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>
88 <li><a href="#i_shl">'<tt>shl</tt>' Instruction</a></li>
89 <li><a href="#i_lshr">'<tt>lshr</tt>' Instruction</a></li>
90 <li><a href="#i_ashr">'<tt>ashr</tt>' Instruction</a></li>
93 <li><a href="#bitwiseops">Bitwise Binary Operations</a>
95 <li><a href="#i_and">'<tt>and</tt>' Instruction</a></li>
96 <li><a href="#i_or">'<tt>or</tt>' Instruction</a></li>
97 <li><a href="#i_xor">'<tt>xor</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>fastcc</tt>" - The fast calling convention</b>:</dt>
535 <dd>This calling convention attempts to make calls as fast as possible
536 (e.g. by passing things in registers). This calling convention allows the
537 target to use whatever tricks it wants to produce fast code for the target,
538 without having to conform to an externally specified ABI. Implementations of
539 this convention should allow arbitrary tail call optimization to be supported.
540 This calling convention does not support varargs and requires the prototype of
541 all callees to exactly match the prototype of the function definition.
544 <dt><b>"<tt>coldcc</tt>" - The cold calling convention</b>:</dt>
546 <dd>This calling convention attempts to make code in the caller as efficient
547 as possible under the assumption that the call is not commonly executed. As
548 such, these calls often preserve all registers so that the call does not break
549 any live ranges in the caller side. This calling convention does not support
550 varargs and requires the prototype of all callees to exactly match the
551 prototype of the function definition.
554 <dt><b>"<tt>cc <<em>n</em>></tt>" - Numbered convention</b>:</dt>
556 <dd>Any calling convention may be specified by number, allowing
557 target-specific calling conventions to be used. Target specific calling
558 conventions start at 64.
562 <p>More calling conventions can be added/defined on an as-needed basis, to
563 support pascal conventions or any other well-known target-independent
568 <!-- ======================================================================= -->
569 <div class="doc_subsection">
570 <a name="visibility">Visibility Styles</a>
573 <div class="doc_text">
576 All Global Variables and Functions have one of the following visibility styles:
580 <dt><b>"<tt>default</tt>" - Default style</b>:</dt>
582 <dd>On ELF, default visibility means that the declaration is visible to other
583 modules and, in shared libraries, means that the declared entity may be
584 overridden. On Darwin, default visibility means that the declaration is
585 visible to other modules. Default visibility corresponds to "external
586 linkage" in the language.
589 <dt><b>"<tt>hidden</tt>" - Hidden style</b>:</dt>
591 <dd>Two declarations of an object with hidden visibility refer to the same
592 object if they are in the same shared object. Usually, hidden visibility
593 indicates that the symbol will not be placed into the dynamic symbol table,
594 so no other module (executable or shared library) can reference it
602 <!-- ======================================================================= -->
603 <div class="doc_subsection">
604 <a name="globalvars">Global Variables</a>
607 <div class="doc_text">
609 <p>Global variables define regions of memory allocated at compilation time
610 instead of run-time. Global variables may optionally be initialized, may have
611 an explicit section to be placed in, and may
612 have an optional explicit alignment specified. A
613 variable may be defined as a global "constant," which indicates that the
614 contents of the variable will <b>never</b> be modified (enabling better
615 optimization, allowing the global data to be placed in the read-only section of
616 an executable, etc). Note that variables that need runtime initialization
617 cannot be marked "constant" as there is a store to the variable.</p>
620 LLVM explicitly allows <em>declarations</em> of global variables to be marked
621 constant, even if the final definition of the global is not. This capability
622 can be used to enable slightly better optimization of the program, but requires
623 the language definition to guarantee that optimizations based on the
624 'constantness' are valid for the translation units that do not include the
628 <p>As SSA values, global variables define pointer values that are in
629 scope (i.e. they dominate) all basic blocks in the program. Global
630 variables always define a pointer to their "content" type because they
631 describe a region of memory, and all memory objects in LLVM are
632 accessed through pointers.</p>
634 <p>LLVM allows an explicit section to be specified for globals. If the target
635 supports it, it will emit globals to the section specified.</p>
637 <p>An explicit alignment may be specified for a global. If not present, or if
638 the alignment is set to zero, the alignment of the global is set by the target
639 to whatever it feels convenient. If an explicit alignment is specified, the
640 global is forced to have at least that much alignment. All alignments must be
643 <p>For example, the following defines a global with an initializer, section,
647 %G = constant float 1.0, section "foo", align 4
653 <!-- ======================================================================= -->
654 <div class="doc_subsection">
655 <a name="functionstructure">Functions</a>
658 <div class="doc_text">
660 <p>LLVM function definitions consist of the "<tt>define</tt>" keyord,
661 an optional <a href="#linkage">linkage type</a>, an optional
662 <a href="#visibility">visibility style</a>, an optional
663 <a href="#callingconv">calling convention</a>, a return type, an optional
664 <a href="#paramattrs">parameter attribute</a> for the return type, a function
665 name, a (possibly empty) argument list (each with optional
666 <a href="#paramattrs">parameter attributes</a>), an optional section, an
667 optional alignment, an opening curly brace, a list of basic blocks, and a
670 LLVM function declarations consist of the "<tt>declare</tt>" keyword, an
671 optional <a href="#linkage">linkage type</a>, an optional
672 <a href="#visibility">visibility style</a>, an optional
673 <a href="#callingconv">calling convention</a>, a return type, an optional
674 <a href="#paramattrs">parameter attribute</a> for the return type, a function
675 name, a possibly empty list of arguments, and an optional alignment.</p>
677 <p>A function definition contains a list of basic blocks, forming the CFG for
678 the function. Each basic block may optionally start with a label (giving the
679 basic block a symbol table entry), contains a list of instructions, and ends
680 with a <a href="#terminators">terminator</a> instruction (such as a branch or
681 function return).</p>
683 <p>The first basic block in a program is special in two ways: it is immediately
684 executed on entrance to the function, and it is not allowed to have predecessor
685 basic blocks (i.e. there can not be any branches to the entry block of a
686 function). Because the block can have no predecessors, it also cannot have any
687 <a href="#i_phi">PHI nodes</a>.</p>
689 <p>LLVM functions are identified by their name and type signature. Hence, two
690 functions with the same name but different parameter lists or return values are
691 considered different functions, and LLVM will resolve references to each
694 <p>LLVM allows an explicit section to be specified for functions. If the target
695 supports it, it will emit functions to the section specified.</p>
697 <p>An explicit alignment may be specified for a function. If not present, or if
698 the alignment is set to zero, the alignment of the function is set by the target
699 to whatever it feels convenient. If an explicit alignment is specified, the
700 function is forced to have at least that much alignment. All alignments must be
705 <!-- ======================================================================= -->
706 <div class="doc_subsection"><a name="paramattrs">Parameter Attributes</a></div>
707 <div class="doc_text">
708 <p>The return type and each parameter of a function type may have a set of
709 <i>parameter attributes</i> associated with them. Parameter attributes are
710 used to communicate additional information about the result or parameters of
711 a function. Parameter attributes are considered to be part of the function
712 type so two functions types that differ only by the parameter attributes
713 are different function types.</p>
715 <p>Parameter attributes are simple keywords that follow the type specified. If
716 multiple parameter attributes are needed, they are space separated. For
718 %someFunc = i16 (i8 sext %someParam) zext
719 %someFunc = i16 (i8 zext %someParam) zext</pre>
720 <p>Note that the two function types above are unique because the parameter has
721 a different attribute (sext in the first one, zext in the second). Also note
722 that the attribute for the function result (zext) comes immediately after the
725 <p>Currently, only the following parameter attributes are defined:</p>
727 <dt><tt>zext</tt></dt>
728 <dd>This indicates that the parameter should be zero extended just before
729 a call to this function.</dd>
730 <dt><tt>sext</tt></dt>
731 <dd>This indicates that the parameter should be sign extended just before
732 a call to this function.</dd>
733 <dt><tt>inreg</tt></dt>
734 <dd>This indicates that the parameter should be placed in register (if
735 possible) during assembling function call. Support for this attribute is
737 <dt><tt>sret</tt></dt>
738 <dd>This indicates that the parameter specifies the address of a structure
739 that is the return value of the function in the source program.
745 <!-- ======================================================================= -->
746 <div class="doc_subsection">
747 <a name="moduleasm">Module-Level Inline Assembly</a>
750 <div class="doc_text">
752 Modules may contain "module-level inline asm" blocks, which corresponds to the
753 GCC "file scope inline asm" blocks. These blocks are internally concatenated by
754 LLVM and treated as a single unit, but may be separated in the .ll file if
755 desired. The syntax is very simple:
758 <div class="doc_code"><pre>
759 module asm "inline asm code goes here"
760 module asm "more can go here"
763 <p>The strings can contain any character by escaping non-printable characters.
764 The escape sequence used is simply "\xx" where "xx" is the two digit hex code
769 The inline asm code is simply printed to the machine code .s file when
770 assembly code is generated.
775 <!-- *********************************************************************** -->
776 <div class="doc_section"> <a name="typesystem">Type System</a> </div>
777 <!-- *********************************************************************** -->
779 <div class="doc_text">
781 <p>The LLVM type system is one of the most important features of the
782 intermediate representation. Being typed enables a number of
783 optimizations to be performed on the IR directly, without having to do
784 extra analyses on the side before the transformation. A strong type
785 system makes it easier to read the generated code and enables novel
786 analyses and transformations that are not feasible to perform on normal
787 three address code representations.</p>
791 <!-- ======================================================================= -->
792 <div class="doc_subsection"> <a name="t_primitive">Primitive Types</a> </div>
793 <div class="doc_text">
794 <p>The primitive types are the fundamental building blocks of the LLVM
795 system. The current set of primitive types is as follows:</p>
797 <table class="layout">
802 <tr><th>Type</th><th>Description</th></tr>
803 <tr><td><tt>void</tt></td><td>No value</td></tr>
804 <tr><td><tt>i8</tt></td><td>8-bit value</td></tr>
805 <tr><td><tt>i32</tt></td><td>32-bit value</td></tr>
806 <tr><td><tt>float</tt></td><td>32-bit floating point value</td></tr>
807 <tr><td><tt>label</tt></td><td>Branch destination</td></tr>
814 <tr><th>Type</th><th>Description</th></tr>
815 <tr><td><tt>i1</tt></td><td>True or False value</td></tr>
816 <tr><td><tt>i16</tt></td><td>16-bit value</td></tr>
817 <tr><td><tt>i64</tt></td><td>64-bit value</td></tr>
818 <tr><td><tt>double</tt></td><td>64-bit floating point value</td></tr>
826 <!-- _______________________________________________________________________ -->
827 <div class="doc_subsubsection"> <a name="t_classifications">Type
828 Classifications</a> </div>
829 <div class="doc_text">
830 <p>These different primitive types fall into a few useful
833 <table border="1" cellspacing="0" cellpadding="4">
835 <tr><th>Classification</th><th>Types</th></tr>
837 <td><a name="t_integer">integer</a></td>
838 <td><tt>i1, i8, i16, i32, i64</tt></td>
841 <td><a name="t_floating">floating point</a></td>
842 <td><tt>float, double</tt></td>
845 <td><a name="t_firstclass">first class</a></td>
846 <td><tt>i1, i8, i16, i32, i64, float, double, <br/>
847 <a href="#t_pointer">pointer</a>,<a href="#t_packed">packed</a></tt>
853 <p>The <a href="#t_firstclass">first class</a> types are perhaps the
854 most important. Values of these types are the only ones which can be
855 produced by instructions, passed as arguments, or used as operands to
856 instructions. This means that all structures and arrays must be
857 manipulated either by pointer or by component.</p>
860 <!-- ======================================================================= -->
861 <div class="doc_subsection"> <a name="t_derived">Derived Types</a> </div>
863 <div class="doc_text">
865 <p>The real power in LLVM comes from the derived types in the system.
866 This is what allows a programmer to represent arrays, functions,
867 pointers, and other useful types. Note that these derived types may be
868 recursive: For example, it is possible to have a two dimensional array.</p>
872 <!-- _______________________________________________________________________ -->
873 <div class="doc_subsubsection"> <a name="t_array">Array Type</a> </div>
875 <div class="doc_text">
879 <p>The array type is a very simple derived type that arranges elements
880 sequentially in memory. The array type requires a size (number of
881 elements) and an underlying data type.</p>
886 [<# elements> x <elementtype>]
889 <p>The number of elements is a constant integer value; elementtype may
890 be any type with a size.</p>
893 <table class="layout">
896 <tt>[40 x i32 ]</tt><br/>
897 <tt>[41 x i32 ]</tt><br/>
898 <tt>[40 x i8]</tt><br/>
901 Array of 40 32-bit integer values.<br/>
902 Array of 41 32-bit integer values.<br/>
903 Array of 40 8-bit integer values.<br/>
907 <p>Here are some examples of multidimensional arrays:</p>
908 <table class="layout">
911 <tt>[3 x [4 x i32]]</tt><br/>
912 <tt>[12 x [10 x float]]</tt><br/>
913 <tt>[2 x [3 x [4 x i16]]]</tt><br/>
916 3x4 array of 32-bit integer values.<br/>
917 12x10 array of single precision floating point values.<br/>
918 2x3x4 array of 16-bit integer values.<br/>
923 <p>Note that 'variable sized arrays' can be implemented in LLVM with a zero
924 length array. Normally, accesses past the end of an array are undefined in
925 LLVM (e.g. it is illegal to access the 5th element of a 3 element array).
926 As a special case, however, zero length arrays are recognized to be variable
927 length. This allows implementation of 'pascal style arrays' with the LLVM
928 type "{ i32, [0 x float]}", for example.</p>
932 <!-- _______________________________________________________________________ -->
933 <div class="doc_subsubsection"> <a name="t_function">Function Type</a> </div>
934 <div class="doc_text">
936 <p>The function type can be thought of as a function signature. It
937 consists of a return type and a list of formal parameter types.
938 Function types are usually used to build virtual function tables
939 (which are structures of pointers to functions), for indirect function
940 calls, and when defining a function.</p>
942 The return type of a function type cannot be an aggregate type.
945 <pre> <returntype> (<parameter list>)<br></pre>
946 <p>...where '<tt><parameter list></tt>' is a comma-separated list of type
947 specifiers. Optionally, the parameter list may include a type <tt>...</tt>,
948 which indicates that the function takes a variable number of arguments.
949 Variable argument functions can access their arguments with the <a
950 href="#int_varargs">variable argument handling intrinsic</a> functions.</p>
952 <table class="layout">
954 <td class="left"><tt>i32 (i32)</tt></td>
955 <td class="left">function taking an <tt>i32</tt>, returning an <tt>i32</tt>
957 </tr><tr class="layout">
958 <td class="left"><tt>float (i16 sext, i32 *) *
960 <td class="left"><a href="#t_pointer">Pointer</a> to a function that takes
961 an <tt>i16</tt> that should be sign extended and a
962 <a href="#t_pointer">pointer</a> to <tt>i32</tt>, returning
965 </tr><tr class="layout">
966 <td class="left"><tt>i32 (i8*, ...)</tt></td>
967 <td class="left">A vararg function that takes at least one
968 <a href="#t_pointer">pointer</a> to <tt>i8 </tt> (char in C),
969 which returns an integer. This is the signature for <tt>printf</tt> in
976 <!-- _______________________________________________________________________ -->
977 <div class="doc_subsubsection"> <a name="t_struct">Structure Type</a> </div>
978 <div class="doc_text">
980 <p>The structure type is used to represent a collection of data members
981 together in memory. The packing of the field types is defined to match
982 the ABI of the underlying processor. The elements of a structure may
983 be any type that has a size.</p>
984 <p>Structures are accessed using '<tt><a href="#i_load">load</a></tt>
985 and '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a
986 field with the '<tt><a href="#i_getelementptr">getelementptr</a></tt>'
989 <pre> { <type list> }<br></pre>
991 <table class="layout">
994 <tt>{ i32, i32, i32 }</tt><br/>
995 <tt>{ float, i32 (i32) * }</tt><br/>
998 a triple of three <tt>i32</tt> values<br/>
999 A pair, where the first element is a <tt>float</tt> and the second element
1000 is a <a href="#t_pointer">pointer</a> to a <a href="#t_function">function</a>
1001 that takes an <tt>i32</tt>, returning an <tt>i32</tt>.<br/>
1007 <!-- _______________________________________________________________________ -->
1008 <div class="doc_subsubsection"> <a name="t_pstruct">Packed Structure Type</a>
1010 <div class="doc_text">
1012 <p>The packed structure type is used to represent a collection of data members
1013 together in memory. There is no padding between fields. Further, the alignment
1014 of a packed structure is 1 byte. The elements of a packed structure may
1015 be any type that has a size.</p>
1016 <p>Structures are accessed using '<tt><a href="#i_load">load</a></tt>
1017 and '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a
1018 field with the '<tt><a href="#i_getelementptr">getelementptr</a></tt>'
1021 <pre> < { <type list> } > <br></pre>
1023 <table class="layout">
1026 <tt> < { i32, i32, i32 } > </tt><br/>
1027 <tt> < { float, i32 (i32) * } > </tt><br/>
1030 a triple of three <tt>i32</tt> values<br/>
1031 A pair, where the first element is a <tt>float</tt> and the second element
1032 is a <a href="#t_pointer">pointer</a> to a <a href="#t_function">function</a>
1033 that takes an <tt>i32</tt>, returning an <tt>i32</tt>.<br/>
1039 <!-- _______________________________________________________________________ -->
1040 <div class="doc_subsubsection"> <a name="t_pointer">Pointer Type</a> </div>
1041 <div class="doc_text">
1043 <p>As in many languages, the pointer type represents a pointer or
1044 reference to another object, which must live in memory.</p>
1046 <pre> <type> *<br></pre>
1048 <table class="layout">
1051 <tt>[4x i32]*</tt><br/>
1052 <tt>i32 (i32 *) *</tt><br/>
1055 A <a href="#t_pointer">pointer</a> to <a href="#t_array">array</a> of
1056 four <tt>i32</tt> values<br/>
1057 A <a href="#t_pointer">pointer</a> to a <a
1058 href="#t_function">function</a> that takes an <tt>i32*</tt>, returning an
1065 <!-- _______________________________________________________________________ -->
1066 <div class="doc_subsubsection"> <a name="t_packed">Packed Type</a> </div>
1067 <div class="doc_text">
1071 <p>A packed type is a simple derived type that represents a vector
1072 of elements. Packed types are used when multiple primitive data
1073 are operated in parallel using a single instruction (SIMD).
