1 <!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN"
2 "http://www.w3.org/TR/html4/strict.dtd">
5 <title>LLVM Assembly Language Reference Manual</title>
6 <meta http-equiv="Content-Type" content="text/html; charset=utf-8">
7 <meta name="author" content="Chris Lattner">
8 <meta name="description"
9 content="LLVM Assembly Language Reference Manual.">
10 <link rel="stylesheet" href="llvm.css" type="text/css">
15 <div class="doc_title"> LLVM Language Reference Manual </div>
17 <li><a href="#abstract">Abstract</a></li>
18 <li><a href="#introduction">Introduction</a></li>
19 <li><a href="#identifiers">Identifiers</a></li>
20 <li><a href="#highlevel">High Level Structure</a>
22 <li><a href="#modulestructure">Module Structure</a></li>
23 <li><a href="#linkage">Linkage Types</a></li>
24 <li><a href="#callingconv">Calling Conventions</a></li>
25 <li><a href="#globalvars">Global Variables</a></li>
26 <li><a href="#functionstructure">Functions</a></li>
27 <li><a href="#aliasstructure">Aliases</a>
28 <li><a href="#paramattrs">Parameter Attributes</a></li>
29 <li><a href="#moduleasm">Module-Level Inline Assembly</a></li>
30 <li><a href="#datalayout">Data Layout</a></li>
33 <li><a href="#typesystem">Type System</a>
35 <li><a href="#t_primitive">Primitive Types</a>
37 <li><a href="#t_classifications">Type Classifications</a></li>
40 <li><a href="#t_derived">Derived Types</a>
42 <li><a href="#t_array">Array Type</a></li>
43 <li><a href="#t_function">Function Type</a></li>
44 <li><a href="#t_pointer">Pointer Type</a></li>
45 <li><a href="#t_struct">Structure Type</a></li>
46 <li><a href="#t_pstruct">Packed Structure Type</a></li>
47 <li><a href="#t_vector">Vector Type</a></li>
48 <li><a href="#t_opaque">Opaque Type</a></li>
53 <li><a href="#constants">Constants</a>
55 <li><a href="#simpleconstants">Simple Constants</a>
56 <li><a href="#aggregateconstants">Aggregate Constants</a>
57 <li><a href="#globalconstants">Global Variable and Function Addresses</a>
58 <li><a href="#undefvalues">Undefined Values</a>
59 <li><a href="#constantexprs">Constant Expressions</a>
62 <li><a href="#othervalues">Other Values</a>
64 <li><a href="#inlineasm">Inline Assembler Expressions</a>
67 <li><a href="#instref">Instruction Reference</a>
69 <li><a href="#terminators">Terminator Instructions</a>
71 <li><a href="#i_ret">'<tt>ret</tt>' Instruction</a></li>
72 <li><a href="#i_br">'<tt>br</tt>' Instruction</a></li>
73 <li><a href="#i_switch">'<tt>switch</tt>' Instruction</a></li>
74 <li><a href="#i_invoke">'<tt>invoke</tt>' Instruction</a></li>
75 <li><a href="#i_unwind">'<tt>unwind</tt>' Instruction</a></li>
76 <li><a href="#i_unreachable">'<tt>unreachable</tt>' Instruction</a></li>
79 <li><a href="#binaryops">Binary Operations</a>
81 <li><a href="#i_add">'<tt>add</tt>' Instruction</a></li>
82 <li><a href="#i_sub">'<tt>sub</tt>' Instruction</a></li>
83 <li><a href="#i_mul">'<tt>mul</tt>' Instruction</a></li>
84 <li><a href="#i_udiv">'<tt>udiv</tt>' Instruction</a></li>
85 <li><a href="#i_sdiv">'<tt>sdiv</tt>' Instruction</a></li>
86 <li><a href="#i_fdiv">'<tt>fdiv</tt>' Instruction</a></li>
87 <li><a href="#i_urem">'<tt>urem</tt>' Instruction</a></li>
88 <li><a href="#i_srem">'<tt>srem</tt>' Instruction</a></li>
89 <li><a href="#i_frem">'<tt>frem</tt>' Instruction</a></li>
92 <li><a href="#bitwiseops">Bitwise Binary Operations</a>
94 <li><a href="#i_shl">'<tt>shl</tt>' Instruction</a></li>
95 <li><a href="#i_lshr">'<tt>lshr</tt>' Instruction</a></li>
96 <li><a href="#i_ashr">'<tt>ashr</tt>' Instruction</a></li>
97 <li><a href="#i_and">'<tt>and</tt>' Instruction</a></li>
98 <li><a href="#i_or">'<tt>or</tt>' Instruction</a></li>
99 <li><a href="#i_xor">'<tt>xor</tt>' Instruction</a></li>
102 <li><a href="#vectorops">Vector Operations</a>
104 <li><a href="#i_extractelement">'<tt>extractelement</tt>' Instruction</a></li>
105 <li><a href="#i_insertelement">'<tt>insertelement</tt>' Instruction</a></li>
106 <li><a href="#i_shufflevector">'<tt>shufflevector</tt>' Instruction</a></li>
109 <li><a href="#memoryops">Memory Access and Addressing Operations</a>
111 <li><a href="#i_malloc">'<tt>malloc</tt>' Instruction</a></li>
112 <li><a href="#i_free">'<tt>free</tt>' Instruction</a></li>
113 <li><a href="#i_alloca">'<tt>alloca</tt>' Instruction</a></li>
114 <li><a href="#i_load">'<tt>load</tt>' Instruction</a></li>
115 <li><a href="#i_store">'<tt>store</tt>' Instruction</a></li>
116 <li><a href="#i_getelementptr">'<tt>getelementptr</tt>' Instruction</a></li>
119 <li><a href="#convertops">Conversion Operations</a>
121 <li><a href="#i_trunc">'<tt>trunc .. to</tt>' Instruction</a></li>
122 <li><a href="#i_zext">'<tt>zext .. to</tt>' Instruction</a></li>
123 <li><a href="#i_sext">'<tt>sext .. to</tt>' Instruction</a></li>
124 <li><a href="#i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a></li>
125 <li><a href="#i_fpext">'<tt>fpext .. to</tt>' Instruction</a></li>
126 <li><a href="#i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a></li>
127 <li><a href="#i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a></li>
128 <li><a href="#i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a></li>
129 <li><a href="#i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a></li>
130 <li><a href="#i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a></li>
131 <li><a href="#i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a></li>
132 <li><a href="#i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a></li>
134 <li><a href="#otherops">Other Operations</a>
136 <li><a href="#i_icmp">'<tt>icmp</tt>' Instruction</a></li>
137 <li><a href="#i_fcmp">'<tt>fcmp</tt>' Instruction</a></li>
138 <li><a href="#i_phi">'<tt>phi</tt>' Instruction</a></li>
139 <li><a href="#i_select">'<tt>select</tt>' Instruction</a></li>
140 <li><a href="#i_call">'<tt>call</tt>' Instruction</a></li>
141 <li><a href="#i_va_arg">'<tt>va_arg</tt>' Instruction</a></li>
146 <li><a href="#intrinsics">Intrinsic Functions</a>
148 <li><a href="#int_varargs">Variable Argument Handling Intrinsics</a>
150 <li><a href="#int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a></li>
151 <li><a href="#int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a></li>
152 <li><a href="#int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a></li>
155 <li><a href="#int_gc">Accurate Garbage Collection Intrinsics</a>
157 <li><a href="#int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a></li>
158 <li><a href="#int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a></li>
159 <li><a href="#int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a></li>
162 <li><a href="#int_codegen">Code Generator Intrinsics</a>
164 <li><a href="#int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a></li>
165 <li><a href="#int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a></li>
166 <li><a href="#int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a></li>
167 <li><a href="#int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a></li>
168 <li><a href="#int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a></li>
169 <li><a href="#int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a></li>
170 <li><a href="#int_readcyclecounter"><tt>llvm.readcyclecounter</tt>' Intrinsic</a></li>
173 <li><a href="#int_libc">Standard C Library Intrinsics</a>
175 <li><a href="#int_memcpy">'<tt>llvm.memcpy.*</tt>' Intrinsic</a></li>
176 <li><a href="#int_memmove">'<tt>llvm.memmove.*</tt>' Intrinsic</a></li>
177 <li><a href="#int_memset">'<tt>llvm.memset.*</tt>' Intrinsic</a></li>
178 <li><a href="#int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a></li>
179 <li><a href="#int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a></li>
182 <li><a href="#int_manip">Bit Manipulation Intrinsics</a>
184 <li><a href="#int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a></li>
185 <li><a href="#int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic </a></li>
186 <li><a href="#int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic </a></li>
187 <li><a href="#int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic </a></li>
188 <li><a href="#int_part_select">'<tt>llvm.part.select.*</tt>' Intrinsic </a></li>
189 <li><a href="#int_part_set">'<tt>llvm.part.set.*</tt>' Intrinsic </a></li>
192 <li><a href="#int_debugger">Debugger intrinsics</a></li>
193 <li><a href="#int_eh">Exception Handling intrinsics</a></li>
194 <li><a href="#int_atomics">Atomic Operations and Synchronization Intrinsics</a>
196 <li><a href="#int_lcs">'<tt>llvm.atomic.lcs.*</tt>' Intrinsic</a></li>
197 <li><a href="#int_ls">'<tt>llvm.atomic.ls.*</tt>' Intrinsic</a></li>
198 <li><a href="#int_las">'<tt>llvm.atomic.las.*</tt>' Intrinsic</a></li>
199 <li><a href="#int_lss">'<tt>llvm.atomic.lss.*</tt>' Intrinsic</a></li>
200 <li><a href="#int_memory_barrier">'<tt>llvm.memory.barrier</tt>' Intrinsic</a></li>
203 <li><a href="#int_trampoline">Trampoline Intrinsic</a>
205 <li><a href="#int_it">'<tt>llvm.init.trampoline</tt>' Intrinsic</a></li>
208 <li><a href="#int_general">General intrinsics</a>
210 <li><a href="#int_var_annotation">
211 <tt>llvm.var.annotation</tt>' Intrinsic</a></li>
214 <li><a href="#int_annotation">
215 <tt>llvm.annotation.*</tt>' Intrinsic</a></li>
222 <div class="doc_author">
223 <p>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a>
224 and <a href="mailto:vadve@cs.uiuc.edu">Vikram Adve</a></p>
227 <!-- *********************************************************************** -->
228 <div class="doc_section"> <a name="abstract">Abstract </a></div>
229 <!-- *********************************************************************** -->
231 <div class="doc_text">
232 <p>This document is a reference manual for the LLVM assembly language.
233 LLVM is an SSA based representation that provides type safety,
234 low-level operations, flexibility, and the capability of representing
235 'all' high-level languages cleanly. It is the common code
236 representation used throughout all phases of the LLVM compilation
240 <!-- *********************************************************************** -->
241 <div class="doc_section"> <a name="introduction">Introduction</a> </div>
242 <!-- *********************************************************************** -->
244 <div class="doc_text">
246 <p>The LLVM code representation is designed to be used in three
247 different forms: as an in-memory compiler IR, as an on-disk bitcode
248 representation (suitable for fast loading by a Just-In-Time compiler),
249 and as a human readable assembly language representation. This allows
250 LLVM to provide a powerful intermediate representation for efficient
251 compiler transformations and analysis, while providing a natural means
252 to debug and visualize the transformations. The three different forms
253 of LLVM are all equivalent. This document describes the human readable
254 representation and notation.</p>
256 <p>The LLVM representation aims to be light-weight and low-level
257 while being expressive, typed, and extensible at the same time. It
258 aims to be a "universal IR" of sorts, by being at a low enough level
259 that high-level ideas may be cleanly mapped to it (similar to how
260 microprocessors are "universal IR's", allowing many source languages to
261 be mapped to them). By providing type information, LLVM can be used as
262 the target of optimizations: for example, through pointer analysis, it
263 can be proven that a C automatic variable is never accessed outside of
264 the current function... allowing it to be promoted to a simple SSA
265 value instead of a memory location.</p>
269 <!-- _______________________________________________________________________ -->
270 <div class="doc_subsubsection"> <a name="wellformed">Well-Formedness</a> </div>
272 <div class="doc_text">
274 <p>It is important to note that this document describes 'well formed'
275 LLVM assembly language. There is a difference between what the parser
276 accepts and what is considered 'well formed'. For example, the
277 following instruction is syntactically okay, but not well formed:</p>
279 <div class="doc_code">
281 %x = <a href="#i_add">add</a> i32 1, %x
285 <p>...because the definition of <tt>%x</tt> does not dominate all of
286 its uses. The LLVM infrastructure provides a verification pass that may
287 be used to verify that an LLVM module is well formed. This pass is
288 automatically run by the parser after parsing input assembly and by
289 the optimizer before it outputs bitcode. The violations pointed out
290 by the verifier pass indicate bugs in transformation passes or input to
294 <!-- Describe the typesetting conventions here. -->
296 <!-- *********************************************************************** -->
297 <div class="doc_section"> <a name="identifiers">Identifiers</a> </div>
298 <!-- *********************************************************************** -->
300 <div class="doc_text">
302 <p>LLVM identifiers come in two basic types: global and local. Global
303 identifiers (functions, global variables) begin with the @ character. Local
304 identifiers (register names, types) begin with the % character. Additionally,
305 there are three different formats for identifiers, for different purposes:
308 <li>Named values are represented as a string of characters with their prefix.
309 For example, %foo, @DivisionByZero, %a.really.long.identifier. The actual
310 regular expression used is '<tt>[%@][a-zA-Z$._][a-zA-Z$._0-9]*</tt>'.
311 Identifiers which require other characters in their names can be surrounded
312 with quotes. In this way, anything except a <tt>"</tt> character can
313 be used in a named value.</li>
315 <li>Unnamed values are represented as an unsigned numeric value with their
316 prefix. For example, %12, @2, %44.</li>
318 <li>Constants, which are described in a <a href="#constants">section about
319 constants</a>, below.</li>
322 <p>LLVM requires that values start with a prefix for two reasons: Compilers
323 don't need to worry about name clashes with reserved words, and the set of
324 reserved words may be expanded in the future without penalty. Additionally,
325 unnamed identifiers allow a compiler to quickly come up with a temporary
326 variable without having to avoid symbol table conflicts.</p>
328 <p>Reserved words in LLVM are very similar to reserved words in other
329 languages. There are keywords for different opcodes
330 ('<tt><a href="#i_add">add</a></tt>',
331 '<tt><a href="#i_bitcast">bitcast</a></tt>',
332 '<tt><a href="#i_ret">ret</a></tt>', etc...), for primitive type names ('<tt><a
333 href="#t_void">void</a></tt>', '<tt><a href="#t_primitive">i32</a></tt>', etc...),
334 and others. These reserved words cannot conflict with variable names, because
335 none of them start with a prefix character ('%' or '@').</p>
337 <p>Here is an example of LLVM code to multiply the integer variable
338 '<tt>%X</tt>' by 8:</p>
342 <div class="doc_code">
344 %result = <a href="#i_mul">mul</a> i32 %X, 8
348 <p>After strength reduction:</p>
350 <div class="doc_code">
352 %result = <a href="#i_shl">shl</a> i32 %X, i8 3
356 <p>And the hard way:</p>
358 <div class="doc_code">
360 <a href="#i_add">add</a> i32 %X, %X <i>; yields {i32}:%0</i>
361 <a href="#i_add">add</a> i32 %0, %0 <i>; yields {i32}:%1</i>
362 %result = <a href="#i_add">add</a> i32 %1, %1
366 <p>This last way of multiplying <tt>%X</tt> by 8 illustrates several
367 important lexical features of LLVM:</p>
371 <li>Comments are delimited with a '<tt>;</tt>' and go until the end of
374 <li>Unnamed temporaries are created when the result of a computation is not
375 assigned to a named value.</li>
377 <li>Unnamed temporaries are numbered sequentially</li>
381 <p>...and it also shows a convention that we follow in this document. When
382 demonstrating instructions, we will follow an instruction with a comment that
383 defines the type and name of value produced. Comments are shown in italic
388 <!-- *********************************************************************** -->
389 <div class="doc_section"> <a name="highlevel">High Level Structure</a> </div>
390 <!-- *********************************************************************** -->
392 <!-- ======================================================================= -->
393 <div class="doc_subsection"> <a name="modulestructure">Module Structure</a>
396 <div class="doc_text">
398 <p>LLVM programs are composed of "Module"s, each of which is a
399 translation unit of the input programs. Each module consists of
400 functions, global variables, and symbol table entries. Modules may be
401 combined together with the LLVM linker, which merges function (and
402 global variable) definitions, resolves forward declarations, and merges
403 symbol table entries. Here is an example of the "hello world" module:</p>
405 <div class="doc_code">
406 <pre><i>; Declare the string constant as a global constant...</i>
407 <a href="#identifiers">@.LC0</a> = <a href="#linkage_internal">internal</a> <a
408 href="#globalvars">constant</a> <a href="#t_array">[13 x i8]</a> c"hello world\0A\00" <i>; [13 x i8]*</i>
410 <i>; External declaration of the puts function</i>
411 <a href="#functionstructure">declare</a> i32 @puts(i8 *) <i>; i32(i8 *)* </i>
413 <i>; Definition of main function</i>
414 define i32 @main() { <i>; i32()* </i>
415 <i>; Convert [13x i8 ]* to i8 *...</i>
417 href="#i_getelementptr">getelementptr</a> [13 x i8 ]* @.LC0, i64 0, i64 0 <i>; i8 *</i>
419 <i>; Call puts function to write out the string to stdout...</i>
421 href="#i_call">call</a> i32 @puts(i8 * %cast210) <i>; i32</i>
423 href="#i_ret">ret</a> i32 0<br>}<br>
427 <p>This example is made up of a <a href="#globalvars">global variable</a>
428 named "<tt>.LC0</tt>", an external declaration of the "<tt>puts</tt>"
429 function, and a <a href="#functionstructure">function definition</a>
430 for "<tt>main</tt>".</p>
432 <p>In general, a module is made up of a list of global values,
433 where both functions and global variables are global values. Global values are
434 represented by a pointer to a memory location (in this case, a pointer to an
435 array of char, and a pointer to a function), and have one of the following <a
436 href="#linkage">linkage types</a>.</p>
440 <!-- ======================================================================= -->
441 <div class="doc_subsection">
442 <a name="linkage">Linkage Types</a>
445 <div class="doc_text">
448 All Global Variables and Functions have one of the following types of linkage:
453 <dt><tt><b><a name="linkage_internal">internal</a></b></tt> </dt>
455 <dd>Global values with internal linkage are only directly accessible by
456 objects in the current module. In particular, linking code into a module with
457 an internal global value may cause the internal to be renamed as necessary to
458 avoid collisions. Because the symbol is internal to the module, all
459 references can be updated. This corresponds to the notion of the
460 '<tt>static</tt>' keyword in C.
463 <dt><tt><b><a name="linkage_linkonce">linkonce</a></b></tt>: </dt>
465 <dd>Globals with "<tt>linkonce</tt>" linkage are merged with other globals of
466 the same name when linkage occurs. This is typically used to implement
467 inline functions, templates, or other code which must be generated in each
468 translation unit that uses it. Unreferenced <tt>linkonce</tt> globals are
469 allowed to be discarded.
472 <dt><tt><b><a name="linkage_weak">weak</a></b></tt>: </dt>
474 <dd>"<tt>weak</tt>" linkage is exactly the same as <tt>linkonce</tt> linkage,
475 except that unreferenced <tt>weak</tt> globals may not be discarded. This is
476 used for globals that may be emitted in multiple translation units, but that
477 are not guaranteed to be emitted into every translation unit that uses them.
478 One example of this are common globals in C, such as "<tt>int X;</tt>" at
482 <dt><tt><b><a name="linkage_appending">appending</a></b></tt>: </dt>
484 <dd>"<tt>appending</tt>" linkage may only be applied to global variables of
485 pointer to array type. When two global variables with appending linkage are
486 linked together, the two global arrays are appended together. This is the
487 LLVM, typesafe, equivalent of having the system linker append together
488 "sections" with identical names when .o files are linked.
491 <dt><tt><b><a name="linkage_externweak">extern_weak</a></b></tt>: </dt>
492 <dd>The semantics of this linkage follow the ELF model: the symbol is weak
493 until linked, if not linked, the symbol becomes null instead of being an
497 <dt><tt><b><a name="linkage_external">externally visible</a></b></tt>:</dt>
499 <dd>If none of the above identifiers are used, the global is externally
500 visible, meaning that it participates in linkage and can be used to resolve
501 external symbol references.
506 The next two types of linkage are targeted for Microsoft Windows platform
507 only. They are designed to support importing (exporting) symbols from (to)
512 <dt><tt><b><a name="linkage_dllimport">dllimport</a></b></tt>: </dt>
514 <dd>"<tt>dllimport</tt>" linkage causes the compiler to reference a function
515 or variable via a global pointer to a pointer that is set up by the DLL
516 exporting the symbol. On Microsoft Windows targets, the pointer name is
517 formed by combining <code>_imp__</code> and the function or variable name.
520 <dt><tt><b><a name="linkage_dllexport">dllexport</a></b></tt>: </dt>
522 <dd>"<tt>dllexport</tt>" linkage causes the compiler to provide a global
523 pointer to a pointer in a DLL, so that it can be referenced with the
524 <tt>dllimport</tt> attribute. On Microsoft Windows targets, the pointer
525 name is formed by combining <code>_imp__</code> and the function or variable
531 <p><a name="linkage_external"></a>For example, since the "<tt>.LC0</tt>"
532 variable is defined to be internal, if another module defined a "<tt>.LC0</tt>"
533 variable and was linked with this one, one of the two would be renamed,
534 preventing a collision. Since "<tt>main</tt>" and "<tt>puts</tt>" are
535 external (i.e., lacking any linkage declarations), they are accessible
536 outside of the current module.</p>
537 <p>It is illegal for a function <i>declaration</i>
538 to have any linkage type other than "externally visible", <tt>dllimport</tt>,
539 or <tt>extern_weak</tt>.</p>
540 <p>Aliases can have only <tt>external</tt>, <tt>internal</tt> and <tt>weak</tt>
544 <!-- ======================================================================= -->
545 <div class="doc_subsection">
546 <a name="callingconv">Calling Conventions</a>
549 <div class="doc_text">
551 <p>LLVM <a href="#functionstructure">functions</a>, <a href="#i_call">calls</a>
552 and <a href="#i_invoke">invokes</a> can all have an optional calling convention
553 specified for the call. The calling convention of any pair of dynamic
554 caller/callee must match, or the behavior of the program is undefined. The
555 following calling conventions are supported by LLVM, and more may be added in
559 <dt><b>"<tt>ccc</tt>" - The C calling convention</b>:</dt>
561 <dd>This calling convention (the default if no other calling convention is
562 specified) matches the target C calling conventions. This calling convention
563 supports varargs function calls and tolerates some mismatch in the declared
564 prototype and implemented declaration of the function (as does normal C).
567 <dt><b>"<tt>fastcc</tt>" - The fast calling convention</b>:</dt>
569 <dd>This calling convention attempts to make calls as fast as possible
570 (e.g. by passing things in registers). This calling convention allows the
571 target to use whatever tricks it wants to produce fast code for the target,
572 without having to conform to an externally specified ABI. Implementations of
573 this convention should allow arbitrary tail call optimization to be supported.
