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="#namedmetadatastructure">Named Metadata</a></li>
47 <li><a href="#paramattrs">Parameter Attributes</a></li>
48 <li><a href="#fnattrs">Function Attributes</a></li>
49 <li><a href="#gc">Garbage Collector Names</a></li>
50 <li><a href="#moduleasm">Module-Level Inline Assembly</a></li>
51 <li><a href="#datalayout">Data Layout</a></li>
52 <li><a href="#pointeraliasing">Pointer Aliasing Rules</a></li>
55 <li><a href="#typesystem">Type System</a>
57 <li><a href="#t_classifications">Type Classifications</a></li>
58 <li><a href="#t_primitive">Primitive Types</a>
60 <li><a href="#t_integer">Integer Type</a></li>
61 <li><a href="#t_floating">Floating Point Types</a></li>
62 <li><a href="#t_void">Void Type</a></li>
63 <li><a href="#t_label">Label Type</a></li>
64 <li><a href="#t_metadata">Metadata Type</a></li>
67 <li><a href="#t_derived">Derived Types</a>
69 <li><a href="#t_array">Array Type</a></li>
70 <li><a href="#t_function">Function Type</a></li>
71 <li><a href="#t_pointer">Pointer Type</a></li>
72 <li><a href="#t_struct">Structure Type</a></li>
73 <li><a href="#t_pstruct">Packed Structure Type</a></li>
74 <li><a href="#t_vector">Vector Type</a></li>
75 <li><a href="#t_opaque">Opaque Type</a></li>
78 <li><a href="#t_uprefs">Type Up-references</a></li>
81 <li><a href="#constants">Constants</a>
83 <li><a href="#simpleconstants">Simple Constants</a></li>
84 <li><a href="#complexconstants">Complex Constants</a></li>
85 <li><a href="#globalconstants">Global Variable and Function Addresses</a></li>
86 <li><a href="#undefvalues">Undefined Values</a></li>
87 <li><a href="#blockaddress">Addresses of Basic Blocks</a></li>
88 <li><a href="#constantexprs">Constant Expressions</a></li>
91 <li><a href="#othervalues">Other Values</a>
93 <li><a href="#inlineasm">Inline Assembler Expressions</a></li>
94 <li><a href="#metadata">Metadata Nodes and Metadata Strings</a></li>
97 <li><a href="#intrinsic_globals">Intrinsic Global Variables</a>
99 <li><a href="#intg_used">The '<tt>llvm.used</tt>' Global Variable</a></li>
100 <li><a href="#intg_compiler_used">The '<tt>llvm.compiler.used</tt>'
101 Global Variable</a></li>
102 <li><a href="#intg_global_ctors">The '<tt>llvm.global_ctors</tt>'
103 Global Variable</a></li>
104 <li><a href="#intg_global_dtors">The '<tt>llvm.global_dtors</tt>'
105 Global Variable</a></li>
108 <li><a href="#instref">Instruction Reference</a>
110 <li><a href="#terminators">Terminator Instructions</a>
112 <li><a href="#i_ret">'<tt>ret</tt>' Instruction</a></li>
113 <li><a href="#i_br">'<tt>br</tt>' Instruction</a></li>
114 <li><a href="#i_switch">'<tt>switch</tt>' Instruction</a></li>
115 <li><a href="#i_indirectbr">'<tt>indirectbr</tt>' Instruction</a></li>
116 <li><a href="#i_invoke">'<tt>invoke</tt>' Instruction</a></li>
117 <li><a href="#i_unwind">'<tt>unwind</tt>' Instruction</a></li>
118 <li><a href="#i_unreachable">'<tt>unreachable</tt>' Instruction</a></li>
121 <li><a href="#binaryops">Binary Operations</a>
123 <li><a href="#i_add">'<tt>add</tt>' Instruction</a></li>
124 <li><a href="#i_fadd">'<tt>fadd</tt>' Instruction</a></li>
125 <li><a href="#i_sub">'<tt>sub</tt>' Instruction</a></li>
126 <li><a href="#i_fsub">'<tt>fsub</tt>' Instruction</a></li>
127 <li><a href="#i_mul">'<tt>mul</tt>' Instruction</a></li>
128 <li><a href="#i_fmul">'<tt>fmul</tt>' Instruction</a></li>
129 <li><a href="#i_udiv">'<tt>udiv</tt>' Instruction</a></li>
130 <li><a href="#i_sdiv">'<tt>sdiv</tt>' Instruction</a></li>
131 <li><a href="#i_fdiv">'<tt>fdiv</tt>' Instruction</a></li>
132 <li><a href="#i_urem">'<tt>urem</tt>' Instruction</a></li>
133 <li><a href="#i_srem">'<tt>srem</tt>' Instruction</a></li>
134 <li><a href="#i_frem">'<tt>frem</tt>' Instruction</a></li>
137 <li><a href="#bitwiseops">Bitwise Binary Operations</a>
139 <li><a href="#i_shl">'<tt>shl</tt>' Instruction</a></li>
140 <li><a href="#i_lshr">'<tt>lshr</tt>' Instruction</a></li>
141 <li><a href="#i_ashr">'<tt>ashr</tt>' Instruction</a></li>
142 <li><a href="#i_and">'<tt>and</tt>' Instruction</a></li>
143 <li><a href="#i_or">'<tt>or</tt>' Instruction</a></li>
144 <li><a href="#i_xor">'<tt>xor</tt>' Instruction</a></li>
147 <li><a href="#vectorops">Vector Operations</a>
149 <li><a href="#i_extractelement">'<tt>extractelement</tt>' Instruction</a></li>
150 <li><a href="#i_insertelement">'<tt>insertelement</tt>' Instruction</a></li>
151 <li><a href="#i_shufflevector">'<tt>shufflevector</tt>' Instruction</a></li>
154 <li><a href="#aggregateops">Aggregate Operations</a>
156 <li><a href="#i_extractvalue">'<tt>extractvalue</tt>' Instruction</a></li>
157 <li><a href="#i_insertvalue">'<tt>insertvalue</tt>' Instruction</a></li>
160 <li><a href="#memoryops">Memory Access and Addressing Operations</a>
162 <li><a href="#i_alloca">'<tt>alloca</tt>' Instruction</a></li>
163 <li><a href="#i_load">'<tt>load</tt>' Instruction</a></li>
164 <li><a href="#i_store">'<tt>store</tt>' Instruction</a></li>
165 <li><a href="#i_getelementptr">'<tt>getelementptr</tt>' Instruction</a></li>
168 <li><a href="#convertops">Conversion Operations</a>
170 <li><a href="#i_trunc">'<tt>trunc .. to</tt>' Instruction</a></li>
171 <li><a href="#i_zext">'<tt>zext .. to</tt>' Instruction</a></li>
172 <li><a href="#i_sext">'<tt>sext .. to</tt>' Instruction</a></li>
173 <li><a href="#i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a></li>
174 <li><a href="#i_fpext">'<tt>fpext .. to</tt>' Instruction</a></li>
175 <li><a href="#i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a></li>
176 <li><a href="#i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a></li>
177 <li><a href="#i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a></li>
178 <li><a href="#i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a></li>
179 <li><a href="#i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a></li>
180 <li><a href="#i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a></li>
181 <li><a href="#i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a></li>
184 <li><a href="#otherops">Other Operations</a>
186 <li><a href="#i_icmp">'<tt>icmp</tt>' Instruction</a></li>
187 <li><a href="#i_fcmp">'<tt>fcmp</tt>' Instruction</a></li>
188 <li><a href="#i_phi">'<tt>phi</tt>' Instruction</a></li>
189 <li><a href="#i_select">'<tt>select</tt>' Instruction</a></li>
190 <li><a href="#i_call">'<tt>call</tt>' Instruction</a></li>
191 <li><a href="#i_va_arg">'<tt>va_arg</tt>' Instruction</a></li>
196 <li><a href="#intrinsics">Intrinsic Functions</a>
198 <li><a href="#int_varargs">Variable Argument Handling Intrinsics</a>
200 <li><a href="#int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a></li>
201 <li><a href="#int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a></li>
202 <li><a href="#int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a></li>
205 <li><a href="#int_gc">Accurate Garbage Collection Intrinsics</a>
207 <li><a href="#int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a></li>
208 <li><a href="#int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a></li>
209 <li><a href="#int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a></li>
212 <li><a href="#int_codegen">Code Generator Intrinsics</a>
214 <li><a href="#int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a></li>
215 <li><a href="#int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a></li>
216 <li><a href="#int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a></li>
217 <li><a href="#int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a></li>
218 <li><a href="#int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a></li>
219 <li><a href="#int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a></li>
220 <li><a href="#int_readcyclecounter"><tt>llvm.readcyclecounter</tt>' Intrinsic</a></li>
223 <li><a href="#int_libc">Standard C Library Intrinsics</a>
225 <li><a href="#int_memcpy">'<tt>llvm.memcpy.*</tt>' Intrinsic</a></li>
226 <li><a href="#int_memmove">'<tt>llvm.memmove.*</tt>' Intrinsic</a></li>
227 <li><a href="#int_memset">'<tt>llvm.memset.*</tt>' Intrinsic</a></li>
228 <li><a href="#int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a></li>
229 <li><a href="#int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a></li>
230 <li><a href="#int_sin">'<tt>llvm.sin.*</tt>' Intrinsic</a></li>
231 <li><a href="#int_cos">'<tt>llvm.cos.*</tt>' Intrinsic</a></li>
232 <li><a href="#int_pow">'<tt>llvm.pow.*</tt>' Intrinsic</a></li>
235 <li><a href="#int_manip">Bit Manipulation Intrinsics</a>
237 <li><a href="#int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a></li>
238 <li><a href="#int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic </a></li>
239 <li><a href="#int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic </a></li>
240 <li><a href="#int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic </a></li>
243 <li><a href="#int_overflow">Arithmetic with Overflow Intrinsics</a>
245 <li><a href="#int_sadd_overflow">'<tt>llvm.sadd.with.overflow.*</tt> Intrinsics</a></li>
246 <li><a href="#int_uadd_overflow">'<tt>llvm.uadd.with.overflow.*</tt> Intrinsics</a></li>
247 <li><a href="#int_ssub_overflow">'<tt>llvm.ssub.with.overflow.*</tt> Intrinsics</a></li>
248 <li><a href="#int_usub_overflow">'<tt>llvm.usub.with.overflow.*</tt> Intrinsics</a></li>
249 <li><a href="#int_smul_overflow">'<tt>llvm.smul.with.overflow.*</tt> Intrinsics</a></li>
250 <li><a href="#int_umul_overflow">'<tt>llvm.umul.with.overflow.*</tt> Intrinsics</a></li>
253 <li><a href="#int_debugger">Debugger intrinsics</a></li>
254 <li><a href="#int_eh">Exception Handling intrinsics</a></li>
255 <li><a href="#int_trampoline">Trampoline Intrinsic</a>
257 <li><a href="#int_it">'<tt>llvm.init.trampoline</tt>' Intrinsic</a></li>
260 <li><a href="#int_atomics">Atomic intrinsics</a>
262 <li><a href="#int_memory_barrier"><tt>llvm.memory_barrier</tt></a></li>
263 <li><a href="#int_atomic_cmp_swap"><tt>llvm.atomic.cmp.swap</tt></a></li>
264 <li><a href="#int_atomic_swap"><tt>llvm.atomic.swap</tt></a></li>
265 <li><a href="#int_atomic_load_add"><tt>llvm.atomic.load.add</tt></a></li>
266 <li><a href="#int_atomic_load_sub"><tt>llvm.atomic.load.sub</tt></a></li>
267 <li><a href="#int_atomic_load_and"><tt>llvm.atomic.load.and</tt></a></li>
268 <li><a href="#int_atomic_load_nand"><tt>llvm.atomic.load.nand</tt></a></li>
269 <li><a href="#int_atomic_load_or"><tt>llvm.atomic.load.or</tt></a></li>
270 <li><a href="#int_atomic_load_xor"><tt>llvm.atomic.load.xor</tt></a></li>
271 <li><a href="#int_atomic_load_max"><tt>llvm.atomic.load.max</tt></a></li>
272 <li><a href="#int_atomic_load_min"><tt>llvm.atomic.load.min</tt></a></li>
273 <li><a href="#int_atomic_load_umax"><tt>llvm.atomic.load.umax</tt></a></li>
274 <li><a href="#int_atomic_load_umin"><tt>llvm.atomic.load.umin</tt></a></li>
277 <li><a href="#int_memorymarkers">Memory Use Markers</a>
279 <li><a href="#int_lifetime_start"><tt>llvm.lifetime.start</tt></a></li>
280 <li><a href="#int_lifetime_end"><tt>llvm.lifetime.end</tt></a></li>
281 <li><a href="#int_invariant_start"><tt>llvm.invariant.start</tt></a></li>
282 <li><a href="#int_invariant_end"><tt>llvm.invariant.end</tt></a></li>
285 <li><a href="#int_general">General intrinsics</a>
287 <li><a href="#int_var_annotation">
288 '<tt>llvm.var.annotation</tt>' Intrinsic</a></li>
289 <li><a href="#int_annotation">
290 '<tt>llvm.annotation.*</tt>' Intrinsic</a></li>
291 <li><a href="#int_trap">
292 '<tt>llvm.trap</tt>' Intrinsic</a></li>
293 <li><a href="#int_stackprotector">
294 '<tt>llvm.stackprotector</tt>' Intrinsic</a></li>
295 <li><a href="#int_objectsize">
296 '<tt>llvm.objectsize</tt>' Intrinsic</a></li>
303 <div class="doc_author">
304 <p>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a>
305 and <a href="mailto:vadve@cs.uiuc.edu">Vikram Adve</a></p>
308 <!-- *********************************************************************** -->
309 <div class="doc_section"> <a name="abstract">Abstract </a></div>
310 <!-- *********************************************************************** -->
312 <div class="doc_text">
314 <p>This document is a reference manual for the LLVM assembly language. LLVM is
315 a Static Single Assignment (SSA) based representation that provides type
316 safety, low-level operations, flexibility, and the capability of representing
317 'all' high-level languages cleanly. It is the common code representation
318 used throughout all phases of the LLVM compilation strategy.</p>
322 <!-- *********************************************************************** -->
323 <div class="doc_section"> <a name="introduction">Introduction</a> </div>
324 <!-- *********************************************************************** -->
326 <div class="doc_text">
328 <p>The LLVM code representation is designed to be used in three different forms:
329 as an in-memory compiler IR, as an on-disk bitcode representation (suitable
330 for fast loading by a Just-In-Time compiler), and as a human readable
331 assembly language representation. This allows LLVM to provide a powerful
332 intermediate representation for efficient compiler transformations and
333 analysis, while providing a natural means to debug and visualize the
334 transformations. The three different forms of LLVM are all equivalent. This
335 document describes the human readable representation and notation.</p>
337 <p>The LLVM representation aims to be light-weight and low-level while being
338 expressive, typed, and extensible at the same time. It aims to be a
339 "universal IR" of sorts, by being at a low enough level that high-level ideas
340 may be cleanly mapped to it (similar to how microprocessors are "universal
341 IR's", allowing many source languages to be mapped to them). By providing
342 type information, LLVM can be used as the target of optimizations: for
343 example, through pointer analysis, it can be proven that a C automatic
344 variable is never accessed outside of the current function, allowing it to
345 be promoted to a simple SSA value instead of a memory location.</p>
349 <!-- _______________________________________________________________________ -->
350 <div class="doc_subsubsection"> <a name="wellformed">Well-Formedness</a> </div>
352 <div class="doc_text">
354 <p>It is important to note that this document describes 'well formed' LLVM
355 assembly language. There is a difference between what the parser accepts and
356 what is considered 'well formed'. For example, the following instruction is
357 syntactically okay, but not well formed:</p>
359 <div class="doc_code">
361 %x = <a href="#i_add">add</a> i32 1, %x
365 <p>because the definition of <tt>%x</tt> does not dominate all of its uses. The
366 LLVM infrastructure provides a verification pass that may be used to verify
367 that an LLVM module is well formed. This pass is automatically run by the
368 parser after parsing input assembly and by the optimizer before it outputs
369 bitcode. The violations pointed out by the verifier pass indicate bugs in
370 transformation passes or input to the parser.</p>
374 <!-- Describe the typesetting conventions here. -->
376 <!-- *********************************************************************** -->
377 <div class="doc_section"> <a name="identifiers">Identifiers</a> </div>
378 <!-- *********************************************************************** -->
380 <div class="doc_text">
382 <p>LLVM identifiers come in two basic types: global and local. Global
383 identifiers (functions, global variables) begin with the <tt>'@'</tt>
384 character. Local identifiers (register names, types) begin with
385 the <tt>'%'</tt> character. Additionally, there are three different formats
386 for identifiers, for different purposes:</p>
389 <li>Named values are represented as a string of characters with their prefix.
390 For example, <tt>%foo</tt>, <tt>@DivisionByZero</tt>,
391 <tt>%a.really.long.identifier</tt>. The actual regular expression used is
392 '<tt>[%@][a-zA-Z$._][a-zA-Z$._0-9]*</tt>'. Identifiers which require
393 other characters in their names can be surrounded with quotes. Special
394 characters may be escaped using <tt>"\xx"</tt> where <tt>xx</tt> is the
395 ASCII code for the character in hexadecimal. In this way, any character
396 can be used in a name value, even quotes themselves.</li>
398 <li>Unnamed values are represented as an unsigned numeric value with their
399 prefix. For example, <tt>%12</tt>, <tt>@2</tt>, <tt>%44</tt>.</li>
401 <li>Constants, which are described in a <a href="#constants">section about
402 constants</a>, below.</li>
405 <p>LLVM requires that values start with a prefix for two reasons: Compilers
406 don't need to worry about name clashes with reserved words, and the set of
407 reserved words may be expanded in the future without penalty. Additionally,
408 unnamed identifiers allow a compiler to quickly come up with a temporary
409 variable without having to avoid symbol table conflicts.</p>
411 <p>Reserved words in LLVM are very similar to reserved words in other
412 languages. There are keywords for different opcodes
413 ('<tt><a href="#i_add">add</a></tt>',
414 '<tt><a href="#i_bitcast">bitcast</a></tt>',
415 '<tt><a href="#i_ret">ret</a></tt>', etc...), for primitive type names
416 ('<tt><a href="#t_void">void</a></tt>',
417 '<tt><a href="#t_primitive">i32</a></tt>', etc...), and others. These
418 reserved words cannot conflict with variable names, because none of them
419 start with a prefix character (<tt>'%'</tt> or <tt>'@'</tt>).</p>
421 <p>Here is an example of LLVM code to multiply the integer variable
422 '<tt>%X</tt>' by 8:</p>
426 <div class="doc_code">
428 %result = <a href="#i_mul">mul</a> i32 %X, 8
432 <p>After strength reduction:</p>
434 <div class="doc_code">
436 %result = <a href="#i_shl">shl</a> i32 %X, i8 3
440 <p>And the hard way:</p>
442 <div class="doc_code">
444 %0 = <a href="#i_add">add</a> i32 %X, %X <i>; yields {i32}:%0</i>
445 %1 = <a href="#i_add">add</a> i32 %0, %0 <i>; yields {i32}:%1</i>
446 %result = <a href="#i_add">add</a> i32 %1, %1
450 <p>This last way of multiplying <tt>%X</tt> by 8 illustrates several important
451 lexical features of LLVM:</p>
454 <li>Comments are delimited with a '<tt>;</tt>' and go until the end of
457 <li>Unnamed temporaries are created when the result of a computation is not
458 assigned to a named value.</li>
460 <li>Unnamed temporaries are numbered sequentially</li>
463 <p>It also shows a convention that we follow in this document. When
464 demonstrating instructions, we will follow an instruction with a comment that
465 defines the type and name of value produced. Comments are shown in italic
470 <!-- *********************************************************************** -->
471 <div class="doc_section"> <a name="highlevel">High Level Structure</a> </div>
472 <!-- *********************************************************************** -->
474 <!-- ======================================================================= -->
475 <div class="doc_subsection"> <a name="modulestructure">Module Structure</a>
478 <div class="doc_text">
480 <p>LLVM programs are composed of "Module"s, each of which is a translation unit
481 of the input programs. Each module consists of functions, global variables,
482 and symbol table entries. Modules may be combined together with the LLVM
483 linker, which merges function (and global variable) definitions, resolves
484 forward declarations, and merges symbol table entries. Here is an example of
485 the "hello world" module:</p>
487 <div class="doc_code">
489 <i>; Declare the string constant as a global constant.</i>
490 <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>
492 <i>; External declaration of the puts function</i>
493 <a href="#functionstructure">declare</a> i32 @puts(i8 *) <i>; i32(i8 *)* </i>
495 <i>; Definition of main function</i>
496 define i32 @main() { <i>; i32()* </i>
497 <i>; Convert [13 x i8]* to i8 *...</i>
498 %cast210 = <a href="#i_getelementptr">getelementptr</a> [13 x i8]* @.LC0, i64 0, i64 0 <i>; i8 *</i>
500 <i>; Call puts function to write out the string to stdout.</i>
501 <a href="#i_call">call</a> i32 @puts(i8 * %cast210) <i>; i32</i>
502 <a href="#i_ret">ret</a> i32 0<br>}
504 <i>; Named metadata</i>
505 !1 = metadata !{i32 41}
510 <p>This example is made up of a <a href="#globalvars">global variable</a> named
511 "<tt>.LC0</tt>", an external declaration of the "<tt>puts</tt>" function,
512 a <a href="#functionstructure">function definition</a> for
513 "<tt>main</tt>" and <a href="#namedmetadatastructure">named metadata</a>
516 <p>In general, a module is made up of a list of global values, where both
517 functions and global variables are global values. Global values are
518 represented by a pointer to a memory location (in this case, a pointer to an
519 array of char, and a pointer to a function), and have one of the
520 following <a href="#linkage">linkage types</a>.</p>
524 <!-- ======================================================================= -->
525 <div class="doc_subsection">
526 <a name="linkage">Linkage Types</a>
529 <div class="doc_text">
531 <p>All Global Variables and Functions have one of the following types of
535 <dt><tt><b><a name="linkage_private">private</a></b></tt></dt>
536 <dd>Global values with private linkage are only directly accessible by objects
537 in the current module. In particular, linking code into a module with an
538 private global value may cause the private to be renamed as necessary to
539 avoid collisions. Because the symbol is private to the module, all
540 references can be updated. This doesn't show up in any symbol table in the
543 <dt><tt><b><a name="linkage_linker_private">linker_private</a></b></tt></dt>
544 <dd>Similar to private, but the symbol is passed through the assembler and
545 removed by the linker after evaluation. Note that (unlike private
546 symbols) linker_private symbols are subject to coalescing by the linker:
547 weak symbols get merged and redefinitions are rejected. However, unlike
548 normal strong symbols, they are removed by the linker from the final
549 linked image (executable or dynamic library).</dd>
551 <dt><tt><b><a name="linkage_internal">internal</a></b></tt></dt>
552 <dd>Similar to private, but the value shows as a local symbol
553 (<tt>STB_LOCAL</tt> in the case of ELF) in the object file. This
554 corresponds to the notion of the '<tt>static</tt>' keyword in C.</dd>
556 <dt><tt><b><a name="linkage_available_externally">available_externally</a></b></tt></dt>
557 <dd>Globals with "<tt>available_externally</tt>" linkage are never emitted
558 into the object file corresponding to the LLVM module. They exist to
559 allow inlining and other optimizations to take place given knowledge of
560 the definition of the global, which is known to be somewhere outside the
561 module. Globals with <tt>available_externally</tt> linkage are allowed to
562 be discarded at will, and are otherwise the same as <tt>linkonce_odr</tt>.
563 This linkage type is only allowed on definitions, not declarations.</dd>
565 <dt><tt><b><a name="linkage_linkonce">linkonce</a></b></tt></dt>
566 <dd>Globals with "<tt>linkonce</tt>" linkage are merged with other globals of
567 the same name when linkage occurs. This can be used to implement
568 some forms of inline functions, templates, or other code which must be
569 generated in each translation unit that uses it, but where the body may
570 be overridden with a more definitive definition later. Unreferenced
571 <tt>linkonce</tt> globals are allowed to be discarded. Note that
572 <tt>linkonce</tt> linkage does not actually allow the optimizer to
573 inline the body of this function into callers because it doesn't know if
574 this definition of the function is the definitive definition within the
575 program or whether it will be overridden by a stronger definition.
576 To enable inlining and other optimizations, use "<tt>linkonce_odr</tt>"
579 <dt><tt><b><a name="linkage_weak">weak</a></b></tt></dt>
580 <dd>"<tt>weak</tt>" linkage has the same merging semantics as
581 <tt>linkonce</tt> linkage, except that unreferenced globals with
582 <tt>weak</tt> linkage may not be discarded. This is used for globals that
583 are declared "weak" in C source code.</dd>
585 <dt><tt><b><a name="linkage_common">common</a></b></tt></dt>
586 <dd>"<tt>common</tt>" linkage is most similar to "<tt>weak</tt>" linkage, but
587 they are used for tentative definitions in C, such as "<tt>int X;</tt>" at
589 Symbols with "<tt>common</tt>" linkage are merged in the same way as
590 <tt>weak symbols</tt>, and they may not be deleted if unreferenced.
591 <tt>common</tt> symbols may not have an explicit section,
592 must have a zero initializer, and may not be marked '<a
593 href="#globalvars"><tt>constant</tt></a>'. Functions and aliases may not
594 have common linkage.</dd>
597 <dt><tt><b><a name="linkage_appending">appending</a></b></tt></dt>
598 <dd>"<tt>appending</tt>" linkage may only be applied to global variables of
599 pointer to array type. When two global variables with appending linkage
600 are linked together, the two global arrays are appended together. This is
601 the LLVM, typesafe, equivalent of having the system linker append together
602 "sections" with identical names when .o files are linked.</dd>
604 <dt><tt><b><a name="linkage_externweak">extern_weak</a></b></tt></dt>
605 <dd>The semantics of this linkage follow the ELF object file model: the symbol
606 is weak until linked, if not linked, the symbol becomes null instead of
607 being an undefined reference.</dd>
609 <dt><tt><b><a name="linkage_linkonce_odr">linkonce_odr</a></b></tt></dt>
610 <dt><tt><b><a name="linkage_weak_odr">weak_odr</a></b></tt></dt>
611 <dd>Some languages allow differing globals to be merged, such as two functions
612 with different semantics. Other languages, such as <tt>C++</tt>, ensure
613 that only equivalent globals are ever merged (the "one definition rule" -
614 "ODR"). Such languages can use the <tt>linkonce_odr</tt>
615 and <tt>weak_odr</tt> linkage types to indicate that the global will only
616 be merged with equivalent globals. These linkage types are otherwise the
617 same as their non-<tt>odr</tt> versions.</dd>
619 <dt><tt><b><a name="linkage_external">externally visible</a></b></tt>:</dt>
620 <dd>If none of the above identifiers are used, the global is externally
621 visible, meaning that it participates in linkage and can be used to
622 resolve external symbol references.</dd>
625 <p>The next two types of linkage are targeted for Microsoft Windows platform
626 only. They are designed to support importing (exporting) symbols from (to)
627 DLLs (Dynamic Link Libraries).</p>
630 <dt><tt><b><a name="linkage_dllimport">dllimport</a></b></tt></dt>
631 <dd>"<tt>dllimport</tt>" linkage causes the compiler to reference a function
632 or variable via a global pointer to a pointer that is set up by the DLL
633 exporting the symbol. On Microsoft Windows targets, the pointer name is
634 formed by combining <code>__imp_</code> and the function or variable
637 <dt><tt><b><a name="linkage_dllexport">dllexport</a></b></tt></dt>
638 <dd>"<tt>dllexport</tt>" linkage causes the compiler to provide a global
639 pointer to a pointer in a DLL, so that it can be referenced with the
640 <tt>dllimport</tt> attribute. On Microsoft Windows targets, the pointer
641 name is formed by combining <code>__imp_</code> and the function or
645 <p>For example, since the "<tt>.LC0</tt>" variable is defined to be internal, if
646 another module defined a "<tt>.LC0</tt>" variable and was linked with this
647 one, one of the two would be renamed, preventing a collision. Since
648 "<tt>main</tt>" and "<tt>puts</tt>" are external (i.e., lacking any linkage
649 declarations), they are accessible outside of the current module.</p>
651 <p>It is illegal for a function <i>declaration</i> to have any linkage type
652 other than "externally visible", <tt>dllimport</tt>
653 or <tt>extern_weak</tt>.</p>
655 <p>Aliases can have only <tt>external</tt>, <tt>internal</tt>, <tt>weak</tt>
656 or <tt>weak_odr</tt> linkages.</p>
660 <!-- ======================================================================= -->
661 <div class="doc_subsection">
662 <a name="callingconv">Calling Conventions</a>
665 <div class="doc_text">
667 <p>LLVM <a href="#functionstructure">functions</a>, <a href="#i_call">calls</a>
668 and <a href="#i_invoke">invokes</a> can all have an optional calling
669 convention specified for the call. The calling convention of any pair of
670 dynamic caller/callee must match, or the behavior of the program is
671 undefined. The following calling conventions are supported by LLVM, and more
672 may be added in the future:</p>
675 <dt><b>"<tt>ccc</tt>" - The C calling convention</b>:</dt>
676 <dd>This calling convention (the default if no other calling convention is
677 specified) matches the target C calling conventions. This calling
678 convention supports varargs function calls and tolerates some mismatch in
679 the declared prototype and implemented declaration of the function (as
682 <dt><b>"<tt>fastcc</tt>" - The fast calling convention</b>:</dt>
683 <dd>This calling convention attempts to make calls as fast as possible
684 (e.g. by passing things in registers). This calling convention allows the
685 target to use whatever tricks it wants to produce fast code for the
686 target, without having to conform to an externally specified ABI
687 (Application Binary Interface).
688 <a href="CodeGenerator.html#tailcallopt">Tail calls can only be optimized
689 when this convention is used.</a> This calling convention does not
690 support varargs and requires the prototype of all callees to exactly match
691 the prototype of the function definition.</dd>
693 <dt><b>"<tt>coldcc</tt>" - The cold calling convention</b>:</dt>
694 <dd>This calling convention attempts to make code in the caller as efficient
695 as possible under the assumption that the call is not commonly executed.
696 As such, these calls often preserve all registers so that the call does
697 not break any live ranges in the caller side. This calling convention
698 does not support varargs and requires the prototype of all callees to
699 exactly match the prototype of the function definition.</dd>
701 <dt><b>"<tt>cc <<em>n</em>></tt>" - Numbered convention</b>:</dt>
702 <dd>Any calling convention may be specified by number, allowing
703 target-specific calling conventions to be used. Target specific calling
704 conventions start at 64.</dd>
707 <p>More calling conventions can be added/defined on an as-needed basis, to
708 support Pascal conventions or any other well-known target-independent
713 <!-- ======================================================================= -->
714 <div class="doc_subsection">
715 <a name="visibility">Visibility Styles</a>
718 <div class="doc_text">
720 <p>All Global Variables and Functions have one of the following visibility
724 <dt><b>"<tt>default</tt>" - Default style</b>:</dt>
725 <dd>On targets that use the ELF object file format, default visibility means
726 that the declaration is visible to other modules and, in shared libraries,
727 means that the declared entity may be overridden. On Darwin, default
728 visibility means that the declaration is visible to other modules. Default
729 visibility corresponds to "external linkage" in the language.</dd>
731 <dt><b>"<tt>hidden</tt>" - Hidden style</b>:</dt>
732 <dd>Two declarations of an object with hidden visibility refer to the same
733 object if they are in the same shared object. Usually, hidden visibility
734 indicates that the symbol will not be placed into the dynamic symbol
735 table, so no other module (executable or shared library) can reference it
738 <dt><b>"<tt>protected</tt>" - Protected style</b>:</dt>
739 <dd>On ELF, protected visibility indicates that the symbol will be placed in
740 the dynamic symbol table, but that references within the defining module
741 will bind to the local symbol. That is, the symbol cannot be overridden by
747 <!-- ======================================================================= -->
748 <div class="doc_subsection">
749 <a name="namedtypes">Named Types</a>
752 <div class="doc_text">
754 <p>LLVM IR allows you to specify name aliases for certain types. This can make
755 it easier to read the IR and make the IR more condensed (particularly when
756 recursive types are involved). An example of a name specification is:</p>
758 <div class="doc_code">
760 %mytype = type { %mytype*, i32 }
764 <p>You may give a name to any <a href="#typesystem">type</a> except
765 "<a href="t_void">void</a>". Type name aliases may be used anywhere a type
766 is expected with the syntax "%mytype".</p>
768 <p>Note that type names are aliases for the structural type that they indicate,
769 and that you can therefore specify multiple names for the same type. This
770 often leads to confusing behavior when dumping out a .ll file. Since LLVM IR
771 uses structural typing, the name is not part of the type. When printing out
772 LLVM IR, the printer will pick <em>one name</em> to render all types of a
773 particular shape. This means that if you have code where two different
774 source types end up having the same LLVM type, that the dumper will sometimes
775 print the "wrong" or unexpected type. This is an important design point and
776 isn't going to change.</p>
780 <!-- ======================================================================= -->
781 <div class="doc_subsection">
782 <a name="globalvars">Global Variables</a>
785 <div class="doc_text">
787 <p>Global variables define regions of memory allocated at compilation time
788 instead of run-time. Global variables may optionally be initialized, may
789 have an explicit section to be placed in, and may have an optional explicit
790 alignment specified. A variable may be defined as "thread_local", which
791 means that it will not be shared by threads (each thread will have a
792 separated copy of the variable). A variable may be defined as a global
793 "constant," which indicates that the contents of the variable
794 will <b>never</b> be modified (enabling better optimization, allowing the
795 global data to be placed in the read-only section of an executable, etc).
