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>
25 <li><a href="#linkage_private">'<tt>private</tt>' Linkage</a></li>
26 <li><a href="#linkage_linker_private">'<tt>linker_private</tt>' Linkage</a></li>
27 <li><a href="#linkage_internal">'<tt>internal</tt>' Linkage</a></li>
28 <li><a href="#linkage_available_externally">'<tt>available_externally</tt>' Linkage</a></li>
29 <li><a href="#linkage_linkonce">'<tt>linkonce</tt>' Linkage</a></li>
30 <li><a href="#linkage_common">'<tt>common</tt>' Linkage</a></li>
31 <li><a href="#linkage_weak">'<tt>weak</tt>' Linkage</a></li>
32 <li><a href="#linkage_appending">'<tt>appending</tt>' Linkage</a></li>
33 <li><a href="#linkage_externweak">'<tt>extern_weak</tt>' Linkage</a></li>
34 <li><a href="#linkage_linkonce_odr">'<tt>linkonce_odr</tt>' Linkage</a></li>
35 <li><a href="#linkage_weak">'<tt>weak_odr</tt>' Linkage</a></li>
36 <li><a href="#linkage_external">'<tt>externally visible</tt>' Linkage</a></li>
37 <li><a href="#linkage_dllimport">'<tt>dllimport</tt>' Linkage</a></li>
38 <li><a href="#linkage_dllexport">'<tt>dllexport</tt>' Linkage</a></li>
41 <li><a href="#callingconv">Calling Conventions</a></li>
42 <li><a href="#namedtypes">Named Types</a></li>
43 <li><a href="#globalvars">Global Variables</a></li>
44 <li><a href="#functionstructure">Functions</a></li>
45 <li><a href="#aliasstructure">Aliases</a></li>
46 <li><a href="#paramattrs">Parameter Attributes</a></li>
47 <li><a href="#fnattrs">Function Attributes</a></li>
48 <li><a href="#gc">Garbage Collector Names</a></li>
49 <li><a href="#moduleasm">Module-Level Inline Assembly</a></li>
50 <li><a href="#datalayout">Data Layout</a></li>
51 <li><a href="#pointeraliasing">Pointer Aliasing Rules</a></li>
54 <li><a href="#typesystem">Type System</a>
56 <li><a href="#t_classifications">Type Classifications</a></li>
57 <li><a href="#t_primitive">Primitive Types</a>
59 <li><a href="#t_integer">Integer Type</a></li>
60 <li><a href="#t_floating">Floating Point Types</a></li>
61 <li><a href="#t_void">Void Type</a></li>
62 <li><a href="#t_label">Label Type</a></li>
63 <li><a href="#t_metadata">Metadata Type</a></li>
66 <li><a href="#t_derived">Derived Types</a>
68 <li><a href="#t_array">Array Type</a></li>
69 <li><a href="#t_function">Function Type</a></li>
70 <li><a href="#t_pointer">Pointer Type</a></li>
71 <li><a href="#t_struct">Structure Type</a></li>
72 <li><a href="#t_pstruct">Packed Structure Type</a></li>
73 <li><a href="#t_vector">Vector Type</a></li>
74 <li><a href="#t_opaque">Opaque Type</a></li>
77 <li><a href="#t_uprefs">Type Up-references</a></li>
80 <li><a href="#constants">Constants</a>
82 <li><a href="#simpleconstants">Simple Constants</a></li>
83 <li><a href="#complexconstants">Complex Constants</a></li>
84 <li><a href="#globalconstants">Global Variable and Function Addresses</a></li>
85 <li><a href="#undefvalues">Undefined Values</a></li>
86 <li><a href="#blockaddress">Addresses of Basic Blocks</a></li>
87 <li><a href="#constantexprs">Constant Expressions</a></li>
88 <li><a href="#metadata">Embedded Metadata</a></li>
91 <li><a href="#othervalues">Other Values</a>
93 <li><a href="#inlineasm">Inline Assembler Expressions</a></li>
96 <li><a href="#intrinsic_globals">Intrinsic Global Variables</a>
98 <li><a href="#intg_used">The '<tt>llvm.used</tt>' Global Variable</a></li>
99 <li><a href="#intg_compiler_used">The '<tt>llvm.compiler.used</tt>'
100 Global Variable</a></li>
101 <li><a href="#intg_global_ctors">The '<tt>llvm.global_ctors</tt>'
102 Global Variable</a></li>
103 <li><a href="#intg_global_dtors">The '<tt>llvm.global_dtors</tt>'
104 Global Variable</a></li>
107 <li><a href="#instref">Instruction Reference</a>
109 <li><a href="#terminators">Terminator Instructions</a>
111 <li><a href="#i_ret">'<tt>ret</tt>' Instruction</a></li>
112 <li><a href="#i_br">'<tt>br</tt>' Instruction</a></li>
113 <li><a href="#i_switch">'<tt>switch</tt>' Instruction</a></li>
114 <li><a href="#i_indirectbr">'<tt>indirectbr</tt>' Instruction</a></li>
115 <li><a href="#i_invoke">'<tt>invoke</tt>' Instruction</a></li>
116 <li><a href="#i_unwind">'<tt>unwind</tt>' Instruction</a></li>
117 <li><a href="#i_unreachable">'<tt>unreachable</tt>' Instruction</a></li>
120 <li><a href="#binaryops">Binary Operations</a>
122 <li><a href="#i_add">'<tt>add</tt>' Instruction</a></li>
123 <li><a href="#i_fadd">'<tt>fadd</tt>' Instruction</a></li>
124 <li><a href="#i_sub">'<tt>sub</tt>' Instruction</a></li>
125 <li><a href="#i_fsub">'<tt>fsub</tt>' Instruction</a></li>
126 <li><a href="#i_mul">'<tt>mul</tt>' Instruction</a></li>
127 <li><a href="#i_fmul">'<tt>fmul</tt>' Instruction</a></li>
128 <li><a href="#i_udiv">'<tt>udiv</tt>' Instruction</a></li>
129 <li><a href="#i_sdiv">'<tt>sdiv</tt>' Instruction</a></li>
130 <li><a href="#i_fdiv">'<tt>fdiv</tt>' Instruction</a></li>
131 <li><a href="#i_urem">'<tt>urem</tt>' Instruction</a></li>
132 <li><a href="#i_srem">'<tt>srem</tt>' Instruction</a></li>
133 <li><a href="#i_frem">'<tt>frem</tt>' Instruction</a></li>
136 <li><a href="#bitwiseops">Bitwise Binary Operations</a>
138 <li><a href="#i_shl">'<tt>shl</tt>' Instruction</a></li>
139 <li><a href="#i_lshr">'<tt>lshr</tt>' Instruction</a></li>
140 <li><a href="#i_ashr">'<tt>ashr</tt>' Instruction</a></li>
141 <li><a href="#i_and">'<tt>and</tt>' Instruction</a></li>
142 <li><a href="#i_or">'<tt>or</tt>' Instruction</a></li>
143 <li><a href="#i_xor">'<tt>xor</tt>' Instruction</a></li>
146 <li><a href="#vectorops">Vector Operations</a>
148 <li><a href="#i_extractelement">'<tt>extractelement</tt>' Instruction</a></li>
149 <li><a href="#i_insertelement">'<tt>insertelement</tt>' Instruction</a></li>
150 <li><a href="#i_shufflevector">'<tt>shufflevector</tt>' Instruction</a></li>
153 <li><a href="#aggregateops">Aggregate Operations</a>
155 <li><a href="#i_extractvalue">'<tt>extractvalue</tt>' Instruction</a></li>
156 <li><a href="#i_insertvalue">'<tt>insertvalue</tt>' Instruction</a></li>
159 <li><a href="#memoryops">Memory Access and Addressing Operations</a>
161 <li><a href="#i_alloca">'<tt>alloca</tt>' Instruction</a></li>
162 <li><a href="#i_load">'<tt>load</tt>' Instruction</a></li>
163 <li><a href="#i_store">'<tt>store</tt>' Instruction</a></li>
164 <li><a href="#i_getelementptr">'<tt>getelementptr</tt>' Instruction</a></li>
167 <li><a href="#convertops">Conversion Operations</a>
169 <li><a href="#i_trunc">'<tt>trunc .. to</tt>' Instruction</a></li>
170 <li><a href="#i_zext">'<tt>zext .. to</tt>' Instruction</a></li>
171 <li><a href="#i_sext">'<tt>sext .. to</tt>' Instruction</a></li>
172 <li><a href="#i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a></li>
173 <li><a href="#i_fpext">'<tt>fpext .. to</tt>' Instruction</a></li>
174 <li><a href="#i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a></li>
175 <li><a href="#i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a></li>
176 <li><a href="#i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a></li>
177 <li><a href="#i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a></li>
178 <li><a href="#i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a></li>
179 <li><a href="#i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a></li>
180 <li><a href="#i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a></li>
183 <li><a href="#otherops">Other Operations</a>
185 <li><a href="#i_icmp">'<tt>icmp</tt>' Instruction</a></li>
186 <li><a href="#i_fcmp">'<tt>fcmp</tt>' Instruction</a></li>
187 <li><a href="#i_phi">'<tt>phi</tt>' Instruction</a></li>
188 <li><a href="#i_select">'<tt>select</tt>' Instruction</a></li>
189 <li><a href="#i_call">'<tt>call</tt>' Instruction</a></li>
190 <li><a href="#i_va_arg">'<tt>va_arg</tt>' Instruction</a></li>
195 <li><a href="#intrinsics">Intrinsic Functions</a>
197 <li><a href="#int_varargs">Variable Argument Handling Intrinsics</a>
199 <li><a href="#int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a></li>
200 <li><a href="#int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a></li>
201 <li><a href="#int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a></li>
204 <li><a href="#int_gc">Accurate Garbage Collection Intrinsics</a>
206 <li><a href="#int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a></li>
207 <li><a href="#int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a></li>
208 <li><a href="#int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a></li>
211 <li><a href="#int_codegen">Code Generator Intrinsics</a>
213 <li><a href="#int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a></li>
214 <li><a href="#int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a></li>
215 <li><a href="#int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a></li>
216 <li><a href="#int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a></li>
217 <li><a href="#int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a></li>
218 <li><a href="#int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a></li>
219 <li><a href="#int_readcyclecounter"><tt>llvm.readcyclecounter</tt>' Intrinsic</a></li>
222 <li><a href="#int_libc">Standard C Library Intrinsics</a>
224 <li><a href="#int_memcpy">'<tt>llvm.memcpy.*</tt>' Intrinsic</a></li>
225 <li><a href="#int_memmove">'<tt>llvm.memmove.*</tt>' Intrinsic</a></li>
226 <li><a href="#int_memset">'<tt>llvm.memset.*</tt>' Intrinsic</a></li>
227 <li><a href="#int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a></li>
228 <li><a href="#int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a></li>
229 <li><a href="#int_sin">'<tt>llvm.sin.*</tt>' Intrinsic</a></li>
230 <li><a href="#int_cos">'<tt>llvm.cos.*</tt>' Intrinsic</a></li>
231 <li><a href="#int_pow">'<tt>llvm.pow.*</tt>' Intrinsic</a></li>
234 <li><a href="#int_manip">Bit Manipulation Intrinsics</a>
236 <li><a href="#int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a></li>
237 <li><a href="#int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic </a></li>
238 <li><a href="#int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic </a></li>
239 <li><a href="#int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic </a></li>
242 <li><a href="#int_overflow">Arithmetic with Overflow Intrinsics</a>
244 <li><a href="#int_sadd_overflow">'<tt>llvm.sadd.with.overflow.*</tt> Intrinsics</a></li>
245 <li><a href="#int_uadd_overflow">'<tt>llvm.uadd.with.overflow.*</tt> Intrinsics</a></li>
246 <li><a href="#int_ssub_overflow">'<tt>llvm.ssub.with.overflow.*</tt> Intrinsics</a></li>
247 <li><a href="#int_usub_overflow">'<tt>llvm.usub.with.overflow.*</tt> Intrinsics</a></li>
248 <li><a href="#int_smul_overflow">'<tt>llvm.smul.with.overflow.*</tt> Intrinsics</a></li>
249 <li><a href="#int_umul_overflow">'<tt>llvm.umul.with.overflow.*</tt> Intrinsics</a></li>
252 <li><a href="#int_debugger">Debugger intrinsics</a></li>
253 <li><a href="#int_eh">Exception Handling intrinsics</a></li>
254 <li><a href="#int_trampoline">Trampoline Intrinsic</a>
256 <li><a href="#int_it">'<tt>llvm.init.trampoline</tt>' Intrinsic</a></li>
259 <li><a href="#int_atomics">Atomic intrinsics</a>
261 <li><a href="#int_memory_barrier"><tt>llvm.memory_barrier</tt></a></li>
262 <li><a href="#int_atomic_cmp_swap"><tt>llvm.atomic.cmp.swap</tt></a></li>
263 <li><a href="#int_atomic_swap"><tt>llvm.atomic.swap</tt></a></li>
264 <li><a href="#int_atomic_load_add"><tt>llvm.atomic.load.add</tt></a></li>
265 <li><a href="#int_atomic_load_sub"><tt>llvm.atomic.load.sub</tt></a></li>
266 <li><a href="#int_atomic_load_and"><tt>llvm.atomic.load.and</tt></a></li>
267 <li><a href="#int_atomic_load_nand"><tt>llvm.atomic.load.nand</tt></a></li>
268 <li><a href="#int_atomic_load_or"><tt>llvm.atomic.load.or</tt></a></li>
269 <li><a href="#int_atomic_load_xor"><tt>llvm.atomic.load.xor</tt></a></li>
270 <li><a href="#int_atomic_load_max"><tt>llvm.atomic.load.max</tt></a></li>
271 <li><a href="#int_atomic_load_min"><tt>llvm.atomic.load.min</tt></a></li>
272 <li><a href="#int_atomic_load_umax"><tt>llvm.atomic.load.umax</tt></a></li>
273 <li><a href="#int_atomic_load_umin"><tt>llvm.atomic.load.umin</tt></a></li>
276 <li><a href="#int_memorymarkers">Memory Use Markers</a>
278 <li><a href="#int_lifetime_start"><tt>llvm.lifetime.start</tt></a></li>
279 <li><a href="#int_lifetime_end"><tt>llvm.lifetime.end</tt></a></li>
280 <li><a href="#int_invariant_start"><tt>llvm.invariant.start</tt></a></li>
281 <li><a href="#int_invariant_end"><tt>llvm.invariant.end</tt></a></li>
284 <li><a href="#int_general">General intrinsics</a>
286 <li><a href="#int_var_annotation">
287 '<tt>llvm.var.annotation</tt>' Intrinsic</a></li>
288 <li><a href="#int_annotation">
289 '<tt>llvm.annotation.*</tt>' Intrinsic</a></li>
290 <li><a href="#int_trap">
291 '<tt>llvm.trap</tt>' Intrinsic</a></li>
292 <li><a href="#int_stackprotector">
293 '<tt>llvm.stackprotector</tt>' Intrinsic</a></li>
300 <div class="doc_author">
301 <p>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a>
302 and <a href="mailto:vadve@cs.uiuc.edu">Vikram Adve</a></p>
305 <!-- *********************************************************************** -->
306 <div class="doc_section"> <a name="abstract">Abstract </a></div>
307 <!-- *********************************************************************** -->
309 <div class="doc_text">
311 <p>This document is a reference manual for the LLVM assembly language. LLVM is
312 a Static Single Assignment (SSA) based representation that provides type
313 safety, low-level operations, flexibility, and the capability of representing
314 'all' high-level languages cleanly. It is the common code representation
315 used throughout all phases of the LLVM compilation strategy.</p>
319 <!-- *********************************************************************** -->
320 <div class="doc_section"> <a name="introduction">Introduction</a> </div>
321 <!-- *********************************************************************** -->
323 <div class="doc_text">
325 <p>The LLVM code representation is designed to be used in three different forms:
326 as an in-memory compiler IR, as an on-disk bitcode representation (suitable
327 for fast loading by a Just-In-Time compiler), and as a human readable
328 assembly language representation. This allows LLVM to provide a powerful
329 intermediate representation for efficient compiler transformations and
330 analysis, while providing a natural means to debug and visualize the
331 transformations. The three different forms of LLVM are all equivalent. This
332 document describes the human readable representation and notation.</p>
334 <p>The LLVM representation aims to be light-weight and low-level while being
335 expressive, typed, and extensible at the same time. It aims to be a
336 "universal IR" of sorts, by being at a low enough level that high-level ideas
337 may be cleanly mapped to it (similar to how microprocessors are "universal
338 IR's", allowing many source languages to be mapped to them). By providing
339 type information, LLVM can be used as the target of optimizations: for
340 example, through pointer analysis, it can be proven that a C automatic
341 variable is never accessed outside of the current function, allowing it to
342 be promoted to a simple SSA value instead of a memory location.</p>
346 <!-- _______________________________________________________________________ -->
347 <div class="doc_subsubsection"> <a name="wellformed">Well-Formedness</a> </div>
349 <div class="doc_text">
351 <p>It is important to note that this document describes 'well formed' LLVM
352 assembly language. There is a difference between what the parser accepts and
353 what is considered 'well formed'. For example, the following instruction is
354 syntactically okay, but not well formed:</p>
356 <div class="doc_code">
358 %x = <a href="#i_add">add</a> i32 1, %x
362 <p>because the definition of <tt>%x</tt> does not dominate all of its uses. The
363 LLVM infrastructure provides a verification pass that may be used to verify
364 that an LLVM module is well formed. This pass is automatically run by the
365 parser after parsing input assembly and by the optimizer before it outputs
366 bitcode. The violations pointed out by the verifier pass indicate bugs in
367 transformation passes or input to the parser.</p>
371 <!-- Describe the typesetting conventions here. -->
373 <!-- *********************************************************************** -->
374 <div class="doc_section"> <a name="identifiers">Identifiers</a> </div>
375 <!-- *********************************************************************** -->
377 <div class="doc_text">
379 <p>LLVM identifiers come in two basic types: global and local. Global
380 identifiers (functions, global variables) begin with the <tt>'@'</tt>
381 character. Local identifiers (register names, types) begin with
382 the <tt>'%'</tt> character. Additionally, there are three different formats
383 for identifiers, for different purposes:</p>
386 <li>Named values are represented as a string of characters with their prefix.
387 For example, <tt>%foo</tt>, <tt>@DivisionByZero</tt>,
388 <tt>%a.really.long.identifier</tt>. The actual regular expression used is
389 '<tt>[%@][a-zA-Z$._][a-zA-Z$._0-9]*</tt>'. Identifiers which require
390 other characters in their names can be surrounded with quotes. Special
391 characters may be escaped using <tt>"\xx"</tt> where <tt>xx</tt> is the
392 ASCII code for the character in hexadecimal. In this way, any character
393 can be used in a name value, even quotes themselves.</li>
395 <li>Unnamed values are represented as an unsigned numeric value with their
396 prefix. For example, <tt>%12</tt>, <tt>@2</tt>, <tt>%44</tt>.</li>
398 <li>Constants, which are described in a <a href="#constants">section about
399 constants</a>, below.</li>
402 <p>LLVM requires that values start with a prefix for two reasons: Compilers
403 don't need to worry about name clashes with reserved words, and the set of
404 reserved words may be expanded in the future without penalty. Additionally,
405 unnamed identifiers allow a compiler to quickly come up with a temporary
406 variable without having to avoid symbol table conflicts.</p>
408 <p>Reserved words in LLVM are very similar to reserved words in other
409 languages. There are keywords for different opcodes
410 ('<tt><a href="#i_add">add</a></tt>',
411 '<tt><a href="#i_bitcast">bitcast</a></tt>',
412 '<tt><a href="#i_ret">ret</a></tt>', etc...), for primitive type names
413 ('<tt><a href="#t_void">void</a></tt>',
414 '<tt><a href="#t_primitive">i32</a></tt>', etc...), and others. These
415 reserved words cannot conflict with variable names, because none of them
416 start with a prefix character (<tt>'%'</tt> or <tt>'@'</tt>).</p>
418 <p>Here is an example of LLVM code to multiply the integer variable
419 '<tt>%X</tt>' by 8:</p>
423 <div class="doc_code">
425 %result = <a href="#i_mul">mul</a> i32 %X, 8
429 <p>After strength reduction:</p>
431 <div class="doc_code">
433 %result = <a href="#i_shl">shl</a> i32 %X, i8 3
437 <p>And the hard way:</p>
439 <div class="doc_code">
441 %0 = <a href="#i_add">add</a> i32 %X, %X <i>; yields {i32}:%0</i>
442 %1 = <a href="#i_add">add</a> i32 %0, %0 <i>; yields {i32}:%1</i>
443 %result = <a href="#i_add">add</a> i32 %1, %1
447 <p>This last way of multiplying <tt>%X</tt> by 8 illustrates several important
448 lexical features of LLVM:</p>
451 <li>Comments are delimited with a '<tt>;</tt>' and go until the end of
454 <li>Unnamed temporaries are created when the result of a computation is not
455 assigned to a named value.</li>
457 <li>Unnamed temporaries are numbered sequentially</li>
460 <p>It also shows a convention that we follow in this document. When
461 demonstrating instructions, we will follow an instruction with a comment that
462 defines the type and name of value produced. Comments are shown in italic
467 <!-- *********************************************************************** -->
468 <div class="doc_section"> <a name="highlevel">High Level Structure</a> </div>
469 <!-- *********************************************************************** -->
471 <!-- ======================================================================= -->
472 <div class="doc_subsection"> <a name="modulestructure">Module Structure</a>
475 <div class="doc_text">
477 <p>LLVM programs are composed of "Module"s, each of which is a translation unit
478 of the input programs. Each module consists of functions, global variables,
479 and symbol table entries. Modules may be combined together with the LLVM
480 linker, which merges function (and global variable) definitions, resolves
481 forward declarations, and merges symbol table entries. Here is an example of
482 the "hello world" module:</p>
484 <div class="doc_code">
486 <i>; Declare the string constant as a global constant.</i>
487 <a href="#identifiers">@.LC0</a> = <a href="#linkage_internal">internal</a> <a href="#globalvars">constant</a> <a href="#t_array">[13 x i8]</a> c"hello world\0A\00" <i>; [13 x i8]*</i>
489 <i>; External declaration of the puts function</i>
490 <a href="#functionstructure">declare</a> i32 @puts(i8 *) <i>; i32(i8 *)* </i>
492 <i>; Definition of main function</i>
493 define i32 @main() { <i>; i32()* </i>
494 <i>; Convert [13 x i8]* to i8 *...</i>
495 %cast210 = <a href="#i_getelementptr">getelementptr</a> [13 x i8]* @.LC0, i64 0, i64 0 <i>; i8 *</i>
497 <i>; Call puts function to write out the string to stdout.</i>
498 <a href="#i_call">call</a> i32 @puts(i8 * %cast210) <i>; i32</i>
499 <a href="#i_ret">ret</a> i32 0<br>}<br>
503 <p>This example is made up of a <a href="#globalvars">global variable</a> named
504 "<tt>.LC0</tt>", an external declaration of the "<tt>puts</tt>" function, and
505 a <a href="#functionstructure">function definition</a> for
508 <p>In general, a module is made up of a list of global values, where both
509 functions and global variables are global values. Global values are
510 represented by a pointer to a memory location (in this case, a pointer to an
511 array of char, and a pointer to a function), and have one of the
512 following <a href="#linkage">linkage types</a>.</p>
516 <!-- ======================================================================= -->
517 <div class="doc_subsection">
518 <a name="linkage">Linkage Types</a>
521 <div class="doc_text">
523 <p>All Global Variables and Functions have one of the following types of
527 <dt><tt><b><a name="linkage_private">private</a></b></tt></dt>
528 <dd>Global values with private linkage are only directly accessible by objects
529 in the current module. In particular, linking code into a module with an
530 private global value may cause the private to be renamed as necessary to
531 avoid collisions. Because the symbol is private to the module, all
532 references can be updated. This doesn't show up in any symbol table in the
535 <dt><tt><b><a name="linkage_linker_private">linker_private</a></b></tt></dt>
536 <dd>Similar to private, but the symbol is passed through the assembler and
537 removed by the linker after evaluation. Note that (unlike private
538 symbols) linker_private symbols are subject to coalescing by the linker:
539 weak symbols get merged and redefinitions are rejected. However, unlike
540 normal strong symbols, they are removed by the linker from the final
541 linked image (executable or dynamic library).</dd>
543 <dt><tt><b><a name="linkage_internal">internal</a></b></tt></dt>
544 <dd>Similar to private, but the value shows as a local symbol
545 (<tt>STB_LOCAL</tt> in the case of ELF) in the object file. This
546 corresponds to the notion of the '<tt>static</tt>' keyword in C.</dd>
548 <dt><tt><b><a name="linkage_available_externally">available_externally</a></b></tt></dt>
549 <dd>Globals with "<tt>available_externally</tt>" linkage are never emitted
550 into the object file corresponding to the LLVM module. They exist to
551 allow inlining and other optimizations to take place given knowledge of
552 the definition of the global, which is known to be somewhere outside the
553 module. Globals with <tt>available_externally</tt> linkage are allowed to
554 be discarded at will, and are otherwise the same as <tt>linkonce_odr</tt>.
555 This linkage type is only allowed on definitions, not declarations.</dd>
557 <dt><tt><b><a name="linkage_linkonce">linkonce</a></b></tt></dt>
558 <dd>Globals with "<tt>linkonce</tt>" linkage are merged with other globals of
559 the same name when linkage occurs. This is typically used to implement
560 inline functions, templates, or other code which must be generated in each
561 translation unit that uses it. Unreferenced <tt>linkonce</tt> globals are
562 allowed to be discarded.</dd>
564 <dt><tt><b><a name="linkage_weak">weak</a></b></tt></dt>
565 <dd>"<tt>weak</tt>" linkage has the same merging semantics as
566 <tt>linkonce</tt> linkage, except that unreferenced globals with
567 <tt>weak</tt> linkage may not be discarded. This is used for globals that
568 are declared "weak" in C source code.</dd>
570 <dt><tt><b><a name="linkage_common">common</a></b></tt></dt>
571 <dd>"<tt>common</tt>" linkage is most similar to "<tt>weak</tt>" linkage, but
572 they are used for tentative definitions in C, such as "<tt>int X;</tt>" at
574 Symbols with "<tt>common</tt>" linkage are merged in the same way as
575 <tt>weak symbols</tt>, and they may not be deleted if unreferenced.
576 <tt>common</tt> symbols may not have an explicit section,
577 must have a zero initializer, and may not be marked '<a
578 href="#globalvars"><tt>constant</tt></a>'. Functions and aliases may not
579 have common linkage.</dd>
582 <dt><tt><b><a name="linkage_appending">appending</a></b></tt></dt>
583 <dd>"<tt>appending</tt>" linkage may only be applied to global variables of
584 pointer to array type. When two global variables with appending linkage
585 are linked together, the two global arrays are appended together. This is
586 the LLVM, typesafe, equivalent of having the system linker append together
587 "sections" with identical names when .o files are linked.</dd>
589 <dt><tt><b><a name="linkage_externweak">extern_weak</a></b></tt></dt>
590 <dd>The semantics of this linkage follow the ELF object file model: the symbol
591 is weak until linked, if not linked, the symbol becomes null instead of
592 being an undefined reference.</dd>
594 <dt><tt><b><a name="linkage_linkonce_odr">linkonce_odr</a></b></tt></dt>
595 <dt><tt><b><a name="linkage_weak_odr">weak_odr</a></b></tt></dt>
596 <dd>Some languages allow differing globals to be merged, such as two functions
597 with different semantics. Other languages, such as <tt>C++</tt>, ensure
598 that only equivalent globals are ever merged (the "one definition rule" -
599 "ODR"). Such languages can use the <tt>linkonce_odr</tt>
600 and <tt>weak_odr</tt> linkage types to indicate that the global will only
601 be merged with equivalent globals. These linkage types are otherwise the
602 same as their non-<tt>odr</tt> versions.</dd>
604 <dt><tt><b><a name="linkage_external">externally visible</a></b></tt>:</dt>
605 <dd>If none of the above identifiers are used, the global is externally
606 visible, meaning that it participates in linkage and can be used to
607 resolve external symbol references.</dd>
610 <p>The next two types of linkage are targeted for Microsoft Windows platform
611 only. They are designed to support importing (exporting) symbols from (to)
612 DLLs (Dynamic Link Libraries).</p>
615 <dt><tt><b><a name="linkage_dllimport">dllimport</a></b></tt></dt>
616 <dd>"<tt>dllimport</tt>" linkage causes the compiler to reference a function
617 or variable via a global pointer to a pointer that is set up by the DLL
618 exporting the symbol. On Microsoft Windows targets, the pointer name is
619 formed by combining <code>__imp_</code> and the function or variable
622 <dt><tt><b><a name="linkage_dllexport">dllexport</a></b></tt></dt>
623 <dd>"<tt>dllexport</tt>" linkage causes the compiler to provide a global
624 pointer to a pointer in a DLL, so that it can be referenced with the
625 <tt>dllimport</tt> attribute. On Microsoft Windows targets, the pointer
626 name is formed by combining <code>__imp_</code> and the function or
630 <p>For example, since the "<tt>.LC0</tt>" variable is defined to be internal, if
631 another module defined a "<tt>.LC0</tt>" variable and was linked with this
632 one, one of the two would be renamed, preventing a collision. Since
633 "<tt>main</tt>" and "<tt>puts</tt>" are external (i.e., lacking any linkage
634 declarations), they are accessible outside of the current module.</p>
636 <p>It is illegal for a function <i>declaration</i> to have any linkage type
637 other than "externally visible", <tt>dllimport</tt>
638 or <tt>extern_weak</tt>.</p>
640 <p>Aliases can have only <tt>external</tt>, <tt>internal</tt>, <tt>weak</tt>
641 or <tt>weak_odr</tt> linkages.</p>
645 <!-- ======================================================================= -->
646 <div class="doc_subsection">
647 <a name="callingconv">Calling Conventions</a>
650 <div class="doc_text">
652 <p>LLVM <a href="#functionstructure">functions</a>, <a href="#i_call">calls</a>
653 and <a href="#i_invoke">invokes</a> can all have an optional calling
654 convention specified for the call. The calling convention of any pair of
655 dynamic caller/callee must match, or the behavior of the program is
656 undefined. The following calling conventions are supported by LLVM, and more
657 may be added in the future:</p>
660 <dt><b>"<tt>ccc</tt>" - The C calling convention</b>:</dt>
661 <dd>This calling convention (the default if no other calling convention is
662 specified) matches the target C calling conventions. This calling
663 convention supports varargs function calls and tolerates some mismatch in
664 the declared prototype and implemented declaration of the function (as
667 <dt><b>"<tt>fastcc</tt>" - The fast calling convention</b>:</dt>
668 <dd>This calling convention attempts to make calls as fast as possible
669 (e.g. by passing things in registers). This calling convention allows the
670 target to use whatever tricks it wants to produce fast code for the
671 target, without having to conform to an externally specified ABI
672 (Application Binary Interface). Implementations of this convention should
673 allow arbitrary <a href="CodeGenerator.html#tailcallopt">tail call
674 optimization</a> to be supported. This calling convention does not
675 support varargs and requires the prototype of all callees to exactly match
676 the prototype of the function definition.</dd>
678 <dt><b>"<tt>coldcc</tt>" - The cold calling convention</b>:</dt>
679 <dd>This calling convention attempts to make code in the caller as efficient
680 as possible under the assumption that the call is not commonly executed.
