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="#moduleasm">Module-Level Inline Assembly</a></li>
30 <li><a href="#typesystem">Type System</a>
32 <li><a href="#t_primitive">Primitive Types</a>
34 <li><a href="#t_classifications">Type Classifications</a></li>
37 <li><a href="#t_derived">Derived Types</a>
39 <li><a href="#t_array">Array Type</a></li>
40 <li><a href="#t_function">Function Type</a></li>
41 <li><a href="#t_pointer">Pointer Type</a></li>
42 <li><a href="#t_struct">Structure Type</a></li>
43 <li><a href="#t_packed">Packed Type</a></li>
44 <li><a href="#t_opaque">Opaque Type</a></li>
49 <li><a href="#constants">Constants</a>
51 <li><a href="#simpleconstants">Simple Constants</a>
52 <li><a href="#aggregateconstants">Aggregate Constants</a>
53 <li><a href="#globalconstants">Global Variable and Function Addresses</a>
54 <li><a href="#undefvalues">Undefined Values</a>
55 <li><a href="#constantexprs">Constant Expressions</a>
58 <li><a href="#othervalues">Other Values</a>
60 <li><a href="#inlineasm">Inline Assembler Expressions</a>
63 <li><a href="#instref">Instruction Reference</a>
65 <li><a href="#terminators">Terminator Instructions</a>
67 <li><a href="#i_ret">'<tt>ret</tt>' Instruction</a></li>
68 <li><a href="#i_br">'<tt>br</tt>' Instruction</a></li>
69 <li><a href="#i_switch">'<tt>switch</tt>' Instruction</a></li>
70 <li><a href="#i_invoke">'<tt>invoke</tt>' Instruction</a></li>
71 <li><a href="#i_unwind">'<tt>unwind</tt>' Instruction</a></li>
72 <li><a href="#i_unreachable">'<tt>unreachable</tt>' Instruction</a></li>
75 <li><a href="#binaryops">Binary Operations</a>
77 <li><a href="#i_add">'<tt>add</tt>' Instruction</a></li>
78 <li><a href="#i_sub">'<tt>sub</tt>' Instruction</a></li>
79 <li><a href="#i_mul">'<tt>mul</tt>' Instruction</a></li>
80 <li><a href="#i_div">'<tt>div</tt>' Instruction</a></li>
81 <li><a href="#i_rem">'<tt>rem</tt>' Instruction</a></li>
82 <li><a href="#i_setcc">'<tt>set<i>cc</i></tt>' Instructions</a></li>
85 <li><a href="#bitwiseops">Bitwise Binary Operations</a>
87 <li><a href="#i_and">'<tt>and</tt>' Instruction</a></li>
88 <li><a href="#i_or">'<tt>or</tt>' Instruction</a></li>
89 <li><a href="#i_xor">'<tt>xor</tt>' Instruction</a></li>
90 <li><a href="#i_shl">'<tt>shl</tt>' Instruction</a></li>
91 <li><a href="#i_shr">'<tt>shr</tt>' Instruction</a></li>
94 <li><a href="#vectorops">Vector Operations</a>
96 <li><a href="#i_extractelement">'<tt>extractelement</tt>' Instruction</a></li>
97 <li><a href="#i_insertelement">'<tt>insertelement</tt>' Instruction</a></li>
98 <li><a href="#i_shufflevector">'<tt>shufflevector</tt>' Instruction</a></li>
99 <li><a href="#i_vsetint">'<tt>vsetint</tt>' Instruction</a></li>
100 <li><a href="#i_vsetfp">'<tt>vsetfp</tt>' Instruction</a></li>
101 <li><a href="#i_vselect">'<tt>vselect</tt>' Instruction</a></li>
104 <li><a href="#memoryops">Memory Access and Addressing Operations</a>
106 <li><a href="#i_malloc">'<tt>malloc</tt>' Instruction</a></li>
107 <li><a href="#i_free">'<tt>free</tt>' Instruction</a></li>
108 <li><a href="#i_alloca">'<tt>alloca</tt>' Instruction</a></li>
109 <li><a href="#i_load">'<tt>load</tt>' Instruction</a></li>
110 <li><a href="#i_store">'<tt>store</tt>' Instruction</a></li>
111 <li><a href="#i_getelementptr">'<tt>getelementptr</tt>' Instruction</a></li>
114 <li><a href="#otherops">Other Operations</a>
116 <li><a href="#i_phi">'<tt>phi</tt>' Instruction</a></li>
117 <li><a href="#i_cast">'<tt>cast .. to</tt>' Instruction</a></li>
118 <li><a href="#i_select">'<tt>select</tt>' Instruction</a></li>
119 <li><a href="#i_call">'<tt>call</tt>' Instruction</a></li>
120 <li><a href="#i_va_arg">'<tt>va_arg</tt>' Instruction</a></li>
125 <li><a href="#intrinsics">Intrinsic Functions</a>
127 <li><a href="#int_varargs">Variable Argument Handling Intrinsics</a>
129 <li><a href="#i_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a></li>
130 <li><a href="#i_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a></li>
131 <li><a href="#i_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a></li>
134 <li><a href="#int_gc">Accurate Garbage Collection Intrinsics</a>
136 <li><a href="#i_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a></li>
137 <li><a href="#i_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a></li>
138 <li><a href="#i_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a></li>
141 <li><a href="#int_codegen">Code Generator Intrinsics</a>
143 <li><a href="#i_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a></li>
144 <li><a href="#i_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a></li>
145 <li><a href="#i_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a></li>
146 <li><a href="#i_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a></li>
147 <li><a href="#i_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a></li>
148 <li><a href="#i_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a></li>
149 <li><a href="#i_readcyclecounter"><tt>llvm.readcyclecounter</tt>' Intrinsic</a></li>
152 <li><a href="#int_libc">Standard C Library Intrinsics</a>
154 <li><a href="#i_memcpy">'<tt>llvm.memcpy.*</tt>' Intrinsic</a></li>
155 <li><a href="#i_memmove">'<tt>llvm.memmove.*</tt>' Intrinsic</a></li>
156 <li><a href="#i_memset">'<tt>llvm.memset.*</tt>' Intrinsic</a></li>
157 <li><a href="#i_isunordered">'<tt>llvm.isunordered.*</tt>' Intrinsic</a></li>
158 <li><a href="#i_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a></li>
162 <li><a href="#int_manip">Bit Manipulation Intrinsics</a>
164 <li><a href="#i_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a></li>
165 <li><a href="#int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic </a></li>
166 <li><a href="#int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic </a></li>
167 <li><a href="#int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic </a></li>
170 <li><a href="#int_debugger">Debugger intrinsics</a></li>
175 <div class="doc_author">
176 <p>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a>
177 and <a href="mailto:vadve@cs.uiuc.edu">Vikram Adve</a></p>
180 <!-- *********************************************************************** -->
181 <div class="doc_section"> <a name="abstract">Abstract </a></div>
182 <!-- *********************************************************************** -->
184 <div class="doc_text">
185 <p>This document is a reference manual for the LLVM assembly language.
186 LLVM is an SSA based representation that provides type safety,
187 low-level operations, flexibility, and the capability of representing
188 'all' high-level languages cleanly. It is the common code
189 representation used throughout all phases of the LLVM compilation
193 <!-- *********************************************************************** -->
194 <div class="doc_section"> <a name="introduction">Introduction</a> </div>
195 <!-- *********************************************************************** -->
197 <div class="doc_text">
199 <p>The LLVM code representation is designed to be used in three
200 different forms: as an in-memory compiler IR, as an on-disk bytecode
201 representation (suitable for fast loading by a Just-In-Time compiler),
202 and as a human readable assembly language representation. This allows
203 LLVM to provide a powerful intermediate representation for efficient
204 compiler transformations and analysis, while providing a natural means
205 to debug and visualize the transformations. The three different forms
206 of LLVM are all equivalent. This document describes the human readable
207 representation and notation.</p>
209 <p>The LLVM representation aims to be light-weight and low-level
210 while being expressive, typed, and extensible at the same time. It
211 aims to be a "universal IR" of sorts, by being at a low enough level
212 that high-level ideas may be cleanly mapped to it (similar to how
213 microprocessors are "universal IR's", allowing many source languages to
214 be mapped to them). By providing type information, LLVM can be used as
215 the target of optimizations: for example, through pointer analysis, it
216 can be proven that a C automatic variable is never accessed outside of
217 the current function... allowing it to be promoted to a simple SSA
218 value instead of a memory location.</p>
222 <!-- _______________________________________________________________________ -->
223 <div class="doc_subsubsection"> <a name="wellformed">Well-Formedness</a> </div>
225 <div class="doc_text">
227 <p>It is important to note that this document describes 'well formed'
228 LLVM assembly language. There is a difference between what the parser
229 accepts and what is considered 'well formed'. For example, the
230 following instruction is syntactically okay, but not well formed:</p>
233 %x = <a href="#i_add">add</a> int 1, %x
236 <p>...because the definition of <tt>%x</tt> does not dominate all of
237 its uses. The LLVM infrastructure provides a verification pass that may
238 be used to verify that an LLVM module is well formed. This pass is
239 automatically run by the parser after parsing input assembly and by
240 the optimizer before it outputs bytecode. The violations pointed out
241 by the verifier pass indicate bugs in transformation passes or input to
244 <!-- Describe the typesetting conventions here. --> </div>
246 <!-- *********************************************************************** -->
247 <div class="doc_section"> <a name="identifiers">Identifiers</a> </div>
248 <!-- *********************************************************************** -->
250 <div class="doc_text">
252 <p>LLVM uses three different forms of identifiers, for different
256 <li>Named values are represented as a string of characters with a '%' prefix.
257 For example, %foo, %DivisionByZero, %a.really.long.identifier. The actual
258 regular expression used is '<tt>%[a-zA-Z$._][a-zA-Z$._0-9]*</tt>'.
259 Identifiers which require other characters in their names can be surrounded
260 with quotes. In this way, anything except a <tt>"</tt> character can be used
263 <li>Unnamed values are represented as an unsigned numeric value with a '%'
264 prefix. For example, %12, %2, %44.</li>
266 <li>Constants, which are described in a <a href="#constants">section about
267 constants</a>, below.</li>
270 <p>LLVM requires that values start with a '%' sign for two reasons: Compilers
271 don't need to worry about name clashes with reserved words, and the set of
272 reserved words may be expanded in the future without penalty. Additionally,
273 unnamed identifiers allow a compiler to quickly come up with a temporary
274 variable without having to avoid symbol table conflicts.</p>
276 <p>Reserved words in LLVM are very similar to reserved words in other
277 languages. There are keywords for different opcodes ('<tt><a
278 href="#i_add">add</a></tt>', '<tt><a href="#i_cast">cast</a></tt>', '<tt><a
279 href="#i_ret">ret</a></tt>', etc...), for primitive type names ('<tt><a
280 href="#t_void">void</a></tt>', '<tt><a href="#t_uint">uint</a></tt>', etc...),
281 and others. These reserved words cannot conflict with variable names, because
282 none of them start with a '%' character.</p>
284 <p>Here is an example of LLVM code to multiply the integer variable
285 '<tt>%X</tt>' by 8:</p>
290 %result = <a href="#i_mul">mul</a> uint %X, 8
293 <p>After strength reduction:</p>
296 %result = <a href="#i_shl">shl</a> uint %X, ubyte 3
299 <p>And the hard way:</p>
302 <a href="#i_add">add</a> uint %X, %X <i>; yields {uint}:%0</i>
303 <a href="#i_add">add</a> uint %0, %0 <i>; yields {uint}:%1</i>
304 %result = <a href="#i_add">add</a> uint %1, %1
307 <p>This last way of multiplying <tt>%X</tt> by 8 illustrates several
308 important lexical features of LLVM:</p>
312 <li>Comments are delimited with a '<tt>;</tt>' and go until the end of
315 <li>Unnamed temporaries are created when the result of a computation is not
316 assigned to a named value.</li>
318 <li>Unnamed temporaries are numbered sequentially</li>
322 <p>...and it also shows a convention that we follow in this document. When
323 demonstrating instructions, we will follow an instruction with a comment that
324 defines the type and name of value produced. Comments are shown in italic
329 <!-- *********************************************************************** -->
330 <div class="doc_section"> <a name="highlevel">High Level Structure</a> </div>
331 <!-- *********************************************************************** -->
333 <!-- ======================================================================= -->
334 <div class="doc_subsection"> <a name="modulestructure">Module Structure</a>
337 <div class="doc_text">
339 <p>LLVM programs are composed of "Module"s, each of which is a
340 translation unit of the input programs. Each module consists of
341 functions, global variables, and symbol table entries. Modules may be
342 combined together with the LLVM linker, which merges function (and
343 global variable) definitions, resolves forward declarations, and merges
344 symbol table entries. Here is an example of the "hello world" module:</p>
346 <pre><i>; Declare the string constant as a global constant...</i>
347 <a href="#identifiers">%.LC0</a> = <a href="#linkage_internal">internal</a> <a
348 href="#globalvars">constant</a> <a href="#t_array">[13 x sbyte]</a> c"hello world\0A\00" <i>; [13 x sbyte]*</i>
350 <i>; External declaration of the puts function</i>
351 <a href="#functionstructure">declare</a> int %puts(sbyte*) <i>; int(sbyte*)* </i>
353 <i>; Global variable / Function body section separator</i>
356 <i>; Definition of main function</i>
357 int %main() { <i>; int()* </i>
358 <i>; Convert [13x sbyte]* to sbyte *...</i>
360 href="#i_getelementptr">getelementptr</a> [13 x sbyte]* %.LC0, long 0, long 0 <i>; sbyte*</i>
362 <i>; Call puts function to write out the string to stdout...</i>
364 href="#i_call">call</a> int %puts(sbyte* %cast210) <i>; int</i>
366 href="#i_ret">ret</a> int 0<br>}<br></pre>
368 <p>This example is made up of a <a href="#globalvars">global variable</a>
369 named "<tt>.LC0</tt>", an external declaration of the "<tt>puts</tt>"
370 function, and a <a href="#functionstructure">function definition</a>
371 for "<tt>main</tt>".</p>
373 <p>In general, a module is made up of a list of global values,
374 where both functions and global variables are global values. Global values are
375 represented by a pointer to a memory location (in this case, a pointer to an
376 array of char, and a pointer to a function), and have one of the following <a
377 href="#linkage">linkage types</a>.</p>
379 <p>Due to a limitation in the current LLVM assembly parser (it is limited by
380 one-token lookahead), modules are split into two pieces by the "implementation"
381 keyword. Global variable prototypes and definitions must occur before the
382 keyword, and function definitions must occur after it. Function prototypes may
383 occur either before or after it. In the future, the implementation keyword may
384 become a noop, if the parser gets smarter.</p>
388 <!-- ======================================================================= -->
389 <div class="doc_subsection">
390 <a name="linkage">Linkage Types</a>
393 <div class="doc_text">
396 All Global Variables and Functions have one of the following types of linkage:
401 <dt><tt><b><a name="linkage_internal">internal</a></b></tt> </dt>
403 <dd>Global values with internal linkage are only directly accessible by
404 objects in the current module. In particular, linking code into a module with
405 an internal global value may cause the internal to be renamed as necessary to
406 avoid collisions. Because the symbol is internal to the module, all
407 references can be updated. This corresponds to the notion of the
408 '<tt>static</tt>' keyword in C, or the idea of "anonymous namespaces" in C++.
411 <dt><tt><b><a name="linkage_linkonce">linkonce</a></b></tt>: </dt>
413 <dd>"<tt>linkonce</tt>" linkage is similar to <tt>internal</tt> linkage, with
414 the twist that linking together two modules defining the same
415 <tt>linkonce</tt> globals will cause one of the globals to be discarded. This
416 is typically used to implement inline functions. Unreferenced
417 <tt>linkonce</tt> globals are allowed to be discarded.
420 <dt><tt><b><a name="linkage_weak">weak</a></b></tt>: </dt>
422 <dd>"<tt>weak</tt>" linkage is exactly the same as <tt>linkonce</tt> linkage,
423 except that unreferenced <tt>weak</tt> globals may not be discarded. This is
424 used to implement constructs in C such as "<tt>int X;</tt>" at global scope.