574 This calling convention does not support varargs and requires the prototype of
575 all callees to exactly match the prototype of the function definition.
578 <dt><b>"<tt>coldcc</tt>" - The cold calling convention</b>:</dt>
580 <dd>This calling convention attempts to make code in the caller as efficient
581 as possible under the assumption that the call is not commonly executed. As
582 such, these calls often preserve all registers so that the call does not break
583 any live ranges in the caller side. This calling convention does not support
584 varargs and requires the prototype of all callees to exactly match the
585 prototype of the function definition.
588 <dt><b>"<tt>cc <<em>n</em>></tt>" - Numbered convention</b>:</dt>
590 <dd>Any calling convention may be specified by number, allowing
591 target-specific calling conventions to be used. Target specific calling
592 conventions start at 64.
596 <p>More calling conventions can be added/defined on an as-needed basis, to
597 support pascal conventions or any other well-known target-independent
602 <!-- ======================================================================= -->
603 <div class="doc_subsection">
604 <a name="visibility">Visibility Styles</a>
607 <div class="doc_text">
610 All Global Variables and Functions have one of the following visibility styles:
614 <dt><b>"<tt>default</tt>" - Default style</b>:</dt>
616 <dd>On ELF, default visibility means that the declaration is visible to other
617 modules and, in shared libraries, means that the declared entity may be
618 overridden. On Darwin, default visibility means that the declaration is
619 visible to other modules. Default visibility corresponds to "external
620 linkage" in the language.
623 <dt><b>"<tt>hidden</tt>" - Hidden style</b>:</dt>
625 <dd>Two declarations of an object with hidden visibility refer to the same
626 object if they are in the same shared object. Usually, hidden visibility
627 indicates that the symbol will not be placed into the dynamic symbol table,
628 so no other module (executable or shared library) can reference it
632 <dt><b>"<tt>protected</tt>" - Protected style</b>:</dt>
634 <dd>On ELF, protected visibility indicates that the symbol will be placed in
635 the dynamic symbol table, but that references within the defining module will
636 bind to the local symbol. That is, the symbol cannot be overridden by another
643 <!-- ======================================================================= -->
644 <div class="doc_subsection">
645 <a name="globalvars">Global Variables</a>
648 <div class="doc_text">
650 <p>Global variables define regions of memory allocated at compilation time
651 instead of run-time. Global variables may optionally be initialized, may have
652 an explicit section to be placed in, and may have an optional explicit alignment
653 specified. A variable may be defined as "thread_local", which means that it
654 will not be shared by threads (each thread will have a separated copy of the
655 variable). A variable may be defined as a global "constant," which indicates
656 that the contents of the variable will <b>never</b> be modified (enabling better
657 optimization, allowing the global data to be placed in the read-only section of
658 an executable, etc). Note that variables that need runtime initialization
659 cannot be marked "constant" as there is a store to the variable.</p>
662 LLVM explicitly allows <em>declarations</em> of global variables to be marked
663 constant, even if the final definition of the global is not. This capability
664 can be used to enable slightly better optimization of the program, but requires
665 the language definition to guarantee that optimizations based on the
666 'constantness' are valid for the translation units that do not include the
670 <p>As SSA values, global variables define pointer values that are in
671 scope (i.e. they dominate) all basic blocks in the program. Global
672 variables always define a pointer to their "content" type because they
673 describe a region of memory, and all memory objects in LLVM are
674 accessed through pointers.</p>
676 <p>LLVM allows an explicit section to be specified for globals. If the target
677 supports it, it will emit globals to the section specified.</p>
679 <p>An explicit alignment may be specified for a global. If not present, or if
680 the alignment is set to zero, the alignment of the global is set by the target
681 to whatever it feels convenient. If an explicit alignment is specified, the
682 global is forced to have at least that much alignment. All alignments must be
685 <p>For example, the following defines a global with an initializer, section,
688 <div class="doc_code">
690 @G = constant float 1.0, section "foo", align 4
697 <!-- ======================================================================= -->
698 <div class="doc_subsection">
699 <a name="functionstructure">Functions</a>
702 <div class="doc_text">
704 <p>LLVM function definitions consist of the "<tt>define</tt>" keyord,
705 an optional <a href="#linkage">linkage type</a>, an optional
706 <a href="#visibility">visibility style</a>, an optional
707 <a href="#callingconv">calling convention</a>, a return type, an optional
708 <a href="#paramattrs">parameter attribute</a> for the return type, a function
709 name, a (possibly empty) argument list (each with optional
710 <a href="#paramattrs">parameter attributes</a>), an optional section, an
711 optional alignment, an opening curly brace, a list of basic blocks, and a
714 LLVM function declarations consist of the "<tt>declare</tt>" keyword, an
715 optional <a href="#linkage">linkage type</a>, an optional
716 <a href="#visibility">visibility style</a>, an optional
717 <a href="#callingconv">calling convention</a>, a return type, an optional
718 <a href="#paramattrs">parameter attribute</a> for the return type, a function
719 name, a possibly empty list of arguments, and an optional alignment.</p>
721 <p>A function definition contains a list of basic blocks, forming the CFG for
722 the function. Each basic block may optionally start with a label (giving the
723 basic block a symbol table entry), contains a list of instructions, and ends
724 with a <a href="#terminators">terminator</a> instruction (such as a branch or
725 function return).</p>
727 <p>The first basic block in a function is special in two ways: it is immediately
728 executed on entrance to the function, and it is not allowed to have predecessor
729 basic blocks (i.e. there can not be any branches to the entry block of a
730 function). Because the block can have no predecessors, it also cannot have any
731 <a href="#i_phi">PHI nodes</a>.</p>
733 <p>LLVM allows an explicit section to be specified for functions. If the target
734 supports it, it will emit functions to the section specified.</p>
736 <p>An explicit alignment may be specified for a function. If not present, or if
737 the alignment is set to zero, the alignment of the function is set by the target
738 to whatever it feels convenient. If an explicit alignment is specified, the
739 function is forced to have at least that much alignment. All alignments must be
745 <!-- ======================================================================= -->
746 <div class="doc_subsection">
747 <a name="aliasstructure">Aliases</a>
749 <div class="doc_text">
750 <p>Aliases act as "second name" for the aliasee value (which can be either
751 function or global variable or bitcast of global value). Aliases may have an
752 optional <a href="#linkage">linkage type</a>, and an
753 optional <a href="#visibility">visibility style</a>.</p>
757 <div class="doc_code">
759 @<Name> = [Linkage] [Visibility] alias <AliaseeTy> @<Aliasee>
767 <!-- ======================================================================= -->
768 <div class="doc_subsection"><a name="paramattrs">Parameter Attributes</a></div>
769 <div class="doc_text">
770 <p>The return type and each parameter of a function type may have a set of
771 <i>parameter attributes</i> associated with them. Parameter attributes are
772 used to communicate additional information about the result or parameters of
773 a function. Parameter attributes are considered to be part of the function
774 type so two functions types that differ only by the parameter attributes
775 are different function types.</p>
777 <p>Parameter attributes are simple keywords that follow the type specified. If
778 multiple parameter attributes are needed, they are space separated. For
781 <div class="doc_code">
783 %someFunc = i16 (i8 signext %someParam) zeroext
784 %someFunc = i16 (i8 zeroext %someParam) zeroext
788 <p>Note that the two function types above are unique because the parameter has
789 a different attribute (<tt>signext</tt> in the first one, <tt>zeroext</tt> in
790 the second). Also note that the attribute for the function result
791 (<tt>zeroext</tt>) comes immediately after the argument list.</p>
793 <p>Currently, only the following parameter attributes are defined:</p>
795 <dt><tt>zeroext</tt></dt>
796 <dd>This indicates that the parameter should be zero extended just before
797 a call to this function.</dd>
798 <dt><tt>signext</tt></dt>
799 <dd>This indicates that the parameter should be sign extended just before
800 a call to this function.</dd>
801 <dt><tt>inreg</tt></dt>
802 <dd>This indicates that the parameter should be placed in register (if
803 possible) during assembling function call. Support for this attribute is
805 <dt><tt>sret</tt></dt>
806 <dd>This indicates that the parameter specifies the address of a structure
807 that is the return value of the function in the source program.</dd>
808 <dt><tt>noalias</tt></dt>
809 <dd>This indicates that the parameter not alias any other object or any
810 other "noalias" objects during the function call.
811 <dt><tt>noreturn</tt></dt>
812 <dd>This function attribute indicates that the function never returns. This
813 indicates to LLVM that every call to this function should be treated as if
814 an <tt>unreachable</tt> instruction immediately followed the call.</dd>
815 <dt><tt>nounwind</tt></dt>
816 <dd>This function attribute indicates that the function type does not use
817 the unwind instruction and does not allow stack unwinding to propagate
819 <dt><tt>nest</tt></dt>
820 <dd>This indicates that the parameter can be excised using the
821 <a href="#int_trampoline">trampoline intrinsics</a>.</dd>
826 <!-- ======================================================================= -->
827 <div class="doc_subsection">
828 <a name="moduleasm">Module-Level Inline Assembly</a>
831 <div class="doc_text">
833 Modules may contain "module-level inline asm" blocks, which corresponds to the
834 GCC "file scope inline asm" blocks. These blocks are internally concatenated by
835 LLVM and treated as a single unit, but may be separated in the .ll file if
836 desired. The syntax is very simple:
839 <div class="doc_code">
841 module asm "inline asm code goes here"
842 module asm "more can go here"
846 <p>The strings can contain any character by escaping non-printable characters.
847 The escape sequence used is simply "\xx" where "xx" is the two digit hex code
852 The inline asm code is simply printed to the machine code .s file when
853 assembly code is generated.
857 <!-- ======================================================================= -->
858 <div class="doc_subsection">
859 <a name="datalayout">Data Layout</a>
862 <div class="doc_text">
863 <p>A module may specify a target specific data layout string that specifies how
864 data is to be laid out in memory. The syntax for the data layout is simply:</p>
865 <pre> target datalayout = "<i>layout specification</i>"</pre>
866 <p>The <i>layout specification</i> consists of a list of specifications
867 separated by the minus sign character ('-'). Each specification starts with a
868 letter and may include other information after the letter to define some
869 aspect of the data layout. The specifications accepted are as follows: </p>
872 <dd>Specifies that the target lays out data in big-endian form. That is, the
873 bits with the most significance have the lowest address location.</dd>
875 <dd>Specifies that hte target lays out data in little-endian form. That is,
876 the bits with the least significance have the lowest address location.</dd>
877 <dt><tt>p:<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
878 <dd>This specifies the <i>size</i> of a pointer and its <i>abi</i> and
879 <i>preferred</i> alignments. All sizes are in bits. Specifying the <i>pref</i>
880 alignment is optional. If omitted, the preceding <tt>:</tt> should be omitted
882 <dt><tt>i<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
883 <dd>This specifies the alignment for an integer type of a given bit
884 <i>size</i>. The value of <i>size</i> must be in the range [1,2^23).</dd>
885 <dt><tt>v<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
886 <dd>This specifies the alignment for a vector type of a given bit
888 <dt><tt>f<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
889 <dd>This specifies the alignment for a floating point type of a given bit
890 <i>size</i>. The value of <i>size</i> must be either 32 (float) or 64
892 <dt><tt>a<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
893 <dd>This specifies the alignment for an aggregate type of a given bit
896 <p>When constructing the data layout for a given target, LLVM starts with a
897 default set of specifications which are then (possibly) overriden by the
898 specifications in the <tt>datalayout</tt> keyword. The default specifications
899 are given in this list:</p>
901 <li><tt>E</tt> - big endian</li>
902 <li><tt>p:32:64:64</tt> - 32-bit pointers with 64-bit alignment</li>
903 <li><tt>i1:8:8</tt> - i1 is 8-bit (byte) aligned</li>
904 <li><tt>i8:8:8</tt> - i8 is 8-bit (byte) aligned</li>
905 <li><tt>i16:16:16</tt> - i16 is 16-bit aligned</li>
906 <li><tt>i32:32:32</tt> - i32 is 32-bit aligned</li>
907 <li><tt>i64:32:64</tt> - i64 has abi alignment of 32-bits but preferred
908 alignment of 64-bits</li>
909 <li><tt>f32:32:32</tt> - float is 32-bit aligned</li>
910 <li><tt>f64:64:64</tt> - double is 64-bit aligned</li>
911 <li><tt>v64:64:64</tt> - 64-bit vector is 64-bit aligned</li>
912 <li><tt>v128:128:128</tt> - 128-bit vector is 128-bit aligned</li>
913 <li><tt>a0:0:1</tt> - aggregates are 8-bit aligned</li>
915 <p>When llvm is determining the alignment for a given type, it uses the
918 <li>If the type sought is an exact match for one of the specifications, that
919 specification is used.</li>
920 <li>If no match is found, and the type sought is an integer type, then the
921 smallest integer type that is larger than the bitwidth of the sought type is
922 used. If none of the specifications are larger than the bitwidth then the the
923 largest integer type is used. For example, given the default specifications
924 above, the i7 type will use the alignment of i8 (next largest) while both
925 i65 and i256 will use the alignment of i64 (largest specified).</li>
926 <li>If no match is found, and the type sought is a vector type, then the
927 largest vector type that is smaller than the sought vector type will be used
928 as a fall back. This happens because <128 x double> can be implemented in
929 terms of 64 <2 x double>, for example.</li>
933 <!-- *********************************************************************** -->
934 <div class="doc_section"> <a name="typesystem">Type System</a> </div>
935 <!-- *********************************************************************** -->
937 <div class="doc_text">
939 <p>The LLVM type system is one of the most important features of the
940 intermediate representation. Being typed enables a number of
941 optimizations to be performed on the IR directly, without having to do
942 extra analyses on the side before the transformation. A strong type
943 system makes it easier to read the generated code and enables novel
944 analyses and transformations that are not feasible to perform on normal
945 three address code representations.</p>
949 <!-- ======================================================================= -->
950 <div class="doc_subsection"> <a name="t_primitive">Primitive Types</a> </div>
951 <div class="doc_text">
952 <p>The primitive types are the fundamental building blocks of the LLVM
953 system. The current set of primitive types is as follows:</p>
955 <table class="layout">
960 <tr><th>Type</th><th>Description</th></tr>
961 <tr><td><tt><a name="t_void">void</a></tt></td><td>No value</td></tr>
962 <tr><td><tt>label</tt></td><td>Branch destination</td></tr>
969 <tr><th>Type</th><th>Description</th></tr>
970 <tr><td><tt>float</tt></td><td>32-bit floating point value</td></tr>
971 <tr><td><tt>double</tt></td><td>64-bit floating point value</td></tr>
979 <!-- _______________________________________________________________________ -->
980 <div class="doc_subsubsection"> <a name="t_classifications">Type
981 Classifications</a> </div>
982 <div class="doc_text">
983 <p>These different primitive types fall into a few useful
986 <table border="1" cellspacing="0" cellpadding="4">
988 <tr><th>Classification</th><th>Types</th></tr>
990 <td><a name="t_integer">integer</a></td>
991 <td><tt>i1, i2, i3, ... i8, ... i16, ... i32, ... i64, ... </tt></td>
994 <td><a name="t_floating">floating point</a></td>
995 <td><tt>float, double</tt></td>
998 <td><a name="t_firstclass">first class</a></td>
999 <td><tt>i1, ..., float, double, <br/>
1000 <a href="#t_pointer">pointer</a>,<a href="#t_vector">vector</a></tt>
1006 <p>The <a href="#t_firstclass">first class</a> types are perhaps the
1007 most important. Values of these types are the only ones which can be
1008 produced by instructions, passed as arguments, or used as operands to
1009 instructions. This means that all structures and arrays must be
1010 manipulated either by pointer or by component.</p>
1013 <!-- ======================================================================= -->
1014 <div class="doc_subsection"> <a name="t_derived">Derived Types</a> </div>
1016 <div class="doc_text">
1018 <p>The real power in LLVM comes from the derived types in the system.
1019 This is what allows a programmer to represent arrays, functions,
1020 pointers, and other useful types. Note that these derived types may be
1021 recursive: For example, it is possible to have a two dimensional array.</p>
1025 <!-- _______________________________________________________________________ -->
1026 <div class="doc_subsubsection"> <a name="t_integer">Integer Type</a> </div>
1028 <div class="doc_text">
1031 <p>The integer type is a very simple derived type that simply specifies an
1032 arbitrary bit width for the integer type desired. Any bit width from 1 bit to
1033 2^23-1 (about 8 million) can be specified.</p>
1041 <p>The number of bits the integer will occupy is specified by the <tt>N</tt>
1045 <table class="layout">
1055 <tt>i1942652</tt><br/>
1058 A boolean integer of 1 bit<br/>
1059 A nibble sized integer of 4 bits.<br/>
1060 A byte sized integer of 8 bits.<br/>
1061 A half word sized integer of 16 bits.<br/>
1062 A word sized integer of 32 bits.<br/>
1063 An integer whose bit width is the answer. <br/>
1064 A double word sized integer of 64 bits.<br/>
1065 A really big integer of over 1 million bits.<br/>
1071 <!-- _______________________________________________________________________ -->
1072 <div class="doc_subsubsection"> <a name="t_array">Array Type</a> </div>
1074 <div class="doc_text">
1078 <p>The array type is a very simple derived type that arranges elements
1079 sequentially in memory. The array type requires a size (number of
1080 elements) and an underlying data type.</p>
1085 [<# elements> x <elementtype>]
1088 <p>The number of elements is a constant integer value; elementtype may
1089 be any type with a size.</p>
1092 <table class="layout">
1095 <tt>[40 x i32 ]</tt><br/>
1096 <tt>[41 x i32 ]</tt><br/>
1097 <tt>[40 x i8]</tt><br/>
1100 Array of 40 32-bit integer values.<br/>
1101 Array of 41 32-bit integer values.<br/>
1102 Array of 40 8-bit integer values.<br/>
1106 <p>Here are some examples of multidimensional arrays:</p>
1107 <table class="layout">
1110 <tt>[3 x [4 x i32]]</tt><br/>
1111 <tt>[12 x [10 x float]]</tt><br/>
1112 <tt>[2 x [3 x [4 x i16]]]</tt><br/>
1115 3x4 array of 32-bit integer values.<br/>
1116 12x10 array of single precision floating point values.<br/>
1117 2x3x4 array of 16-bit integer values.<br/>
1122 <p>Note that 'variable sized arrays' can be implemented in LLVM with a zero
1123 length array. Normally, accesses past the end of an array are undefined in
1124 LLVM (e.g. it is illegal to access the 5th element of a 3 element array).
1125 As a special case, however, zero length arrays are recognized to be variable
1126 length. This allows implementation of 'pascal style arrays' with the LLVM
1127 type "{ i32, [0 x float]}", for example.</p>
1131 <!-- _______________________________________________________________________ -->
1132 <div class="doc_subsubsection"> <a name="t_function">Function Type</a> </div>
1133 <div class="doc_text">
1135 <p>The function type can be thought of as a function signature. It
1136 consists of a return type and a list of formal parameter types.
1137 Function types are usually used to build virtual function tables
1138 (which are structures of pointers to functions), for indirect function
1139 calls, and when defining a function.</p>
1141 The return type of a function type cannot be an aggregate type.
1144 <pre> <returntype> (<parameter list>)<br></pre>
1145 <p>...where '<tt><parameter list></tt>' is a comma-separated list of type
1146 specifiers. Optionally, the parameter list may include a type <tt>...</tt>,
1147 which indicates that the function takes a variable number of arguments.
1148 Variable argument functions can access their arguments with the <a
1149 href="#int_varargs">variable argument handling intrinsic</a> functions.</p>
1151 <table class="layout">
1153 <td class="left"><tt>i32 (i32)</tt></td>
1154 <td class="left">function taking an <tt>i32</tt>, returning an <tt>i32</tt>
1156 </tr><tr class="layout">
1157 <td class="left"><tt>float (i16 signext, i32 *) *
1159 <td class="left"><a href="#t_pointer">Pointer</a> to a function that takes
1160 an <tt>i16</tt> that should be sign extended and a
1161 <a href="#t_pointer">pointer</a> to <tt>i32</tt>, returning
1164 </tr><tr class="layout">
1165 <td class="left"><tt>i32 (i8*, ...)</tt></td>
1166 <td class="left">A vararg function that takes at least one
1167 <a href="#t_pointer">pointer</a> to <tt>i8 </tt> (char in C),
1168 which returns an integer. This is the signature for <tt>printf</tt> in
1175 <!-- _______________________________________________________________________ -->
1176 <div class="doc_subsubsection"> <a name="t_struct">Structure Type</a> </div>
1177 <div class="doc_text">
1179 <p>The structure type is used to represent a collection of data members
1180 together in memory. The packing of the field types is defined to match
1181 the ABI of the underlying processor. The elements of a structure may
1182 be any type that has a size.</p>
1183 <p>Structures are accessed using '<tt><a href="#i_load">load</a></tt>
1184 and '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a
1185 field with the '<tt><a href="#i_getelementptr">getelementptr</a></tt>'
1188 <pre> { <type list> }<br></pre>
1190 <table class="layout">
1192 <td class="left"><tt>{ i32, i32, i32 }</tt></td>
1193 <td class="left">A triple of three <tt>i32</tt> values</td>
1194 </tr><tr class="layout">
1195 <td class="left"><tt>{ float, i32 (i32) * }</tt></td>
1196 <td class="left">A pair, where the first element is a <tt>float</tt> and the
1197 second element is a <a href="#t_pointer">pointer</a> to a
1198 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
1199 an <tt>i32</tt>.</td>
1204 <!-- _______________________________________________________________________ -->
1205 <div class="doc_subsubsection"> <a name="t_pstruct">Packed Structure Type</a>
1207 <div class="doc_text">
1209 <p>The packed structure type is used to represent a collection of data members
1210 together in memory. There is no padding between fields. Further, the alignment
1211 of a packed structure is 1 byte. The elements of a packed structure may
1212 be any type that has a size.</p>
1213 <p>Structures are accessed using '<tt><a href="#i_load">load</a></tt>
1214 and '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a
1215 field with the '<tt><a href="#i_getelementptr">getelementptr</a></tt>'
1218 <pre> < { <type list> } > <br></pre>
1220 <table class="layout">
1222 <td class="left"><tt>< { i32, i32, i32 } ></tt></td>
1223 <td class="left">A triple of three <tt>i32</tt> values</td>
1224 </tr><tr class="layout">
1225 <td class="left"><tt>< { float, i32 (i32) * } ></tt></td>
1226 <td class="left">A pair, where the first element is a <tt>float</tt> and the
1227 second element is a <a href="#t_pointer">pointer</a> to a
1228 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
1229 an <tt>i32</tt>.</td>
1234 <!-- _______________________________________________________________________ -->
1235 <div class="doc_subsubsection"> <a name="t_pointer">Pointer Type</a> </div>
1236 <div class="doc_text">
1238 <p>As in many languages, the pointer type represents a pointer or
1239 reference to another object, which must live in memory.</p>
1241 <pre> <type> *<br></pre>
1243 <table class="layout">
1246 <tt>[4x i32]*</tt><br/>
1247 <tt>i32 (i32 *) *</tt><br/>
1250 A <a href="#t_pointer">pointer</a> to <a href="#t_array">array</a> of
1251 four <tt>i32</tt> values<br/>
1252 A <a href="#t_pointer">pointer</a> to a <a
1253 href="#t_function">function</a> that takes an <tt>i32*</tt>, returning an
1260 <!-- _______________________________________________________________________ -->
1261 <div class="doc_subsubsection"> <a name="t_vector">Vector Type</a> </div>
1262 <div class="doc_text">
1266 <p>A vector type is a simple derived type that represents a vector
1267 of elements. Vector types are used when multiple primitive data
1268 are operated in parallel using a single instruction (SIMD).