681 As such, these calls often preserve all registers so that the call does
682 not break any live ranges in the caller side. This calling convention
683 does not support varargs and requires the prototype of all callees to
684 exactly match the prototype of the function definition.</dd>
686 <dt><b>"<tt>cc <<em>n</em>></tt>" - Numbered convention</b>:</dt>
687 <dd>Any calling convention may be specified by number, allowing
688 target-specific calling conventions to be used. Target specific calling
689 conventions start at 64.</dd>
692 <p>More calling conventions can be added/defined on an as-needed basis, to
693 support Pascal conventions or any other well-known target-independent
698 <!-- ======================================================================= -->
699 <div class="doc_subsection">
700 <a name="visibility">Visibility Styles</a>
703 <div class="doc_text">
705 <p>All Global Variables and Functions have one of the following visibility
709 <dt><b>"<tt>default</tt>" - Default style</b>:</dt>
710 <dd>On targets that use the ELF object file format, default visibility means
711 that the declaration is visible to other modules and, in shared libraries,
712 means that the declared entity may be overridden. On Darwin, default
713 visibility means that the declaration is visible to other modules. Default
714 visibility corresponds to "external linkage" in the language.</dd>
716 <dt><b>"<tt>hidden</tt>" - Hidden style</b>:</dt>
717 <dd>Two declarations of an object with hidden visibility refer to the same
718 object if they are in the same shared object. Usually, hidden visibility
719 indicates that the symbol will not be placed into the dynamic symbol
720 table, so no other module (executable or shared library) can reference it
723 <dt><b>"<tt>protected</tt>" - Protected style</b>:</dt>
724 <dd>On ELF, protected visibility indicates that the symbol will be placed in
725 the dynamic symbol table, but that references within the defining module
726 will bind to the local symbol. That is, the symbol cannot be overridden by
732 <!-- ======================================================================= -->
733 <div class="doc_subsection">
734 <a name="namedtypes">Named Types</a>
737 <div class="doc_text">
739 <p>LLVM IR allows you to specify name aliases for certain types. This can make
740 it easier to read the IR and make the IR more condensed (particularly when
741 recursive types are involved). An example of a name specification is:</p>
743 <div class="doc_code">
745 %mytype = type { %mytype*, i32 }
749 <p>You may give a name to any <a href="#typesystem">type</a> except
750 "<a href="t_void">void</a>". Type name aliases may be used anywhere a type
751 is expected with the syntax "%mytype".</p>
753 <p>Note that type names are aliases for the structural type that they indicate,
754 and that you can therefore specify multiple names for the same type. This
755 often leads to confusing behavior when dumping out a .ll file. Since LLVM IR
756 uses structural typing, the name is not part of the type. When printing out
757 LLVM IR, the printer will pick <em>one name</em> to render all types of a
758 particular shape. This means that if you have code where two different
759 source types end up having the same LLVM type, that the dumper will sometimes
760 print the "wrong" or unexpected type. This is an important design point and
761 isn't going to change.</p>
765 <!-- ======================================================================= -->
766 <div class="doc_subsection">
767 <a name="globalvars">Global Variables</a>
770 <div class="doc_text">
772 <p>Global variables define regions of memory allocated at compilation time
773 instead of run-time. Global variables may optionally be initialized, may
774 have an explicit section to be placed in, and may have an optional explicit
775 alignment specified. A variable may be defined as "thread_local", which
776 means that it will not be shared by threads (each thread will have a
777 separated copy of the variable). A variable may be defined as a global
778 "constant," which indicates that the contents of the variable
779 will <b>never</b> be modified (enabling better optimization, allowing the
780 global data to be placed in the read-only section of an executable, etc).
781 Note that variables that need runtime initialization cannot be marked
782 "constant" as there is a store to the variable.</p>
784 <p>LLVM explicitly allows <em>declarations</em> of global variables to be marked
785 constant, even if the final definition of the global is not. This capability
786 can be used to enable slightly better optimization of the program, but
787 requires the language definition to guarantee that optimizations based on the
788 'constantness' are valid for the translation units that do not include the
791 <p>As SSA values, global variables define pointer values that are in scope
792 (i.e. they dominate) all basic blocks in the program. Global variables
793 always define a pointer to their "content" type because they describe a
794 region of memory, and all memory objects in LLVM are accessed through
797 <p>A global variable may be declared to reside in a target-specific numbered
798 address space. For targets that support them, address spaces may affect how
799 optimizations are performed and/or what target instructions are used to
800 access the variable. The default address space is zero. The address space
801 qualifier must precede any other attributes.</p>
803 <p>LLVM allows an explicit section to be specified for globals. If the target
804 supports it, it will emit globals to the section specified.</p>
806 <p>An explicit alignment may be specified for a global. If not present, or if
807 the alignment is set to zero, the alignment of the global is set by the
808 target to whatever it feels convenient. If an explicit alignment is
809 specified, the global is forced to have at least that much alignment. All
810 alignments must be a power of 2.</p>
812 <p>For example, the following defines a global in a numbered address space with
813 an initializer, section, and alignment:</p>
815 <div class="doc_code">
817 @G = addrspace(5) constant float 1.0, section "foo", align 4
824 <!-- ======================================================================= -->
825 <div class="doc_subsection">
826 <a name="functionstructure">Functions</a>
829 <div class="doc_text">
831 <p>LLVM function definitions consist of the "<tt>define</tt>" keyord, an
832 optional <a href="#linkage">linkage type</a>, an optional
833 <a href="#visibility">visibility style</a>, an optional
834 <a href="#callingconv">calling convention</a>, a return type, an optional
835 <a href="#paramattrs">parameter attribute</a> for the return type, a function
836 name, a (possibly empty) argument list (each with optional
837 <a href="#paramattrs">parameter attributes</a>), optional
838 <a href="#fnattrs">function attributes</a>, an optional section, an optional
839 alignment, an optional <a href="#gc">garbage collector name</a>, an opening
840 curly brace, a list of basic blocks, and a closing curly brace.</p>
842 <p>LLVM function declarations consist of the "<tt>declare</tt>" keyword, an
843 optional <a href="#linkage">linkage type</a>, an optional
844 <a href="#visibility">visibility style</a>, an optional
845 <a href="#callingconv">calling convention</a>, a return type, an optional
846 <a href="#paramattrs">parameter attribute</a> for the return type, a function
847 name, a possibly empty list of arguments, an optional alignment, and an
848 optional <a href="#gc">garbage collector name</a>.</p>
850 <p>A function definition contains a list of basic blocks, forming the CFG
851 (Control Flow Graph) for the function. Each basic block may optionally start
852 with a label (giving the basic block a symbol table entry), contains a list
853 of instructions, and ends with a <a href="#terminators">terminator</a>
854 instruction (such as a branch or function return).</p>
856 <p>The first basic block in a function is special in two ways: it is immediately
857 executed on entrance to the function, and it is not allowed to have
858 predecessor basic blocks (i.e. there can not be any branches to the entry
859 block of a function). Because the block can have no predecessors, it also
860 cannot have any <a href="#i_phi">PHI nodes</a>.</p>
862 <p>LLVM allows an explicit section to be specified for functions. If the target
863 supports it, it will emit functions to the section specified.</p>
865 <p>An explicit alignment may be specified for a function. If not present, or if
866 the alignment is set to zero, the alignment of the function is set by the
867 target to whatever it feels convenient. If an explicit alignment is
868 specified, the function is forced to have at least that much alignment. All
869 alignments must be a power of 2.</p>
872 <div class="doc_code">
874 define [<a href="#linkage">linkage</a>] [<a href="#visibility">visibility</a>]
875 [<a href="#callingconv">cconv</a>] [<a href="#paramattrs">ret attrs</a>]
876 <ResultType> @<FunctionName> ([argument list])
877 [<a href="#fnattrs">fn Attrs</a>] [section "name"] [align N]
878 [<a href="#gc">gc</a>] { ... }
884 <!-- ======================================================================= -->
885 <div class="doc_subsection">
886 <a name="aliasstructure">Aliases</a>
889 <div class="doc_text">
891 <p>Aliases act as "second name" for the aliasee value (which can be either
892 function, global variable, another alias or bitcast of global value). Aliases
893 may have an optional <a href="#linkage">linkage type</a>, and an
894 optional <a href="#visibility">visibility style</a>.</p>
897 <div class="doc_code">
899 @<Name> = alias [Linkage] [Visibility] <AliaseeTy> @<Aliasee>
905 <!-- ======================================================================= -->
906 <div class="doc_subsection"><a name="paramattrs">Parameter Attributes</a></div>
908 <div class="doc_text">
910 <p>The return type and each parameter of a function type may have a set of
911 <i>parameter attributes</i> associated with them. Parameter attributes are
912 used to communicate additional information about the result or parameters of
913 a function. Parameter attributes are considered to be part of the function,
914 not of the function type, so functions with different parameter attributes
915 can have the same function type.</p>
917 <p>Parameter attributes are simple keywords that follow the type specified. If
918 multiple parameter attributes are needed, they are space separated. For
921 <div class="doc_code">
923 declare i32 @printf(i8* noalias nocapture, ...)
924 declare i32 @atoi(i8 zeroext)
925 declare signext i8 @returns_signed_char()
929 <p>Note that any attributes for the function result (<tt>nounwind</tt>,
930 <tt>readonly</tt>) come immediately after the argument list.</p>
932 <p>Currently, only the following parameter attributes are defined:</p>
935 <dt><tt><b>zeroext</b></tt></dt>
936 <dd>This indicates to the code generator that the parameter or return value
937 should be zero-extended to a 32-bit value by the caller (for a parameter)
938 or the callee (for a return value).</dd>
940 <dt><tt><b>signext</b></tt></dt>
941 <dd>This indicates to the code generator that the parameter or return value
942 should be sign-extended to a 32-bit value by the caller (for a parameter)
943 or the callee (for a return value).</dd>
945 <dt><tt><b>inreg</b></tt></dt>
946 <dd>This indicates that this parameter or return value should be treated in a
947 special target-dependent fashion during while emitting code for a function
948 call or return (usually, by putting it in a register as opposed to memory,
949 though some targets use it to distinguish between two different kinds of
950 registers). Use of this attribute is target-specific.</dd>
952 <dt><tt><b><a name="byval">byval</a></b></tt></dt>
953 <dd>This indicates that the pointer parameter should really be passed by value
954 to the function. The attribute implies that a hidden copy of the pointee
955 is made between the caller and the callee, so the callee is unable to
956 modify the value in the callee. This attribute is only valid on LLVM
957 pointer arguments. It is generally used to pass structs and arrays by
958 value, but is also valid on pointers to scalars. The copy is considered
959 to belong to the caller not the callee (for example,
960 <tt><a href="#readonly">readonly</a></tt> functions should not write to
961 <tt>byval</tt> parameters). This is not a valid attribute for return
962 values. The byval attribute also supports specifying an alignment with
963 the align attribute. This has a target-specific effect on the code
964 generator that usually indicates a desired alignment for the synthesized
967 <dt><tt><b>sret</b></tt></dt>
968 <dd>This indicates that the pointer parameter specifies the address of a
969 structure that is the return value of the function in the source program.
970 This pointer must be guaranteed by the caller to be valid: loads and
971 stores to the structure may be assumed by the callee to not to trap. This
972 may only be applied to the first parameter. This is not a valid attribute
973 for return values. </dd>
975 <dt><tt><b>noalias</b></tt></dt>
976 <dd>This indicates that the pointer does not alias any global or any other
977 parameter. The caller is responsible for ensuring that this is the
978 case. On a function return value, <tt>noalias</tt> additionally indicates
979 that the pointer does not alias any other pointers visible to the
980 caller. For further details, please see the discussion of the NoAlias
982 <a href="http://llvm.org/docs/AliasAnalysis.html#MustMayNo">alias
985 <dt><tt><b>nocapture</b></tt></dt>
986 <dd>This indicates that the callee does not make any copies of the pointer
987 that outlive the callee itself. This is not a valid attribute for return
990 <dt><tt><b>nest</b></tt></dt>
991 <dd>This indicates that the pointer parameter can be excised using the
992 <a href="#int_trampoline">trampoline intrinsics</a>. This is not a valid
993 attribute for return values.</dd>
998 <!-- ======================================================================= -->
999 <div class="doc_subsection">
1000 <a name="gc">Garbage Collector Names</a>
1003 <div class="doc_text">
1005 <p>Each function may specify a garbage collector name, which is simply a
1008 <div class="doc_code">
1010 define void @f() gc "name" { ... }
1014 <p>The compiler declares the supported values of <i>name</i>. Specifying a
1015 collector which will cause the compiler to alter its output in order to
1016 support the named garbage collection algorithm.</p>
1020 <!-- ======================================================================= -->
1021 <div class="doc_subsection">
1022 <a name="fnattrs">Function Attributes</a>
1025 <div class="doc_text">
1027 <p>Function attributes are set to communicate additional information about a
1028 function. Function attributes are considered to be part of the function, not
1029 of the function type, so functions with different parameter attributes can
1030 have the same function type.</p>
1032 <p>Function attributes are simple keywords that follow the type specified. If
1033 multiple attributes are needed, they are space separated. For example:</p>
1035 <div class="doc_code">
1037 define void @f() noinline { ... }
1038 define void @f() alwaysinline { ... }
1039 define void @f() alwaysinline optsize { ... }
1040 define void @f() optsize { ... }
1045 <dt><tt><b>alwaysinline</b></tt></dt>
1046 <dd>This attribute indicates that the inliner should attempt to inline this
1047 function into callers whenever possible, ignoring any active inlining size
1048 threshold for this caller.</dd>
1050 <dt><tt><b>inlinehint</b></tt></dt>
1051 <dd>This attribute indicates that the source code contained a hint that inlining
1052 this function is desirable (such as the "inline" keyword in C/C++). It
1053 is just a hint; it imposes no requirements on the inliner.</dd>
1055 <dt><tt><b>noinline</b></tt></dt>
1056 <dd>This attribute indicates that the inliner should never inline this
1057 function in any situation. This attribute may not be used together with
1058 the <tt>alwaysinline</tt> attribute.</dd>
1060 <dt><tt><b>optsize</b></tt></dt>
1061 <dd>This attribute suggests that optimization passes and code generator passes
1062 make choices that keep the code size of this function low, and otherwise
1063 do optimizations specifically to reduce code size.</dd>
1065 <dt><tt><b>noreturn</b></tt></dt>
1066 <dd>This function attribute indicates that the function never returns
1067 normally. This produces undefined behavior at runtime if the function
1068 ever does dynamically return.</dd>
1070 <dt><tt><b>nounwind</b></tt></dt>
1071 <dd>This function attribute indicates that the function never returns with an
1072 unwind or exceptional control flow. If the function does unwind, its
1073 runtime behavior is undefined.</dd>
1075 <dt><tt><b>readnone</b></tt></dt>
1076 <dd>This attribute indicates that the function computes its result (or decides
1077 to unwind an exception) based strictly on its arguments, without
1078 dereferencing any pointer arguments or otherwise accessing any mutable
1079 state (e.g. memory, control registers, etc) visible to caller functions.
1080 It does not write through any pointer arguments
1081 (including <tt><a href="#byval">byval</a></tt> arguments) and never
1082 changes any state visible to callers. This means that it cannot unwind
1083 exceptions by calling the <tt>C++</tt> exception throwing methods, but
1084 could use the <tt>unwind</tt> instruction.</dd>
1086 <dt><tt><b><a name="readonly">readonly</a></b></tt></dt>
1087 <dd>This attribute indicates that the function does not write through any
1088 pointer arguments (including <tt><a href="#byval">byval</a></tt>
1089 arguments) or otherwise modify any state (e.g. memory, control registers,
1090 etc) visible to caller functions. It may dereference pointer arguments
1091 and read state that may be set in the caller. A readonly function always
1092 returns the same value (or unwinds an exception identically) when called
1093 with the same set of arguments and global state. It cannot unwind an
1094 exception by calling the <tt>C++</tt> exception throwing methods, but may
1095 use the <tt>unwind</tt> instruction.</dd>
1097 <dt><tt><b><a name="ssp">ssp</a></b></tt></dt>
1098 <dd>This attribute indicates that the function should emit a stack smashing
1099 protector. It is in the form of a "canary"—a random value placed on
1100 the stack before the local variables that's checked upon return from the
1101 function to see if it has been overwritten. A heuristic is used to
1102 determine if a function needs stack protectors or not.<br>
1104 If a function that has an <tt>ssp</tt> attribute is inlined into a
1105 function that doesn't have an <tt>ssp</tt> attribute, then the resulting
1106 function will have an <tt>ssp</tt> attribute.</dd>
1108 <dt><tt><b>sspreq</b></tt></dt>
1109 <dd>This attribute indicates that the function should <em>always</em> emit a
1110 stack smashing protector. This overrides
1111 the <tt><a href="#ssp">ssp</a></tt> function attribute.<br>
1113 If a function that has an <tt>sspreq</tt> attribute is inlined into a
1114 function that doesn't have an <tt>sspreq</tt> attribute or which has
1115 an <tt>ssp</tt> attribute, then the resulting function will have
1116 an <tt>sspreq</tt> attribute.</dd>
1118 <dt><tt><b>noredzone</b></tt></dt>
1119 <dd>This attribute indicates that the code generator should not use a red
1120 zone, even if the target-specific ABI normally permits it.</dd>
1122 <dt><tt><b>noimplicitfloat</b></tt></dt>
1123 <dd>This attributes disables implicit floating point instructions.</dd>
1125 <dt><tt><b>naked</b></tt></dt>
1126 <dd>This attribute disables prologue / epilogue emission for the function.
1127 This can have very system-specific consequences.</dd>
1132 <!-- ======================================================================= -->
1133 <div class="doc_subsection">
1134 <a name="moduleasm">Module-Level Inline Assembly</a>
1137 <div class="doc_text">
1139 <p>Modules may contain "module-level inline asm" blocks, which corresponds to
1140 the GCC "file scope inline asm" blocks. These blocks are internally
1141 concatenated by LLVM and treated as a single unit, but may be separated in
1142 the <tt>.ll</tt> file if desired. The syntax is very simple:</p>
1144 <div class="doc_code">
1146 module asm "inline asm code goes here"
1147 module asm "more can go here"
1151 <p>The strings can contain any character by escaping non-printable characters.
1152 The escape sequence used is simply "\xx" where "xx" is the two digit hex code
1155 <p>The inline asm code is simply printed to the machine code .s file when
1156 assembly code is generated.</p>
1160 <!-- ======================================================================= -->
1161 <div class="doc_subsection">
1162 <a name="datalayout">Data Layout</a>
1165 <div class="doc_text">
1167 <p>A module may specify a target specific data layout string that specifies how
1168 data is to be laid out in memory. The syntax for the data layout is
1171 <div class="doc_code">
1173 target datalayout = "<i>layout specification</i>"
1177 <p>The <i>layout specification</i> consists of a list of specifications
1178 separated by the minus sign character ('-'). Each specification starts with
1179 a letter and may include other information after the letter to define some
1180 aspect of the data layout. The specifications accepted are as follows:</p>
1184 <dd>Specifies that the target lays out data in big-endian form. That is, the
1185 bits with the most significance have the lowest address location.</dd>
1188 <dd>Specifies that the target lays out data in little-endian form. That is,
1189 the bits with the least significance have the lowest address
1192 <dt><tt>p:<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1193 <dd>This specifies the <i>size</i> of a pointer and its <i>abi</i> and
1194 <i>preferred</i> alignments. All sizes are in bits. Specifying
1195 the <i>pref</i> alignment is optional. If omitted, the
1196 preceding <tt>:</tt> should be omitted too.</dd>
1198 <dt><tt>i<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1199 <dd>This specifies the alignment for an integer type of a given bit
1200 <i>size</i>. The value of <i>size</i> must be in the range [1,2^23).</dd>
1202 <dt><tt>v<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1203 <dd>This specifies the alignment for a vector type of a given bit
1206 <dt><tt>f<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1207 <dd>This specifies the alignment for a floating point type of a given bit
1208 <i>size</i>. The value of <i>size</i> must be either 32 (float) or 64
1211 <dt><tt>a<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1212 <dd>This specifies the alignment for an aggregate type of a given bit
1215 <dt><tt>s<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1216 <dd>This specifies the alignment for a stack object of a given bit
1219 <dt><tt>n<i>size1</i>:<i>size2</i>:<i>size3</i>...</tt></dt>
1220 <dd>This specifies a set of native integer widths for the target CPU
1221 in bits. For example, it might contain "n32" for 32-bit PowerPC,
1222 "n32:64" for PowerPC 64, or "n8:16:32:64" for X86-64. Elements of
1223 this set are considered to support most general arithmetic
1224 operations efficiently.</dd>
1227 <p>When constructing the data layout for a given target, LLVM starts with a
1228 default set of specifications which are then (possibly) overriden by the
1229 specifications in the <tt>datalayout</tt> keyword. The default specifications
1230 are given in this list:</p>
1233 <li><tt>E</tt> - big endian</li>
1234 <li><tt>p:32:64:64</tt> - 32-bit pointers with 64-bit alignment</li>
1235 <li><tt>i1:8:8</tt> - i1 is 8-bit (byte) aligned</li>
1236 <li><tt>i8:8:8</tt> - i8 is 8-bit (byte) aligned</li>
1237 <li><tt>i16:16:16</tt> - i16 is 16-bit aligned</li>
1238 <li><tt>i32:32:32</tt> - i32 is 32-bit aligned</li>
1239 <li><tt>i64:32:64</tt> - i64 has ABI alignment of 32-bits but preferred
1240 alignment of 64-bits</li>
1241 <li><tt>f32:32:32</tt> - float is 32-bit aligned</li>
1242 <li><tt>f64:64:64</tt> - double is 64-bit aligned</li>
1243 <li><tt>v64:64:64</tt> - 64-bit vector is 64-bit aligned</li>
1244 <li><tt>v128:128:128</tt> - 128-bit vector is 128-bit aligned</li>
1245 <li><tt>a0:0:1</tt> - aggregates are 8-bit aligned</li>
1246 <li><tt>s0:64:64</tt> - stack objects are 64-bit aligned</li>
1249 <p>When LLVM is determining the alignment for a given type, it uses the
1250 following rules:</p>
1253 <li>If the type sought is an exact match for one of the specifications, that
1254 specification is used.</li>
1256 <li>If no match is found, and the type sought is an integer type, then the
1257 smallest integer type that is larger than the bitwidth of the sought type
1258 is used. If none of the specifications are larger than the bitwidth then
1259 the the largest integer type is used. For example, given the default
1260 specifications above, the i7 type will use the alignment of i8 (next
1261 largest) while both i65 and i256 will use the alignment of i64 (largest
1264 <li>If no match is found, and the type sought is a vector type, then the
1265 largest vector type that is smaller than the sought vector type will be
1266 used as a fall back. This happens because <128 x double> can be
1267 implemented in terms of 64 <2 x double>, for example.</li>
1272 <!-- ======================================================================= -->
1273 <div class="doc_subsection">
1274 <a name="pointeraliasing">Pointer Aliasing Rules</a>
1277 <div class="doc_text">
1279 <p>Any memory access must be done through a pointer value associated
1280 with an address range of the memory access, otherwise the behavior
1281 is undefined. Pointer values are associated with address ranges
1282 according to the following rules:</p>
1285 <li>A pointer value formed from a
1286 <tt><a href="#i_getelementptr">getelementptr</a></tt> instruction
1287 is associated with the addresses associated with the first operand
1288 of the <tt>getelementptr</tt>.</li>
1289 <li>An address of a global variable is associated with the address
1290 range of the variable's storage.</li>
1291 <li>The result value of an allocation instruction is associated with
1292 the address range of the allocated storage.</li>
1293 <li>A null pointer in the default address-space is associated with
1295 <li>A pointer value formed by an
1296 <tt><a href="#i_inttoptr">inttoptr</a></tt> is associated with all
1297 address ranges of all pointer values that contribute (directly or
1298 indirectly) to the computation of the pointer's value.</li>
1299 <li>The result value of a
1300 <tt><a href="#i_bitcast">bitcast</a></tt> is associated with all
1301 addresses associated with the operand of the <tt>bitcast</tt>.</li>
1302 <li>An integer constant other than zero or a pointer value returned
1303 from a function not defined within LLVM may be associated with address
1304 ranges allocated through mechanisms other than those provided by
1305 LLVM. Such ranges shall not overlap with any ranges of addresses
1306 allocated by mechanisms provided by LLVM.</li>
1309 <p>LLVM IR does not associate types with memory. The result type of a
1310 <tt><a href="#i_load">load</a></tt> merely indicates the size and
1311 alignment of the memory from which to load, as well as the
1312 interpretation of the value. The first operand of a
1313 <tt><a href="#i_store">store</a></tt> similarly only indicates the size
1314 and alignment of the store.</p>
1316 <p>Consequently, type-based alias analysis, aka TBAA, aka
1317 <tt>-fstrict-aliasing</tt>, is not applicable to general unadorned
1318 LLVM IR. <a href="#metadata">Metadata</a> may be used to encode
1319 additional information which specialized optimization passes may use
1320 to implement type-based alias analysis.</p>
1324 <!-- *********************************************************************** -->
1325 <div class="doc_section"> <a name="typesystem">Type System</a> </div>
1326 <!-- *********************************************************************** -->
1328 <div class="doc_text">
1330 <p>The LLVM type system is one of the most important features of the
1331 intermediate representation. Being typed enables a number of optimizations
1332 to be performed on the intermediate representation directly, without having
1333 to do extra analyses on the side before the transformation. A strong type
1334 system makes it easier to read the generated code and enables novel analyses
1335 and transformations that are not feasible to perform on normal three address
1336 code representations.</p>
1340 <!-- ======================================================================= -->
1341 <div class="doc_subsection"> <a name="t_classifications">Type
1342 Classifications</a> </div>
1344 <div class="doc_text">
1346 <p>The types fall into a few useful classifications:</p>
1348 <table border="1" cellspacing="0" cellpadding="4">
1350 <tr><th>Classification</th><th>Types</th></tr>
1352 <td><a href="#t_integer">integer</a></td>
1353 <td><tt>i1, i2, i3, ... i8, ... i16, ... i32, ... i64, ... </tt></td>
1356 <td><a href="#t_floating">floating point</a></td>
1357 <td><tt>float, double, x86_fp80, fp128, ppc_fp128</tt></td>
1360 <td><a name="t_firstclass">first class</a></td>
1361 <td><a href="#t_integer">integer</a>,
1362 <a href="#t_floating">floating point</a>,
1363 <a href="#t_pointer">pointer</a>,
1364 <a href="#t_vector">vector</a>,
1365 <a href="#t_struct">structure</a>,
1366 <a href="#t_array">array</a>,
1367 <a href="#t_label">label</a>,
1368 <a href="#t_metadata">metadata</a>.
1372 <td><a href="#t_primitive">primitive</a></td>
1373 <td><a href="#t_label">label</a>,
1374 <a href="#t_void">void</a>,
1375 <a href="#t_floating">floating point</a>,
1376 <a href="#t_metadata">metadata</a>.</td>
1379 <td><a href="#t_derived">derived</a></td>
1380 <td><a href="#t_integer">integer</a>,
1381 <a href="#t_array">array</a>,
1382 <a href="#t_function">function</a>,
1383 <a href="#t_pointer">pointer</a>,
1384 <a href="#t_struct">structure</a>,
1385 <a href="#t_pstruct">packed structure</a>,
1386 <a href="#t_vector">vector</a>,
1387 <a href="#t_opaque">opaque</a>.
1393 <p>The <a href="#t_firstclass">first class</a> types are perhaps the most
1394 important. Values of these types are the only ones which can be produced by
1399 <!-- ======================================================================= -->
1400 <div class="doc_subsection"> <a name="t_primitive">Primitive Types</a> </div>
1402 <div class="doc_text">
1404 <p>The primitive types are the fundamental building blocks of the LLVM
1409 <!-- _______________________________________________________________________ -->
1410 <div class="doc_subsubsection"> <a name="t_integer">Integer Type</a> </div>
1412 <div class="doc_text">
1415 <p>The integer type is a very simple type that simply specifies an arbitrary
1416 bit width for the integer type desired. Any bit width from 1 bit to
1417 2<sup>23</sup>-1 (about 8 million) can be specified.</p>
1424 <p>The number of bits the integer will occupy is specified by the <tt>N</tt>
1428 <table class="layout">
1430 <td class="left"><tt>i1</tt></td>
1431 <td class="left">a single-bit integer.</td>
1434 <td class="left"><tt>i32</tt></td>
1435 <td class="left">a 32-bit integer.</td>
1438 <td class="left"><tt>i1942652</tt></td>
1439 <td class="left">a really big integer of over 1 million bits.</td>
1445 <!-- _______________________________________________________________________ -->
1446 <div class="doc_subsubsection"> <a name="t_floating">Floating Point Types</a> </div>
1448 <div class="doc_text">
1452 <tr><th>Type</th><th>Description</th></tr>
1453 <tr><td><tt>float</tt></td><td>32-bit floating point value</td></tr>
1454 <tr><td><tt>double</tt></td><td>64-bit floating point value</td></tr>
1455 <tr><td><tt>fp128</tt></td><td>128-bit floating point value (112-bit mantissa)</td></tr>
1456 <tr><td><tt>x86_fp80</tt></td><td>80-bit floating point value (X87)</td></tr>
1457 <tr><td><tt>ppc_fp128</tt></td><td>128-bit floating point value (two 64-bits)</td></tr>
1463 <!-- _______________________________________________________________________ -->
1464 <div class="doc_subsubsection"> <a name="t_void">Void Type</a> </div>
1466 <div class="doc_text">
1469 <p>The void type does not represent any value and has no size.</p>
1478 <!-- _______________________________________________________________________ -->
1479 <div class="doc_subsubsection"> <a name="t_label">Label Type</a> </div>
1481 <div class="doc_text">
1484 <p>The label type represents code labels.</p>
1493 <!-- _______________________________________________________________________ -->
1494 <div class="doc_subsubsection"> <a name="t_metadata">Metadata Type</a> </div>
1496 <div class="doc_text">
1499 <p>The metadata type represents embedded metadata. No derived types may be
1500 created from metadata except for <a href="#t_function">function</a>
1511 <!-- ======================================================================= -->
1512 <div class="doc_subsection"> <a name="t_derived">Derived Types</a> </div>
1514 <div class="doc_text">
1516 <p>The real power in LLVM comes from the derived types in the system. This is
1517 what allows a programmer to represent arrays, functions, pointers, and other
1518 useful types. Each of these types contain one or more element types which
1519 may be a primitive type, or another derived type. For example, it is
1520 possible to have a two dimensional array, using an array as the element type
1521 of another array.</p>
1525 <!-- _______________________________________________________________________ -->
1526 <div class="doc_subsubsection"> <a name="t_array">Array Type</a> </div>
1528 <div class="doc_text">
1531 <p>The array type is a very simple derived type that arranges elements
1532 sequentially in memory. The array type requires a size (number of elements)
1533 and an underlying data type.</p>
1537 [<# elements> x <elementtype>]
1540 <p>The number of elements is a constant integer value; <tt>elementtype</tt> may
1541 be any type with a size.</p>
1544 <table class="layout">
1546 <td class="left"><tt>[40 x i32]</tt></td>
1547 <td class="left">Array of 40 32-bit integer values.</td>
1550 <td class="left"><tt>[41 x i32]</tt></td>
1551 <td class="left">Array of 41 32-bit integer values.</td>
1554 <td class="left"><tt>[4 x i8]</tt></td>
1555 <td class="left">Array of 4 8-bit integer values.</td>
1558 <p>Here are some examples of multidimensional arrays:</p>
1559 <table class="layout">
1561 <td class="left"><tt>[3 x [4 x i32]]</tt></td>
1562 <td class="left">3x4 array of 32-bit integer values.</td>
1565 <td class="left"><tt>[12 x [10 x float]]</tt></td>
1566 <td class="left">12x10 array of single precision floating point values.</td>
1569 <td class="left"><tt>[2 x [3 x [4 x i16]]]</tt></td>
1570 <td class="left">2x3x4 array of 16-bit integer values.</td>
1574 <p>There is no restriction on indexing beyond the end of the array implied by
1575 a static type (though there are restrictions on indexing beyond the bounds
1576 of an allocated object in some cases). This means that single-dimension
1577 'variable sized array' addressing can be implemented in LLVM with a zero
1578 length array type. An implementation of 'pascal style arrays' in LLVM could
1579 use the type "<tt>{ i32, [0 x float]}</tt>", for example.</p>
1583 <!-- _______________________________________________________________________ -->
1584 <div class="doc_subsubsection"> <a name="t_function">Function Type</a> </div>
1586 <div class="doc_text">
1589 <p>The function type can be thought of as a function signature. It consists of
1590 a return type and a list of formal parameter types. The return type of a
1591 function type is a scalar type, a void type, or a struct type. If the return
1592 type is a struct type then all struct elements must be of first class types,
1593 and the struct must have at least one element.</p>
1597 <returntype> (<parameter list>)
1600 <p>...where '<tt><parameter list></tt>' is a comma-separated list of type
1601 specifiers. Optionally, the parameter list may include a type <tt>...</tt>,
1602 which indicates that the function takes a variable number of arguments.