427 <dt><tt><b><a name="linkage_appending">appending</a></b></tt>: </dt>
429 <dd>"<tt>appending</tt>" linkage may only be applied to global variables of
430 pointer to array type. When two global variables with appending linkage are
431 linked together, the two global arrays are appended together. This is the
432 LLVM, typesafe, equivalent of having the system linker append together
433 "sections" with identical names when .o files are linked.
436 <dt><tt><b><a name="linkage_external">externally visible</a></b></tt>:</dt>
438 <dd>If none of the above identifiers are used, the global is externally
439 visible, meaning that it participates in linkage and can be used to resolve
440 external symbol references.
444 <p><a name="linkage_external">For example, since the "<tt>.LC0</tt>"
445 variable is defined to be internal, if another module defined a "<tt>.LC0</tt>"
446 variable and was linked with this one, one of the two would be renamed,
447 preventing a collision. Since "<tt>main</tt>" and "<tt>puts</tt>" are
448 external (i.e., lacking any linkage declarations), they are accessible
449 outside of the current module. It is illegal for a function <i>declaration</i>
450 to have any linkage type other than "externally visible".</a></p>
454 <!-- ======================================================================= -->
455 <div class="doc_subsection">
456 <a name="callingconv">Calling Conventions</a>
459 <div class="doc_text">
461 <p>LLVM <a href="#functionstructure">functions</a>, <a href="#i_call">calls</a>
462 and <a href="#i_invoke">invokes</a> can all have an optional calling convention
463 specified for the call. The calling convention of any pair of dynamic
464 caller/callee must match, or the behavior of the program is undefined. The
465 following calling conventions are supported by LLVM, and more may be added in
469 <dt><b>"<tt>ccc</tt>" - The C calling convention</b>:</dt>
471 <dd>This calling convention (the default if no other calling convention is
472 specified) matches the target C calling conventions. This calling convention
473 supports varargs function calls and tolerates some mismatch in the declared
474 prototype and implemented declaration of the function (as does normal C).
477 <dt><b>"<tt>csretcc</tt>" - The C struct return calling convention</b>:</dt>
479 <dd>This calling convention matches the target C calling conventions, except
480 that functions with this convention are required to take a pointer as their
481 first argument, and the return type of the function must be void. This is
482 used for C functions that return aggregates by-value. In this case, the
483 function has been transformed to take a pointer to the struct as the first
484 argument to the function. For targets where the ABI specifies specific
485 behavior for structure-return calls, the calling convention can be used to
486 distinguish between struct return functions and other functions that take a
487 pointer to a struct as the first argument.
490 <dt><b>"<tt>fastcc</tt>" - The fast calling convention</b>:</dt>
492 <dd>This calling convention attempts to make calls as fast as possible
493 (e.g. by passing things in registers). This calling convention allows the
494 target to use whatever tricks it wants to produce fast code for the target,
495 without having to conform to an externally specified ABI. Implementations of
496 this convention should allow arbitrary tail call optimization to be supported.
497 This calling convention does not support varargs and requires the prototype of
498 all callees to exactly match the prototype of the function definition.
501 <dt><b>"<tt>coldcc</tt>" - The cold calling convention</b>:</dt>
503 <dd>This calling convention attempts to make code in the caller as efficient
504 as possible under the assumption that the call is not commonly executed. As
505 such, these calls often preserve all registers so that the call does not break
506 any live ranges in the caller side. This calling convention does not support
507 varargs and requires the prototype of all callees to exactly match the
508 prototype of the function definition.
511 <dt><b>"<tt>cc <<em>n</em>></tt>" - Numbered convention</b>:</dt>
513 <dd>Any calling convention may be specified by number, allowing
514 target-specific calling conventions to be used. Target specific calling
515 conventions start at 64.
519 <p>More calling conventions can be added/defined on an as-needed basis, to
520 support pascal conventions or any other well-known target-independent
525 <!-- ======================================================================= -->
526 <div class="doc_subsection">
527 <a name="globalvars">Global Variables</a>
530 <div class="doc_text">
532 <p>Global variables define regions of memory allocated at compilation time
533 instead of run-time. Global variables may optionally be initialized, may have
534 an explicit section to be placed in, and may
535 have an optional explicit alignment specified. A
536 variable may be defined as a global "constant," which indicates that the
537 contents of the variable will <b>never</b> be modified (enabling better
538 optimization, allowing the global data to be placed in the read-only section of
539 an executable, etc). Note that variables that need runtime initialization
540 cannot be marked "constant" as there is a store to the variable.</p>
543 LLVM explicitly allows <em>declarations</em> of global variables to be marked
544 constant, even if the final definition of the global is not. This capability
545 can be used to enable slightly better optimization of the program, but requires
546 the language definition to guarantee that optimizations based on the
547 'constantness' are valid for the translation units that do not include the
551 <p>As SSA values, global variables define pointer values that are in
552 scope (i.e. they dominate) all basic blocks in the program. Global
553 variables always define a pointer to their "content" type because they
554 describe a region of memory, and all memory objects in LLVM are
555 accessed through pointers.</p>
557 <p>LLVM allows an explicit section to be specified for globals. If the target
558 supports it, it will emit globals to the section specified.</p>
560 <p>An explicit alignment may be specified for a global. If not present, or if
561 the alignment is set to zero, the alignment of the global is set by the target
562 to whatever it feels convenient. If an explicit alignment is specified, the
563 global is forced to have at least that much alignment. All alignments must be
569 <!-- ======================================================================= -->
570 <div class="doc_subsection">
571 <a name="functionstructure">Functions</a>
574 <div class="doc_text">
576 <p>LLVM function definitions consist of an optional <a href="#linkage">linkage
577 type</a>, an optional <a href="#callingconv">calling convention</a>, a return
578 type, a function name, a (possibly empty) argument list, an optional section,
579 an optional alignment, an opening curly brace,
580 a list of basic blocks, and a closing curly brace. LLVM function declarations
581 are defined with the "<tt>declare</tt>" keyword, an optional <a
582 href="#callingconv">calling convention</a>, a return type, a function name,
583 a possibly empty list of arguments, and an optional alignment.</p>
585 <p>A function definition contains a list of basic blocks, forming the CFG for
586 the function. Each basic block may optionally start with a label (giving the
587 basic block a symbol table entry), contains a list of instructions, and ends
588 with a <a href="#terminators">terminator</a> instruction (such as a branch or
589 function return).</p>
591 <p>The first basic block in a program is special in two ways: it is immediately
592 executed on entrance to the function, and it is not allowed to have predecessor
593 basic blocks (i.e. there can not be any branches to the entry block of a
594 function). Because the block can have no predecessors, it also cannot have any
595 <a href="#i_phi">PHI nodes</a>.</p>
597 <p>LLVM functions are identified by their name and type signature. Hence, two
598 functions with the same name but different parameter lists or return values are
599 considered different functions, and LLVM will resolve references to each
602 <p>LLVM allows an explicit section to be specified for functions. If the target
603 supports it, it will emit functions to the section specified.</p>
605 <p>An explicit alignment may be specified for a function. If not present, or if
606 the alignment is set to zero, the alignment of the function is set by the target
607 to whatever it feels convenient. If an explicit alignment is specified, the
608 function is forced to have at least that much alignment. All alignments must be
613 <!-- ======================================================================= -->
614 <div class="doc_subsection">
615 <a name="moduleasm">Module-Level Inline Assembly</a>
618 <div class="doc_text">
620 Modules may contain "module-level inline asm" blocks, which corresponds to the
621 GCC "file scope inline asm" blocks. These blocks are internally concatenated by
622 LLVM and treated as a single unit, but may be separated in the .ll file if
623 desired. The syntax is very simple:
626 <div class="doc_code"><pre>
627 module asm "inline asm code goes here"
628 module asm "more can go here"
631 <p>The strings can contain any character by escaping non-printable characters.
632 The escape sequence used is simply "\xx" where "xx" is the two digit hex code
637 The inline asm code is simply printed to the machine code .s file when
638 assembly code is generated.
643 <!-- *********************************************************************** -->
644 <div class="doc_section"> <a name="typesystem">Type System</a> </div>
645 <!-- *********************************************************************** -->
647 <div class="doc_text">
649 <p>The LLVM type system is one of the most important features of the
650 intermediate representation. Being typed enables a number of
651 optimizations to be performed on the IR directly, without having to do
652 extra analyses on the side before the transformation. A strong type
653 system makes it easier to read the generated code and enables novel
654 analyses and transformations that are not feasible to perform on normal
655 three address code representations.</p>
659 <!-- ======================================================================= -->
660 <div class="doc_subsection"> <a name="t_primitive">Primitive Types</a> </div>
661 <div class="doc_text">
662 <p>The primitive types are the fundamental building blocks of the LLVM
663 system. The current set of primitive types is as follows:</p>
665 <table class="layout">
670 <tr><th>Type</th><th>Description</th></tr>
671 <tr><td><tt>void</tt></td><td>No value</td></tr>
672 <tr><td><tt>ubyte</tt></td><td>Unsigned 8-bit value</td></tr>
673 <tr><td><tt>ushort</tt></td><td>Unsigned 16-bit value</td></tr>
674 <tr><td><tt>uint</tt></td><td>Unsigned 32-bit value</td></tr>
675 <tr><td><tt>ulong</tt></td><td>Unsigned 64-bit value</td></tr>
676 <tr><td><tt>float</tt></td><td>32-bit floating point value</td></tr>
677 <tr><td><tt>label</tt></td><td>Branch destination</td></tr>
684 <tr><th>Type</th><th>Description</th></tr>
685 <tr><td><tt>bool</tt></td><td>True or False value</td></tr>
686 <tr><td><tt>sbyte</tt></td><td>Signed 8-bit value</td></tr>
687 <tr><td><tt>short</tt></td><td>Signed 16-bit value</td></tr>
688 <tr><td><tt>int</tt></td><td>Signed 32-bit value</td></tr>
689 <tr><td><tt>long</tt></td><td>Signed 64-bit value</td></tr>
690 <tr><td><tt>double</tt></td><td>64-bit floating point value</td></tr>
698 <!-- _______________________________________________________________________ -->
699 <div class="doc_subsubsection"> <a name="t_classifications">Type
700 Classifications</a> </div>
701 <div class="doc_text">
702 <p>These different primitive types fall into a few useful
705 <table border="1" cellspacing="0" cellpadding="4">
707 <tr><th>Classification</th><th>Types</th></tr>
709 <td><a name="t_signed">signed</a></td>
710 <td><tt>sbyte, short, int, long, float, double</tt></td>
713 <td><a name="t_unsigned">unsigned</a></td>
714 <td><tt>ubyte, ushort, uint, ulong</tt></td>
717 <td><a name="t_integer">integer</a></td>
718 <td><tt>ubyte, sbyte, ushort, short, uint, int, ulong, long</tt></td>
721 <td><a name="t_integral">integral</a></td>
722 <td><tt>bool, ubyte, sbyte, ushort, short, uint, int, ulong, long</tt>
726 <td><a name="t_floating">floating point</a></td>
727 <td><tt>float, double</tt></td>
730 <td><a name="t_firstclass">first class</a></td>
731 <td><tt>bool, ubyte, sbyte, ushort, short, uint, int, ulong, long,<br>
732 float, double, <a href="#t_pointer">pointer</a>,
733 <a href="#t_packed">packed</a></tt></td>
738 <p>The <a href="#t_firstclass">first class</a> types are perhaps the
739 most important. Values of these types are the only ones which can be
740 produced by instructions, passed as arguments, or used as operands to
741 instructions. This means that all structures and arrays must be
742 manipulated either by pointer or by component.</p>
745 <!-- ======================================================================= -->
746 <div class="doc_subsection"> <a name="t_derived">Derived Types</a> </div>
748 <div class="doc_text">
750 <p>The real power in LLVM comes from the derived types in the system.
751 This is what allows a programmer to represent arrays, functions,
752 pointers, and other useful types. Note that these derived types may be
753 recursive: For example, it is possible to have a two dimensional array.</p>
757 <!-- _______________________________________________________________________ -->
758 <div class="doc_subsubsection"> <a name="t_array">Array Type</a> </div>
760 <div class="doc_text">
764 <p>The array type is a very simple derived type that arranges elements
765 sequentially in memory. The array type requires a size (number of
766 elements) and an underlying data type.</p>
771 [<# elements> x <elementtype>]
774 <p>The number of elements is a constant integer value; elementtype may
775 be any type with a size.</p>
778 <table class="layout">
781 <tt>[40 x int ]</tt><br/>
782 <tt>[41 x int ]</tt><br/>
783 <tt>[40 x uint]</tt><br/>
786 Array of 40 integer values.<br/>
787 Array of 41 integer values.<br/>
788 Array of 40 unsigned integer values.<br/>
792 <p>Here are some examples of multidimensional arrays:</p>
793 <table class="layout">
796 <tt>[3 x [4 x int]]</tt><br/>
797 <tt>[12 x [10 x float]]</tt><br/>
798 <tt>[2 x [3 x [4 x uint]]]</tt><br/>
801 3x4 array of integer values.<br/>
802 12x10 array of single precision floating point values.<br/>
803 2x3x4 array of unsigned integer values.<br/>
808 <p>Note that 'variable sized arrays' can be implemented in LLVM with a zero
809 length array. Normally, accesses past the end of an array are undefined in
810 LLVM (e.g. it is illegal to access the 5th element of a 3 element array).
811 As a special case, however, zero length arrays are recognized to be variable
812 length. This allows implementation of 'pascal style arrays' with the LLVM
813 type "{ int, [0 x float]}", for example.</p>
817 <!-- _______________________________________________________________________ -->
818 <div class="doc_subsubsection"> <a name="t_function">Function Type</a> </div>
819 <div class="doc_text">
821 <p>The function type can be thought of as a function signature. It
822 consists of a return type and a list of formal parameter types.
823 Function types are usually used to build virtual function tables
824 (which are structures of pointers to functions), for indirect function
825 calls, and when defining a function.</p>
827 The return type of a function type cannot be an aggregate type.
830 <pre> <returntype> (<parameter list>)<br></pre>
831 <p>...where '<tt><parameter list></tt>' is a comma-separated list of type
832 specifiers. Optionally, the parameter list may include a type <tt>...</tt>,
833 which indicates that the function takes a variable number of arguments.
834 Variable argument functions can access their arguments with the <a
835 href="#int_varargs">variable argument handling intrinsic</a> functions.</p>
837 <table class="layout">
840 <tt>int (int)</tt> <br/>
841 <tt>float (int, int *) *</tt><br/>
842 <tt>int (sbyte *, ...)</tt><br/>
845 function taking an <tt>int</tt>, returning an <tt>int</tt><br/>
846 <a href="#t_pointer">Pointer</a> to a function that takes an
847 <tt>int</tt> and a <a href="#t_pointer">pointer</a> to <tt>int</tt>,
848 returning <tt>float</tt>.<br/>
849 A vararg function that takes at least one <a href="#t_pointer">pointer</a>
850 to <tt>sbyte</tt> (signed char in C), which returns an integer. This is
851 the signature for <tt>printf</tt> in LLVM.<br/>
857 <!-- _______________________________________________________________________ -->
858 <div class="doc_subsubsection"> <a name="t_struct">Structure Type</a> </div>
859 <div class="doc_text">
861 <p>The structure type is used to represent a collection of data members
862 together in memory. The packing of the field types is defined to match
863 the ABI of the underlying processor. The elements of a structure may
864 be any type that has a size.</p>
865 <p>Structures are accessed using '<tt><a href="#i_load">load</a></tt>
866 and '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a
867 field with the '<tt><a href="#i_getelementptr">getelementptr</a></tt>'
870 <pre> { <type list> }<br></pre>
872 <table class="layout">
875 <tt>{ int, int, int }</tt><br/>
876 <tt>{ float, int (int) * }</tt><br/>
879 a triple of three <tt>int</tt> values<br/>
880 A pair, where the first element is a <tt>float</tt> and the second element
881 is a <a href="#t_pointer">pointer</a> to a <a href="#t_function">function</a>
882 that takes an <tt>int</tt>, returning an <tt>int</tt>.<br/>
888 <!-- _______________________________________________________________________ -->
889 <div class="doc_subsubsection"> <a name="t_pointer">Pointer Type</a> </div>
890 <div class="doc_text">
892 <p>As in many languages, the pointer type represents a pointer or
893 reference to another object, which must live in memory.</p>
895 <pre> <type> *<br></pre>
897 <table class="layout">
900 <tt>[4x int]*</tt><br/>
901 <tt>int (int *) *</tt><br/>
904 A <a href="#t_pointer">pointer</a> to <a href="#t_array">array</a> of
905 four <tt>int</tt> values<br/>
906 A <a href="#t_pointer">pointer</a> to a <a
907 href="#t_function">function</a> that takes an <tt>int*</tt>, returning an
914 <!-- _______________________________________________________________________ -->
915 <div class="doc_subsubsection"> <a name="t_packed">Packed Type</a> </div>
916 <div class="doc_text">
920 <p>A packed type is a simple derived type that represents a vector
921 of elements. Packed types are used when multiple primitive data
922 are operated in parallel using a single instruction (SIMD).
