.atomic.load.umin.i32.p0i32( i32* %ptr, i32 10 )
6393 <i>; yields {i32}:result2 = 8</i>
6394 %result3 = call i32 @llvm.atomic.load.umax.i32.p0i32( i32* %ptr, i32 30 )
6395 <i>; yields {i32}:result3 = 8</i>
6396 %memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 30</i>
6400 <!-- ======================================================================= -->
6401 <div class="doc_subsection">
6402 <a name="int_general">General Intrinsics</a>
6405 <div class="doc_text">
6406 <p> This class of intrinsics is designed to be generic and has
6407 no specific purpose. </p>
6410 <!-- _______________________________________________________________________ -->
6411 <div class="doc_subsubsection">
6412 <a name="int_var_annotation">'<tt>llvm.var.annotation</tt>' Intrinsic</a>
6415 <div class="doc_text">
6419 declare void @llvm.var.annotation(i8* <val>, i8* <str>, i8* <str>, i32 <int> )
6425 The '<tt>llvm.var.annotation</tt>' intrinsic
6431 The first argument is a pointer to a value, the second is a pointer to a
6432 global string, the third is a pointer to a global string which is the source
6433 file name, and the last argument is the line number.
6439 This intrinsic allows annotation of local variables with arbitrary strings.
6440 This can be useful for special purpose optimizations that want to look for these
6441 annotations. These have no other defined use, they are ignored by code
6442 generation and optimization.
6446 <!-- _______________________________________________________________________ -->
6447 <div class="doc_subsubsection">
6448 <a name="int_annotation">'<tt>llvm.annotation.*</tt>' Intrinsic</a>
6451 <div class="doc_text">
6454 <p>This is an overloaded intrinsic. You can use '<tt>llvm.annotation</tt>' on
6455 any integer bit width.
6458 declare i8 @llvm.annotation.i8(i8 <val>, i8* <str>, i8* <str>, i32 <int> )
6459 declare i16 @llvm.annotation.i16(i16 <val>, i8* <str>, i8* <str>, i32 <int> )
6460 declare i32 @llvm.annotation.i32(i32 <val>, i8* <str>, i8* <str>, i32 <int> )
6461 declare i64 @llvm.annotation.i64(i64 <val>, i8* <str>, i8* <str>, i32 <int> )
6462 declare i256 @llvm.annotation.i256(i256 <val>, i8* <str>, i8* <str>, i32 <int> )
6468 The '<tt>llvm.annotation</tt>' intrinsic.
6474 The first argument is an integer value (result of some expression),
6475 the second is a pointer to a global string, the third is a pointer to a global
6476 string which is the source file name, and the last argument is the line number.
6477 It returns the value of the first argument.
6483 This intrinsic allows annotations to be put on arbitrary expressions
6484 with arbitrary strings. This can be useful for special purpose optimizations
6485 that want to look for these annotations. These have no other defined use, they
6486 are ignored by code generation and optimization.
6490 <!-- _______________________________________________________________________ -->
6491 <div class="doc_subsubsection">
6492 <a name="int_trap">'<tt>llvm.trap</tt>' Intrinsic</a>
6495 <div class="doc_text">
6499 declare void @llvm.trap()
6505 The '<tt>llvm.trap</tt>' intrinsic
6517 This intrinsics is lowered to the target dependent trap instruction. If the
6518 target does not have a trap instruction, this intrinsic will be lowered to the
6519 call of the abort() function.
6523 <!-- _______________________________________________________________________ -->
6524 <div class="doc_subsubsection">
6525 <a name="int_stackprotector">'<tt>llvm.stackprotector</tt>' Intrinsic</a>
6527 <div class="doc_text">
6530 declare void @llvm.stackprotector( i8* <guard>, i8** <slot> )
6535 The <tt>llvm.stackprotector</tt> intrinsic takes the <tt>guard</tt> and stores
6536 it onto the stack at <tt>slot</tt>. The stack slot is adjusted to ensure that
6537 it is placed on the stack before local variables.
6541 The <tt>llvm.stackprotector</tt> intrinsic requires two pointer arguments. The
6542 first argument is the value loaded from the stack guard
6543 <tt>@__stack_chk_guard</tt>. The second variable is an <tt>alloca</tt> that
6544 has enough space to hold the value of the guard.
6548 This intrinsic causes the prologue/epilogue inserter to force the position of
6549 the <tt>AllocaInst</tt> stack slot to be before local variables on the
6550 stack. This is to ensure that if a local variable on the stack is overwritten,
6551 it will destroy the value of the guard. When the function exits, the guard on
6552 the stack is checked against the original guard. If they're different, then
6553 the program aborts by calling the <tt>__stack_chk_fail()</tt> function.
6557 <!-- *********************************************************************** -->
6560 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
6561 src="http://jigsaw.w3.org/css-validator/images/vcss-blue" alt="Valid CSS"></a>
6562 <a href="http://validator.w3.org/check/referer"><img
6563 src="http://www.w3.org/Icons/valid-html401-blue" alt="Valid HTML 4.01"></a>
6565 <a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
6566 <a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br>
6567 Last modified: $Date$