1269 A vector type requires a size (number of
1270 elements) and an underlying primitive data type. Vectors must have a power
1271 of two length (1, 2, 4, 8, 16 ...). Vector types are
1272 considered <a href="#t_firstclass">first class</a>.</p>
1277 < <# elements> x <elementtype> >
1280 <p>The number of elements is a constant integer value; elementtype may
1281 be any integer or floating point type.</p>
1285 <table class="layout">
1288 <tt><4 x i32></tt><br/>
1289 <tt><8 x float></tt><br/>
1290 <tt><2 x i64></tt><br/>
1293 Vector of 4 32-bit integer values.<br/>
1294 Vector of 8 floating-point values.<br/>
1295 Vector of 2 64-bit integer values.<br/>
1301 <!-- _______________________________________________________________________ -->
1302 <div class="doc_subsubsection"> <a name="t_opaque">Opaque Type</a> </div>
1303 <div class="doc_text">
1307 <p>Opaque types are used to represent unknown types in the system. This
1308 corresponds (for example) to the C notion of a foward declared structure type.
1309 In LLVM, opaque types can eventually be resolved to any type (not just a
1310 structure type).</p>
1320 <table class="layout">
1326 An opaque type.<br/>
1333 <!-- *********************************************************************** -->
1334 <div class="doc_section"> <a name="constants">Constants</a> </div>
1335 <!-- *********************************************************************** -->
1337 <div class="doc_text">
1339 <p>LLVM has several different basic types of constants. This section describes
1340 them all and their syntax.</p>
1344 <!-- ======================================================================= -->
1345 <div class="doc_subsection"><a name="simpleconstants">Simple Constants</a></div>
1347 <div class="doc_text">
1350 <dt><b>Boolean constants</b></dt>
1352 <dd>The two strings '<tt>true</tt>' and '<tt>false</tt>' are both valid
1353 constants of the <tt><a href="#t_primitive">i1</a></tt> type.
1356 <dt><b>Integer constants</b></dt>
1358 <dd>Standard integers (such as '4') are constants of the <a
1359 href="#t_integer">integer</a> type. Negative numbers may be used with
1363 <dt><b>Floating point constants</b></dt>
1365 <dd>Floating point constants use standard decimal notation (e.g. 123.421),
1366 exponential notation (e.g. 1.23421e+2), or a more precise hexadecimal
1367 notation (see below). Floating point constants must have a <a
1368 href="#t_floating">floating point</a> type. </dd>
1370 <dt><b>Null pointer constants</b></dt>
1372 <dd>The identifier '<tt>null</tt>' is recognized as a null pointer constant
1373 and must be of <a href="#t_pointer">pointer type</a>.</dd>
1377 <p>The one non-intuitive notation for constants is the optional hexadecimal form
1378 of floating point constants. For example, the form '<tt>double
1379 0x432ff973cafa8000</tt>' is equivalent to (but harder to read than) '<tt>double
1380 4.5e+15</tt>'. The only time hexadecimal floating point constants are required
1381 (and the only time that they are generated by the disassembler) is when a
1382 floating point constant must be emitted but it cannot be represented as a
1383 decimal floating point number. For example, NaN's, infinities, and other
1384 special values are represented in their IEEE hexadecimal format so that
1385 assembly and disassembly do not cause any bits to change in the constants.</p>
1389 <!-- ======================================================================= -->
1390 <div class="doc_subsection"><a name="aggregateconstants">Aggregate Constants</a>
1393 <div class="doc_text">
1394 <p>Aggregate constants arise from aggregation of simple constants
1395 and smaller aggregate constants.</p>
1398 <dt><b>Structure constants</b></dt>
1400 <dd>Structure constants are represented with notation similar to structure
1401 type definitions (a comma separated list of elements, surrounded by braces
1402 (<tt>{}</tt>)). For example: "<tt>{ i32 4, float 17.0, i32* %G }</tt>",
1403 where "<tt>%G</tt>" is declared as "<tt>@G = external global i32</tt>". Structure constants
1404 must have <a href="#t_struct">structure type</a>, and the number and
1405 types of elements must match those specified by the type.
1408 <dt><b>Array constants</b></dt>
1410 <dd>Array constants are represented with notation similar to array type
1411 definitions (a comma separated list of elements, surrounded by square brackets
1412 (<tt>[]</tt>)). For example: "<tt>[ i32 42, i32 11, i32 74 ]</tt>". Array
1413 constants must have <a href="#t_array">array type</a>, and the number and
1414 types of elements must match those specified by the type.
1417 <dt><b>Vector constants</b></dt>
1419 <dd>Vector constants are represented with notation similar to vector type
1420 definitions (a comma separated list of elements, surrounded by
1421 less-than/greater-than's (<tt><></tt>)). For example: "<tt>< i32 42,
1422 i32 11, i32 74, i32 100 ></tt>". Vector constants must have <a
1423 href="#t_vector">vector type</a>, and the number and types of elements must
1424 match those specified by the type.
1427 <dt><b>Zero initialization</b></dt>
1429 <dd>The string '<tt>zeroinitializer</tt>' can be used to zero initialize a
1430 value to zero of <em>any</em> type, including scalar and aggregate types.
1431 This is often used to avoid having to print large zero initializers (e.g. for
1432 large arrays) and is always exactly equivalent to using explicit zero
1439 <!-- ======================================================================= -->
1440 <div class="doc_subsection">
1441 <a name="globalconstants">Global Variable and Function Addresses</a>
1444 <div class="doc_text">
1446 <p>The addresses of <a href="#globalvars">global variables</a> and <a
1447 href="#functionstructure">functions</a> are always implicitly valid (link-time)
1448 constants. These constants are explicitly referenced when the <a
1449 href="#identifiers">identifier for the global</a> is used and always have <a
1450 href="#t_pointer">pointer</a> type. For example, the following is a legal LLVM
1453 <div class="doc_code">
1457 @Z = global [2 x i32*] [ i32* @X, i32* @Y ]
1463 <!-- ======================================================================= -->
1464 <div class="doc_subsection"><a name="undefvalues">Undefined Values</a></div>
1465 <div class="doc_text">
1466 <p>The string '<tt>undef</tt>' is recognized as a type-less constant that has
1467 no specific value. Undefined values may be of any type and be used anywhere
1468 a constant is permitted.</p>
1470 <p>Undefined values indicate to the compiler that the program is well defined
1471 no matter what value is used, giving the compiler more freedom to optimize.
1475 <!-- ======================================================================= -->
1476 <div class="doc_subsection"><a name="constantexprs">Constant Expressions</a>
1479 <div class="doc_text">
1481 <p>Constant expressions are used to allow expressions involving other constants
1482 to be used as constants. Constant expressions may be of any <a
1483 href="#t_firstclass">first class</a> type and may involve any LLVM operation
1484 that does not have side effects (e.g. load and call are not supported). The
1485 following is the syntax for constant expressions:</p>
1488 <dt><b><tt>trunc ( CST to TYPE )</tt></b></dt>
1489 <dd>Truncate a constant to another type. The bit size of CST must be larger
1490 than the bit size of TYPE. Both types must be integers.</dd>
1492 <dt><b><tt>zext ( CST to TYPE )</tt></b></dt>
1493 <dd>Zero extend a constant to another type. The bit size of CST must be
1494 smaller or equal to the bit size of TYPE. Both types must be integers.</dd>
1496 <dt><b><tt>sext ( CST to TYPE )</tt></b></dt>
1497 <dd>Sign extend a constant to another type. The bit size of CST must be
1498 smaller or equal to the bit size of TYPE. Both types must be integers.</dd>
1500 <dt><b><tt>fptrunc ( CST to TYPE )</tt></b></dt>
1501 <dd>Truncate a floating point constant to another floating point type. The
1502 size of CST must be larger than the size of TYPE. Both types must be
1503 floating point.</dd>
1505 <dt><b><tt>fpext ( CST to TYPE )</tt></b></dt>
1506 <dd>Floating point extend a constant to another type. The size of CST must be
1507 smaller or equal to the size of TYPE. Both types must be floating point.</dd>
1509 <dt><b><tt>fptoui ( CST to TYPE )</tt></b></dt>
1510 <dd>Convert a floating point constant to the corresponding unsigned integer
1511 constant. TYPE must be an integer type. CST must be floating point. If the
1512 value won't fit in the integer type, the results are undefined.</dd>
1514 <dt><b><tt>fptosi ( CST to TYPE )</tt></b></dt>
1515 <dd>Convert a floating point constant to the corresponding signed integer
1516 constant. TYPE must be an integer type. CST must be floating point. If the
1517 value won't fit in the integer type, the results are undefined.</dd>
1519 <dt><b><tt>uitofp ( CST to TYPE )</tt></b></dt>
1520 <dd>Convert an unsigned integer constant to the corresponding floating point
1521 constant. TYPE must be floating point. CST must be of integer type. If the
1522 value won't fit in the floating point type, the results are undefined.</dd>
1524 <dt><b><tt>sitofp ( CST to TYPE )</tt></b></dt>
1525 <dd>Convert a signed integer constant to the corresponding floating point
1526 constant. TYPE must be floating point. CST must be of integer type. If the
1527 value won't fit in the floating point type, the results are undefined.</dd>
1529 <dt><b><tt>ptrtoint ( CST to TYPE )</tt></b></dt>
1530 <dd>Convert a pointer typed constant to the corresponding integer constant
1531 TYPE must be an integer type. CST must be of pointer type. The CST value is
1532 zero extended, truncated, or unchanged to make it fit in TYPE.</dd>
1534 <dt><b><tt>inttoptr ( CST to TYPE )</tt></b></dt>
1535 <dd>Convert a integer constant to a pointer constant. TYPE must be a
1536 pointer type. CST must be of integer type. The CST value is zero extended,
1537 truncated, or unchanged to make it fit in a pointer size. This one is
1538 <i>really</i> dangerous!</dd>
1540 <dt><b><tt>bitcast ( CST to TYPE )</tt></b></dt>
1541 <dd>Convert a constant, CST, to another TYPE. The size of CST and TYPE must be
1542 identical (same number of bits). The conversion is done as if the CST value
1543 was stored to memory and read back as TYPE. In other words, no bits change
1544 with this operator, just the type. This can be used for conversion of
1545 vector types to any other type, as long as they have the same bit width. For
1546 pointers it is only valid to cast to another pointer type.
1549 <dt><b><tt>getelementptr ( CSTPTR, IDX0, IDX1, ... )</tt></b></dt>
1551 <dd>Perform the <a href="#i_getelementptr">getelementptr operation</a> on
1552 constants. As with the <a href="#i_getelementptr">getelementptr</a>
1553 instruction, the index list may have zero or more indexes, which are required
1554 to make sense for the type of "CSTPTR".</dd>
1556 <dt><b><tt>select ( COND, VAL1, VAL2 )</tt></b></dt>
1558 <dd>Perform the <a href="#i_select">select operation</a> on
1561 <dt><b><tt>icmp COND ( VAL1, VAL2 )</tt></b></dt>
1562 <dd>Performs the <a href="#i_icmp">icmp operation</a> on constants.</dd>
1564 <dt><b><tt>fcmp COND ( VAL1, VAL2 )</tt></b></dt>
1565 <dd>Performs the <a href="#i_fcmp">fcmp operation</a> on constants.</dd>
1567 <dt><b><tt>extractelement ( VAL, IDX )</tt></b></dt>
1569 <dd>Perform the <a href="#i_extractelement">extractelement
1570 operation</a> on constants.
1572 <dt><b><tt>insertelement ( VAL, ELT, IDX )</tt></b></dt>
1574 <dd>Perform the <a href="#i_insertelement">insertelement
1575 operation</a> on constants.</dd>
1578 <dt><b><tt>shufflevector ( VEC1, VEC2, IDXMASK )</tt></b></dt>
1580 <dd>Perform the <a href="#i_shufflevector">shufflevector
1581 operation</a> on constants.</dd>
1583 <dt><b><tt>OPCODE ( LHS, RHS )</tt></b></dt>
1585 <dd>Perform the specified operation of the LHS and RHS constants. OPCODE may
1586 be any of the <a href="#binaryops">binary</a> or <a href="#bitwiseops">bitwise
1587 binary</a> operations. The constraints on operands are the same as those for
1588 the corresponding instruction (e.g. no bitwise operations on floating point
1589 values are allowed).</dd>
1593 <!-- *********************************************************************** -->
1594 <div class="doc_section"> <a name="othervalues">Other Values</a> </div>
1595 <!-- *********************************************************************** -->
1597 <!-- ======================================================================= -->
1598 <div class="doc_subsection">
1599 <a name="inlineasm">Inline Assembler Expressions</a>
1602 <div class="doc_text">
1605 LLVM supports inline assembler expressions (as opposed to <a href="#moduleasm">
1606 Module-Level Inline Assembly</a>) through the use of a special value. This
1607 value represents the inline assembler as a string (containing the instructions
1608 to emit), a list of operand constraints (stored as a string), and a flag that
1609 indicates whether or not the inline asm expression has side effects. An example
1610 inline assembler expression is:
1613 <div class="doc_code">
1615 i32 (i32) asm "bswap $0", "=r,r"
1620 Inline assembler expressions may <b>only</b> be used as the callee operand of
1621 a <a href="#i_call"><tt>call</tt> instruction</a>. Thus, typically we have:
1624 <div class="doc_code">
1626 %X = call i32 asm "<a href="#int_bswap">bswap</a> $0", "=r,r"(i32 %Y)
1631 Inline asms with side effects not visible in the constraint list must be marked
1632 as having side effects. This is done through the use of the
1633 '<tt>sideeffect</tt>' keyword, like so:
1636 <div class="doc_code">
1638 call void asm sideeffect "eieio", ""()
1642 <p>TODO: The format of the asm and constraints string still need to be
1643 documented here. Constraints on what can be done (e.g. duplication, moving, etc
1644 need to be documented).
1649 <!-- *********************************************************************** -->
1650 <div class="doc_section"> <a name="instref">Instruction Reference</a> </div>
1651 <!-- *********************************************************************** -->
1653 <div class="doc_text">
1655 <p>The LLVM instruction set consists of several different
1656 classifications of instructions: <a href="#terminators">terminator
1657 instructions</a>, <a href="#binaryops">binary instructions</a>,
1658 <a href="#bitwiseops">bitwise binary instructions</a>, <a
1659 href="#memoryops">memory instructions</a>, and <a href="#otherops">other
1660 instructions</a>.</p>
1664 <!-- ======================================================================= -->
1665 <div class="doc_subsection"> <a name="terminators">Terminator
1666 Instructions</a> </div>
1668 <div class="doc_text">
1670 <p>As mentioned <a href="#functionstructure">previously</a>, every
1671 basic block in a program ends with a "Terminator" instruction, which
1672 indicates which block should be executed after the current block is
1673 finished. These terminator instructions typically yield a '<tt>void</tt>'
1674 value: they produce control flow, not values (the one exception being
1675 the '<a href="#i_invoke"><tt>invoke</tt></a>' instruction).</p>
1676 <p>There are six different terminator instructions: the '<a
1677 href="#i_ret"><tt>ret</tt></a>' instruction, the '<a href="#i_br"><tt>br</tt></a>'
1678 instruction, the '<a href="#i_switch"><tt>switch</tt></a>' instruction,
1679 the '<a href="#i_invoke"><tt>invoke</tt></a>' instruction, the '<a
1680 href="#i_unwind"><tt>unwind</tt></a>' instruction, and the '<a
1681 href="#i_unreachable"><tt>unreachable</tt></a>' instruction.</p>
1685 <!-- _______________________________________________________________________ -->
1686 <div class="doc_subsubsection"> <a name="i_ret">'<tt>ret</tt>'
1687 Instruction</a> </div>
1688 <div class="doc_text">
1690 <pre> ret <type> <value> <i>; Return a value from a non-void function</i>
1691 ret void <i>; Return from void function</i>
1694 <p>The '<tt>ret</tt>' instruction is used to return control flow (and a
1695 value) from a function back to the caller.</p>
1696 <p>There are two forms of the '<tt>ret</tt>' instruction: one that
1697 returns a value and then causes control flow, and one that just causes
1698 control flow to occur.</p>
1700 <p>The '<tt>ret</tt>' instruction may return any '<a
1701 href="#t_firstclass">first class</a>' type. Notice that a function is
1702 not <a href="#wellformed">well formed</a> if there exists a '<tt>ret</tt>'
1703 instruction inside of the function that returns a value that does not
1704 match the return type of the function.</p>
1706 <p>When the '<tt>ret</tt>' instruction is executed, control flow
1707 returns back to the calling function's context. If the caller is a "<a
1708 href="#i_call"><tt>call</tt></a>" instruction, execution continues at
1709 the instruction after the call. If the caller was an "<a
1710 href="#i_invoke"><tt>invoke</tt></a>" instruction, execution continues
1711 at the beginning of the "normal" destination block. If the instruction
1712 returns a value, that value shall set the call or invoke instruction's
1715 <pre> ret i32 5 <i>; Return an integer value of 5</i>
1716 ret void <i>; Return from a void function</i>
1719 <!-- _______________________________________________________________________ -->
1720 <div class="doc_subsubsection"> <a name="i_br">'<tt>br</tt>' Instruction</a> </div>
1721 <div class="doc_text">
1723 <pre> br i1 <cond>, label <iftrue>, label <iffalse><br> br label <dest> <i>; Unconditional branch</i>
1726 <p>The '<tt>br</tt>' instruction is used to cause control flow to
1727 transfer to a different basic block in the current function. There are
1728 two forms of this instruction, corresponding to a conditional branch
1729 and an unconditional branch.</p>
1731 <p>The conditional branch form of the '<tt>br</tt>' instruction takes a
1732 single '<tt>i1</tt>' value and two '<tt>label</tt>' values. The
1733 unconditional form of the '<tt>br</tt>' instruction takes a single
1734 '<tt>label</tt>' value as a target.</p>
1736 <p>Upon execution of a conditional '<tt>br</tt>' instruction, the '<tt>i1</tt>'
1737 argument is evaluated. If the value is <tt>true</tt>, control flows
1738 to the '<tt>iftrue</tt>' <tt>label</tt> argument. If "cond" is <tt>false</tt>,
1739 control flows to the '<tt>iffalse</tt>' <tt>label</tt> argument.</p>
1741 <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
1742 href="#i_ret">ret</a> i32 1<br>IfUnequal:<br> <a href="#i_ret">ret</a> i32 0<br></pre>
1744 <!-- _______________________________________________________________________ -->
1745 <div class="doc_subsubsection">
1746 <a name="i_switch">'<tt>switch</tt>' Instruction</a>
1749 <div class="doc_text">
1753 switch <intty> <value>, label <defaultdest> [ <intty> <val>, label <dest> ... ]
1758 <p>The '<tt>switch</tt>' instruction is used to transfer control flow to one of
1759 several different places. It is a generalization of the '<tt>br</tt>'
1760 instruction, allowing a branch to occur to one of many possible
1766 <p>The '<tt>switch</tt>' instruction uses three parameters: an integer
1767 comparison value '<tt>value</tt>', a default '<tt>label</tt>' destination, and
1768 an array of pairs of comparison value constants and '<tt>label</tt>'s. The
1769 table is not allowed to contain duplicate constant entries.</p>
1773 <p>The <tt>switch</tt> instruction specifies a table of values and
1774 destinations. When the '<tt>switch</tt>' instruction is executed, this
1775 table is searched for the given value. If the value is found, control flow is
1776 transfered to the corresponding destination; otherwise, control flow is
1777 transfered to the default destination.</p>
1779 <h5>Implementation:</h5>
1781 <p>Depending on properties of the target machine and the particular
1782 <tt>switch</tt> instruction, this instruction may be code generated in different
1783 ways. For example, it could be generated as a series of chained conditional
1784 branches or with a lookup table.</p>
1789 <i>; Emulate a conditional br instruction</i>
1790 %Val = <a href="#i_zext">zext</a> i1 %value to i32
1791 switch i32 %Val, label %truedest [i32 0, label %falsedest ]
1793 <i>; Emulate an unconditional br instruction</i>
1794 switch i32 0, label %dest [ ]
1796 <i>; Implement a jump table:</i>
1797 switch i32 %val, label %otherwise [ i32 0, label %onzero
1799 i32 2, label %ontwo ]
1803 <!-- _______________________________________________________________________ -->
1804 <div class="doc_subsubsection">
1805 <a name="i_invoke">'<tt>invoke</tt>' Instruction</a>
1808 <div class="doc_text">
1813 <result> = invoke [<a href="#callingconv">cconv</a>] <ptr to function ty> %<function ptr val>(<function args>)
1814 to label <normal label> unwind label <exception label>
1819 <p>The '<tt>invoke</tt>' instruction causes control to transfer to a specified
1820 function, with the possibility of control flow transfer to either the
1821 '<tt>normal</tt>' label or the
1822 '<tt>exception</tt>' label. If the callee function returns with the
1823 "<tt><a href="#i_ret">ret</a></tt>" instruction, control flow will return to the
1824 "normal" label. If the callee (or any indirect callees) returns with the "<a
1825 href="#i_unwind"><tt>unwind</tt></a>" instruction, control is interrupted and
1826 continued at the dynamically nearest "exception" label.</p>
1830 <p>This instruction requires several arguments:</p>
1834 The optional "cconv" marker indicates which <a href="#callingconv">calling
1835 convention</a> the call should use. If none is specified, the call defaults
1836 to using C calling conventions.
1838 <li>'<tt>ptr to function ty</tt>': shall be the signature of the pointer to
1839 function value being invoked. In most cases, this is a direct function
1840 invocation, but indirect <tt>invoke</tt>s are just as possible, branching off
1841 an arbitrary pointer to function value.