796 Note that variables that need runtime initialization cannot be marked
797 "constant" as there is a store to the variable.</p>
799 <p>LLVM explicitly allows <em>declarations</em> of global variables to be marked
800 constant, even if the final definition of the global is not. This capability
801 can be used to enable slightly better optimization of the program, but
802 requires the language definition to guarantee that optimizations based on the
803 'constantness' are valid for the translation units that do not include the
806 <p>As SSA values, global variables define pointer values that are in scope
807 (i.e. they dominate) all basic blocks in the program. Global variables
808 always define a pointer to their "content" type because they describe a
809 region of memory, and all memory objects in LLVM are accessed through
812 <p>A global variable may be declared to reside in a target-specific numbered
813 address space. For targets that support them, address spaces may affect how
814 optimizations are performed and/or what target instructions are used to
815 access the variable. The default address space is zero. The address space
816 qualifier must precede any other attributes.</p>
818 <p>LLVM allows an explicit section to be specified for globals. If the target
819 supports it, it will emit globals to the section specified.</p>
821 <p>An explicit alignment may be specified for a global. If not present, or if
822 the alignment is set to zero, the alignment of the global is set by the
823 target to whatever it feels convenient. If an explicit alignment is
824 specified, the global is forced to have at least that much alignment. All
825 alignments must be a power of 2.</p>
827 <p>For example, the following defines a global in a numbered address space with
828 an initializer, section, and alignment:</p>
830 <div class="doc_code">
832 @G = addrspace(5) constant float 1.0, section "foo", align 4
839 <!-- ======================================================================= -->
840 <div class="doc_subsection">
841 <a name="functionstructure">Functions</a>
844 <div class="doc_text">
846 <p>LLVM function definitions consist of the "<tt>define</tt>" keyord, an
847 optional <a href="#linkage">linkage type</a>, an optional
848 <a href="#visibility">visibility style</a>, an optional
849 <a href="#callingconv">calling convention</a>, a return type, an optional
850 <a href="#paramattrs">parameter attribute</a> for the return type, a function
851 name, a (possibly empty) argument list (each with optional
852 <a href="#paramattrs">parameter attributes</a>), optional
853 <a href="#fnattrs">function attributes</a>, an optional section, an optional
854 alignment, an optional <a href="#gc">garbage collector name</a>, an opening
855 curly brace, a list of basic blocks, and a closing curly brace.</p>
857 <p>LLVM function declarations consist of the "<tt>declare</tt>" keyword, an
858 optional <a href="#linkage">linkage type</a>, an optional
859 <a href="#visibility">visibility style</a>, an optional
860 <a href="#callingconv">calling convention</a>, a return type, an optional
861 <a href="#paramattrs">parameter attribute</a> for the return type, a function
862 name, a possibly empty list of arguments, an optional alignment, and an
863 optional <a href="#gc">garbage collector name</a>.</p>
865 <p>A function definition contains a list of basic blocks, forming the CFG
866 (Control Flow Graph) for the function. Each basic block may optionally start
867 with a label (giving the basic block a symbol table entry), contains a list
868 of instructions, and ends with a <a href="#terminators">terminator</a>
869 instruction (such as a branch or function return).</p>
871 <p>The first basic block in a function is special in two ways: it is immediately
872 executed on entrance to the function, and it is not allowed to have
873 predecessor basic blocks (i.e. there can not be any branches to the entry
874 block of a function). Because the block can have no predecessors, it also
875 cannot have any <a href="#i_phi">PHI nodes</a>.</p>
877 <p>LLVM allows an explicit section to be specified for functions. If the target
878 supports it, it will emit functions to the section specified.</p>
880 <p>An explicit alignment may be specified for a function. If not present, or if
881 the alignment is set to zero, the alignment of the function is set by the
882 target to whatever it feels convenient. If an explicit alignment is
883 specified, the function is forced to have at least that much alignment. All
884 alignments must be a power of 2.</p>
887 <div class="doc_code">
889 define [<a href="#linkage">linkage</a>] [<a href="#visibility">visibility</a>]
890 [<a href="#callingconv">cconv</a>] [<a href="#paramattrs">ret attrs</a>]
891 <ResultType> @<FunctionName> ([argument list])
892 [<a href="#fnattrs">fn Attrs</a>] [section "name"] [align N]
893 [<a href="#gc">gc</a>] { ... }
899 <!-- ======================================================================= -->
900 <div class="doc_subsection">
901 <a name="aliasstructure">Aliases</a>
904 <div class="doc_text">
906 <p>Aliases act as "second name" for the aliasee value (which can be either
907 function, global variable, another alias or bitcast of global value). Aliases
908 may have an optional <a href="#linkage">linkage type</a>, and an
909 optional <a href="#visibility">visibility style</a>.</p>
912 <div class="doc_code">
914 @<Name> = alias [Linkage] [Visibility] <AliaseeTy> @<Aliasee>
920 <!-- ======================================================================= -->
921 <div class="doc_subsection">
922 <a name="namedmetadatastructure">Named Metadata</a>
925 <div class="doc_text">
927 <p>Named metadata is a collection of metadata. <a href="#metadata">Metadata
928 nodes</a> (but not metadata strings) and null are the only valid operands for
929 a named metadata.</p>
932 <div class="doc_code">
934 !1 = metadata !{metadata !"one"}
941 <!-- ======================================================================= -->
942 <div class="doc_subsection"><a name="paramattrs">Parameter Attributes</a></div>
944 <div class="doc_text">
946 <p>The return type and each parameter of a function type may have a set of
947 <i>parameter attributes</i> associated with them. Parameter attributes are
948 used to communicate additional information about the result or parameters of
949 a function. Parameter attributes are considered to be part of the function,
950 not of the function type, so functions with different parameter attributes
951 can have the same function type.</p>
953 <p>Parameter attributes are simple keywords that follow the type specified. If
954 multiple parameter attributes are needed, they are space separated. For
957 <div class="doc_code">
959 declare i32 @printf(i8* noalias nocapture, ...)
960 declare i32 @atoi(i8 zeroext)
961 declare signext i8 @returns_signed_char()
965 <p>Note that any attributes for the function result (<tt>nounwind</tt>,
966 <tt>readonly</tt>) come immediately after the argument list.</p>
968 <p>Currently, only the following parameter attributes are defined:</p>
971 <dt><tt><b>zeroext</b></tt></dt>
972 <dd>This indicates to the code generator that the parameter or return value
973 should be zero-extended to a 32-bit value by the caller (for a parameter)
974 or the callee (for a return value).</dd>
976 <dt><tt><b>signext</b></tt></dt>
977 <dd>This indicates to the code generator that the parameter or return value
978 should be sign-extended to a 32-bit value by the caller (for a parameter)
979 or the callee (for a return value).</dd>
981 <dt><tt><b>inreg</b></tt></dt>
982 <dd>This indicates that this parameter or return value should be treated in a
983 special target-dependent fashion during while emitting code for a function
984 call or return (usually, by putting it in a register as opposed to memory,
985 though some targets use it to distinguish between two different kinds of
986 registers). Use of this attribute is target-specific.</dd>
988 <dt><tt><b><a name="byval">byval</a></b></tt></dt>
989 <dd>This indicates that the pointer parameter should really be passed by value
990 to the function. The attribute implies that a hidden copy of the pointee
991 is made between the caller and the callee, so the callee is unable to
992 modify the value in the callee. This attribute is only valid on LLVM
993 pointer arguments. It is generally used to pass structs and arrays by
994 value, but is also valid on pointers to scalars. The copy is considered
995 to belong to the caller not the callee (for example,
996 <tt><a href="#readonly">readonly</a></tt> functions should not write to
997 <tt>byval</tt> parameters). This is not a valid attribute for return
998 values. The byval attribute also supports specifying an alignment with
999 the align attribute. This has a target-specific effect on the code
1000 generator that usually indicates a desired alignment for the synthesized
1003 <dt><tt><b>sret</b></tt></dt>
1004 <dd>This indicates that the pointer parameter specifies the address of a
1005 structure that is the return value of the function in the source program.
1006 This pointer must be guaranteed by the caller to be valid: loads and
1007 stores to the structure may be assumed by the callee to not to trap. This
1008 may only be applied to the first parameter. This is not a valid attribute
1009 for return values. </dd>
1011 <dt><tt><b>noalias</b></tt></dt>
1012 <dd>This indicates that the pointer does not alias any global or any other
1013 parameter. The caller is responsible for ensuring that this is the
1014 case. On a function return value, <tt>noalias</tt> additionally indicates
1015 that the pointer does not alias any other pointers visible to the
1016 caller. For further details, please see the discussion of the NoAlias
1018 <a href="http://llvm.org/docs/AliasAnalysis.html#MustMayNo">alias
1021 <dt><tt><b>nocapture</b></tt></dt>
1022 <dd>This indicates that the callee does not make any copies of the pointer
1023 that outlive the callee itself. This is not a valid attribute for return
1026 <dt><tt><b>nest</b></tt></dt>
1027 <dd>This indicates that the pointer parameter can be excised using the
1028 <a href="#int_trampoline">trampoline intrinsics</a>. This is not a valid
1029 attribute for return values.</dd>
1034 <!-- ======================================================================= -->
1035 <div class="doc_subsection">
1036 <a name="gc">Garbage Collector Names</a>
1039 <div class="doc_text">
1041 <p>Each function may specify a garbage collector name, which is simply a
1044 <div class="doc_code">
1046 define void @f() gc "name" { ... }
1050 <p>The compiler declares the supported values of <i>name</i>. Specifying a
1051 collector which will cause the compiler to alter its output in order to
1052 support the named garbage collection algorithm.</p>
1056 <!-- ======================================================================= -->
1057 <div class="doc_subsection">
1058 <a name="fnattrs">Function Attributes</a>
1061 <div class="doc_text">
1063 <p>Function attributes are set to communicate additional information about a
1064 function. Function attributes are considered to be part of the function, not
1065 of the function type, so functions with different parameter attributes can
1066 have the same function type.</p>
1068 <p>Function attributes are simple keywords that follow the type specified. If
1069 multiple attributes are needed, they are space separated. For example:</p>
1071 <div class="doc_code">
1073 define void @f() noinline { ... }
1074 define void @f() alwaysinline { ... }
1075 define void @f() alwaysinline optsize { ... }
1076 define void @f() optsize { ... }
1081 <dt><tt><b>alwaysinline</b></tt></dt>
1082 <dd>This attribute indicates that the inliner should attempt to inline this
1083 function into callers whenever possible, ignoring any active inlining size
1084 threshold for this caller.</dd>
1086 <dt><tt><b>noinline</b></tt></dt>
1087 <dd>This attribute indicates that the inliner should never inline this
1088 function in any situation. This attribute may not be used together with
1089 the <tt>alwaysinline</tt> attribute.</dd>
1091 <dt><tt><b>optsize</b></tt></dt>
1092 <dd>This attribute suggests that optimization passes and code generator passes
1093 make choices that keep the code size of this function low, and otherwise
1094 do optimizations specifically to reduce code size.</dd>
1096 <dt><tt><b>noreturn</b></tt></dt>
1097 <dd>This function attribute indicates that the function never returns
1098 normally. This produces undefined behavior at runtime if the function
1099 ever does dynamically return.</dd>
1101 <dt><tt><b>nounwind</b></tt></dt>
1102 <dd>This function attribute indicates that the function never returns with an
1103 unwind or exceptional control flow. If the function does unwind, its
1104 runtime behavior is undefined.</dd>
1106 <dt><tt><b>readnone</b></tt></dt>
1107 <dd>This attribute indicates that the function computes its result (or decides
1108 to unwind an exception) based strictly on its arguments, without
1109 dereferencing any pointer arguments or otherwise accessing any mutable
1110 state (e.g. memory, control registers, etc) visible to caller functions.
1111 It does not write through any pointer arguments
1112 (including <tt><a href="#byval">byval</a></tt> arguments) and never
1113 changes any state visible to callers. This means that it cannot unwind
1114 exceptions by calling the <tt>C++</tt> exception throwing methods, but
1115 could use the <tt>unwind</tt> instruction.</dd>
1117 <dt><tt><b><a name="readonly">readonly</a></b></tt></dt>
1118 <dd>This attribute indicates that the function does not write through any
1119 pointer arguments (including <tt><a href="#byval">byval</a></tt>
1120 arguments) or otherwise modify any state (e.g. memory, control registers,
1121 etc) visible to caller functions. It may dereference pointer arguments
1122 and read state that may be set in the caller. A readonly function always
1123 returns the same value (or unwinds an exception identically) when called
1124 with the same set of arguments and global state. It cannot unwind an
1125 exception by calling the <tt>C++</tt> exception throwing methods, but may
1126 use the <tt>unwind</tt> instruction.</dd>
1128 <dt><tt><b><a name="ssp">ssp</a></b></tt></dt>
1129 <dd>This attribute indicates that the function should emit a stack smashing
1130 protector. It is in the form of a "canary"—a random value placed on
1131 the stack before the local variables that's checked upon return from the
1132 function to see if it has been overwritten. A heuristic is used to
1133 determine if a function needs stack protectors or not.<br>
1135 If a function that has an <tt>ssp</tt> attribute is inlined into a
1136 function that doesn't have an <tt>ssp</tt> attribute, then the resulting
1137 function will have an <tt>ssp</tt> attribute.</dd>
1139 <dt><tt><b>sspreq</b></tt></dt>
1140 <dd>This attribute indicates that the function should <em>always</em> emit a
1141 stack smashing protector. This overrides
1142 the <tt><a href="#ssp">ssp</a></tt> function attribute.<br>
1144 If a function that has an <tt>sspreq</tt> attribute is inlined into a
1145 function that doesn't have an <tt>sspreq</tt> attribute or which has
1146 an <tt>ssp</tt> attribute, then the resulting function will have
1147 an <tt>sspreq</tt> attribute.</dd>
1149 <dt><tt><b>noredzone</b></tt></dt>
1150 <dd>This attribute indicates that the code generator should not use a red
1151 zone, even if the target-specific ABI normally permits it.</dd>
1153 <dt><tt><b>noimplicitfloat</b></tt></dt>
1154 <dd>This attributes disables implicit floating point instructions.</dd>
1156 <dt><tt><b>naked</b></tt></dt>
1157 <dd>This attribute disables prologue / epilogue emission for the function.
1158 This can have very system-specific consequences.</dd>
1163 <!-- ======================================================================= -->
1164 <div class="doc_subsection">
1165 <a name="moduleasm">Module-Level Inline Assembly</a>
1168 <div class="doc_text">
1170 <p>Modules may contain "module-level inline asm" blocks, which corresponds to
1171 the GCC "file scope inline asm" blocks. These blocks are internally
1172 concatenated by LLVM and treated as a single unit, but may be separated in
1173 the <tt>.ll</tt> file if desired. The syntax is very simple:</p>
1175 <div class="doc_code">
1177 module asm "inline asm code goes here"
1178 module asm "more can go here"
1182 <p>The strings can contain any character by escaping non-printable characters.
1183 The escape sequence used is simply "\xx" where "xx" is the two digit hex code
1186 <p>The inline asm code is simply printed to the machine code .s file when
1187 assembly code is generated.</p>
1191 <!-- ======================================================================= -->
1192 <div class="doc_subsection">
1193 <a name="datalayout">Data Layout</a>
1196 <div class="doc_text">
1198 <p>A module may specify a target specific data layout string that specifies how
1199 data is to be laid out in memory. The syntax for the data layout is
1202 <div class="doc_code">
1204 target datalayout = "<i>layout specification</i>"
1208 <p>The <i>layout specification</i> consists of a list of specifications
1209 separated by the minus sign character ('-'). Each specification starts with
1210 a letter and may include other information after the letter to define some
1211 aspect of the data layout. The specifications accepted are as follows:</p>
1215 <dd>Specifies that the target lays out data in big-endian form. That is, the
1216 bits with the most significance have the lowest address location.</dd>
1219 <dd>Specifies that the target lays out data in little-endian form. That is,
1220 the bits with the least significance have the lowest address
1223 <dt><tt>p:<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1224 <dd>This specifies the <i>size</i> of a pointer and its <i>abi</i> and
1225 <i>preferred</i> alignments. All sizes are in bits. Specifying
1226 the <i>pref</i> alignment is optional. If omitted, the
1227 preceding <tt>:</tt> should be omitted too.</dd>
1229 <dt><tt>i<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1230 <dd>This specifies the alignment for an integer type of a given bit
1231 <i>size</i>. The value of <i>size</i> must be in the range [1,2^23).</dd>
1233 <dt><tt>v<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1234 <dd>This specifies the alignment for a vector type of a given bit
1237 <dt><tt>f<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1238 <dd>This specifies the alignment for a floating point type of a given bit
1239 <i>size</i>. The value of <i>size</i> must be either 32 (float) or 64
1242 <dt><tt>a<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1243 <dd>This specifies the alignment for an aggregate type of a given bit
1246 <dt><tt>s<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1247 <dd>This specifies the alignment for a stack object of a given bit
1250 <dt><tt>n<i>size1</i>:<i>size2</i>:<i>size3</i>...</tt></dt>
1251 <dd>This specifies a set of native integer widths for the target CPU
1252 in bits. For example, it might contain "n32" for 32-bit PowerPC,
1253 "n32:64" for PowerPC 64, or "n8:16:32:64" for X86-64. Elements of
1254 this set are considered to support most general arithmetic
1255 operations efficiently.</dd>
1258 <p>When constructing the data layout for a given target, LLVM starts with a
1259 default set of specifications which are then (possibly) overriden by the
1260 specifications in the <tt>datalayout</tt> keyword. The default specifications
1261 are given in this list:</p>
1264 <li><tt>E</tt> - big endian</li>
1265 <li><tt>p:32:64:64</tt> - 32-bit pointers with 64-bit alignment</li>
1266 <li><tt>i1:8:8</tt> - i1 is 8-bit (byte) aligned</li>
1267 <li><tt>i8:8:8</tt> - i8 is 8-bit (byte) aligned</li>
1268 <li><tt>i16:16:16</tt> - i16 is 16-bit aligned</li>
1269 <li><tt>i32:32:32</tt> - i32 is 32-bit aligned</li>
1270 <li><tt>i64:32:64</tt> - i64 has ABI alignment of 32-bits but preferred
1271 alignment of 64-bits</li>
1272 <li><tt>f32:32:32</tt> - float is 32-bit aligned</li>
1273 <li><tt>f64:64:64</tt> - double is 64-bit aligned</li>
1274 <li><tt>v64:64:64</tt> - 64-bit vector is 64-bit aligned</li>
1275 <li><tt>v128:128:128</tt> - 128-bit vector is 128-bit aligned</li>
1276 <li><tt>a0:0:1</tt> - aggregates are 8-bit aligned</li>
1277 <li><tt>s0:64:64</tt> - stack objects are 64-bit aligned</li>
1280 <p>When LLVM is determining the alignment for a given type, it uses the
1281 following rules:</p>
1284 <li>If the type sought is an exact match for one of the specifications, that
1285 specification is used.</li>
1287 <li>If no match is found, and the type sought is an integer type, then the
1288 smallest integer type that is larger than the bitwidth of the sought type
1289 is used. If none of the specifications are larger than the bitwidth then
1290 the the largest integer type is used. For example, given the default
1291 specifications above, the i7 type will use the alignment of i8 (next
1292 largest) while both i65 and i256 will use the alignment of i64 (largest
1295 <li>If no match is found, and the type sought is a vector type, then the
1296 largest vector type that is smaller than the sought vector type will be
1297 used as a fall back. This happens because <128 x double> can be
1298 implemented in terms of 64 <2 x double>, for example.</li>
1303 <!-- ======================================================================= -->
1304 <div class="doc_subsection">
1305 <a name="pointeraliasing">Pointer Aliasing Rules</a>
1308 <div class="doc_text">
1310 <p>Any memory access must be done through a pointer value associated
1311 with an address range of the memory access, otherwise the behavior
1312 is undefined. Pointer values are associated with address ranges
1313 according to the following rules:</p>
1316 <li>A pointer value formed from a
1317 <tt><a href="#i_getelementptr">getelementptr</a></tt> instruction
1318 is associated with the addresses associated with the first operand
1319 of the <tt>getelementptr</tt>.</li>
1320 <li>An address of a global variable is associated with the address
1321 range of the variable's storage.</li>
1322 <li>The result value of an allocation instruction is associated with
1323 the address range of the allocated storage.</li>
1324 <li>A null pointer in the default address-space is associated with
1326 <li>A pointer value formed by an
1327 <tt><a href="#i_inttoptr">inttoptr</a></tt> is associated with all
1328 address ranges of all pointer values that contribute (directly or
1329 indirectly) to the computation of the pointer's value.</li>
1330 <li>The result value of a
1331 <tt><a href="#i_bitcast">bitcast</a></tt> is associated with all
1332 addresses associated with the operand of the <tt>bitcast</tt>.</li>
1333 <li>An integer constant other than zero or a pointer value returned
1334 from a function not defined within LLVM may be associated with address
1335 ranges allocated through mechanisms other than those provided by
1336 LLVM. Such ranges shall not overlap with any ranges of addresses
1337 allocated by mechanisms provided by LLVM.</li>
1340 <p>LLVM IR does not associate types with memory. The result type of a
1341 <tt><a href="#i_load">load</a></tt> merely indicates the size and
1342 alignment of the memory from which to load, as well as the
1343 interpretation of the value. The first operand of a
1344 <tt><a href="#i_store">store</a></tt> similarly only indicates the size
1345 and alignment of the store.</p>
1347 <p>Consequently, type-based alias analysis, aka TBAA, aka
1348 <tt>-fstrict-aliasing</tt>, is not applicable to general unadorned
1349 LLVM IR. <a href="#metadata">Metadata</a> may be used to encode
1350 additional information which specialized optimization passes may use
1351 to implement type-based alias analysis.</p>
1355 <!-- *********************************************************************** -->
1356 <div class="doc_section"> <a name="typesystem">Type System</a> </div>
1357 <!-- *********************************************************************** -->
1359 <div class="doc_text">
1361 <p>The LLVM type system is one of the most important features of the
1362 intermediate representation. Being typed enables a number of optimizations
1363 to be performed on the intermediate representation directly, without having
1364 to do extra analyses on the side before the transformation. A strong type
1365 system makes it easier to read the generated code and enables novel analyses
1366 and transformations that are not feasible to perform on normal three address
1367 code representations.</p>
1371 <!-- ======================================================================= -->
1372 <div class="doc_subsection"> <a name="t_classifications">Type
1373 Classifications</a> </div>
1375 <div class="doc_text">
1377 <p>The types fall into a few useful classifications:</p>
1379 <table border="1" cellspacing="0" cellpadding="4">
1381 <tr><th>Classification</th><th>Types</th></tr>
1383 <td><a href="#t_integer">integer</a></td>
1384 <td><tt>i1, i2, i3, ... i8, ... i16, ... i32, ... i64, ... </tt></td>
1387 <td><a href="#t_floating">floating point</a></td>
1388 <td><tt>float, double, x86_fp80, fp128, ppc_fp128</tt></td>
1391 <td><a name="t_firstclass">first class</a></td>
1392 <td><a href="#t_integer">integer</a>,
1393 <a href="#t_floating">floating point</a>,
1394 <a href="#t_pointer">pointer</a>,
1395 <a href="#t_vector">vector</a>,
1396 <a href="#t_struct">structure</a>,
1397 <a href="#t_array">array</a>,
1398 <a href="#t_label">label</a>,
1399 <a href="#t_metadata">metadata</a>.
1403 <td><a href="#t_primitive">primitive</a></td>
1404 <td><a href="#t_label">label</a>,
1405 <a href="#t_void">void</a>,
1406 <a href="#t_floating">floating point</a>,
1407 <a href="#t_metadata">metadata</a>.</td>
1410 <td><a href="#t_derived">derived</a></td>
1411 <td><a href="#t_integer">integer</a>,
1412 <a href="#t_array">array</a>,
1413 <a href="#t_function">function</a>,
1414 <a href="#t_pointer">pointer</a>,
1415 <a href="#t_struct">structure</a>,
1416 <a href="#t_pstruct">packed structure</a>,
1417 <a href="#t_vector">vector</a>,
1418 <a href="#t_opaque">opaque</a>.
1424 <p>The <a href="#t_firstclass">first class</a> types are perhaps the most
1425 important. Values of these types are the only ones which can be produced by
1430 <!-- ======================================================================= -->
1431 <div class="doc_subsection"> <a name="t_primitive">Primitive Types</a> </div>
1433 <div class="doc_text">
1435 <p>The primitive types are the fundamental building blocks of the LLVM
1440 <!-- _______________________________________________________________________ -->
1441 <div class="doc_subsubsection"> <a name="t_integer">Integer Type</a> </div>
1443 <div class="doc_text">
1446 <p>The integer type is a very simple type that simply specifies an arbitrary
1447 bit width for the integer type desired. Any bit width from 1 bit to
1448 2<sup>23</sup>-1 (about 8 million) can be specified.</p>
1455 <p>The number of bits the integer will occupy is specified by the <tt>N</tt>
1459 <table class="layout">
1461 <td class="left"><tt>i1</tt></td>
1462 <td class="left">a single-bit integer.</td>
1465 <td class="left"><tt>i32</tt></td>
1466 <td class="left">a 32-bit integer.</td>
1469 <td class="left"><tt>i1942652</tt></td>
1470 <td class="left">a really big integer of over 1 million bits.</td>
1476 <!-- _______________________________________________________________________ -->
1477 <div class="doc_subsubsection"> <a name="t_floating">Floating Point Types</a> </div>
1479 <div class="doc_text">
1483 <tr><th>Type</th><th>Description</th></tr>
1484 <tr><td><tt>float</tt></td><td>32-bit floating point value</td></tr>
1485 <tr><td><tt>double</tt></td><td>64-bit floating point value</td></tr>
1486 <tr><td><tt>fp128</tt></td><td>128-bit floating point value (112-bit mantissa)</td></tr>
1487 <tr><td><tt>x86_fp80</tt></td><td>80-bit floating point value (X87)</td></tr>
1488 <tr><td><tt>ppc_fp128</tt></td><td>128-bit floating point value (two 64-bits)</td></tr>
1494 <!-- _______________________________________________________________________ -->
1495 <div class="doc_subsubsection"> <a name="t_void">Void Type</a> </div>
1497 <div class="doc_text">
1500 <p>The void type does not represent any value and has no size.</p>
1509 <!-- _______________________________________________________________________ -->
1510 <div class="doc_subsubsection"> <a name="t_label">Label Type</a> </div>
1512 <div class="doc_text">
1515 <p>The label type represents code labels.</p>
1524 <!-- _______________________________________________________________________ -->
1525 <div class="doc_subsubsection"> <a name="t_metadata">Metadata Type</a> </div>
1527 <div class="doc_text">
1530 <p>The metadata type represents embedded metadata. No derived types may be
1531 created from metadata except for <a href="#t_function">function</a>
1542 <!-- ======================================================================= -->
1543 <div class="doc_subsection"> <a name="t_derived">Derived Types</a> </div>
1545 <div class="doc_text">
1547 <p>The real power in LLVM comes from the derived types in the system. This is
1548 what allows a programmer to represent arrays, functions, pointers, and other
1549 useful types. Each of these types contain one or more element types which
1550 may be a primitive type, or another derived type. For example, it is
1551 possible to have a two dimensional array, using an array as the element type
1552 of another array.</p>
1556 <!-- _______________________________________________________________________ -->
1557 <div class="doc_subsubsection"> <a name="t_array">Array Type</a> </div>
1559 <div class="doc_text">
1562 <p>The array type is a very simple derived type that arranges elements
1563 sequentially in memory. The array type requires a size (number of elements)
1564 and an underlying data type.</p>
1568 [<# elements> x <elementtype>]
1571 <p>The number of elements is a constant integer value; <tt>elementtype</tt> may
1572 be any type with a size.</p>
1575 <table class="layout">
1577 <td class="left"><tt>[40 x i32]</tt></td>
1578 <td class="left">Array of 40 32-bit integer values.</td>
1581 <td class="left"><tt>[41 x i32]</tt></td>
1582 <td class="left">Array of 41 32-bit integer values.</td>
1585 <td class="left"><tt>[4 x i8]</tt></td>
1586 <td class="left">Array of 4 8-bit integer values.</td>
1589 <p>Here are some examples of multidimensional arrays:</p>
1590 <table class="layout">
1592 <td class="left"><tt>[3 x [4 x i32]]</tt></td>
1593 <td class="left">3x4 array of 32-bit integer values.</td>
1596 <td class="left"><tt>[12 x [10 x float]]</tt></td>
1597 <td class="left">12x10 array of single precision floating point values.</td>
1600 <td class="left"><tt>[2 x [3 x [4 x i16]]]</tt></td>
1601 <td class="left">2x3x4 array of 16-bit integer values.</td>
1605 <p>There is no restriction on indexing beyond the end of the array implied by
1606 a static type (though there are restrictions on indexing beyond the bounds
1607 of an allocated object in some cases). This means that single-dimension
1608 'variable sized array' addressing can be implemented in LLVM with a zero
1609 length array type. An implementation of 'pascal style arrays' in LLVM could
1610 use the type "<tt>{ i32, [0 x float]}</tt>", for example.</p>
1614 <!-- _______________________________________________________________________ -->
1615 <div class="doc_subsubsection"> <a name="t_function">Function Type</a> </div>
1617 <div class="doc_text">
1620 <p>The function type can be thought of as a function signature. It consists of
1621 a return type and a list of formal parameter types. The return type of a
1622 function type is a scalar type, a void type, or a struct type. If the return
1623 type is a struct type then all struct elements must be of first class types,
1624 and the struct must have at least one element.</p>
1628 <returntype> (<parameter list>)
1631 <p>...where '<tt><parameter list></tt>' is a comma-separated list of type
1632 specifiers. Optionally, the parameter list may include a type <tt>...</tt>,
1633 which indicates that the function takes a variable number of arguments.
1634 Variable argument functions can access their arguments with
1635 the <a href="#int_varargs">variable argument handling intrinsic</a>
1636 functions. '<tt><returntype></tt>' is a any type except
1637 <a href="#t_label">label</a>.</p>
1640 <table class="layout">
1642 <td class="left"><tt>i32 (i32)</tt></td>
1643 <td class="left">function taking an <tt>i32</tt>, returning an <tt>i32</tt>
1645 </tr><tr class="layout">
1646 <td class="left"><tt>float (i16 signext, i32 *) *
1648 <td class="left"><a href="#t_pointer">Pointer</a> to a function that takes
1649 an <tt>i16</tt> that should be sign extended and a
1650 <a href="#t_pointer">pointer</a> to <tt>i32</tt>, returning
1653 </tr><tr class="layout">
1654 <td class="left"><tt>i32 (i8*, ...)</tt></td>
1655 <td class="left">A vararg function that takes at least one
1656 <a href="#t_pointer">pointer</a> to <tt>i8 </tt> (char in C),
1657 which returns an integer. This is the signature for <tt>printf</tt> in
1660 </tr><tr class="layout">
1661 <td class="left"><tt>{i32, i32} (i32)</tt></td>
1662 <td class="left">A function taking an <tt>i32</tt>, returning a
1663 <a href="#t_struct">structure</a> containing two <tt>i32</tt> values
1670 <!-- _______________________________________________________________________ -->
1671 <div class="doc_subsubsection"> <a name="t_struct">Structure Type</a> </div>
1673 <div class="doc_text">
1676 <p>The structure type is used to represent a collection of data members together
1677 in memory. The packing of the field types is defined to match the ABI of the
1678 underlying processor. The elements of a structure may be any type that has a
1681 <p>Structures in memory are accessed using '<tt><a href="#i_load">load</a></tt>'
1682 and '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a field
1683 with the '<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.
