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="#globalvars">Global Variables</a></li>
26 <li><a href="#functionstructure">Functions</a></li>
27 <li><a href="#aliasstructure">Aliases</a>
28 <li><a href="#paramattrs">Parameter Attributes</a></li>
29 <li><a href="#moduleasm">Module-Level Inline Assembly</a></li>
30 <li><a href="#datalayout">Data Layout</a></li>
33 <li><a href="#typesystem">Type System</a>
35 <li><a href="#t_primitive">Primitive Types</a>
37 <li><a href="#t_classifications">Type Classifications</a></li>
40 <li><a href="#t_derived">Derived Types</a>
42 <li><a href="#t_array">Array Type</a></li>
43 <li><a href="#t_function">Function Type</a></li>
44 <li><a href="#t_pointer">Pointer Type</a></li>
45 <li><a href="#t_struct">Structure Type</a></li>
46 <li><a href="#t_pstruct">Packed Structure Type</a></li>
47 <li><a href="#t_vector">Vector Type</a></li>
48 <li><a href="#t_opaque">Opaque Type</a></li>
53 <li><a href="#constants">Constants</a>
55 <li><a href="#simpleconstants">Simple Constants</a>
56 <li><a href="#aggregateconstants">Aggregate Constants</a>
57 <li><a href="#globalconstants">Global Variable and Function Addresses</a>
58 <li><a href="#undefvalues">Undefined Values</a>
59 <li><a href="#constantexprs">Constant Expressions</a>
62 <li><a href="#othervalues">Other Values</a>
64 <li><a href="#inlineasm">Inline Assembler Expressions</a>
67 <li><a href="#instref">Instruction Reference</a>
69 <li><a href="#terminators">Terminator Instructions</a>
71 <li><a href="#i_ret">'<tt>ret</tt>' Instruction</a></li>
72 <li><a href="#i_br">'<tt>br</tt>' Instruction</a></li>
73 <li><a href="#i_switch">'<tt>switch</tt>' Instruction</a></li>
74 <li><a href="#i_invoke">'<tt>invoke</tt>' Instruction</a></li>
75 <li><a href="#i_unwind">'<tt>unwind</tt>' Instruction</a></li>
76 <li><a href="#i_unreachable">'<tt>unreachable</tt>' Instruction</a></li>
79 <li><a href="#binaryops">Binary Operations</a>
81 <li><a href="#i_add">'<tt>add</tt>' Instruction</a></li>
82 <li><a href="#i_sub">'<tt>sub</tt>' Instruction</a></li>
83 <li><a href="#i_mul">'<tt>mul</tt>' Instruction</a></li>
84 <li><a href="#i_udiv">'<tt>udiv</tt>' Instruction</a></li>
85 <li><a href="#i_sdiv">'<tt>sdiv</tt>' Instruction</a></li>
86 <li><a href="#i_fdiv">'<tt>fdiv</tt>' Instruction</a></li>
87 <li><a href="#i_urem">'<tt>urem</tt>' Instruction</a></li>
88 <li><a href="#i_srem">'<tt>srem</tt>' Instruction</a></li>
89 <li><a href="#i_frem">'<tt>frem</tt>' Instruction</a></li>
92 <li><a href="#bitwiseops">Bitwise Binary Operations</a>
94 <li><a href="#i_shl">'<tt>shl</tt>' Instruction</a></li>
95 <li><a href="#i_lshr">'<tt>lshr</tt>' Instruction</a></li>
96 <li><a href="#i_ashr">'<tt>ashr</tt>' Instruction</a></li>
97 <li><a href="#i_and">'<tt>and</tt>' Instruction</a></li>
98 <li><a href="#i_or">'<tt>or</tt>' Instruction</a></li>
99 <li><a href="#i_xor">'<tt>xor</tt>' Instruction</a></li>
102 <li><a href="#vectorops">Vector Operations</a>
104 <li><a href="#i_extractelement">'<tt>extractelement</tt>' Instruction</a></li>
105 <li><a href="#i_insertelement">'<tt>insertelement</tt>' Instruction</a></li>
106 <li><a href="#i_shufflevector">'<tt>shufflevector</tt>' Instruction</a></li>
109 <li><a href="#memoryops">Memory Access and Addressing Operations</a>
111 <li><a href="#i_malloc">'<tt>malloc</tt>' Instruction</a></li>
112 <li><a href="#i_free">'<tt>free</tt>' Instruction</a></li>
113 <li><a href="#i_alloca">'<tt>alloca</tt>' Instruction</a></li>
114 <li><a href="#i_load">'<tt>load</tt>' Instruction</a></li>
115 <li><a href="#i_store">'<tt>store</tt>' Instruction</a></li>
116 <li><a href="#i_getelementptr">'<tt>getelementptr</tt>' Instruction</a></li>
119 <li><a href="#convertops">Conversion Operations</a>
121 <li><a href="#i_trunc">'<tt>trunc .. to</tt>' Instruction</a></li>
122 <li><a href="#i_zext">'<tt>zext .. to</tt>' Instruction</a></li>
123 <li><a href="#i_sext">'<tt>sext .. to</tt>' Instruction</a></li>
124 <li><a href="#i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a></li>
125 <li><a href="#i_fpext">'<tt>fpext .. to</tt>' Instruction</a></li>
126 <li><a href="#i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a></li>
127 <li><a href="#i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a></li>
128 <li><a href="#i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a></li>
129 <li><a href="#i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a></li>
130 <li><a href="#i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a></li>
131 <li><a href="#i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a></li>
132 <li><a href="#i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a></li>
134 <li><a href="#otherops">Other Operations</a>
136 <li><a href="#i_icmp">'<tt>icmp</tt>' Instruction</a></li>
137 <li><a href="#i_fcmp">'<tt>fcmp</tt>' Instruction</a></li>
138 <li><a href="#i_phi">'<tt>phi</tt>' Instruction</a></li>
139 <li><a href="#i_select">'<tt>select</tt>' Instruction</a></li>
140 <li><a href="#i_call">'<tt>call</tt>' Instruction</a></li>
141 <li><a href="#i_va_arg">'<tt>va_arg</tt>' Instruction</a></li>
146 <li><a href="#intrinsics">Intrinsic Functions</a>
148 <li><a href="#int_varargs">Variable Argument Handling Intrinsics</a>
150 <li><a href="#int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a></li>
151 <li><a href="#int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a></li>
152 <li><a href="#int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a></li>
155 <li><a href="#int_gc">Accurate Garbage Collection Intrinsics</a>
157 <li><a href="#int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a></li>
158 <li><a href="#int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a></li>
159 <li><a href="#int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a></li>
162 <li><a href="#int_codegen">Code Generator Intrinsics</a>
164 <li><a href="#int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a></li>
165 <li><a href="#int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a></li>
166 <li><a href="#int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a></li>
167 <li><a href="#int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a></li>
168 <li><a href="#int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a></li>
169 <li><a href="#int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a></li>
170 <li><a href="#int_readcyclecounter"><tt>llvm.readcyclecounter</tt>' Intrinsic</a></li>
173 <li><a href="#int_libc">Standard C Library Intrinsics</a>
175 <li><a href="#int_memcpy">'<tt>llvm.memcpy.*</tt>' Intrinsic</a></li>
176 <li><a href="#int_memmove">'<tt>llvm.memmove.*</tt>' Intrinsic</a></li>
177 <li><a href="#int_memset">'<tt>llvm.memset.*</tt>' Intrinsic</a></li>
178 <li><a href="#int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a></li>
179 <li><a href="#int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a></li>
182 <li><a href="#int_manip">Bit Manipulation Intrinsics</a>
184 <li><a href="#int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a></li>
185 <li><a href="#int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic </a></li>
186 <li><a href="#int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic </a></li>
187 <li><a href="#int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic </a></li>
188 <li><a href="#int_part_select">'<tt>llvm.part.select.*</tt>' Intrinsic </a></li>
189 <li><a href="#int_part_set">'<tt>llvm.part.set.*</tt>' Intrinsic </a></li>
192 <li><a href="#int_debugger">Debugger intrinsics</a></li>
193 <li><a href="#int_eh">Exception Handling intrinsics</a></li>
198 <div class="doc_author">
199 <p>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a>
200 and <a href="mailto:vadve@cs.uiuc.edu">Vikram Adve</a></p>
203 <!-- *********************************************************************** -->
204 <div class="doc_section"> <a name="abstract">Abstract </a></div>
205 <!-- *********************************************************************** -->
207 <div class="doc_text">
208 <p>This document is a reference manual for the LLVM assembly language.
209 LLVM is an SSA based representation that provides type safety,
210 low-level operations, flexibility, and the capability of representing
211 'all' high-level languages cleanly. It is the common code
212 representation used throughout all phases of the LLVM compilation
216 <!-- *********************************************************************** -->
217 <div class="doc_section"> <a name="introduction">Introduction</a> </div>
218 <!-- *********************************************************************** -->
220 <div class="doc_text">
222 <p>The LLVM code representation is designed to be used in three
223 different forms: as an in-memory compiler IR, as an on-disk bytecode
224 representation (suitable for fast loading by a Just-In-Time compiler),
225 and as a human readable assembly language representation. This allows
226 LLVM to provide a powerful intermediate representation for efficient
227 compiler transformations and analysis, while providing a natural means
228 to debug and visualize the transformations. The three different forms
229 of LLVM are all equivalent. This document describes the human readable
230 representation and notation.</p>
232 <p>The LLVM representation aims to be light-weight and low-level
233 while being expressive, typed, and extensible at the same time. It
234 aims to be a "universal IR" of sorts, by being at a low enough level
235 that high-level ideas may be cleanly mapped to it (similar to how
236 microprocessors are "universal IR's", allowing many source languages to
237 be mapped to them). By providing type information, LLVM can be used as
238 the target of optimizations: for example, through pointer analysis, it
239 can be proven that a C automatic variable is never accessed outside of
240 the current function... allowing it to be promoted to a simple SSA
241 value instead of a memory location.</p>
245 <!-- _______________________________________________________________________ -->
246 <div class="doc_subsubsection"> <a name="wellformed">Well-Formedness</a> </div>
248 <div class="doc_text">
250 <p>It is important to note that this document describes 'well formed'
251 LLVM assembly language. There is a difference between what the parser
252 accepts and what is considered 'well formed'. For example, the
253 following instruction is syntactically okay, but not well formed:</p>
256 %x = <a href="#i_add">add</a> i32 1, %x
259 <p>...because the definition of <tt>%x</tt> does not dominate all of
260 its uses. The LLVM infrastructure provides a verification pass that may
261 be used to verify that an LLVM module is well formed. This pass is
262 automatically run by the parser after parsing input assembly and by
263 the optimizer before it outputs bytecode. The violations pointed out
264 by the verifier pass indicate bugs in transformation passes or input to
267 <!-- Describe the typesetting conventions here. --> </div>
269 <!-- *********************************************************************** -->
270 <div class="doc_section"> <a name="identifiers">Identifiers</a> </div>
271 <!-- *********************************************************************** -->
273 <div class="doc_text">
275 <p>LLVM uses three different forms of identifiers, for different
279 <li>Named values are represented as a string of characters with a '%' prefix.
280 For example, %foo, %DivisionByZero, %a.really.long.identifier. The actual
281 regular expression used is '<tt>%[a-zA-Z$._][a-zA-Z$._0-9]*</tt>'.
282 Identifiers which require other characters in their names can be surrounded
283 with quotes. In this way, anything except a <tt>"</tt> character can be used
286 <li>Unnamed values are represented as an unsigned numeric value with a '%'
287 prefix. For example, %12, %2, %44.</li>
289 <li>Constants, which are described in a <a href="#constants">section about
290 constants</a>, below.</li>
293 <p>LLVM requires that values start with a '%' sign for two reasons: Compilers
294 don't need to worry about name clashes with reserved words, and the set of
295 reserved words may be expanded in the future without penalty. Additionally,
296 unnamed identifiers allow a compiler to quickly come up with a temporary
297 variable without having to avoid symbol table conflicts.</p>
299 <p>Reserved words in LLVM are very similar to reserved words in other
300 languages. There are keywords for different opcodes
301 ('<tt><a href="#i_add">add</a></tt>',
302 '<tt><a href="#i_bitcast">bitcast</a></tt>',
303 '<tt><a href="#i_ret">ret</a></tt>', etc...), for primitive type names ('<tt><a
304 href="#t_void">void</a></tt>', '<tt><a href="#t_primitive">i32</a></tt>', etc...),
305 and others. These reserved words cannot conflict with variable names, because
306 none of them start with a '%' character.</p>
308 <p>Here is an example of LLVM code to multiply the integer variable
309 '<tt>%X</tt>' by 8:</p>
314 %result = <a href="#i_mul">mul</a> i32 %X, 8
317 <p>After strength reduction:</p>
320 %result = <a href="#i_shl">shl</a> i32 %X, i8 3
323 <p>And the hard way:</p>
326 <a href="#i_add">add</a> i32 %X, %X <i>; yields {i32}:%0</i>
327 <a href="#i_add">add</a> i32 %0, %0 <i>; yields {i32}:%1</i>
328 %result = <a href="#i_add">add</a> i32 %1, %1
331 <p>This last way of multiplying <tt>%X</tt> by 8 illustrates several
332 important lexical features of LLVM:</p>
336 <li>Comments are delimited with a '<tt>;</tt>' and go until the end of
339 <li>Unnamed temporaries are created when the result of a computation is not
340 assigned to a named value.</li>
342 <li>Unnamed temporaries are numbered sequentially</li>
346 <p>...and it also shows a convention that we follow in this document. When
347 demonstrating instructions, we will follow an instruction with a comment that
348 defines the type and name of value produced. Comments are shown in italic
353 <!-- *********************************************************************** -->
354 <div class="doc_section"> <a name="highlevel">High Level Structure</a> </div>
355 <!-- *********************************************************************** -->
357 <!-- ======================================================================= -->
358 <div class="doc_subsection"> <a name="modulestructure">Module Structure</a>
361 <div class="doc_text">
363 <p>LLVM programs are composed of "Module"s, each of which is a
364 translation unit of the input programs. Each module consists of
365 functions, global variables, and symbol table entries. Modules may be
366 combined together with the LLVM linker, which merges function (and
367 global variable) definitions, resolves forward declarations, and merges
368 symbol table entries. Here is an example of the "hello world" module:</p>
370 <pre><i>; Declare the string constant as a global constant...</i>
371 <a href="#identifiers">%.LC0</a> = <a href="#linkage_internal">internal</a> <a
372 href="#globalvars">constant</a> <a href="#t_array">[13 x i8 ]</a> c"hello world\0A\00" <i>; [13 x i8 ]*</i>
374 <i>; External declaration of the puts function</i>
375 <a href="#functionstructure">declare</a> i32 %puts(i8 *) <i>; i32(i8 *)* </i>
377 <i>; Definition of main function</i>
378 define i32 %main() { <i>; i32()* </i>
379 <i>; Convert [13x i8 ]* to i8 *...</i>
381 href="#i_getelementptr">getelementptr</a> [13 x i8 ]* %.LC0, i64 0, i64 0 <i>; i8 *</i>
383 <i>; Call puts function to write out the string to stdout...</i>
385 href="#i_call">call</a> i32 %puts(i8 * %cast210) <i>; i32</i>
387 href="#i_ret">ret</a> i32 0<br>}<br></pre>
389 <p>This example is made up of a <a href="#globalvars">global variable</a>
390 named "<tt>.LC0</tt>", an external declaration of the "<tt>puts</tt>"
391 function, and a <a href="#functionstructure">function definition</a>
392 for "<tt>main</tt>".</p>
394 <p>In general, a module is made up of a list of global values,
395 where both functions and global variables are global values. Global values are
396 represented by a pointer to a memory location (in this case, a pointer to an
397 array of char, and a pointer to a function), and have one of the following <a
398 href="#linkage">linkage types</a>.</p>
402 <!-- ======================================================================= -->
403 <div class="doc_subsection">
404 <a name="linkage">Linkage Types</a>
407 <div class="doc_text">
410 All Global Variables and Functions have one of the following types of linkage:
415 <dt><tt><b><a name="linkage_internal">internal</a></b></tt> </dt>
417 <dd>Global values with internal linkage are only directly accessible by
418 objects in the current module. In particular, linking code into a module with
419 an internal global value may cause the internal to be renamed as necessary to
420 avoid collisions. Because the symbol is internal to the module, all
421 references can be updated. This corresponds to the notion of the
422 '<tt>static</tt>' keyword in C.
425 <dt><tt><b><a name="linkage_linkonce">linkonce</a></b></tt>: </dt>
427 <dd>Globals with "<tt>linkonce</tt>" linkage are merged with other globals of
428 the same name when linkage occurs. This is typically used to implement
429 inline functions, templates, or other code which must be generated in each
430 translation unit that uses it. Unreferenced <tt>linkonce</tt> globals are
431 allowed to be discarded.
434 <dt><tt><b><a name="linkage_weak">weak</a></b></tt>: </dt>
436 <dd>"<tt>weak</tt>" linkage is exactly the same as <tt>linkonce</tt> linkage,
437 except that unreferenced <tt>weak</tt> globals may not be discarded. This is
438 used for globals that may be emitted in multiple translation units, but that
439 are not guaranteed to be emitted into every translation unit that uses them.
440 One example of this are common globals in C, such as "<tt>int X;</tt>" at
444 <dt><tt><b><a name="linkage_appending">appending</a></b></tt>: </dt>
446 <dd>"<tt>appending</tt>" linkage may only be applied to global variables of
447 pointer to array type. When two global variables with appending linkage are
448 linked together, the two global arrays are appended together. This is the
449 LLVM, typesafe, equivalent of having the system linker append together
450 "sections" with identical names when .o files are linked.
453 <dt><tt><b><a name="linkage_externweak">extern_weak</a></b></tt>: </dt>
454 <dd>The semantics of this linkage follow the ELF model: the symbol is weak
455 until linked, if not linked, the symbol becomes null instead of being an
459 <dt><tt><b><a name="linkage_external">externally visible</a></b></tt>:</dt>
461 <dd>If none of the above identifiers are used, the global is externally
462 visible, meaning that it participates in linkage and can be used to resolve
463 external symbol references.
468 The next two types of linkage are targeted for Microsoft Windows platform
469 only. They are designed to support importing (exporting) symbols from (to)
474 <dt><tt><b><a name="linkage_dllimport">dllimport</a></b></tt>: </dt>
476 <dd>"<tt>dllimport</tt>" linkage causes the compiler to reference a function
477 or variable via a global pointer to a pointer that is set up by the DLL
478 exporting the symbol. On Microsoft Windows targets, the pointer name is
479 formed by combining <code>_imp__</code> and the function or variable name.
482 <dt><tt><b><a name="linkage_dllexport">dllexport</a></b></tt>: </dt>
484 <dd>"<tt>dllexport</tt>" linkage causes the compiler to provide a global
485 pointer to a pointer in a DLL, so that it can be referenced with the
486 <tt>dllimport</tt> attribute. On Microsoft Windows targets, the pointer
487 name is formed by combining <code>_imp__</code> and the function or variable
493 <p><a name="linkage_external"></a>For example, since the "<tt>.LC0</tt>"
494 variable is defined to be internal, if another module defined a "<tt>.LC0</tt>"
495 variable and was linked with this one, one of the two would be renamed,
496 preventing a collision. Since "<tt>main</tt>" and "<tt>puts</tt>" are
497 external (i.e., lacking any linkage declarations), they are accessible
498 outside of the current module.</p>
499 <p>It is illegal for a function <i>declaration</i>
500 to have any linkage type other than "externally visible", <tt>dllimport</tt>,
501 or <tt>extern_weak</tt>.</p>
502 <p>Aliases can have only <tt>external</tt>, <tt>internal</tt> and <tt>weak</tt>
506 <!-- ======================================================================= -->
507 <div class="doc_subsection">
508 <a name="callingconv">Calling Conventions</a>
511 <div class="doc_text">
513 <p>LLVM <a href="#functionstructure">functions</a>, <a href="#i_call">calls</a>
514 and <a href="#i_invoke">invokes</a> can all have an optional calling convention
515 specified for the call. The calling convention of any pair of dynamic
516 caller/callee must match, or the behavior of the program is undefined. The
517 following calling conventions are supported by LLVM, and more may be added in
521 <dt><b>"<tt>ccc</tt>" - The C calling convention</b>:</dt>
523 <dd>This calling convention (the default if no other calling convention is
524 specified) matches the target C calling conventions. This calling convention
525 supports varargs function calls and tolerates some mismatch in the declared
526 prototype and implemented declaration of the function (as does normal C).
