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="#paramattrs">Parameter Attributes</a></li>
28 <li><a href="#moduleasm">Module-Level Inline Assembly</a></li>
31 <li><a href="#typesystem">Type System</a>
33 <li><a href="#t_primitive">Primitive Types</a>
35 <li><a href="#t_classifications">Type Classifications</a></li>
38 <li><a href="#t_derived">Derived Types</a>
40 <li><a href="#t_array">Array Type</a></li>
41 <li><a href="#t_function">Function Type</a></li>
42 <li><a href="#t_pointer">Pointer Type</a></li>
43 <li><a href="#t_struct">Structure Type</a></li>
44 <li><a href="#t_pstruct">Packed Structure Type</a></li>
45 <li><a href="#t_packed">Packed Type</a></li>
46 <li><a href="#t_opaque">Opaque Type</a></li>
51 <li><a href="#constants">Constants</a>
53 <li><a href="#simpleconstants">Simple Constants</a>
54 <li><a href="#aggregateconstants">Aggregate Constants</a>
55 <li><a href="#globalconstants">Global Variable and Function Addresses</a>
56 <li><a href="#undefvalues">Undefined Values</a>
57 <li><a href="#constantexprs">Constant Expressions</a>
60 <li><a href="#othervalues">Other Values</a>
62 <li><a href="#inlineasm">Inline Assembler Expressions</a>
65 <li><a href="#instref">Instruction Reference</a>
67 <li><a href="#terminators">Terminator Instructions</a>
69 <li><a href="#i_ret">'<tt>ret</tt>' Instruction</a></li>
70 <li><a href="#i_br">'<tt>br</tt>' Instruction</a></li>
71 <li><a href="#i_switch">'<tt>switch</tt>' Instruction</a></li>
72 <li><a href="#i_invoke">'<tt>invoke</tt>' Instruction</a></li>
73 <li><a href="#i_unwind">'<tt>unwind</tt>' Instruction</a></li>
74 <li><a href="#i_unreachable">'<tt>unreachable</tt>' Instruction</a></li>
77 <li><a href="#binaryops">Binary Operations</a>
79 <li><a href="#i_add">'<tt>add</tt>' Instruction</a></li>
80 <li><a href="#i_sub">'<tt>sub</tt>' Instruction</a></li>
81 <li><a href="#i_mul">'<tt>mul</tt>' Instruction</a></li>
82 <li><a href="#i_udiv">'<tt>udiv</tt>' Instruction</a></li>
83 <li><a href="#i_sdiv">'<tt>sdiv</tt>' Instruction</a></li>
84 <li><a href="#i_fdiv">'<tt>fdiv</tt>' Instruction</a></li>
85 <li><a href="#i_urem">'<tt>urem</tt>' Instruction</a></li>
86 <li><a href="#i_srem">'<tt>srem</tt>' Instruction</a></li>
87 <li><a href="#i_frem">'<tt>frem</tt>' Instruction</a></li>
90 <li><a href="#bitwiseops">Bitwise Binary Operations</a>
92 <li><a href="#i_and">'<tt>and</tt>' Instruction</a></li>
93 <li><a href="#i_or">'<tt>or</tt>' Instruction</a></li>
94 <li><a href="#i_xor">'<tt>xor</tt>' Instruction</a></li>
95 <li><a href="#i_shl">'<tt>shl</tt>' Instruction</a></li>
96 <li><a href="#i_lshr">'<tt>lshr</tt>' Instruction</a></li>
97 <li><a href="#i_ashr">'<tt>ashr</tt>' Instruction</a></li>
100 <li><a href="#vectorops">Vector Operations</a>
102 <li><a href="#i_extractelement">'<tt>extractelement</tt>' Instruction</a></li>
103 <li><a href="#i_insertelement">'<tt>insertelement</tt>' Instruction</a></li>
104 <li><a href="#i_shufflevector">'<tt>shufflevector</tt>' Instruction</a></li>
107 <li><a href="#memoryops">Memory Access and Addressing Operations</a>
109 <li><a href="#i_malloc">'<tt>malloc</tt>' Instruction</a></li>
110 <li><a href="#i_free">'<tt>free</tt>' Instruction</a></li>
111 <li><a href="#i_alloca">'<tt>alloca</tt>' Instruction</a></li>
112 <li><a href="#i_load">'<tt>load</tt>' Instruction</a></li>
113 <li><a href="#i_store">'<tt>store</tt>' Instruction</a></li>
114 <li><a href="#i_getelementptr">'<tt>getelementptr</tt>' Instruction</a></li>
117 <li><a href="#convertops">Conversion Operations</a>
119 <li><a href="#i_trunc">'<tt>trunc .. to</tt>' Instruction</a></li>
120 <li><a href="#i_zext">'<tt>zext .. to</tt>' Instruction</a></li>
121 <li><a href="#i_sext">'<tt>sext .. to</tt>' Instruction</a></li>
122 <li><a href="#i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a></li>
123 <li><a href="#i_fpext">'<tt>fpext .. to</tt>' Instruction</a></li>
124 <li><a href="#i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a></li>
125 <li><a href="#i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a></li>
126 <li><a href="#i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a></li>
127 <li><a href="#i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a></li>
128 <li><a href="#i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a></li>
129 <li><a href="#i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a></li>
130 <li><a href="#i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a></li>
132 <li><a href="#otherops">Other Operations</a>
134 <li><a href="#i_icmp">'<tt>icmp</tt>' Instruction</a></li>
135 <li><a href="#i_fcmp">'<tt>fcmp</tt>' Instruction</a></li>
136 <li><a href="#i_phi">'<tt>phi</tt>' Instruction</a></li>
137 <li><a href="#i_select">'<tt>select</tt>' Instruction</a></li>
138 <li><a href="#i_call">'<tt>call</tt>' Instruction</a></li>
139 <li><a href="#i_va_arg">'<tt>va_arg</tt>' Instruction</a></li>
144 <li><a href="#intrinsics">Intrinsic Functions</a>
146 <li><a href="#int_varargs">Variable Argument Handling Intrinsics</a>
148 <li><a href="#i_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a></li>
149 <li><a href="#i_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a></li>
150 <li><a href="#i_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a></li>
153 <li><a href="#int_gc">Accurate Garbage Collection Intrinsics</a>
155 <li><a href="#i_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a></li>
156 <li><a href="#i_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a></li>
157 <li><a href="#i_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a></li>
160 <li><a href="#int_codegen">Code Generator Intrinsics</a>
162 <li><a href="#i_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a></li>
163 <li><a href="#i_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a></li>
164 <li><a href="#i_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a></li>
165 <li><a href="#i_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a></li>
166 <li><a href="#i_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a></li>
167 <li><a href="#i_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a></li>
168 <li><a href="#i_readcyclecounter"><tt>llvm.readcyclecounter</tt>' Intrinsic</a></li>
171 <li><a href="#int_libc">Standard C Library Intrinsics</a>
173 <li><a href="#i_memcpy">'<tt>llvm.memcpy.*</tt>' Intrinsic</a></li>
174 <li><a href="#i_memmove">'<tt>llvm.memmove.*</tt>' Intrinsic</a></li>
175 <li><a href="#i_memset">'<tt>llvm.memset.*</tt>' Intrinsic</a></li>
176 <li><a href="#i_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a></li>
177 <li><a href="#i_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a></li>
180 <li><a href="#int_manip">Bit Manipulation Intrinsics</a>
182 <li><a href="#i_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a></li>
183 <li><a href="#int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic </a></li>
184 <li><a href="#int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic </a></li>
185 <li><a href="#int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic </a></li>
188 <li><a href="#int_debugger">Debugger intrinsics</a></li>
193 <div class="doc_author">
194 <p>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a>
195 and <a href="mailto:vadve@cs.uiuc.edu">Vikram Adve</a></p>
198 <!-- *********************************************************************** -->
199 <div class="doc_section"> <a name="abstract">Abstract </a></div>
200 <!-- *********************************************************************** -->
202 <div class="doc_text">
203 <p>This document is a reference manual for the LLVM assembly language.
204 LLVM is an SSA based representation that provides type safety,
205 low-level operations, flexibility, and the capability of representing
206 'all' high-level languages cleanly. It is the common code
207 representation used throughout all phases of the LLVM compilation
211 <!-- *********************************************************************** -->
212 <div class="doc_section"> <a name="introduction">Introduction</a> </div>
213 <!-- *********************************************************************** -->
215 <div class="doc_text">
217 <p>The LLVM code representation is designed to be used in three
218 different forms: as an in-memory compiler IR, as an on-disk bytecode
219 representation (suitable for fast loading by a Just-In-Time compiler),
220 and as a human readable assembly language representation. This allows
221 LLVM to provide a powerful intermediate representation for efficient
222 compiler transformations and analysis, while providing a natural means
223 to debug and visualize the transformations. The three different forms
224 of LLVM are all equivalent. This document describes the human readable
225 representation and notation.</p>
227 <p>The LLVM representation aims to be light-weight and low-level
228 while being expressive, typed, and extensible at the same time. It
229 aims to be a "universal IR" of sorts, by being at a low enough level
230 that high-level ideas may be cleanly mapped to it (similar to how
231 microprocessors are "universal IR's", allowing many source languages to
232 be mapped to them). By providing type information, LLVM can be used as
233 the target of optimizations: for example, through pointer analysis, it
234 can be proven that a C automatic variable is never accessed outside of
235 the current function... allowing it to be promoted to a simple SSA
236 value instead of a memory location.</p>
240 <!-- _______________________________________________________________________ -->
241 <div class="doc_subsubsection"> <a name="wellformed">Well-Formedness</a> </div>
243 <div class="doc_text">
245 <p>It is important to note that this document describes 'well formed'
246 LLVM assembly language. There is a difference between what the parser
247 accepts and what is considered 'well formed'. For example, the
248 following instruction is syntactically okay, but not well formed:</p>
251 %x = <a href="#i_add">add</a> i32 1, %x
254 <p>...because the definition of <tt>%x</tt> does not dominate all of
255 its uses. The LLVM infrastructure provides a verification pass that may
256 be used to verify that an LLVM module is well formed. This pass is
257 automatically run by the parser after parsing input assembly and by
258 the optimizer before it outputs bytecode. The violations pointed out
259 by the verifier pass indicate bugs in transformation passes or input to
262 <!-- Describe the typesetting conventions here. --> </div>
264 <!-- *********************************************************************** -->
265 <div class="doc_section"> <a name="identifiers">Identifiers</a> </div>
266 <!-- *********************************************************************** -->
268 <div class="doc_text">
270 <p>LLVM uses three different forms of identifiers, for different
274 <li>Named values are represented as a string of characters with a '%' prefix.
275 For example, %foo, %DivisionByZero, %a.really.long.identifier. The actual
276 regular expression used is '<tt>%[a-zA-Z$._][a-zA-Z$._0-9]*</tt>'.
277 Identifiers which require other characters in their names can be surrounded
278 with quotes. In this way, anything except a <tt>"</tt> character can be used
281 <li>Unnamed values are represented as an unsigned numeric value with a '%'
282 prefix. For example, %12, %2, %44.</li>
284 <li>Constants, which are described in a <a href="#constants">section about
285 constants</a>, below.</li>
288 <p>LLVM requires that values start with a '%' sign for two reasons: Compilers
289 don't need to worry about name clashes with reserved words, and the set of
290 reserved words may be expanded in the future without penalty. Additionally,
291 unnamed identifiers allow a compiler to quickly come up with a temporary
292 variable without having to avoid symbol table conflicts.</p>
294 <p>Reserved words in LLVM are very similar to reserved words in other
295 languages. There are keywords for different opcodes
296 ('<tt><a href="#i_add">add</a></tt>',
297 '<tt><a href="#i_bitcast">bitcast</a></tt>',
298 '<tt><a href="#i_ret">ret</a></tt>', etc...), for primitive type names ('<tt><a
299 href="#t_void">void</a></tt>', '<tt><a href="#t_primitive">i32</a></tt>', etc...),
300 and others. These reserved words cannot conflict with variable names, because
301 none of them start with a '%' character.</p>
303 <p>Here is an example of LLVM code to multiply the integer variable
304 '<tt>%X</tt>' by 8:</p>
309 %result = <a href="#i_mul">mul</a> i32 %X, 8
312 <p>After strength reduction:</p>
315 %result = <a href="#i_shl">shl</a> i32 %X, i8 3
318 <p>And the hard way:</p>
321 <a href="#i_add">add</a> i32 %X, %X <i>; yields {i32}:%0</i>
322 <a href="#i_add">add</a> i32 %0, %0 <i>; yields {i32}:%1</i>
323 %result = <a href="#i_add">add</a> i32 %1, %1
326 <p>This last way of multiplying <tt>%X</tt> by 8 illustrates several
327 important lexical features of LLVM:</p>
331 <li>Comments are delimited with a '<tt>;</tt>' and go until the end of
334 <li>Unnamed temporaries are created when the result of a computation is not
335 assigned to a named value.</li>
337 <li>Unnamed temporaries are numbered sequentially</li>
341 <p>...and it also shows a convention that we follow in this document. When
342 demonstrating instructions, we will follow an instruction with a comment that
343 defines the type and name of value produced. Comments are shown in italic
348 <!-- *********************************************************************** -->
349 <div class="doc_section"> <a name="highlevel">High Level Structure</a> </div>
350 <!-- *********************************************************************** -->
352 <!-- ======================================================================= -->
353 <div class="doc_subsection"> <a name="modulestructure">Module Structure</a>
356 <div class="doc_text">
358 <p>LLVM programs are composed of "Module"s, each of which is a
359 translation unit of the input programs. Each module consists of
360 functions, global variables, and symbol table entries. Modules may be
361 combined together with the LLVM linker, which merges function (and
362 global variable) definitions, resolves forward declarations, and merges
363 symbol table entries. Here is an example of the "hello world" module:</p>
365 <pre><i>; Declare the string constant as a global constant...</i>
366 <a href="#identifiers">%.LC0</a> = <a href="#linkage_internal">internal</a> <a
367 href="#globalvars">constant</a> <a href="#t_array">[13 x i8 ]</a> c"hello world\0A\00" <i>; [13 x i8 ]*</i>
369 <i>; External declaration of the puts function</i>
370 <a href="#functionstructure">declare</a> i32 %puts(i8 *) <i>; i32(i8 *)* </i>
372 <i>; Global variable / Function body section separator</i>
375 <i>; Definition of main function</i>
376 define i32 %main() { <i>; i32()* </i>
377 <i>; Convert [13x i8 ]* to i8 *...</i>
379 href="#i_getelementptr">getelementptr</a> [13 x i8 ]* %.LC0, i64 0, i64 0 <i>; i8 *</i>
381 <i>; Call puts function to write out the string to stdout...</i>
383 href="#i_call">call</a> i32 %puts(i8 * %cast210) <i>; i32</i>
385 href="#i_ret">ret</a> i32 0<br>}<br></pre>
387 <p>This example is made up of a <a href="#globalvars">global variable</a>
388 named "<tt>.LC0</tt>", an external declaration of the "<tt>puts</tt>"
389 function, and a <a href="#functionstructure">function definition</a>
390 for "<tt>main</tt>".</p>
392 <p>In general, a module is made up of a list of global values,
393 where both functions and global variables are global values. Global values are
394 represented by a pointer to a memory location (in this case, a pointer to an
395 array of char, and a pointer to a function), and have one of the following <a
396 href="#linkage">linkage types</a>.</p>
398 <p>Due to a limitation in the current LLVM assembly parser (it is limited by
399 one-token lookahead), modules are split into two pieces by the "implementation"
400 keyword. Global variable prototypes and definitions must occur before the
401 keyword, and function definitions must occur after it. Function prototypes may
402 occur either before or after it. In the future, the implementation keyword may
403 become a noop, if the parser gets smarter.</p>
407 <!-- ======================================================================= -->
408 <div class="doc_subsection">
409 <a name="linkage">Linkage Types</a>
412 <div class="doc_text">
415 All Global Variables and Functions have one of the following types of linkage:
420 <dt><tt><b><a name="linkage_internal">internal</a></b></tt> </dt>
422 <dd>Global values with internal linkage are only directly accessible by
423 objects in the current module. In particular, linking code into a module with
424 an internal global value may cause the internal to be renamed as necessary to
425 avoid collisions. Because the symbol is internal to the module, all
426 references can be updated. This corresponds to the notion of the
427 '<tt>static</tt>' keyword in C.
430 <dt><tt><b><a name="linkage_linkonce">linkonce</a></b></tt>: </dt>
432 <dd>Globals with "<tt>linkonce</tt>" linkage are merged with other globals of
433 the same name when linkage occurs. This is typically used to implement
434 inline functions, templates, or other code which must be generated in each
435 translation unit that uses it. Unreferenced <tt>linkonce</tt> globals are
436 allowed to be discarded.
439 <dt><tt><b><a name="linkage_weak">weak</a></b></tt>: </dt>
441 <dd>"<tt>weak</tt>" linkage is exactly the same as <tt>linkonce</tt> linkage,
442 except that unreferenced <tt>weak</tt> globals may not be discarded. This is
443 used for globals that may be emitted in multiple translation units, but that
444 are not guaranteed to be emitted into every translation unit that uses them.
445 One example of this are common globals in C, such as "<tt>int X;</tt>" at
449 <dt><tt><b><a name="linkage_appending">appending</a></b></tt>: </dt>
451 <dd>"<tt>appending</tt>" linkage may only be applied to global variables of
452 pointer to array type. When two global variables with appending linkage are
453 linked together, the two global arrays are appended together. This is the
454 LLVM, typesafe, equivalent of having the system linker append together
455 "sections" with identical names when .o files are linked.
458 <dt><tt><b><a name="linkage_externweak">extern_weak</a></b></tt>: </dt>
459 <dd>The semantics of this linkage follow the ELF model: the symbol is weak
460 until linked, if not linked, the symbol becomes null instead of being an
465 <dt><tt><b><a name="linkage_external">externally visible</a></b></tt>:</dt>
467 <dd>If none of the above identifiers are used, the global is externally
468 visible, meaning that it participates in linkage and can be used to resolve
469 external symbol references.
473 The next two types of linkage are targeted for Microsoft Windows platform
474 only. They are designed to support importing (exporting) symbols from (to)
479 <dt><tt><b><a name="linkage_dllimport">dllimport</a></b></tt>: </dt>
481 <dd>"<tt>dllimport</tt>" linkage causes the compiler to reference a function
482 or variable via a global pointer to a pointer that is set up by the DLL
483 exporting the symbol. On Microsoft Windows targets, the pointer name is
484 formed by combining <code>_imp__</code> and the function or variable name.