1684 Structures in registers are accessed using the
1685 '<tt><a href="#i_extractvalue">extractvalue</a></tt>' and
1686 '<tt><a href="#i_insertvalue">insertvalue</a></tt>' instructions.</p>
1689 { <type list> }
1693 <table class="layout">
1695 <td class="left"><tt>{ i32, i32, i32 }</tt></td>
1696 <td class="left">A triple of three <tt>i32</tt> values</td>
1697 </tr><tr class="layout">
1698 <td class="left"><tt>{ float, i32 (i32) * }</tt></td>
1699 <td class="left">A pair, where the first element is a <tt>float</tt> and the
1700 second element is a <a href="#t_pointer">pointer</a> to a
1701 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
1702 an <tt>i32</tt>.</td>
1708 <!-- _______________________________________________________________________ -->
1709 <div class="doc_subsubsection"> <a name="t_pstruct">Packed Structure Type</a>
1712 <div class="doc_text">
1715 <p>The packed structure type is used to represent a collection of data members
1716 together in memory. There is no padding between fields. Further, the
1717 alignment of a packed structure is 1 byte. The elements of a packed
1718 structure may be any type that has a size.</p>
1720 <p>Structures are accessed using '<tt><a href="#i_load">load</a></tt> and
1721 '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a field with
1722 the '<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.</p>
1726 < { <type list> } >
1730 <table class="layout">
1732 <td class="left"><tt>< { i32, i32, i32 } ></tt></td>
1733 <td class="left">A triple of three <tt>i32</tt> values</td>
1734 </tr><tr class="layout">
1736 <tt>< { float, i32 (i32)* } ></tt></td>
1737 <td class="left">A pair, where the first element is a <tt>float</tt> and the
1738 second element is a <a href="#t_pointer">pointer</a> to a
1739 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
1740 an <tt>i32</tt>.</td>
1746 <!-- _______________________________________________________________________ -->
1747 <div class="doc_subsubsection"> <a name="t_pointer">Pointer Type</a> </div>
1749 <div class="doc_text">
1752 <p>As in many languages, the pointer type represents a pointer or reference to
1753 another object, which must live in memory. Pointer types may have an optional
1754 address space attribute defining the target-specific numbered address space
1755 where the pointed-to object resides. The default address space is zero.</p>
1757 <p>Note that LLVM does not permit pointers to void (<tt>void*</tt>) nor does it
1758 permit pointers to labels (<tt>label*</tt>). Use <tt>i8*</tt> instead.</p>
1766 <table class="layout">
1768 <td class="left"><tt>[4 x i32]*</tt></td>
1769 <td class="left">A <a href="#t_pointer">pointer</a> to <a
1770 href="#t_array">array</a> of four <tt>i32</tt> values.</td>
1773 <td class="left"><tt>i32 (i32 *) *</tt></td>
1774 <td class="left"> A <a href="#t_pointer">pointer</a> to a <a
1775 href="#t_function">function</a> that takes an <tt>i32*</tt>, returning an
1779 <td class="left"><tt>i32 addrspace(5)*</tt></td>
1780 <td class="left">A <a href="#t_pointer">pointer</a> to an <tt>i32</tt> value
1781 that resides in address space #5.</td>
1787 <!-- _______________________________________________________________________ -->
1788 <div class="doc_subsubsection"> <a name="t_vector">Vector Type</a> </div>
1790 <div class="doc_text">
1793 <p>A vector type is a simple derived type that represents a vector of elements.
1794 Vector types are used when multiple primitive data are operated in parallel
1795 using a single instruction (SIMD). A vector type requires a size (number of
1796 elements) and an underlying primitive data type. Vector types are considered
1797 <a href="#t_firstclass">first class</a>.</p>
1801 < <# elements> x <elementtype> >
1804 <p>The number of elements is a constant integer value; elementtype may be any
1805 integer or floating point type.</p>
1808 <table class="layout">
1810 <td class="left"><tt><4 x i32></tt></td>
1811 <td class="left">Vector of 4 32-bit integer values.</td>
1814 <td class="left"><tt><8 x float></tt></td>
1815 <td class="left">Vector of 8 32-bit floating-point values.</td>
1818 <td class="left"><tt><2 x i64></tt></td>
1819 <td class="left">Vector of 2 64-bit integer values.</td>
1825 <!-- _______________________________________________________________________ -->
1826 <div class="doc_subsubsection"> <a name="t_opaque">Opaque Type</a> </div>
1827 <div class="doc_text">
1830 <p>Opaque types are used to represent unknown types in the system. This
1831 corresponds (for example) to the C notion of a forward declared structure
1832 type. In LLVM, opaque types can eventually be resolved to any type (not just
1833 a structure type).</p>
1841 <table class="layout">
1843 <td class="left"><tt>opaque</tt></td>
1844 <td class="left">An opaque type.</td>
1850 <!-- ======================================================================= -->
1851 <div class="doc_subsection">
1852 <a name="t_uprefs">Type Up-references</a>
1855 <div class="doc_text">
1858 <p>An "up reference" allows you to refer to a lexically enclosing type without
1859 requiring it to have a name. For instance, a structure declaration may
1860 contain a pointer to any of the types it is lexically a member of. Example
1861 of up references (with their equivalent as named type declarations)
1865 { \2 * } %x = type { %x* }
1866 { \2 }* %y = type { %y }*
1870 <p>An up reference is needed by the asmprinter for printing out cyclic types
1871 when there is no declared name for a type in the cycle. Because the
1872 asmprinter does not want to print out an infinite type string, it needs a
1873 syntax to handle recursive types that have no names (all names are optional
1881 <p>The level is the count of the lexical type that is being referred to.</p>
1884 <table class="layout">
1886 <td class="left"><tt>\1*</tt></td>
1887 <td class="left">Self-referential pointer.</td>
1890 <td class="left"><tt>{ { \3*, i8 }, i32 }</tt></td>
1891 <td class="left">Recursive structure where the upref refers to the out-most
1898 <!-- *********************************************************************** -->
1899 <div class="doc_section"> <a name="constants">Constants</a> </div>
1900 <!-- *********************************************************************** -->
1902 <div class="doc_text">
1904 <p>LLVM has several different basic types of constants. This section describes
1905 them all and their syntax.</p>
1909 <!-- ======================================================================= -->
1910 <div class="doc_subsection"><a name="simpleconstants">Simple Constants</a></div>
1912 <div class="doc_text">
1915 <dt><b>Boolean constants</b></dt>
1916 <dd>The two strings '<tt>true</tt>' and '<tt>false</tt>' are both valid
1917 constants of the <tt><a href="#t_integer">i1</a></tt> type.</dd>
1919 <dt><b>Integer constants</b></dt>
1920 <dd>Standard integers (such as '4') are constants of
1921 the <a href="#t_integer">integer</a> type. Negative numbers may be used
1922 with integer types.</dd>
1924 <dt><b>Floating point constants</b></dt>
1925 <dd>Floating point constants use standard decimal notation (e.g. 123.421),
1926 exponential notation (e.g. 1.23421e+2), or a more precise hexadecimal
1927 notation (see below). The assembler requires the exact decimal value of a
1928 floating-point constant. For example, the assembler accepts 1.25 but
1929 rejects 1.3 because 1.3 is a repeating decimal in binary. Floating point
1930 constants must have a <a href="#t_floating">floating point</a> type. </dd>
1932 <dt><b>Null pointer constants</b></dt>
1933 <dd>The identifier '<tt>null</tt>' is recognized as a null pointer constant
1934 and must be of <a href="#t_pointer">pointer type</a>.</dd>
1937 <p>The one non-intuitive notation for constants is the hexadecimal form of
1938 floating point constants. For example, the form '<tt>double
1939 0x432ff973cafa8000</tt>' is equivalent to (but harder to read than)
1940 '<tt>double 4.5e+15</tt>'. The only time hexadecimal floating point
1941 constants are required (and the only time that they are generated by the
1942 disassembler) is when a floating point constant must be emitted but it cannot
1943 be represented as a decimal floating point number in a reasonable number of
1944 digits. For example, NaN's, infinities, and other special values are
1945 represented in their IEEE hexadecimal format so that assembly and disassembly
1946 do not cause any bits to change in the constants.</p>
1948 <p>When using the hexadecimal form, constants of types float and double are
1949 represented using the 16-digit form shown above (which matches the IEEE754
1950 representation for double); float values must, however, be exactly
1951 representable as IEE754 single precision. Hexadecimal format is always used
1952 for long double, and there are three forms of long double. The 80-bit format
1953 used by x86 is represented as <tt>0xK</tt> followed by 20 hexadecimal digits.
1954 The 128-bit format used by PowerPC (two adjacent doubles) is represented
1955 by <tt>0xM</tt> followed by 32 hexadecimal digits. The IEEE 128-bit format
1956 is represented by <tt>0xL</tt> followed by 32 hexadecimal digits; no
1957 currently supported target uses this format. Long doubles will only work if
1958 they match the long double format on your target. All hexadecimal formats
1959 are big-endian (sign bit at the left).</p>
1963 <!-- ======================================================================= -->
1964 <div class="doc_subsection">
1965 <a name="aggregateconstants"></a> <!-- old anchor -->
1966 <a name="complexconstants">Complex Constants</a>
1969 <div class="doc_text">
1971 <p>Complex constants are a (potentially recursive) combination of simple
1972 constants and smaller complex constants.</p>
1975 <dt><b>Structure constants</b></dt>
1976 <dd>Structure constants are represented with notation similar to structure
1977 type definitions (a comma separated list of elements, surrounded by braces
1978 (<tt>{}</tt>)). For example: "<tt>{ i32 4, float 17.0, i32* @G }</tt>",
1979 where "<tt>@G</tt>" is declared as "<tt>@G = external global i32</tt>".
1980 Structure constants must have <a href="#t_struct">structure type</a>, and
1981 the number and types of elements must match those specified by the
1984 <dt><b>Array constants</b></dt>
1985 <dd>Array constants are represented with notation similar to array type
1986 definitions (a comma separated list of elements, surrounded by square
1987 brackets (<tt>[]</tt>)). For example: "<tt>[ i32 42, i32 11, i32 74
1988 ]</tt>". Array constants must have <a href="#t_array">array type</a>, and
1989 the number and types of elements must match those specified by the
1992 <dt><b>Vector constants</b></dt>
1993 <dd>Vector constants are represented with notation similar to vector type
1994 definitions (a comma separated list of elements, surrounded by
1995 less-than/greater-than's (<tt><></tt>)). For example: "<tt>< i32
1996 42, i32 11, i32 74, i32 100 ></tt>". Vector constants must
1997 have <a href="#t_vector">vector type</a>, and the number and types of
1998 elements must match those specified by the type.</dd>
2000 <dt><b>Zero initialization</b></dt>
2001 <dd>The string '<tt>zeroinitializer</tt>' can be used to zero initialize a
2002 value to zero of <em>any</em> type, including scalar and aggregate types.
2003 This is often used to avoid having to print large zero initializers
2004 (e.g. for large arrays) and is always exactly equivalent to using explicit
2005 zero initializers.</dd>
2007 <dt><b>Metadata node</b></dt>
2008 <dd>A metadata node is a structure-like constant with
2009 <a href="#t_metadata">metadata type</a>. For example: "<tt>metadata !{
2010 i32 0, metadata !"test" }</tt>". Unlike other constants that are meant to
2011 be interpreted as part of the instruction stream, metadata is a place to
2012 attach additional information such as debug info.</dd>
2017 <!-- ======================================================================= -->
2018 <div class="doc_subsection">
2019 <a name="globalconstants">Global Variable and Function Addresses</a>
2022 <div class="doc_text">
2024 <p>The addresses of <a href="#globalvars">global variables</a>
2025 and <a href="#functionstructure">functions</a> are always implicitly valid
2026 (link-time) constants. These constants are explicitly referenced when
2027 the <a href="#identifiers">identifier for the global</a> is used and always
2028 have <a href="#t_pointer">pointer</a> type. For example, the following is a
2029 legal LLVM file:</p>
2031 <div class="doc_code">
2035 @Z = global [2 x i32*] [ i32* @X, i32* @Y ]
2041 <!-- ======================================================================= -->
2042 <div class="doc_subsection"><a name="undefvalues">Undefined Values</a></div>
2043 <div class="doc_text">
2045 <p>The string '<tt>undef</tt>' can be used anywhere a constant is expected, and
2046 indicates that the user of the value may receive an unspecified bit-pattern.
2047 Undefined values may be of any type (other than label or void) and be used
2048 anywhere a constant is permitted.</p>
2050 <p>Undefined values are useful because they indicate to the compiler that the
2051 program is well defined no matter what value is used. This gives the
2052 compiler more freedom to optimize. Here are some examples of (potentially
2053 surprising) transformations that are valid (in pseudo IR):</p>
2056 <div class="doc_code">
2068 <p>This is safe because all of the output bits are affected by the undef bits.
2069 Any output bit can have a zero or one depending on the input bits.</p>
2071 <div class="doc_code">
2084 <p>These logical operations have bits that are not always affected by the input.
2085 For example, if "%X" has a zero bit, then the output of the 'and' operation will
2086 always be a zero, no matter what the corresponding bit from the undef is. As
2087 such, it is unsafe to optimize or assume that the result of the and is undef.
2088 However, it is safe to assume that all bits of the undef could be 0, and
2089 optimize the and to 0. Likewise, it is safe to assume that all the bits of
2090 the undef operand to the or could be set, allowing the or to be folded to
2093 <div class="doc_code">
2095 %A = select undef, %X, %Y
2096 %B = select undef, 42, %Y
2097 %C = select %X, %Y, undef
2109 <p>This set of examples show that undefined select (and conditional branch)
2110 conditions can go "either way" but they have to come from one of the two
2111 operands. In the %A example, if %X and %Y were both known to have a clear low
2112 bit, then %A would have to have a cleared low bit. However, in the %C example,
2113 the optimizer is allowed to assume that the undef operand could be the same as
2114 %Y, allowing the whole select to be eliminated.</p>
2117 <div class="doc_code">
2119 %A = xor undef, undef
2138 <p>This example points out that two undef operands are not necessarily the same.
2139 This can be surprising to people (and also matches C semantics) where they
2140 assume that "X^X" is always zero, even if X is undef. This isn't true for a
2141 number of reasons, but the short answer is that an undef "variable" can
2142 arbitrarily change its value over its "live range". This is true because the
2143 "variable" doesn't actually <em>have a live range</em>. Instead, the value is
2144 logically read from arbitrary registers that happen to be around when needed,
2145 so the value is not necessarily consistent over time. In fact, %A and %C need
2146 to have the same semantics or the core LLVM "replace all uses with" concept
2149 <div class="doc_code">
2159 <p>These examples show the crucial difference between an <em>undefined
2160 value</em> and <em>undefined behavior</em>. An undefined value (like undef) is
2161 allowed to have an arbitrary bit-pattern. This means that the %A operation
2162 can be constant folded to undef because the undef could be an SNaN, and fdiv is
2163 not (currently) defined on SNaN's. However, in the second example, we can make
2164 a more aggressive assumption: because the undef is allowed to be an arbitrary
2165 value, we are allowed to assume that it could be zero. Since a divide by zero
2166 has <em>undefined behavior</em>, we are allowed to assume that the operation
2167 does not execute at all. This allows us to delete the divide and all code after
2168 it: since the undefined operation "can't happen", the optimizer can assume that
2169 it occurs in dead code.
2172 <div class="doc_code">
2174 a: store undef -> %X
2175 b: store %X -> undef
2182 <p>These examples reiterate the fdiv example: a store "of" an undefined value
2183 can be assumed to not have any effect: we can assume that the value is
2184 overwritten with bits that happen to match what was already there. However, a
2185 store "to" an undefined location could clobber arbitrary memory, therefore, it
2186 has undefined behavior.</p>
2190 <!-- ======================================================================= -->
2191 <div class="doc_subsection"><a name="blockaddress">Addresses of Basic
2193 <div class="doc_text">
2195 <p><b><tt>blockaddress(@function, %block)</tt></b></p>
2197 <p>The '<tt>blockaddress</tt>' constant computes the address of the specified
2198 basic block in the specified function, and always has an i8* type. Taking
2199 the address of the entry block is illegal.</p>
2201 <p>This value only has defined behavior when used as an operand to the
2202 '<a href="#i_indirectbr"><tt>indirectbr</tt></a>' instruction or for comparisons
2203 against null. Pointer equality tests between labels addresses is undefined
2204 behavior - though, again, comparison against null is ok, and no label is
2205 equal to the null pointer. This may also be passed around as an opaque
2206 pointer sized value as long as the bits are not inspected. This allows
2207 <tt>ptrtoint</tt> and arithmetic to be performed on these values so long as
2208 the original value is reconstituted before the <tt>indirectbr</tt>.</p>
2210 <p>Finally, some targets may provide defined semantics when
2211 using the value as the operand to an inline assembly, but that is target
2218 <!-- ======================================================================= -->
2219 <div class="doc_subsection"><a name="constantexprs">Constant Expressions</a>
2222 <div class="doc_text">
2224 <p>Constant expressions are used to allow expressions involving other constants
2225 to be used as constants. Constant expressions may be of
2226 any <a href="#t_firstclass">first class</a> type and may involve any LLVM
2227 operation that does not have side effects (e.g. load and call are not
2228 supported). The following is the syntax for constant expressions:</p>
2231 <dt><b><tt>trunc ( CST to TYPE )</tt></b></dt>
2232 <dd>Truncate a constant to another type. The bit size of CST must be larger
2233 than the bit size of TYPE. Both types must be integers.</dd>
2235 <dt><b><tt>zext ( CST to TYPE )</tt></b></dt>
2236 <dd>Zero extend a constant to another type. The bit size of CST must be
2237 smaller or equal to the bit size of TYPE. Both types must be
2240 <dt><b><tt>sext ( CST to TYPE )</tt></b></dt>
2241 <dd>Sign extend a constant to another type. The bit size of CST must be
2242 smaller or equal to the bit size of TYPE. Both types must be
2245 <dt><b><tt>fptrunc ( CST to TYPE )</tt></b></dt>
2246 <dd>Truncate a floating point constant to another floating point type. The
2247 size of CST must be larger than the size of TYPE. Both types must be
2248 floating point.</dd>
2250 <dt><b><tt>fpext ( CST to TYPE )</tt></b></dt>
2251 <dd>Floating point extend a constant to another type. The size of CST must be
2252 smaller or equal to the size of TYPE. Both types must be floating
2255 <dt><b><tt>fptoui ( CST to TYPE )</tt></b></dt>
2256 <dd>Convert a floating point constant to the corresponding unsigned integer
2257 constant. TYPE must be a scalar or vector integer type. CST must be of
2258 scalar or vector floating point type. Both CST and TYPE must be scalars,
2259 or vectors of the same number of elements. If the value won't fit in the
2260 integer type, the results are undefined.</dd>
2262 <dt><b><tt>fptosi ( CST to TYPE )</tt></b></dt>
2263 <dd>Convert a floating point constant to the corresponding signed integer
2264 constant. TYPE must be a scalar or vector integer type. CST must be of
2265 scalar or vector floating point type. Both CST and TYPE must be scalars,
2266 or vectors of the same number of elements. If the value won't fit in the
2267 integer type, the results are undefined.</dd>
2269 <dt><b><tt>uitofp ( CST to TYPE )</tt></b></dt>
2270 <dd>Convert an unsigned integer constant to the corresponding floating point
2271 constant. TYPE must be a scalar or vector floating point type. CST must be
2272 of scalar or vector integer type. Both CST and TYPE must be scalars, or
2273 vectors of the same number of elements. If the value won't fit in the
2274 floating point type, the results are undefined.</dd>
2276 <dt><b><tt>sitofp ( CST to TYPE )</tt></b></dt>
2277 <dd>Convert a signed integer constant to the corresponding floating point
2278 constant. TYPE must be a scalar or vector floating point type. CST must be
2279 of scalar or vector integer type. Both CST and TYPE must be scalars, or
2280 vectors of the same number of elements. If the value won't fit in the
2281 floating point type, the results are undefined.</dd>
2283 <dt><b><tt>ptrtoint ( CST to TYPE )</tt></b></dt>
2284 <dd>Convert a pointer typed constant to the corresponding integer constant
2285 <tt>TYPE</tt> must be an integer type. <tt>CST</tt> must be of pointer
2286 type. The <tt>CST</tt> value is zero extended, truncated, or unchanged to
2287 make it fit in <tt>TYPE</tt>.</dd>
2289 <dt><b><tt>inttoptr ( CST to TYPE )</tt></b></dt>
2290 <dd>Convert a integer constant to a pointer constant. TYPE must be a pointer
2291 type. CST must be of integer type. The CST value is zero extended,
2292 truncated, or unchanged to make it fit in a pointer size. This one is
2293 <i>really</i> dangerous!</dd>
2295 <dt><b><tt>bitcast ( CST to TYPE )</tt></b></dt>
2296 <dd>Convert a constant, CST, to another TYPE. The constraints of the operands
2297 are the same as those for the <a href="#i_bitcast">bitcast
2298 instruction</a>.</dd>
2300 <dt><b><tt>getelementptr ( CSTPTR, IDX0, IDX1, ... )</tt></b></dt>
2301 <dt><b><tt>getelementptr inbounds ( CSTPTR, IDX0, IDX1, ... )</tt></b></dt>
2302 <dd>Perform the <a href="#i_getelementptr">getelementptr operation</a> on
2303 constants. As with the <a href="#i_getelementptr">getelementptr</a>
2304 instruction, the index list may have zero or more indexes, which are
2305 required to make sense for the type of "CSTPTR".</dd>
2307 <dt><b><tt>select ( COND, VAL1, VAL2 )</tt></b></dt>
2308 <dd>Perform the <a href="#i_select">select operation</a> on constants.</dd>
2310 <dt><b><tt>icmp COND ( VAL1, VAL2 )</tt></b></dt>
2311 <dd>Performs the <a href="#i_icmp">icmp operation</a> on constants.</dd>
2313 <dt><b><tt>fcmp COND ( VAL1, VAL2 )</tt></b></dt>
2314 <dd>Performs the <a href="#i_fcmp">fcmp operation</a> on constants.</dd>
2316 <dt><b><tt>extractelement ( VAL, IDX )</tt></b></dt>
2317 <dd>Perform the <a href="#i_extractelement">extractelement operation</a> on
2320 <dt><b><tt>insertelement ( VAL, ELT, IDX )</tt></b></dt>
2321 <dd>Perform the <a href="#i_insertelement">insertelement operation</a> on
2324 <dt><b><tt>shufflevector ( VEC1, VEC2, IDXMASK )</tt></b></dt>
2325 <dd>Perform the <a href="#i_shufflevector">shufflevector operation</a> on
2328 <dt><b><tt>OPCODE ( LHS, RHS )</tt></b></dt>
2329 <dd>Perform the specified operation of the LHS and RHS constants. OPCODE may
2330 be any of the <a href="#binaryops">binary</a>
2331 or <a href="#bitwiseops">bitwise binary</a> operations. The constraints
2332 on operands are the same as those for the corresponding instruction
2333 (e.g. no bitwise operations on floating point values are allowed).</dd>
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>
2407 <!-- ======================================================================= -->
2408 <div class="doc_subsection"><a name="metadata">Metadata Nodes and Metadata
2412 <div class="doc_text">
2414 <p>LLVM IR allows metadata to be attached to instructions in the program that
2415 can convey extra information about the code to the optimizers and code
2416 generator. One example application of metadata is source-level debug
2417 information. There are two metadata primitives: strings and nodes. All
2418 metadata has the <tt>metadata</tt> type and is identified in syntax by a
2419 preceding exclamation point ('<tt>!</tt>').</p>
2421 <p>A metadata string is a string surrounded by double quotes. It can contain
2422 any character by escaping non-printable characters with "\xx" where "xx" is
2423 the two digit hex code. For example: "<tt>!"test\00"</tt>".</p>
2425 <p>Metadata nodes are represented with notation similar to structure constants
2426 (a comma separated list of elements, surrounded by braces and preceded by an
2427 exclamation point). For example: "<tt>!{ metadata !"test\00", i32
2428 10}</tt>". Metadata nodes can have any values as their operand.</p>
2430 <p>A <a href="#namedmetadatastructure">named metadata</a> is a collection of
2431 metadata nodes, which can be looked up in the module symbol table. For
2432 example: "<tt>!foo = metadata !{!4, !3}</tt>".
2437 <!-- *********************************************************************** -->
2438 <div class="doc_section">
2439 <a name="intrinsic_globals">Intrinsic Global Variables</a>
2441 <!-- *********************************************************************** -->
2443 <p>LLVM has a number of "magic" global variables that contain data that affect
2444 code generation or other IR semantics. These are documented here. All globals
2445 of this sort should have a section specified as "<tt>llvm.metadata</tt>". This
2446 section and all globals that start with "<tt>llvm.</tt>" are reserved for use
2449 <!-- ======================================================================= -->
2450 <div class="doc_subsection">
2451 <a name="intg_used">The '<tt>llvm.used</tt>' Global Variable</a>
2454 <div class="doc_text">
2456 <p>The <tt>@llvm.used</tt> global is an array with i8* element type which has <a
2457 href="#linkage_appending">appending linkage</a>. This array contains a list of
2458 pointers to global variables and functions which may optionally have a pointer
2459 cast formed of bitcast or getelementptr. For example, a legal use of it is:</p>
2465 @llvm.used = appending global [2 x i8*] [
2467 i8* bitcast (i32* @Y to i8*)
2468 ], section "llvm.metadata"
2471 <p>If a global variable appears in the <tt>@llvm.used</tt> list, then the
2472 compiler, assembler, and linker are required to treat the symbol as if there is
2473 a reference to the global that it cannot see. For example, if a variable has
2474 internal linkage and no references other than that from the <tt>@llvm.used</tt>
2475 list, it cannot be deleted. This is commonly used to represent references from
2476 inline asms and other things the compiler cannot "see", and corresponds to
2477 "attribute((used))" in GNU C.</p>
2479 <p>On some targets, the code generator must emit a directive to the assembler or
2480 object file to prevent the assembler and linker from molesting the symbol.</p>
2484 <!-- ======================================================================= -->
2485 <div class="doc_subsection">
2486 <a name="intg_compiler_used">The '<tt>llvm.compiler.used</tt>' Global Variable</a>
2489 <div class="doc_text">
2491 <p>The <tt>@llvm.compiler.used</tt> directive is the same as the
2492 <tt>@llvm.used</tt> directive, except that it only prevents the compiler from
2493 touching the symbol. On targets that support it, this allows an intelligent
2494 linker to optimize references to the symbol without being impeded as it would be
2495 by <tt>@llvm.used</tt>.</p>
2497 <p>This is a rare construct that should only be used in rare circumstances, and
2498 should not be exposed to source languages.</p>
2502 <!-- ======================================================================= -->
2503 <div class="doc_subsection">
2504 <a name="intg_global_ctors">The '<tt>llvm.global_ctors</tt>' Global Variable</a>
2507 <div class="doc_text">
2509 <p>TODO: Describe this.</p>
2513 <!-- ======================================================================= -->
2514 <div class="doc_subsection">
2515 <a name="intg_global_dtors">The '<tt>llvm.global_dtors</tt>' Global Variable</a>
2518 <div class="doc_text">
2520 <p>TODO: Describe this.</p>
2525 <!-- *********************************************************************** -->
2526 <div class="doc_section"> <a name="instref">Instruction Reference</a> </div>
2527 <!-- *********************************************************************** -->
2529 <div class="doc_text">
2531 <p>The LLVM instruction set consists of several different classifications of
2532 instructions: <a href="#terminators">terminator
2533 instructions</a>, <a href="#binaryops">binary instructions</a>,
2534 <a href="#bitwiseops">bitwise binary instructions</a>,
2535 <a href="#memoryops">memory instructions</a>, and
2536 <a href="#otherops">other instructions</a>.</p>
2540 <!-- ======================================================================= -->
2541 <div class="doc_subsection"> <a name="terminators">Terminator
2542 Instructions</a> </div>
2544 <div class="doc_text">
2546 <p>As mentioned <a href="#functionstructure">previously</a>, every basic block
2547 in a program ends with a "Terminator" instruction, which indicates which
2548 block should be executed after the current block is finished. These
2549 terminator instructions typically yield a '<tt>void</tt>' value: they produce
2550 control flow, not values (the one exception being the
2551 '<a href="#i_invoke"><tt>invoke</tt></a>' instruction).</p>
2553 <p>There are six different terminator instructions: the
2554 '<a href="#i_ret"><tt>ret</tt></a>' instruction, the
2555 '<a href="#i_br"><tt>br</tt></a>' instruction, the
2556 '<a href="#i_switch"><tt>switch</tt></a>' instruction, the
2557 '<a href="#i_indirectbr">'<tt>indirectbr</tt></a>' Instruction, the
2558 '<a href="#i_invoke"><tt>invoke</tt></a>' instruction, the
2559 '<a href="#i_unwind"><tt>unwind</tt></a>' instruction, and the
2560 '<a href="#i_unreachable"><tt>unreachable</tt></a>' instruction.</p>
2564 <!-- _______________________________________________________________________ -->
2565 <div class="doc_subsubsection"> <a name="i_ret">'<tt>ret</tt>'
2566 Instruction</a> </div>
2568 <div class="doc_text">
2572 ret <type> <value> <i>; Return a value from a non-void function</i>
2573 ret void <i>; Return from void function</i>
2577 <p>The '<tt>ret</tt>' instruction is used to return control flow (and optionally
2578 a value) from a function back to the caller.</p>
2580 <p>There are two forms of the '<tt>ret</tt>' instruction: one that returns a
2581 value and then causes control flow, and one that just causes control flow to
2585 <p>The '<tt>ret</tt>' instruction optionally accepts a single argument, the
2586 return value. The type of the return value must be a
2587 '<a href="#t_firstclass">first class</a>' type.</p>
2589 <p>A function is not <a href="#wellformed">well formed</a> if it it has a
2590 non-void return type and contains a '<tt>ret</tt>' instruction with no return
2591 value or a return value with a type that does not match its type, or if it
2592 has a void return type and contains a '<tt>ret</tt>' instruction with a
2596 <p>When the '<tt>ret</tt>' instruction is executed, control flow returns back to
2597 the calling function's context. If the caller is a
2598 "<a href="#i_call"><tt>call</tt></a>" instruction, execution continues at the
2599 instruction after the call. If the caller was an
2600 "<a href="#i_invoke"><tt>invoke</tt></a>" instruction, execution continues at
2601 the beginning of the "normal" destination block. If the instruction returns
2602 a value, that value shall set the call or invoke instruction's return
2607 ret i32 5 <i>; Return an integer value of 5</i>
2608 ret void <i>; Return from a void function</i>
2609 ret { i32, i8 } { i32 4, i8 2 } <i>; Return a struct of values 4 and 2</i>
2613 <!-- _______________________________________________________________________ -->
2614 <div class="doc_subsubsection"> <a name="i_br">'<tt>br</tt>' Instruction</a> </div>
2616 <div class="doc_text">
2620 br i1 <cond>, label <iftrue>, label <iffalse><br> br label <dest> <i>; Unconditional branch</i>
2624 <p>The '<tt>br</tt>' instruction is used to cause control flow to transfer to a
2625 different basic block in the current function. There are two forms of this
2626 instruction, corresponding to a conditional branch and an unconditional
2630 <p>The conditional branch form of the '<tt>br</tt>' instruction takes a single
2631 '<tt>i1</tt>' value and two '<tt>label</tt>' values. The unconditional form
2632 of the '<tt>br</tt>' instruction takes a single '<tt>label</tt>' value as a
2636 <p>Upon execution of a conditional '<tt>br</tt>' instruction, the '<tt>i1</tt>'
2637 argument is evaluated. If the value is <tt>true</tt>, control flows to the
2638 '<tt>iftrue</tt>' <tt>label</tt> argument. If "cond" is <tt>false</tt>,
2639 control flows to the '<tt>iffalse</tt>' <tt>label</tt> argument.</p>
2644 %cond = <a href="#i_icmp">icmp</a> eq i32 %a, %b
2645 br i1 %cond, label %IfEqual, label %IfUnequal
2647 <a href="#i_ret">ret</a> i32 1
2649 <a href="#i_ret">ret</a> i32 0
2654 <!-- _______________________________________________________________________ -->
2655 <div class="doc_subsubsection">
2656 <a name="i_switch">'<tt>switch</tt>' Instruction</a>
2659 <div class="doc_text">
2663 switch <intty> <value>, label <defaultdest> [ <intty> <val>, label <dest> ... ]
2667 <p>The '<tt>switch</tt>' instruction is used to transfer control flow to one of
2668 several different places. It is a generalization of the '<tt>br</tt>'
2669 instruction, allowing a branch to occur to one of many possible
2673 <p>The '<tt>switch</tt>' instruction uses three parameters: an integer
2674 comparison value '<tt>value</tt>', a default '<tt>label</tt>' destination,
2675 and an array of pairs of comparison value constants and '<tt>label</tt>'s.