1603 Variable argument functions can access their arguments with
1604 the <a href="#int_varargs">variable argument handling intrinsic</a>
1605 functions. '<tt><returntype></tt>' is a any type except
1606 <a href="#t_label">label</a>.</p>
1609 <table class="layout">
1611 <td class="left"><tt>i32 (i32)</tt></td>
1612 <td class="left">function taking an <tt>i32</tt>, returning an <tt>i32</tt>
1614 </tr><tr class="layout">
1615 <td class="left"><tt>float (i16 signext, i32 *) *
1617 <td class="left"><a href="#t_pointer">Pointer</a> to a function that takes
1618 an <tt>i16</tt> that should be sign extended and a
1619 <a href="#t_pointer">pointer</a> to <tt>i32</tt>, returning
1622 </tr><tr class="layout">
1623 <td class="left"><tt>i32 (i8*, ...)</tt></td>
1624 <td class="left">A vararg function that takes at least one
1625 <a href="#t_pointer">pointer</a> to <tt>i8 </tt> (char in C),
1626 which returns an integer. This is the signature for <tt>printf</tt> in
1629 </tr><tr class="layout">
1630 <td class="left"><tt>{i32, i32} (i32)</tt></td>
1631 <td class="left">A function taking an <tt>i32</tt>, returning a
1632 <a href="#t_struct">structure</a> containing two <tt>i32</tt> values
1639 <!-- _______________________________________________________________________ -->
1640 <div class="doc_subsubsection"> <a name="t_struct">Structure Type</a> </div>
1642 <div class="doc_text">
1645 <p>The structure type is used to represent a collection of data members together
1646 in memory. The packing of the field types is defined to match the ABI of the
1647 underlying processor. The elements of a structure may be any type that has a
1650 <p>Structures are accessed using '<tt><a href="#i_load">load</a></tt> and
1651 '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a field with
1652 the '<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.</p>
1656 { <type list> }
1660 <table class="layout">
1662 <td class="left"><tt>{ i32, i32, i32 }</tt></td>
1663 <td class="left">A triple of three <tt>i32</tt> values</td>
1664 </tr><tr class="layout">
1665 <td class="left"><tt>{ float, i32 (i32) * }</tt></td>
1666 <td class="left">A pair, where the first element is a <tt>float</tt> and the
1667 second element is a <a href="#t_pointer">pointer</a> to a
1668 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
1669 an <tt>i32</tt>.</td>
1675 <!-- _______________________________________________________________________ -->
1676 <div class="doc_subsubsection"> <a name="t_pstruct">Packed Structure Type</a>
1679 <div class="doc_text">
1682 <p>The packed structure type is used to represent a collection of data members
1683 together in memory. There is no padding between fields. Further, the
1684 alignment of a packed structure is 1 byte. The elements of a packed
1685 structure may be any type that has a size.</p>
1687 <p>Structures are accessed using '<tt><a href="#i_load">load</a></tt> and
1688 '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a field with
1689 the '<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.</p>
1693 < { <type list> } >
1697 <table class="layout">
1699 <td class="left"><tt>< { i32, i32, i32 } ></tt></td>
1700 <td class="left">A triple of three <tt>i32</tt> values</td>
1701 </tr><tr class="layout">
1703 <tt>< { float, i32 (i32)* } ></tt></td>
1704 <td class="left">A pair, where the first element is a <tt>float</tt> and the
1705 second element is a <a href="#t_pointer">pointer</a> to a
1706 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
1707 an <tt>i32</tt>.</td>
1713 <!-- _______________________________________________________________________ -->
1714 <div class="doc_subsubsection"> <a name="t_pointer">Pointer Type</a> </div>
1716 <div class="doc_text">
1719 <p>As in many languages, the pointer type represents a pointer or reference to
1720 another object, which must live in memory. Pointer types may have an optional
1721 address space attribute defining the target-specific numbered address space
1722 where the pointed-to object resides. The default address space is zero.</p>
1724 <p>Note that LLVM does not permit pointers to void (<tt>void*</tt>) nor does it
1725 permit pointers to labels (<tt>label*</tt>). Use <tt>i8*</tt> instead.</p>
1733 <table class="layout">
1735 <td class="left"><tt>[4 x i32]*</tt></td>
1736 <td class="left">A <a href="#t_pointer">pointer</a> to <a
1737 href="#t_array">array</a> of four <tt>i32</tt> values.</td>
1740 <td class="left"><tt>i32 (i32 *) *</tt></td>
1741 <td class="left"> A <a href="#t_pointer">pointer</a> to a <a
1742 href="#t_function">function</a> that takes an <tt>i32*</tt>, returning an
1746 <td class="left"><tt>i32 addrspace(5)*</tt></td>
1747 <td class="left">A <a href="#t_pointer">pointer</a> to an <tt>i32</tt> value
1748 that resides in address space #5.</td>
1754 <!-- _______________________________________________________________________ -->
1755 <div class="doc_subsubsection"> <a name="t_vector">Vector Type</a> </div>
1757 <div class="doc_text">
1760 <p>A vector type is a simple derived type that represents a vector of elements.
1761 Vector types are used when multiple primitive data are operated in parallel
1762 using a single instruction (SIMD). A vector type requires a size (number of
1763 elements) and an underlying primitive data type. Vectors must have a power
1764 of two length (1, 2, 4, 8, 16 ...). Vector types are considered
1765 <a href="#t_firstclass">first class</a>.</p>
1769 < <# elements> x <elementtype> >
1772 <p>The number of elements is a constant integer value; elementtype may be any
1773 integer or floating point type.</p>
1776 <table class="layout">
1778 <td class="left"><tt><4 x i32></tt></td>
1779 <td class="left">Vector of 4 32-bit integer values.</td>
1782 <td class="left"><tt><8 x float></tt></td>
1783 <td class="left">Vector of 8 32-bit floating-point values.</td>
1786 <td class="left"><tt><2 x i64></tt></td>
1787 <td class="left">Vector of 2 64-bit integer values.</td>
1793 <!-- _______________________________________________________________________ -->
1794 <div class="doc_subsubsection"> <a name="t_opaque">Opaque Type</a> </div>
1795 <div class="doc_text">
1798 <p>Opaque types are used to represent unknown types in the system. This
1799 corresponds (for example) to the C notion of a forward declared structure
1800 type. In LLVM, opaque types can eventually be resolved to any type (not just
1801 a structure type).</p>
1809 <table class="layout">
1811 <td class="left"><tt>opaque</tt></td>
1812 <td class="left">An opaque type.</td>
1818 <!-- ======================================================================= -->
1819 <div class="doc_subsection">
1820 <a name="t_uprefs">Type Up-references</a>
1823 <div class="doc_text">
1826 <p>An "up reference" allows you to refer to a lexically enclosing type without
1827 requiring it to have a name. For instance, a structure declaration may
1828 contain a pointer to any of the types it is lexically a member of. Example
1829 of up references (with their equivalent as named type declarations)
1833 { \2 * } %x = type { %x* }
1834 { \2 }* %y = type { %y }*
1838 <p>An up reference is needed by the asmprinter for printing out cyclic types
1839 when there is no declared name for a type in the cycle. Because the
1840 asmprinter does not want to print out an infinite type string, it needs a
1841 syntax to handle recursive types that have no names (all names are optional
1849 <p>The level is the count of the lexical type that is being referred to.</p>
1852 <table class="layout">
1854 <td class="left"><tt>\1*</tt></td>
1855 <td class="left">Self-referential pointer.</td>
1858 <td class="left"><tt>{ { \3*, i8 }, i32 }</tt></td>
1859 <td class="left">Recursive structure where the upref refers to the out-most
1866 <!-- *********************************************************************** -->
1867 <div class="doc_section"> <a name="constants">Constants</a> </div>
1868 <!-- *********************************************************************** -->
1870 <div class="doc_text">
1872 <p>LLVM has several different basic types of constants. This section describes
1873 them all and their syntax.</p>
1877 <!-- ======================================================================= -->
1878 <div class="doc_subsection"><a name="simpleconstants">Simple Constants</a></div>
1880 <div class="doc_text">
1883 <dt><b>Boolean constants</b></dt>
1884 <dd>The two strings '<tt>true</tt>' and '<tt>false</tt>' are both valid
1885 constants of the <tt><a href="#t_integer">i1</a></tt> type.</dd>
1887 <dt><b>Integer constants</b></dt>
1888 <dd>Standard integers (such as '4') are constants of
1889 the <a href="#t_integer">integer</a> type. Negative numbers may be used
1890 with integer types.</dd>
1892 <dt><b>Floating point constants</b></dt>
1893 <dd>Floating point constants use standard decimal notation (e.g. 123.421),
1894 exponential notation (e.g. 1.23421e+2), or a more precise hexadecimal
1895 notation (see below). The assembler requires the exact decimal value of a
1896 floating-point constant. For example, the assembler accepts 1.25 but
1897 rejects 1.3 because 1.3 is a repeating decimal in binary. Floating point
1898 constants must have a <a href="#t_floating">floating point</a> type. </dd>
1900 <dt><b>Null pointer constants</b></dt>
1901 <dd>The identifier '<tt>null</tt>' is recognized as a null pointer constant
1902 and must be of <a href="#t_pointer">pointer type</a>.</dd>
1905 <p>The one non-intuitive notation for constants is the hexadecimal form of
1906 floating point constants. For example, the form '<tt>double
1907 0x432ff973cafa8000</tt>' is equivalent to (but harder to read than)
1908 '<tt>double 4.5e+15</tt>'. The only time hexadecimal floating point
1909 constants are required (and the only time that they are generated by the
1910 disassembler) is when a floating point constant must be emitted but it cannot
1911 be represented as a decimal floating point number in a reasonable number of
1912 digits. For example, NaN's, infinities, and other special values are
1913 represented in their IEEE hexadecimal format so that assembly and disassembly
1914 do not cause any bits to change in the constants.</p>
1916 <p>When using the hexadecimal form, constants of types float and double are
1917 represented using the 16-digit form shown above (which matches the IEEE754
1918 representation for double); float values must, however, be exactly
1919 representable as IEE754 single precision. Hexadecimal format is always used
1920 for long double, and there are three forms of long double. The 80-bit format
1921 used by x86 is represented as <tt>0xK</tt> followed by 20 hexadecimal digits.
1922 The 128-bit format used by PowerPC (two adjacent doubles) is represented
1923 by <tt>0xM</tt> followed by 32 hexadecimal digits. The IEEE 128-bit format
1924 is represented by <tt>0xL</tt> followed by 32 hexadecimal digits; no
1925 currently supported target uses this format. Long doubles will only work if
1926 they match the long double format on your target. All hexadecimal formats
1927 are big-endian (sign bit at the left).</p>
1931 <!-- ======================================================================= -->
1932 <div class="doc_subsection">
1933 <a name="aggregateconstants"></a> <!-- old anchor -->
1934 <a name="complexconstants">Complex Constants</a>
1937 <div class="doc_text">
1939 <p>Complex constants are a (potentially recursive) combination of simple
1940 constants and smaller complex constants.</p>
1943 <dt><b>Structure constants</b></dt>
1944 <dd>Structure constants are represented with notation similar to structure
1945 type definitions (a comma separated list of elements, surrounded by braces
1946 (<tt>{}</tt>)). For example: "<tt>{ i32 4, float 17.0, i32* @G }</tt>",
1947 where "<tt>@G</tt>" is declared as "<tt>@G = external global i32</tt>".
1948 Structure constants must have <a href="#t_struct">structure type</a>, and
1949 the number and types of elements must match those specified by the
1952 <dt><b>Array constants</b></dt>
1953 <dd>Array constants are represented with notation similar to array type
1954 definitions (a comma separated list of elements, surrounded by square
1955 brackets (<tt>[]</tt>)). For example: "<tt>[ i32 42, i32 11, i32 74
1956 ]</tt>". Array constants must have <a href="#t_array">array type</a>, and
1957 the number and types of elements must match those specified by the
1960 <dt><b>Vector constants</b></dt>
1961 <dd>Vector constants are represented with notation similar to vector type
1962 definitions (a comma separated list of elements, surrounded by
1963 less-than/greater-than's (<tt><></tt>)). For example: "<tt>< i32
1964 42, i32 11, i32 74, i32 100 ></tt>". Vector constants must
1965 have <a href="#t_vector">vector type</a>, and the number and types of
1966 elements must match those specified by the type.</dd>
1968 <dt><b>Zero initialization</b></dt>
1969 <dd>The string '<tt>zeroinitializer</tt>' can be used to zero initialize a
1970 value to zero of <em>any</em> type, including scalar and aggregate types.
1971 This is often used to avoid having to print large zero initializers
1972 (e.g. for large arrays) and is always exactly equivalent to using explicit
1973 zero initializers.</dd>
1975 <dt><b>Metadata node</b></dt>
1976 <dd>A metadata node is a structure-like constant with
1977 <a href="#t_metadata">metadata type</a>. For example: "<tt>metadata !{
1978 i32 0, metadata !"test" }</tt>". Unlike other constants that are meant to
1979 be interpreted as part of the instruction stream, metadata is a place to
1980 attach additional information such as debug info.</dd>
1985 <!-- ======================================================================= -->
1986 <div class="doc_subsection">
1987 <a name="globalconstants">Global Variable and Function Addresses</a>
1990 <div class="doc_text">
1992 <p>The addresses of <a href="#globalvars">global variables</a>
1993 and <a href="#functionstructure">functions</a> are always implicitly valid
1994 (link-time) constants. These constants are explicitly referenced when
1995 the <a href="#identifiers">identifier for the global</a> is used and always
1996 have <a href="#t_pointer">pointer</a> type. For example, the following is a
1997 legal LLVM file:</p>
1999 <div class="doc_code">
2003 @Z = global [2 x i32*] [ i32* @X, i32* @Y ]
2009 <!-- ======================================================================= -->
2010 <div class="doc_subsection"><a name="undefvalues">Undefined Values</a></div>
2011 <div class="doc_text">
2013 <p>The string '<tt>undef</tt>' can be used anywhere a constant is expected, and
2014 indicates that the user of the value may receive an unspecified bit-pattern.
2015 Undefined values may be of any type (other than label or void) and be used
2016 anywhere a constant is permitted.</p>
2018 <p>Undefined values are useful because they indicate to the compiler that the
2019 program is well defined no matter what value is used. This gives the
2020 compiler more freedom to optimize. Here are some examples of (potentially
2021 surprising) transformations that are valid (in pseudo IR):</p>
2024 <div class="doc_code">
2036 <p>This is safe because all of the output bits are affected by the undef bits.
2037 Any output bit can have a zero or one depending on the input bits.</p>
2039 <div class="doc_code">
2052 <p>These logical operations have bits that are not always affected by the input.
2053 For example, if "%X" has a zero bit, then the output of the 'and' operation will
2054 always be a zero, no matter what the corresponding bit from the undef is. As
2055 such, it is unsafe to optimize or assume that the result of the and is undef.
2056 However, it is safe to assume that all bits of the undef could be 0, and
2057 optimize the and to 0. Likewise, it is safe to assume that all the bits of
2058 the undef operand to the or could be set, allowing the or to be folded to
2061 <div class="doc_code">
2063 %A = select undef, %X, %Y
2064 %B = select undef, 42, %Y
2065 %C = select %X, %Y, undef
2077 <p>This set of examples show that undefined select (and conditional branch)
2078 conditions can go "either way" but they have to come from one of the two
2079 operands. In the %A example, if %X and %Y were both known to have a clear low
2080 bit, then %A would have to have a cleared low bit. However, in the %C example,
2081 the optimizer is allowed to assume that the undef operand could be the same as
2082 %Y, allowing the whole select to be eliminated.</p>
2085 <div class="doc_code">
2087 %A = xor undef, undef
2106 <p>This example points out that two undef operands are not necessarily the same.
2107 This can be surprising to people (and also matches C semantics) where they
2108 assume that "X^X" is always zero, even if X is undef. This isn't true for a
2109 number of reasons, but the short answer is that an undef "variable" can
2110 arbitrarily change its value over its "live range". This is true because the
2111 "variable" doesn't actually <em>have a live range</em>. Instead, the value is
2112 logically read from arbitrary registers that happen to be around when needed,
2113 so the value is not necessarily consistent over time. In fact, %A and %C need
2114 to have the same semantics or the core LLVM "replace all uses with" concept
2117 <div class="doc_code">
2127 <p>These examples show the crucial difference between an <em>undefined
2128 value</em> and <em>undefined behavior</em>. An undefined value (like undef) is
2129 allowed to have an arbitrary bit-pattern. This means that the %A operation
2130 can be constant folded to undef because the undef could be an SNaN, and fdiv is
2131 not (currently) defined on SNaN's. However, in the second example, we can make
2132 a more aggressive assumption: because the undef is allowed to be an arbitrary
2133 value, we are allowed to assume that it could be zero. Since a divide by zero
2134 has <em>undefined behavior</em>, we are allowed to assume that the operation
2135 does not execute at all. This allows us to delete the divide and all code after
2136 it: since the undefined operation "can't happen", the optimizer can assume that
2137 it occurs in dead code.
2140 <div class="doc_code">
2142 a: store undef -> %X
2143 b: store %X -> undef
2150 <p>These examples reiterate the fdiv example: a store "of" an undefined value
2151 can be assumed to not have any effect: we can assume that the value is
2152 overwritten with bits that happen to match what was already there. However, a
2153 store "to" an undefined location could clobber arbitrary memory, therefore, it
2154 has undefined behavior.</p>
2158 <!-- ======================================================================= -->
2159 <div class="doc_subsection"><a name="blockaddress">Addresses of Basic
2161 <div class="doc_text">
2163 <p><b><tt>blockaddress(@function, %block)</tt></b></p>
2165 <p>The '<tt>blockaddress</tt>' constant computes the address of the specified
2166 basic block in the specified function, and always has an i8* type. Taking
2167 the address of the entry block is illegal.</p>
2169 <p>This value only has defined behavior when used as an operand to the
2170 '<a href="#i_indirectbr"><tt>indirectbr</tt></a>' instruction or for comparisons
2171 against null. Pointer equality tests between labels addresses is undefined
2172 behavior - though, again, comparison against null is ok, and no label is
2173 equal to the null pointer. This may also be passed around as an opaque
2174 pointer sized value as long as the bits are not inspected. This allows
2175 <tt>ptrtoint</tt> and arithmetic to be performed on these values so long as
2176 the original value is reconstituted before the <tt>indirectbr</tt>.</p>
2178 <p>Finally, some targets may provide defined semantics when
2179 using the value as the operand to an inline assembly, but that is target
2186 <!-- ======================================================================= -->
2187 <div class="doc_subsection"><a name="constantexprs">Constant Expressions</a>
2190 <div class="doc_text">
2192 <p>Constant expressions are used to allow expressions involving other constants
2193 to be used as constants. Constant expressions may be of
2194 any <a href="#t_firstclass">first class</a> type and may involve any LLVM
2195 operation that does not have side effects (e.g. load and call are not
2196 supported). The following is the syntax for constant expressions:</p>
2199 <dt><b><tt>trunc ( CST to TYPE )</tt></b></dt>
2200 <dd>Truncate a constant to another type. The bit size of CST must be larger
2201 than the bit size of TYPE. Both types must be integers.</dd>
2203 <dt><b><tt>zext ( CST to TYPE )</tt></b></dt>
2204 <dd>Zero extend a constant to another type. The bit size of CST must be
2205 smaller or equal to the bit size of TYPE. Both types must be
2208 <dt><b><tt>sext ( CST to TYPE )</tt></b></dt>
2209 <dd>Sign extend a constant to another type. The bit size of CST must be
2210 smaller or equal to the bit size of TYPE. Both types must be
2213 <dt><b><tt>fptrunc ( CST to TYPE )</tt></b></dt>
2214 <dd>Truncate a floating point constant to another floating point type. The
2215 size of CST must be larger than the size of TYPE. Both types must be
2216 floating point.</dd>
2218 <dt><b><tt>fpext ( CST to TYPE )</tt></b></dt>
2219 <dd>Floating point extend a constant to another type. The size of CST must be
2220 smaller or equal to the size of TYPE. Both types must be floating
2223 <dt><b><tt>fptoui ( CST to TYPE )</tt></b></dt>
2224 <dd>Convert a floating point constant to the corresponding unsigned integer
2225 constant. TYPE must be a scalar or vector integer type. CST must be of
2226 scalar or vector floating point type. Both CST and TYPE must be scalars,
2227 or vectors of the same number of elements. If the value won't fit in the
2228 integer type, the results are undefined.</dd>
2230 <dt><b><tt>fptosi ( CST to TYPE )</tt></b></dt>
2231 <dd>Convert a floating point constant to the corresponding signed integer
2232 constant. TYPE must be a scalar or vector integer type. CST must be of
2233 scalar or vector floating point type. Both CST and TYPE must be scalars,
2234 or vectors of the same number of elements. If the value won't fit in the
2235 integer type, the results are undefined.</dd>
2237 <dt><b><tt>uitofp ( CST to TYPE )</tt></b></dt>
2238 <dd>Convert an unsigned integer constant to the corresponding floating point
2239 constant. TYPE must be a scalar or vector floating point type. CST must be
2240 of scalar or vector integer type. Both CST and TYPE must be scalars, or
2241 vectors of the same number of elements. If the value won't fit in the
2242 floating point type, the results are undefined.</dd>
2244 <dt><b><tt>sitofp ( CST to TYPE )</tt></b></dt>
2245 <dd>Convert a signed integer constant to the corresponding floating point
2246 constant. TYPE must be a scalar or vector floating point type. CST must be
2247 of scalar or vector integer type. Both CST and TYPE must be scalars, or
2248 vectors of the same number of elements. If the value won't fit in the
2249 floating point type, the results are undefined.</dd>
2251 <dt><b><tt>ptrtoint ( CST to TYPE )</tt></b></dt>
2252 <dd>Convert a pointer typed constant to the corresponding integer constant
2253 <tt>TYPE</tt> must be an integer type. <tt>CST</tt> must be of pointer
2254 type. The <tt>CST</tt> value is zero extended, truncated, or unchanged to
2255 make it fit in <tt>TYPE</tt>.</dd>
2257 <dt><b><tt>inttoptr ( CST to TYPE )</tt></b></dt>
2258 <dd>Convert a integer constant to a pointer constant. TYPE must be a pointer
2259 type. CST must be of integer type. The CST value is zero extended,
2260 truncated, or unchanged to make it fit in a pointer size. This one is
2261 <i>really</i> dangerous!</dd>
2263 <dt><b><tt>bitcast ( CST to TYPE )</tt></b></dt>
2264 <dd>Convert a constant, CST, to another TYPE. The constraints of the operands
2265 are the same as those for the <a href="#i_bitcast">bitcast
2266 instruction</a>.</dd>
2268 <dt><b><tt>getelementptr ( CSTPTR, IDX0, IDX1, ... )</tt></b></dt>
2269 <dt><b><tt>getelementptr inbounds ( CSTPTR, IDX0, IDX1, ... )</tt></b></dt>
2270 <dd>Perform the <a href="#i_getelementptr">getelementptr operation</a> on
2271 constants. As with the <a href="#i_getelementptr">getelementptr</a>
2272 instruction, the index list may have zero or more indexes, which are
2273 required to make sense for the type of "CSTPTR".</dd>
2275 <dt><b><tt>select ( COND, VAL1, VAL2 )</tt></b></dt>
2276 <dd>Perform the <a href="#i_select">select operation</a> on constants.</dd>
2278 <dt><b><tt>icmp COND ( VAL1, VAL2 )</tt></b></dt>
2279 <dd>Performs the <a href="#i_icmp">icmp operation</a> on constants.</dd>
2281 <dt><b><tt>fcmp COND ( VAL1, VAL2 )</tt></b></dt>
2282 <dd>Performs the <a href="#i_fcmp">fcmp operation</a> on constants.</dd>
2284 <dt><b><tt>extractelement ( VAL, IDX )</tt></b></dt>
2285 <dd>Perform the <a href="#i_extractelement">extractelement operation</a> on
2288 <dt><b><tt>insertelement ( VAL, ELT, IDX )</tt></b></dt>
2289 <dd>Perform the <a href="#i_insertelement">insertelement operation</a> on
2292 <dt><b><tt>shufflevector ( VEC1, VEC2, IDXMASK )</tt></b></dt>
2293 <dd>Perform the <a href="#i_shufflevector">shufflevector operation</a> on
2296 <dt><b><tt>OPCODE ( LHS, RHS )</tt></b></dt>
2297 <dd>Perform the specified operation of the LHS and RHS constants. OPCODE may
2298 be any of the <a href="#binaryops">binary</a>
2299 or <a href="#bitwiseops">bitwise binary</a> operations. The constraints
2300 on operands are the same as those for the corresponding instruction
2301 (e.g. no bitwise operations on floating point values are allowed).</dd>
2306 <!-- ======================================================================= -->
2307 <div class="doc_subsection"><a name="metadata">Embedded Metadata</a>
2310 <div class="doc_text">
2312 <p>Embedded metadata provides a way to attach arbitrary data to the instruction
2313 stream without affecting the behaviour of the program. There are two
2314 metadata primitives, strings and nodes. All metadata has the
2315 <tt>metadata</tt> type and is identified in syntax by a preceding exclamation
2316 point ('<tt>!</tt>').</p>
2318 <p>A metadata string is a string surrounded by double quotes. It can contain
2319 any character by escaping non-printable characters with "\xx" where "xx" is
2320 the two digit hex code. For example: "<tt>!"test\00"</tt>".</p>
2322 <p>Metadata nodes are represented with notation similar to structure constants
2323 (a comma separated list of elements, surrounded by braces and preceded by an
2324 exclamation point). For example: "<tt>!{ metadata !"test\00", i32
2327 <p>A metadata node will attempt to track changes to the values it holds. In the
2328 event that a value is deleted, it will be replaced with a typeless
2329 "<tt>null</tt>", such as "<tt>metadata !{null, i32 10}</tt>".</p>
2331 <p>Optimizations may rely on metadata to provide additional information about
2332 the program that isn't available in the instructions, or that isn't easily
2333 computable. Similarly, the code generator may expect a certain metadata
2334 format to be used to express debugging information.</p>
2338 <!-- *********************************************************************** -->
2339 <div class="doc_section"> <a name="othervalues">Other Values</a> </div>
2340 <!-- *********************************************************************** -->
2342 <!-- ======================================================================= -->
2343 <div class="doc_subsection">
2344 <a name="inlineasm">Inline Assembler Expressions</a>
2347 <div class="doc_text">
2349 <p>LLVM supports inline assembler expressions (as opposed
2350 to <a href="#moduleasm"> Module-Level Inline Assembly</a>) through the use of
2351 a special value. This value represents the inline assembler as a string
2352 (containing the instructions to emit), a list of operand constraints (stored
2353 as a string), a flag that indicates whether or not the inline asm
2354 expression has side effects, and a flag indicating whether the function
2355 containing the asm needs to align its stack conservatively. An example
2356 inline assembler expression is:</p>
2358 <div class="doc_code">
2360 i32 (i32) asm "bswap $0", "=r,r"
2364 <p>Inline assembler expressions may <b>only</b> be used as the callee operand of
2365 a <a href="#i_call"><tt>call</tt> instruction</a>. Thus, typically we
2368 <div class="doc_code">
2370 %X = call i32 asm "<a href="#int_bswap">bswap</a> $0", "=r,r"(i32 %Y)
2374 <p>Inline asms with side effects not visible in the constraint list must be
2375 marked as having side effects. This is done through the use of the
2376 '<tt>sideeffect</tt>' keyword, like so:</p>
2378 <div class="doc_code">
2380 call void asm sideeffect "eieio", ""()
2384 <p>In some cases inline asms will contain code that will not work unless the
2385 stack is aligned in some way, such as calls or SSE instructions on x86,
2386 yet will not contain code that does that alignment within the asm.
2387 The compiler should make conservative assumptions about what the asm might
2388 contain and should generate its usual stack alignment code in the prologue
2389 if the '<tt>alignstack</tt>' keyword is present:</p>
2391 <div class="doc_code">
2393 call void asm alignstack "eieio", ""()
2397 <p>If both keywords appear the '<tt>sideeffect</tt>' keyword must come
2400 <p>TODO: The format of the asm and constraints string still need to be
2401 documented here. Constraints on what can be done (e.g. duplication, moving,
2402 etc need to be documented). This is probably best done by reference to
2403 another document that covers inline asm from a holistic perspective.</p>
2408 <!-- *********************************************************************** -->
2409 <div class="doc_section">
2410 <a name="intrinsic_globals">Intrinsic Global Variables</a>
2412 <!-- *********************************************************************** -->
2414 <p>LLVM has a number of "magic" global variables that contain data that affect
2415 code generation or other IR semantics. These are documented here. All globals
2416 of this sort should have a section specified as "<tt>llvm.metadata</tt>". This
2417 section and all globals that start with "<tt>llvm.</tt>" are reserved for use
2420 <!-- ======================================================================= -->
2421 <div class="doc_subsection">
2422 <a name="intg_used">The '<tt>llvm.used</tt>' Global Variable</a>
2425 <div class="doc_text">
2427 <p>The <tt>@llvm.used</tt> global is an array with i8* element type which has <a
2428 href="#linkage_appending">appending linkage</a>. This array contains a list of
2429 pointers to global variables and functions which may optionally have a pointer
2430 cast formed of bitcast or getelementptr. For example, a legal use of it is:</p>
2436 @llvm.used = appending global [2 x i8*] [
2438 i8* bitcast (i32* @Y to i8*)
2439 ], section "llvm.metadata"
2442 <p>If a global variable appears in the <tt>@llvm.used</tt> list, then the
2443 compiler, assembler, and linker are required to treat the symbol as if there is
2444 a reference to the global that it cannot see. For example, if a variable has
2445 internal linkage and no references other than that from the <tt>@llvm.used</tt>
2446 list, it cannot be deleted. This is commonly used to represent references from
2447 inline asms and other things the compiler cannot "see", and corresponds to
2448 "attribute((used))" in GNU C.</p>
2450 <p>On some targets, the code generator must emit a directive to the assembler or
2451 object file to prevent the assembler and linker from molesting the symbol.</p>
2455 <!-- ======================================================================= -->
2456 <div class="doc_subsection">
2457 <a name="intg_compiler_used">The '<tt>llvm.compiler.used</tt>' Global Variable</a>
2460 <div class="doc_text">
2462 <p>The <tt>@llvm.compiler.used</tt> directive is the same as the
2463 <tt>@llvm.used</tt> directive, except that it only prevents the compiler from
2464 touching the symbol. On targets that support it, this allows an intelligent
2465 linker to optimize references to the symbol without being impeded as it would be
2466 by <tt>@llvm.used</tt>.</p>
2468 <p>This is a rare construct that should only be used in rare circumstances, and
2469 should not be exposed to source languages.</p>
2473 <!-- ======================================================================= -->
2474 <div class="doc_subsection">
2475 <a name="intg_global_ctors">The '<tt>llvm.global_ctors</tt>' Global Variable</a>
2478 <div class="doc_text">
2480 <p>TODO: Describe this.</p>
2484 <!-- ======================================================================= -->
2485 <div class="doc_subsection">
2486 <a name="intg_global_dtors">The '<tt>llvm.global_dtors</tt>' Global Variable</a>
2489 <div class="doc_text">
2491 <p>TODO: Describe this.</p>
2496 <!-- *********************************************************************** -->
2497 <div class="doc_section"> <a name="instref">Instruction Reference</a> </div>
2498 <!-- *********************************************************************** -->
2500 <div class="doc_text">
2502 <p>The LLVM instruction set consists of several different classifications of
2503 instructions: <a href="#terminators">terminator
2504 instructions</a>, <a href="#binaryops">binary instructions</a>,
2505 <a href="#bitwiseops">bitwise binary instructions</a>,
2506 <a href="#memoryops">memory instructions</a>, and
2507 <a href="#otherops">other instructions</a>.</p>
2511 <!-- ======================================================================= -->
2512 <div class="doc_subsection"> <a name="terminators">Terminator
2513 Instructions</a> </div>
2515 <div class="doc_text">
2517 <p>As mentioned <a href="#functionstructure">previously</a>, every basic block
2518 in a program ends with a "Terminator" instruction, which indicates which
2519 block should be executed after the current block is finished. These
2520 terminator instructions typically yield a '<tt>void</tt>' value: they produce
2521 control flow, not values (the one exception being the
2522 '<a href="#i_invoke"><tt>invoke</tt></a>' instruction).</p>
2524 <p>There are six different terminator instructions: the
2525 '<a href="#i_ret"><tt>ret</tt></a>' instruction, the
2526 '<a href="#i_br"><tt>br</tt></a>' instruction, the
2527 '<a href="#i_switch"><tt>switch</tt></a>' instruction, the
2528 '<a href="#i_indirectbr">'<tt>indirectbr</tt></a>' Instruction, the
2529 '<a href="#i_invoke"><tt>invoke</tt></a>' instruction, the
2530 '<a href="#i_unwind"><tt>unwind</tt></a>' instruction, and the
2531 '<a href="#i_unreachable"><tt>unreachable</tt></a>' instruction.</p>
2535 <!-- _______________________________________________________________________ -->
2536 <div class="doc_subsubsection"> <a name="i_ret">'<tt>ret</tt>'
2537 Instruction</a> </div>
2539 <div class="doc_text">
2543 ret <type> <value> <i>; Return a value from a non-void function</i>
2544 ret void <i>; Return from void function</i>
2548 <p>The '<tt>ret</tt>' instruction is used to return control flow (and optionally
2549 a value) from a function back to the caller.</p>
2551 <p>There are two forms of the '<tt>ret</tt>' instruction: one that returns a
2552 value and then causes control flow, and one that just causes control flow to
2556 <p>The '<tt>ret</tt>' instruction optionally accepts a single argument, the
2557 return value. The type of the return value must be a
2558 '<a href="#t_firstclass">first class</a>' type.</p>
2560 <p>A function is not <a href="#wellformed">well formed</a> if it it has a
2561 non-void return type and contains a '<tt>ret</tt>' instruction with no return
2562 value or a return value with a type that does not match its type, or if it
2563 has a void return type and contains a '<tt>ret</tt>' instruction with a
2567 <p>When the '<tt>ret</tt>' instruction is executed, control flow returns back to
2568 the calling function's context. If the caller is a
2569 "<a href="#i_call"><tt>call</tt></a>" instruction, execution continues at the
2570 instruction after the call. If the caller was an
2571 "<a href="#i_invoke"><tt>invoke</tt></a>" instruction, execution continues at
2572 the beginning of the "normal" destination block. If the instruction returns
2573 a value, that value shall set the call or invoke instruction's return
2578 ret i32 5 <i>; Return an integer value of 5</i>
2579 ret void <i>; Return from a void function</i>
2580 ret { i32, i8 } { i32 4, i8 2 } <i>; Return a struct of values 4 and 2</i>
2584 <!-- _______________________________________________________________________ -->
2585 <div class="doc_subsubsection"> <a name="i_br">'<tt>br</tt>' Instruction</a> </div>
2587 <div class="doc_text">
2591 br i1 <cond>, label <iftrue>, label <iffalse><br> br label <dest> <i>; Unconditional branch</i>
2595 <p>The '<tt>br</tt>' instruction is used to cause control flow to transfer to a
2596 different basic block in the current function. There are two forms of this
2597 instruction, corresponding to a conditional branch and an unconditional
2601 <p>The conditional branch form of the '<tt>br</tt>' instruction takes a single
2602 '<tt>i1</tt>' value and two '<tt>label</tt>' values. The unconditional form
2603 of the '<tt>br</tt>' instruction takes a single '<tt>label</tt>' value as a
2607 <p>Upon execution of a conditional '<tt>br</tt>' instruction, the '<tt>i1</tt>'
2608 argument is evaluated. If the value is <tt>true</tt>, control flows to the
2609 '<tt>iftrue</tt>' <tt>label</tt> argument. If "cond" is <tt>false</tt>,
2610 control flows to the '<tt>iffalse</tt>' <tt>label</tt> argument.</p>
2615 %cond = <a href="#i_icmp">icmp</a> eq i32 %a, %b
2616 br i1 %cond, label %IfEqual, label %IfUnequal
2618 <a href="#i_ret">ret</a> i32 1
2620 <a href="#i_ret">ret</a> i32 0
2625 <!-- _______________________________________________________________________ -->
2626 <div class="doc_subsubsection">
2627 <a name="i_switch">'<tt>switch</tt>' Instruction</a>
2630 <div class="doc_text">
2634 switch <intty> <value>, label <defaultdest> [ <intty> <val>, label <dest> ... ]
2638 <p>The '<tt>switch</tt>' instruction is used to transfer control flow to one of
2639 several different places. It is a generalization of the '<tt>br</tt>'
2640 instruction, allowing a branch to occur to one of many possible
2644 <p>The '<tt>switch</tt>' instruction uses three parameters: an integer
2645 comparison value '<tt>value</tt>', a default '<tt>label</tt>' destination,
2646 and an array of pairs of comparison value constants and '<tt>label</tt>'s.
