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 </a>
928 node and null are the only valid named metadata operands.
929 Metadata strings are not allowed as an named metadata operand.</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>inlinehint</b></tt></dt>
1087 <dd>This attribute indicates that the source code contained a hint that inlining
1088 this function is desirable (such as the "inline" keyword in C/C++). It
1089 is just a hint; it imposes no requirements on the inliner.</dd>
1091 <dt><tt><b>noinline</b></tt></dt>
1092 <dd>This attribute indicates that the inliner should never inline this
1093 function in any situation. This attribute may not be used together with
1094 the <tt>alwaysinline</tt> attribute.</dd>
1096 <dt><tt><b>optsize</b></tt></dt>
1097 <dd>This attribute suggests that optimization passes and code generator passes
1098 make choices that keep the code size of this function low, and otherwise
1099 do optimizations specifically to reduce code size.</dd>
1101 <dt><tt><b>noreturn</b></tt></dt>
1102 <dd>This function attribute indicates that the function never returns
1103 normally. This produces undefined behavior at runtime if the function
1104 ever does dynamically return.</dd>
1106 <dt><tt><b>nounwind</b></tt></dt>
1107 <dd>This function attribute indicates that the function never returns with an
1108 unwind or exceptional control flow. If the function does unwind, its
1109 runtime behavior is undefined.</dd>
1111 <dt><tt><b>readnone</b></tt></dt>
1112 <dd>This attribute indicates that the function computes its result (or decides
1113 to unwind an exception) based strictly on its arguments, without
1114 dereferencing any pointer arguments or otherwise accessing any mutable
1115 state (e.g. memory, control registers, etc) visible to caller functions.
1116 It does not write through any pointer arguments
1117 (including <tt><a href="#byval">byval</a></tt> arguments) and never
1118 changes any state visible to callers. This means that it cannot unwind
1119 exceptions by calling the <tt>C++</tt> exception throwing methods, but
1120 could use the <tt>unwind</tt> instruction.</dd>
1122 <dt><tt><b><a name="readonly">readonly</a></b></tt></dt>
1123 <dd>This attribute indicates that the function does not write through any
1124 pointer arguments (including <tt><a href="#byval">byval</a></tt>
1125 arguments) or otherwise modify any state (e.g. memory, control registers,
1126 etc) visible to caller functions. It may dereference pointer arguments
1127 and read state that may be set in the caller. A readonly function always
1128 returns the same value (or unwinds an exception identically) when called
1129 with the same set of arguments and global state. It cannot unwind an
1130 exception by calling the <tt>C++</tt> exception throwing methods, but may
1131 use the <tt>unwind</tt> instruction.</dd>
1133 <dt><tt><b><a name="ssp">ssp</a></b></tt></dt>
1134 <dd>This attribute indicates that the function should emit a stack smashing
1135 protector. It is in the form of a "canary"—a random value placed on
1136 the stack before the local variables that's checked upon return from the
1137 function to see if it has been overwritten. A heuristic is used to
1138 determine if a function needs stack protectors or not.<br>
1140 If a function that has an <tt>ssp</tt> attribute is inlined into a
1141 function that doesn't have an <tt>ssp</tt> attribute, then the resulting
1142 function will have an <tt>ssp</tt> attribute.</dd>
1144 <dt><tt><b>sspreq</b></tt></dt>
1145 <dd>This attribute indicates that the function should <em>always</em> emit a
1146 stack smashing protector. This overrides
1147 the <tt><a href="#ssp">ssp</a></tt> function attribute.<br>
1149 If a function that has an <tt>sspreq</tt> attribute is inlined into a
1150 function that doesn't have an <tt>sspreq</tt> attribute or which has
1151 an <tt>ssp</tt> attribute, then the resulting function will have
1152 an <tt>sspreq</tt> attribute.</dd>
1154 <dt><tt><b>noredzone</b></tt></dt>
1155 <dd>This attribute indicates that the code generator should not use a red
1156 zone, even if the target-specific ABI normally permits it.</dd>
1158 <dt><tt><b>noimplicitfloat</b></tt></dt>
1159 <dd>This attributes disables implicit floating point instructions.</dd>
1161 <dt><tt><b>naked</b></tt></dt>
1162 <dd>This attribute disables prologue / epilogue emission for the function.
1163 This can have very system-specific consequences.</dd>
1168 <!-- ======================================================================= -->
1169 <div class="doc_subsection">
1170 <a name="moduleasm">Module-Level Inline Assembly</a>
1173 <div class="doc_text">
1175 <p>Modules may contain "module-level inline asm" blocks, which corresponds to
1176 the GCC "file scope inline asm" blocks. These blocks are internally
1177 concatenated by LLVM and treated as a single unit, but may be separated in
1178 the <tt>.ll</tt> file if desired. The syntax is very simple:</p>
1180 <div class="doc_code">
1182 module asm "inline asm code goes here"
1183 module asm "more can go here"
1187 <p>The strings can contain any character by escaping non-printable characters.
1188 The escape sequence used is simply "\xx" where "xx" is the two digit hex code
1191 <p>The inline asm code is simply printed to the machine code .s file when
1192 assembly code is generated.</p>
1196 <!-- ======================================================================= -->
1197 <div class="doc_subsection">
1198 <a name="datalayout">Data Layout</a>
1201 <div class="doc_text">
1203 <p>A module may specify a target specific data layout string that specifies how
1204 data is to be laid out in memory. The syntax for the data layout is
1207 <div class="doc_code">
1209 target datalayout = "<i>layout specification</i>"
1213 <p>The <i>layout specification</i> consists of a list of specifications
1214 separated by the minus sign character ('-'). Each specification starts with
1215 a letter and may include other information after the letter to define some
1216 aspect of the data layout. The specifications accepted are as follows:</p>
1220 <dd>Specifies that the target lays out data in big-endian form. That is, the
1221 bits with the most significance have the lowest address location.</dd>
1224 <dd>Specifies that the target lays out data in little-endian form. That is,
1225 the bits with the least significance have the lowest address
1228 <dt><tt>p:<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1229 <dd>This specifies the <i>size</i> of a pointer and its <i>abi</i> and
1230 <i>preferred</i> alignments. All sizes are in bits. Specifying
1231 the <i>pref</i> alignment is optional. If omitted, the
1232 preceding <tt>:</tt> should be omitted too.</dd>
1234 <dt><tt>i<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1235 <dd>This specifies the alignment for an integer type of a given bit
1236 <i>size</i>. The value of <i>size</i> must be in the range [1,2^23).</dd>
1238 <dt><tt>v<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1239 <dd>This specifies the alignment for a vector type of a given bit
1242 <dt><tt>f<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1243 <dd>This specifies the alignment for a floating point type of a given bit
1244 <i>size</i>. The value of <i>size</i> must be either 32 (float) or 64
1247 <dt><tt>a<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1248 <dd>This specifies the alignment for an aggregate type of a given bit
1251 <dt><tt>s<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1252 <dd>This specifies the alignment for a stack object of a given bit
1255 <dt><tt>n<i>size1</i>:<i>size2</i>:<i>size3</i>...</tt></dt>
1256 <dd>This specifies a set of native integer widths for the target CPU
1257 in bits. For example, it might contain "n32" for 32-bit PowerPC,
1258 "n32:64" for PowerPC 64, or "n8:16:32:64" for X86-64. Elements of
1259 this set are considered to support most general arithmetic
1260 operations efficiently.</dd>
1263 <p>When constructing the data layout for a given target, LLVM starts with a
1264 default set of specifications which are then (possibly) overriden by the
1265 specifications in the <tt>datalayout</tt> keyword. The default specifications
1266 are given in this list:</p>
1269 <li><tt>E</tt> - big endian</li>
1270 <li><tt>p:32:64:64</tt> - 32-bit pointers with 64-bit alignment</li>
1271 <li><tt>i1:8:8</tt> - i1 is 8-bit (byte) aligned</li>
1272 <li><tt>i8:8:8</tt> - i8 is 8-bit (byte) aligned</li>
1273 <li><tt>i16:16:16</tt> - i16 is 16-bit aligned</li>
1274 <li><tt>i32:32:32</tt> - i32 is 32-bit aligned</li>
1275 <li><tt>i64:32:64</tt> - i64 has ABI alignment of 32-bits but preferred
1276 alignment of 64-bits</li>
1277 <li><tt>f32:32:32</tt> - float is 32-bit aligned</li>
1278 <li><tt>f64:64:64</tt> - double is 64-bit aligned</li>
1279 <li><tt>v64:64:64</tt> - 64-bit vector is 64-bit aligned</li>
1280 <li><tt>v128:128:128</tt> - 128-bit vector is 128-bit aligned</li>
1281 <li><tt>a0:0:1</tt> - aggregates are 8-bit aligned</li>
1282 <li><tt>s0:64:64</tt> - stack objects are 64-bit aligned</li>
1285 <p>When LLVM is determining the alignment for a given type, it uses the
1286 following rules:</p>
1289 <li>If the type sought is an exact match for one of the specifications, that
1290 specification is used.</li>
1292 <li>If no match is found, and the type sought is an integer type, then the
1293 smallest integer type that is larger than the bitwidth of the sought type
1294 is used. If none of the specifications are larger than the bitwidth then
1295 the the largest integer type is used. For example, given the default
1296 specifications above, the i7 type will use the alignment of i8 (next
1297 largest) while both i65 and i256 will use the alignment of i64 (largest
1300 <li>If no match is found, and the type sought is a vector type, then the
1301 largest vector type that is smaller than the sought vector type will be
1302 used as a fall back. This happens because <128 x double> can be
1303 implemented in terms of 64 <2 x double>, for example.</li>
1308 <!-- ======================================================================= -->
1309 <div class="doc_subsection">
1310 <a name="pointeraliasing">Pointer Aliasing Rules</a>
1313 <div class="doc_text">
1315 <p>Any memory access must be done through a pointer value associated
1316 with an address range of the memory access, otherwise the behavior
1317 is undefined. Pointer values are associated with address ranges