529 <dt><b>"<tt>fastcc</tt>" - The fast calling convention</b>:</dt>
531 <dd>This calling convention attempts to make calls as fast as possible
532 (e.g. by passing things in registers). This calling convention allows the
533 target to use whatever tricks it wants to produce fast code for the target,
534 without having to conform to an externally specified ABI. Implementations of
535 this convention should allow arbitrary tail call optimization to be supported.
536 This calling convention does not support varargs and requires the prototype of
537 all callees to exactly match the prototype of the function definition.
540 <dt><b>"<tt>coldcc</tt>" - The cold calling convention</b>:</dt>
542 <dd>This calling convention attempts to make code in the caller as efficient
543 as possible under the assumption that the call is not commonly executed. As
544 such, these calls often preserve all registers so that the call does not break
545 any live ranges in the caller side. This calling convention does not support
546 varargs and requires the prototype of all callees to exactly match the
547 prototype of the function definition.
550 <dt><b>"<tt>cc <<em>n</em>></tt>" - Numbered convention</b>:</dt>
552 <dd>Any calling convention may be specified by number, allowing
553 target-specific calling conventions to be used. Target specific calling
554 conventions start at 64.
558 <p>More calling conventions can be added/defined on an as-needed basis, to
559 support pascal conventions or any other well-known target-independent
564 <!-- ======================================================================= -->
565 <div class="doc_subsection">
566 <a name="visibility">Visibility Styles</a>
569 <div class="doc_text">
572 All Global Variables and Functions have one of the following visibility styles:
576 <dt><b>"<tt>default</tt>" - Default style</b>:</dt>
578 <dd>On ELF, default visibility means that the declaration is visible to other
579 modules and, in shared libraries, means that the declared entity may be
580 overridden. On Darwin, default visibility means that the declaration is
581 visible to other modules. Default visibility corresponds to "external
582 linkage" in the language.
585 <dt><b>"<tt>hidden</tt>" - Hidden style</b>:</dt>
587 <dd>Two declarations of an object with hidden visibility refer to the same
588 object if they are in the same shared object. Usually, hidden visibility
589 indicates that the symbol will not be placed into the dynamic symbol table,
590 so no other module (executable or shared library) can reference it
594 <dt><b>"<tt>protected</tt>" - Protected style</b>:</dt>
596 <dd>On ELF, protected visibility indicates that the symbol will be placed in
597 the dynamic symbol table, but that references within the defining module will
598 bind to the local symbol. That is, the symbol cannot be overridden by another
605 <!-- ======================================================================= -->
606 <div class="doc_subsection">
607 <a name="globalvars">Global Variables</a>
610 <div class="doc_text">
612 <p>Global variables define regions of memory allocated at compilation time
613 instead of run-time. Global variables may optionally be initialized, may have
614 an explicit section to be placed in, and may have an optional explicit alignment
615 specified. A variable may be defined as "thread_local", which means that it
616 will not be shared by threads (each thread will have a separated copy of the
617 variable). A variable may be defined as a global "constant," which indicates
618 that the contents of the variable will <b>never</b> be modified (enabling better
619 optimization, allowing the global data to be placed in the read-only section of
620 an executable, etc). Note that variables that need runtime initialization
621 cannot be marked "constant" as there is a store to the variable.</p>
624 LLVM explicitly allows <em>declarations</em> of global variables to be marked
625 constant, even if the final definition of the global is not. This capability
626 can be used to enable slightly better optimization of the program, but requires
627 the language definition to guarantee that optimizations based on the
628 'constantness' are valid for the translation units that do not include the
632 <p>As SSA values, global variables define pointer values that are in
633 scope (i.e. they dominate) all basic blocks in the program. Global
634 variables always define a pointer to their "content" type because they
635 describe a region of memory, and all memory objects in LLVM are
636 accessed through pointers.</p>
638 <p>LLVM allows an explicit section to be specified for globals. If the target
639 supports it, it will emit globals to the section specified.</p>
641 <p>An explicit alignment may be specified for a global. If not present, or if
642 the alignment is set to zero, the alignment of the global is set by the target
643 to whatever it feels convenient. If an explicit alignment is specified, the
644 global is forced to have at least that much alignment. All alignments must be
647 <p>For example, the following defines a global with an initializer, section,
651 %G = constant float 1.0, section "foo", align 4
657 <!-- ======================================================================= -->
658 <div class="doc_subsection">
659 <a name="functionstructure">Functions</a>
662 <div class="doc_text">
664 <p>LLVM function definitions consist of the "<tt>define</tt>" keyord,
665 an optional <a href="#linkage">linkage type</a>, an optional
666 <a href="#visibility">visibility style</a>, an optional
667 <a href="#callingconv">calling convention</a>, a return type, an optional
668 <a href="#paramattrs">parameter attribute</a> for the return type, a function
669 name, a (possibly empty) argument list (each with optional
670 <a href="#paramattrs">parameter attributes</a>), an optional section, an
671 optional alignment, an opening curly brace, a list of basic blocks, and a
674 LLVM function declarations consist of the "<tt>declare</tt>" keyword, an
675 optional <a href="#linkage">linkage type</a>, an optional
676 <a href="#visibility">visibility style</a>, an optional
677 <a href="#callingconv">calling convention</a>, a return type, an optional
678 <a href="#paramattrs">parameter attribute</a> for the return type, a function
679 name, a possibly empty list of arguments, and an optional alignment.</p>
681 <p>A function definition contains a list of basic blocks, forming the CFG for
682 the function. Each basic block may optionally start with a label (giving the
683 basic block a symbol table entry), contains a list of instructions, and ends
684 with a <a href="#terminators">terminator</a> instruction (such as a branch or
685 function return).</p>
687 <p>The first basic block in a program is special in two ways: it is immediately
688 executed on entrance to the function, and it is not allowed to have predecessor
689 basic blocks (i.e. there can not be any branches to the entry block of a
690 function). Because the block can have no predecessors, it also cannot have any
691 <a href="#i_phi">PHI nodes</a>.</p>
693 <p>LLVM functions are identified by their name and type signature. Hence, two
694 functions with the same name but different parameter lists or return values are
695 considered different functions, and LLVM will resolve references to each
698 <p>LLVM allows an explicit section to be specified for functions. If the target
699 supports it, it will emit functions to the section specified.</p>
701 <p>An explicit alignment may be specified for a function. If not present, or if
702 the alignment is set to zero, the alignment of the function is set by the target
703 to whatever it feels convenient. If an explicit alignment is specified, the
704 function is forced to have at least that much alignment. All alignments must be
710 <!-- ======================================================================= -->
711 <div class="doc_subsection">
712 <a name="aliasstructure">Aliases</a>
714 <div class="doc_text">
715 <p>Aliases act as "second name" for the aliasee value (which can be either
716 function or global variable or bitcast of global value). Aliases may have an
717 optional <a href="#linkage">linkage type</a>, and an
718 optional <a href="#visibility">visibility style</a>.</p>
723 @<Name> = [Linkage] [Visibility] alias <AliaseeTy> @<Aliasee>
730 <!-- ======================================================================= -->
731 <div class="doc_subsection"><a name="paramattrs">Parameter Attributes</a></div>
732 <div class="doc_text">
733 <p>The return type and each parameter of a function type may have a set of
734 <i>parameter attributes</i> associated with them. Parameter attributes are
735 used to communicate additional information about the result or parameters of
736 a function. Parameter attributes are considered to be part of the function
737 type so two functions types that differ only by the parameter attributes
738 are different function types.</p>
740 <p>Parameter attributes are simple keywords that follow the type specified. If
741 multiple parameter attributes are needed, they are space separated. For
743 %someFunc = i16 (i8 sext %someParam) zext
744 %someFunc = i16 (i8 zext %someParam) zext</pre>
745 <p>Note that the two function types above are unique because the parameter has
746 a different attribute (sext in the first one, zext in the second). Also note
747 that the attribute for the function result (zext) comes immediately after the
750 <p>Currently, only the following parameter attributes are defined:</p>
752 <dt><tt>zext</tt></dt>
753 <dd>This indicates that the parameter should be zero extended just before
754 a call to this function.</dd>
755 <dt><tt>sext</tt></dt>
756 <dd>This indicates that the parameter should be sign extended just before
757 a call to this function.</dd>
758 <dt><tt>inreg</tt></dt>
759 <dd>This indicates that the parameter should be placed in register (if
760 possible) during assembling function call. Support for this attribute is
762 <dt><tt>sret</tt></dt>
763 <dd>This indicates that the parameter specifies the address of a structure
764 that is the return value of the function in the source program.</dd>
765 <dt><tt>noreturn</tt></dt>
766 <dd>This function attribute indicates that the function never returns. This
767 indicates to LLVM that every call to this function should be treated as if
768 an <tt>unreachable</tt> instruction immediately followed the call.</dd>
769 <dt><tt>nounwind</tt></dt>
770 <dd>This function attribute indicates that the function type does not use
771 the unwind instruction and does not allow stack unwinding to propagate
777 <!-- ======================================================================= -->
778 <div class="doc_subsection">
779 <a name="moduleasm">Module-Level Inline Assembly</a>
782 <div class="doc_text">
784 Modules may contain "module-level inline asm" blocks, which corresponds to the
785 GCC "file scope inline asm" blocks. These blocks are internally concatenated by
786 LLVM and treated as a single unit, but may be separated in the .ll file if
787 desired. The syntax is very simple:
790 <div class="doc_code"><pre>
791 module asm "inline asm code goes here"
792 module asm "more can go here"
795 <p>The strings can contain any character by escaping non-printable characters.
796 The escape sequence used is simply "\xx" where "xx" is the two digit hex code
801 The inline asm code is simply printed to the machine code .s file when
802 assembly code is generated.
806 <!-- ======================================================================= -->
807 <div class="doc_subsection">
808 <a name="datalayout">Data Layout</a>
811 <div class="doc_text">
812 <p>A module may specify a target specific data layout string that specifies how
813 data is to be laid out in memory. The syntax for the data layout is simply:</p>
814 <pre> target datalayout = "<i>layout specification</i>"</pre>
815 <p>The <i>layout specification</i> consists of a list of specifications
816 separated by the minus sign character ('-'). Each specification starts with a
817 letter and may include other information after the letter to define some
818 aspect of the data layout. The specifications accepted are as follows: </p>
821 <dd>Specifies that the target lays out data in big-endian form. That is, the
822 bits with the most significance have the lowest address location.</dd>
824 <dd>Specifies that hte target lays out data in little-endian form. That is,
825 the bits with the least significance have the lowest address location.</dd>
826 <dt><tt>p:<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
827 <dd>This specifies the <i>size</i> of a pointer and its <i>abi</i> and
828 <i>preferred</i> alignments. All sizes are in bits. Specifying the <i>pref</i>
829 alignment is optional. If omitted, the preceding <tt>:</tt> should be omitted
831 <dt><tt>i<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
832 <dd>This specifies the alignment for an integer type of a given bit
833 <i>size</i>. The value of <i>size</i> must be in the range [1,2^23).</dd>
834 <dt><tt>v<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
835 <dd>This specifies the alignment for a vector type of a given bit
837 <dt><tt>f<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
838 <dd>This specifies the alignment for a floating point type of a given bit
839 <i>size</i>. The value of <i>size</i> must be either 32 (float) or 64
841 <dt><tt>a<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
842 <dd>This specifies the alignment for an aggregate type of a given bit
845 <p>When constructing the data layout for a given target, LLVM starts with a
846 default set of specifications which are then (possibly) overriden by the
847 specifications in the <tt>datalayout</tt> keyword. The default specifications
848 are given in this list:</p>
850 <li><tt>E</tt> - big endian</li>
851 <li><tt>p:32:64:64</tt> - 32-bit pointers with 64-bit alignment</li>
852 <li><tt>i1:8:8</tt> - i1 is 8-bit (byte) aligned</li>
853 <li><tt>i8:8:8</tt> - i8 is 8-bit (byte) aligned</li>
854 <li><tt>i16:16:16</tt> - i16 is 16-bit aligned</li>
855 <li><tt>i32:32:32</tt> - i32 is 32-bit aligned</li>
856 <li><tt>i64:32:64</tt> - i64 has abi alignment of 32-bits but preferred
857 alignment of 64-bits</li>
858 <li><tt>f32:32:32</tt> - float is 32-bit aligned</li>
859 <li><tt>f64:64:64</tt> - double is 64-bit aligned</li>
860 <li><tt>v64:64:64</tt> - 64-bit vector is 64-bit aligned</li>
861 <li><tt>v128:128:128</tt> - 128-bit vector is 128-bit aligned</li>
862 <li><tt>a0:0:1</tt> - aggregates are 8-bit aligned</li>
864 <p>When llvm is determining the alignment for a given type, it uses the
867 <li>If the type sought is an exact match for one of the specifications, that
868 specification is used.</li>
869 <li>If no match is found, and the type sought is an integer type, then the
870 smallest integer type that is larger than the bitwidth of the sought type is
871 used. If none of the specifications are larger than the bitwidth then the the
872 largest integer type is used. For example, given the default specifications
873 above, the i7 type will use the alignment of i8 (next largest) while both
874 i65 and i256 will use the alignment of i64 (largest specified).</li>
875 <li>If no match is found, and the type sought is a vector type, then the
876 largest vector type that is smaller than the sought vector type will be used
877 as a fall back. This happens because <128 x double> can be implemented in
878 terms of 64 <2 x double>, for example.</li>
882 <!-- *********************************************************************** -->
883 <div class="doc_section"> <a name="typesystem">Type System</a> </div>
884 <!-- *********************************************************************** -->
886 <div class="doc_text">
888 <p>The LLVM type system is one of the most important features of the
889 intermediate representation. Being typed enables a number of
890 optimizations to be performed on the IR directly, without having to do
891 extra analyses on the side before the transformation. A strong type
892 system makes it easier to read the generated code and enables novel
893 analyses and transformations that are not feasible to perform on normal
894 three address code representations.</p>
898 <!-- ======================================================================= -->
899 <div class="doc_subsection"> <a name="t_primitive">Primitive Types</a> </div>
900 <div class="doc_text">
901 <p>The primitive types are the fundamental building blocks of the LLVM
902 system. The current set of primitive types is as follows:</p>
904 <table class="layout">
909 <tr><th>Type</th><th>Description</th></tr>
910 <tr><td><tt><a name="t_void">void</a></tt></td><td>No value</td></tr>
911 <tr><td><tt>i8</tt></td><td>8-bit value</td></tr>
912 <tr><td><tt>i32</tt></td><td>32-bit value</td></tr>
913 <tr><td><tt>float</tt></td><td>32-bit floating point value</td></tr>
914 <tr><td><tt>label</tt></td><td>Branch destination</td></tr>
921 <tr><th>Type</th><th>Description</th></tr>
922 <tr><td><tt>i1</tt></td><td>True or False value</td></tr>
923 <tr><td><tt>i16</tt></td><td>16-bit value</td></tr>
924 <tr><td><tt>i64</tt></td><td>64-bit value</td></tr>
925 <tr><td><tt>double</tt></td><td>64-bit floating point value</td></tr>
933 <!-- _______________________________________________________________________ -->
934 <div class="doc_subsubsection"> <a name="t_classifications">Type
935 Classifications</a> </div>
936 <div class="doc_text">
937 <p>These different primitive types fall into a few useful
940 <table border="1" cellspacing="0" cellpadding="4">
942 <tr><th>Classification</th><th>Types</th></tr>
944 <td><a name="t_integer">integer</a></td>
945 <td><tt>i1, i8, i16, i32, i64</tt></td>
948 <td><a name="t_floating">floating point</a></td>
949 <td><tt>float, double</tt></td>
952 <td><a name="t_firstclass">first class</a></td>
953 <td><tt>i1, i8, i16, i32, i64, float, double, <br/>
954 <a href="#t_pointer">pointer</a>,<a href="#t_vector">vector</a></tt>
960 <p>The <a href="#t_firstclass">first class</a> types are perhaps the
961 most important. Values of these types are the only ones which can be
962 produced by instructions, passed as arguments, or used as operands to
963 instructions. This means that all structures and arrays must be
964 manipulated either by pointer or by component.</p>
967 <!-- ======================================================================= -->
968 <div class="doc_subsection"> <a name="t_derived">Derived Types</a> </div>
970 <div class="doc_text">
972 <p>The real power in LLVM comes from the derived types in the system.
973 This is what allows a programmer to represent arrays, functions,
974 pointers, and other useful types. Note that these derived types may be
975 recursive: For example, it is possible to have a two dimensional array.</p>
979 <!-- _______________________________________________________________________ -->
980 <div class="doc_subsubsection"> <a name="t_array">Array Type</a> </div>
982 <div class="doc_text">
986 <p>The array type is a very simple derived type that arranges elements
987 sequentially in memory. The array type requires a size (number of
988 elements) and an underlying data type.</p>
993 [<# elements> x <elementtype>]
996 <p>The number of elements is a constant integer value; elementtype may
997 be any type with a size.</p>
1000 <table class="layout">
1003 <tt>[40 x i32 ]</tt><br/>
1004 <tt>[41 x i32 ]</tt><br/>
1005 <tt>[40 x i8]</tt><br/>
1008 Array of 40 32-bit integer values.<br/>
1009 Array of 41 32-bit integer values.<br/>
1010 Array of 40 8-bit integer values.<br/>
1014 <p>Here are some examples of multidimensional arrays:</p>
1015 <table class="layout">
1018 <tt>[3 x [4 x i32]]</tt><br/>
1019 <tt>[12 x [10 x float]]</tt><br/>
1020 <tt>[2 x [3 x [4 x i16]]]</tt><br/>
1023 3x4 array of 32-bit integer values.<br/>
1024 12x10 array of single precision floating point values.<br/>
1025 2x3x4 array of 16-bit integer values.<br/>
1030 <p>Note that 'variable sized arrays' can be implemented in LLVM with a zero
1031 length array. Normally, accesses past the end of an array are undefined in
1032 LLVM (e.g. it is illegal to access the 5th element of a 3 element array).
1033 As a special case, however, zero length arrays are recognized to be variable
1034 length. This allows implementation of 'pascal style arrays' with the LLVM
1035 type "{ i32, [0 x float]}", for example.</p>
1039 <!-- _______________________________________________________________________ -->
1040 <div class="doc_subsubsection"> <a name="t_function">Function Type</a> </div>
1041 <div class="doc_text">
1043 <p>The function type can be thought of as a function signature. It
1044 consists of a return type and a list of formal parameter types.
1045 Function types are usually used to build virtual function tables
1046 (which are structures of pointers to functions), for indirect function
1047 calls, and when defining a function.</p>
1049 The return type of a function type cannot be an aggregate type.
1052 <pre> <returntype> (<parameter list>)<br></pre>
1053 <p>...where '<tt><parameter list></tt>' is a comma-separated list of type
1054 specifiers. Optionally, the parameter list may include a type <tt>...</tt>,
1055 which indicates that the function takes a variable number of arguments.