487 <dt><tt><b><a name="linkage_dllexport">dllexport</a></b></tt>: </dt>
489 <dd>"<tt>dllexport</tt>" linkage causes the compiler to provide a global
490 pointer to a pointer in a DLL, so that it can be referenced with the
491 <tt>dllimport</tt> attribute. On Microsoft Windows targets, the pointer
492 name is formed by combining <code>_imp__</code> and the function or variable
498 <p><a name="linkage_external"></a>For example, since the "<tt>.LC0</tt>"
499 variable is defined to be internal, if another module defined a "<tt>.LC0</tt>"
500 variable and was linked with this one, one of the two would be renamed,
501 preventing a collision. Since "<tt>main</tt>" and "<tt>puts</tt>" are
502 external (i.e., lacking any linkage declarations), they are accessible
503 outside of the current module.</p>
504 <p>It is illegal for a function <i>declaration</i>
505 to have any linkage type other than "externally visible", <tt>dllimport</tt>,
506 or <tt>extern_weak</tt>.</p>
510 <!-- ======================================================================= -->
511 <div class="doc_subsection">
512 <a name="callingconv">Calling Conventions</a>
515 <div class="doc_text">
517 <p>LLVM <a href="#functionstructure">functions</a>, <a href="#i_call">calls</a>
518 and <a href="#i_invoke">invokes</a> can all have an optional calling convention
519 specified for the call. The calling convention of any pair of dynamic
520 caller/callee must match, or the behavior of the program is undefined. The
521 following calling conventions are supported by LLVM, and more may be added in
525 <dt><b>"<tt>ccc</tt>" - The C calling convention</b>:</dt>
527 <dd>This calling convention (the default if no other calling convention is
528 specified) matches the target C calling conventions. This calling convention
529 supports varargs function calls and tolerates some mismatch in the declared
530 prototype and implemented declaration of the function (as does normal C).
533 <dt><b>"<tt>csretcc</tt>" - The C struct return calling convention</b>:</dt>
535 <dd>This calling convention matches the target C calling conventions, except
536 that functions with this convention are required to take a pointer as their
537 first argument, and the return type of the function must be void. This is
538 used for C functions that return aggregates by-value. In this case, the
539 function has been transformed to take a pointer to the struct as the first
540 argument to the function. For targets where the ABI specifies specific
541 behavior for structure-return calls, the calling convention can be used to
542 distinguish between struct return functions and other functions that take a
543 pointer to a struct as the first argument.
546 <dt><b>"<tt>fastcc</tt>" - The fast calling convention</b>:</dt>
548 <dd>This calling convention attempts to make calls as fast as possible
549 (e.g. by passing things in registers). This calling convention allows the
550 target to use whatever tricks it wants to produce fast code for the target,
551 without having to conform to an externally specified ABI. Implementations of
552 this convention should allow arbitrary tail call optimization to be supported.
553 This calling convention does not support varargs and requires the prototype of
554 all callees to exactly match the prototype of the function definition.
557 <dt><b>"<tt>coldcc</tt>" - The cold calling convention</b>:</dt>
559 <dd>This calling convention attempts to make code in the caller as efficient
560 as possible under the assumption that the call is not commonly executed. As
561 such, these calls often preserve all registers so that the call does not break
562 any live ranges in the caller side. This calling convention does not support
563 varargs and requires the prototype of all callees to exactly match the
564 prototype of the function definition.
567 <dt><b>"<tt>cc <<em>n</em>></tt>" - Numbered convention</b>:</dt>
569 <dd>Any calling convention may be specified by number, allowing
570 target-specific calling conventions to be used. Target specific calling
571 conventions start at 64.
575 <p>More calling conventions can be added/defined on an as-needed basis, to
576 support pascal conventions or any other well-known target-independent
581 <!-- ======================================================================= -->
582 <div class="doc_subsection">
583 <a name="visibility">Visibility Styles</a>
586 <div class="doc_text">
589 All Global Variables and Functions have one of the following visibility styles:
593 <dt><b>"<tt>default</tt>" - Default style</b>:</dt>
595 <dd>On ELF, default visibility means that the declaration is visible to other
596 modules and, in shared libraries, means that the declared entity may be
597 overridden. On Darwin, default visibility means that the declaration is
598 visible to other modules. Default visibility corresponds to "external
599 linkage" in the language.
602 <dt><b>"<tt>hidden</tt>" - Hidden style</b>:</dt>
604 <dd>Two declarations of an object with hidden visibility refer to the same
605 object if they are in the same shared object. Usually, hidden visibility
606 indicates that the symbol will not be placed into the dynamic symbol table,
607 so no other module (executable or shared library) can reference it
615 <!-- ======================================================================= -->
616 <div class="doc_subsection">
617 <a name="globalvars">Global Variables</a>
620 <div class="doc_text">
622 <p>Global variables define regions of memory allocated at compilation time
623 instead of run-time. Global variables may optionally be initialized, may have
624 an explicit section to be placed in, and may
625 have an optional explicit alignment specified. A
626 variable may be defined as a global "constant," which indicates that the
627 contents of the variable will <b>never</b> be modified (enabling better
628 optimization, allowing the global data to be placed in the read-only section of
629 an executable, etc). Note that variables that need runtime initialization
630 cannot be marked "constant" as there is a store to the variable.</p>
633 LLVM explicitly allows <em>declarations</em> of global variables to be marked
634 constant, even if the final definition of the global is not. This capability
635 can be used to enable slightly better optimization of the program, but requires
636 the language definition to guarantee that optimizations based on the
637 'constantness' are valid for the translation units that do not include the
641 <p>As SSA values, global variables define pointer values that are in
642 scope (i.e. they dominate) all basic blocks in the program. Global
643 variables always define a pointer to their "content" type because they
644 describe a region of memory, and all memory objects in LLVM are
645 accessed through pointers.</p>
647 <p>LLVM allows an explicit section to be specified for globals. If the target
648 supports it, it will emit globals to the section specified.</p>
650 <p>An explicit alignment may be specified for a global. If not present, or if
651 the alignment is set to zero, the alignment of the global is set by the target
652 to whatever it feels convenient. If an explicit alignment is specified, the
653 global is forced to have at least that much alignment. All alignments must be
656 <p>For example, the following defines a global with an initializer, section,
660 %G = constant float 1.0, section "foo", align 4
666 <!-- ======================================================================= -->
667 <div class="doc_subsection">
668 <a name="functionstructure">Functions</a>
671 <div class="doc_text">
673 <p>LLVM function definitions consist of the "<tt>define</tt>" keyord,
674 an optional <a href="#linkage">linkage type</a>, an optional
675 <a href="#visibility">visibility style</a>, an optional
676 <a href="#callingconv">calling convention</a>, a return type, an optional
677 <a href="#paramattrs">parameter attribute</a> for the return type, a function
678 name, a (possibly empty) argument list (each with optional
679 <a href="#paramattrs">parameter attributes</a>), an optional section, an
680 optional alignment, an opening curly brace, a list of basic blocks, and a
683 LLVM function declarations consist of the "<tt>declare</tt>" keyword, an
684 optional <a href="#linkage">linkage type</a>, an optional
685 <a href="#visibility">visibility style</a>, an optional
686 <a href="#callingconv">calling convention</a>, a return type, an optional
687 <a href="#paramattrs">parameter attribute</a> for the return type, a function
688 name, a possibly empty list of arguments, and an optional alignment.</p>
690 <p>A function definition contains a list of basic blocks, forming the CFG for
691 the function. Each basic block may optionally start with a label (giving the
692 basic block a symbol table entry), contains a list of instructions, and ends
693 with a <a href="#terminators">terminator</a> instruction (such as a branch or
694 function return).</p>
696 <p>The first basic block in a program is special in two ways: it is immediately
697 executed on entrance to the function, and it is not allowed to have predecessor
698 basic blocks (i.e. there can not be any branches to the entry block of a
699 function). Because the block can have no predecessors, it also cannot have any
700 <a href="#i_phi">PHI nodes</a>.</p>
702 <p>LLVM functions are identified by their name and type signature. Hence, two
703 functions with the same name but different parameter lists or return values are
704 considered different functions, and LLVM will resolve references to each
707 <p>LLVM allows an explicit section to be specified for functions. If the target
708 supports it, it will emit functions to the section specified.</p>
710 <p>An explicit alignment may be specified for a function. If not present, or if
711 the alignment is set to zero, the alignment of the function is set by the target
712 to whatever it feels convenient. If an explicit alignment is specified, the
713 function is forced to have at least that much alignment. All alignments must be
718 <!-- ======================================================================= -->
719 <div class="doc_subsection"><a name="paramattrs">Parameter Attributes</a></div>
720 <div class="doc_text">
721 <p>The return type and each parameter of a function type may have a set of
722 <i>parameter attributes</i> associated with them. Parameter attributes are
723 used to communicate additional information about the result or parameters of
724 a function. Parameter attributes are considered to be part of the function
725 type so two functions types that differ only by the parameter attributes
726 are different function types.</p>
728 <p>Parameter attributes are simple keywords that follow the type specified. If
729 multiple parameter attributes are needed, they are space separated. For
731 %someFunc = i16 (i8 sext %someParam) zext
732 %someFunc = i16 (i8 zext %someParam) zext</pre>
733 <p>Note that the two function types above are unique because the parameter has
734 a different attribute (sext in the first one, zext in the second). Also note
735 that the attribute for the function result (zext) comes immediately after the
738 <p>Currently, only the following parameter attributes are defined:</p>
740 <dt><tt>zext</tt></dt>
741 <dd>This indicates that the parameter should be zero extended just before
742 a call to this function.</dd>
743 <dt><tt>sext</tt></dt>
744 <dd>This indicates that the parameter should be sign extended just before
745 a call to this function.</dd>
748 <p>The current motivation for parameter attributes is to enable the sign and
749 zero extend information necessary for the C calling convention to be passed
750 from the front end to LLVM. The <tt>zext</tt> and <tt>sext</tt> attributes
751 are used by the code generator to perform the required extension. However,
752 parameter attributes are an orthogonal feature to calling conventions and
753 may be used for other purposes in the future.</p>
756 <!-- ======================================================================= -->
757 <div class="doc_subsection">
758 <a name="moduleasm">Module-Level Inline Assembly</a>
761 <div class="doc_text">
763 Modules may contain "module-level inline asm" blocks, which corresponds to the
764 GCC "file scope inline asm" blocks. These blocks are internally concatenated by
765 LLVM and treated as a single unit, but may be separated in the .ll file if
766 desired. The syntax is very simple:
769 <div class="doc_code"><pre>
770 module asm "inline asm code goes here"
771 module asm "more can go here"
774 <p>The strings can contain any character by escaping non-printable characters.
775 The escape sequence used is simply "\xx" where "xx" is the two digit hex code
780 The inline asm code is simply printed to the machine code .s file when
781 assembly code is generated.
786 <!-- *********************************************************************** -->
787 <div class="doc_section"> <a name="typesystem">Type System</a> </div>
788 <!-- *********************************************************************** -->
790 <div class="doc_text">
792 <p>The LLVM type system is one of the most important features of the
793 intermediate representation. Being typed enables a number of
794 optimizations to be performed on the IR directly, without having to do
795 extra analyses on the side before the transformation. A strong type
796 system makes it easier to read the generated code and enables novel
797 analyses and transformations that are not feasible to perform on normal
798 three address code representations.</p>
802 <!-- ======================================================================= -->
803 <div class="doc_subsection"> <a name="t_primitive">Primitive Types</a> </div>
804 <div class="doc_text">
805 <p>The primitive types are the fundamental building blocks of the LLVM
806 system. The current set of primitive types is as follows:</p>
808 <table class="layout">
813 <tr><th>Type</th><th>Description</th></tr>
814 <tr><td><tt>void</tt></td><td>No value</td></tr>
815 <tr><td><tt>i8</tt></td><td>8-bit value</td></tr>
816 <tr><td><tt>i32</tt></td><td>32-bit value</td></tr>
817 <tr><td><tt>float</tt></td><td>32-bit floating point value</td></tr>
818 <tr><td><tt>label</tt></td><td>Branch destination</td></tr>
825 <tr><th>Type</th><th>Description</th></tr>
826 <tr><td><tt>i1</tt></td><td>True or False value</td></tr>
827 <tr><td><tt>i16</tt></td><td>16-bit value</td></tr>
828 <tr><td><tt>i64</tt></td><td>64-bit value</td></tr>
829 <tr><td><tt>double</tt></td><td>64-bit floating point value</td></tr>
837 <!-- _______________________________________________________________________ -->
838 <div class="doc_subsubsection"> <a name="t_classifications">Type
839 Classifications</a> </div>
840 <div class="doc_text">
841 <p>These different primitive types fall into a few useful
844 <table border="1" cellspacing="0" cellpadding="4">
846 <tr><th>Classification</th><th>Types</th></tr>
848 <td><a name="t_integer">integer</a></td>
849 <td><tt>i1, i8, i16, i32, i64</tt></td>
852 <td><a name="t_floating">floating point</a></td>
853 <td><tt>float, double</tt></td>
856 <td><a name="t_firstclass">first class</a></td>
857 <td><tt>i1, i8, i16, i32, i64, float, double, <br/>
858 <a href="#t_pointer">pointer</a>,<a href="#t_packed">packed</a></tt>
864 <p>The <a href="#t_firstclass">first class</a> types are perhaps the
865 most important. Values of these types are the only ones which can be
866 produced by instructions, passed as arguments, or used as operands to
867 instructions. This means that all structures and arrays must be
868 manipulated either by pointer or by component.</p>
871 <!-- ======================================================================= -->
872 <div class="doc_subsection"> <a name="t_derived">Derived Types</a> </div>
874 <div class="doc_text">
876 <p>The real power in LLVM comes from the derived types in the system.
877 This is what allows a programmer to represent arrays, functions,
878 pointers, and other useful types. Note that these derived types may be
879 recursive: For example, it is possible to have a two dimensional array.</p>
883 <!-- _______________________________________________________________________ -->
884 <div class="doc_subsubsection"> <a name="t_array">Array Type</a> </div>
886 <div class="doc_text">
890 <p>The array type is a very simple derived type that arranges elements
891 sequentially in memory. The array type requires a size (number of
892 elements) and an underlying data type.</p>
897 [<# elements> x <elementtype>]
900 <p>The number of elements is a constant integer value; elementtype may
901 be any type with a size.</p>
904 <table class="layout">
907 <tt>[40 x i32 ]</tt><br/>
908 <tt>[41 x i32 ]</tt><br/>
909 <tt>[40 x i8]</tt><br/>
912 Array of 40 32-bit integer values.<br/>
913 Array of 41 32-bit integer values.<br/>
914 Array of 40 8-bit integer values.<br/>
918 <p>Here are some examples of multidimensional arrays:</p>
919 <table class="layout">
922 <tt>[3 x [4 x i32]]</tt><br/>
923 <tt>[12 x [10 x float]]</tt><br/>
924 <tt>[2 x [3 x [4 x i16]]]</tt><br/>
927 3x4 array of 32-bit integer values.<br/>
928 12x10 array of single precision floating point values.<br/>
929 2x3x4 array of 16-bit integer values.<br/>
934 <p>Note that 'variable sized arrays' can be implemented in LLVM with a zero
935 length array. Normally, accesses past the end of an array are undefined in
936 LLVM (e.g. it is illegal to access the 5th element of a 3 element array).
937 As a special case, however, zero length arrays are recognized to be variable
938 length. This allows implementation of 'pascal style arrays' with the LLVM
939 type "{ i32, [0 x float]}", for example.</p>
943 <!-- _______________________________________________________________________ -->
944 <div class="doc_subsubsection"> <a name="t_function">Function Type</a> </div>
945 <div class="doc_text">
947 <p>The function type can be thought of as a function signature. It
948 consists of a return type and a list of formal parameter types.
949 Function types are usually used to build virtual function tables
950 (which are structures of pointers to functions), for indirect function
951 calls, and when defining a function.</p>
953 The return type of a function type cannot be an aggregate type.
956 <pre> <returntype> (<parameter list>)<br></pre>
957 <p>...where '<tt><parameter list></tt>' is a comma-separated list of type
958 specifiers. Optionally, the parameter list may include a type <tt>...</tt>,
959 which indicates that the function takes a variable number of arguments.
960 Variable argument functions can access their arguments with the <a
961 href="#int_varargs">variable argument handling intrinsic</a> functions.</p>
963 <table class="layout">
965 <td class="left"><tt>i32 (i32)</tt></td>
966 <td class="left">function taking an <tt>i32</tt>, returning an <tt>i32</tt>
968 </tr><tr class="layout">
969 <td class="left"><tt>float (i16 sext, i32 *) *
971 <td class="left"><a href="#t_pointer">Pointer</a> to a function that takes
972 an <tt>i16</tt> that should be sign extended and a
973 <a href="#t_pointer">pointer</a> to <tt>i32</tt>, returning
976 </tr><tr class="layout">
977 <td class="left"><tt>i32 (i8*, ...)</tt></td>
978 <td class="left">A vararg function that takes at least one
979 <a href="#t_pointer">pointer</a> to <tt>i8 </tt> (char in C),
980 which returns an integer. This is the signature for <tt>printf</tt> in
987 <!-- _______________________________________________________________________ -->
988 <div class="doc_subsubsection"> <a name="t_struct">Structure Type</a> </div>
989 <div class="doc_text">
991 <p>The structure type is used to represent a collection of data members
992 together in memory. The packing of the field types is defined to match
993 the ABI of the underlying processor. The elements of a structure may
994 be any type that has a size.</p>
995 <p>Structures are accessed using '<tt><a href="#i_load">load</a></tt>
996 and '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a
997 field with the '<tt><a href="#i_getelementptr">getelementptr</a></tt>'
1000 <pre> { <type list> }<br></pre>
1002 <table class="layout">
1005 <tt>{ i32, i32, i32 }</tt><br/>
1006 <tt>{ float, i32 (i32) * }</tt><br/>
1009 a triple of three <tt>i32</tt> values<br/>
1010 A pair, where the first element is a <tt>float</tt> and the second element
1011 is a <a href="#t_pointer">pointer</a> to a <a href="#t_function">function</a>
1012 that takes an <tt>i32</tt>, returning an <tt>i32</tt>.<br/>
1018 <!-- _______________________________________________________________________ -->
1019 <div class="doc_subsubsection"> <a name="t_pstruct">Packed Structure Type</a>
1021 <div class="doc_text">
1023 <p>The packed structure type is used to represent a collection of data members
1024 together in memory. There is no padding between fields. Further, the alignment
1025 of a packed structure is 1 byte. The elements of a packed structure may
1026 be any type that has a size.</p>
1027 <p>Structures are accessed using '<tt><a href="#i_load">load</a></tt>
1028 and '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a
1029 field with the '<tt><a href="#i_getelementptr">getelementptr</a></tt>'
1032 <pre> < { <type list> } > <br></pre>
1034 <table class="layout">
1037 <tt> < { i32, i32, i32 } > </tt><br/>
1038 <tt> < { float, i32 (i32) * } > </tt><br/>
1041 a triple of three <tt>i32</tt> values<br/>
1042 A pair, where the first element is a <tt>float</tt> and the second element
1043 is a <a href="#t_pointer">pointer</a> to a <a href="#t_function">function</a>
1044 that takes an <tt>i32</tt>, returning an <tt>i32</tt>.<br/>
1050 <!-- _______________________________________________________________________ -->
1051 <div class="doc_subsubsection"> <a name="t_pointer">Pointer Type</a> </div>
1052 <div class="doc_text">
1054 <p>As in many languages, the pointer type represents a pointer or
1055 reference to another object, which must live in memory.</p>
1057 <pre> <type> *<br></pre>
1059 <table class="layout">
1062 <tt>[4x i32]*</tt><br/>
1063 <tt>i32 (i32 *) *</tt><br/>
1066 A <a href="#t_pointer">pointer</a> to <a href="#t_array">array</a> of
1067 four <tt>i32</tt> values<br/>
1068 A <a href="#t_pointer">pointer</a> to a <a
1069 href="#t_function">function</a> that takes an <tt>i32*</tt>, returning an
1076 <!-- _______________________________________________________________________ -->
1077 <div class="doc_subsubsection"> <a name="t_packed">Packed Type</a> </div>
1078 <div class="doc_text">
1082 <p>A packed type is a simple derived type that represents a vector
1083 of elements. Packed types are used when multiple primitive data
1084 are operated in parallel using a single instruction (SIMD).