1318 according to the following rules:</p>
1321 <li>A pointer value formed from a
1322 <tt><a href="#i_getelementptr">getelementptr</a></tt> instruction
1323 is associated with the addresses associated with the first operand
1324 of the <tt>getelementptr</tt>.</li>
1325 <li>An address of a global variable is associated with the address
1326 range of the variable's storage.</li>
1327 <li>The result value of an allocation instruction is associated with
1328 the address range of the allocated storage.</li>
1329 <li>A null pointer in the default address-space is associated with
1331 <li>A pointer value formed by an
1332 <tt><a href="#i_inttoptr">inttoptr</a></tt> is associated with all
1333 address ranges of all pointer values that contribute (directly or
1334 indirectly) to the computation of the pointer's value.</li>
1335 <li>The result value of a
1336 <tt><a href="#i_bitcast">bitcast</a></tt> is associated with all
1337 addresses associated with the operand of the <tt>bitcast</tt>.</li>
1338 <li>An integer constant other than zero or a pointer value returned
1339 from a function not defined within LLVM may be associated with address
1340 ranges allocated through mechanisms other than those provided by
1341 LLVM. Such ranges shall not overlap with any ranges of addresses
1342 allocated by mechanisms provided by LLVM.</li>
1345 <p>LLVM IR does not associate types with memory. The result type of a
1346 <tt><a href="#i_load">load</a></tt> merely indicates the size and
1347 alignment of the memory from which to load, as well as the
1348 interpretation of the value. The first operand of a
1349 <tt><a href="#i_store">store</a></tt> similarly only indicates the size
1350 and alignment of the store.</p>
1352 <p>Consequently, type-based alias analysis, aka TBAA, aka
1353 <tt>-fstrict-aliasing</tt>, is not applicable to general unadorned
1354 LLVM IR. <a href="#metadata">Metadata</a> may be used to encode
1355 additional information which specialized optimization passes may use
1356 to implement type-based alias analysis.</p>
1360 <!-- *********************************************************************** -->
1361 <div class="doc_section"> <a name="typesystem">Type System</a> </div>
1362 <!-- *********************************************************************** -->
1364 <div class="doc_text">
1366 <p>The LLVM type system is one of the most important features of the
1367 intermediate representation. Being typed enables a number of optimizations
1368 to be performed on the intermediate representation directly, without having
1369 to do extra analyses on the side before the transformation. A strong type
1370 system makes it easier to read the generated code and enables novel analyses
1371 and transformations that are not feasible to perform on normal three address
1372 code representations.</p>
1376 <!-- ======================================================================= -->
1377 <div class="doc_subsection"> <a name="t_classifications">Type
1378 Classifications</a> </div>
1380 <div class="doc_text">
1382 <p>The types fall into a few useful classifications:</p>
1384 <table border="1" cellspacing="0" cellpadding="4">
1386 <tr><th>Classification</th><th>Types</th></tr>
1388 <td><a href="#t_integer">integer</a></td>
1389 <td><tt>i1, i2, i3, ... i8, ... i16, ... i32, ... i64, ... </tt></td>
1392 <td><a href="#t_floating">floating point</a></td>
1393 <td><tt>float, double, x86_fp80, fp128, ppc_fp128</tt></td>
1396 <td><a name="t_firstclass">first class</a></td>
1397 <td><a href="#t_integer">integer</a>,
1398 <a href="#t_floating">floating point</a>,
1399 <a href="#t_pointer">pointer</a>,
1400 <a href="#t_vector">vector</a>,
1401 <a href="#t_struct">structure</a>,
1402 <a href="#t_array">array</a>,
1403 <a href="#t_label">label</a>,
1404 <a href="#t_metadata">metadata</a>.
1408 <td><a href="#t_primitive">primitive</a></td>
1409 <td><a href="#t_label">label</a>,
1410 <a href="#t_void">void</a>,
1411 <a href="#t_floating">floating point</a>,
1412 <a href="#t_metadata">metadata</a>.</td>
1415 <td><a href="#t_derived">derived</a></td>
1416 <td><a href="#t_integer">integer</a>,
1417 <a href="#t_array">array</a>,
1418 <a href="#t_function">function</a>,
1419 <a href="#t_pointer">pointer</a>,
1420 <a href="#t_struct">structure</a>,
1421 <a href="#t_pstruct">packed structure</a>,
1422 <a href="#t_vector">vector</a>,
1423 <a href="#t_opaque">opaque</a>.
1429 <p>The <a href="#t_firstclass">first class</a> types are perhaps the most
1430 important. Values of these types are the only ones which can be produced by
1435 <!-- ======================================================================= -->
1436 <div class="doc_subsection"> <a name="t_primitive">Primitive Types</a> </div>
1438 <div class="doc_text">
1440 <p>The primitive types are the fundamental building blocks of the LLVM
1445 <!-- _______________________________________________________________________ -->
1446 <div class="doc_subsubsection"> <a name="t_integer">Integer Type</a> </div>
1448 <div class="doc_text">
1451 <p>The integer type is a very simple type that simply specifies an arbitrary
1452 bit width for the integer type desired. Any bit width from 1 bit to
1453 2<sup>23</sup>-1 (about 8 million) can be specified.</p>
1460 <p>The number of bits the integer will occupy is specified by the <tt>N</tt>
1464 <table class="layout">
1466 <td class="left"><tt>i1</tt></td>
1467 <td class="left">a single-bit integer.</td>
1470 <td class="left"><tt>i32</tt></td>
1471 <td class="left">a 32-bit integer.</td>
1474 <td class="left"><tt>i1942652</tt></td>
1475 <td class="left">a really big integer of over 1 million bits.</td>
1481 <!-- _______________________________________________________________________ -->
1482 <div class="doc_subsubsection"> <a name="t_floating">Floating Point Types</a> </div>
1484 <div class="doc_text">
1488 <tr><th>Type</th><th>Description</th></tr>
1489 <tr><td><tt>float</tt></td><td>32-bit floating point value</td></tr>
1490 <tr><td><tt>double</tt></td><td>64-bit floating point value</td></tr>
1491 <tr><td><tt>fp128</tt></td><td>128-bit floating point value (112-bit mantissa)</td></tr>
1492 <tr><td><tt>x86_fp80</tt></td><td>80-bit floating point value (X87)</td></tr>
1493 <tr><td><tt>ppc_fp128</tt></td><td>128-bit floating point value (two 64-bits)</td></tr>
1499 <!-- _______________________________________________________________________ -->
1500 <div class="doc_subsubsection"> <a name="t_void">Void Type</a> </div>
1502 <div class="doc_text">
1505 <p>The void type does not represent any value and has no size.</p>
1514 <!-- _______________________________________________________________________ -->
1515 <div class="doc_subsubsection"> <a name="t_label">Label Type</a> </div>
1517 <div class="doc_text">
1520 <p>The label type represents code labels.</p>
1529 <!-- _______________________________________________________________________ -->
1530 <div class="doc_subsubsection"> <a name="t_metadata">Metadata Type</a> </div>
1532 <div class="doc_text">
1535 <p>The metadata type represents embedded metadata. No derived types may be
1536 created from metadata except for <a href="#t_function">function</a>
1547 <!-- ======================================================================= -->
1548 <div class="doc_subsection"> <a name="t_derived">Derived Types</a> </div>
1550 <div class="doc_text">
1552 <p>The real power in LLVM comes from the derived types in the system. This is
1553 what allows a programmer to represent arrays, functions, pointers, and other
1554 useful types. Each of these types contain one or more element types which
1555 may be a primitive type, or another derived type. For example, it is
1556 possible to have a two dimensional array, using an array as the element type
1557 of another array.</p>
1561 <!-- _______________________________________________________________________ -->
1562 <div class="doc_subsubsection"> <a name="t_array">Array Type</a> </div>
1564 <div class="doc_text">
1567 <p>The array type is a very simple derived type that arranges elements
1568 sequentially in memory. The array type requires a size (number of elements)
1569 and an underlying data type.</p>
1573 [<# elements> x <elementtype>]
1576 <p>The number of elements is a constant integer value; <tt>elementtype</tt> may
1577 be any type with a size.</p>
1580 <table class="layout">
1582 <td class="left"><tt>[40 x i32]</tt></td>
1583 <td class="left">Array of 40 32-bit integer values.</td>
1586 <td class="left"><tt>[41 x i32]</tt></td>
1587 <td class="left">Array of 41 32-bit integer values.</td>
1590 <td class="left"><tt>[4 x i8]</tt></td>
1591 <td class="left">Array of 4 8-bit integer values.</td>
1594 <p>Here are some examples of multidimensional arrays:</p>
1595 <table class="layout">
1597 <td class="left"><tt>[3 x [4 x i32]]</tt></td>
1598 <td class="left">3x4 array of 32-bit integer values.</td>
1601 <td class="left"><tt>[12 x [10 x float]]</tt></td>
1602 <td class="left">12x10 array of single precision floating point values.</td>
1605 <td class="left"><tt>[2 x [3 x [4 x i16]]]</tt></td>
1606 <td class="left">2x3x4 array of 16-bit integer values.</td>
1610 <p>There is no restriction on indexing beyond the end of the array implied by
1611 a static type (though there are restrictions on indexing beyond the bounds
1612 of an allocated object in some cases). This means that single-dimension
1613 'variable sized array' addressing can be implemented in LLVM with a zero
1614 length array type. An implementation of 'pascal style arrays' in LLVM could
1615 use the type "<tt>{ i32, [0 x float]}</tt>", for example.</p>
1619 <!-- _______________________________________________________________________ -->
1620 <div class="doc_subsubsection"> <a name="t_function">Function Type</a> </div>
1622 <div class="doc_text">
1625 <p>The function type can be thought of as a function signature. It consists of
1626 a return type and a list of formal parameter types. The return type of a
1627 function type is a scalar type, a void type, or a struct type. If the return
1628 type is a struct type then all struct elements must be of first class types,
1629 and the struct must have at least one element.</p>
1633 <returntype> (<parameter list>)
1636 <p>...where '<tt><parameter list></tt>' is a comma-separated list of type
1637 specifiers. Optionally, the parameter list may include a type <tt>...</tt>,
1638 which indicates that the function takes a variable number of arguments.