2647 The table is not allowed to contain duplicate constant entries.</p>
2650 <p>The <tt>switch</tt> instruction specifies a table of values and
2651 destinations. When the '<tt>switch</tt>' instruction is executed, this table
2652 is searched for the given value. If the value is found, control flow is
2653 transferred to the corresponding destination; otherwise, control flow is
2654 transferred to the default destination.</p>
2656 <h5>Implementation:</h5>
2657 <p>Depending on properties of the target machine and the particular
2658 <tt>switch</tt> instruction, this instruction may be code generated in
2659 different ways. For example, it could be generated as a series of chained
2660 conditional branches or with a lookup table.</p>
2664 <i>; Emulate a conditional br instruction</i>
2665 %Val = <a href="#i_zext">zext</a> i1 %value to i32
2666 switch i32 %Val, label %truedest [ i32 0, label %falsedest ]
2668 <i>; Emulate an unconditional br instruction</i>
2669 switch i32 0, label %dest [ ]
2671 <i>; Implement a jump table:</i>
2672 switch i32 %val, label %otherwise [ i32 0, label %onzero
2674 i32 2, label %ontwo ]
2680 <!-- _______________________________________________________________________ -->
2681 <div class="doc_subsubsection">
2682 <a name="i_indirectbr">'<tt>indirectbr</tt>' Instruction</a>
2685 <div class="doc_text">
2689 indirectbr <somety>* <address>, [ label <dest1>, label <dest2>, ... ]
2694 <p>The '<tt>indirectbr</tt>' instruction implements an indirect branch to a label
2695 within the current function, whose address is specified by
2696 "<tt>address</tt>". Address must be derived from a <a
2697 href="#blockaddress">blockaddress</a> constant.</p>
2701 <p>The '<tt>address</tt>' argument is the address of the label to jump to. The
2702 rest of the arguments indicate the full set of possible destinations that the
2703 address may point to. Blocks are allowed to occur multiple times in the
2704 destination list, though this isn't particularly useful.</p>
2706 <p>This destination list is required so that dataflow analysis has an accurate
2707 understanding of the CFG.</p>
2711 <p>Control transfers to the block specified in the address argument. All
2712 possible destination blocks must be listed in the label list, otherwise this
2713 instruction has undefined behavior. This implies that jumps to labels
2714 defined in other functions have undefined behavior as well.</p>
2716 <h5>Implementation:</h5>
2718 <p>This is typically implemented with a jump through a register.</p>
2722 indirectbr i8* %Addr, [ label %bb1, label %bb2, label %bb3 ]
2728 <!-- _______________________________________________________________________ -->
2729 <div class="doc_subsubsection">
2730 <a name="i_invoke">'<tt>invoke</tt>' Instruction</a>
2733 <div class="doc_text">
2737 <result> = invoke [<a href="#callingconv">cconv</a>] [<a href="#paramattrs">ret attrs</a>] <ptr to function ty> <function ptr val>(<function args>) [<a href="#fnattrs">fn attrs</a>]
2738 to label <normal label> unwind label <exception label>
2742 <p>The '<tt>invoke</tt>' instruction causes control to transfer to a specified
2743 function, with the possibility of control flow transfer to either the
2744 '<tt>normal</tt>' label or the '<tt>exception</tt>' label. If the callee
2745 function returns with the "<tt><a href="#i_ret">ret</a></tt>" instruction,
2746 control flow will return to the "normal" label. If the callee (or any
2747 indirect callees) returns with the "<a href="#i_unwind"><tt>unwind</tt></a>"
2748 instruction, control is interrupted and continued at the dynamically nearest
2749 "exception" label.</p>
2752 <p>This instruction requires several arguments:</p>
2755 <li>The optional "cconv" marker indicates which <a href="#callingconv">calling
2756 convention</a> the call should use. If none is specified, the call
2757 defaults to using C calling conventions.</li>
2759 <li>The optional <a href="#paramattrs">Parameter Attributes</a> list for
2760 return values. Only '<tt>zeroext</tt>', '<tt>signext</tt>', and
2761 '<tt>inreg</tt>' attributes are valid here.</li>
2763 <li>'<tt>ptr to function ty</tt>': shall be the signature of the pointer to
2764 function value being invoked. In most cases, this is a direct function
2765 invocation, but indirect <tt>invoke</tt>s are just as possible, branching
2766 off an arbitrary pointer to function value.</li>
2768 <li>'<tt>function ptr val</tt>': An LLVM value containing a pointer to a
2769 function to be invoked. </li>
2771 <li>'<tt>function args</tt>': argument list whose types match the function
2772 signature argument types. If the function signature indicates the
2773 function accepts a variable number of arguments, the extra arguments can
2776 <li>'<tt>normal label</tt>': the label reached when the called function
2777 executes a '<tt><a href="#i_ret">ret</a></tt>' instruction. </li>
2779 <li>'<tt>exception label</tt>': the label reached when a callee returns with
2780 the <a href="#i_unwind"><tt>unwind</tt></a> instruction. </li>
2782 <li>The optional <a href="#fnattrs">function attributes</a> list. Only
2783 '<tt>noreturn</tt>', '<tt>nounwind</tt>', '<tt>readonly</tt>' and
2784 '<tt>readnone</tt>' attributes are valid here.</li>
2788 <p>This instruction is designed to operate as a standard
2789 '<tt><a href="#i_call">call</a></tt>' instruction in most regards. The
2790 primary difference is that it establishes an association with a label, which
2791 is used by the runtime library to unwind the stack.</p>
2793 <p>This instruction is used in languages with destructors to ensure that proper
2794 cleanup is performed in the case of either a <tt>longjmp</tt> or a thrown
2795 exception. Additionally, this is important for implementation of
2796 '<tt>catch</tt>' clauses in high-level languages that support them.</p>
2798 <p>For the purposes of the SSA form, the definition of the value returned by the
2799 '<tt>invoke</tt>' instruction is deemed to occur on the edge from the current
2800 block to the "normal" label. If the callee unwinds then no return value is
2805 %retval = invoke i32 @Test(i32 15) to label %Continue
2806 unwind label %TestCleanup <i>; {i32}:retval set</i>
2807 %retval = invoke <a href="#callingconv">coldcc</a> i32 %Testfnptr(i32 15) to label %Continue
2808 unwind label %TestCleanup <i>; {i32}:retval set</i>
2813 <!-- _______________________________________________________________________ -->
2815 <div class="doc_subsubsection"> <a name="i_unwind">'<tt>unwind</tt>'
2816 Instruction</a> </div>
2818 <div class="doc_text">
2826 <p>The '<tt>unwind</tt>' instruction unwinds the stack, continuing control flow
2827 at the first callee in the dynamic call stack which used
2828 an <a href="#i_invoke"><tt>invoke</tt></a> instruction to perform the call.
2829 This is primarily used to implement exception handling.</p>
2832 <p>The '<tt>unwind</tt>' instruction causes execution of the current function to
2833 immediately halt. The dynamic call stack is then searched for the
2834 first <a href="#i_invoke"><tt>invoke</tt></a> instruction on the call stack.
2835 Once found, execution continues at the "exceptional" destination block
2836 specified by the <tt>invoke</tt> instruction. If there is no <tt>invoke</tt>
2837 instruction in the dynamic call chain, undefined behavior results.</p>
2841 <!-- _______________________________________________________________________ -->
2843 <div class="doc_subsubsection"> <a name="i_unreachable">'<tt>unreachable</tt>'
2844 Instruction</a> </div>
2846 <div class="doc_text">
2854 <p>The '<tt>unreachable</tt>' instruction has no defined semantics. This
2855 instruction is used to inform the optimizer that a particular portion of the
2856 code is not reachable. This can be used to indicate that the code after a
2857 no-return function cannot be reached, and other facts.</p>
2860 <p>The '<tt>unreachable</tt>' instruction has no defined semantics.</p>
2864 <!-- ======================================================================= -->
2865 <div class="doc_subsection"> <a name="binaryops">Binary Operations</a> </div>
2867 <div class="doc_text">
2869 <p>Binary operators are used to do most of the computation in a program. They
2870 require two operands of the same type, execute an operation on them, and
2871 produce a single value. The operands might represent multiple data, as is
2872 the case with the <a href="#t_vector">vector</a> data type. The result value
2873 has the same type as its operands.</p>
2875 <p>There are several different binary operators:</p>
2879 <!-- _______________________________________________________________________ -->
2880 <div class="doc_subsubsection">
2881 <a name="i_add">'<tt>add</tt>' Instruction</a>
2884 <div class="doc_text">
2888 <result> = add <ty> <op1>, <op2> <i>; yields {ty}:result</i>
2889 <result> = add nuw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
2890 <result> = add nsw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
2891 <result> = add nuw nsw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
2895 <p>The '<tt>add</tt>' instruction returns the sum of its two operands.</p>
2898 <p>The two arguments to the '<tt>add</tt>' instruction must
2899 be <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
2900 integer values. Both arguments must have identical types.</p>
2903 <p>The value produced is the integer sum of the two operands.</p>
2905 <p>If the sum has unsigned overflow, the result returned is the mathematical
2906 result modulo 2<sup>n</sup>, where n is the bit width of the result.</p>
2908 <p>Because LLVM integers use a two's complement representation, this instruction
2909 is appropriate for both signed and unsigned integers.</p>
2911 <p><tt>nuw</tt> and <tt>nsw</tt> stand for "No Unsigned Wrap"
2912 and "No Signed Wrap", respectively. If the <tt>nuw</tt> and/or
2913 <tt>nsw</tt> keywords are present, the result value of the <tt>add</tt>
2914 is undefined if unsigned and/or signed overflow, respectively, occurs.</p>
2918 <result> = add i32 4, %var <i>; yields {i32}:result = 4 + %var</i>
2923 <!-- _______________________________________________________________________ -->
2924 <div class="doc_subsubsection">
2925 <a name="i_fadd">'<tt>fadd</tt>' Instruction</a>
2928 <div class="doc_text">
2932 <result> = fadd <ty> <op1>, <op2> <i>; yields {ty}:result</i>
2936 <p>The '<tt>fadd</tt>' instruction returns the sum of its two operands.</p>
2939 <p>The two arguments to the '<tt>fadd</tt>' instruction must be
2940 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
2941 floating point values. Both arguments must have identical types.</p>
2944 <p>The value produced is the floating point sum of the two operands.</p>
2948 <result> = fadd float 4.0, %var <i>; yields {float}:result = 4.0 + %var</i>
2953 <!-- _______________________________________________________________________ -->
2954 <div class="doc_subsubsection">
2955 <a name="i_sub">'<tt>sub</tt>' Instruction</a>
2958 <div class="doc_text">
2962 <result> = sub <ty> <op1>, <op2> <i>; yields {ty}:result</i>
2963 <result> = sub nuw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
2964 <result> = sub nsw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
2965 <result> = sub nuw nsw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
2969 <p>The '<tt>sub</tt>' instruction returns the difference of its two
2972 <p>Note that the '<tt>sub</tt>' instruction is used to represent the
2973 '<tt>neg</tt>' instruction present in most other intermediate
2974 representations.</p>
2977 <p>The two arguments to the '<tt>sub</tt>' instruction must
2978 be <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
2979 integer values. Both arguments must have identical types.</p>
2982 <p>The value produced is the integer difference of the two operands.</p>
2984 <p>If the difference has unsigned overflow, the result returned is the
2985 mathematical result modulo 2<sup>n</sup>, where n is the bit width of the
2988 <p>Because LLVM integers use a two's complement representation, this instruction
2989 is appropriate for both signed and unsigned integers.</p>
2991 <p><tt>nuw</tt> and <tt>nsw</tt> stand for "No Unsigned Wrap"
2992 and "No Signed Wrap", respectively. If the <tt>nuw</tt> and/or
2993 <tt>nsw</tt> keywords are present, the result value of the <tt>sub</tt>
2994 is undefined if unsigned and/or signed overflow, respectively, occurs.</p>
2998 <result> = sub i32 4, %var <i>; yields {i32}:result = 4 - %var</i>
2999 <result> = sub i32 0, %val <i>; yields {i32}:result = -%var</i>
3004 <!-- _______________________________________________________________________ -->
3005 <div class="doc_subsubsection">
3006 <a name="i_fsub">'<tt>fsub</tt>' Instruction</a>
3009 <div class="doc_text">
3013 <result> = fsub <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3017 <p>The '<tt>fsub</tt>' instruction returns the difference of its two
3020 <p>Note that the '<tt>fsub</tt>' instruction is used to represent the
3021 '<tt>fneg</tt>' instruction present in most other intermediate
3022 representations.</p>
3025 <p>The two arguments to the '<tt>fsub</tt>' instruction must be
3026 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3027 floating point values. Both arguments must have identical types.</p>
3030 <p>The value produced is the floating point difference of the two operands.</p>
3034 <result> = fsub float 4.0, %var <i>; yields {float}:result = 4.0 - %var</i>
3035 <result> = fsub float -0.0, %val <i>; yields {float}:result = -%var</i>
3040 <!-- _______________________________________________________________________ -->
3041 <div class="doc_subsubsection">
3042 <a name="i_mul">'<tt>mul</tt>' Instruction</a>
3045 <div class="doc_text">
3049 <result> = mul <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3050 <result> = mul nuw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3051 <result> = mul nsw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3052 <result> = mul nuw nsw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3056 <p>The '<tt>mul</tt>' instruction returns the product of its two operands.</p>
3059 <p>The two arguments to the '<tt>mul</tt>' instruction must
3060 be <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
3061 integer values. Both arguments must have identical types.</p>
3064 <p>The value produced is the integer product of the two operands.</p>
3066 <p>If the result of the multiplication has unsigned overflow, the result
3067 returned is the mathematical result modulo 2<sup>n</sup>, where n is the bit
3068 width of the result.</p>
3070 <p>Because LLVM integers use a two's complement representation, and the result
3071 is the same width as the operands, this instruction returns the correct
3072 result for both signed and unsigned integers. If a full product
3073 (e.g. <tt>i32</tt>x<tt>i32</tt>-><tt>i64</tt>) is needed, the operands should
3074 be sign-extended or zero-extended as appropriate to the width of the full
3077 <p><tt>nuw</tt> and <tt>nsw</tt> stand for "No Unsigned Wrap"
3078 and "No Signed Wrap", respectively. If the <tt>nuw</tt> and/or
3079 <tt>nsw</tt> keywords are present, the result value of the <tt>mul</tt>
3080 is undefined if unsigned and/or signed overflow, respectively, occurs.</p>
3084 <result> = mul i32 4, %var <i>; yields {i32}:result = 4 * %var</i>
3089 <!-- _______________________________________________________________________ -->
3090 <div class="doc_subsubsection">
3091 <a name="i_fmul">'<tt>fmul</tt>' Instruction</a>
3094 <div class="doc_text">
3098 <result> = fmul <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3102 <p>The '<tt>fmul</tt>' instruction returns the product of its two operands.</p>
3105 <p>The two arguments to the '<tt>fmul</tt>' instruction must be
3106 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3107 floating point values. Both arguments must have identical types.</p>
3110 <p>The value produced is the floating point product of the two operands.</p>
3114 <result> = fmul float 4.0, %var <i>; yields {float}:result = 4.0 * %var</i>
3119 <!-- _______________________________________________________________________ -->
3120 <div class="doc_subsubsection"> <a name="i_udiv">'<tt>udiv</tt>' Instruction
3123 <div class="doc_text">
3127 <result> = udiv <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3131 <p>The '<tt>udiv</tt>' instruction returns the quotient of its two operands.</p>
3134 <p>The two arguments to the '<tt>udiv</tt>' instruction must be
3135 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3136 values. Both arguments must have identical types.</p>
3139 <p>The value produced is the unsigned integer quotient of the two operands.</p>
3141 <p>Note that unsigned integer division and signed integer division are distinct
3142 operations; for signed integer division, use '<tt>sdiv</tt>'.</p>
3144 <p>Division by zero leads to undefined behavior.</p>
3148 <result> = udiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
3153 <!-- _______________________________________________________________________ -->
3154 <div class="doc_subsubsection"> <a name="i_sdiv">'<tt>sdiv</tt>' Instruction
3157 <div class="doc_text">
3161 <result> = sdiv <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3162 <result> = sdiv exact <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3166 <p>The '<tt>sdiv</tt>' instruction returns the quotient of its two operands.</p>
3169 <p>The two arguments to the '<tt>sdiv</tt>' instruction must be
3170 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3171 values. Both arguments must have identical types.</p>
3174 <p>The value produced is the signed integer quotient of the two operands rounded
3177 <p>Note that signed integer division and unsigned integer division are distinct
3178 operations; for unsigned integer division, use '<tt>udiv</tt>'.</p>
3180 <p>Division by zero leads to undefined behavior. Overflow also leads to
3181 undefined behavior; this is a rare case, but can occur, for example, by doing
3182 a 32-bit division of -2147483648 by -1.</p>
3184 <p>If the <tt>exact</tt> keyword is present, the result value of the
3185 <tt>sdiv</tt> is undefined if the result would be rounded or if overflow
3190 <result> = sdiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
3195 <!-- _______________________________________________________________________ -->
3196 <div class="doc_subsubsection"> <a name="i_fdiv">'<tt>fdiv</tt>'
3197 Instruction</a> </div>
3199 <div class="doc_text">
3203 <result> = fdiv <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3207 <p>The '<tt>fdiv</tt>' instruction returns the quotient of its two operands.</p>
3210 <p>The two arguments to the '<tt>fdiv</tt>' instruction must be
3211 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3212 floating point values. Both arguments must have identical types.</p>
3215 <p>The value produced is the floating point quotient of the two operands.</p>
3219 <result> = fdiv float 4.0, %var <i>; yields {float}:result = 4.0 / %var</i>
3224 <!-- _______________________________________________________________________ -->
3225 <div class="doc_subsubsection"> <a name="i_urem">'<tt>urem</tt>' Instruction</a>
3228 <div class="doc_text">
3232 <result> = urem <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3236 <p>The '<tt>urem</tt>' instruction returns the remainder from the unsigned
3237 division of its two arguments.</p>
3240 <p>The two arguments to the '<tt>urem</tt>' instruction must be
3241 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3242 values. Both arguments must have identical types.</p>
3245 <p>This instruction returns the unsigned integer <i>remainder</i> of a division.
3246 This instruction always performs an unsigned division to get the
3249 <p>Note that unsigned integer remainder and signed integer remainder are
3250 distinct operations; for signed integer remainder, use '<tt>srem</tt>'.</p>
3252 <p>Taking the remainder of a division by zero leads to undefined behavior.</p>
3256 <result> = urem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
3261 <!-- _______________________________________________________________________ -->
3262 <div class="doc_subsubsection">
3263 <a name="i_srem">'<tt>srem</tt>' Instruction</a>
3266 <div class="doc_text">
3270 <result> = srem <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3274 <p>The '<tt>srem</tt>' instruction returns the remainder from the signed
3275 division of its two operands. This instruction can also take
3276 <a href="#t_vector">vector</a> versions of the values in which case the
3277 elements must be integers.</p>
3280 <p>The two arguments to the '<tt>srem</tt>' instruction must be
3281 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3282 values. Both arguments must have identical types.</p>
3285 <p>This instruction returns the <i>remainder</i> of a division (where the result
3286 has the same sign as the dividend, <tt>op1</tt>), not the <i>modulo</i>
3287 operator (where the result has the same sign as the divisor, <tt>op2</tt>) of
3288 a value. For more information about the difference,
3289 see <a href="http://mathforum.org/dr.math/problems/anne.4.28.99.html">The
3290 Math Forum</a>. For a table of how this is implemented in various languages,
3291 please see <a href="http://en.wikipedia.org/wiki/Modulo_operation">
3292 Wikipedia: modulo operation</a>.</p>
3294 <p>Note that signed integer remainder and unsigned integer remainder are
3295 distinct operations; for unsigned integer remainder, use '<tt>urem</tt>'.</p>
3297 <p>Taking the remainder of a division by zero leads to undefined behavior.
3298 Overflow also leads to undefined behavior; this is a rare case, but can
3299 occur, for example, by taking the remainder of a 32-bit division of
3300 -2147483648 by -1. (The remainder doesn't actually overflow, but this rule
3301 lets srem be implemented using instructions that return both the result of
3302 the division and the remainder.)</p>
3306 <result> = srem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
3311 <!-- _______________________________________________________________________ -->
3312 <div class="doc_subsubsection">
3313 <a name="i_frem">'<tt>frem</tt>' Instruction</a> </div>
3315 <div class="doc_text">
3319 <result> = frem <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3323 <p>The '<tt>frem</tt>' instruction returns the remainder from the division of
3324 its two operands.</p>
3327 <p>The two arguments to the '<tt>frem</tt>' instruction must be
3328 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3329 floating point values. Both arguments must have identical types.</p>
3332 <p>This instruction returns the <i>remainder</i> of a division. The remainder
3333 has the same sign as the dividend.</p>
3337 <result> = frem float 4.0, %var <i>; yields {float}:result = 4.0 % %var</i>
3342 <!-- ======================================================================= -->
3343 <div class="doc_subsection"> <a name="bitwiseops">Bitwise Binary
3344 Operations</a> </div>
3346 <div class="doc_text">
3348 <p>Bitwise binary operators are used to do various forms of bit-twiddling in a
3349 program. They are generally very efficient instructions and can commonly be
3350 strength reduced from other instructions. They require two operands of the
3351 same type, execute an operation on them, and produce a single value. The
3352 resulting value is the same type as its operands.</p>
3356 <!-- _______________________________________________________________________ -->
3357 <div class="doc_subsubsection"> <a name="i_shl">'<tt>shl</tt>'
3358 Instruction</a> </div>
3360 <div class="doc_text">
3364 <result> = shl <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3368 <p>The '<tt>shl</tt>' instruction returns the first operand shifted to the left
3369 a specified number of bits.</p>
3372 <p>Both arguments to the '<tt>shl</tt>' instruction must be the
3373 same <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
3374 integer type. '<tt>op2</tt>' is treated as an unsigned value.</p>
3377 <p>The value produced is <tt>op1</tt> * 2<sup><tt>op2</tt></sup> mod
3378 2<sup>n</sup>, where <tt>n</tt> is the width of the result. If <tt>op2</tt>
3379 is (statically or dynamically) negative or equal to or larger than the number
3380 of bits in <tt>op1</tt>, the result is undefined. If the arguments are
3381 vectors, each vector element of <tt>op1</tt> is shifted by the corresponding
3382 shift amount in <tt>op2</tt>.</p>
3386 <result> = shl i32 4, %var <i>; yields {i32}: 4 << %var</i>
3387 <result> = shl i32 4, 2 <i>; yields {i32}: 16</i>
3388 <result> = shl i32 1, 10 <i>; yields {i32}: 1024</i>
3389 <result> = shl i32 1, 32 <i>; undefined</i>
3390 <result> = shl <2 x i32> < i32 1, i32 1>, < i32 1, i32 2> <i>; yields: result=<2 x i32> < i32 2, i32 4></i>
3395 <!-- _______________________________________________________________________ -->
3396 <div class="doc_subsubsection"> <a name="i_lshr">'<tt>lshr</tt>'
3397 Instruction</a> </div>
3399 <div class="doc_text">
3403 <result> = lshr <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3407 <p>The '<tt>lshr</tt>' instruction (logical shift right) returns the first
3408 operand shifted to the right a specified number of bits with zero fill.</p>
3411 <p>Both arguments to the '<tt>lshr</tt>' instruction must be the same
3412 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3413 type. '<tt>op2</tt>' is treated as an unsigned value.</p>
3416 <p>This instruction always performs a logical shift right operation. The most
3417 significant bits of the result will be filled with zero bits after the shift.
3418 If <tt>op2</tt> is (statically or dynamically) equal to or larger than the
3419 number of bits in <tt>op1</tt>, the result is undefined. If the arguments are
3420 vectors, each vector element of <tt>op1</tt> is shifted by the corresponding
3421 shift amount in <tt>op2</tt>.</p>
3425 <result> = lshr i32 4, 1 <i>; yields {i32}:result = 2</i>
3426 <result> = lshr i32 4, 2 <i>; yields {i32}:result = 1</i>
3427 <result> = lshr i8 4, 3 <i>; yields {i8}:result = 0</i>
3428 <result> = lshr i8 -2, 1 <i>; yields {i8}:result = 0x7FFFFFFF </i>
3429 <result> = lshr i32 1, 32 <i>; undefined</i>
3430 <result> = lshr <2 x i32> < i32 -2, i32 4>, < i32 1, i32 2> <i>; yields: result=<2 x i32> < i32 0x7FFFFFFF, i32 1></i>
3435 <!-- _______________________________________________________________________ -->
3436 <div class="doc_subsubsection"> <a name="i_ashr">'<tt>ashr</tt>'
3437 Instruction</a> </div>
3438 <div class="doc_text">
3442 <result> = ashr <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3446 <p>The '<tt>ashr</tt>' instruction (arithmetic shift right) returns the first
3447 operand shifted to the right a specified number of bits with sign
3451 <p>Both arguments to the '<tt>ashr</tt>' instruction must be the same
3452 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3453 type. '<tt>op2</tt>' is treated as an unsigned value.</p>
3456 <p>This instruction always performs an arithmetic shift right operation, The
3457 most significant bits of the result will be filled with the sign bit
3458 of <tt>op1</tt>. If <tt>op2</tt> is (statically or dynamically) equal to or
3459 larger than the number of bits in <tt>op1</tt>, the result is undefined. If
3460 the arguments are vectors, each vector element of <tt>op1</tt> is shifted by
3461 the corresponding shift amount in <tt>op2</tt>.</p>
3465 <result> = ashr i32 4, 1 <i>; yields {i32}:result = 2</i>
3466 <result> = ashr i32 4, 2 <i>; yields {i32}:result = 1</i>
3467 <result> = ashr i8 4, 3 <i>; yields {i8}:result = 0</i>
3468 <result> = ashr i8 -2, 1 <i>; yields {i8}:result = -1</i>
3469 <result> = ashr i32 1, 32 <i>; undefined</i>
3470 <result> = ashr <2 x i32> < i32 -2, i32 4>, < i32 1, i32 3> <i>; yields: result=<2 x i32> < i32 -1, i32 0></i>
3475 <!-- _______________________________________________________________________ -->
3476 <div class="doc_subsubsection"> <a name="i_and">'<tt>and</tt>'
3477 Instruction</a> </div>
3479 <div class="doc_text">
3483 <result> = and <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3487 <p>The '<tt>and</tt>' instruction returns the bitwise logical and of its two
3491 <p>The two arguments to the '<tt>and</tt>' instruction must be
3492 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3493 values. Both arguments must have identical types.</p>
3496 <p>The truth table used for the '<tt>and</tt>' instruction is:</p>
3498 <table border="1" cellspacing="0" cellpadding="4">
3530 <result> = and i32 4, %var <i>; yields {i32}:result = 4 & %var</i>
3531 <result> = and i32 15, 40 <i>; yields {i32}:result = 8</i>
3532 <result> = and i32 4, 8 <i>; yields {i32}:result = 0</i>
3535 <!-- _______________________________________________________________________ -->
3536 <div class="doc_subsubsection"> <a name="i_or">'<tt>or</tt>' Instruction</a> </div>
3538 <div class="doc_text">
3542 <result> = or <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3546 <p>The '<tt>or</tt>' instruction returns the bitwise logical inclusive or of its
3550 <p>The two arguments to the '<tt>or</tt>' instruction must be
3551 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3552 values. Both arguments must have identical types.</p>
3555 <p>The truth table used for the '<tt>or</tt>' instruction is:</p>
3557 <table border="1" cellspacing="0" cellpadding="4">
3589 <result> = or i32 4, %var <i>; yields {i32}:result = 4 | %var</i>
3590 <result> = or i32 15, 40 <i>; yields {i32}:result = 47</i>
3591 <result> = or i32 4, 8 <i>; yields {i32}:result = 12</i>
3596 <!-- _______________________________________________________________________ -->
3597 <div class="doc_subsubsection"> <a name="i_xor">'<tt>xor</tt>'
3598 Instruction</a> </div>
3600 <div class="doc_text">
3604 <result> = xor <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3608 <p>The '<tt>xor</tt>' instruction returns the bitwise logical exclusive or of
3609 its two operands. The <tt>xor</tt> is used to implement the "one's
3610 complement" operation, which is the "~" operator in C.</p>
3613 <p>The two arguments to the '<tt>xor</tt>' instruction must be
3614 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3615 values. Both arguments must have identical types.</p>
3618 <p>The truth table used for the '<tt>xor</tt>' instruction is:</p>
3620 <table border="1" cellspacing="0" cellpadding="4">
3652 <result> = xor i32 4, %var <i>; yields {i32}:result = 4 ^ %var</i>
3653 <result> = xor i32 15, 40 <i>; yields {i32}:result = 39</i>
3654 <result> = xor i32 4, 8 <i>; yields {i32}:result = 12</i>
3655 <result> = xor i32 %V, -1 <i>; yields {i32}:result = ~%V</i>
3660 <!-- ======================================================================= -->
3661 <div class="doc_subsection">
3662 <a name="vectorops">Vector Operations</a>
3665 <div class="doc_text">
3667 <p>LLVM supports several instructions to represent vector operations in a
3668 target-independent manner. These instructions cover the element-access and
3669 vector-specific operations needed to process vectors effectively. While LLVM
3670 does directly support these vector operations, many sophisticated algorithms
3671 will want to use target-specific intrinsics to take full advantage of a
3672 specific target.</p>
3676 <!-- _______________________________________________________________________ -->
3677 <div class="doc_subsubsection">
3678 <a name="i_extractelement">'<tt>extractelement</tt>' Instruction</a>
3681 <div class="doc_text">
3685 <result> = extractelement <n x <ty>> <val>, i32 <idx> <i>; yields <ty></i>
3689 <p>The '<tt>extractelement</tt>' instruction extracts a single scalar element
3690 from a vector at a specified index.</p>
3694 <p>The first operand of an '<tt>extractelement</tt>' instruction is a value
3695 of <a href="#t_vector">vector</a> type. The second operand is an index
3696 indicating the position from which to extract the element. The index may be
3700 <p>The result is a scalar of the same type as the element type of
3701 <tt>val</tt>. Its value is the value at position <tt>idx</tt> of
3702 <tt>val</tt>. If <tt>idx</tt> exceeds the length of <tt>val</tt>, the
3703 results are undefined.</p>
3707 <result> = extractelement <4 x i32> %vec, i32 0 <i>; yields i32</i>
3712 <!-- _______________________________________________________________________ -->
3713 <div class="doc_subsubsection">
3714 <a name="i_insertelement">'<tt>insertelement</tt>' Instruction</a>
3717 <div class="doc_text">
3721 <result> = insertelement <n x <ty>> <val>, <ty> <elt>, i32 <idx> <i>; yields <n x <ty>></i>
3725 <p>The '<tt>insertelement</tt>' instruction inserts a scalar element into a
3726 vector at a specified index.</p>
3729 <p>The first operand of an '<tt>insertelement</tt>' instruction is a value
3730 of <a href="#t_vector">vector</a> type. The second operand is a scalar value
3731 whose type must equal the element type of the first operand. The third
3732 operand is an index indicating the position at which to insert the value.