1085 A packed type requires a size (number of
1086 elements) and an underlying primitive data type. Vectors must have a power
1087 of two length (1, 2, 4, 8, 16 ...). Packed types are
1088 considered <a href="#t_firstclass">first class</a>.</p>
1093 < <# elements> x <elementtype> >
1096 <p>The number of elements is a constant integer value; elementtype may
1097 be any integer or floating point type.</p>
1101 <table class="layout">
1104 <tt><4 x i32></tt><br/>
1105 <tt><8 x float></tt><br/>
1106 <tt><2 x i64></tt><br/>
1109 Packed vector of 4 32-bit integer values.<br/>
1110 Packed vector of 8 floating-point values.<br/>
1111 Packed vector of 2 64-bit integer values.<br/>
1117 <!-- _______________________________________________________________________ -->
1118 <div class="doc_subsubsection"> <a name="t_opaque">Opaque Type</a> </div>
1119 <div class="doc_text">
1123 <p>Opaque types are used to represent unknown types in the system. This
1124 corresponds (for example) to the C notion of a foward declared structure type.
1125 In LLVM, opaque types can eventually be resolved to any type (not just a
1126 structure type).</p>
1136 <table class="layout">
1142 An opaque type.<br/>
1149 <!-- *********************************************************************** -->
1150 <div class="doc_section"> <a name="constants">Constants</a> </div>
1151 <!-- *********************************************************************** -->
1153 <div class="doc_text">
1155 <p>LLVM has several different basic types of constants. This section describes
1156 them all and their syntax.</p>
1160 <!-- ======================================================================= -->
1161 <div class="doc_subsection"><a name="simpleconstants">Simple Constants</a></div>
1163 <div class="doc_text">
1166 <dt><b>Boolean constants</b></dt>
1168 <dd>The two strings '<tt>true</tt>' and '<tt>false</tt>' are both valid
1169 constants of the <tt><a href="#t_primitive">i1</a></tt> type.
1172 <dt><b>Integer constants</b></dt>
1174 <dd>Standard integers (such as '4') are constants of the <a
1175 href="#t_integer">integer</a> type. Negative numbers may be used with
1179 <dt><b>Floating point constants</b></dt>
1181 <dd>Floating point constants use standard decimal notation (e.g. 123.421),
1182 exponential notation (e.g. 1.23421e+2), or a more precise hexadecimal
1183 notation (see below). Floating point constants must have a <a
1184 href="#t_floating">floating point</a> type. </dd>
1186 <dt><b>Null pointer constants</b></dt>
1188 <dd>The identifier '<tt>null</tt>' is recognized as a null pointer constant
1189 and must be of <a href="#t_pointer">pointer type</a>.</dd>
1193 <p>The one non-intuitive notation for constants is the optional hexadecimal form
1194 of floating point constants. For example, the form '<tt>double
1195 0x432ff973cafa8000</tt>' is equivalent to (but harder to read than) '<tt>double
1196 4.5e+15</tt>'. The only time hexadecimal floating point constants are required
1197 (and the only time that they are generated by the disassembler) is when a
1198 floating point constant must be emitted but it cannot be represented as a
1199 decimal floating point number. For example, NaN's, infinities, and other
1200 special values are represented in their IEEE hexadecimal format so that
1201 assembly and disassembly do not cause any bits to change in the constants.</p>
1205 <!-- ======================================================================= -->
1206 <div class="doc_subsection"><a name="aggregateconstants">Aggregate Constants</a>
1209 <div class="doc_text">
1210 <p>Aggregate constants arise from aggregation of simple constants
1211 and smaller aggregate constants.</p>
1214 <dt><b>Structure constants</b></dt>
1216 <dd>Structure constants are represented with notation similar to structure
1217 type definitions (a comma separated list of elements, surrounded by braces
1218 (<tt>{}</tt>)). For example: "<tt>{ i32 4, float 17.0, i32* %G }</tt>",
1219 where "<tt>%G</tt>" is declared as "<tt>%G = external global i32</tt>". Structure constants
1220 must have <a href="#t_struct">structure type</a>, and the number and
1221 types of elements must match those specified by the type.
1224 <dt><b>Array constants</b></dt>
1226 <dd>Array constants are represented with notation similar to array type
1227 definitions (a comma separated list of elements, surrounded by square brackets
1228 (<tt>[]</tt>)). For example: "<tt>[ i32 42, i32 11, i32 74 ]</tt>". Array
1229 constants must have <a href="#t_array">array type</a>, and the number and
1230 types of elements must match those specified by the type.
1233 <dt><b>Packed constants</b></dt>
1235 <dd>Packed constants are represented with notation similar to packed type
1236 definitions (a comma separated list of elements, surrounded by
1237 less-than/greater-than's (<tt><></tt>)). For example: "<tt>< i32 42,
1238 i32 11, i32 74, i32 100 ></tt>". Packed constants must have <a
1239 href="#t_packed">packed type</a>, and the number and types of elements must
1240 match those specified by the type.
1243 <dt><b>Zero initialization</b></dt>
1245 <dd>The string '<tt>zeroinitializer</tt>' can be used to zero initialize a
1246 value to zero of <em>any</em> type, including scalar and aggregate types.
1247 This is often used to avoid having to print large zero initializers (e.g. for
1248 large arrays) and is always exactly equivalent to using explicit zero
1255 <!-- ======================================================================= -->
1256 <div class="doc_subsection">
1257 <a name="globalconstants">Global Variable and Function Addresses</a>
1260 <div class="doc_text">
1262 <p>The addresses of <a href="#globalvars">global variables</a> and <a
1263 href="#functionstructure">functions</a> are always implicitly valid (link-time)
1264 constants. These constants are explicitly referenced when the <a
1265 href="#identifiers">identifier for the global</a> is used and always have <a
1266 href="#t_pointer">pointer</a> type. For example, the following is a legal LLVM
1272 %Z = global [2 x i32*] [ i32* %X, i32* %Y ]
1277 <!-- ======================================================================= -->
1278 <div class="doc_subsection"><a name="undefvalues">Undefined Values</a></div>
1279 <div class="doc_text">
1280 <p>The string '<tt>undef</tt>' is recognized as a type-less constant that has
1281 no specific value. Undefined values may be of any type and be used anywhere
1282 a constant is permitted.</p>
1284 <p>Undefined values indicate to the compiler that the program is well defined
1285 no matter what value is used, giving the compiler more freedom to optimize.
1289 <!-- ======================================================================= -->
1290 <div class="doc_subsection"><a name="constantexprs">Constant Expressions</a>
1293 <div class="doc_text">
1295 <p>Constant expressions are used to allow expressions involving other constants
1296 to be used as constants. Constant expressions may be of any <a
1297 href="#t_firstclass">first class</a> type and may involve any LLVM operation
1298 that does not have side effects (e.g. load and call are not supported). The
1299 following is the syntax for constant expressions:</p>
1302 <dt><b><tt>trunc ( CST to TYPE )</tt></b></dt>
1303 <dd>Truncate a constant to another type. The bit size of CST must be larger
1304 than the bit size of TYPE. Both types must be integers.</dd>
1306 <dt><b><tt>zext ( CST to TYPE )</tt></b></dt>
1307 <dd>Zero extend a constant to another type. The bit size of CST must be
1308 smaller or equal to the bit size of TYPE. Both types must be integers.</dd>
1310 <dt><b><tt>sext ( CST to TYPE )</tt></b></dt>
1311 <dd>Sign extend a constant to another type. The bit size of CST must be
1312 smaller or equal to the bit size of TYPE. Both types must be integers.</dd>
1314 <dt><b><tt>fptrunc ( CST to TYPE )</tt></b></dt>
1315 <dd>Truncate a floating point constant to another floating point type. The
1316 size of CST must be larger than the size of TYPE. Both types must be
1317 floating point.</dd>
1319 <dt><b><tt>fpext ( CST to TYPE )</tt></b></dt>
1320 <dd>Floating point extend a constant to another type. The size of CST must be
1321 smaller or equal to the size of TYPE. Both types must be floating point.</dd>
1323 <dt><b><tt>fp2uint ( CST to TYPE )</tt></b></dt>
1324 <dd>Convert a floating point constant to the corresponding unsigned integer
1325 constant. TYPE must be an integer type. CST must be floating point. If the
1326 value won't fit in the integer type, the results are undefined.</dd>
1328 <dt><b><tt>fptosi ( CST to TYPE )</tt></b></dt>
1329 <dd>Convert a floating point constant to the corresponding signed integer
1330 constant. TYPE must be an integer type. CST must be floating point. If the
1331 value won't fit in the integer type, the results are undefined.</dd>
1333 <dt><b><tt>uitofp ( CST to TYPE )</tt></b></dt>
1334 <dd>Convert an unsigned integer constant to the corresponding floating point
1335 constant. TYPE must be floating point. CST must be of integer type. If the
1336 value won't fit in the floating point type, the results are undefined.</dd>
1338 <dt><b><tt>sitofp ( CST to TYPE )</tt></b></dt>
1339 <dd>Convert a signed integer constant to the corresponding floating point
1340 constant. TYPE must be floating point. CST must be of integer type. If the
1341 value won't fit in the floating point type, the results are undefined.</dd>
1343 <dt><b><tt>ptrtoint ( CST to TYPE )</tt></b></dt>
1344 <dd>Convert a pointer typed constant to the corresponding integer constant
1345 TYPE must be an integer type. CST must be of pointer type. The CST value is
1346 zero extended, truncated, or unchanged to make it fit in TYPE.</dd>
1348 <dt><b><tt>inttoptr ( CST to TYPE )</tt></b></dt>
1349 <dd>Convert a integer constant to a pointer constant. TYPE must be a
1350 pointer type. CST must be of integer type. The CST value is zero extended,
1351 truncated, or unchanged to make it fit in a pointer size. This one is
1352 <i>really</i> dangerous!</dd>
1354 <dt><b><tt>bitcast ( CST to TYPE )</tt></b></dt>
1355 <dd>Convert a constant, CST, to another TYPE. The size of CST and TYPE must be
1356 identical (same number of bits). The conversion is done as if the CST value
1357 was stored to memory and read back as TYPE. In other words, no bits change
1358 with this operator, just the type. This can be used for conversion of
1359 packed types to any other type, as long as they have the same bit width. For
1360 pointers it is only valid to cast to another pointer type.
1363 <dt><b><tt>getelementptr ( CSTPTR, IDX0, IDX1, ... )</tt></b></dt>
1365 <dd>Perform the <a href="#i_getelementptr">getelementptr operation</a> on
1366 constants. As with the <a href="#i_getelementptr">getelementptr</a>
1367 instruction, the index list may have zero or more indexes, which are required
1368 to make sense for the type of "CSTPTR".</dd>
1370 <dt><b><tt>select ( COND, VAL1, VAL2 )</tt></b></dt>
1372 <dd>Perform the <a href="#i_select">select operation</a> on
1375 <dt><b><tt>icmp COND ( VAL1, VAL2 )</tt></b></dt>
1376 <dd>Performs the <a href="#i_icmp">icmp operation</a> on constants.</dd>
1378 <dt><b><tt>fcmp COND ( VAL1, VAL2 )</tt></b></dt>
1379 <dd>Performs the <a href="#i_fcmp">fcmp operation</a> on constants.</dd>
1381 <dt><b><tt>extractelement ( VAL, IDX )</tt></b></dt>
1383 <dd>Perform the <a href="#i_extractelement">extractelement
1384 operation</a> on constants.
1386 <dt><b><tt>insertelement ( VAL, ELT, IDX )</tt></b></dt>
1388 <dd>Perform the <a href="#i_insertelement">insertelement
1389 operation</a> on constants.</dd>
1392 <dt><b><tt>shufflevector ( VEC1, VEC2, IDXMASK )</tt></b></dt>
1394 <dd>Perform the <a href="#i_shufflevector">shufflevector
1395 operation</a> on constants.</dd>
1397 <dt><b><tt>OPCODE ( LHS, RHS )</tt></b></dt>
1399 <dd>Perform the specified operation of the LHS and RHS constants. OPCODE may
1400 be any of the <a href="#binaryops">binary</a> or <a href="#bitwiseops">bitwise
1401 binary</a> operations. The constraints on operands are the same as those for
1402 the corresponding instruction (e.g. no bitwise operations on floating point
1403 values are allowed).</dd>
1407 <!-- *********************************************************************** -->
1408 <div class="doc_section"> <a name="othervalues">Other Values</a> </div>
1409 <!-- *********************************************************************** -->
1411 <!-- ======================================================================= -->
1412 <div class="doc_subsection">
1413 <a name="inlineasm">Inline Assembler Expressions</a>
1416 <div class="doc_text">
1419 LLVM supports inline assembler expressions (as opposed to <a href="#moduleasm">
1420 Module-Level Inline Assembly</a>) through the use of a special value. This
1421 value represents the inline assembler as a string (containing the instructions
1422 to emit), a list of operand constraints (stored as a string), and a flag that
1423 indicates whether or not the inline asm expression has side effects. An example
1424 inline assembler expression is:
1428 i32 (i32) asm "bswap $0", "=r,r"
1432 Inline assembler expressions may <b>only</b> be used as the callee operand of
1433 a <a href="#i_call"><tt>call</tt> instruction</a>. Thus, typically we have:
1437 %X = call i32 asm "<a href="#i_bswap">bswap</a> $0", "=r,r"(i32 %Y)
1441 Inline asms with side effects not visible in the constraint list must be marked
1442 as having side effects. This is done through the use of the
1443 '<tt>sideeffect</tt>' keyword, like so:
1447 call void asm sideeffect "eieio", ""()
1450 <p>TODO: The format of the asm and constraints string still need to be
1451 documented here. Constraints on what can be done (e.g. duplication, moving, etc
1452 need to be documented).
1457 <!-- *********************************************************************** -->
1458 <div class="doc_section"> <a name="instref">Instruction Reference</a> </div>
1459 <!-- *********************************************************************** -->
1461 <div class="doc_text">
1463 <p>The LLVM instruction set consists of several different
1464 classifications of instructions: <a href="#terminators">terminator
1465 instructions</a>, <a href="#binaryops">binary instructions</a>,
1466 <a href="#bitwiseops">bitwise binary instructions</a>, <a
1467 href="#memoryops">memory instructions</a>, and <a href="#otherops">other
1468 instructions</a>.</p>
1472 <!-- ======================================================================= -->
1473 <div class="doc_subsection"> <a name="terminators">Terminator
1474 Instructions</a> </div>
1476 <div class="doc_text">
1478 <p>As mentioned <a href="#functionstructure">previously</a>, every
1479 basic block in a program ends with a "Terminator" instruction, which
1480 indicates which block should be executed after the current block is
1481 finished. These terminator instructions typically yield a '<tt>void</tt>'
1482 value: they produce control flow, not values (the one exception being
1483 the '<a href="#i_invoke"><tt>invoke</tt></a>' instruction).</p>
1484 <p>There are six different terminator instructions: the '<a
1485 href="#i_ret"><tt>ret</tt></a>' instruction, the '<a href="#i_br"><tt>br</tt></a>'
1486 instruction, the '<a href="#i_switch"><tt>switch</tt></a>' instruction,
1487 the '<a href="#i_invoke"><tt>invoke</tt></a>' instruction, the '<a
1488 href="#i_unwind"><tt>unwind</tt></a>' instruction, and the '<a
1489 href="#i_unreachable"><tt>unreachable</tt></a>' instruction.</p>
1493 <!-- _______________________________________________________________________ -->
1494 <div class="doc_subsubsection"> <a name="i_ret">'<tt>ret</tt>'
1495 Instruction</a> </div>
1496 <div class="doc_text">
1498 <pre> ret <type> <value> <i>; Return a value from a non-void function</i>
1499 ret void <i>; Return from void function</i>
1502 <p>The '<tt>ret</tt>' instruction is used to return control flow (and a
1503 value) from a function back to the caller.</p>
1504 <p>There are two forms of the '<tt>ret</tt>' instruction: one that
1505 returns a value and then causes control flow, and one that just causes
1506 control flow to occur.</p>
1508 <p>The '<tt>ret</tt>' instruction may return any '<a
1509 href="#t_firstclass">first class</a>' type. Notice that a function is
1510 not <a href="#wellformed">well formed</a> if there exists a '<tt>ret</tt>'
1511 instruction inside of the function that returns a value that does not
1512 match the return type of the function.</p>
1514 <p>When the '<tt>ret</tt>' instruction is executed, control flow
1515 returns back to the calling function's context. If the caller is a "<a
1516 href="#i_call"><tt>call</tt></a>" instruction, execution continues at
1517 the instruction after the call. If the caller was an "<a
1518 href="#i_invoke"><tt>invoke</tt></a>" instruction, execution continues
1519 at the beginning of the "normal" destination block. If the instruction
1520 returns a value, that value shall set the call or invoke instruction's
1523 <pre> ret i32 5 <i>; Return an integer value of 5</i>
1524 ret void <i>; Return from a void function</i>
1527 <!-- _______________________________________________________________________ -->
1528 <div class="doc_subsubsection"> <a name="i_br">'<tt>br</tt>' Instruction</a> </div>
1529 <div class="doc_text">
1531 <pre> br i1 <cond>, label <iftrue>, label <iffalse><br> br label <dest> <i>; Unconditional branch</i>
1534 <p>The '<tt>br</tt>' instruction is used to cause control flow to
1535 transfer to a different basic block in the current function. There are
1536 two forms of this instruction, corresponding to a conditional branch
1537 and an unconditional branch.</p>
1539 <p>The conditional branch form of the '<tt>br</tt>' instruction takes a
1540 single '<tt>i1</tt>' value and two '<tt>label</tt>' values. The
1541 unconditional form of the '<tt>br</tt>' instruction takes a single '<tt>label</tt>'
1542 value as a target.</p>
1544 <p>Upon execution of a conditional '<tt>br</tt>' instruction, the '<tt>i1</tt>'
1545 argument is evaluated. If the value is <tt>true</tt>, control flows
1546 to the '<tt>iftrue</tt>' <tt>label</tt> argument. If "cond" is <tt>false</tt>,
1547 control flows to the '<tt>iffalse</tt>' <tt>label</tt> argument.</p>
1549 <pre>Test:<br> %cond = <a href="#i_icmp">icmp</a> eq, i32 %a, %b<br> br i1 %cond, label %IfEqual, label %IfUnequal<br>IfEqual:<br> <a
1550 href="#i_ret">ret</a> i32 1<br>IfUnequal:<br> <a href="#i_ret">ret</a> i32 0<br></pre>
1552 <!-- _______________________________________________________________________ -->
1553 <div class="doc_subsubsection">
1554 <a name="i_switch">'<tt>switch</tt>' Instruction</a>
1557 <div class="doc_text">
1561 switch <intty> <value>, label <defaultdest> [ <intty> <val>, label <dest> ... ]
1566 <p>The '<tt>switch</tt>' instruction is used to transfer control flow to one of
1567 several different places. It is a generalization of the '<tt>br</tt>'
1568 instruction, allowing a branch to occur to one of many possible
1574 <p>The '<tt>switch</tt>' instruction uses three parameters: an integer
1575 comparison value '<tt>value</tt>', a default '<tt>label</tt>' destination, and
1576 an array of pairs of comparison value constants and '<tt>label</tt>'s. The
1577 table is not allowed to contain duplicate constant entries.</p>
1581 <p>The <tt>switch</tt> instruction specifies a table of values and
1582 destinations. When the '<tt>switch</tt>' instruction is executed, this
1583 table is searched for the given value. If the value is found, control flow is
1584 transfered to the corresponding destination; otherwise, control flow is
1585 transfered to the default destination.</p>
1587 <h5>Implementation:</h5>
1589 <p>Depending on properties of the target machine and the particular
1590 <tt>switch</tt> instruction, this instruction may be code generated in different
1591 ways. For example, it could be generated as a series of chained conditional
1592 branches or with a lookup table.</p>
1597 <i>; Emulate a conditional br instruction</i>
1598 %Val = <a href="#i_zext">zext</a> i1 %value to i32
1599 switch i32 %Val, label %truedest [i32 0, label %falsedest ]
1601 <i>; Emulate an unconditional br instruction</i>
1602 switch i32 0, label %dest [ ]
1604 <i>; Implement a jump table:</i>
1605 switch i32 %val, label %otherwise [ i32 0, label %onzero
1607 i32 2, label %ontwo ]
1611 <!-- _______________________________________________________________________ -->
1612 <div class="doc_subsubsection">
1613 <a name="i_invoke">'<tt>invoke</tt>' Instruction</a>
1616 <div class="doc_text">
1621 <result> = invoke [<a href="#callingconv">cconv</a>] <ptr to function ty> %<function ptr val>(<function args>)
1622 to label <normal label> unwind label <exception label>
1627 <p>The '<tt>invoke</tt>' instruction causes control to transfer to a specified
1628 function, with the possibility of control flow transfer to either the
1629 '<tt>normal</tt>' label or the
1630 '<tt>exception</tt>' label. If the callee function returns with the
1631 "<tt><a href="#i_ret">ret</a></tt>" instruction, control flow will return to the
1632 "normal" label. If the callee (or any indirect callees) returns with the "<a
1633 href="#i_unwind"><tt>unwind</tt></a>" instruction, control is interrupted and
1634 continued at the dynamically nearest "exception" label.</p>
1638 <p>This instruction requires several arguments:</p>
1642 The optional "cconv" marker indicates which <a href="callingconv">calling
1643 convention</a> the call should use. If none is specified, the call defaults
1644 to using C calling conventions.