1639 Variable argument functions can access their arguments with
1640 the <a href="#int_varargs">variable argument handling intrinsic</a>
1641 functions. '<tt><returntype></tt>' is a any type except
1642 <a href="#t_label">label</a>.</p>
1645 <table class="layout">
1647 <td class="left"><tt>i32 (i32)</tt></td>
1648 <td class="left">function taking an <tt>i32</tt>, returning an <tt>i32</tt>
1650 </tr><tr class="layout">
1651 <td class="left"><tt>float (i16 signext, i32 *) *
1653 <td class="left"><a href="#t_pointer">Pointer</a> to a function that takes
1654 an <tt>i16</tt> that should be sign extended and a
1655 <a href="#t_pointer">pointer</a> to <tt>i32</tt>, returning
1658 </tr><tr class="layout">
1659 <td class="left"><tt>i32 (i8*, ...)</tt></td>
1660 <td class="left">A vararg function that takes at least one
1661 <a href="#t_pointer">pointer</a> to <tt>i8 </tt> (char in C),
1662 which returns an integer. This is the signature for <tt>printf</tt> in
1665 </tr><tr class="layout">
1666 <td class="left"><tt>{i32, i32} (i32)</tt></td>
1667 <td class="left">A function taking an <tt>i32</tt>, returning a
1668 <a href="#t_struct">structure</a> containing two <tt>i32</tt> values
1675 <!-- _______________________________________________________________________ -->
1676 <div class="doc_subsubsection"> <a name="t_struct">Structure Type</a> </div>
1678 <div class="doc_text">
1681 <p>The structure type is used to represent a collection of data members together
1682 in memory. The packing of the field types is defined to match the ABI of the
1683 underlying processor. The elements of a structure may be any type that has a
1686 <p>Structures in memory are accessed using '<tt><a href="#i_load">load</a></tt>'
1687 and '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a field
1688 with the '<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.
1689 Structures in registers are accessed using the
1690 '<tt><a href="#i_extractvalue">extractvalue</a></tt>' and
1691 '<tt><a href="#i_insertvalue">insertvalue</a></tt>' instructions.</p>
1694 { <type list> }
1698 <table class="layout">
1700 <td class="left"><tt>{ i32, i32, i32 }</tt></td>
1701 <td class="left">A triple of three <tt>i32</tt> values</td>
1702 </tr><tr class="layout">
1703 <td class="left"><tt>{ float, i32 (i32) * }</tt></td>
1704 <td class="left">A pair, where the first element is a <tt>float</tt> and the
1705 second element is a <a href="#t_pointer">pointer</a> to a
1706 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
1707 an <tt>i32</tt>.</td>
1713 <!-- _______________________________________________________________________ -->
1714 <div class="doc_subsubsection"> <a name="t_pstruct">Packed Structure Type</a>
1717 <div class="doc_text">
1720 <p>The packed structure type is used to represent a collection of data members
1721 together in memory. There is no padding between fields. Further, the
1722 alignment of a packed structure is 1 byte. The elements of a packed
1723 structure may be any type that has a size.</p>
1725 <p>Structures are accessed using '<tt><a href="#i_load">load</a></tt> and
1726 '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a field with
1727 the '<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.</p>
1731 < { <type list> } >
1735 <table class="layout">
1737 <td class="left"><tt>< { i32, i32, i32 } ></tt></td>
1738 <td class="left">A triple of three <tt>i32</tt> values</td>
1739 </tr><tr class="layout">
1741 <tt>< { float, i32 (i32)* } ></tt></td>
1742 <td class="left">A pair, where the first element is a <tt>float</tt> and the
1743 second element is a <a href="#t_pointer">pointer</a> to a
1744 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
1745 an <tt>i32</tt>.</td>
1751 <!-- _______________________________________________________________________ -->
1752 <div class="doc_subsubsection"> <a name="t_pointer">Pointer Type</a> </div>
1754 <div class="doc_text">
1757 <p>As in many languages, the pointer type represents a pointer or reference to
1758 another object, which must live in memory. Pointer types may have an optional
1759 address space attribute defining the target-specific numbered address space
1760 where the pointed-to object resides. The default address space is zero.</p>
1762 <p>Note that LLVM does not permit pointers to void (<tt>void*</tt>) nor does it
1763 permit pointers to labels (<tt>label*</tt>). Use <tt>i8*</tt> instead.</p>
1771 <table class="layout">
1773 <td class="left"><tt>[4 x i32]*</tt></td>
1774 <td class="left">A <a href="#t_pointer">pointer</a> to <a
1775 href="#t_array">array</a> of four <tt>i32</tt> values.</td>
1778 <td class="left"><tt>i32 (i32 *) *</tt></td>
1779 <td class="left"> A <a href="#t_pointer">pointer</a> to a <a
1780 href="#t_function">function</a> that takes an <tt>i32*</tt>, returning an
1784 <td class="left"><tt>i32 addrspace(5)*</tt></td>
1785 <td class="left">A <a href="#t_pointer">pointer</a> to an <tt>i32</tt> value
1786 that resides in address space #5.</td>
1792 <!-- _______________________________________________________________________ -->
1793 <div class="doc_subsubsection"> <a name="t_vector">Vector Type</a> </div>
1795 <div class="doc_text">
1798 <p>A vector type is a simple derived type that represents a vector of elements.
1799 Vector types are used when multiple primitive data are operated in parallel
1800 using a single instruction (SIMD). A vector type requires a size (number of
1801 elements) and an underlying primitive data type. Vector types are considered
1802 <a href="#t_firstclass">first class</a>.</p>
1806 < <# elements> x <elementtype> >
1809 <p>The number of elements is a constant integer value; elementtype may be any
1810 integer or floating point type.</p>
1813 <table class="layout">
1815 <td class="left"><tt><4 x i32></tt></td>
1816 <td class="left">Vector of 4 32-bit integer values.</td>
1819 <td class="left"><tt><8 x float></tt></td>
1820 <td class="left">Vector of 8 32-bit floating-point values.</td>
1823 <td class="left"><tt><2 x i64></tt></td>
1824 <td class="left">Vector of 2 64-bit integer values.</td>
1830 <!-- _______________________________________________________________________ -->
1831 <div class="doc_subsubsection"> <a name="t_opaque">Opaque Type</a> </div>
1832 <div class="doc_text">
1835 <p>Opaque types are used to represent unknown types in the system. This
1836 corresponds (for example) to the C notion of a forward declared structure
1837 type. In LLVM, opaque types can eventually be resolved to any type (not just
1838 a structure type).</p>
1846 <table class="layout">
1848 <td class="left"><tt>opaque</tt></td>
1849 <td class="left">An opaque type.</td>
1855 <!-- ======================================================================= -->
1856 <div class="doc_subsection">
1857 <a name="t_uprefs">Type Up-references</a>
1860 <div class="doc_text">
1863 <p>An "up reference" allows you to refer to a lexically enclosing type without
1864 requiring it to have a name. For instance, a structure declaration may
1865 contain a pointer to any of the types it is lexically a member of. Example
1866 of up references (with their equivalent as named type declarations)
1870 { \2 * } %x = type { %x* }
1871 { \2 }* %y = type { %y }*
1875 <p>An up reference is needed by the asmprinter for printing out cyclic types
1876 when there is no declared name for a type in the cycle. Because the
1877 asmprinter does not want to print out an infinite type string, it needs a
1878 syntax to handle recursive types that have no names (all names are optional
1886 <p>The level is the count of the lexical type that is being referred to.</p>
1889 <table class="layout">
1891 <td class="left"><tt>\1*</tt></td>
1892 <td class="left">Self-referential pointer.</td>
1895 <td class="left"><tt>{ { \3*, i8 }, i32 }</tt></td>
1896 <td class="left">Recursive structure where the upref refers to the out-most
1903 <!-- *********************************************************************** -->
1904 <div class="doc_section"> <a name="constants">Constants</a> </div>
1905 <!-- *********************************************************************** -->
1907 <div class="doc_text">
1909 <p>LLVM has several different basic types of constants. This section describes
1910 them all and their syntax.</p>
1914 <!-- ======================================================================= -->
1915 <div class="doc_subsection"><a name="simpleconstants">Simple Constants</a></div>
1917 <div class="doc_text">
1920 <dt><b>Boolean constants</b></dt>
1921 <dd>The two strings '<tt>true</tt>' and '<tt>false</tt>' are both valid
1922 constants of the <tt><a href="#t_integer">i1</a></tt> type.</dd>
1924 <dt><b>Integer constants</b></dt>
1925 <dd>Standard integers (such as '4') are constants of
1926 the <a href="#t_integer">integer</a> type. Negative numbers may be used
1927 with integer types.</dd>
1929 <dt><b>Floating point constants</b></dt>
1930 <dd>Floating point constants use standard decimal notation (e.g. 123.421),
1931 exponential notation (e.g. 1.23421e+2), or a more precise hexadecimal
1932 notation (see below). The assembler requires the exact decimal value of a
1933 floating-point constant. For example, the assembler accepts 1.25 but
1934 rejects 1.3 because 1.3 is a repeating decimal in binary. Floating point
1935 constants must have a <a href="#t_floating">floating point</a> type. </dd>
1937 <dt><b>Null pointer constants</b></dt>
1938 <dd>The identifier '<tt>null</tt>' is recognized as a null pointer constant
1939 and must be of <a href="#t_pointer">pointer type</a>.</dd>
1942 <p>The one non-intuitive notation for constants is the hexadecimal form of
1943 floating point constants. For example, the form '<tt>double
1944 0x432ff973cafa8000</tt>' is equivalent to (but harder to read than)
1945 '<tt>double 4.5e+15</tt>'. The only time hexadecimal floating point
1946 constants are required (and the only time that they are generated by the
1947 disassembler) is when a floating point constant must be emitted but it cannot
1948 be represented as a decimal floating point number in a reasonable number of
1949 digits. For example, NaN's, infinities, and other special values are
1950 represented in their IEEE hexadecimal format so that assembly and disassembly
1951 do not cause any bits to change in the constants.</p>
1953 <p>When using the hexadecimal form, constants of types float and double are
1954 represented using the 16-digit form shown above (which matches the IEEE754
1955 representation for double); float values must, however, be exactly
1956 representable as IEE754 single precision. Hexadecimal format is always used
1957 for long double, and there are three forms of long double. The 80-bit format
1958 used by x86 is represented as <tt>0xK</tt> followed by 20 hexadecimal digits.