3733 The index may be a variable.</p>
3736 <p>The result is a vector of the same type as <tt>val</tt>. Its element values
3737 are those of <tt>val</tt> except at position <tt>idx</tt>, where it gets the
3738 value <tt>elt</tt>. If <tt>idx</tt> exceeds the length of <tt>val</tt>, the
3739 results are undefined.</p>
3743 <result> = insertelement <4 x i32> %vec, i32 1, i32 0 <i>; yields <4 x i32></i>
3748 <!-- _______________________________________________________________________ -->
3749 <div class="doc_subsubsection">
3750 <a name="i_shufflevector">'<tt>shufflevector</tt>' Instruction</a>
3753 <div class="doc_text">
3757 <result> = shufflevector <n x <ty>> <v1>, <n x <ty>> <v2>, <m x i32> <mask> <i>; yields <m x <ty>></i>
3761 <p>The '<tt>shufflevector</tt>' instruction constructs a permutation of elements
3762 from two input vectors, returning a vector with the same element type as the
3763 input and length that is the same as the shuffle mask.</p>
3766 <p>The first two operands of a '<tt>shufflevector</tt>' instruction are vectors
3767 with types that match each other. The third argument is a shuffle mask whose
3768 element type is always 'i32'. The result of the instruction is a vector
3769 whose length is the same as the shuffle mask and whose element type is the
3770 same as the element type of the first two operands.</p>
3772 <p>The shuffle mask operand is required to be a constant vector with either
3773 constant integer or undef values.</p>
3776 <p>The elements of the two input vectors are numbered from left to right across
3777 both of the vectors. The shuffle mask operand specifies, for each element of
3778 the result vector, which element of the two input vectors the result element
3779 gets. The element selector may be undef (meaning "don't care") and the
3780 second operand may be undef if performing a shuffle from only one vector.</p>
3784 <result> = shufflevector <4 x i32> %v1, <4 x i32> %v2,
3785 <4 x i32> <i32 0, i32 4, i32 1, i32 5> <i>; yields <4 x i32></i>
3786 <result> = shufflevector <4 x i32> %v1, <4 x i32> undef,
3787 <4 x i32> <i32 0, i32 1, i32 2, i32 3> <i>; yields <4 x i32></i> - Identity shuffle.
3788 <result> = shufflevector <8 x i32> %v1, <8 x i32> undef,
3789 <4 x i32> <i32 0, i32 1, i32 2, i32 3> <i>; yields <4 x i32></i>
3790 <result> = shufflevector <4 x i32> %v1, <4 x i32> %v2,
3791 <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7 > <i>; yields <8 x i32></i>
3796 <!-- ======================================================================= -->
3797 <div class="doc_subsection">
3798 <a name="aggregateops">Aggregate Operations</a>
3801 <div class="doc_text">
3803 <p>LLVM supports several instructions for working with aggregate values.</p>
3807 <!-- _______________________________________________________________________ -->
3808 <div class="doc_subsubsection">
3809 <a name="i_extractvalue">'<tt>extractvalue</tt>' Instruction</a>
3812 <div class="doc_text">
3816 <result> = extractvalue <aggregate type> <val>, <idx>{, <idx>}*
3820 <p>The '<tt>extractvalue</tt>' instruction extracts the value of a struct field
3821 or array element from an aggregate value.</p>
3824 <p>The first operand of an '<tt>extractvalue</tt>' instruction is a value
3825 of <a href="#t_struct">struct</a> or <a href="#t_array">array</a> type. The
3826 operands are constant indices to specify which value to extract in a similar
3827 manner as indices in a
3828 '<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.</p>
3831 <p>The result is the value at the position in the aggregate specified by the
3836 <result> = extractvalue {i32, float} %agg, 0 <i>; yields i32</i>
3841 <!-- _______________________________________________________________________ -->
3842 <div class="doc_subsubsection">
3843 <a name="i_insertvalue">'<tt>insertvalue</tt>' Instruction</a>
3846 <div class="doc_text">
3850 <result> = insertvalue <aggregate type> <val>, <ty> <val>, <idx> <i>; yields <n x <ty>></i>
3854 <p>The '<tt>insertvalue</tt>' instruction inserts a value into a struct field or
3855 array element in an aggregate.</p>
3859 <p>The first operand of an '<tt>insertvalue</tt>' instruction is a value
3860 of <a href="#t_struct">struct</a> or <a href="#t_array">array</a> type. The
3861 second operand is a first-class value to insert. The following operands are
3862 constant indices indicating the position at which to insert the value in a
3863 similar manner as indices in a
3864 '<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction. The
3865 value to insert must have the same type as the value identified by the
3869 <p>The result is an aggregate of the same type as <tt>val</tt>. Its value is
3870 that of <tt>val</tt> except that the value at the position specified by the
3871 indices is that of <tt>elt</tt>.</p>
3875 <result> = insertvalue {i32, float} %agg, i32 1, 0 <i>; yields {i32, float}</i>
3881 <!-- ======================================================================= -->
3882 <div class="doc_subsection">
3883 <a name="memoryops">Memory Access and Addressing Operations</a>
3886 <div class="doc_text">
3888 <p>A key design point of an SSA-based representation is how it represents
3889 memory. In LLVM, no memory locations are in SSA form, which makes things
3890 very simple. This section describes how to read, write, and allocate
3895 <!-- _______________________________________________________________________ -->
3896 <div class="doc_subsubsection">
3897 <a name="i_alloca">'<tt>alloca</tt>' Instruction</a>
3900 <div class="doc_text">
3904 <result> = alloca <type>[, i32 <NumElements>][, align <alignment>] <i>; yields {type*}:result</i>
3908 <p>The '<tt>alloca</tt>' instruction allocates memory on the stack frame of the
3909 currently executing function, to be automatically released when this function
3910 returns to its caller. The object is always allocated in the generic address
3911 space (address space zero).</p>
3914 <p>The '<tt>alloca</tt>' instruction
3915 allocates <tt>sizeof(<type>)*NumElements</tt> bytes of memory on the
3916 runtime stack, returning a pointer of the appropriate type to the program.
3917 If "NumElements" is specified, it is the number of elements allocated,
3918 otherwise "NumElements" is defaulted to be one. If a constant alignment is
3919 specified, the value result of the allocation is guaranteed to be aligned to
3920 at least that boundary. If not specified, or if zero, the target can choose
3921 to align the allocation on any convenient boundary compatible with the
3924 <p>'<tt>type</tt>' may be any sized type.</p>
3927 <p>Memory is allocated; a pointer is returned. The operation is undefined if
3928 there is insufficient stack space for the allocation. '<tt>alloca</tt>'d
3929 memory is automatically released when the function returns. The
3930 '<tt>alloca</tt>' instruction is commonly used to represent automatic
3931 variables that must have an address available. When the function returns
3932 (either with the <tt><a href="#i_ret">ret</a></tt>
3933 or <tt><a href="#i_unwind">unwind</a></tt> instructions), the memory is
3934 reclaimed. Allocating zero bytes is legal, but the result is undefined.</p>
3938 %ptr = alloca i32 <i>; yields {i32*}:ptr</i>
3939 %ptr = alloca i32, i32 4 <i>; yields {i32*}:ptr</i>
3940 %ptr = alloca i32, i32 4, align 1024 <i>; yields {i32*}:ptr</i>
3941 %ptr = alloca i32, align 1024 <i>; yields {i32*}:ptr</i>
3946 <!-- _______________________________________________________________________ -->
3947 <div class="doc_subsubsection"> <a name="i_load">'<tt>load</tt>'
3948 Instruction</a> </div>
3950 <div class="doc_text">
3954 <result> = load <ty>* <pointer>[, align <alignment>]
3955 <result> = volatile load <ty>* <pointer>[, align <alignment>]
3959 <p>The '<tt>load</tt>' instruction is used to read from memory.</p>
3962 <p>The argument to the '<tt>load</tt>' instruction specifies the memory address
3963 from which to load. The pointer must point to
3964 a <a href="#t_firstclass">first class</a> type. If the <tt>load</tt> is
3965 marked as <tt>volatile</tt>, then the optimizer is not allowed to modify the
3966 number or order of execution of this <tt>load</tt> with other
3967 volatile <tt>load</tt> and <tt><a href="#i_store">store</a></tt>
3970 <p>The optional constant "align" argument specifies the alignment of the
3971 operation (that is, the alignment of the memory address). A value of 0 or an
3972 omitted "align" argument means that the operation has the preferential
3973 alignment for the target. It is the responsibility of the code emitter to
3974 ensure that the alignment information is correct. Overestimating the
3975 alignment results in an undefined behavior. Underestimating the alignment may
3976 produce less efficient code. An alignment of 1 is always safe.</p>
3979 <p>The location of memory pointed to is loaded. If the value being loaded is of
3980 scalar type then the number of bytes read does not exceed the minimum number
3981 of bytes needed to hold all bits of the type. For example, loading an
3982 <tt>i24</tt> reads at most three bytes. When loading a value of a type like
3983 <tt>i20</tt> with a size that is not an integral number of bytes, the result
3984 is undefined if the value was not originally written using a store of the
3989 %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
3990 <a href="#i_store">store</a> i32 3, i32* %ptr <i>; yields {void}</i>
3991 %val = load i32* %ptr <i>; yields {i32}:val = i32 3</i>
3996 <!-- _______________________________________________________________________ -->
3997 <div class="doc_subsubsection"> <a name="i_store">'<tt>store</tt>'
3998 Instruction</a> </div>
4000 <div class="doc_text">
4004 store <ty> <value>, <ty>* <pointer>[, align <alignment>] <i>; yields {void}</i>
4005 volatile store <ty> <value>, <ty>* <pointer>[, align <alignment>] <i>; yields {void}</i>
4009 <p>The '<tt>store</tt>' instruction is used to write to memory.</p>
4012 <p>There are two arguments to the '<tt>store</tt>' instruction: a value to store
4013 and an address at which to store it. The type of the
4014 '<tt><pointer></tt>' operand must be a pointer to
4015 the <a href="#t_firstclass">first class</a> type of the
4016 '<tt><value></tt>' operand. If the <tt>store</tt> is marked
4017 as <tt>volatile</tt>, then the optimizer is not allowed to modify the number
4018 or order of execution of this <tt>store</tt> with other
4019 volatile <tt>load</tt> and <tt><a href="#i_store">store</a></tt>
4022 <p>The optional constant "align" argument specifies the alignment of the
4023 operation (that is, the alignment of the memory address). A value of 0 or an
4024 omitted "align" argument means that the operation has the preferential
4025 alignment for the target. It is the responsibility of the code emitter to
4026 ensure that the alignment information is correct. Overestimating the
4027 alignment results in an undefined behavior. Underestimating the alignment may
4028 produce less efficient code. An alignment of 1 is always safe.</p>
4031 <p>The contents of memory are updated to contain '<tt><value></tt>' at the
4032 location specified by the '<tt><pointer></tt>' operand. If
4033 '<tt><value></tt>' is of scalar type then the number of bytes written
4034 does not exceed the minimum number of bytes needed to hold all bits of the
4035 type. For example, storing an <tt>i24</tt> writes at most three bytes. When
4036 writing a value of a type like <tt>i20</tt> with a size that is not an
4037 integral number of bytes, it is unspecified what happens to the extra bits
4038 that do not belong to the type, but they will typically be overwritten.</p>
4042 %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
4043 store i32 3, i32* %ptr <i>; yields {void}</i>
4044 %val = <a href="#i_load">load</a> i32* %ptr <i>; yields {i32}:val = i32 3</i>
4049 <!-- _______________________________________________________________________ -->
4050 <div class="doc_subsubsection">
4051 <a name="i_getelementptr">'<tt>getelementptr</tt>' Instruction</a>
4054 <div class="doc_text">
4058 <result> = getelementptr <pty>* <ptrval>{, <ty> <idx>}*
4059 <result> = getelementptr inbounds <pty>* <ptrval>{, <ty> <idx>}*
4063 <p>The '<tt>getelementptr</tt>' instruction is used to get the address of a
4064 subelement of an aggregate data structure. It performs address calculation
4065 only and does not access memory.</p>
4068 <p>The first argument is always a pointer, and forms the basis of the
4069 calculation. The remaining arguments are indices that indicate which of the
4070 elements of the aggregate object are indexed. The interpretation of each
4071 index is dependent on the type being indexed into. The first index always
4072 indexes the pointer value given as the first argument, the second index
4073 indexes a value of the type pointed to (not necessarily the value directly
4074 pointed to, since the first index can be non-zero), etc. The first type
4075 indexed into must be a pointer value, subsequent types can be arrays, vectors
4076 and structs. Note that subsequent types being indexed into can never be
4077 pointers, since that would require loading the pointer before continuing
4080 <p>The type of each index argument depends on the type it is indexing into.
4081 When indexing into a (optionally packed) structure, only <tt>i32</tt> integer
4082 <b>constants</b> are allowed. When indexing into an array, pointer or
4083 vector, integers of any width are allowed, and they are not required to be
4086 <p>For example, let's consider a C code fragment and how it gets compiled to
4089 <div class="doc_code">
4102 int *foo(struct ST *s) {
4103 return &s[1].Z.B[5][13];
4108 <p>The LLVM code generated by the GCC frontend is:</p>
4110 <div class="doc_code">
4112 %RT = <a href="#namedtypes">type</a> { i8 , [10 x [20 x i32]], i8 }
4113 %ST = <a href="#namedtypes">type</a> { i32, double, %RT }
4115 define i32* @foo(%ST* %s) {
4117 %reg = getelementptr %ST* %s, i32 1, i32 2, i32 1, i32 5, i32 13
4124 <p>In the example above, the first index is indexing into the '<tt>%ST*</tt>'
4125 type, which is a pointer, yielding a '<tt>%ST</tt>' = '<tt>{ i32, double, %RT
4126 }</tt>' type, a structure. The second index indexes into the third element
4127 of the structure, yielding a '<tt>%RT</tt>' = '<tt>{ i8 , [10 x [20 x i32]],
4128 i8 }</tt>' type, another structure. The third index indexes into the second
4129 element of the structure, yielding a '<tt>[10 x [20 x i32]]</tt>' type, an
4130 array. The two dimensions of the array are subscripted into, yielding an
4131 '<tt>i32</tt>' type. The '<tt>getelementptr</tt>' instruction returns a
4132 pointer to this element, thus computing a value of '<tt>i32*</tt>' type.</p>
4134 <p>Note that it is perfectly legal to index partially through a structure,
4135 returning a pointer to an inner element. Because of this, the LLVM code for
4136 the given testcase is equivalent to:</p>
4139 define i32* @foo(%ST* %s) {
4140 %t1 = getelementptr %ST* %s, i32 1 <i>; yields %ST*:%t1</i>
4141 %t2 = getelementptr %ST* %t1, i32 0, i32 2 <i>; yields %RT*:%t2</i>
4142 %t3 = getelementptr %RT* %t2, i32 0, i32 1 <i>; yields [10 x [20 x i32]]*:%t3</i>
4143 %t4 = getelementptr [10 x [20 x i32]]* %t3, i32 0, i32 5 <i>; yields [20 x i32]*:%t4</i>
4144 %t5 = getelementptr [20 x i32]* %t4, i32 0, i32 13 <i>; yields i32*:%t5</i>
4149 <p>If the <tt>inbounds</tt> keyword is present, the result value of the
4150 <tt>getelementptr</tt> is undefined if the base pointer is not an
4151 <i>in bounds</i> address of an allocated object, or if any of the addresses
4152 that would be formed by successive addition of the offsets implied by the
4153 indices to the base address with infinitely precise arithmetic are not an
4154 <i>in bounds</i> address of that allocated object.
4155 The <i>in bounds</i> addresses for an allocated object are all the addresses
4156 that point into the object, plus the address one byte past the end.</p>
4158 <p>If the <tt>inbounds</tt> keyword is not present, the offsets are added to
4159 the base address with silently-wrapping two's complement arithmetic, and
4160 the result value of the <tt>getelementptr</tt> may be outside the object
4161 pointed to by the base pointer. The result value may not necessarily be
4162 used to access memory though, even if it happens to point into allocated
4163 storage. See the <a href="#pointeraliasing">Pointer Aliasing Rules</a>
4164 section for more information.</p>
4166 <p>The getelementptr instruction is often confusing. For some more insight into
4167 how it works, see <a href="GetElementPtr.html">the getelementptr FAQ</a>.</p>
4171 <i>; yields [12 x i8]*:aptr</i>
4172 %aptr = getelementptr {i32, [12 x i8]}* %saptr, i64 0, i32 1
4173 <i>; yields i8*:vptr</i>
4174 %vptr = getelementptr {i32, <2 x i8>}* %svptr, i64 0, i32 1, i32 1
4175 <i>; yields i8*:eptr</i>
4176 %eptr = getelementptr [12 x i8]* %aptr, i64 0, i32 1
4177 <i>; yields i32*:iptr</i>
4178 %iptr = getelementptr [10 x i32]* @arr, i16 0, i16 0
4183 <!-- ======================================================================= -->
4184 <div class="doc_subsection"> <a name="convertops">Conversion Operations</a>
4187 <div class="doc_text">
4189 <p>The instructions in this category are the conversion instructions (casting)
4190 which all take a single operand and a type. They perform various bit
4191 conversions on the operand.</p>
4195 <!-- _______________________________________________________________________ -->
4196 <div class="doc_subsubsection">
4197 <a name="i_trunc">'<tt>trunc .. to</tt>' Instruction</a>
4199 <div class="doc_text">
4203 <result> = trunc <ty> <value> to <ty2> <i>; yields ty2</i>
4207 <p>The '<tt>trunc</tt>' instruction truncates its operand to the
4208 type <tt>ty2</tt>.</p>
4211 <p>The '<tt>trunc</tt>' instruction takes a <tt>value</tt> to trunc, which must
4212 be an <a href="#t_integer">integer</a> type, and a type that specifies the
4213 size and type of the result, which must be
4214 an <a href="#t_integer">integer</a> type. The bit size of <tt>value</tt> must
4215 be larger than the bit size of <tt>ty2</tt>. Equal sized types are not
4219 <p>The '<tt>trunc</tt>' instruction truncates the high order bits
4220 in <tt>value</tt> and converts the remaining bits to <tt>ty2</tt>. Since the
4221 source size must be larger than the destination size, <tt>trunc</tt> cannot
4222 be a <i>no-op cast</i>. It will always truncate bits.</p>
4226 %X = trunc i32 257 to i8 <i>; yields i8:1</i>
4227 %Y = trunc i32 123 to i1 <i>; yields i1:true</i>
4228 %Z = trunc i32 122 to i1 <i>; yields i1:false</i>
4233 <!-- _______________________________________________________________________ -->
4234 <div class="doc_subsubsection">
4235 <a name="i_zext">'<tt>zext .. to</tt>' Instruction</a>
4237 <div class="doc_text">
4241 <result> = zext <ty> <value> to <ty2> <i>; yields ty2</i>
4245 <p>The '<tt>zext</tt>' instruction zero extends its operand to type
4250 <p>The '<tt>zext</tt>' instruction takes a value to cast, which must be of
4251 <a href="#t_integer">integer</a> type, and a type to cast it to, which must
4252 also be of <a href="#t_integer">integer</a> type. The bit size of the
4253 <tt>value</tt> must be smaller than the bit size of the destination type,
4257 <p>The <tt>zext</tt> fills the high order bits of the <tt>value</tt> with zero
4258 bits until it reaches the size of the destination type, <tt>ty2</tt>.</p>
4260 <p>When zero extending from i1, the result will always be either 0 or 1.</p>
4264 %X = zext i32 257 to i64 <i>; yields i64:257</i>
4265 %Y = zext i1 true to i32 <i>; yields i32:1</i>
4270 <!-- _______________________________________________________________________ -->
4271 <div class="doc_subsubsection">
4272 <a name="i_sext">'<tt>sext .. to</tt>' Instruction</a>
4274 <div class="doc_text">
4278 <result> = sext <ty> <value> to <ty2> <i>; yields ty2</i>
4282 <p>The '<tt>sext</tt>' sign extends <tt>value</tt> to the type <tt>ty2</tt>.</p>
4285 <p>The '<tt>sext</tt>' instruction takes a value to cast, which must be of
4286 <a href="#t_integer">integer</a> type, and a type to cast it to, which must
4287 also be of <a href="#t_integer">integer</a> type. The bit size of the
4288 <tt>value</tt> must be smaller than the bit size of the destination type,
4292 <p>The '<tt>sext</tt>' instruction performs a sign extension by copying the sign
4293 bit (highest order bit) of the <tt>value</tt> until it reaches the bit size
4294 of the type <tt>ty2</tt>.</p>
4296 <p>When sign extending from i1, the extension always results in -1 or 0.</p>
4300 %X = sext i8 -1 to i16 <i>; yields i16 :65535</i>
4301 %Y = sext i1 true to i32 <i>; yields i32:-1</i>
4306 <!-- _______________________________________________________________________ -->
4307 <div class="doc_subsubsection">
4308 <a name="i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a>
4311 <div class="doc_text">
4315 <result> = fptrunc <ty> <value> to <ty2> <i>; yields ty2</i>
4319 <p>The '<tt>fptrunc</tt>' instruction truncates <tt>value</tt> to type
4323 <p>The '<tt>fptrunc</tt>' instruction takes a <a href="#t_floating">floating
4324 point</a> value to cast and a <a href="#t_floating">floating point</a> type
4325 to cast it to. The size of <tt>value</tt> must be larger than the size of
4326 <tt>ty2</tt>. This implies that <tt>fptrunc</tt> cannot be used to make a
4327 <i>no-op cast</i>.</p>
4330 <p>The '<tt>fptrunc</tt>' instruction truncates a <tt>value</tt> from a larger
4331 <a href="#t_floating">floating point</a> type to a smaller
4332 <a href="#t_floating">floating point</a> type. If the value cannot fit
4333 within the destination type, <tt>ty2</tt>, then the results are
4338 %X = fptrunc double 123.0 to float <i>; yields float:123.0</i>
4339 %Y = fptrunc double 1.0E+300 to float <i>; yields undefined</i>
4344 <!-- _______________________________________________________________________ -->
4345 <div class="doc_subsubsection">
4346 <a name="i_fpext">'<tt>fpext .. to</tt>' Instruction</a>
4348 <div class="doc_text">
4352 <result> = fpext <ty> <value> to <ty2> <i>; yields ty2</i>
4356 <p>The '<tt>fpext</tt>' extends a floating point <tt>value</tt> to a larger
4357 floating point value.</p>
4360 <p>The '<tt>fpext</tt>' instruction takes a
4361 <a href="#t_floating">floating point</a> <tt>value</tt> to cast, and
4362 a <a href="#t_floating">floating point</a> type to cast it to. The source
4363 type must be smaller than the destination type.</p>
4366 <p>The '<tt>fpext</tt>' instruction extends the <tt>value</tt> from a smaller
4367 <a href="#t_floating">floating point</a> type to a larger
4368 <a href="#t_floating">floating point</a> type. The <tt>fpext</tt> cannot be
4369 used to make a <i>no-op cast</i> because it always changes bits. Use
4370 <tt>bitcast</tt> to make a <i>no-op cast</i> for a floating point cast.</p>
4374 %X = fpext float 3.1415 to double <i>; yields double:3.1415</i>
4375 %Y = fpext float 1.0 to float <i>; yields float:1.0 (no-op)</i>
4380 <!-- _______________________________________________________________________ -->
4381 <div class="doc_subsubsection">
4382 <a name="i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a>
4384 <div class="doc_text">
4388 <result> = fptoui <ty> <value> to <ty2> <i>; yields ty2</i>
4392 <p>The '<tt>fptoui</tt>' converts a floating point <tt>value</tt> to its
4393 unsigned integer equivalent of type <tt>ty2</tt>.</p>
4396 <p>The '<tt>fptoui</tt>' instruction takes a value to cast, which must be a
4397 scalar or vector <a href="#t_floating">floating point</a> value, and a type
4398 to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
4399 type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
4400 vector integer type with the same number of elements as <tt>ty</tt></p>
4403 <p>The '<tt>fptoui</tt>' instruction converts its
4404 <a href="#t_floating">floating point</a> operand into the nearest (rounding
4405 towards zero) unsigned integer value. If the value cannot fit
4406 in <tt>ty2</tt>, the results are undefined.</p>
4410 %X = fptoui double 123.0 to i32 <i>; yields i32:123</i>
4411 %Y = fptoui float 1.0E+300 to i1 <i>; yields undefined:1</i>
4412 %Z = fptoui float 1.04E+17 to i8 <i>; yields undefined:1</i>
4417 <!-- _______________________________________________________________________ -->
4418 <div class="doc_subsubsection">
4419 <a name="i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a>
4421 <div class="doc_text">
4425 <result> = fptosi <ty> <value> to <ty2> <i>; yields ty2</i>
4429 <p>The '<tt>fptosi</tt>' instruction converts
4430 <a href="#t_floating">floating point</a> <tt>value</tt> to
4431 type <tt>ty2</tt>.</p>
4434 <p>The '<tt>fptosi</tt>' instruction takes a value to cast, which must be a
4435 scalar or vector <a href="#t_floating">floating point</a> value, and a type
4436 to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
4437 type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
4438 vector integer type with the same number of elements as <tt>ty</tt></p>
4441 <p>The '<tt>fptosi</tt>' instruction converts its
4442 <a href="#t_floating">floating point</a> operand into the nearest (rounding
4443 towards zero) signed integer value. If the value cannot fit in <tt>ty2</tt>,
4444 the results are undefined.</p>
4448 %X = fptosi double -123.0 to i32 <i>; yields i32:-123</i>
4449 %Y = fptosi float 1.0E-247 to i1 <i>; yields undefined:1</i>
4450 %Z = fptosi float 1.04E+17 to i8 <i>; yields undefined:1</i>
4455 <!-- _______________________________________________________________________ -->
4456 <div class="doc_subsubsection">
4457 <a name="i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a>
4459 <div class="doc_text">
4463 <result> = uitofp <ty> <value> to <ty2> <i>; yields ty2</i>
4467 <p>The '<tt>uitofp</tt>' instruction regards <tt>value</tt> as an unsigned
4468 integer and converts that value to the <tt>ty2</tt> type.</p>
4471 <p>The '<tt>uitofp</tt>' instruction takes a value to cast, which must be a
4472 scalar or vector <a href="#t_integer">integer</a> value, and a type to cast
4473 it to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
4474 type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
4475 floating point type with the same number of elements as <tt>ty</tt></p>
4478 <p>The '<tt>uitofp</tt>' instruction interprets its operand as an unsigned
4479 integer quantity and converts it to the corresponding floating point
4480 value. If the value cannot fit in the floating point value, the results are
4485 %X = uitofp i32 257 to float <i>; yields float:257.0</i>
4486 %Y = uitofp i8 -1 to double <i>; yields double:255.0</i>
4491 <!-- _______________________________________________________________________ -->
4492 <div class="doc_subsubsection">
4493 <a name="i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a>
4495 <div class="doc_text">
4499 <result> = sitofp <ty> <value> to <ty2> <i>; yields ty2</i>
4503 <p>The '<tt>sitofp</tt>' instruction regards <tt>value</tt> as a signed integer
4504 and converts that value to the <tt>ty2</tt> type.</p>
4507 <p>The '<tt>sitofp</tt>' instruction takes a value to cast, which must be a
4508 scalar or vector <a href="#t_integer">integer</a> value, and a type to cast
4509 it to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
4510 type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
4511 floating point type with the same number of elements as <tt>ty</tt></p>
4514 <p>The '<tt>sitofp</tt>' instruction interprets its operand as a signed integer
4515 quantity and converts it to the corresponding floating point value. If the
4516 value cannot fit in the floating point value, the results are undefined.</p>
4520 %X = sitofp i32 257 to float <i>; yields float:257.0</i>
4521 %Y = sitofp i8 -1 to double <i>; yields double:-1.0</i>
4526 <!-- _______________________________________________________________________ -->
4527 <div class="doc_subsubsection">
4528 <a name="i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a>
4530 <div class="doc_text">
4534 <result> = ptrtoint <ty> <value> to <ty2> <i>; yields ty2</i>
4538 <p>The '<tt>ptrtoint</tt>' instruction converts the pointer <tt>value</tt> to
4539 the integer type <tt>ty2</tt>.</p>
4542 <p>The '<tt>ptrtoint</tt>' instruction takes a <tt>value</tt> to cast, which
4543 must be a <a href="#t_pointer">pointer</a> value, and a type to cast it to
4544 <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a> type.</p>
4547 <p>The '<tt>ptrtoint</tt>' instruction converts <tt>value</tt> to integer type
4548 <tt>ty2</tt> by interpreting the pointer value as an integer and either
4549 truncating or zero extending that value to the size of the integer type. If
4550 <tt>value</tt> is smaller than <tt>ty2</tt> then a zero extension is done. If
4551 <tt>value</tt> is larger than <tt>ty2</tt> then a truncation is done. If they
4552 are the same size, then nothing is done (<i>no-op cast</i>) other than a type
4557 %X = ptrtoint i32* %X to i8 <i>; yields truncation on 32-bit architecture</i>
4558 %Y = ptrtoint i32* %x to i64 <i>; yields zero extension on 32-bit architecture</i>
4563 <!-- _______________________________________________________________________ -->
4564 <div class="doc_subsubsection">
4565 <a name="i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a>
4567 <div class="doc_text">
4571 <result> = inttoptr <ty> <value> to <ty2> <i>; yields ty2</i>
4575 <p>The '<tt>inttoptr</tt>' instruction converts an integer <tt>value</tt> to a
4576 pointer type, <tt>ty2</tt>.</p>
4579 <p>The '<tt>inttoptr</tt>' instruction takes an <a href="#t_integer">integer</a>
4580 value to cast, and a type to cast it to, which must be a
4581 <a href="#t_pointer">pointer</a> type.</p>
4584 <p>The '<tt>inttoptr</tt>' instruction converts <tt>value</tt> to type
4585 <tt>ty2</tt> by applying either a zero extension or a truncation depending on
4586 the size of the integer <tt>value</tt>. If <tt>value</tt> is larger than the
4587 size of a pointer then a truncation is done. If <tt>value</tt> is smaller
4588 than the size of a pointer then a zero extension is done. If they are the
4589 same size, nothing is done (<i>no-op cast</i>).</p>
4593 %X = inttoptr i32 255 to i32* <i>; yields zero extension on 64-bit architecture</i>
4594 %Y = inttoptr i32 255 to i32* <i>; yields no-op on 32-bit architecture</i>
4595 %Z = inttoptr i64 0 to i32* <i>; yields truncation on 32-bit architecture</i>
4600 <!-- _______________________________________________________________________ -->
4601 <div class="doc_subsubsection">
4602 <a name="i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a>
4604 <div class="doc_text">
4608 <result> = bitcast <ty> <value> to <ty2> <i>; yields ty2</i>
4612 <p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
4613 <tt>ty2</tt> without changing any bits.</p>
4616 <p>The '<tt>bitcast</tt>' instruction takes a value to cast, which must be a
4617 non-aggregate first class value, and a type to cast it to, which must also be
4618 a non-aggregate <a href="#t_firstclass">first class</a> type. The bit sizes
4619 of <tt>value</tt> and the destination type, <tt>ty2</tt>, must be
4620 identical. If the source type is a pointer, the destination type must also be
4621 a pointer. This instruction supports bitwise conversion of vectors to
4622 integers and to vectors of other types (as long as they have the same
4626 <p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
4627 <tt>ty2</tt>. It is always a <i>no-op cast</i> because no bits change with
4628 this conversion. The conversion is done as if the <tt>value</tt> had been
4629 stored to memory and read back as type <tt>ty2</tt>. Pointer types may only
4630 be converted to other pointer types with this instruction. To convert
4631 pointers to other types, use the <a href="#i_inttoptr">inttoptr</a> or
4632 <a href="#i_ptrtoint">ptrtoint</a> instructions first.</p>
4636 %X = bitcast i8 255 to i8 <i>; yields i8 :-1</i>
4637 %Y = bitcast i32* %x to sint* <i>; yields sint*:%x</i>
4638 %Z = bitcast <2 x int> %V to i64; <i>; yields i64: %V</i>
4643 <!-- ======================================================================= -->
4644 <div class="doc_subsection"> <a name="otherops">Other Operations</a> </div>
4646 <div class="doc_text">
4648 <p>The instructions in this category are the "miscellaneous" instructions, which
4649 defy better classification.</p>
4653 <!-- _______________________________________________________________________ -->
4654 <div class="doc_subsubsection"><a name="i_icmp">'<tt>icmp</tt>' Instruction</a>