1646 <li>'<tt>ptr to function ty</tt>': shall be the signature of the pointer to
1647 function value being invoked. In most cases, this is a direct function
1648 invocation, but indirect <tt>invoke</tt>s are just as possible, branching off
1649 an arbitrary pointer to function value.
1652 <li>'<tt>function ptr val</tt>': An LLVM value containing a pointer to a
1653 function to be invoked. </li>
1655 <li>'<tt>function args</tt>': argument list whose types match the function
1656 signature argument types. If the function signature indicates the function
1657 accepts a variable number of arguments, the extra arguments can be
1660 <li>'<tt>normal label</tt>': the label reached when the called function
1661 executes a '<tt><a href="#i_ret">ret</a></tt>' instruction. </li>
1663 <li>'<tt>exception label</tt>': the label reached when a callee returns with
1664 the <a href="#i_unwind"><tt>unwind</tt></a> instruction. </li>
1670 <p>This instruction is designed to operate as a standard '<tt><a
1671 href="#i_call">call</a></tt>' instruction in most regards. The primary
1672 difference is that it establishes an association with a label, which is used by
1673 the runtime library to unwind the stack.</p>
1675 <p>This instruction is used in languages with destructors to ensure that proper
1676 cleanup is performed in the case of either a <tt>longjmp</tt> or a thrown
1677 exception. Additionally, this is important for implementation of
1678 '<tt>catch</tt>' clauses in high-level languages that support them.</p>
1682 %retval = invoke i32 %Test(i32 15) to label %Continue
1683 unwind label %TestCleanup <i>; {i32}:retval set</i>
1684 %retval = invoke <a href="#callingconv">coldcc</a> i32 %Test(i32 15) to label %Continue
1685 unwind label %TestCleanup <i>; {i32}:retval set</i>
1690 <!-- _______________________________________________________________________ -->
1692 <div class="doc_subsubsection"> <a name="i_unwind">'<tt>unwind</tt>'
1693 Instruction</a> </div>
1695 <div class="doc_text">
1704 <p>The '<tt>unwind</tt>' instruction unwinds the stack, continuing control flow
1705 at the first callee in the dynamic call stack which used an <a
1706 href="#i_invoke"><tt>invoke</tt></a> instruction to perform the call. This is
1707 primarily used to implement exception handling.</p>
1711 <p>The '<tt>unwind</tt>' intrinsic causes execution of the current function to
1712 immediately halt. The dynamic call stack is then searched for the first <a
1713 href="#i_invoke"><tt>invoke</tt></a> instruction on the call stack. Once found,
1714 execution continues at the "exceptional" destination block specified by the
1715 <tt>invoke</tt> instruction. If there is no <tt>invoke</tt> instruction in the
1716 dynamic call chain, undefined behavior results.</p>
1719 <!-- _______________________________________________________________________ -->
1721 <div class="doc_subsubsection"> <a name="i_unreachable">'<tt>unreachable</tt>'
1722 Instruction</a> </div>
1724 <div class="doc_text">
1733 <p>The '<tt>unreachable</tt>' instruction has no defined semantics. This
1734 instruction is used to inform the optimizer that a particular portion of the
1735 code is not reachable. This can be used to indicate that the code after a
1736 no-return function cannot be reached, and other facts.</p>
1740 <p>The '<tt>unreachable</tt>' instruction has no defined semantics.</p>
1745 <!-- ======================================================================= -->
1746 <div class="doc_subsection"> <a name="binaryops">Binary Operations</a> </div>
1747 <div class="doc_text">
1748 <p>Binary operators are used to do most of the computation in a
1749 program. They require two operands, execute an operation on them, and
1750 produce a single value. The operands might represent
1751 multiple data, as is the case with the <a href="#t_packed">packed</a> data type.
1752 The result value of a binary operator is not
1753 necessarily the same type as its operands.</p>
1754 <p>There are several different binary operators:</p>
1756 <!-- _______________________________________________________________________ -->
1757 <div class="doc_subsubsection"> <a name="i_add">'<tt>add</tt>'
1758 Instruction</a> </div>
1759 <div class="doc_text">
1761 <pre> <result> = add <ty> <var1>, <var2> <i>; yields {ty}:result</i>
1764 <p>The '<tt>add</tt>' instruction returns the sum of its two operands.</p>
1766 <p>The two arguments to the '<tt>add</tt>' instruction must be either <a
1767 href="#t_integer">integer</a> or <a href="#t_floating">floating point</a> values.
1768 This instruction can also take <a href="#t_packed">packed</a> versions of the values.
1769 Both arguments must have identical types.</p>
1771 <p>The value produced is the integer or floating point sum of the two
1774 <pre> <result> = add i32 4, %var <i>; yields {i32}:result = 4 + %var</i>
1777 <!-- _______________________________________________________________________ -->
1778 <div class="doc_subsubsection"> <a name="i_sub">'<tt>sub</tt>'
1779 Instruction</a> </div>
1780 <div class="doc_text">
1782 <pre> <result> = sub <ty> <var1>, <var2> <i>; yields {ty}:result</i>
1785 <p>The '<tt>sub</tt>' instruction returns the difference of its two
1787 <p>Note that the '<tt>sub</tt>' instruction is used to represent the '<tt>neg</tt>'
1788 instruction present in most other intermediate representations.</p>
1790 <p>The two arguments to the '<tt>sub</tt>' instruction must be either <a
1791 href="#t_integer">integer</a> or <a href="#t_floating">floating point</a>
1793 This instruction can also take <a href="#t_packed">packed</a> versions of the values.
1794 Both arguments must have identical types.</p>
1796 <p>The value produced is the integer or floating point difference of
1797 the two operands.</p>
1799 <pre> <result> = sub i32 4, %var <i>; yields {i32}:result = 4 - %var</i>
1800 <result> = sub i32 0, %val <i>; yields {i32}:result = -%var</i>
1803 <!-- _______________________________________________________________________ -->
1804 <div class="doc_subsubsection"> <a name="i_mul">'<tt>mul</tt>'
1805 Instruction</a> </div>
1806 <div class="doc_text">
1808 <pre> <result> = mul <ty> <var1>, <var2> <i>; yields {ty}:result</i>
1811 <p>The '<tt>mul</tt>' instruction returns the product of its two
1814 <p>The two arguments to the '<tt>mul</tt>' instruction must be either <a
1815 href="#t_integer">integer</a> or <a href="#t_floating">floating point</a>
1817 This instruction can also take <a href="#t_packed">packed</a> versions of the values.
1818 Both arguments must have identical types.</p>
1820 <p>The value produced is the integer or floating point product of the
1822 <p>Because the operands are the same width, the result of an integer
1823 multiplication is the same whether the operands should be deemed unsigned or
1826 <pre> <result> = mul i32 4, %var <i>; yields {i32}:result = 4 * %var</i>
1829 <!-- _______________________________________________________________________ -->
1830 <div class="doc_subsubsection"> <a name="i_udiv">'<tt>udiv</tt>' Instruction
1832 <div class="doc_text">
1834 <pre> <result> = udiv <ty> <var1>, <var2> <i>; yields {ty}:result</i>
1837 <p>The '<tt>udiv</tt>' instruction returns the quotient of its two
1840 <p>The two arguments to the '<tt>udiv</tt>' instruction must be
1841 <a href="#t_integer">integer</a> values. Both arguments must have identical
1842 types. This instruction can also take <a href="#t_packed">packed</a> versions
1843 of the values in which case the elements must be integers.</p>
1845 <p>The value produced is the unsigned integer quotient of the two operands. This
1846 instruction always performs an unsigned division operation, regardless of
1847 whether the arguments are unsigned or not.</p>
1849 <pre> <result> = udiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
1852 <!-- _______________________________________________________________________ -->
1853 <div class="doc_subsubsection"> <a name="i_sdiv">'<tt>sdiv</tt>' Instruction
1855 <div class="doc_text">
1857 <pre> <result> = sdiv <ty> <var1>, <var2> <i>; yields {ty}:result</i>
1860 <p>The '<tt>sdiv</tt>' instruction returns the quotient of its two
1863 <p>The two arguments to the '<tt>sdiv</tt>' instruction must be
1864 <a href="#t_integer">integer</a> values. Both arguments must have identical
1865 types. This instruction can also take <a href="#t_packed">packed</a> versions
1866 of the values in which case the elements must be integers.</p>
1868 <p>The value produced is the signed integer quotient of the two operands. This
1869 instruction always performs a signed division operation, regardless of whether
1870 the arguments are signed or not.</p>
1872 <pre> <result> = sdiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
1875 <!-- _______________________________________________________________________ -->
1876 <div class="doc_subsubsection"> <a name="i_fdiv">'<tt>fdiv</tt>'
1877 Instruction</a> </div>
1878 <div class="doc_text">
1880 <pre> <result> = fdiv <ty> <var1>, <var2> <i>; yields {ty}:result</i>
1883 <p>The '<tt>fdiv</tt>' instruction returns the quotient of its two
1886 <p>The two arguments to the '<tt>div</tt>' instruction must be
1887 <a href="#t_floating">floating point</a> values. Both arguments must have
1888 identical types. This instruction can also take <a href="#t_packed">packed</a>
1889 versions of the values in which case the elements must be floating point.</p>
1891 <p>The value produced is the floating point quotient of the two operands.</p>
1893 <pre> <result> = fdiv float 4.0, %var <i>; yields {float}:result = 4.0 / %var</i>
1896 <!-- _______________________________________________________________________ -->
1897 <div class="doc_subsubsection"> <a name="i_urem">'<tt>urem</tt>' Instruction</a>
1899 <div class="doc_text">
1901 <pre> <result> = urem <ty> <var1>, <var2> <i>; yields {ty}:result</i>
1904 <p>The '<tt>urem</tt>' instruction returns the remainder from the
1905 unsigned division of its two arguments.</p>
1907 <p>The two arguments to the '<tt>urem</tt>' instruction must be
1908 <a href="#t_integer">integer</a> values. Both arguments must have identical
1911 <p>This instruction returns the unsigned integer <i>remainder</i> of a division.
1912 This instruction always performs an unsigned division to get the remainder,
1913 regardless of whether the arguments are unsigned or not.</p>
1915 <pre> <result> = urem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
1919 <!-- _______________________________________________________________________ -->
1920 <div class="doc_subsubsection"> <a name="i_srem">'<tt>srem</tt>'
1921 Instruction</a> </div>
1922 <div class="doc_text">
1924 <pre> <result> = srem <ty> <var1>, <var2> <i>; yields {ty}:result</i>
1927 <p>The '<tt>srem</tt>' instruction returns the remainder from the
1928 signed division of its two operands.</p>
1930 <p>The two arguments to the '<tt>srem</tt>' instruction must be
1931 <a href="#t_integer">integer</a> values. Both arguments must have identical
1934 <p>This instruction returns the <i>remainder</i> of a division (where the result
1935 has the same sign as the divisor), not the <i>modulus</i> (where the
1936 result has the same sign as the dividend) of a value. For more
1937 information about the difference, see <a
1938 href="http://mathforum.org/dr.math/problems/anne.4.28.99.html">The
1941 <pre> <result> = srem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
1945 <!-- _______________________________________________________________________ -->
1946 <div class="doc_subsubsection"> <a name="i_frem">'<tt>frem</tt>'
1947 Instruction</a> </div>
1948 <div class="doc_text">
1950 <pre> <result> = frem <ty> <var1>, <var2> <i>; yields {ty}:result</i>
1953 <p>The '<tt>frem</tt>' instruction returns the remainder from the
1954 division of its two operands.</p>
1956 <p>The two arguments to the '<tt>frem</tt>' instruction must be
1957 <a href="#t_floating">floating point</a> values. Both arguments must have
1958 identical types.</p>
1960 <p>This instruction returns the <i>remainder</i> of a division.</p>
1962 <pre> <result> = frem float 4.0, %var <i>; yields {float}:result = 4.0 % %var</i>
1966 <!-- ======================================================================= -->
1967 <div class="doc_subsection"> <a name="bitwiseops">Bitwise Binary
1968 Operations</a> </div>
1969 <div class="doc_text">
1970 <p>Bitwise binary operators are used to do various forms of
1971 bit-twiddling in a program. They are generally very efficient
1972 instructions and can commonly be strength reduced from other
1973 instructions. They require two operands, execute an operation on them,
1974 and produce a single value. The resulting value of the bitwise binary
1975 operators is always the same type as its first operand.</p>
1977 <!-- _______________________________________________________________________ -->
1978 <div class="doc_subsubsection"> <a name="i_and">'<tt>and</tt>'
1979 Instruction</a> </div>
1980 <div class="doc_text">
1982 <pre> <result> = and <ty> <var1>, <var2> <i>; yields {ty}:result</i>
1985 <p>The '<tt>and</tt>' instruction returns the bitwise logical and of
1986 its two operands.</p>
1988 <p>The two arguments to the '<tt>and</tt>' instruction must be <a
1989 href="#t_integer">integer</a> values. Both arguments must have
1990 identical types.</p>
1992 <p>The truth table used for the '<tt>and</tt>' instruction is:</p>
1994 <div style="align: center">
1995 <table border="1" cellspacing="0" cellpadding="4">
2026 <pre> <result> = and i32 4, %var <i>; yields {i32}:result = 4 & %var</i>
2027 <result> = and i32 15, 40 <i>; yields {i32}:result = 8</i>
2028 <result> = and i32 4, 8 <i>; yields {i32}:result = 0</i>
2031 <!-- _______________________________________________________________________ -->
2032 <div class="doc_subsubsection"> <a name="i_or">'<tt>or</tt>' Instruction</a> </div>
2033 <div class="doc_text">
2035 <pre> <result> = or <ty> <var1>, <var2> <i>; yields {ty}:result</i>
2038 <p>The '<tt>or</tt>' instruction returns the bitwise logical inclusive
2039 or of its two operands.</p>
2041 <p>The two arguments to the '<tt>or</tt>' instruction must be <a
2042 href="#t_integer">integer</a> values. Both arguments must have
2043 identical types.</p>
2045 <p>The truth table used for the '<tt>or</tt>' instruction is:</p>
2047 <div style="align: center">
2048 <table border="1" cellspacing="0" cellpadding="4">
2079 <pre> <result> = or i32 4, %var <i>; yields {i32}:result = 4 | %var</i>
2080 <result> = or i32 15, 40 <i>; yields {i32}:result = 47</i>
2081 <result> = or i32 4, 8 <i>; yields {i32}:result = 12</i>
2084 <!-- _______________________________________________________________________ -->
2085 <div class="doc_subsubsection"> <a name="i_xor">'<tt>xor</tt>'
2086 Instruction</a> </div>
2087 <div class="doc_text">
2089 <pre> <result> = xor <ty> <var1>, <var2> <i>; yields {ty}:result</i>
2092 <p>The '<tt>xor</tt>' instruction returns the bitwise logical exclusive
2093 or of its two operands. The <tt>xor</tt> is used to implement the
2094 "one's complement" operation, which is the "~" operator in C.</p>
2096 <p>The two arguments to the '<tt>xor</tt>' instruction must be <a
2097 href="#t_integer">integer</a> values. Both arguments must have
2098 identical types.</p>
2100 <p>The truth table used for the '<tt>xor</tt>' instruction is:</p>
2102 <div style="align: center">
2103 <table border="1" cellspacing="0" cellpadding="4">
2135 <pre> <result> = xor i32 4, %var <i>; yields {i32}:result = 4 ^ %var</i>
2136 <result> = xor i32 15, 40 <i>; yields {i32}:result = 39</i>
2137 <result> = xor i32 4, 8 <i>; yields {i32}:result = 12</i>
2138 <result> = xor i32 %V, -1 <i>; yields {i32}:result = ~%V</i>
2141 <!-- _______________________________________________________________________ -->
2142 <div class="doc_subsubsection"> <a name="i_shl">'<tt>shl</tt>'
2143 Instruction</a> </div>
2144 <div class="doc_text">
2146 <pre> <result> = shl <ty> <var1>, i8 <var2> <i>; yields {ty}:result</i>
2149 <p>The '<tt>shl</tt>' instruction returns the first operand shifted to
2150 the left a specified number of bits.</p>
2152 <p>The first argument to the '<tt>shl</tt>' instruction must be an <a
2153 href="#t_integer">integer</a> type. The second argument must be an '<tt>i8</tt>'
2156 <p>The value produced is <tt>var1</tt> * 2<sup><tt>var2</tt></sup>.</p>
2158 <pre> <result> = shl i32 4, i8 %var <i>; yields {i32}:result = 4 << %var</i>
2159 <result> = shl i32 4, i8 2 <i>; yields {i32}:result = 16</i>
2160 <result> = shl i32 1, i8 10 <i>; yields {i32}:result = 1024</i>
2163 <!-- _______________________________________________________________________ -->
2164 <div class="doc_subsubsection"> <a name="i_lshr">'<tt>lshr</tt>'
2165 Instruction</a> </div>
2166 <div class="doc_text">
2168 <pre> <result> = lshr <ty> <var1>, i8 <var2> <i>; yields {ty}:result</i>
2172 <p>The '<tt>lshr</tt>' instruction (logical shift right) returns the first
2173 operand shifted to the right a specified number of bits.</p>
2176 <p>The first argument to the '<tt>lshr</tt>' instruction must be an <a
2177 href="#t_integer">integer</a> type. The second argument must be an '<tt>i8</tt>' type.</p>
2180 <p>This instruction always performs a logical shift right operation. The
2181 <tt>var2</tt> most significant bits will be filled with zero bits after the
2186 <result> = lshr i32 4, i8 1 <i>; yields {i32}:result = 2</i>
2187 <result> = lshr i32 4, i8 2 <i>; yields {i32}:result = 1</i>
2188 <result> = lshr i8 4, i8 3 <i>; yields {i8 }:result = 0</i>
2189 <result> = lshr i8 -2, i8 1 <i>; yields {i8 }:result = 0x7FFFFFFF </i>
2193 <!-- ======================================================================= -->
2194 <div class="doc_subsubsection"> <a name="i_ashr">'<tt>ashr</tt>'
2195 Instruction</a> </div>
2196 <div class="doc_text">
2199 <pre> <result> = ashr <ty> <var1>, i8 <var2> <i>; yields {ty}:result</i>
2203 <p>The '<tt>ashr</tt>' instruction (arithmetic shift right) returns the first
2204 operand shifted to the right a specified number of bits.</p>
2207 <p>The first argument to the '<tt>ashr</tt>' instruction must be an
2208 <a href="#t_integer">integer</a> type. The second argument must be an
2209 '<tt>i8</tt>' type.</p>
2212 <p>This instruction always performs an arithmetic shift right operation,
2213 regardless of whether the arguments are signed or not. The <tt>var2</tt> most
2214 significant bits will be filled with the sign bit of <tt>var1</tt>.</p>
2218 <result> = ashr i32 4, i8 1 <i>; yields {i32}:result = 2</i>
2219 <result> = ashr i32 4, i8 2 <i>; yields {i32}:result = 1</i>
2220 <result> = ashr i8 4, i8 3 <i>; yields {i8}:result = 0</i>
2221 <result> = ashr i8 -2, i8 1 <i>; yields {i8 }:result = -1</i>
2225 <!-- ======================================================================= -->
2226 <div class="doc_subsection">
2227 <a name="vectorops">Vector Operations</a>
2230 <div class="doc_text">
2232 <p>LLVM supports several instructions to represent vector operations in a
2233 target-independent manner. This instructions cover the element-access and
2234 vector-specific operations needed to process vectors effectively. While LLVM
2235 does directly support these vector operations, many sophisticated algorithms
2236 will want to use target-specific intrinsics to take full advantage of a specific
2241 <!-- _______________________________________________________________________ -->
2242 <div class="doc_subsubsection">
2243 <a name="i_extractelement">'<tt>extractelement</tt>' Instruction</a>
2246 <div class="doc_text">
2251 <result> = extractelement <n x <ty>> <val>, i32 <idx> <i>; yields <ty></i>
2257 The '<tt>extractelement</tt>' instruction extracts a single scalar
2258 element from a packed vector at a specified index.