1959 The 128-bit format used by PowerPC (two adjacent doubles) is represented
1960 by <tt>0xM</tt> followed by 32 hexadecimal digits. The IEEE 128-bit format
1961 is represented by <tt>0xL</tt> followed by 32 hexadecimal digits; no
1962 currently supported target uses this format. Long doubles will only work if
1963 they match the long double format on your target. All hexadecimal formats
1964 are big-endian (sign bit at the left).</p>
1968 <!-- ======================================================================= -->
1969 <div class="doc_subsection">
1970 <a name="aggregateconstants"></a> <!-- old anchor -->
1971 <a name="complexconstants">Complex Constants</a>
1974 <div class="doc_text">
1976 <p>Complex constants are a (potentially recursive) combination of simple
1977 constants and smaller complex constants.</p>
1980 <dt><b>Structure constants</b></dt>
1981 <dd>Structure constants are represented with notation similar to structure
1982 type definitions (a comma separated list of elements, surrounded by braces
1983 (<tt>{}</tt>)). For example: "<tt>{ i32 4, float 17.0, i32* @G }</tt>",
1984 where "<tt>@G</tt>" is declared as "<tt>@G = external global i32</tt>".
1985 Structure constants must have <a href="#t_struct">structure type</a>, and
1986 the number and types of elements must match those specified by the
1989 <dt><b>Array constants</b></dt>
1990 <dd>Array constants are represented with notation similar to array type
1991 definitions (a comma separated list of elements, surrounded by square
1992 brackets (<tt>[]</tt>)). For example: "<tt>[ i32 42, i32 11, i32 74
1993 ]</tt>". Array constants must have <a href="#t_array">array type</a>, and
1994 the number and types of elements must match those specified by the
1997 <dt><b>Vector constants</b></dt>
1998 <dd>Vector constants are represented with notation similar to vector type
1999 definitions (a comma separated list of elements, surrounded by
2000 less-than/greater-than's (<tt><></tt>)). For example: "<tt>< i32
2001 42, i32 11, i32 74, i32 100 ></tt>". Vector constants must
2002 have <a href="#t_vector">vector type</a>, and the number and types of
2003 elements must match those specified by the type.</dd>
2005 <dt><b>Zero initialization</b></dt>
2006 <dd>The string '<tt>zeroinitializer</tt>' can be used to zero initialize a
2007 value to zero of <em>any</em> type, including scalar and aggregate types.
2008 This is often used to avoid having to print large zero initializers
2009 (e.g. for large arrays) and is always exactly equivalent to using explicit
2010 zero initializers.</dd>
2012 <dt><b>Metadata node</b></dt>
2013 <dd>A metadata node is a structure-like constant with
2014 <a href="#t_metadata">metadata type</a>. For example: "<tt>metadata !{
2015 i32 0, metadata !"test" }</tt>". Unlike other constants that are meant to
2016 be interpreted as part of the instruction stream, metadata is a place to
2017 attach additional information such as debug info.</dd>
2022 <!-- ======================================================================= -->
2023 <div class="doc_subsection">
2024 <a name="globalconstants">Global Variable and Function Addresses</a>
2027 <div class="doc_text">
2029 <p>The addresses of <a href="#globalvars">global variables</a>
2030 and <a href="#functionstructure">functions</a> are always implicitly valid
2031 (link-time) constants. These constants are explicitly referenced when
2032 the <a href="#identifiers">identifier for the global</a> is used and always
2033 have <a href="#t_pointer">pointer</a> type. For example, the following is a
2034 legal LLVM file:</p>
2036 <div class="doc_code">
2040 @Z = global [2 x i32*] [ i32* @X, i32* @Y ]
2046 <!-- ======================================================================= -->
2047 <div class="doc_subsection"><a name="undefvalues">Undefined Values</a></div>
2048 <div class="doc_text">
2050 <p>The string '<tt>undef</tt>' can be used anywhere a constant is expected, and
2051 indicates that the user of the value may receive an unspecified bit-pattern.
2052 Undefined values may be of any type (other than label or void) and be used
2053 anywhere a constant is permitted.</p>
2055 <p>Undefined values are useful because they indicate to the compiler that the
2056 program is well defined no matter what value is used. This gives the
2057 compiler more freedom to optimize. Here are some examples of (potentially
2058 surprising) transformations that are valid (in pseudo IR):</p>
2061 <div class="doc_code">
2073 <p>This is safe because all of the output bits are affected by the undef bits.
2074 Any output bit can have a zero or one depending on the input bits.</p>
2076 <div class="doc_code">
2089 <p>These logical operations have bits that are not always affected by the input.
2090 For example, if "%X" has a zero bit, then the output of the 'and' operation will
2091 always be a zero, no matter what the corresponding bit from the undef is. As
2092 such, it is unsafe to optimize or assume that the result of the and is undef.
2093 However, it is safe to assume that all bits of the undef could be 0, and
2094 optimize the and to 0. Likewise, it is safe to assume that all the bits of
2095 the undef operand to the or could be set, allowing the or to be folded to
2098 <div class="doc_code">
2100 %A = select undef, %X, %Y
2101 %B = select undef, 42, %Y
2102 %C = select %X, %Y, undef
2114 <p>This set of examples show that undefined select (and conditional branch)
2115 conditions can go "either way" but they have to come from one of the two
2116 operands. In the %A example, if %X and %Y were both known to have a clear low
2117 bit, then %A would have to have a cleared low bit. However, in the %C example,
2118 the optimizer is allowed to assume that the undef operand could be the same as
2119 %Y, allowing the whole select to be eliminated.</p>
2122 <div class="doc_code">
2124 %A = xor undef, undef
2143 <p>This example points out that two undef operands are not necessarily the same.
2144 This can be surprising to people (and also matches C semantics) where they
2145 assume that "X^X" is always zero, even if X is undef. This isn't true for a
2146 number of reasons, but the short answer is that an undef "variable" can
2147 arbitrarily change its value over its "live range". This is true because the
2148 "variable" doesn't actually <em>have a live range</em>. Instead, the value is
2149 logically read from arbitrary registers that happen to be around when needed,
2150 so the value is not necessarily consistent over time. In fact, %A and %C need
2151 to have the same semantics or the core LLVM "replace all uses with" concept
2154 <div class="doc_code">
2164 <p>These examples show the crucial difference between an <em>undefined
2165 value</em> and <em>undefined behavior</em>. An undefined value (like undef) is
2166 allowed to have an arbitrary bit-pattern. This means that the %A operation
2167 can be constant folded to undef because the undef could be an SNaN, and fdiv is
2168 not (currently) defined on SNaN's. However, in the second example, we can make
2169 a more aggressive assumption: because the undef is allowed to be an arbitrary
2170 value, we are allowed to assume that it could be zero. Since a divide by zero
2171 has <em>undefined behavior</em>, we are allowed to assume that the operation
2172 does not execute at all. This allows us to delete the divide and all code after
2173 it: since the undefined operation "can't happen", the optimizer can assume that
2174 it occurs in dead code.