1056 Variable argument functions can access their arguments with the <a
1057 href="#int_varargs">variable argument handling intrinsic</a> functions.</p>
1059 <table class="layout">
1061 <td class="left"><tt>i32 (i32)</tt></td>
1062 <td class="left">function taking an <tt>i32</tt>, returning an <tt>i32</tt>
1064 </tr><tr class="layout">
1065 <td class="left"><tt>float (i16 sext, i32 *) *
1067 <td class="left"><a href="#t_pointer">Pointer</a> to a function that takes
1068 an <tt>i16</tt> that should be sign extended and a
1069 <a href="#t_pointer">pointer</a> to <tt>i32</tt>, returning
1072 </tr><tr class="layout">
1073 <td class="left"><tt>i32 (i8*, ...)</tt></td>
1074 <td class="left">A vararg function that takes at least one
1075 <a href="#t_pointer">pointer</a> to <tt>i8 </tt> (char in C),
1076 which returns an integer. This is the signature for <tt>printf</tt> in
1083 <!-- _______________________________________________________________________ -->
1084 <div class="doc_subsubsection"> <a name="t_struct">Structure Type</a> </div>
1085 <div class="doc_text">
1087 <p>The structure type is used to represent a collection of data members
1088 together in memory. The packing of the field types is defined to match
1089 the ABI of the underlying processor. The elements of a structure may
1090 be any type that has a size.</p>
1091 <p>Structures are accessed using '<tt><a href="#i_load">load</a></tt>
1092 and '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a
1093 field with the '<tt><a href="#i_getelementptr">getelementptr</a></tt>'
1096 <pre> { <type list> }<br></pre>
1098 <table class="layout">
1100 <td class="left"><tt>{ i32, i32, i32 }</tt></td>
1101 <td class="left">A triple of three <tt>i32</tt> values</td>
1102 </tr><tr class="layout">
1103 <td class="left"><tt>{ float, i32 (i32) * }</tt></td>
1104 <td class="left">A pair, where the first element is a <tt>float</tt> and the
1105 second element is a <a href="#t_pointer">pointer</a> to a
1106 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
1107 an <tt>i32</tt>.</td>
1112 <!-- _______________________________________________________________________ -->
1113 <div class="doc_subsubsection"> <a name="t_pstruct">Packed Structure Type</a>
1115 <div class="doc_text">
1117 <p>The packed structure type is used to represent a collection of data members
1118 together in memory. There is no padding between fields. Further, the alignment
1119 of a packed structure is 1 byte. The elements of a packed structure may
1120 be any type that has a size.</p>
1121 <p>Structures are accessed using '<tt><a href="#i_load">load</a></tt>
1122 and '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a
1123 field with the '<tt><a href="#i_getelementptr">getelementptr</a></tt>'
1126 <pre> < { <type list> } > <br></pre>
1128 <table class="layout">
1130 <td class="left"><tt>< { i32, i32, i32 } ></tt></td>
1131 <td class="left">A triple of three <tt>i32</tt> values</td>
1132 </tr><tr class="layout">
1133 <td class="left"><tt>< { float, i32 (i32) * } ></tt></td>
1134 <td class="left">A pair, where the first element is a <tt>float</tt> and the
1135 second element is a <a href="#t_pointer">pointer</a> to a
1136 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
1137 an <tt>i32</tt>.</td>
1142 <!-- _______________________________________________________________________ -->
1143 <div class="doc_subsubsection"> <a name="t_pointer">Pointer Type</a> </div>
1144 <div class="doc_text">
1146 <p>As in many languages, the pointer type represents a pointer or
1147 reference to another object, which must live in memory.</p>
1149 <pre> <type> *<br></pre>
1151 <table class="layout">
1154 <tt>[4x i32]*</tt><br/>
1155 <tt>i32 (i32 *) *</tt><br/>
1158 A <a href="#t_pointer">pointer</a> to <a href="#t_array">array</a> of
1159 four <tt>i32</tt> values<br/>
1160 A <a href="#t_pointer">pointer</a> to a <a
1161 href="#t_function">function</a> that takes an <tt>i32*</tt>, returning an
1168 <!-- _______________________________________________________________________ -->
1169 <div class="doc_subsubsection"> <a name="t_vector">Vector Type</a> </div>
1170 <div class="doc_text">
1174 <p>A vector type is a simple derived type that represents a vector
1175 of elements. Vector types are used when multiple primitive data
1176 are operated in parallel using a single instruction (SIMD).
1177 A vector type requires a size (number of
1178 elements) and an underlying primitive data type. Vectors must have a power
1179 of two length (1, 2, 4, 8, 16 ...). Vector types are
1180 considered <a href="#t_firstclass">first class</a>.</p>
1185 < <# elements> x <elementtype> >
1188 <p>The number of elements is a constant integer value; elementtype may
1189 be any integer or floating point type.</p>
1193 <table class="layout">
1196 <tt><4 x i32></tt><br/>
1197 <tt><8 x float></tt><br/>
1198 <tt><2 x i64></tt><br/>
1201 Vector of 4 32-bit integer values.<br/>
1202 Vector of 8 floating-point values.<br/>
1203 Vector of 2 64-bit integer values.<br/>
1209 <!-- _______________________________________________________________________ -->
1210 <div class="doc_subsubsection"> <a name="t_opaque">Opaque Type</a> </div>
1211 <div class="doc_text">
1215 <p>Opaque types are used to represent unknown types in the system. This
1216 corresponds (for example) to the C notion of a foward declared structure type.
1217 In LLVM, opaque types can eventually be resolved to any type (not just a
1218 structure type).</p>
1228 <table class="layout">
1234 An opaque type.<br/>
1241 <!-- *********************************************************************** -->
1242 <div class="doc_section"> <a name="constants">Constants</a> </div>
1243 <!-- *********************************************************************** -->
1245 <div class="doc_text">
1247 <p>LLVM has several different basic types of constants. This section describes
1248 them all and their syntax.</p>
1252 <!-- ======================================================================= -->
1253 <div class="doc_subsection"><a name="simpleconstants">Simple Constants</a></div>
1255 <div class="doc_text">
1258 <dt><b>Boolean constants</b></dt>
1260 <dd>The two strings '<tt>true</tt>' and '<tt>false</tt>' are both valid
1261 constants of the <tt><a href="#t_primitive">i1</a></tt> type.
1264 <dt><b>Integer constants</b></dt>
1266 <dd>Standard integers (such as '4') are constants of the <a
1267 href="#t_integer">integer</a> type. Negative numbers may be used with
1271 <dt><b>Floating point constants</b></dt>
1273 <dd>Floating point constants use standard decimal notation (e.g. 123.421),
1274 exponential notation (e.g. 1.23421e+2), or a more precise hexadecimal
1275 notation (see below). Floating point constants must have a <a
1276 href="#t_floating">floating point</a> type. </dd>
1278 <dt><b>Null pointer constants</b></dt>
1280 <dd>The identifier '<tt>null</tt>' is recognized as a null pointer constant
1281 and must be of <a href="#t_pointer">pointer type</a>.</dd>
1285 <p>The one non-intuitive notation for constants is the optional hexadecimal form
1286 of floating point constants. For example, the form '<tt>double
1287 0x432ff973cafa8000</tt>' is equivalent to (but harder to read than) '<tt>double
1288 4.5e+15</tt>'. The only time hexadecimal floating point constants are required
1289 (and the only time that they are generated by the disassembler) is when a
1290 floating point constant must be emitted but it cannot be represented as a
1291 decimal floating point number. For example, NaN's, infinities, and other
1292 special values are represented in their IEEE hexadecimal format so that
1293 assembly and disassembly do not cause any bits to change in the constants.</p>
1297 <!-- ======================================================================= -->
1298 <div class="doc_subsection"><a name="aggregateconstants">Aggregate Constants</a>
1301 <div class="doc_text">
1302 <p>Aggregate constants arise from aggregation of simple constants
1303 and smaller aggregate constants.</p>
1306 <dt><b>Structure constants</b></dt>
1308 <dd>Structure constants are represented with notation similar to structure
1309 type definitions (a comma separated list of elements, surrounded by braces
1310 (<tt>{}</tt>)). For example: "<tt>{ i32 4, float 17.0, i32* %G }</tt>",
1311 where "<tt>%G</tt>" is declared as "<tt>%G = external global i32</tt>". Structure constants
1312 must have <a href="#t_struct">structure type</a>, and the number and
1313 types of elements must match those specified by the type.
1316 <dt><b>Array constants</b></dt>
1318 <dd>Array constants are represented with notation similar to array type
1319 definitions (a comma separated list of elements, surrounded by square brackets
1320 (<tt>[]</tt>)). For example: "<tt>[ i32 42, i32 11, i32 74 ]</tt>". Array
1321 constants must have <a href="#t_array">array type</a>, and the number and
1322 types of elements must match those specified by the type.
1325 <dt><b>Vector constants</b></dt>
1327 <dd>Vector constants are represented with notation similar to vector type
1328 definitions (a comma separated list of elements, surrounded by
1329 less-than/greater-than's (<tt><></tt>)). For example: "<tt>< i32 42,
1330 i32 11, i32 74, i32 100 ></tt>". Vector constants must have <a
1331 href="#t_vector">vector type</a>, and the number and types of elements must
1332 match those specified by the type.
1335 <dt><b>Zero initialization</b></dt>
1337 <dd>The string '<tt>zeroinitializer</tt>' can be used to zero initialize a
1338 value to zero of <em>any</em> type, including scalar and aggregate types.
1339 This is often used to avoid having to print large zero initializers (e.g. for
1340 large arrays) and is always exactly equivalent to using explicit zero
1347 <!-- ======================================================================= -->
1348 <div class="doc_subsection">
1349 <a name="globalconstants">Global Variable and Function Addresses</a>
1352 <div class="doc_text">
1354 <p>The addresses of <a href="#globalvars">global variables</a> and <a
1355 href="#functionstructure">functions</a> are always implicitly valid (link-time)
1356 constants. These constants are explicitly referenced when the <a
1357 href="#identifiers">identifier for the global</a> is used and always have <a
1358 href="#t_pointer">pointer</a> type. For example, the following is a legal LLVM
1364 %Z = global [2 x i32*] [ i32* %X, i32* %Y ]
1369 <!-- ======================================================================= -->
1370 <div class="doc_subsection"><a name="undefvalues">Undefined Values</a></div>
1371 <div class="doc_text">
1372 <p>The string '<tt>undef</tt>' is recognized as a type-less constant that has
1373 no specific value. Undefined values may be of any type and be used anywhere
1374 a constant is permitted.</p>
1376 <p>Undefined values indicate to the compiler that the program is well defined
1377 no matter what value is used, giving the compiler more freedom to optimize.
1381 <!-- ======================================================================= -->
1382 <div class="doc_subsection"><a name="constantexprs">Constant Expressions</a>
1385 <div class="doc_text">
1387 <p>Constant expressions are used to allow expressions involving other constants
1388 to be used as constants. Constant expressions may be of any <a
1389 href="#t_firstclass">first class</a> type and may involve any LLVM operation
1390 that does not have side effects (e.g. load and call are not supported). The
1391 following is the syntax for constant expressions:</p>
1394 <dt><b><tt>trunc ( CST to TYPE )</tt></b></dt>
1395 <dd>Truncate a constant to another type. The bit size of CST must be larger
1396 than the bit size of TYPE. Both types must be integers.</dd>
1398 <dt><b><tt>zext ( CST to TYPE )</tt></b></dt>
1399 <dd>Zero extend a constant to another type. The bit size of CST must be
1400 smaller or equal to the bit size of TYPE. Both types must be integers.</dd>
1402 <dt><b><tt>sext ( CST to TYPE )</tt></b></dt>
1403 <dd>Sign extend a constant to another type. The bit size of CST must be
1404 smaller or equal to the bit size of TYPE. Both types must be integers.</dd>
1406 <dt><b><tt>fptrunc ( CST to TYPE )</tt></b></dt>
1407 <dd>Truncate a floating point constant to another floating point type. The
1408 size of CST must be larger than the size of TYPE. Both types must be
1409 floating point.</dd>
1411 <dt><b><tt>fpext ( CST to TYPE )</tt></b></dt>
1412 <dd>Floating point extend a constant to another type. The size of CST must be
1413 smaller or equal to the size of TYPE. Both types must be floating point.</dd>
1415 <dt><b><tt>fp2uint ( CST to TYPE )</tt></b></dt>
1416 <dd>Convert a floating point constant to the corresponding unsigned integer
1417 constant. TYPE must be an integer type. CST must be floating point. If the
1418 value won't fit in the integer type, the results are undefined.</dd>
1420 <dt><b><tt>fptosi ( CST to TYPE )</tt></b></dt>
1421 <dd>Convert a floating point constant to the corresponding signed integer
1422 constant. TYPE must be an integer type. CST must be floating point. If the
1423 value won't fit in the integer type, the results are undefined.</dd>
1425 <dt><b><tt>uitofp ( CST to TYPE )</tt></b></dt>
1426 <dd>Convert an unsigned integer constant to the corresponding floating point
1427 constant. TYPE must be floating point. CST must be of integer type. If the
1428 value won't fit in the floating point type, the results are undefined.</dd>
1430 <dt><b><tt>sitofp ( CST to TYPE )</tt></b></dt>
1431 <dd>Convert a signed integer constant to the corresponding floating point
1432 constant. TYPE must be floating point. CST must be of integer type. If the
1433 value won't fit in the floating point type, the results are undefined.</dd>
1435 <dt><b><tt>ptrtoint ( CST to TYPE )</tt></b></dt>
1436 <dd>Convert a pointer typed constant to the corresponding integer constant
1437 TYPE must be an integer type. CST must be of pointer type. The CST value is
1438 zero extended, truncated, or unchanged to make it fit in TYPE.</dd>
1440 <dt><b><tt>inttoptr ( CST to TYPE )</tt></b></dt>
1441 <dd>Convert a integer constant to a pointer constant. TYPE must be a
1442 pointer type. CST must be of integer type. The CST value is zero extended,
1443 truncated, or unchanged to make it fit in a pointer size. This one is
1444 <i>really</i> dangerous!</dd>
1446 <dt><b><tt>bitcast ( CST to TYPE )</tt></b></dt>
1447 <dd>Convert a constant, CST, to another TYPE. The size of CST and TYPE must be
1448 identical (same number of bits). The conversion is done as if the CST value
1449 was stored to memory and read back as TYPE. In other words, no bits change
1450 with this operator, just the type. This can be used for conversion of
1451 vector types to any other type, as long as they have the same bit width. For
1452 pointers it is only valid to cast to another pointer type.
1455 <dt><b><tt>getelementptr ( CSTPTR, IDX0, IDX1, ... )</tt></b></dt>
1457 <dd>Perform the <a href="#i_getelementptr">getelementptr operation</a> on
1458 constants. As with the <a href="#i_getelementptr">getelementptr</a>
1459 instruction, the index list may have zero or more indexes, which are required
1460 to make sense for the type of "CSTPTR".</dd>
1462 <dt><b><tt>select ( COND, VAL1, VAL2 )</tt></b></dt>
1464 <dd>Perform the <a href="#i_select">select operation</a> on
1467 <dt><b><tt>icmp COND ( VAL1, VAL2 )</tt></b></dt>
1468 <dd>Performs the <a href="#i_icmp">icmp operation</a> on constants.</dd>
1470 <dt><b><tt>fcmp COND ( VAL1, VAL2 )</tt></b></dt>
1471 <dd>Performs the <a href="#i_fcmp">fcmp operation</a> on constants.</dd>
1473 <dt><b><tt>extractelement ( VAL, IDX )</tt></b></dt>
1475 <dd>Perform the <a href="#i_extractelement">extractelement
1476 operation</a> on constants.
1478 <dt><b><tt>insertelement ( VAL, ELT, IDX )</tt></b></dt>
1480 <dd>Perform the <a href="#i_insertelement">insertelement
1481 operation</a> on constants.</dd>
1484 <dt><b><tt>shufflevector ( VEC1, VEC2, IDXMASK )</tt></b></dt>
1486 <dd>Perform the <a href="#i_shufflevector">shufflevector
1487 operation</a> on constants.</dd>
1489 <dt><b><tt>OPCODE ( LHS, RHS )</tt></b></dt>
1491 <dd>Perform the specified operation of the LHS and RHS constants. OPCODE may
1492 be any of the <a href="#binaryops">binary</a> or <a href="#bitwiseops">bitwise
1493 binary</a> operations. The constraints on operands are the same as those for
1494 the corresponding instruction (e.g. no bitwise operations on floating point
1495 values are allowed).</dd>
1499 <!-- *********************************************************************** -->
1500 <div class="doc_section"> <a name="othervalues">Other Values</a> </div>
1501 <!-- *********************************************************************** -->
1503 <!-- ======================================================================= -->
1504 <div class="doc_subsection">
1505 <a name="inlineasm">Inline Assembler Expressions</a>
1508 <div class="doc_text">
1511 LLVM supports inline assembler expressions (as opposed to <a href="#moduleasm">
1512 Module-Level Inline Assembly</a>) through the use of a special value. This
1513 value represents the inline assembler as a string (containing the instructions
1514 to emit), a list of operand constraints (stored as a string), and a flag that
1515 indicates whether or not the inline asm expression has side effects. An example
1516 inline assembler expression is:
1520 i32 (i32) asm "bswap $0", "=r,r"
1524 Inline assembler expressions may <b>only</b> be used as the callee operand of
1525 a <a href="#i_call"><tt>call</tt> instruction</a>. Thus, typically we have:
1529 %X = call i32 asm "<a href="#int_bswap">bswap</a> $0", "=r,r"(i32 %Y)
1533 Inline asms with side effects not visible in the constraint list must be marked
1534 as having side effects. This is done through the use of the
1535 '<tt>sideeffect</tt>' keyword, like so:
1539 call void asm sideeffect "eieio", ""()
1542 <p>TODO: The format of the asm and constraints string still need to be
1543 documented here. Constraints on what can be done (e.g. duplication, moving, etc
1544 need to be documented).
1549 <!-- *********************************************************************** -->
1550 <div class="doc_section"> <a name="instref">Instruction Reference</a> </div>
1551 <!-- *********************************************************************** -->
1553 <div class="doc_text">
1555 <p>The LLVM instruction set consists of several different
1556 classifications of instructions: <a href="#terminators">terminator
1557 instructions</a>, <a href="#binaryops">binary instructions</a>,
1558 <a href="#bitwiseops">bitwise binary instructions</a>, <a
1559 href="#memoryops">memory instructions</a>, and <a href="#otherops">other
1560 instructions</a>.</p>
1564 <!-- ======================================================================= -->
1565 <div class="doc_subsection"> <a name="terminators">Terminator
1566 Instructions</a> </div>
1568 <div class="doc_text">
1570 <p>As mentioned <a href="#functionstructure">previously</a>, every
1571 basic block in a program ends with a "Terminator" instruction, which
1572 indicates which block should be executed after the current block is
1573 finished. These terminator instructions typically yield a '<tt>void</tt>'
1574 value: they produce control flow, not values (the one exception being
1575 the '<a href="#i_invoke"><tt>invoke</tt></a>' instruction).</p>
1576 <p>There are six different terminator instructions: the '<a
1577 href="#i_ret"><tt>ret</tt></a>' instruction, the '<a href="#i_br"><tt>br</tt></a>'
1578 instruction, the '<a href="#i_switch"><tt>switch</tt></a>' instruction,
1579 the '<a href="#i_invoke"><tt>invoke</tt></a>' instruction, the '<a
1580 href="#i_unwind"><tt>unwind</tt></a>' instruction, and the '<a
1581 href="#i_unreachable"><tt>unreachable</tt></a>' instruction.</p>
1585 <!-- _______________________________________________________________________ -->
1586 <div class="doc_subsubsection"> <a name="i_ret">'<tt>ret</tt>'
1587 Instruction</a> </div>
1588 <div class="doc_text">
1590 <pre> ret <type> <value> <i>; Return a value from a non-void function</i>
1591 ret void <i>; Return from void function</i>
1594 <p>The '<tt>ret</tt>' instruction is used to return control flow (and a
1595 value) from a function back to the caller.</p>
1596 <p>There are two forms of the '<tt>ret</tt>' instruction: one that
1597 returns a value and then causes control flow, and one that just causes
1598 control flow to occur.</p>
1600 <p>The '<tt>ret</tt>' instruction may return any '<a
1601 href="#t_firstclass">first class</a>' type. Notice that a function is
1602 not <a href="#wellformed">well formed</a> if there exists a '<tt>ret</tt>'
1603 instruction inside of the function that returns a value that does not
1604 match the return type of the function.</p>
1606 <p>When the '<tt>ret</tt>' instruction is executed, control flow
1607 returns back to the calling function's context. If the caller is a "<a
1608 href="#i_call"><tt>call</tt></a>" instruction, execution continues at
1609 the instruction after the call. If the caller was an "<a
1610 href="#i_invoke"><tt>invoke</tt></a>" instruction, execution continues
1611 at the beginning of the "normal" destination block. If the instruction
1612 returns a value, that value shall set the call or invoke instruction's
1615 <pre> ret i32 5 <i>; Return an integer value of 5</i>
1616 ret void <i>; Return from a void function</i>
1619 <!-- _______________________________________________________________________ -->
1620 <div class="doc_subsubsection"> <a name="i_br">'<tt>br</tt>' Instruction</a> </div>
1621 <div class="doc_text">
1623 <pre> br i1 <cond>, label <iftrue>, label <iffalse><br> br label <dest> <i>; Unconditional branch</i>
1626 <p>The '<tt>br</tt>' instruction is used to cause control flow to
1627 transfer to a different basic block in the current function. There are
1628 two forms of this instruction, corresponding to a conditional branch
1629 and an unconditional branch.</p>
1631 <p>The conditional branch form of the '<tt>br</tt>' instruction takes a
1632 single '<tt>i1</tt>' value and two '<tt>label</tt>' values. The
1633 unconditional form of the '<tt>br</tt>' instruction takes a single
1634 '<tt>label</tt>' value as a target.</p>
1636 <p>Upon execution of a conditional '<tt>br</tt>' instruction, the '<tt>i1</tt>'
1637 argument is evaluated. If the value is <tt>true</tt>, control flows
1638 to the '<tt>iftrue</tt>' <tt>label</tt> argument. If "cond" is <tt>false</tt>,
1639 control flows to the '<tt>iffalse</tt>' <tt>label</tt> argument.</p>
1641 <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
1642 href="#i_ret">ret</a> i32 1<br>IfUnequal:<br> <a href="#i_ret">ret</a> i32 0<br></pre>
1644 <!-- _______________________________________________________________________ -->
1645 <div class="doc_subsubsection">
1646 <a name="i_switch">'<tt>switch</tt>' Instruction</a>
1649 <div class="doc_text">
1653 switch <intty> <value>, label <defaultdest> [ <intty> <val>, label <dest> ... ]
1658 <p>The '<tt>switch</tt>' instruction is used to transfer control flow to one of
1659 several different places. It is a generalization of the '<tt>br</tt>'
1660 instruction, allowing a branch to occur to one of many possible
1666 <p>The '<tt>switch</tt>' instruction uses three parameters: an integer
1667 comparison value '<tt>value</tt>', a default '<tt>label</tt>' destination, and
1668 an array of pairs of comparison value constants and '<tt>label</tt>'s. The
1669 table is not allowed to contain duplicate constant entries.</p>
1673 <p>The <tt>switch</tt> instruction specifies a table of values and
1674 destinations. When the '<tt>switch</tt>' instruction is executed, this
1675 table is searched for the given value. If the value is found, control flow is
1676 transfered to the corresponding destination; otherwise, control flow is
1677 transfered to the default destination.</p>
1679 <h5>Implementation:</h5>
1681 <p>Depending on properties of the target machine and the particular
1682 <tt>switch</tt> instruction, this instruction may be code generated in different
1683 ways. For example, it could be generated as a series of chained conditional
1684 branches or with a lookup table.</p>
1689 <i>; Emulate a conditional br instruction</i>
1690 %Val = <a href="#i_zext">zext</a> i1 %value to i32
1691 switch i32 %Val, label %truedest [i32 0, label %falsedest ]
1693 <i>; Emulate an unconditional br instruction</i>
1694 switch i32 0, label %dest [ ]
1696 <i>; Implement a jump table:</i>
1697 switch i32 %val, label %otherwise [ i32 0, label %onzero
1699 i32 2, label %ontwo ]
1703 <!-- _______________________________________________________________________ -->
1704 <div class="doc_subsubsection">
1705 <a name="i_invoke">'<tt>invoke</tt>' Instruction</a>
1708 <div class="doc_text">
1713 <result> = invoke [<a href="#callingconv">cconv</a>] <ptr to function ty> %<function ptr val>(<function args>)
1714 to label <normal label> unwind label <exception label>
1719 <p>The '<tt>invoke</tt>' instruction causes control to transfer to a specified
1720 function, with the possibility of control flow transfer to either the
1721 '<tt>normal</tt>' label or the
1722 '<tt>exception</tt>' label. If the callee function returns with the
1723 "<tt><a href="#i_ret">ret</a></tt>" instruction, control flow will return to the
1724 "normal" label. If the callee (or any indirect callees) returns with the "<a
1725 href="#i_unwind"><tt>unwind</tt></a>" instruction, control is interrupted and
1726 continued at the dynamically nearest "exception" label.</p>
1730 <p>This instruction requires several arguments:</p>
1734 The optional "cconv" marker indicates which <a href="#callingconv">calling
1735 convention</a> the call should use. If none is specified, the call defaults
1736 to using C calling conventions.