2676 The table is not allowed to contain duplicate constant entries.</p>
2679 <p>The <tt>switch</tt> instruction specifies a table of values and
2680 destinations. When the '<tt>switch</tt>' instruction is executed, this table
2681 is searched for the given value. If the value is found, control flow is
2682 transferred to the corresponding destination; otherwise, control flow is
2683 transferred to the default destination.</p>
2685 <h5>Implementation:</h5>
2686 <p>Depending on properties of the target machine and the particular
2687 <tt>switch</tt> instruction, this instruction may be code generated in
2688 different ways. For example, it could be generated as a series of chained
2689 conditional branches or with a lookup table.</p>
2693 <i>; Emulate a conditional br instruction</i>
2694 %Val = <a href="#i_zext">zext</a> i1 %value to i32
2695 switch i32 %Val, label %truedest [ i32 0, label %falsedest ]
2697 <i>; Emulate an unconditional br instruction</i>
2698 switch i32 0, label %dest [ ]
2700 <i>; Implement a jump table:</i>
2701 switch i32 %val, label %otherwise [ i32 0, label %onzero
2703 i32 2, label %ontwo ]
2709 <!-- _______________________________________________________________________ -->
2710 <div class="doc_subsubsection">
2711 <a name="i_indirectbr">'<tt>indirectbr</tt>' Instruction</a>
2714 <div class="doc_text">
2718 indirectbr <somety>* <address>, [ label <dest1>, label <dest2>, ... ]
2723 <p>The '<tt>indirectbr</tt>' instruction implements an indirect branch to a label
2724 within the current function, whose address is specified by
2725 "<tt>address</tt>". Address must be derived from a <a
2726 href="#blockaddress">blockaddress</a> constant.</p>
2730 <p>The '<tt>address</tt>' argument is the address of the label to jump to. The
2731 rest of the arguments indicate the full set of possible destinations that the
2732 address may point to. Blocks are allowed to occur multiple times in the
2733 destination list, though this isn't particularly useful.</p>
2735 <p>This destination list is required so that dataflow analysis has an accurate
2736 understanding of the CFG.</p>
2740 <p>Control transfers to the block specified in the address argument. All
2741 possible destination blocks must be listed in the label list, otherwise this
2742 instruction has undefined behavior. This implies that jumps to labels
2743 defined in other functions have undefined behavior as well.</p>
2745 <h5>Implementation:</h5>
2747 <p>This is typically implemented with a jump through a register.</p>
2751 indirectbr i8* %Addr, [ label %bb1, label %bb2, label %bb3 ]
2757 <!-- _______________________________________________________________________ -->
2758 <div class="doc_subsubsection">
2759 <a name="i_invoke">'<tt>invoke</tt>' Instruction</a>
2762 <div class="doc_text">
2766 <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>]
2767 to label <normal label> unwind label <exception label>
2771 <p>The '<tt>invoke</tt>' instruction causes control to transfer to a specified
2772 function, with the possibility of control flow transfer to either the
2773 '<tt>normal</tt>' label or the '<tt>exception</tt>' label. If the callee
2774 function returns with the "<tt><a href="#i_ret">ret</a></tt>" instruction,
2775 control flow will return to the "normal" label. If the callee (or any
2776 indirect callees) returns with the "<a href="#i_unwind"><tt>unwind</tt></a>"
2777 instruction, control is interrupted and continued at the dynamically nearest
2778 "exception" label.</p>
2781 <p>This instruction requires several arguments:</p>
2784 <li>The optional "cconv" marker indicates which <a href="#callingconv">calling
2785 convention</a> the call should use. If none is specified, the call
2786 defaults to using C calling conventions.</li>
2788 <li>The optional <a href="#paramattrs">Parameter Attributes</a> list for
2789 return values. Only '<tt>zeroext</tt>', '<tt>signext</tt>', and
2790 '<tt>inreg</tt>' attributes are valid here.</li>
2792 <li>'<tt>ptr to function ty</tt>': shall be the signature of the pointer to
2793 function value being invoked. In most cases, this is a direct function
2794 invocation, but indirect <tt>invoke</tt>s are just as possible, branching
2795 off an arbitrary pointer to function value.</li>
2797 <li>'<tt>function ptr val</tt>': An LLVM value containing a pointer to a
2798 function to be invoked. </li>
2800 <li>'<tt>function args</tt>': argument list whose types match the function
2801 signature argument types. If the function signature indicates the
2802 function accepts a variable number of arguments, the extra arguments can
2805 <li>'<tt>normal label</tt>': the label reached when the called function
2806 executes a '<tt><a href="#i_ret">ret</a></tt>' instruction. </li>
2808 <li>'<tt>exception label</tt>': the label reached when a callee returns with
2809 the <a href="#i_unwind"><tt>unwind</tt></a> instruction. </li>
2811 <li>The optional <a href="#fnattrs">function attributes</a> list. Only
2812 '<tt>noreturn</tt>', '<tt>nounwind</tt>', '<tt>readonly</tt>' and
2813 '<tt>readnone</tt>' attributes are valid here.</li>
2817 <p>This instruction is designed to operate as a standard
2818 '<tt><a href="#i_call">call</a></tt>' instruction in most regards. The
2819 primary difference is that it establishes an association with a label, which
2820 is used by the runtime library to unwind the stack.</p>
2822 <p>This instruction is used in languages with destructors to ensure that proper
2823 cleanup is performed in the case of either a <tt>longjmp</tt> or a thrown
2824 exception. Additionally, this is important for implementation of
2825 '<tt>catch</tt>' clauses in high-level languages that support them.</p>
2827 <p>For the purposes of the SSA form, the definition of the value returned by the
2828 '<tt>invoke</tt>' instruction is deemed to occur on the edge from the current
2829 block to the "normal" label. If the callee unwinds then no return value is
2832 <p>Note that the code generator does not yet completely support unwind, and
2833 that the invoke/unwind semantics are likely to change in future versions.</p>
2837 %retval = invoke i32 @Test(i32 15) to label %Continue
2838 unwind label %TestCleanup <i>; {i32}:retval set</i>
2839 %retval = invoke <a href="#callingconv">coldcc</a> i32 %Testfnptr(i32 15) to label %Continue
2840 unwind label %TestCleanup <i>; {i32}:retval set</i>
2845 <!-- _______________________________________________________________________ -->
2847 <div class="doc_subsubsection"> <a name="i_unwind">'<tt>unwind</tt>'
2848 Instruction</a> </div>
2850 <div class="doc_text">
2858 <p>The '<tt>unwind</tt>' instruction unwinds the stack, continuing control flow
2859 at the first callee in the dynamic call stack which used
2860 an <a href="#i_invoke"><tt>invoke</tt></a> instruction to perform the call.
2861 This is primarily used to implement exception handling.</p>
2864 <p>The '<tt>unwind</tt>' instruction causes execution of the current function to
2865 immediately halt. The dynamic call stack is then searched for the
2866 first <a href="#i_invoke"><tt>invoke</tt></a> instruction on the call stack.
2867 Once found, execution continues at the "exceptional" destination block
2868 specified by the <tt>invoke</tt> instruction. If there is no <tt>invoke</tt>
2869 instruction in the dynamic call chain, undefined behavior results.</p>
2871 <p>Note that the code generator does not yet completely support unwind, and
2872 that the invoke/unwind semantics are likely to change in future versions.</p>
2876 <!-- _______________________________________________________________________ -->
2878 <div class="doc_subsubsection"> <a name="i_unreachable">'<tt>unreachable</tt>'
2879 Instruction</a> </div>
2881 <div class="doc_text">
2889 <p>The '<tt>unreachable</tt>' instruction has no defined semantics. This
2890 instruction is used to inform the optimizer that a particular portion of the
2891 code is not reachable. This can be used to indicate that the code after a
2892 no-return function cannot be reached, and other facts.</p>
2895 <p>The '<tt>unreachable</tt>' instruction has no defined semantics.</p>
2899 <!-- ======================================================================= -->
2900 <div class="doc_subsection"> <a name="binaryops">Binary Operations</a> </div>
2902 <div class="doc_text">
2904 <p>Binary operators are used to do most of the computation in a program. They
2905 require two operands of the same type, execute an operation on them, and
2906 produce a single value. The operands might represent multiple data, as is
2907 the case with the <a href="#t_vector">vector</a> data type. The result value
2908 has the same type as its operands.</p>
2910 <p>There are several different binary operators:</p>
2914 <!-- _______________________________________________________________________ -->
2915 <div class="doc_subsubsection">
2916 <a name="i_add">'<tt>add</tt>' Instruction</a>
2919 <div class="doc_text">
2923 <result> = add <ty> <op1>, <op2> <i>; yields {ty}:result</i>
2924 <result> = add nuw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
2925 <result> = add nsw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
2926 <result> = add nuw nsw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
2930 <p>The '<tt>add</tt>' instruction returns the sum of its two operands.</p>
2933 <p>The two arguments to the '<tt>add</tt>' instruction must
2934 be <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
2935 integer values. Both arguments must have identical types.</p>
2938 <p>The value produced is the integer sum of the two operands.</p>
2940 <p>If the sum has unsigned overflow, the result returned is the mathematical
2941 result modulo 2<sup>n</sup>, where n is the bit width of the result.</p>
2943 <p>Because LLVM integers use a two's complement representation, this instruction
2944 is appropriate for both signed and unsigned integers.</p>
2946 <p><tt>nuw</tt> and <tt>nsw</tt> stand for "No Unsigned Wrap"
2947 and "No Signed Wrap", respectively. If the <tt>nuw</tt> and/or
2948 <tt>nsw</tt> keywords are present, the result value of the <tt>add</tt>
2949 is undefined if unsigned and/or signed overflow, respectively, occurs.</p>
2953 <result> = add i32 4, %var <i>; yields {i32}:result = 4 + %var</i>
2958 <!-- _______________________________________________________________________ -->
2959 <div class="doc_subsubsection">
2960 <a name="i_fadd">'<tt>fadd</tt>' Instruction</a>
2963 <div class="doc_text">
2967 <result> = fadd <ty> <op1>, <op2> <i>; yields {ty}:result</i>
2971 <p>The '<tt>fadd</tt>' instruction returns the sum of its two operands.</p>
2974 <p>The two arguments to the '<tt>fadd</tt>' instruction must be
2975 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
2976 floating point values. Both arguments must have identical types.</p>
2979 <p>The value produced is the floating point sum of the two operands.</p>
2983 <result> = fadd float 4.0, %var <i>; yields {float}:result = 4.0 + %var</i>
2988 <!-- _______________________________________________________________________ -->
2989 <div class="doc_subsubsection">
2990 <a name="i_sub">'<tt>sub</tt>' Instruction</a>
2993 <div class="doc_text">
2997 <result> = sub <ty> <op1>, <op2> <i>; yields {ty}:result</i>
2998 <result> = sub nuw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
2999 <result> = sub nsw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3000 <result> = sub nuw nsw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3004 <p>The '<tt>sub</tt>' instruction returns the difference of its two
3007 <p>Note that the '<tt>sub</tt>' instruction is used to represent the
3008 '<tt>neg</tt>' instruction present in most other intermediate
3009 representations.</p>
3012 <p>The two arguments to the '<tt>sub</tt>' instruction must
3013 be <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
3014 integer values. Both arguments must have identical types.</p>
3017 <p>The value produced is the integer difference of the two operands.</p>
3019 <p>If the difference has unsigned overflow, the result returned is the
3020 mathematical result modulo 2<sup>n</sup>, where n is the bit width of the
3023 <p>Because LLVM integers use a two's complement representation, this instruction
3024 is appropriate for both signed and unsigned integers.</p>
3026 <p><tt>nuw</tt> and <tt>nsw</tt> stand for "No Unsigned Wrap"
3027 and "No Signed Wrap", respectively. If the <tt>nuw</tt> and/or
3028 <tt>nsw</tt> keywords are present, the result value of the <tt>sub</tt>
3029 is undefined if unsigned and/or signed overflow, respectively, occurs.</p>
3033 <result> = sub i32 4, %var <i>; yields {i32}:result = 4 - %var</i>
3034 <result> = sub i32 0, %val <i>; yields {i32}:result = -%var</i>
3039 <!-- _______________________________________________________________________ -->
3040 <div class="doc_subsubsection">
3041 <a name="i_fsub">'<tt>fsub</tt>' Instruction</a>
3044 <div class="doc_text">
3048 <result> = fsub <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3052 <p>The '<tt>fsub</tt>' instruction returns the difference of its two
3055 <p>Note that the '<tt>fsub</tt>' instruction is used to represent the
3056 '<tt>fneg</tt>' instruction present in most other intermediate
3057 representations.</p>
3060 <p>The two arguments to the '<tt>fsub</tt>' instruction must be
3061 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3062 floating point values. Both arguments must have identical types.</p>
3065 <p>The value produced is the floating point difference of the two operands.</p>
3069 <result> = fsub float 4.0, %var <i>; yields {float}:result = 4.0 - %var</i>
3070 <result> = fsub float -0.0, %val <i>; yields {float}:result = -%var</i>
3075 <!-- _______________________________________________________________________ -->
3076 <div class="doc_subsubsection">
3077 <a name="i_mul">'<tt>mul</tt>' Instruction</a>
3080 <div class="doc_text">
3084 <result> = mul <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3085 <result> = mul nuw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3086 <result> = mul nsw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3087 <result> = mul nuw nsw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3091 <p>The '<tt>mul</tt>' instruction returns the product of its two operands.</p>
3094 <p>The two arguments to the '<tt>mul</tt>' instruction must
3095 be <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
3096 integer values. Both arguments must have identical types.</p>
3099 <p>The value produced is the integer product of the two operands.</p>
3101 <p>If the result of the multiplication has unsigned overflow, the result
3102 returned is the mathematical result modulo 2<sup>n</sup>, where n is the bit
3103 width of the result.</p>
3105 <p>Because LLVM integers use a two's complement representation, and the result
3106 is the same width as the operands, this instruction returns the correct
3107 result for both signed and unsigned integers. If a full product
3108 (e.g. <tt>i32</tt>x<tt>i32</tt>-><tt>i64</tt>) is needed, the operands should
3109 be sign-extended or zero-extended as appropriate to the width of the full
3112 <p><tt>nuw</tt> and <tt>nsw</tt> stand for "No Unsigned Wrap"
3113 and "No Signed Wrap", respectively. If the <tt>nuw</tt> and/or
3114 <tt>nsw</tt> keywords are present, the result value of the <tt>mul</tt>
3115 is undefined if unsigned and/or signed overflow, respectively, occurs.</p>
3119 <result> = mul i32 4, %var <i>; yields {i32}:result = 4 * %var</i>
3124 <!-- _______________________________________________________________________ -->
3125 <div class="doc_subsubsection">
3126 <a name="i_fmul">'<tt>fmul</tt>' Instruction</a>
3129 <div class="doc_text">
3133 <result> = fmul <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3137 <p>The '<tt>fmul</tt>' instruction returns the product of its two operands.</p>
3140 <p>The two arguments to the '<tt>fmul</tt>' instruction must be
3141 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3142 floating point values. Both arguments must have identical types.</p>
3145 <p>The value produced is the floating point product of the two operands.</p>
3149 <result> = fmul float 4.0, %var <i>; yields {float}:result = 4.0 * %var</i>
3154 <!-- _______________________________________________________________________ -->
3155 <div class="doc_subsubsection"> <a name="i_udiv">'<tt>udiv</tt>' Instruction
3158 <div class="doc_text">
3162 <result> = udiv <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3166 <p>The '<tt>udiv</tt>' instruction returns the quotient of its two operands.</p>
3169 <p>The two arguments to the '<tt>udiv</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 unsigned integer quotient of the two operands.</p>
3176 <p>Note that unsigned integer division and signed integer division are distinct
3177 operations; for signed integer division, use '<tt>sdiv</tt>'.</p>
3179 <p>Division by zero leads to undefined behavior.</p>
3183 <result> = udiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
3188 <!-- _______________________________________________________________________ -->
3189 <div class="doc_subsubsection"> <a name="i_sdiv">'<tt>sdiv</tt>' Instruction
3192 <div class="doc_text">
3196 <result> = sdiv <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3197 <result> = sdiv exact <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3201 <p>The '<tt>sdiv</tt>' instruction returns the quotient of its two operands.</p>
3204 <p>The two arguments to the '<tt>sdiv</tt>' instruction must be
3205 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3206 values. Both arguments must have identical types.</p>
3209 <p>The value produced is the signed integer quotient of the two operands rounded
3212 <p>Note that signed integer division and unsigned integer division are distinct
3213 operations; for unsigned integer division, use '<tt>udiv</tt>'.</p>
3215 <p>Division by zero leads to undefined behavior. Overflow also leads to
3216 undefined behavior; this is a rare case, but can occur, for example, by doing
3217 a 32-bit division of -2147483648 by -1.</p>
3219 <p>If the <tt>exact</tt> keyword is present, the result value of the
3220 <tt>sdiv</tt> is undefined if the result would be rounded or if overflow
3225 <result> = sdiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
3230 <!-- _______________________________________________________________________ -->
3231 <div class="doc_subsubsection"> <a name="i_fdiv">'<tt>fdiv</tt>'
3232 Instruction</a> </div>
3234 <div class="doc_text">
3238 <result> = fdiv <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3242 <p>The '<tt>fdiv</tt>' instruction returns the quotient of its two operands.</p>
3245 <p>The two arguments to the '<tt>fdiv</tt>' instruction must be
3246 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3247 floating point values. Both arguments must have identical types.</p>
3250 <p>The value produced is the floating point quotient of the two operands.</p>
3254 <result> = fdiv float 4.0, %var <i>; yields {float}:result = 4.0 / %var</i>
3259 <!-- _______________________________________________________________________ -->
3260 <div class="doc_subsubsection"> <a name="i_urem">'<tt>urem</tt>' Instruction</a>
3263 <div class="doc_text">
3267 <result> = urem <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3271 <p>The '<tt>urem</tt>' instruction returns the remainder from the unsigned
3272 division of its two arguments.</p>
3275 <p>The two arguments to the '<tt>urem</tt>' instruction must be
3276 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3277 values. Both arguments must have identical types.</p>
3280 <p>This instruction returns the unsigned integer <i>remainder</i> of a division.
3281 This instruction always performs an unsigned division to get the
3284 <p>Note that unsigned integer remainder and signed integer remainder are
3285 distinct operations; for signed integer remainder, use '<tt>srem</tt>'.</p>
3287 <p>Taking the remainder of a division by zero leads to undefined behavior.</p>
3291 <result> = urem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
3296 <!-- _______________________________________________________________________ -->
3297 <div class="doc_subsubsection">
3298 <a name="i_srem">'<tt>srem</tt>' Instruction</a>
3301 <div class="doc_text">
3305 <result> = srem <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3309 <p>The '<tt>srem</tt>' instruction returns the remainder from the signed
3310 division of its two operands. This instruction can also take
3311 <a href="#t_vector">vector</a> versions of the values in which case the
3312 elements must be integers.</p>
3315 <p>The two arguments to the '<tt>srem</tt>' instruction must be
3316 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3317 values. Both arguments must have identical types.</p>
3320 <p>This instruction returns the <i>remainder</i> of a division (where the result
3321 has the same sign as the dividend, <tt>op1</tt>), not the <i>modulo</i>
3322 operator (where the result has the same sign as the divisor, <tt>op2</tt>) of
3323 a value. For more information about the difference,
3324 see <a href="http://mathforum.org/dr.math/problems/anne.4.28.99.html">The
3325 Math Forum</a>. For a table of how this is implemented in various languages,
3326 please see <a href="http://en.wikipedia.org/wiki/Modulo_operation">
3327 Wikipedia: modulo operation</a>.</p>
3329 <p>Note that signed integer remainder and unsigned integer remainder are
3330 distinct operations; for unsigned integer remainder, use '<tt>urem</tt>'.</p>
3332 <p>Taking the remainder of a division by zero leads to undefined behavior.
3333 Overflow also leads to undefined behavior; this is a rare case, but can
3334 occur, for example, by taking the remainder of a 32-bit division of
3335 -2147483648 by -1. (The remainder doesn't actually overflow, but this rule
3336 lets srem be implemented using instructions that return both the result of
3337 the division and the remainder.)</p>
3341 <result> = srem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
3346 <!-- _______________________________________________________________________ -->
3347 <div class="doc_subsubsection">
3348 <a name="i_frem">'<tt>frem</tt>' Instruction</a> </div>
3350 <div class="doc_text">
3354 <result> = frem <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3358 <p>The '<tt>frem</tt>' instruction returns the remainder from the division of
3359 its two operands.</p>
3362 <p>The two arguments to the '<tt>frem</tt>' instruction must be
3363 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3364 floating point values. Both arguments must have identical types.</p>
3367 <p>This instruction returns the <i>remainder</i> of a division. The remainder
3368 has the same sign as the dividend.</p>
3372 <result> = frem float 4.0, %var <i>; yields {float}:result = 4.0 % %var</i>
3377 <!-- ======================================================================= -->
3378 <div class="doc_subsection"> <a name="bitwiseops">Bitwise Binary
3379 Operations</a> </div>
3381 <div class="doc_text">
3383 <p>Bitwise binary operators are used to do various forms of bit-twiddling in a
3384 program. They are generally very efficient instructions and can commonly be
3385 strength reduced from other instructions. They require two operands of the
3386 same type, execute an operation on them, and produce a single value. The
3387 resulting value is the same type as its operands.</p>
3391 <!-- _______________________________________________________________________ -->
3392 <div class="doc_subsubsection"> <a name="i_shl">'<tt>shl</tt>'
3393 Instruction</a> </div>
3395 <div class="doc_text">
3399 <result> = shl <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3403 <p>The '<tt>shl</tt>' instruction returns the first operand shifted to the left
3404 a specified number of bits.</p>
3407 <p>Both arguments to the '<tt>shl</tt>' instruction must be the
3408 same <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
3409 integer type. '<tt>op2</tt>' is treated as an unsigned value.</p>
3412 <p>The value produced is <tt>op1</tt> * 2<sup><tt>op2</tt></sup> mod
3413 2<sup>n</sup>, where <tt>n</tt> is the width of the result. If <tt>op2</tt>
3414 is (statically or dynamically) negative or equal to or larger than the number
3415 of bits in <tt>op1</tt>, the result is undefined. If the arguments are
3416 vectors, each vector element of <tt>op1</tt> is shifted by the corresponding
3417 shift amount in <tt>op2</tt>.</p>
3421 <result> = shl i32 4, %var <i>; yields {i32}: 4 << %var</i>
3422 <result> = shl i32 4, 2 <i>; yields {i32}: 16</i>
3423 <result> = shl i32 1, 10 <i>; yields {i32}: 1024</i>
3424 <result> = shl i32 1, 32 <i>; undefined</i>
3425 <result> = shl <2 x i32> < i32 1, i32 1>, < i32 1, i32 2> <i>; yields: result=<2 x i32> < i32 2, i32 4></i>
3430 <!-- _______________________________________________________________________ -->
3431 <div class="doc_subsubsection"> <a name="i_lshr">'<tt>lshr</tt>'
3432 Instruction</a> </div>
3434 <div class="doc_text">
3438 <result> = lshr <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3442 <p>The '<tt>lshr</tt>' instruction (logical shift right) returns the first
3443 operand shifted to the right a specified number of bits with zero fill.</p>
3446 <p>Both arguments to the '<tt>lshr</tt>' instruction must be the same
3447 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3448 type. '<tt>op2</tt>' is treated as an unsigned value.</p>
3451 <p>This instruction always performs a logical shift right operation. The most
3452 significant bits of the result will be filled with zero bits after the shift.
3453 If <tt>op2</tt> is (statically or dynamically) equal to or larger than the
3454 number of bits in <tt>op1</tt>, the result is undefined. If the arguments are
3455 vectors, each vector element of <tt>op1</tt> is shifted by the corresponding
3456 shift amount in <tt>op2</tt>.</p>
3460 <result> = lshr i32 4, 1 <i>; yields {i32}:result = 2</i>
3461 <result> = lshr i32 4, 2 <i>; yields {i32}:result = 1</i>
3462 <result> = lshr i8 4, 3 <i>; yields {i8}:result = 0</i>
3463 <result> = lshr i8 -2, 1 <i>; yields {i8}:result = 0x7FFFFFFF </i>
3464 <result> = lshr i32 1, 32 <i>; undefined</i>
3465 <result> = lshr <2 x i32> < i32 -2, i32 4>, < i32 1, i32 2> <i>; yields: result=<2 x i32> < i32 0x7FFFFFFF, i32 1></i>
3470 <!-- _______________________________________________________________________ -->
3471 <div class="doc_subsubsection"> <a name="i_ashr">'<tt>ashr</tt>'
3472 Instruction</a> </div>
3473 <div class="doc_text">
3477 <result> = ashr <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3481 <p>The '<tt>ashr</tt>' instruction (arithmetic shift right) returns the first
3482 operand shifted to the right a specified number of bits with sign
3486 <p>Both arguments to the '<tt>ashr</tt>' instruction must be the same
3487 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3488 type. '<tt>op2</tt>' is treated as an unsigned value.</p>
3491 <p>This instruction always performs an arithmetic shift right operation, The
3492 most significant bits of the result will be filled with the sign bit
3493 of <tt>op1</tt>. If <tt>op2</tt> is (statically or dynamically) equal to or
3494 larger than the number of bits in <tt>op1</tt>, the result is undefined. If
3495 the arguments are vectors, each vector element of <tt>op1</tt> is shifted by
3496 the corresponding shift amount in <tt>op2</tt>.</p>
3500 <result> = ashr i32 4, 1 <i>; yields {i32}:result = 2</i>
3501 <result> = ashr i32 4, 2 <i>; yields {i32}:result = 1</i>
3502 <result> = ashr i8 4, 3 <i>; yields {i8}:result = 0</i>
3503 <result> = ashr i8 -2, 1 <i>; yields {i8}:result = -1</i>
3504 <result> = ashr i32 1, 32 <i>; undefined</i>
3505 <result> = ashr <2 x i32> < i32 -2, i32 4>, < i32 1, i32 3> <i>; yields: result=<2 x i32> < i32 -1, i32 0></i>
3510 <!-- _______________________________________________________________________ -->
3511 <div class="doc_subsubsection"> <a name="i_and">'<tt>and</tt>'
3512 Instruction</a> </div>
3514 <div class="doc_text">
3518 <result> = and <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3522 <p>The '<tt>and</tt>' instruction returns the bitwise logical and of its two
3526 <p>The two arguments to the '<tt>and</tt>' instruction must be
3527 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3528 values. Both arguments must have identical types.</p>
3531 <p>The truth table used for the '<tt>and</tt>' instruction is:</p>
3533 <table border="1" cellspacing="0" cellpadding="4">
3565 <result> = and i32 4, %var <i>; yields {i32}:result = 4 & %var</i>
3566 <result> = and i32 15, 40 <i>; yields {i32}:result = 8</i>
3567 <result> = and i32 4, 8 <i>; yields {i32}:result = 0</i>
3570 <!-- _______________________________________________________________________ -->
3571 <div class="doc_subsubsection"> <a name="i_or">'<tt>or</tt>' Instruction</a> </div>
3573 <div class="doc_text">
3577 <result> = or <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3581 <p>The '<tt>or</tt>' instruction returns the bitwise logical inclusive or of its
3585 <p>The two arguments to the '<tt>or</tt>' instruction must be
3586 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3587 values. Both arguments must have identical types.</p>
3590 <p>The truth table used for the '<tt>or</tt>' instruction is:</p>
3592 <table border="1" cellspacing="0" cellpadding="4">
3624 <result> = or i32 4, %var <i>; yields {i32}:result = 4 | %var</i>
3625 <result> = or i32 15, 40 <i>; yields {i32}:result = 47</i>
3626 <result> = or i32 4, 8 <i>; yields {i32}:result = 12</i>
3631 <!-- _______________________________________________________________________ -->
3632 <div class="doc_subsubsection"> <a name="i_xor">'<tt>xor</tt>'
3633 Instruction</a> </div>
3635 <div class="doc_text">
3639 <result> = xor <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3643 <p>The '<tt>xor</tt>' instruction returns the bitwise logical exclusive or of
3644 its two operands. The <tt>xor</tt> is used to implement the "one's
3645 complement" operation, which is the "~" operator in C.</p>
3648 <p>The two arguments to the '<tt>xor</tt>' instruction must be
3649 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3650 values. Both arguments must have identical types.</p>
3653 <p>The truth table used for the '<tt>xor</tt>' instruction is:</p>
3655 <table border="1" cellspacing="0" cellpadding="4">
3687 <result> = xor i32 4, %var <i>; yields {i32}:result = 4 ^ %var</i>
3688 <result> = xor i32 15, 40 <i>; yields {i32}:result = 39</i>
3689 <result> = xor i32 4, 8 <i>; yields {i32}:result = 12</i>
3690 <result> = xor i32 %V, -1 <i>; yields {i32}:result = ~%V</i>
3695 <!-- ======================================================================= -->
3696 <div class="doc_subsection">
3697 <a name="vectorops">Vector Operations</a>
3700 <div class="doc_text">
3702 <p>LLVM supports several instructions to represent vector operations in a
3703 target-independent manner. These instructions cover the element-access and
3704 vector-specific operations needed to process vectors effectively. While LLVM
3705 does directly support these vector operations, many sophisticated algorithms
3706 will want to use target-specific intrinsics to take full advantage of a
3707 specific target.</p>
3711 <!-- _______________________________________________________________________ -->
3712 <div class="doc_subsubsection">
3713 <a name="i_extractelement">'<tt>extractelement</tt>' Instruction</a>
3716 <div class="doc_text">
3720 <result> = extractelement <n x <ty>> <val>, i32 <idx> <i>; yields <ty></i>
3724 <p>The '<tt>extractelement</tt>' instruction extracts a single scalar element
3725 from a vector at a specified index.</p>
3729 <p>The first operand of an '<tt>extractelement</tt>' instruction is a value
3730 of <a href="#t_vector">vector</a> type. The second operand is an index
3731 indicating the position from which to extract the element. The index may be
3735 <p>The result is a scalar of the same type as the element type of
3736 <tt>val</tt>. Its value is the value at position <tt>idx</tt> of
3737 <tt>val</tt>. If <tt>idx</tt> exceeds the length of <tt>val</tt>, the
3738 results are undefined.</p>
3742 <result> = extractelement <4 x i32> %vec, i32 0 <i>; yields i32</i>
3747 <!-- _______________________________________________________________________ -->
3748 <div class="doc_subsubsection">
3749 <a name="i_insertelement">'<tt>insertelement</tt>' Instruction</a>
3752 <div class="doc_text">
3756 <result> = insertelement <n x <ty>> <val>, <ty> <elt>, i32 <idx> <i>; yields <n x <ty>></i>
3760 <p>The '<tt>insertelement</tt>' instruction inserts a scalar element into a
3761 vector at a specified index.</p>
3764 <p>The first operand of an '<tt>insertelement</tt>' instruction is a value
3765 of <a href="#t_vector">vector</a> type. The second operand is a scalar value
3766 whose type must equal the element type of the first operand. The third
3767 operand is an index indicating the position at which to insert the value.
3768 The index may be a variable.</p>
3771 <p>The result is a vector of the same type as <tt>val</tt>. Its element values
3772 are those of <tt>val</tt> except at position <tt>idx</tt>, where it gets the
3773 value <tt>elt</tt>. If <tt>idx</tt> exceeds the length of <tt>val</tt>, the
3774 results are undefined.</p>
3778 <result> = insertelement <4 x i32> %vec, i32 1, i32 0 <i>; yields <4 x i32></i>
3783 <!-- _______________________________________________________________________ -->
3784 <div class="doc_subsubsection">
3785 <a name="i_shufflevector">'<tt>shufflevector</tt>' Instruction</a>
3788 <div class="doc_text">
3792 <result> = shufflevector <n x <ty>> <v1>, <n x <ty>> <v2>, <m x i32> <mask> <i>; yields <m x <ty>></i>
3796 <p>The '<tt>shufflevector</tt>' instruction constructs a permutation of elements
3797 from two input vectors, returning a vector with the same element type as the
3798 input and length that is the same as the shuffle mask.</p>
3801 <p>The first two operands of a '<tt>shufflevector</tt>' instruction are vectors
3802 with types that match each other. The third argument is a shuffle mask whose
3803 element type is always 'i32'. The result of the instruction is a vector
3804 whose length is the same as the shuffle mask and whose element type is the
3805 same as the element type of the first two operands.</p>
3807 <p>The shuffle mask operand is required to be a constant vector with either
3808 constant integer or undef values.</p>
3811 <p>The elements of the two input vectors are numbered from left to right across
3812 both of the vectors. The shuffle mask operand specifies, for each element of
3813 the result vector, which element of the two input vectors the result element
3814 gets. The element selector may be undef (meaning "don't care") and the
3815 second operand may be undef if performing a shuffle from only one vector.</p>
3819 <result> = shufflevector <4 x i32> %v1, <4 x i32> %v2,
3820 <4 x i32> <i32 0, i32 4, i32 1, i32 5> <i>; yields <4 x i32></i>
3821 <result> = shufflevector <4 x i32> %v1, <4 x i32> undef,
3822 <4 x i32> <i32 0, i32 1, i32 2, i32 3> <i>; yields <4 x i32></i> - Identity shuffle.