2177 <div class="doc_code">
2179 a: store undef -> %X
2180 b: store %X -> undef
2187 <p>These examples reiterate the fdiv example: a store "of" an undefined value
2188 can be assumed to not have any effect: we can assume that the value is
2189 overwritten with bits that happen to match what was already there. However, a
2190 store "to" an undefined location could clobber arbitrary memory, therefore, it
2191 has undefined behavior.</p>
2195 <!-- ======================================================================= -->
2196 <div class="doc_subsection"><a name="blockaddress">Addresses of Basic
2198 <div class="doc_text">
2200 <p><b><tt>blockaddress(@function, %block)</tt></b></p>
2202 <p>The '<tt>blockaddress</tt>' constant computes the address of the specified
2203 basic block in the specified function, and always has an i8* type. Taking
2204 the address of the entry block is illegal.</p>
2206 <p>This value only has defined behavior when used as an operand to the
2207 '<a href="#i_indirectbr"><tt>indirectbr</tt></a>' instruction or for comparisons
2208 against null. Pointer equality tests between labels addresses is undefined
2209 behavior - though, again, comparison against null is ok, and no label is
2210 equal to the null pointer. This may also be passed around as an opaque
2211 pointer sized value as long as the bits are not inspected. This allows
2212 <tt>ptrtoint</tt> and arithmetic to be performed on these values so long as
2213 the original value is reconstituted before the <tt>indirectbr</tt>.</p>
2215 <p>Finally, some targets may provide defined semantics when
2216 using the value as the operand to an inline assembly, but that is target
2223 <!-- ======================================================================= -->
2224 <div class="doc_subsection"><a name="constantexprs">Constant Expressions</a>
2227 <div class="doc_text">
2229 <p>Constant expressions are used to allow expressions involving other constants
2230 to be used as constants. Constant expressions may be of
2231 any <a href="#t_firstclass">first class</a> type and may involve any LLVM
2232 operation that does not have side effects (e.g. load and call are not
2233 supported). The following is the syntax for constant expressions:</p>
2236 <dt><b><tt>trunc ( CST to TYPE )</tt></b></dt>
2237 <dd>Truncate a constant to another type. The bit size of CST must be larger
2238 than the bit size of TYPE. Both types must be integers.</dd>
2240 <dt><b><tt>zext ( CST to TYPE )</tt></b></dt>
2241 <dd>Zero 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>sext ( CST to TYPE )</tt></b></dt>
2246 <dd>Sign extend a constant to another type. The bit size of CST must be
2247 smaller or equal to the bit size of TYPE. Both types must be
2250 <dt><b><tt>fptrunc ( CST to TYPE )</tt></b></dt>
2251 <dd>Truncate a floating point constant to another floating point type. The
2252 size of CST must be larger than the size of TYPE. Both types must be
2253 floating point.</dd>
2255 <dt><b><tt>fpext ( CST to TYPE )</tt></b></dt>
2256 <dd>Floating point extend a constant to another type. The size of CST must be
2257 smaller or equal to the size of TYPE. Both types must be floating
2260 <dt><b><tt>fptoui ( CST to TYPE )</tt></b></dt>
2261 <dd>Convert a floating point constant to the corresponding unsigned integer
2262 constant. TYPE must be a scalar or vector integer type. CST must be of
2263 scalar or vector floating point type. Both CST and TYPE must be scalars,
2264 or vectors of the same number of elements. If the value won't fit in the
2265 integer type, the results are undefined.</dd>
2267 <dt><b><tt>fptosi ( CST to TYPE )</tt></b></dt>
2268 <dd>Convert a floating point constant to the corresponding signed integer
2269 constant. TYPE must be a scalar or vector integer type. CST must be of
2270 scalar or vector floating point type. Both CST and TYPE must be scalars,
2271 or vectors of the same number of elements. If the value won't fit in the
2272 integer type, the results are undefined.</dd>
2274 <dt><b><tt>uitofp ( CST to TYPE )</tt></b></dt>
2275 <dd>Convert an unsigned integer constant to the corresponding floating point
2276 constant. TYPE must be a scalar or vector floating point type. CST must be
2277 of scalar or vector integer type. Both CST and TYPE must be scalars, or
2278 vectors of the same number of elements. If the value won't fit in the
2279 floating point type, the results are undefined.</dd>
2281 <dt><b><tt>sitofp ( CST to TYPE )</tt></b></dt>
2282 <dd>Convert a signed integer constant to the corresponding floating point
2283 constant. TYPE must be a scalar or vector floating point type. CST must be
2284 of scalar or vector integer type. Both CST and TYPE must be scalars, or
2285 vectors of the same number of elements. If the value won't fit in the
2286 floating point type, the results are undefined.</dd>
2288 <dt><b><tt>ptrtoint ( CST to TYPE )</tt></b></dt>
2289 <dd>Convert a pointer typed constant to the corresponding integer constant
2290 <tt>TYPE</tt> must be an integer type. <tt>CST</tt> must be of pointer
2291 type. The <tt>CST</tt> value is zero extended, truncated, or unchanged to
2292 make it fit in <tt>TYPE</tt>.</dd>
2294 <dt><b><tt>inttoptr ( CST to TYPE )</tt></b></dt>
2295 <dd>Convert a integer constant to a pointer constant. TYPE must be a pointer
2296 type. CST must be of integer type. The CST value is zero extended,
2297 truncated, or unchanged to make it fit in a pointer size. This one is
2298 <i>really</i> dangerous!</dd>
2300 <dt><b><tt>bitcast ( CST to TYPE )</tt></b></dt>
2301 <dd>Convert a constant, CST, to another TYPE. The constraints of the operands
2302 are the same as those for the <a href="#i_bitcast">bitcast
2303 instruction</a>.</dd>
2305 <dt><b><tt>getelementptr ( CSTPTR, IDX0, IDX1, ... )</tt></b></dt>
2306 <dt><b><tt>getelementptr inbounds ( CSTPTR, IDX0, IDX1, ... )</tt></b></dt>
2307 <dd>Perform the <a href="#i_getelementptr">getelementptr operation</a> on
2308 constants. As with the <a href="#i_getelementptr">getelementptr</a>
2309 instruction, the index list may have zero or more indexes, which are
2310 required to make sense for the type of "CSTPTR".</dd>
2312 <dt><b><tt>select ( COND, VAL1, VAL2 )</tt></b></dt>
2313 <dd>Perform the <a href="#i_select">select operation</a> on constants.</dd>
2315 <dt><b><tt>icmp COND ( VAL1, VAL2 )</tt></b></dt>
2316 <dd>Performs the <a href="#i_icmp">icmp operation</a> on constants.</dd>
2318 <dt><b><tt>fcmp COND ( VAL1, VAL2 )</tt></b></dt>
2319 <dd>Performs the <a href="#i_fcmp">fcmp operation</a> on constants.</dd>
2321 <dt><b><tt>extractelement ( VAL, IDX )</tt></b></dt>
2322 <dd>Perform the <a href="#i_extractelement">extractelement operation</a> on
2325 <dt><b><tt>insertelement ( VAL, ELT, IDX )</tt></b></dt>
2326 <dd>Perform the <a href="#i_insertelement">insertelement operation</a> on
2329 <dt><b><tt>shufflevector ( VEC1, VEC2, IDXMASK )</tt></b></dt>
2330 <dd>Perform the <a href="#i_shufflevector">shufflevector operation</a> on
2333 <dt><b><tt>OPCODE ( LHS, RHS )</tt></b></dt>
2334 <dd>Perform the specified operation of the LHS and RHS constants. OPCODE may
2335 be any of the <a href="#binaryops">binary</a>
2336 or <a href="#bitwiseops">bitwise binary</a> operations. The constraints
2337 on operands are the same as those for the corresponding instruction
2338 (e.g. no bitwise operations on floating point values are allowed).</dd>
2343 <!-- ======================================================================= -->
2344 <div class="doc_subsection"><a name="metadata">Metadata Nodes and Metadata Strings</a>
2347 <div class="doc_text">
2349 <p>Metadata provides a way to attach arbitrary data to the instruction
2350 stream without affecting the behaviour of the program. There are two
2351 metadata primitives, strings and nodes. All metadata has the
2352 <tt>metadata</tt> type and is identified in syntax by a preceding exclamation
2353 point ('<tt>!</tt>').</p>
2355 <p>A metadata string is a string surrounded by double quotes. It can contain
2356 any character by escaping non-printable characters with "\xx" where "xx" is
2357 the two digit hex code. For example: "<tt>!"test\00"</tt>".</p>
2359 <p>Metadata nodes are represented with notation similar to structure constants
2360 (a comma separated list of elements, surrounded by braces and preceded by an
2361 exclamation point). For example: "<tt>!{ metadata !"test\00", i32
2364 <p>A metadata node will attempt to track changes to the values it holds. In the
2365 event that a value is deleted, it will be replaced with a typeless
2366 "<tt>null</tt>", such as "<tt>metadata !{null, i32 10}</tt>".</p>
2368 <p>A <a href="#namedmetadatastructure">named metadata</a> is a collection of
2369 metadata nodes. For example: "<tt>!foo = metadata !{!4, !3}</tt>".