1844 <li>'<tt>function ptr val</tt>': An LLVM value containing a pointer to a
1845 function to be invoked. </li>
1847 <li>'<tt>function args</tt>': argument list whose types match the function
1848 signature argument types. If the function signature indicates the function
1849 accepts a variable number of arguments, the extra arguments can be
1852 <li>'<tt>normal label</tt>': the label reached when the called function
1853 executes a '<tt><a href="#i_ret">ret</a></tt>' instruction. </li>
1855 <li>'<tt>exception label</tt>': the label reached when a callee returns with
1856 the <a href="#i_unwind"><tt>unwind</tt></a> instruction. </li>
1862 <p>This instruction is designed to operate as a standard '<tt><a
1863 href="#i_call">call</a></tt>' instruction in most regards. The primary
1864 difference is that it establishes an association with a label, which is used by
1865 the runtime library to unwind the stack.</p>
1867 <p>This instruction is used in languages with destructors to ensure that proper
1868 cleanup is performed in the case of either a <tt>longjmp</tt> or a thrown
1869 exception. Additionally, this is important for implementation of
1870 '<tt>catch</tt>' clauses in high-level languages that support them.</p>
1874 %retval = invoke i32 %Test(i32 15) to label %Continue
1875 unwind label %TestCleanup <i>; {i32}:retval set</i>
1876 %retval = invoke <a href="#callingconv">coldcc</a> i32 %Test(i32 15) to label %Continue
1877 unwind label %TestCleanup <i>; {i32}:retval set</i>
1882 <!-- _______________________________________________________________________ -->
1884 <div class="doc_subsubsection"> <a name="i_unwind">'<tt>unwind</tt>'
1885 Instruction</a> </div>
1887 <div class="doc_text">
1896 <p>The '<tt>unwind</tt>' instruction unwinds the stack, continuing control flow
1897 at the first callee in the dynamic call stack which used an <a
1898 href="#i_invoke"><tt>invoke</tt></a> instruction to perform the call. This is
1899 primarily used to implement exception handling.</p>
1903 <p>The '<tt>unwind</tt>' intrinsic causes execution of the current function to
1904 immediately halt. The dynamic call stack is then searched for the first <a
1905 href="#i_invoke"><tt>invoke</tt></a> instruction on the call stack. Once found,
1906 execution continues at the "exceptional" destination block specified by the
1907 <tt>invoke</tt> instruction. If there is no <tt>invoke</tt> instruction in the
1908 dynamic call chain, undefined behavior results.</p>
1911 <!-- _______________________________________________________________________ -->
1913 <div class="doc_subsubsection"> <a name="i_unreachable">'<tt>unreachable</tt>'
1914 Instruction</a> </div>
1916 <div class="doc_text">
1925 <p>The '<tt>unreachable</tt>' instruction has no defined semantics. This
1926 instruction is used to inform the optimizer that a particular portion of the
1927 code is not reachable. This can be used to indicate that the code after a
1928 no-return function cannot be reached, and other facts.</p>
1932 <p>The '<tt>unreachable</tt>' instruction has no defined semantics.</p>
1937 <!-- ======================================================================= -->
1938 <div class="doc_subsection"> <a name="binaryops">Binary Operations</a> </div>
1939 <div class="doc_text">
1940 <p>Binary operators are used to do most of the computation in a
1941 program. They require two operands, execute an operation on them, and
1942 produce a single value. The operands might represent
1943 multiple data, as is the case with the <a href="#t_vector">vector</a> data type.
1944 The result value of a binary operator is not
1945 necessarily the same type as its operands.</p>
1946 <p>There are several different binary operators:</p>
1948 <!-- _______________________________________________________________________ -->
1949 <div class="doc_subsubsection"> <a name="i_add">'<tt>add</tt>'
1950 Instruction</a> </div>
1951 <div class="doc_text">
1953 <pre> <result> = add <ty> <var1>, <var2> <i>; yields {ty}:result</i>
1956 <p>The '<tt>add</tt>' instruction returns the sum of its two operands.</p>
1958 <p>The two arguments to the '<tt>add</tt>' instruction must be either <a
1959 href="#t_integer">integer</a> or <a href="#t_floating">floating point</a> values.
1960 This instruction can also take <a href="#t_vector">vector</a> versions of the values.
1961 Both arguments must have identical types.</p>
1963 <p>The value produced is the integer or floating point sum of the two
1966 <pre> <result> = add i32 4, %var <i>; yields {i32}:result = 4 + %var</i>
1969 <!-- _______________________________________________________________________ -->
1970 <div class="doc_subsubsection"> <a name="i_sub">'<tt>sub</tt>'
1971 Instruction</a> </div>
1972 <div class="doc_text">
1974 <pre> <result> = sub <ty> <var1>, <var2> <i>; yields {ty}:result</i>
1977 <p>The '<tt>sub</tt>' instruction returns the difference of its two
1979 <p>Note that the '<tt>sub</tt>' instruction is used to represent the '<tt>neg</tt>'
1980 instruction present in most other intermediate representations.</p>
1982 <p>The two arguments to the '<tt>sub</tt>' instruction must be either <a
1983 href="#t_integer">integer</a> or <a href="#t_floating">floating point</a>
1985 This instruction can also take <a href="#t_vector">vector</a> versions of the values.
1986 Both arguments must have identical types.</p>
1988 <p>The value produced is the integer or floating point difference of
1989 the two operands.</p>
1992 <result> = sub i32 4, %var <i>; yields {i32}:result = 4 - %var</i>
1993 <result> = sub i32 0, %val <i>; yields {i32}:result = -%var</i>
1996 <!-- _______________________________________________________________________ -->
1997 <div class="doc_subsubsection"> <a name="i_mul">'<tt>mul</tt>'
1998 Instruction</a> </div>
1999 <div class="doc_text">
2001 <pre> <result> = mul <ty> <var1>, <var2> <i>; yields {ty}:result</i>
2004 <p>The '<tt>mul</tt>' instruction returns the product of its two
2007 <p>The two arguments to the '<tt>mul</tt>' instruction must be either <a
2008 href="#t_integer">integer</a> or <a href="#t_floating">floating point</a>
2010 This instruction can also take <a href="#t_vector">vector</a> versions of the values.
2011 Both arguments must have identical types.</p>
2013 <p>The value produced is the integer or floating point product of the
2015 <p>Because the operands are the same width, the result of an integer
2016 multiplication is the same whether the operands should be deemed unsigned or
2019 <pre> <result> = mul i32 4, %var <i>; yields {i32}:result = 4 * %var</i>
2022 <!-- _______________________________________________________________________ -->
2023 <div class="doc_subsubsection"> <a name="i_udiv">'<tt>udiv</tt>' Instruction
2025 <div class="doc_text">
2027 <pre> <result> = udiv <ty> <var1>, <var2> <i>; yields {ty}:result</i>
2030 <p>The '<tt>udiv</tt>' instruction returns the quotient of its two
2033 <p>The two arguments to the '<tt>udiv</tt>' instruction must be
2034 <a href="#t_integer">integer</a> values. Both arguments must have identical
2035 types. This instruction can also take <a href="#t_vector">vector</a> versions
2036 of the values in which case the elements must be integers.</p>
2038 <p>The value produced is the unsigned integer quotient of the two operands. This
2039 instruction always performs an unsigned division operation, regardless of
2040 whether the arguments are unsigned or not.</p>
2042 <pre> <result> = udiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
2045 <!-- _______________________________________________________________________ -->
2046 <div class="doc_subsubsection"> <a name="i_sdiv">'<tt>sdiv</tt>' Instruction
2048 <div class="doc_text">
2050 <pre> <result> = sdiv <ty> <var1>, <var2> <i>; yields {ty}:result</i>
2053 <p>The '<tt>sdiv</tt>' instruction returns the quotient of its two
2056 <p>The two arguments to the '<tt>sdiv</tt>' instruction must be
2057 <a href="#t_integer">integer</a> values. Both arguments must have identical
2058 types. This instruction can also take <a href="#t_vector">vector</a> versions
2059 of the values in which case the elements must be integers.</p>
2061 <p>The value produced is the signed integer quotient of the two operands. This
2062 instruction always performs a signed division operation, regardless of whether
2063 the arguments are signed or not.</p>
2065 <pre> <result> = sdiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
2068 <!-- _______________________________________________________________________ -->
2069 <div class="doc_subsubsection"> <a name="i_fdiv">'<tt>fdiv</tt>'
2070 Instruction</a> </div>
2071 <div class="doc_text">
2073 <pre> <result> = fdiv <ty> <var1>, <var2> <i>; yields {ty}:result</i>
2076 <p>The '<tt>fdiv</tt>' instruction returns the quotient of its two
2079 <p>The two arguments to the '<tt>fdiv</tt>' instruction must be
2080 <a href="#t_floating">floating point</a> values. Both arguments must have
2081 identical types. This instruction can also take <a href="#t_vector">vector</a>
2082 versions of floating point values.</p>
2084 <p>The value produced is the floating point quotient of the two operands.</p>
2086 <pre> <result> = fdiv float 4.0, %var <i>; yields {float}:result = 4.0 / %var</i>
2089 <!-- _______________________________________________________________________ -->
2090 <div class="doc_subsubsection"> <a name="i_urem">'<tt>urem</tt>' Instruction</a>
2092 <div class="doc_text">
2094 <pre> <result> = urem <ty> <var1>, <var2> <i>; yields {ty}:result</i>
2097 <p>The '<tt>urem</tt>' instruction returns the remainder from the
2098 unsigned division of its two arguments.</p>
2100 <p>The two arguments to the '<tt>urem</tt>' instruction must be
2101 <a href="#t_integer">integer</a> values. Both arguments must have identical
2104 <p>This instruction returns the unsigned integer <i>remainder</i> of a division.
2105 This instruction always performs an unsigned division to get the remainder,
2106 regardless of whether the arguments are unsigned or not.</p>
2108 <pre> <result> = urem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
2112 <!-- _______________________________________________________________________ -->
2113 <div class="doc_subsubsection"> <a name="i_srem">'<tt>srem</tt>'
2114 Instruction</a> </div>
2115 <div class="doc_text">
2117 <pre> <result> = srem <ty> <var1>, <var2> <i>; yields {ty}:result</i>
2120 <p>The '<tt>srem</tt>' instruction returns the remainder from the
2121 signed division of its two operands.</p>
2123 <p>The two arguments to the '<tt>srem</tt>' instruction must be
2124 <a href="#t_integer">integer</a> values. Both arguments must have identical
2127 <p>This instruction returns the <i>remainder</i> of a division (where the result
2128 has the same sign as the dividend, <tt>var1</tt>), not the <i>modulo</i>
2129 operator (where the result has the same sign as the divisor, <tt>var2</tt>) of
2130 a value. For more information about the difference, see <a
2131 href="http://mathforum.org/dr.math/problems/anne.4.28.99.html">The
2132 Math Forum</a>. For a table of how this is implemented in various languages,
2133 please see <a href="http://en.wikipedia.org/wiki/Modulo_operation">
2134 Wikipedia: modulo operation</a>.</p>
2136 <pre> <result> = srem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
2140 <!-- _______________________________________________________________________ -->
2141 <div class="doc_subsubsection"> <a name="i_frem">'<tt>frem</tt>'
2142 Instruction</a> </div>
2143 <div class="doc_text">
2145 <pre> <result> = frem <ty> <var1>, <var2> <i>; yields {ty}:result</i>
2148 <p>The '<tt>frem</tt>' instruction returns the remainder from the
2149 division of its two operands.</p>
2151 <p>The two arguments to the '<tt>frem</tt>' instruction must be
2152 <a href="#t_floating">floating point</a> values. Both arguments must have
2153 identical types.</p>
2155 <p>This instruction returns the <i>remainder</i> of a division.</p>
2157 <pre> <result> = frem float 4.0, %var <i>; yields {float}:result = 4.0 % %var</i>
2161 <!-- ======================================================================= -->
2162 <div class="doc_subsection"> <a name="bitwiseops">Bitwise Binary
2163 Operations</a> </div>
2164 <div class="doc_text">
2165 <p>Bitwise binary operators are used to do various forms of
2166 bit-twiddling in a program. They are generally very efficient
2167 instructions and can commonly be strength reduced from other
2168 instructions. They require two operands, execute an operation on them,
2169 and produce a single value. The resulting value of the bitwise binary
2170 operators is always the same type as its first operand.</p>
2173 <!-- _______________________________________________________________________ -->
2174 <div class="doc_subsubsection"> <a name="i_shl">'<tt>shl</tt>'
2175 Instruction</a> </div>
2176 <div class="doc_text">
2178 <pre> <result> = shl <ty> <var1>, <var2> <i>; yields {ty}:result</i>
2181 <p>The '<tt>shl</tt>' instruction returns the first operand shifted to
2182 the left a specified number of bits.</p>
2184 <p>Both arguments to the '<tt>shl</tt>' instruction must be the same <a
2185 href="#t_integer">integer</a> type.</p>
2187 <p>The value produced is <tt>var1</tt> * 2<sup><tt>var2</tt></sup>.</p>
2188 <h5>Example:</h5><pre>
2189 <result> = shl i32 4, %var <i>; yields {i32}: 4 << %var</i>
2190 <result> = shl i32 4, 2 <i>; yields {i32}: 16</i>
2191 <result> = shl i32 1, 10 <i>; yields {i32}: 1024</i>
2194 <!-- _______________________________________________________________________ -->
2195 <div class="doc_subsubsection"> <a name="i_lshr">'<tt>lshr</tt>'
2196 Instruction</a> </div>
2197 <div class="doc_text">
2199 <pre> <result> = lshr <ty> <var1>, <var2> <i>; yields {ty}:result</i>
2203 <p>The '<tt>lshr</tt>' instruction (logical shift right) returns the first
2204 operand shifted to the right a specified number of bits with zero fill.</p>
2207 <p>Both arguments to the '<tt>lshr</tt>' instruction must be the same
2208 <a href="#t_integer">integer</a> type.</p>
2211 <p>This instruction always performs a logical shift right operation. The most
2212 significant bits of the result will be filled with zero bits after the
2217 <result> = lshr i32 4, 1 <i>; yields {i32}:result = 2</i>
2218 <result> = lshr i32 4, 2 <i>; yields {i32}:result = 1</i>
2219 <result> = lshr i8 4, 3 <i>; yields {i8}:result = 0</i>
2220 <result> = lshr i8 -2, 1 <i>; yields {i8}:result = 0x7FFFFFFF </i>
2224 <!-- _______________________________________________________________________ -->
2225 <div class="doc_subsubsection"> <a name="i_ashr">'<tt>ashr</tt>'
2226 Instruction</a> </div>
2227 <div class="doc_text">
2230 <pre> <result> = ashr <ty> <var1>, <var2> <i>; yields {ty}:result</i>
2234 <p>The '<tt>ashr</tt>' instruction (arithmetic shift right) returns the first
2235 operand shifted to the right a specified number of bits with sign extension.</p>
2238 <p>Both arguments to the '<tt>ashr</tt>' instruction must be the same
2239 <a href="#t_integer">integer</a> type.</p>
2242 <p>This instruction always performs an arithmetic shift right operation,
2243 The most significant bits of the result will be filled with the sign bit
2244 of <tt>var1</tt>.</p>
2248 <result> = ashr i32 4, 1 <i>; yields {i32}:result = 2</i>
2249 <result> = ashr i32 4, 2 <i>; yields {i32}:result = 1</i>
2250 <result> = ashr i8 4, 3 <i>; yields {i8}:result = 0</i>
2251 <result> = ashr i8 -2, 1 <i>; yields {i8}:result = -1</i>
2255 <!-- _______________________________________________________________________ -->
2256 <div class="doc_subsubsection"> <a name="i_and">'<tt>and</tt>'
2257 Instruction</a> </div>
2258 <div class="doc_text">
2260 <pre> <result> = and <ty> <var1>, <var2> <i>; yields {ty}:result</i>
2263 <p>The '<tt>and</tt>' instruction returns the bitwise logical and of
2264 its two operands.</p>
2266 <p>The two arguments to the '<tt>and</tt>' instruction must be <a
2267 href="#t_integer">integer</a> values. Both arguments must have
2268 identical types.</p>
2270 <p>The truth table used for the '<tt>and</tt>' instruction is:</p>
2272 <div style="align: center">
2273 <table border="1" cellspacing="0" cellpadding="4">
2304 <pre> <result> = and i32 4, %var <i>; yields {i32}:result = 4 & %var</i>
2305 <result> = and i32 15, 40 <i>; yields {i32}:result = 8</i>
2306 <result> = and i32 4, 8 <i>; yields {i32}:result = 0</i>
2309 <!-- _______________________________________________________________________ -->
2310 <div class="doc_subsubsection"> <a name="i_or">'<tt>or</tt>' Instruction</a> </div>
2311 <div class="doc_text">
2313 <pre> <result> = or <ty> <var1>, <var2> <i>; yields {ty}:result</i>
2316 <p>The '<tt>or</tt>' instruction returns the bitwise logical inclusive
2317 or of its two operands.</p>
2319 <p>The two arguments to the '<tt>or</tt>' instruction must be <a
2320 href="#t_integer">integer</a> values. Both arguments must have
2321 identical types.</p>
2323 <p>The truth table used for the '<tt>or</tt>' instruction is:</p>
2325 <div style="align: center">
2326 <table border="1" cellspacing="0" cellpadding="4">
2357 <pre> <result> = or i32 4, %var <i>; yields {i32}:result = 4 | %var</i>
2358 <result> = or i32 15, 40 <i>; yields {i32}:result = 47</i>
2359 <result> = or i32 4, 8 <i>; yields {i32}:result = 12</i>
2362 <!-- _______________________________________________________________________ -->
2363 <div class="doc_subsubsection"> <a name="i_xor">'<tt>xor</tt>'
2364 Instruction</a> </div>
2365 <div class="doc_text">
2367 <pre> <result> = xor <ty> <var1>, <var2> <i>; yields {ty}:result</i>
2370 <p>The '<tt>xor</tt>' instruction returns the bitwise logical exclusive
2371 or of its two operands. The <tt>xor</tt> is used to implement the
2372 "one's complement" operation, which is the "~" operator in C.</p>
2374 <p>The two arguments to the '<tt>xor</tt>' instruction must be <a
2375 href="#t_integer">integer</a> values. Both arguments must have
2376 identical types.</p>
2378 <p>The truth table used for the '<tt>xor</tt>' instruction is:</p>
2380 <div style="align: center">
2381 <table border="1" cellspacing="0" cellpadding="4">
2413 <pre> <result> = xor i32 4, %var <i>; yields {i32}:result = 4 ^ %var</i>
2414 <result> = xor i32 15, 40 <i>; yields {i32}:result = 39</i>
2415 <result> = xor i32 4, 8 <i>; yields {i32}:result = 12</i>
2416 <result> = xor i32 %V, -1 <i>; yields {i32}:result = ~%V</i>
2420 <!-- ======================================================================= -->
2421 <div class="doc_subsection">
2422 <a name="vectorops">Vector Operations</a>
2425 <div class="doc_text">
2427 <p>LLVM supports several instructions to represent vector operations in a
2428 target-independent manner. These instructions cover the element-access and
2429 vector-specific operations needed to process vectors effectively. While LLVM
2430 does directly support these vector operations, many sophisticated algorithms
2431 will want to use target-specific intrinsics to take full advantage of a specific
2436 <!-- _______________________________________________________________________ -->
2437 <div class="doc_subsubsection">
2438 <a name="i_extractelement">'<tt>extractelement</tt>' Instruction</a>
2441 <div class="doc_text">
2446 <result> = extractelement <n x <ty>> <val>, i32 <idx> <i>; yields <ty></i>
2452 The '<tt>extractelement</tt>' instruction extracts a single scalar
2453 element from a vector at a specified index.
2460 The first operand of an '<tt>extractelement</tt>' instruction is a
2461 value of <a href="#t_vector">vector</a> type. The second operand is
2462 an index indicating the position from which to extract the element.
2463 The index may be a variable.</p>
2468 The result is a scalar of the same type as the element type of
2469 <tt>val</tt>. Its value is the value at position <tt>idx</tt> of
2470 <tt>val</tt>. If <tt>idx</tt> exceeds the length of <tt>val</tt>, the
2471 results are undefined.
2477 %result = extractelement <4 x i32> %vec, i32 0 <i>; yields i32</i>
2482 <!-- _______________________________________________________________________ -->
2483 <div class="doc_subsubsection">
2484 <a name="i_insertelement">'<tt>insertelement</tt>' Instruction</a>
2487 <div class="doc_text">
2492 <result> = insertelement <n x <ty>> <val>, <ty> <elt>, i32 <idx> <i>; yields <n x <ty>></i>
2498 The '<tt>insertelement</tt>' instruction inserts a scalar
2499 element into a vector at a specified index.
2506 The first operand of an '<tt>insertelement</tt>' instruction is a
2507 value of <a href="#t_vector">vector</a> type. The second operand is a
2508 scalar value whose type must equal the element type of the first
2509 operand. The third operand is an index indicating the position at
2510 which to insert the value. The index may be a variable.</p>
2515 The result is a vector of the same type as <tt>val</tt>. Its
2516 element values are those of <tt>val</tt> except at position
2517 <tt>idx</tt>, where it gets the value <tt>elt</tt>. If <tt>idx</tt>
2518 exceeds the length of <tt>val</tt>, the results are undefined.
2524 %result = insertelement <4 x i32> %vec, i32 1, i32 0 <i>; yields <4 x i32></i>
2528 <!-- _______________________________________________________________________ -->
2529 <div class="doc_subsubsection">
2530 <a name="i_shufflevector">'<tt>shufflevector</tt>' Instruction</a>
2533 <div class="doc_text">
2538 <result> = shufflevector <n x <ty>> <v1>, <n x <ty>> <v2>, <n x i32> <mask> <i>; yields <n x <ty>></i>
2544 The '<tt>shufflevector</tt>' instruction constructs a permutation of elements
2545 from two input vectors, returning a vector of the same type.
2551 The first two operands of a '<tt>shufflevector</tt>' instruction are vectors
2552 with types that match each other and types that match the result of the
2553 instruction. The third argument is a shuffle mask, which has the same number
2554 of elements as the other vector type, but whose element type is always 'i32'.
2558 The shuffle mask operand is required to be a constant vector with either
2559 constant integer or undef values.
2565 The elements of the two input vectors are numbered from left to right across
2566 both of the vectors. The shuffle mask operand specifies, for each element of
2567 the result vector, which element of the two input registers the result element
2568 gets. The element selector may be undef (meaning "don't care") and the second
2569 operand may be undef if performing a shuffle from only one vector.
2575 %result = shufflevector <4 x i32> %v1, <4 x i32> %v2,
2576 <4 x i32> <i32 0, i32 4, i32 1, i32 5> <i>; yields <4 x i32></i>
2577 %result = shufflevector <4 x i32> %v1, <4 x i32> undef,
2578 <4 x i32> <i32 0, i32 1, i32 2, i32 3> <i>; yields <4 x i32></i> - Identity shuffle.