923 A packed type requires a size (number of
924 elements) and an underlying primitive data type. Vectors must have a power
925 of two length (1, 2, 4, 8, 16 ...). Packed types are
926 considered <a href="#t_firstclass">first class</a>.</p>
931 < <# elements> x <elementtype> >
934 <p>The number of elements is a constant integer value; elementtype may
935 be any integral or floating point type.</p>
939 <table class="layout">
942 <tt><4 x int></tt><br/>
943 <tt><8 x float></tt><br/>
944 <tt><2 x uint></tt><br/>
947 Packed vector of 4 integer values.<br/>
948 Packed vector of 8 floating-point values.<br/>
949 Packed vector of 2 unsigned integer values.<br/>
955 <!-- _______________________________________________________________________ -->
956 <div class="doc_subsubsection"> <a name="t_opaque">Opaque Type</a> </div>
957 <div class="doc_text">
961 <p>Opaque types are used to represent unknown types in the system. This
962 corresponds (for example) to the C notion of a foward declared structure type.
963 In LLVM, opaque types can eventually be resolved to any type (not just a
974 <table class="layout">
987 <!-- *********************************************************************** -->
988 <div class="doc_section"> <a name="constants">Constants</a> </div>
989 <!-- *********************************************************************** -->
991 <div class="doc_text">
993 <p>LLVM has several different basic types of constants. This section describes
994 them all and their syntax.</p>
998 <!-- ======================================================================= -->
999 <div class="doc_subsection"><a name="simpleconstants">Simple Constants</a></div>
1001 <div class="doc_text">
1004 <dt><b>Boolean constants</b></dt>
1006 <dd>The two strings '<tt>true</tt>' and '<tt>false</tt>' are both valid
1007 constants of the <tt><a href="#t_primitive">bool</a></tt> type.
1010 <dt><b>Integer constants</b></dt>
1012 <dd>Standard integers (such as '4') are constants of the <a
1013 href="#t_integer">integer</a> type. Negative numbers may be used with signed
1017 <dt><b>Floating point constants</b></dt>
1019 <dd>Floating point constants use standard decimal notation (e.g. 123.421),
1020 exponential notation (e.g. 1.23421e+2), or a more precise hexadecimal
1021 notation (see below). Floating point constants must have a <a
1022 href="#t_floating">floating point</a> type. </dd>
1024 <dt><b>Null pointer constants</b></dt>
1026 <dd>The identifier '<tt>null</tt>' is recognized as a null pointer constant
1027 and must be of <a href="#t_pointer">pointer type</a>.</dd>
1031 <p>The one non-intuitive notation for constants is the optional hexadecimal form
1032 of floating point constants. For example, the form '<tt>double
1033 0x432ff973cafa8000</tt>' is equivalent to (but harder to read than) '<tt>double
1034 4.5e+15</tt>'. The only time hexadecimal floating point constants are required
1035 (and the only time that they are generated by the disassembler) is when a
1036 floating point constant must be emitted but it cannot be represented as a
1037 decimal floating point number. For example, NaN's, infinities, and other
1038 special values are represented in their IEEE hexadecimal format so that
1039 assembly and disassembly do not cause any bits to change in the constants.</p>
1043 <!-- ======================================================================= -->
1044 <div class="doc_subsection"><a name="aggregateconstants">Aggregate Constants</a>
1047 <div class="doc_text">
1048 <p>Aggregate constants arise from aggregation of simple constants
1049 and smaller aggregate constants.</p>
1052 <dt><b>Structure constants</b></dt>
1054 <dd>Structure constants are represented with notation similar to structure
1055 type definitions (a comma separated list of elements, surrounded by braces
1056 (<tt>{}</tt>)). For example: "<tt>{ int 4, float 17.0, int* %G }</tt>",
1057 where "<tt>%G</tt>" is declared as "<tt>%G = external global int</tt>". Structure constants
1058 must have <a href="#t_struct">structure type</a>, and the number and
1059 types of elements must match those specified by the type.
1062 <dt><b>Array constants</b></dt>
1064 <dd>Array constants are represented with notation similar to array type
1065 definitions (a comma separated list of elements, surrounded by square brackets
1066 (<tt>[]</tt>)). For example: "<tt>[ int 42, int 11, int 74 ]</tt>". Array
1067 constants must have <a href="#t_array">array type</a>, and the number and
1068 types of elements must match those specified by the type.
1071 <dt><b>Packed constants</b></dt>
1073 <dd>Packed constants are represented with notation similar to packed type
1074 definitions (a comma separated list of elements, surrounded by
1075 less-than/greater-than's (<tt><></tt>)). For example: "<tt>< int 42,
1076 int 11, int 74, int 100 ></tt>". Packed constants must have <a
1077 href="#t_packed">packed type</a>, and the number and types of elements must
1078 match those specified by the type.
1081 <dt><b>Zero initialization</b></dt>
1083 <dd>The string '<tt>zeroinitializer</tt>' can be used to zero initialize a
1084 value to zero of <em>any</em> type, including scalar and aggregate types.
1085 This is often used to avoid having to print large zero initializers (e.g. for
1086 large arrays) and is always exactly equivalent to using explicit zero
1093 <!-- ======================================================================= -->
1094 <div class="doc_subsection">
1095 <a name="globalconstants">Global Variable and Function Addresses</a>
1098 <div class="doc_text">
1100 <p>The addresses of <a href="#globalvars">global variables</a> and <a
1101 href="#functionstructure">functions</a> are always implicitly valid (link-time)
1102 constants. These constants are explicitly referenced when the <a
1103 href="#identifiers">identifier for the global</a> is used and always have <a
1104 href="#t_pointer">pointer</a> type. For example, the following is a legal LLVM
1110 %Z = global [2 x int*] [ int* %X, int* %Y ]
1115 <!-- ======================================================================= -->
1116 <div class="doc_subsection"><a name="undefvalues">Undefined Values</a></div>
1117 <div class="doc_text">
1118 <p>The string '<tt>undef</tt>' is recognized as a type-less constant that has
1119 no specific value. Undefined values may be of any type and be used anywhere
1120 a constant is permitted.</p>
1122 <p>Undefined values indicate to the compiler that the program is well defined
1123 no matter what value is used, giving the compiler more freedom to optimize.
1127 <!-- ======================================================================= -->
1128 <div class="doc_subsection"><a name="constantexprs">Constant Expressions</a>
1131 <div class="doc_text">
1133 <p>Constant expressions are used to allow expressions involving other constants
1134 to be used as constants. Constant expressions may be of any <a
1135 href="#t_firstclass">first class</a> type and may involve any LLVM operation
1136 that does not have side effects (e.g. load and call are not supported). The
1137 following is the syntax for constant expressions:</p>
1140 <dt><b><tt>cast ( CST to TYPE )</tt></b></dt>
1142 <dd>Cast a constant to another type.</dd>
1144 <dt><b><tt>getelementptr ( CSTPTR, IDX0, IDX1, ... )</tt></b></dt>
1146 <dd>Perform the <a href="#i_getelementptr">getelementptr operation</a> on
1147 constants. As with the <a href="#i_getelementptr">getelementptr</a>
1148 instruction, the index list may have zero or more indexes, which are required
1149 to make sense for the type of "CSTPTR".</dd>
1151 <dt><b><tt>select ( COND, VAL1, VAL2 )</tt></b></dt>
1153 <dd>Perform the <a href="#i_select">select operation</a> on
1156 <dt><b><tt>extractelement ( VAL, IDX )</tt></b></dt>
1158 <dd>Perform the <a href="#i_extractelement">extractelement
1159 operation</a> on constants.
1161 <dt><b><tt>insertelement ( VAL, ELT, IDX )</tt></b></dt>
1163 <dd>Perform the <a href="#i_insertelement">insertelement
1164 operation</a> on constants.
1167 <dt><b><tt>shufflevector ( VEC1, VEC2, IDXMASK )</tt></b></dt>
1169 <dd>Perform the <a href="#i_shufflevector">shufflevector
1170 operation</a> on constants.
1172 <dt><b><tt>OPCODE ( LHS, RHS )</tt></b></dt>
1174 <dd>Perform the specified operation of the LHS and RHS constants. OPCODE may
1175 be any of the <a href="#binaryops">binary</a> or <a href="#bitwiseops">bitwise
1176 binary</a> operations. The constraints on operands are the same as those for
1177 the corresponding instruction (e.g. no bitwise operations on floating point
1178 values are allowed).</dd>
1182 <!-- *********************************************************************** -->
1183 <div class="doc_section"> <a name="othervalues">Other Values</a> </div>
1184 <!-- *********************************************************************** -->
1186 <!-- ======================================================================= -->
1187 <div class="doc_subsection">
1188 <a name="inlineasm">Inline Assembler Expressions</a>
1191 <div class="doc_text">
1194 LLVM supports inline assembler expressions (as opposed to <a href="#moduleasm">
1195 Module-Level Inline Assembly</a>) through the use of a special value. This
1196 value represents the inline assembler as a string (containing the instructions
1197 to emit), a list of operand constraints (stored as a string), and a flag that
1198 indicates whether or not the inline asm expression has side effects. An example
1199 inline assembler expression is:
1203 int(int) asm "bswap $0", "=r,r"
1207 Inline assembler expressions may <b>only</b> be used as the callee operand of
1208 a <a href="#i_call"><tt>call</tt> instruction</a>. Thus, typically we have:
1212 %X = call int asm "<a href="#i_bswap">bswap</a> $0", "=r,r"(int %Y)
1216 Inline asms with side effects not visible in the constraint list must be marked
1217 as having side effects. This is done through the use of the
1218 '<tt>sideeffect</tt>' keyword, like so:
1222 call void asm sideeffect "eieio", ""()
1225 <p>TODO: The format of the asm and constraints string still need to be
1226 documented here. Constraints on what can be done (e.g. duplication, moving, etc
1227 need to be documented).
1232 <!-- *********************************************************************** -->
1233 <div class="doc_section"> <a name="instref">Instruction Reference</a> </div>
1234 <!-- *********************************************************************** -->
1236 <div class="doc_text">
1238 <p>The LLVM instruction set consists of several different
1239 classifications of instructions: <a href="#terminators">terminator
1240 instructions</a>, <a href="#binaryops">binary instructions</a>,
1241 <a href="#bitwiseops">bitwise binary instructions</a>, <a
1242 href="#memoryops">memory instructions</a>, and <a href="#otherops">other
1243 instructions</a>.</p>
1247 <!-- ======================================================================= -->
1248 <div class="doc_subsection"> <a name="terminators">Terminator
1249 Instructions</a> </div>
1251 <div class="doc_text">
1253 <p>As mentioned <a href="#functionstructure">previously</a>, every
1254 basic block in a program ends with a "Terminator" instruction, which
1255 indicates which block should be executed after the current block is
1256 finished. These terminator instructions typically yield a '<tt>void</tt>'
1257 value: they produce control flow, not values (the one exception being
1258 the '<a href="#i_invoke"><tt>invoke</tt></a>' instruction).</p>
1259 <p>There are six different terminator instructions: the '<a
1260 href="#i_ret"><tt>ret</tt></a>' instruction, the '<a href="#i_br"><tt>br</tt></a>'
1261 instruction, the '<a href="#i_switch"><tt>switch</tt></a>' instruction,
1262 the '<a href="#i_invoke"><tt>invoke</tt></a>' instruction, the '<a
1263 href="#i_unwind"><tt>unwind</tt></a>' instruction, and the '<a
1264 href="#i_unreachable"><tt>unreachable</tt></a>' instruction.</p>
1268 <!-- _______________________________________________________________________ -->
1269 <div class="doc_subsubsection"> <a name="i_ret">'<tt>ret</tt>'
1270 Instruction</a> </div>
1271 <div class="doc_text">
1273 <pre> ret <type> <value> <i>; Return a value from a non-void function</i>
1274 ret void <i>; Return from void function</i>
1277 <p>The '<tt>ret</tt>' instruction is used to return control flow (and a
1278 value) from a function back to the caller.</p>
1279 <p>There are two forms of the '<tt>ret</tt>' instruction: one that
1280 returns a value and then causes control flow, and one that just causes
1281 control flow to occur.</p>
1283 <p>The '<tt>ret</tt>' instruction may return any '<a
1284 href="#t_firstclass">first class</a>' type. Notice that a function is
1285 not <a href="#wellformed">well formed</a> if there exists a '<tt>ret</tt>'
1286 instruction inside of the function that returns a value that does not
1287 match the return type of the function.</p>
1289 <p>When the '<tt>ret</tt>' instruction is executed, control flow
1290 returns back to the calling function's context. If the caller is a "<a
1291 href="#i_call"><tt>call</tt></a>" instruction, execution continues at
1292 the instruction after the call. If the caller was an "<a
1293 href="#i_invoke"><tt>invoke</tt></a>" instruction, execution continues
1294 at the beginning of the "normal" destination block. If the instruction
1295 returns a value, that value shall set the call or invoke instruction's
1298 <pre> ret int 5 <i>; Return an integer value of 5</i>
1299 ret void <i>; Return from a void function</i>
1302 <!-- _______________________________________________________________________ -->
1303 <div class="doc_subsubsection"> <a name="i_br">'<tt>br</tt>' Instruction</a> </div>
1304 <div class="doc_text">
1306 <pre> br bool <cond>, label <iftrue>, label <iffalse><br> br label <dest> <i>; Unconditional branch</i>
1309 <p>The '<tt>br</tt>' instruction is used to cause control flow to
1310 transfer to a different basic block in the current function. There are
1311 two forms of this instruction, corresponding to a conditional branch
1312 and an unconditional branch.</p>
1314 <p>The conditional branch form of the '<tt>br</tt>' instruction takes a
1315 single '<tt>bool</tt>' value and two '<tt>label</tt>' values. The
1316 unconditional form of the '<tt>br</tt>' instruction takes a single '<tt>label</tt>'
1317 value as a target.</p>
1319 <p>Upon execution of a conditional '<tt>br</tt>' instruction, the '<tt>bool</tt>'
1320 argument is evaluated. If the value is <tt>true</tt>, control flows
1321 to the '<tt>iftrue</tt>' <tt>label</tt> argument. If "cond" is <tt>false</tt>,
1322 control flows to the '<tt>iffalse</tt>' <tt>label</tt> argument.</p>
1324 <pre>Test:<br> %cond = <a href="#i_setcc">seteq</a> int %a, %b<br> br bool %cond, label %IfEqual, label %IfUnequal<br>IfEqual:<br> <a
1325 href="#i_ret">ret</a> int 1<br>IfUnequal:<br> <a href="#i_ret">ret</a> int 0<br></pre>
1327 <!-- _______________________________________________________________________ -->
1328 <div class="doc_subsubsection">
1329 <a name="i_switch">'<tt>switch</tt>' Instruction</a>
1332 <div class="doc_text">
1336 switch <intty> <value>, label <defaultdest> [ <intty> <val>, label <dest> ... ]
1341 <p>The '<tt>switch</tt>' instruction is used to transfer control flow to one of
1342 several different places. It is a generalization of the '<tt>br</tt>'
1343 instruction, allowing a branch to occur to one of many possible
1349 <p>The '<tt>switch</tt>' instruction uses three parameters: an integer
1350 comparison value '<tt>value</tt>', a default '<tt>label</tt>' destination, and
1351 an array of pairs of comparison value constants and '<tt>label</tt>'s. The
1352 table is not allowed to contain duplicate constant entries.</p>
1356 <p>The <tt>switch</tt> instruction specifies a table of values and
1357 destinations. When the '<tt>switch</tt>' instruction is executed, this
1358 table is searched for the given value. If the value is found, control flow is
1359 transfered to the corresponding destination; otherwise, control flow is
1360 transfered to the default destination.</p>
1362 <h5>Implementation:</h5>
1364 <p>Depending on properties of the target machine and the particular
1365 <tt>switch</tt> instruction, this instruction may be code generated in different
1366 ways. For example, it could be generated as a series of chained conditional
1367 branches or with a lookup table.</p>
1372 <i>; Emulate a conditional br instruction</i>
1373 %Val = <a href="#i_cast">cast</a> bool %value to int
1374 switch int %Val, label %truedest [int 0, label %falsedest ]
1376 <i>; Emulate an unconditional br instruction</i>
1377 switch uint 0, label %dest [ ]
1379 <i>; Implement a jump table:</i>
1380 switch uint %val, label %otherwise [ uint 0, label %onzero
1381 uint 1, label %onone
1382 uint 2, label %ontwo ]
1386 <!-- _______________________________________________________________________ -->
1387 <div class="doc_subsubsection">
1388 <a name="i_invoke">'<tt>invoke</tt>' Instruction</a>
1391 <div class="doc_text">
1396 <result> = invoke [<a href="#callingconv">cconv</a>] <ptr to function ty> %<function ptr val>(<function args>)
1397 to label <normal label> unwind label <exception label>
1402 <p>The '<tt>invoke</tt>' instruction causes control to transfer to a specified
1403 function, with the possibility of control flow transfer to either the
1404 '<tt>normal</tt>' label or the
1405 '<tt>exception</tt>' label. If the callee function returns with the
1406 "<tt><a href="#i_ret">ret</a></tt>" instruction, control flow will return to the
1407 "normal" label. If the callee (or any indirect callees) returns with the "<a
1408 href="#i_unwind"><tt>unwind</tt></a>" instruction, control is interrupted and
1409 continued at the dynamically nearest "exception" label.</p>
1413 <p>This instruction requires several arguments:</p>
1417 The optional "cconv" marker indicates which <a href="callingconv">calling
1418 convention</a> the call should use. If none is specified, the call defaults
1419 to using C calling conventions.