2265 The first operand of an '<tt>extractelement</tt>' instruction is a
2266 value of <a href="#t_packed">packed</a> type. The second operand is
2267 an index indicating the position from which to extract the element.
2268 The index may be a variable.</p>
2273 The result is a scalar of the same type as the element type of
2274 <tt>val</tt>. Its value is the value at position <tt>idx</tt> of
2275 <tt>val</tt>. If <tt>idx</tt> exceeds the length of <tt>val</tt>, the
2276 results are undefined.
2282 %result = extractelement <4 x i32> %vec, i32 0 <i>; yields i32</i>
2287 <!-- _______________________________________________________________________ -->
2288 <div class="doc_subsubsection">
2289 <a name="i_insertelement">'<tt>insertelement</tt>' Instruction</a>
2292 <div class="doc_text">
2297 <result> = insertelement <n x <ty>> <val>, <ty> <elt>, i32 <idx> <i>; yields <n x <ty>></i>
2303 The '<tt>insertelement</tt>' instruction inserts a scalar
2304 element into a packed vector at a specified index.
2311 The first operand of an '<tt>insertelement</tt>' instruction is a
2312 value of <a href="#t_packed">packed</a> type. The second operand is a
2313 scalar value whose type must equal the element type of the first
2314 operand. The third operand is an index indicating the position at
2315 which to insert the value. The index may be a variable.</p>
2320 The result is a packed vector of the same type as <tt>val</tt>. Its
2321 element values are those of <tt>val</tt> except at position
2322 <tt>idx</tt>, where it gets the value <tt>elt</tt>. If <tt>idx</tt>
2323 exceeds the length of <tt>val</tt>, the results are undefined.
2329 %result = insertelement <4 x i32> %vec, i32 1, i32 0 <i>; yields <4 x i32></i>
2333 <!-- _______________________________________________________________________ -->
2334 <div class="doc_subsubsection">
2335 <a name="i_shufflevector">'<tt>shufflevector</tt>' Instruction</a>
2338 <div class="doc_text">
2343 <result> = shufflevector <n x <ty>> <v1>, <n x <ty>> <v2>, <n x i32> <mask> <i>; yields <n x <ty>></i>
2349 The '<tt>shufflevector</tt>' instruction constructs a permutation of elements
2350 from two input vectors, returning a vector of the same type.
2356 The first two operands of a '<tt>shufflevector</tt>' instruction are vectors
2357 with types that match each other and types that match the result of the
2358 instruction. The third argument is a shuffle mask, which has the same number
2359 of elements as the other vector type, but whose element type is always 'i32'.
2363 The shuffle mask operand is required to be a constant vector with either
2364 constant integer or undef values.
2370 The elements of the two input vectors are numbered from left to right across
2371 both of the vectors. The shuffle mask operand specifies, for each element of
2372 the result vector, which element of the two input registers the result element
2373 gets. The element selector may be undef (meaning "don't care") and the second
2374 operand may be undef if performing a shuffle from only one vector.
2380 %result = shufflevector <4 x i32> %v1, <4 x i32> %v2,
2381 <4 x i32> <i32 0, i32 4, i32 1, i32 5> <i>; yields <4 x i32></i>
2382 %result = shufflevector <4 x i32> %v1, <4 x i32> undef,
2383 <4 x i32> <i32 0, i32 1, i32 2, i32 3> <i>; yields <4 x i32></i> - Identity shuffle.
2388 <!-- ======================================================================= -->
2389 <div class="doc_subsection">
2390 <a name="memoryops">Memory Access and Addressing Operations</a>
2393 <div class="doc_text">
2395 <p>A key design point of an SSA-based representation is how it
2396 represents memory. In LLVM, no memory locations are in SSA form, which
2397 makes things very simple. This section describes how to read, write,
2398 allocate, and free memory in LLVM.</p>
2402 <!-- _______________________________________________________________________ -->
2403 <div class="doc_subsubsection">
2404 <a name="i_malloc">'<tt>malloc</tt>' Instruction</a>
2407 <div class="doc_text">
2412 <result> = malloc <type>[, i32 <NumElements>][, align <alignment>] <i>; yields {type*}:result</i>
2417 <p>The '<tt>malloc</tt>' instruction allocates memory from the system
2418 heap and returns a pointer to it.</p>
2422 <p>The '<tt>malloc</tt>' instruction allocates
2423 <tt>sizeof(<type>)*NumElements</tt>
2424 bytes of memory from the operating system and returns a pointer of the
2425 appropriate type to the program. If "NumElements" is specified, it is the
2426 number of elements allocated. If an alignment is specified, the value result
2427 of the allocation is guaranteed to be aligned to at least that boundary. If
2428 not specified, or if zero, the target can choose to align the allocation on any
2429 convenient boundary.</p>
2431 <p>'<tt>type</tt>' must be a sized type.</p>
2435 <p>Memory is allocated using the system "<tt>malloc</tt>" function, and
2436 a pointer is returned.</p>
2441 %array = malloc [4 x i8 ] <i>; yields {[%4 x i8]*}:array</i>
2443 %size = <a href="#i_add">add</a> i32 2, 2 <i>; yields {i32}:size = i32 4</i>
2444 %array1 = malloc i8, i32 4 <i>; yields {i8*}:array1</i>
2445 %array2 = malloc [12 x i8], i32 %size <i>; yields {[12 x i8]*}:array2</i>
2446 %array3 = malloc i32, i32 4, align 1024 <i>; yields {i32*}:array3</i>
2447 %array4 = malloc i32, align 1024 <i>; yields {i32*}:array4</i>
2451 <!-- _______________________________________________________________________ -->
2452 <div class="doc_subsubsection">
2453 <a name="i_free">'<tt>free</tt>' Instruction</a>
2456 <div class="doc_text">
2461 free <type> <value> <i>; yields {void}</i>
2466 <p>The '<tt>free</tt>' instruction returns memory back to the unused
2467 memory heap to be reallocated in the future.</p>
2471 <p>'<tt>value</tt>' shall be a pointer value that points to a value
2472 that was allocated with the '<tt><a href="#i_malloc">malloc</a></tt>'
2477 <p>Access to the memory pointed to by the pointer is no longer defined
2478 after this instruction executes.</p>
2483 %array = <a href="#i_malloc">malloc</a> [4 x i8] <i>; yields {[4 x i8]*}:array</i>
2484 free [4 x i8]* %array
2488 <!-- _______________________________________________________________________ -->
2489 <div class="doc_subsubsection">
2490 <a name="i_alloca">'<tt>alloca</tt>' Instruction</a>
2493 <div class="doc_text">
2498 <result> = alloca <type>[, i32 <NumElements>][, align <alignment>] <i>; yields {type*}:result</i>
2503 <p>The '<tt>alloca</tt>' instruction allocates memory on the current
2504 stack frame of the procedure that is live until the current function
2505 returns to its caller.</p>
2509 <p>The '<tt>alloca</tt>' instruction allocates <tt>sizeof(<type>)*NumElements</tt>
2510 bytes of memory on the runtime stack, returning a pointer of the
2511 appropriate type to the program. If "NumElements" is specified, it is the
2512 number of elements allocated. If an alignment is specified, the value result
2513 of the allocation is guaranteed to be aligned to at least that boundary. If
2514 not specified, or if zero, the target can choose to align the allocation on any
2515 convenient boundary.</p>
2517 <p>'<tt>type</tt>' may be any sized type.</p>
2521 <p>Memory is allocated; a pointer is returned. '<tt>alloca</tt>'d
2522 memory is automatically released when the function returns. The '<tt>alloca</tt>'
2523 instruction is commonly used to represent automatic variables that must
2524 have an address available. When the function returns (either with the <tt><a
2525 href="#i_ret">ret</a></tt> or <tt><a href="#i_unwind">unwind</a></tt>
2526 instructions), the memory is reclaimed.</p>
2531 %ptr = alloca i32 <i>; yields {i32*}:ptr</i>
2532 %ptr = alloca i32, i32 4 <i>; yields {i32*}:ptr</i>
2533 %ptr = alloca i32, i32 4, align 1024 <i>; yields {i32*}:ptr</i>
2534 %ptr = alloca i32, align 1024 <i>; yields {i32*}:ptr</i>
2538 <!-- _______________________________________________________________________ -->
2539 <div class="doc_subsubsection"> <a name="i_load">'<tt>load</tt>'
2540 Instruction</a> </div>
2541 <div class="doc_text">
2543 <pre> <result> = load <ty>* <pointer><br> <result> = volatile load <ty>* <pointer><br></pre>
2545 <p>The '<tt>load</tt>' instruction is used to read from memory.</p>
2547 <p>The argument to the '<tt>load</tt>' instruction specifies the memory
2548 address from which to load. The pointer must point to a <a
2549 href="#t_firstclass">first class</a> type. If the <tt>load</tt> is
2550 marked as <tt>volatile</tt>, then the optimizer is not allowed to modify
2551 the number or order of execution of this <tt>load</tt> with other
2552 volatile <tt>load</tt> and <tt><a href="#i_store">store</a></tt>
2555 <p>The location of memory pointed to is loaded.</p>
2557 <pre> %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
2559 href="#i_store">store</a> i32 3, i32* %ptr <i>; yields {void}</i>
2560 %val = load i32* %ptr <i>; yields {i32}:val = i32 3</i>
2563 <!-- _______________________________________________________________________ -->
2564 <div class="doc_subsubsection"> <a name="i_store">'<tt>store</tt>'
2565 Instruction</a> </div>
2566 <div class="doc_text">
2568 <pre> store <ty> <value>, <ty>* <pointer> <i>; yields {void}</i>
2569 volatile store <ty> <value>, <ty>* <pointer> <i>; yields {void}</i>
2572 <p>The '<tt>store</tt>' instruction is used to write to memory.</p>
2574 <p>There are two arguments to the '<tt>store</tt>' instruction: a value
2575 to store and an address in which to store it. The type of the '<tt><pointer></tt>'
2576 operand must be a pointer to the type of the '<tt><value></tt>'
2577 operand. If the <tt>store</tt> is marked as <tt>volatile</tt>, then the
2578 optimizer is not allowed to modify the number or order of execution of
2579 this <tt>store</tt> with other volatile <tt>load</tt> and <tt><a
2580 href="#i_store">store</a></tt> instructions.</p>
2582 <p>The contents of memory are updated to contain '<tt><value></tt>'
2583 at the location specified by the '<tt><pointer></tt>' operand.</p>
2585 <pre> %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
2587 href="#i_store">store</a> i32 3, i32* %ptr <i>; yields {void}</i>
2588 %val = load i32* %ptr <i>; yields {i32}:val = i32 3</i>
2592 <!-- _______________________________________________________________________ -->
2593 <div class="doc_subsubsection">
2594 <a name="i_getelementptr">'<tt>getelementptr</tt>' Instruction</a>
2597 <div class="doc_text">
2600 <result> = getelementptr <ty>* <ptrval>{, <ty> <idx>}*
2606 The '<tt>getelementptr</tt>' instruction is used to get the address of a
2607 subelement of an aggregate data structure.</p>
2611 <p>This instruction takes a list of integer operands that indicate what
2612 elements of the aggregate object to index to. The actual types of the arguments
2613 provided depend on the type of the first pointer argument. The
2614 '<tt>getelementptr</tt>' instruction is used to index down through the type
2615 levels of a structure or to a specific index in an array. When indexing into a
2616 structure, only <tt>i32</tt> integer constants are allowed. When indexing
2617 into an array or pointer, only integers of 32 or 64 bits are allowed, and will
2618 be sign extended to 64-bit values.</p>
2620 <p>For example, let's consider a C code fragment and how it gets
2621 compiled to LLVM:</p>
2635 define i32 *foo(struct ST *s) {
2636 return &s[1].Z.B[5][13];
2640 <p>The LLVM code generated by the GCC frontend is:</p>
2643 %RT = type { i8 , [10 x [20 x i32]], i8 }
2644 %ST = type { i32, double, %RT }
2648 define i32* %foo(%ST* %s) {
2650 %reg = getelementptr %ST* %s, i32 1, i32 2, i32 1, i32 5, i32 13
2657 <p>The index types specified for the '<tt>getelementptr</tt>' instruction depend
2658 on the pointer type that is being indexed into. <a href="#t_pointer">Pointer</a>
2659 and <a href="#t_array">array</a> types can use a 32-bit or 64-bit
2660 <a href="#t_integer">integer</a> type but the value will always be sign extended
2661 to 64-bits. <a href="#t_struct">Structure</a> types, require <tt>i32</tt>
2662 <b>constants</b>.</p>
2664 <p>In the example above, the first index is indexing into the '<tt>%ST*</tt>'
2665 type, which is a pointer, yielding a '<tt>%ST</tt>' = '<tt>{ i32, double, %RT
2666 }</tt>' type, a structure. The second index indexes into the third element of
2667 the structure, yielding a '<tt>%RT</tt>' = '<tt>{ i8 , [10 x [20 x i32]],
2668 i8 }</tt>' type, another structure. The third index indexes into the second
2669 element of the structure, yielding a '<tt>[10 x [20 x i32]]</tt>' type, an
2670 array. The two dimensions of the array are subscripted into, yielding an
2671 '<tt>i32</tt>' type. The '<tt>getelementptr</tt>' instruction returns a pointer
2672 to this element, thus computing a value of '<tt>i32*</tt>' type.</p>
2674 <p>Note that it is perfectly legal to index partially through a
2675 structure, returning a pointer to an inner element. Because of this,
2676 the LLVM code for the given testcase is equivalent to:</p>
2679 define i32* %foo(%ST* %s) {
2680 %t1 = getelementptr %ST* %s, i32 1 <i>; yields %ST*:%t1</i>
2681 %t2 = getelementptr %ST* %t1, i32 0, i32 2 <i>; yields %RT*:%t2</i>
2682 %t3 = getelementptr %RT* %t2, i32 0, i32 1 <i>; yields [10 x [20 x i32]]*:%t3</i>
2683 %t4 = getelementptr [10 x [20 x i32]]* %t3, i32 0, i32 5 <i>; yields [20 x i32]*:%t4</i>
2684 %t5 = getelementptr [20 x i32]* %t4, i32 0, i32 13 <i>; yields i32*:%t5</i>
2689 <p>Note that it is undefined to access an array out of bounds: array and
2690 pointer indexes must always be within the defined bounds of the array type.