2583 <!-- ======================================================================= -->
2584 <div class="doc_subsection">
2585 <a name="memoryops">Memory Access and Addressing Operations</a>
2588 <div class="doc_text">
2590 <p>A key design point of an SSA-based representation is how it
2591 represents memory. In LLVM, no memory locations are in SSA form, which
2592 makes things very simple. This section describes how to read, write,
2593 allocate, and free memory in LLVM.</p>
2597 <!-- _______________________________________________________________________ -->
2598 <div class="doc_subsubsection">
2599 <a name="i_malloc">'<tt>malloc</tt>' Instruction</a>
2602 <div class="doc_text">
2607 <result> = malloc <type>[, i32 <NumElements>][, align <alignment>] <i>; yields {type*}:result</i>
2612 <p>The '<tt>malloc</tt>' instruction allocates memory from the system
2613 heap and returns a pointer to it.</p>
2617 <p>The '<tt>malloc</tt>' instruction allocates
2618 <tt>sizeof(<type>)*NumElements</tt>
2619 bytes of memory from the operating system and returns a pointer of the
2620 appropriate type to the program. If "NumElements" is specified, it is the
2621 number of elements allocated. If an alignment is specified, the value result
2622 of the allocation is guaranteed to be aligned to at least that boundary. If
2623 not specified, or if zero, the target can choose to align the allocation on any
2624 convenient boundary.</p>
2626 <p>'<tt>type</tt>' must be a sized type.</p>
2630 <p>Memory is allocated using the system "<tt>malloc</tt>" function, and
2631 a pointer is returned.</p>
2636 %array = malloc [4 x i8 ] <i>; yields {[%4 x i8]*}:array</i>
2638 %size = <a href="#i_add">add</a> i32 2, 2 <i>; yields {i32}:size = i32 4</i>
2639 %array1 = malloc i8, i32 4 <i>; yields {i8*}:array1</i>
2640 %array2 = malloc [12 x i8], i32 %size <i>; yields {[12 x i8]*}:array2</i>
2641 %array3 = malloc i32, i32 4, align 1024 <i>; yields {i32*}:array3</i>
2642 %array4 = malloc i32, align 1024 <i>; yields {i32*}:array4</i>
2646 <!-- _______________________________________________________________________ -->
2647 <div class="doc_subsubsection">
2648 <a name="i_free">'<tt>free</tt>' Instruction</a>
2651 <div class="doc_text">
2656 free <type> <value> <i>; yields {void}</i>
2661 <p>The '<tt>free</tt>' instruction returns memory back to the unused
2662 memory heap to be reallocated in the future.</p>
2666 <p>'<tt>value</tt>' shall be a pointer value that points to a value
2667 that was allocated with the '<tt><a href="#i_malloc">malloc</a></tt>'
2672 <p>Access to the memory pointed to by the pointer is no longer defined
2673 after this instruction executes.</p>
2678 %array = <a href="#i_malloc">malloc</a> [4 x i8] <i>; yields {[4 x i8]*}:array</i>
2679 free [4 x i8]* %array
2683 <!-- _______________________________________________________________________ -->
2684 <div class="doc_subsubsection">
2685 <a name="i_alloca">'<tt>alloca</tt>' Instruction</a>
2688 <div class="doc_text">
2693 <result> = alloca <type>[, i32 <NumElements>][, align <alignment>] <i>; yields {type*}:result</i>
2698 <p>The '<tt>alloca</tt>' instruction allocates memory on the stack frame of the
2699 currently executing function, to be automatically released when this function
2700 returns to its caller.</p>
2704 <p>The '<tt>alloca</tt>' instruction allocates <tt>sizeof(<type>)*NumElements</tt>
2705 bytes of memory on the runtime stack, returning a pointer of the
2706 appropriate type to the program. If "NumElements" is specified, it is the
2707 number of elements allocated. If an alignment is specified, the value result
2708 of the allocation is guaranteed to be aligned to at least that boundary. If
2709 not specified, or if zero, the target can choose to align the allocation on any
2710 convenient boundary.</p>
2712 <p>'<tt>type</tt>' may be any sized type.</p>
2716 <p>Memory is allocated; a pointer is returned. '<tt>alloca</tt>'d
2717 memory is automatically released when the function returns. The '<tt>alloca</tt>'
2718 instruction is commonly used to represent automatic variables that must
2719 have an address available. When the function returns (either with the <tt><a
2720 href="#i_ret">ret</a></tt> or <tt><a href="#i_unwind">unwind</a></tt>
2721 instructions), the memory is reclaimed.</p>
2726 %ptr = alloca i32 <i>; yields {i32*}:ptr</i>
2727 %ptr = alloca i32, i32 4 <i>; yields {i32*}:ptr</i>
2728 %ptr = alloca i32, i32 4, align 1024 <i>; yields {i32*}:ptr</i>
2729 %ptr = alloca i32, align 1024 <i>; yields {i32*}:ptr</i>
2733 <!-- _______________________________________________________________________ -->
2734 <div class="doc_subsubsection"> <a name="i_load">'<tt>load</tt>'
2735 Instruction</a> </div>
2736 <div class="doc_text">
2738 <pre> <result> = load <ty>* <pointer>[, align <alignment>]<br> <result> = volatile load <ty>* <pointer>[, align <alignment>]<br></pre>
2740 <p>The '<tt>load</tt>' instruction is used to read from memory.</p>
2742 <p>The argument to the '<tt>load</tt>' instruction specifies the memory
2743 address from which to load. The pointer must point to a <a
2744 href="#t_firstclass">first class</a> type. If the <tt>load</tt> is
2745 marked as <tt>volatile</tt>, then the optimizer is not allowed to modify
2746 the number or order of execution of this <tt>load</tt> with other
2747 volatile <tt>load</tt> and <tt><a href="#i_store">store</a></tt>
2750 <p>The location of memory pointed to is loaded.</p>
2752 <pre> %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
2754 href="#i_store">store</a> i32 3, i32* %ptr <i>; yields {void}</i>
2755 %val = load i32* %ptr <i>; yields {i32}:val = i32 3</i>
2758 <!-- _______________________________________________________________________ -->
2759 <div class="doc_subsubsection"> <a name="i_store">'<tt>store</tt>'
2760 Instruction</a> </div>
2761 <div class="doc_text">
2763 <pre> store <ty> <value>, <ty>* <pointer>[, align <alignment>] <i>; yields {void}</i>
2764 volatile store <ty> <value>, <ty>* <pointer>[, align <alignment>] <i>; yields {void}</i>
2767 <p>The '<tt>store</tt>' instruction is used to write to memory.</p>
2769 <p>There are two arguments to the '<tt>store</tt>' instruction: a value
2770 to store and an address at which to store it. The type of the '<tt><pointer></tt>'
2771 operand must be a pointer to the type of the '<tt><value></tt>'
2772 operand. If the <tt>store</tt> is marked as <tt>volatile</tt>, then the
2773 optimizer is not allowed to modify the number or order of execution of
2774 this <tt>store</tt> with other volatile <tt>load</tt> and <tt><a
2775 href="#i_store">store</a></tt> instructions.</p>
2777 <p>The contents of memory are updated to contain '<tt><value></tt>'
2778 at the location specified by the '<tt><pointer></tt>' operand.</p>
2780 <pre> %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
2782 href="#i_store">store</a> i32 3, i32* %ptr <i>; yields {void}</i>
2783 %val = load i32* %ptr <i>; yields {i32}:val = i32 3</i>
2787 <!-- _______________________________________________________________________ -->
2788 <div class="doc_subsubsection">
2789 <a name="i_getelementptr">'<tt>getelementptr</tt>' Instruction</a>
2792 <div class="doc_text">
2795 <result> = getelementptr <ty>* <ptrval>{, <ty> <idx>}*
2801 The '<tt>getelementptr</tt>' instruction is used to get the address of a
2802 subelement of an aggregate data structure.</p>
2806 <p>This instruction takes a list of integer operands that indicate what
2807 elements of the aggregate object to index to. The actual types of the arguments
2808 provided depend on the type of the first pointer argument. The
2809 '<tt>getelementptr</tt>' instruction is used to index down through the type
2810 levels of a structure or to a specific index in an array. When indexing into a
2811 structure, only <tt>i32</tt> integer constants are allowed. When indexing
2812 into an array or pointer, only integers of 32 or 64 bits are allowed, and will
2813 be sign extended to 64-bit values.</p>
2815 <p>For example, let's consider a C code fragment and how it gets
2816 compiled to LLVM:</p>
2818 <div class="doc_code">
2831 int *foo(struct ST *s) {
2832 return &s[1].Z.B[5][13];
2837 <p>The LLVM code generated by the GCC frontend is:</p>
2839 <div class="doc_code">
2841 %RT = type { i8 , [10 x [20 x i32]], i8 }
2842 %ST = type { i32, double, %RT }
2844 define i32* %foo(%ST* %s) {
2846 %reg = getelementptr %ST* %s, i32 1, i32 2, i32 1, i32 5, i32 13
2854 <p>The index types specified for the '<tt>getelementptr</tt>' instruction depend
2855 on the pointer type that is being indexed into. <a href="#t_pointer">Pointer</a>
2856 and <a href="#t_array">array</a> types can use a 32-bit or 64-bit
2857 <a href="#t_integer">integer</a> type but the value will always be sign extended
2858 to 64-bits. <a href="#t_struct">Structure</a> types require <tt>i32</tt>
2859 <b>constants</b>.</p>
2861 <p>In the example above, the first index is indexing into the '<tt>%ST*</tt>'
2862 type, which is a pointer, yielding a '<tt>%ST</tt>' = '<tt>{ i32, double, %RT
2863 }</tt>' type, a structure. The second index indexes into the third element of
2864 the structure, yielding a '<tt>%RT</tt>' = '<tt>{ i8 , [10 x [20 x i32]],
2865 i8 }</tt>' type, another structure. The third index indexes into the second
2866 element of the structure, yielding a '<tt>[10 x [20 x i32]]</tt>' type, an
2867 array. The two dimensions of the array are subscripted into, yielding an
2868 '<tt>i32</tt>' type. The '<tt>getelementptr</tt>' instruction returns a pointer
2869 to this element, thus computing a value of '<tt>i32*</tt>' type.</p>
2871 <p>Note that it is perfectly legal to index partially through a
2872 structure, returning a pointer to an inner element. Because of this,
2873 the LLVM code for the given testcase is equivalent to:</p>
2876 define i32* %foo(%ST* %s) {
2877 %t1 = getelementptr %ST* %s, i32 1 <i>; yields %ST*:%t1</i>
2878 %t2 = getelementptr %ST* %t1, i32 0, i32 2 <i>; yields %RT*:%t2</i>
2879 %t3 = getelementptr %RT* %t2, i32 0, i32 1 <i>; yields [10 x [20 x i32]]*:%t3</i>
2880 %t4 = getelementptr [10 x [20 x i32]]* %t3, i32 0, i32 5 <i>; yields [20 x i32]*:%t4</i>
2881 %t5 = getelementptr [20 x i32]* %t4, i32 0, i32 13 <i>; yields i32*:%t5</i>
2886 <p>Note that it is undefined to access an array out of bounds: array and
2887 pointer indexes must always be within the defined bounds of the array type.
2888 The one exception for this rules is zero length arrays. These arrays are
2889 defined to be accessible as variable length arrays, which requires access
2890 beyond the zero'th element.</p>
2892 <p>The getelementptr instruction is often confusing. For some more insight
2893 into how it works, see <a href="GetElementPtr.html">the getelementptr
2899 <i>; yields [12 x i8]*:aptr</i>
2900 %aptr = getelementptr {i32, [12 x i8]}* %sptr, i64 0, i32 1
2904 <!-- ======================================================================= -->
2905 <div class="doc_subsection"> <a name="convertops">Conversion Operations</a>
2907 <div class="doc_text">
2908 <p>The instructions in this category are the conversion instructions (casting)
2909 which all take a single operand and a type. They perform various bit conversions
2913 <!-- _______________________________________________________________________ -->
2914 <div class="doc_subsubsection">
2915 <a name="i_trunc">'<tt>trunc .. to</tt>' Instruction</a>
2917 <div class="doc_text">
2921 <result> = trunc <ty> <value> to <ty2> <i>; yields ty2</i>
2926 The '<tt>trunc</tt>' instruction truncates its operand to the type <tt>ty2</tt>.
2931 The '<tt>trunc</tt>' instruction takes a <tt>value</tt> to trunc, which must
2932 be an <a href="#t_integer">integer</a> type, and a type that specifies the size
2933 and type of the result, which must be an <a href="#t_integer">integer</a>
2934 type. The bit size of <tt>value</tt> must be larger than the bit size of
2935 <tt>ty2</tt>. Equal sized types are not allowed.</p>
2939 The '<tt>trunc</tt>' instruction truncates the high order bits in <tt>value</tt>
2940 and converts the remaining bits to <tt>ty2</tt>. Since the source size must be
2941 larger than the destination size, <tt>trunc</tt> cannot be a <i>no-op cast</i>.
2942 It will always truncate bits.</p>
2946 %X = trunc i32 257 to i8 <i>; yields i8:1</i>
2947 %Y = trunc i32 123 to i1 <i>; yields i1:true</i>
2948 %Y = trunc i32 122 to i1 <i>; yields i1:false</i>
2952 <!-- _______________________________________________________________________ -->
2953 <div class="doc_subsubsection">
2954 <a name="i_zext">'<tt>zext .. to</tt>' Instruction</a>
2956 <div class="doc_text">
2960 <result> = zext <ty> <value> to <ty2> <i>; yields ty2</i>
2964 <p>The '<tt>zext</tt>' instruction zero extends its operand to type
2969 <p>The '<tt>zext</tt>' instruction takes a value to cast, which must be of
2970 <a href="#t_integer">integer</a> type, and a type to cast it to, which must
2971 also be of <a href="#t_integer">integer</a> type. The bit size of the
2972 <tt>value</tt> must be smaller than the bit size of the destination type,
2976 <p>The <tt>zext</tt> fills the high order bits of the <tt>value</tt> with zero
2977 bits until it reaches the size of the destination type, <tt>ty2</tt>.</p>
2979 <p>When zero extending from i1, the result will always be either 0 or 1.</p>
2983 %X = zext i32 257 to i64 <i>; yields i64:257</i>
2984 %Y = zext i1 true to i32 <i>; yields i32:1</i>
2988 <!-- _______________________________________________________________________ -->
2989 <div class="doc_subsubsection">
2990 <a name="i_sext">'<tt>sext .. to</tt>' Instruction</a>
2992 <div class="doc_text">
2996 <result> = sext <ty> <value> to <ty2> <i>; yields ty2</i>
3000 <p>The '<tt>sext</tt>' sign extends <tt>value</tt> to the type <tt>ty2</tt>.</p>
3004 The '<tt>sext</tt>' instruction takes a value to cast, which must be of
3005 <a href="#t_integer">integer</a> type, and a type to cast it to, which must
3006 also be of <a href="#t_integer">integer</a> type. The bit size of the
3007 <tt>value</tt> must be smaller than the bit size of the destination type,
3012 The '<tt>sext</tt>' instruction performs a sign extension by copying the sign
3013 bit (highest order bit) of the <tt>value</tt> until it reaches the bit size of
3014 the type <tt>ty2</tt>.</p>
3016 <p>When sign extending from i1, the extension always results in -1 or 0.</p>
3020 %X = sext i8 -1 to i16 <i>; yields i16 :65535</i>
3021 %Y = sext i1 true to i32 <i>; yields i32:-1</i>
3025 <!-- _______________________________________________________________________ -->
3026 <div class="doc_subsubsection">
3027 <a name="i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a>
3030 <div class="doc_text">
3035 <result> = fptrunc <ty> <value> to <ty2> <i>; yields ty2</i>
3039 <p>The '<tt>fptrunc</tt>' instruction truncates <tt>value</tt> to type
3044 <p>The '<tt>fptrunc</tt>' instruction takes a <a href="#t_floating">floating
3045 point</a> value to cast and a <a href="#t_floating">floating point</a> type to
3046 cast it to. The size of <tt>value</tt> must be larger than the size of
3047 <tt>ty2</tt>. This implies that <tt>fptrunc</tt> cannot be used to make a
3048 <i>no-op cast</i>.</p>
3051 <p> The '<tt>fptrunc</tt>' instruction truncates a <tt>value</tt> from a larger
3052 <a href="#t_floating">floating point</a> type to a smaller
3053 <a href="#t_floating">floating point</a> type. If the value cannot fit within
3054 the destination type, <tt>ty2</tt>, then the results are undefined.</p>
3058 %X = fptrunc double 123.0 to float <i>; yields float:123.0</i>
3059 %Y = fptrunc double 1.0E+300 to float <i>; yields undefined</i>
3063 <!-- _______________________________________________________________________ -->
3064 <div class="doc_subsubsection">
3065 <a name="i_fpext">'<tt>fpext .. to</tt>' Instruction</a>
3067 <div class="doc_text">
3071 <result> = fpext <ty> <value> to <ty2> <i>; yields ty2</i>
3075 <p>The '<tt>fpext</tt>' extends a floating point <tt>value</tt> to a larger
3076 floating point value.</p>
3079 <p>The '<tt>fpext</tt>' instruction takes a
3080 <a href="#t_floating">floating point</a> <tt>value</tt> to cast,
3081 and a <a href="#t_floating">floating point</a> type to cast it to. The source
3082 type must be smaller than the destination type.</p>
3085 <p>The '<tt>fpext</tt>' instruction extends the <tt>value</tt> from a smaller
3086 <a href="#t_floating">floating point</a> type to a larger
3087 <a href="#t_floating">floating point</a> type. The <tt>fpext</tt> cannot be
3088 used to make a <i>no-op cast</i> because it always changes bits. Use
3089 <tt>bitcast</tt> to make a <i>no-op cast</i> for a floating point cast.</p>
3093 %X = fpext float 3.1415 to double <i>; yields double:3.1415</i>
3094 %Y = fpext float 1.0 to float <i>; yields float:1.0 (no-op)</i>
3098 <!-- _______________________________________________________________________ -->
3099 <div class="doc_subsubsection">
3100 <a name="i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a>
3102 <div class="doc_text">
3106 <result> = fptoui <ty> <value> to <ty2> <i>; yields ty2</i>
3110 <p>The '<tt>fptoui</tt>' converts a floating point <tt>value</tt> to its
3111 unsigned integer equivalent of type <tt>ty2</tt>.
3115 <p>The '<tt>fptoui</tt>' instruction takes a value to cast, which must be a
3116 <a href="#t_floating">floating point</a> value, and a type to cast it to, which
3117 must be an <a href="#t_integer">integer</a> type.</p>
3120 <p> The '<tt>fptoui</tt>' instruction converts its
3121 <a href="#t_floating">floating point</a> operand into the nearest (rounding
3122 towards zero) unsigned integer value. If the value cannot fit in <tt>ty2</tt>,
3123 the results are undefined.</p>
3127 %X = fptoui double 123.0 to i32 <i>; yields i32:123</i>
3128 %Y = fptoui float 1.0E+300 to i1 <i>; yields undefined:1</i>
3129 %X = fptoui float 1.04E+17 to i8 <i>; yields undefined:1</i>
3133 <!-- _______________________________________________________________________ -->
3134 <div class="doc_subsubsection">
3135 <a name="i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a>
3137 <div class="doc_text">
3141 <result> = fptosi <ty> <value> to <ty2> <i>; yields ty2</i>
3145 <p>The '<tt>fptosi</tt>' instruction converts
3146 <a href="#t_floating">floating point</a> <tt>value</tt> to type <tt>ty2</tt>.
3151 <p> The '<tt>fptosi</tt>' instruction takes a value to cast, which must be a
3152 <a href="#t_floating">floating point</a> value, and a type to cast it to, which
3153 must also be an <a href="#t_integer">integer</a> type.</p>
3156 <p>The '<tt>fptosi</tt>' instruction converts its
3157 <a href="#t_floating">floating point</a> operand into the nearest (rounding
3158 towards zero) signed integer value. If the value cannot fit in <tt>ty2</tt>,
3159 the results are undefined.</p>
3163 %X = fptosi double -123.0 to i32 <i>; yields i32:-123</i>
3164 %Y = fptosi float 1.0E-247 to i1 <i>; yields undefined:1</i>
3165 %X = fptosi float 1.04E+17 to i8 <i>; yields undefined:1</i>
3169 <!-- _______________________________________________________________________ -->
3170 <div class="doc_subsubsection">
3171 <a name="i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a>
3173 <div class="doc_text">
3177 <result> = uitofp <ty> <value> to <ty2> <i>; yields ty2</i>
3181 <p>The '<tt>uitofp</tt>' instruction regards <tt>value</tt> as an unsigned
3182 integer and converts that value to the <tt>ty2</tt> type.</p>
3186 <p>The '<tt>uitofp</tt>' instruction takes a value to cast, which must be an
3187 <a href="#t_integer">integer</a> value, and a type to cast it to, which must
3188 be a <a href="#t_floating">floating point</a> type.</p>
3191 <p>The '<tt>uitofp</tt>' instruction interprets its operand as an unsigned
3192 integer quantity and converts it to the corresponding floating point value. If
3193 the value cannot fit in the floating point value, the results are undefined.</p>
3198 %X = uitofp i32 257 to float <i>; yields float:257.0</i>
3199 %Y = uitofp i8 -1 to double <i>; yields double:255.0</i>
3203 <!-- _______________________________________________________________________ -->
3204 <div class="doc_subsubsection">
3205 <a name="i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a>
3207 <div class="doc_text">
3211 <result> = sitofp <ty> <value> to <ty2> <i>; yields ty2</i>
3215 <p>The '<tt>sitofp</tt>' instruction regards <tt>value</tt> as a signed
3216 integer and converts that value to the <tt>ty2</tt> type.</p>
3219 <p>The '<tt>sitofp</tt>' instruction takes a value to cast, which must be an
3220 <a href="#t_integer">integer</a> value, and a type to cast it to, which must be
3221 a <a href="#t_floating">floating point</a> type.</p>
3224 <p>The '<tt>sitofp</tt>' instruction interprets its operand as a signed
3225 integer quantity and converts it to the corresponding floating point value. If
3226 the value cannot fit in the floating point value, the results are undefined.</p>
3230 %X = sitofp i32 257 to float <i>; yields float:257.0</i>
3231 %Y = sitofp i8 -1 to double <i>; yields double:-1.0</i>
3235 <!-- _______________________________________________________________________ -->
3236 <div class="doc_subsubsection">
3237 <a name="i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a>
3239 <div class="doc_text">
3243 <result> = ptrtoint <ty> <value> to <ty2> <i>; yields ty2</i>
3247 <p>The '<tt>ptrtoint</tt>' instruction converts the pointer <tt>value</tt> to
3248 the integer type <tt>ty2</tt>.</p>
3251 <p>The '<tt>ptrtoint</tt>' instruction takes a <tt>value</tt> to cast, which
3252 must be a <a href="#t_pointer">pointer</a> value, and a type to cast it to
3253 <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a> type.
3256 <p>The '<tt>ptrtoint</tt>' instruction converts <tt>value</tt> to integer type
3257 <tt>ty2</tt> by interpreting the pointer value as an integer and either
3258 truncating or zero extending that value to the size of the integer type. If
3259 <tt>value</tt> is smaller than <tt>ty2</tt> then a zero extension is done. If
3260 <tt>value</tt> is larger than <tt>ty2</tt> then a truncation is done. If they
3261 are the same size, then nothing is done (<i>no-op cast</i>) other than a type
3266 %X = ptrtoint i32* %X to i8 <i>; yields truncation on 32-bit architecture</i>
3267 %Y = ptrtoint i32* %x to i64 <i>; yields zero extension on 32-bit architecture</i>
3271 <!-- _______________________________________________________________________ -->
3272 <div class="doc_subsubsection">
3273 <a name="i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a>
3275 <div class="doc_text">
3279 <result> = inttoptr <ty> <value> to <ty2> <i>; yields ty2</i>
3283 <p>The '<tt>inttoptr</tt>' instruction converts an integer <tt>value</tt> to
3284 a pointer type, <tt>ty2</tt>.</p>
3287 <p>The '<tt>inttoptr</tt>' instruction takes an <a href="#t_integer">integer</a>
3288 value to cast, and a type to cast it to, which must be a
3289 <a href="#t_pointer">pointer</a> type.