1738 <li>'<tt>ptr to function ty</tt>': shall be the signature of the pointer to
1739 function value being invoked. In most cases, this is a direct function
1740 invocation, but indirect <tt>invoke</tt>s are just as possible, branching off
1741 an arbitrary pointer to function value.
1744 <li>'<tt>function ptr val</tt>': An LLVM value containing a pointer to a
1745 function to be invoked. </li>
1747 <li>'<tt>function args</tt>': argument list whose types match the function
1748 signature argument types. If the function signature indicates the function
1749 accepts a variable number of arguments, the extra arguments can be
1752 <li>'<tt>normal label</tt>': the label reached when the called function
1753 executes a '<tt><a href="#i_ret">ret</a></tt>' instruction. </li>
1755 <li>'<tt>exception label</tt>': the label reached when a callee returns with
1756 the <a href="#i_unwind"><tt>unwind</tt></a> instruction. </li>
1762 <p>This instruction is designed to operate as a standard '<tt><a
1763 href="#i_call">call</a></tt>' instruction in most regards. The primary
1764 difference is that it establishes an association with a label, which is used by
1765 the runtime library to unwind the stack.</p>
1767 <p>This instruction is used in languages with destructors to ensure that proper
1768 cleanup is performed in the case of either a <tt>longjmp</tt> or a thrown
1769 exception. Additionally, this is important for implementation of
1770 '<tt>catch</tt>' clauses in high-level languages that support them.</p>
1774 %retval = invoke i32 %Test(i32 15) to label %Continue
1775 unwind label %TestCleanup <i>; {i32}:retval set</i>
1776 %retval = invoke <a href="#callingconv">coldcc</a> i32 %Test(i32 15) to label %Continue
1777 unwind label %TestCleanup <i>; {i32}:retval set</i>
1782 <!-- _______________________________________________________________________ -->
1784 <div class="doc_subsubsection"> <a name="i_unwind">'<tt>unwind</tt>'
1785 Instruction</a> </div>
1787 <div class="doc_text">
1796 <p>The '<tt>unwind</tt>' instruction unwinds the stack, continuing control flow
1797 at the first callee in the dynamic call stack which used an <a
1798 href="#i_invoke"><tt>invoke</tt></a> instruction to perform the call. This is
1799 primarily used to implement exception handling.</p>
1803 <p>The '<tt>unwind</tt>' intrinsic causes execution of the current function to
1804 immediately halt. The dynamic call stack is then searched for the first <a
1805 href="#i_invoke"><tt>invoke</tt></a> instruction on the call stack. Once found,
1806 execution continues at the "exceptional" destination block specified by the
1807 <tt>invoke</tt> instruction. If there is no <tt>invoke</tt> instruction in the
1808 dynamic call chain, undefined behavior results.</p>
1811 <!-- _______________________________________________________________________ -->
1813 <div class="doc_subsubsection"> <a name="i_unreachable">'<tt>unreachable</tt>'
1814 Instruction</a> </div>
1816 <div class="doc_text">
1825 <p>The '<tt>unreachable</tt>' instruction has no defined semantics. This
1826 instruction is used to inform the optimizer that a particular portion of the
1827 code is not reachable. This can be used to indicate that the code after a
1828 no-return function cannot be reached, and other facts.</p>
1832 <p>The '<tt>unreachable</tt>' instruction has no defined semantics.</p>
1837 <!-- ======================================================================= -->
1838 <div class="doc_subsection"> <a name="binaryops">Binary Operations</a> </div>
1839 <div class="doc_text">
1840 <p>Binary operators are used to do most of the computation in a
1841 program. They require two operands, execute an operation on them, and
1842 produce a single value. The operands might represent
1843 multiple data, as is the case with the <a href="#t_vector">vector</a> data type.
1844 The result value of a binary operator is not
1845 necessarily the same type as its operands.</p>
1846 <p>There are several different binary operators:</p>
1848 <!-- _______________________________________________________________________ -->
1849 <div class="doc_subsubsection"> <a name="i_add">'<tt>add</tt>'
1850 Instruction</a> </div>
1851 <div class="doc_text">
1853 <pre> <result> = add <ty> <var1>, <var2> <i>; yields {ty}:result</i>
1856 <p>The '<tt>add</tt>' instruction returns the sum of its two operands.</p>
1858 <p>The two arguments to the '<tt>add</tt>' instruction must be either <a
1859 href="#t_integer">integer</a> or <a href="#t_floating">floating point</a> values.
1860 This instruction can also take <a href="#t_vector">vector</a> versions of the values.
1861 Both arguments must have identical types.</p>
1863 <p>The value produced is the integer or floating point sum of the two
1866 <pre> <result> = add i32 4, %var <i>; yields {i32}:result = 4 + %var</i>
1869 <!-- _______________________________________________________________________ -->
1870 <div class="doc_subsubsection"> <a name="i_sub">'<tt>sub</tt>'
1871 Instruction</a> </div>
1872 <div class="doc_text">
1874 <pre> <result> = sub <ty> <var1>, <var2> <i>; yields {ty}:result</i>
1877 <p>The '<tt>sub</tt>' instruction returns the difference of its two
1879 <p>Note that the '<tt>sub</tt>' instruction is used to represent the '<tt>neg</tt>'
1880 instruction present in most other intermediate representations.</p>
1882 <p>The two arguments to the '<tt>sub</tt>' instruction must be either <a
1883 href="#t_integer">integer</a> or <a href="#t_floating">floating point</a>
1885 This instruction can also take <a href="#t_vector">vector</a> versions of the values.
1886 Both arguments must have identical types.</p>
1888 <p>The value produced is the integer or floating point difference of
1889 the two operands.</p>
1891 <pre> <result> = sub i32 4, %var <i>; yields {i32}:result = 4 - %var</i>
1892 <result> = sub i32 0, %val <i>; yields {i32}:result = -%var</i>
1895 <!-- _______________________________________________________________________ -->
1896 <div class="doc_subsubsection"> <a name="i_mul">'<tt>mul</tt>'
1897 Instruction</a> </div>
1898 <div class="doc_text">
1900 <pre> <result> = mul <ty> <var1>, <var2> <i>; yields {ty}:result</i>
1903 <p>The '<tt>mul</tt>' instruction returns the product of its two
1906 <p>The two arguments to the '<tt>mul</tt>' instruction must be either <a
1907 href="#t_integer">integer</a> or <a href="#t_floating">floating point</a>
1909 This instruction can also take <a href="#t_vector">vector</a> versions of the values.
1910 Both arguments must have identical types.</p>
1912 <p>The value produced is the integer or floating point product of the
1914 <p>Because the operands are the same width, the result of an integer
1915 multiplication is the same whether the operands should be deemed unsigned or
1918 <pre> <result> = mul i32 4, %var <i>; yields {i32}:result = 4 * %var</i>
1921 <!-- _______________________________________________________________________ -->
1922 <div class="doc_subsubsection"> <a name="i_udiv">'<tt>udiv</tt>' Instruction
1924 <div class="doc_text">
1926 <pre> <result> = udiv <ty> <var1>, <var2> <i>; yields {ty}:result</i>
1929 <p>The '<tt>udiv</tt>' instruction returns the quotient of its two
1932 <p>The two arguments to the '<tt>udiv</tt>' instruction must be
1933 <a href="#t_integer">integer</a> values. Both arguments must have identical
1934 types. This instruction can also take <a href="#t_vector">vector</a> versions
1935 of the values in which case the elements must be integers.</p>
1937 <p>The value produced is the unsigned integer quotient of the two operands. This
1938 instruction always performs an unsigned division operation, regardless of
1939 whether the arguments are unsigned or not.</p>
1941 <pre> <result> = udiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
1944 <!-- _______________________________________________________________________ -->
1945 <div class="doc_subsubsection"> <a name="i_sdiv">'<tt>sdiv</tt>' Instruction
1947 <div class="doc_text">
1949 <pre> <result> = sdiv <ty> <var1>, <var2> <i>; yields {ty}:result</i>
1952 <p>The '<tt>sdiv</tt>' instruction returns the quotient of its two
1955 <p>The two arguments to the '<tt>sdiv</tt>' instruction must be
1956 <a href="#t_integer">integer</a> values. Both arguments must have identical
1957 types. This instruction can also take <a href="#t_vector">vector</a> versions
1958 of the values in which case the elements must be integers.</p>
1960 <p>The value produced is the signed integer quotient of the two operands. This
1961 instruction always performs a signed division operation, regardless of whether
1962 the arguments are signed or not.</p>
1964 <pre> <result> = sdiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
1967 <!-- _______________________________________________________________________ -->
1968 <div class="doc_subsubsection"> <a name="i_fdiv">'<tt>fdiv</tt>'
1969 Instruction</a> </div>
1970 <div class="doc_text">
1972 <pre> <result> = fdiv <ty> <var1>, <var2> <i>; yields {ty}:result</i>
1975 <p>The '<tt>fdiv</tt>' instruction returns the quotient of its two
1978 <p>The two arguments to the '<tt>fdiv</tt>' instruction must be
1979 <a href="#t_floating">floating point</a> values. Both arguments must have
1980 identical types. This instruction can also take <a href="#t_vector">vector</a>
1981 versions of floating point values.</p>
1983 <p>The value produced is the floating point quotient of the two operands.</p>
1985 <pre> <result> = fdiv float 4.0, %var <i>; yields {float}:result = 4.0 / %var</i>
1988 <!-- _______________________________________________________________________ -->
1989 <div class="doc_subsubsection"> <a name="i_urem">'<tt>urem</tt>' Instruction</a>
1991 <div class="doc_text">
1993 <pre> <result> = urem <ty> <var1>, <var2> <i>; yields {ty}:result</i>
1996 <p>The '<tt>urem</tt>' instruction returns the remainder from the
1997 unsigned division of its two arguments.</p>
1999 <p>The two arguments to the '<tt>urem</tt>' instruction must be
2000 <a href="#t_integer">integer</a> values. Both arguments must have identical
2003 <p>This instruction returns the unsigned integer <i>remainder</i> of a division.
2004 This instruction always performs an unsigned division to get the remainder,
2005 regardless of whether the arguments are unsigned or not.</p>
2007 <pre> <result> = urem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
2011 <!-- _______________________________________________________________________ -->
2012 <div class="doc_subsubsection"> <a name="i_srem">'<tt>srem</tt>'
2013 Instruction</a> </div>
2014 <div class="doc_text">
2016 <pre> <result> = srem <ty> <var1>, <var2> <i>; yields {ty}:result</i>
2019 <p>The '<tt>srem</tt>' instruction returns the remainder from the
2020 signed division of its two operands.</p>
2022 <p>The two arguments to the '<tt>srem</tt>' instruction must be
2023 <a href="#t_integer">integer</a> values. Both arguments must have identical
2026 <p>This instruction returns the <i>remainder</i> of a division (where the result
2027 has the same sign as the dividend, <tt>var1</tt>), not the <i>modulo</i>
2028 operator (where the result has the same sign as the divisor, <tt>var2</tt>) of
2029 a value. For more information about the difference, see <a
2030 href="http://mathforum.org/dr.math/problems/anne.4.28.99.html">The
2031 Math Forum</a>. For a table of how this is implemented in various languages,
2032 please see <a href="http://en.wikipedia.org/wiki/Modulo_operation">
2033 Wikipedia: modulo operation</a>.</p>
2035 <pre> <result> = srem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
2039 <!-- _______________________________________________________________________ -->
2040 <div class="doc_subsubsection"> <a name="i_frem">'<tt>frem</tt>'
2041 Instruction</a> </div>
2042 <div class="doc_text">
2044 <pre> <result> = frem <ty> <var1>, <var2> <i>; yields {ty}:result</i>
2047 <p>The '<tt>frem</tt>' instruction returns the remainder from the
2048 division of its two operands.</p>
2050 <p>The two arguments to the '<tt>frem</tt>' instruction must be
2051 <a href="#t_floating">floating point</a> values. Both arguments must have
2052 identical types.</p>
2054 <p>This instruction returns the <i>remainder</i> of a division.</p>
2056 <pre> <result> = frem float 4.0, %var <i>; yields {float}:result = 4.0 % %var</i>
2060 <!-- ======================================================================= -->
2061 <div class="doc_subsection"> <a name="bitwiseops">Bitwise Binary
2062 Operations</a> </div>
2063 <div class="doc_text">
2064 <p>Bitwise binary operators are used to do various forms of
2065 bit-twiddling in a program. They are generally very efficient
2066 instructions and can commonly be strength reduced from other
2067 instructions. They require two operands, execute an operation on them,
2068 and produce a single value. The resulting value of the bitwise binary
2069 operators is always the same type as its first operand.</p>
2072 <!-- _______________________________________________________________________ -->
2073 <div class="doc_subsubsection"> <a name="i_shl">'<tt>shl</tt>'
2074 Instruction</a> </div>
2075 <div class="doc_text">
2077 <pre> <result> = shl <ty> <var1>, <var2> <i>; yields {ty}:result</i>
2080 <p>The '<tt>shl</tt>' instruction returns the first operand shifted to
2081 the left a specified number of bits.</p>
2083 <p>Both arguments to the '<tt>shl</tt>' instruction must be the same <a
2084 href="#t_integer">integer</a> type.</p>
2086 <p>The value produced is <tt>var1</tt> * 2<sup><tt>var2</tt></sup>.</p>
2087 <h5>Example:</h5><pre>
2088 <result> = shl i32 4, %var <i>; yields {i32}: 4 << %var</i>
2089 <result> = shl i32 4, 2 <i>; yields {i32}: 16</i>
2090 <result> = shl i32 1, 10 <i>; yields {i32}: 1024</i>
2093 <!-- _______________________________________________________________________ -->
2094 <div class="doc_subsubsection"> <a name="i_lshr">'<tt>lshr</tt>'
2095 Instruction</a> </div>
2096 <div class="doc_text">
2098 <pre> <result> = lshr <ty> <var1>, <var2> <i>; yields {ty}:result</i>
2102 <p>The '<tt>lshr</tt>' instruction (logical shift right) returns the first
2103 operand shifted to the right a specified number of bits with zero fill.</p>
2106 <p>Both arguments to the '<tt>lshr</tt>' instruction must be the same
2107 <a href="#t_integer">integer</a> type.</p>
2110 <p>This instruction always performs a logical shift right operation. The most
2111 significant bits of the result will be filled with zero bits after the
2116 <result> = lshr i32 4, 1 <i>; yields {i32}:result = 2</i>
2117 <result> = lshr i32 4, 2 <i>; yields {i32}:result = 1</i>
2118 <result> = lshr i8 4, 3 <i>; yields {i8}:result = 0</i>
2119 <result> = lshr i8 -2, 1 <i>; yields {i8}:result = 0x7FFFFFFF </i>
2123 <!-- _______________________________________________________________________ -->
2124 <div class="doc_subsubsection"> <a name="i_ashr">'<tt>ashr</tt>'
2125 Instruction</a> </div>
2126 <div class="doc_text">
2129 <pre> <result> = ashr <ty> <var1>, <var2> <i>; yields {ty}:result</i>
2133 <p>The '<tt>ashr</tt>' instruction (arithmetic shift right) returns the first
2134 operand shifted to the right a specified number of bits with sign extension.</p>
2137 <p>Both arguments to the '<tt>ashr</tt>' instruction must be the same
2138 <a href="#t_integer">integer</a> type.</p>
2141 <p>This instruction always performs an arithmetic shift right operation,
2142 The most significant bits of the result will be filled with the sign bit
2143 of <tt>var1</tt>.</p>
2147 <result> = ashr i32 4, 1 <i>; yields {i32}:result = 2</i>
2148 <result> = ashr i32 4, 2 <i>; yields {i32}:result = 1</i>
2149 <result> = ashr i8 4, 3 <i>; yields {i8}:result = 0</i>
2150 <result> = ashr i8 -2, 1 <i>; yields {i8}:result = -1</i>
2154 <!-- _______________________________________________________________________ -->
2155 <div class="doc_subsubsection"> <a name="i_and">'<tt>and</tt>'
2156 Instruction</a> </div>
2157 <div class="doc_text">
2159 <pre> <result> = and <ty> <var1>, <var2> <i>; yields {ty}:result</i>
2162 <p>The '<tt>and</tt>' instruction returns the bitwise logical and of
2163 its two operands.</p>
2165 <p>The two arguments to the '<tt>and</tt>' instruction must be <a
2166 href="#t_integer">integer</a> values. Both arguments must have
2167 identical types.</p>
2169 <p>The truth table used for the '<tt>and</tt>' instruction is:</p>
2171 <div style="align: center">
2172 <table border="1" cellspacing="0" cellpadding="4">
2203 <pre> <result> = and i32 4, %var <i>; yields {i32}:result = 4 & %var</i>
2204 <result> = and i32 15, 40 <i>; yields {i32}:result = 8</i>
2205 <result> = and i32 4, 8 <i>; yields {i32}:result = 0</i>
2208 <!-- _______________________________________________________________________ -->
2209 <div class="doc_subsubsection"> <a name="i_or">'<tt>or</tt>' Instruction</a> </div>
2210 <div class="doc_text">
2212 <pre> <result> = or <ty> <var1>, <var2> <i>; yields {ty}:result</i>
2215 <p>The '<tt>or</tt>' instruction returns the bitwise logical inclusive
2216 or of its two operands.</p>
2218 <p>The two arguments to the '<tt>or</tt>' instruction must be <a
2219 href="#t_integer">integer</a> values. Both arguments must have
2220 identical types.</p>
2222 <p>The truth table used for the '<tt>or</tt>' instruction is:</p>
2224 <div style="align: center">
2225 <table border="1" cellspacing="0" cellpadding="4">
2256 <pre> <result> = or i32 4, %var <i>; yields {i32}:result = 4 | %var</i>
2257 <result> = or i32 15, 40 <i>; yields {i32}:result = 47</i>
2258 <result> = or i32 4, 8 <i>; yields {i32}:result = 12</i>
2261 <!-- _______________________________________________________________________ -->
2262 <div class="doc_subsubsection"> <a name="i_xor">'<tt>xor</tt>'
2263 Instruction</a> </div>
2264 <div class="doc_text">
2266 <pre> <result> = xor <ty> <var1>, <var2> <i>; yields {ty}:result</i>
2269 <p>The '<tt>xor</tt>' instruction returns the bitwise logical exclusive
2270 or of its two operands. The <tt>xor</tt> is used to implement the
2271 "one's complement" operation, which is the "~" operator in C.</p>
2273 <p>The two arguments to the '<tt>xor</tt>' instruction must be <a
2274 href="#t_integer">integer</a> values. Both arguments must have
2275 identical types.</p>
2277 <p>The truth table used for the '<tt>xor</tt>' instruction is:</p>
2279 <div style="align: center">
2280 <table border="1" cellspacing="0" cellpadding="4">
2312 <pre> <result> = xor i32 4, %var <i>; yields {i32}:result = 4 ^ %var</i>
2313 <result> = xor i32 15, 40 <i>; yields {i32}:result = 39</i>
2314 <result> = xor i32 4, 8 <i>; yields {i32}:result = 12</i>
2315 <result> = xor i32 %V, -1 <i>; yields {i32}:result = ~%V</i>
2319 <!-- ======================================================================= -->
2320 <div class="doc_subsection">
2321 <a name="vectorops">Vector Operations</a>
2324 <div class="doc_text">
2326 <p>LLVM supports several instructions to represent vector operations in a
2327 target-independent manner. These instructions cover the element-access and
2328 vector-specific operations needed to process vectors effectively. While LLVM
2329 does directly support these vector operations, many sophisticated algorithms
2330 will want to use target-specific intrinsics to take full advantage of a specific
2335 <!-- _______________________________________________________________________ -->
2336 <div class="doc_subsubsection">
2337 <a name="i_extractelement">'<tt>extractelement</tt>' Instruction</a>
2340 <div class="doc_text">
2345 <result> = extractelement <n x <ty>> <val>, i32 <idx> <i>; yields <ty></i>
2351 The '<tt>extractelement</tt>' instruction extracts a single scalar
2352 element from a vector at a specified index.