1421 <li>'<tt>ptr to function ty</tt>': shall be the signature of the pointer to
1422 function value being invoked. In most cases, this is a direct function
1423 invocation, but indirect <tt>invoke</tt>s are just as possible, branching off
1424 an arbitrary pointer to function value.
1427 <li>'<tt>function ptr val</tt>': An LLVM value containing a pointer to a
1428 function to be invoked. </li>
1430 <li>'<tt>function args</tt>': argument list whose types match the function
1431 signature argument types. If the function signature indicates the function
1432 accepts a variable number of arguments, the extra arguments can be
1435 <li>'<tt>normal label</tt>': the label reached when the called function
1436 executes a '<tt><a href="#i_ret">ret</a></tt>' instruction. </li>
1438 <li>'<tt>exception label</tt>': the label reached when a callee returns with
1439 the <a href="#i_unwind"><tt>unwind</tt></a> instruction. </li>
1445 <p>This instruction is designed to operate as a standard '<tt><a
1446 href="#i_call">call</a></tt>' instruction in most regards. The primary
1447 difference is that it establishes an association with a label, which is used by
1448 the runtime library to unwind the stack.</p>
1450 <p>This instruction is used in languages with destructors to ensure that proper
1451 cleanup is performed in the case of either a <tt>longjmp</tt> or a thrown
1452 exception. Additionally, this is important for implementation of
1453 '<tt>catch</tt>' clauses in high-level languages that support them.</p>
1457 %retval = invoke int %Test(int 15) to label %Continue
1458 unwind label %TestCleanup <i>; {int}:retval set</i>
1459 %retval = invoke <a href="#callingconv">coldcc</a> int %Test(int 15) to label %Continue
1460 unwind label %TestCleanup <i>; {int}:retval set</i>
1465 <!-- _______________________________________________________________________ -->
1467 <div class="doc_subsubsection"> <a name="i_unwind">'<tt>unwind</tt>'
1468 Instruction</a> </div>
1470 <div class="doc_text">
1479 <p>The '<tt>unwind</tt>' instruction unwinds the stack, continuing control flow
1480 at the first callee in the dynamic call stack which used an <a
1481 href="#i_invoke"><tt>invoke</tt></a> instruction to perform the call. This is
1482 primarily used to implement exception handling.</p>
1486 <p>The '<tt>unwind</tt>' intrinsic causes execution of the current function to
1487 immediately halt. The dynamic call stack is then searched for the first <a
1488 href="#i_invoke"><tt>invoke</tt></a> instruction on the call stack. Once found,
1489 execution continues at the "exceptional" destination block specified by the
1490 <tt>invoke</tt> instruction. If there is no <tt>invoke</tt> instruction in the
1491 dynamic call chain, undefined behavior results.</p>
1494 <!-- _______________________________________________________________________ -->
1496 <div class="doc_subsubsection"> <a name="i_unreachable">'<tt>unreachable</tt>'
1497 Instruction</a> </div>
1499 <div class="doc_text">
1508 <p>The '<tt>unreachable</tt>' instruction has no defined semantics. This
1509 instruction is used to inform the optimizer that a particular portion of the
1510 code is not reachable. This can be used to indicate that the code after a
1511 no-return function cannot be reached, and other facts.</p>
1515 <p>The '<tt>unreachable</tt>' instruction has no defined semantics.</p>
1520 <!-- ======================================================================= -->
1521 <div class="doc_subsection"> <a name="binaryops">Binary Operations</a> </div>
1522 <div class="doc_text">
1523 <p>Binary operators are used to do most of the computation in a
1524 program. They require two operands, execute an operation on them, and
1525 produce a single value. The operands might represent
1526 multiple data, as is the case with the <a href="#t_packed">packed</a> data type.
1527 The result value of a binary operator is not
1528 necessarily the same type as its operands.</p>
1529 <p>There are several different binary operators:</p>
1531 <!-- _______________________________________________________________________ -->
1532 <div class="doc_subsubsection"> <a name="i_add">'<tt>add</tt>'
1533 Instruction</a> </div>
1534 <div class="doc_text">
1536 <pre> <result> = add <ty> <var1>, <var2> <i>; yields {ty}:result</i>
1539 <p>The '<tt>add</tt>' instruction returns the sum of its two operands.</p>
1541 <p>The two arguments to the '<tt>add</tt>' instruction must be either <a
1542 href="#t_integer">integer</a> or <a href="#t_floating">floating point</a> values.
1543 This instruction can also take <a href="#t_packed">packed</a> versions of the values.
1544 Both arguments must have identical types.</p>
1546 <p>The value produced is the integer or floating point sum of the two
1549 <pre> <result> = add int 4, %var <i>; yields {int}:result = 4 + %var</i>
1552 <!-- _______________________________________________________________________ -->
1553 <div class="doc_subsubsection"> <a name="i_sub">'<tt>sub</tt>'
1554 Instruction</a> </div>
1555 <div class="doc_text">
1557 <pre> <result> = sub <ty> <var1>, <var2> <i>; yields {ty}:result</i>
1560 <p>The '<tt>sub</tt>' instruction returns the difference of its two
1562 <p>Note that the '<tt>sub</tt>' instruction is used to represent the '<tt>neg</tt>'
1563 instruction present in most other intermediate representations.</p>
1565 <p>The two arguments to the '<tt>sub</tt>' instruction must be either <a
1566 href="#t_integer">integer</a> or <a href="#t_floating">floating point</a>
1568 This instruction can also take <a href="#t_packed">packed</a> versions of the values.
1569 Both arguments must have identical types.</p>
1571 <p>The value produced is the integer or floating point difference of
1572 the two operands.</p>
1574 <pre> <result> = sub int 4, %var <i>; yields {int}:result = 4 - %var</i>
1575 <result> = sub int 0, %val <i>; yields {int}:result = -%var</i>
1578 <!-- _______________________________________________________________________ -->
1579 <div class="doc_subsubsection"> <a name="i_mul">'<tt>mul</tt>'
1580 Instruction</a> </div>
1581 <div class="doc_text">
1583 <pre> <result> = mul <ty> <var1>, <var2> <i>; yields {ty}:result</i>
1586 <p>The '<tt>mul</tt>' instruction returns the product of its two
1589 <p>The two arguments to the '<tt>mul</tt>' instruction must be either <a
1590 href="#t_integer">integer</a> or <a href="#t_floating">floating point</a>
1592 This instruction can also take <a href="#t_packed">packed</a> versions of the values.
1593 Both arguments must have identical types.</p>
1595 <p>The value produced is the integer or floating point product of the
1597 <p>There is no signed vs unsigned multiplication. The appropriate
1598 action is taken based on the type of the operand.</p>
1600 <pre> <result> = mul int 4, %var <i>; yields {int}:result = 4 * %var</i>
1603 <!-- _______________________________________________________________________ -->
1604 <div class="doc_subsubsection"> <a name="i_div">'<tt>div</tt>'
1605 Instruction</a> </div>
1606 <div class="doc_text">
1608 <pre> <result> = div <ty> <var1>, <var2> <i>; yields {ty}:result</i>
1611 <p>The '<tt>div</tt>' instruction returns the quotient of its two
1614 <p>The two arguments to the '<tt>div</tt>' instruction must be either <a
1615 href="#t_integer">integer</a> or <a href="#t_floating">floating point</a>
1617 This instruction can also take <a href="#t_packed">packed</a> versions of the values.
1618 Both arguments must have identical types.</p>
1620 <p>The value produced is the integer or floating point quotient of the
1623 <pre> <result> = div int 4, %var <i>; yields {int}:result = 4 / %var</i>
1626 <!-- _______________________________________________________________________ -->
1627 <div class="doc_subsubsection"> <a name="i_rem">'<tt>rem</tt>'
1628 Instruction</a> </div>
1629 <div class="doc_text">
1631 <pre> <result> = rem <ty> <var1>, <var2> <i>; yields {ty}:result</i>
1634 <p>The '<tt>rem</tt>' instruction returns the remainder from the
1635 division of its two operands.</p>
1637 <p>The two arguments to the '<tt>rem</tt>' instruction must be either <a
1638 href="#t_integer">integer</a> or <a href="#t_floating">floating point</a>
1640 This instruction can also take <a href="#t_packed">packed</a> versions of the values.
1641 Both arguments must have identical types.</p>
1643 <p>This returns the <i>remainder</i> of a division (where the result
1644 has the same sign as the divisor), not the <i>modulus</i> (where the
1645 result has the same sign as the dividend) of a value. For more
1646 information about the difference, see <a
1647 href="http://mathforum.org/dr.math/problems/anne.4.28.99.html">The
1650 <pre> <result> = rem int 4, %var <i>; yields {int}:result = 4 % %var</i>
1654 <!-- _______________________________________________________________________ -->
1655 <div class="doc_subsubsection"> <a name="i_setcc">'<tt>set<i>cc</i></tt>'
1656 Instructions</a> </div>
1657 <div class="doc_text">
1659 <pre> <result> = seteq <ty> <var1>, <var2> <i>; yields {bool}:result</i>
1660 <result> = setne <ty> <var1>, <var2> <i>; yields {bool}:result</i>
1661 <result> = setlt <ty> <var1>, <var2> <i>; yields {bool}:result</i>
1662 <result> = setgt <ty> <var1>, <var2> <i>; yields {bool}:result</i>
1663 <result> = setle <ty> <var1>, <var2> <i>; yields {bool}:result</i>
1664 <result> = setge <ty> <var1>, <var2> <i>; yields {bool}:result</i>
1667 <p>The '<tt>set<i>cc</i></tt>' family of instructions returns a boolean
1668 value based on a comparison of their two operands.</p>
1670 <p>The two arguments to the '<tt>set<i>cc</i></tt>' instructions must
1671 be of <a href="#t_firstclass">first class</a> type (it is not possible
1672 to compare '<tt>label</tt>'s, '<tt>array</tt>'s, '<tt>structure</tt>'
1673 or '<tt>void</tt>' values, etc...). Both arguments must have identical
1676 <p>The '<tt>seteq</tt>' instruction yields a <tt>true</tt> '<tt>bool</tt>'
1677 value if both operands are equal.<br>
1678 The '<tt>setne</tt>' instruction yields a <tt>true</tt> '<tt>bool</tt>'
1679 value if both operands are unequal.<br>
1680 The '<tt>setlt</tt>' instruction yields a <tt>true</tt> '<tt>bool</tt>'
1681 value if the first operand is less than the second operand.<br>
1682 The '<tt>setgt</tt>' instruction yields a <tt>true</tt> '<tt>bool</tt>'
1683 value if the first operand is greater than the second operand.<br>
1684 The '<tt>setle</tt>' instruction yields a <tt>true</tt> '<tt>bool</tt>'
1685 value if the first operand is less than or equal to the second operand.<br>
1686 The '<tt>setge</tt>' instruction yields a <tt>true</tt> '<tt>bool</tt>'
1687 value if the first operand is greater than or equal to the second
1690 <pre> <result> = seteq int 4, 5 <i>; yields {bool}:result = false</i>
1691 <result> = setne float 4, 5 <i>; yields {bool}:result = true</i>
1692 <result> = setlt uint 4, 5 <i>; yields {bool}:result = true</i>
1693 <result> = setgt sbyte 4, 5 <i>; yields {bool}:result = false</i>
1694 <result> = setle sbyte 4, 5 <i>; yields {bool}:result = true</i>
1695 <result> = setge sbyte 4, 5 <i>; yields {bool}:result = false</i>
1699 <!-- ======================================================================= -->
1700 <div class="doc_subsection"> <a name="bitwiseops">Bitwise Binary
1701 Operations</a> </div>
1702 <div class="doc_text">
1703 <p>Bitwise binary operators are used to do various forms of
1704 bit-twiddling in a program. They are generally very efficient
1705 instructions and can commonly be strength reduced from other
1706 instructions. They require two operands, execute an operation on them,
1707 and produce a single value. The resulting value of the bitwise binary
1708 operators is always the same type as its first operand.</p>
1710 <!-- _______________________________________________________________________ -->
1711 <div class="doc_subsubsection"> <a name="i_and">'<tt>and</tt>'
1712 Instruction</a> </div>
1713 <div class="doc_text">
1715 <pre> <result> = and <ty> <var1>, <var2> <i>; yields {ty}:result</i>
1718 <p>The '<tt>and</tt>' instruction returns the bitwise logical and of
1719 its two operands.</p>
1721 <p>The two arguments to the '<tt>and</tt>' instruction must be <a
1722 href="#t_integral">integral</a> values. Both arguments must have
1723 identical types.</p>
1725 <p>The truth table used for the '<tt>and</tt>' instruction is:</p>
1727 <div style="align: center">
1728 <table border="1" cellspacing="0" cellpadding="4">
1759 <pre> <result> = and int 4, %var <i>; yields {int}:result = 4 & %var</i>
1760 <result> = and int 15, 40 <i>; yields {int}:result = 8</i>
1761 <result> = and int 4, 8 <i>; yields {int}:result = 0</i>
1764 <!-- _______________________________________________________________________ -->
1765 <div class="doc_subsubsection"> <a name="i_or">'<tt>or</tt>' Instruction</a> </div>
1766 <div class="doc_text">
1768 <pre> <result> = or <ty> <var1>, <var2> <i>; yields {ty}:result</i>
1771 <p>The '<tt>or</tt>' instruction returns the bitwise logical inclusive
1772 or of its two operands.</p>
1774 <p>The two arguments to the '<tt>or</tt>' instruction must be <a
1775 href="#t_integral">integral</a> values. Both arguments must have
1776 identical types.</p>
1778 <p>The truth table used for the '<tt>or</tt>' instruction is:</p>
1780 <div style="align: center">
1781 <table border="1" cellspacing="0" cellpadding="4">
1812 <pre> <result> = or int 4, %var <i>; yields {int}:result = 4 | %var</i>
1813 <result> = or int 15, 40 <i>; yields {int}:result = 47</i>
1814 <result> = or int 4, 8 <i>; yields {int}:result = 12</i>
1817 <!-- _______________________________________________________________________ -->
1818 <div class="doc_subsubsection"> <a name="i_xor">'<tt>xor</tt>'
1819 Instruction</a> </div>
1820 <div class="doc_text">
1822 <pre> <result> = xor <ty> <var1>, <var2> <i>; yields {ty}:result</i>
1825 <p>The '<tt>xor</tt>' instruction returns the bitwise logical exclusive
1826 or of its two operands. The <tt>xor</tt> is used to implement the
1827 "one's complement" operation, which is the "~" operator in C.</p>
1829 <p>The two arguments to the '<tt>xor</tt>' instruction must be <a
1830 href="#t_integral">integral</a> values. Both arguments must have
1831 identical types.</p>
1833 <p>The truth table used for the '<tt>xor</tt>' instruction is:</p>
1835 <div style="align: center">
1836 <table border="1" cellspacing="0" cellpadding="4">
1868 <pre> <result> = xor int 4, %var <i>; yields {int}:result = 4 ^ %var</i>
1869 <result> = xor int 15, 40 <i>; yields {int}:result = 39</i>
1870 <result> = xor int 4, 8 <i>; yields {int}:result = 12</i>
1871 <result> = xor int %V, -1 <i>; yields {int}:result = ~%V</i>
1874 <!-- _______________________________________________________________________ -->
1875 <div class="doc_subsubsection"> <a name="i_shl">'<tt>shl</tt>'
1876 Instruction</a> </div>
1877 <div class="doc_text">
1879 <pre> <result> = shl <ty> <var1>, ubyte <var2> <i>; yields {ty}:result</i>
1882 <p>The '<tt>shl</tt>' instruction returns the first operand shifted to
1883 the left a specified number of bits.</p>
1885 <p>The first argument to the '<tt>shl</tt>' instruction must be an <a
1886 href="#t_integer">integer</a> type. The second argument must be an '<tt>ubyte</tt>'
1889 <p>The value produced is <tt>var1</tt> * 2<sup><tt>var2</tt></sup>.</p>
1891 <pre> <result> = shl int 4, ubyte %var <i>; yields {int}:result = 4 << %var</i>
1892 <result> = shl int 4, ubyte 2 <i>; yields {int}:result = 16</i>
1893 <result> = shl int 1, ubyte 10 <i>; yields {int}:result = 1024</i>
1896 <!-- _______________________________________________________________________ -->
1897 <div class="doc_subsubsection"> <a name="i_shr">'<tt>shr</tt>'
1898 Instruction</a> </div>
1899 <div class="doc_text">
1901 <pre> <result> = shr <ty> <var1>, ubyte <var2> <i>; yields {ty}:result</i>
1904 <p>The '<tt>shr</tt>' instruction returns the first operand shifted to
1905 the right a specified number of bits.</p>
1907 <p>The first argument to the '<tt>shr</tt>' instruction must be an <a
1908 href="#t_integer">integer</a> type. The second argument must be an '<tt>ubyte</tt>'
1911 <p>If the first argument is a <a href="#t_signed">signed</a> type, the
1912 most significant bit is duplicated in the newly free'd bit positions.