2371 <p>Optimizations may rely on metadata to provide additional information about
2372 the program that isn't available in the instructions, or that isn't easily
2373 computable. Similarly, the code generator may expect a certain metadata
2374 format to be used to express debugging information.</p>
2378 <!-- *********************************************************************** -->
2379 <div class="doc_section"> <a name="othervalues">Other Values</a> </div>
2380 <!-- *********************************************************************** -->
2382 <!-- ======================================================================= -->
2383 <div class="doc_subsection">
2384 <a name="inlineasm">Inline Assembler Expressions</a>
2387 <div class="doc_text">
2389 <p>LLVM supports inline assembler expressions (as opposed
2390 to <a href="#moduleasm"> Module-Level Inline Assembly</a>) through the use of
2391 a special value. This value represents the inline assembler as a string
2392 (containing the instructions to emit), a list of operand constraints (stored
2393 as a string), a flag that indicates whether or not the inline asm
2394 expression has side effects, and a flag indicating whether the function
2395 containing the asm needs to align its stack conservatively. An example
2396 inline assembler expression is:</p>
2398 <div class="doc_code">
2400 i32 (i32) asm "bswap $0", "=r,r"
2404 <p>Inline assembler expressions may <b>only</b> be used as the callee operand of
2405 a <a href="#i_call"><tt>call</tt> instruction</a>. Thus, typically we
2408 <div class="doc_code">
2410 %X = call i32 asm "<a href="#int_bswap">bswap</a> $0", "=r,r"(i32 %Y)
2414 <p>Inline asms with side effects not visible in the constraint list must be
2415 marked as having side effects. This is done through the use of the
2416 '<tt>sideeffect</tt>' keyword, like so:</p>
2418 <div class="doc_code">
2420 call void asm sideeffect "eieio", ""()
2424 <p>In some cases inline asms will contain code that will not work unless the
2425 stack is aligned in some way, such as calls or SSE instructions on x86,
2426 yet will not contain code that does that alignment within the asm.
2427 The compiler should make conservative assumptions about what the asm might
2428 contain and should generate its usual stack alignment code in the prologue
2429 if the '<tt>alignstack</tt>' keyword is present:</p>
2431 <div class="doc_code">
2433 call void asm alignstack "eieio", ""()
2437 <p>If both keywords appear the '<tt>sideeffect</tt>' keyword must come
2440 <p>TODO: The format of the asm and constraints string still need to be
2441 documented here. Constraints on what can be done (e.g. duplication, moving,
2442 etc need to be documented). This is probably best done by reference to
2443 another document that covers inline asm from a holistic perspective.</p>
2448 <!-- *********************************************************************** -->
2449 <div class="doc_section">
2450 <a name="intrinsic_globals">Intrinsic Global Variables</a>
2452 <!-- *********************************************************************** -->
2454 <p>LLVM has a number of "magic" global variables that contain data that affect
2455 code generation or other IR semantics. These are documented here. All globals
2456 of this sort should have a section specified as "<tt>llvm.metadata</tt>". This
2457 section and all globals that start with "<tt>llvm.</tt>" are reserved for use
2460 <!-- ======================================================================= -->
2461 <div class="doc_subsection">
2462 <a name="intg_used">The '<tt>llvm.used</tt>' Global Variable</a>
2465 <div class="doc_text">
2467 <p>The <tt>@llvm.used</tt> global is an array with i8* element type which has <a
2468 href="#linkage_appending">appending linkage</a>. This array contains a list of
2469 pointers to global variables and functions which may optionally have a pointer
2470 cast formed of bitcast or getelementptr. For example, a legal use of it is:</p>
2476 @llvm.used = appending global [2 x i8*] [
2478 i8* bitcast (i32* @Y to i8*)
2479 ], section "llvm.metadata"
2482 <p>If a global variable appears in the <tt>@llvm.used</tt> list, then the
2483 compiler, assembler, and linker are required to treat the symbol as if there is
2484 a reference to the global that it cannot see. For example, if a variable has
2485 internal linkage and no references other than that from the <tt>@llvm.used</tt>
2486 list, it cannot be deleted. This is commonly used to represent references from
2487 inline asms and other things the compiler cannot "see", and corresponds to
2488 "attribute((used))" in GNU C.</p>
2490 <p>On some targets, the code generator must emit a directive to the assembler or
2491 object file to prevent the assembler and linker from molesting the symbol.</p>
2495 <!-- ======================================================================= -->
2496 <div class="doc_subsection">
2497 <a name="intg_compiler_used">The '<tt>llvm.compiler.used</tt>' Global Variable</a>
2500 <div class="doc_text">
2502 <p>The <tt>@llvm.compiler.used</tt> directive is the same as the
2503 <tt>@llvm.used</tt> directive, except that it only prevents the compiler from
2504 touching the symbol. On targets that support it, this allows an intelligent
2505 linker to optimize references to the symbol without being impeded as it would be
2506 by <tt>@llvm.used</tt>.</p>
2508 <p>This is a rare construct that should only be used in rare circumstances, and
2509 should not be exposed to source languages.</p>
2513 <!-- ======================================================================= -->
2514 <div class="doc_subsection">
2515 <a name="intg_global_ctors">The '<tt>llvm.global_ctors</tt>' Global Variable</a>
2518 <div class="doc_text">
2520 <p>TODO: Describe this.</p>
2524 <!-- ======================================================================= -->
2525 <div class="doc_subsection">
2526 <a name="intg_global_dtors">The '<tt>llvm.global_dtors</tt>' Global Variable</a>
2529 <div class="doc_text">
2531 <p>TODO: Describe this.</p>
2536 <!-- *********************************************************************** -->
2537 <div class="doc_section"> <a name="instref">Instruction Reference</a> </div>
2538 <!-- *********************************************************************** -->
2540 <div class="doc_text">
2542 <p>The LLVM instruction set consists of several different classifications of
2543 instructions: <a href="#terminators">terminator
2544 instructions</a>, <a href="#binaryops">binary instructions</a>,
2545 <a href="#bitwiseops">bitwise binary instructions</a>,
2546 <a href="#memoryops">memory instructions</a>, and
2547 <a href="#otherops">other instructions</a>.</p>
2551 <!-- ======================================================================= -->
2552 <div class="doc_subsection"> <a name="terminators">Terminator
2553 Instructions</a> </div>
2555 <div class="doc_text">
2557 <p>As mentioned <a href="#functionstructure">previously</a>, every basic block
2558 in a program ends with a "Terminator" instruction, which indicates which
2559 block should be executed after the current block is finished. These
2560 terminator instructions typically yield a '<tt>void</tt>' value: they produce
2561 control flow, not values (the one exception being the
2562 '<a href="#i_invoke"><tt>invoke</tt></a>' instruction).</p>
2564 <p>There are six different terminator instructions: the
2565 '<a href="#i_ret"><tt>ret</tt></a>' instruction, the
2566 '<a href="#i_br"><tt>br</tt></a>' instruction, the
2567 '<a href="#i_switch"><tt>switch</tt></a>' instruction, the
2568 '<a href="#i_indirectbr">'<tt>indirectbr</tt></a>' Instruction, the
2569 '<a href="#i_invoke"><tt>invoke</tt></a>' instruction, the
2570 '<a href="#i_unwind"><tt>unwind</tt></a>' instruction, and the
2571 '<a href="#i_unreachable"><tt>unreachable</tt></a>' instruction.</p>
2575 <!-- _______________________________________________________________________ -->
2576 <div class="doc_subsubsection"> <a name="i_ret">'<tt>ret</tt>'
2577 Instruction</a> </div>
2579 <div class="doc_text">
2583 ret <type> <value> <i>; Return a value from a non-void function</i>
2584 ret void <i>; Return from void function</i>
2588 <p>The '<tt>ret</tt>' instruction is used to return control flow (and optionally
2589 a value) from a function back to the caller.</p>
2591 <p>There are two forms of the '<tt>ret</tt>' instruction: one that returns a
2592 value and then causes control flow, and one that just causes control flow to
2596 <p>The '<tt>ret</tt>' instruction optionally accepts a single argument, the
2597 return value. The type of the return value must be a
2598 '<a href="#t_firstclass">first class</a>' type.</p>
2600 <p>A function is not <a href="#wellformed">well formed</a> if it it has a
2601 non-void return type and contains a '<tt>ret</tt>' instruction with no return
2602 value or a return value with a type that does not match its type, or if it
2603 has a void return type and contains a '<tt>ret</tt>' instruction with a
2607 <p>When the '<tt>ret</tt>' instruction is executed, control flow returns back to
2608 the calling function's context. If the caller is a
2609 "<a href="#i_call"><tt>call</tt></a>" instruction, execution continues at the
2610 instruction after the call. If the caller was an
2611 "<a href="#i_invoke"><tt>invoke</tt></a>" instruction, execution continues at
2612 the beginning of the "normal" destination block. If the instruction returns
2613 a value, that value shall set the call or invoke instruction's return
2618 ret i32 5 <i>; Return an integer value of 5</i>
2619 ret void <i>; Return from a void function</i>
2620 ret { i32, i8 } { i32 4, i8 2 } <i>; Return a struct of values 4 and 2</i>
2624 <!-- _______________________________________________________________________ -->
2625 <div class="doc_subsubsection"> <a name="i_br">'<tt>br</tt>' Instruction</a> </div>
2627 <div class="doc_text">
2631 br i1 <cond>, label <iftrue>, label <iffalse><br> br label <dest> <i>; Unconditional branch</i>
2635 <p>The '<tt>br</tt>' instruction is used to cause control flow to transfer to a
2636 different basic block in the current function. There are two forms of this
2637 instruction, corresponding to a conditional branch and an unconditional
2641 <p>The conditional branch form of the '<tt>br</tt>' instruction takes a single
2642 '<tt>i1</tt>' value and two '<tt>label</tt>' values. The unconditional form
2643 of the '<tt>br</tt>' instruction takes a single '<tt>label</tt>' value as a
2647 <p>Upon execution of a conditional '<tt>br</tt>' instruction, the '<tt>i1</tt>'
2648 argument is evaluated. If the value is <tt>true</tt>, control flows to the
2649 '<tt>iftrue</tt>' <tt>label</tt> argument. If "cond" is <tt>false</tt>,
2650 control flows to the '<tt>iffalse</tt>' <tt>label</tt> argument.</p>
2655 %cond = <a href="#i_icmp">icmp</a> eq i32 %a, %b
2656 br i1 %cond, label %IfEqual, label %IfUnequal
2658 <a href="#i_ret">ret</a> i32 1
2660 <a href="#i_ret">ret</a> i32 0
2665 <!-- _______________________________________________________________________ -->
2666 <div class="doc_subsubsection">
2667 <a name="i_switch">'<tt>switch</tt>' Instruction</a>
2670 <div class="doc_text">
2674 switch <intty> <value>, label <defaultdest> [ <intty> <val>, label <dest> ... ]
2678 <p>The '<tt>switch</tt>' instruction is used to transfer control flow to one of
2679 several different places. It is a generalization of the '<tt>br</tt>'
2680 instruction, allowing a branch to occur to one of many possible
2684 <p>The '<tt>switch</tt>' instruction uses three parameters: an integer
2685 comparison value '<tt>value</tt>', a default '<tt>label</tt>' destination,
2686 and an array of pairs of comparison value constants and '<tt>label</tt>'s.