1074 A packed type requires a size (number of
1075 elements) and an underlying primitive data type. Vectors must have a power
1076 of two length (1, 2, 4, 8, 16 ...). Packed types are
1077 considered <a href="#t_firstclass">first class</a>.</p>
1082 < <# elements> x <elementtype> >
1085 <p>The number of elements is a constant integer value; elementtype may
1086 be any integer or floating point type.</p>
1090 <table class="layout">
1093 <tt><4 x i32></tt><br/>
1094 <tt><8 x float></tt><br/>
1095 <tt><2 x i64></tt><br/>
1098 Packed vector of 4 32-bit integer values.<br/>
1099 Packed vector of 8 floating-point values.<br/>
1100 Packed vector of 2 64-bit integer values.<br/>
1106 <!-- _______________________________________________________________________ -->
1107 <div class="doc_subsubsection"> <a name="t_opaque">Opaque Type</a> </div>
1108 <div class="doc_text">
1112 <p>Opaque types are used to represent unknown types in the system. This
1113 corresponds (for example) to the C notion of a foward declared structure type.
1114 In LLVM, opaque types can eventually be resolved to any type (not just a
1115 structure type).</p>
1125 <table class="layout">
1131 An opaque type.<br/>
1138 <!-- *********************************************************************** -->
1139 <div class="doc_section"> <a name="constants">Constants</a> </div>
1140 <!-- *********************************************************************** -->
1142 <div class="doc_text">
1144 <p>LLVM has several different basic types of constants. This section describes
1145 them all and their syntax.</p>
1149 <!-- ======================================================================= -->
1150 <div class="doc_subsection"><a name="simpleconstants">Simple Constants</a></div>
1152 <div class="doc_text">
1155 <dt><b>Boolean constants</b></dt>
1157 <dd>The two strings '<tt>true</tt>' and '<tt>false</tt>' are both valid
1158 constants of the <tt><a href="#t_primitive">i1</a></tt> type.
1161 <dt><b>Integer constants</b></dt>
1163 <dd>Standard integers (such as '4') are constants of the <a
1164 href="#t_integer">integer</a> type. Negative numbers may be used with
1168 <dt><b>Floating point constants</b></dt>
1170 <dd>Floating point constants use standard decimal notation (e.g. 123.421),
1171 exponential notation (e.g. 1.23421e+2), or a more precise hexadecimal
1172 notation (see below). Floating point constants must have a <a
1173 href="#t_floating">floating point</a> type. </dd>
1175 <dt><b>Null pointer constants</b></dt>
1177 <dd>The identifier '<tt>null</tt>' is recognized as a null pointer constant
1178 and must be of <a href="#t_pointer">pointer type</a>.</dd>
1182 <p>The one non-intuitive notation for constants is the optional hexadecimal form
1183 of floating point constants. For example, the form '<tt>double
1184 0x432ff973cafa8000</tt>' is equivalent to (but harder to read than) '<tt>double
1185 4.5e+15</tt>'. The only time hexadecimal floating point constants are required
1186 (and the only time that they are generated by the disassembler) is when a
1187 floating point constant must be emitted but it cannot be represented as a
1188 decimal floating point number. For example, NaN's, infinities, and other
1189 special values are represented in their IEEE hexadecimal format so that
1190 assembly and disassembly do not cause any bits to change in the constants.</p>
1194 <!-- ======================================================================= -->
1195 <div class="doc_subsection"><a name="aggregateconstants">Aggregate Constants</a>
1198 <div class="doc_text">
1199 <p>Aggregate constants arise from aggregation of simple constants
1200 and smaller aggregate constants.</p>
1203 <dt><b>Structure constants</b></dt>
1205 <dd>Structure constants are represented with notation similar to structure
1206 type definitions (a comma separated list of elements, surrounded by braces
1207 (<tt>{}</tt>)). For example: "<tt>{ i32 4, float 17.0, i32* %G }</tt>",
1208 where "<tt>%G</tt>" is declared as "<tt>%G = external global i32</tt>". Structure constants
1209 must have <a href="#t_struct">structure type</a>, and the number and
1210 types of elements must match those specified by the type.
1213 <dt><b>Array constants</b></dt>
1215 <dd>Array constants are represented with notation similar to array type
1216 definitions (a comma separated list of elements, surrounded by square brackets
1217 (<tt>[]</tt>)). For example: "<tt>[ i32 42, i32 11, i32 74 ]</tt>". Array
1218 constants must have <a href="#t_array">array type</a>, and the number and
1219 types of elements must match those specified by the type.
1222 <dt><b>Packed constants</b></dt>
1224 <dd>Packed constants are represented with notation similar to packed type
1225 definitions (a comma separated list of elements, surrounded by
1226 less-than/greater-than's (<tt><></tt>)). For example: "<tt>< i32 42,
1227 i32 11, i32 74, i32 100 ></tt>". Packed constants must have <a
1228 href="#t_packed">packed type</a>, and the number and types of elements must
1229 match those specified by the type.
1232 <dt><b>Zero initialization</b></dt>
1234 <dd>The string '<tt>zeroinitializer</tt>' can be used to zero initialize a
1235 value to zero of <em>any</em> type, including scalar and aggregate types.
1236 This is often used to avoid having to print large zero initializers (e.g. for
1237 large arrays) and is always exactly equivalent to using explicit zero
1244 <!-- ======================================================================= -->
1245 <div class="doc_subsection">
1246 <a name="globalconstants">Global Variable and Function Addresses</a>
1249 <div class="doc_text">
1251 <p>The addresses of <a href="#globalvars">global variables</a> and <a
1252 href="#functionstructure">functions</a> are always implicitly valid (link-time)
1253 constants. These constants are explicitly referenced when the <a
1254 href="#identifiers">identifier for the global</a> is used and always have <a
1255 href="#t_pointer">pointer</a> type. For example, the following is a legal LLVM
1261 %Z = global [2 x i32*] [ i32* %X, i32* %Y ]
1266 <!-- ======================================================================= -->
1267 <div class="doc_subsection"><a name="undefvalues">Undefined Values</a></div>
1268 <div class="doc_text">
1269 <p>The string '<tt>undef</tt>' is recognized as a type-less constant that has
1270 no specific value. Undefined values may be of any type and be used anywhere
1271 a constant is permitted.</p>
1273 <p>Undefined values indicate to the compiler that the program is well defined
1274 no matter what value is used, giving the compiler more freedom to optimize.
1278 <!-- ======================================================================= -->
1279 <div class="doc_subsection"><a name="constantexprs">Constant Expressions</a>
1282 <div class="doc_text">
1284 <p>Constant expressions are used to allow expressions involving other constants
1285 to be used as constants. Constant expressions may be of any <a
1286 href="#t_firstclass">first class</a> type and may involve any LLVM operation
1287 that does not have side effects (e.g. load and call are not supported). The
1288 following is the syntax for constant expressions:</p>
1291 <dt><b><tt>trunc ( CST to TYPE )</tt></b></dt>
1292 <dd>Truncate a constant to another type. The bit size of CST must be larger
1293 than the bit size of TYPE. Both types must be integers.</dd>
1295 <dt><b><tt>zext ( CST to TYPE )</tt></b></dt>
1296 <dd>Zero extend a constant to another type. The bit size of CST must be
1297 smaller or equal to the bit size of TYPE. Both types must be integers.</dd>
1299 <dt><b><tt>sext ( CST to TYPE )</tt></b></dt>
1300 <dd>Sign extend a constant to another type. The bit size of CST must be
1301 smaller or equal to the bit size of TYPE. Both types must be integers.</dd>
1303 <dt><b><tt>fptrunc ( CST to TYPE )</tt></b></dt>
1304 <dd>Truncate a floating point constant to another floating point type. The
1305 size of CST must be larger than the size of TYPE. Both types must be
1306 floating point.</dd>
1308 <dt><b><tt>fpext ( CST to TYPE )</tt></b></dt>
1309 <dd>Floating point extend a constant to another type. The size of CST must be
1310 smaller or equal to the size of TYPE. Both types must be floating point.</dd>
1312 <dt><b><tt>fp2uint ( CST to TYPE )</tt></b></dt>
1313 <dd>Convert a floating point constant to the corresponding unsigned integer
1314 constant. TYPE must be an integer type. CST must be floating point. If the
1315 value won't fit in the integer type, the results are undefined.</dd>
1317 <dt><b><tt>fptosi ( CST to TYPE )</tt></b></dt>
1318 <dd>Convert a floating point constant to the corresponding signed integer
1319 constant. TYPE must be an integer type. CST must be floating point. If the
1320 value won't fit in the integer type, the results are undefined.</dd>
1322 <dt><b><tt>uitofp ( CST to TYPE )</tt></b></dt>
1323 <dd>Convert an unsigned integer constant to the corresponding floating point
1324 constant. TYPE must be floating point. CST must be of integer type. If the
1325 value won't fit in the floating point type, the results are undefined.</dd>
1327 <dt><b><tt>sitofp ( CST to TYPE )</tt></b></dt>
1328 <dd>Convert a signed integer constant to the corresponding floating point
1329 constant. TYPE must be floating point. CST must be of integer type. If the
1330 value won't fit in the floating point type, the results are undefined.</dd>
1332 <dt><b><tt>ptrtoint ( CST to TYPE )</tt></b></dt>
1333 <dd>Convert a pointer typed constant to the corresponding integer constant
1334 TYPE must be an integer type. CST must be of pointer type. The CST value is
1335 zero extended, truncated, or unchanged to make it fit in TYPE.</dd>
1337 <dt><b><tt>inttoptr ( CST to TYPE )</tt></b></dt>
1338 <dd>Convert a integer constant to a pointer constant. TYPE must be a
1339 pointer type. CST must be of integer type. The CST value is zero extended,
1340 truncated, or unchanged to make it fit in a pointer size. This one is
1341 <i>really</i> dangerous!</dd>
1343 <dt><b><tt>bitcast ( CST to TYPE )</tt></b></dt>
1344 <dd>Convert a constant, CST, to another TYPE. The size of CST and TYPE must be
1345 identical (same number of bits). The conversion is done as if the CST value
1346 was stored to memory and read back as TYPE. In other words, no bits change
1347 with this operator, just the type. This can be used for conversion of
1348 packed types to any other type, as long as they have the same bit width. For
1349 pointers it is only valid to cast to another pointer type.
1352 <dt><b><tt>getelementptr ( CSTPTR, IDX0, IDX1, ... )</tt></b></dt>
1354 <dd>Perform the <a href="#i_getelementptr">getelementptr operation</a> on
1355 constants. As with the <a href="#i_getelementptr">getelementptr</a>
1356 instruction, the index list may have zero or more indexes, which are required
1357 to make sense for the type of "CSTPTR".</dd>
1359 <dt><b><tt>select ( COND, VAL1, VAL2 )</tt></b></dt>
1361 <dd>Perform the <a href="#i_select">select operation</a> on
1364 <dt><b><tt>icmp COND ( VAL1, VAL2 )</tt></b></dt>
1365 <dd>Performs the <a href="#i_icmp">icmp operation</a> on constants.</dd>
1367 <dt><b><tt>fcmp COND ( VAL1, VAL2 )</tt></b></dt>
1368 <dd>Performs the <a href="#i_fcmp">fcmp operation</a> on constants.</dd>
1370 <dt><b><tt>extractelement ( VAL, IDX )</tt></b></dt>
1372 <dd>Perform the <a href="#i_extractelement">extractelement
1373 operation</a> on constants.
1375 <dt><b><tt>insertelement ( VAL, ELT, IDX )</tt></b></dt>
1377 <dd>Perform the <a href="#i_insertelement">insertelement
1378 operation</a> on constants.</dd>
1381 <dt><b><tt>shufflevector ( VEC1, VEC2, IDXMASK )</tt></b></dt>
1383 <dd>Perform the <a href="#i_shufflevector">shufflevector
1384 operation</a> on constants.</dd>
1386 <dt><b><tt>OPCODE ( LHS, RHS )</tt></b></dt>
1388 <dd>Perform the specified operation of the LHS and RHS constants. OPCODE may
1389 be any of the <a href="#binaryops">binary</a> or <a href="#bitwiseops">bitwise
1390 binary</a> operations. The constraints on operands are the same as those for
1391 the corresponding instruction (e.g. no bitwise operations on floating point
1392 values are allowed).</dd>
1396 <!-- *********************************************************************** -->
1397 <div class="doc_section"> <a name="othervalues">Other Values</a> </div>
1398 <!-- *********************************************************************** -->
1400 <!-- ======================================================================= -->
1401 <div class="doc_subsection">
1402 <a name="inlineasm">Inline Assembler Expressions</a>
1405 <div class="doc_text">
1408 LLVM supports inline assembler expressions (as opposed to <a href="#moduleasm">
1409 Module-Level Inline Assembly</a>) through the use of a special value. This
1410 value represents the inline assembler as a string (containing the instructions
1411 to emit), a list of operand constraints (stored as a string), and a flag that
1412 indicates whether or not the inline asm expression has side effects. An example
1413 inline assembler expression is:
1417 i32 (i32) asm "bswap $0", "=r,r"
1421 Inline assembler expressions may <b>only</b> be used as the callee operand of
1422 a <a href="#i_call"><tt>call</tt> instruction</a>. Thus, typically we have:
1426 %X = call i32 asm "<a href="#i_bswap">bswap</a> $0", "=r,r"(i32 %Y)
1430 Inline asms with side effects not visible in the constraint list must be marked
1431 as having side effects. This is done through the use of the
1432 '<tt>sideeffect</tt>' keyword, like so:
1436 call void asm sideeffect "eieio", ""()
1439 <p>TODO: The format of the asm and constraints string still need to be
1440 documented here. Constraints on what can be done (e.g. duplication, moving, etc
1441 need to be documented).