3292 <p>The '<tt>inttoptr</tt>' instruction converts <tt>value</tt> to type
3293 <tt>ty2</tt> by applying either a zero extension or a truncation depending on
3294 the size of the integer <tt>value</tt>. If <tt>value</tt> is larger than the
3295 size of a pointer then a truncation is done. If <tt>value</tt> is smaller than
3296 the size of a pointer then a zero extension is done. If they are the same size,
3297 nothing is done (<i>no-op cast</i>).</p>
3301 %X = inttoptr i32 255 to i32* <i>; yields zero extension on 64-bit architecture</i>
3302 %X = inttoptr i32 255 to i32* <i>; yields no-op on 32-bit architecture</i>
3303 %Y = inttoptr i64 0 to i32* <i>; yields truncation on 32-bit architecture</i>
3307 <!-- _______________________________________________________________________ -->
3308 <div class="doc_subsubsection">
3309 <a name="i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a>
3311 <div class="doc_text">
3315 <result> = bitcast <ty> <value> to <ty2> <i>; yields ty2</i>
3319 <p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
3320 <tt>ty2</tt> without changing any bits.</p>
3323 <p>The '<tt>bitcast</tt>' instruction takes a value to cast, which must be
3324 a first class value, and a type to cast it to, which must also be a <a
3325 href="#t_firstclass">first class</a> type. The bit sizes of <tt>value</tt>
3326 and the destination type, <tt>ty2</tt>, must be identical. If the source
3327 type is a pointer, the destination type must also be a pointer.</p>
3330 <p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
3331 <tt>ty2</tt>. It is always a <i>no-op cast</i> because no bits change with
3332 this conversion. The conversion is done as if the <tt>value</tt> had been
3333 stored to memory and read back as type <tt>ty2</tt>. Pointer types may only be
3334 converted to other pointer types with this instruction. To convert pointers to
3335 other types, use the <a href="#i_inttoptr">inttoptr</a> or
3336 <a href="#i_ptrtoint">ptrtoint</a> instructions first.</p>
3340 %X = bitcast i8 255 to i8 <i>; yields i8 :-1</i>
3341 %Y = bitcast i32* %x to sint* <i>; yields sint*:%x</i>
3342 %Z = bitcast <2xint> %V to i64; <i>; yields i64: %V</i>
3346 <!-- ======================================================================= -->
3347 <div class="doc_subsection"> <a name="otherops">Other Operations</a> </div>
3348 <div class="doc_text">
3349 <p>The instructions in this category are the "miscellaneous"
3350 instructions, which defy better classification.</p>
3353 <!-- _______________________________________________________________________ -->
3354 <div class="doc_subsubsection"><a name="i_icmp">'<tt>icmp</tt>' Instruction</a>
3356 <div class="doc_text">
3358 <pre> <result> = icmp <cond> <ty> <var1>, <var2> <i>; yields {i1}:result</i>
3361 <p>The '<tt>icmp</tt>' instruction returns a boolean value based on comparison
3362 of its two integer operands.</p>
3364 <p>The '<tt>icmp</tt>' instruction takes three operands. The first operand is
3365 the condition code indicating the kind of comparison to perform. It is not
3366 a value, just a keyword. The possible condition code are:
3368 <li><tt>eq</tt>: equal</li>
3369 <li><tt>ne</tt>: not equal </li>
3370 <li><tt>ugt</tt>: unsigned greater than</li>
3371 <li><tt>uge</tt>: unsigned greater or equal</li>
3372 <li><tt>ult</tt>: unsigned less than</li>
3373 <li><tt>ule</tt>: unsigned less or equal</li>
3374 <li><tt>sgt</tt>: signed greater than</li>
3375 <li><tt>sge</tt>: signed greater or equal</li>
3376 <li><tt>slt</tt>: signed less than</li>
3377 <li><tt>sle</tt>: signed less or equal</li>
3379 <p>The remaining two arguments must be <a href="#t_integer">integer</a> or
3380 <a href="#t_pointer">pointer</a> typed. They must also be identical types.</p>
3382 <p>The '<tt>icmp</tt>' compares <tt>var1</tt> and <tt>var2</tt> according to
3383 the condition code given as <tt>cond</tt>. The comparison performed always
3384 yields a <a href="#t_primitive">i1</a> result, as follows:
3386 <li><tt>eq</tt>: yields <tt>true</tt> if the operands are equal,
3387 <tt>false</tt> otherwise. No sign interpretation is necessary or performed.
3389 <li><tt>ne</tt>: yields <tt>true</tt> if the operands are unequal,
3390 <tt>false</tt> otherwise. No sign interpretation is necessary or performed.
3391 <li><tt>ugt</tt>: interprets the operands as unsigned values and yields
3392 <tt>true</tt> if <tt>var1</tt> is greater than <tt>var2</tt>.</li>
3393 <li><tt>uge</tt>: interprets the operands as unsigned values and yields
3394 <tt>true</tt> if <tt>var1</tt> is greater than or equal to <tt>var2</tt>.</li>
3395 <li><tt>ult</tt>: interprets the operands as unsigned values and yields
3396 <tt>true</tt> if <tt>var1</tt> is less than <tt>var2</tt>.</li>
3397 <li><tt>ule</tt>: interprets the operands as unsigned values and yields
3398 <tt>true</tt> if <tt>var1</tt> is less than or equal to <tt>var2</tt>.</li>
3399 <li><tt>sgt</tt>: interprets the operands as signed values and yields
3400 <tt>true</tt> if <tt>var1</tt> is greater than <tt>var2</tt>.</li>
3401 <li><tt>sge</tt>: interprets the operands as signed values and yields
3402 <tt>true</tt> if <tt>var1</tt> is greater than or equal to <tt>var2</tt>.</li>
3403 <li><tt>slt</tt>: interprets the operands as signed values and yields
3404 <tt>true</tt> if <tt>var1</tt> is less than <tt>var2</tt>.</li>
3405 <li><tt>sle</tt>: interprets the operands as signed values and yields
3406 <tt>true</tt> if <tt>var1</tt> is less than or equal to <tt>var2</tt>.</li>
3408 <p>If the operands are <a href="#t_pointer">pointer</a> typed, the pointer
3409 values are compared as if they were integers.</p>
3412 <pre> <result> = icmp eq i32 4, 5 <i>; yields: result=false</i>
3413 <result> = icmp ne float* %X, %X <i>; yields: result=false</i>
3414 <result> = icmp ult i16 4, 5 <i>; yields: result=true</i>
3415 <result> = icmp sgt i16 4, 5 <i>; yields: result=false</i>
3416 <result> = icmp ule i16 -4, 5 <i>; yields: result=false</i>
3417 <result> = icmp sge i16 4, 5 <i>; yields: result=false</i>
3421 <!-- _______________________________________________________________________ -->
3422 <div class="doc_subsubsection"><a name="i_fcmp">'<tt>fcmp</tt>' Instruction</a>
3424 <div class="doc_text">
3426 <pre> <result> = fcmp <cond> <ty> <var1>, <var2> <i>; yields {i1}:result</i>
3429 <p>The '<tt>fcmp</tt>' instruction returns a boolean value based on comparison
3430 of its floating point operands.</p>
3432 <p>The '<tt>fcmp</tt>' instruction takes three operands. The first operand is
3433 the condition code indicating the kind of comparison to perform. It is not
3434 a value, just a keyword. The possible condition code are:
3436 <li><tt>false</tt>: no comparison, always returns false</li>
3437 <li><tt>oeq</tt>: ordered and equal</li>
3438 <li><tt>ogt</tt>: ordered and greater than </li>
3439 <li><tt>oge</tt>: ordered and greater than or equal</li>
3440 <li><tt>olt</tt>: ordered and less than </li>
3441 <li><tt>ole</tt>: ordered and less than or equal</li>
3442 <li><tt>one</tt>: ordered and not equal</li>
3443 <li><tt>ord</tt>: ordered (no nans)</li>
3444 <li><tt>ueq</tt>: unordered or equal</li>
3445 <li><tt>ugt</tt>: unordered or greater than </li>
3446 <li><tt>uge</tt>: unordered or greater than or equal</li>
3447 <li><tt>ult</tt>: unordered or less than </li>
3448 <li><tt>ule</tt>: unordered or less than or equal</li>
3449 <li><tt>une</tt>: unordered or not equal</li>
3450 <li><tt>uno</tt>: unordered (either nans)</li>
3451 <li><tt>true</tt>: no comparison, always returns true</li>
3453 <p><i>Ordered</i> means that neither operand is a QNAN while
3454 <i>unordered</i> means that either operand may be a QNAN.</p>
3455 <p>The <tt>val1</tt> and <tt>val2</tt> arguments must be
3456 <a href="#t_floating">floating point</a> typed. They must have identical
3459 <p>The '<tt>fcmp</tt>' compares <tt>var1</tt> and <tt>var2</tt> according to
3460 the condition code given as <tt>cond</tt>. The comparison performed always
3461 yields a <a href="#t_primitive">i1</a> result, as follows:
3463 <li><tt>false</tt>: always yields <tt>false</tt>, regardless of operands.</li>
3464 <li><tt>oeq</tt>: yields <tt>true</tt> if both operands are not a QNAN and
3465 <tt>var1</tt> is equal to <tt>var2</tt>.</li>
3466 <li><tt>ogt</tt>: yields <tt>true</tt> if both operands are not a QNAN and
3467 <tt>var1</tt> is greather than <tt>var2</tt>.</li>
3468 <li><tt>oge</tt>: yields <tt>true</tt> if both operands are not a QNAN and
3469 <tt>var1</tt> is greater than or equal to <tt>var2</tt>.</li>
3470 <li><tt>olt</tt>: yields <tt>true</tt> if both operands are not a QNAN and
3471 <tt>var1</tt> is less than <tt>var2</tt>.</li>
3472 <li><tt>ole</tt>: yields <tt>true</tt> if both operands are not a QNAN and
3473 <tt>var1</tt> is less than or equal to <tt>var2</tt>.</li>
3474 <li><tt>one</tt>: yields <tt>true</tt> if both operands are not a QNAN and
3475 <tt>var1</tt> is not equal to <tt>var2</tt>.</li>
3476 <li><tt>ord</tt>: yields <tt>true</tt> if both operands are not a QNAN.</li>
3477 <li><tt>ueq</tt>: yields <tt>true</tt> if either operand is a QNAN or
3478 <tt>var1</tt> is equal to <tt>var2</tt>.</li>
3479 <li><tt>ugt</tt>: yields <tt>true</tt> if either operand is a QNAN or
3480 <tt>var1</tt> is greater than <tt>var2</tt>.</li>
3481 <li><tt>uge</tt>: yields <tt>true</tt> if either operand is a QNAN or
3482 <tt>var1</tt> is greater than or equal to <tt>var2</tt>.</li>
3483 <li><tt>ult</tt>: yields <tt>true</tt> if either operand is a QNAN or
3484 <tt>var1</tt> is less than <tt>var2</tt>.</li>
3485 <li><tt>ule</tt>: yields <tt>true</tt> if either operand is a QNAN or
3486 <tt>var1</tt> is less than or equal to <tt>var2</tt>.</li>
3487 <li><tt>une</tt>: yields <tt>true</tt> if either operand is a QNAN or
3488 <tt>var1</tt> is not equal to <tt>var2</tt>.</li>
3489 <li><tt>uno</tt>: yields <tt>true</tt> if either operand is a QNAN.</li>
3490 <li><tt>true</tt>: always yields <tt>true</tt>, regardless of operands.</li>
3494 <pre> <result> = fcmp oeq float 4.0, 5.0 <i>; yields: result=false</i>
3495 <result> = icmp one float 4.0, 5.0 <i>; yields: result=true</i>
3496 <result> = icmp olt float 4.0, 5.0 <i>; yields: result=true</i>
3497 <result> = icmp ueq double 1.0, 2.0 <i>; yields: result=false</i>
3501 <!-- _______________________________________________________________________ -->
3502 <div class="doc_subsubsection"> <a name="i_phi">'<tt>phi</tt>'
3503 Instruction</a> </div>
3504 <div class="doc_text">
3506 <pre> <result> = phi <ty> [ <val0>, <label0>], ...<br></pre>
3508 <p>The '<tt>phi</tt>' instruction is used to implement the φ node in
3509 the SSA graph representing the function.</p>
3511 <p>The type of the incoming values is specified with the first type
3512 field. After this, the '<tt>phi</tt>' instruction takes a list of pairs
3513 as arguments, with one pair for each predecessor basic block of the
3514 current block. Only values of <a href="#t_firstclass">first class</a>
3515 type may be used as the value arguments to the PHI node. Only labels
3516 may be used as the label arguments.</p>
3517 <p>There must be no non-phi instructions between the start of a basic
3518 block and the PHI instructions: i.e. PHI instructions must be first in
3521 <p>At runtime, the '<tt>phi</tt>' instruction logically takes on the value
3522 specified by the pair corresponding to the predecessor basic block that executed
3523 just prior to the current block.</p>
3525 <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>
3528 <!-- _______________________________________________________________________ -->
3529 <div class="doc_subsubsection">
3530 <a name="i_select">'<tt>select</tt>' Instruction</a>
3533 <div class="doc_text">
3538 <result> = select i1 <cond>, <ty> <val1>, <ty> <val2> <i>; yields ty</i>
3544 The '<tt>select</tt>' instruction is used to choose one value based on a
3545 condition, without branching.
3552 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.
3558 If the boolean condition evaluates to true, the instruction returns the first
3559 value argument; otherwise, it returns the second value argument.
3565 %X = select i1 true, i8 17, i8 42 <i>; yields i8:17</i>
3570 <!-- _______________________________________________________________________ -->
3571 <div class="doc_subsubsection">
3572 <a name="i_call">'<tt>call</tt>' Instruction</a>
3575 <div class="doc_text">
3579 <result> = [tail] call [<a href="#callingconv">cconv</a>] <ty> [<fnty>*] <fnptrval>(<param list>)
3584 <p>The '<tt>call</tt>' instruction represents a simple function call.</p>
3588 <p>This instruction requires several arguments:</p>
3592 <p>The optional "tail" marker indicates whether the callee function accesses
3593 any allocas or varargs in the caller. If the "tail" marker is present, the
3594 function call is eligible for tail call optimization. Note that calls may
3595 be marked "tail" even if they do not occur before a <a
3596 href="#i_ret"><tt>ret</tt></a> instruction.
3599 <p>The optional "cconv" marker indicates which <a href="#callingconv">calling
3600 convention</a> the call should use. If none is specified, the call defaults
3601 to using C calling conventions.
3604 <p>'<tt>ty</tt>': the type of the call instruction itself which is also
3605 the type of the return value. Functions that return no value are marked
3606 <tt><a href="#t_void">void</a></tt>.</p>
3609 <p>'<tt>fnty</tt>': shall be the signature of the pointer to function
3610 value being invoked. The argument types must match the types implied by
3611 this signature. This type can be omitted if the function is not varargs
3612 and if the function type does not return a pointer to a function.</p>
3615 <p>'<tt>fnptrval</tt>': An LLVM value containing a pointer to a function to
3616 be invoked. In most cases, this is a direct function invocation, but
3617 indirect <tt>call</tt>s are just as possible, calling an arbitrary pointer
3618 to function value.</p>
3621 <p>'<tt>function args</tt>': argument list whose types match the
3622 function signature argument types. All arguments must be of
3623 <a href="#t_firstclass">first class</a> type. If the function signature
3624 indicates the function accepts a variable number of arguments, the extra
3625 arguments can be specified.</p>
3631 <p>The '<tt>call</tt>' instruction is used to cause control flow to
3632 transfer to a specified function, with its incoming arguments bound to
3633 the specified values. Upon a '<tt><a href="#i_ret">ret</a></tt>'
3634 instruction in the called function, control flow continues with the
3635 instruction after the function call, and the return value of the
3636 function is bound to the result argument. This is a simpler case of
3637 the <a href="#i_invoke">invoke</a> instruction.</p>
3642 %retval = call i32 @test(i32 %argc)
3643 call i32 (i8 *, ...)* @printf(i8 * %msg, i32 12, i8 42);
3644 %X = tail call i32 @foo()
3645 %Y = tail call <a href="#callingconv">fastcc</a> i32 @foo()
3646 %Z = call void %foo(i8 97 signext)
3651 <!-- _______________________________________________________________________ -->
3652 <div class="doc_subsubsection">
3653 <a name="i_va_arg">'<tt>va_arg</tt>' Instruction</a>
3656 <div class="doc_text">
3661 <resultval> = va_arg <va_list*> <arglist>, <argty>
3666 <p>The '<tt>va_arg</tt>' instruction is used to access arguments passed through
3667 the "variable argument" area of a function call. It is used to implement the
3668 <tt>va_arg</tt> macro in C.</p>
3672 <p>This instruction takes a <tt>va_list*</tt> value and the type of
3673 the argument. It returns a value of the specified argument type and
3674 increments the <tt>va_list</tt> to point to the next argument. The
3675 actual type of <tt>va_list</tt> is target specific.</p>
3679 <p>The '<tt>va_arg</tt>' instruction loads an argument of the specified
3680 type from the specified <tt>va_list</tt> and causes the
3681 <tt>va_list</tt> to point to the next argument. For more information,
3682 see the variable argument handling <a href="#int_varargs">Intrinsic
3685 <p>It is legal for this instruction to be called in a function which does not
3686 take a variable number of arguments, for example, the <tt>vfprintf</tt>
3689 <p><tt>va_arg</tt> is an LLVM instruction instead of an <a
3690 href="#intrinsics">intrinsic function</a> because it takes a type as an
3695 <p>See the <a href="#int_varargs">variable argument processing</a> section.</p>
3699 <!-- *********************************************************************** -->
3700 <div class="doc_section"> <a name="intrinsics">Intrinsic Functions</a> </div>
3701 <!-- *********************************************************************** -->
3703 <div class="doc_text">
3705 <p>LLVM supports the notion of an "intrinsic function". These functions have
3706 well known names and semantics and are required to follow certain restrictions.
3707 Overall, these intrinsics represent an extension mechanism for the LLVM
3708 language that does not require changing all of the transformations in LLVM when
3709 adding to the language (or the bitcode reader/writer, the parser, etc...).</p>
3711 <p>Intrinsic function names must all start with an "<tt>llvm.</tt>" prefix. This
3712 prefix is reserved in LLVM for intrinsic names; thus, function names may not
3713 begin with this prefix. Intrinsic functions must always be external functions:
3714 you cannot define the body of intrinsic functions. Intrinsic functions may
3715 only be used in call or invoke instructions: it is illegal to take the address
3716 of an intrinsic function. Additionally, because intrinsic functions are part
3717 of the LLVM language, it is required if any are added that they be documented
3720 <p>Some intrinsic functions can be overloaded, i.e., the intrinsic represents
3721 a family of functions that perform the same operation but on different data
3722 types. Because LLVM can represent over 8 million different integer types,
3723 overloading is used commonly to allow an intrinsic function to operate on any
3724 integer type. One or more of the argument types or the result type can be
3725 overloaded to accept any integer type. Argument types may also be defined as
3726 exactly matching a previous argument's type or the result type. This allows an
3727 intrinsic function which accepts multiple arguments, but needs all of them to
3728 be of the same type, to only be overloaded with respect to a single argument or
3731 <p>Overloaded intrinsics will have the names of its overloaded argument types
3732 encoded into its function name, each preceded by a period. Only those types
3733 which are overloaded result in a name suffix. Arguments whose type is matched
3734 against another type do not. For example, the <tt>llvm.ctpop</tt> function can
3735 take an integer of any width and returns an integer of exactly the same integer
3736 width. This leads to a family of functions such as
3737 <tt>i8 @llvm.ctpop.i8(i8 %val)</tt> and <tt>i29 @llvm.ctpop.i29(i29 %val)</tt>.
3738 Only one type, the return type, is overloaded, and only one type suffix is
3739 required. Because the argument's type is matched against the return type, it
3740 does not require its own name suffix.</p>
3742 <p>To learn how to add an intrinsic function, please see the
3743 <a href="ExtendingLLVM.html">Extending LLVM Guide</a>.
3748 <!-- ======================================================================= -->
3749 <div class="doc_subsection">
3750 <a name="int_varargs">Variable Argument Handling Intrinsics</a>
3753 <div class="doc_text">
3755 <p>Variable argument support is defined in LLVM with the <a
3756 href="#i_va_arg"><tt>va_arg</tt></a> instruction and these three
3757 intrinsic functions. These functions are related to the similarly
3758 named macros defined in the <tt><stdarg.h></tt> header file.</p>
3760 <p>All of these functions operate on arguments that use a
3761 target-specific value type "<tt>va_list</tt>". The LLVM assembly
3762 language reference manual does not define what this type is, so all
3763 transformations should be prepared to handle these functions regardless of
3766 <p>This example shows how the <a href="#i_va_arg"><tt>va_arg</tt></a>
3767 instruction and the variable argument handling intrinsic functions are
3770 <div class="doc_code">
3772 define i32 @test(i32 %X, ...) {
3773 ; Initialize variable argument processing
3775 %ap2 = bitcast i8** %ap to i8*
3776 call void @llvm.va_start(i8* %ap2)
3778 ; Read a single integer argument
3779 %tmp = va_arg i8** %ap, i32
3781 ; Demonstrate usage of llvm.va_copy and llvm.va_end
3783 %aq2 = bitcast i8** %aq to i8*
3784 call void @llvm.va_copy(i8* %aq2, i8* %ap2)
3785 call void @llvm.va_end(i8* %aq2)
3787 ; Stop processing of arguments.
3788 call void @llvm.va_end(i8* %ap2)
3792 declare void @llvm.va_start(i8*)
3793 declare void @llvm.va_copy(i8*, i8*)
3794 declare void @llvm.va_end(i8*)
3800 <!-- _______________________________________________________________________ -->
3801 <div class="doc_subsubsection">
3802 <a name="int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a>
3806 <div class="doc_text">
3808 <pre> declare void %llvm.va_start(i8* <arglist>)<br></pre>
3810 <P>The '<tt>llvm.va_start</tt>' intrinsic initializes
3811 <tt>*<arglist></tt> for subsequent use by <tt><a
3812 href="#i_va_arg">va_arg</a></tt>.</p>
3816 <P>The argument is a pointer to a <tt>va_list</tt> element to initialize.</p>
3820 <P>The '<tt>llvm.va_start</tt>' intrinsic works just like the <tt>va_start</tt>
3821 macro available in C. In a target-dependent way, it initializes the
3822 <tt>va_list</tt> element to which the argument points, so that the next call to
3823 <tt>va_arg</tt> will produce the first variable argument passed to the function.