3823 <result> = shufflevector <8 x i32> %v1, <8 x i32> undef,
3824 <4 x i32> <i32 0, i32 1, i32 2, i32 3> <i>; yields <4 x i32></i>
3825 <result> = shufflevector <4 x i32> %v1, <4 x i32> %v2,
3826 <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>
3831 <!-- ======================================================================= -->
3832 <div class="doc_subsection">
3833 <a name="aggregateops">Aggregate Operations</a>
3836 <div class="doc_text">
3838 <p>LLVM supports several instructions for working with aggregate values.</p>
3842 <!-- _______________________________________________________________________ -->
3843 <div class="doc_subsubsection">
3844 <a name="i_extractvalue">'<tt>extractvalue</tt>' Instruction</a>
3847 <div class="doc_text">
3851 <result> = extractvalue <aggregate type> <val>, <idx>{, <idx>}*
3855 <p>The '<tt>extractvalue</tt>' instruction extracts the value of a struct field
3856 or array element from an aggregate value.</p>
3859 <p>The first operand of an '<tt>extractvalue</tt>' instruction is a value
3860 of <a href="#t_struct">struct</a> or <a href="#t_array">array</a> type. The
3861 operands are constant indices to specify which value to extract in a similar
3862 manner as indices in a
3863 '<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.</p>
3866 <p>The result is the value at the position in the aggregate specified by the
3871 <result> = extractvalue {i32, float} %agg, 0 <i>; yields i32</i>
3876 <!-- _______________________________________________________________________ -->
3877 <div class="doc_subsubsection">
3878 <a name="i_insertvalue">'<tt>insertvalue</tt>' Instruction</a>
3881 <div class="doc_text">
3885 <result> = insertvalue <aggregate type> <val>, <ty> <elt>, <idx> <i>; yields <aggregate type></i>
3889 <p>The '<tt>insertvalue</tt>' instruction inserts a value into a struct field or
3890 array element in an aggregate.</p>
3894 <p>The first operand of an '<tt>insertvalue</tt>' instruction is a value
3895 of <a href="#t_struct">struct</a> or <a href="#t_array">array</a> type. The
3896 second operand is a first-class value to insert. The following operands are
3897 constant indices indicating the position at which to insert the value in a
3898 similar manner as indices in a
3899 '<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction. The
3900 value to insert must have the same type as the value identified by the
3904 <p>The result is an aggregate of the same type as <tt>val</tt>. Its value is
3905 that of <tt>val</tt> except that the value at the position specified by the
3906 indices is that of <tt>elt</tt>.</p>
3910 %agg1 = insertvalue {i32, float} undef, i32 1, 0 <i>; yields {i32 1, float undef}</i>
3911 %agg2 = insertvalue {i32, float} %agg1, float %val, 1 <i>; yields {i32 1, float %val}</i>
3917 <!-- ======================================================================= -->
3918 <div class="doc_subsection">
3919 <a name="memoryops">Memory Access and Addressing Operations</a>
3922 <div class="doc_text">
3924 <p>A key design point of an SSA-based representation is how it represents
3925 memory. In LLVM, no memory locations are in SSA form, which makes things
3926 very simple. This section describes how to read, write, and allocate
3931 <!-- _______________________________________________________________________ -->
3932 <div class="doc_subsubsection">
3933 <a name="i_alloca">'<tt>alloca</tt>' Instruction</a>
3936 <div class="doc_text">
3940 <result> = alloca <type>[, i32 <NumElements>][, align <alignment>] <i>; yields {type*}:result</i>
3944 <p>The '<tt>alloca</tt>' instruction allocates memory on the stack frame of the
3945 currently executing function, to be automatically released when this function
3946 returns to its caller. The object is always allocated in the generic address
3947 space (address space zero).</p>
3950 <p>The '<tt>alloca</tt>' instruction
3951 allocates <tt>sizeof(<type>)*NumElements</tt> bytes of memory on the
3952 runtime stack, returning a pointer of the appropriate type to the program.
3953 If "NumElements" is specified, it is the number of elements allocated,
3954 otherwise "NumElements" is defaulted to be one. If a constant alignment is
3955 specified, the value result of the allocation is guaranteed to be aligned to
3956 at least that boundary. If not specified, or if zero, the target can choose
3957 to align the allocation on any convenient boundary compatible with the
3960 <p>'<tt>type</tt>' may be any sized type.</p>
3963 <p>Memory is allocated; a pointer is returned. The operation is undefined if
3964 there is insufficient stack space for the allocation. '<tt>alloca</tt>'d
3965 memory is automatically released when the function returns. The
3966 '<tt>alloca</tt>' instruction is commonly used to represent automatic
3967 variables that must have an address available. When the function returns
3968 (either with the <tt><a href="#i_ret">ret</a></tt>
3969 or <tt><a href="#i_unwind">unwind</a></tt> instructions), the memory is
3970 reclaimed. Allocating zero bytes is legal, but the result is undefined.</p>
3974 %ptr = alloca i32 <i>; yields {i32*}:ptr</i>
3975 %ptr = alloca i32, i32 4 <i>; yields {i32*}:ptr</i>
3976 %ptr = alloca i32, i32 4, align 1024 <i>; yields {i32*}:ptr</i>
3977 %ptr = alloca i32, align 1024 <i>; yields {i32*}:ptr</i>
3982 <!-- _______________________________________________________________________ -->
3983 <div class="doc_subsubsection"> <a name="i_load">'<tt>load</tt>'
3984 Instruction</a> </div>
3986 <div class="doc_text">
3990 <result> = load <ty>* <pointer>[, align <alignment>]
3991 <result> = volatile load <ty>* <pointer>[, align <alignment>]
3995 <p>The '<tt>load</tt>' instruction is used to read from memory.</p>
3998 <p>The argument to the '<tt>load</tt>' instruction specifies the memory address
3999 from which to load. The pointer must point to
4000 a <a href="#t_firstclass">first class</a> type. If the <tt>load</tt> is
4001 marked as <tt>volatile</tt>, then the optimizer is not allowed to modify the
4002 number or order of execution of this <tt>load</tt> with other
4003 volatile <tt>load</tt> and <tt><a href="#i_store">store</a></tt>
4006 <p>The optional constant "align" argument specifies the alignment of the
4007 operation (that is, the alignment of the memory address). A value of 0 or an
4008 omitted "align" argument means that the operation has the preferential
4009 alignment for the target. It is the responsibility of the code emitter to
4010 ensure that the alignment information is correct. Overestimating the
4011 alignment results in an undefined behavior. Underestimating the alignment may
4012 produce less efficient code. An alignment of 1 is always safe.</p>
4015 <p>The location of memory pointed to is loaded. If the value being loaded is of
4016 scalar type then the number of bytes read does not exceed the minimum number
4017 of bytes needed to hold all bits of the type. For example, loading an
4018 <tt>i24</tt> reads at most three bytes. When loading a value of a type like
4019 <tt>i20</tt> with a size that is not an integral number of bytes, the result
4020 is undefined if the value was not originally written using a store of the
4025 %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
4026 <a href="#i_store">store</a> i32 3, i32* %ptr <i>; yields {void}</i>
4027 %val = load i32* %ptr <i>; yields {i32}:val = i32 3</i>
4032 <!-- _______________________________________________________________________ -->
4033 <div class="doc_subsubsection"> <a name="i_store">'<tt>store</tt>'
4034 Instruction</a> </div>
4036 <div class="doc_text">
4040 store <ty> <value>, <ty>* <pointer>[, align <alignment>] <i>; yields {void}</i>
4041 volatile store <ty> <value>, <ty>* <pointer>[, align <alignment>] <i>; yields {void}</i>
4045 <p>The '<tt>store</tt>' instruction is used to write to memory.</p>
4048 <p>There are two arguments to the '<tt>store</tt>' instruction: a value to store
4049 and an address at which to store it. The type of the
4050 '<tt><pointer></tt>' operand must be a pointer to
4051 the <a href="#t_firstclass">first class</a> type of the
4052 '<tt><value></tt>' operand. If the <tt>store</tt> is marked
4053 as <tt>volatile</tt>, then the optimizer is not allowed to modify the number
4054 or order of execution of this <tt>store</tt> with other
4055 volatile <tt>load</tt> and <tt><a href="#i_store">store</a></tt>
4058 <p>The optional constant "align" argument specifies the alignment of the
4059 operation (that is, the alignment of the memory address). A value of 0 or an
4060 omitted "align" argument means that the operation has the preferential
4061 alignment for the target. It is the responsibility of the code emitter to
4062 ensure that the alignment information is correct. Overestimating the
4063 alignment results in an undefined behavior. Underestimating the alignment may
4064 produce less efficient code. An alignment of 1 is always safe.</p>
4067 <p>The contents of memory are updated to contain '<tt><value></tt>' at the
4068 location specified by the '<tt><pointer></tt>' operand. If
4069 '<tt><value></tt>' is of scalar type then the number of bytes written
4070 does not exceed the minimum number of bytes needed to hold all bits of the
4071 type. For example, storing an <tt>i24</tt> writes at most three bytes. When
4072 writing a value of a type like <tt>i20</tt> with a size that is not an
4073 integral number of bytes, it is unspecified what happens to the extra bits
4074 that do not belong to the type, but they will typically be overwritten.</p>
4078 %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
4079 store i32 3, i32* %ptr <i>; yields {void}</i>
4080 %val = <a href="#i_load">load</a> i32* %ptr <i>; yields {i32}:val = i32 3</i>
4085 <!-- _______________________________________________________________________ -->
4086 <div class="doc_subsubsection">
4087 <a name="i_getelementptr">'<tt>getelementptr</tt>' Instruction</a>
4090 <div class="doc_text">
4094 <result> = getelementptr <pty>* <ptrval>{, <ty> <idx>}*
4095 <result> = getelementptr inbounds <pty>* <ptrval>{, <ty> <idx>}*
4099 <p>The '<tt>getelementptr</tt>' instruction is used to get the address of a
4100 subelement of an aggregate data structure. It performs address calculation
4101 only and does not access memory.</p>
4104 <p>The first argument is always a pointer, and forms the basis of the
4105 calculation. The remaining arguments are indices that indicate which of the
4106 elements of the aggregate object are indexed. The interpretation of each
4107 index is dependent on the type being indexed into. The first index always
4108 indexes the pointer value given as the first argument, the second index
4109 indexes a value of the type pointed to (not necessarily the value directly
4110 pointed to, since the first index can be non-zero), etc. The first type
4111 indexed into must be a pointer value, subsequent types can be arrays, vectors
4112 and structs. Note that subsequent types being indexed into can never be
4113 pointers, since that would require loading the pointer before continuing
4116 <p>The type of each index argument depends on the type it is indexing into.
4117 When indexing into a (optionally packed) structure, only <tt>i32</tt> integer
4118 <b>constants</b> are allowed. When indexing into an array, pointer or
4119 vector, integers of any width are allowed, and they are not required to be
4122 <p>For example, let's consider a C code fragment and how it gets compiled to
4125 <div class="doc_code">
4138 int *foo(struct ST *s) {
4139 return &s[1].Z.B[5][13];
4144 <p>The LLVM code generated by the GCC frontend is:</p>
4146 <div class="doc_code">
4148 %RT = <a href="#namedtypes">type</a> { i8 , [10 x [20 x i32]], i8 }
4149 %ST = <a href="#namedtypes">type</a> { i32, double, %RT }
4151 define i32* @foo(%ST* %s) {
4153 %reg = getelementptr %ST* %s, i32 1, i32 2, i32 1, i32 5, i32 13
4160 <p>In the example above, the first index is indexing into the '<tt>%ST*</tt>'
4161 type, which is a pointer, yielding a '<tt>%ST</tt>' = '<tt>{ i32, double, %RT
4162 }</tt>' type, a structure. The second index indexes into the third element
4163 of the structure, yielding a '<tt>%RT</tt>' = '<tt>{ i8 , [10 x [20 x i32]],
4164 i8 }</tt>' type, another structure. The third index indexes into the second
4165 element of the structure, yielding a '<tt>[10 x [20 x i32]]</tt>' type, an
4166 array. The two dimensions of the array are subscripted into, yielding an
4167 '<tt>i32</tt>' type. The '<tt>getelementptr</tt>' instruction returns a
4168 pointer to this element, thus computing a value of '<tt>i32*</tt>' type.</p>
4170 <p>Note that it is perfectly legal to index partially through a structure,
4171 returning a pointer to an inner element. Because of this, the LLVM code for
4172 the given testcase is equivalent to:</p>
4175 define i32* @foo(%ST* %s) {
4176 %t1 = getelementptr %ST* %s, i32 1 <i>; yields %ST*:%t1</i>
4177 %t2 = getelementptr %ST* %t1, i32 0, i32 2 <i>; yields %RT*:%t2</i>
4178 %t3 = getelementptr %RT* %t2, i32 0, i32 1 <i>; yields [10 x [20 x i32]]*:%t3</i>
4179 %t4 = getelementptr [10 x [20 x i32]]* %t3, i32 0, i32 5 <i>; yields [20 x i32]*:%t4</i>
4180 %t5 = getelementptr [20 x i32]* %t4, i32 0, i32 13 <i>; yields i32*:%t5</i>
4185 <p>If the <tt>inbounds</tt> keyword is present, the result value of the
4186 <tt>getelementptr</tt> is undefined if the base pointer is not an
4187 <i>in bounds</i> address of an allocated object, or if any of the addresses
4188 that would be formed by successive addition of the offsets implied by the
4189 indices to the base address with infinitely precise arithmetic are not an
4190 <i>in bounds</i> address of that allocated object.
4191 The <i>in bounds</i> addresses for an allocated object are all the addresses
4192 that point into the object, plus the address one byte past the end.</p>
4194 <p>If the <tt>inbounds</tt> keyword is not present, the offsets are added to
4195 the base address with silently-wrapping two's complement arithmetic, and
4196 the result value of the <tt>getelementptr</tt> may be outside the object
4197 pointed to by the base pointer. The result value may not necessarily be
4198 used to access memory though, even if it happens to point into allocated
4199 storage. See the <a href="#pointeraliasing">Pointer Aliasing Rules</a>
4200 section for more information.</p>
4202 <p>The getelementptr instruction is often confusing. For some more insight into
4203 how it works, see <a href="GetElementPtr.html">the getelementptr FAQ</a>.</p>
4207 <i>; yields [12 x i8]*:aptr</i>
4208 %aptr = getelementptr {i32, [12 x i8]}* %saptr, i64 0, i32 1
4209 <i>; yields i8*:vptr</i>
4210 %vptr = getelementptr {i32, <2 x i8>}* %svptr, i64 0, i32 1, i32 1
4211 <i>; yields i8*:eptr</i>
4212 %eptr = getelementptr [12 x i8]* %aptr, i64 0, i32 1
4213 <i>; yields i32*:iptr</i>
4214 %iptr = getelementptr [10 x i32]* @arr, i16 0, i16 0
4219 <!-- ======================================================================= -->
4220 <div class="doc_subsection"> <a name="convertops">Conversion Operations</a>
4223 <div class="doc_text">
4225 <p>The instructions in this category are the conversion instructions (casting)
4226 which all take a single operand and a type. They perform various bit
4227 conversions on the operand.</p>
4231 <!-- _______________________________________________________________________ -->
4232 <div class="doc_subsubsection">
4233 <a name="i_trunc">'<tt>trunc .. to</tt>' Instruction</a>
4235 <div class="doc_text">
4239 <result> = trunc <ty> <value> to <ty2> <i>; yields ty2</i>
4243 <p>The '<tt>trunc</tt>' instruction truncates its operand to the
4244 type <tt>ty2</tt>.</p>
4247 <p>The '<tt>trunc</tt>' instruction takes a <tt>value</tt> to trunc, which must
4248 be an <a href="#t_integer">integer</a> type, and a type that specifies the
4249 size and type of the result, which must be
4250 an <a href="#t_integer">integer</a> type. The bit size of <tt>value</tt> must
4251 be larger than the bit size of <tt>ty2</tt>. Equal sized types are not
4255 <p>The '<tt>trunc</tt>' instruction truncates the high order bits
4256 in <tt>value</tt> and converts the remaining bits to <tt>ty2</tt>. Since the
4257 source size must be larger than the destination size, <tt>trunc</tt> cannot
4258 be a <i>no-op cast</i>. It will always truncate bits.</p>
4262 %X = trunc i32 257 to i8 <i>; yields i8:1</i>
4263 %Y = trunc i32 123 to i1 <i>; yields i1:true</i>
4264 %Z = trunc i32 122 to i1 <i>; yields i1:false</i>
4269 <!-- _______________________________________________________________________ -->
4270 <div class="doc_subsubsection">
4271 <a name="i_zext">'<tt>zext .. to</tt>' Instruction</a>
4273 <div class="doc_text">
4277 <result> = zext <ty> <value> to <ty2> <i>; yields ty2</i>
4281 <p>The '<tt>zext</tt>' instruction zero extends its operand to type
4286 <p>The '<tt>zext</tt>' instruction takes a value to cast, which must be of
4287 <a href="#t_integer">integer</a> type, and a type to cast it to, which must
4288 also be of <a href="#t_integer">integer</a> type. The bit size of the
4289 <tt>value</tt> must be smaller than the bit size of the destination type,
4293 <p>The <tt>zext</tt> fills the high order bits of the <tt>value</tt> with zero
4294 bits until it reaches the size of the destination type, <tt>ty2</tt>.</p>
4296 <p>When zero extending from i1, the result will always be either 0 or 1.</p>
4300 %X = zext i32 257 to i64 <i>; yields i64:257</i>
4301 %Y = zext i1 true to i32 <i>; yields i32:1</i>
4306 <!-- _______________________________________________________________________ -->
4307 <div class="doc_subsubsection">
4308 <a name="i_sext">'<tt>sext .. to</tt>' Instruction</a>
4310 <div class="doc_text">
4314 <result> = sext <ty> <value> to <ty2> <i>; yields ty2</i>
4318 <p>The '<tt>sext</tt>' sign extends <tt>value</tt> to the type <tt>ty2</tt>.</p>
4321 <p>The '<tt>sext</tt>' instruction takes a value to cast, which must be of
4322 <a href="#t_integer">integer</a> type, and a type to cast it to, which must
4323 also be of <a href="#t_integer">integer</a> type. The bit size of the
4324 <tt>value</tt> must be smaller than the bit size of the destination type,
4328 <p>The '<tt>sext</tt>' instruction performs a sign extension by copying the sign
4329 bit (highest order bit) of the <tt>value</tt> until it reaches the bit size
4330 of the type <tt>ty2</tt>.</p>
4332 <p>When sign extending from i1, the extension always results in -1 or 0.</p>
4336 %X = sext i8 -1 to i16 <i>; yields i16 :65535</i>
4337 %Y = sext i1 true to i32 <i>; yields i32:-1</i>
4342 <!-- _______________________________________________________________________ -->
4343 <div class="doc_subsubsection">
4344 <a name="i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a>
4347 <div class="doc_text">
4351 <result> = fptrunc <ty> <value> to <ty2> <i>; yields ty2</i>
4355 <p>The '<tt>fptrunc</tt>' instruction truncates <tt>value</tt> to type
4359 <p>The '<tt>fptrunc</tt>' instruction takes a <a href="#t_floating">floating
4360 point</a> value to cast and a <a href="#t_floating">floating point</a> type
4361 to cast it to. The size of <tt>value</tt> must be larger than the size of
4362 <tt>ty2</tt>. This implies that <tt>fptrunc</tt> cannot be used to make a
4363 <i>no-op cast</i>.</p>
4366 <p>The '<tt>fptrunc</tt>' instruction truncates a <tt>value</tt> from a larger
4367 <a href="#t_floating">floating point</a> type to a smaller
4368 <a href="#t_floating">floating point</a> type. If the value cannot fit
4369 within the destination type, <tt>ty2</tt>, then the results are
4374 %X = fptrunc double 123.0 to float <i>; yields float:123.0</i>
4375 %Y = fptrunc double 1.0E+300 to float <i>; yields undefined</i>
4380 <!-- _______________________________________________________________________ -->
4381 <div class="doc_subsubsection">
4382 <a name="i_fpext">'<tt>fpext .. to</tt>' Instruction</a>
4384 <div class="doc_text">
4388 <result> = fpext <ty> <value> to <ty2> <i>; yields ty2</i>
4392 <p>The '<tt>fpext</tt>' extends a floating point <tt>value</tt> to a larger
4393 floating point value.</p>
4396 <p>The '<tt>fpext</tt>' instruction takes a
4397 <a href="#t_floating">floating point</a> <tt>value</tt> to cast, and
4398 a <a href="#t_floating">floating point</a> type to cast it to. The source
4399 type must be smaller than the destination type.</p>
4402 <p>The '<tt>fpext</tt>' instruction extends the <tt>value</tt> from a smaller
4403 <a href="#t_floating">floating point</a> type to a larger
4404 <a href="#t_floating">floating point</a> type. The <tt>fpext</tt> cannot be
4405 used to make a <i>no-op cast</i> because it always changes bits. Use
4406 <tt>bitcast</tt> to make a <i>no-op cast</i> for a floating point cast.</p>
4410 %X = fpext float 3.1415 to double <i>; yields double:3.1415</i>
4411 %Y = fpext float 1.0 to float <i>; yields float:1.0 (no-op)</i>
4416 <!-- _______________________________________________________________________ -->
4417 <div class="doc_subsubsection">
4418 <a name="i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a>
4420 <div class="doc_text">
4424 <result> = fptoui <ty> <value> to <ty2> <i>; yields ty2</i>
4428 <p>The '<tt>fptoui</tt>' converts a floating point <tt>value</tt> to its
4429 unsigned integer equivalent of type <tt>ty2</tt>.</p>
4432 <p>The '<tt>fptoui</tt>' instruction takes a value to cast, which must be a
4433 scalar or vector <a href="#t_floating">floating point</a> value, and a type
4434 to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
4435 type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
4436 vector integer type with the same number of elements as <tt>ty</tt></p>
4439 <p>The '<tt>fptoui</tt>' instruction converts its
4440 <a href="#t_floating">floating point</a> operand into the nearest (rounding
4441 towards zero) unsigned integer value. If the value cannot fit
4442 in <tt>ty2</tt>, the results are undefined.</p>
4446 %X = fptoui double 123.0 to i32 <i>; yields i32:123</i>
4447 %Y = fptoui float 1.0E+300 to i1 <i>; yields undefined:1</i>
4448 %Z = fptoui float 1.04E+17 to i8 <i>; yields undefined:1</i>
4453 <!-- _______________________________________________________________________ -->
4454 <div class="doc_subsubsection">
4455 <a name="i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a>
4457 <div class="doc_text">
4461 <result> = fptosi <ty> <value> to <ty2> <i>; yields ty2</i>
4465 <p>The '<tt>fptosi</tt>' instruction converts
4466 <a href="#t_floating">floating point</a> <tt>value</tt> to
4467 type <tt>ty2</tt>.</p>
4470 <p>The '<tt>fptosi</tt>' instruction takes a value to cast, which must be a
4471 scalar or vector <a href="#t_floating">floating point</a> value, and a type
4472 to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
4473 type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
4474 vector integer type with the same number of elements as <tt>ty</tt></p>
4477 <p>The '<tt>fptosi</tt>' instruction converts its
4478 <a href="#t_floating">floating point</a> operand into the nearest (rounding
4479 towards zero) signed integer value. If the value cannot fit in <tt>ty2</tt>,
4480 the results are undefined.</p>
4484 %X = fptosi double -123.0 to i32 <i>; yields i32:-123</i>
4485 %Y = fptosi float 1.0E-247 to i1 <i>; yields undefined:1</i>
4486 %Z = fptosi float 1.04E+17 to i8 <i>; yields undefined:1</i>
4491 <!-- _______________________________________________________________________ -->
4492 <div class="doc_subsubsection">
4493 <a name="i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a>
4495 <div class="doc_text">
4499 <result> = uitofp <ty> <value> to <ty2> <i>; yields ty2</i>
4503 <p>The '<tt>uitofp</tt>' instruction regards <tt>value</tt> as an unsigned
4504 integer and converts that value to the <tt>ty2</tt> type.</p>
4507 <p>The '<tt>uitofp</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>uitofp</tt>' instruction interprets its operand as an unsigned
4515 integer quantity and converts it to the corresponding floating point
4516 value. If the value cannot fit in the floating point value, the results are
4521 %X = uitofp i32 257 to float <i>; yields float:257.0</i>
4522 %Y = uitofp i8 -1 to double <i>; yields double:255.0</i>
4527 <!-- _______________________________________________________________________ -->
4528 <div class="doc_subsubsection">
4529 <a name="i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a>
4531 <div class="doc_text">
4535 <result> = sitofp <ty> <value> to <ty2> <i>; yields ty2</i>
4539 <p>The '<tt>sitofp</tt>' instruction regards <tt>value</tt> as a signed integer
4540 and converts that value to the <tt>ty2</tt> type.</p>
4543 <p>The '<tt>sitofp</tt>' instruction takes a value to cast, which must be a
4544 scalar or vector <a href="#t_integer">integer</a> value, and a type to cast
4545 it to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
4546 type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
4547 floating point type with the same number of elements as <tt>ty</tt></p>
4550 <p>The '<tt>sitofp</tt>' instruction interprets its operand as a signed integer
4551 quantity and converts it to the corresponding floating point value. If the
4552 value cannot fit in the floating point value, the results are undefined.</p>
4556 %X = sitofp i32 257 to float <i>; yields float:257.0</i>
4557 %Y = sitofp i8 -1 to double <i>; yields double:-1.0</i>
4562 <!-- _______________________________________________________________________ -->
4563 <div class="doc_subsubsection">
4564 <a name="i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a>
4566 <div class="doc_text">
4570 <result> = ptrtoint <ty> <value> to <ty2> <i>; yields ty2</i>
4574 <p>The '<tt>ptrtoint</tt>' instruction converts the pointer <tt>value</tt> to
4575 the integer type <tt>ty2</tt>.</p>
4578 <p>The '<tt>ptrtoint</tt>' instruction takes a <tt>value</tt> to cast, which
4579 must be a <a href="#t_pointer">pointer</a> value, and a type to cast it to
4580 <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a> type.</p>
4583 <p>The '<tt>ptrtoint</tt>' instruction converts <tt>value</tt> to integer type
4584 <tt>ty2</tt> by interpreting the pointer value as an integer and either
4585 truncating or zero extending that value to the size of the integer type. If
4586 <tt>value</tt> is smaller than <tt>ty2</tt> then a zero extension is done. If
4587 <tt>value</tt> is larger than <tt>ty2</tt> then a truncation is done. If they
4588 are the same size, then nothing is done (<i>no-op cast</i>) other than a type
4593 %X = ptrtoint i32* %X to i8 <i>; yields truncation on 32-bit architecture</i>
4594 %Y = ptrtoint i32* %x to i64 <i>; yields zero extension on 32-bit architecture</i>
4599 <!-- _______________________________________________________________________ -->
4600 <div class="doc_subsubsection">
4601 <a name="i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a>
4603 <div class="doc_text">
4607 <result> = inttoptr <ty> <value> to <ty2> <i>; yields ty2</i>
4611 <p>The '<tt>inttoptr</tt>' instruction converts an integer <tt>value</tt> to a
4612 pointer type, <tt>ty2</tt>.</p>
4615 <p>The '<tt>inttoptr</tt>' instruction takes an <a href="#t_integer">integer</a>
4616 value to cast, and a type to cast it to, which must be a
4617 <a href="#t_pointer">pointer</a> type.</p>
4620 <p>The '<tt>inttoptr</tt>' instruction converts <tt>value</tt> to type
4621 <tt>ty2</tt> by applying either a zero extension or a truncation depending on
4622 the size of the integer <tt>value</tt>. If <tt>value</tt> is larger than the
4623 size of a pointer then a truncation is done. If <tt>value</tt> is smaller
4624 than the size of a pointer then a zero extension is done. If they are the
4625 same size, nothing is done (<i>no-op cast</i>).</p>
4629 %X = inttoptr i32 255 to i32* <i>; yields zero extension on 64-bit architecture</i>
4630 %Y = inttoptr i32 255 to i32* <i>; yields no-op on 32-bit architecture</i>
4631 %Z = inttoptr i64 0 to i32* <i>; yields truncation on 32-bit architecture</i>
4636 <!-- _______________________________________________________________________ -->
4637 <div class="doc_subsubsection">
4638 <a name="i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a>
4640 <div class="doc_text">
4644 <result> = bitcast <ty> <value> to <ty2> <i>; yields ty2</i>
4648 <p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
4649 <tt>ty2</tt> without changing any bits.</p>
4652 <p>The '<tt>bitcast</tt>' instruction takes a value to cast, which must be a
4653 non-aggregate first class value, and a type to cast it to, which must also be
4654 a non-aggregate <a href="#t_firstclass">first class</a> type. The bit sizes
4655 of <tt>value</tt> and the destination type, <tt>ty2</tt>, must be
4656 identical. If the source type is a pointer, the destination type must also be
4657 a pointer. This instruction supports bitwise conversion of vectors to
4658 integers and to vectors of other types (as long as they have the same
4662 <p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
4663 <tt>ty2</tt>. It is always a <i>no-op cast</i> because no bits change with
4664 this conversion. The conversion is done as if the <tt>value</tt> had been
4665 stored to memory and read back as type <tt>ty2</tt>. Pointer types may only
4666 be converted to other pointer types with this instruction. To convert
4667 pointers to other types, use the <a href="#i_inttoptr">inttoptr</a> or
4668 <a href="#i_ptrtoint">ptrtoint</a> instructions first.</p>
4672 %X = bitcast i8 255 to i8 <i>; yields i8 :-1</i>
4673 %Y = bitcast i32* %x to sint* <i>; yields sint*:%x</i>
4674 %Z = bitcast <2 x int> %V to i64; <i>; yields i64: %V</i>
4679 <!-- ======================================================================= -->
4680 <div class="doc_subsection"> <a name="otherops">Other Operations</a> </div>
4682 <div class="doc_text">
4684 <p>The instructions in this category are the "miscellaneous" instructions, which
4685 defy better classification.</p>
4689 <!-- _______________________________________________________________________ -->
4690 <div class="doc_subsubsection"><a name="i_icmp">'<tt>icmp</tt>' Instruction</a>
4693 <div class="doc_text">
4697 <result> = icmp <cond> <ty> <op1>, <op2> <i>; yields {i1} or {<N x i1>}:result</i>
4701 <p>The '<tt>icmp</tt>' instruction returns a boolean value or a vector of
4702 boolean values based on comparison of its two integer, integer vector, or
4703 pointer operands.</p>
4706 <p>The '<tt>icmp</tt>' instruction takes three operands. The first operand is
4707 the condition code indicating the kind of comparison to perform. It is not a
4708 value, just a keyword. The possible condition code are:</p>
4711 <li><tt>eq</tt>: equal</li>
4712 <li><tt>ne</tt>: not equal </li>
4713 <li><tt>ugt</tt>: unsigned greater than</li>
4714 <li><tt>uge</tt>: unsigned greater or equal</li>
4715 <li><tt>ult</tt>: unsigned less than</li>
4716 <li><tt>ule</tt>: unsigned less or equal</li>
4717 <li><tt>sgt</tt>: signed greater than</li>
4718 <li><tt>sge</tt>: signed greater or equal</li>
4719 <li><tt>slt</tt>: signed less than</li>
4720 <li><tt>sle</tt>: signed less or equal</li>
4723 <p>The remaining two arguments must be <a href="#t_integer">integer</a> or
4724 <a href="#t_pointer">pointer</a> or integer <a href="#t_vector">vector</a>
4725 typed. They must also be identical types.</p>
4728 <p>The '<tt>icmp</tt>' compares <tt>op1</tt> and <tt>op2</tt> according to the
4729 condition code given as <tt>cond</tt>. The comparison performed always yields
4730 either an <a href="#t_integer"><tt>i1</tt></a> or vector of <tt>i1</tt>
4731 result, as follows:</p>
4734 <li><tt>eq</tt>: yields <tt>true</tt> if the operands are equal,
4735 <tt>false</tt> otherwise. No sign interpretation is necessary or
4738 <li><tt>ne</tt>: yields <tt>true</tt> if the operands are unequal,
4739 <tt>false</tt> otherwise. No sign interpretation is necessary or
4742 <li><tt>ugt</tt>: interprets the operands as unsigned values and yields
4743 <tt>true</tt> if <tt>op1</tt> is greater than <tt>op2</tt>.</li>
4745 <li><tt>uge</tt>: interprets the operands as unsigned values and yields
4746 <tt>true</tt> if <tt>op1</tt> is greater than or equal
4747 to <tt>op2</tt>.</li>
4749 <li><tt>ult</tt>: interprets the operands as unsigned values and yields
4750 <tt>true</tt> if <tt>op1</tt> is less than <tt>op2</tt>.</li>
4752 <li><tt>ule</tt>: interprets the operands as unsigned values and yields
4753 <tt>true</tt> if <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
4755 <li><tt>sgt</tt>: interprets the operands as signed values and yields
4756 <tt>true</tt> if <tt>op1</tt> is greater than <tt>op2</tt>.</li>
4758 <li><tt>sge</tt>: interprets the operands as signed values and yields
4759 <tt>true</tt> if <tt>op1</tt> is greater than or equal
4760 to <tt>op2</tt>.</li>
4762 <li><tt>slt</tt>: interprets the operands as signed values and yields
4763 <tt>true</tt> if <tt>op1</tt> is less than <tt>op2</tt>.</li>
4765 <li><tt>sle</tt>: interprets the operands as signed values and yields
4766 <tt>true</tt> if <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
4769 <p>If the operands are <a href="#t_pointer">pointer</a> typed, the pointer
4770 values are compared as if they were integers.</p>
4772 <p>If the operands are integer vectors, then they are compared element by
4773 element. The result is an <tt>i1</tt> vector with the same number of elements
4774 as the values being compared. Otherwise, the result is an <tt>i1</tt>.</p>
4778 <result> = icmp eq i32 4, 5 <i>; yields: result=false</i>
4779 <result> = icmp ne float* %X, %X <i>; yields: result=false</i>
4780 <result> = icmp ult i16 4, 5 <i>; yields: result=true</i>
4781 <result> = icmp sgt i16 4, 5 <i>; yields: result=false</i>
4782 <result> = icmp ule i16 -4, 5 <i>; yields: result=false</i>
4783 <result> = icmp sge i16 4, 5 <i>; yields: result=false</i>
4786 <p>Note that the code generator does not yet support vector types with
4787 the <tt>icmp</tt> instruction.</p>
4791 <!-- _______________________________________________________________________ -->
4792 <div class="doc_subsubsection"><a name="i_fcmp">'<tt>fcmp</tt>' Instruction</a>
4795 <div class="doc_text">
4799 <result> = fcmp <cond> <ty> <op1>, <op2> <i>; yields {i1} or {<N x i1>}:result</i>
4803 <p>The '<tt>fcmp</tt>' instruction returns a boolean value or vector of boolean
4804 values based on comparison of its operands.</p>
4806 <p>If the operands are floating point scalars, then the result type is a boolean
4807 (<a href="#t_integer"><tt>i1</tt></a>).</p>
4809 <p>If the operands are floating point vectors, then the result type is a vector
4810 of boolean with the same number of elements as the operands being
4814 <p>The '<tt>fcmp</tt>' instruction takes three operands. The first operand is
4815 the condition code indicating the kind of comparison to perform. It is not a
4816 value, just a keyword. The possible condition code are:</p>
4819 <li><tt>false</tt>: no comparison, always returns false</li>
4820 <li><tt>oeq</tt>: ordered and equal</li>
4821 <li><tt>ogt</tt>: ordered and greater than </li>
4822 <li><tt>oge</tt>: ordered and greater than or equal</li>
4823 <li><tt>olt</tt>: ordered and less than </li>
4824 <li><tt>ole</tt>: ordered and less than or equal</li>
4825 <li><tt>one</tt>: ordered and not equal</li>
4826 <li><tt>ord</tt>: ordered (no nans)</li>
4827 <li><tt>ueq</tt>: unordered or equal</li>
4828 <li><tt>ugt</tt>: unordered or greater than </li>
4829 <li><tt>uge</tt>: unordered or greater than or equal</li>
4830 <li><tt>ult</tt>: unordered or less than </li>
4831 <li><tt>ule</tt>: unordered or less than or equal</li>
4832 <li><tt>une</tt>: unordered or not equal</li>
4833 <li><tt>uno</tt>: unordered (either nans)</li>
4834 <li><tt>true</tt>: no comparison, always returns true</li>
4837 <p><i>Ordered</i> means that neither operand is a QNAN while
4838 <i>unordered</i> means that either operand may be a QNAN.</p>
4840 <p>Each of <tt>val1</tt> and <tt>val2</tt> arguments must be either
4841 a <a href="#t_floating">floating point</a> type or
4842 a <a href="#t_vector">vector</a> of floating point type. They must have
4843 identical types.</p>
4846 <p>The '<tt>fcmp</tt>' instruction compares <tt>op1</tt> and <tt>op2</tt>
4847 according to the condition code given as <tt>cond</tt>. If the operands are
4848 vectors, then the vectors are compared element by element. Each comparison
4849 performed always yields an <a href="#t_integer">i1</a> result, as
4853 <li><tt>false</tt>: always yields <tt>false</tt>, regardless of operands.</li>
4855 <li><tt>oeq</tt>: yields <tt>true</tt> if both operands are not a QNAN and
4856 <tt>op1</tt> is equal to <tt>op2</tt>.</li>
4858 <li><tt>ogt</tt>: yields <tt>true</tt> if both operands are not a QNAN and
4859 <tt>op1</tt> is greather than <tt>op2</tt>.</li>
4861 <li><tt>oge</tt>: yields <tt>true</tt> if both operands are not a QNAN and
4862 <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
4864 <li><tt>olt</tt>: yields <tt>true</tt> if both operands are not a QNAN and
4865 <tt>op1</tt> is less than <tt>op2</tt>.</li>
4867 <li><tt>ole</tt>: yields <tt>true</tt> if both operands are not a QNAN and
4868 <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
4870 <li><tt>one</tt>: yields <tt>true</tt> if both operands are not a QNAN and
4871 <tt>op1</tt> is not equal to <tt>op2</tt>.</li>
4873 <li><tt>ord</tt>: yields <tt>true</tt> if both operands are not a QNAN.</li>
4875 <li><tt>ueq</tt>: yields <tt>true</tt> if either operand is a QNAN or
4876 <tt>op1</tt> is equal to <tt>op2</tt>.</li>
4878 <li><tt>ugt</tt>: yields <tt>true</tt> if either operand is a QNAN or
4879 <tt>op1</tt> is greater than <tt>op2</tt>.</li>
4881 <li><tt>uge</tt>: yields <tt>true</tt> if either operand is a QNAN or
4882 <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
4884 <li><tt>ult</tt>: yields <tt>true</tt> if either operand is a QNAN or
4885 <tt>op1</tt> is less than <tt>op2</tt>.</li>
4887 <li><tt>ule</tt>: yields <tt>true</tt> if either operand is a QNAN or
4888 <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
4890 <li><tt>une</tt>: yields <tt>true</tt> if either operand is a QNAN or
4891 <tt>op1</tt> is not equal to <tt>op2</tt>.</li>
4893 <li><tt>uno</tt>: yields <tt>true</tt> if either operand is a QNAN.</li>
4895 <li><tt>true</tt>: always yields <tt>true</tt>, regardless of operands.</li>
4900 <result> = fcmp oeq float 4.0, 5.0 <i>; yields: result=false</i>
4901 <result> = fcmp one float 4.0, 5.0 <i>; yields: result=true</i>
4902 <result> = fcmp olt float 4.0, 5.0 <i>; yields: result=true</i>
4903 <result> = fcmp ueq double 1.0, 2.0 <i>; yields: result=false</i>
4906 <p>Note that the code generator does not yet support vector types with
4907 the <tt>fcmp</tt> instruction.</p>
4911 <!-- _______________________________________________________________________ -->
4912 <div class="doc_subsubsection">
4913 <a name="i_phi">'<tt>phi</tt>' Instruction</a>
4916 <div class="doc_text">
4920 <result> = phi <ty> [ <val0>, <label0>], ...