1446 <!-- *********************************************************************** -->
1447 <div class="doc_section"> <a name="instref">Instruction Reference</a> </div>
1448 <!-- *********************************************************************** -->
1450 <div class="doc_text">
1452 <p>The LLVM instruction set consists of several different
1453 classifications of instructions: <a href="#terminators">terminator
1454 instructions</a>, <a href="#binaryops">binary instructions</a>,
1455 <a href="#bitwiseops">bitwise binary instructions</a>, <a
1456 href="#memoryops">memory instructions</a>, and <a href="#otherops">other
1457 instructions</a>.</p>
1461 <!-- ======================================================================= -->
1462 <div class="doc_subsection"> <a name="terminators">Terminator
1463 Instructions</a> </div>
1465 <div class="doc_text">
1467 <p>As mentioned <a href="#functionstructure">previously</a>, every
1468 basic block in a program ends with a "Terminator" instruction, which
1469 indicates which block should be executed after the current block is
1470 finished. These terminator instructions typically yield a '<tt>void</tt>'
1471 value: they produce control flow, not values (the one exception being
1472 the '<a href="#i_invoke"><tt>invoke</tt></a>' instruction).</p>
1473 <p>There are six different terminator instructions: the '<a
1474 href="#i_ret"><tt>ret</tt></a>' instruction, the '<a href="#i_br"><tt>br</tt></a>'
1475 instruction, the '<a href="#i_switch"><tt>switch</tt></a>' instruction,
1476 the '<a href="#i_invoke"><tt>invoke</tt></a>' instruction, the '<a
1477 href="#i_unwind"><tt>unwind</tt></a>' instruction, and the '<a
1478 href="#i_unreachable"><tt>unreachable</tt></a>' instruction.</p>
1482 <!-- _______________________________________________________________________ -->
1483 <div class="doc_subsubsection"> <a name="i_ret">'<tt>ret</tt>'
1484 Instruction</a> </div>
1485 <div class="doc_text">
1487 <pre> ret <type> <value> <i>; Return a value from a non-void function</i>
1488 ret void <i>; Return from void function</i>
1491 <p>The '<tt>ret</tt>' instruction is used to return control flow (and a
1492 value) from a function back to the caller.</p>
1493 <p>There are two forms of the '<tt>ret</tt>' instruction: one that
1494 returns a value and then causes control flow, and one that just causes
1495 control flow to occur.</p>
1497 <p>The '<tt>ret</tt>' instruction may return any '<a
1498 href="#t_firstclass">first class</a>' type. Notice that a function is
1499 not <a href="#wellformed">well formed</a> if there exists a '<tt>ret</tt>'
1500 instruction inside of the function that returns a value that does not
1501 match the return type of the function.</p>
1503 <p>When the '<tt>ret</tt>' instruction is executed, control flow
1504 returns back to the calling function's context. If the caller is a "<a
1505 href="#i_call"><tt>call</tt></a>" instruction, execution continues at
1506 the instruction after the call. If the caller was an "<a
1507 href="#i_invoke"><tt>invoke</tt></a>" instruction, execution continues
1508 at the beginning of the "normal" destination block. If the instruction
1509 returns a value, that value shall set the call or invoke instruction's
1512 <pre> ret i32 5 <i>; Return an integer value of 5</i>
1513 ret void <i>; Return from a void function</i>
1516 <!-- _______________________________________________________________________ -->
1517 <div class="doc_subsubsection"> <a name="i_br">'<tt>br</tt>' Instruction</a> </div>
1518 <div class="doc_text">
1520 <pre> br i1 <cond>, label <iftrue>, label <iffalse><br> br label <dest> <i>; Unconditional branch</i>
1523 <p>The '<tt>br</tt>' instruction is used to cause control flow to
1524 transfer to a different basic block in the current function. There are
1525 two forms of this instruction, corresponding to a conditional branch
1526 and an unconditional branch.</p>
1528 <p>The conditional branch form of the '<tt>br</tt>' instruction takes a
1529 single '<tt>i1</tt>' value and two '<tt>label</tt>' values. The
1530 unconditional form of the '<tt>br</tt>' instruction takes a single '<tt>label</tt>'
1531 value as a target.</p>
1533 <p>Upon execution of a conditional '<tt>br</tt>' instruction, the '<tt>i1</tt>'
1534 argument is evaluated. If the value is <tt>true</tt>, control flows
1535 to the '<tt>iftrue</tt>' <tt>label</tt> argument. If "cond" is <tt>false</tt>,
1536 control flows to the '<tt>iffalse</tt>' <tt>label</tt> argument.</p>
1538 <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
1539 href="#i_ret">ret</a> i32 1<br>IfUnequal:<br> <a href="#i_ret">ret</a> i32 0<br></pre>
1541 <!-- _______________________________________________________________________ -->
1542 <div class="doc_subsubsection">
1543 <a name="i_switch">'<tt>switch</tt>' Instruction</a>
1546 <div class="doc_text">
1550 switch <intty> <value>, label <defaultdest> [ <intty> <val>, label <dest> ... ]
1555 <p>The '<tt>switch</tt>' instruction is used to transfer control flow to one of
1556 several different places. It is a generalization of the '<tt>br</tt>'
1557 instruction, allowing a branch to occur to one of many possible
1563 <p>The '<tt>switch</tt>' instruction uses three parameters: an integer
1564 comparison value '<tt>value</tt>', a default '<tt>label</tt>' destination, and
1565 an array of pairs of comparison value constants and '<tt>label</tt>'s. The
1566 table is not allowed to contain duplicate constant entries.</p>
1570 <p>The <tt>switch</tt> instruction specifies a table of values and
1571 destinations. When the '<tt>switch</tt>' instruction is executed, this
1572 table is searched for the given value. If the value is found, control flow is
1573 transfered to the corresponding destination; otherwise, control flow is
1574 transfered to the default destination.</p>
1576 <h5>Implementation:</h5>
1578 <p>Depending on properties of the target machine and the particular
1579 <tt>switch</tt> instruction, this instruction may be code generated in different
1580 ways. For example, it could be generated as a series of chained conditional
1581 branches or with a lookup table.</p>
1586 <i>; Emulate a conditional br instruction</i>
1587 %Val = <a href="#i_zext">zext</a> i1 %value to i32
1588 switch i32 %Val, label %truedest [i32 0, label %falsedest ]
1590 <i>; Emulate an unconditional br instruction</i>
1591 switch i32 0, label %dest [ ]
1593 <i>; Implement a jump table:</i>
1594 switch i32 %val, label %otherwise [ i32 0, label %onzero
1596 i32 2, label %ontwo ]
1600 <!-- _______________________________________________________________________ -->
1601 <div class="doc_subsubsection">
1602 <a name="i_invoke">'<tt>invoke</tt>' Instruction</a>
1605 <div class="doc_text">
1610 <result> = invoke [<a href="#callingconv">cconv</a>] <ptr to function ty> %<function ptr val>(<function args>)
1611 to label <normal label> unwind label <exception label>
1616 <p>The '<tt>invoke</tt>' instruction causes control to transfer to a specified
1617 function, with the possibility of control flow transfer to either the
1618 '<tt>normal</tt>' label or the
1619 '<tt>exception</tt>' label. If the callee function returns with the
1620 "<tt><a href="#i_ret">ret</a></tt>" instruction, control flow will return to the
1621 "normal" label. If the callee (or any indirect callees) returns with the "<a
1622 href="#i_unwind"><tt>unwind</tt></a>" instruction, control is interrupted and
1623 continued at the dynamically nearest "exception" label.</p>
1627 <p>This instruction requires several arguments:</p>
1631 The optional "cconv" marker indicates which <a href="callingconv">calling
1632 convention</a> the call should use. If none is specified, the call defaults
1633 to using C calling conventions.
1635 <li>'<tt>ptr to function ty</tt>': shall be the signature of the pointer to
1636 function value being invoked. In most cases, this is a direct function
1637 invocation, but indirect <tt>invoke</tt>s are just as possible, branching off
1638 an arbitrary pointer to function value.
1641 <li>'<tt>function ptr val</tt>': An LLVM value containing a pointer to a
1642 function to be invoked. </li>
1644 <li>'<tt>function args</tt>': argument list whose types match the function
1645 signature argument types. If the function signature indicates the function
1646 accepts a variable number of arguments, the extra arguments can be
1649 <li>'<tt>normal label</tt>': the label reached when the called function
1650 executes a '<tt><a href="#i_ret">ret</a></tt>' instruction. </li>
1652 <li>'<tt>exception label</tt>': the label reached when a callee returns with
1653 the <a href="#i_unwind"><tt>unwind</tt></a> instruction. </li>
1659 <p>This instruction is designed to operate as a standard '<tt><a
1660 href="#i_call">call</a></tt>' instruction in most regards. The primary
1661 difference is that it establishes an association with a label, which is used by
1662 the runtime library to unwind the stack.</p>
1664 <p>This instruction is used in languages with destructors to ensure that proper
1665 cleanup is performed in the case of either a <tt>longjmp</tt> or a thrown
1666 exception. Additionally, this is important for implementation of
1667 '<tt>catch</tt>' clauses in high-level languages that support them.</p>
1671 %retval = invoke i32 %Test(i32 15) to label %Continue
1672 unwind label %TestCleanup <i>; {i32}:retval set</i>
1673 %retval = invoke <a href="#callingconv">coldcc</a> i32 %Test(i32 15) to label %Continue
1674 unwind label %TestCleanup <i>; {i32}:retval set</i>
1679 <!-- _______________________________________________________________________ -->
1681 <div class="doc_subsubsection"> <a name="i_unwind">'<tt>unwind</tt>'
1682 Instruction</a> </div>
1684 <div class="doc_text">
1693 <p>The '<tt>unwind</tt>' instruction unwinds the stack, continuing control flow
1694 at the first callee in the dynamic call stack which used an <a
1695 href="#i_invoke"><tt>invoke</tt></a> instruction to perform the call. This is
1696 primarily used to implement exception handling.</p>
1700 <p>The '<tt>unwind</tt>' intrinsic causes execution of the current function to
1701 immediately halt. The dynamic call stack is then searched for the first <a
1702 href="#i_invoke"><tt>invoke</tt></a> instruction on the call stack. Once found,
1703 execution continues at the "exceptional" destination block specified by the
1704 <tt>invoke</tt> instruction. If there is no <tt>invoke</tt> instruction in the
1705 dynamic call chain, undefined behavior results.</p>
1708 <!-- _______________________________________________________________________ -->
1710 <div class="doc_subsubsection"> <a name="i_unreachable">'<tt>unreachable</tt>'
1711 Instruction</a> </div>
1713 <div class="doc_text">
1722 <p>The '<tt>unreachable</tt>' instruction has no defined semantics. This
1723 instruction is used to inform the optimizer that a particular portion of the
1724 code is not reachable. This can be used to indicate that the code after a
1725 no-return function cannot be reached, and other facts.</p>
1729 <p>The '<tt>unreachable</tt>' instruction has no defined semantics.</p>
1734 <!-- ======================================================================= -->
1735 <div class="doc_subsection"> <a name="binaryops">Binary Operations</a> </div>
1736 <div class="doc_text">
1737 <p>Binary operators are used to do most of the computation in a
1738 program. They require two operands, execute an operation on them, and
1739 produce a single value. The operands might represent
1740 multiple data, as is the case with the <a href="#t_packed">packed</a> data type.
1741 The result value of a binary operator is not
1742 necessarily the same type as its operands.</p>
1743 <p>There are several different binary operators:</p>
1745 <!-- _______________________________________________________________________ -->
1746 <div class="doc_subsubsection"> <a name="i_add">'<tt>add</tt>'
1747 Instruction</a> </div>
1748 <div class="doc_text">
1750 <pre> <result> = add <ty> <var1>, <var2> <i>; yields {ty}:result</i>
1753 <p>The '<tt>add</tt>' instruction returns the sum of its two operands.</p>
1755 <p>The two arguments to the '<tt>add</tt>' instruction must be either <a
1756 href="#t_integer">integer</a> or <a href="#t_floating">floating point</a> values.
1757 This instruction can also take <a href="#t_packed">packed</a> versions of the values.
1758 Both arguments must have identical types.</p>
1760 <p>The value produced is the integer or floating point sum of the two
1763 <pre> <result> = add i32 4, %var <i>; yields {i32}:result = 4 + %var</i>
1766 <!-- _______________________________________________________________________ -->
1767 <div class="doc_subsubsection"> <a name="i_sub">'<tt>sub</tt>'
1768 Instruction</a> </div>
1769 <div class="doc_text">
1771 <pre> <result> = sub <ty> <var1>, <var2> <i>; yields {ty}:result</i>
1774 <p>The '<tt>sub</tt>' instruction returns the difference of its two
1776 <p>Note that the '<tt>sub</tt>' instruction is used to represent the '<tt>neg</tt>'
1777 instruction present in most other intermediate representations.</p>
1779 <p>The two arguments to the '<tt>sub</tt>' instruction must be either <a
1780 href="#t_integer">integer</a> or <a href="#t_floating">floating point</a>
1782 This instruction can also take <a href="#t_packed">packed</a> versions of the values.
1783 Both arguments must have identical types.</p>
1785 <p>The value produced is the integer or floating point difference of
1786 the two operands.</p>
1788 <pre> <result> = sub i32 4, %var <i>; yields {i32}:result = 4 - %var</i>
1789 <result> = sub i32 0, %val <i>; yields {i32}:result = -%var</i>
1792 <!-- _______________________________________________________________________ -->
1793 <div class="doc_subsubsection"> <a name="i_mul">'<tt>mul</tt>'
1794 Instruction</a> </div>
1795 <div class="doc_text">
1797 <pre> <result> = mul <ty> <var1>, <var2> <i>; yields {ty}:result</i>
1800 <p>The '<tt>mul</tt>' instruction returns the product of its two
1803 <p>The two arguments to the '<tt>mul</tt>' instruction must be either <a
1804 href="#t_integer">integer</a> or <a href="#t_floating">floating point</a>
1806 This instruction can also take <a href="#t_packed">packed</a> versions of the values.
1807 Both arguments must have identical types.</p>
1809 <p>The value produced is the integer or floating point product of the
1811 <p>Because the operands are the same width, the result of an integer
1812 multiplication is the same whether the operands should be deemed unsigned or
1815 <pre> <result> = mul i32 4, %var <i>; yields {i32}:result = 4 * %var</i>
1818 <!-- _______________________________________________________________________ -->
1819 <div class="doc_subsubsection"> <a name="i_udiv">'<tt>udiv</tt>' Instruction
1821 <div class="doc_text">
1823 <pre> <result> = udiv <ty> <var1>, <var2> <i>; yields {ty}:result</i>
1826 <p>The '<tt>udiv</tt>' instruction returns the quotient of its two
1829 <p>The two arguments to the '<tt>udiv</tt>' instruction must be
1830 <a href="#t_integer">integer</a> values. Both arguments must have identical
1831 types. This instruction can also take <a href="#t_packed">packed</a> versions
1832 of the values in which case the elements must be integers.</p>
1834 <p>The value produced is the unsigned integer quotient of the two operands. This
1835 instruction always performs an unsigned division operation, regardless of
1836 whether the arguments are unsigned or not.</p>
1838 <pre> <result> = udiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
1841 <!-- _______________________________________________________________________ -->
1842 <div class="doc_subsubsection"> <a name="i_sdiv">'<tt>sdiv</tt>' Instruction
1844 <div class="doc_text">
1846 <pre> <result> = sdiv <ty> <var1>, <var2> <i>; yields {ty}:result</i>
1849 <p>The '<tt>sdiv</tt>' instruction returns the quotient of its two
1852 <p>The two arguments to the '<tt>sdiv</tt>' instruction must be
1853 <a href="#t_integer">integer</a> values. Both arguments must have identical
1854 types. This instruction can also take <a href="#t_packed">packed</a> versions
1855 of the values in which case the elements must be integers.</p>
1857 <p>The value produced is the signed integer quotient of the two operands. This
1858 instruction always performs a signed division operation, regardless of whether
1859 the arguments are signed or not.</p>
1861 <pre> <result> = sdiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
1864 <!-- _______________________________________________________________________ -->
1865 <div class="doc_subsubsection"> <a name="i_fdiv">'<tt>fdiv</tt>'
1866 Instruction</a> </div>
1867 <div class="doc_text">
1869 <pre> <result> = fdiv <ty> <var1>, <var2> <i>; yields {ty}:result</i>
1872 <p>The '<tt>fdiv</tt>' instruction returns the quotient of its two
1875 <p>The two arguments to the '<tt>div</tt>' instruction must be
1876 <a href="#t_floating">floating point</a> values. Both arguments must have
1877 identical types. This instruction can also take <a href="#t_packed">packed</a>
1878 versions of the values in which case the elements must be floating point.</p>
1880 <p>The value produced is the floating point quotient of the two operands.</p>
1882 <pre> <result> = fdiv float 4.0, %var <i>; yields {float}:result = 4.0 / %var</i>
1885 <!-- _______________________________________________________________________ -->
1886 <div class="doc_subsubsection"> <a name="i_urem">'<tt>urem</tt>' Instruction</a>
1888 <div class="doc_text">
1890 <pre> <result> = urem <ty> <var1>, <var2> <i>; yields {ty}:result</i>
1893 <p>The '<tt>urem</tt>' instruction returns the remainder from the
1894 unsigned division of its two arguments.</p>
1896 <p>The two arguments to the '<tt>urem</tt>' instruction must be
1897 <a href="#t_integer">integer</a> values. Both arguments must have identical
1900 <p>This instruction returns the unsigned integer <i>remainder</i> of a division.