2691 The one exception for this rules is zero length arrays. These arrays are
2692 defined to be accessible as variable length arrays, which requires access
2693 beyond the zero'th element.</p>
2695 <p>The getelementptr instruction is often confusing. For some more insight
2696 into how it works, see <a href="GetElementPtr.html">the getelementptr
2702 <i>; yields [12 x i8]*:aptr</i>
2703 %aptr = getelementptr {i32, [12 x i8]}* %sptr, i64 0, i32 1
2707 <!-- ======================================================================= -->
2708 <div class="doc_subsection"> <a name="convertops">Conversion Operations</a>
2710 <div class="doc_text">
2711 <p>The instructions in this category are the conversion instructions (casting)
2712 which all take a single operand and a type. They perform various bit conversions
2716 <!-- _______________________________________________________________________ -->
2717 <div class="doc_subsubsection">
2718 <a name="i_trunc">'<tt>trunc .. to</tt>' Instruction</a>
2720 <div class="doc_text">
2724 <result> = trunc <ty> <value> to <ty2> <i>; yields ty2</i>
2729 The '<tt>trunc</tt>' instruction truncates its operand to the type <tt>ty2</tt>.
2734 The '<tt>trunc</tt>' instruction takes a <tt>value</tt> to trunc, which must
2735 be an <a href="#t_integer">integer</a> type, and a type that specifies the size
2736 and type of the result, which must be an <a href="#t_integer">integer</a>
2737 type. The bit size of <tt>value</tt> must be larger than the bit size of
2738 <tt>ty2</tt>. Equal sized types are not allowed.</p>
2742 The '<tt>trunc</tt>' instruction truncates the high order bits in <tt>value</tt>
2743 and converts the remaining bits to <tt>ty2</tt>. Since the source size must be
2744 larger than the destination size, <tt>trunc</tt> cannot be a <i>no-op cast</i>.
2745 It will always truncate bits.</p>
2749 %X = trunc i32 257 to i8 <i>; yields i8:1</i>
2750 %Y = trunc i32 123 to i1 <i>; yields i1:true</i>
2751 %Y = trunc i32 122 to i1 <i>; yields i1:false</i>
2755 <!-- _______________________________________________________________________ -->
2756 <div class="doc_subsubsection">
2757 <a name="i_zext">'<tt>zext .. to</tt>' Instruction</a>
2759 <div class="doc_text">
2763 <result> = zext <ty> <value> to <ty2> <i>; yields ty2</i>
2767 <p>The '<tt>zext</tt>' instruction zero extends its operand to type
2772 <p>The '<tt>zext</tt>' instruction takes a value to cast, which must be of
2773 <a href="#t_integer">integer</a> type, and a type to cast it to, which must
2774 also be of <a href="#t_integer">integer</a> type. The bit size of the
2775 <tt>value</tt> must be smaller than the bit size of the destination type,
2779 <p>The <tt>zext</tt> fills the high order bits of the <tt>value</tt> with zero
2780 bits until it reaches the size of the destination type, <tt>ty2</tt>. When the
2781 the operand and the type are the same size, no bit filling is done and the
2782 cast is considered a <i>no-op cast</i> because no bits change (only the type
2785 <p>When zero extending from i1, the result will always be either 0 or 1.</p>
2789 %X = zext i32 257 to i64 <i>; yields i64:257</i>
2790 %Y = zext i1 true to i32 <i>; yields i32:1</i>
2794 <!-- _______________________________________________________________________ -->
2795 <div class="doc_subsubsection">
2796 <a name="i_sext">'<tt>sext .. to</tt>' Instruction</a>
2798 <div class="doc_text">
2802 <result> = sext <ty> <value> to <ty2> <i>; yields ty2</i>
2806 <p>The '<tt>sext</tt>' sign extends <tt>value</tt> to the type <tt>ty2</tt>.</p>
2810 The '<tt>sext</tt>' instruction takes a value to cast, which must be of
2811 <a href="#t_integer">integer</a> type, and a type to cast it to, which must
2812 also be of <a href="#t_integer">integer</a> type. The bit size of the
2813 <tt>value</tt> must be smaller than the bit size of the destination type,
2818 The '<tt>sext</tt>' instruction performs a sign extension by copying the sign
2819 bit (highest order bit) of the <tt>value</tt> until it reaches the bit size of
2820 the type <tt>ty2</tt>. When the the operand and the type are the same size,
2821 no bit filling is done and the cast is considered a <i>no-op cast</i> because
2822 no bits change (only the type changes).</p>
2824 <p>When sign extending from i1, the extension always results in -1 or 0.</p>
2828 %X = sext i8 -1 to i16 <i>; yields i16 :65535</i>
2829 %Y = sext i1 true to i32 <i>; yields i32:-1</i>
2833 <!-- _______________________________________________________________________ -->
2834 <div class="doc_subsubsection">
2835 <a name="i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a>
2838 <div class="doc_text">
2843 <result> = fptrunc <ty> <value> to <ty2> <i>; yields ty2</i>
2847 <p>The '<tt>fptrunc</tt>' instruction truncates <tt>value</tt> to type
2852 <p>The '<tt>fptrunc</tt>' instruction takes a <a href="#t_floating">floating
2853 point</a> value to cast and a <a href="#t_floating">floating point</a> type to
2854 cast it to. The size of <tt>value</tt> must be larger than the size of
2855 <tt>ty2</tt>. This implies that <tt>fptrunc</tt> cannot be used to make a
2856 <i>no-op cast</i>.</p>
2859 <p> The '<tt>fptrunc</tt>' instruction truncates a <tt>value</tt> from a larger
2860 <a href="#t_floating">floating point</a> type to a smaller
2861 <a href="#t_floating">floating point</a> type. If the value cannot fit within
2862 the destination type, <tt>ty2</tt>, then the results are undefined.</p>
2866 %X = fptrunc double 123.0 to float <i>; yields float:123.0</i>
2867 %Y = fptrunc double 1.0E+300 to float <i>; yields undefined</i>
2871 <!-- _______________________________________________________________________ -->
2872 <div class="doc_subsubsection">
2873 <a name="i_fpext">'<tt>fpext .. to</tt>' Instruction</a>
2875 <div class="doc_text">
2879 <result> = fpext <ty> <value> to <ty2> <i>; yields ty2</i>
2883 <p>The '<tt>fpext</tt>' extends a floating point <tt>value</tt> to a larger
2884 floating point value.</p>
2887 <p>The '<tt>fpext</tt>' instruction takes a
2888 <a href="#t_floating">floating point</a> <tt>value</tt> to cast,
2889 and a <a href="#t_floating">floating point</a> type to cast it to. The source
2890 type must be smaller than the destination type.</p>
2893 <p>The '<tt>fpext</tt>' instruction extends the <tt>value</tt> from a smaller
2894 <a href="t_floating">floating point</a> type to a larger
2895 <a href="t_floating">floating point</a> type. The <tt>fpext</tt> cannot be
2896 used to make a <i>no-op cast</i> because it always changes bits. Use
2897 <tt>bitcast</tt> to make a <i>no-op cast</i> for a floating point cast.</p>
2901 %X = fpext float 3.1415 to double <i>; yields double:3.1415</i>
2902 %Y = fpext float 1.0 to float <i>; yields float:1.0 (no-op)</i>
2906 <!-- _______________________________________________________________________ -->
2907 <div class="doc_subsubsection">
2908 <a name="i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a>
2910 <div class="doc_text">
2914 <result> = fp2uint <ty> <value> to <ty2> <i>; yields ty2</i>
2918 <p>The '<tt>fp2uint</tt>' converts a floating point <tt>value</tt> to its
2919 unsigned integer equivalent of type <tt>ty2</tt>.
2923 <p>The '<tt>fp2uint</tt>' instruction takes a value to cast, which must be a
2924 <a href="#t_floating">floating point</a> value, and a type to cast it to, which
2925 must be an <a href="#t_integer">integer</a> type.</p>
2928 <p> The '<tt>fp2uint</tt>' instruction converts its
2929 <a href="#t_floating">floating point</a> operand into the nearest (rounding
2930 towards zero) unsigned integer value. If the value cannot fit in <tt>ty2</tt>,
2931 the results are undefined.</p>
2933 <p>When converting to i1, the conversion is done as a comparison against
2934 zero. If the <tt>value</tt> was zero, the i1 result will be <tt>false</tt>.
2935 If the <tt>value</tt> was non-zero, the i1 result will be <tt>true</tt>.</p>
2939 %X = fp2uint double 123.0 to i32 <i>; yields i32:123</i>
2940 %Y = fp2uint float 1.0E+300 to i1 <i>; yields i1:true</i>
2941 %X = fp2uint float 1.04E+17 to i8 <i>; yields undefined:1</i>
2945 <!-- _______________________________________________________________________ -->
2946 <div class="doc_subsubsection">
2947 <a name="i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a>
2949 <div class="doc_text">
2953 <result> = fptosi <ty> <value> to <ty2> <i>; yields ty2</i>
2957 <p>The '<tt>fptosi</tt>' instruction converts
2958 <a href="#t_floating">floating point</a> <tt>value</tt> to type <tt>ty2</tt>.
2963 <p> The '<tt>fptosi</tt>' instruction takes a value to cast, which must be a
2964 <a href="#t_floating">floating point</a> value, and a type to cast it to, which
2965 must also be an <a href="#t_integer">integer</a> type.</p>
2968 <p>The '<tt>fptosi</tt>' instruction converts its
2969 <a href="#t_floating">floating point</a> operand into the nearest (rounding
2970 towards zero) signed integer value. If the value cannot fit in <tt>ty2</tt>,
2971 the results are undefined.</p>
2973 <p>When converting to i1, the conversion is done as a comparison against
2974 zero. If the <tt>value</tt> was zero, the i1 result will be <tt>false</tt>.
2975 If the <tt>value</tt> was non-zero, the i1 result will be <tt>true</tt>.</p>
2979 %X = fptosi double -123.0 to i32 <i>; yields i32:-123</i>
2980 %Y = fptosi float 1.0E-247 to i1 <i>; yields i1:true</i>
2981 %X = fptosi float 1.04E+17 to i8 <i>; yields undefined:1</i>
2985 <!-- _______________________________________________________________________ -->
2986 <div class="doc_subsubsection">
2987 <a name="i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a>
2989 <div class="doc_text">
2993 <result> = uitofp <ty> <value> to <ty2> <i>; yields ty2</i>
2997 <p>The '<tt>uitofp</tt>' instruction regards <tt>value</tt> as an unsigned
2998 integer and converts that value to the <tt>ty2</tt> type.</p>
3002 <p>The '<tt>uitofp</tt>' instruction takes a value to cast, which must be an
3003 <a href="#t_integer">integer</a> value, and a type to cast it to, which must
3004 be a <a href="#t_floating">floating point</a> type.</p>
3007 <p>The '<tt>uitofp</tt>' instruction interprets its operand as an unsigned
3008 integer quantity and converts it to the corresponding floating point value. If
3009 the value cannot fit in the floating point value, the results are undefined.</p>
3014 %X = uitofp i32 257 to float <i>; yields float:257.0</i>
3015 %Y = uitofp i8 -1 to double <i>; yields double:255.0</i>
3019 <!-- _______________________________________________________________________ -->
3020 <div class="doc_subsubsection">
3021 <a name="i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a>
3023 <div class="doc_text">
3027 <result> = sitofp <ty> <value> to <ty2> <i>; yields ty2</i>
3031 <p>The '<tt>sitofp</tt>' instruction regards <tt>value</tt> as a signed
3032 integer and converts that value to the <tt>ty2</tt> type.</p>
3035 <p>The '<tt>sitofp</tt>' instruction takes a value to cast, which must be an
3036 <a href="#t_integer">integer</a> value, and a type to cast it to, which must be
3037 a <a href="#t_floating">floating point</a> type.</p>
3040 <p>The '<tt>sitofp</tt>' instruction interprets its operand as a signed
3041 integer quantity and converts it to the corresponding floating point value. If
3042 the value cannot fit in the floating point value, the results are undefined.</p>
3046 %X = sitofp i32 257 to float <i>; yields float:257.0</i>
3047 %Y = sitofp i8 -1 to double <i>; yields double:-1.0</i>
3051 <!-- _______________________________________________________________________ -->
3052 <div class="doc_subsubsection">
3053 <a name="i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a>
3055 <div class="doc_text">
3059 <result> = ptrtoint <ty> <value> to <ty2> <i>; yields ty2</i>
3063 <p>The '<tt>ptrtoint</tt>' instruction converts the pointer <tt>value</tt> to
3064 the integer type <tt>ty2</tt>.</p>
3067 <p>The '<tt>ptrtoint</tt>' instruction takes a <tt>value</tt> to cast, which
3068 must be a <a href="t_pointer">pointer</a> value, and a type to cast it to
3069 <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a> type.
3072 <p>The '<tt>ptrtoint</tt>' instruction converts <tt>value</tt> to integer type
3073 <tt>ty2</tt> by interpreting the pointer value as an integer and either
3074 truncating or zero extending that value to the size of the integer type. If
3075 <tt>value</tt> is smaller than <tt>ty2</tt> then a zero extension is done. If
3076 <tt>value</tt> is larger than <tt>ty2</tt> then a truncation is done. If they
3077 are the same size, then nothing is done (<i>no-op cast</i>).</p>
3081 %X = ptrtoint i32* %X to i8 <i>; yields truncation on 32-bit</i>
3082 %Y = ptrtoint i32* %x to i64 <i>; yields zero extend on 32-bit</i>
3086 <!-- _______________________________________________________________________ -->
3087 <div class="doc_subsubsection">
3088 <a name="i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a>
3090 <div class="doc_text">
3094 <result> = inttoptr <ty> <value> to <ty2> <i>; yields ty2</i>
3098 <p>The '<tt>inttoptr</tt>' instruction converts an integer <tt>value</tt> to
3099 a pointer type, <tt>ty2</tt>.</p>
3102 <p>The '<tt>inttoptr</tt>' instruction takes an <a href="i_integer">integer</a>
3103 value to cast, and a type to cast it to, which must be a
3104 <a href="#t_pointer">pointer</a> type.
3107 <p>The '<tt>inttoptr</tt>' instruction converts <tt>value</tt> to type
3108 <tt>ty2</tt> by applying either a zero extension or a truncation depending on
3109 the size of the integer <tt>value</tt>. If <tt>value</tt> is larger than the
3110 size of a pointer then a truncation is done. If <tt>value</tt> is smaller than
3111 the size of a pointer then a zero extension is done. If they are the same size,
3112 nothing is done (<i>no-op cast</i>).</p>
3116 %X = inttoptr i32 255 to i32* <i>; yields zero extend on 64-bit</i>
3117 %X = inttoptr i32 255 to i32* <i>; yields no-op on 32-bit </i>
3118 %Y = inttoptr i16 0 to i32* <i>; yields zero extend on 32-bit</i>
3122 <!-- _______________________________________________________________________ -->
3123 <div class="doc_subsubsection">
3124 <a name="i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a>
3126 <div class="doc_text">
3130 <result> = bitcast <ty> <value> to <ty2> <i>; yields ty2</i>
3134 <p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
3135 <tt>ty2</tt> without changing any bits.</p>
3138 <p>The '<tt>bitcast</tt>' instruction takes a value to cast, which must be
3139 a first class value, and a type to cast it to, which must also be a <a
3140 href="#t_firstclass">first class</a> type. The bit sizes of <tt>value</tt>
3141 and the destination type, <tt>ty2</tt>, must be identical. If the source
3142 type is a pointer, the destination type must also be a pointer.</p>
3145 <p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
3146 <tt>ty2</tt>. It is always a <i>no-op cast</i> because no bits change with
3147 this conversion. The conversion is done as if the <tt>value</tt> had been
3148 stored to memory and read back as type <tt>ty2</tt>. Pointer types may only be
3149 converted to other pointer types with this instruction. To convert pointers to
3150 other types, use the <a href="#i_inttoptr">inttoptr</a> or
3151 <a href="#i_ptrtoint">ptrtoint</a> instructions first.</p>
3155 %X = bitcast i8 255 to i8 <i>; yields i8 :-1</i>
3156 %Y = bitcast i32* %x to sint* <i>; yields sint*:%x</i>
3157 %Z = bitcast <2xint> %V to i64; <i>; yields i64: %V</i>
3161 <!-- ======================================================================= -->
3162 <div class="doc_subsection"> <a name="otherops">Other Operations</a> </div>
3163 <div class="doc_text">
3164 <p>The instructions in this category are the "miscellaneous"
3165 instructions, which defy better classification.</p>
3168 <!-- _______________________________________________________________________ -->
3169 <div class="doc_subsubsection"><a name="i_icmp">'<tt>icmp</tt>' Instruction</a>
3171 <div class="doc_text">
3173 <pre> <result> = icmp <cond> <ty> <var1>, <var2>
3174 <i>; yields {i1}:result</i>
3177 <p>The '<tt>icmp</tt>' instruction returns a boolean value based on comparison
3178 of its two integer operands.</p>
3180 <p>The '<tt>icmp</tt>' instruction takes three operands. The first operand is
3181 the condition code which indicates the kind of comparison to perform. It is not
3182 a value, just a keyword. The possibilities for the condition code are:
3184 <li><tt>eq</tt>: equal</li>
3185 <li><tt>ne</tt>: not equal </li>
3186 <li><tt>ugt</tt>: unsigned greater than</li>
3187 <li><tt>uge</tt>: unsigned greater or equal</li>
3188 <li><tt>ult</tt>: unsigned less than</li>
3189 <li><tt>ule</tt>: unsigned less or equal</li>
3190 <li><tt>sgt</tt>: signed greater than</li>
3191 <li><tt>sge</tt>: signed greater or equal</li>
3192 <li><tt>slt</tt>: signed less than</li>
3193 <li><tt>sle</tt>: signed less or equal</li>
3195 <p>The remaining two arguments must be <a href="#t_integer">integer</a> or
3196 <a href="#t_pointer">pointer</a> typed. They must also be identical types.</p>
3198 <p>The '<tt>icmp</tt>' compares <tt>var1</tt> and <tt>var2</tt> according to
3199 the condition code given as <tt>cond</tt>. The comparison performed always
3200 yields a <a href="#t_primitive">i1</a> result, as follows:
3202 <li><tt>eq</tt>: yields <tt>true</tt> if the operands are equal,
3203 <tt>false</tt> otherwise. No sign interpretation is necessary or performed.
3205 <li><tt>ne</tt>: yields <tt>true</tt> if the operands are unequal,
3206 <tt>false</tt> otherwise. No sign interpretation is necessary or performed.