2359 The first operand of an '<tt>extractelement</tt>' instruction is a
2360 value of <a href="#t_vector">vector</a> type. The second operand is
2361 an index indicating the position from which to extract the element.
2362 The index may be a variable.</p>
2367 The result is a scalar of the same type as the element type of
2368 <tt>val</tt>. Its value is the value at position <tt>idx</tt> of
2369 <tt>val</tt>. If <tt>idx</tt> exceeds the length of <tt>val</tt>, the
2370 results are undefined.
2376 %result = extractelement <4 x i32> %vec, i32 0 <i>; yields i32</i>
2381 <!-- _______________________________________________________________________ -->
2382 <div class="doc_subsubsection">
2383 <a name="i_insertelement">'<tt>insertelement</tt>' Instruction</a>
2386 <div class="doc_text">
2391 <result> = insertelement <n x <ty>> <val>, <ty> <elt>, i32 <idx> <i>; yields <n x <ty>></i>
2397 The '<tt>insertelement</tt>' instruction inserts a scalar
2398 element into a vector at a specified index.
2405 The first operand of an '<tt>insertelement</tt>' instruction is a
2406 value of <a href="#t_vector">vector</a> type. The second operand is a
2407 scalar value whose type must equal the element type of the first
2408 operand. The third operand is an index indicating the position at
2409 which to insert the value. The index may be a variable.</p>
2414 The result is a vector of the same type as <tt>val</tt>. Its
2415 element values are those of <tt>val</tt> except at position
2416 <tt>idx</tt>, where it gets the value <tt>elt</tt>. If <tt>idx</tt>
2417 exceeds the length of <tt>val</tt>, the results are undefined.
2423 %result = insertelement <4 x i32> %vec, i32 1, i32 0 <i>; yields <4 x i32></i>
2427 <!-- _______________________________________________________________________ -->
2428 <div class="doc_subsubsection">
2429 <a name="i_shufflevector">'<tt>shufflevector</tt>' Instruction</a>
2432 <div class="doc_text">
2437 <result> = shufflevector <n x <ty>> <v1>, <n x <ty>> <v2>, <n x i32> <mask> <i>; yields <n x <ty>></i>
2443 The '<tt>shufflevector</tt>' instruction constructs a permutation of elements
2444 from two input vectors, returning a vector of the same type.
2450 The first two operands of a '<tt>shufflevector</tt>' instruction are vectors
2451 with types that match each other and types that match the result of the
2452 instruction. The third argument is a shuffle mask, which has the same number
2453 of elements as the other vector type, but whose element type is always 'i32'.
2457 The shuffle mask operand is required to be a constant vector with either
2458 constant integer or undef values.
2464 The elements of the two input vectors are numbered from left to right across
2465 both of the vectors. The shuffle mask operand specifies, for each element of
2466 the result vector, which element of the two input registers the result element
2467 gets. The element selector may be undef (meaning "don't care") and the second
2468 operand may be undef if performing a shuffle from only one vector.
2474 %result = shufflevector <4 x i32> %v1, <4 x i32> %v2,
2475 <4 x i32> <i32 0, i32 4, i32 1, i32 5> <i>; yields <4 x i32></i>
2476 %result = shufflevector <4 x i32> %v1, <4 x i32> undef,
2477 <4 x i32> <i32 0, i32 1, i32 2, i32 3> <i>; yields <4 x i32></i> - Identity shuffle.
2482 <!-- ======================================================================= -->
2483 <div class="doc_subsection">
2484 <a name="memoryops">Memory Access and Addressing Operations</a>
2487 <div class="doc_text">
2489 <p>A key design point of an SSA-based representation is how it
2490 represents memory. In LLVM, no memory locations are in SSA form, which
2491 makes things very simple. This section describes how to read, write,
2492 allocate, and free memory in LLVM.</p>
2496 <!-- _______________________________________________________________________ -->
2497 <div class="doc_subsubsection">
2498 <a name="i_malloc">'<tt>malloc</tt>' Instruction</a>
2501 <div class="doc_text">
2506 <result> = malloc <type>[, i32 <NumElements>][, align <alignment>] <i>; yields {type*}:result</i>
2511 <p>The '<tt>malloc</tt>' instruction allocates memory from the system
2512 heap and returns a pointer to it.</p>
2516 <p>The '<tt>malloc</tt>' instruction allocates
2517 <tt>sizeof(<type>)*NumElements</tt>
2518 bytes of memory from the operating system and returns a pointer of the
2519 appropriate type to the program. If "NumElements" is specified, it is the
2520 number of elements allocated. If an alignment is specified, the value result
2521 of the allocation is guaranteed to be aligned to at least that boundary. If
2522 not specified, or if zero, the target can choose to align the allocation on any
2523 convenient boundary.</p>
2525 <p>'<tt>type</tt>' must be a sized type.</p>
2529 <p>Memory is allocated using the system "<tt>malloc</tt>" function, and
2530 a pointer is returned.</p>
2535 %array = malloc [4 x i8 ] <i>; yields {[%4 x i8]*}:array</i>
2537 %size = <a href="#i_add">add</a> i32 2, 2 <i>; yields {i32}:size = i32 4</i>
2538 %array1 = malloc i8, i32 4 <i>; yields {i8*}:array1</i>
2539 %array2 = malloc [12 x i8], i32 %size <i>; yields {[12 x i8]*}:array2</i>
2540 %array3 = malloc i32, i32 4, align 1024 <i>; yields {i32*}:array3</i>
2541 %array4 = malloc i32, align 1024 <i>; yields {i32*}:array4</i>
2545 <!-- _______________________________________________________________________ -->
2546 <div class="doc_subsubsection">
2547 <a name="i_free">'<tt>free</tt>' Instruction</a>
2550 <div class="doc_text">
2555 free <type> <value> <i>; yields {void}</i>
2560 <p>The '<tt>free</tt>' instruction returns memory back to the unused
2561 memory heap to be reallocated in the future.</p>
2565 <p>'<tt>value</tt>' shall be a pointer value that points to a value
2566 that was allocated with the '<tt><a href="#i_malloc">malloc</a></tt>'
2571 <p>Access to the memory pointed to by the pointer is no longer defined
2572 after this instruction executes.</p>
2577 %array = <a href="#i_malloc">malloc</a> [4 x i8] <i>; yields {[4 x i8]*}:array</i>
2578 free [4 x i8]* %array
2582 <!-- _______________________________________________________________________ -->
2583 <div class="doc_subsubsection">
2584 <a name="i_alloca">'<tt>alloca</tt>' Instruction</a>
2587 <div class="doc_text">
2592 <result> = alloca <type>[, i32 <NumElements>][, align <alignment>] <i>; yields {type*}:result</i>
2597 <p>The '<tt>alloca</tt>' instruction allocates memory on the stack frame of the
2598 currently executing function, to be automatically released when this function
2599 returns to its caller.</p>
2603 <p>The '<tt>alloca</tt>' instruction allocates <tt>sizeof(<type>)*NumElements</tt>
2604 bytes of memory on the runtime stack, returning a pointer of the
2605 appropriate type to the program. If "NumElements" is specified, it is the
2606 number of elements allocated. If an alignment is specified, the value result
2607 of the allocation is guaranteed to be aligned to at least that boundary. If
2608 not specified, or if zero, the target can choose to align the allocation on any
2609 convenient boundary.</p>
2611 <p>'<tt>type</tt>' may be any sized type.</p>
2615 <p>Memory is allocated; a pointer is returned. '<tt>alloca</tt>'d
2616 memory is automatically released when the function returns. The '<tt>alloca</tt>'
2617 instruction is commonly used to represent automatic variables that must
2618 have an address available. When the function returns (either with the <tt><a
2619 href="#i_ret">ret</a></tt> or <tt><a href="#i_unwind">unwind</a></tt>
2620 instructions), the memory is reclaimed.</p>
2625 %ptr = alloca i32 <i>; yields {i32*}:ptr</i>
2626 %ptr = alloca i32, i32 4 <i>; yields {i32*}:ptr</i>
2627 %ptr = alloca i32, i32 4, align 1024 <i>; yields {i32*}:ptr</i>
2628 %ptr = alloca i32, align 1024 <i>; yields {i32*}:ptr</i>
2632 <!-- _______________________________________________________________________ -->
2633 <div class="doc_subsubsection"> <a name="i_load">'<tt>load</tt>'
2634 Instruction</a> </div>
2635 <div class="doc_text">
2637 <pre> <result> = load <ty>* <pointer>[, align <alignment>]<br> <result> = volatile load <ty>* <pointer>[, align <alignment>]<br></pre>
2639 <p>The '<tt>load</tt>' instruction is used to read from memory.</p>
2641 <p>The argument to the '<tt>load</tt>' instruction specifies the memory
2642 address from which to load. The pointer must point to a <a
2643 href="#t_firstclass">first class</a> type. If the <tt>load</tt> is
2644 marked as <tt>volatile</tt>, then the optimizer is not allowed to modify
2645 the number or order of execution of this <tt>load</tt> with other
2646 volatile <tt>load</tt> and <tt><a href="#i_store">store</a></tt>
2649 <p>The location of memory pointed to is loaded.</p>
2651 <pre> %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
2653 href="#i_store">store</a> i32 3, i32* %ptr <i>; yields {void}</i>
2654 %val = load i32* %ptr <i>; yields {i32}:val = i32 3</i>
2657 <!-- _______________________________________________________________________ -->
2658 <div class="doc_subsubsection"> <a name="i_store">'<tt>store</tt>'
2659 Instruction</a> </div>
2660 <div class="doc_text">
2662 <pre> store <ty> <value>, <ty>* <pointer>[, align <alignment>] <i>; yields {void}</i>
2663 volatile store <ty> <value>, <ty>* <pointer>[, align <alignment>] <i>; yields {void}</i>
2666 <p>The '<tt>store</tt>' instruction is used to write to memory.</p>
2668 <p>There are two arguments to the '<tt>store</tt>' instruction: a value
2669 to store and an address at which to store it. The type of the '<tt><pointer></tt>'
2670 operand must be a pointer to the type of the '<tt><value></tt>'
2671 operand. If the <tt>store</tt> is marked as <tt>volatile</tt>, then the
2672 optimizer is not allowed to modify the number or order of execution of
2673 this <tt>store</tt> with other volatile <tt>load</tt> and <tt><a
2674 href="#i_store">store</a></tt> instructions.</p>
2676 <p>The contents of memory are updated to contain '<tt><value></tt>'
2677 at the location specified by the '<tt><pointer></tt>' operand.</p>
2679 <pre> %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
2681 href="#i_store">store</a> i32 3, i32* %ptr <i>; yields {void}</i>
2682 %val = load i32* %ptr <i>; yields {i32}:val = i32 3</i>
2686 <!-- _______________________________________________________________________ -->
2687 <div class="doc_subsubsection">
2688 <a name="i_getelementptr">'<tt>getelementptr</tt>' Instruction</a>
2691 <div class="doc_text">
2694 <result> = getelementptr <ty>* <ptrval>{, <ty> <idx>}*
2700 The '<tt>getelementptr</tt>' instruction is used to get the address of a
2701 subelement of an aggregate data structure.</p>
2705 <p>This instruction takes a list of integer operands that indicate what
2706 elements of the aggregate object to index to. The actual types of the arguments
2707 provided depend on the type of the first pointer argument. The
2708 '<tt>getelementptr</tt>' instruction is used to index down through the type
2709 levels of a structure or to a specific index in an array. When indexing into a
2710 structure, only <tt>i32</tt> integer constants are allowed. When indexing
2711 into an array or pointer, only integers of 32 or 64 bits are allowed, and will
2712 be sign extended to 64-bit values.</p>
2714 <p>For example, let's consider a C code fragment and how it gets
2715 compiled to LLVM:</p>
2729 define i32 *foo(struct ST *s) {
2730 return &s[1].Z.B[5][13];
2734 <p>The LLVM code generated by the GCC frontend is:</p>
2737 %RT = type { i8 , [10 x [20 x i32]], i8 }
2738 %ST = type { i32, double, %RT }
2740 define i32* %foo(%ST* %s) {
2742 %reg = getelementptr %ST* %s, i32 1, i32 2, i32 1, i32 5, i32 13
2749 <p>The index types specified for the '<tt>getelementptr</tt>' instruction depend
2750 on the pointer type that is being indexed into. <a href="#t_pointer">Pointer</a>
2751 and <a href="#t_array">array</a> types can use a 32-bit or 64-bit
2752 <a href="#t_integer">integer</a> type but the value will always be sign extended
2753 to 64-bits. <a href="#t_struct">Structure</a> types require <tt>i32</tt>
2754 <b>constants</b>.</p>
2756 <p>In the example above, the first index is indexing into the '<tt>%ST*</tt>'
2757 type, which is a pointer, yielding a '<tt>%ST</tt>' = '<tt>{ i32, double, %RT
2758 }</tt>' type, a structure. The second index indexes into the third element of
2759 the structure, yielding a '<tt>%RT</tt>' = '<tt>{ i8 , [10 x [20 x i32]],
2760 i8 }</tt>' type, another structure. The third index indexes into the second
2761 element of the structure, yielding a '<tt>[10 x [20 x i32]]</tt>' type, an
2762 array. The two dimensions of the array are subscripted into, yielding an
2763 '<tt>i32</tt>' type. The '<tt>getelementptr</tt>' instruction returns a pointer
2764 to this element, thus computing a value of '<tt>i32*</tt>' type.</p>
2766 <p>Note that it is perfectly legal to index partially through a
2767 structure, returning a pointer to an inner element. Because of this,
2768 the LLVM code for the given testcase is equivalent to:</p>
2771 define i32* %foo(%ST* %s) {
2772 %t1 = getelementptr %ST* %s, i32 1 <i>; yields %ST*:%t1</i>
2773 %t2 = getelementptr %ST* %t1, i32 0, i32 2 <i>; yields %RT*:%t2</i>
2774 %t3 = getelementptr %RT* %t2, i32 0, i32 1 <i>; yields [10 x [20 x i32]]*:%t3</i>
2775 %t4 = getelementptr [10 x [20 x i32]]* %t3, i32 0, i32 5 <i>; yields [20 x i32]*:%t4</i>
2776 %t5 = getelementptr [20 x i32]* %t4, i32 0, i32 13 <i>; yields i32*:%t5</i>
2781 <p>Note that it is undefined to access an array out of bounds: array and
2782 pointer indexes must always be within the defined bounds of the array type.
2783 The one exception for this rules is zero length arrays. These arrays are
2784 defined to be accessible as variable length arrays, which requires access
2785 beyond the zero'th element.</p>
2787 <p>The getelementptr instruction is often confusing. For some more insight
2788 into how it works, see <a href="GetElementPtr.html">the getelementptr
2794 <i>; yields [12 x i8]*:aptr</i>
2795 %aptr = getelementptr {i32, [12 x i8]}* %sptr, i64 0, i32 1
2799 <!-- ======================================================================= -->
2800 <div class="doc_subsection"> <a name="convertops">Conversion Operations</a>
2802 <div class="doc_text">
2803 <p>The instructions in this category are the conversion instructions (casting)
2804 which all take a single operand and a type. They perform various bit conversions
2808 <!-- _______________________________________________________________________ -->
2809 <div class="doc_subsubsection">
2810 <a name="i_trunc">'<tt>trunc .. to</tt>' Instruction</a>
2812 <div class="doc_text">
2816 <result> = trunc <ty> <value> to <ty2> <i>; yields ty2</i>
2821 The '<tt>trunc</tt>' instruction truncates its operand to the type <tt>ty2</tt>.
2826 The '<tt>trunc</tt>' instruction takes a <tt>value</tt> to trunc, which must
2827 be an <a href="#t_integer">integer</a> type, and a type that specifies the size
2828 and type of the result, which must be an <a href="#t_integer">integer</a>
2829 type. The bit size of <tt>value</tt> must be larger than the bit size of
2830 <tt>ty2</tt>. Equal sized types are not allowed.</p>
2834 The '<tt>trunc</tt>' instruction truncates the high order bits in <tt>value</tt>
2835 and converts the remaining bits to <tt>ty2</tt>. Since the source size must be
2836 larger than the destination size, <tt>trunc</tt> cannot be a <i>no-op cast</i>.