1901 This instruction always performs an unsigned division to get the remainder,
1902 regardless of whether the arguments are unsigned or not.</p>
1904 <pre> <result> = urem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
1908 <!-- _______________________________________________________________________ -->
1909 <div class="doc_subsubsection"> <a name="i_srem">'<tt>srem</tt>'
1910 Instruction</a> </div>
1911 <div class="doc_text">
1913 <pre> <result> = srem <ty> <var1>, <var2> <i>; yields {ty}:result</i>
1916 <p>The '<tt>srem</tt>' instruction returns the remainder from the
1917 signed division of its two operands.</p>
1919 <p>The two arguments to the '<tt>srem</tt>' instruction must be
1920 <a href="#t_integer">integer</a> values. Both arguments must have identical
1923 <p>This instruction returns the <i>remainder</i> of a division (where the result
1924 has the same sign as the divisor), not the <i>modulus</i> (where the
1925 result has the same sign as the dividend) of a value. For more
1926 information about the difference, see <a
1927 href="http://mathforum.org/dr.math/problems/anne.4.28.99.html">The
1930 <pre> <result> = srem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
1934 <!-- _______________________________________________________________________ -->
1935 <div class="doc_subsubsection"> <a name="i_frem">'<tt>frem</tt>'
1936 Instruction</a> </div>
1937 <div class="doc_text">
1939 <pre> <result> = frem <ty> <var1>, <var2> <i>; yields {ty}:result</i>
1942 <p>The '<tt>frem</tt>' instruction returns the remainder from the
1943 division of its two operands.</p>
1945 <p>The two arguments to the '<tt>frem</tt>' instruction must be
1946 <a href="#t_floating">floating point</a> values. Both arguments must have
1947 identical types.</p>
1949 <p>This instruction returns the <i>remainder</i> of a division.</p>
1951 <pre> <result> = frem float 4.0, %var <i>; yields {float}:result = 4.0 % %var</i>
1955 <!-- _______________________________________________________________________ -->
1956 <div class="doc_subsubsection"> <a name="i_shl">'<tt>shl</tt>'
1957 Instruction</a> </div>
1958 <div class="doc_text">
1960 <pre> <result> = shl <ty> <var1>, <var2> <i>; yields {ty}:result</i>
1963 <p>The '<tt>shl</tt>' instruction returns the first operand shifted to
1964 the left a specified number of bits.</p>
1966 <p>Both arguments to the '<tt>shl</tt>' instruction must be the same <a
1967 href="#t_integer">integer</a> type.</p>
1969 <p>The value produced is <tt>var1</tt> * 2<sup><tt>var2</tt></sup>.</p>
1970 <h5>Example:</h5><pre>
1971 <result> = shl i32 4, %var <i>; yields {i32}: 4 << %var</i>
1972 <result> = shl i32 4, 2 <i>; yields {i32}: 16</i>
1973 <result> = shl i32 1, 10 <i>; yields {i32}: 1024</i>
1976 <!-- _______________________________________________________________________ -->
1977 <div class="doc_subsubsection"> <a name="i_lshr">'<tt>lshr</tt>'
1978 Instruction</a> </div>
1979 <div class="doc_text">
1981 <pre> <result> = lshr <ty> <var1>, <var2> <i>; yields {ty}:result</i>
1985 <p>The '<tt>lshr</tt>' instruction (logical shift right) returns the first
1986 operand shifted to the right a specified number of bits.</p>
1989 <p>Both arguments to the '<tt>lshr</tt>' instruction must be the same
1990 <a href="#t_integer">integer</a> type.</p>
1993 <p>This instruction always performs a logical shift right operation. The most
1994 significant bits of the result will be filled with zero bits after the
1999 <result> = lshr i32 4, 1 <i>; yields {i32}:result = 2</i>
2000 <result> = lshr i32 4, 2 <i>; yields {i32}:result = 1</i>
2001 <result> = lshr i8 4, 3 <i>; yields {i8}:result = 0</i>
2002 <result> = lshr i8 -2, 1 <i>; yields {i8}:result = 0x7FFFFFFF </i>
2006 <!-- ======================================================================= -->
2007 <div class="doc_subsubsection"> <a name="i_ashr">'<tt>ashr</tt>'
2008 Instruction</a> </div>
2009 <div class="doc_text">
2012 <pre> <result> = ashr <ty> <var1>, <var2> <i>; yields {ty}:result</i>
2016 <p>The '<tt>ashr</tt>' instruction (arithmetic shift right) returns the first
2017 operand shifted to the right a specified number of bits.</p>
2020 <p>Both arguments to the '<tt>ashr</tt>' instruction must be the same
2021 <a href="#t_integer">integer</a> type.</p>
2024 <p>This instruction always performs an arithmetic shift right operation,
2025 The most significant bits of the result will be filled with the sign bit
2026 of <tt>var1</tt>.</p>
2030 <result> = ashr i32 4, 1 <i>; yields {i32}:result = 2</i>
2031 <result> = ashr i32 4, 2 <i>; yields {i32}:result = 1</i>
2032 <result> = ashr i8 4, 3 <i>; yields {i8}:result = 0</i>
2033 <result> = ashr i8 -2, 1 <i>; yields {i8}:result = -1</i>
2037 <!-- ======================================================================= -->
2038 <div class="doc_subsection"> <a name="bitwiseops">Bitwise Binary
2039 Operations</a> </div>
2040 <div class="doc_text">
2041 <p>Bitwise binary operators are used to do various forms of
2042 bit-twiddling in a program. They are generally very efficient
2043 instructions and can commonly be strength reduced from other
2044 instructions. They require two operands, execute an operation on them,
2045 and produce a single value. The resulting value of the bitwise binary
2046 operators is always the same type as its first operand.</p>
2048 <!-- _______________________________________________________________________ -->
2049 <div class="doc_subsubsection"> <a name="i_and">'<tt>and</tt>'
2050 Instruction</a> </div>
2051 <div class="doc_text">
2053 <pre> <result> = and <ty> <var1>, <var2> <i>; yields {ty}:result</i>
2056 <p>The '<tt>and</tt>' instruction returns the bitwise logical and of
2057 its two operands.</p>
2059 <p>The two arguments to the '<tt>and</tt>' instruction must be <a
2060 href="#t_integer">integer</a> values. Both arguments must have
2061 identical types.</p>
2063 <p>The truth table used for the '<tt>and</tt>' instruction is:</p>
2065 <div style="align: center">
2066 <table border="1" cellspacing="0" cellpadding="4">
2097 <pre> <result> = and i32 4, %var <i>; yields {i32}:result = 4 & %var</i>
2098 <result> = and i32 15, 40 <i>; yields {i32}:result = 8</i>
2099 <result> = and i32 4, 8 <i>; yields {i32}:result = 0</i>
2102 <!-- _______________________________________________________________________ -->
2103 <div class="doc_subsubsection"> <a name="i_or">'<tt>or</tt>' Instruction</a> </div>
2104 <div class="doc_text">
2106 <pre> <result> = or <ty> <var1>, <var2> <i>; yields {ty}:result</i>
2109 <p>The '<tt>or</tt>' instruction returns the bitwise logical inclusive
2110 or of its two operands.</p>
2112 <p>The two arguments to the '<tt>or</tt>' instruction must be <a
2113 href="#t_integer">integer</a> values. Both arguments must have
2114 identical types.</p>
2116 <p>The truth table used for the '<tt>or</tt>' instruction is:</p>
2118 <div style="align: center">
2119 <table border="1" cellspacing="0" cellpadding="4">
2150 <pre> <result> = or i32 4, %var <i>; yields {i32}:result = 4 | %var</i>
2151 <result> = or i32 15, 40 <i>; yields {i32}:result = 47</i>
2152 <result> = or i32 4, 8 <i>; yields {i32}:result = 12</i>
2155 <!-- _______________________________________________________________________ -->
2156 <div class="doc_subsubsection"> <a name="i_xor">'<tt>xor</tt>'
2157 Instruction</a> </div>
2158 <div class="doc_text">
2160 <pre> <result> = xor <ty> <var1>, <var2> <i>; yields {ty}:result</i>
2163 <p>The '<tt>xor</tt>' instruction returns the bitwise logical exclusive
2164 or of its two operands. The <tt>xor</tt> is used to implement the
2165 "one's complement" operation, which is the "~" operator in C.</p>
2167 <p>The two arguments to the '<tt>xor</tt>' instruction must be <a
2168 href="#t_integer">integer</a> values. Both arguments must have
2169 identical types.</p>
2171 <p>The truth table used for the '<tt>xor</tt>' instruction is:</p>
2173 <div style="align: center">
2174 <table border="1" cellspacing="0" cellpadding="4">
2206 <pre> <result> = xor i32 4, %var <i>; yields {i32}:result = 4 ^ %var</i>
2207 <result> = xor i32 15, 40 <i>; yields {i32}:result = 39</i>
2208 <result> = xor i32 4, 8 <i>; yields {i32}:result = 12</i>
2209 <result> = xor i32 %V, -1 <i>; yields {i32}:result = ~%V</i>
2213 <!-- ======================================================================= -->
2214 <div class="doc_subsection">
2215 <a name="vectorops">Vector Operations</a>
2218 <div class="doc_text">
2220 <p>LLVM supports several instructions to represent vector operations in a
2221 target-independent manner. This instructions cover the element-access and
2222 vector-specific operations needed to process vectors effectively. While LLVM
2223 does directly support these vector operations, many sophisticated algorithms
2224 will want to use target-specific intrinsics to take full advantage of a specific
2229 <!-- _______________________________________________________________________ -->
2230 <div class="doc_subsubsection">
2231 <a name="i_extractelement">'<tt>extractelement</tt>' Instruction</a>
2234 <div class="doc_text">
2239 <result> = extractelement <n x <ty>> <val>, i32 <idx> <i>; yields <ty></i>
2245 The '<tt>extractelement</tt>' instruction extracts a single scalar
2246 element from a packed vector at a specified index.
2253 The first operand of an '<tt>extractelement</tt>' instruction is a
2254 value of <a href="#t_packed">packed</a> type. The second operand is
2255 an index indicating the position from which to extract the element.
2256 The index may be a variable.</p>
2261 The result is a scalar of the same type as the element type of
2262 <tt>val</tt>. Its value is the value at position <tt>idx</tt> of
2263 <tt>val</tt>. If <tt>idx</tt> exceeds the length of <tt>val</tt>, the
2264 results are undefined.
2270 %result = extractelement <4 x i32> %vec, i32 0 <i>; yields i32</i>
2275 <!-- _______________________________________________________________________ -->
2276 <div class="doc_subsubsection">
2277 <a name="i_insertelement">'<tt>insertelement</tt>' Instruction</a>
2280 <div class="doc_text">
2285 <result> = insertelement <n x <ty>> <val>, <ty> <elt>, i32 <idx> <i>; yields <n x <ty>></i>
2291 The '<tt>insertelement</tt>' instruction inserts a scalar
2292 element into a packed vector at a specified index.
2299 The first operand of an '<tt>insertelement</tt>' instruction is a
2300 value of <a href="#t_packed">packed</a> type. The second operand is a
2301 scalar value whose type must equal the element type of the first
2302 operand. The third operand is an index indicating the position at
2303 which to insert the value. The index may be a variable.</p>
2308 The result is a packed vector of the same type as <tt>val</tt>. Its
2309 element values are those of <tt>val</tt> except at position
2310 <tt>idx</tt>, where it gets the value <tt>elt</tt>. If <tt>idx</tt>
2311 exceeds the length of <tt>val</tt>, the results are undefined.
2317 %result = insertelement <4 x i32> %vec, i32 1, i32 0 <i>; yields <4 x i32></i>
2321 <!-- _______________________________________________________________________ -->
2322 <div class="doc_subsubsection">
2323 <a name="i_shufflevector">'<tt>shufflevector</tt>' Instruction</a>
2326 <div class="doc_text">
2331 <result> = shufflevector <n x <ty>> <v1>, <n x <ty>> <v2>, <n x i32> <mask> <i>; yields <n x <ty>></i>
2337 The '<tt>shufflevector</tt>' instruction constructs a permutation of elements
2338 from two input vectors, returning a vector of the same type.
2344 The first two operands of a '<tt>shufflevector</tt>' instruction are vectors
2345 with types that match each other and types that match the result of the
2346 instruction. The third argument is a shuffle mask, which has the same number
2347 of elements as the other vector type, but whose element type is always 'i32'.
2351 The shuffle mask operand is required to be a constant vector with either
2352 constant integer or undef values.
2358 The elements of the two input vectors are numbered from left to right across
2359 both of the vectors. The shuffle mask operand specifies, for each element of
2360 the result vector, which element of the two input registers the result element
2361 gets. The element selector may be undef (meaning "don't care") and the second
2362 operand may be undef if performing a shuffle from only one vector.
2368 %result = shufflevector <4 x i32> %v1, <4 x i32> %v2,
2369 <4 x i32> <i32 0, i32 4, i32 1, i32 5> <i>; yields <4 x i32></i>
2370 %result = shufflevector <4 x i32> %v1, <4 x i32> undef,
2371 <4 x i32> <i32 0, i32 1, i32 2, i32 3> <i>; yields <4 x i32></i> - Identity shuffle.