1913 If the first argument is unsigned, zero bits shall fill the empty
1916 <pre> <result> = shr int 4, ubyte %var <i>; yields {int}:result = 4 >> %var</i>
1917 <result> = shr uint 4, ubyte 1 <i>; yields {uint}:result = 2</i>
1918 <result> = shr int 4, ubyte 2 <i>; yields {int}:result = 1</i>
1919 <result> = shr sbyte 4, ubyte 3 <i>; yields {sbyte}:result = 0</i>
1920 <result> = shr sbyte -2, ubyte 1 <i>; yields {sbyte}:result = -1</i>
1924 <!-- ======================================================================= -->
1925 <div class="doc_subsection">
1926 <a name="vectorops">Vector Operations</a>
1929 <div class="doc_text">
1931 <p>LLVM supports several instructions to represent vector operations in a
1932 target-independent manner. This instructions cover the element-access and
1933 vector-specific operations needed to process vectors effectively. While LLVM
1934 does directly support these vector operations, many sophisticated algorithms
1935 will want to use target-specific intrinsics to take full advantage of a specific
1940 <!-- _______________________________________________________________________ -->
1941 <div class="doc_subsubsection">
1942 <a name="i_extractelement">'<tt>extractelement</tt>' Instruction</a>
1945 <div class="doc_text">
1950 <result> = extractelement <n x <ty>> <val>, uint <idx> <i>; yields <ty></i>
1956 The '<tt>extractelement</tt>' instruction extracts a single scalar
1957 element from a packed vector at a specified index.
1964 The first operand of an '<tt>extractelement</tt>' instruction is a
1965 value of <a href="#t_packed">packed</a> type. The second operand is
1966 an index indicating the position from which to extract the element.
1967 The index may be a variable.</p>
1972 The result is a scalar of the same type as the element type of
1973 <tt>val</tt>. Its value is the value at position <tt>idx</tt> of
1974 <tt>val</tt>. If <tt>idx</tt> exceeds the length of <tt>val</tt>, the
1975 results are undefined.
1981 %result = extractelement <4 x int> %vec, uint 0 <i>; yields int</i>
1986 <!-- _______________________________________________________________________ -->
1987 <div class="doc_subsubsection">
1988 <a name="i_insertelement">'<tt>insertelement</tt>' Instruction</a>
1991 <div class="doc_text">
1996 <result> = insertelement <n x <ty>> <val>, <ty> <elt>, uint <idx> <i>; yields <n x <ty>></i>
2002 The '<tt>insertelement</tt>' instruction inserts a scalar
2003 element into a packed vector at a specified index.
2010 The first operand of an '<tt>insertelement</tt>' instruction is a
2011 value of <a href="#t_packed">packed</a> type. The second operand is a
2012 scalar value whose type must equal the element type of the first
2013 operand. The third operand is an index indicating the position at
2014 which to insert the value. The index may be a variable.</p>
2019 The result is a packed vector of the same type as <tt>val</tt>. Its
2020 element values are those of <tt>val</tt> except at position
2021 <tt>idx</tt>, where it gets the value <tt>elt</tt>. If <tt>idx</tt>
2022 exceeds the length of <tt>val</tt>, the results are undefined.
2028 %result = insertelement <4 x int> %vec, int 1, uint 0 <i>; yields <4 x int></i>
2032 <!-- _______________________________________________________________________ -->
2033 <div class="doc_subsubsection">
2034 <a name="i_shufflevector">'<tt>shufflevector</tt>' Instruction</a>
2037 <div class="doc_text">
2042 <result> = shufflevector <n x <ty>> <v1>, <n x <ty>> <v2>, <n x uint> <mask> <i>; yields <n x <ty>></i>
2048 The '<tt>shufflevector</tt>' instruction constructs a permutation of elements
2049 from two input vectors, returning a vector of the same type.
2055 The first two operands of a '<tt>shufflevector</tt>' instruction are vectors
2056 with types that match each other and types that match the result of the
2057 instruction. The third argument is a shuffle mask, which has the same number
2058 of elements as the other vector type, but whose element type is always 'uint'.
2062 The shuffle mask operand is required to be a constant vector with either
2063 constant integer or undef values.
2069 The elements of the two input vectors are numbered from left to right across
2070 both of the vectors. The shuffle mask operand specifies, for each element of
2071 the result vector, which element of the two input registers the result element
2072 gets. The element selector may be undef (meaning "don't care") and the second
2073 operand may be undef if performing a shuffle from only one vector.
2079 %result = shufflevector <4 x int> %v1, <4 x int> %v2,
2080 <4 x uint> <uint 0, uint 4, uint 1, uint 5> <i>; yields <4 x int></i>
2081 %result = shufflevector <4 x int> %v1, <4 x int> undef,
2082 <4 x uint> <uint 0, uint 1, uint 2, uint 3> <i>; yields <4 x int></i> - Identity shuffle.
2087 <!-- _______________________________________________________________________ -->
2088 <div class="doc_subsubsection"> <a name="i_vsetint">'<tt>vsetint</tt>'
2089 Instruction</a> </div>
2090 <div class="doc_text">
2092 <pre><result> = vsetint <op>, <n x <ty>> <var1>, <var2> <i>; yields <n x bool></i>
2097 <p>The '<tt>vsetint</tt>' instruction takes two integer vectors and
2098 returns a vector of boolean values representing, at each position, the
2099 result of the comparison between the values at that position in the
2104 <p>The arguments to a '<tt>vsetint</tt>' instruction are a comparison
2105 operation and two value arguments. The value arguments must be of <a
2106 href="#t_integral">integral</a> <a href="#t_packed">packed</a> type,
2107 and they must have identical types. The operation argument must be
2108 one of <tt>eq</tt>, <tt>ne</tt>, <tt>slt</tt>, <tt>sgt</tt>,
2109 <tt>sle</tt>, <tt>sge</tt>, <tt>ult</tt>, <tt>ugt</tt>, <tt>ule</tt>,
2110 <tt>uge</tt>, <tt>true</tt>, and <tt>false</tt>. The result is a
2111 packed <tt>bool</tt> value with the same length as each operand.</p>
2115 <p>The following table shows the semantics of '<tt>vsetint</tt>'. For
2116 each position of the result, the comparison is done on the
2117 corresponding positions of the two value arguments. Note that the
2118 signedness of the comparison depends on the comparison opcode and
2119 <i>not</i> on the signedness of the value operands. E.g., <tt>vsetint
2120 slt <4 x unsigned> %x, %y</tt> does an elementwise <i>signed</i>
2121 comparison of <tt>%x</tt> and <tt>%y</tt>.</p>
2123 <table border="1" cellspacing="0" cellpadding="4">
2125 <tr><th>Operation</th><th>Result is true iff</th><th>Comparison is</th></tr>
2126 <tr><td><tt>eq</tt></td><td>var1 == var2</td><td>--</td></tr>
2127 <tr><td><tt>ne</tt></td><td>var1 != var2</td><td>--</td></tr>
2128 <tr><td><tt>slt</tt></td><td>var1 < var2</td><td>signed</td></tr>
2129 <tr><td><tt>sgt</tt></td><td>var1 > var2</td><td>signed</td></tr>
2130 <tr><td><tt>sle</tt></td><td>var1 <= var2</td><td>signed</td></tr>
2131 <tr><td><tt>sge</tt></td><td>var1 >= var2</td><td>signed</td></tr>
2132 <tr><td><tt>ult</tt></td><td>var1 < var2</td><td>unsigned</td></tr>
2133 <tr><td><tt>ugt</tt></td><td>var1 > var2</td><td>unsigned</td></tr>
2134 <tr><td><tt>ule</tt></td><td>var1 <= var2</td><td>unsigned</td></tr>
2135 <tr><td><tt>uge</tt></td><td>var1 >= var2</td><td>unsigned</td></tr>
2136 <tr><td><tt>true</tt></td><td>always</td><td>--</td></tr>
2137 <tr><td><tt>false</tt></td><td>never</td><td>--</td></tr>
2142 <pre> <result> = vsetint eq <2 x int> <int 0, int 1>, <int 1, int 0> <i>; yields {<2 x bool>}:result = false, false</i>
2143 <result> = vsetint ne <2 x int> <int 0, int 1>, <int 1, int 0> <i>; yields {<2 x bool>}:result = true, true</i>
2144 <result> = vsetint slt <2 x int> <int 0, int 1>, <int 1, int 0> <i>; yields {<2 x bool>}:result = true, false</i>
2145 <result> = vsetint sgt <2 x int> <int 0, int 1>, <int 1, int 0> <i>; yields {<2 x bool>}:result = false, true</i>
2146 <result> = vsetint sle <2 x int> <int 0, int 1>, <int 1, int 0> <i>; yields {<2 x bool>}:result = true, false</i>
2147 <result> = vsetint sge <2 x int> <int 0, int 1>, <int 1, int 0> <i>; yields {<2 x bool>}:result = false, true</i>
2151 <!-- _______________________________________________________________________ -->
2152 <div class="doc_subsubsection"> <a name="i_vsetfp">'<tt>vsetfp</tt>'
2153 Instruction</a> </div>
2154 <div class="doc_text">
2156 <pre><result> = vsetfp <op>, <n x <ty>> <var1>, <var2> <i>; yields <n x bool></i>
2161 <p>The '<tt>vsetfp</tt>' instruction takes two floating point vector
2162 arguments and returns a vector of boolean values representing, at each
2163 position, the result of the comparison between the values at that
2164 position in the two operands.</p>
2168 <p>The arguments to a '<tt>vsetfp</tt>' instruction are a comparison
2169 operation and two value arguments. The value arguments must be of <a
2170 href="t_floating">floating point</a> <a href="#t_packed">packed</a>
2171 type, and they must have identical types. The operation argument must
2172 be one of <tt>eq</tt>, <tt>ne</tt>, <tt>lt</tt>, <tt>gt</tt>,
2173 <tt>le</tt>, <tt>ge</tt>, <tt>oeq</tt>, <tt>one</tt>, <tt>olt</tt>,
2174 <tt>ogt</tt>, <tt>ole</tt>, <tt>oge</tt>, <tt>ueq</tt>, <tt>une</tt>,
2175 <tt>ult</tt>, <tt>ugt</tt>, <tt>ule</tt>, <tt>uge</tt>, <tt>o</tt>,
2176 <tt>u</tt>, <tt>true</tt>, and <tt>false</tt>. The result is a packed
2177 <tt>bool</tt> value with the same length as each operand.</p>
2181 <p>The following table shows the semantics of '<tt>vsetfp</tt>' for
2182 floating point types. If either operand is a floating point Not a
2183 Number (NaN) value, the operation is unordered, and the value in the
2184 first column below is produced at that position. Otherwise, the
2185 operation is ordered, and the value in the second column is
2188 <table border="1" cellspacing="0" cellpadding="4">
2190 <tr><th>Operation</th><th>If unordered<th>Otherwise true iff</th></tr>
2191 <tr><td><tt>eq</tt></td><td>undefined</td><td>var1 == var2</td></tr>
2192 <tr><td><tt>ne</tt></td><td>undefined</td><td>var1 != var2</td></tr>
2193 <tr><td><tt>lt</tt></td><td>undefined</td><td>var1 < var2</td></tr>
2194 <tr><td><tt>gt</tt></td><td>undefined</td><td>var1 > var2</td></tr>
2195 <tr><td><tt>le</tt></td><td>undefined</td><td>var1 <= var2</td></tr>
2196 <tr><td><tt>ge</tt></td><td>undefined</td><td>var1 >= var2</td></tr>
2197 <tr><td><tt>oeq</tt></td><td>false</td><td>var1 == var2</td></tr>
2198 <tr><td><tt>one</tt></td><td>false</td><td>var1 != var2</td></tr>
2199 <tr><td><tt>olt</tt></td><td>false</td><td>var1 < var2</td></tr>
2200 <tr><td><tt>ogt</tt></td><td>false</td><td>var1 > var2</td></tr>
2201 <tr><td><tt>ole</tt></td><td>false</td><td>var1 <= var2</td></tr>
2202 <tr><td><tt>oge</tt></td><td>false</td><td>var1 >= var2</td></tr>
2203 <tr><td><tt>ueq</tt></td><td>true</td><td>var1 == var2</td></tr>
2204 <tr><td><tt>une</tt></td><td>true</td><td>var1 != var2</td></tr>
2205 <tr><td><tt>ult</tt></td><td>true</td><td>var1 < var2</td></tr>
2206 <tr><td><tt>ugt</tt></td><td>true</td><td>var1 > var2</td></tr>
2207 <tr><td><tt>ule</tt></td><td>true</td><td>var1 <= var2</td></tr>
2208 <tr><td><tt>uge</tt></td><td>true</td><td>var1 >= var2</td></tr>
2209 <tr><td><tt>o</tt></td><td>false</td><td>always</td></tr>
2210 <tr><td><tt>u</tt></td><td>true</td><td>never</td></tr>
2211 <tr><td><tt>true</tt></td><td>true</td><td>always</td></tr>
2212 <tr><td><tt>false</tt></td><td>false</td><td>never</td></tr>
2217 <pre> <result> = vsetfp eq <2 x float> <float 0.0, float 1.0>, <float 1.0, float 0.0> <i>; yields {<2 x bool>}:result = false, false</i>
2218 <result> = vsetfp ne <2 x float> <float 0.0, float 1.0>, <float 1.0, float 0.0> <i>; yields {<2 x bool>}:result = true, true</i>
2219 <result> = vsetfp lt <2 x float> <float 0.0, float 1.0>, <float 1.0, float 0.0> <i>; yields {<2 x bool>}:result = true, false</i>
2220 <result> = vsetfp gt <2 x float> <float 0.0, float 1.0>, <float 1.0, float 0.0> <i>; yields {<2 x bool>}:result = false, true</i>
2221 <result> = vsetfp le <2 x float> <float 0.0, float 1.0>, <float 1.0, float 0.0> <i>; yields {<2 x bool>}:result = true, false</i>
2222 <result> = vsetfp ge <2 x float> <float 0.0, float 1.0>, <float 1.0, float 0.0> <i>; yields {<2 x bool>}:result = false, true</i>
2226 <!-- _______________________________________________________________________ -->
2227 <div class="doc_subsubsection">
2228 <a name="i_vselect">'<tt>vselect</tt>' Instruction</a>
2231 <div class="doc_text">
2236 <result> = vselect <n x bool> <cond>, <n x <ty>> <val1>, <n x <ty>> <val2> <i>; yields <n x <ty>></i>
2242 The '<tt>vselect</tt>' instruction chooses one value at each position
2243 of a vector based on a condition.