2837 It will always truncate bits.</p>
2841 %X = trunc i32 257 to i8 <i>; yields i8:1</i>
2842 %Y = trunc i32 123 to i1 <i>; yields i1:true</i>
2843 %Y = trunc i32 122 to i1 <i>; yields i1:false</i>
2847 <!-- _______________________________________________________________________ -->
2848 <div class="doc_subsubsection">
2849 <a name="i_zext">'<tt>zext .. to</tt>' Instruction</a>
2851 <div class="doc_text">
2855 <result> = zext <ty> <value> to <ty2> <i>; yields ty2</i>
2859 <p>The '<tt>zext</tt>' instruction zero extends its operand to type
2864 <p>The '<tt>zext</tt>' instruction takes a value to cast, which must be of
2865 <a href="#t_integer">integer</a> type, and a type to cast it to, which must
2866 also be of <a href="#t_integer">integer</a> type. The bit size of the
2867 <tt>value</tt> must be smaller than the bit size of the destination type,
2871 <p>The <tt>zext</tt> fills the high order bits of the <tt>value</tt> with zero
2872 bits until it reaches the size of the destination type, <tt>ty2</tt>. When the
2873 the operand and the type are the same size, no bit filling is done and the
2874 cast is considered a <i>no-op cast</i> because no bits change (only the type
2877 <p>When zero extending from i1, the result will always be either 0 or 1.</p>
2881 %X = zext i32 257 to i64 <i>; yields i64:257</i>
2882 %Y = zext i1 true to i32 <i>; yields i32:1</i>
2886 <!-- _______________________________________________________________________ -->
2887 <div class="doc_subsubsection">
2888 <a name="i_sext">'<tt>sext .. to</tt>' Instruction</a>
2890 <div class="doc_text">
2894 <result> = sext <ty> <value> to <ty2> <i>; yields ty2</i>
2898 <p>The '<tt>sext</tt>' sign extends <tt>value</tt> to the type <tt>ty2</tt>.</p>
2902 The '<tt>sext</tt>' instruction takes a value to cast, which must be of
2903 <a href="#t_integer">integer</a> type, and a type to cast it to, which must
2904 also be of <a href="#t_integer">integer</a> type. The bit size of the
2905 <tt>value</tt> must be smaller than the bit size of the destination type,
2910 The '<tt>sext</tt>' instruction performs a sign extension by copying the sign
2911 bit (highest order bit) of the <tt>value</tt> until it reaches the bit size of
2912 the type <tt>ty2</tt>. When the the operand and the type are the same size,
2913 no bit filling is done and the cast is considered a <i>no-op cast</i> because
2914 no bits change (only the type changes).</p>
2916 <p>When sign extending from i1, the extension always results in -1 or 0.</p>
2920 %X = sext i8 -1 to i16 <i>; yields i16 :65535</i>
2921 %Y = sext i1 true to i32 <i>; yields i32:-1</i>
2925 <!-- _______________________________________________________________________ -->
2926 <div class="doc_subsubsection">
2927 <a name="i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a>
2930 <div class="doc_text">
2935 <result> = fptrunc <ty> <value> to <ty2> <i>; yields ty2</i>
2939 <p>The '<tt>fptrunc</tt>' instruction truncates <tt>value</tt> to type
2944 <p>The '<tt>fptrunc</tt>' instruction takes a <a href="#t_floating">floating
2945 point</a> value to cast and a <a href="#t_floating">floating point</a> type to
2946 cast it to. The size of <tt>value</tt> must be larger than the size of
2947 <tt>ty2</tt>. This implies that <tt>fptrunc</tt> cannot be used to make a
2948 <i>no-op cast</i>.</p>
2951 <p> The '<tt>fptrunc</tt>' instruction truncates a <tt>value</tt> from a larger
2952 <a href="#t_floating">floating point</a> type to a smaller
2953 <a href="#t_floating">floating point</a> type. If the value cannot fit within
2954 the destination type, <tt>ty2</tt>, then the results are undefined.</p>
2958 %X = fptrunc double 123.0 to float <i>; yields float:123.0</i>
2959 %Y = fptrunc double 1.0E+300 to float <i>; yields undefined</i>
2963 <!-- _______________________________________________________________________ -->
2964 <div class="doc_subsubsection">
2965 <a name="i_fpext">'<tt>fpext .. to</tt>' Instruction</a>
2967 <div class="doc_text">
2971 <result> = fpext <ty> <value> to <ty2> <i>; yields ty2</i>
2975 <p>The '<tt>fpext</tt>' extends a floating point <tt>value</tt> to a larger
2976 floating point value.</p>
2979 <p>The '<tt>fpext</tt>' instruction takes a
2980 <a href="#t_floating">floating point</a> <tt>value</tt> to cast,
2981 and a <a href="#t_floating">floating point</a> type to cast it to. The source
2982 type must be smaller than the destination type.</p>
2985 <p>The '<tt>fpext</tt>' instruction extends the <tt>value</tt> from a smaller
2986 <a href="#t_floating">floating point</a> type to a larger
2987 <a href="#t_floating">floating point</a> type. The <tt>fpext</tt> cannot be
2988 used to make a <i>no-op cast</i> because it always changes bits. Use
2989 <tt>bitcast</tt> to make a <i>no-op cast</i> for a floating point cast.</p>
2993 %X = fpext float 3.1415 to double <i>; yields double:3.1415</i>
2994 %Y = fpext float 1.0 to float <i>; yields float:1.0 (no-op)</i>
2998 <!-- _______________________________________________________________________ -->
2999 <div class="doc_subsubsection">
3000 <a name="i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a>
3002 <div class="doc_text">
3006 <result> = fp2uint <ty> <value> to <ty2> <i>; yields ty2</i>
3010 <p>The '<tt>fp2uint</tt>' converts a floating point <tt>value</tt> to its
3011 unsigned integer equivalent of type <tt>ty2</tt>.
3015 <p>The '<tt>fp2uint</tt>' instruction takes a value to cast, which must be a
3016 <a href="#t_floating">floating point</a> value, and a type to cast it to, which
3017 must be an <a href="#t_integer">integer</a> type.</p>
3020 <p> The '<tt>fp2uint</tt>' instruction converts its
3021 <a href="#t_floating">floating point</a> operand into the nearest (rounding
3022 towards zero) unsigned integer value. If the value cannot fit in <tt>ty2</tt>,
3023 the results are undefined.</p>
3025 <p>When converting to i1, the conversion is done as a comparison against
3026 zero. If the <tt>value</tt> was zero, the i1 result will be <tt>false</tt>.
3027 If the <tt>value</tt> was non-zero, the i1 result will be <tt>true</tt>.</p>
3031 %X = fp2uint double 123.0 to i32 <i>; yields i32:123</i>
3032 %Y = fp2uint float 1.0E+300 to i1 <i>; yields i1:true</i>
3033 %X = fp2uint float 1.04E+17 to i8 <i>; yields undefined:1</i>
3037 <!-- _______________________________________________________________________ -->
3038 <div class="doc_subsubsection">
3039 <a name="i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a>
3041 <div class="doc_text">
3045 <result> = fptosi <ty> <value> to <ty2> <i>; yields ty2</i>
3049 <p>The '<tt>fptosi</tt>' instruction converts
3050 <a href="#t_floating">floating point</a> <tt>value</tt> to type <tt>ty2</tt>.
3055 <p> The '<tt>fptosi</tt>' instruction takes a value to cast, which must be a
3056 <a href="#t_floating">floating point</a> value, and a type to cast it to, which
3057 must also be an <a href="#t_integer">integer</a> type.</p>
3060 <p>The '<tt>fptosi</tt>' instruction converts its
3061 <a href="#t_floating">floating point</a> operand into the nearest (rounding
3062 towards zero) signed integer value. If the value cannot fit in <tt>ty2</tt>,
3063 the results are undefined.</p>
3065 <p>When converting to i1, the conversion is done as a comparison against
3066 zero. If the <tt>value</tt> was zero, the i1 result will be <tt>false</tt>.
3067 If the <tt>value</tt> was non-zero, the i1 result will be <tt>true</tt>.</p>
3071 %X = fptosi double -123.0 to i32 <i>; yields i32:-123</i>
3072 %Y = fptosi float 1.0E-247 to i1 <i>; yields i1:true</i>
3073 %X = fptosi float 1.04E+17 to i8 <i>; yields undefined:1</i>
3077 <!-- _______________________________________________________________________ -->
3078 <div class="doc_subsubsection">
3079 <a name="i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a>
3081 <div class="doc_text">
3085 <result> = uitofp <ty> <value> to <ty2> <i>; yields ty2</i>
3089 <p>The '<tt>uitofp</tt>' instruction regards <tt>value</tt> as an unsigned
3090 integer and converts that value to the <tt>ty2</tt> type.</p>
3094 <p>The '<tt>uitofp</tt>' instruction takes a value to cast, which must be an
3095 <a href="#t_integer">integer</a> value, and a type to cast it to, which must
3096 be a <a href="#t_floating">floating point</a> type.</p>
3099 <p>The '<tt>uitofp</tt>' instruction interprets its operand as an unsigned
3100 integer quantity and converts it to the corresponding floating point value. If
3101 the value cannot fit in the floating point value, the results are undefined.</p>
3106 %X = uitofp i32 257 to float <i>; yields float:257.0</i>
3107 %Y = uitofp i8 -1 to double <i>; yields double:255.0</i>
3111 <!-- _______________________________________________________________________ -->
3112 <div class="doc_subsubsection">
3113 <a name="i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a>
3115 <div class="doc_text">
3119 <result> = sitofp <ty> <value> to <ty2> <i>; yields ty2</i>
3123 <p>The '<tt>sitofp</tt>' instruction regards <tt>value</tt> as a signed
3124 integer and converts that value to the <tt>ty2</tt> type.</p>
3127 <p>The '<tt>sitofp</tt>' instruction takes a value to cast, which must be an
3128 <a href="#t_integer">integer</a> value, and a type to cast it to, which must be
3129 a <a href="#t_floating">floating point</a> type.</p>
3132 <p>The '<tt>sitofp</tt>' instruction interprets its operand as a signed
3133 integer quantity and converts it to the corresponding floating point value. If
3134 the value cannot fit in the floating point value, the results are undefined.</p>
3138 %X = sitofp i32 257 to float <i>; yields float:257.0</i>
3139 %Y = sitofp i8 -1 to double <i>; yields double:-1.0</i>
3143 <!-- _______________________________________________________________________ -->
3144 <div class="doc_subsubsection">
3145 <a name="i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a>
3147 <div class="doc_text">
3151 <result> = ptrtoint <ty> <value> to <ty2> <i>; yields ty2</i>
3155 <p>The '<tt>ptrtoint</tt>' instruction converts the pointer <tt>value</tt> to
3156 the integer type <tt>ty2</tt>.</p>
3159 <p>The '<tt>ptrtoint</tt>' instruction takes a <tt>value</tt> to cast, which
3160 must be a <a href="#t_pointer">pointer</a> value, and a type to cast it to
3161 <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a> type.
3164 <p>The '<tt>ptrtoint</tt>' instruction converts <tt>value</tt> to integer type
3165 <tt>ty2</tt> by interpreting the pointer value as an integer and either
3166 truncating or zero extending that value to the size of the integer type. If
3167 <tt>value</tt> is smaller than <tt>ty2</tt> then a zero extension is done. If
3168 <tt>value</tt> is larger than <tt>ty2</tt> then a truncation is done. If they
3169 are the same size, then nothing is done (<i>no-op cast</i>) other than a type
3174 %X = ptrtoint i32* %X to i8 <i>; yields truncation on 32-bit architecture</i>
3175 %Y = ptrtoint i32* %x to i64 <i>; yields zero extension on 32-bit architecture</i>
3179 <!-- _______________________________________________________________________ -->
3180 <div class="doc_subsubsection">
3181 <a name="i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a>
3183 <div class="doc_text">
3187 <result> = inttoptr <ty> <value> to <ty2> <i>; yields ty2</i>
3191 <p>The '<tt>inttoptr</tt>' instruction converts an integer <tt>value</tt> to
3192 a pointer type, <tt>ty2</tt>.</p>
3195 <p>The '<tt>inttoptr</tt>' instruction takes an <a href="#t_integer">integer</a>
3196 value to cast, and a type to cast it to, which must be a
3197 <a href="#t_pointer">pointer</a> type.
3200 <p>The '<tt>inttoptr</tt>' instruction converts <tt>value</tt> to type
3201 <tt>ty2</tt> by applying either a zero extension or a truncation depending on
3202 the size of the integer <tt>value</tt>. If <tt>value</tt> is larger than the
3203 size of a pointer then a truncation is done. If <tt>value</tt> is smaller than
3204 the size of a pointer then a zero extension is done. If they are the same size,
3205 nothing is done (<i>no-op cast</i>).</p>
3209 %X = inttoptr i32 255 to i32* <i>; yields zero extension on 64-bit architecture</i>
3210 %X = inttoptr i32 255 to i32* <i>; yields no-op on 32-bit architecture</i>
3211 %Y = inttoptr i64 0 to i32* <i>; yields truncation on 32-bit architecture</i>
3215 <!-- _______________________________________________________________________ -->
3216 <div class="doc_subsubsection">
3217 <a name="i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a>
3219 <div class="doc_text">
3223 <result> = bitcast <ty> <value> to <ty2> <i>; yields ty2</i>
3227 <p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
3228 <tt>ty2</tt> without changing any bits.</p>
3231 <p>The '<tt>bitcast</tt>' instruction takes a value to cast, which must be
3232 a first class value, and a type to cast it to, which must also be a <a
3233 href="#t_firstclass">first class</a> type. The bit sizes of <tt>value</tt>
3234 and the destination type, <tt>ty2</tt>, must be identical. If the source
3235 type is a pointer, the destination type must also be a pointer.</p>
3238 <p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
3239 <tt>ty2</tt>. It is always a <i>no-op cast</i> because no bits change with
3240 this conversion. The conversion is done as if the <tt>value</tt> had been
3241 stored to memory and read back as type <tt>ty2</tt>. Pointer types may only be
3242 converted to other pointer types with this instruction. To convert pointers to
3243 other types, use the <a href="#i_inttoptr">inttoptr</a> or
3244 <a href="#i_ptrtoint">ptrtoint</a> instructions first.</p>
3248 %X = bitcast i8 255 to i8 <i>; yields i8 :-1</i>
3249 %Y = bitcast i32* %x to sint* <i>; yields sint*:%x</i>
3250 %Z = bitcast <2xint> %V to i64; <i>; yields i64: %V</i>
3254 <!-- ======================================================================= -->
3255 <div class="doc_subsection"> <a name="otherops">Other Operations</a> </div>
3256 <div class="doc_text">
3257 <p>The instructions in this category are the "miscellaneous"
3258 instructions, which defy better classification.</p>
3261 <!-- _______________________________________________________________________ -->
3262 <div class="doc_subsubsection"><a name="i_icmp">'<tt>icmp</tt>' Instruction</a>
3264 <div class="doc_text">
3266 <pre> <result> = icmp <cond> <ty> <var1>, <var2> <i>; yields {i1}:result</i>
3269 <p>The '<tt>icmp</tt>' instruction returns a boolean value based on comparison
3270 of its two integer operands.</p>
3272 <p>The '<tt>icmp</tt>' instruction takes three operands. The first operand is
3273 the condition code indicating the kind of comparison to perform. It is not
3274 a value, just a keyword. The possible condition code are:
3276 <li><tt>eq</tt>: equal</li>
3277 <li><tt>ne</tt>: not equal </li>
3278 <li><tt>ugt</tt>: unsigned greater than</li>
3279 <li><tt>uge</tt>: unsigned greater or equal</li>
3280 <li><tt>ult</tt>: unsigned less than</li>
3281 <li><tt>ule</tt>: unsigned less or equal</li>
3282 <li><tt>sgt</tt>: signed greater than</li>
3283 <li><tt>sge</tt>: signed greater or equal</li>
3284 <li><tt>slt</tt>: signed less than</li>
3285 <li><tt>sle</tt>: signed less or equal</li>
3287 <p>The remaining two arguments must be <a href="#t_integer">integer</a> or
3288 <a href="#t_pointer">pointer</a> typed. They must also be identical types.</p>
3290 <p>The '<tt>icmp</tt>' compares <tt>var1</tt> and <tt>var2</tt> according to
3291 the condition code given as <tt>cond</tt>. The comparison performed always
3292 yields a <a href="#t_primitive">i1</a> result, as follows:
3294 <li><tt>eq</tt>: yields <tt>true</tt> if the operands are equal,
3295 <tt>false</tt> otherwise. No sign interpretation is necessary or performed.
3297 <li><tt>ne</tt>: yields <tt>true</tt> if the operands are unequal,
3298 <tt>false</tt> otherwise. No sign interpretation is necessary or performed.
3299 <li><tt>ugt</tt>: interprets the operands as unsigned values and yields
3300 <tt>true</tt> if <tt>var1</tt> is greater than <tt>var2</tt>.</li>
3301 <li><tt>uge</tt>: interprets the operands as unsigned values and yields
3302 <tt>true</tt> if <tt>var1</tt> is greater than or equal to <tt>var2</tt>.</li>
3303 <li><tt>ult</tt>: interprets the operands as unsigned values and yields
3304 <tt>true</tt> if <tt>var1</tt> is less than <tt>var2</tt>.</li>
3305 <li><tt>ule</tt>: interprets the operands as unsigned values and yields
3306 <tt>true</tt> if <tt>var1</tt> is less than or equal to <tt>var2</tt>.</li>
3307 <li><tt>sgt</tt>: interprets the operands as signed values and yields
3308 <tt>true</tt> if <tt>var1</tt> is greater than <tt>var2</tt>.</li>
3309 <li><tt>sge</tt>: interprets the operands as signed values and yields
3310 <tt>true</tt> if <tt>var1</tt> is greater than or equal to <tt>var2</tt>.</li>
3311 <li><tt>slt</tt>: interprets the operands as signed values and yields
3312 <tt>true</tt> if <tt>var1</tt> is less than <tt>var2</tt>.</li>
3313 <li><tt>sle</tt>: interprets the operands as signed values and yields
3314 <tt>true</tt> if <tt>var1</tt> is less than or equal to <tt>var2</tt>.</li>
3316 <p>If the operands are <a href="#t_pointer">pointer</a> typed, the pointer
3317 values are compared as if they were integers.</p>
3320 <pre> <result> = icmp eq i32 4, 5 <i>; yields: result=false</i>
3321 <result> = icmp ne float* %X, %X <i>; yields: result=false</i>
3322 <result> = icmp ult i16 4, 5 <i>; yields: result=true</i>
3323 <result> = icmp sgt i16 4, 5 <i>; yields: result=false</i>
3324 <result> = icmp ule i16 -4, 5 <i>; yields: result=false</i>
3325 <result> = icmp sge i16 4, 5 <i>; yields: result=false</i>
3329 <!-- _______________________________________________________________________ -->
3330 <div class="doc_subsubsection"><a name="i_fcmp">'<tt>fcmp</tt>' Instruction</a>
3332 <div class="doc_text">
3334 <pre> <result> = fcmp <cond> <ty> <var1>, <var2> <i>; yields {i1}:result</i>
3337 <p>The '<tt>fcmp</tt>' instruction returns a boolean value based on comparison
3338 of its floating point operands.</p>
3340 <p>The '<tt>fcmp</tt>' instruction takes three operands. The first operand is
3341 the condition code indicating the kind of comparison to perform. It is not
3342 a value, just a keyword. The possible condition code are:
3344 <li><tt>false</tt>: no comparison, always returns false</li>
3345 <li><tt>oeq</tt>: ordered and equal</li>
3346 <li><tt>ogt</tt>: ordered and greater than </li>
3347 <li><tt>oge</tt>: ordered and greater than or equal</li>
3348 <li><tt>olt</tt>: ordered and less than </li>
3349 <li><tt>ole</tt>: ordered and less than or equal</li>
3350 <li><tt>one</tt>: ordered and not equal</li>
3351 <li><tt>ord</tt>: ordered (no nans)</li>
3352 <li><tt>ueq</tt>: unordered or equal</li>
3353 <li><tt>ugt</tt>: unordered or greater than </li>
3354 <li><tt>uge</tt>: unordered or greater than or equal</li>
3355 <li><tt>ult</tt>: unordered or less than </li>
3356 <li><tt>ule</tt>: unordered or less than or equal</li>
3357 <li><tt>une</tt>: unordered or not equal</li>
3358 <li><tt>uno</tt>: unordered (either nans)</li>
3359 <li><tt>true</tt>: no comparison, always returns true</li>
3361 <p><i>Ordered</i> means that neither operand is a QNAN while
3362 <i>unordered</i> means that either operand may be a QNAN.</p>
3363 <p>The <tt>val1</tt> and <tt>val2</tt> arguments must be
3364 <a href="#t_floating">floating point</a> typed. They must have identical
3367 <p>The '<tt>fcmp</tt>' compares <tt>var1</tt> and <tt>var2</tt> according to
3368 the condition code given as <tt>cond</tt>. The comparison performed always
3369 yields a <a href="#t_primitive">i1</a> result, as follows:
3371 <li><tt>false</tt>: always yields <tt>false</tt>, regardless of operands.</li>
3372 <li><tt>oeq</tt>: yields <tt>true</tt> if both operands are not a QNAN and
3373 <tt>var1</tt> is equal to <tt>var2</tt>.</li>
3374 <li><tt>ogt</tt>: yields <tt>true</tt> if both operands are not a QNAN and
3375 <tt>var1</tt> is greather than <tt>var2</tt>.</li>
3376 <li><tt>oge</tt>: yields <tt>true</tt> if both operands are not a QNAN and
3377 <tt>var1</tt> is greater than or equal to <tt>var2</tt>.</li>
3378 <li><tt>olt</tt>: yields <tt>true</tt> if both operands are not a QNAN and
3379 <tt>var1</tt> is less than <tt>var2</tt>.</li>
3380 <li><tt>ole</tt>: yields <tt>true</tt> if both operands are not a QNAN and
3381 <tt>var1</tt> is less than or equal to <tt>var2</tt>.</li>
3382 <li><tt>one</tt>: yields <tt>true</tt> if both operands are not a QNAN and
3383 <tt>var1</tt> is not equal to <tt>var2</tt>.</li>
3384 <li><tt>ord</tt>: yields <tt>true</tt> if both operands are not a QNAN.</li>
3385 <li><tt>ueq</tt>: yields <tt>true</tt> if either operand is a QNAN or
3386 <tt>var1</tt> is equal to <tt>var2</tt>.</li>
3387 <li><tt>ugt</tt>: yields <tt>true</tt> if either operand is a QNAN or
3388 <tt>var1</tt> is greater than <tt>var2</tt>.</li>
3389 <li><tt>uge</tt>: yields <tt>true</tt> if either operand is a QNAN or
3390 <tt>var1</tt> is greater than or equal to <tt>var2</tt>.</li>
3391 <li><tt>ult</tt>: yields <tt>true</tt> if either operand is a QNAN or
3392 <tt>var1</tt> is less than <tt>var2</tt>.</li>
3393 <li><tt>ule</tt>: yields <tt>true</tt> if either operand is a QNAN or
3394 <tt>var1</tt> is less than or equal to <tt>var2</tt>.</li>
3395 <li><tt>une</tt>: yields <tt>true</tt> if either operand is a QNAN or
3396 <tt>var1</tt> is not equal to <tt>var2</tt>.</li>
3397 <li><tt>uno</tt>: yields <tt>true</tt> if either operand is a QNAN.</li>
3398 <li><tt>true</tt>: always yields <tt>true</tt>, regardless of operands.</li>
3402 <pre> <result> = fcmp oeq float 4.0, 5.0 <i>; yields: result=false</i>
3403 <result> = icmp one float 4.0, 5.0 <i>; yields: result=true</i>
3404 <result> = icmp olt float 4.0, 5.0 <i>; yields: result=true</i>
3405 <result> = icmp ueq double 1.0, 2.0 <i>; yields: result=false</i>
3409 <!-- _______________________________________________________________________ -->
3410 <div class="doc_subsubsection"> <a name="i_phi">'<tt>phi</tt>'
3411 Instruction</a> </div>
3412 <div class="doc_text">
3414 <pre> <result> = phi <ty> [ <val0>, <label0>], ...<br></pre>
3416 <p>The '<tt>phi</tt>' instruction is used to implement the φ node in
3417 the SSA graph representing the function.</p>
3419 <p>The type of the incoming values is specified with the first type
3420 field. After this, the '<tt>phi</tt>' instruction takes a list of pairs
3421 as arguments, with one pair for each predecessor basic block of the
3422 current block. Only values of <a href="#t_firstclass">first class</a>
3423 type may be used as the value arguments to the PHI node. Only labels
3424 may be used as the label arguments.</p>
3425 <p>There must be no non-phi instructions between the start of a basic
3426 block and the PHI instructions: i.e. PHI instructions must be first in
3429 <p>At runtime, the '<tt>phi</tt>' instruction logically takes on the value
3430 specified by the pair corresponding to the predecessor basic block that executed
3431 just prior to the current block.</p>
3433 <pre>Loop: ; Infinite loop that counts from 0 on up...<br> %indvar = phi i32 [ 0, %LoopHeader ], [ %nextindvar, %Loop ]<br> %nextindvar = add i32 %indvar, 1<br> br label %Loop<br></pre>
3436 <!-- _______________________________________________________________________ -->
3437 <div class="doc_subsubsection">
3438 <a name="i_select">'<tt>select</tt>' Instruction</a>
3441 <div class="doc_text">
3446 <result> = select i1 <cond>, <ty> <val1>, <ty> <val2> <i>; yields ty</i>
3452 The '<tt>select</tt>' instruction is used to choose one value based on a
3453 condition, without branching.