2376 <!-- ======================================================================= -->
2377 <div class="doc_subsection">
2378 <a name="memoryops">Memory Access and Addressing Operations</a>
2381 <div class="doc_text">
2383 <p>A key design point of an SSA-based representation is how it
2384 represents memory. In LLVM, no memory locations are in SSA form, which
2385 makes things very simple. This section describes how to read, write,
2386 allocate, and free memory in LLVM.</p>
2390 <!-- _______________________________________________________________________ -->
2391 <div class="doc_subsubsection">
2392 <a name="i_malloc">'<tt>malloc</tt>' Instruction</a>
2395 <div class="doc_text">
2400 <result> = malloc <type>[, i32 <NumElements>][, align <alignment>] <i>; yields {type*}:result</i>
2405 <p>The '<tt>malloc</tt>' instruction allocates memory from the system
2406 heap and returns a pointer to it.</p>
2410 <p>The '<tt>malloc</tt>' instruction allocates
2411 <tt>sizeof(<type>)*NumElements</tt>
2412 bytes of memory from the operating system and returns a pointer of the
2413 appropriate type to the program. If "NumElements" is specified, it is the
2414 number of elements allocated. If an alignment is specified, the value result
2415 of the allocation is guaranteed to be aligned to at least that boundary. If
2416 not specified, or if zero, the target can choose to align the allocation on any
2417 convenient boundary.</p>
2419 <p>'<tt>type</tt>' must be a sized type.</p>
2423 <p>Memory is allocated using the system "<tt>malloc</tt>" function, and
2424 a pointer is returned.</p>
2429 %array = malloc [4 x i8 ] <i>; yields {[%4 x i8]*}:array</i>
2431 %size = <a href="#i_add">add</a> i32 2, 2 <i>; yields {i32}:size = i32 4</i>
2432 %array1 = malloc i8, i32 4 <i>; yields {i8*}:array1</i>
2433 %array2 = malloc [12 x i8], i32 %size <i>; yields {[12 x i8]*}:array2</i>
2434 %array3 = malloc i32, i32 4, align 1024 <i>; yields {i32*}:array3</i>
2435 %array4 = malloc i32, align 1024 <i>; yields {i32*}:array4</i>
2439 <!-- _______________________________________________________________________ -->
2440 <div class="doc_subsubsection">
2441 <a name="i_free">'<tt>free</tt>' Instruction</a>
2444 <div class="doc_text">
2449 free <type> <value> <i>; yields {void}</i>
2454 <p>The '<tt>free</tt>' instruction returns memory back to the unused
2455 memory heap to be reallocated in the future.</p>
2459 <p>'<tt>value</tt>' shall be a pointer value that points to a value
2460 that was allocated with the '<tt><a href="#i_malloc">malloc</a></tt>'
2465 <p>Access to the memory pointed to by the pointer is no longer defined
2466 after this instruction executes.</p>
2471 %array = <a href="#i_malloc">malloc</a> [4 x i8] <i>; yields {[4 x i8]*}:array</i>
2472 free [4 x i8]* %array
2476 <!-- _______________________________________________________________________ -->
2477 <div class="doc_subsubsection">
2478 <a name="i_alloca">'<tt>alloca</tt>' Instruction</a>
2481 <div class="doc_text">
2486 <result> = alloca <type>[, i32 <NumElements>][, align <alignment>] <i>; yields {type*}:result</i>
2491 <p>The '<tt>alloca</tt>' instruction allocates memory on the current
2492 stack frame of the procedure that is live until the current function
2493 returns to its caller.</p>
2497 <p>The '<tt>alloca</tt>' instruction allocates <tt>sizeof(<type>)*NumElements</tt>
2498 bytes of memory on the runtime stack, returning a pointer of the
2499 appropriate type to the program. If "NumElements" is specified, it is the
2500 number of elements allocated. If an alignment is specified, the value result
2501 of the allocation is guaranteed to be aligned to at least that boundary. If
2502 not specified, or if zero, the target can choose to align the allocation on any
2503 convenient boundary.</p>
2505 <p>'<tt>type</tt>' may be any sized type.</p>
2509 <p>Memory is allocated; a pointer is returned. '<tt>alloca</tt>'d
2510 memory is automatically released when the function returns. The '<tt>alloca</tt>'
2511 instruction is commonly used to represent automatic variables that must
2512 have an address available. When the function returns (either with the <tt><a
2513 href="#i_ret">ret</a></tt> or <tt><a href="#i_unwind">unwind</a></tt>
2514 instructions), the memory is reclaimed.</p>
2519 %ptr = alloca i32 <i>; yields {i32*}:ptr</i>
2520 %ptr = alloca i32, i32 4 <i>; yields {i32*}:ptr</i>
2521 %ptr = alloca i32, i32 4, align 1024 <i>; yields {i32*}:ptr</i>
2522 %ptr = alloca i32, align 1024 <i>; yields {i32*}:ptr</i>
2526 <!-- _______________________________________________________________________ -->
2527 <div class="doc_subsubsection"> <a name="i_load">'<tt>load</tt>'
2528 Instruction</a> </div>
2529 <div class="doc_text">
2531 <pre> <result> = load <ty>* <pointer><br> <result> = volatile load <ty>* <pointer><br></pre>
2533 <p>The '<tt>load</tt>' instruction is used to read from memory.</p>
2535 <p>The argument to the '<tt>load</tt>' instruction specifies the memory
2536 address from which to load. The pointer must point to a <a
2537 href="#t_firstclass">first class</a> type. If the <tt>load</tt> is
2538 marked as <tt>volatile</tt>, then the optimizer is not allowed to modify
2539 the number or order of execution of this <tt>load</tt> with other
2540 volatile <tt>load</tt> and <tt><a href="#i_store">store</a></tt>
2543 <p>The location of memory pointed to is loaded.</p>
2545 <pre> %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
2547 href="#i_store">store</a> i32 3, i32* %ptr <i>; yields {void}</i>
2548 %val = load i32* %ptr <i>; yields {i32}:val = i32 3</i>
2551 <!-- _______________________________________________________________________ -->
2552 <div class="doc_subsubsection"> <a name="i_store">'<tt>store</tt>'
2553 Instruction</a> </div>
2554 <div class="doc_text">
2556 <pre> store <ty> <value>, <ty>* <pointer> <i>; yields {void}</i>
2557 volatile store <ty> <value>, <ty>* <pointer> <i>; yields {void}</i>
2560 <p>The '<tt>store</tt>' instruction is used to write to memory.</p>
2562 <p>There are two arguments to the '<tt>store</tt>' instruction: a value
2563 to store and an address in which to store it. The type of the '<tt><pointer></tt>'
2564 operand must be a pointer to the type of the '<tt><value></tt>'
2565 operand. If the <tt>store</tt> is marked as <tt>volatile</tt>, then the
2566 optimizer is not allowed to modify the number or order of execution of
2567 this <tt>store</tt> with other volatile <tt>load</tt> and <tt><a
2568 href="#i_store">store</a></tt> instructions.</p>
2570 <p>The contents of memory are updated to contain '<tt><value></tt>'
2571 at the location specified by the '<tt><pointer></tt>' operand.</p>
2573 <pre> %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
2575 href="#i_store">store</a> i32 3, i32* %ptr <i>; yields {void}</i>
2576 %val = load i32* %ptr <i>; yields {i32}:val = i32 3</i>
2580 <!-- _______________________________________________________________________ -->
2581 <div class="doc_subsubsection">
2582 <a name="i_getelementptr">'<tt>getelementptr</tt>' Instruction</a>
2585 <div class="doc_text">
2588 <result> = getelementptr <ty>* <ptrval>{, <ty> <idx>}*
2594 The '<tt>getelementptr</tt>' instruction is used to get the address of a
2595 subelement of an aggregate data structure.</p>
2599 <p>This instruction takes a list of integer operands that indicate what
2600 elements of the aggregate object to index to. The actual types of the arguments
2601 provided depend on the type of the first pointer argument. The
2602 '<tt>getelementptr</tt>' instruction is used to index down through the type
2603 levels of a structure or to a specific index in an array. When indexing into a
2604 structure, only <tt>i32</tt> integer constants are allowed. When indexing
2605 into an array or pointer, only integers of 32 or 64 bits are allowed, and will
2606 be sign extended to 64-bit values.</p>
2608 <p>For example, let's consider a C code fragment and how it gets
2609 compiled to LLVM:</p>
2623 define i32 *foo(struct ST *s) {
2624 return &s[1].Z.B[5][13];
2628 <p>The LLVM code generated by the GCC frontend is:</p>
2631 %RT = type { i8 , [10 x [20 x i32]], i8 }
2632 %ST = type { i32, double, %RT }
2636 define i32* %foo(%ST* %s) {
2638 %reg = getelementptr %ST* %s, i32 1, i32 2, i32 1, i32 5, i32 13
2645 <p>The index types specified for the '<tt>getelementptr</tt>' instruction depend
2646 on the pointer type that is being indexed into. <a href="#t_pointer">Pointer</a>
2647 and <a href="#t_array">array</a> types can use a 32-bit or 64-bit
2648 <a href="#t_integer">integer</a> type but the value will always be sign extended
2649 to 64-bits. <a href="#t_struct">Structure</a> types, require <tt>i32</tt>
2650 <b>constants</b>.</p>
2652 <p>In the example above, the first index is indexing into the '<tt>%ST*</tt>'
2653 type, which is a pointer, yielding a '<tt>%ST</tt>' = '<tt>{ i32, double, %RT
2654 }</tt>' type, a structure. The second index indexes into the third element of
2655 the structure, yielding a '<tt>%RT</tt>' = '<tt>{ i8 , [10 x [20 x i32]],
2656 i8 }</tt>' type, another structure. The third index indexes into the second
2657 element of the structure, yielding a '<tt>[10 x [20 x i32]]</tt>' type, an
2658 array. The two dimensions of the array are subscripted into, yielding an
2659 '<tt>i32</tt>' type. The '<tt>getelementptr</tt>' instruction returns a pointer
2660 to this element, thus computing a value of '<tt>i32*</tt>' type.</p>
2662 <p>Note that it is perfectly legal to index partially through a
2663 structure, returning a pointer to an inner element. Because of this,
2664 the LLVM code for the given testcase is equivalent to:</p>
2667 define i32* %foo(%ST* %s) {
2668 %t1 = getelementptr %ST* %s, i32 1 <i>; yields %ST*:%t1</i>
2669 %t2 = getelementptr %ST* %t1, i32 0, i32 2 <i>; yields %RT*:%t2</i>
2670 %t3 = getelementptr %RT* %t2, i32 0, i32 1 <i>; yields [10 x [20 x i32]]*:%t3</i>
2671 %t4 = getelementptr [10 x [20 x i32]]* %t3, i32 0, i32 5 <i>; yields [20 x i32]*:%t4</i>
2672 %t5 = getelementptr [20 x i32]* %t4, i32 0, i32 13 <i>; yields i32*:%t5</i>
2677 <p>Note that it is undefined to access an array out of bounds: array and
2678 pointer indexes must always be within the defined bounds of the array type.
2679 The one exception for this rules is zero length arrays. These arrays are
2680 defined to be accessible as variable length arrays, which requires access
2681 beyond the zero'th element.</p>
2683 <p>The getelementptr instruction is often confusing. For some more insight
2684 into how it works, see <a href="GetElementPtr.html">the getelementptr
2690 <i>; yields [12 x i8]*:aptr</i>
2691 %aptr = getelementptr {i32, [12 x i8]}* %sptr, i64 0, i32 1
2695 <!-- ======================================================================= -->
2696 <div class="doc_subsection"> <a name="convertops">Conversion Operations</a>
2698 <div class="doc_text">
2699 <p>The instructions in this category are the conversion instructions (casting)
2700 which all take a single operand and a type. They perform various bit conversions
2704 <!-- _______________________________________________________________________ -->
2705 <div class="doc_subsubsection">
2706 <a name="i_trunc">'<tt>trunc .. to</tt>' Instruction</a>
2708 <div class="doc_text">
2712 <result> = trunc <ty> <value> to <ty2> <i>; yields ty2</i>
2717 The '<tt>trunc</tt>' instruction truncates its operand to the type <tt>ty2</tt>.
2722 The '<tt>trunc</tt>' instruction takes a <tt>value</tt> to trunc, which must
2723 be an <a href="#t_integer">integer</a> type, and a type that specifies the size
2724 and type of the result, which must be an <a href="#t_integer">integer</a>
2725 type. The bit size of <tt>value</tt> must be larger than the bit size of
2726 <tt>ty2</tt>. Equal sized types are not allowed.</p>
2730 The '<tt>trunc</tt>' instruction truncates the high order bits in <tt>value</tt>
2731 and converts the remaining bits to <tt>ty2</tt>. Since the source size must be
2732 larger than the destination size, <tt>trunc</tt> cannot be a <i>no-op cast</i>.
2733 It will always truncate bits.</p>
2737 %X = trunc i32 257 to i8 <i>; yields i8:1</i>
2738 %Y = trunc i32 123 to i1 <i>; yields i1:true</i>
2739 %Y = trunc i32 122 to i1 <i>; yields i1:false</i>
2743 <!-- _______________________________________________________________________ -->
2744 <div class="doc_subsubsection">
2745 <a name="i_zext">'<tt>zext .. to</tt>' Instruction</a>
2747 <div class="doc_text">
2751 <result> = zext <ty> <value> to <ty2> <i>; yields ty2</i>
2755 <p>The '<tt>zext</tt>' instruction zero extends its operand to type
2760 <p>The '<tt>zext</tt>' instruction takes a value to cast, which must be of
2761 <a href="#t_integer">integer</a> type, and a type to cast it to, which must
2762 also be of <a href="#t_integer">integer</a> type. The bit size of the
2763 <tt>value</tt> must be smaller than the bit size of the destination type,
2767 <p>The <tt>zext</tt> fills the high order bits of the <tt>value</tt> with zero
2768 bits until it reaches the size of the destination type, <tt>ty2</tt>. When the
2769 the operand and the type are the same size, no bit filling is done and the
2770 cast is considered a <i>no-op cast</i> because no bits change (only the type
2773 <p>When zero extending from i1, the result will always be either 0 or 1.</p>
2777 %X = zext i32 257 to i64 <i>; yields i64:257</i>
2778 %Y = zext i1 true to i32 <i>; yields i32:1</i>
2782 <!-- _______________________________________________________________________ -->
2783 <div class="doc_subsubsection">
2784 <a name="i_sext">'<tt>sext .. to</tt>' Instruction</a>
2786 <div class="doc_text">
2790 <result> = sext <ty> <value> to <ty2> <i>; yields ty2</i>
2794 <p>The '<tt>sext</tt>' sign extends <tt>value</tt> to the type <tt>ty2</tt>.</p>
2798 The '<tt>sext</tt>' instruction takes a value to cast, which must be of
2799 <a href="#t_integer">integer</a> type, and a type to cast it to, which must
2800 also be of <a href="#t_integer">integer</a> type. The bit size of the
2801 <tt>value</tt> must be smaller than the bit size of the destination type,
2806 The '<tt>sext</tt>' instruction performs a sign extension by copying the sign
2807 bit (highest order bit) of the <tt>value</tt> until it reaches the bit size of
2808 the type <tt>ty2</tt>. When the the operand and the type are the same size,
2809 no bit filling is done and the cast is considered a <i>no-op cast</i> because
2810 no bits change (only the type changes).</p>
2812 <p>When sign extending from i1, the extension always results in -1 or 0.</p>
2816 %X = sext i8 -1 to i16 <i>; yields i16 :65535</i>
2817 %Y = sext i1 true to i32 <i>; yields i32:-1</i>
2821 <!-- _______________________________________________________________________ -->
2822 <div class="doc_subsubsection">
2823 <a name="i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a>
2826 <div class="doc_text">
2831 <result> = fptrunc <ty> <value> to <ty2> <i>; yields ty2</i>
2835 <p>The '<tt>fptrunc</tt>' instruction truncates <tt>value</tt> to type
2840 <p>The '<tt>fptrunc</tt>' instruction takes a <a href="#t_floating">floating
2841 point</a> value to cast and a <a href="#t_floating">floating point</a> type to
2842 cast it to. The size of <tt>value</tt> must be larger than the size of
2843 <tt>ty2</tt>. This implies that <tt>fptrunc</tt> cannot be used to make a
2844 <i>no-op cast</i>.</p>
2847 <p> The '<tt>fptrunc</tt>' instruction truncates a <tt>value</tt> from a larger
2848 <a href="#t_floating">floating point</a> type to a smaller
2849 <a href="#t_floating">floating point</a> type. If the value cannot fit within
2850 the destination type, <tt>ty2</tt>, then the results are undefined.</p>
2854 %X = fptrunc double 123.0 to float <i>; yields float:123.0</i>
2855 %Y = fptrunc double 1.0E+300 to float <i>; yields undefined</i>
2859 <!-- _______________________________________________________________________ -->
2860 <div class="doc_subsubsection">
2861 <a name="i_fpext">'<tt>fpext .. to</tt>' Instruction</a>
2863 <div class="doc_text">
2867 <result> = fpext <ty> <value> to <ty2> <i>; yields ty2</i>
2871 <p>The '<tt>fpext</tt>' extends a floating point <tt>value</tt> to a larger
2872 floating point value.</p>
2875 <p>The '<tt>fpext</tt>' instruction takes a
2876 <a href="#t_floating">floating point</a> <tt>value</tt> to cast,
2877 and a <a href="#t_floating">floating point</a> type to cast it to. The source
2878 type must be smaller than the destination type.</p>
2881 <p>The '<tt>fpext</tt>' instruction extends the <tt>value</tt> from a smaller
2882 <a href="t_floating">floating point</a> type to a larger
2883 <a href="t_floating">floating point</a> type. The <tt>fpext</tt> cannot be
2884 used to make a <i>no-op cast</i> because it always changes bits. Use
2885 <tt>bitcast</tt> to make a <i>no-op cast</i> for a floating point cast.</p>
2889 %X = fpext float 3.1415 to double <i>; yields double:3.1415</i>
2890 %Y = fpext float 1.0 to float <i>; yields float:1.0 (no-op)</i>
2894 <!-- _______________________________________________________________________ -->
2895 <div class="doc_subsubsection">
2896 <a name="i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a>
2898 <div class="doc_text">
2902 <result> = fp2uint <ty> <value> to <ty2> <i>; yields ty2</i>
2906 <p>The '<tt>fp2uint</tt>' converts a floating point <tt>value</tt> to its
2907 unsigned integer equivalent of type <tt>ty2</tt>.
2911 <p>The '<tt>fp2uint</tt>' instruction takes a value to cast, which must be a
2912 <a href="#t_floating">floating point</a> value, and a type to cast it to, which
2913 must be an <a href="#t_integer">integer</a> type.</p>
2916 <p> The '<tt>fp2uint</tt>' instruction converts its
2917 <a href="#t_floating">floating point</a> operand into the nearest (rounding
2918 towards zero) unsigned integer value. If the value cannot fit in <tt>ty2</tt>,
2919 the results are undefined.</p>
2921 <p>When converting to i1, the conversion is done as a comparison against
2922 zero. If the <tt>value</tt> was zero, the i1 result will be <tt>false</tt>.