4924 <p>The '<tt>phi</tt>' instruction is used to implement the φ node in the
4925 SSA graph representing the function.</p>
4928 <p>The type of the incoming values is specified with the first type field. After
4929 this, the '<tt>phi</tt>' instruction takes a list of pairs as arguments, with
4930 one pair for each predecessor basic block of the current block. Only values
4931 of <a href="#t_firstclass">first class</a> type may be used as the value
4932 arguments to the PHI node. Only labels may be used as the label
4935 <p>There must be no non-phi instructions between the start of a basic block and
4936 the PHI instructions: i.e. PHI instructions must be first in a basic
4939 <p>For the purposes of the SSA form, the use of each incoming value is deemed to
4940 occur on the edge from the corresponding predecessor block to the current
4941 block (but after any definition of an '<tt>invoke</tt>' instruction's return
4942 value on the same edge).</p>
4945 <p>At runtime, the '<tt>phi</tt>' instruction logically takes on the value
4946 specified by the pair corresponding to the predecessor basic block that
4947 executed just prior to the current block.</p>
4951 Loop: ; Infinite loop that counts from 0 on up...
4952 %indvar = phi i32 [ 0, %LoopHeader ], [ %nextindvar, %Loop ]
4953 %nextindvar = add i32 %indvar, 1
4959 <!-- _______________________________________________________________________ -->
4960 <div class="doc_subsubsection">
4961 <a name="i_select">'<tt>select</tt>' Instruction</a>
4964 <div class="doc_text">
4968 <result> = select <i>selty</i> <cond>, <ty> <val1>, <ty> <val2> <i>; yields ty</i>
4970 <i>selty</i> is either i1 or {<N x i1>}
4974 <p>The '<tt>select</tt>' instruction is used to choose one value based on a
4975 condition, without branching.</p>
4979 <p>The '<tt>select</tt>' instruction requires an 'i1' value or a vector of 'i1'
4980 values indicating the condition, and two values of the
4981 same <a href="#t_firstclass">first class</a> type. If the val1/val2 are
4982 vectors and the condition is a scalar, then entire vectors are selected, not
4983 individual elements.</p>
4986 <p>If the condition is an i1 and it evaluates to 1, the instruction returns the
4987 first value argument; otherwise, it returns the second value argument.</p>
4989 <p>If the condition is a vector of i1, then the value arguments must be vectors
4990 of the same size, and the selection is done element by element.</p>
4994 %X = select i1 true, i8 17, i8 42 <i>; yields i8:17</i>
4997 <p>Note that the code generator does not yet support conditions
4998 with vector type.</p>
5002 <!-- _______________________________________________________________________ -->
5003 <div class="doc_subsubsection">
5004 <a name="i_call">'<tt>call</tt>' Instruction</a>
5007 <div class="doc_text">
5011 <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>]
5015 <p>The '<tt>call</tt>' instruction represents a simple function call.</p>
5018 <p>This instruction requires several arguments:</p>
5021 <li>The optional "tail" marker indicates that the callee function does not
5022 access any allocas or varargs in the caller. Note that calls may be
5023 marked "tail" even if they do not occur before
5024 a <a href="#i_ret"><tt>ret</tt></a> instruction. If the "tail" marker is
5025 present, the function call is eligible for tail call optimization,
5026 but <a href="CodeGenerator.html#tailcallopt">might not in fact be
5027 optimized into a jump</a>. As of this writing, the extra requirements for
5028 a call to actually be optimized are:
5030 <li>Caller and callee both have the calling
5031 convention <tt>fastcc</tt>.</li>
5032 <li>The call is in tail position (ret immediately follows call and ret
5033 uses value of call or is void).</li>
5034 <li>Option <tt>-tailcallopt</tt> is enabled,
5035 or <code>llvm::PerformTailCallOpt</code> is <code>true</code>.</li>
5036 <li><a href="CodeGenerator.html#tailcallopt">Platform specific
5037 constraints are met.</a></li>
5041 <li>The optional "cconv" marker indicates which <a href="#callingconv">calling
5042 convention</a> the call should use. If none is specified, the call
5043 defaults to using C calling conventions. The calling convention of the
5044 call must match the calling convention of the target function, or else the
5045 behavior is undefined.</li>
5047 <li>The optional <a href="#paramattrs">Parameter Attributes</a> list for
5048 return values. Only '<tt>zeroext</tt>', '<tt>signext</tt>', and
5049 '<tt>inreg</tt>' attributes are valid here.</li>
5051 <li>'<tt>ty</tt>': the type of the call instruction itself which is also the
5052 type of the return value. Functions that return no value are marked
5053 <tt><a href="#t_void">void</a></tt>.</li>
5055 <li>'<tt>fnty</tt>': shall be the signature of the pointer to function value
5056 being invoked. The argument types must match the types implied by this
5057 signature. This type can be omitted if the function is not varargs and if
5058 the function type does not return a pointer to a function.</li>
5060 <li>'<tt>fnptrval</tt>': An LLVM value containing a pointer to a function to
5061 be invoked. In most cases, this is a direct function invocation, but
5062 indirect <tt>call</tt>s are just as possible, calling an arbitrary pointer
5063 to function value.</li>
5065 <li>'<tt>function args</tt>': argument list whose types match the function
5066 signature argument types. All arguments must be of
5067 <a href="#t_firstclass">first class</a> type. If the function signature
5068 indicates the function accepts a variable number of arguments, the extra
5069 arguments can be specified.</li>
5071 <li>The optional <a href="#fnattrs">function attributes</a> list. Only
5072 '<tt>noreturn</tt>', '<tt>nounwind</tt>', '<tt>readonly</tt>' and
5073 '<tt>readnone</tt>' attributes are valid here.</li>
5077 <p>The '<tt>call</tt>' instruction is used to cause control flow to transfer to
5078 a specified function, with its incoming arguments bound to the specified
5079 values. Upon a '<tt><a href="#i_ret">ret</a></tt>' instruction in the called
5080 function, control flow continues with the instruction after the function
5081 call, and the return value of the function is bound to the result
5086 %retval = call i32 @test(i32 %argc)
5087 call i32 (i8 *, ...)* @printf(i8 * %msg, i32 12, i8 42) <i>; yields i32</i>
5088 %X = tail call i32 @foo() <i>; yields i32</i>
5089 %Y = tail call <a href="#callingconv">fastcc</a> i32 @foo() <i>; yields i32</i>
5090 call void %foo(i8 97 signext)
5092 %struct.A = type { i32, i8 }
5093 %r = call %struct.A @foo() <i>; yields { 32, i8 }</i>
5094 %gr = extractvalue %struct.A %r, 0 <i>; yields i32</i>
5095 %gr1 = extractvalue %struct.A %r, 1 <i>; yields i8</i>
5096 %Z = call void @foo() noreturn <i>; indicates that %foo never returns normally</i>
5097 %ZZ = call zeroext i32 @bar() <i>; Return value is %zero extended</i>
5100 <p>llvm treats calls to some functions with names and arguments that match the
5101 standard C99 library as being the C99 library functions, and may perform
5102 optimizations or generate code for them under that assumption. This is
5103 something we'd like to change in the future to provide better support for
5104 freestanding environments and non-C-based langauges.</p>
5108 <!-- _______________________________________________________________________ -->
5109 <div class="doc_subsubsection">
5110 <a name="i_va_arg">'<tt>va_arg</tt>' Instruction</a>
5113 <div class="doc_text">
5117 <resultval> = va_arg <va_list*> <arglist>, <argty>
5121 <p>The '<tt>va_arg</tt>' instruction is used to access arguments passed through
5122 the "variable argument" area of a function call. It is used to implement the
5123 <tt>va_arg</tt> macro in C.</p>
5126 <p>This instruction takes a <tt>va_list*</tt> value and the type of the
5127 argument. It returns a value of the specified argument type and increments
5128 the <tt>va_list</tt> to point to the next argument. The actual type
5129 of <tt>va_list</tt> is target specific.</p>
5132 <p>The '<tt>va_arg</tt>' instruction loads an argument of the specified type
5133 from the specified <tt>va_list</tt> and causes the <tt>va_list</tt> to point
5134 to the next argument. For more information, see the variable argument
5135 handling <a href="#int_varargs">Intrinsic Functions</a>.</p>
5137 <p>It is legal for this instruction to be called in a function which does not
5138 take a variable number of arguments, for example, the <tt>vfprintf</tt>
5141 <p><tt>va_arg</tt> is an LLVM instruction instead of
5142 an <a href="#intrinsics">intrinsic function</a> because it takes a type as an
5146 <p>See the <a href="#int_varargs">variable argument processing</a> section.</p>
5148 <p>Note that the code generator does not yet fully support va_arg on many
5149 targets. Also, it does not currently support va_arg with aggregate types on
5154 <!-- *********************************************************************** -->
5155 <div class="doc_section"> <a name="intrinsics">Intrinsic Functions</a> </div>
5156 <!-- *********************************************************************** -->
5158 <div class="doc_text">
5160 <p>LLVM supports the notion of an "intrinsic function". These functions have
5161 well known names and semantics and are required to follow certain
5162 restrictions. Overall, these intrinsics represent an extension mechanism for
5163 the LLVM language that does not require changing all of the transformations
5164 in LLVM when adding to the language (or the bitcode reader/writer, the
5165 parser, etc...).</p>
5167 <p>Intrinsic function names must all start with an "<tt>llvm.</tt>" prefix. This
5168 prefix is reserved in LLVM for intrinsic names; thus, function names may not
5169 begin with this prefix. Intrinsic functions must always be external
5170 functions: you cannot define the body of intrinsic functions. Intrinsic
5171 functions may only be used in call or invoke instructions: it is illegal to
5172 take the address of an intrinsic function. Additionally, because intrinsic
5173 functions are part of the LLVM language, it is required if any are added that
5174 they be documented here.</p>
5176 <p>Some intrinsic functions can be overloaded, i.e., the intrinsic represents a
5177 family of functions that perform the same operation but on different data
5178 types. Because LLVM can represent over 8 million different integer types,
5179 overloading is used commonly to allow an intrinsic function to operate on any
5180 integer type. One or more of the argument types or the result type can be
5181 overloaded to accept any integer type. Argument types may also be defined as
5182 exactly matching a previous argument's type or the result type. This allows
5183 an intrinsic function which accepts multiple arguments, but needs all of them
5184 to be of the same type, to only be overloaded with respect to a single
5185 argument or the result.</p>
5187 <p>Overloaded intrinsics will have the names of its overloaded argument types
5188 encoded into its function name, each preceded by a period. Only those types
5189 which are overloaded result in a name suffix. Arguments whose type is matched
5190 against another type do not. For example, the <tt>llvm.ctpop</tt> function
5191 can take an integer of any width and returns an integer of exactly the same
5192 integer width. This leads to a family of functions such as
5193 <tt>i8 @llvm.ctpop.i8(i8 %val)</tt> and <tt>i29 @llvm.ctpop.i29(i29
5194 %val)</tt>. Only one type, the return type, is overloaded, and only one type
5195 suffix is required. Because the argument's type is matched against the return
5196 type, it does not require its own name suffix.</p>
5198 <p>To learn how to add an intrinsic function, please see the
5199 <a href="ExtendingLLVM.html">Extending LLVM Guide</a>.</p>
5203 <!-- ======================================================================= -->
5204 <div class="doc_subsection">
5205 <a name="int_varargs">Variable Argument Handling Intrinsics</a>
5208 <div class="doc_text">
5210 <p>Variable argument support is defined in LLVM with
5211 the <a href="#i_va_arg"><tt>va_arg</tt></a> instruction and these three
5212 intrinsic functions. These functions are related to the similarly named
5213 macros defined in the <tt><stdarg.h></tt> header file.</p>
5215 <p>All of these functions operate on arguments that use a target-specific value
5216 type "<tt>va_list</tt>". The LLVM assembly language reference manual does
5217 not define what this type is, so all transformations should be prepared to
5218 handle these functions regardless of the type used.</p>
5220 <p>This example shows how the <a href="#i_va_arg"><tt>va_arg</tt></a>
5221 instruction and the variable argument handling intrinsic functions are
5224 <div class="doc_code">
5226 define i32 @test(i32 %X, ...) {
5227 ; Initialize variable argument processing
5229 %ap2 = bitcast i8** %ap to i8*
5230 call void @llvm.va_start(i8* %ap2)
5232 ; Read a single integer argument
5233 %tmp = va_arg i8** %ap, i32
5235 ; Demonstrate usage of llvm.va_copy and llvm.va_end
5237 %aq2 = bitcast i8** %aq to i8*
5238 call void @llvm.va_copy(i8* %aq2, i8* %ap2)
5239 call void @llvm.va_end(i8* %aq2)
5241 ; Stop processing of arguments.
5242 call void @llvm.va_end(i8* %ap2)
5246 declare void @llvm.va_start(i8*)
5247 declare void @llvm.va_copy(i8*, i8*)
5248 declare void @llvm.va_end(i8*)
5254 <!-- _______________________________________________________________________ -->
5255 <div class="doc_subsubsection">
5256 <a name="int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a>
5260 <div class="doc_text">
5264 declare void %llvm.va_start(i8* <arglist>)
5268 <p>The '<tt>llvm.va_start</tt>' intrinsic initializes <tt>*<arglist></tt>
5269 for subsequent use by <tt><a href="#i_va_arg">va_arg</a></tt>.</p>
5272 <p>The argument is a pointer to a <tt>va_list</tt> element to initialize.</p>
5275 <p>The '<tt>llvm.va_start</tt>' intrinsic works just like the <tt>va_start</tt>
5276 macro available in C. In a target-dependent way, it initializes
5277 the <tt>va_list</tt> element to which the argument points, so that the next
5278 call to <tt>va_arg</tt> will produce the first variable argument passed to
5279 the function. Unlike the C <tt>va_start</tt> macro, this intrinsic does not
5280 need to know the last argument of the function as the compiler can figure
5285 <!-- _______________________________________________________________________ -->
5286 <div class="doc_subsubsection">
5287 <a name="int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a>
5290 <div class="doc_text">
5294 declare void @llvm.va_end(i8* <arglist>)
5298 <p>The '<tt>llvm.va_end</tt>' intrinsic destroys <tt>*<arglist></tt>,
5299 which has been initialized previously
5300 with <tt><a href="#int_va_start">llvm.va_start</a></tt>
5301 or <tt><a href="#i_va_copy">llvm.va_copy</a></tt>.</p>
5304 <p>The argument is a pointer to a <tt>va_list</tt> to destroy.</p>
5307 <p>The '<tt>llvm.va_end</tt>' intrinsic works just like the <tt>va_end</tt>
5308 macro available in C. In a target-dependent way, it destroys
5309 the <tt>va_list</tt> element to which the argument points. Calls
5310 to <a href="#int_va_start"><tt>llvm.va_start</tt></a>
5311 and <a href="#int_va_copy"> <tt>llvm.va_copy</tt></a> must be matched exactly
5312 with calls to <tt>llvm.va_end</tt>.</p>
5316 <!-- _______________________________________________________________________ -->
5317 <div class="doc_subsubsection">
5318 <a name="int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a>
5321 <div class="doc_text">
5325 declare void @llvm.va_copy(i8* <destarglist>, i8* <srcarglist>)
5329 <p>The '<tt>llvm.va_copy</tt>' intrinsic copies the current argument position
5330 from the source argument list to the destination argument list.</p>
5333 <p>The first argument is a pointer to a <tt>va_list</tt> element to initialize.
5334 The second argument is a pointer to a <tt>va_list</tt> element to copy
5338 <p>The '<tt>llvm.va_copy</tt>' intrinsic works just like the <tt>va_copy</tt>
5339 macro available in C. In a target-dependent way, it copies the
5340 source <tt>va_list</tt> element into the destination <tt>va_list</tt>
5341 element. This intrinsic is necessary because
5342 the <tt><a href="#int_va_start"> llvm.va_start</a></tt> intrinsic may be
5343 arbitrarily complex and require, for example, memory allocation.</p>
5347 <!-- ======================================================================= -->
5348 <div class="doc_subsection">
5349 <a name="int_gc">Accurate Garbage Collection Intrinsics</a>
5352 <div class="doc_text">
5354 <p>LLVM support for <a href="GarbageCollection.html">Accurate Garbage
5355 Collection</a> (GC) requires the implementation and generation of these
5356 intrinsics. These intrinsics allow identification of <a href="#int_gcroot">GC
5357 roots on the stack</a>, as well as garbage collector implementations that
5358 require <a href="#int_gcread">read</a> and <a href="#int_gcwrite">write</a>
5359 barriers. Front-ends for type-safe garbage collected languages should generate
5360 these intrinsics to make use of the LLVM garbage collectors. For more details,
5361 see <a href="GarbageCollection.html">Accurate Garbage Collection with
5364 <p>The garbage collection intrinsics only operate on objects in the generic
5365 address space (address space zero).</p>
5369 <!-- _______________________________________________________________________ -->
5370 <div class="doc_subsubsection">
5371 <a name="int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a>
5374 <div class="doc_text">
5378 declare void @llvm.gcroot(i8** %ptrloc, i8* %metadata)
5382 <p>The '<tt>llvm.gcroot</tt>' intrinsic declares the existence of a GC root to
5383 the code generator, and allows some metadata to be associated with it.</p>
5386 <p>The first argument specifies the address of a stack object that contains the
5387 root pointer. The second pointer (which must be either a constant or a
5388 global value address) contains the meta-data to be associated with the
5392 <p>At runtime, a call to this intrinsic stores a null pointer into the "ptrloc"
5393 location. At compile-time, the code generator generates information to allow
5394 the runtime to find the pointer at GC safe points. The '<tt>llvm.gcroot</tt>'
5395 intrinsic may only be used in a function which <a href="#gc">specifies a GC
5400 <!-- _______________________________________________________________________ -->
5401 <div class="doc_subsubsection">
5402 <a name="int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a>
5405 <div class="doc_text">
5409 declare i8* @llvm.gcread(i8* %ObjPtr, i8** %Ptr)
5413 <p>The '<tt>llvm.gcread</tt>' intrinsic identifies reads of references from heap
5414 locations, allowing garbage collector implementations that require read
5418 <p>The second argument is the address to read from, which should be an address
5419 allocated from the garbage collector. The first object is a pointer to the
5420 start of the referenced object, if needed by the language runtime (otherwise
5424 <p>The '<tt>llvm.gcread</tt>' intrinsic has the same semantics as a load
5425 instruction, but may be replaced with substantially more complex code by the
5426 garbage collector runtime, as needed. The '<tt>llvm.gcread</tt>' intrinsic
5427 may only be used in a function which <a href="#gc">specifies a GC
5432 <!-- _______________________________________________________________________ -->
5433 <div class="doc_subsubsection">
5434 <a name="int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a>
5437 <div class="doc_text">
5441 declare void @llvm.gcwrite(i8* %P1, i8* %Obj, i8** %P2)
5445 <p>The '<tt>llvm.gcwrite</tt>' intrinsic identifies writes of references to heap
5446 locations, allowing garbage collector implementations that require write
5447 barriers (such as generational or reference counting collectors).</p>
5450 <p>The first argument is the reference to store, the second is the start of the
5451 object to store it to, and the third is the address of the field of Obj to
5452 store to. If the runtime does not require a pointer to the object, Obj may
5456 <p>The '<tt>llvm.gcwrite</tt>' intrinsic has the same semantics as a store
5457 instruction, but may be replaced with substantially more complex code by the
5458 garbage collector runtime, as needed. The '<tt>llvm.gcwrite</tt>' intrinsic
5459 may only be used in a function which <a href="#gc">specifies a GC
5464 <!-- ======================================================================= -->
5465 <div class="doc_subsection">
5466 <a name="int_codegen">Code Generator Intrinsics</a>
5469 <div class="doc_text">
5471 <p>These intrinsics are provided by LLVM to expose special features that may
5472 only be implemented with code generator support.</p>
5476 <!-- _______________________________________________________________________ -->
5477 <div class="doc_subsubsection">
5478 <a name="int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a>
5481 <div class="doc_text">
5485 declare i8 *@llvm.returnaddress(i32 <level>)
5489 <p>The '<tt>llvm.returnaddress</tt>' intrinsic attempts to compute a
5490 target-specific value indicating the return address of the current function
5491 or one of its callers.</p>
5494 <p>The argument to this intrinsic indicates which function to return the address
5495 for. Zero indicates the calling function, one indicates its caller, etc.
5496 The argument is <b>required</b> to be a constant integer value.</p>
5499 <p>The '<tt>llvm.returnaddress</tt>' intrinsic either returns a pointer
5500 indicating the return address of the specified call frame, or zero if it
5501 cannot be identified. The value returned by this intrinsic is likely to be
5502 incorrect or 0 for arguments other than zero, so it should only be used for
5503 debugging purposes.</p>
5505 <p>Note that calling this intrinsic does not prevent function inlining or other
5506 aggressive transformations, so the value returned may not be that of the
5507 obvious source-language caller.</p>
5511 <!-- _______________________________________________________________________ -->
5512 <div class="doc_subsubsection">
5513 <a name="int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a>
5516 <div class="doc_text">
5520 declare i8 *@llvm.frameaddress(i32 <level>)
5524 <p>The '<tt>llvm.frameaddress</tt>' intrinsic attempts to return the
5525 target-specific frame pointer value for the specified stack frame.</p>
5528 <p>The argument to this intrinsic indicates which function to return the frame
5529 pointer for. Zero indicates the calling function, one indicates its caller,
5530 etc. The argument is <b>required</b> to be a constant integer value.</p>
5533 <p>The '<tt>llvm.frameaddress</tt>' intrinsic either returns a pointer
5534 indicating the frame address of the specified call frame, or zero if it
5535 cannot be identified. The value returned by this intrinsic is likely to be
5536 incorrect or 0 for arguments other than zero, so it should only be used for
5537 debugging purposes.</p>
5539 <p>Note that calling this intrinsic does not prevent function inlining or other
5540 aggressive transformations, so the value returned may not be that of the
5541 obvious source-language caller.</p>
5545 <!-- _______________________________________________________________________ -->
5546 <div class="doc_subsubsection">
5547 <a name="int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a>
5550 <div class="doc_text">
5554 declare i8 *@llvm.stacksave()
5558 <p>The '<tt>llvm.stacksave</tt>' intrinsic is used to remember the current state
5559 of the function stack, for use
5560 with <a href="#int_stackrestore"> <tt>llvm.stackrestore</tt></a>. This is
5561 useful for implementing language features like scoped automatic variable
5562 sized arrays in C99.</p>
5565 <p>This intrinsic returns a opaque pointer value that can be passed
5566 to <a href="#int_stackrestore"><tt>llvm.stackrestore</tt></a>. When
5567 an <tt>llvm.stackrestore</tt> intrinsic is executed with a value saved
5568 from <tt>llvm.stacksave</tt>, it effectively restores the state of the stack
5569 to the state it was in when the <tt>llvm.stacksave</tt> intrinsic executed.
5570 In practice, this pops any <a href="#i_alloca">alloca</a> blocks from the
5571 stack that were allocated after the <tt>llvm.stacksave</tt> was executed.</p>
5575 <!-- _______________________________________________________________________ -->
5576 <div class="doc_subsubsection">
5577 <a name="int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a>
5580 <div class="doc_text">
5584 declare void @llvm.stackrestore(i8 * %ptr)
5588 <p>The '<tt>llvm.stackrestore</tt>' intrinsic is used to restore the state of
5589 the function stack to the state it was in when the
5590 corresponding <a href="#int_stacksave"><tt>llvm.stacksave</tt></a> intrinsic
5591 executed. This is useful for implementing language features like scoped
5592 automatic variable sized arrays in C99.</p>
5595 <p>See the description
5596 for <a href="#int_stacksave"><tt>llvm.stacksave</tt></a>.</p>
5600 <!-- _______________________________________________________________________ -->
5601 <div class="doc_subsubsection">
5602 <a name="int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a>
5605 <div class="doc_text">
5609 declare void @llvm.prefetch(i8* <address>, i32 <rw>, i32 <locality>)
5613 <p>The '<tt>llvm.prefetch</tt>' intrinsic is a hint to the code generator to
5614 insert a prefetch instruction if supported; otherwise, it is a noop.