3460 The '<tt>select</tt>' instruction requires a boolean value indicating the condition, and two values of the same <a href="#t_firstclass">first class</a> type.
3466 If the boolean condition evaluates to true, the instruction returns the first
3467 value argument; otherwise, it returns the second value argument.
3473 %X = select i1 true, i8 17, i8 42 <i>; yields i8:17</i>
3478 <!-- _______________________________________________________________________ -->
3479 <div class="doc_subsubsection">
3480 <a name="i_call">'<tt>call</tt>' Instruction</a>
3483 <div class="doc_text">
3487 <result> = [tail] call [<a href="#callingconv">cconv</a>] <ty>* <fnptrval>(<param list>)
3492 <p>The '<tt>call</tt>' instruction represents a simple function call.</p>
3496 <p>This instruction requires several arguments:</p>
3500 <p>The optional "tail" marker indicates whether the callee function accesses
3501 any allocas or varargs in the caller. If the "tail" marker is present, the
3502 function call is eligible for tail call optimization. Note that calls may
3503 be marked "tail" even if they do not occur before a <a
3504 href="#i_ret"><tt>ret</tt></a> instruction.
3507 <p>The optional "cconv" marker indicates which <a href="#callingconv">calling
3508 convention</a> the call should use. If none is specified, the call defaults
3509 to using C calling conventions.
3512 <p>'<tt>ty</tt>': shall be the signature of the pointer to function value
3513 being invoked. The argument types must match the types implied by this
3514 signature. This type can be omitted if the function is not varargs and
3515 if the function type does not return a pointer to a function.</p>
3518 <p>'<tt>fnptrval</tt>': An LLVM value containing a pointer to a function to
3519 be invoked. In most cases, this is a direct function invocation, but
3520 indirect <tt>call</tt>s are just as possible, calling an arbitrary pointer
3521 to function value.</p>
3524 <p>'<tt>function args</tt>': argument list whose types match the
3525 function signature argument types. All arguments must be of
3526 <a href="#t_firstclass">first class</a> type. If the function signature
3527 indicates the function accepts a variable number of arguments, the extra
3528 arguments can be specified.</p>
3534 <p>The '<tt>call</tt>' instruction is used to cause control flow to
3535 transfer to a specified function, with its incoming arguments bound to
3536 the specified values. Upon a '<tt><a href="#i_ret">ret</a></tt>'
3537 instruction in the called function, control flow continues with the
3538 instruction after the function call, and the return value of the
3539 function is bound to the result argument. This is a simpler case of
3540 the <a href="#i_invoke">invoke</a> instruction.</p>
3545 %retval = call i32 %test(i32 %argc)
3546 call i32(i8 *, ...) *%printf(i8 * %msg, i32 12, i8 42);
3547 %X = tail call i32 %foo()
3548 %Y = tail call <a href="#callingconv">fastcc</a> i32 %foo()
3553 <!-- _______________________________________________________________________ -->
3554 <div class="doc_subsubsection">
3555 <a name="i_va_arg">'<tt>va_arg</tt>' Instruction</a>
3558 <div class="doc_text">
3563 <resultval> = va_arg <va_list*> <arglist>, <argty>
3568 <p>The '<tt>va_arg</tt>' instruction is used to access arguments passed through
3569 the "variable argument" area of a function call. It is used to implement the
3570 <tt>va_arg</tt> macro in C.</p>
3574 <p>This instruction takes a <tt>va_list*</tt> value and the type of
3575 the argument. It returns a value of the specified argument type and
3576 increments the <tt>va_list</tt> to point to the next argument. The
3577 actual type of <tt>va_list</tt> is target specific.</p>
3581 <p>The '<tt>va_arg</tt>' instruction loads an argument of the specified
3582 type from the specified <tt>va_list</tt> and causes the
3583 <tt>va_list</tt> to point to the next argument. For more information,
3584 see the variable argument handling <a href="#int_varargs">Intrinsic
3587 <p>It is legal for this instruction to be called in a function which does not
3588 take a variable number of arguments, for example, the <tt>vfprintf</tt>
3591 <p><tt>va_arg</tt> is an LLVM instruction instead of an <a
3592 href="#intrinsics">intrinsic function</a> because it takes a type as an
3597 <p>See the <a href="#int_varargs">variable argument processing</a> section.</p>
3601 <!-- *********************************************************************** -->
3602 <div class="doc_section"> <a name="intrinsics">Intrinsic Functions</a> </div>
3603 <!-- *********************************************************************** -->
3605 <div class="doc_text">
3607 <p>LLVM supports the notion of an "intrinsic function". These functions have
3608 well known names and semantics and are required to follow certain restrictions.
3609 Overall, these intrinsics represent an extension mechanism for the LLVM
3610 language that does not require changing all of the transformations in LLVM when
3611 adding to the language (or the bytecode reader/writer, the parser, etc...).</p>
3613 <p>Intrinsic function names must all start with an "<tt>llvm.</tt>" prefix. This
3614 prefix is reserved in LLVM for intrinsic names; thus, function names may not
3615 begin with this prefix. Intrinsic functions must always be external functions:
3616 you cannot define the body of intrinsic functions. Intrinsic functions may
3617 only be used in call or invoke instructions: it is illegal to take the address
3618 of an intrinsic function. Additionally, because intrinsic functions are part
3619 of the LLVM language, it is required if any are added that they be documented
3622 <p>Some intrinsic functions can be overloaded, i.e., the intrinsic represents
3623 a family of functions that perform the same operation but on different data
3624 types. This is most frequent with the integer types. Since LLVM can represent
3625 over 8 million different integer types, there is a way to declare an intrinsic
3626 that can be overloaded based on its arguments. Such an intrinsic will have the
3627 names of its argument types encoded into its function name, each
3628 preceded by a period. For example, the <tt>llvm.ctpop</tt> function can take an
3629 integer of any width. This leads to a family of functions such as
3630 <tt>i32 @llvm.ctpop.i8(i8 %val)</tt> and <tt>i32 @llvm.ctpop.i29(i29 %val)</tt>.
3634 <p>To learn how to add an intrinsic function, please see the
3635 <a href="ExtendingLLVM.html">Extending LLVM Guide</a>.
3640 <!-- ======================================================================= -->
3641 <div class="doc_subsection">
3642 <a name="int_varargs">Variable Argument Handling Intrinsics</a>
3645 <div class="doc_text">
3647 <p>Variable argument support is defined in LLVM with the <a
3648 href="#i_va_arg"><tt>va_arg</tt></a> instruction and these three
3649 intrinsic functions. These functions are related to the similarly
3650 named macros defined in the <tt><stdarg.h></tt> header file.</p>
3652 <p>All of these functions operate on arguments that use a
3653 target-specific value type "<tt>va_list</tt>". The LLVM assembly
3654 language reference manual does not define what this type is, so all
3655 transformations should be prepared to handle these functions regardless of
3658 <p>This example shows how the <a href="#i_va_arg"><tt>va_arg</tt></a>
3659 instruction and the variable argument handling intrinsic functions are
3663 define i32 @test(i32 %X, ...) {
3664 ; Initialize variable argument processing
3666 %ap2 = bitcast i8** %ap to i8*
3667 call void @llvm.va_start(i8* %ap2)
3669 ; Read a single integer argument
3670 %tmp = va_arg i8** %ap, i32
3672 ; Demonstrate usage of llvm.va_copy and llvm.va_end
3674 %aq2 = bitcast i8** %aq to i8*
3675 call void @llvm.va_copy(i8* %aq2, i8* %ap2)
3676 call void @llvm.va_end(i8* %aq2)
3678 ; Stop processing of arguments.
3679 call void @llvm.va_end(i8* %ap2)
3683 declare void @llvm.va_start(i8*)
3684 declare void @llvm.va_copy(i8*, i8*)
3685 declare void @llvm.va_end(i8*)
3689 <!-- _______________________________________________________________________ -->
3690 <div class="doc_subsubsection">
3691 <a name="int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a>
3695 <div class="doc_text">
3697 <pre> declare void %llvm.va_start(i8* <arglist>)<br></pre>
3699 <P>The '<tt>llvm.va_start</tt>' intrinsic initializes
3700 <tt>*<arglist></tt> for subsequent use by <tt><a
3701 href="#i_va_arg">va_arg</a></tt>.</p>
3705 <P>The argument is a pointer to a <tt>va_list</tt> element to initialize.</p>
3709 <P>The '<tt>llvm.va_start</tt>' intrinsic works just like the <tt>va_start</tt>
3710 macro available in C. In a target-dependent way, it initializes the
3711 <tt>va_list</tt> element to which the argument points, so that the next call to
3712 <tt>va_arg</tt> will produce the first variable argument passed to the function.
3713 Unlike the C <tt>va_start</tt> macro, this intrinsic does not need to know the
3714 last argument of the function as the compiler can figure that out.</p>
3718 <!-- _______________________________________________________________________ -->
3719 <div class="doc_subsubsection">
3720 <a name="int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a>
3723 <div class="doc_text">
3725 <pre> declare void @llvm.va_end(i8* <arglist>)<br></pre>
3728 <p>The '<tt>llvm.va_end</tt>' intrinsic destroys <tt>*<arglist></tt>,
3729 which has been initialized previously with <tt><a href="#int_va_start">llvm.va_start</a></tt>
3730 or <tt><a href="#i_va_copy">llvm.va_copy</a></tt>.</p>
3734 <p>The argument is a pointer to a <tt>va_list</tt> to destroy.</p>
3738 <p>The '<tt>llvm.va_end</tt>' intrinsic works just like the <tt>va_end</tt>
3739 macro available in C. In a target-dependent way, it destroys the
3740 <tt>va_list</tt> element to which the argument points. Calls to <a
3741 href="#int_va_start"><tt>llvm.va_start</tt></a> and <a href="#int_va_copy">
3742 <tt>llvm.va_copy</tt></a> must be matched exactly with calls to
3743 <tt>llvm.va_end</tt>.</p>
3747 <!-- _______________________________________________________________________ -->
3748 <div class="doc_subsubsection">
3749 <a name="int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a>
3752 <div class="doc_text">
3757 declare void @llvm.va_copy(i8* <destarglist>, i8* <srcarglist>)
3762 <p>The '<tt>llvm.va_copy</tt>' intrinsic copies the current argument position
3763 from the source argument list to the destination argument list.</p>
3767 <p>The first argument is a pointer to a <tt>va_list</tt> element to initialize.
3768 The second argument is a pointer to a <tt>va_list</tt> element to copy from.</p>
3773 <p>The '<tt>llvm.va_copy</tt>' intrinsic works just like the <tt>va_copy</tt>
3774 macro available in C. In a target-dependent way, it copies the source
3775 <tt>va_list</tt> element into the destination <tt>va_list</tt> element. This
3776 intrinsic is necessary because the <tt><a href="#int_va_start">
3777 llvm.va_start</a></tt> intrinsic may be arbitrarily complex and require, for
3778 example, memory allocation.</p>
3782 <!-- ======================================================================= -->
3783 <div class="doc_subsection">
3784 <a name="int_gc">Accurate Garbage Collection Intrinsics</a>
3787 <div class="doc_text">
3790 LLVM support for <a href="GarbageCollection.html">Accurate Garbage
3791 Collection</a> requires the implementation and generation of these intrinsics.
3792 These intrinsics allow identification of <a href="#int_gcroot">GC roots on the
3793 stack</a>, as well as garbage collector implementations that require <a
3794 href="#int_gcread">read</a> and <a href="#int_gcwrite">write</a> barriers.
3795 Front-ends for type-safe garbage collected languages should generate these
3796 intrinsics to make use of the LLVM garbage collectors. For more details, see <a
3797 href="GarbageCollection.html">Accurate Garbage Collection with LLVM</a>.
3801 <!-- _______________________________________________________________________ -->
3802 <div class="doc_subsubsection">
3803 <a name="int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a>
3806 <div class="doc_text">
3811 declare void @llvm.gcroot(<ty>** %ptrloc, <ty2>* %metadata)
3816 <p>The '<tt>llvm.gcroot</tt>' intrinsic declares the existence of a GC root to
3817 the code generator, and allows some metadata to be associated with it.</p>
3821 <p>The first argument specifies the address of a stack object that contains the
3822 root pointer. The second pointer (which must be either a constant or a global
3823 value address) contains the meta-data to be associated with the root.</p>
3827 <p>At runtime, a call to this intrinsics stores a null pointer into the "ptrloc"
3828 location. At compile-time, the code generator generates information to allow
3829 the runtime to find the pointer at GC safe points.
3835 <!-- _______________________________________________________________________ -->
3836 <div class="doc_subsubsection">
3837 <a name="int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a>
3840 <div class="doc_text">
3845 declare i8 * @llvm.gcread(i8 * %ObjPtr, i8 ** %Ptr)
3850 <p>The '<tt>llvm.gcread</tt>' intrinsic identifies reads of references from heap
3851 locations, allowing garbage collector implementations that require read
3856 <p>The second argument is the address to read from, which should be an address
3857 allocated from the garbage collector. The first object is a pointer to the
3858 start of the referenced object, if needed by the language runtime (otherwise
3863 <p>The '<tt>llvm.gcread</tt>' intrinsic has the same semantics as a load
3864 instruction, but may be replaced with substantially more complex code by the
3865 garbage collector runtime, as needed.</p>
3870 <!-- _______________________________________________________________________ -->
3871 <div class="doc_subsubsection">
3872 <a name="int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a>
3875 <div class="doc_text">
3880 declare void @llvm.gcwrite(i8 * %P1, i8 * %Obj, i8 ** %P2)
3885 <p>The '<tt>llvm.gcwrite</tt>' intrinsic identifies writes of references to heap
3886 locations, allowing garbage collector implementations that require write
3887 barriers (such as generational or reference counting collectors).</p>
3891 <p>The first argument is the reference to store, the second is the start of the
3892 object to store it to, and the third is the address of the field of Obj to
3893 store to. If the runtime does not require a pointer to the object, Obj may be
3898 <p>The '<tt>llvm.gcwrite</tt>' intrinsic has the same semantics as a store
3899 instruction, but may be replaced with substantially more complex code by the
3900 garbage collector runtime, as needed.</p>
3906 <!-- ======================================================================= -->
3907 <div class="doc_subsection">
3908 <a name="int_codegen">Code Generator Intrinsics</a>
3911 <div class="doc_text">
3913 These intrinsics are provided by LLVM to expose special features that may only
3914 be implemented with code generator support.
3919 <!-- _______________________________________________________________________ -->
3920 <div class="doc_subsubsection">
3921 <a name="int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a>
3924 <div class="doc_text">
3928 declare i8 *@llvm.returnaddress(i32 <level>)
3934 The '<tt>llvm.returnaddress</tt>' intrinsic attempts to compute a
3935 target-specific value indicating the return address of the current function
3936 or one of its callers.
3942 The argument to this intrinsic indicates which function to return the address
3943 for. Zero indicates the calling function, one indicates its caller, etc. The
3944 argument is <b>required</b> to be a constant integer value.
3950 The '<tt>llvm.returnaddress</tt>' intrinsic either returns a pointer indicating
3951 the return address of the specified call frame, or zero if it cannot be
3952 identified. The value returned by this intrinsic is likely to be incorrect or 0
3953 for arguments other than zero, so it should only be used for debugging purposes.
3957 Note that calling this intrinsic does not prevent function inlining or other
3958 aggressive transformations, so the value returned may not be that of the obvious
3959 source-language caller.
3964 <!-- _______________________________________________________________________ -->
3965 <div class="doc_subsubsection">
3966 <a name="int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a>
3969 <div class="doc_text">
3973 declare i8 *@llvm.frameaddress(i32 <level>)
3979 The '<tt>llvm.frameaddress</tt>' intrinsic attempts to return the
3980 target-specific frame pointer value for the specified stack frame.
3986 The argument to this intrinsic indicates which function to return the frame
3987 pointer for. Zero indicates the calling function, one indicates its caller,
3988 etc. The argument is <b>required</b> to be a constant integer value.
3994 The '<tt>llvm.frameaddress</tt>' intrinsic either returns a pointer indicating
3995 the frame address of the specified call frame, or zero if it cannot be
3996 identified. The value returned by this intrinsic is likely to be incorrect or 0
3997 for arguments other than zero, so it should only be used for debugging purposes.
4001 Note that calling this intrinsic does not prevent function inlining or other
4002 aggressive transformations, so the value returned may not be that of the obvious
4003 source-language caller.
4007 <!-- _______________________________________________________________________ -->
4008 <div class="doc_subsubsection">
4009 <a name="int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a>
4012 <div class="doc_text">
4016 declare i8 *@llvm.stacksave()
4022 The '<tt>llvm.stacksave</tt>' intrinsic is used to remember the current state of
4023 the function stack, for use with <a href="#int_stackrestore">
4024 <tt>llvm.stackrestore</tt></a>. This is useful for implementing language
4025 features like scoped automatic variable sized arrays in C99.
4031 This intrinsic returns a opaque pointer value that can be passed to <a
4032 href="#int_stackrestore"><tt>llvm.stackrestore</tt></a>. When an
4033 <tt>llvm.stackrestore</tt> intrinsic is executed with a value saved from
4034 <tt>llvm.stacksave</tt>, it effectively restores the state of the stack to the
4035 state it was in when the <tt>llvm.stacksave</tt> intrinsic executed. In
4036 practice, this pops any <a href="#i_alloca">alloca</a> blocks from the stack
4037 that were allocated after the <tt>llvm.stacksave</tt> was executed.