2250 The '<tt>vselect</tt>' instruction requires a <a
2251 href="#t_packed">packed</a> <tt>bool</tt> value indicating the
2252 condition at each vector position, and two values of the same packed
2253 type. All three operands must have the same length. The type of the
2254 result is the same as the type of the two value operands.</p>
2259 At each position where the <tt>bool</tt> vector is true, that position
2260 of the result gets its value from the first value argument; otherwise,
2261 it gets its value from the second value argument.
2267 %X = vselect bool <2 x bool> <bool true, bool false>, <2 x ubyte> <ubyte 17, ubyte 17>,
2268 <2 x ubyte> <ubyte 42, ubyte 42> <i>; yields <2 x ubyte>:17, 42</i>
2274 <!-- ======================================================================= -->
2275 <div class="doc_subsection">
2276 <a name="memoryops">Memory Access and Addressing Operations</a>
2279 <div class="doc_text">
2281 <p>A key design point of an SSA-based representation is how it
2282 represents memory. In LLVM, no memory locations are in SSA form, which
2283 makes things very simple. This section describes how to read, write,
2284 allocate, and free memory in LLVM.</p>
2288 <!-- _______________________________________________________________________ -->
2289 <div class="doc_subsubsection">
2290 <a name="i_malloc">'<tt>malloc</tt>' Instruction</a>
2293 <div class="doc_text">
2298 <result> = malloc <type>[, uint <NumElements>][, align <alignment>] <i>; yields {type*}:result</i>
2303 <p>The '<tt>malloc</tt>' instruction allocates memory from the system
2304 heap and returns a pointer to it.</p>
2308 <p>The '<tt>malloc</tt>' instruction allocates
2309 <tt>sizeof(<type>)*NumElements</tt>
2310 bytes of memory from the operating system and returns a pointer of the
2311 appropriate type to the program. If "NumElements" is specified, it is the
2312 number of elements allocated. If an alignment is specified, the value result
2313 of the allocation is guaranteed to be aligned to at least that boundary. If
2314 not specified, or if zero, the target can choose to align the allocation on any
2315 convenient boundary.</p>
2317 <p>'<tt>type</tt>' must be a sized type.</p>
2321 <p>Memory is allocated using the system "<tt>malloc</tt>" function, and
2322 a pointer is returned.</p>
2327 %array = malloc [4 x ubyte ] <i>; yields {[%4 x ubyte]*}:array</i>
2329 %size = <a href="#i_add">add</a> uint 2, 2 <i>; yields {uint}:size = uint 4</i>
2330 %array1 = malloc ubyte, uint 4 <i>; yields {ubyte*}:array1</i>
2331 %array2 = malloc [12 x ubyte], uint %size <i>; yields {[12 x ubyte]*}:array2</i>
2332 %array3 = malloc int, uint 4, align 1024 <i>; yields {int*}:array3</i>
2333 %array4 = malloc int, align 1024 <i>; yields {int*}:array4</i>
2337 <!-- _______________________________________________________________________ -->
2338 <div class="doc_subsubsection">
2339 <a name="i_free">'<tt>free</tt>' Instruction</a>
2342 <div class="doc_text">
2347 free <type> <value> <i>; yields {void}</i>
2352 <p>The '<tt>free</tt>' instruction returns memory back to the unused
2353 memory heap to be reallocated in the future.</p>
2357 <p>'<tt>value</tt>' shall be a pointer value that points to a value
2358 that was allocated with the '<tt><a href="#i_malloc">malloc</a></tt>'
2363 <p>Access to the memory pointed to by the pointer is no longer defined
2364 after this instruction executes.</p>
2369 %array = <a href="#i_malloc">malloc</a> [4 x ubyte] <i>; yields {[4 x ubyte]*}:array</i>
2370 free [4 x ubyte]* %array
2374 <!-- _______________________________________________________________________ -->
2375 <div class="doc_subsubsection">
2376 <a name="i_alloca">'<tt>alloca</tt>' Instruction</a>
2379 <div class="doc_text">
2384 <result> = alloca <type>[, uint <NumElements>][, align <alignment>] <i>; yields {type*}:result</i>
2389 <p>The '<tt>alloca</tt>' instruction allocates memory on the current
2390 stack frame of the procedure that is live until the current function
2391 returns to its caller.</p>
2395 <p>The '<tt>alloca</tt>' instruction allocates <tt>sizeof(<type>)*NumElements</tt>
2396 bytes of memory on the runtime stack, returning a pointer of the
2397 appropriate type to the program. If "NumElements" is specified, it is the
2398 number of elements allocated. If an alignment is specified, the value result
2399 of the allocation is guaranteed to be aligned to at least that boundary. If
2400 not specified, or if zero, the target can choose to align the allocation on any
2401 convenient boundary.</p>
2403 <p>'<tt>type</tt>' may be any sized type.</p>
2407 <p>Memory is allocated; a pointer is returned. '<tt>alloca</tt>'d
2408 memory is automatically released when the function returns. The '<tt>alloca</tt>'
2409 instruction is commonly used to represent automatic variables that must
2410 have an address available. When the function returns (either with the <tt><a
2411 href="#i_ret">ret</a></tt> or <tt><a href="#i_unwind">unwind</a></tt>
2412 instructions), the memory is reclaimed.</p>
2417 %ptr = alloca int <i>; yields {int*}:ptr</i>
2418 %ptr = alloca int, uint 4 <i>; yields {int*}:ptr</i>
2419 %ptr = alloca int, uint 4, align 1024 <i>; yields {int*}:ptr</i>
2420 %ptr = alloca int, align 1024 <i>; yields {int*}:ptr</i>
2424 <!-- _______________________________________________________________________ -->
2425 <div class="doc_subsubsection"> <a name="i_load">'<tt>load</tt>'
2426 Instruction</a> </div>
2427 <div class="doc_text">
2429 <pre> <result> = load <ty>* <pointer><br> <result> = volatile load <ty>* <pointer><br></pre>
2431 <p>The '<tt>load</tt>' instruction is used to read from memory.</p>
2433 <p>The argument to the '<tt>load</tt>' instruction specifies the memory
2434 address from which to load. The pointer must point to a <a
2435 href="#t_firstclass">first class</a> type. If the <tt>load</tt> is
2436 marked as <tt>volatile</tt>, then the optimizer is not allowed to modify
2437 the number or order of execution of this <tt>load</tt> with other
2438 volatile <tt>load</tt> and <tt><a href="#i_store">store</a></tt>
2441 <p>The location of memory pointed to is loaded.</p>
2443 <pre> %ptr = <a href="#i_alloca">alloca</a> int <i>; yields {int*}:ptr</i>
2445 href="#i_store">store</a> int 3, int* %ptr <i>; yields {void}</i>
2446 %val = load int* %ptr <i>; yields {int}:val = int 3</i>
2449 <!-- _______________________________________________________________________ -->
2450 <div class="doc_subsubsection"> <a name="i_store">'<tt>store</tt>'
2451 Instruction</a> </div>
2453 <pre> store <ty> <value>, <ty>* <pointer> <i>; yields {void}</i>
2454 volatile store <ty> <value>, <ty>* <pointer> <i>; yields {void}</i>
2457 <p>The '<tt>store</tt>' instruction is used to write to memory.</p>
2459 <p>There are two arguments to the '<tt>store</tt>' instruction: a value
2460 to store and an address in which to store it. The type of the '<tt><pointer></tt>'
2461 operand must be a pointer to the type of the '<tt><value></tt>'
2462 operand. If the <tt>store</tt> is marked as <tt>volatile</tt>, then the
2463 optimizer is not allowed to modify the number or order of execution of
2464 this <tt>store</tt> with other volatile <tt>load</tt> and <tt><a
2465 href="#i_store">store</a></tt> instructions.</p>
2467 <p>The contents of memory are updated to contain '<tt><value></tt>'
2468 at the location specified by the '<tt><pointer></tt>' operand.</p>
2470 <pre> %ptr = <a href="#i_alloca">alloca</a> int <i>; yields {int*}:ptr</i>
2472 href="#i_store">store</a> int 3, int* %ptr <i>; yields {void}</i>
2473 %val = load int* %ptr <i>; yields {int}:val = int 3</i>
2475 <!-- _______________________________________________________________________ -->
2476 <div class="doc_subsubsection">
2477 <a name="i_getelementptr">'<tt>getelementptr</tt>' Instruction</a>
2480 <div class="doc_text">
2483 <result> = getelementptr <ty>* <ptrval>{, <ty> <idx>}*
2489 The '<tt>getelementptr</tt>' instruction is used to get the address of a
2490 subelement of an aggregate data structure.</p>
2494 <p>This instruction takes a list of integer constants that indicate what
2495 elements of the aggregate object to index to. The actual types of the arguments
2496 provided depend on the type of the first pointer argument. The
2497 '<tt>getelementptr</tt>' instruction is used to index down through the type
2498 levels of a structure or to a specific index in an array. When indexing into a
2499 structure, only <tt>uint</tt>
2500 integer constants are allowed. When indexing into an array or pointer,
2501 <tt>int</tt> and <tt>long</tt> indexes are allowed of any sign.</p>
2503 <p>For example, let's consider a C code fragment and how it gets
2504 compiled to LLVM:</p>
2518 int *foo(struct ST *s) {
2519 return &s[1].Z.B[5][13];
2523 <p>The LLVM code generated by the GCC frontend is:</p>
2526 %RT = type { sbyte, [10 x [20 x int]], sbyte }
2527 %ST = type { int, double, %RT }
2531 int* %foo(%ST* %s) {
2533 %reg = getelementptr %ST* %s, int 1, uint 2, uint 1, int 5, int 13
2540 <p>The index types specified for the '<tt>getelementptr</tt>' instruction depend
2541 on the pointer type that is being indexed into. <a href="#t_pointer">Pointer</a>
2542 and <a href="#t_array">array</a> types require <tt>uint</tt>, <tt>int</tt>,
2543 <tt>ulong</tt>, or <tt>long</tt> values, and <a href="#t_struct">structure</a>
2544 types require <tt>uint</tt> <b>constants</b>.</p>
2546 <p>In the example above, the first index is indexing into the '<tt>%ST*</tt>'
2547 type, which is a pointer, yielding a '<tt>%ST</tt>' = '<tt>{ int, double, %RT
2548 }</tt>' type, a structure. The second index indexes into the third element of
2549 the structure, yielding a '<tt>%RT</tt>' = '<tt>{ sbyte, [10 x [20 x int]],
2550 sbyte }</tt>' type, another structure. The third index indexes into the second
2551 element of the structure, yielding a '<tt>[10 x [20 x int]]</tt>' type, an
2552 array. The two dimensions of the array are subscripted into, yielding an
2553 '<tt>int</tt>' type. The '<tt>getelementptr</tt>' instruction returns a pointer
2554 to this element, thus computing a value of '<tt>int*</tt>' type.</p>
2556 <p>Note that it is perfectly legal to index partially through a
2557 structure, returning a pointer to an inner element. Because of this,
2558 the LLVM code for the given testcase is equivalent to:</p>
2561 int* %foo(%ST* %s) {
2562 %t1 = getelementptr %ST* %s, int 1 <i>; yields %ST*:%t1</i>
2563 %t2 = getelementptr %ST* %t1, int 0, uint 2 <i>; yields %RT*:%t2</i>
2564 %t3 = getelementptr %RT* %t2, int 0, uint 1 <i>; yields [10 x [20 x int]]*:%t3</i>
2565 %t4 = getelementptr [10 x [20 x int]]* %t3, int 0, int 5 <i>; yields [20 x int]*:%t4</i>
2566 %t5 = getelementptr [20 x int]* %t4, int 0, int 13 <i>; yields int*:%t5</i>
2571 <p>Note that it is undefined to access an array out of bounds: array and
2572 pointer indexes must always be within the defined bounds of the array type.
2573 The one exception for this rules is zero length arrays. These arrays are
2574 defined to be accessible as variable length arrays, which requires access
2575 beyond the zero'th element.</p>
2577 <p>The getelementptr instruction is often confusing. For some more insight
2578 into how it works, see <a href="GetElementPtr.html">the getelementptr
2584 <i>; yields [12 x ubyte]*:aptr</i>
2585 %aptr = getelementptr {int, [12 x ubyte]}* %sptr, long 0, uint 1
2589 <!-- ======================================================================= -->
2590 <div class="doc_subsection"> <a name="otherops">Other Operations</a> </div>
2591 <div class="doc_text">
2592 <p>The instructions in this category are the "miscellaneous"
2593 instructions, which defy better classification.</p>
2595 <!-- _______________________________________________________________________ -->
2596 <div class="doc_subsubsection"> <a name="i_phi">'<tt>phi</tt>'
2597 Instruction</a> </div>
2598 <div class="doc_text">
2600 <pre> <result> = phi <ty> [ <val0>, <label0>], ...<br></pre>
2602 <p>The '<tt>phi</tt>' instruction is used to implement the φ node in
2603 the SSA graph representing the function.</p>
2605 <p>The type of the incoming values are specified with the first type
2606 field. After this, the '<tt>phi</tt>' instruction takes a list of pairs
2607 as arguments, with one pair for each predecessor basic block of the
2608 current block. Only values of <a href="#t_firstclass">first class</a>
2609 type may be used as the value arguments to the PHI node. Only labels
2610 may be used as the label arguments.</p>
2611 <p>There must be no non-phi instructions between the start of a basic
2612 block and the PHI instructions: i.e. PHI instructions must be first in
2615 <p>At runtime, the '<tt>phi</tt>' instruction logically takes on the
2616 value specified by the parameter, depending on which basic block we
2617 came from in the last <a href="#terminators">terminator</a> instruction.</p>
2619 <pre>Loop: ; Infinite loop that counts from 0 on up...<br> %indvar = phi uint [ 0, %LoopHeader ], [ %nextindvar, %Loop ]<br> %nextindvar = add uint %indvar, 1<br> br label %Loop<br></pre>
2622 <!-- _______________________________________________________________________ -->
2623 <div class="doc_subsubsection">
2624 <a name="i_cast">'<tt>cast .. to</tt>' Instruction</a>
2627 <div class="doc_text">
2632 <result> = cast <ty> <value> to <ty2> <i>; yields ty2</i>
2638 The '<tt>cast</tt>' instruction is used as the primitive means to convert
2639 integers to floating point, change data type sizes, and break type safety (by
2647 The '<tt>cast</tt>' instruction takes a value to cast, which must be a first
2648 class value, and a type to cast it to, which must also be a <a
2649 href="#t_firstclass">first class</a> type.