2923 If the <tt>value</tt> was non-zero, the i1 result will be <tt>true</tt>.</p>
2927 %X = fp2uint double 123.0 to i32 <i>; yields i32:123</i>
2928 %Y = fp2uint float 1.0E+300 to i1 <i>; yields i1:true</i>
2929 %X = fp2uint float 1.04E+17 to i8 <i>; yields undefined:1</i>
2933 <!-- _______________________________________________________________________ -->
2934 <div class="doc_subsubsection">
2935 <a name="i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a>
2937 <div class="doc_text">
2941 <result> = fptosi <ty> <value> to <ty2> <i>; yields ty2</i>
2945 <p>The '<tt>fptosi</tt>' instruction converts
2946 <a href="#t_floating">floating point</a> <tt>value</tt> to type <tt>ty2</tt>.
2951 <p> The '<tt>fptosi</tt>' instruction takes a value to cast, which must be a
2952 <a href="#t_floating">floating point</a> value, and a type to cast it to, which
2953 must also be an <a href="#t_integer">integer</a> type.</p>
2956 <p>The '<tt>fptosi</tt>' instruction converts its
2957 <a href="#t_floating">floating point</a> operand into the nearest (rounding
2958 towards zero) signed integer value. If the value cannot fit in <tt>ty2</tt>,
2959 the results are undefined.</p>
2961 <p>When converting to i1, the conversion is done as a comparison against
2962 zero. If the <tt>value</tt> was zero, the i1 result will be <tt>false</tt>.
2963 If the <tt>value</tt> was non-zero, the i1 result will be <tt>true</tt>.</p>
2967 %X = fptosi double -123.0 to i32 <i>; yields i32:-123</i>
2968 %Y = fptosi float 1.0E-247 to i1 <i>; yields i1:true</i>
2969 %X = fptosi float 1.04E+17 to i8 <i>; yields undefined:1</i>
2973 <!-- _______________________________________________________________________ -->
2974 <div class="doc_subsubsection">
2975 <a name="i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a>
2977 <div class="doc_text">
2981 <result> = uitofp <ty> <value> to <ty2> <i>; yields ty2</i>
2985 <p>The '<tt>uitofp</tt>' instruction regards <tt>value</tt> as an unsigned
2986 integer and converts that value to the <tt>ty2</tt> type.</p>
2990 <p>The '<tt>uitofp</tt>' instruction takes a value to cast, which must be an
2991 <a href="#t_integer">integer</a> value, and a type to cast it to, which must
2992 be a <a href="#t_floating">floating point</a> type.</p>
2995 <p>The '<tt>uitofp</tt>' instruction interprets its operand as an unsigned
2996 integer quantity and converts it to the corresponding floating point value. If
2997 the value cannot fit in the floating point value, the results are undefined.</p>
3002 %X = uitofp i32 257 to float <i>; yields float:257.0</i>
3003 %Y = uitofp i8 -1 to double <i>; yields double:255.0</i>
3007 <!-- _______________________________________________________________________ -->
3008 <div class="doc_subsubsection">
3009 <a name="i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a>
3011 <div class="doc_text">
3015 <result> = sitofp <ty> <value> to <ty2> <i>; yields ty2</i>
3019 <p>The '<tt>sitofp</tt>' instruction regards <tt>value</tt> as a signed
3020 integer and converts that value to the <tt>ty2</tt> type.</p>
3023 <p>The '<tt>sitofp</tt>' instruction takes a value to cast, which must be an
3024 <a href="#t_integer">integer</a> value, and a type to cast it to, which must be
3025 a <a href="#t_floating">floating point</a> type.</p>
3028 <p>The '<tt>sitofp</tt>' instruction interprets its operand as a signed
3029 integer quantity and converts it to the corresponding floating point value. If
3030 the value cannot fit in the floating point value, the results are undefined.</p>
3034 %X = sitofp i32 257 to float <i>; yields float:257.0</i>
3035 %Y = sitofp i8 -1 to double <i>; yields double:-1.0</i>
3039 <!-- _______________________________________________________________________ -->
3040 <div class="doc_subsubsection">
3041 <a name="i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a>
3043 <div class="doc_text">
3047 <result> = ptrtoint <ty> <value> to <ty2> <i>; yields ty2</i>
3051 <p>The '<tt>ptrtoint</tt>' instruction converts the pointer <tt>value</tt> to
3052 the integer type <tt>ty2</tt>.</p>
3055 <p>The '<tt>ptrtoint</tt>' instruction takes a <tt>value</tt> to cast, which
3056 must be a <a href="t_pointer">pointer</a> value, and a type to cast it to
3057 <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a> type.
3060 <p>The '<tt>ptrtoint</tt>' instruction converts <tt>value</tt> to integer type
3061 <tt>ty2</tt> by interpreting the pointer value as an integer and either
3062 truncating or zero extending that value to the size of the integer type. If
3063 <tt>value</tt> is smaller than <tt>ty2</tt> then a zero extension is done. If
3064 <tt>value</tt> is larger than <tt>ty2</tt> then a truncation is done. If they
3065 are the same size, then nothing is done (<i>no-op cast</i>).</p>
3069 %X = ptrtoint i32* %X to i8 <i>; yields truncation on 32-bit</i>
3070 %Y = ptrtoint i32* %x to i64 <i>; yields zero extend on 32-bit</i>
3074 <!-- _______________________________________________________________________ -->
3075 <div class="doc_subsubsection">
3076 <a name="i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a>
3078 <div class="doc_text">
3082 <result> = inttoptr <ty> <value> to <ty2> <i>; yields ty2</i>
3086 <p>The '<tt>inttoptr</tt>' instruction converts an integer <tt>value</tt> to
3087 a pointer type, <tt>ty2</tt>.</p>
3090 <p>The '<tt>inttoptr</tt>' instruction takes an <a href="i_integer">integer</a>
3091 value to cast, and a type to cast it to, which must be a
3092 <a href="#t_pointer">pointer</a> type.
3095 <p>The '<tt>inttoptr</tt>' instruction converts <tt>value</tt> to type
3096 <tt>ty2</tt> by applying either a zero extension or a truncation depending on
3097 the size of the integer <tt>value</tt>. If <tt>value</tt> is larger than the
3098 size of a pointer then a truncation is done. If <tt>value</tt> is smaller than
3099 the size of a pointer then a zero extension is done. If they are the same size,
3100 nothing is done (<i>no-op cast</i>).</p>
3104 %X = inttoptr i32 255 to i32* <i>; yields zero extend on 64-bit</i>
3105 %X = inttoptr i32 255 to i32* <i>; yields no-op on 32-bit </i>
3106 %Y = inttoptr i16 0 to i32* <i>; yields zero extend on 32-bit</i>
3110 <!-- _______________________________________________________________________ -->
3111 <div class="doc_subsubsection">
3112 <a name="i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a>
3114 <div class="doc_text">
3118 <result> = bitcast <ty> <value> to <ty2> <i>; yields ty2</i>
3122 <p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
3123 <tt>ty2</tt> without changing any bits.</p>
3126 <p>The '<tt>bitcast</tt>' instruction takes a value to cast, which must be
3127 a first class value, and a type to cast it to, which must also be a <a
3128 href="#t_firstclass">first class</a> type. The bit sizes of <tt>value</tt>
3129 and the destination type, <tt>ty2</tt>, must be identical. If the source
3130 type is a pointer, the destination type must also be a pointer.</p>
3133 <p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
3134 <tt>ty2</tt>. It is always a <i>no-op cast</i> because no bits change with
3135 this conversion. The conversion is done as if the <tt>value</tt> had been
3136 stored to memory and read back as type <tt>ty2</tt>. Pointer types may only be
3137 converted to other pointer types with this instruction. To convert pointers to
3138 other types, use the <a href="#i_inttoptr">inttoptr</a> or
3139 <a href="#i_ptrtoint">ptrtoint</a> instructions first.</p>
3143 %X = bitcast i8 255 to i8 <i>; yields i8 :-1</i>
3144 %Y = bitcast i32* %x to sint* <i>; yields sint*:%x</i>
3145 %Z = bitcast <2xint> %V to i64; <i>; yields i64: %V</i>
3149 <!-- ======================================================================= -->
3150 <div class="doc_subsection"> <a name="otherops">Other Operations</a> </div>
3151 <div class="doc_text">
3152 <p>The instructions in this category are the "miscellaneous"
3153 instructions, which defy better classification.</p>
3156 <!-- _______________________________________________________________________ -->
3157 <div class="doc_subsubsection"><a name="i_icmp">'<tt>icmp</tt>' Instruction</a>
3159 <div class="doc_text">
3161 <pre> <result> = icmp <cond> <ty> <var1>, <var2>
3162 <i>; yields {i1}:result</i>
3165 <p>The '<tt>icmp</tt>' instruction returns a boolean value based on comparison
3166 of its two integer operands.</p>
3168 <p>The '<tt>icmp</tt>' instruction takes three operands. The first operand is
3169 the condition code which indicates the kind of comparison to perform. It is not
3170 a value, just a keyword. The possibilities for the condition code are:
3172 <li><tt>eq</tt>: equal</li>
3173 <li><tt>ne</tt>: not equal </li>
3174 <li><tt>ugt</tt>: unsigned greater than</li>
3175 <li><tt>uge</tt>: unsigned greater or equal</li>
3176 <li><tt>ult</tt>: unsigned less than</li>
3177 <li><tt>ule</tt>: unsigned less or equal</li>
3178 <li><tt>sgt</tt>: signed greater than</li>
3179 <li><tt>sge</tt>: signed greater or equal</li>
3180 <li><tt>slt</tt>: signed less than</li>
3181 <li><tt>sle</tt>: signed less or equal</li>
3183 <p>The remaining two arguments must be <a href="#t_integer">integer</a> or
3184 <a href="#t_pointer">pointer</a> typed. They must also be identical types.</p>
3186 <p>The '<tt>icmp</tt>' compares <tt>var1</tt> and <tt>var2</tt> according to
3187 the condition code given as <tt>cond</tt>. The comparison performed always
3188 yields a <a href="#t_primitive">i1</a> result, as follows:
3190 <li><tt>eq</tt>: yields <tt>true</tt> if the operands are equal,
3191 <tt>false</tt> otherwise. No sign interpretation is necessary or performed.
3193 <li><tt>ne</tt>: yields <tt>true</tt> if the operands are unequal,
3194 <tt>false</tt> otherwise. No sign interpretation is necessary or performed.
3195 <li><tt>ugt</tt>: interprets the operands as unsigned values and yields
3196 <tt>true</tt> if <tt>var1</tt> is greater than <tt>var2</tt>.</li>
3197 <li><tt>uge</tt>: interprets the operands as unsigned values and yields
3198 <tt>true</tt> if <tt>var1</tt> is greater than or equal to <tt>var2</tt>.</li>
3199 <li><tt>ult</tt>: interprets the operands as unsigned values and yields
3200 <tt>true</tt> if <tt>var1</tt> is less than <tt>var2</tt>.</li>
3201 <li><tt>ule</tt>: interprets the operands as unsigned values and yields
3202 <tt>true</tt> if <tt>var1</tt> is less than or equal to <tt>var2</tt>.</li>
3203 <li><tt>sgt</tt>: interprets the operands as signed values and yields
3204 <tt>true</tt> if <tt>var1</tt> is greater than <tt>var2</tt>.</li>
3205 <li><tt>sge</tt>: interprets the operands as signed values and yields
3206 <tt>true</tt> if <tt>var1</tt> is greater than or equal to <tt>var2</tt>.</li>
3207 <li><tt>slt</tt>: interprets the operands as signed values and yields
3208 <tt>true</tt> if <tt>var1</tt> is less than <tt>var2</tt>.</li>
3209 <li><tt>sle</tt>: interprets the operands as signed values and yields
3210 <tt>true</tt> if <tt>var1</tt> is less than or equal to <tt>var2</tt>.</li>
3212 <p>If the operands are <a href="#t_pointer">pointer</a> typed, the pointer
3213 values are treated as integers and then compared.</p>
3216 <pre> <result> = icmp eq i32 4, 5 <i>; yields: result=false</i>
3217 <result> = icmp ne float* %X, %X <i>; yields: result=false</i>
3218 <result> = icmp ult i16 4, 5 <i>; yields: result=true</i>
3219 <result> = icmp sgt i16 4, 5 <i>; yields: result=false</i>
3220 <result> = icmp ule i16 -4, 5 <i>; yields: result=false</i>
3221 <result> = icmp sge i16 4, 5 <i>; yields: result=false</i>
3225 <!-- _______________________________________________________________________ -->
3226 <div class="doc_subsubsection"><a name="i_fcmp">'<tt>fcmp</tt>' Instruction</a>
3228 <div class="doc_text">
3230 <pre> <result> = fcmp <cond> <ty> <var1>, <var2>
3231 <i>; yields {i1}:result</i>
3234 <p>The '<tt>fcmp</tt>' instruction returns a boolean value based on comparison
3235 of its floating point operands.</p>
3237 <p>The '<tt>fcmp</tt>' instruction takes three operands. The first operand is
3238 the condition code which indicates the kind of comparison to perform. It is not
3239 a value, just a keyword. The possibilities for the condition code are:
3241 <li><tt>false</tt>: no comparison, always returns false</li>
3242 <li><tt>oeq</tt>: ordered and equal</li>
3243 <li><tt>ogt</tt>: ordered and greater than </li>
3244 <li><tt>oge</tt>: ordered and greater than or equal</li>
3245 <li><tt>olt</tt>: ordered and less than </li>
3246 <li><tt>ole</tt>: ordered and less than or equal</li>
3247 <li><tt>one</tt>: ordered and not equal</li>
3248 <li><tt>ord</tt>: ordered (no nans)</li>
3249 <li><tt>ueq</tt>: unordered or equal</li>
3250 <li><tt>ugt</tt>: unordered or greater than </li>
3251 <li><tt>uge</tt>: unordered or greater than or equal</li>
3252 <li><tt>ult</tt>: unordered or less than </li>
3253 <li><tt>ule</tt>: unordered or less than or equal</li>
3254 <li><tt>une</tt>: unordered or not equal</li>
3255 <li><tt>uno</tt>: unordered (either nans)</li>
3256 <li><tt>true</tt>: no comparison, always returns true</li>
3258 <p>In the preceding, <i>ordered</i> means that neither operand is a QNAN while
3259 <i>unordered</i> means that either operand may be a QNAN.</p>
3260 <p>The <tt>val1</tt> and <tt>val2</tt> arguments must be
3261 <a href="#t_floating">floating point</a> typed. They must have identical
3263 <p>In the foregoing, <i>ordered</i> means that neither operand is a QNAN and
3264 <i>unordered</i> means that either operand is a QNAN.</p>
3266 <p>The '<tt>fcmp</tt>' compares <tt>var1</tt> and <tt>var2</tt> according to
3267 the condition code given as <tt>cond</tt>. The comparison performed always
3268 yields a <a href="#t_primitive">i1</a> result, as follows:
3270 <li><tt>false</tt>: always yields <tt>false</tt>, regardless of operands.</li>
3271 <li><tt>oeq</tt>: yields <tt>true</tt> if both operands are not a QNAN and
3272 <tt>var1</tt> is equal to <tt>var2</tt>.</li>
3273 <li><tt>ogt</tt>: yields <tt>true</tt> if both operands are not a QNAN and
3274 <tt>var1</tt> is greather than <tt>var2</tt>.</li>
3275 <li><tt>oge</tt>: yields <tt>true</tt> if both operands are not a QNAN and
3276 <tt>var1</tt> is greater than or equal to <tt>var2</tt>.</li>
3277 <li><tt>olt</tt>: yields <tt>true</tt> if both operands are not a QNAN and
3278 <tt>var1</tt> is less than <tt>var2</tt>.</li>
3279 <li><tt>ole</tt>: yields <tt>true</tt> if both operands are not a QNAN and
3280 <tt>var1</tt> is less than or equal to <tt>var2</tt>.</li>
3281 <li><tt>one</tt>: yields <tt>true</tt> if both operands are not a QNAN and
3282 <tt>var1</tt> is not equal to <tt>var2</tt>.</li>
3283 <li><tt>ord</tt>: yields <tt>true</tt> if both operands are not a QNAN.</li>
3284 <li><tt>ueq</tt>: yields <tt>true</tt> if either operand is a QNAN or
3285 <tt>var1</tt> is equal to <tt>var2</tt>.</li>
3286 <li><tt>ugt</tt>: yields <tt>true</tt> if either operand is a QNAN or
3287 <tt>var1</tt> is greater than <tt>var2</tt>.</li>
3288 <li><tt>uge</tt>: yields <tt>true</tt> if either operand is a QNAN or
3289 <tt>var1</tt> is greater than or equal to <tt>var2</tt>.</li>
3290 <li><tt>ult</tt>: yields <tt>true</tt> if either operand is a QNAN or
3291 <tt>var1</tt> is less than <tt>var2</tt>.</li>
3292 <li><tt>ule</tt>: yields <tt>true</tt> if either operand is a QNAN or
3293 <tt>var1</tt> is less than or equal to <tt>var2</tt>.</li>
3294 <li><tt>une</tt>: yields <tt>true</tt> if either operand is a QNAN or
3295 <tt>var1</tt> is not equal to <tt>var2</tt>.</li>
3296 <li><tt>uno</tt>: yields <tt>true</tt> if either operand is a QNAN.</li>
3297 <li><tt>true</tt>: always yields <tt>true</tt>, regardless of operands.</li>
3301 <pre> <result> = fcmp oeq float 4.0, 5.0 <i>; yields: result=false</i>
3302 <result> = icmp one float 4.0, 5.0 <i>; yields: result=true</i>
3303 <result> = icmp olt float 4.0, 5.0 <i>; yields: result=true</i>
3304 <result> = icmp ueq double 1.0, 2.0 <i>; yields: result=false</i>
3308 <!-- _______________________________________________________________________ -->
3309 <div class="doc_subsubsection"> <a name="i_phi">'<tt>phi</tt>'
3310 Instruction</a> </div>
3311 <div class="doc_text">
3313 <pre> <result> = phi <ty> [ <val0>, <label0>], ...<br></pre>
3315 <p>The '<tt>phi</tt>' instruction is used to implement the φ node in
3316 the SSA graph representing the function.</p>
3318 <p>The type of the incoming values are specified with the first type
3319 field. After this, the '<tt>phi</tt>' instruction takes a list of pairs
3320 as arguments, with one pair for each predecessor basic block of the
3321 current block. Only values of <a href="#t_firstclass">first class</a>
3322 type may be used as the value arguments to the PHI node. Only labels
3323 may be used as the label arguments.</p>
3324 <p>There must be no non-phi instructions between the start of a basic
3325 block and the PHI instructions: i.e. PHI instructions must be first in
3328 <p>At runtime, the '<tt>phi</tt>' instruction logically takes on the
3329 value specified by the parameter, depending on which basic block we
3330 came from in the last <a href="#terminators">terminator</a> instruction.</p>
3332 <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>
3335 <!-- _______________________________________________________________________ -->
3336 <div class="doc_subsubsection">
3337 <a name="i_select">'<tt>select</tt>' Instruction</a>
3340 <div class="doc_text">
3345 <result> = select i1 <cond>, <ty> <val1>, <ty> <val2> <i>; yields ty</i>
3351 The '<tt>select</tt>' instruction is used to choose one value based on a
3352 condition, without branching.