3207 <li><tt>ugt</tt>: interprets the operands as unsigned values and yields
3208 <tt>true</tt> if <tt>var1</tt> is greater than <tt>var2</tt>.</li>
3209 <li><tt>uge</tt>: interprets the operands as unsigned values and yields
3210 <tt>true</tt> if <tt>var1</tt> is greater than or equal to <tt>var2</tt>.</li>
3211 <li><tt>ult</tt>: interprets the operands as unsigned values and yields
3212 <tt>true</tt> if <tt>var1</tt> is less than <tt>var2</tt>.</li>
3213 <li><tt>ule</tt>: interprets the operands as unsigned values and yields
3214 <tt>true</tt> if <tt>var1</tt> is less than or equal to <tt>var2</tt>.</li>
3215 <li><tt>sgt</tt>: interprets the operands as signed values and yields
3216 <tt>true</tt> if <tt>var1</tt> is greater than <tt>var2</tt>.</li>
3217 <li><tt>sge</tt>: interprets the operands as signed values and yields
3218 <tt>true</tt> if <tt>var1</tt> is greater than or equal to <tt>var2</tt>.</li>
3219 <li><tt>slt</tt>: interprets the operands as signed values and yields
3220 <tt>true</tt> if <tt>var1</tt> is less than <tt>var2</tt>.</li>
3221 <li><tt>sle</tt>: interprets the operands as signed values and yields
3222 <tt>true</tt> if <tt>var1</tt> is less than or equal to <tt>var2</tt>.</li>
3224 <p>If the operands are <a href="#t_pointer">pointer</a> typed, the pointer
3225 values are treated as integers and then compared.</p>
3228 <pre> <result> = icmp eq i32 4, 5 <i>; yields: result=false</i>
3229 <result> = icmp ne float* %X, %X <i>; yields: result=false</i>
3230 <result> = icmp ult i16 4, 5 <i>; yields: result=true</i>
3231 <result> = icmp sgt i16 4, 5 <i>; yields: result=false</i>
3232 <result> = icmp ule i16 -4, 5 <i>; yields: result=false</i>
3233 <result> = icmp sge i16 4, 5 <i>; yields: result=false</i>
3237 <!-- _______________________________________________________________________ -->
3238 <div class="doc_subsubsection"><a name="i_fcmp">'<tt>fcmp</tt>' Instruction</a>
3240 <div class="doc_text">
3242 <pre> <result> = fcmp <cond> <ty> <var1>, <var2>
3243 <i>; yields {i1}:result</i>
3246 <p>The '<tt>fcmp</tt>' instruction returns a boolean value based on comparison
3247 of its floating point operands.</p>
3249 <p>The '<tt>fcmp</tt>' instruction takes three operands. The first operand is
3250 the condition code which indicates the kind of comparison to perform. It is not
3251 a value, just a keyword. The possibilities for the condition code are:
3253 <li><tt>false</tt>: no comparison, always returns false</li>
3254 <li><tt>oeq</tt>: ordered and equal</li>
3255 <li><tt>ogt</tt>: ordered and greater than </li>
3256 <li><tt>oge</tt>: ordered and greater than or equal</li>
3257 <li><tt>olt</tt>: ordered and less than </li>
3258 <li><tt>ole</tt>: ordered and less than or equal</li>
3259 <li><tt>one</tt>: ordered and not equal</li>
3260 <li><tt>ord</tt>: ordered (no nans)</li>
3261 <li><tt>ueq</tt>: unordered or equal</li>
3262 <li><tt>ugt</tt>: unordered or greater than </li>
3263 <li><tt>uge</tt>: unordered or greater than or equal</li>
3264 <li><tt>ult</tt>: unordered or less than </li>
3265 <li><tt>ule</tt>: unordered or less than or equal</li>
3266 <li><tt>une</tt>: unordered or not equal</li>
3267 <li><tt>uno</tt>: unordered (either nans)</li>
3268 <li><tt>true</tt>: no comparison, always returns true</li>
3270 <p>In the preceding, <i>ordered</i> means that neither operand is a QNAN while
3271 <i>unordered</i> means that either operand may be a QNAN.</p>
3272 <p>The <tt>val1</tt> and <tt>val2</tt> arguments must be
3273 <a href="#t_floating">floating point</a> typed. They must have identical
3275 <p>In the foregoing, <i>ordered</i> means that neither operand is a QNAN and
3276 <i>unordered</i> means that either operand is a QNAN.</p>
3278 <p>The '<tt>fcmp</tt>' compares <tt>var1</tt> and <tt>var2</tt> according to
3279 the condition code given as <tt>cond</tt>. The comparison performed always
3280 yields a <a href="#t_primitive">i1</a> result, as follows:
3282 <li><tt>false</tt>: always yields <tt>false</tt>, regardless of operands.</li>
3283 <li><tt>oeq</tt>: yields <tt>true</tt> if both operands are not a QNAN and
3284 <tt>var1</tt> is equal to <tt>var2</tt>.</li>
3285 <li><tt>ogt</tt>: yields <tt>true</tt> if both operands are not a QNAN and
3286 <tt>var1</tt> is greather than <tt>var2</tt>.</li>
3287 <li><tt>oge</tt>: yields <tt>true</tt> if both operands are not a QNAN and
3288 <tt>var1</tt> is greater than or equal to <tt>var2</tt>.</li>
3289 <li><tt>olt</tt>: yields <tt>true</tt> if both operands are not a QNAN and
3290 <tt>var1</tt> is less than <tt>var2</tt>.</li>
3291 <li><tt>ole</tt>: yields <tt>true</tt> if both operands are not a QNAN and
3292 <tt>var1</tt> is less than or equal to <tt>var2</tt>.</li>
3293 <li><tt>one</tt>: yields <tt>true</tt> if both operands are not a QNAN and
3294 <tt>var1</tt> is not equal to <tt>var2</tt>.</li>
3295 <li><tt>ord</tt>: yields <tt>true</tt> if both operands are not a QNAN.</li>
3296 <li><tt>ueq</tt>: yields <tt>true</tt> if either operand is a QNAN or
3297 <tt>var1</tt> is equal to <tt>var2</tt>.</li>
3298 <li><tt>ugt</tt>: yields <tt>true</tt> if either operand is a QNAN or
3299 <tt>var1</tt> is greater than <tt>var2</tt>.</li>
3300 <li><tt>uge</tt>: yields <tt>true</tt> if either operand is a QNAN or
3301 <tt>var1</tt> is greater than or equal to <tt>var2</tt>.</li>
3302 <li><tt>ult</tt>: yields <tt>true</tt> if either operand is a QNAN or
3303 <tt>var1</tt> is less than <tt>var2</tt>.</li>
3304 <li><tt>ule</tt>: yields <tt>true</tt> if either operand is a QNAN or
3305 <tt>var1</tt> is less than or equal to <tt>var2</tt>.</li>
3306 <li><tt>une</tt>: yields <tt>true</tt> if either operand is a QNAN or
3307 <tt>var1</tt> is not equal to <tt>var2</tt>.</li>
3308 <li><tt>uno</tt>: yields <tt>true</tt> if either operand is a QNAN.</li>
3309 <li><tt>true</tt>: always yields <tt>true</tt>, regardless of operands.</li>
3313 <pre> <result> = fcmp oeq float 4.0, 5.0 <i>; yields: result=false</i>
3314 <result> = icmp one float 4.0, 5.0 <i>; yields: result=true</i>
3315 <result> = icmp olt float 4.0, 5.0 <i>; yields: result=true</i>
3316 <result> = icmp ueq double 1.0, 2.0 <i>; yields: result=false</i>
3320 <!-- _______________________________________________________________________ -->
3321 <div class="doc_subsubsection"> <a name="i_phi">'<tt>phi</tt>'
3322 Instruction</a> </div>
3323 <div class="doc_text">
3325 <pre> <result> = phi <ty> [ <val0>, <label0>], ...<br></pre>
3327 <p>The '<tt>phi</tt>' instruction is used to implement the φ node in
3328 the SSA graph representing the function.</p>
3330 <p>The type of the incoming values are specified with the first type
3331 field. After this, the '<tt>phi</tt>' instruction takes a list of pairs
3332 as arguments, with one pair for each predecessor basic block of the
3333 current block. Only values of <a href="#t_firstclass">first class</a>
3334 type may be used as the value arguments to the PHI node. Only labels
3335 may be used as the label arguments.</p>
3336 <p>There must be no non-phi instructions between the start of a basic
3337 block and the PHI instructions: i.e. PHI instructions must be first in
3340 <p>At runtime, the '<tt>phi</tt>' instruction logically takes on the
3341 value specified by the parameter, depending on which basic block we
3342 came from in the last <a href="#terminators">terminator</a> instruction.</p>
3344 <pre>Loop: ; Infinite loop that counts from 0 on up...<br> %indvar = phi i32 [ 0, %LoopHeader ], [ %nextindvar, %Loop ]<br> %nextindvar = add i32 %indvar, 1<br> br label %Loop<br></pre>
3347 <!-- _______________________________________________________________________ -->
3348 <div class="doc_subsubsection">
3349 <a name="i_select">'<tt>select</tt>' Instruction</a>
3352 <div class="doc_text">
3357 <result> = select i1 <cond>, <ty> <val1>, <ty> <val2> <i>; yields ty</i>
3363 The '<tt>select</tt>' instruction is used to choose one value based on a
3364 condition, without branching.
3371 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.
3377 If the boolean condition evaluates to true, the instruction returns the first
3378 value argument; otherwise, it returns the second value argument.
3384 %X = select i1 true, i8 17, i8 42 <i>; yields i8:17</i>
3389 <!-- _______________________________________________________________________ -->
3390 <div class="doc_subsubsection">
3391 <a name="i_call">'<tt>call</tt>' Instruction</a>
3394 <div class="doc_text">
3398 <result> = [tail] call [<a href="#callingconv">cconv</a>] <ty>* <fnptrval>(<param list>)
3403 <p>The '<tt>call</tt>' instruction represents a simple function call.</p>
3407 <p>This instruction requires several arguments:</p>
3411 <p>The optional "tail" marker indicates whether the callee function accesses
3412 any allocas or varargs in the caller. If the "tail" marker is present, the
3413 function call is eligible for tail call optimization. Note that calls may
3414 be marked "tail" even if they do not occur before a <a
3415 href="#i_ret"><tt>ret</tt></a> instruction.
3418 <p>The optional "cconv" marker indicates which <a href="callingconv">calling
3419 convention</a> the call should use. If none is specified, the call defaults
3420 to using C calling conventions.
3423 <p>'<tt>ty</tt>': shall be the signature of the pointer to function value
3424 being invoked. The argument types must match the types implied by this
3425 signature. This type can be omitted if the function is not varargs and
3426 if the function type does not return a pointer to a function.</p>
3429 <p>'<tt>fnptrval</tt>': An LLVM value containing a pointer to a function to
3430 be invoked. In most cases, this is a direct function invocation, but
3431 indirect <tt>call</tt>s are just as possible, calling an arbitrary pointer
3432 to function value.</p>
3435 <p>'<tt>function args</tt>': argument list whose types match the
3436 function signature argument types. All arguments must be of
3437 <a href="#t_firstclass">first class</a> type. If the function signature
3438 indicates the function accepts a variable number of arguments, the extra
3439 arguments can be specified.</p>
3445 <p>The '<tt>call</tt>' instruction is used to cause control flow to
3446 transfer to a specified function, with its incoming arguments bound to
3447 the specified values. Upon a '<tt><a href="#i_ret">ret</a></tt>'
3448 instruction in the called function, control flow continues with the
3449 instruction after the function call, and the return value of the
3450 function is bound to the result argument. This is a simpler case of
3451 the <a href="#i_invoke">invoke</a> instruction.</p>
3456 %retval = call i32 %test(i32 %argc)
3457 call i32(i8 *, ...) *%printf(i8 * %msg, i32 12, i8 42);
3458 %X = tail call i32 %foo()
3459 %Y = tail call <a href="#callingconv">fastcc</a> i32 %foo()
3464 <!-- _______________________________________________________________________ -->
3465 <div class="doc_subsubsection">
3466 <a name="i_va_arg">'<tt>va_arg</tt>' Instruction</a>
3469 <div class="doc_text">
3474 <resultval> = va_arg <va_list*> <arglist>, <argty>
3479 <p>The '<tt>va_arg</tt>' instruction is used to access arguments passed through
3480 the "variable argument" area of a function call. It is used to implement the
3481 <tt>va_arg</tt> macro in C.</p>
3485 <p>This instruction takes a <tt>va_list*</tt> value and the type of
3486 the argument. It returns a value of the specified argument type and
3487 increments the <tt>va_list</tt> to point to the next argument. Again, the
3488 actual type of <tt>va_list</tt> is target specific.</p>
3492 <p>The '<tt>va_arg</tt>' instruction loads an argument of the specified
3493 type from the specified <tt>va_list</tt> and causes the
3494 <tt>va_list</tt> to point to the next argument. For more information,
3495 see the variable argument handling <a href="#int_varargs">Intrinsic
3498 <p>It is legal for this instruction to be called in a function which does not
3499 take a variable number of arguments, for example, the <tt>vfprintf</tt>
3502 <p><tt>va_arg</tt> is an LLVM instruction instead of an <a
3503 href="#intrinsics">intrinsic function</a> because it takes a type as an
3508 <p>See the <a href="#int_varargs">variable argument processing</a> section.</p>
3512 <!-- *********************************************************************** -->
3513 <div class="doc_section"> <a name="intrinsics">Intrinsic Functions</a> </div>
3514 <!-- *********************************************************************** -->
3516 <div class="doc_text">
3518 <p>LLVM supports the notion of an "intrinsic function". These functions have
3519 well known names and semantics and are required to follow certain
3520 restrictions. Overall, these instructions represent an extension mechanism for
3521 the LLVM language that does not require changing all of the transformations in
3522 LLVM to add to the language (or the bytecode reader/writer, the parser,
3525 <p>Intrinsic function names must all start with an "<tt>llvm.</tt>" prefix. This
3526 prefix is reserved in LLVM for intrinsic names; thus, functions may not be named
3527 this. Intrinsic functions must always be external functions: you cannot define
3528 the body of intrinsic functions. Intrinsic functions may only be used in call
3529 or invoke instructions: it is illegal to take the address of an intrinsic
3530 function. Additionally, because intrinsic functions are part of the LLVM
3531 language, it is required that they all be documented here if any are added.</p>
3534 <p>To learn how to add an intrinsic function, please see the <a
3535 href="ExtendingLLVM.html">Extending LLVM Guide</a>.
3540 <!-- ======================================================================= -->
3541 <div class="doc_subsection">
3542 <a name="int_varargs">Variable Argument Handling Intrinsics</a>
3545 <div class="doc_text">
3547 <p>Variable argument support is defined in LLVM with the <a
3548 href="#i_va_arg"><tt>va_arg</tt></a> instruction and these three
3549 intrinsic functions. These functions are related to the similarly
3550 named macros defined in the <tt><stdarg.h></tt> header file.</p>
3552 <p>All of these functions operate on arguments that use a
3553 target-specific value type "<tt>va_list</tt>". The LLVM assembly
3554 language reference manual does not define what this type is, so all
3555 transformations should be prepared to handle intrinsics with any type
3558 <p>This example shows how the <a href="#i_va_arg"><tt>va_arg</tt></a>
3559 instruction and the variable argument handling intrinsic functions are
3563 define i32 %test(i32 %X, ...) {
3564 ; Initialize variable argument processing
3566 %ap2 = bitcast i8** %ap to i8*
3567 call void %<a href="#i_va_start">llvm.va_start</a>(i8* %ap2)
3569 ; Read a single integer argument
3570 %tmp = va_arg i8 ** %ap, i32
3572 ; Demonstrate usage of llvm.va_copy and llvm.va_end
3574 %aq2 = bitcast i8** %aq to i8*
3575 call void %<a href="#i_va_copy">llvm.va_copy</a>(i8 *%aq2, i8* %ap2)
3576 call void %<a href="#i_va_end">llvm.va_end</a>(i8* %aq2)
3578 ; Stop processing of arguments.
3579 call void %<a href="#i_va_end">llvm.va_end</a>(i8* %ap2)
3585 <!-- _______________________________________________________________________ -->
3586 <div class="doc_subsubsection">
3587 <a name="i_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a>
3591 <div class="doc_text">
3593 <pre> declare void %llvm.va_start(i8* <arglist>)<br></pre>
3595 <P>The '<tt>llvm.va_start</tt>' intrinsic initializes
3596 <tt>*<arglist></tt> for subsequent use by <tt><a
3597 href="#i_va_arg">va_arg</a></tt>.</p>
3601 <P>The argument is a pointer to a <tt>va_list</tt> element to initialize.</p>
3605 <P>The '<tt>llvm.va_start</tt>' intrinsic works just like the <tt>va_start</tt>
3606 macro available in C. In a target-dependent way, it initializes the
3607 <tt>va_list</tt> element the argument points to, so that the next call to
3608 <tt>va_arg</tt> will produce the first variable argument passed to the function.
3609 Unlike the C <tt>va_start</tt> macro, this intrinsic does not need to know the
3610 last argument of the function, the compiler can figure that out.</p>
3614 <!-- _______________________________________________________________________ -->
3615 <div class="doc_subsubsection">
3616 <a name="i_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a>
3619 <div class="doc_text">
3621 <pre> declare void %llvm.va_end(i8* <arglist>)<br></pre>
3624 <p>The '<tt>llvm.va_end</tt>' intrinsic destroys <tt><arglist></tt>
3625 which has been initialized previously with <tt><a href="#i_va_start">llvm.va_start</a></tt>
3626 or <tt><a href="#i_va_copy">llvm.va_copy</a></tt>.</p>
3630 <p>The argument is a <tt>va_list</tt> to destroy.</p>
3634 <p>The '<tt>llvm.va_end</tt>' intrinsic works just like the <tt>va_end</tt>
3635 macro available in C. In a target-dependent way, it destroys the <tt>va_list</tt>.
3636 Calls to <a href="#i_va_start"><tt>llvm.va_start</tt></a> and <a
3637 href="#i_va_copy"><tt>llvm.va_copy</tt></a> must be matched exactly
3638 with calls to <tt>llvm.va_end</tt>.</p>
3642 <!-- _______________________________________________________________________ -->
3643 <div class="doc_subsubsection">
3644 <a name="i_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a>
3647 <div class="doc_text">
3652 declare void %llvm.va_copy(i8* <destarglist>, i8* <srcarglist>)
3657 <p>The '<tt>llvm.va_copy</tt>' intrinsic copies the current argument position from
3658 the source argument list to the destination argument list.</p>
3662 <p>The first argument is a pointer to a <tt>va_list</tt> element to initialize.
3663 The second argument is a pointer to a <tt>va_list</tt> element to copy from.</p>
3668 <p>The '<tt>llvm.va_copy</tt>' intrinsic works just like the <tt>va_copy</tt> macro
3669 available in C. In a target-dependent way, it copies the source
3670 <tt>va_list</tt> element into the destination list. This intrinsic is necessary
3671 because the <tt><a href="i_va_begin">llvm.va_begin</a></tt> intrinsic may be
3672 arbitrarily complex and require memory allocation, for example.</p>
3676 <!-- ======================================================================= -->
3677 <div class="doc_subsection">
3678 <a name="int_gc">Accurate Garbage Collection Intrinsics</a>
3681 <div class="doc_text">
3684 LLVM support for <a href="GarbageCollection.html">Accurate Garbage
3685 Collection</a> requires the implementation and generation of these intrinsics.
3686 These intrinsics allow identification of <a href="#i_gcroot">GC roots on the
3687 stack</a>, as well as garbage collector implementations that require <a
3688 href="#i_gcread">read</a> and <a href="#i_gcwrite">write</a> barriers.
3689 Front-ends for type-safe garbage collected languages should generate these
3690 intrinsics to make use of the LLVM garbage collectors. For more details, see <a
3691 href="GarbageCollection.html">Accurate Garbage Collection with LLVM</a>.
3695 <!-- _______________________________________________________________________ -->
3696 <div class="doc_subsubsection">
3697 <a name="i_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a>
3700 <div class="doc_text">
3705 declare void %llvm.gcroot(<ty>** %ptrloc, <ty2>* %metadata)
3710 <p>The '<tt>llvm.gcroot</tt>' intrinsic declares the existence of a GC root to
3711 the code generator, and allows some metadata to be associated with it.</p>
3715 <p>The first argument specifies the address of a stack object that contains the
3716 root pointer. The second pointer (which must be either a constant or a global
3717 value address) contains the meta-data to be associated with the root.</p>
3721 <p>At runtime, a call to this intrinsics stores a null pointer into the "ptrloc"
3722 location. At compile-time, the code generator generates information to allow
3723 the runtime to find the pointer at GC safe points.