4657 <div class="doc_text">
4661 <result> = icmp <cond> <ty> <op1>, <op2> <i>; yields {i1} or {<N x i1>}:result</i>
4665 <p>The '<tt>icmp</tt>' instruction returns a boolean value or a vector of
4666 boolean values based on comparison of its two integer, integer vector, or
4667 pointer operands.</p>
4670 <p>The '<tt>icmp</tt>' instruction takes three operands. The first operand is
4671 the condition code indicating the kind of comparison to perform. It is not a
4672 value, just a keyword. The possible condition code are:</p>
4675 <li><tt>eq</tt>: equal</li>
4676 <li><tt>ne</tt>: not equal </li>
4677 <li><tt>ugt</tt>: unsigned greater than</li>
4678 <li><tt>uge</tt>: unsigned greater or equal</li>
4679 <li><tt>ult</tt>: unsigned less than</li>
4680 <li><tt>ule</tt>: unsigned less or equal</li>
4681 <li><tt>sgt</tt>: signed greater than</li>
4682 <li><tt>sge</tt>: signed greater or equal</li>
4683 <li><tt>slt</tt>: signed less than</li>
4684 <li><tt>sle</tt>: signed less or equal</li>
4687 <p>The remaining two arguments must be <a href="#t_integer">integer</a> or
4688 <a href="#t_pointer">pointer</a> or integer <a href="#t_vector">vector</a>
4689 typed. They must also be identical types.</p>
4692 <p>The '<tt>icmp</tt>' compares <tt>op1</tt> and <tt>op2</tt> according to the
4693 condition code given as <tt>cond</tt>. The comparison performed always yields
4694 either an <a href="#t_integer"><tt>i1</tt></a> or vector of <tt>i1</tt>
4695 result, as follows:</p>
4698 <li><tt>eq</tt>: yields <tt>true</tt> if the operands are equal,
4699 <tt>false</tt> otherwise. No sign interpretation is necessary or
4702 <li><tt>ne</tt>: yields <tt>true</tt> if the operands are unequal,
4703 <tt>false</tt> otherwise. No sign interpretation is necessary or
4706 <li><tt>ugt</tt>: interprets the operands as unsigned values and yields
4707 <tt>true</tt> if <tt>op1</tt> is greater than <tt>op2</tt>.</li>
4709 <li><tt>uge</tt>: interprets the operands as unsigned values and yields
4710 <tt>true</tt> if <tt>op1</tt> is greater than or equal
4711 to <tt>op2</tt>.</li>
4713 <li><tt>ult</tt>: interprets the operands as unsigned values and yields
4714 <tt>true</tt> if <tt>op1</tt> is less than <tt>op2</tt>.</li>
4716 <li><tt>ule</tt>: interprets the operands as unsigned values and yields
4717 <tt>true</tt> if <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
4719 <li><tt>sgt</tt>: interprets the operands as signed values and yields
4720 <tt>true</tt> if <tt>op1</tt> is greater than <tt>op2</tt>.</li>
4722 <li><tt>sge</tt>: interprets the operands as signed values and yields
4723 <tt>true</tt> if <tt>op1</tt> is greater than or equal
4724 to <tt>op2</tt>.</li>
4726 <li><tt>slt</tt>: interprets the operands as signed values and yields
4727 <tt>true</tt> if <tt>op1</tt> is less than <tt>op2</tt>.</li>
4729 <li><tt>sle</tt>: interprets the operands as signed values and yields
4730 <tt>true</tt> if <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
4733 <p>If the operands are <a href="#t_pointer">pointer</a> typed, the pointer
4734 values are compared as if they were integers.</p>
4736 <p>If the operands are integer vectors, then they are compared element by
4737 element. The result is an <tt>i1</tt> vector with the same number of elements
4738 as the values being compared. Otherwise, the result is an <tt>i1</tt>.</p>
4742 <result> = icmp eq i32 4, 5 <i>; yields: result=false</i>
4743 <result> = icmp ne float* %X, %X <i>; yields: result=false</i>
4744 <result> = icmp ult i16 4, 5 <i>; yields: result=true</i>
4745 <result> = icmp sgt i16 4, 5 <i>; yields: result=false</i>
4746 <result> = icmp ule i16 -4, 5 <i>; yields: result=false</i>
4747 <result> = icmp sge i16 4, 5 <i>; yields: result=false</i>
4750 <p>Note that the code generator does not yet support vector types with
4751 the <tt>icmp</tt> instruction.</p>
4755 <!-- _______________________________________________________________________ -->
4756 <div class="doc_subsubsection"><a name="i_fcmp">'<tt>fcmp</tt>' Instruction</a>
4759 <div class="doc_text">
4763 <result> = fcmp <cond> <ty> <op1>, <op2> <i>; yields {i1} or {<N x i1>}:result</i>
4767 <p>The '<tt>fcmp</tt>' instruction returns a boolean value or vector of boolean
4768 values based on comparison of its operands.</p>
4770 <p>If the operands are floating point scalars, then the result type is a boolean
4771 (<a href="#t_integer"><tt>i1</tt></a>).</p>
4773 <p>If the operands are floating point vectors, then the result type is a vector
4774 of boolean with the same number of elements as the operands being
4778 <p>The '<tt>fcmp</tt>' instruction takes three operands. The first operand is
4779 the condition code indicating the kind of comparison to perform. It is not a
4780 value, just a keyword. The possible condition code are:</p>
4783 <li><tt>false</tt>: no comparison, always returns false</li>
4784 <li><tt>oeq</tt>: ordered and equal</li>
4785 <li><tt>ogt</tt>: ordered and greater than </li>
4786 <li><tt>oge</tt>: ordered and greater than or equal</li>
4787 <li><tt>olt</tt>: ordered and less than </li>
4788 <li><tt>ole</tt>: ordered and less than or equal</li>
4789 <li><tt>one</tt>: ordered and not equal</li>
4790 <li><tt>ord</tt>: ordered (no nans)</li>
4791 <li><tt>ueq</tt>: unordered or equal</li>
4792 <li><tt>ugt</tt>: unordered or greater than </li>
4793 <li><tt>uge</tt>: unordered or greater than or equal</li>
4794 <li><tt>ult</tt>: unordered or less than </li>
4795 <li><tt>ule</tt>: unordered or less than or equal</li>
4796 <li><tt>une</tt>: unordered or not equal</li>
4797 <li><tt>uno</tt>: unordered (either nans)</li>
4798 <li><tt>true</tt>: no comparison, always returns true</li>
4801 <p><i>Ordered</i> means that neither operand is a QNAN while
4802 <i>unordered</i> means that either operand may be a QNAN.</p>
4804 <p>Each of <tt>val1</tt> and <tt>val2</tt> arguments must be either
4805 a <a href="#t_floating">floating point</a> type or
4806 a <a href="#t_vector">vector</a> of floating point type. They must have
4807 identical types.</p>
4810 <p>The '<tt>fcmp</tt>' instruction compares <tt>op1</tt> and <tt>op2</tt>
4811 according to the condition code given as <tt>cond</tt>. If the operands are
4812 vectors, then the vectors are compared element by element. Each comparison
4813 performed always yields an <a href="#t_integer">i1</a> result, as
4817 <li><tt>false</tt>: always yields <tt>false</tt>, regardless of operands.</li>
4819 <li><tt>oeq</tt>: yields <tt>true</tt> if both operands are not a QNAN and
4820 <tt>op1</tt> is equal to <tt>op2</tt>.</li>
4822 <li><tt>ogt</tt>: yields <tt>true</tt> if both operands are not a QNAN and
4823 <tt>op1</tt> is greather than <tt>op2</tt>.</li>
4825 <li><tt>oge</tt>: yields <tt>true</tt> if both operands are not a QNAN and
4826 <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
4828 <li><tt>olt</tt>: yields <tt>true</tt> if both operands are not a QNAN and
4829 <tt>op1</tt> is less than <tt>op2</tt>.</li>
4831 <li><tt>ole</tt>: yields <tt>true</tt> if both operands are not a QNAN and
4832 <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
4834 <li><tt>one</tt>: yields <tt>true</tt> if both operands are not a QNAN and
4835 <tt>op1</tt> is not equal to <tt>op2</tt>.</li>
4837 <li><tt>ord</tt>: yields <tt>true</tt> if both operands are not a QNAN.</li>
4839 <li><tt>ueq</tt>: yields <tt>true</tt> if either operand is a QNAN or
4840 <tt>op1</tt> is equal to <tt>op2</tt>.</li>
4842 <li><tt>ugt</tt>: yields <tt>true</tt> if either operand is a QNAN or
4843 <tt>op1</tt> is greater than <tt>op2</tt>.</li>
4845 <li><tt>uge</tt>: yields <tt>true</tt> if either operand is a QNAN or
4846 <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
4848 <li><tt>ult</tt>: yields <tt>true</tt> if either operand is a QNAN or
4849 <tt>op1</tt> is less than <tt>op2</tt>.</li>
4851 <li><tt>ule</tt>: yields <tt>true</tt> if either operand is a QNAN or
4852 <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
4854 <li><tt>une</tt>: yields <tt>true</tt> if either operand is a QNAN or
4855 <tt>op1</tt> is not equal to <tt>op2</tt>.</li>
4857 <li><tt>uno</tt>: yields <tt>true</tt> if either operand is a QNAN.</li>
4859 <li><tt>true</tt>: always yields <tt>true</tt>, regardless of operands.</li>
4864 <result> = fcmp oeq float 4.0, 5.0 <i>; yields: result=false</i>
4865 <result> = fcmp one float 4.0, 5.0 <i>; yields: result=true</i>
4866 <result> = fcmp olt float 4.0, 5.0 <i>; yields: result=true</i>
4867 <result> = fcmp ueq double 1.0, 2.0 <i>; yields: result=false</i>
4870 <p>Note that the code generator does not yet support vector types with
4871 the <tt>fcmp</tt> instruction.</p>
4875 <!-- _______________________________________________________________________ -->
4876 <div class="doc_subsubsection">
4877 <a name="i_phi">'<tt>phi</tt>' Instruction</a>
4880 <div class="doc_text">
4884 <result> = phi <ty> [ <val0>, <label0>], ...
4888 <p>The '<tt>phi</tt>' instruction is used to implement the φ node in the
4889 SSA graph representing the function.</p>
4892 <p>The type of the incoming values is specified with the first type field. After
4893 this, the '<tt>phi</tt>' instruction takes a list of pairs as arguments, with
4894 one pair for each predecessor basic block of the current block. Only values
4895 of <a href="#t_firstclass">first class</a> type may be used as the value
4896 arguments to the PHI node. Only labels may be used as the label
4899 <p>There must be no non-phi instructions between the start of a basic block and
4900 the PHI instructions: i.e. PHI instructions must be first in a basic
4903 <p>For the purposes of the SSA form, the use of each incoming value is deemed to
4904 occur on the edge from the corresponding predecessor block to the current
4905 block (but after any definition of an '<tt>invoke</tt>' instruction's return
4906 value on the same edge).</p>
4909 <p>At runtime, the '<tt>phi</tt>' instruction logically takes on the value
4910 specified by the pair corresponding to the predecessor basic block that
4911 executed just prior to the current block.</p>
4915 Loop: ; Infinite loop that counts from 0 on up...
4916 %indvar = phi i32 [ 0, %LoopHeader ], [ %nextindvar, %Loop ]
4917 %nextindvar = add i32 %indvar, 1
4923 <!-- _______________________________________________________________________ -->
4924 <div class="doc_subsubsection">
4925 <a name="i_select">'<tt>select</tt>' Instruction</a>
4928 <div class="doc_text">
4932 <result> = select <i>selty</i> <cond>, <ty> <val1>, <ty> <val2> <i>; yields ty</i>
4934 <i>selty</i> is either i1 or {<N x i1>}
4938 <p>The '<tt>select</tt>' instruction is used to choose one value based on a
4939 condition, without branching.</p>
4943 <p>The '<tt>select</tt>' instruction requires an 'i1' value or a vector of 'i1'
4944 values indicating the condition, and two values of the
4945 same <a href="#t_firstclass">first class</a> type. If the val1/val2 are
4946 vectors and the condition is a scalar, then entire vectors are selected, not
4947 individual elements.</p>
4950 <p>If the condition is an i1 and it evaluates to 1, the instruction returns the
4951 first value argument; otherwise, it returns the second value argument.</p>
4953 <p>If the condition is a vector of i1, then the value arguments must be vectors
4954 of the same size, and the selection is done element by element.</p>
4958 %X = select i1 true, i8 17, i8 42 <i>; yields i8:17</i>
4961 <p>Note that the code generator does not yet support conditions
4962 with vector type.</p>
4966 <!-- _______________________________________________________________________ -->
4967 <div class="doc_subsubsection">
4968 <a name="i_call">'<tt>call</tt>' Instruction</a>
4971 <div class="doc_text">
4975 <result> = [tail] call [<a href="#callingconv">cconv</a>] [<a href="#paramattrs">ret attrs</a>] <ty> [<fnty>*] <fnptrval>(<function args>) [<a href="#fnattrs">fn attrs</a>]
4979 <p>The '<tt>call</tt>' instruction represents a simple function call.</p>
4982 <p>This instruction requires several arguments:</p>
4985 <li>The optional "tail" marker indicates whether the callee function accesses
4986 any allocas or varargs in the caller. If the "tail" marker is present,
4987 the function call is eligible for tail call optimization. Note that calls
4988 may be marked "tail" even if they do not occur before
4989 a <a href="#i_ret"><tt>ret</tt></a> instruction.</li>
4991 <li>The optional "cconv" marker indicates which <a href="#callingconv">calling
4992 convention</a> the call should use. If none is specified, the call
4993 defaults to using C calling conventions.</li>
4995 <li>The optional <a href="#paramattrs">Parameter Attributes</a> list for
4996 return values. Only '<tt>zeroext</tt>', '<tt>signext</tt>', and
4997 '<tt>inreg</tt>' attributes are valid here.</li>
4999 <li>'<tt>ty</tt>': the type of the call instruction itself which is also the
5000 type of the return value. Functions that return no value are marked
5001 <tt><a href="#t_void">void</a></tt>.</li>
5003 <li>'<tt>fnty</tt>': shall be the signature of the pointer to function value
5004 being invoked. The argument types must match the types implied by this
5005 signature. This type can be omitted if the function is not varargs and if
5006 the function type does not return a pointer to a function.</li>
5008 <li>'<tt>fnptrval</tt>': An LLVM value containing a pointer to a function to
5009 be invoked. In most cases, this is a direct function invocation, but
5010 indirect <tt>call</tt>s are just as possible, calling an arbitrary pointer
5011 to function value.</li>
5013 <li>'<tt>function args</tt>': argument list whose types match the function
5014 signature argument types. All arguments must be of
5015 <a href="#t_firstclass">first class</a> type. If the function signature
5016 indicates the function accepts a variable number of arguments, the extra
5017 arguments can be specified.</li>
5019 <li>The optional <a href="#fnattrs">function attributes</a> list. Only
5020 '<tt>noreturn</tt>', '<tt>nounwind</tt>', '<tt>readonly</tt>' and
5021 '<tt>readnone</tt>' attributes are valid here.</li>
5025 <p>The '<tt>call</tt>' instruction is used to cause control flow to transfer to
5026 a specified function, with its incoming arguments bound to the specified
5027 values. Upon a '<tt><a href="#i_ret">ret</a></tt>' instruction in the called
5028 function, control flow continues with the instruction after the function
5029 call, and the return value of the function is bound to the result
5034 %retval = call i32 @test(i32 %argc)
5035 call i32 (i8 *, ...)* @printf(i8 * %msg, i32 12, i8 42) <i>; yields i32</i>
5036 %X = tail call i32 @foo() <i>; yields i32</i>
5037 %Y = tail call <a href="#callingconv">fastcc</a> i32 @foo() <i>; yields i32</i>
5038 call void %foo(i8 97 signext)
5040 %struct.A = type { i32, i8 }
5041 %r = call %struct.A @foo() <i>; yields { 32, i8 }</i>
5042 %gr = extractvalue %struct.A %r, 0 <i>; yields i32</i>
5043 %gr1 = extractvalue %struct.A %r, 1 <i>; yields i8</i>
5044 %Z = call void @foo() noreturn <i>; indicates that %foo never returns normally</i>
5045 %ZZ = call zeroext i32 @bar() <i>; Return value is %zero extended</i>
5048 <p>llvm treats calls to some functions with names and arguments that match the
5049 standard C99 library as being the C99 library functions, and may perform
5050 optimizations or generate code for them under that assumption. This is
5051 something we'd like to change in the future to provide better support for
5052 freestanding environments and non-C-based langauges.</p>
5056 <!-- _______________________________________________________________________ -->
5057 <div class="doc_subsubsection">
5058 <a name="i_va_arg">'<tt>va_arg</tt>' Instruction</a>
5061 <div class="doc_text">
5065 <resultval> = va_arg <va_list*> <arglist>, <argty>
5069 <p>The '<tt>va_arg</tt>' instruction is used to access arguments passed through
5070 the "variable argument" area of a function call. It is used to implement the
5071 <tt>va_arg</tt> macro in C.</p>
5074 <p>This instruction takes a <tt>va_list*</tt> value and the type of the
5075 argument. It returns a value of the specified argument type and increments
5076 the <tt>va_list</tt> to point to the next argument. The actual type
5077 of <tt>va_list</tt> is target specific.</p>
5080 <p>The '<tt>va_arg</tt>' instruction loads an argument of the specified type
5081 from the specified <tt>va_list</tt> and causes the <tt>va_list</tt> to point
5082 to the next argument. For more information, see the variable argument
5083 handling <a href="#int_varargs">Intrinsic Functions</a>.</p>
5085 <p>It is legal for this instruction to be called in a function which does not
5086 take a variable number of arguments, for example, the <tt>vfprintf</tt>
5089 <p><tt>va_arg</tt> is an LLVM instruction instead of
5090 an <a href="#intrinsics">intrinsic function</a> because it takes a type as an
5094 <p>See the <a href="#int_varargs">variable argument processing</a> section.</p>
5096 <p>Note that the code generator does not yet fully support va_arg on many
5097 targets. Also, it does not currently support va_arg with aggregate types on
5102 <!-- *********************************************************************** -->
5103 <div class="doc_section"> <a name="intrinsics">Intrinsic Functions</a> </div>
5104 <!-- *********************************************************************** -->
5106 <div class="doc_text">
5108 <p>LLVM supports the notion of an "intrinsic function". These functions have
5109 well known names and semantics and are required to follow certain
5110 restrictions. Overall, these intrinsics represent an extension mechanism for
5111 the LLVM language that does not require changing all of the transformations
5112 in LLVM when adding to the language (or the bitcode reader/writer, the
5113 parser, etc...).</p>
5115 <p>Intrinsic function names must all start with an "<tt>llvm.</tt>" prefix. This
5116 prefix is reserved in LLVM for intrinsic names; thus, function names may not
5117 begin with this prefix. Intrinsic functions must always be external
5118 functions: you cannot define the body of intrinsic functions. Intrinsic
5119 functions may only be used in call or invoke instructions: it is illegal to
5120 take the address of an intrinsic function. Additionally, because intrinsic
5121 functions are part of the LLVM language, it is required if any are added that
5122 they be documented here.</p>
5124 <p>Some intrinsic functions can be overloaded, i.e., the intrinsic represents a
5125 family of functions that perform the same operation but on different data
5126 types. Because LLVM can represent over 8 million different integer types,
5127 overloading is used commonly to allow an intrinsic function to operate on any
5128 integer type. One or more of the argument types or the result type can be
5129 overloaded to accept any integer type. Argument types may also be defined as
5130 exactly matching a previous argument's type or the result type. This allows
5131 an intrinsic function which accepts multiple arguments, but needs all of them
5132 to be of the same type, to only be overloaded with respect to a single
5133 argument or the result.</p>
5135 <p>Overloaded intrinsics will have the names of its overloaded argument types
5136 encoded into its function name, each preceded by a period. Only those types
5137 which are overloaded result in a name suffix. Arguments whose type is matched
5138 against another type do not. For example, the <tt>llvm.ctpop</tt> function
5139 can take an integer of any width and returns an integer of exactly the same
5140 integer width. This leads to a family of functions such as
5141 <tt>i8 @llvm.ctpop.i8(i8 %val)</tt> and <tt>i29 @llvm.ctpop.i29(i29
5142 %val)</tt>. Only one type, the return type, is overloaded, and only one type
5143 suffix is required. Because the argument's type is matched against the return
5144 type, it does not require its own name suffix.</p>
5146 <p>To learn how to add an intrinsic function, please see the
5147 <a href="ExtendingLLVM.html">Extending LLVM Guide</a>.</p>
5151 <!-- ======================================================================= -->
5152 <div class="doc_subsection">
5153 <a name="int_varargs">Variable Argument Handling Intrinsics</a>
5156 <div class="doc_text">
5158 <p>Variable argument support is defined in LLVM with
5159 the <a href="#i_va_arg"><tt>va_arg</tt></a> instruction and these three
5160 intrinsic functions. These functions are related to the similarly named
5161 macros defined in the <tt><stdarg.h></tt> header file.</p>
5163 <p>All of these functions operate on arguments that use a target-specific value
5164 type "<tt>va_list</tt>". The LLVM assembly language reference manual does
5165 not define what this type is, so all transformations should be prepared to
5166 handle these functions regardless of the type used.</p>
5168 <p>This example shows how the <a href="#i_va_arg"><tt>va_arg</tt></a>
5169 instruction and the variable argument handling intrinsic functions are
5172 <div class="doc_code">
5174 define i32 @test(i32 %X, ...) {
5175 ; Initialize variable argument processing
5177 %ap2 = bitcast i8** %ap to i8*
5178 call void @llvm.va_start(i8* %ap2)
5180 ; Read a single integer argument
5181 %tmp = va_arg i8** %ap, i32
5183 ; Demonstrate usage of llvm.va_copy and llvm.va_end
5185 %aq2 = bitcast i8** %aq to i8*
5186 call void @llvm.va_copy(i8* %aq2, i8* %ap2)
5187 call void @llvm.va_end(i8* %aq2)
5189 ; Stop processing of arguments.
5190 call void @llvm.va_end(i8* %ap2)
5194 declare void @llvm.va_start(i8*)
5195 declare void @llvm.va_copy(i8*, i8*)
5196 declare void @llvm.va_end(i8*)
5202 <!-- _______________________________________________________________________ -->
5203 <div class="doc_subsubsection">
5204 <a name="int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a>
5208 <div class="doc_text">
5212 declare void %llvm.va_start(i8* <arglist>)
5216 <p>The '<tt>llvm.va_start</tt>' intrinsic initializes <tt>*<arglist></tt>
5217 for subsequent use by <tt><a href="#i_va_arg">va_arg</a></tt>.</p>
5220 <p>The argument is a pointer to a <tt>va_list</tt> element to initialize.</p>
5223 <p>The '<tt>llvm.va_start</tt>' intrinsic works just like the <tt>va_start</tt>
5224 macro available in C. In a target-dependent way, it initializes
5225 the <tt>va_list</tt> element to which the argument points, so that the next
5226 call to <tt>va_arg</tt> will produce the first variable argument passed to
5227 the function. Unlike the C <tt>va_start</tt> macro, this intrinsic does not
5228 need to know the last argument of the function as the compiler can figure
5233 <!-- _______________________________________________________________________ -->
5234 <div class="doc_subsubsection">
5235 <a name="int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a>
5238 <div class="doc_text">
5242 declare void @llvm.va_end(i8* <arglist>)
5246 <p>The '<tt>llvm.va_end</tt>' intrinsic destroys <tt>*<arglist></tt>,
5247 which has been initialized previously
5248 with <tt><a href="#int_va_start">llvm.va_start</a></tt>
5249 or <tt><a href="#i_va_copy">llvm.va_copy</a></tt>.</p>
5252 <p>The argument is a pointer to a <tt>va_list</tt> to destroy.</p>
5255 <p>The '<tt>llvm.va_end</tt>' intrinsic works just like the <tt>va_end</tt>
5256 macro available in C. In a target-dependent way, it destroys
5257 the <tt>va_list</tt> element to which the argument points. Calls
5258 to <a href="#int_va_start"><tt>llvm.va_start</tt></a>
5259 and <a href="#int_va_copy"> <tt>llvm.va_copy</tt></a> must be matched exactly
5260 with calls to <tt>llvm.va_end</tt>.</p>
5264 <!-- _______________________________________________________________________ -->
5265 <div class="doc_subsubsection">
5266 <a name="int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a>
5269 <div class="doc_text">
5273 declare void @llvm.va_copy(i8* <destarglist>, i8* <srcarglist>)
5277 <p>The '<tt>llvm.va_copy</tt>' intrinsic copies the current argument position
5278 from the source argument list to the destination argument list.</p>
5281 <p>The first argument is a pointer to a <tt>va_list</tt> element to initialize.
5282 The second argument is a pointer to a <tt>va_list</tt> element to copy
5286 <p>The '<tt>llvm.va_copy</tt>' intrinsic works just like the <tt>va_copy</tt>
5287 macro available in C. In a target-dependent way, it copies the
5288 source <tt>va_list</tt> element into the destination <tt>va_list</tt>
5289 element. This intrinsic is necessary because
5290 the <tt><a href="#int_va_start"> llvm.va_start</a></tt> intrinsic may be
5291 arbitrarily complex and require, for example, memory allocation.</p>
5295 <!-- ======================================================================= -->
5296 <div class="doc_subsection">
5297 <a name="int_gc">Accurate Garbage Collection Intrinsics</a>
5300 <div class="doc_text">
5302 <p>LLVM support for <a href="GarbageCollection.html">Accurate Garbage
5303 Collection</a> (GC) requires the implementation and generation of these
5304 intrinsics. These intrinsics allow identification of <a href="#int_gcroot">GC
5305 roots on the stack</a>, as well as garbage collector implementations that
5306 require <a href="#int_gcread">read</a> and <a href="#int_gcwrite">write</a>
5307 barriers. Front-ends for type-safe garbage collected languages should generate
5308 these intrinsics to make use of the LLVM garbage collectors. For more details,
5309 see <a href="GarbageCollection.html">Accurate Garbage Collection with
5312 <p>The garbage collection intrinsics only operate on objects in the generic
5313 address space (address space zero).</p>
5317 <!-- _______________________________________________________________________ -->
5318 <div class="doc_subsubsection">
5319 <a name="int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a>
5322 <div class="doc_text">
5326 declare void @llvm.gcroot(i8** %ptrloc, i8* %metadata)
5330 <p>The '<tt>llvm.gcroot</tt>' intrinsic declares the existence of a GC root to
5331 the code generator, and allows some metadata to be associated with it.</p>
5334 <p>The first argument specifies the address of a stack object that contains the
5335 root pointer. The second pointer (which must be either a constant or a
5336 global value address) contains the meta-data to be associated with the
5340 <p>At runtime, a call to this intrinsic stores a null pointer into the "ptrloc"
5341 location. At compile-time, the code generator generates information to allow
5342 the runtime to find the pointer at GC safe points. The '<tt>llvm.gcroot</tt>'
5343 intrinsic may only be used in a function which <a href="#gc">specifies a GC
5348 <!-- _______________________________________________________________________ -->
5349 <div class="doc_subsubsection">
5350 <a name="int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a>
5353 <div class="doc_text">
5357 declare i8* @llvm.gcread(i8* %ObjPtr, i8** %Ptr)
5361 <p>The '<tt>llvm.gcread</tt>' intrinsic identifies reads of references from heap
5362 locations, allowing garbage collector implementations that require read
5366 <p>The second argument is the address to read from, which should be an address
5367 allocated from the garbage collector. The first object is a pointer to the
5368 start of the referenced object, if needed by the language runtime (otherwise
5372 <p>The '<tt>llvm.gcread</tt>' intrinsic has the same semantics as a load
5373 instruction, but may be replaced with substantially more complex code by the
5374 garbage collector runtime, as needed. The '<tt>llvm.gcread</tt>' intrinsic
5375 may only be used in a function which <a href="#gc">specifies a GC
5380 <!-- _______________________________________________________________________ -->
5381 <div class="doc_subsubsection">
5382 <a name="int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a>
5385 <div class="doc_text">
5389 declare void @llvm.gcwrite(i8* %P1, i8* %Obj, i8** %P2)
5393 <p>The '<tt>llvm.gcwrite</tt>' intrinsic identifies writes of references to heap
5394 locations, allowing garbage collector implementations that require write
5395 barriers (such as generational or reference counting collectors).</p>
5398 <p>The first argument is the reference to store, the second is the start of the
5399 object to store it to, and the third is the address of the field of Obj to
5400 store to. If the runtime does not require a pointer to the object, Obj may
5404 <p>The '<tt>llvm.gcwrite</tt>' intrinsic has the same semantics as a store
5405 instruction, but may be replaced with substantially more complex code by the
5406 garbage collector runtime, as needed. The '<tt>llvm.gcwrite</tt>' intrinsic
5407 may only be used in a function which <a href="#gc">specifies a GC
5412 <!-- ======================================================================= -->
5413 <div class="doc_subsection">
5414 <a name="int_codegen">Code Generator Intrinsics</a>
5417 <div class="doc_text">
5419 <p>These intrinsics are provided by LLVM to expose special features that may
5420 only be implemented with code generator support.</p>
5424 <!-- _______________________________________________________________________ -->
5425 <div class="doc_subsubsection">
5426 <a name="int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a>
5429 <div class="doc_text">
5433 declare i8 *@llvm.returnaddress(i32 <level>)
5437 <p>The '<tt>llvm.returnaddress</tt>' intrinsic attempts to compute a
5438 target-specific value indicating the return address of the current function
5439 or one of its callers.</p>
5442 <p>The argument to this intrinsic indicates which function to return the address
5443 for. Zero indicates the calling function, one indicates its caller, etc.
5444 The argument is <b>required</b> to be a constant integer value.</p>
5447 <p>The '<tt>llvm.returnaddress</tt>' intrinsic either returns a pointer
5448 indicating the return address of the specified call frame, or zero if it
5449 cannot be identified. The value returned by this intrinsic is likely to be
5450 incorrect or 0 for arguments other than zero, so it should only be used for
5451 debugging purposes.</p>
5453 <p>Note that calling this intrinsic does not prevent function inlining or other
5454 aggressive transformations, so the value returned may not be that of the
5455 obvious source-language caller.</p>
5459 <!-- _______________________________________________________________________ -->
5460 <div class="doc_subsubsection">
5461 <a name="int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a>
5464 <div class="doc_text">
5468 declare i8 *@llvm.frameaddress(i32 <level>)
5472 <p>The '<tt>llvm.frameaddress</tt>' intrinsic attempts to return the
5473 target-specific frame pointer value for the specified stack frame.</p>
5476 <p>The argument to this intrinsic indicates which function to return the frame
5477 pointer for. Zero indicates the calling function, one indicates its caller,
5478 etc. The argument is <b>required</b> to be a constant integer value.</p>
5481 <p>The '<tt>llvm.frameaddress</tt>' intrinsic either returns a pointer
5482 indicating the frame address of the specified call frame, or zero if it
5483 cannot be identified. The value returned by this intrinsic is likely to be
5484 incorrect or 0 for arguments other than zero, so it should only be used for
5485 debugging purposes.</p>
5487 <p>Note that calling this intrinsic does not prevent function inlining or other
5488 aggressive transformations, so the value returned may not be that of the
5489 obvious source-language caller.</p>
5493 <!-- _______________________________________________________________________ -->
5494 <div class="doc_subsubsection">
5495 <a name="int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a>
5498 <div class="doc_text">
5502 declare i8 *@llvm.stacksave()
5506 <p>The '<tt>llvm.stacksave</tt>' intrinsic is used to remember the current state
5507 of the function stack, for use
5508 with <a href="#int_stackrestore"> <tt>llvm.stackrestore</tt></a>. This is
5509 useful for implementing language features like scoped automatic variable
5510 sized arrays in C99.</p>
5513 <p>This intrinsic returns a opaque pointer value that can be passed
5514 to <a href="#int_stackrestore"><tt>llvm.stackrestore</tt></a>. When
5515 an <tt>llvm.stackrestore</tt> intrinsic is executed with a value saved
5516 from <tt>llvm.stacksave</tt>, it effectively restores the state of the stack
5517 to the state it was in when the <tt>llvm.stacksave</tt> intrinsic executed.