3359 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.
3365 If the boolean condition evaluates to true, the instruction returns the first
3366 value argument; otherwise, it returns the second value argument.
3372 %X = select i1 true, i8 17, i8 42 <i>; yields i8:17</i>
3377 <!-- _______________________________________________________________________ -->
3378 <div class="doc_subsubsection">
3379 <a name="i_call">'<tt>call</tt>' Instruction</a>
3382 <div class="doc_text">
3386 <result> = [tail] call [<a href="#callingconv">cconv</a>] <ty>* <fnptrval>(<param list>)
3391 <p>The '<tt>call</tt>' instruction represents a simple function call.</p>
3395 <p>This instruction requires several arguments:</p>
3399 <p>The optional "tail" marker indicates whether the callee function accesses
3400 any allocas or varargs in the caller. If the "tail" marker is present, the
3401 function call is eligible for tail call optimization. Note that calls may
3402 be marked "tail" even if they do not occur before a <a
3403 href="#i_ret"><tt>ret</tt></a> instruction.
3406 <p>The optional "cconv" marker indicates which <a href="callingconv">calling
3407 convention</a> the call should use. If none is specified, the call defaults
3408 to using C calling conventions.
3411 <p>'<tt>ty</tt>': shall be the signature of the pointer to function value
3412 being invoked. The argument types must match the types implied by this
3413 signature. This type can be omitted if the function is not varargs and
3414 if the function type does not return a pointer to a function.</p>
3417 <p>'<tt>fnptrval</tt>': An LLVM value containing a pointer to a function to
3418 be invoked. In most cases, this is a direct function invocation, but
3419 indirect <tt>call</tt>s are just as possible, calling an arbitrary pointer
3420 to function value.</p>
3423 <p>'<tt>function args</tt>': argument list whose types match the
3424 function signature argument types. All arguments must be of
3425 <a href="#t_firstclass">first class</a> type. If the function signature
3426 indicates the function accepts a variable number of arguments, the extra
3427 arguments can be specified.</p>
3433 <p>The '<tt>call</tt>' instruction is used to cause control flow to
3434 transfer to a specified function, with its incoming arguments bound to
3435 the specified values. Upon a '<tt><a href="#i_ret">ret</a></tt>'
3436 instruction in the called function, control flow continues with the
3437 instruction after the function call, and the return value of the
3438 function is bound to the result argument. This is a simpler case of
3439 the <a href="#i_invoke">invoke</a> instruction.</p>
3444 %retval = call i32 %test(i32 %argc)
3445 call i32(i8 *, ...) *%printf(i8 * %msg, i32 12, i8 42);
3446 %X = tail call i32 %foo()
3447 %Y = tail call <a href="#callingconv">fastcc</a> i32 %foo()
3452 <!-- _______________________________________________________________________ -->
3453 <div class="doc_subsubsection">
3454 <a name="i_va_arg">'<tt>va_arg</tt>' Instruction</a>
3457 <div class="doc_text">
3462 <resultval> = va_arg <va_list*> <arglist>, <argty>
3467 <p>The '<tt>va_arg</tt>' instruction is used to access arguments passed through
3468 the "variable argument" area of a function call. It is used to implement the
3469 <tt>va_arg</tt> macro in C.</p>
3473 <p>This instruction takes a <tt>va_list*</tt> value and the type of
3474 the argument. It returns a value of the specified argument type and
3475 increments the <tt>va_list</tt> to point to the next argument. Again, the
3476 actual type of <tt>va_list</tt> is target specific.</p>
3480 <p>The '<tt>va_arg</tt>' instruction loads an argument of the specified
3481 type from the specified <tt>va_list</tt> and causes the
3482 <tt>va_list</tt> to point to the next argument. For more information,
3483 see the variable argument handling <a href="#int_varargs">Intrinsic
3486 <p>It is legal for this instruction to be called in a function which does not
3487 take a variable number of arguments, for example, the <tt>vfprintf</tt>
3490 <p><tt>va_arg</tt> is an LLVM instruction instead of an <a
3491 href="#intrinsics">intrinsic function</a> because it takes a type as an
3496 <p>See the <a href="#int_varargs">variable argument processing</a> section.</p>
3500 <!-- *********************************************************************** -->
3501 <div class="doc_section"> <a name="intrinsics">Intrinsic Functions</a> </div>
3502 <!-- *********************************************************************** -->
3504 <div class="doc_text">
3506 <p>LLVM supports the notion of an "intrinsic function". These functions have
3507 well known names and semantics and are required to follow certain
3508 restrictions. Overall, these instructions represent an extension mechanism for
3509 the LLVM language that does not require changing all of the transformations in
3510 LLVM to add to the language (or the bytecode reader/writer, the parser,
3513 <p>Intrinsic function names must all start with an "<tt>llvm.</tt>" prefix. This
3514 prefix is reserved in LLVM for intrinsic names; thus, functions may not be named
3515 this. Intrinsic functions must always be external functions: you cannot define
3516 the body of intrinsic functions. Intrinsic functions may only be used in call
3517 or invoke instructions: it is illegal to take the address of an intrinsic
3518 function. Additionally, because intrinsic functions are part of the LLVM
3519 language, it is required that they all be documented here if any are added.</p>
3522 <p>To learn how to add an intrinsic function, please see the <a
3523 href="ExtendingLLVM.html">Extending LLVM Guide</a>.
3528 <!-- ======================================================================= -->
3529 <div class="doc_subsection">
3530 <a name="int_varargs">Variable Argument Handling Intrinsics</a>
3533 <div class="doc_text">
3535 <p>Variable argument support is defined in LLVM with the <a
3536 href="#i_va_arg"><tt>va_arg</tt></a> instruction and these three
3537 intrinsic functions. These functions are related to the similarly
3538 named macros defined in the <tt><stdarg.h></tt> header file.</p>
3540 <p>All of these functions operate on arguments that use a
3541 target-specific value type "<tt>va_list</tt>". The LLVM assembly
3542 language reference manual does not define what this type is, so all
3543 transformations should be prepared to handle intrinsics with any type
3546 <p>This example shows how the <a href="#i_va_arg"><tt>va_arg</tt></a>
3547 instruction and the variable argument handling intrinsic functions are
3551 define i32 %test(i32 %X, ...) {
3552 ; Initialize variable argument processing
3554 %ap2 = bitcast i8** %ap to i8*
3555 call void %<a href="#i_va_start">llvm.va_start</a>(i8* %ap2)
3557 ; Read a single integer argument
3558 %tmp = va_arg i8 ** %ap, i32
3560 ; Demonstrate usage of llvm.va_copy and llvm.va_end
3562 %aq2 = bitcast i8** %aq to i8*
3563 call void %<a href="#i_va_copy">llvm.va_copy</a>(i8 *%aq2, i8* %ap2)
3564 call void %<a href="#i_va_end">llvm.va_end</a>(i8* %aq2)
3566 ; Stop processing of arguments.
3567 call void %<a href="#i_va_end">llvm.va_end</a>(i8* %ap2)
3573 <!-- _______________________________________________________________________ -->
3574 <div class="doc_subsubsection">
3575 <a name="i_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a>
3579 <div class="doc_text">
3581 <pre> declare void %llvm.va_start(i8* <arglist>)<br></pre>
3583 <P>The '<tt>llvm.va_start</tt>' intrinsic initializes
3584 <tt>*<arglist></tt> for subsequent use by <tt><a
3585 href="#i_va_arg">va_arg</a></tt>.</p>
3589 <P>The argument is a pointer to a <tt>va_list</tt> element to initialize.</p>
3593 <P>The '<tt>llvm.va_start</tt>' intrinsic works just like the <tt>va_start</tt>
3594 macro available in C. In a target-dependent way, it initializes the
3595 <tt>va_list</tt> element the argument points to, so that the next call to
3596 <tt>va_arg</tt> will produce the first variable argument passed to the function.
3597 Unlike the C <tt>va_start</tt> macro, this intrinsic does not need to know the
3598 last argument of the function, the compiler can figure that out.</p>
3602 <!-- _______________________________________________________________________ -->
3603 <div class="doc_subsubsection">
3604 <a name="i_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a>
3607 <div class="doc_text">
3609 <pre> declare void %llvm.va_end(i8* <arglist>)<br></pre>
3612 <p>The '<tt>llvm.va_end</tt>' intrinsic destroys <tt><arglist></tt>
3613 which has been initialized previously with <tt><a href="#i_va_start">llvm.va_start</a></tt>
3614 or <tt><a href="#i_va_copy">llvm.va_copy</a></tt>.</p>
3618 <p>The argument is a <tt>va_list</tt> to destroy.</p>
3622 <p>The '<tt>llvm.va_end</tt>' intrinsic works just like the <tt>va_end</tt>
3623 macro available in C. In a target-dependent way, it destroys the <tt>va_list</tt>.
3624 Calls to <a href="#i_va_start"><tt>llvm.va_start</tt></a> and <a
3625 href="#i_va_copy"><tt>llvm.va_copy</tt></a> must be matched exactly
3626 with calls to <tt>llvm.va_end</tt>.</p>
3630 <!-- _______________________________________________________________________ -->
3631 <div class="doc_subsubsection">
3632 <a name="i_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a>
3635 <div class="doc_text">
3640 declare void %llvm.va_copy(i8* <destarglist>, i8* <srcarglist>)
3645 <p>The '<tt>llvm.va_copy</tt>' intrinsic copies the current argument position from
3646 the source argument list to the destination argument list.</p>
3650 <p>The first argument is a pointer to a <tt>va_list</tt> element to initialize.
3651 The second argument is a pointer to a <tt>va_list</tt> element to copy from.</p>
3656 <p>The '<tt>llvm.va_copy</tt>' intrinsic works just like the <tt>va_copy</tt> macro
3657 available in C. In a target-dependent way, it copies the source
3658 <tt>va_list</tt> element into the destination list. This intrinsic is necessary
3659 because the <tt><a href="i_va_begin">llvm.va_begin</a></tt> intrinsic may be
3660 arbitrarily complex and require memory allocation, for example.</p>
3664 <!-- ======================================================================= -->
3665 <div class="doc_subsection">
3666 <a name="int_gc">Accurate Garbage Collection Intrinsics</a>
3669 <div class="doc_text">
3672 LLVM support for <a href="GarbageCollection.html">Accurate Garbage
3673 Collection</a> requires the implementation and generation of these intrinsics.
3674 These intrinsics allow identification of <a href="#i_gcroot">GC roots on the
3675 stack</a>, as well as garbage collector implementations that require <a
3676 href="#i_gcread">read</a> and <a href="#i_gcwrite">write</a> barriers.
3677 Front-ends for type-safe garbage collected languages should generate these
3678 intrinsics to make use of the LLVM garbage collectors. For more details, see <a
3679 href="GarbageCollection.html">Accurate Garbage Collection with LLVM</a>.
3683 <!-- _______________________________________________________________________ -->
3684 <div class="doc_subsubsection">
3685 <a name="i_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a>
3688 <div class="doc_text">
3693 declare void %llvm.gcroot(<ty>** %ptrloc, <ty2>* %metadata)
3698 <p>The '<tt>llvm.gcroot</tt>' intrinsic declares the existence of a GC root to
3699 the code generator, and allows some metadata to be associated with it.</p>
3703 <p>The first argument specifies the address of a stack object that contains the
3704 root pointer. The second pointer (which must be either a constant or a global
3705 value address) contains the meta-data to be associated with the root.</p>
3709 <p>At runtime, a call to this intrinsics stores a null pointer into the "ptrloc"
3710 location. At compile-time, the code generator generates information to allow
3711 the runtime to find the pointer at GC safe points.
3717 <!-- _______________________________________________________________________ -->
3718 <div class="doc_subsubsection">
3719 <a name="i_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a>
3722 <div class="doc_text">
3727 declare i8 * %llvm.gcread(i8 * %ObjPtr, i8 ** %Ptr)
3732 <p>The '<tt>llvm.gcread</tt>' intrinsic identifies reads of references from heap
3733 locations, allowing garbage collector implementations that require read
3738 <p>The second argument is the address to read from, which should be an address
3739 allocated from the garbage collector. The first object is a pointer to the
3740 start of the referenced object, if needed by the language runtime (otherwise
3745 <p>The '<tt>llvm.gcread</tt>' intrinsic has the same semantics as a load
3746 instruction, but may be replaced with substantially more complex code by the
3747 garbage collector runtime, as needed.</p>
3752 <!-- _______________________________________________________________________ -->
3753 <div class="doc_subsubsection">
3754 <a name="i_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a>
3757 <div class="doc_text">
3762 declare void %llvm.gcwrite(i8 * %P1, i8 * %Obj, i8 ** %P2)
3767 <p>The '<tt>llvm.gcwrite</tt>' intrinsic identifies writes of references to heap
3768 locations, allowing garbage collector implementations that require write
3769 barriers (such as generational or reference counting collectors).</p>
3773 <p>The first argument is the reference to store, the second is the start of the
3774 object to store it to, and the third is the address of the field of Obj to
3775 store to. If the runtime does not require a pointer to the object, Obj may be
3780 <p>The '<tt>llvm.gcwrite</tt>' intrinsic has the same semantics as a store
3781 instruction, but may be replaced with substantially more complex code by the
3782 garbage collector runtime, as needed.</p>
3788 <!-- ======================================================================= -->
3789 <div class="doc_subsection">
3790 <a name="int_codegen">Code Generator Intrinsics</a>
3793 <div class="doc_text">
3795 These intrinsics are provided by LLVM to expose special features that may only
3796 be implemented with code generator support.
3801 <!-- _______________________________________________________________________ -->
3802 <div class="doc_subsubsection">
3803 <a name="i_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a>
3806 <div class="doc_text">
3810 declare i8 *%llvm.returnaddress(i32 <level>)
3816 The '<tt>llvm.returnaddress</tt>' intrinsic attempts to compute a
3817 target-specific value indicating the return address of the current function
3818 or one of its callers.
3824 The argument to this intrinsic indicates which function to return the address
3825 for. Zero indicates the calling function, one indicates its caller, etc. The
3826 argument is <b>required</b> to be a constant integer value.