2655 This instruction follows the C rules for explicit casts when determining how the
2656 data being cast must change to fit in its new container.
2660 When casting to bool, any value that would be considered true in the context of
2661 a C '<tt>if</tt>' condition is converted to the boolean '<tt>true</tt>' values,
2662 all else are '<tt>false</tt>'.
2666 When extending an integral value from a type of one signness to another (for
2667 example '<tt>sbyte</tt>' to '<tt>ulong</tt>'), the value is sign-extended if the
2668 <b>source</b> value is signed, and zero-extended if the source value is
2669 unsigned. <tt>bool</tt> values are always zero extended into either zero or
2676 %X = cast int 257 to ubyte <i>; yields ubyte:1</i>
2677 %Y = cast int 123 to bool <i>; yields bool:true</i>
2681 <!-- _______________________________________________________________________ -->
2682 <div class="doc_subsubsection">
2683 <a name="i_select">'<tt>select</tt>' Instruction</a>
2686 <div class="doc_text">
2691 <result> = select bool <cond>, <ty> <val1>, <ty> <val2> <i>; yields ty</i>
2697 The '<tt>select</tt>' instruction is used to choose one value based on a
2698 condition, without branching.
2705 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.
2711 If the boolean condition evaluates to true, the instruction returns the first
2712 value argument; otherwise, it returns the second value argument.
2718 %X = select bool true, ubyte 17, ubyte 42 <i>; yields ubyte:17</i>
2723 <!-- _______________________________________________________________________ -->
2724 <div class="doc_subsubsection">
2725 <a name="i_call">'<tt>call</tt>' Instruction</a>
2728 <div class="doc_text">
2732 <result> = [tail] call [<a href="#callingconv">cconv</a>] <ty>* <fnptrval>(<param list>)
2737 <p>The '<tt>call</tt>' instruction represents a simple function call.</p>
2741 <p>This instruction requires several arguments:</p>
2745 <p>The optional "tail" marker indicates whether the callee function accesses
2746 any allocas or varargs in the caller. If the "tail" marker is present, the
2747 function call is eligible for tail call optimization. Note that calls may
2748 be marked "tail" even if they do not occur before a <a
2749 href="#i_ret"><tt>ret</tt></a> instruction.
2752 <p>The optional "cconv" marker indicates which <a href="callingconv">calling
2753 convention</a> the call should use. If none is specified, the call defaults
2754 to using C calling conventions.
2757 <p>'<tt>ty</tt>': shall be the signature of the pointer to function value
2758 being invoked. The argument types must match the types implied by this
2759 signature. This type can be omitted if the function is not varargs and
2760 if the function type does not return a pointer to a function.</p>
2763 <p>'<tt>fnptrval</tt>': An LLVM value containing a pointer to a function to
2764 be invoked. In most cases, this is a direct function invocation, but
2765 indirect <tt>call</tt>s are just as possible, calling an arbitrary pointer
2766 to function value.</p>
2769 <p>'<tt>function args</tt>': argument list whose types match the
2770 function signature argument types. All arguments must be of
2771 <a href="#t_firstclass">first class</a> type. If the function signature
2772 indicates the function accepts a variable number of arguments, the extra
2773 arguments can be specified.</p>
2779 <p>The '<tt>call</tt>' instruction is used to cause control flow to
2780 transfer to a specified function, with its incoming arguments bound to
2781 the specified values. Upon a '<tt><a href="#i_ret">ret</a></tt>'
2782 instruction in the called function, control flow continues with the
2783 instruction after the function call, and the return value of the
2784 function is bound to the result argument. This is a simpler case of
2785 the <a href="#i_invoke">invoke</a> instruction.</p>
2790 %retval = call int %test(int %argc)
2791 call int(sbyte*, ...) *%printf(sbyte* %msg, int 12, sbyte 42);
2792 %X = tail call int %foo()
2793 %Y = tail call <a href="#callingconv">fastcc</a> int %foo()
2798 <!-- _______________________________________________________________________ -->
2799 <div class="doc_subsubsection">
2800 <a name="i_va_arg">'<tt>va_arg</tt>' Instruction</a>
2803 <div class="doc_text">
2808 <resultval> = va_arg <va_list*> <arglist>, <argty>
2813 <p>The '<tt>va_arg</tt>' instruction is used to access arguments passed through
2814 the "variable argument" area of a function call. It is used to implement the
2815 <tt>va_arg</tt> macro in C.</p>
2819 <p>This instruction takes a <tt>va_list*</tt> value and the type of
2820 the argument. It returns a value of the specified argument type and
2821 increments the <tt>va_list</tt> to point to the next argument. Again, the
2822 actual type of <tt>va_list</tt> is target specific.</p>
2826 <p>The '<tt>va_arg</tt>' instruction loads an argument of the specified
2827 type from the specified <tt>va_list</tt> and causes the
2828 <tt>va_list</tt> to point to the next argument. For more information,
2829 see the variable argument handling <a href="#int_varargs">Intrinsic
2832 <p>It is legal for this instruction to be called in a function which does not
2833 take a variable number of arguments, for example, the <tt>vfprintf</tt>
2836 <p><tt>va_arg</tt> is an LLVM instruction instead of an <a
2837 href="#intrinsics">intrinsic function</a> because it takes a type as an
2842 <p>See the <a href="#int_varargs">variable argument processing</a> section.</p>
2846 <!-- *********************************************************************** -->
2847 <div class="doc_section"> <a name="intrinsics">Intrinsic Functions</a> </div>
2848 <!-- *********************************************************************** -->
2850 <div class="doc_text">
2852 <p>LLVM supports the notion of an "intrinsic function". These functions have
2853 well known names and semantics and are required to follow certain
2854 restrictions. Overall, these instructions represent an extension mechanism for
2855 the LLVM language that does not require changing all of the transformations in
2856 LLVM to add to the language (or the bytecode reader/writer, the parser,
2859 <p>Intrinsic function names must all start with an "<tt>llvm.</tt>" prefix. This
2860 prefix is reserved in LLVM for intrinsic names; thus, functions may not be named
2861 this. Intrinsic functions must always be external functions: you cannot define
2862 the body of intrinsic functions. Intrinsic functions may only be used in call
2863 or invoke instructions: it is illegal to take the address of an intrinsic
2864 function. Additionally, because intrinsic functions are part of the LLVM
2865 language, it is required that they all be documented here if any are added.</p>
2868 <p>To learn how to add an intrinsic function, please see the <a
2869 href="ExtendingLLVM.html">Extending LLVM Guide</a>.
2874 <!-- ======================================================================= -->
2875 <div class="doc_subsection">
2876 <a name="int_varargs">Variable Argument Handling Intrinsics</a>
2879 <div class="doc_text">
2881 <p>Variable argument support is defined in LLVM with the <a
2882 href="#i_va_arg"><tt>va_arg</tt></a> instruction and these three
2883 intrinsic functions. These functions are related to the similarly
2884 named macros defined in the <tt><stdarg.h></tt> header file.</p>
2886 <p>All of these functions operate on arguments that use a
2887 target-specific value type "<tt>va_list</tt>". The LLVM assembly
2888 language reference manual does not define what this type is, so all
2889 transformations should be prepared to handle intrinsics with any type
2892 <p>This example shows how the <a href="#i_va_arg"><tt>va_arg</tt></a>
2893 instruction and the variable argument handling intrinsic functions are
2897 int %test(int %X, ...) {
2898 ; Initialize variable argument processing
2900 call void %<a href="#i_va_start">llvm.va_start</a>(sbyte** %ap)
2902 ; Read a single integer argument
2903 %tmp = va_arg sbyte** %ap, int
2905 ; Demonstrate usage of llvm.va_copy and llvm.va_end
2907 call void %<a href="#i_va_copy">llvm.va_copy</a>(sbyte** %aq, sbyte** %ap)
2908 call void %<a href="#i_va_end">llvm.va_end</a>(sbyte** %aq)
2910 ; Stop processing of arguments.
2911 call void %<a href="#i_va_end">llvm.va_end</a>(sbyte** %ap)
2917 <!-- _______________________________________________________________________ -->
2918 <div class="doc_subsubsection">
2919 <a name="i_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a>
2923 <div class="doc_text">
2925 <pre> declare void %llvm.va_start(<va_list>* <arglist>)<br></pre>
2927 <P>The '<tt>llvm.va_start</tt>' intrinsic initializes
2928 <tt>*<arglist></tt> for subsequent use by <tt><a
2929 href="#i_va_arg">va_arg</a></tt>.</p>
2933 <P>The argument is a pointer to a <tt>va_list</tt> element to initialize.</p>
2937 <P>The '<tt>llvm.va_start</tt>' intrinsic works just like the <tt>va_start</tt>
2938 macro available in C. In a target-dependent way, it initializes the
2939 <tt>va_list</tt> element the argument points to, so that the next call to
2940 <tt>va_arg</tt> will produce the first variable argument passed to the function.
2941 Unlike the C <tt>va_start</tt> macro, this intrinsic does not need to know the
2942 last argument of the function, the compiler can figure that out.</p>
2946 <!-- _______________________________________________________________________ -->
2947 <div class="doc_subsubsection">
2948 <a name="i_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a>
2951 <div class="doc_text">
2953 <pre> declare void %llvm.va_end(<va_list*> <arglist>)<br></pre>
2955 <p>The '<tt>llvm.va_end</tt>' intrinsic destroys <tt><arglist></tt>
2956 which has been initialized previously with <tt><a href="#i_va_start">llvm.va_start</a></tt>
2957 or <tt><a href="#i_va_copy">llvm.va_copy</a></tt>.</p>
2959 <p>The argument is a <tt>va_list</tt> to destroy.</p>
2961 <p>The '<tt>llvm.va_end</tt>' intrinsic works just like the <tt>va_end</tt>
2962 macro available in C. In a target-dependent way, it destroys the <tt>va_list</tt>.
2963 Calls to <a href="#i_va_start"><tt>llvm.va_start</tt></a> and <a
2964 href="#i_va_copy"><tt>llvm.va_copy</tt></a> must be matched exactly
2965 with calls to <tt>llvm.va_end</tt>.</p>
2968 <!-- _______________________________________________________________________ -->
2969 <div class="doc_subsubsection">
2970 <a name="i_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a>
2973 <div class="doc_text">
2978 declare void %llvm.va_copy(<va_list>* <destarglist>,
2979 <va_list>* <srcarglist>)
2984 <p>The '<tt>llvm.va_copy</tt>' intrinsic copies the current argument position from
2985 the source argument list to the destination argument list.</p>
2989 <p>The first argument is a pointer to a <tt>va_list</tt> element to initialize.
2990 The second argument is a pointer to a <tt>va_list</tt> element to copy from.</p>
2995 <p>The '<tt>llvm.va_copy</tt>' intrinsic works just like the <tt>va_copy</tt> macro
2996 available in C. In a target-dependent way, it copies the source
2997 <tt>va_list</tt> element into the destination list. This intrinsic is necessary
2998 because the <tt><a href="i_va_begin">llvm.va_begin</a></tt> intrinsic may be
2999 arbitrarily complex and require memory allocation, for example.</p>
3003 <!-- ======================================================================= -->
3004 <div class="doc_subsection">
3005 <a name="int_gc">Accurate Garbage Collection Intrinsics</a>
3008 <div class="doc_text">
3011 LLVM support for <a href="GarbageCollection.html">Accurate Garbage
3012 Collection</a> requires the implementation and generation of these intrinsics.
3013 These intrinsics allow identification of <a href="#i_gcroot">GC roots on the
3014 stack</a>, as well as garbage collector implementations that require <a
3015 href="#i_gcread">read</a> and <a href="#i_gcwrite">write</a> barriers.
3016 Front-ends for type-safe garbage collected languages should generate these
3017 intrinsics to make use of the LLVM garbage collectors. For more details, see <a
3018 href="GarbageCollection.html">Accurate Garbage Collection with LLVM</a>.
3022 <!-- _______________________________________________________________________ -->
3023 <div class="doc_subsubsection">
3024 <a name="i_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a>
3027 <div class="doc_text">
3032 declare void %llvm.gcroot(<ty>** %ptrloc, <ty2>* %metadata)
3037 <p>The '<tt>llvm.gcroot</tt>' intrinsic declares the existence of a GC root to
3038 the code generator, and allows some metadata to be associated with it.</p>
3042 <p>The first argument specifies the address of a stack object that contains the
3043 root pointer. The second pointer (which must be either a constant or a global
3044 value address) contains the meta-data to be associated with the root.</p>
3048 <p>At runtime, a call to this intrinsics stores a null pointer into the "ptrloc"
3049 location. At compile-time, the code generator generates information to allow
3050 the runtime to find the pointer at GC safe points.
3056 <!-- _______________________________________________________________________ -->
3057 <div class="doc_subsubsection">
3058 <a name="i_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a>
3061 <div class="doc_text">
3066 declare sbyte* %llvm.gcread(sbyte* %ObjPtr, sbyte** %Ptr)
3071 <p>The '<tt>llvm.gcread</tt>' intrinsic identifies reads of references from heap
3072 locations, allowing garbage collector implementations that require read
3077 <p>The second argument is the address to read from, which should be an address
3078 allocated from the garbage collector. The first object is a pointer to the
3079 start of the referenced object, if needed by the language runtime (otherwise
3084 <p>The '<tt>llvm.gcread</tt>' intrinsic has the same semantics as a load
3085 instruction, but may be replaced with substantially more complex code by the
3086 garbage collector runtime, as needed.</p>
3091 <!-- _______________________________________________________________________ -->
3092 <div class="doc_subsubsection">
3093 <a name="i_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a>
3096 <div class="doc_text">
3101 declare void %llvm.gcwrite(sbyte* %P1, sbyte* %Obj, sbyte** %P2)
3106 <p>The '<tt>llvm.gcwrite</tt>' intrinsic identifies writes of references to heap
3107 locations, allowing garbage collector implementations that require write
3108 barriers (such as generational or reference counting collectors).</p>
3112 <p>The first argument is the reference to store, the second is the start of the
3113 object to store it to, and the third is the address of the field of Obj to
3114 store to. If the runtime does not require a pointer to the object, Obj may be
3119 <p>The '<tt>llvm.gcwrite</tt>' intrinsic has the same semantics as a store
3120 instruction, but may be replaced with substantially more complex code by the
3121 garbage collector runtime, as needed.</p>
3127 <!-- ======================================================================= -->
3128 <div class="doc_subsection">
3129 <a name="int_codegen">Code Generator Intrinsics</a>
3132 <div class="doc_text">
3134 These intrinsics are provided by LLVM to expose special features that may only
3135 be implemented with code generator support.
3140 <!-- _______________________________________________________________________ -->
3141 <div class="doc_subsubsection">
3142 <a name="i_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a>
3145 <div class="doc_text">
3149 declare sbyte *%llvm.returnaddress(uint <level>)
3155 The '<tt>llvm.returnaddress</tt>' intrinsic returns a target-specific value
3156 indicating the return address of the current function or one of its callers.
3162 The argument to this intrinsic indicates which function to return the address
3163 for. Zero indicates the calling function, one indicates its caller, etc. The
3164 argument is <b>required</b> to be a constant integer value.
3170 The '<tt>llvm.returnaddress</tt>' intrinsic either returns a pointer indicating
3171 the return address of the specified call frame, or zero if it cannot be
3172 identified. The value returned by this intrinsic is likely to be incorrect or 0
3173 for arguments other than zero, so it should only be used for debugging purposes.
3177 Note that calling this intrinsic does not prevent function inlining or other
3178 aggressive transformations, so the value returned may not be that of the obvious
3179 source-language caller.