5518 In practice, this pops any <a href="#i_alloca">alloca</a> blocks from the
5519 stack that were allocated after the <tt>llvm.stacksave</tt> was executed.</p>
5523 <!-- _______________________________________________________________________ -->
5524 <div class="doc_subsubsection">
5525 <a name="int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a>
5528 <div class="doc_text">
5532 declare void @llvm.stackrestore(i8 * %ptr)
5536 <p>The '<tt>llvm.stackrestore</tt>' intrinsic is used to restore the state of
5537 the function stack to the state it was in when the
5538 corresponding <a href="#int_stacksave"><tt>llvm.stacksave</tt></a> intrinsic
5539 executed. This is useful for implementing language features like scoped
5540 automatic variable sized arrays in C99.</p>
5543 <p>See the description
5544 for <a href="#int_stacksave"><tt>llvm.stacksave</tt></a>.</p>
5548 <!-- _______________________________________________________________________ -->
5549 <div class="doc_subsubsection">
5550 <a name="int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a>
5553 <div class="doc_text">
5557 declare void @llvm.prefetch(i8* <address>, i32 <rw>, i32 <locality>)
5561 <p>The '<tt>llvm.prefetch</tt>' intrinsic is a hint to the code generator to
5562 insert a prefetch instruction if supported; otherwise, it is a noop.
5563 Prefetches have no effect on the behavior of the program but can change its
5564 performance characteristics.</p>
5567 <p><tt>address</tt> is the address to be prefetched, <tt>rw</tt> is the
5568 specifier determining if the fetch should be for a read (0) or write (1),
5569 and <tt>locality</tt> is a temporal locality specifier ranging from (0) - no
5570 locality, to (3) - extremely local keep in cache. The <tt>rw</tt>
5571 and <tt>locality</tt> arguments must be constant integers.</p>
5574 <p>This intrinsic does not modify the behavior of the program. In particular,
5575 prefetches cannot trap and do not produce a value. On targets that support
5576 this intrinsic, the prefetch can provide hints to the processor cache for
5577 better performance.</p>
5581 <!-- _______________________________________________________________________ -->
5582 <div class="doc_subsubsection">
5583 <a name="int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a>
5586 <div class="doc_text">
5590 declare void @llvm.pcmarker(i32 <id>)
5594 <p>The '<tt>llvm.pcmarker</tt>' intrinsic is a method to export a Program
5595 Counter (PC) in a region of code to simulators and other tools. The method
5596 is target specific, but it is expected that the marker will use exported
5597 symbols to transmit the PC of the marker. The marker makes no guarantees
5598 that it will remain with any specific instruction after optimizations. It is
5599 possible that the presence of a marker will inhibit optimizations. The
5600 intended use is to be inserted after optimizations to allow correlations of
5601 simulation runs.</p>
5604 <p><tt>id</tt> is a numerical id identifying the marker.</p>
5607 <p>This intrinsic does not modify the behavior of the program. Backends that do
5608 not support this intrinisic may ignore it.</p>
5612 <!-- _______________________________________________________________________ -->
5613 <div class="doc_subsubsection">
5614 <a name="int_readcyclecounter">'<tt>llvm.readcyclecounter</tt>' Intrinsic</a>
5617 <div class="doc_text">
5621 declare i64 @llvm.readcyclecounter( )
5625 <p>The '<tt>llvm.readcyclecounter</tt>' intrinsic provides access to the cycle
5626 counter register (or similar low latency, high accuracy clocks) on those
5627 targets that support it. On X86, it should map to RDTSC. On Alpha, it
5628 should map to RPCC. As the backing counters overflow quickly (on the order
5629 of 9 seconds on alpha), this should only be used for small timings.</p>
5632 <p>When directly supported, reading the cycle counter should not modify any
5633 memory. Implementations are allowed to either return a application specific
5634 value or a system wide value. On backends without support, this is lowered
5635 to a constant 0.</p>
5639 <!-- ======================================================================= -->
5640 <div class="doc_subsection">
5641 <a name="int_libc">Standard C Library Intrinsics</a>
5644 <div class="doc_text">
5646 <p>LLVM provides intrinsics for a few important standard C library functions.
5647 These intrinsics allow source-language front-ends to pass information about
5648 the alignment of the pointer arguments to the code generator, providing
5649 opportunity for more efficient code generation.</p>
5653 <!-- _______________________________________________________________________ -->
5654 <div class="doc_subsubsection">
5655 <a name="int_memcpy">'<tt>llvm.memcpy</tt>' Intrinsic</a>
5658 <div class="doc_text">
5661 <p>This is an overloaded intrinsic. You can use <tt>llvm.memcpy</tt> on any
5662 integer bit width. Not all targets support all bit widths however.</p>
5665 declare void @llvm.memcpy.i8(i8 * <dest>, i8 * <src>,
5666 i8 <len>, i32 <align>)
5667 declare void @llvm.memcpy.i16(i8 * <dest>, i8 * <src>,
5668 i16 <len>, i32 <align>)
5669 declare void @llvm.memcpy.i32(i8 * <dest>, i8 * <src>,
5670 i32 <len>, i32 <align>)
5671 declare void @llvm.memcpy.i64(i8 * <dest>, i8 * <src>,
5672 i64 <len>, i32 <align>)
5676 <p>The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the
5677 source location to the destination location.</p>
5679 <p>Note that, unlike the standard libc function, the <tt>llvm.memcpy.*</tt>
5680 intrinsics do not return a value, and takes an extra alignment argument.</p>
5683 <p>The first argument is a pointer to the destination, the second is a pointer
5684 to the source. The third argument is an integer argument specifying the
5685 number of bytes to copy, and the fourth argument is the alignment of the
5686 source and destination locations.</p>
5688 <p>If the call to this intrinisic has an alignment value that is not 0 or 1,
5689 then the caller guarantees that both the source and destination pointers are
5690 aligned to that boundary.</p>
5693 <p>The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the
5694 source location to the destination location, which are not allowed to
5695 overlap. It copies "len" bytes of memory over. If the argument is known to
5696 be aligned to some boundary, this can be specified as the fourth argument,
5697 otherwise it should be set to 0 or 1.</p>
5701 <!-- _______________________________________________________________________ -->
5702 <div class="doc_subsubsection">
5703 <a name="int_memmove">'<tt>llvm.memmove</tt>' Intrinsic</a>
5706 <div class="doc_text">
5709 <p>This is an overloaded intrinsic. You can use llvm.memmove on any integer bit
5710 width. Not all targets support all bit widths however.</p>
5713 declare void @llvm.memmove.i8(i8 * <dest>, i8 * <src>,
5714 i8 <len>, i32 <align>)
5715 declare void @llvm.memmove.i16(i8 * <dest>, i8 * <src>,
5716 i16 <len>, i32 <align>)
5717 declare void @llvm.memmove.i32(i8 * <dest>, i8 * <src>,
5718 i32 <len>, i32 <align>)
5719 declare void @llvm.memmove.i64(i8 * <dest>, i8 * <src>,
5720 i64 <len>, i32 <align>)
5724 <p>The '<tt>llvm.memmove.*</tt>' intrinsics move a block of memory from the
5725 source location to the destination location. It is similar to the
5726 '<tt>llvm.memcpy</tt>' intrinsic but allows the two memory locations to
5729 <p>Note that, unlike the standard libc function, the <tt>llvm.memmove.*</tt>
5730 intrinsics do not return a value, and takes an extra alignment argument.</p>
5733 <p>The first argument is a pointer to the destination, the second is a pointer
5734 to the source. The third argument is an integer argument specifying the
5735 number of bytes to copy, and the fourth argument is the alignment of the
5736 source and destination locations.</p>
5738 <p>If the call to this intrinisic has an alignment value that is not 0 or 1,
5739 then the caller guarantees that the source and destination pointers are
5740 aligned to that boundary.</p>
5743 <p>The '<tt>llvm.memmove.*</tt>' intrinsics copy a block of memory from the
5744 source location to the destination location, which may overlap. It copies
5745 "len" bytes of memory over. If the argument is known to be aligned to some
5746 boundary, this can be specified as the fourth argument, otherwise it should
5747 be set to 0 or 1.</p>
5751 <!-- _______________________________________________________________________ -->
5752 <div class="doc_subsubsection">
5753 <a name="int_memset">'<tt>llvm.memset.*</tt>' Intrinsics</a>
5756 <div class="doc_text">
5759 <p>This is an overloaded intrinsic. You can use llvm.memset on any integer bit
5760 width. Not all targets support all bit widths however.</p>
5763 declare void @llvm.memset.i8(i8 * <dest>, i8 <val>,
5764 i8 <len>, i32 <align>)
5765 declare void @llvm.memset.i16(i8 * <dest>, i8 <val>,
5766 i16 <len>, i32 <align>)
5767 declare void @llvm.memset.i32(i8 * <dest>, i8 <val>,
5768 i32 <len>, i32 <align>)
5769 declare void @llvm.memset.i64(i8 * <dest>, i8 <val>,
5770 i64 <len>, i32 <align>)
5774 <p>The '<tt>llvm.memset.*</tt>' intrinsics fill a block of memory with a
5775 particular byte value.</p>
5777 <p>Note that, unlike the standard libc function, the <tt>llvm.memset</tt>
5778 intrinsic does not return a value, and takes an extra alignment argument.</p>
5781 <p>The first argument is a pointer to the destination to fill, the second is the
5782 byte value to fill it with, the third argument is an integer argument
5783 specifying the number of bytes to fill, and the fourth argument is the known
5784 alignment of destination location.</p>
5786 <p>If the call to this intrinisic has an alignment value that is not 0 or 1,
5787 then the caller guarantees that the destination pointer is aligned to that
5791 <p>The '<tt>llvm.memset.*</tt>' intrinsics fill "len" bytes of memory starting
5792 at the destination location. If the argument is known to be aligned to some
5793 boundary, this can be specified as the fourth argument, otherwise it should
5794 be set to 0 or 1.</p>
5798 <!-- _______________________________________________________________________ -->
5799 <div class="doc_subsubsection">
5800 <a name="int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a>
5803 <div class="doc_text">
5806 <p>This is an overloaded intrinsic. You can use <tt>llvm.sqrt</tt> on any
5807 floating point or vector of floating point type. Not all targets support all
5811 declare float @llvm.sqrt.f32(float %Val)
5812 declare double @llvm.sqrt.f64(double %Val)
5813 declare x86_fp80 @llvm.sqrt.f80(x86_fp80 %Val)
5814 declare fp128 @llvm.sqrt.f128(fp128 %Val)
5815 declare ppc_fp128 @llvm.sqrt.ppcf128(ppc_fp128 %Val)
5819 <p>The '<tt>llvm.sqrt</tt>' intrinsics return the sqrt of the specified operand,
5820 returning the same value as the libm '<tt>sqrt</tt>' functions would.
5821 Unlike <tt>sqrt</tt> in libm, however, <tt>llvm.sqrt</tt> has undefined
5822 behavior for negative numbers other than -0.0 (which allows for better
5823 optimization, because there is no need to worry about errno being
5824 set). <tt>llvm.sqrt(-0.0)</tt> is defined to return -0.0 like IEEE sqrt.</p>
5827 <p>The argument and return value are floating point numbers of the same
5831 <p>This function returns the sqrt of the specified operand if it is a
5832 nonnegative floating point number.</p>
5836 <!-- _______________________________________________________________________ -->
5837 <div class="doc_subsubsection">
5838 <a name="int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a>
5841 <div class="doc_text">
5844 <p>This is an overloaded intrinsic. You can use <tt>llvm.powi</tt> on any
5845 floating point or vector of floating point type. Not all targets support all
5849 declare float @llvm.powi.f32(float %Val, i32 %power)
5850 declare double @llvm.powi.f64(double %Val, i32 %power)
5851 declare x86_fp80 @llvm.powi.f80(x86_fp80 %Val, i32 %power)
5852 declare fp128 @llvm.powi.f128(fp128 %Val, i32 %power)
5853 declare ppc_fp128 @llvm.powi.ppcf128(ppc_fp128 %Val, i32 %power)
5857 <p>The '<tt>llvm.powi.*</tt>' intrinsics return the first operand raised to the
5858 specified (positive or negative) power. The order of evaluation of
5859 multiplications is not defined. When a vector of floating point type is
5860 used, the second argument remains a scalar integer value.</p>
5863 <p>The second argument is an integer power, and the first is a value to raise to
5867 <p>This function returns the first value raised to the second power with an
5868 unspecified sequence of rounding operations.</p>
5872 <!-- _______________________________________________________________________ -->
5873 <div class="doc_subsubsection">
5874 <a name="int_sin">'<tt>llvm.sin.*</tt>' Intrinsic</a>
5877 <div class="doc_text">
5880 <p>This is an overloaded intrinsic. You can use <tt>llvm.sin</tt> on any
5881 floating point or vector of floating point type. Not all targets support all
5885 declare float @llvm.sin.f32(float %Val)
5886 declare double @llvm.sin.f64(double %Val)
5887 declare x86_fp80 @llvm.sin.f80(x86_fp80 %Val)
5888 declare fp128 @llvm.sin.f128(fp128 %Val)
5889 declare ppc_fp128 @llvm.sin.ppcf128(ppc_fp128 %Val)
5893 <p>The '<tt>llvm.sin.*</tt>' intrinsics return the sine of the operand.</p>
5896 <p>The argument and return value are floating point numbers of the same
5900 <p>This function returns the sine of the specified operand, returning the same
5901 values as the libm <tt>sin</tt> functions would, and handles error conditions
5902 in the same way.</p>
5906 <!-- _______________________________________________________________________ -->
5907 <div class="doc_subsubsection">
5908 <a name="int_cos">'<tt>llvm.cos.*</tt>' Intrinsic</a>
5911 <div class="doc_text">
5914 <p>This is an overloaded intrinsic. You can use <tt>llvm.cos</tt> on any
5915 floating point or vector of floating point type. Not all targets support all
5919 declare float @llvm.cos.f32(float %Val)
5920 declare double @llvm.cos.f64(double %Val)
5921 declare x86_fp80 @llvm.cos.f80(x86_fp80 %Val)
5922 declare fp128 @llvm.cos.f128(fp128 %Val)
5923 declare ppc_fp128 @llvm.cos.ppcf128(ppc_fp128 %Val)
5927 <p>The '<tt>llvm.cos.*</tt>' intrinsics return the cosine of the operand.</p>
5930 <p>The argument and return value are floating point numbers of the same
5934 <p>This function returns the cosine of the specified operand, returning the same
5935 values as the libm <tt>cos</tt> functions would, and handles error conditions
5936 in the same way.</p>
5940 <!-- _______________________________________________________________________ -->
5941 <div class="doc_subsubsection">
5942 <a name="int_pow">'<tt>llvm.pow.*</tt>' Intrinsic</a>
5945 <div class="doc_text">
5948 <p>This is an overloaded intrinsic. You can use <tt>llvm.pow</tt> on any
5949 floating point or vector of floating point type. Not all targets support all
5953 declare float @llvm.pow.f32(float %Val, float %Power)
5954 declare double @llvm.pow.f64(double %Val, double %Power)
5955 declare x86_fp80 @llvm.pow.f80(x86_fp80 %Val, x86_fp80 %Power)
5956 declare fp128 @llvm.pow.f128(fp128 %Val, fp128 %Power)
5957 declare ppc_fp128 @llvm.pow.ppcf128(ppc_fp128 %Val, ppc_fp128 Power)
5961 <p>The '<tt>llvm.pow.*</tt>' intrinsics return the first operand raised to the
5962 specified (positive or negative) power.</p>
5965 <p>The second argument is a floating point power, and the first is a value to
5966 raise to that power.</p>
5969 <p>This function returns the first value raised to the second power, returning
5970 the same values as the libm <tt>pow</tt> functions would, and handles error
5971 conditions in the same way.</p>
5975 <!-- ======================================================================= -->
5976 <div class="doc_subsection">
5977 <a name="int_manip">Bit Manipulation Intrinsics</a>
5980 <div class="doc_text">
5982 <p>LLVM provides intrinsics for a few important bit manipulation operations.
5983 These allow efficient code generation for some algorithms.</p>
5987 <!-- _______________________________________________________________________ -->
5988 <div class="doc_subsubsection">
5989 <a name="int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a>
5992 <div class="doc_text">
5995 <p>This is an overloaded intrinsic function. You can use bswap on any integer
5996 type that is an even number of bytes (i.e. BitWidth % 16 == 0).</p>
5999 declare i16 @llvm.bswap.i16(i16 <id>)
6000 declare i32 @llvm.bswap.i32(i32 <id>)
6001 declare i64 @llvm.bswap.i64(i64 <id>)
6005 <p>The '<tt>llvm.bswap</tt>' family of intrinsics is used to byte swap integer
6006 values with an even number of bytes (positive multiple of 16 bits). These
6007 are useful for performing operations on data that is not in the target's
6008 native byte order.</p>
6011 <p>The <tt>llvm.bswap.i16</tt> intrinsic returns an i16 value that has the high
6012 and low byte of the input i16 swapped. Similarly,
6013 the <tt>llvm.bswap.i32</tt> intrinsic returns an i32 value that has the four
6014 bytes of the input i32 swapped, so that if the input bytes are numbered 0, 1,
6015 2, 3 then the returned i32 will have its bytes in 3, 2, 1, 0 order.
6016 The <tt>llvm.bswap.i48</tt>, <tt>llvm.bswap.i64</tt> and other intrinsics
6017 extend this concept to additional even-byte lengths (6 bytes, 8 bytes and
6018 more, respectively).</p>
6022 <!-- _______________________________________________________________________ -->
6023 <div class="doc_subsubsection">
6024 <a name="int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic</a>
6027 <div class="doc_text">
6030 <p>This is an overloaded intrinsic. You can use llvm.ctpop on any integer bit
6031 width. Not all targets support all bit widths however.</p>
6034 declare i8 @llvm.ctpop.i8(i8 <src>)
6035 declare i16 @llvm.ctpop.i16(i16 <src>)
6036 declare i32 @llvm.ctpop.i32(i32 <src>)
6037 declare i64 @llvm.ctpop.i64(i64 <src>)
6038 declare i256 @llvm.ctpop.i256(i256 <src>)
6042 <p>The '<tt>llvm.ctpop</tt>' family of intrinsics counts the number of bits set
6046 <p>The only argument is the value to be counted. The argument may be of any
6047 integer type. The return type must match the argument type.</p>
6050 <p>The '<tt>llvm.ctpop</tt>' intrinsic counts the 1's in a variable.</p>
6054 <!-- _______________________________________________________________________ -->
6055 <div class="doc_subsubsection">
6056 <a name="int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic</a>
6059 <div class="doc_text">
6062 <p>This is an overloaded intrinsic. You can use <tt>llvm.ctlz</tt> on any
6063 integer bit width. Not all targets support all bit widths however.</p>
6066 declare i8 @llvm.ctlz.i8 (i8 <src>)
6067 declare i16 @llvm.ctlz.i16(i16 <src>)
6068 declare i32 @llvm.ctlz.i32(i32 <src>)
6069 declare i64 @llvm.ctlz.i64(i64 <src>)
6070 declare i256 @llvm.ctlz.i256(i256 <src>)
6074 <p>The '<tt>llvm.ctlz</tt>' family of intrinsic functions counts the number of
6075 leading zeros in a variable.</p>
6078 <p>The only argument is the value to be counted. The argument may be of any
6079 integer type. The return type must match the argument type.</p>
6082 <p>The '<tt>llvm.ctlz</tt>' intrinsic counts the leading (most significant)
6083 zeros in a variable. If the src == 0 then the result is the size in bits of
6084 the type of src. For example, <tt>llvm.ctlz(i32 2) = 30</tt>.</p>
6088 <!-- _______________________________________________________________________ -->
6089 <div class="doc_subsubsection">
6090 <a name="int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic</a>
6093 <div class="doc_text">
6096 <p>This is an overloaded intrinsic. You can use <tt>llvm.cttz</tt> on any
6097 integer bit width. Not all targets support all bit widths however.</p>
6100 declare i8 @llvm.cttz.i8 (i8 <src>)
6101 declare i16 @llvm.cttz.i16(i16 <src>)
6102 declare i32 @llvm.cttz.i32(i32 <src>)
6103 declare i64 @llvm.cttz.i64(i64 <src>)
6104 declare i256 @llvm.cttz.i256(i256 <src>)
6108 <p>The '<tt>llvm.cttz</tt>' family of intrinsic functions counts the number of
6112 <p>The only argument is the value to be counted. The argument may be of any
6113 integer type. The return type must match the argument type.</p>
6116 <p>The '<tt>llvm.cttz</tt>' intrinsic counts the trailing (least significant)
6117 zeros in a variable. If the src == 0 then the result is the size in bits of
6118 the type of src. For example, <tt>llvm.cttz(2) = 1</tt>.</p>
6122 <!-- ======================================================================= -->
6123 <div class="doc_subsection">
6124 <a name="int_overflow">Arithmetic with Overflow Intrinsics</a>
6127 <div class="doc_text">
6129 <p>LLVM provides intrinsics for some arithmetic with overflow operations.</p>
6133 <!-- _______________________________________________________________________ -->
6134 <div class="doc_subsubsection">
6135 <a name="int_sadd_overflow">'<tt>llvm.sadd.with.overflow.*</tt>' Intrinsics</a>
6138 <div class="doc_text">
6141 <p>This is an overloaded intrinsic. You can use <tt>llvm.sadd.with.overflow</tt>
6142 on any integer bit width.</p>
6145 declare {i16, i1} @llvm.sadd.with.overflow.i16(i16 %a, i16 %b)
6146 declare {i32, i1} @llvm.sadd.with.overflow.i32(i32 %a, i32 %b)
6147 declare {i64, i1} @llvm.sadd.with.overflow.i64(i64 %a, i64 %b)
6151 <p>The '<tt>llvm.sadd.with.overflow</tt>' family of intrinsic functions perform
6152 a signed addition of the two arguments, and indicate whether an overflow
6153 occurred during the signed summation.</p>
6156 <p>The arguments (%a and %b) and the first element of the result structure may
6157 be of integer types of any bit width, but they must have the same bit
6158 width. The second element of the result structure must be of
6159 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
6160 undergo signed addition.</p>
6163 <p>The '<tt>llvm.sadd.with.overflow</tt>' family of intrinsic functions perform
6164 a signed addition of the two variables. They return a structure — the
6165 first element of which is the signed summation, and the second element of
6166 which is a bit specifying if the signed summation resulted in an
6171 %res = call {i32, i1} @llvm.sadd.with.overflow.i32(i32 %a, i32 %b)
6172 %sum = extractvalue {i32, i1} %res, 0
6173 %obit = extractvalue {i32, i1} %res, 1
6174 br i1 %obit, label %overflow, label %normal
6179 <!-- _______________________________________________________________________ -->
6180 <div class="doc_subsubsection">
6181 <a name="int_uadd_overflow">'<tt>llvm.uadd.with.overflow.*</tt>' Intrinsics</a>
6184 <div class="doc_text">
6187 <p>This is an overloaded intrinsic. You can use <tt>llvm.uadd.with.overflow</tt>
6188 on any integer bit width.</p>
6191 declare {i16, i1} @llvm.uadd.with.overflow.i16(i16 %a, i16 %b)
6192 declare {i32, i1} @llvm.uadd.with.overflow.i32(i32 %a, i32 %b)
6193 declare {i64, i1} @llvm.uadd.with.overflow.i64(i64 %a, i64 %b)
6197 <p>The '<tt>llvm.uadd.with.overflow</tt>' family of intrinsic functions perform
6198 an unsigned addition of the two arguments, and indicate whether a carry
6199 occurred during the unsigned summation.</p>
6202 <p>The arguments (%a and %b) and the first element of the result structure may
6203 be of integer types of any bit width, but they must have the same bit
6204 width. The second element of the result structure must be of
6205 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
6206 undergo unsigned addition.</p>
6209 <p>The '<tt>llvm.uadd.with.overflow</tt>' family of intrinsic functions perform
6210 an unsigned addition of the two arguments. They return a structure —
6211 the first element of which is the sum, and the second element of which is a
6212 bit specifying if the unsigned summation resulted in a carry.</p>
6216 %res = call {i32, i1} @llvm.uadd.with.overflow.i32(i32 %a, i32 %b)
6217 %sum = extractvalue {i32, i1} %res, 0
6218 %obit = extractvalue {i32, i1} %res, 1
6219 br i1 %obit, label %carry, label %normal
6224 <!-- _______________________________________________________________________ -->
6225 <div class="doc_subsubsection">
6226 <a name="int_ssub_overflow">'<tt>llvm.ssub.with.overflow.*</tt>' Intrinsics</a>
6229 <div class="doc_text">
6232 <p>This is an overloaded intrinsic. You can use <tt>llvm.ssub.with.overflow</tt>
6233 on any integer bit width.</p>
6236 declare {i16, i1} @llvm.ssub.with.overflow.i16(i16 %a, i16 %b)
6237 declare {i32, i1} @llvm.ssub.with.overflow.i32(i32 %a, i32 %b)
6238 declare {i64, i1} @llvm.ssub.with.overflow.i64(i64 %a, i64 %b)
6242 <p>The '<tt>llvm.ssub.with.overflow</tt>' family of intrinsic functions perform
6243 a signed subtraction of the two arguments, and indicate whether an overflow
6244 occurred during the signed subtraction.</p>
6247 <p>The arguments (%a and %b) and the first element of the result structure may
6248 be of integer types of any bit width, but they must have the same bit
6249 width. The second element of the result structure must be of
6250 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
6251 undergo signed subtraction.</p>
6254 <p>The '<tt>llvm.ssub.with.overflow</tt>' family of intrinsic functions perform
6255 a signed subtraction of the two arguments. They return a structure —
6256 the first element of which is the subtraction, and the second element of
6257 which is a bit specifying if the signed subtraction resulted in an
6262 %res = call {i32, i1} @llvm.ssub.with.overflow.i32(i32 %a, i32 %b)
6263 %sum = extractvalue {i32, i1} %res, 0
6264 %obit = extractvalue {i32, i1} %res, 1
6265 br i1 %obit, label %overflow, label %normal
6270 <!-- _______________________________________________________________________ -->
6271 <div class="doc_subsubsection">
6272 <a name="int_usub_overflow">'<tt>llvm.usub.with.overflow.*</tt>' Intrinsics</a>
6275 <div class="doc_text">
6278 <p>This is an overloaded intrinsic. You can use <tt>llvm.usub.with.overflow</tt>
6279 on any integer bit width.</p>
6282 declare {i16, i1} @llvm.usub.with.overflow.i16(i16 %a, i16 %b)
6283 declare {i32, i1} @llvm.usub.with.overflow.i32(i32 %a, i32 %b)
6284 declare {i64, i1} @llvm.usub.with.overflow.i64(i64 %a, i64 %b)
6288 <p>The '<tt>llvm.usub.with.overflow</tt>' family of intrinsic functions perform
6289 an unsigned subtraction of the two arguments, and indicate whether an
6290 overflow occurred during the unsigned subtraction.</p>
6293 <p>The arguments (%a and %b) and the first element of the result structure may
6294 be of integer types of any bit width, but they must have the same bit
6295 width. The second element of the result structure must be of
6296 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
6297 undergo unsigned subtraction.</p>
6300 <p>The '<tt>llvm.usub.with.overflow</tt>' family of intrinsic functions perform
6301 an unsigned subtraction of the two arguments. They return a structure —
6302 the first element of which is the subtraction, and the second element of
6303 which is a bit specifying if the unsigned subtraction resulted in an
6308 %res = call {i32, i1} @llvm.usub.with.overflow.i32(i32 %a, i32 %b)
6309 %sum = extractvalue {i32, i1} %res, 0
6310 %obit = extractvalue {i32, i1} %res, 1
6311 br i1 %obit, label %overflow, label %normal
6316 <!-- _______________________________________________________________________ -->
6317 <div class="doc_subsubsection">
6318 <a name="int_smul_overflow">'<tt>llvm.smul.with.overflow.*</tt>' Intrinsics</a>
6321 <div class="doc_text">
6324 <p>This is an overloaded intrinsic. You can use <tt>llvm.smul.with.overflow</tt>
6325 on any integer bit width.</p>
6328 declare {i16, i1} @llvm.smul.with.overflow.i16(i16 %a, i16 %b)
6329 declare {i32, i1} @llvm.smul.with.overflow.i32(i32 %a, i32 %b)
6330 declare {i64, i1} @llvm.smul.with.overflow.i64(i64 %a, i64 %b)
6335 <p>The '<tt>llvm.smul.with.overflow</tt>' family of intrinsic functions perform
6336 a signed multiplication of the two arguments, and indicate whether an
6337 overflow occurred during the signed multiplication.</p>
6340 <p>The arguments (%a and %b) and the first element of the result structure may
6341 be of integer types of any bit width, but they must have the same bit
6342 width. The second element of the result structure must be of
6343 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
6344 undergo signed multiplication.</p>
6347 <p>The '<tt>llvm.smul.with.overflow</tt>' family of intrinsic functions perform
6348 a signed multiplication of the two arguments. They return a structure —
6349 the first element of which is the multiplication, and the second element of
6350 which is a bit specifying if the signed multiplication resulted in an
6355 %res = call {i32, i1} @llvm.smul.with.overflow.i32(i32 %a, i32 %b)
6356 %sum = extractvalue {i32, i1} %res, 0
6357 %obit = extractvalue {i32, i1} %res, 1
6358 br i1 %obit, label %overflow, label %normal
6363 <!-- _______________________________________________________________________ -->
6364 <div class="doc_subsubsection">
6365 <a name="int_umul_overflow">'<tt>llvm.umul.with.overflow.*</tt>' Intrinsics</a>
6368 <div class="doc_text">
6371 <p>This is an overloaded intrinsic. You can use <tt>llvm.umul.with.overflow</tt>
6372 on any integer bit width.</p>
6375 declare {i16, i1} @llvm.umul.with.overflow.i16(i16 %a, i16 %b)
6376 declare {i32, i1} @llvm.umul.with.overflow.i32(i32 %a, i32 %b)
6377 declare {i64, i1} @llvm.umul.with.overflow.i64(i64 %a, i64 %b)
6381 <p>The '<tt>llvm.umul.with.overflow</tt>' family of intrinsic functions perform
6382 a unsigned multiplication of the two arguments, and indicate whether an
6383 overflow occurred during the unsigned multiplication.</p>
6386 <p>The arguments (%a and %b) and the first element of the result structure may
6387 be of integer types of any bit width, but they must have the same bit
6388 width. The second element of the result structure must be of
6389 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
6390 undergo unsigned multiplication.</p>
6393 <p>The '<tt>llvm.umul.with.overflow</tt>' family of intrinsic functions perform
6394 an unsigned multiplication of the two arguments. They return a structure
6395 — the first element of which is the multiplication, and the second
6396 element of which is a bit specifying if the unsigned multiplication resulted
6401 %res = call {i32, i1} @llvm.umul.with.overflow.i32(i32 %a, i32 %b)
6402 %sum = extractvalue {i32, i1} %res, 0
6403 %obit = extractvalue {i32, i1} %res, 1
6404 br i1 %obit, label %overflow, label %normal
6409 <!-- ======================================================================= -->
6410 <div class="doc_subsection">
6411 <a name="int_debugger">Debugger Intrinsics</a>
6414 <div class="doc_text">
6416 <p>The LLVM debugger intrinsics (which all start with <tt>llvm.dbg.</tt>
6417 prefix), are described in
6418 the <a href="SourceLevelDebugging.html#format_common_intrinsics">LLVM Source
6419 Level Debugging</a> document.</p>
6423 <!-- ======================================================================= -->
6424 <div class="doc_subsection">
6425 <a name="int_eh">Exception Handling Intrinsics</a>
6428 <div class="doc_text">
6430 <p>The LLVM exception handling intrinsics (which all start with
6431 <tt>llvm.eh.</tt> prefix), are described in
6432 the <a href="ExceptionHandling.html#format_common_intrinsics">LLVM Exception
6433 Handling</a> document.</p>
6437 <!-- ======================================================================= -->
6438 <div class="doc_subsection">
6439 <a name="int_trampoline">Trampoline Intrinsic</a>
6442 <div class="doc_text">
6444 <p>This intrinsic makes it possible to excise one parameter, marked with
6445 the <tt>nest</tt> attribute, from a function. The result is a callable
6446 function pointer lacking the nest parameter - the caller does not need to
6447 provide a value for it. Instead, the value to use is stored in advance in a
6448 "trampoline", a block of memory usually allocated on the stack, which also
6449 contains code to splice the nest value into the argument list. This is used
6450 to implement the GCC nested function address extension.</p>
6452 <p>For example, if the function is
6453 <tt>i32 f(i8* nest %c, i32 %x, i32 %y)</tt> then the resulting function
6454 pointer has signature <tt>i32 (i32, i32)*</tt>. It can be created as
6457 <div class="doc_code">
6459 %tramp = alloca [10 x i8], align 4 ; size and alignment only correct for X86
6460 %tramp1 = getelementptr [10 x i8]* %tramp, i32 0, i32 0
6461 %p = call i8* @llvm.init.trampoline( i8* %tramp1, i8* bitcast (i32 (i8* nest , i32, i32)* @f to i8*), i8* %nval )
6462 %fp = bitcast i8* %p to i32 (i32, i32)*
6466 <p>The call <tt>%val = call i32 %fp( i32 %x, i32 %y )</tt> is then equivalent
6467 to <tt>%val = call i32 %f( i8* %nval, i32 %x, i32 %y )</tt>.</p>
6471 <!-- _______________________________________________________________________ -->
6472 <div class="doc_subsubsection">
6473 <a name="int_it">'<tt>llvm.init.trampoline</tt>' Intrinsic</a>
6476 <div class="doc_text">
6480 declare i8* @llvm.init.trampoline(i8* <tramp>, i8* <func>, i8* <nval>)
6484 <p>This fills the memory pointed to by <tt>tramp</tt> with code and returns a
6485 function pointer suitable for executing it.</p>
6488 <p>The <tt>llvm.init.trampoline</tt> intrinsic takes three arguments, all
6489 pointers. The <tt>tramp</tt> argument must point to a sufficiently large and
6490 sufficiently aligned block of memory; this memory is written to by the
6491 intrinsic. Note that the size and the alignment are target-specific - LLVM
6492 currently provides no portable way of determining them, so a front-end that
6493 generates this intrinsic needs to have some target-specific knowledge.