3832 The '<tt>llvm.returnaddress</tt>' intrinsic either returns a pointer indicating
3833 the return address of the specified call frame, or zero if it cannot be
3834 identified. The value returned by this intrinsic is likely to be incorrect or 0
3835 for arguments other than zero, so it should only be used for debugging purposes.
3839 Note that calling this intrinsic does not prevent function inlining or other
3840 aggressive transformations, so the value returned may not be that of the obvious
3841 source-language caller.
3846 <!-- _______________________________________________________________________ -->
3847 <div class="doc_subsubsection">
3848 <a name="i_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a>
3851 <div class="doc_text">
3855 declare i8 *%llvm.frameaddress(i32 <level>)
3861 The '<tt>llvm.frameaddress</tt>' intrinsic attempts to return the
3862 target-specific frame pointer value for the specified stack frame.
3868 The argument to this intrinsic indicates which function to return the frame
3869 pointer for. Zero indicates the calling function, one indicates its caller,
3870 etc. The argument is <b>required</b> to be a constant integer value.
3876 The '<tt>llvm.frameaddress</tt>' intrinsic either returns a pointer indicating
3877 the frame address of the specified call frame, or zero if it cannot be
3878 identified. The value returned by this intrinsic is likely to be incorrect or 0
3879 for arguments other than zero, so it should only be used for debugging purposes.
3883 Note that calling this intrinsic does not prevent function inlining or other
3884 aggressive transformations, so the value returned may not be that of the obvious
3885 source-language caller.
3889 <!-- _______________________________________________________________________ -->
3890 <div class="doc_subsubsection">
3891 <a name="i_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a>
3894 <div class="doc_text">
3898 declare i8 *%llvm.stacksave()
3904 The '<tt>llvm.stacksave</tt>' intrinsic is used to remember the current state of
3905 the function stack, for use with <a href="#i_stackrestore">
3906 <tt>llvm.stackrestore</tt></a>. This is useful for implementing language
3907 features like scoped automatic variable sized arrays in C99.
3913 This intrinsic returns a opaque pointer value that can be passed to <a
3914 href="#i_stackrestore"><tt>llvm.stackrestore</tt></a>. When an
3915 <tt>llvm.stackrestore</tt> intrinsic is executed with a value saved from
3916 <tt>llvm.stacksave</tt>, it effectively restores the state of the stack to the
3917 state it was in when the <tt>llvm.stacksave</tt> intrinsic executed. In
3918 practice, this pops any <a href="#i_alloca">alloca</a> blocks from the stack
3919 that were allocated after the <tt>llvm.stacksave</tt> was executed.
3924 <!-- _______________________________________________________________________ -->
3925 <div class="doc_subsubsection">
3926 <a name="i_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a>
3929 <div class="doc_text">
3933 declare void %llvm.stackrestore(i8 * %ptr)
3939 The '<tt>llvm.stackrestore</tt>' intrinsic is used to restore the state of
3940 the function stack to the state it was in when the corresponding <a
3941 href="#llvm.stacksave"><tt>llvm.stacksave</tt></a> intrinsic executed. This is
3942 useful for implementing language features like scoped automatic variable sized
3949 See the description for <a href="#i_stacksave"><tt>llvm.stacksave</tt></a>.
3955 <!-- _______________________________________________________________________ -->
3956 <div class="doc_subsubsection">
3957 <a name="i_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a>
3960 <div class="doc_text">
3964 declare void %llvm.prefetch(i8 * <address>,
3965 i32 <rw>, i32 <locality>)
3972 The '<tt>llvm.prefetch</tt>' intrinsic is a hint to the code generator to insert
3973 a prefetch instruction if supported; otherwise, it is a noop. Prefetches have
3975 effect on the behavior of the program but can change its performance
3982 <tt>address</tt> is the address to be prefetched, <tt>rw</tt> is the specifier
3983 determining if the fetch should be for a read (0) or write (1), and
3984 <tt>locality</tt> is a temporal locality specifier ranging from (0) - no
3985 locality, to (3) - extremely local keep in cache. The <tt>rw</tt> and
3986 <tt>locality</tt> arguments must be constant integers.
3992 This intrinsic does not modify the behavior of the program. In particular,
3993 prefetches cannot trap and do not produce a value. On targets that support this
3994 intrinsic, the prefetch can provide hints to the processor cache for better
4000 <!-- _______________________________________________________________________ -->
4001 <div class="doc_subsubsection">
4002 <a name="i_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a>
4005 <div class="doc_text">
4009 declare void %llvm.pcmarker( i32 <id> )
4016 The '<tt>llvm.pcmarker</tt>' intrinsic is a method to export a Program Counter
4018 code to simulators and other tools. The method is target specific, but it is
4019 expected that the marker will use exported symbols to transmit the PC of the marker.
4020 The marker makes no guarantees that it will remain with any specific instruction
4021 after optimizations. It is possible that the presence of a marker will inhibit
4022 optimizations. The intended use is to be inserted after optimizations to allow
4023 correlations of simulation runs.
4029 <tt>id</tt> is a numerical id identifying the marker.
4035 This intrinsic does not modify the behavior of the program. Backends that do not
4036 support this intrinisic may ignore it.
4041 <!-- _______________________________________________________________________ -->
4042 <div class="doc_subsubsection">
4043 <a name="i_readcyclecounter">'<tt>llvm.readcyclecounter</tt>' Intrinsic</a>
4046 <div class="doc_text">
4050 declare i64 %llvm.readcyclecounter( )
4057 The '<tt>llvm.readcyclecounter</tt>' intrinsic provides access to the cycle
4058 counter register (or similar low latency, high accuracy clocks) on those targets
4059 that support it. On X86, it should map to RDTSC. On Alpha, it should map to RPCC.
4060 As the backing counters overflow quickly (on the order of 9 seconds on alpha), this
4061 should only be used for small timings.
4067 When directly supported, reading the cycle counter should not modify any memory.
4068 Implementations are allowed to either return a application specific value or a
4069 system wide value. On backends without support, this is lowered to a constant 0.
4074 <!-- ======================================================================= -->
4075 <div class="doc_subsection">
4076 <a name="int_libc">Standard C Library Intrinsics</a>
4079 <div class="doc_text">
4081 LLVM provides intrinsics for a few important standard C library functions.
4082 These intrinsics allow source-language front-ends to pass information about the
4083 alignment of the pointer arguments to the code generator, providing opportunity
4084 for more efficient code generation.
4089 <!-- _______________________________________________________________________ -->
4090 <div class="doc_subsubsection">
4091 <a name="i_memcpy">'<tt>llvm.memcpy</tt>' Intrinsic</a>
4094 <div class="doc_text">
4098 declare void %llvm.memcpy.i32(i8 * <dest>, i8 * <src>,
4099 i32 <len>, i32 <align>)
4100 declare void %llvm.memcpy.i64(i8 * <dest>, i8 * <src>,
4101 i64 <len>, i32 <align>)
4107 The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the source
4108 location to the destination location.
4112 Note that, unlike the standard libc function, the <tt>llvm.memcpy.*</tt>
4113 intrinsics do not return a value, and takes an extra alignment argument.
4119 The first argument is a pointer to the destination, the second is a pointer to
4120 the source. The third argument is an integer argument
4121 specifying the number of bytes to copy, and the fourth argument is the alignment
4122 of the source and destination locations.
4126 If the call to this intrinisic has an alignment value that is not 0 or 1, then
4127 the caller guarantees that both the source and destination pointers are aligned
4134 The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the source
4135 location to the destination location, which are not allowed to overlap. It
4136 copies "len" bytes of memory over. If the argument is known to be aligned to
4137 some boundary, this can be specified as the fourth argument, otherwise it should
4143 <!-- _______________________________________________________________________ -->
4144 <div class="doc_subsubsection">
4145 <a name="i_memmove">'<tt>llvm.memmove</tt>' Intrinsic</a>
4148 <div class="doc_text">
4152 declare void %llvm.memmove.i32(i8 * <dest>, i8 * <src>,
4153 i32 <len>, i32 <align>)
4154 declare void %llvm.memmove.i64(i8 * <dest>, i8 * <src>,
4155 i64 <len>, i32 <align>)
4161 The '<tt>llvm.memmove.*</tt>' intrinsics move a block of memory from the source
4162 location to the destination location. It is similar to the
4163 '<tt>llvm.memcmp</tt>' intrinsic but allows the two memory locations to overlap.
4167 Note that, unlike the standard libc function, the <tt>llvm.memmove.*</tt>
4168 intrinsics do not return a value, and takes an extra alignment argument.
4174 The first argument is a pointer to the destination, the second is a pointer to
4175 the source. The third argument is an integer argument
4176 specifying the number of bytes to copy, and the fourth argument is the alignment
4177 of the source and destination locations.
4181 If the call to this intrinisic has an alignment value that is not 0 or 1, then
4182 the caller guarantees that the source and destination pointers are aligned to
4189 The '<tt>llvm.memmove.*</tt>' intrinsics copy a block of memory from the source
4190 location to the destination location, which may overlap. It
4191 copies "len" bytes of memory over. If the argument is known to be aligned to
4192 some boundary, this can be specified as the fourth argument, otherwise it should
4198 <!-- _______________________________________________________________________ -->
4199 <div class="doc_subsubsection">
4200 <a name="i_memset">'<tt>llvm.memset.*</tt>' Intrinsics</a>
4203 <div class="doc_text">
4207 declare void %llvm.memset.i32(i8 * <dest>, i8 <val>,
4208 i32 <len>, i32 <align>)
4209 declare void %llvm.memset.i64(i8 * <dest>, i8 <val>,
4210 i64 <len>, i32 <align>)
4216 The '<tt>llvm.memset.*</tt>' intrinsics fill a block of memory with a particular
4221 Note that, unlike the standard libc function, the <tt>llvm.memset</tt> intrinsic
4222 does not return a value, and takes an extra alignment argument.
4228 The first argument is a pointer to the destination to fill, the second is the
4229 byte value to fill it with, the third argument is an integer
4230 argument specifying the number of bytes to fill, and the fourth argument is the
4231 known alignment of destination location.
4235 If the call to this intrinisic has an alignment value that is not 0 or 1, then
4236 the caller guarantees that the destination pointer is aligned to that boundary.
4242 The '<tt>llvm.memset.*</tt>' intrinsics fill "len" bytes of memory starting at
4244 destination location. If the argument is known to be aligned to some boundary,
4245 this can be specified as the fourth argument, otherwise it should be set to 0 or
4251 <!-- _______________________________________________________________________ -->
4252 <div class="doc_subsubsection">
4253 <a name="i_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a>
4256 <div class="doc_text">
4260 declare float %llvm.sqrt.f32(float %Val)
4261 declare double %llvm.sqrt.f64(double %Val)
4267 The '<tt>llvm.sqrt</tt>' intrinsics return the sqrt of the specified operand,
4268 returning the same value as the libm '<tt>sqrt</tt>' function would. Unlike
4269 <tt>sqrt</tt> in libm, however, <tt>llvm.sqrt</tt> has undefined behavior for
4270 negative numbers (which allows for better optimization).
4276 The argument and return value are floating point numbers of the same type.
4282 This function returns the sqrt of the specified operand if it is a positive
4283 floating point number.
4287 <!-- _______________________________________________________________________ -->
4288 <div class="doc_subsubsection">
4289 <a name="i_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a>
4292 <div class="doc_text">
4296 declare float %llvm.powi.f32(float %Val, i32 %power)
4297 declare double %llvm.powi.f64(double %Val, i32 %power)
4303 The '<tt>llvm.powi.*</tt>' intrinsics return the first operand raised to the
4304 specified (positive or negative) power. The order of evaluation of
4305 multiplications is not defined.
4311 The second argument is an integer power, and the first is a value to raise to
4318 This function returns the first value raised to the second power with an
4319 unspecified sequence of rounding operations.</p>
4323 <!-- ======================================================================= -->
4324 <div class="doc_subsection">
4325 <a name="int_manip">Bit Manipulation Intrinsics</a>
4328 <div class="doc_text">
4330 LLVM provides intrinsics for a few important bit manipulation operations.
4331 These allow efficient code generation for some algorithms.
4336 <!-- _______________________________________________________________________ -->
4337 <div class="doc_subsubsection">
4338 <a name="i_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a>
4341 <div class="doc_text">
4345 declare i16 %llvm.bswap.i16(i16 <id>)
4346 declare i32 %llvm.bswap.i32(i32 <id>)
4347 declare i64 %llvm.bswap.i64(i64 <id>)
4353 The '<tt>llvm.bwsap</tt>' family of intrinsics is used to byteswap a 16, 32 or
4354 64 bit quantity. These are useful for performing operations on data that is not
4355 in the target's native byte order.
4361 The <tt>llvm.bswap.16</tt> intrinsic returns an i16 value that has the high
4362 and low byte of the input i16 swapped. Similarly, the <tt>llvm.bswap.i32</tt>
4363 intrinsic returns an i32 value that has the four bytes of the input i32
4364 swapped, so that if the input bytes are numbered 0, 1, 2, 3 then the returned
4365 i32 will have its bytes in 3, 2, 1, 0 order. The <tt>llvm.bswap.i64</tt>
4366 intrinsic extends this concept to 64 bits.
4371 <!-- _______________________________________________________________________ -->
4372 <div class="doc_subsubsection">
4373 <a name="int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic</a>
4376 <div class="doc_text">
4380 declare i8 %llvm.ctpop.i8 (i8 <src>)
4381 declare i16 %llvm.ctpop.i16(i16 <src>)
4382 declare i32 %llvm.ctpop.i32(i32 <src>)
4383 declare i64 %llvm.ctpop.i64(i64 <src>)
4389 The '<tt>llvm.ctpop</tt>' family of intrinsics counts the number of bits set in a
4396 The only argument is the value to be counted. The argument may be of any
4397 integer type. The return type must match the argument type.
4403 The '<tt>llvm.ctpop</tt>' intrinsic counts the 1's in a variable.
4407 <!-- _______________________________________________________________________ -->
4408 <div class="doc_subsubsection">
4409 <a name="int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic</a>
4412 <div class="doc_text">
4416 declare i8 %llvm.ctlz.i8 (i8 <src>)
4417 declare i16 %llvm.ctlz.i16(i16 <src>)
4418 declare i32 %llvm.ctlz.i32(i32 <src>)
4419 declare i64 %llvm.ctlz.i64(i64 <src>)
4425 The '<tt>llvm.ctlz</tt>' family of intrinsic functions counts the number of
4426 leading zeros in a variable.
4432 The only argument is the value to be counted. The argument may be of any
4433 integer type. The return type must match the argument type.
4439 The '<tt>llvm.ctlz</tt>' intrinsic counts the leading (most significant) zeros
4440 in a variable. If the src == 0 then the result is the size in bits of the type
4441 of src. For example, <tt>llvm.ctlz(i32 2) = 30</tt>.
4447 <!-- _______________________________________________________________________ -->
4448 <div class="doc_subsubsection">
4449 <a name="int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic</a>
4452 <div class="doc_text">
4456 declare i8 %llvm.cttz.i8 (i8 <src>)
4457 declare i16 %llvm.cttz.i16(i16 <src>)
4458 declare i32 %llvm.cttz.i32(i32 <src>)
4459 declare i64 %llvm.cttz.i64(i64 <src>)
4465 The '<tt>llvm.cttz</tt>' family of intrinsic functions counts the number of
4472 The only argument is the value to be counted. The argument may be of any
4473 integer type. The return type must match the argument type.
4479 The '<tt>llvm.cttz</tt>' intrinsic counts the trailing (least significant) zeros
4480 in a variable. If the src == 0 then the result is the size in bits of the type
4481 of src. For example, <tt>llvm.cttz(2) = 1</tt>.
4485 <!-- ======================================================================= -->
4486 <div class="doc_subsection">
4487 <a name="int_debugger">Debugger Intrinsics</a>
4490 <div class="doc_text">
4492 The LLVM debugger intrinsics (which all start with <tt>llvm.dbg.</tt> prefix),
4493 are described in the <a
4494 href="SourceLevelDebugging.html#format_common_intrinsics">LLVM Source Level
4495 Debugging</a> document.
4500 <!-- *********************************************************************** -->
4503 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
4504 src="http://jigsaw.w3.org/css-validator/images/vcss" alt="Valid CSS!"></a>
4505 <a href="http://validator.w3.org/check/referer"><img
4506 src="http://www.w3.org/Icons/valid-html401" alt="Valid HTML 4.01!" /></a>
4508 <a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
4509 <a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br>
4510 Last modified: $Date$