4042 <!-- _______________________________________________________________________ -->
4043 <div class="doc_subsubsection">
4044 <a name="int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a>
4047 <div class="doc_text">
4051 declare void @llvm.stackrestore(i8 * %ptr)
4057 The '<tt>llvm.stackrestore</tt>' intrinsic is used to restore the state of
4058 the function stack to the state it was in when the corresponding <a
4059 href="#int_stacksave"><tt>llvm.stacksave</tt></a> intrinsic executed. This is
4060 useful for implementing language features like scoped automatic variable sized
4067 See the description for <a href="#int_stacksave"><tt>llvm.stacksave</tt></a>.
4073 <!-- _______________________________________________________________________ -->
4074 <div class="doc_subsubsection">
4075 <a name="int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a>
4078 <div class="doc_text">
4082 declare void @llvm.prefetch(i8 * <address>,
4083 i32 <rw>, i32 <locality>)
4090 The '<tt>llvm.prefetch</tt>' intrinsic is a hint to the code generator to insert
4091 a prefetch instruction if supported; otherwise, it is a noop. Prefetches have
4093 effect on the behavior of the program but can change its performance
4100 <tt>address</tt> is the address to be prefetched, <tt>rw</tt> is the specifier
4101 determining if the fetch should be for a read (0) or write (1), and
4102 <tt>locality</tt> is a temporal locality specifier ranging from (0) - no
4103 locality, to (3) - extremely local keep in cache. The <tt>rw</tt> and
4104 <tt>locality</tt> arguments must be constant integers.
4110 This intrinsic does not modify the behavior of the program. In particular,
4111 prefetches cannot trap and do not produce a value. On targets that support this
4112 intrinsic, the prefetch can provide hints to the processor cache for better
4118 <!-- _______________________________________________________________________ -->
4119 <div class="doc_subsubsection">
4120 <a name="int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a>
4123 <div class="doc_text">
4127 declare void @llvm.pcmarker( i32 <id> )
4134 The '<tt>llvm.pcmarker</tt>' intrinsic is a method to export a Program Counter
4136 code to simulators and other tools. The method is target specific, but it is
4137 expected that the marker will use exported symbols to transmit the PC of the marker.
4138 The marker makes no guarantees that it will remain with any specific instruction
4139 after optimizations. It is possible that the presence of a marker will inhibit
4140 optimizations. The intended use is to be inserted after optimizations to allow
4141 correlations of simulation runs.
4147 <tt>id</tt> is a numerical id identifying the marker.
4153 This intrinsic does not modify the behavior of the program. Backends that do not
4154 support this intrinisic may ignore it.
4159 <!-- _______________________________________________________________________ -->
4160 <div class="doc_subsubsection">
4161 <a name="int_readcyclecounter">'<tt>llvm.readcyclecounter</tt>' Intrinsic</a>
4164 <div class="doc_text">
4168 declare i64 @llvm.readcyclecounter( )
4175 The '<tt>llvm.readcyclecounter</tt>' intrinsic provides access to the cycle
4176 counter register (or similar low latency, high accuracy clocks) on those targets
4177 that support it. On X86, it should map to RDTSC. On Alpha, it should map to RPCC.
4178 As the backing counters overflow quickly (on the order of 9 seconds on alpha), this
4179 should only be used for small timings.
4185 When directly supported, reading the cycle counter should not modify any memory.
4186 Implementations are allowed to either return a application specific value or a
4187 system wide value. On backends without support, this is lowered to a constant 0.
4192 <!-- ======================================================================= -->
4193 <div class="doc_subsection">
4194 <a name="int_libc">Standard C Library Intrinsics</a>
4197 <div class="doc_text">
4199 LLVM provides intrinsics for a few important standard C library functions.
4200 These intrinsics allow source-language front-ends to pass information about the
4201 alignment of the pointer arguments to the code generator, providing opportunity
4202 for more efficient code generation.
4207 <!-- _______________________________________________________________________ -->
4208 <div class="doc_subsubsection">
4209 <a name="int_memcpy">'<tt>llvm.memcpy</tt>' Intrinsic</a>
4212 <div class="doc_text">
4216 declare void @llvm.memcpy.i32(i8 * <dest>, i8 * <src>,
4217 i32 <len>, i32 <align>)
4218 declare void @llvm.memcpy.i64(i8 * <dest>, i8 * <src>,
4219 i64 <len>, i32 <align>)
4225 The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the source
4226 location to the destination location.
4230 Note that, unlike the standard libc function, the <tt>llvm.memcpy.*</tt>
4231 intrinsics do not return a value, and takes an extra alignment argument.
4237 The first argument is a pointer to the destination, the second is a pointer to
4238 the source. The third argument is an integer argument
4239 specifying the number of bytes to copy, and the fourth argument is the alignment
4240 of the source and destination locations.
4244 If the call to this intrinisic has an alignment value that is not 0 or 1, then
4245 the caller guarantees that both the source and destination pointers are aligned
4252 The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the source
4253 location to the destination location, which are not allowed to overlap. It
4254 copies "len" bytes of memory over. If the argument is known to be aligned to
4255 some boundary, this can be specified as the fourth argument, otherwise it should
4261 <!-- _______________________________________________________________________ -->
4262 <div class="doc_subsubsection">
4263 <a name="int_memmove">'<tt>llvm.memmove</tt>' Intrinsic</a>
4266 <div class="doc_text">
4270 declare void @llvm.memmove.i32(i8 * <dest>, i8 * <src>,
4271 i32 <len>, i32 <align>)
4272 declare void @llvm.memmove.i64(i8 * <dest>, i8 * <src>,
4273 i64 <len>, i32 <align>)
4279 The '<tt>llvm.memmove.*</tt>' intrinsics move a block of memory from the source
4280 location to the destination location. It is similar to the
4281 '<tt>llvm.memcmp</tt>' intrinsic but allows the two memory locations to overlap.
4285 Note that, unlike the standard libc function, the <tt>llvm.memmove.*</tt>
4286 intrinsics do not return a value, and takes an extra alignment argument.
4292 The first argument is a pointer to the destination, the second is a pointer to
4293 the source. The third argument is an integer argument
4294 specifying the number of bytes to copy, and the fourth argument is the alignment
4295 of the source and destination locations.
4299 If the call to this intrinisic has an alignment value that is not 0 or 1, then
4300 the caller guarantees that the source and destination pointers are aligned to
4307 The '<tt>llvm.memmove.*</tt>' intrinsics copy a block of memory from the source
4308 location to the destination location, which may overlap. It
4309 copies "len" bytes of memory over. If the argument is known to be aligned to
4310 some boundary, this can be specified as the fourth argument, otherwise it should
4316 <!-- _______________________________________________________________________ -->
4317 <div class="doc_subsubsection">
4318 <a name="int_memset">'<tt>llvm.memset.*</tt>' Intrinsics</a>
4321 <div class="doc_text">
4325 declare void @llvm.memset.i32(i8 * <dest>, i8 <val>,
4326 i32 <len>, i32 <align>)
4327 declare void @llvm.memset.i64(i8 * <dest>, i8 <val>,
4328 i64 <len>, i32 <align>)
4334 The '<tt>llvm.memset.*</tt>' intrinsics fill a block of memory with a particular
4339 Note that, unlike the standard libc function, the <tt>llvm.memset</tt> intrinsic
4340 does not return a value, and takes an extra alignment argument.
4346 The first argument is a pointer to the destination to fill, the second is the
4347 byte value to fill it with, the third argument is an integer
4348 argument specifying the number of bytes to fill, and the fourth argument is the
4349 known alignment of destination location.
4353 If the call to this intrinisic has an alignment value that is not 0 or 1, then
4354 the caller guarantees that the destination pointer is aligned to that boundary.
4360 The '<tt>llvm.memset.*</tt>' intrinsics fill "len" bytes of memory starting at
4362 destination location. If the argument is known to be aligned to some boundary,
4363 this can be specified as the fourth argument, otherwise it should be set to 0 or
4369 <!-- _______________________________________________________________________ -->
4370 <div class="doc_subsubsection">
4371 <a name="int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a>
4374 <div class="doc_text">
4378 declare float @llvm.sqrt.f32(float %Val)
4379 declare double @llvm.sqrt.f64(double %Val)
4385 The '<tt>llvm.sqrt</tt>' intrinsics return the sqrt of the specified operand,
4386 returning the same value as the libm '<tt>sqrt</tt>' function would. Unlike
4387 <tt>sqrt</tt> in libm, however, <tt>llvm.sqrt</tt> has undefined behavior for
4388 negative numbers (which allows for better optimization).
4394 The argument and return value are floating point numbers of the same type.
4400 This function returns the sqrt of the specified operand if it is a positive
4401 floating point number.
4405 <!-- _______________________________________________________________________ -->
4406 <div class="doc_subsubsection">
4407 <a name="int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a>
4410 <div class="doc_text">
4414 declare float @llvm.powi.f32(float %Val, i32 %power)
4415 declare double @llvm.powi.f64(double %Val, i32 %power)
4421 The '<tt>llvm.powi.*</tt>' intrinsics return the first operand raised to the
4422 specified (positive or negative) power. The order of evaluation of
4423 multiplications is not defined.
4429 The second argument is an integer power, and the first is a value to raise to
4436 This function returns the first value raised to the second power with an
4437 unspecified sequence of rounding operations.</p>
4441 <!-- ======================================================================= -->
4442 <div class="doc_subsection">
4443 <a name="int_manip">Bit Manipulation Intrinsics</a>
4446 <div class="doc_text">
4448 LLVM provides intrinsics for a few important bit manipulation operations.
4449 These allow efficient code generation for some algorithms.
4454 <!-- _______________________________________________________________________ -->
4455 <div class="doc_subsubsection">
4456 <a name="int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a>
4459 <div class="doc_text">
4462 <p>This is an overloaded intrinsic function. You can use bswap on any integer
4463 type that is an even number of bytes (i.e. BitWidth % 16 == 0). Note the suffix
4464 that includes the type for the result and the operand.
4466 declare i16 @llvm.bswap.i16.i16(i16 <id>)
4467 declare i32 @llvm.bswap.i32.i32(i32 <id>)
4468 declare i64 @llvm.bswap.i64.i64(i64 <id>)
4474 The '<tt>llvm.bswap</tt>' family of intrinsics is used to byte swap integer
4475 values with an even number of bytes (positive multiple of 16 bits). These are
4476 useful for performing operations on data that is not in the target's native
4483 The <tt>llvm.bswap.16.i16</tt> intrinsic returns an i16 value that has the high
4484 and low byte of the input i16 swapped. Similarly, the <tt>llvm.bswap.i32</tt>
4485 intrinsic returns an i32 value that has the four bytes of the input i32
4486 swapped, so that if the input bytes are numbered 0, 1, 2, 3 then the returned
4487 i32 will have its bytes in 3, 2, 1, 0 order. The <tt>llvm.bswap.i48.i48</tt>,
4488 <tt>llvm.bswap.i64.i64</tt> and other intrinsics extend this concept to
4489 additional even-byte lengths (6 bytes, 8 bytes and more, respectively).
4494 <!-- _______________________________________________________________________ -->
4495 <div class="doc_subsubsection">
4496 <a name="int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic</a>
4499 <div class="doc_text">
4502 <p>This is an overloaded intrinsic. You can use llvm.ctpop on any integer bit
4503 width. Not all targets support all bit widths however.
4505 declare i32 @llvm.ctpop.i8 (i8 <src>)
4506 declare i32 @llvm.ctpop.i16(i16 <src>)
4507 declare i32 @llvm.ctpop.i32(i32 <src>)
4508 declare i32 @llvm.ctpop.i64(i64 <src>)
4509 declare i32 @llvm.ctpop.i256(i256 <src>)
4515 The '<tt>llvm.ctpop</tt>' family of intrinsics counts the number of bits set in a
4522 The only argument is the value to be counted. The argument may be of any
4523 integer type. The return type must match the argument type.
4529 The '<tt>llvm.ctpop</tt>' intrinsic counts the 1's in a variable.
4533 <!-- _______________________________________________________________________ -->
4534 <div class="doc_subsubsection">
4535 <a name="int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic</a>
4538 <div class="doc_text">
4541 <p>This is an overloaded intrinsic. You can use <tt>llvm.ctlz</tt> on any
4542 integer bit width. Not all targets support all bit widths however.
4544 declare i32 @llvm.ctlz.i8 (i8 <src>)
4545 declare i32 @llvm.ctlz.i16(i16 <src>)
4546 declare i32 @llvm.ctlz.i32(i32 <src>)
4547 declare i32 @llvm.ctlz.i64(i64 <src>)
4548 declare i32 @llvm.ctlz.i256(i256 <src>)
4554 The '<tt>llvm.ctlz</tt>' family of intrinsic functions counts the number of
4555 leading zeros in a variable.
4561 The only argument is the value to be counted. The argument may be of any
4562 integer type. The return type must match the argument type.
4568 The '<tt>llvm.ctlz</tt>' intrinsic counts the leading (most significant) zeros
4569 in a variable. If the src == 0 then the result is the size in bits of the type
4570 of src. For example, <tt>llvm.ctlz(i32 2) = 30</tt>.
4576 <!-- _______________________________________________________________________ -->
4577 <div class="doc_subsubsection">
4578 <a name="int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic</a>
4581 <div class="doc_text">
4584 <p>This is an overloaded intrinsic. You can use <tt>llvm.cttz</tt> on any
4585 integer bit width. Not all targets support all bit widths however.
4587 declare i32 @llvm.cttz.i8 (i8 <src>)
4588 declare i32 @llvm.cttz.i16(i16 <src>)
4589 declare i32 @llvm.cttz.i32(i32 <src>)
4590 declare i32 @llvm.cttz.i64(i64 <src>)
4591 declare i32 @llvm.cttz.i256(i256 <src>)
4597 The '<tt>llvm.cttz</tt>' family of intrinsic functions counts the number of
4604 The only argument is the value to be counted. The argument may be of any
4605 integer type. The return type must match the argument type.
4611 The '<tt>llvm.cttz</tt>' intrinsic counts the trailing (least significant) zeros
4612 in a variable. If the src == 0 then the result is the size in bits of the type
4613 of src. For example, <tt>llvm.cttz(2) = 1</tt>.
4617 <!-- _______________________________________________________________________ -->
4618 <div class="doc_subsubsection">
4619 <a name="int_part_select">'<tt>llvm.part.select.*</tt>' Intrinsic</a>
4622 <div class="doc_text">
4625 <p>This is an overloaded intrinsic. You can use <tt>llvm.part.select</tt>
4626 on any integer bit width.
4628 declare i17 @llvm.part.select.i17.i17 (i17 %val, i32 %loBit, i32 %hiBit)
4629 declare i29 @llvm.part.select.i29.i29 (i29 %val, i32 %loBit, i32 %hiBit)
4633 <p>The '<tt>llvm.part.select</tt>' family of intrinsic functions selects a
4634 range of bits from an integer value and returns them in the same bit width as
4635 the original value.</p>
4638 <p>The first argument, <tt>%val</tt> and the result may be integer types of
4639 any bit width but they must have the same bit width. The second and third
4640 arguments must be <tt>i32</tt> type since they specify only a bit index.</p>
4643 <p>The operation of the '<tt>llvm.part.select</tt>' intrinsic has two modes
4644 of operation: forwards and reverse. If <tt>%loBit</tt> is greater than
4645 <tt>%hiBits</tt> then the intrinsic operates in reverse mode. Otherwise it
4646 operates in forward mode.</p>
4647 <p>In forward mode, this intrinsic is the equivalent of shifting <tt>%val</tt>
4648 right by <tt>%loBit</tt> bits and then ANDing it with a mask with
4649 only the <tt>%hiBit - %loBit</tt> bits set, as follows:</p>
4651 <li>The <tt>%val</tt> is shifted right (LSHR) by the number of bits specified
4652 by <tt>%loBits</tt>. This normalizes the value to the low order bits.</li>
4653 <li>The <tt>%loBits</tt> value is subtracted from the <tt>%hiBits</tt> value
4654 to determine the number of bits to retain.</li>
4655 <li>A mask of the retained bits is created by shifting a -1 value.</li>
4656 <li>The mask is ANDed with <tt>%val</tt> to produce the result.
4658 <p>In reverse mode, a similar computation is made except that the bits are
4659 returned in the reverse order. So, for example, if <tt>X</tt> has the value
4660 <tt>i16 0x0ACF (101011001111)</tt> and we apply
4661 <tt>part.select(i16 X, 8, 3)</tt> to it, we get back the value
4662 <tt>i16 0x0026 (000000100110)</tt>.</p>
4665 <div class="doc_subsubsection">
4666 <a name="int_part_set">'<tt>llvm.part.set.*</tt>' Intrinsic</a>
4669 <div class="doc_text">
4672 <p>This is an overloaded intrinsic. You can use <tt>llvm.part.set</tt>
4673 on any integer bit width.
4675 declare i17 @llvm.part.set.i17.i17.i9 (i17 %val, i9 %repl, i32 %lo, i32 %hi)
4676 declare i29 @llvm.part.set.i29.i29.i9 (i29 %val, i9 %repl, i32 %lo, i32 %hi)
4680 <p>The '<tt>llvm.part.set</tt>' family of intrinsic functions replaces a range
4681 of bits in an integer value with another integer value. It returns the integer
4682 with the replaced bits.</p>
4685 <p>The first argument, <tt>%val</tt> and the result may be integer types of
4686 any bit width but they must have the same bit width. <tt>%val</tt> is the value
4687 whose bits will be replaced. The second argument, <tt>%repl</tt> may be an
4688 integer of any bit width. The third and fourth arguments must be <tt>i32</tt>
4689 type since they specify only a bit index.</p>
4692 <p>The operation of the '<tt>llvm.part.set</tt>' intrinsic has two modes
4693 of operation: forwards and reverse. If <tt>%lo</tt> is greater than
4694 <tt>%hi</tt> then the intrinsic operates in reverse mode. Otherwise it
4695 operates in forward mode.</p>
4696 <p>For both modes, the <tt>%repl</tt> value is prepared for use by either
4697 truncating it down to the size of the replacement area or zero extending it
4698 up to that size.</p>
4699 <p>In forward mode, the bits between <tt>%lo</tt> and <tt>%hi</tt> (inclusive)
4700 are replaced with corresponding bits from <tt>%repl</tt>. That is the 0th bit
4701 in <tt>%repl</tt> replaces the <tt>%lo</tt>th bit in <tt>%val</tt> and etc. up
4702 to the <tt>%hi</tt>th bit.
4703 <p>In reverse mode, a similar computation is made except that the bits replaced
4704 wrap around to include both the highest and lowest bits. For example, if a
4705 16 bit value is being replaced then <tt>%lo=8</tt> and <tt>%hi=4</tt> would
4706 cause these bits to be set: <tt>0xFF1F</tt>.</p>
4709 llvm.part.set(0xFFFF, 0, 4, 7) -> 0xFF0F
4710 llvm.part.set(0xFFFF, 0, 7, 4) -> 0x0060
4711 llvm.part.set(0xFFFF, 0, 8, 3) -> 0x00F0
4712 llvm.part.set(0xFFFF, 0, 3, 8) -> 0xFE07
4716 <!-- ======================================================================= -->
4717 <div class="doc_subsection">
4718 <a name="int_debugger">Debugger Intrinsics</a>
4721 <div class="doc_text">
4723 The LLVM debugger intrinsics (which all start with <tt>llvm.dbg.</tt> prefix),
4724 are described in the <a
4725 href="SourceLevelDebugging.html#format_common_intrinsics">LLVM Source Level
4726 Debugging</a> document.
4731 <!-- ======================================================================= -->
4732 <div class="doc_subsection">
4733 <a name="int_eh">Exception Handling Intrinsics</a>
4736 <div class="doc_text">
4737 <p> The LLVM exception handling intrinsics (which all start with
4738 <tt>llvm.eh.</tt> prefix), are described in the <a
4739 href="ExceptionHandling.html#format_common_intrinsics">LLVM Exception
4740 Handling</a> document. </p>
4744 <!-- *********************************************************************** -->
4747 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
4748 src="http://jigsaw.w3.org/css-validator/images/vcss" alt="Valid CSS!"></a>
4749 <a href="http://validator.w3.org/check/referer"><img
4750 src="http://www.w3.org/Icons/valid-html401" alt="Valid HTML 4.01!" /></a>
4752 <a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
4753 <a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br>
4754 Last modified: $Date$