5615 Prefetches have no effect on the behavior of the program but can change its
5616 performance characteristics.</p>
5619 <p><tt>address</tt> is the address to be prefetched, <tt>rw</tt> is the
5620 specifier determining if the fetch should be for a read (0) or write (1),
5621 and <tt>locality</tt> is a temporal locality specifier ranging from (0) - no
5622 locality, to (3) - extremely local keep in cache. The <tt>rw</tt>
5623 and <tt>locality</tt> arguments must be constant integers.</p>
5626 <p>This intrinsic does not modify the behavior of the program. In particular,
5627 prefetches cannot trap and do not produce a value. On targets that support
5628 this intrinsic, the prefetch can provide hints to the processor cache for
5629 better performance.</p>
5633 <!-- _______________________________________________________________________ -->
5634 <div class="doc_subsubsection">
5635 <a name="int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a>
5638 <div class="doc_text">
5642 declare void @llvm.pcmarker(i32 <id>)
5646 <p>The '<tt>llvm.pcmarker</tt>' intrinsic is a method to export a Program
5647 Counter (PC) in a region of code to simulators and other tools. The method
5648 is target specific, but it is expected that the marker will use exported
5649 symbols to transmit the PC of the marker. The marker makes no guarantees
5650 that it will remain with any specific instruction after optimizations. It is
5651 possible that the presence of a marker will inhibit optimizations. The
5652 intended use is to be inserted after optimizations to allow correlations of
5653 simulation runs.</p>
5656 <p><tt>id</tt> is a numerical id identifying the marker.</p>
5659 <p>This intrinsic does not modify the behavior of the program. Backends that do
5660 not support this intrinisic may ignore it.</p>
5664 <!-- _______________________________________________________________________ -->
5665 <div class="doc_subsubsection">
5666 <a name="int_readcyclecounter">'<tt>llvm.readcyclecounter</tt>' Intrinsic</a>
5669 <div class="doc_text">
5673 declare i64 @llvm.readcyclecounter( )
5677 <p>The '<tt>llvm.readcyclecounter</tt>' intrinsic provides access to the cycle
5678 counter register (or similar low latency, high accuracy clocks) on those
5679 targets that support it. On X86, it should map to RDTSC. On Alpha, it
5680 should map to RPCC. As the backing counters overflow quickly (on the order
5681 of 9 seconds on alpha), this should only be used for small timings.</p>
5684 <p>When directly supported, reading the cycle counter should not modify any
5685 memory. Implementations are allowed to either return a application specific
5686 value or a system wide value. On backends without support, this is lowered
5687 to a constant 0.</p>
5691 <!-- ======================================================================= -->
5692 <div class="doc_subsection">
5693 <a name="int_libc">Standard C Library Intrinsics</a>
5696 <div class="doc_text">
5698 <p>LLVM provides intrinsics for a few important standard C library functions.
5699 These intrinsics allow source-language front-ends to pass information about
5700 the alignment of the pointer arguments to the code generator, providing
5701 opportunity for more efficient code generation.</p>
5705 <!-- _______________________________________________________________________ -->
5706 <div class="doc_subsubsection">
5707 <a name="int_memcpy">'<tt>llvm.memcpy</tt>' Intrinsic</a>
5710 <div class="doc_text">
5713 <p>This is an overloaded intrinsic. You can use <tt>llvm.memcpy</tt> on any
5714 integer bit width. Not all targets support all bit widths however.</p>
5717 declare void @llvm.memcpy.i8(i8 * <dest>, i8 * <src>,
5718 i8 <len>, i32 <align>)
5719 declare void @llvm.memcpy.i16(i8 * <dest>, i8 * <src>,
5720 i16 <len>, i32 <align>)
5721 declare void @llvm.memcpy.i32(i8 * <dest>, i8 * <src>,
5722 i32 <len>, i32 <align>)
5723 declare void @llvm.memcpy.i64(i8 * <dest>, i8 * <src>,
5724 i64 <len>, i32 <align>)
5728 <p>The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the
5729 source location to the destination location.</p>
5731 <p>Note that, unlike the standard libc function, the <tt>llvm.memcpy.*</tt>
5732 intrinsics do not return a value, and takes an extra alignment argument.</p>
5735 <p>The first argument is a pointer to the destination, the second is a pointer
5736 to the source. The third argument is an integer argument specifying the
5737 number of bytes to copy, and the fourth argument is the alignment of the
5738 source and destination locations.</p>
5740 <p>If the call to this intrinisic has an alignment value that is not 0 or 1,
5741 then the caller guarantees that both the source and destination pointers are
5742 aligned to that boundary.</p>
5745 <p>The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the
5746 source location to the destination location, which are not allowed to
5747 overlap. It copies "len" bytes of memory over. If the argument is known to
5748 be aligned to some boundary, this can be specified as the fourth argument,
5749 otherwise it should be set to 0 or 1.</p>
5753 <!-- _______________________________________________________________________ -->
5754 <div class="doc_subsubsection">
5755 <a name="int_memmove">'<tt>llvm.memmove</tt>' Intrinsic</a>
5758 <div class="doc_text">
5761 <p>This is an overloaded intrinsic. You can use llvm.memmove on any integer bit
5762 width. Not all targets support all bit widths however.</p>
5765 declare void @llvm.memmove.i8(i8 * <dest>, i8 * <src>,
5766 i8 <len>, i32 <align>)
5767 declare void @llvm.memmove.i16(i8 * <dest>, i8 * <src>,
5768 i16 <len>, i32 <align>)
5769 declare void @llvm.memmove.i32(i8 * <dest>, i8 * <src>,
5770 i32 <len>, i32 <align>)
5771 declare void @llvm.memmove.i64(i8 * <dest>, i8 * <src>,
5772 i64 <len>, i32 <align>)
5776 <p>The '<tt>llvm.memmove.*</tt>' intrinsics move a block of memory from the
5777 source location to the destination location. It is similar to the
5778 '<tt>llvm.memcpy</tt>' intrinsic but allows the two memory locations to
5781 <p>Note that, unlike the standard libc function, the <tt>llvm.memmove.*</tt>
5782 intrinsics do not return a value, and takes an extra alignment argument.</p>
5785 <p>The first argument is a pointer to the destination, the second is a pointer
5786 to the source. The third argument is an integer argument specifying the
5787 number of bytes to copy, and the fourth argument is the alignment of the
5788 source and destination locations.</p>
5790 <p>If the call to this intrinisic has an alignment value that is not 0 or 1,
5791 then the caller guarantees that the source and destination pointers are
5792 aligned to that boundary.</p>
5795 <p>The '<tt>llvm.memmove.*</tt>' intrinsics copy a block of memory from the
5796 source location to the destination location, which may overlap. It copies
5797 "len" bytes of memory over. If the argument is known to be aligned to some
5798 boundary, this can be specified as the fourth argument, otherwise it should
5799 be set to 0 or 1.</p>
5803 <!-- _______________________________________________________________________ -->
5804 <div class="doc_subsubsection">
5805 <a name="int_memset">'<tt>llvm.memset.*</tt>' Intrinsics</a>
5808 <div class="doc_text">
5811 <p>This is an overloaded intrinsic. You can use llvm.memset on any integer bit
5812 width. Not all targets support all bit widths however.</p>
5815 declare void @llvm.memset.i8(i8 * <dest>, i8 <val>,
5816 i8 <len>, i32 <align>)
5817 declare void @llvm.memset.i16(i8 * <dest>, i8 <val>,
5818 i16 <len>, i32 <align>)
5819 declare void @llvm.memset.i32(i8 * <dest>, i8 <val>,
5820 i32 <len>, i32 <align>)
5821 declare void @llvm.memset.i64(i8 * <dest>, i8 <val>,
5822 i64 <len>, i32 <align>)
5826 <p>The '<tt>llvm.memset.*</tt>' intrinsics fill a block of memory with a
5827 particular byte value.</p>
5829 <p>Note that, unlike the standard libc function, the <tt>llvm.memset</tt>
5830 intrinsic does not return a value, and takes an extra alignment argument.</p>
5833 <p>The first argument is a pointer to the destination to fill, the second is the
5834 byte value to fill it with, the third argument is an integer argument
5835 specifying the number of bytes to fill, and the fourth argument is the known
5836 alignment of destination location.</p>
5838 <p>If the call to this intrinisic has an alignment value that is not 0 or 1,
5839 then the caller guarantees that the destination pointer is aligned to that
5843 <p>The '<tt>llvm.memset.*</tt>' intrinsics fill "len" bytes of memory starting
5844 at the destination location. If the argument is known to be aligned to some
5845 boundary, this can be specified as the fourth argument, otherwise it should
5846 be set to 0 or 1.</p>
5850 <!-- _______________________________________________________________________ -->
5851 <div class="doc_subsubsection">
5852 <a name="int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a>
5855 <div class="doc_text">
5858 <p>This is an overloaded intrinsic. You can use <tt>llvm.sqrt</tt> on any
5859 floating point or vector of floating point type. Not all targets support all
5863 declare float @llvm.sqrt.f32(float %Val)
5864 declare double @llvm.sqrt.f64(double %Val)
5865 declare x86_fp80 @llvm.sqrt.f80(x86_fp80 %Val)
5866 declare fp128 @llvm.sqrt.f128(fp128 %Val)
5867 declare ppc_fp128 @llvm.sqrt.ppcf128(ppc_fp128 %Val)
5871 <p>The '<tt>llvm.sqrt</tt>' intrinsics return the sqrt of the specified operand,
5872 returning the same value as the libm '<tt>sqrt</tt>' functions would.
5873 Unlike <tt>sqrt</tt> in libm, however, <tt>llvm.sqrt</tt> has undefined
5874 behavior for negative numbers other than -0.0 (which allows for better
5875 optimization, because there is no need to worry about errno being
5876 set). <tt>llvm.sqrt(-0.0)</tt> is defined to return -0.0 like IEEE sqrt.</p>
5879 <p>The argument and return value are floating point numbers of the same
5883 <p>This function returns the sqrt of the specified operand if it is a
5884 nonnegative floating point number.</p>
5888 <!-- _______________________________________________________________________ -->
5889 <div class="doc_subsubsection">
5890 <a name="int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a>
5893 <div class="doc_text">
5896 <p>This is an overloaded intrinsic. You can use <tt>llvm.powi</tt> on any
5897 floating point or vector of floating point type. Not all targets support all
5901 declare float @llvm.powi.f32(float %Val, i32 %power)
5902 declare double @llvm.powi.f64(double %Val, i32 %power)
5903 declare x86_fp80 @llvm.powi.f80(x86_fp80 %Val, i32 %power)
5904 declare fp128 @llvm.powi.f128(fp128 %Val, i32 %power)
5905 declare ppc_fp128 @llvm.powi.ppcf128(ppc_fp128 %Val, i32 %power)
5909 <p>The '<tt>llvm.powi.*</tt>' intrinsics return the first operand raised to the
5910 specified (positive or negative) power. The order of evaluation of
5911 multiplications is not defined. When a vector of floating point type is
5912 used, the second argument remains a scalar integer value.</p>
5915 <p>The second argument is an integer power, and the first is a value to raise to
5919 <p>This function returns the first value raised to the second power with an
5920 unspecified sequence of rounding operations.</p>
5924 <!-- _______________________________________________________________________ -->
5925 <div class="doc_subsubsection">
5926 <a name="int_sin">'<tt>llvm.sin.*</tt>' Intrinsic</a>
5929 <div class="doc_text">
5932 <p>This is an overloaded intrinsic. You can use <tt>llvm.sin</tt> on any
5933 floating point or vector of floating point type. Not all targets support all
5937 declare float @llvm.sin.f32(float %Val)
5938 declare double @llvm.sin.f64(double %Val)
5939 declare x86_fp80 @llvm.sin.f80(x86_fp80 %Val)
5940 declare fp128 @llvm.sin.f128(fp128 %Val)
5941 declare ppc_fp128 @llvm.sin.ppcf128(ppc_fp128 %Val)
5945 <p>The '<tt>llvm.sin.*</tt>' intrinsics return the sine of the operand.</p>
5948 <p>The argument and return value are floating point numbers of the same
5952 <p>This function returns the sine of the specified operand, returning the same
5953 values as the libm <tt>sin</tt> functions would, and handles error conditions
5954 in the same way.</p>
5958 <!-- _______________________________________________________________________ -->
5959 <div class="doc_subsubsection">
5960 <a name="int_cos">'<tt>llvm.cos.*</tt>' Intrinsic</a>
5963 <div class="doc_text">
5966 <p>This is an overloaded intrinsic. You can use <tt>llvm.cos</tt> on any
5967 floating point or vector of floating point type. Not all targets support all
5971 declare float @llvm.cos.f32(float %Val)
5972 declare double @llvm.cos.f64(double %Val)
5973 declare x86_fp80 @llvm.cos.f80(x86_fp80 %Val)
5974 declare fp128 @llvm.cos.f128(fp128 %Val)
5975 declare ppc_fp128 @llvm.cos.ppcf128(ppc_fp128 %Val)
5979 <p>The '<tt>llvm.cos.*</tt>' intrinsics return the cosine of the operand.</p>
5982 <p>The argument and return value are floating point numbers of the same
5986 <p>This function returns the cosine of the specified operand, returning the same
5987 values as the libm <tt>cos</tt> functions would, and handles error conditions
5988 in the same way.</p>
5992 <!-- _______________________________________________________________________ -->
5993 <div class="doc_subsubsection">
5994 <a name="int_pow">'<tt>llvm.pow.*</tt>' Intrinsic</a>
5997 <div class="doc_text">
6000 <p>This is an overloaded intrinsic. You can use <tt>llvm.pow</tt> on any
6001 floating point or vector of floating point type. Not all targets support all
6005 declare float @llvm.pow.f32(float %Val, float %Power)
6006 declare double @llvm.pow.f64(double %Val, double %Power)
6007 declare x86_fp80 @llvm.pow.f80(x86_fp80 %Val, x86_fp80 %Power)
6008 declare fp128 @llvm.pow.f128(fp128 %Val, fp128 %Power)
6009 declare ppc_fp128 @llvm.pow.ppcf128(ppc_fp128 %Val, ppc_fp128 Power)
6013 <p>The '<tt>llvm.pow.*</tt>' intrinsics return the first operand raised to the
6014 specified (positive or negative) power.</p>
6017 <p>The second argument is a floating point power, and the first is a value to
6018 raise to that power.</p>
6021 <p>This function returns the first value raised to the second power, returning
6022 the same values as the libm <tt>pow</tt> functions would, and handles error
6023 conditions in the same way.</p>
6027 <!-- ======================================================================= -->
6028 <div class="doc_subsection">
6029 <a name="int_manip">Bit Manipulation Intrinsics</a>
6032 <div class="doc_text">
6034 <p>LLVM provides intrinsics for a few important bit manipulation operations.
6035 These allow efficient code generation for some algorithms.</p>
6039 <!-- _______________________________________________________________________ -->
6040 <div class="doc_subsubsection">
6041 <a name="int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a>
6044 <div class="doc_text">
6047 <p>This is an overloaded intrinsic function. You can use bswap on any integer
6048 type that is an even number of bytes (i.e. BitWidth % 16 == 0).</p>
6051 declare i16 @llvm.bswap.i16(i16 <id>)
6052 declare i32 @llvm.bswap.i32(i32 <id>)
6053 declare i64 @llvm.bswap.i64(i64 <id>)
6057 <p>The '<tt>llvm.bswap</tt>' family of intrinsics is used to byte swap integer
6058 values with an even number of bytes (positive multiple of 16 bits). These
6059 are useful for performing operations on data that is not in the target's
6060 native byte order.</p>
6063 <p>The <tt>llvm.bswap.i16</tt> intrinsic returns an i16 value that has the high
6064 and low byte of the input i16 swapped. Similarly,
6065 the <tt>llvm.bswap.i32</tt> intrinsic returns an i32 value that has the four
6066 bytes of the input i32 swapped, so that if the input bytes are numbered 0, 1,
6067 2, 3 then the returned i32 will have its bytes in 3, 2, 1, 0 order.
6068 The <tt>llvm.bswap.i48</tt>, <tt>llvm.bswap.i64</tt> and other intrinsics
6069 extend this concept to additional even-byte lengths (6 bytes, 8 bytes and
6070 more, respectively).</p>
6074 <!-- _______________________________________________________________________ -->
6075 <div class="doc_subsubsection">
6076 <a name="int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic</a>
6079 <div class="doc_text">
6082 <p>This is an overloaded intrinsic. You can use llvm.ctpop on any integer bit
6083 width. Not all targets support all bit widths however.</p>
6086 declare i8 @llvm.ctpop.i8(i8 <src>)
6087 declare i16 @llvm.ctpop.i16(i16 <src>)
6088 declare i32 @llvm.ctpop.i32(i32 <src>)
6089 declare i64 @llvm.ctpop.i64(i64 <src>)
6090 declare i256 @llvm.ctpop.i256(i256 <src>)
6094 <p>The '<tt>llvm.ctpop</tt>' family of intrinsics counts the number of bits set
6098 <p>The only argument is the value to be counted. The argument may be of any
6099 integer type. The return type must match the argument type.</p>
6102 <p>The '<tt>llvm.ctpop</tt>' intrinsic counts the 1's in a variable.</p>
6106 <!-- _______________________________________________________________________ -->
6107 <div class="doc_subsubsection">
6108 <a name="int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic</a>
6111 <div class="doc_text">
6114 <p>This is an overloaded intrinsic. You can use <tt>llvm.ctlz</tt> on any
6115 integer bit width. Not all targets support all bit widths however.</p>
6118 declare i8 @llvm.ctlz.i8 (i8 <src>)
6119 declare i16 @llvm.ctlz.i16(i16 <src>)
6120 declare i32 @llvm.ctlz.i32(i32 <src>)
6121 declare i64 @llvm.ctlz.i64(i64 <src>)
6122 declare i256 @llvm.ctlz.i256(i256 <src>)
6126 <p>The '<tt>llvm.ctlz</tt>' family of intrinsic functions counts the number of
6127 leading zeros in a variable.</p>
6130 <p>The only argument is the value to be counted. The argument may be of any
6131 integer type. The return type must match the argument type.</p>
6134 <p>The '<tt>llvm.ctlz</tt>' intrinsic counts the leading (most significant)
6135 zeros in a variable. If the src == 0 then the result is the size in bits of
6136 the type of src. For example, <tt>llvm.ctlz(i32 2) = 30</tt>.</p>
6140 <!-- _______________________________________________________________________ -->
6141 <div class="doc_subsubsection">
6142 <a name="int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic</a>
6145 <div class="doc_text">
6148 <p>This is an overloaded intrinsic. You can use <tt>llvm.cttz</tt> on any
6149 integer bit width. Not all targets support all bit widths however.</p>
6152 declare i8 @llvm.cttz.i8 (i8 <src>)
6153 declare i16 @llvm.cttz.i16(i16 <src>)
6154 declare i32 @llvm.cttz.i32(i32 <src>)
6155 declare i64 @llvm.cttz.i64(i64 <src>)
6156 declare i256 @llvm.cttz.i256(i256 <src>)
6160 <p>The '<tt>llvm.cttz</tt>' family of intrinsic functions counts the number of
6164 <p>The only argument is the value to be counted. The argument may be of any
6165 integer type. The return type must match the argument type.</p>
6168 <p>The '<tt>llvm.cttz</tt>' intrinsic counts the trailing (least significant)
6169 zeros in a variable. If the src == 0 then the result is the size in bits of
6170 the type of src. For example, <tt>llvm.cttz(2) = 1</tt>.</p>
6174 <!-- ======================================================================= -->
6175 <div class="doc_subsection">
6176 <a name="int_overflow">Arithmetic with Overflow Intrinsics</a>
6179 <div class="doc_text">
6181 <p>LLVM provides intrinsics for some arithmetic with overflow operations.</p>
6185 <!-- _______________________________________________________________________ -->
6186 <div class="doc_subsubsection">
6187 <a name="int_sadd_overflow">'<tt>llvm.sadd.with.overflow.*</tt>' Intrinsics</a>
6190 <div class="doc_text">
6193 <p>This is an overloaded intrinsic. You can use <tt>llvm.sadd.with.overflow</tt>
6194 on any integer bit width.</p>
6197 declare {i16, i1} @llvm.sadd.with.overflow.i16(i16 %a, i16 %b)
6198 declare {i32, i1} @llvm.sadd.with.overflow.i32(i32 %a, i32 %b)
6199 declare {i64, i1} @llvm.sadd.with.overflow.i64(i64 %a, i64 %b)
6203 <p>The '<tt>llvm.sadd.with.overflow</tt>' family of intrinsic functions perform
6204 a signed addition of the two arguments, and indicate whether an overflow
6205 occurred during the signed summation.</p>
6208 <p>The arguments (%a and %b) and the first element of the result structure may
6209 be of integer types of any bit width, but they must have the same bit
6210 width. The second element of the result structure must be of
6211 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
6212 undergo signed addition.</p>
6215 <p>The '<tt>llvm.sadd.with.overflow</tt>' family of intrinsic functions perform
6216 a signed addition of the two variables. They return a structure — the
6217 first element of which is the signed summation, and the second element of
6218 which is a bit specifying if the signed summation resulted in an
6223 %res = call {i32, i1} @llvm.sadd.with.overflow.i32(i32 %a, i32 %b)
6224 %sum = extractvalue {i32, i1} %res, 0
6225 %obit = extractvalue {i32, i1} %res, 1
6226 br i1 %obit, label %overflow, label %normal
6231 <!-- _______________________________________________________________________ -->
6232 <div class="doc_subsubsection">
6233 <a name="int_uadd_overflow">'<tt>llvm.uadd.with.overflow.*</tt>' Intrinsics</a>
6236 <div class="doc_text">
6239 <p>This is an overloaded intrinsic. You can use <tt>llvm.uadd.with.overflow</tt>
6240 on any integer bit width.</p>
6243 declare {i16, i1} @llvm.uadd.with.overflow.i16(i16 %a, i16 %b)
6244 declare {i32, i1} @llvm.uadd.with.overflow.i32(i32 %a, i32 %b)
6245 declare {i64, i1} @llvm.uadd.with.overflow.i64(i64 %a, i64 %b)
6249 <p>The '<tt>llvm.uadd.with.overflow</tt>' family of intrinsic functions perform
6250 an unsigned addition of the two arguments, and indicate whether a carry
6251 occurred during the unsigned summation.</p>
6254 <p>The arguments (%a and %b) and the first element of the result structure may
6255 be of integer types of any bit width, but they must have the same bit
6256 width. The second element of the result structure must be of
6257 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
6258 undergo unsigned addition.</p>
6261 <p>The '<tt>llvm.uadd.with.overflow</tt>' family of intrinsic functions perform
6262 an unsigned addition of the two arguments. They return a structure —
6263 the first element of which is the sum, and the second element of which is a
6264 bit specifying if the unsigned summation resulted in a carry.</p>
6268 %res = call {i32, i1} @llvm.uadd.with.overflow.i32(i32 %a, i32 %b)
6269 %sum = extractvalue {i32, i1} %res, 0
6270 %obit = extractvalue {i32, i1} %res, 1
6271 br i1 %obit, label %carry, label %normal
6276 <!-- _______________________________________________________________________ -->
6277 <div class="doc_subsubsection">
6278 <a name="int_ssub_overflow">'<tt>llvm.ssub.with.overflow.*</tt>' Intrinsics</a>
6281 <div class="doc_text">
6284 <p>This is an overloaded intrinsic. You can use <tt>llvm.ssub.with.overflow</tt>
6285 on any integer bit width.</p>
6288 declare {i16, i1} @llvm.ssub.with.overflow.i16(i16 %a, i16 %b)
6289 declare {i32, i1} @llvm.ssub.with.overflow.i32(i32 %a, i32 %b)
6290 declare {i64, i1} @llvm.ssub.with.overflow.i64(i64 %a, i64 %b)
6294 <p>The '<tt>llvm.ssub.with.overflow</tt>' family of intrinsic functions perform
6295 a signed subtraction of the two arguments, and indicate whether an overflow
6296 occurred during the signed subtraction.</p>
6299 <p>The arguments (%a and %b) and the first element of the result structure may
6300 be of integer types of any bit width, but they must have the same bit
6301 width. The second element of the result structure must be of
6302 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
6303 undergo signed subtraction.</p>
6306 <p>The '<tt>llvm.ssub.with.overflow</tt>' family of intrinsic functions perform
6307 a signed subtraction of the two arguments. They return a structure —
6308 the first element of which is the subtraction, and the second element of
6309 which is a bit specifying if the signed subtraction resulted in an
6314 %res = call {i32, i1} @llvm.ssub.with.overflow.i32(i32 %a, i32 %b)
6315 %sum = extractvalue {i32, i1} %res, 0
6316 %obit = extractvalue {i32, i1} %res, 1
6317 br i1 %obit, label %overflow, label %normal
6322 <!-- _______________________________________________________________________ -->
6323 <div class="doc_subsubsection">
6324 <a name="int_usub_overflow">'<tt>llvm.usub.with.overflow.*</tt>' Intrinsics</a>
6327 <div class="doc_text">
6330 <p>This is an overloaded intrinsic. You can use <tt>llvm.usub.with.overflow</tt>
6331 on any integer bit width.</p>
6334 declare {i16, i1} @llvm.usub.with.overflow.i16(i16 %a, i16 %b)
6335 declare {i32, i1} @llvm.usub.with.overflow.i32(i32 %a, i32 %b)
6336 declare {i64, i1} @llvm.usub.with.overflow.i64(i64 %a, i64 %b)
6340 <p>The '<tt>llvm.usub.with.overflow</tt>' family of intrinsic functions perform
6341 an unsigned subtraction of the two arguments, and indicate whether an
6342 overflow occurred during the unsigned subtraction.</p>
6345 <p>The arguments (%a and %b) and the first element of the result structure may
6346 be of integer types of any bit width, but they must have the same bit
6347 width. The second element of the result structure must be of
6348 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
6349 undergo unsigned subtraction.</p>
6352 <p>The '<tt>llvm.usub.with.overflow</tt>' family of intrinsic functions perform
6353 an unsigned subtraction of the two arguments. They return a structure —
6354 the first element of which is the subtraction, and the second element of
6355 which is a bit specifying if the unsigned subtraction resulted in an
6360 %res = call {i32, i1} @llvm.usub.with.overflow.i32(i32 %a, i32 %b)
6361 %sum = extractvalue {i32, i1} %res, 0
6362 %obit = extractvalue {i32, i1} %res, 1
6363 br i1 %obit, label %overflow, label %normal
6368 <!-- _______________________________________________________________________ -->
6369 <div class="doc_subsubsection">
6370 <a name="int_smul_overflow">'<tt>llvm.smul.with.overflow.*</tt>' Intrinsics</a>
6373 <div class="doc_text">
6376 <p>This is an overloaded intrinsic. You can use <tt>llvm.smul.with.overflow</tt>
6377 on any integer bit width.</p>
6380 declare {i16, i1} @llvm.smul.with.overflow.i16(i16 %a, i16 %b)
6381 declare {i32, i1} @llvm.smul.with.overflow.i32(i32 %a, i32 %b)
6382 declare {i64, i1} @llvm.smul.with.overflow.i64(i64 %a, i64 %b)
6387 <p>The '<tt>llvm.smul.with.overflow</tt>' family of intrinsic functions perform
6388 a signed multiplication of the two arguments, and indicate whether an
6389 overflow occurred during the signed multiplication.</p>
6392 <p>The arguments (%a and %b) and the first element of the result structure may
6393 be of integer types of any bit width, but they must have the same bit
6394 width. The second element of the result structure must be of
6395 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
6396 undergo signed multiplication.</p>
6399 <p>The '<tt>llvm.smul.with.overflow</tt>' family of intrinsic functions perform
6400 a signed multiplication of the two arguments. They return a structure —
6401 the first element of which is the multiplication, and the second element of
6402 which is a bit specifying if the signed multiplication resulted in an
6407 %res = call {i32, i1} @llvm.smul.with.overflow.i32(i32 %a, i32 %b)
6408 %sum = extractvalue {i32, i1} %res, 0
6409 %obit = extractvalue {i32, i1} %res, 1
6410 br i1 %obit, label %overflow, label %normal
6415 <!-- _______________________________________________________________________ -->
6416 <div class="doc_subsubsection">
6417 <a name="int_umul_overflow">'<tt>llvm.umul.with.overflow.*</tt>' Intrinsics</a>
6420 <div class="doc_text">
6423 <p>This is an overloaded intrinsic. You can use <tt>llvm.umul.with.overflow</tt>
6424 on any integer bit width.</p>
6427 declare {i16, i1} @llvm.umul.with.overflow.i16(i16 %a, i16 %b)
6428 declare {i32, i1} @llvm.umul.with.overflow.i32(i32 %a, i32 %b)
6429 declare {i64, i1} @llvm.umul.with.overflow.i64(i64 %a, i64 %b)
6433 <p>The '<tt>llvm.umul.with.overflow</tt>' family of intrinsic functions perform
6434 a unsigned multiplication of the two arguments, and indicate whether an
6435 overflow occurred during the unsigned multiplication.</p>
6438 <p>The arguments (%a and %b) and the first element of the result structure may
6439 be of integer types of any bit width, but they must have the same bit
6440 width. The second element of the result structure must be of
6441 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
6442 undergo unsigned multiplication.</p>
6445 <p>The '<tt>llvm.umul.with.overflow</tt>' family of intrinsic functions perform
6446 an unsigned multiplication of the two arguments. They return a structure
6447 — the first element of which is the multiplication, and the second
6448 element of which is a bit specifying if the unsigned multiplication resulted
6453 %res = call {i32, i1} @llvm.umul.with.overflow.i32(i32 %a, i32 %b)
6454 %sum = extractvalue {i32, i1} %res, 0
6455 %obit = extractvalue {i32, i1} %res, 1
6456 br i1 %obit, label %overflow, label %normal
6461 <!-- ======================================================================= -->
6462 <div class="doc_subsection">
6463 <a name="int_debugger">Debugger Intrinsics</a>
6466 <div class="doc_text">
6468 <p>The LLVM debugger intrinsics (which all start with <tt>llvm.dbg.</tt>
6469 prefix), are described in
6470 the <a href="SourceLevelDebugging.html#format_common_intrinsics">LLVM Source
6471 Level Debugging</a> document.</p>
6475 <!-- ======================================================================= -->
6476 <div class="doc_subsection">
6477 <a name="int_eh">Exception Handling Intrinsics</a>
6480 <div class="doc_text">
6482 <p>The LLVM exception handling intrinsics (which all start with
6483 <tt>llvm.eh.</tt> prefix), are described in
6484 the <a href="ExceptionHandling.html#format_common_intrinsics">LLVM Exception
6485 Handling</a> document.</p>
6489 <!-- ======================================================================= -->
6490 <div class="doc_subsection">
6491 <a name="int_trampoline">Trampoline Intrinsic</a>
6494 <div class="doc_text">
6496 <p>This intrinsic makes it possible to excise one parameter, marked with
6497 the <tt>nest</tt> attribute, from a function. The result is a callable
6498 function pointer lacking the nest parameter - the caller does not need to
6499 provide a value for it. Instead, the value to use is stored in advance in a
6500 "trampoline", a block of memory usually allocated on the stack, which also
6501 contains code to splice the nest value into the argument list. This is used
6502 to implement the GCC nested function address extension.</p>
6504 <p>For example, if the function is
6505 <tt>i32 f(i8* nest %c, i32 %x, i32 %y)</tt> then the resulting function
6506 pointer has signature <tt>i32 (i32, i32)*</tt>. It can be created as
6509 <div class="doc_code">
6511 %tramp = alloca [10 x i8], align 4 ; size and alignment only correct for X86
6512 %tramp1 = getelementptr [10 x i8]* %tramp, i32 0, i32 0
6513 %p = call i8* @llvm.init.trampoline( i8* %tramp1, i8* bitcast (i32 (i8* nest , i32, i32)* @f to i8*), i8* %nval )
6514 %fp = bitcast i8* %p to i32 (i32, i32)*
6518 <p>The call <tt>%val = call i32 %fp( i32 %x, i32 %y )</tt> is then equivalent
6519 to <tt>%val = call i32 %f( i8* %nval, i32 %x, i32 %y )</tt>.</p>
6523 <!-- _______________________________________________________________________ -->
6524 <div class="doc_subsubsection">
6525 <a name="int_it">'<tt>llvm.init.trampoline</tt>' Intrinsic</a>
6528 <div class="doc_text">
6532 declare i8* @llvm.init.trampoline(i8* <tramp>, i8* <func>, i8* <nval>)
6536 <p>This fills the memory pointed to by <tt>tramp</tt> with code and returns a
6537 function pointer suitable for executing it.</p>
6540 <p>The <tt>llvm.init.trampoline</tt> intrinsic takes three arguments, all
6541 pointers. The <tt>tramp</tt> argument must point to a sufficiently large and
6542 sufficiently aligned block of memory; this memory is written to by the
6543 intrinsic. Note that the size and the alignment are target-specific - LLVM
6544 currently provides no portable way of determining them, so a front-end that
6545 generates this intrinsic needs to have some target-specific knowledge.