3729 <!-- _______________________________________________________________________ -->
3730 <div class="doc_subsubsection">
3731 <a name="i_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a>
3734 <div class="doc_text">
3739 declare i8 * %llvm.gcread(i8 * %ObjPtr, i8 ** %Ptr)
3744 <p>The '<tt>llvm.gcread</tt>' intrinsic identifies reads of references from heap
3745 locations, allowing garbage collector implementations that require read
3750 <p>The second argument is the address to read from, which should be an address
3751 allocated from the garbage collector. The first object is a pointer to the
3752 start of the referenced object, if needed by the language runtime (otherwise
3757 <p>The '<tt>llvm.gcread</tt>' intrinsic has the same semantics as a load
3758 instruction, but may be replaced with substantially more complex code by the
3759 garbage collector runtime, as needed.</p>
3764 <!-- _______________________________________________________________________ -->
3765 <div class="doc_subsubsection">
3766 <a name="i_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a>
3769 <div class="doc_text">
3774 declare void %llvm.gcwrite(i8 * %P1, i8 * %Obj, i8 ** %P2)
3779 <p>The '<tt>llvm.gcwrite</tt>' intrinsic identifies writes of references to heap
3780 locations, allowing garbage collector implementations that require write
3781 barriers (such as generational or reference counting collectors).</p>
3785 <p>The first argument is the reference to store, the second is the start of the
3786 object to store it to, and the third is the address of the field of Obj to
3787 store to. If the runtime does not require a pointer to the object, Obj may be
3792 <p>The '<tt>llvm.gcwrite</tt>' intrinsic has the same semantics as a store
3793 instruction, but may be replaced with substantially more complex code by the
3794 garbage collector runtime, as needed.</p>
3800 <!-- ======================================================================= -->
3801 <div class="doc_subsection">
3802 <a name="int_codegen">Code Generator Intrinsics</a>
3805 <div class="doc_text">
3807 These intrinsics are provided by LLVM to expose special features that may only
3808 be implemented with code generator support.
3813 <!-- _______________________________________________________________________ -->
3814 <div class="doc_subsubsection">
3815 <a name="i_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a>
3818 <div class="doc_text">
3822 declare i8 *%llvm.returnaddress(i32 <level>)
3828 The '<tt>llvm.returnaddress</tt>' intrinsic attempts to compute a
3829 target-specific value indicating the return address of the current function
3830 or one of its callers.
3836 The argument to this intrinsic indicates which function to return the address
3837 for. Zero indicates the calling function, one indicates its caller, etc. The
3838 argument is <b>required</b> to be a constant integer value.
3844 The '<tt>llvm.returnaddress</tt>' intrinsic either returns a pointer indicating
3845 the return address of the specified call frame, or zero if it cannot be
3846 identified. The value returned by this intrinsic is likely to be incorrect or 0
3847 for arguments other than zero, so it should only be used for debugging purposes.
3851 Note that calling this intrinsic does not prevent function inlining or other
3852 aggressive transformations, so the value returned may not be that of the obvious
3853 source-language caller.
3858 <!-- _______________________________________________________________________ -->
3859 <div class="doc_subsubsection">
3860 <a name="i_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a>
3863 <div class="doc_text">
3867 declare i8 *%llvm.frameaddress(i32 <level>)
3873 The '<tt>llvm.frameaddress</tt>' intrinsic attempts to return the
3874 target-specific frame pointer value for the specified stack frame.
3880 The argument to this intrinsic indicates which function to return the frame
3881 pointer for. Zero indicates the calling function, one indicates its caller,
3882 etc. The argument is <b>required</b> to be a constant integer value.
3888 The '<tt>llvm.frameaddress</tt>' intrinsic either returns a pointer indicating
3889 the frame address of the specified call frame, or zero if it cannot be
3890 identified. The value returned by this intrinsic is likely to be incorrect or 0
3891 for arguments other than zero, so it should only be used for debugging purposes.
3895 Note that calling this intrinsic does not prevent function inlining or other
3896 aggressive transformations, so the value returned may not be that of the obvious
3897 source-language caller.
3901 <!-- _______________________________________________________________________ -->
3902 <div class="doc_subsubsection">
3903 <a name="i_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a>
3906 <div class="doc_text">
3910 declare i8 *%llvm.stacksave()
3916 The '<tt>llvm.stacksave</tt>' intrinsic is used to remember the current state of
3917 the function stack, for use with <a href="#i_stackrestore">
3918 <tt>llvm.stackrestore</tt></a>. This is useful for implementing language
3919 features like scoped automatic variable sized arrays in C99.
3925 This intrinsic returns a opaque pointer value that can be passed to <a
3926 href="#i_stackrestore"><tt>llvm.stackrestore</tt></a>. When an
3927 <tt>llvm.stackrestore</tt> intrinsic is executed with a value saved from
3928 <tt>llvm.stacksave</tt>, it effectively restores the state of the stack to the
3929 state it was in when the <tt>llvm.stacksave</tt> intrinsic executed. In
3930 practice, this pops any <a href="#i_alloca">alloca</a> blocks from the stack
3931 that were allocated after the <tt>llvm.stacksave</tt> was executed.
3936 <!-- _______________________________________________________________________ -->
3937 <div class="doc_subsubsection">
3938 <a name="i_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a>
3941 <div class="doc_text">
3945 declare void %llvm.stackrestore(i8 * %ptr)
3951 The '<tt>llvm.stackrestore</tt>' intrinsic is used to restore the state of
3952 the function stack to the state it was in when the corresponding <a
3953 href="#llvm.stacksave"><tt>llvm.stacksave</tt></a> intrinsic executed. This is
3954 useful for implementing language features like scoped automatic variable sized
3961 See the description for <a href="#i_stacksave"><tt>llvm.stacksave</tt></a>.
3967 <!-- _______________________________________________________________________ -->
3968 <div class="doc_subsubsection">
3969 <a name="i_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a>
3972 <div class="doc_text">
3976 declare void %llvm.prefetch(i8 * <address>,
3977 i32 <rw>, i32 <locality>)
3984 The '<tt>llvm.prefetch</tt>' intrinsic is a hint to the code generator to insert
3985 a prefetch instruction if supported; otherwise, it is a noop. Prefetches have
3987 effect on the behavior of the program but can change its performance
3994 <tt>address</tt> is the address to be prefetched, <tt>rw</tt> is the specifier
3995 determining if the fetch should be for a read (0) or write (1), and
3996 <tt>locality</tt> is a temporal locality specifier ranging from (0) - no
3997 locality, to (3) - extremely local keep in cache. The <tt>rw</tt> and
3998 <tt>locality</tt> arguments must be constant integers.
4004 This intrinsic does not modify the behavior of the program. In particular,
4005 prefetches cannot trap and do not produce a value. On targets that support this
4006 intrinsic, the prefetch can provide hints to the processor cache for better
4012 <!-- _______________________________________________________________________ -->
4013 <div class="doc_subsubsection">
4014 <a name="i_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a>
4017 <div class="doc_text">
4021 declare void %llvm.pcmarker( i32 <id> )
4028 The '<tt>llvm.pcmarker</tt>' intrinsic is a method to export a Program Counter
4030 code to simulators and other tools. The method is target specific, but it is
4031 expected that the marker will use exported symbols to transmit the PC of the marker.
4032 The marker makes no guarantees that it will remain with any specific instruction
4033 after optimizations. It is possible that the presence of a marker will inhibit
4034 optimizations. The intended use is to be inserted after optimizations to allow
4035 correlations of simulation runs.
4041 <tt>id</tt> is a numerical id identifying the marker.
4047 This intrinsic does not modify the behavior of the program. Backends that do not
4048 support this intrinisic may ignore it.
4053 <!-- _______________________________________________________________________ -->
4054 <div class="doc_subsubsection">
4055 <a name="i_readcyclecounter">'<tt>llvm.readcyclecounter</tt>' Intrinsic</a>
4058 <div class="doc_text">
4062 declare i64 %llvm.readcyclecounter( )
4069 The '<tt>llvm.readcyclecounter</tt>' intrinsic provides access to the cycle
4070 counter register (or similar low latency, high accuracy clocks) on those targets
4071 that support it. On X86, it should map to RDTSC. On Alpha, it should map to RPCC.
4072 As the backing counters overflow quickly (on the order of 9 seconds on alpha), this
4073 should only be used for small timings.
4079 When directly supported, reading the cycle counter should not modify any memory.
4080 Implementations are allowed to either return a application specific value or a
4081 system wide value. On backends without support, this is lowered to a constant 0.
4086 <!-- ======================================================================= -->
4087 <div class="doc_subsection">
4088 <a name="int_libc">Standard C Library Intrinsics</a>
4091 <div class="doc_text">
4093 LLVM provides intrinsics for a few important standard C library functions.
4094 These intrinsics allow source-language front-ends to pass information about the
4095 alignment of the pointer arguments to the code generator, providing opportunity
4096 for more efficient code generation.
4101 <!-- _______________________________________________________________________ -->
4102 <div class="doc_subsubsection">
4103 <a name="i_memcpy">'<tt>llvm.memcpy</tt>' Intrinsic</a>
4106 <div class="doc_text">
4110 declare void %llvm.memcpy.i32(i8 * <dest>, i8 * <src>,
4111 i32 <len>, i32 <align>)
4112 declare void %llvm.memcpy.i64(i8 * <dest>, i8 * <src>,
4113 i64 <len>, i32 <align>)
4119 The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the source
4120 location to the destination location.
4124 Note that, unlike the standard libc function, the <tt>llvm.memcpy.*</tt>
4125 intrinsics do not return a value, and takes an extra alignment argument.
4131 The first argument is a pointer to the destination, the second is a pointer to
4132 the source. The third argument is an integer argument
4133 specifying the number of bytes to copy, and the fourth argument is the alignment
4134 of the source and destination locations.
4138 If the call to this intrinisic has an alignment value that is not 0 or 1, then
4139 the caller guarantees that both the source and destination pointers are aligned
4146 The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the source
4147 location to the destination location, which are not allowed to overlap. It
4148 copies "len" bytes of memory over. If the argument is known to be aligned to
4149 some boundary, this can be specified as the fourth argument, otherwise it should
4155 <!-- _______________________________________________________________________ -->
4156 <div class="doc_subsubsection">
4157 <a name="i_memmove">'<tt>llvm.memmove</tt>' Intrinsic</a>
4160 <div class="doc_text">
4164 declare void %llvm.memmove.i32(i8 * <dest>, i8 * <src>,
4165 i32 <len>, i32 <align>)
4166 declare void %llvm.memmove.i64(i8 * <dest>, i8 * <src>,
4167 i64 <len>, i32 <align>)
4173 The '<tt>llvm.memmove.*</tt>' intrinsics move a block of memory from the source
4174 location to the destination location. It is similar to the
4175 '<tt>llvm.memcmp</tt>' intrinsic but allows the two memory locations to overlap.
4179 Note that, unlike the standard libc function, the <tt>llvm.memmove.*</tt>
4180 intrinsics do not return a value, and takes an extra alignment argument.
4186 The first argument is a pointer to the destination, the second is a pointer to
4187 the source. The third argument is an integer argument
4188 specifying the number of bytes to copy, and the fourth argument is the alignment
4189 of the source and destination locations.
4193 If the call to this intrinisic has an alignment value that is not 0 or 1, then
4194 the caller guarantees that the source and destination pointers are aligned to
4201 The '<tt>llvm.memmove.*</tt>' intrinsics copy a block of memory from the source
4202 location to the destination location, which may overlap. It
4203 copies "len" bytes of memory over. If the argument is known to be aligned to
4204 some boundary, this can be specified as the fourth argument, otherwise it should
4210 <!-- _______________________________________________________________________ -->
4211 <div class="doc_subsubsection">
4212 <a name="i_memset">'<tt>llvm.memset.*</tt>' Intrinsics</a>
4215 <div class="doc_text">
4219 declare void %llvm.memset.i32(i8 * <dest>, i8 <val>,
4220 i32 <len>, i32 <align>)
4221 declare void %llvm.memset.i64(i8 * <dest>, i8 <val>,
4222 i64 <len>, i32 <align>)
4228 The '<tt>llvm.memset.*</tt>' intrinsics fill a block of memory with a particular
4233 Note that, unlike the standard libc function, the <tt>llvm.memset</tt> intrinsic
4234 does not return a value, and takes an extra alignment argument.
4240 The first argument is a pointer to the destination to fill, the second is the
4241 byte value to fill it with, the third argument is an integer
4242 argument specifying the number of bytes to fill, and the fourth argument is the
4243 known alignment of destination location.
4247 If the call to this intrinisic has an alignment value that is not 0 or 1, then
4248 the caller guarantees that the destination pointer is aligned to that boundary.
4254 The '<tt>llvm.memset.*</tt>' intrinsics fill "len" bytes of memory starting at
4256 destination location. If the argument is known to be aligned to some boundary,
4257 this can be specified as the fourth argument, otherwise it should be set to 0 or
4263 <!-- _______________________________________________________________________ -->
4264 <div class="doc_subsubsection">
4265 <a name="i_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a>
4268 <div class="doc_text">
4272 declare float %llvm.sqrt.f32(float %Val)
4273 declare double %llvm.sqrt.f64(double %Val)
4279 The '<tt>llvm.sqrt</tt>' intrinsics return the sqrt of the specified operand,
4280 returning the same value as the libm '<tt>sqrt</tt>' function would. Unlike
4281 <tt>sqrt</tt> in libm, however, <tt>llvm.sqrt</tt> has undefined behavior for
4282 negative numbers (which allows for better optimization).
4288 The argument and return value are floating point numbers of the same type.
4294 This function returns the sqrt of the specified operand if it is a positive
4295 floating point number.
4299 <!-- _______________________________________________________________________ -->
4300 <div class="doc_subsubsection">
4301 <a name="i_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a>
4304 <div class="doc_text">
4308 declare float %llvm.powi.f32(float %Val, i32 %power)
4309 declare double %llvm.powi.f64(double %Val, i32 %power)
4315 The '<tt>llvm.powi.*</tt>' intrinsics return the first operand raised to the
4316 specified (positive or negative) power. The order of evaluation of
4317 multiplications is not defined.
4323 The second argument is an integer power, and the first is a value to raise to
4330 This function returns the first value raised to the second power with an
4331 unspecified sequence of rounding operations.</p>
4335 <!-- ======================================================================= -->
4336 <div class="doc_subsection">
4337 <a name="int_manip">Bit Manipulation Intrinsics</a>
4340 <div class="doc_text">
4342 LLVM provides intrinsics for a few important bit manipulation operations.
4343 These allow efficient code generation for some algorithms.
4348 <!-- _______________________________________________________________________ -->
4349 <div class="doc_subsubsection">
4350 <a name="i_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a>
4353 <div class="doc_text">
4357 declare i16 %llvm.bswap.i16(i16 <id>)
4358 declare i32 %llvm.bswap.i32(i32 <id>)
4359 declare i64 %llvm.bswap.i64(i64 <id>)
4365 The '<tt>llvm.bwsap</tt>' family of intrinsics is used to byteswap a 16, 32 or
4366 64 bit quantity. These are useful for performing operations on data that is not
4367 in the target's native byte order.
4373 The <tt>llvm.bswap.16</tt> intrinsic returns an i16 value that has the high
4374 and low byte of the input i16 swapped. Similarly, the <tt>llvm.bswap.i32</tt>
4375 intrinsic returns an i32 value that has the four bytes of the input i32
4376 swapped, so that if the input bytes are numbered 0, 1, 2, 3 then the returned
4377 i32 will have its bytes in 3, 2, 1, 0 order. The <tt>llvm.bswap.i64</tt>
4378 intrinsic extends this concept to 64 bits.
4383 <!-- _______________________________________________________________________ -->
4384 <div class="doc_subsubsection">
4385 <a name="int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic</a>
4388 <div class="doc_text">
4392 declare i8 %llvm.ctpop.i8 (i8 <src>)
4393 declare i16 %llvm.ctpop.i16(i16 <src>)
4394 declare i32 %llvm.ctpop.i32(i32 <src>)
4395 declare i64 %llvm.ctpop.i64(i64 <src>)
4401 The '<tt>llvm.ctpop</tt>' family of intrinsics counts the number of bits set in a
4408 The only argument is the value to be counted. The argument may be of any
4409 integer type. The return type must match the argument type.
4415 The '<tt>llvm.ctpop</tt>' intrinsic counts the 1's in a variable.
4419 <!-- _______________________________________________________________________ -->
4420 <div class="doc_subsubsection">
4421 <a name="int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic</a>
4424 <div class="doc_text">
4428 declare i8 %llvm.ctlz.i8 (i8 <src>)
4429 declare i16 %llvm.ctlz.i16(i16 <src>)
4430 declare i32 %llvm.ctlz.i32(i32 <src>)
4431 declare i64 %llvm.ctlz.i64(i64 <src>)
4437 The '<tt>llvm.ctlz</tt>' family of intrinsic functions counts the number of
4438 leading zeros in a variable.
4444 The only argument is the value to be counted. The argument may be of any
4445 integer type. The return type must match the argument type.
4451 The '<tt>llvm.ctlz</tt>' intrinsic counts the leading (most significant) zeros
4452 in a variable. If the src == 0 then the result is the size in bits of the type
4453 of src. For example, <tt>llvm.ctlz(i32 2) = 30</tt>.
4459 <!-- _______________________________________________________________________ -->
4460 <div class="doc_subsubsection">
4461 <a name="int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic</a>
4464 <div class="doc_text">
4468 declare i8 %llvm.cttz.i8 (i8 <src>)
4469 declare i16 %llvm.cttz.i16(i16 <src>)
4470 declare i32 %llvm.cttz.i32(i32 <src>)
4471 declare i64 %llvm.cttz.i64(i64 <src>)
4477 The '<tt>llvm.cttz</tt>' family of intrinsic functions counts the number of
4484 The only argument is the value to be counted. The argument may be of any
4485 integer type. The return type must match the argument type.
4491 The '<tt>llvm.cttz</tt>' intrinsic counts the trailing (least significant) zeros
4492 in a variable. If the src == 0 then the result is the size in bits of the type
4493 of src. For example, <tt>llvm.cttz(2) = 1</tt>.
4497 <!-- ======================================================================= -->
4498 <div class="doc_subsection">
4499 <a name="int_debugger">Debugger Intrinsics</a>
4502 <div class="doc_text">
4504 The LLVM debugger intrinsics (which all start with <tt>llvm.dbg.</tt> prefix),
4505 are described in the <a
4506 href="SourceLevelDebugging.html#format_common_intrinsics">LLVM Source Level
4507 Debugging</a> document.
4512 <!-- *********************************************************************** -->
4515 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
4516 src="http://jigsaw.w3.org/css-validator/images/vcss" alt="Valid CSS!"></a>
4517 <a href="http://validator.w3.org/check/referer"><img
4518 src="http://www.w3.org/Icons/valid-html401" alt="Valid HTML 4.01!" /></a>
4520 <a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
4521 <a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br>
4522 Last modified: $Date$