3824 Unlike the C <tt>va_start</tt> macro, this intrinsic does not need to know the
3825 last argument of the function as the compiler can figure that out.</p>
3829 <!-- _______________________________________________________________________ -->
3830 <div class="doc_subsubsection">
3831 <a name="int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a>
3834 <div class="doc_text">
3836 <pre> declare void @llvm.va_end(i8* <arglist>)<br></pre>
3839 <p>The '<tt>llvm.va_end</tt>' intrinsic destroys <tt>*<arglist></tt>,
3840 which has been initialized previously with <tt><a href="#int_va_start">llvm.va_start</a></tt>
3841 or <tt><a href="#i_va_copy">llvm.va_copy</a></tt>.</p>
3845 <p>The argument is a pointer to a <tt>va_list</tt> to destroy.</p>
3849 <p>The '<tt>llvm.va_end</tt>' intrinsic works just like the <tt>va_end</tt>
3850 macro available in C. In a target-dependent way, it destroys the
3851 <tt>va_list</tt> element to which the argument points. Calls to <a
3852 href="#int_va_start"><tt>llvm.va_start</tt></a> and <a href="#int_va_copy">
3853 <tt>llvm.va_copy</tt></a> must be matched exactly with calls to
3854 <tt>llvm.va_end</tt>.</p>
3858 <!-- _______________________________________________________________________ -->
3859 <div class="doc_subsubsection">
3860 <a name="int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a>
3863 <div class="doc_text">
3868 declare void @llvm.va_copy(i8* <destarglist>, i8* <srcarglist>)
3873 <p>The '<tt>llvm.va_copy</tt>' intrinsic copies the current argument position
3874 from the source argument list to the destination argument list.</p>
3878 <p>The first argument is a pointer to a <tt>va_list</tt> element to initialize.
3879 The second argument is a pointer to a <tt>va_list</tt> element to copy from.</p>
3884 <p>The '<tt>llvm.va_copy</tt>' intrinsic works just like the <tt>va_copy</tt>
3885 macro available in C. In a target-dependent way, it copies the source
3886 <tt>va_list</tt> element into the destination <tt>va_list</tt> element. This
3887 intrinsic is necessary because the <tt><a href="#int_va_start">
3888 llvm.va_start</a></tt> intrinsic may be arbitrarily complex and require, for
3889 example, memory allocation.</p>
3893 <!-- ======================================================================= -->
3894 <div class="doc_subsection">
3895 <a name="int_gc">Accurate Garbage Collection Intrinsics</a>
3898 <div class="doc_text">
3901 LLVM support for <a href="GarbageCollection.html">Accurate Garbage
3902 Collection</a> requires the implementation and generation of these intrinsics.
3903 These intrinsics allow identification of <a href="#int_gcroot">GC roots on the
3904 stack</a>, as well as garbage collector implementations that require <a
3905 href="#int_gcread">read</a> and <a href="#int_gcwrite">write</a> barriers.
3906 Front-ends for type-safe garbage collected languages should generate these
3907 intrinsics to make use of the LLVM garbage collectors. For more details, see <a
3908 href="GarbageCollection.html">Accurate Garbage Collection with LLVM</a>.
3912 <!-- _______________________________________________________________________ -->
3913 <div class="doc_subsubsection">
3914 <a name="int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a>
3917 <div class="doc_text">
3922 declare void @llvm.gcroot(i8** %ptrloc, i8* %metadata)
3927 <p>The '<tt>llvm.gcroot</tt>' intrinsic declares the existence of a GC root to
3928 the code generator, and allows some metadata to be associated with it.</p>
3932 <p>The first argument specifies the address of a stack object that contains the
3933 root pointer. The second pointer (which must be either a constant or a global
3934 value address) contains the meta-data to be associated with the root.</p>
3938 <p>At runtime, a call to this intrinsics stores a null pointer into the "ptrloc"
3939 location. At compile-time, the code generator generates information to allow
3940 the runtime to find the pointer at GC safe points.
3946 <!-- _______________________________________________________________________ -->
3947 <div class="doc_subsubsection">
3948 <a name="int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a>
3951 <div class="doc_text">
3956 declare i8* @llvm.gcread(i8* %ObjPtr, i8** %Ptr)
3961 <p>The '<tt>llvm.gcread</tt>' intrinsic identifies reads of references from heap
3962 locations, allowing garbage collector implementations that require read
3967 <p>The second argument is the address to read from, which should be an address
3968 allocated from the garbage collector. The first object is a pointer to the
3969 start of the referenced object, if needed by the language runtime (otherwise
3974 <p>The '<tt>llvm.gcread</tt>' intrinsic has the same semantics as a load
3975 instruction, but may be replaced with substantially more complex code by the
3976 garbage collector runtime, as needed.</p>
3981 <!-- _______________________________________________________________________ -->
3982 <div class="doc_subsubsection">
3983 <a name="int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a>
3986 <div class="doc_text">
3991 declare void @llvm.gcwrite(i8* %P1, i8* %Obj, i8** %P2)
3996 <p>The '<tt>llvm.gcwrite</tt>' intrinsic identifies writes of references to heap
3997 locations, allowing garbage collector implementations that require write
3998 barriers (such as generational or reference counting collectors).</p>
4002 <p>The first argument is the reference to store, the second is the start of the
4003 object to store it to, and the third is the address of the field of Obj to
4004 store to. If the runtime does not require a pointer to the object, Obj may be
4009 <p>The '<tt>llvm.gcwrite</tt>' intrinsic has the same semantics as a store
4010 instruction, but may be replaced with substantially more complex code by the
4011 garbage collector runtime, as needed.</p>
4017 <!-- ======================================================================= -->
4018 <div class="doc_subsection">
4019 <a name="int_codegen">Code Generator Intrinsics</a>
4022 <div class="doc_text">
4024 These intrinsics are provided by LLVM to expose special features that may only
4025 be implemented with code generator support.
4030 <!-- _______________________________________________________________________ -->
4031 <div class="doc_subsubsection">
4032 <a name="int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a>
4035 <div class="doc_text">
4039 declare i8 *@llvm.returnaddress(i32 <level>)
4045 The '<tt>llvm.returnaddress</tt>' intrinsic attempts to compute a
4046 target-specific value indicating the return address of the current function
4047 or one of its callers.
4053 The argument to this intrinsic indicates which function to return the address
4054 for. Zero indicates the calling function, one indicates its caller, etc. The
4055 argument is <b>required</b> to be a constant integer value.
4061 The '<tt>llvm.returnaddress</tt>' intrinsic either returns a pointer indicating
4062 the return address of the specified call frame, or zero if it cannot be
4063 identified. The value returned by this intrinsic is likely to be incorrect or 0
4064 for arguments other than zero, so it should only be used for debugging purposes.
4068 Note that calling this intrinsic does not prevent function inlining or other
4069 aggressive transformations, so the value returned may not be that of the obvious
4070 source-language caller.
4075 <!-- _______________________________________________________________________ -->
4076 <div class="doc_subsubsection">
4077 <a name="int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a>
4080 <div class="doc_text">
4084 declare i8 *@llvm.frameaddress(i32 <level>)
4090 The '<tt>llvm.frameaddress</tt>' intrinsic attempts to return the
4091 target-specific frame pointer value for the specified stack frame.
4097 The argument to this intrinsic indicates which function to return the frame
4098 pointer for. Zero indicates the calling function, one indicates its caller,
4099 etc. The argument is <b>required</b> to be a constant integer value.
4105 The '<tt>llvm.frameaddress</tt>' intrinsic either returns a pointer indicating
4106 the frame address of the specified call frame, or zero if it cannot be
4107 identified. The value returned by this intrinsic is likely to be incorrect or 0
4108 for arguments other than zero, so it should only be used for debugging purposes.
4112 Note that calling this intrinsic does not prevent function inlining or other
4113 aggressive transformations, so the value returned may not be that of the obvious
4114 source-language caller.
4118 <!-- _______________________________________________________________________ -->
4119 <div class="doc_subsubsection">
4120 <a name="int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a>
4123 <div class="doc_text">
4127 declare i8 *@llvm.stacksave()
4133 The '<tt>llvm.stacksave</tt>' intrinsic is used to remember the current state of
4134 the function stack, for use with <a href="#int_stackrestore">
4135 <tt>llvm.stackrestore</tt></a>. This is useful for implementing language
4136 features like scoped automatic variable sized arrays in C99.
4142 This intrinsic returns a opaque pointer value that can be passed to <a
4143 href="#int_stackrestore"><tt>llvm.stackrestore</tt></a>. When an
4144 <tt>llvm.stackrestore</tt> intrinsic is executed with a value saved from
4145 <tt>llvm.stacksave</tt>, it effectively restores the state of the stack to the
4146 state it was in when the <tt>llvm.stacksave</tt> intrinsic executed. In
4147 practice, this pops any <a href="#i_alloca">alloca</a> blocks from the stack
4148 that were allocated after the <tt>llvm.stacksave</tt> was executed.
4153 <!-- _______________________________________________________________________ -->
4154 <div class="doc_subsubsection">
4155 <a name="int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a>
4158 <div class="doc_text">
4162 declare void @llvm.stackrestore(i8 * %ptr)
4168 The '<tt>llvm.stackrestore</tt>' intrinsic is used to restore the state of
4169 the function stack to the state it was in when the corresponding <a
4170 href="#int_stacksave"><tt>llvm.stacksave</tt></a> intrinsic executed. This is
4171 useful for implementing language features like scoped automatic variable sized
4178 See the description for <a href="#int_stacksave"><tt>llvm.stacksave</tt></a>.
4184 <!-- _______________________________________________________________________ -->
4185 <div class="doc_subsubsection">
4186 <a name="int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a>
4189 <div class="doc_text">
4193 declare void @llvm.prefetch(i8* <address>, i32 <rw>, i32 <locality>)
4200 The '<tt>llvm.prefetch</tt>' intrinsic is a hint to the code generator to insert
4201 a prefetch instruction if supported; otherwise, it is a noop. Prefetches have
4203 effect on the behavior of the program but can change its performance
4210 <tt>address</tt> is the address to be prefetched, <tt>rw</tt> is the specifier
4211 determining if the fetch should be for a read (0) or write (1), and
4212 <tt>locality</tt> is a temporal locality specifier ranging from (0) - no
4213 locality, to (3) - extremely local keep in cache. The <tt>rw</tt> and
4214 <tt>locality</tt> arguments must be constant integers.
4220 This intrinsic does not modify the behavior of the program. In particular,
4221 prefetches cannot trap and do not produce a value. On targets that support this
4222 intrinsic, the prefetch can provide hints to the processor cache for better
4228 <!-- _______________________________________________________________________ -->
4229 <div class="doc_subsubsection">
4230 <a name="int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a>
4233 <div class="doc_text">
4237 declare void @llvm.pcmarker(i32 <id>)
4244 The '<tt>llvm.pcmarker</tt>' intrinsic is a method to export a Program Counter
4246 code to simulators and other tools. The method is target specific, but it is
4247 expected that the marker will use exported symbols to transmit the PC of the marker.
4248 The marker makes no guarantees that it will remain with any specific instruction
4249 after optimizations. It is possible that the presence of a marker will inhibit
4250 optimizations. The intended use is to be inserted after optimizations to allow
4251 correlations of simulation runs.
4257 <tt>id</tt> is a numerical id identifying the marker.
4263 This intrinsic does not modify the behavior of the program. Backends that do not
4264 support this intrinisic may ignore it.
4269 <!-- _______________________________________________________________________ -->
4270 <div class="doc_subsubsection">
4271 <a name="int_readcyclecounter">'<tt>llvm.readcyclecounter</tt>' Intrinsic</a>
4274 <div class="doc_text">
4278 declare i64 @llvm.readcyclecounter( )
4285 The '<tt>llvm.readcyclecounter</tt>' intrinsic provides access to the cycle
4286 counter register (or similar low latency, high accuracy clocks) on those targets
4287 that support it. On X86, it should map to RDTSC. On Alpha, it should map to RPCC.
4288 As the backing counters overflow quickly (on the order of 9 seconds on alpha), this
4289 should only be used for small timings.
4295 When directly supported, reading the cycle counter should not modify any memory.
4296 Implementations are allowed to either return a application specific value or a
4297 system wide value. On backends without support, this is lowered to a constant 0.
4302 <!-- ======================================================================= -->
4303 <div class="doc_subsection">
4304 <a name="int_libc">Standard C Library Intrinsics</a>
4307 <div class="doc_text">
4309 LLVM provides intrinsics for a few important standard C library functions.
4310 These intrinsics allow source-language front-ends to pass information about the
4311 alignment of the pointer arguments to the code generator, providing opportunity
4312 for more efficient code generation.
4317 <!-- _______________________________________________________________________ -->
4318 <div class="doc_subsubsection">
4319 <a name="int_memcpy">'<tt>llvm.memcpy</tt>' Intrinsic</a>
4322 <div class="doc_text">
4326 declare void @llvm.memcpy.i32(i8 * <dest>, i8 * <src>,
4327 i32 <len>, i32 <align>)
4328 declare void @llvm.memcpy.i64(i8 * <dest>, i8 * <src>,
4329 i64 <len>, i32 <align>)
4335 The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the source
4336 location to the destination location.
4340 Note that, unlike the standard libc function, the <tt>llvm.memcpy.*</tt>
4341 intrinsics do not return a value, and takes an extra alignment argument.
4347 The first argument is a pointer to the destination, the second is a pointer to
4348 the source. The third argument is an integer argument
4349 specifying the number of bytes to copy, and the fourth argument is the alignment
4350 of the source and destination locations.
4354 If the call to this intrinisic has an alignment value that is not 0 or 1, then
4355 the caller guarantees that both the source and destination pointers are aligned
4362 The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the source
4363 location to the destination location, which are not allowed to overlap. It
4364 copies "len" bytes of memory over. If the argument is known to be aligned to
4365 some boundary, this can be specified as the fourth argument, otherwise it should
4371 <!-- _______________________________________________________________________ -->
4372 <div class="doc_subsubsection">
4373 <a name="int_memmove">'<tt>llvm.memmove</tt>' Intrinsic</a>
4376 <div class="doc_text">
4380 declare void @llvm.memmove.i32(i8 * <dest>, i8 * <src>,
4381 i32 <len>, i32 <align>)
4382 declare void @llvm.memmove.i64(i8 * <dest>, i8 * <src>,
4383 i64 <len>, i32 <align>)
4389 The '<tt>llvm.memmove.*</tt>' intrinsics move a block of memory from the source
4390 location to the destination location. It is similar to the
4391 '<tt>llvm.memcmp</tt>' intrinsic but allows the two memory locations to overlap.
4395 Note that, unlike the standard libc function, the <tt>llvm.memmove.*</tt>
4396 intrinsics do not return a value, and takes an extra alignment argument.
4402 The first argument is a pointer to the destination, the second is a pointer to
4403 the source. The third argument is an integer argument
4404 specifying the number of bytes to copy, and the fourth argument is the alignment
4405 of the source and destination locations.
4409 If the call to this intrinisic has an alignment value that is not 0 or 1, then
4410 the caller guarantees that the source and destination pointers are aligned to
4417 The '<tt>llvm.memmove.*</tt>' intrinsics copy a block of memory from the source
4418 location to the destination location, which may overlap. It
4419 copies "len" bytes of memory over. If the argument is known to be aligned to
4420 some boundary, this can be specified as the fourth argument, otherwise it should
4426 <!-- _______________________________________________________________________ -->
4427 <div class="doc_subsubsection">
4428 <a name="int_memset">'<tt>llvm.memset.*</tt>' Intrinsics</a>
4431 <div class="doc_text">
4435 declare void @llvm.memset.i32(i8 * <dest>, i8 <val>,
4436 i32 <len>, i32 <align>)
4437 declare void @llvm.memset.i64(i8 * <dest>, i8 <val>,
4438 i64 <len>, i32 <align>)
4444 The '<tt>llvm.memset.*</tt>' intrinsics fill a block of memory with a particular
4449 Note that, unlike the standard libc function, the <tt>llvm.memset</tt> intrinsic
4450 does not return a value, and takes an extra alignment argument.
4456 The first argument is a pointer to the destination to fill, the second is the
4457 byte value to fill it with, the third argument is an integer
4458 argument specifying the number of bytes to fill, and the fourth argument is the
4459 known alignment of destination location.
4463 If the call to this intrinisic has an alignment value that is not 0 or 1, then
4464 the caller guarantees that the destination pointer is aligned to that boundary.
4470 The '<tt>llvm.memset.*</tt>' intrinsics fill "len" bytes of memory starting at
4472 destination location. If the argument is known to be aligned to some boundary,
4473 this can be specified as the fourth argument, otherwise it should be set to 0 or
4479 <!-- _______________________________________________________________________ -->
4480 <div class="doc_subsubsection">
4481 <a name="int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a>
4484 <div class="doc_text">
4488 declare float @llvm.sqrt.f32(float %Val)
4489 declare double @llvm.sqrt.f64(double %Val)
4495 The '<tt>llvm.sqrt</tt>' intrinsics return the sqrt of the specified operand,
4496 returning the same value as the libm '<tt>sqrt</tt>' function would. Unlike
4497 <tt>sqrt</tt> in libm, however, <tt>llvm.sqrt</tt> has undefined behavior for
4498 negative numbers (which allows for better optimization).
4504 The argument and return value are floating point numbers of the same type.
4510 This function returns the sqrt of the specified operand if it is a nonnegative
4511 floating point number.
4515 <!-- _______________________________________________________________________ -->
4516 <div class="doc_subsubsection">
4517 <a name="int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a>
4520 <div class="doc_text">
4524 declare float @llvm.powi.f32(float %Val, i32 %power)
4525 declare double @llvm.powi.f64(double %Val, i32 %power)
4531 The '<tt>llvm.powi.*</tt>' intrinsics return the first operand raised to the
4532 specified (positive or negative) power. The order of evaluation of
4533 multiplications is not defined.
4539 The second argument is an integer power, and the first is a value to raise to
4546 This function returns the first value raised to the second power with an
4547 unspecified sequence of rounding operations.</p>
4551 <!-- ======================================================================= -->
4552 <div class="doc_subsection">
4553 <a name="int_manip">Bit Manipulation Intrinsics</a>
4556 <div class="doc_text">
4558 LLVM provides intrinsics for a few important bit manipulation operations.
4559 These allow efficient code generation for some algorithms.
4564 <!-- _______________________________________________________________________ -->
4565 <div class="doc_subsubsection">
4566 <a name="int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a>
4569 <div class="doc_text">
4572 <p>This is an overloaded intrinsic function. You can use bswap on any integer
4573 type that is an even number of bytes (i.e. BitWidth % 16 == 0).
4575 declare i16 @llvm.bswap.i16(i16 <id>)
4576 declare i32 @llvm.bswap.i32(i32 <id>)
4577 declare i64 @llvm.bswap.i64(i64 <id>)
4583 The '<tt>llvm.bswap</tt>' family of intrinsics is used to byte swap integer
4584 values with an even number of bytes (positive multiple of 16 bits). These are
4585 useful for performing operations on data that is not in the target's native
4592 The <tt>llvm.bswap.i16</tt> intrinsic returns an i16 value that has the high
4593 and low byte of the input i16 swapped. Similarly, the <tt>llvm.bswap.i32</tt>
4594 intrinsic returns an i32 value that has the four bytes of the input i32
4595 swapped, so that if the input bytes are numbered 0, 1, 2, 3 then the returned
4596 i32 will have its bytes in 3, 2, 1, 0 order. The <tt>llvm.bswap.i48</tt>,
4597 <tt>llvm.bswap.i64</tt> and other intrinsics extend this concept to
4598 additional even-byte lengths (6 bytes, 8 bytes and more, respectively).
4603 <!-- _______________________________________________________________________ -->
4604 <div class="doc_subsubsection">
4605 <a name="int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic</a>
4608 <div class="doc_text">
4611 <p>This is an overloaded intrinsic. You can use llvm.ctpop on any integer bit
4612 width. Not all targets support all bit widths however.
4614 declare i8 @llvm.ctpop.i8 (i8 <src>)
4615 declare i16 @llvm.ctpop.i16(i16 <src>)
4616 declare i32 @llvm.ctpop.i32(i32 <src>)
4617 declare i64 @llvm.ctpop.i64(i64 <src>)
4618 declare i256 @llvm.ctpop.i256(i256 <src>)
4624 The '<tt>llvm.ctpop</tt>' family of intrinsics counts the number of bits set in a
4631 The only argument is the value to be counted. The argument may be of any
4632 integer type. The return type must match the argument type.
4638 The '<tt>llvm.ctpop</tt>' intrinsic counts the 1's in a variable.
4642 <!-- _______________________________________________________________________ -->
4643 <div class="doc_subsubsection">
4644 <a name="int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic</a>
4647 <div class="doc_text">
4650 <p>This is an overloaded intrinsic. You can use <tt>llvm.ctlz</tt> on any
4651 integer bit width. Not all targets support all bit widths however.
4653 declare i8 @llvm.ctlz.i8 (i8 <src>)
4654 declare i16 @llvm.ctlz.i16(i16 <src>)
4655 declare i32 @llvm.ctlz.i32(i32 <src>)
4656 declare i64 @llvm.ctlz.i64(i64 <src>)
4657 declare i256 @llvm.ctlz.i256(i256 <src>)
4663 The '<tt>llvm.ctlz</tt>' family of intrinsic functions counts the number of
4664 leading zeros in a variable.
4670 The only argument is the value to be counted. The argument may be of any
4671 integer type. The return type must match the argument type.
4677 The '<tt>llvm.ctlz</tt>' intrinsic counts the leading (most significant) zeros
4678 in a variable. If the src == 0 then the result is the size in bits of the type
4679 of src. For example, <tt>llvm.ctlz(i32 2) = 30</tt>.
4685 <!-- _______________________________________________________________________ -->
4686 <div class="doc_subsubsection">
4687 <a name="int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic</a>
4690 <div class="doc_text">
4693 <p>This is an overloaded intrinsic. You can use <tt>llvm.cttz</tt> on any
4694 integer bit width. Not all targets support all bit widths however.
4696 declare i8 @llvm.cttz.i8 (i8 <src>)
4697 declare i16 @llvm.cttz.i16(i16 <src>)
4698 declare i32 @llvm.cttz.i32(i32 <src>)
4699 declare i64 @llvm.cttz.i64(i64 <src>)
4700 declare i256 @llvm.cttz.i256(i256 <src>)
4706 The '<tt>llvm.cttz</tt>' family of intrinsic functions counts the number of
4713 The only argument is the value to be counted. The argument may be of any
4714 integer type. The return type must match the argument type.
4720 The '<tt>llvm.cttz</tt>' intrinsic counts the trailing (least significant) zeros
4721 in a variable. If the src == 0 then the result is the size in bits of the type
4722 of src. For example, <tt>llvm.cttz(2) = 1</tt>.
4726 <!-- _______________________________________________________________________ -->
4727 <div class="doc_subsubsection">
4728 <a name="int_part_select">'<tt>llvm.part.select.*</tt>' Intrinsic</a>
4731 <div class="doc_text">
4734 <p>This is an overloaded intrinsic. You can use <tt>llvm.part.select</tt>
4735 on any integer bit width.