6494 The <tt>func</tt> argument must hold a function bitcast to
6495 an <tt>i8*</tt>.</p>
6498 <p>The block of memory pointed to by <tt>tramp</tt> is filled with target
6499 dependent code, turning it into a function. A pointer to this function is
6500 returned, but needs to be bitcast to an <a href="#int_trampoline">appropriate
6501 function pointer type</a> before being called. The new function's signature
6502 is the same as that of <tt>func</tt> with any arguments marked with
6503 the <tt>nest</tt> attribute removed. At most one such <tt>nest</tt> argument
6504 is allowed, and it must be of pointer type. Calling the new function is
6505 equivalent to calling <tt>func</tt> with the same argument list, but
6506 with <tt>nval</tt> used for the missing <tt>nest</tt> argument. If, after
6507 calling <tt>llvm.init.trampoline</tt>, the memory pointed to
6508 by <tt>tramp</tt> is modified, then the effect of any later call to the
6509 returned function pointer is undefined.</p>
6513 <!-- ======================================================================= -->
6514 <div class="doc_subsection">
6515 <a name="int_atomics">Atomic Operations and Synchronization Intrinsics</a>
6518 <div class="doc_text">
6520 <p>These intrinsic functions expand the "universal IR" of LLVM to represent
6521 hardware constructs for atomic operations and memory synchronization. This
6522 provides an interface to the hardware, not an interface to the programmer. It
6523 is aimed at a low enough level to allow any programming models or APIs
6524 (Application Programming Interfaces) which need atomic behaviors to map
6525 cleanly onto it. It is also modeled primarily on hardware behavior. Just as
6526 hardware provides a "universal IR" for source languages, it also provides a
6527 starting point for developing a "universal" atomic operation and
6528 synchronization IR.</p>
6530 <p>These do <em>not</em> form an API such as high-level threading libraries,
6531 software transaction memory systems, atomic primitives, and intrinsic
6532 functions as found in BSD, GNU libc, atomic_ops, APR, and other system and
6533 application libraries. The hardware interface provided by LLVM should allow
6534 a clean implementation of all of these APIs and parallel programming models.
6535 No one model or paradigm should be selected above others unless the hardware
6536 itself ubiquitously does so.</p>
6540 <!-- _______________________________________________________________________ -->
6541 <div class="doc_subsubsection">
6542 <a name="int_memory_barrier">'<tt>llvm.memory.barrier</tt>' Intrinsic</a>
6544 <div class="doc_text">
6547 declare void @llvm.memory.barrier( i1 <ll>, i1 <ls>, i1 <sl>, i1 <ss>, i1 <device> )
6551 <p>The <tt>llvm.memory.barrier</tt> intrinsic guarantees ordering between
6552 specific pairs of memory access types.</p>
6555 <p>The <tt>llvm.memory.barrier</tt> intrinsic requires five boolean arguments.
6556 The first four arguments enables a specific barrier as listed below. The
6557 fith argument specifies that the barrier applies to io or device or uncached
6561 <li><tt>ll</tt>: load-load barrier</li>
6562 <li><tt>ls</tt>: load-store barrier</li>
6563 <li><tt>sl</tt>: store-load barrier</li>
6564 <li><tt>ss</tt>: store-store barrier</li>
6565 <li><tt>device</tt>: barrier applies to device and uncached memory also.</li>
6569 <p>This intrinsic causes the system to enforce some ordering constraints upon
6570 the loads and stores of the program. This barrier does not
6571 indicate <em>when</em> any events will occur, it only enforces
6572 an <em>order</em> in which they occur. For any of the specified pairs of load
6573 and store operations (f.ex. load-load, or store-load), all of the first
6574 operations preceding the barrier will complete before any of the second
6575 operations succeeding the barrier begin. Specifically the semantics for each
6576 pairing is as follows:</p>
6579 <li><tt>ll</tt>: All loads before the barrier must complete before any load
6580 after the barrier begins.</li>
6581 <li><tt>ls</tt>: All loads before the barrier must complete before any
6582 store after the barrier begins.</li>
6583 <li><tt>ss</tt>: All stores before the barrier must complete before any
6584 store after the barrier begins.</li>
6585 <li><tt>sl</tt>: All stores before the barrier must complete before any
6586 load after the barrier begins.</li>
6589 <p>These semantics are applied with a logical "and" behavior when more than one
6590 is enabled in a single memory barrier intrinsic.</p>
6592 <p>Backends may implement stronger barriers than those requested when they do
6593 not support as fine grained a barrier as requested. Some architectures do
6594 not need all types of barriers and on such architectures, these become
6599 %mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
6600 %ptr = bitcast i8* %mallocP to i32*
6603 %result1 = load i32* %ptr <i>; yields {i32}:result1 = 4</i>
6604 call void @llvm.memory.barrier( i1 false, i1 true, i1 false, i1 false )
6605 <i>; guarantee the above finishes</i>
6606 store i32 8, %ptr <i>; before this begins</i>
6611 <!-- _______________________________________________________________________ -->
6612 <div class="doc_subsubsection">
6613 <a name="int_atomic_cmp_swap">'<tt>llvm.atomic.cmp.swap.*</tt>' Intrinsic</a>
6616 <div class="doc_text">
6619 <p>This is an overloaded intrinsic. You can use <tt>llvm.atomic.cmp.swap</tt> on
6620 any integer bit width and for different address spaces. Not all targets
6621 support all bit widths however.</p>
6624 declare i8 @llvm.atomic.cmp.swap.i8.p0i8( i8* <ptr>, i8 <cmp>, i8 <val> )
6625 declare i16 @llvm.atomic.cmp.swap.i16.p0i16( i16* <ptr>, i16 <cmp>, i16 <val> )
6626 declare i32 @llvm.atomic.cmp.swap.i32.p0i32( i32* <ptr>, i32 <cmp>, i32 <val> )
6627 declare i64 @llvm.atomic.cmp.swap.i64.p0i64( i64* <ptr>, i64 <cmp>, i64 <val> )
6631 <p>This loads a value in memory and compares it to a given value. If they are
6632 equal, it stores a new value into the memory.</p>
6635 <p>The <tt>llvm.atomic.cmp.swap</tt> intrinsic takes three arguments. The result
6636 as well as both <tt>cmp</tt> and <tt>val</tt> must be integer values with the
6637 same bit width. The <tt>ptr</tt> argument must be a pointer to a value of
6638 this integer type. While any bit width integer may be used, targets may only
6639 lower representations they support in hardware.</p>
6642 <p>This entire intrinsic must be executed atomically. It first loads the value
6643 in memory pointed to by <tt>ptr</tt> and compares it with the
6644 value <tt>cmp</tt>. If they are equal, <tt>val</tt> is stored into the
6645 memory. The loaded value is yielded in all cases. This provides the
6646 equivalent of an atomic compare-and-swap operation within the SSA
6651 %mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
6652 %ptr = bitcast i8* %mallocP to i32*
6655 %val1 = add i32 4, 4
6656 %result1 = call i32 @llvm.atomic.cmp.swap.i32.p0i32( i32* %ptr, i32 4, %val1 )
6657 <i>; yields {i32}:result1 = 4</i>
6658 %stored1 = icmp eq i32 %result1, 4 <i>; yields {i1}:stored1 = true</i>
6659 %memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 8</i>
6661 %val2 = add i32 1, 1
6662 %result2 = call i32 @llvm.atomic.cmp.swap.i32.p0i32( i32* %ptr, i32 5, %val2 )
6663 <i>; yields {i32}:result2 = 8</i>
6664 %stored2 = icmp eq i32 %result2, 5 <i>; yields {i1}:stored2 = false</i>
6666 %memval2 = load i32* %ptr <i>; yields {i32}:memval2 = 8</i>
6671 <!-- _______________________________________________________________________ -->
6672 <div class="doc_subsubsection">
6673 <a name="int_atomic_swap">'<tt>llvm.atomic.swap.*</tt>' Intrinsic</a>
6675 <div class="doc_text">
6678 <p>This is an overloaded intrinsic. You can use <tt>llvm.atomic.swap</tt> on any
6679 integer bit width. Not all targets support all bit widths however.</p>
6682 declare i8 @llvm.atomic.swap.i8.p0i8( i8* <ptr>, i8 <val> )
6683 declare i16 @llvm.atomic.swap.i16.p0i16( i16* <ptr>, i16 <val> )
6684 declare i32 @llvm.atomic.swap.i32.p0i32( i32* <ptr>, i32 <val> )
6685 declare i64 @llvm.atomic.swap.i64.p0i64( i64* <ptr>, i64 <val> )
6689 <p>This intrinsic loads the value stored in memory at <tt>ptr</tt> and yields
6690 the value from memory. It then stores the value in <tt>val</tt> in the memory
6691 at <tt>ptr</tt>.</p>
6694 <p>The <tt>llvm.atomic.swap</tt> intrinsic takes two arguments. Both
6695 the <tt>val</tt> argument and the result must be integers of the same bit
6696 width. The first argument, <tt>ptr</tt>, must be a pointer to a value of this
6697 integer type. The targets may only lower integer representations they
6701 <p>This intrinsic loads the value pointed to by <tt>ptr</tt>, yields it, and
6702 stores <tt>val</tt> back into <tt>ptr</tt> atomically. This provides the
6703 equivalent of an atomic swap operation within the SSA framework.</p>
6707 %mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
6708 %ptr = bitcast i8* %mallocP to i32*
6711 %val1 = add i32 4, 4
6712 %result1 = call i32 @llvm.atomic.swap.i32.p0i32( i32* %ptr, i32 %val1 )
6713 <i>; yields {i32}:result1 = 4</i>
6714 %stored1 = icmp eq i32 %result1, 4 <i>; yields {i1}:stored1 = true</i>
6715 %memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 8</i>
6717 %val2 = add i32 1, 1
6718 %result2 = call i32 @llvm.atomic.swap.i32.p0i32( i32* %ptr, i32 %val2 )
6719 <i>; yields {i32}:result2 = 8</i>
6721 %stored2 = icmp eq i32 %result2, 8 <i>; yields {i1}:stored2 = true</i>
6722 %memval2 = load i32* %ptr <i>; yields {i32}:memval2 = 2</i>
6727 <!-- _______________________________________________________________________ -->
6728 <div class="doc_subsubsection">
6729 <a name="int_atomic_load_add">'<tt>llvm.atomic.load.add.*</tt>' Intrinsic</a>
6733 <div class="doc_text">
6736 <p>This is an overloaded intrinsic. You can use <tt>llvm.atomic.load.add</tt> on
6737 any integer bit width. Not all targets support all bit widths however.</p>
6740 declare i8 @llvm.atomic.load.add.i8..p0i8( i8* <ptr>, i8 <delta> )
6741 declare i16 @llvm.atomic.load.add.i16..p0i16( i16* <ptr>, i16 <delta> )
6742 declare i32 @llvm.atomic.load.add.i32..p0i32( i32* <ptr>, i32 <delta> )
6743 declare i64 @llvm.atomic.load.add.i64..p0i64( i64* <ptr>, i64 <delta> )
6747 <p>This intrinsic adds <tt>delta</tt> to the value stored in memory
6748 at <tt>ptr</tt>. It yields the original value at <tt>ptr</tt>.</p>
6751 <p>The intrinsic takes two arguments, the first a pointer to an integer value
6752 and the second an integer value. The result is also an integer value. These
6753 integer types can have any bit width, but they must all have the same bit
6754 width. The targets may only lower integer representations they support.</p>
6757 <p>This intrinsic does a series of operations atomically. It first loads the
6758 value stored at <tt>ptr</tt>. It then adds <tt>delta</tt>, stores the result
6759 to <tt>ptr</tt>. It yields the original value stored at <tt>ptr</tt>.</p>
6763 %mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
6764 %ptr = bitcast i8* %mallocP to i32*
6766 %result1 = call i32 @llvm.atomic.load.add.i32.p0i32( i32* %ptr, i32 4 )
6767 <i>; yields {i32}:result1 = 4</i>
6768 %result2 = call i32 @llvm.atomic.load.add.i32.p0i32( i32* %ptr, i32 2 )
6769 <i>; yields {i32}:result2 = 8</i>
6770 %result3 = call i32 @llvm.atomic.load.add.i32.p0i32( i32* %ptr, i32 5 )
6771 <i>; yields {i32}:result3 = 10</i>
6772 %memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 15</i>
6777 <!-- _______________________________________________________________________ -->
6778 <div class="doc_subsubsection">
6779 <a name="int_atomic_load_sub">'<tt>llvm.atomic.load.sub.*</tt>' Intrinsic</a>
6783 <div class="doc_text">
6786 <p>This is an overloaded intrinsic. You can use <tt>llvm.atomic.load.sub</tt> on
6787 any integer bit width and for different address spaces. Not all targets
6788 support all bit widths however.</p>
6791 declare i8 @llvm.atomic.load.sub.i8.p0i32( i8* <ptr>, i8 <delta> )
6792 declare i16 @llvm.atomic.load.sub.i16.p0i32( i16* <ptr>, i16 <delta> )
6793 declare i32 @llvm.atomic.load.sub.i32.p0i32( i32* <ptr>, i32 <delta> )
6794 declare i64 @llvm.atomic.load.sub.i64.p0i32( i64* <ptr>, i64 <delta> )
6798 <p>This intrinsic subtracts <tt>delta</tt> to the value stored in memory at
6799 <tt>ptr</tt>. It yields the original value at <tt>ptr</tt>.</p>
6802 <p>The intrinsic takes two arguments, the first a pointer to an integer value
6803 and the second an integer value. The result is also an integer value. These
6804 integer types can have any bit width, but they must all have the same bit
6805 width. The targets may only lower integer representations they support.</p>
6808 <p>This intrinsic does a series of operations atomically. It first loads the
6809 value stored at <tt>ptr</tt>. It then subtracts <tt>delta</tt>, stores the
6810 result to <tt>ptr</tt>. It yields the original value stored
6811 at <tt>ptr</tt>.</p>
6815 %mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
6816 %ptr = bitcast i8* %mallocP to i32*
6818 %result1 = call i32 @llvm.atomic.load.sub.i32.p0i32( i32* %ptr, i32 4 )
6819 <i>; yields {i32}:result1 = 8</i>
6820 %result2 = call i32 @llvm.atomic.load.sub.i32.p0i32( i32* %ptr, i32 2 )
6821 <i>; yields {i32}:result2 = 4</i>
6822 %result3 = call i32 @llvm.atomic.load.sub.i32.p0i32( i32* %ptr, i32 5 )
6823 <i>; yields {i32}:result3 = 2</i>
6824 %memval1 = load i32* %ptr <i>; yields {i32}:memval1 = -3</i>
6829 <!-- _______________________________________________________________________ -->
6830 <div class="doc_subsubsection">
6831 <a name="int_atomic_load_and">'<tt>llvm.atomic.load.and.*</tt>' Intrinsic</a><br>
6832 <a name="int_atomic_load_nand">'<tt>llvm.atomic.load.nand.*</tt>' Intrinsic</a><br>
6833 <a name="int_atomic_load_or">'<tt>llvm.atomic.load.or.*</tt>' Intrinsic</a><br>
6834 <a name="int_atomic_load_xor">'<tt>llvm.atomic.load.xor.*</tt>' Intrinsic</a><br>
6837 <div class="doc_text">
6840 <p>These are overloaded intrinsics. You can
6841 use <tt>llvm.atomic.load_and</tt>, <tt>llvm.atomic.load_nand</tt>,
6842 <tt>llvm.atomic.load_or</tt>, and <tt>llvm.atomic.load_xor</tt> on any integer
6843 bit width and for different address spaces. Not all targets support all bit
6847 declare i8 @llvm.atomic.load.and.i8.p0i8( i8* <ptr>, i8 <delta> )
6848 declare i16 @llvm.atomic.load.and.i16.p0i16( i16* <ptr>, i16 <delta> )
6849 declare i32 @llvm.atomic.load.and.i32.p0i32( i32* <ptr>, i32 <delta> )
6850 declare i64 @llvm.atomic.load.and.i64.p0i64( i64* <ptr>, i64 <delta> )
6854 declare i8 @llvm.atomic.load.or.i8.p0i8( i8* <ptr>, i8 <delta> )
6855 declare i16 @llvm.atomic.load.or.i16.p0i16( i16* <ptr>, i16 <delta> )
6856 declare i32 @llvm.atomic.load.or.i32.p0i32( i32* <ptr>, i32 <delta> )
6857 declare i64 @llvm.atomic.load.or.i64.p0i64( i64* <ptr>, i64 <delta> )
6861 declare i8 @llvm.atomic.load.nand.i8.p0i32( i8* <ptr>, i8 <delta> )
6862 declare i16 @llvm.atomic.load.nand.i16.p0i32( i16* <ptr>, i16 <delta> )
6863 declare i32 @llvm.atomic.load.nand.i32.p0i32( i32* <ptr>, i32 <delta> )
6864 declare i64 @llvm.atomic.load.nand.i64.p0i32( i64* <ptr>, i64 <delta> )
6868 declare i8 @llvm.atomic.load.xor.i8.p0i32( i8* <ptr>, i8 <delta> )
6869 declare i16 @llvm.atomic.load.xor.i16.p0i32( i16* <ptr>, i16 <delta> )
6870 declare i32 @llvm.atomic.load.xor.i32.p0i32( i32* <ptr>, i32 <delta> )
6871 declare i64 @llvm.atomic.load.xor.i64.p0i32( i64* <ptr>, i64 <delta> )
6875 <p>These intrinsics bitwise the operation (and, nand, or, xor) <tt>delta</tt> to
6876 the value stored in memory at <tt>ptr</tt>. It yields the original value
6877 at <tt>ptr</tt>.</p>
6880 <p>These intrinsics take two arguments, the first a pointer to an integer value
6881 and the second an integer value. The result is also an integer value. These
6882 integer types can have any bit width, but they must all have the same bit
6883 width. The targets may only lower integer representations they support.</p>
6886 <p>These intrinsics does a series of operations atomically. They first load the
6887 value stored at <tt>ptr</tt>. They then do the bitwise
6888 operation <tt>delta</tt>, store the result to <tt>ptr</tt>. They yield the
6889 original value stored at <tt>ptr</tt>.</p>
6893 %mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
6894 %ptr = bitcast i8* %mallocP to i32*
6895 store i32 0x0F0F, %ptr
6896 %result0 = call i32 @llvm.atomic.load.nand.i32.p0i32( i32* %ptr, i32 0xFF )
6897 <i>; yields {i32}:result0 = 0x0F0F</i>
6898 %result1 = call i32 @llvm.atomic.load.and.i32.p0i32( i32* %ptr, i32 0xFF )
6899 <i>; yields {i32}:result1 = 0xFFFFFFF0</i>
6900 %result2 = call i32 @llvm.atomic.load.or.i32.p0i32( i32* %ptr, i32 0F )
6901 <i>; yields {i32}:result2 = 0xF0</i>
6902 %result3 = call i32 @llvm.atomic.load.xor.i32.p0i32( i32* %ptr, i32 0F )
6903 <i>; yields {i32}:result3 = FF</i>
6904 %memval1 = load i32* %ptr <i>; yields {i32}:memval1 = F0</i>
6909 <!-- _______________________________________________________________________ -->
6910 <div class="doc_subsubsection">
6911 <a name="int_atomic_load_max">'<tt>llvm.atomic.load.max.*</tt>' Intrinsic</a><br>
6912 <a name="int_atomic_load_min">'<tt>llvm.atomic.load.min.*</tt>' Intrinsic</a><br>
6913 <a name="int_atomic_load_umax">'<tt>llvm.atomic.load.umax.*</tt>' Intrinsic</a><br>
6914 <a name="int_atomic_load_umin">'<tt>llvm.atomic.load.umin.*</tt>' Intrinsic</a><br>
6917 <div class="doc_text">
6920 <p>These are overloaded intrinsics. You can use <tt>llvm.atomic.load_max</tt>,
6921 <tt>llvm.atomic.load_min</tt>, <tt>llvm.atomic.load_umax</tt>, and
6922 <tt>llvm.atomic.load_umin</tt> on any integer bit width and for different
6923 address spaces. Not all targets support all bit widths however.</p>
6926 declare i8 @llvm.atomic.load.max.i8.p0i8( i8* <ptr>, i8 <delta> )
6927 declare i16 @llvm.atomic.load.max.i16.p0i16( i16* <ptr>, i16 <delta> )
6928 declare i32 @llvm.atomic.load.max.i32.p0i32( i32* <ptr>, i32 <delta> )
6929 declare i64 @llvm.atomic.load.max.i64.p0i64( i64* <ptr>, i64 <delta> )
6933 declare i8 @llvm.atomic.load.min.i8.p0i8( i8* <ptr>, i8 <delta> )
6934 declare i16 @llvm.atomic.load.min.i16.p0i16( i16* <ptr>, i16 <delta> )
6935 declare i32 @llvm.atomic.load.min.i32..p0i32( i32* <ptr>, i32 <delta> )
6936 declare i64 @llvm.atomic.load.min.i64..p0i64( i64* <ptr>, i64 <delta> )
6940 declare i8 @llvm.atomic.load.umax.i8.p0i8( i8* <ptr>, i8 <delta> )
6941 declare i16 @llvm.atomic.load.umax.i16.p0i16( i16* <ptr>, i16 <delta> )
6942 declare i32 @llvm.atomic.load.umax.i32.p0i32( i32* <ptr>, i32 <delta> )
6943 declare i64 @llvm.atomic.load.umax.i64.p0i64( i64* <ptr>, i64 <delta> )
6947 declare i8 @llvm.atomic.load.umin.i8..p0i8( i8* <ptr>, i8 <delta> )
6948 declare i16 @llvm.atomic.load.umin.i16.p0i16( i16* <ptr>, i16 <delta> )
6949 declare i32 @llvm.atomic.load.umin.i32..p0i32( i32* <ptr>, i32 <delta> )
6950 declare i64 @llvm.atomic.load.umin.i64..p0i64( i64* <ptr>, i64 <delta> )
6954 <p>These intrinsics takes the signed or unsigned minimum or maximum of
6955 <tt>delta</tt> and the value stored in memory at <tt>ptr</tt>. It yields the
6956 original value at <tt>ptr</tt>.</p>
6959 <p>These intrinsics take two arguments, the first a pointer to an integer value
6960 and the second an integer value. The result is also an integer value. These
6961 integer types can have any bit width, but they must all have the same bit
6962 width. The targets may only lower integer representations they support.</p>
6965 <p>These intrinsics does a series of operations atomically. They first load the
6966 value stored at <tt>ptr</tt>. They then do the signed or unsigned min or
6967 max <tt>delta</tt> and the value, store the result to <tt>ptr</tt>. They
6968 yield the original value stored at <tt>ptr</tt>.</p>
6972 %mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
6973 %ptr = bitcast i8* %mallocP to i32*
6975 %result0 = call i32 @llvm.atomic.load.min.i32.p0i32( i32* %ptr, i32 -2 )
6976 <i>; yields {i32}:result0 = 7</i>
6977 %result1 = call i32 @llvm.atomic.load.max.i32.p0i32( i32* %ptr, i32 8 )
6978 <i>; yields {i32}:result1 = -2</i>
6979 %result2 = call i32 @llvm.atomic.load.umin.i32.p0i32( i32* %ptr, i32 10 )
6980 <i>; yields {i32}:result2 = 8</i>
6981 %result3 = call i32 @llvm.atomic.load.umax.i32.p0i32( i32* %ptr, i32 30 )
6982 <i>; yields {i32}:result3 = 8</i>
6983 %memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 30</i>
6989 <!-- ======================================================================= -->
6990 <div class="doc_subsection">
6991 <a name="int_memorymarkers">Memory Use Markers</a>
6994 <div class="doc_text">
6996 <p>This class of intrinsics exists to information about the lifetime of memory
6997 objects and ranges where variables are immutable.</p>
7001 <!-- _______________________________________________________________________ -->
7002 <div class="doc_subsubsection">
7003 <a name="int_lifetime_start">'<tt>llvm.lifetime.start</tt>' Intrinsic</a>
7006 <div class="doc_text">
7010 declare void @llvm.lifetime.start(i64 <size>, i8* nocapture <ptr>)
7014 <p>The '<tt>llvm.lifetime.start</tt>' intrinsic specifies the start of a memory
7015 object's lifetime.</p>
7018 <p>The first argument is a constant integer representing the size of the
7019 object, or -1 if it is variable sized. The second argument is a pointer to
7023 <p>This intrinsic indicates that before this point in the code, the value of the
7024 memory pointed to by <tt>ptr</tt> is dead. This means that it is known to
7025 never be used and has an undefined value. A load from the pointer that
7026 precedes this intrinsic can be replaced with
7027 <tt>'<a href="#undefvalues">undef</a>'</tt>.</p>
7031 <!-- _______________________________________________________________________ -->
7032 <div class="doc_subsubsection">
7033 <a name="int_lifetime_end">'<tt>llvm.lifetime.end</tt>' Intrinsic</a>
7036 <div class="doc_text">
7040 declare void @llvm.lifetime.end(i64 <size>, i8* nocapture <ptr>)
7044 <p>The '<tt>llvm.lifetime.end</tt>' intrinsic specifies the end of a memory
7045 object's lifetime.</p>
7048 <p>The first argument is a constant integer representing the size of the
7049 object, or -1 if it is variable sized. The second argument is a pointer to
7053 <p>This intrinsic indicates that after this point in the code, the value of the
7054 memory pointed to by <tt>ptr</tt> is dead. This means that it is known to
7055 never be used and has an undefined value. Any stores into the memory object
7056 following this intrinsic may be removed as dead.
7060 <!-- _______________________________________________________________________ -->
7061 <div class="doc_subsubsection">
7062 <a name="int_invariant_start">'<tt>llvm.invariant.start</tt>' Intrinsic</a>
7065 <div class="doc_text">
7069 declare {}* @llvm.invariant.start(i64 <size>, i8* nocapture <ptr>) readonly
7073 <p>The '<tt>llvm.invariant.start</tt>' intrinsic specifies that the contents of
7074 a memory object will not change.</p>
7077 <p>The first argument is a constant integer representing the size of the
7078 object, or -1 if it is variable sized. The second argument is a pointer to
7082 <p>This intrinsic indicates that until an <tt>llvm.invariant.end</tt> that uses
7083 the return value, the referenced memory location is constant and
7088 <!-- _______________________________________________________________________ -->
7089 <div class="doc_subsubsection">
7090 <a name="int_invariant_end">'<tt>llvm.invariant.end</tt>' Intrinsic</a>
7093 <div class="doc_text">
7097 declare void @llvm.invariant.end({}* <start>, i64 <size>, i8* nocapture <ptr>)
7101 <p>The '<tt>llvm.invariant.end</tt>' intrinsic specifies that the contents of
7102 a memory object are mutable.</p>
7105 <p>The first argument is the matching <tt>llvm.invariant.start</tt> intrinsic.
7106 The second argument is a constant integer representing the size of the
7107 object, or -1 if it is variable sized and the third argument is a pointer
7111 <p>This intrinsic indicates that the memory is mutable again.</p>
7115 <!-- ======================================================================= -->
7116 <div class="doc_subsection">
7117 <a name="int_general">General Intrinsics</a>
7120 <div class="doc_text">
7122 <p>This class of intrinsics is designed to be generic and has no specific
7127 <!-- _______________________________________________________________________ -->
7128 <div class="doc_subsubsection">
7129 <a name="int_var_annotation">'<tt>llvm.var.annotation</tt>' Intrinsic</a>
7132 <div class="doc_text">
7136 declare void @llvm.var.annotation(i8* <val>, i8* <str>, i8* <str>, i32 <int> )
7140 <p>The '<tt>llvm.var.annotation</tt>' intrinsic.</p>
7143 <p>The first argument is a pointer to a value, the second is a pointer to a
7144 global string, the third is a pointer to a global string which is the source
7145 file name, and the last argument is the line number.</p>
7148 <p>This intrinsic allows annotation of local variables with arbitrary strings.
7149 This can be useful for special purpose optimizations that want to look for
7150 these annotations. These have no other defined use, they are ignored by code
7151 generation and optimization.</p>
7155 <!-- _______________________________________________________________________ -->
7156 <div class="doc_subsubsection">
7157 <a name="int_annotation">'<tt>llvm.annotation.*</tt>' Intrinsic</a>
7160 <div class="doc_text">
7163 <p>This is an overloaded intrinsic. You can use '<tt>llvm.annotation</tt>' on
7164 any integer bit width.</p>
7167 declare i8 @llvm.annotation.i8(i8 <val>, i8* <str>, i8* <str>, i32 <int> )
7168 declare i16 @llvm.annotation.i16(i16 <val>, i8* <str>, i8* <str>, i32 <int> )
7169 declare i32 @llvm.annotation.i32(i32 <val>, i8* <str>, i8* <str>, i32 <int> )
7170 declare i64 @llvm.annotation.i64(i64 <val>, i8* <str>, i8* <str>, i32 <int> )
7171 declare i256 @llvm.annotation.i256(i256 <val>, i8* <str>, i8* <str>, i32 <int> )
7175 <p>The '<tt>llvm.annotation</tt>' intrinsic.</p>
7178 <p>The first argument is an integer value (result of some expression), the
7179 second is a pointer to a global string, the third is a pointer to a global
7180 string which is the source file name, and the last argument is the line
7181 number. It returns the value of the first argument.</p>
7184 <p>This intrinsic allows annotations to be put on arbitrary expressions with
7185 arbitrary strings. This can be useful for special purpose optimizations that
7186 want to look for these annotations. These have no other defined use, they
7187 are ignored by code generation and optimization.</p>
7191 <!-- _______________________________________________________________________ -->
7192 <div class="doc_subsubsection">
7193 <a name="int_trap">'<tt>llvm.trap</tt>' Intrinsic</a>
7196 <div class="doc_text">
7200 declare void @llvm.trap()
7204 <p>The '<tt>llvm.trap</tt>' intrinsic.</p>
7210 <p>This intrinsics is lowered to the target dependent trap instruction. If the
7211 target does not have a trap instruction, this intrinsic will be lowered to
7212 the call of the <tt>abort()</tt> function.</p>
7216 <!-- _______________________________________________________________________ -->
7217 <div class="doc_subsubsection">
7218 <a name="int_stackprotector">'<tt>llvm.stackprotector</tt>' Intrinsic</a>
7221 <div class="doc_text">
7225 declare void @llvm.stackprotector( i8* <guard>, i8** <slot> )
7229 <p>The <tt>llvm.stackprotector</tt> intrinsic takes the <tt>guard</tt> and
7230 stores it onto the stack at <tt>slot</tt>. The stack slot is adjusted to
7231 ensure that it is placed on the stack before local variables.</p>
7234 <p>The <tt>llvm.stackprotector</tt> intrinsic requires two pointer
7235 arguments. The first argument is the value loaded from the stack
7236 guard <tt>@__stack_chk_guard</tt>. The second variable is an <tt>alloca</tt>
7237 that has enough space to hold the value of the guard.</p>
7240 <p>This intrinsic causes the prologue/epilogue inserter to force the position of
7241 the <tt>AllocaInst</tt> stack slot to be before local variables on the
7242 stack. This is to ensure that if a local variable on the stack is
7243 overwritten, it will destroy the value of the guard. When the function exits,
7244 the guard on the stack is checked against the original guard. If they're
7245 different, then the program aborts by calling the <tt>__stack_chk_fail()</tt>
7250 <!-- *********************************************************************** -->
7253 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
7254 src="http://jigsaw.w3.org/css-validator/images/vcss-blue" alt="Valid CSS"></a>
7255 <a href="http://validator.w3.org/check/referer"><img
7256 src="http://www.w3.org/Icons/valid-html401-blue" alt="Valid HTML 4.01"></a>
7258 <a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
7259 <a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br>
7260 Last modified: $Date$