2687 The table is not allowed to contain duplicate constant entries.</p>
2690 <p>The <tt>switch</tt> instruction specifies a table of values and
2691 destinations. When the '<tt>switch</tt>' instruction is executed, this table
2692 is searched for the given value. If the value is found, control flow is
2693 transferred to the corresponding destination; otherwise, control flow is
2694 transferred to the default destination.</p>
2696 <h5>Implementation:</h5>
2697 <p>Depending on properties of the target machine and the particular
2698 <tt>switch</tt> instruction, this instruction may be code generated in
2699 different ways. For example, it could be generated as a series of chained
2700 conditional branches or with a lookup table.</p>
2704 <i>; Emulate a conditional br instruction</i>
2705 %Val = <a href="#i_zext">zext</a> i1 %value to i32
2706 switch i32 %Val, label %truedest [ i32 0, label %falsedest ]
2708 <i>; Emulate an unconditional br instruction</i>
2709 switch i32 0, label %dest [ ]
2711 <i>; Implement a jump table:</i>
2712 switch i32 %val, label %otherwise [ i32 0, label %onzero
2714 i32 2, label %ontwo ]
2720 <!-- _______________________________________________________________________ -->
2721 <div class="doc_subsubsection">
2722 <a name="i_indirectbr">'<tt>indirectbr</tt>' Instruction</a>
2725 <div class="doc_text">
2729 indirectbr <somety>* <address>, [ label <dest1>, label <dest2>, ... ]
2734 <p>The '<tt>indirectbr</tt>' instruction implements an indirect branch to a label
2735 within the current function, whose address is specified by
2736 "<tt>address</tt>". Address must be derived from a <a
2737 href="#blockaddress">blockaddress</a> constant.</p>
2741 <p>The '<tt>address</tt>' argument is the address of the label to jump to. The
2742 rest of the arguments indicate the full set of possible destinations that the
2743 address may point to. Blocks are allowed to occur multiple times in the
2744 destination list, though this isn't particularly useful.</p>
2746 <p>This destination list is required so that dataflow analysis has an accurate
2747 understanding of the CFG.</p>
2751 <p>Control transfers to the block specified in the address argument. All
2752 possible destination blocks must be listed in the label list, otherwise this
2753 instruction has undefined behavior. This implies that jumps to labels
2754 defined in other functions have undefined behavior as well.</p>
2756 <h5>Implementation:</h5>
2758 <p>This is typically implemented with a jump through a register.</p>
2762 indirectbr i8* %Addr, [ label %bb1, label %bb2, label %bb3 ]
2768 <!-- _______________________________________________________________________ -->
2769 <div class="doc_subsubsection">
2770 <a name="i_invoke">'<tt>invoke</tt>' Instruction</a>
2773 <div class="doc_text">
2777 <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>]
2778 to label <normal label> unwind label <exception label>
2782 <p>The '<tt>invoke</tt>' instruction causes control to transfer to a specified
2783 function, with the possibility of control flow transfer to either the
2784 '<tt>normal</tt>' label or the '<tt>exception</tt>' label. If the callee
2785 function returns with the "<tt><a href="#i_ret">ret</a></tt>" instruction,
2786 control flow will return to the "normal" label. If the callee (or any
2787 indirect callees) returns with the "<a href="#i_unwind"><tt>unwind</tt></a>"
2788 instruction, control is interrupted and continued at the dynamically nearest
2789 "exception" label.</p>
2792 <p>This instruction requires several arguments:</p>
2795 <li>The optional "cconv" marker indicates which <a href="#callingconv">calling
2796 convention</a> the call should use. If none is specified, the call
2797 defaults to using C calling conventions.</li>
2799 <li>The optional <a href="#paramattrs">Parameter Attributes</a> list for
2800 return values. Only '<tt>zeroext</tt>', '<tt>signext</tt>', and
2801 '<tt>inreg</tt>' attributes are valid here.</li>
2803 <li>'<tt>ptr to function ty</tt>': shall be the signature of the pointer to
2804 function value being invoked. In most cases, this is a direct function
2805 invocation, but indirect <tt>invoke</tt>s are just as possible, branching
2806 off an arbitrary pointer to function value.</li>
2808 <li>'<tt>function ptr val</tt>': An LLVM value containing a pointer to a
2809 function to be invoked. </li>
2811 <li>'<tt>function args</tt>': argument list whose types match the function
2812 signature argument types. If the function signature indicates the
2813 function accepts a variable number of arguments, the extra arguments can
2816 <li>'<tt>normal label</tt>': the label reached when the called function
2817 executes a '<tt><a href="#i_ret">ret</a></tt>' instruction. </li>
2819 <li>'<tt>exception label</tt>': the label reached when a callee returns with
2820 the <a href="#i_unwind"><tt>unwind</tt></a> instruction. </li>
2822 <li>The optional <a href="#fnattrs">function attributes</a> list. Only
2823 '<tt>noreturn</tt>', '<tt>nounwind</tt>', '<tt>readonly</tt>' and
2824 '<tt>readnone</tt>' attributes are valid here.</li>
2828 <p>This instruction is designed to operate as a standard
2829 '<tt><a href="#i_call">call</a></tt>' instruction in most regards. The
2830 primary difference is that it establishes an association with a label, which
2831 is used by the runtime library to unwind the stack.</p>
2833 <p>This instruction is used in languages with destructors to ensure that proper
2834 cleanup is performed in the case of either a <tt>longjmp</tt> or a thrown
2835 exception. Additionally, this is important for implementation of
2836 '<tt>catch</tt>' clauses in high-level languages that support them.</p>
2838 <p>For the purposes of the SSA form, the definition of the value returned by the
2839 '<tt>invoke</tt>' instruction is deemed to occur on the edge from the current
2840 block to the "normal" label. If the callee unwinds then no return value is
2843 <p>Note that the code generator does not yet completely support unwind, and
2844 that the invoke/unwind semantics are likely to change in future versions.</p>
2848 %retval = invoke i32 @Test(i32 15) to label %Continue
2849 unwind label %TestCleanup <i>; {i32}:retval set</i>
2850 %retval = invoke <a href="#callingconv">coldcc</a> i32 %Testfnptr(i32 15) to label %Continue
2851 unwind label %TestCleanup <i>; {i32}:retval set</i>
2856 <!-- _______________________________________________________________________ -->
2858 <div class="doc_subsubsection"> <a name="i_unwind">'<tt>unwind</tt>'
2859 Instruction</a> </div>
2861 <div class="doc_text">
2869 <p>The '<tt>unwind</tt>' instruction unwinds the stack, continuing control flow
2870 at the first callee in the dynamic call stack which used
2871 an <a href="#i_invoke"><tt>invoke</tt></a> instruction to perform the call.
2872 This is primarily used to implement exception handling.</p>
2875 <p>The '<tt>unwind</tt>' instruction causes execution of the current function to
2876 immediately halt. The dynamic call stack is then searched for the
2877 first <a href="#i_invoke"><tt>invoke</tt></a> instruction on the call stack.