6546 The <tt>func</tt> argument must hold a function bitcast to
6547 an <tt>i8*</tt>.</p>
6550 <p>The block of memory pointed to by <tt>tramp</tt> is filled with target
6551 dependent code, turning it into a function. A pointer to this function is
6552 returned, but needs to be bitcast to an <a href="#int_trampoline">appropriate
6553 function pointer type</a> before being called. The new function's signature
6554 is the same as that of <tt>func</tt> with any arguments marked with
6555 the <tt>nest</tt> attribute removed. At most one such <tt>nest</tt> argument
6556 is allowed, and it must be of pointer type. Calling the new function is
6557 equivalent to calling <tt>func</tt> with the same argument list, but
6558 with <tt>nval</tt> used for the missing <tt>nest</tt> argument. If, after
6559 calling <tt>llvm.init.trampoline</tt>, the memory pointed to
6560 by <tt>tramp</tt> is modified, then the effect of any later call to the
6561 returned function pointer is undefined.</p>
6565 <!-- ======================================================================= -->
6566 <div class="doc_subsection">
6567 <a name="int_atomics">Atomic Operations and Synchronization Intrinsics</a>
6570 <div class="doc_text">
6572 <p>These intrinsic functions expand the "universal IR" of LLVM to represent
6573 hardware constructs for atomic operations and memory synchronization. This
6574 provides an interface to the hardware, not an interface to the programmer. It
6575 is aimed at a low enough level to allow any programming models or APIs
6576 (Application Programming Interfaces) which need atomic behaviors to map
6577 cleanly onto it. It is also modeled primarily on hardware behavior. Just as
6578 hardware provides a "universal IR" for source languages, it also provides a
6579 starting point for developing a "universal" atomic operation and
6580 synchronization IR.</p>
6582 <p>These do <em>not</em> form an API such as high-level threading libraries,
6583 software transaction memory systems, atomic primitives, and intrinsic
6584 functions as found in BSD, GNU libc, atomic_ops, APR, and other system and
6585 application libraries. The hardware interface provided by LLVM should allow
6586 a clean implementation of all of these APIs and parallel programming models.
6587 No one model or paradigm should be selected above others unless the hardware
6588 itself ubiquitously does so.</p>
6592 <!-- _______________________________________________________________________ -->
6593 <div class="doc_subsubsection">
6594 <a name="int_memory_barrier">'<tt>llvm.memory.barrier</tt>' Intrinsic</a>
6596 <div class="doc_text">
6599 declare void @llvm.memory.barrier( i1 <ll>, i1 <ls>, i1 <sl>, i1 <ss>, i1 <device> )
6603 <p>The <tt>llvm.memory.barrier</tt> intrinsic guarantees ordering between
6604 specific pairs of memory access types.</p>
6607 <p>The <tt>llvm.memory.barrier</tt> intrinsic requires five boolean arguments.
6608 The first four arguments enables a specific barrier as listed below. The
6609 fith argument specifies that the barrier applies to io or device or uncached
6613 <li><tt>ll</tt>: load-load barrier</li>
6614 <li><tt>ls</tt>: load-store barrier</li>
6615 <li><tt>sl</tt>: store-load barrier</li>
6616 <li><tt>ss</tt>: store-store barrier</li>
6617 <li><tt>device</tt>: barrier applies to device and uncached memory also.</li>
6621 <p>This intrinsic causes the system to enforce some ordering constraints upon
6622 the loads and stores of the program. This barrier does not
6623 indicate <em>when</em> any events will occur, it only enforces
6624 an <em>order</em> in which they occur. For any of the specified pairs of load
6625 and store operations (f.ex. load-load, or store-load), all of the first
6626 operations preceding the barrier will complete before any of the second
6627 operations succeeding the barrier begin. Specifically the semantics for each
6628 pairing is as follows:</p>
6631 <li><tt>ll</tt>: All loads before the barrier must complete before any load
6632 after the barrier begins.</li>
6633 <li><tt>ls</tt>: All loads before the barrier must complete before any
6634 store after the barrier begins.</li>
6635 <li><tt>ss</tt>: All stores before the barrier must complete before any
6636 store after the barrier begins.</li>
6637 <li><tt>sl</tt>: All stores before the barrier must complete before any
6638 load after the barrier begins.</li>
6641 <p>These semantics are applied with a logical "and" behavior when more than one
6642 is enabled in a single memory barrier intrinsic.</p>
6644 <p>Backends may implement stronger barriers than those requested when they do
6645 not support as fine grained a barrier as requested. Some architectures do
6646 not need all types of barriers and on such architectures, these become
6651 %mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
6652 %ptr = bitcast i8* %mallocP to i32*
6655 %result1 = load i32* %ptr <i>; yields {i32}:result1 = 4</i>
6656 call void @llvm.memory.barrier( i1 false, i1 true, i1 false, i1 false )
6657 <i>; guarantee the above finishes</i>
6658 store i32 8, %ptr <i>; before this begins</i>
6663 <!-- _______________________________________________________________________ -->
6664 <div class="doc_subsubsection">
6665 <a name="int_atomic_cmp_swap">'<tt>llvm.atomic.cmp.swap.*</tt>' Intrinsic</a>
6668 <div class="doc_text">
6671 <p>This is an overloaded intrinsic. You can use <tt>llvm.atomic.cmp.swap</tt> on
6672 any integer bit width and for different address spaces. Not all targets
6673 support all bit widths however.</p>
6676 declare i8 @llvm.atomic.cmp.swap.i8.p0i8( i8* <ptr>, i8 <cmp>, i8 <val> )
6677 declare i16 @llvm.atomic.cmp.swap.i16.p0i16( i16* <ptr>, i16 <cmp>, i16 <val> )
6678 declare i32 @llvm.atomic.cmp.swap.i32.p0i32( i32* <ptr>, i32 <cmp>, i32 <val> )
6679 declare i64 @llvm.atomic.cmp.swap.i64.p0i64( i64* <ptr>, i64 <cmp>, i64 <val> )
6683 <p>This loads a value in memory and compares it to a given value. If they are
6684 equal, it stores a new value into the memory.</p>
6687 <p>The <tt>llvm.atomic.cmp.swap</tt> intrinsic takes three arguments. The result
6688 as well as both <tt>cmp</tt> and <tt>val</tt> must be integer values with the
6689 same bit width. The <tt>ptr</tt> argument must be a pointer to a value of
6690 this integer type. While any bit width integer may be used, targets may only
6691 lower representations they support in hardware.</p>
6694 <p>This entire intrinsic must be executed atomically. It first loads the value
6695 in memory pointed to by <tt>ptr</tt> and compares it with the
6696 value <tt>cmp</tt>. If they are equal, <tt>val</tt> is stored into the
6697 memory. The loaded value is yielded in all cases. This provides the
6698 equivalent of an atomic compare-and-swap operation within the SSA
6703 %mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
6704 %ptr = bitcast i8* %mallocP to i32*
6707 %val1 = add i32 4, 4
6708 %result1 = call i32 @llvm.atomic.cmp.swap.i32.p0i32( i32* %ptr, i32 4, %val1 )
6709 <i>; yields {i32}:result1 = 4</i>
6710 %stored1 = icmp eq i32 %result1, 4 <i>; yields {i1}:stored1 = true</i>
6711 %memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 8</i>
6713 %val2 = add i32 1, 1
6714 %result2 = call i32 @llvm.atomic.cmp.swap.i32.p0i32( i32* %ptr, i32 5, %val2 )
6715 <i>; yields {i32}:result2 = 8</i>
6716 %stored2 = icmp eq i32 %result2, 5 <i>; yields {i1}:stored2 = false</i>
6718 %memval2 = load i32* %ptr <i>; yields {i32}:memval2 = 8</i>
6723 <!-- _______________________________________________________________________ -->
6724 <div class="doc_subsubsection">
6725 <a name="int_atomic_swap">'<tt>llvm.atomic.swap.*</tt>' Intrinsic</a>
6727 <div class="doc_text">
6730 <p>This is an overloaded intrinsic. You can use <tt>llvm.atomic.swap</tt> on any
6731 integer bit width. Not all targets support all bit widths however.</p>
6734 declare i8 @llvm.atomic.swap.i8.p0i8( i8* <ptr>, i8 <val> )
6735 declare i16 @llvm.atomic.swap.i16.p0i16( i16* <ptr>, i16 <val> )
6736 declare i32 @llvm.atomic.swap.i32.p0i32( i32* <ptr>, i32 <val> )
6737 declare i64 @llvm.atomic.swap.i64.p0i64( i64* <ptr>, i64 <val> )
6741 <p>This intrinsic loads the value stored in memory at <tt>ptr</tt> and yields
6742 the value from memory. It then stores the value in <tt>val</tt> in the memory
6743 at <tt>ptr</tt>.</p>
6746 <p>The <tt>llvm.atomic.swap</tt> intrinsic takes two arguments. Both
6747 the <tt>val</tt> argument and the result must be integers of the same bit
6748 width. The first argument, <tt>ptr</tt>, must be a pointer to a value of this
6749 integer type. The targets may only lower integer representations they
6753 <p>This intrinsic loads the value pointed to by <tt>ptr</tt>, yields it, and
6754 stores <tt>val</tt> back into <tt>ptr</tt> atomically. This provides the
6755 equivalent of an atomic swap operation within the SSA framework.</p>
6759 %mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
6760 %ptr = bitcast i8* %mallocP to i32*
6763 %val1 = add i32 4, 4
6764 %result1 = call i32 @llvm.atomic.swap.i32.p0i32( i32* %ptr, i32 %val1 )
6765 <i>; yields {i32}:result1 = 4</i>
6766 %stored1 = icmp eq i32 %result1, 4 <i>; yields {i1}:stored1 = true</i>
6767 %memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 8</i>
6769 %val2 = add i32 1, 1
6770 %result2 = call i32 @llvm.atomic.swap.i32.p0i32( i32* %ptr, i32 %val2 )
6771 <i>; yields {i32}:result2 = 8</i>
6773 %stored2 = icmp eq i32 %result2, 8 <i>; yields {i1}:stored2 = true</i>
6774 %memval2 = load i32* %ptr <i>; yields {i32}:memval2 = 2</i>
6779 <!-- _______________________________________________________________________ -->
6780 <div class="doc_subsubsection">
6781 <a name="int_atomic_load_add">'<tt>llvm.atomic.load.add.*</tt>' Intrinsic</a>
6785 <div class="doc_text">
6788 <p>This is an overloaded intrinsic. You can use <tt>llvm.atomic.load.add</tt> on
6789 any integer bit width. Not all targets support all bit widths however.</p>
6792 declare i8 @llvm.atomic.load.add.i8..p0i8( i8* <ptr>, i8 <delta> )
6793 declare i16 @llvm.atomic.load.add.i16..p0i16( i16* <ptr>, i16 <delta> )
6794 declare i32 @llvm.atomic.load.add.i32..p0i32( i32* <ptr>, i32 <delta> )
6795 declare i64 @llvm.atomic.load.add.i64..p0i64( i64* <ptr>, i64 <delta> )
6799 <p>This intrinsic adds <tt>delta</tt> to the value stored in memory
6800 at <tt>ptr</tt>. It yields the original value at <tt>ptr</tt>.</p>
6803 <p>The intrinsic takes two arguments, the first a pointer to an integer value
6804 and the second an integer value. The result is also an integer value. These
6805 integer types can have any bit width, but they must all have the same bit
6806 width. The targets may only lower integer representations they support.</p>
6809 <p>This intrinsic does a series of operations atomically. It first loads the
6810 value stored at <tt>ptr</tt>. It then adds <tt>delta</tt>, stores the result
6811 to <tt>ptr</tt>. It yields the original value stored 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.add.i32.p0i32( i32* %ptr, i32 4 )
6819 <i>; yields {i32}:result1 = 4</i>
6820 %result2 = call i32 @llvm.atomic.load.add.i32.p0i32( i32* %ptr, i32 2 )
6821 <i>; yields {i32}:result2 = 8</i>
6822 %result3 = call i32 @llvm.atomic.load.add.i32.p0i32( i32* %ptr, i32 5 )
6823 <i>; yields {i32}:result3 = 10</i>
6824 %memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 15</i>
6829 <!-- _______________________________________________________________________ -->
6830 <div class="doc_subsubsection">
6831 <a name="int_atomic_load_sub">'<tt>llvm.atomic.load.sub.*</tt>' Intrinsic</a>
6835 <div class="doc_text">
6838 <p>This is an overloaded intrinsic. You can use <tt>llvm.atomic.load.sub</tt> on
6839 any integer bit width and for different address spaces. Not all targets
6840 support all bit widths however.</p>
6843 declare i8 @llvm.atomic.load.sub.i8.p0i32( i8* <ptr>, i8 <delta> )
6844 declare i16 @llvm.atomic.load.sub.i16.p0i32( i16* <ptr>, i16 <delta> )
6845 declare i32 @llvm.atomic.load.sub.i32.p0i32( i32* <ptr>, i32 <delta> )
6846 declare i64 @llvm.atomic.load.sub.i64.p0i32( i64* <ptr>, i64 <delta> )
6850 <p>This intrinsic subtracts <tt>delta</tt> to the value stored in memory at
6851 <tt>ptr</tt>. It yields the original value at <tt>ptr</tt>.</p>
6854 <p>The intrinsic takes two arguments, the first a pointer to an integer value
6855 and the second an integer value. The result is also an integer value. These
6856 integer types can have any bit width, but they must all have the same bit
6857 width. The targets may only lower integer representations they support.</p>
6860 <p>This intrinsic does a series of operations atomically. It first loads the
6861 value stored at <tt>ptr</tt>. It then subtracts <tt>delta</tt>, stores the
6862 result to <tt>ptr</tt>. It yields the original value stored
6863 at <tt>ptr</tt>.</p>
6867 %mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
6868 %ptr = bitcast i8* %mallocP to i32*
6870 %result1 = call i32 @llvm.atomic.load.sub.i32.p0i32( i32* %ptr, i32 4 )
6871 <i>; yields {i32}:result1 = 8</i>
6872 %result2 = call i32 @llvm.atomic.load.sub.i32.p0i32( i32* %ptr, i32 2 )
6873 <i>; yields {i32}:result2 = 4</i>
6874 %result3 = call i32 @llvm.atomic.load.sub.i32.p0i32( i32* %ptr, i32 5 )
6875 <i>; yields {i32}:result3 = 2</i>
6876 %memval1 = load i32* %ptr <i>; yields {i32}:memval1 = -3</i>
6881 <!-- _______________________________________________________________________ -->
6882 <div class="doc_subsubsection">
6883 <a name="int_atomic_load_and">'<tt>llvm.atomic.load.and.*</tt>' Intrinsic</a><br>
6884 <a name="int_atomic_load_nand">'<tt>llvm.atomic.load.nand.*</tt>' Intrinsic</a><br>
6885 <a name="int_atomic_load_or">'<tt>llvm.atomic.load.or.*</tt>' Intrinsic</a><br>
6886 <a name="int_atomic_load_xor">'<tt>llvm.atomic.load.xor.*</tt>' Intrinsic</a><br>
6889 <div class="doc_text">
6892 <p>These are overloaded intrinsics. You can
6893 use <tt>llvm.atomic.load_and</tt>, <tt>llvm.atomic.load_nand</tt>,
6894 <tt>llvm.atomic.load_or</tt>, and <tt>llvm.atomic.load_xor</tt> on any integer
6895 bit width and for different address spaces. Not all targets support all bit
6899 declare i8 @llvm.atomic.load.and.i8.p0i8( i8* <ptr>, i8 <delta> )
6900 declare i16 @llvm.atomic.load.and.i16.p0i16( i16* <ptr>, i16 <delta> )
6901 declare i32 @llvm.atomic.load.and.i32.p0i32( i32* <ptr>, i32 <delta> )
6902 declare i64 @llvm.atomic.load.and.i64.p0i64( i64* <ptr>, i64 <delta> )
6906 declare i8 @llvm.atomic.load.or.i8.p0i8( i8* <ptr>, i8 <delta> )
6907 declare i16 @llvm.atomic.load.or.i16.p0i16( i16* <ptr>, i16 <delta> )
6908 declare i32 @llvm.atomic.load.or.i32.p0i32( i32* <ptr>, i32 <delta> )
6909 declare i64 @llvm.atomic.load.or.i64.p0i64( i64* <ptr>, i64 <delta> )
6913 declare i8 @llvm.atomic.load.nand.i8.p0i32( i8* <ptr>, i8 <delta> )
6914 declare i16 @llvm.atomic.load.nand.i16.p0i32( i16* <ptr>, i16 <delta> )
6915 declare i32 @llvm.atomic.load.nand.i32.p0i32( i32* <ptr>, i32 <delta> )
6916 declare i64 @llvm.atomic.load.nand.i64.p0i32( i64* <ptr>, i64 <delta> )
6920 declare i8 @llvm.atomic.load.xor.i8.p0i32( i8* <ptr>, i8 <delta> )
6921 declare i16 @llvm.atomic.load.xor.i16.p0i32( i16* <ptr>, i16 <delta> )
6922 declare i32 @llvm.atomic.load.xor.i32.p0i32( i32* <ptr>, i32 <delta> )
6923 declare i64 @llvm.atomic.load.xor.i64.p0i32( i64* <ptr>, i64 <delta> )
6927 <p>These intrinsics bitwise the operation (and, nand, or, xor) <tt>delta</tt> to
6928 the value stored in memory at <tt>ptr</tt>. It yields the original value
6929 at <tt>ptr</tt>.</p>
6932 <p>These intrinsics take two arguments, the first a pointer to an integer value
6933 and the second an integer value. The result is also an integer value. These
6934 integer types can have any bit width, but they must all have the same bit
6935 width. The targets may only lower integer representations they support.</p>
6938 <p>These intrinsics does a series of operations atomically. They first load the
6939 value stored at <tt>ptr</tt>. They then do the bitwise
6940 operation <tt>delta</tt>, store the result to <tt>ptr</tt>. They yield the
6941 original value stored at <tt>ptr</tt>.</p>
6945 %mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
6946 %ptr = bitcast i8* %mallocP to i32*
6947 store i32 0x0F0F, %ptr
6948 %result0 = call i32 @llvm.atomic.load.nand.i32.p0i32( i32* %ptr, i32 0xFF )
6949 <i>; yields {i32}:result0 = 0x0F0F</i>
6950 %result1 = call i32 @llvm.atomic.load.and.i32.p0i32( i32* %ptr, i32 0xFF )
6951 <i>; yields {i32}:result1 = 0xFFFFFFF0</i>
6952 %result2 = call i32 @llvm.atomic.load.or.i32.p0i32( i32* %ptr, i32 0F )
6953 <i>; yields {i32}:result2 = 0xF0</i>
6954 %result3 = call i32 @llvm.atomic.load.xor.i32.p0i32( i32* %ptr, i32 0F )
6955 <i>; yields {i32}:result3 = FF</i>
6956 %memval1 = load i32* %ptr <i>; yields {i32}:memval1 = F0</i>
6961 <!-- _______________________________________________________________________ -->
6962 <div class="doc_subsubsection">
6963 <a name="int_atomic_load_max">'<tt>llvm.atomic.load.max.*</tt>' Intrinsic</a><br>
6964 <a name="int_atomic_load_min">'<tt>llvm.atomic.load.min.*</tt>' Intrinsic</a><br>
6965 <a name="int_atomic_load_umax">'<tt>llvm.atomic.load.umax.*</tt>' Intrinsic</a><br>
6966 <a name="int_atomic_load_umin">'<tt>llvm.atomic.load.umin.*</tt>' Intrinsic</a><br>
6969 <div class="doc_text">
6972 <p>These are overloaded intrinsics. You can use <tt>llvm.atomic.load_max</tt>,
6973 <tt>llvm.atomic.load_min</tt>, <tt>llvm.atomic.load_umax</tt>, and
6974 <tt>llvm.atomic.load_umin</tt> on any integer bit width and for different
6975 address spaces. Not all targets support all bit widths however.</p>
6978 declare i8 @llvm.atomic.load.max.i8.p0i8( i8* <ptr>, i8 <delta> )
6979 declare i16 @llvm.atomic.load.max.i16.p0i16( i16* <ptr>, i16 <delta> )
6980 declare i32 @llvm.atomic.load.max.i32.p0i32( i32* <ptr>, i32 <delta> )
6981 declare i64 @llvm.atomic.load.max.i64.p0i64( i64* <ptr>, i64 <delta> )
6985 declare i8 @llvm.atomic.load.min.i8.p0i8( i8* <ptr>, i8 <delta> )
6986 declare i16 @llvm.atomic.load.min.i16.p0i16( i16* <ptr>, i16 <delta> )
6987 declare i32 @llvm.atomic.load.min.i32..p0i32( i32* <ptr>, i32 <delta> )
6988 declare i64 @llvm.atomic.load.min.i64..p0i64( i64* <ptr>, i64 <delta> )
6992 declare i8 @llvm.atomic.load.umax.i8.p0i8( i8* <ptr>, i8 <delta> )
6993 declare i16 @llvm.atomic.load.umax.i16.p0i16( i16* <ptr>, i16 <delta> )
6994 declare i32 @llvm.atomic.load.umax.i32.p0i32( i32* <ptr>, i32 <delta> )
6995 declare i64 @llvm.atomic.load.umax.i64.p0i64( i64* <ptr>, i64 <delta> )
6999 declare i8 @llvm.atomic.load.umin.i8..p0i8( i8* <ptr>, i8 <delta> )
7000 declare i16 @llvm.atomic.load.umin.i16.p0i16( i16* <ptr>, i16 <delta> )
7001 declare i32 @llvm.atomic.load.umin.i32..p0i32( i32* <ptr>, i32 <delta> )
7002 declare i64 @llvm.atomic.load.umin.i64..p0i64( i64* <ptr>, i64 <delta> )
7006 <p>These intrinsics takes the signed or unsigned minimum or maximum of
7007 <tt>delta</tt> and the value stored in memory at <tt>ptr</tt>. It yields the
7008 original value at <tt>ptr</tt>.</p>
7011 <p>These intrinsics take two arguments, the first a pointer to an integer value
7012 and the second an integer value. The result is also an integer value. These
7013 integer types can have any bit width, but they must all have the same bit
7014 width. The targets may only lower integer representations they support.</p>
7017 <p>These intrinsics does a series of operations atomically. They first load the
7018 value stored at <tt>ptr</tt>. They then do the signed or unsigned min or
7019 max <tt>delta</tt> and the value, store the result to <tt>ptr</tt>. They
7020 yield the original value stored at <tt>ptr</tt>.</p>
7024 %mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
7025 %ptr = bitcast i8* %mallocP to i32*
7027 %result0 = call i32 @llvm.atomic.load.min.i32.p0i32( i32* %ptr, i32 -2 )
7028 <i>; yields {i32}:result0 = 7</i>
7029 %result1 = call i32 @llvm.atomic.load.max.i32.p0i32( i32* %ptr, i32 8 )
7030 <i>; yields {i32}:result1 = -2</i>
7031 %result2 = call i32 @llvm.atomic.load.umin.i32.p0i32( i32* %ptr, i32 10 )
7032 <i>; yields {i32}:result2 = 8</i>
7033 %result3 = call i32 @llvm.atomic.load.umax.i32.p0i32( i32* %ptr, i32 30 )
7034 <i>; yields {i32}:result3 = 8</i>
7035 %memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 30</i>
7041 <!-- ======================================================================= -->
7042 <div class="doc_subsection">
7043 <a name="int_memorymarkers">Memory Use Markers</a>
7046 <div class="doc_text">
7048 <p>This class of intrinsics exists to information about the lifetime of memory
7049 objects and ranges where variables are immutable.</p>
7053 <!-- _______________________________________________________________________ -->
7054 <div class="doc_subsubsection">
7055 <a name="int_lifetime_start">'<tt>llvm.lifetime.start</tt>' Intrinsic</a>
7058 <div class="doc_text">
7062 declare void @llvm.lifetime.start(i64 <size>, i8* nocapture <ptr>)
7066 <p>The '<tt>llvm.lifetime.start</tt>' intrinsic specifies the start of a memory
7067 object's lifetime.</p>
7070 <p>The first argument is a constant integer representing the size of the
7071 object, or -1 if it is variable sized. The second argument is a pointer to
7075 <p>This intrinsic indicates that before this point in the code, the value of the
7076 memory pointed to by <tt>ptr</tt> is dead. This means that it is known to
7077 never be used and has an undefined value. A load from the pointer that
7078 precedes this intrinsic can be replaced with
7079 <tt>'<a href="#undefvalues">undef</a>'</tt>.</p>
7083 <!-- _______________________________________________________________________ -->
7084 <div class="doc_subsubsection">
7085 <a name="int_lifetime_end">'<tt>llvm.lifetime.end</tt>' Intrinsic</a>
7088 <div class="doc_text">
7092 declare void @llvm.lifetime.end(i64 <size>, i8* nocapture <ptr>)
7096 <p>The '<tt>llvm.lifetime.end</tt>' intrinsic specifies the end of a memory
7097 object's lifetime.</p>
7100 <p>The first argument is a constant integer representing the size of the
7101 object, or -1 if it is variable sized. The second argument is a pointer to
7105 <p>This intrinsic indicates that after this point in the code, the value of the
7106 memory pointed to by <tt>ptr</tt> is dead. This means that it is known to
7107 never be used and has an undefined value. Any stores into the memory object
7108 following this intrinsic may be removed as dead.
7112 <!-- _______________________________________________________________________ -->
7113 <div class="doc_subsubsection">
7114 <a name="int_invariant_start">'<tt>llvm.invariant.start</tt>' Intrinsic</a>
7117 <div class="doc_text">
7121 declare {}* @llvm.invariant.start(i64 <size>, i8* nocapture <ptr>) readonly
7125 <p>The '<tt>llvm.invariant.start</tt>' intrinsic specifies that the contents of
7126 a memory object will not change.</p>
7129 <p>The first argument is a constant integer representing the size of the
7130 object, or -1 if it is variable sized. The second argument is a pointer to
7134 <p>This intrinsic indicates that until an <tt>llvm.invariant.end</tt> that uses
7135 the return value, the referenced memory location is constant and
7140 <!-- _______________________________________________________________________ -->
7141 <div class="doc_subsubsection">
7142 <a name="int_invariant_end">'<tt>llvm.invariant.end</tt>' Intrinsic</a>
7145 <div class="doc_text">
7149 declare void @llvm.invariant.end({}* <start>, i64 <size>, i8* nocapture <ptr>)
7153 <p>The '<tt>llvm.invariant.end</tt>' intrinsic specifies that the contents of
7154 a memory object are mutable.</p>
7157 <p>The first argument is the matching <tt>llvm.invariant.start</tt> intrinsic.
7158 The second argument is a constant integer representing the size of the
7159 object, or -1 if it is variable sized and the third argument is a pointer
7163 <p>This intrinsic indicates that the memory is mutable again.</p>
7167 <!-- ======================================================================= -->
7168 <div class="doc_subsection">
7169 <a name="int_general">General Intrinsics</a>
7172 <div class="doc_text">
7174 <p>This class of intrinsics is designed to be generic and has no specific
7179 <!-- _______________________________________________________________________ -->
7180 <div class="doc_subsubsection">
7181 <a name="int_var_annotation">'<tt>llvm.var.annotation</tt>' Intrinsic</a>
7184 <div class="doc_text">
7188 declare void @llvm.var.annotation(i8* <val>, i8* <str>, i8* <str>, i32 <int> )
7192 <p>The '<tt>llvm.var.annotation</tt>' intrinsic.</p>
7195 <p>The first argument is a pointer to a value, the second is a pointer to a
7196 global string, the third is a pointer to a global string which is the source
7197 file name, and the last argument is the line number.</p>
7200 <p>This intrinsic allows annotation of local variables with arbitrary strings.
7201 This can be useful for special purpose optimizations that want to look for
7202 these annotations. These have no other defined use, they are ignored by code
7203 generation and optimization.</p>
7207 <!-- _______________________________________________________________________ -->
7208 <div class="doc_subsubsection">
7209 <a name="int_annotation">'<tt>llvm.annotation.*</tt>' Intrinsic</a>
7212 <div class="doc_text">
7215 <p>This is an overloaded intrinsic. You can use '<tt>llvm.annotation</tt>' on
7216 any integer bit width.</p>
7219 declare i8 @llvm.annotation.i8(i8 <val>, i8* <str>, i8* <str>, i32 <int> )
7220 declare i16 @llvm.annotation.i16(i16 <val>, i8* <str>, i8* <str>, i32 <int> )
7221 declare i32 @llvm.annotation.i32(i32 <val>, i8* <str>, i8* <str>, i32 <int> )
7222 declare i64 @llvm.annotation.i64(i64 <val>, i8* <str>, i8* <str>, i32 <int> )
7223 declare i256 @llvm.annotation.i256(i256 <val>, i8* <str>, i8* <str>, i32 <int> )
7227 <p>The '<tt>llvm.annotation</tt>' intrinsic.</p>
7230 <p>The first argument is an integer value (result of some expression), the
7231 second is a pointer to a global string, the third is a pointer to a global
7232 string which is the source file name, and the last argument is the line
7233 number. It returns the value of the first argument.</p>
7236 <p>This intrinsic allows annotations to be put on arbitrary expressions with
7237 arbitrary strings. This can be useful for special purpose optimizations that
7238 want to look for these annotations. These have no other defined use, they
7239 are ignored by code generation and optimization.</p>
7243 <!-- _______________________________________________________________________ -->
7244 <div class="doc_subsubsection">
7245 <a name="int_trap">'<tt>llvm.trap</tt>' Intrinsic</a>
7248 <div class="doc_text">
7252 declare void @llvm.trap()
7256 <p>The '<tt>llvm.trap</tt>' intrinsic.</p>
7262 <p>This intrinsics is lowered to the target dependent trap instruction. If the
7263 target does not have a trap instruction, this intrinsic will be lowered to
7264 the call of the <tt>abort()</tt> function.</p>
7268 <!-- _______________________________________________________________________ -->
7269 <div class="doc_subsubsection">
7270 <a name="int_stackprotector">'<tt>llvm.stackprotector</tt>' Intrinsic</a>
7273 <div class="doc_text">
7277 declare void @llvm.stackprotector( i8* <guard>, i8** <slot> )
7281 <p>The <tt>llvm.stackprotector</tt> intrinsic takes the <tt>guard</tt> and
7282 stores it onto the stack at <tt>slot</tt>. The stack slot is adjusted to
7283 ensure that it is placed on the stack before local variables.</p>
7286 <p>The <tt>llvm.stackprotector</tt> intrinsic requires two pointer
7287 arguments. The first argument is the value loaded from the stack
7288 guard <tt>@__stack_chk_guard</tt>. The second variable is an <tt>alloca</tt>
7289 that has enough space to hold the value of the guard.</p>
7292 <p>This intrinsic causes the prologue/epilogue inserter to force the position of
7293 the <tt>AllocaInst</tt> stack slot to be before local variables on the
7294 stack. This is to ensure that if a local variable on the stack is
7295 overwritten, it will destroy the value of the guard. When the function exits,
7296 the guard on the stack is checked against the original guard. If they're
7297 different, then the program aborts by calling the <tt>__stack_chk_fail()</tt>
7302 <!-- _______________________________________________________________________ -->
7303 <div class="doc_subsubsection">
7304 <a name="int_objectsize">'<tt>llvm.objectsize</tt>' Intrinsic</a>
7307 <div class="doc_text">
7311 declare i32 @llvm.objectsize.i32( i8* <object>, i1 <type> )
7312 declare i64 @llvm.objectsize.i64( i8* <object>, i1 <type> )
7316 <p>The <tt>llvm.objectsize</tt> intrinsic is designed to provide information
7317 to the optimizers to discover at compile time either a) when an
7318 operation like memcpy will either overflow a buffer that corresponds to
7319 an object, or b) to determine that a runtime check for overflow isn't
7320 necessary. An object in this context means an allocation of a
7321 specific class, structure, array, or other object.</p>
7324 <p>The <tt>llvm.objectsize</tt> intrinsic takes two arguments. The first
7325 argument is a pointer to or into the <tt>object</tt>. The second argument
7326 is a boolean 0 or 1. This argument determines whether you want the
7327 maximum (0) or minimum (1) bytes remaining. This needs to be a literal 0 or
7328 1, variables are not allowed.</p>
7331 <p>The <tt>llvm.objectsize</tt> intrinsic is lowered to either a constant
7332 representing the size of the object concerned or <tt>i32/i64 -1 or 0</tt>
7333 (depending on the <tt>type</tt> argument if the size cannot be determined
7334 at compile time.</p>
7338 <!-- *********************************************************************** -->
7341 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
7342 src="http://jigsaw.w3.org/css-validator/images/vcss-blue" alt="Valid CSS"></a>
7343 <a href="http://validator.w3.org/check/referer"><img
7344 src="http://www.w3.org/Icons/valid-html401-blue" alt="Valid HTML 4.01"></a>
7346 <a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
7347 <a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br>
7348 Last modified: $Date$