3184 <!-- _______________________________________________________________________ -->
3185 <div class="doc_subsubsection">
3186 <a name="i_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a>
3189 <div class="doc_text">
3193 declare sbyte *%llvm.frameaddress(uint <level>)
3199 The '<tt>llvm.frameaddress</tt>' intrinsic returns the target-specific frame
3200 pointer value for the specified stack frame.
3206 The argument to this intrinsic indicates which function to return the frame
3207 pointer for. Zero indicates the calling function, one indicates its caller,
3208 etc. The argument is <b>required</b> to be a constant integer value.
3214 The '<tt>llvm.frameaddress</tt>' intrinsic either returns a pointer indicating
3215 the frame address of the specified call frame, or zero if it cannot be
3216 identified. The value returned by this intrinsic is likely to be incorrect or 0
3217 for arguments other than zero, so it should only be used for debugging purposes.
3221 Note that calling this intrinsic does not prevent function inlining or other
3222 aggressive transformations, so the value returned may not be that of the obvious
3223 source-language caller.
3227 <!-- _______________________________________________________________________ -->
3228 <div class="doc_subsubsection">
3229 <a name="i_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a>
3232 <div class="doc_text">
3236 declare sbyte *%llvm.stacksave()
3242 The '<tt>llvm.stacksave</tt>' intrinsic is used to remember the current state of
3243 the function stack, for use with <a href="#i_stackrestore">
3244 <tt>llvm.stackrestore</tt></a>. This is useful for implementing language
3245 features like scoped automatic variable sized arrays in C99.
3251 This intrinsic returns a opaque pointer value that can be passed to <a
3252 href="#i_stackrestore"><tt>llvm.stackrestore</tt></a>. When an
3253 <tt>llvm.stackrestore</tt> intrinsic is executed with a value saved from
3254 <tt>llvm.stacksave</tt>, it effectively restores the state of the stack to the
3255 state it was in when the <tt>llvm.stacksave</tt> intrinsic executed. In
3256 practice, this pops any <a href="#i_alloca">alloca</a> blocks from the stack
3257 that were allocated after the <tt>llvm.stacksave</tt> was executed.
3262 <!-- _______________________________________________________________________ -->
3263 <div class="doc_subsubsection">
3264 <a name="i_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a>
3267 <div class="doc_text">
3271 declare void %llvm.stackrestore(sbyte* %ptr)
3277 The '<tt>llvm.stackrestore</tt>' intrinsic is used to restore the state of
3278 the function stack to the state it was in when the corresponding <a
3279 href="#llvm.stacksave"><tt>llvm.stacksave</tt></a> intrinsic executed. This is
3280 useful for implementing language features like scoped automatic variable sized
3287 See the description for <a href="#i_stacksave"><tt>llvm.stacksave</tt></a>.
3293 <!-- _______________________________________________________________________ -->
3294 <div class="doc_subsubsection">
3295 <a name="i_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a>
3298 <div class="doc_text">
3302 declare void %llvm.prefetch(sbyte * <address>,
3303 uint <rw>, uint <locality>)
3310 The '<tt>llvm.prefetch</tt>' intrinsic is a hint to the code generator to insert
3311 a prefetch instruction if supported; otherwise, it is a noop. Prefetches have
3313 effect on the behavior of the program but can change its performance
3320 <tt>address</tt> is the address to be prefetched, <tt>rw</tt> is the specifier
3321 determining if the fetch should be for a read (0) or write (1), and
3322 <tt>locality</tt> is a temporal locality specifier ranging from (0) - no
3323 locality, to (3) - extremely local keep in cache. The <tt>rw</tt> and
3324 <tt>locality</tt> arguments must be constant integers.
3330 This intrinsic does not modify the behavior of the program. In particular,
3331 prefetches cannot trap and do not produce a value. On targets that support this
3332 intrinsic, the prefetch can provide hints to the processor cache for better
3338 <!-- _______________________________________________________________________ -->
3339 <div class="doc_subsubsection">
3340 <a name="i_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a>
3343 <div class="doc_text">
3347 declare void %llvm.pcmarker( uint <id> )
3354 The '<tt>llvm.pcmarker</tt>' intrinsic is a method to export a Program Counter
3356 code to simulators and other tools. The method is target specific, but it is
3357 expected that the marker will use exported symbols to transmit the PC of the marker.
3358 The marker makes no guarantees that it will remain with any specific instruction
3359 after optimizations. It is possible that the presence of a marker will inhibit
3360 optimizations. The intended use is to be inserted after optimizations to allow
3361 correlations of simulation runs.
3367 <tt>id</tt> is a numerical id identifying the marker.
3373 This intrinsic does not modify the behavior of the program. Backends that do not
3374 support this intrinisic may ignore it.
3379 <!-- _______________________________________________________________________ -->
3380 <div class="doc_subsubsection">
3381 <a name="i_readcyclecounter">'<tt>llvm.readcyclecounter</tt>' Intrinsic</a>
3384 <div class="doc_text">
3388 declare ulong %llvm.readcyclecounter( )
3395 The '<tt>llvm.readcyclecounter</tt>' intrinsic provides access to the cycle
3396 counter register (or similar low latency, high accuracy clocks) on those targets
3397 that support it. On X86, it should map to RDTSC. On Alpha, it should map to RPCC.
3398 As the backing counters overflow quickly (on the order of 9 seconds on alpha), this
3399 should only be used for small timings.
3405 When directly supported, reading the cycle counter should not modify any memory.
3406 Implementations are allowed to either return a application specific value or a
3407 system wide value. On backends without support, this is lowered to a constant 0.
3412 <!-- ======================================================================= -->
3413 <div class="doc_subsection">
3414 <a name="int_libc">Standard C Library Intrinsics</a>
3417 <div class="doc_text">
3419 LLVM provides intrinsics for a few important standard C library functions.
3420 These intrinsics allow source-language front-ends to pass information about the
3421 alignment of the pointer arguments to the code generator, providing opportunity
3422 for more efficient code generation.
3427 <!-- _______________________________________________________________________ -->
3428 <div class="doc_subsubsection">
3429 <a name="i_memcpy">'<tt>llvm.memcpy</tt>' Intrinsic</a>
3432 <div class="doc_text">
3436 declare void %llvm.memcpy.i32(sbyte* <dest>, sbyte* <src>,
3437 uint <len>, uint <align>)
3438 declare void %llvm.memcpy.i64(sbyte* <dest>, sbyte* <src>,
3439 ulong <len>, uint <align>)
3445 The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the source
3446 location to the destination location.
3450 Note that, unlike the standard libc function, the <tt>llvm.memcpy.*</tt>
3451 intrinsics do not return a value, and takes an extra alignment argument.
3457 The first argument is a pointer to the destination, the second is a pointer to
3458 the source. The third argument is an integer argument
3459 specifying the number of bytes to copy, and the fourth argument is the alignment
3460 of the source and destination locations.
3464 If the call to this intrinisic has an alignment value that is not 0 or 1, then
3465 the caller guarantees that both the source and destination pointers are aligned
3472 The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the source
3473 location to the destination location, which are not allowed to overlap. It
3474 copies "len" bytes of memory over. If the argument is known to be aligned to
3475 some boundary, this can be specified as the fourth argument, otherwise it should
3481 <!-- _______________________________________________________________________ -->
3482 <div class="doc_subsubsection">
3483 <a name="i_memmove">'<tt>llvm.memmove</tt>' Intrinsic</a>
3486 <div class="doc_text">
3490 declare void %llvm.memmove.i32(sbyte* <dest>, sbyte* <src>,
3491 uint <len>, uint <align>)
3492 declare void %llvm.memmove.i64(sbyte* <dest>, sbyte* <src>,
3493 ulong <len>, uint <align>)
3499 The '<tt>llvm.memmove.*</tt>' intrinsics move a block of memory from the source
3500 location to the destination location. It is similar to the
3501 '<tt>llvm.memcmp</tt>' intrinsic but allows the two memory locations to overlap.
3505 Note that, unlike the standard libc function, the <tt>llvm.memmove.*</tt>
3506 intrinsics do not return a value, and takes an extra alignment argument.
3512 The first argument is a pointer to the destination, the second is a pointer to
3513 the source. The third argument is an integer argument
3514 specifying the number of bytes to copy, and the fourth argument is the alignment
3515 of the source and destination locations.
3519 If the call to this intrinisic has an alignment value that is not 0 or 1, then
3520 the caller guarantees that the source and destination pointers are aligned to
3527 The '<tt>llvm.memmove.*</tt>' intrinsics copy a block of memory from the source
3528 location to the destination location, which may overlap. It
3529 copies "len" bytes of memory over. If the argument is known to be aligned to
3530 some boundary, this can be specified as the fourth argument, otherwise it should
3536 <!-- _______________________________________________________________________ -->
3537 <div class="doc_subsubsection">
3538 <a name="i_memset">'<tt>llvm.memset.*</tt>' Intrinsics</a>
3541 <div class="doc_text">
3545 declare void %llvm.memset.i32(sbyte* <dest>, ubyte <val>,
3546 uint <len>, uint <align>)
3547 declare void %llvm.memset.i64(sbyte* <dest>, ubyte <val>,
3548 ulong <len>, uint <align>)
3554 The '<tt>llvm.memset.*</tt>' intrinsics fill a block of memory with a particular
3559 Note that, unlike the standard libc function, the <tt>llvm.memset</tt> intrinsic
3560 does not return a value, and takes an extra alignment argument.
3566 The first argument is a pointer to the destination to fill, the second is the
3567 byte value to fill it with, the third argument is an integer
3568 argument specifying the number of bytes to fill, and the fourth argument is the
3569 known alignment of destination location.
3573 If the call to this intrinisic has an alignment value that is not 0 or 1, then
3574 the caller guarantees that the destination pointer is aligned to that boundary.
3580 The '<tt>llvm.memset.*</tt>' intrinsics fill "len" bytes of memory starting at
3582 destination location. If the argument is known to be aligned to some boundary,
3583 this can be specified as the fourth argument, otherwise it should be set to 0 or
3589 <!-- _______________________________________________________________________ -->
3590 <div class="doc_subsubsection">
3591 <a name="i_isunordered">'<tt>llvm.isunordered.*</tt>' Intrinsic</a>
3594 <div class="doc_text">
3598 declare bool %llvm.isunordered.f32(float Val1, float Val2)
3599 declare bool %llvm.isunordered.f64(double Val1, double Val2)
3605 The '<tt>llvm.isunordered</tt>' intrinsics return true if either or both of the
3606 specified floating point values is a NAN.
3612 The arguments are floating point numbers of the same type.
3618 If either or both of the arguments is a SNAN or QNAN, it returns true, otherwise
3624 <!-- _______________________________________________________________________ -->
3625 <div class="doc_subsubsection">
3626 <a name="i_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a>
3629 <div class="doc_text">
3633 declare double %llvm.sqrt.f32(float Val)
3634 declare double %llvm.sqrt.f64(double Val)
3640 The '<tt>llvm.sqrt</tt>' intrinsics return the sqrt of the specified operand,
3641 returning the same value as the libm '<tt>sqrt</tt>' function would. Unlike
3642 <tt>sqrt</tt> in libm, however, <tt>llvm.sqrt</tt> has undefined behavior for
3643 negative numbers (which allows for better optimization).
3649 The argument and return value are floating point numbers of the same type.
3655 This function returns the sqrt of the specified operand if it is a positive
3656 floating point number.
3660 <!-- ======================================================================= -->
3661 <div class="doc_subsection">
3662 <a name="int_manip">Bit Manipulation Intrinsics</a>
3665 <div class="doc_text">
3667 LLVM provides intrinsics for a few important bit manipulation operations.
3668 These allow efficient code generation for some algorithms.
3673 <!-- _______________________________________________________________________ -->
3674 <div class="doc_subsubsection">
3675 <a name="i_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a>
3678 <div class="doc_text">
3682 declare ushort %llvm.bswap.i16(ushort <id>)
3683 declare uint %llvm.bswap.i32(uint <id>)
3684 declare ulong %llvm.bswap.i64(ulong <id>)
3690 The '<tt>llvm.bwsap</tt>' family of intrinsics is used to byteswap a 16, 32 or
3691 64 bit quantity. These are useful for performing operations on data that is not
3692 in the target's native byte order.
3698 The <tt>llvm.bswap.16</tt> intrinsic returns a ushort value that has the high and low
3699 byte of the input ushort swapped. Similarly, the <tt>llvm.bswap.i32</tt> intrinsic
3700 returns a uint value that has the four bytes of the input uint swapped, so that
3701 if the input bytes are numbered 0, 1, 2, 3 then the returned uint will have its
3702 bytes in 3, 2, 1, 0 order. The <tt>llvm.bswap.i64</tt> intrinsic extends this concept
3708 <!-- _______________________________________________________________________ -->
3709 <div class="doc_subsubsection">
3710 <a name="int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic</a>
3713 <div class="doc_text">
3717 declare ubyte %llvm.ctpop.i8 (ubyte <src>)
3718 declare ushort %llvm.ctpop.i16(ushort <src>)
3719 declare uint %llvm.ctpop.i32(uint <src>)
3720 declare ulong %llvm.ctpop.i64(ulong <src>)
3726 The '<tt>llvm.ctpop</tt>' family of intrinsics counts the number of bits set in a
3733 The only argument is the value to be counted. The argument may be of any
3734 unsigned integer type. The return type must match the argument type.
3740 The '<tt>llvm.ctpop</tt>' intrinsic counts the 1's in a variable.
3744 <!-- _______________________________________________________________________ -->
3745 <div class="doc_subsubsection">
3746 <a name="int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic</a>
3749 <div class="doc_text">
3753 declare ubyte %llvm.ctlz.i8 (ubyte <src>)
3754 declare ushort %llvm.ctlz.i16(ushort <src>)
3755 declare uint %llvm.ctlz.i32(uint <src>)
3756 declare ulong %llvm.ctlz.i64(ulong <src>)
3762 The '<tt>llvm.ctlz</tt>' family of intrinsic functions counts the number of
3763 leading zeros in a variable.
3769 The only argument is the value to be counted. The argument may be of any
3770 unsigned integer type. The return type must match the argument type.
3776 The '<tt>llvm.ctlz</tt>' intrinsic counts the leading (most significant) zeros
3777 in a variable. If the src == 0 then the result is the size in bits of the type
3778 of src. For example, <tt>llvm.ctlz(int 2) = 30</tt>.
3784 <!-- _______________________________________________________________________ -->
3785 <div class="doc_subsubsection">
3786 <a name="int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic</a>
3789 <div class="doc_text">
3793 declare ubyte %llvm.cttz.i8 (ubyte <src>)
3794 declare ushort %llvm.cttz.i16(ushort <src>)
3795 declare uint %llvm.cttz.i32(uint <src>)
3796 declare ulong %llvm.cttz.i64(ulong <src>)
3802 The '<tt>llvm.cttz</tt>' family of intrinsic functions counts the number of
3809 The only argument is the value to be counted. The argument may be of any
3810 unsigned integer type. The return type must match the argument type.
3816 The '<tt>llvm.cttz</tt>' intrinsic counts the trailing (least significant) zeros
3817 in a variable. If the src == 0 then the result is the size in bits of the type
3818 of src. For example, <tt>llvm.cttz(2) = 1</tt>.
3822 <!-- ======================================================================= -->
3823 <div class="doc_subsection">
3824 <a name="int_debugger">Debugger Intrinsics</a>
3827 <div class="doc_text">
3829 The LLVM debugger intrinsics (which all start with <tt>llvm.dbg.</tt> prefix),
3830 are described in the <a
3831 href="SourceLevelDebugging.html#format_common_intrinsics">LLVM Source Level
3832 Debugging</a> document.
3837 <!-- *********************************************************************** -->
3840 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
3841 src="http://jigsaw.w3.org/css-validator/images/vcss" alt="Valid CSS!"></a>
3842 <a href="http://validator.w3.org/check/referer"><img
3843 src="http://www.w3.org/Icons/valid-html401" alt="Valid HTML 4.01!" /></a>
3845 <a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
3846 <a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br>
3847 Last modified: $Date$