4737 declare i17 @llvm.part.select.i17 (i17 %val, i32 %loBit, i32 %hiBit)
4738 declare i29 @llvm.part.select.i29 (i29 %val, i32 %loBit, i32 %hiBit)
4742 <p>The '<tt>llvm.part.select</tt>' family of intrinsic functions selects a
4743 range of bits from an integer value and returns them in the same bit width as
4744 the original value.</p>
4747 <p>The first argument, <tt>%val</tt> and the result may be integer types of
4748 any bit width but they must have the same bit width. The second and third
4749 arguments must be <tt>i32</tt> type since they specify only a bit index.</p>
4752 <p>The operation of the '<tt>llvm.part.select</tt>' intrinsic has two modes
4753 of operation: forwards and reverse. If <tt>%loBit</tt> is greater than
4754 <tt>%hiBits</tt> then the intrinsic operates in reverse mode. Otherwise it
4755 operates in forward mode.</p>
4756 <p>In forward mode, this intrinsic is the equivalent of shifting <tt>%val</tt>
4757 right by <tt>%loBit</tt> bits and then ANDing it with a mask with
4758 only the <tt>%hiBit - %loBit</tt> bits set, as follows:</p>
4760 <li>The <tt>%val</tt> is shifted right (LSHR) by the number of bits specified
4761 by <tt>%loBits</tt>. This normalizes the value to the low order bits.</li>
4762 <li>The <tt>%loBits</tt> value is subtracted from the <tt>%hiBits</tt> value
4763 to determine the number of bits to retain.</li>
4764 <li>A mask of the retained bits is created by shifting a -1 value.</li>
4765 <li>The mask is ANDed with <tt>%val</tt> to produce the result.
4767 <p>In reverse mode, a similar computation is made except that the bits are
4768 returned in the reverse order. So, for example, if <tt>X</tt> has the value
4769 <tt>i16 0x0ACF (101011001111)</tt> and we apply
4770 <tt>part.select(i16 X, 8, 3)</tt> to it, we get back the value
4771 <tt>i16 0x0026 (000000100110)</tt>.</p>
4774 <div class="doc_subsubsection">
4775 <a name="int_part_set">'<tt>llvm.part.set.*</tt>' Intrinsic</a>
4778 <div class="doc_text">
4781 <p>This is an overloaded intrinsic. You can use <tt>llvm.part.set</tt>
4782 on any integer bit width.
4784 declare i17 @llvm.part.set.i17.i9 (i17 %val, i9 %repl, i32 %lo, i32 %hi)
4785 declare i29 @llvm.part.set.i29.i9 (i29 %val, i9 %repl, i32 %lo, i32 %hi)
4789 <p>The '<tt>llvm.part.set</tt>' family of intrinsic functions replaces a range
4790 of bits in an integer value with another integer value. It returns the integer
4791 with the replaced bits.</p>
4794 <p>The first argument, <tt>%val</tt> and the result may be integer types of
4795 any bit width but they must have the same bit width. <tt>%val</tt> is the value
4796 whose bits will be replaced. The second argument, <tt>%repl</tt> may be an
4797 integer of any bit width. The third and fourth arguments must be <tt>i32</tt>
4798 type since they specify only a bit index.</p>
4801 <p>The operation of the '<tt>llvm.part.set</tt>' intrinsic has two modes
4802 of operation: forwards and reverse. If <tt>%lo</tt> is greater than
4803 <tt>%hi</tt> then the intrinsic operates in reverse mode. Otherwise it
4804 operates in forward mode.</p>
4805 <p>For both modes, the <tt>%repl</tt> value is prepared for use by either
4806 truncating it down to the size of the replacement area or zero extending it
4807 up to that size.</p>
4808 <p>In forward mode, the bits between <tt>%lo</tt> and <tt>%hi</tt> (inclusive)
4809 are replaced with corresponding bits from <tt>%repl</tt>. That is the 0th bit
4810 in <tt>%repl</tt> replaces the <tt>%lo</tt>th bit in <tt>%val</tt> and etc. up
4811 to the <tt>%hi</tt>th bit.
4812 <p>In reverse mode, a similar computation is made except that the bits are
4813 reversed. That is, the <tt>0</tt>th bit in <tt>%repl</tt> replaces the
4814 <tt>%hi</tt> bit in <tt>%val</tt> and etc. down to the <tt>%lo</tt>th bit.
4817 llvm.part.set(0xFFFF, 0, 4, 7) -> 0xFF0F
4818 llvm.part.set(0xFFFF, 0, 7, 4) -> 0xFF0F
4819 llvm.part.set(0xFFFF, 1, 7, 4) -> 0xFF8F
4820 llvm.part.set(0xFFFF, F, 8, 3) -> 0xFFE7
4821 llvm.part.set(0xFFFF, 0, 3, 8) -> 0xFE07
4825 <!-- ======================================================================= -->
4826 <div class="doc_subsection">
4827 <a name="int_debugger">Debugger Intrinsics</a>
4830 <div class="doc_text">
4832 The LLVM debugger intrinsics (which all start with <tt>llvm.dbg.</tt> prefix),
4833 are described in the <a
4834 href="SourceLevelDebugging.html#format_common_intrinsics">LLVM Source Level
4835 Debugging</a> document.
4840 <!-- ======================================================================= -->
4841 <div class="doc_subsection">
4842 <a name="int_eh">Exception Handling Intrinsics</a>
4845 <div class="doc_text">
4846 <p> The LLVM exception handling intrinsics (which all start with
4847 <tt>llvm.eh.</tt> prefix), are described in the <a
4848 href="ExceptionHandling.html#format_common_intrinsics">LLVM Exception
4849 Handling</a> document. </p>
4852 <!-- ======================================================================= -->
4853 <div class="doc_subsection">
4854 <a name="int_atomics">Atomic Operations and Synchronization Intrinsics</a>
4857 <div class="doc_text">
4859 These intrinsic functions expand the "universal IR" of LLVM to represent
4860 hardware constructs for atomic operations and memory synchronization. This
4861 provides an interface to the hardware, not an interface to the programmer. It
4862 is aimed at a low enough level to allow any programming models or APIs which
4863 need atomic behaviors to map cleanly onto it. It is also modeled primarily on
4864 hardware behavior. Just as hardware provides a "universal IR" for source
4865 languages, it also provides a starting point for developing a "universal"
4866 atomic operation and synchronization IR.
4869 These do <em>not</em> form an API such as high-level threading libraries,
4870 software transaction memory systems, atomic primitives, and intrinsic
4871 functions as found in BSD, GNU libc, atomic_ops, APR, and other system and
4872 application libraries. The hardware interface provided by LLVM should allow
4873 a clean implementation of all of these APIs and parallel programming models.
4874 No one model or paradigm should be selected above others unless the hardware
4875 itself ubiquitously does so.
4879 <!-- _______________________________________________________________________ -->
4880 <div class="doc_subsubsection">
4881 <a name="int_lcs">'<tt>llvm.atomic.lcs.*</tt>' Intrinsic</a>
4883 <div class="doc_text">
4886 This is an overloaded intrinsic. You can use <tt>llvm.atomic.lcs</tt> on any
4887 integer bit width. Not all targets support all bit widths however.</p>
4889 declare i8 @llvm.atomic.lcs.i8.i8p.i8.i8( i8* <ptr>, i8 <cmp>, i8 <val> )
4890 declare i16 @llvm.atomic.lcs.i16.i16p.i16.i16( i16* <ptr>, i16 <cmp>, i16 <val> )
4891 declare i32 @llvm.atomic.lcs.i32.i32p.i32.i32( i32* <ptr>, i32 <cmp>, i32 <val> )
4892 declare i64 @llvm.atomic.lcs.i64.i64p.i64.i64( i64* <ptr>, i64 <cmp>, i64 <val> )
4896 This loads a value in memory and compares it to a given value. If they are
4897 equal, it stores a new value into the memory.
4901 The <tt>llvm.atomic.lcs</tt> intrinsic takes three arguments. The result as
4902 well as both <tt>cmp</tt> and <tt>val</tt> must be integer values with the
4903 same bit width. The <tt>ptr</tt> argument must be a pointer to a value of
4904 this integer type. While any bit width integer may be used, targets may only
4905 lower representations they support in hardware.
4909 This entire intrinsic must be executed atomically. It first loads the value
4910 in memory pointed to by <tt>ptr</tt> and compares it with the value
4911 <tt>cmp</tt>. If they are equal, <tt>val</tt> is stored into the memory. The
4912 loaded value is yielded in all cases. This provides the equivalent of an
4913 atomic compare-and-swap operation within the SSA framework.
4920 %val1 = add i32 4, 4
4921 %result1 = call i32 @llvm.atomic.lcs( i32* %ptr, i32 4, %val1 )
4922 <i>; yields {i32}:result1 = 4</i>
4923 %stored1 = icmp eq i32 %result1, 4 <i>; yields {i1}:stored1 = true</i>
4924 %memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 8</i>
4926 %val2 = add i32 1, 1
4927 %result2 = call i32 @llvm.atomic.lcs( i32* %ptr, i32 5, %val2 )
4928 <i>; yields {i32}:result2 = 8</i>
4929 %stored2 = icmp eq i32 %result2, 5 <i>; yields {i1}:stored2 = false</i>
4930 %memval2 = load i32* %ptr <i>; yields {i32}:memval2 = 8</i>
4934 <!-- _______________________________________________________________________ -->
4935 <div class="doc_subsubsection">
4936 <a name="int_ls">'<tt>llvm.atomic.ls.*</tt>' Intrinsic</a>
4938 <div class="doc_text">
4941 This is an overloaded intrinsic. You can use <tt>llvm.atomic.ls</tt> on any
4942 integer bit width. Not all targets support all bit widths however.</p>
4944 declare i8 @llvm.atomic.ls.i8.i8p.i8( i8* <ptr>, i8 <val> )
4945 declare i16 @llvm.atomic.ls.i16.i16p.i16( i16* <ptr>, i16 <val> )
4946 declare i32 @llvm.atomic.ls.i32.i32p.i32( i32* <ptr>, i32 <val> )
4947 declare i64 @llvm.atomic.ls.i64.i64p.i64( i64* <ptr>, i64 <val> )
4951 This intrinsic loads the value stored in memory at <tt>ptr</tt> and yields
4952 the value from memory. It then stores the value in <tt>val</tt> in the memory
4957 The <tt>llvm.atomic.ls</tt> intrinsic takes two arguments. Both the
4958 <tt>val</tt> argument and the result must be integers of the same bit width.
4959 The first argument, <tt>ptr</tt>, must be a pointer to a value of this
4960 integer type. The targets may only lower integer representations they
4965 This intrinsic loads the value pointed to by <tt>ptr</tt>, yields it, and
4966 stores <tt>val</tt> back into <tt>ptr</tt> atomically. This provides the
4967 equivalent of an atomic swap operation within the SSA framework.
4974 %val1 = add i32 4, 4
4975 %result1 = call i32 @llvm.atomic.ls( i32* %ptr, i32 %val1 )
4976 <i>; yields {i32}:result1 = 4</i>
4977 %stored1 = icmp eq i32 %result1, 4 <i>; yields {i1}:stored1 = true</i>
4978 %memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 8</i>
4980 %val2 = add i32 1, 1
4981 %result2 = call i32 @llvm.atomic.ls( i32* %ptr, i32 %val2 )
4982 <i>; yields {i32}:result2 = 8</i>
4983 %stored2 = icmp eq i32 %result2, 8 <i>; yields {i1}:stored2 = true</i>
4984 %memval2 = load i32* %ptr <i>; yields {i32}:memval2 = 2</i>
4988 <!-- _______________________________________________________________________ -->
4989 <div class="doc_subsubsection">
4990 <a name="int_las">'<tt>llvm.atomic.las.*</tt>' Intrinsic</a>
4992 <div class="doc_text">
4995 This is an overloaded intrinsic. You can use <tt>llvm.atomic.las</tt> on any
4996 integer bit width. Not all targets support all bit widths however.</p>
4998 declare i8 @llvm.atomic.las.i8.i8p.i8( i8* <ptr>, i8 <delta> )
4999 declare i16 @llvm.atomic.las.i16.i16p.i16( i16* <ptr>, i16 <delta> )
5000 declare i32 @llvm.atomic.las.i32.i32p.i32( i32* <ptr>, i32 <delta> )
5001 declare i64 @llvm.atomic.las.i64.i64p.i64( i64* <ptr>, i64 <delta> )
5005 This intrinsic adds <tt>delta</tt> to the value stored in memory at
5006 <tt>ptr</tt>. It yields the original value at <tt>ptr</tt>.
5010 The intrinsic takes two arguments, the first a pointer to an integer value
5011 and the second an integer value. The result is also an integer value. These
5012 integer types can have any bit width, but they must all have the same bit
5013 width. The targets may only lower integer representations they support.
5017 This intrinsic does a series of operations atomically. It first loads the
5018 value stored at <tt>ptr</tt>. It then adds <tt>delta</tt>, stores the result
5019 to <tt>ptr</tt>. It yields the original value stored at <tt>ptr</tt>.
5025 %result1 = call i32 @llvm.atomic.las( i32* %ptr, i32 4 )
5026 <i>; yields {i32}:result1 = 4</i>
5027 %result2 = call i32 @llvm.atomic.las( i32* %ptr, i32 2 )
5028 <i>; yields {i32}:result2 = 8</i>
5029 %result3 = call i32 @llvm.atomic.las( i32* %ptr, i32 5 )
5030 <i>; yields {i32}:result3 = 10</i>
5031 %memval = load i32* %ptr <i>; yields {i32}:memval1 = 15</i>
5035 <!-- _______________________________________________________________________ -->
5036 <div class="doc_subsubsection">
5037 <a name="int_lss">'<tt>llvm.atomic.lss.*</tt>' Intrinsic</a>
5039 <div class="doc_text">
5042 This is an overloaded intrinsic. You can use <tt>llvm.atomic.lss</tt> on any
5043 integer bit width. Not all targets support all bit widths however.</p>
5045 declare i8 @llvm.atomic.lss.i8.i8.i8( i8* <ptr>, i8 <delta> )
5046 declare i16 @llvm.atomic.lss.i16.i16.i16( i16* <ptr>, i16 <delta> )
5047 declare i32 @llvm.atomic.lss.i32.i32.i32( i32* <ptr>, i32 <delta> )
5048 declare i64 @llvm.atomic.lss.i64.i64.i64( i64* <ptr>, i64 <delta> )
5052 This intrinsic subtracts <tt>delta</tt> from the value stored in memory at
5053 <tt>ptr</tt>. It yields the original value at <tt>ptr</tt>.
5057 The intrinsic takes two arguments, the first a pointer to an integer value
5058 and the second an integer value. The result is also an integer value. These
5059 integer types can have any bit width, but they must all have the same bit
5060 width. The targets may only lower integer representations they support.
5064 This intrinsic does a series of operations atomically. It first loads the
5065 value stored at <tt>ptr</tt>. It then subtracts <tt>delta</tt>,
5066 stores the result to <tt>ptr</tt>. It yields the original value stored
5073 %result1 = call i32 @llvm.atomic.lss( i32* %ptr, i32 4 )
5074 <i>; yields {i32}:result1 = 32</i>
5075 %result2 = call i32 @llvm.atomic.lss( i32* %ptr, i32 2 )
5076 <i>; yields {i32}:result2 = 28</i>
5077 %result3 = call i32 @llvm.atomic.lss( i32* %ptr, i32 5 )
5078 <i>; yields {i32}:result3 = 26</i>
5079 %memval = load i32* %ptr <i>; yields {i32}:memval1 = 21</i>
5083 <!-- _______________________________________________________________________ -->
5084 <div class="doc_subsubsection">
5085 <a name="int_memory_barrier">'<tt>llvm.memory.barrier</tt>' Intrinsic</a>
5087 <div class="doc_text">
5090 declare void @llvm.memory.barrier( i1 <ll>, i1 <ls>, i1 <sl>, i1 <ss> )
5094 The <tt>llvm.memory.barrier</tt> intrinsic guarantees ordering between
5095 specific pairs of memory access types.
5099 The <tt>llvm.memory.barrier</tt> intrinsic requires four boolean arguments.
5100 Each argument enables a specific barrier as listed below.
5103 <li><tt>ll</tt>: load-load barrier</li>
5104 <li><tt>ls</tt>: load-store barrier</li>
5105 <li><tt>sl</tt>: store-load barrier</li>
5106 <li><tt>ss</tt>: store-store barrier</li>
5110 This intrinsic causes the system to enforce some ordering constraints upon
5111 the loads and stores of the program. This barrier does not indicate
5112 <em>when</em> any events will occur, it only enforces an <em>order</em> in
5113 which they occur. For any of the specified pairs of load and store operations
5114 (f.ex. load-load, or store-load), all of the first operations preceding the
5115 barrier will complete before any of the second operations succeeding the
5116 barrier begin. Specifically the semantics for each pairing is as follows:
5119 <li><tt>ll</tt>: All loads before the barrier must complete before any load
5120 after the barrier begins.</li>
5121 <li><tt>ls</tt>: All loads before the barrier must complete before any
5122 store after the barrier begins.</li>
5123 <li><tt>ss</tt>: All stores before the barrier must complete before any
5124 store after the barrier begins.</li>
5125 <li><tt>sl</tt>: All stores before the barrier must complete before any
5126 load after the barrier begins.</li>
5129 These semantics are applied with a logical "and" behavior when more than one
5130 is enabled in a single memory barrier intrinsic.
5137 %result1 = load i32* %ptr <i>; yields {i32}:result1 = 4</i>
5138 call void @llvm.memory.barrier( i1 false, i1 true, i1 false, i1 false )
5139 <i>; guarantee the above finishes</i>
5140 store i32 8, %ptr <i>; before this begins</i>
5144 <!-- ======================================================================= -->
5145 <div class="doc_subsection">
5146 <a name="int_trampoline">Trampoline Intrinsic</a>
5149 <div class="doc_text">
5151 This intrinsic makes it possible to excise one parameter, marked with
5152 the <tt>nest</tt> attribute, from a function. The result is a callable
5153 function pointer lacking the nest parameter - the caller does not need
5154 to provide a value for it. Instead, the value to use is stored in
5155 advance in a "trampoline", a block of memory usually allocated
5156 on the stack, which also contains code to splice the nest value into the
5157 argument list. This is used to implement the GCC nested function address
5161 For example, if the function is
5162 <tt>i32 f(i8* nest %c, i32 %x, i32 %y)</tt> then the resulting function
5163 pointer has signature <tt>i32 (i32, i32)*</tt>. It can be created as follows:</p>
5165 %tramp = alloca [10 x i8], align 4 ; size and alignment only correct for X86
5166 %tramp1 = getelementptr [10 x i8]* %tramp, i32 0, i32 0
5167 %p = call i8* @llvm.init.trampoline( i8* %tramp1, i8* bitcast (i32 (i8* nest , i32, i32)* @f to i8*), i8* %nval )
5168 %fp = bitcast i8* %p to i32 (i32, i32)*
5170 <p>The call <tt>%val = call i32 %fp( i32 %x, i32 %y )</tt> is then equivalent
5171 to <tt>%val = call i32 %f( i8* %nval, i32 %x, i32 %y )</tt>.</p>
5174 <!-- _______________________________________________________________________ -->
5175 <div class="doc_subsubsection">
5176 <a name="int_it">'<tt>llvm.init.trampoline</tt>' Intrinsic</a>
5178 <div class="doc_text">
5181 declare i8* @llvm.init.trampoline(i8* <tramp>, i8* <func>, i8* <nval>)
5185 This fills the memory pointed to by <tt>tramp</tt> with code
5186 and returns a function pointer suitable for executing it.
5190 The <tt>llvm.init.trampoline</tt> intrinsic takes three arguments, all
5191 pointers. The <tt>tramp</tt> argument must point to a sufficiently large
5192 and sufficiently aligned block of memory; this memory is written to by the
5193 intrinsic. Note that the size and the alignment are target-specific - LLVM
5194 currently provides no portable way of determining them, so a front-end that
5195 generates this intrinsic needs to have some target-specific knowledge.
5196 The <tt>func</tt> argument must hold a function bitcast to an <tt>i8*</tt>.
5200 The block of memory pointed to by <tt>tramp</tt> is filled with target
5201 dependent code, turning it into a function. A pointer to this function is
5202 returned, but needs to be bitcast to an
5203 <a href="#int_trampoline">appropriate function pointer type</a>
5204 before being called. The new function's signature is the same as that of
5205 <tt>func</tt> with any arguments marked with the <tt>nest</tt> attribute
5206 removed. At most one such <tt>nest</tt> argument is allowed, and it must be
5207 of pointer type. Calling the new function is equivalent to calling
5208 <tt>func</tt> with the same argument list, but with <tt>nval</tt> used for the
5209 missing <tt>nest</tt> argument. If, after calling
5210 <tt>llvm.init.trampoline</tt>, the memory pointed to by <tt>tramp</tt> is
5211 modified, then the effect of any later call to the returned function pointer is
5216 <!-- ======================================================================= -->
5217 <div class="doc_subsection">
5218 <a name="int_general">General Intrinsics</a>
5221 <div class="doc_text">
5222 <p> This class of intrinsics is designed to be generic and has
5223 no specific purpose. </p>
5226 <!-- _______________________________________________________________________ -->
5227 <div class="doc_subsubsection">
5228 <a name="int_var_annotation">'<tt>llvm.var.annotation</tt>' Intrinsic</a>
5231 <div class="doc_text">
5235 declare void @llvm.var.annotation(i8* <val>, i8* <str>, i8* <str>, i32 <int> )
5241 The '<tt>llvm.var.annotation</tt>' intrinsic
5247 The first argument is a pointer to a value, the second is a pointer to a
5248 global string, the third is a pointer to a global string which is the source
5249 file name, and the last argument is the line number.
5255 This intrinsic allows annotation of local variables with arbitrary strings.
5256 This can be useful for special purpose optimizations that want to look for these
5257 annotations. These have no other defined use, they are ignored by code
5258 generation and optimization.
5261 <!-- _______________________________________________________________________ -->
5262 <div class="doc_subsubsection">
5263 <a name="int_annotation">'<tt>llvm.annotation.*</tt>' Intrinsic</a>
5266 <div class="doc_text">
5269 <p>This is an overloaded intrinsic. You can use '<tt>llvm.annotation</tt>' on
5270 any integer bit width.
5273 declare i8 @llvm.annotation.i8(i8 <val>, i8* <str>, i8* <str>, i32 <int> )
5274 declare i16 @llvm.annotation.i16(i16 <val>, i8* <str>, i8* <str>, i32 <int> )
5275 declare i32 @llvm.annotation.i32(i32 <val>, i8* <str>, i8* <str>, i32 <int> )
5276 declare i64 @llvm.annotation.i64(i64 <val>, i8* <str>, i8* <str>, i32 <int> )
5277 declare i256 @llvm.annotation.i256(i256 <val>, i8* <str>, i8* <str>, i32 <int> )
5283 The '<tt>llvm.annotation</tt>' intrinsic.
5289 The first argument is an integer value (result of some expression),
5290 the second is a pointer to a global string, the third is a pointer to a global
5291 string which is the source file name, and the last argument is the line number.
5292 It returns the value of the first argument.
5298 This intrinsic allows annotations to be put on arbitrary expressions
5299 with arbitrary strings. This can be useful for special purpose optimizations
5300 that want to look for these annotations. These have no other defined use, they
5301 are ignored by code generation and optimization.
5304 <!-- *********************************************************************** -->
5307 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
5308 src="http://jigsaw.w3.org/css-validator/images/vcss" alt="Valid CSS!"></a>
5309 <a href="http://validator.w3.org/check/referer"><img
5310 src="http://www.w3.org/Icons/valid-html401" alt="Valid HTML 4.01!" /></a>
5312 <a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
5313 <a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br>
5314 Last modified: $Date$