2878 Once found, execution continues at the "exceptional" destination block
2879 specified by the <tt>invoke</tt> instruction. If there is no <tt>invoke</tt>
2880 instruction in the dynamic call chain, undefined behavior results.</p>
2882 <p>Note that the code generator does not yet completely support unwind, and
2883 that the invoke/unwind semantics are likely to change in future versions.</p>
2887 <!-- _______________________________________________________________________ -->
2889 <div class="doc_subsubsection"> <a name="i_unreachable">'<tt>unreachable</tt>'
2890 Instruction</a> </div>
2892 <div class="doc_text">
2900 <p>The '<tt>unreachable</tt>' instruction has no defined semantics. This
2901 instruction is used to inform the optimizer that a particular portion of the
2902 code is not reachable. This can be used to indicate that the code after a
2903 no-return function cannot be reached, and other facts.</p>
2906 <p>The '<tt>unreachable</tt>' instruction has no defined semantics.</p>
2910 <!-- ======================================================================= -->
2911 <div class="doc_subsection"> <a name="binaryops">Binary Operations</a> </div>
2913 <div class="doc_text">
2915 <p>Binary operators are used to do most of the computation in a program. They
2916 require two operands of the same type, execute an operation on them, and
2917 produce a single value. The operands might represent multiple data, as is
2918 the case with the <a href="#t_vector">vector</a> data type. The result value
2919 has the same type as its operands.</p>
2921 <p>There are several different binary operators:</p>
2925 <!-- _______________________________________________________________________ -->
2926 <div class="doc_subsubsection">
2927 <a name="i_add">'<tt>add</tt>' Instruction</a>
2930 <div class="doc_text">
2934 <result> = add <ty> <op1>, <op2> <i>; yields {ty}:result</i>
2935 <result> = add nuw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
2936 <result> = add nsw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
2937 <result> = add nuw nsw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
2941 <p>The '<tt>add</tt>' instruction returns the sum of its two operands.</p>
2944 <p>The two arguments to the '<tt>add</tt>' instruction must
2945 be <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
2946 integer values. Both arguments must have identical types.</p>
2949 <p>The value produced is the integer sum of the two operands.</p>
2951 <p>If the sum has unsigned overflow, the result returned is the mathematical
2952 result modulo 2<sup>n</sup>, where n is the bit width of the result.</p>
2954 <p>Because LLVM integers use a two's complement representation, this instruction
2955 is appropriate for both signed and unsigned integers.</p>
2957 <p><tt>nuw</tt> and <tt>nsw</tt> stand for "No Unsigned Wrap"
2958 and "No Signed Wrap", respectively. If the <tt>nuw</tt> and/or
2959 <tt>nsw</tt> keywords are present, the result value of the <tt>add</tt>
2960 is undefined if unsigned and/or signed overflow, respectively, occurs.</p>
2964 <result> = add i32 4, %var <i>; yields {i32}:result = 4 + %var</i>
2969 <!-- _______________________________________________________________________ -->
2970 <div class="doc_subsubsection">
2971 <a name="i_fadd">'<tt>fadd</tt>' Instruction</a>
2974 <div class="doc_text">
2978 <result> = fadd <ty> <op1>, <op2> <i>; yields {ty}:result</i>
2982 <p>The '<tt>fadd</tt>' instruction returns the sum of its two operands.</p>
2985 <p>The two arguments to the '<tt>fadd</tt>' instruction must be
2986 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
2987 floating point values. Both arguments must have identical types.</p>
2990 <p>The value produced is the floating point sum of the two operands.</p>
2994 <result> = fadd float 4.0, %var <i>; yields {float}:result = 4.0 + %var</i>
2999 <!-- _______________________________________________________________________ -->
3000 <div class="doc_subsubsection">
3001 <a name="i_sub">'<tt>sub</tt>' Instruction</a>
3004 <div class="doc_text">
3008 <result> = sub <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3009 <result> = sub nuw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3010 <result> = sub nsw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3011 <result> = sub nuw nsw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3015 <p>The '<tt>sub</tt>' instruction returns the difference of its two
3018 <p>Note that the '<tt>sub</tt>' instruction is used to represent the
3019 '<tt>neg</tt>' instruction present in most other intermediate
3020 representations.</p>
3023 <p>The two arguments to the '<tt>sub</tt>' instruction must
3024 be <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
3025 integer values. Both arguments must have identical types.</p>
3028 <p>The value produced is the integer difference of the two operands.</p>
3030 <p>If the difference has unsigned overflow, the result returned is the
3031 mathematical result modulo 2<sup>n</sup>, where n is the bit width of the
3034 <p>Because LLVM integers use a two's complement representation, this instruction
3035 is appropriate for both signed and unsigned integers.</p>
3037 <p><tt>nuw</tt> and <tt>nsw</tt> stand for "No Unsigned Wrap"
3038 and "No Signed Wrap", respectively. If the <tt>nuw</tt> and/or
3039 <tt>nsw</tt> keywords are present, the result value of the <tt>sub</tt>
3040 is undefined if unsigned and/or signed overflow, respectively, occurs.</p>
3044 <result> = sub i32 4, %var <i>; yields {i32}:result = 4 - %var</i>
3045 <result> = sub i32 0, %val <i>; yields {i32}:result = -%var</i>
3050 <!-- _______________________________________________________________________ -->
3051 <div class="doc_subsubsection">
3052 <a name="i_fsub">'<tt>fsub</tt>' Instruction</a>
3055 <div class="doc_text">
3059 <result> = fsub <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3063 <p>The '<tt>fsub</tt>' instruction returns the difference of its two
3066 <p>Note that the '<tt>fsub</tt>' instruction is used to represent the
3067 '<tt>fneg</tt>' instruction present in most other intermediate
3068 representations.</p>
3071 <p>The two arguments to the '<tt>fsub</tt>' instruction must be
3072 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3073 floating point values. Both arguments must have identical types.</p>
3076 <p>The value produced is the floating point difference of the two operands.</p>
3080 <result> = fsub float 4.0, %var <i>; yields {float}:result = 4.0 - %var</i>
3081 <result> = fsub float -0.0, %val <i>; yields {float}:result = -%var</i>
3086 <!-- _______________________________________________________________________ -->
3087 <div class="doc_subsubsection">
3088 <a name="i_mul">'<tt>mul</tt>' Instruction</a>
3091 <div class="doc_text">
3095 <result> = mul <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3096 <result> = mul nuw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3097 <result> = mul nsw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3098 <result> = mul nuw nsw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3102 <p>The '<tt>mul</tt>' instruction returns the product of its two operands.</p>
3105 <p>The two arguments to the '<tt>mul</tt>' instruction must
3106 be <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
3107 integer values. Both arguments must have identical types.</p>
3110 <p>The value produced is the integer product of the two operands.</p>
3112 <p>If the result of the multiplication has unsigned overflow, the result
3113 returned is the mathematical result modulo 2<sup>n</sup>, where n is the bit
3114 width of the result.</p>
3116 <p>Because LLVM integers use a two's complement representation, and the result
3117 is the same width as the operands, this instruction returns the correct
3118 result for both signed and unsigned integers. If a full product
3119 (e.g. <tt>i32</tt>x<tt>i32</tt>-><tt>i64</tt>) is needed, the operands should
3120 be sign-extended or zero-extended as appropriate to the width of the full
3123 <p><tt>nuw</tt> and <tt>nsw</tt> stand for "No Unsigned Wrap"
3124 and "No Signed Wrap", respectively. If the <tt>nuw</tt> and/or
3125 <tt>nsw</tt> keywords are present, the result value of the <tt>mul</tt>
3126 is undefined if unsigned and/or signed overflow, respectively, occurs.</p>
3130 <result> = mul i32 4, %var <i>; yields {i32}:result = 4 * %var</i>
3135 <!-- _______________________________________________________________________ -->
3136 <div class="doc_subsubsection">
3137 <a name="i_fmul">'<tt>fmul</tt>' Instruction</a>
3140 <div class="doc_text">
3144 <result> = fmul <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3148 <p>The '<tt>fmul</tt>' instruction returns the product of its two operands.</p>
3151 <p>The two arguments to the '<tt>fmul</tt>' instruction must be
3152 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3153 floating point values. Both arguments must have identical types.</p>
3156 <p>The value produced is the floating point product of the two operands.</p>
3160 <result> = fmul float 4.0, %var <i>; yields {float}:result = 4.0 * %var</i>
3165 <!-- _______________________________________________________________________ -->
3166 <div class="doc_subsubsection"> <a name="i_udiv">'<tt>udiv</tt>' Instruction
3169 <div class="doc_text">
3173 <result> = udiv <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3177 <p>The '<tt>udiv</tt>' instruction returns the quotient of its two operands.</p>
3180 <p>The two arguments to the '<tt>udiv</tt>' instruction must be
3181 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3182 values. Both arguments must have identical types.</p>
3185 <p>The value produced is the unsigned integer quotient of the two operands.</p>
3187 <p>Note that unsigned integer division and signed integer division are distinct
3188 operations; for signed integer division, use '<tt>sdiv</tt>'.</p>
3190 <p>Division by zero leads to undefined behavior.</p>
3194 <result> = udiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
3199 <!-- _______________________________________________________________________ -->
3200 <div class="doc_subsubsection"> <a name="i_sdiv">'<tt>sdiv</tt>' Instruction
3203 <div class="doc_text">
3207 <result> = sdiv <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3208 <result> = sdiv exact <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3212 <p>The '<tt>sdiv</tt>' instruction returns the quotient of its two operands.</p>
3215 <p>The two arguments to the '<tt>sdiv</tt>' instruction must be
3216 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3217 values. Both arguments must have identical types.</p>
3220 <p>The value produced is the signed integer quotient of the two operands rounded
3223 <p>Note that signed integer division and unsigned integer division are distinct
3224 operations; for unsigned integer division, use '<tt>udiv</tt>'.</p>
3226 <p>Division by zero leads to undefined behavior. Overflow also leads to
3227 undefined behavior; this is a rare case, but can occur, for example, by doing
3228 a 32-bit division of -2147483648 by -1.</p>
3230 <p>If the <tt>exact</tt> keyword is present, the result value of the
3231 <tt>sdiv</tt> is undefined if the result would be rounded or if overflow
3236 <result> = sdiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
3241 <!-- _______________________________________________________________________ -->
3242 <div class="doc_subsubsection"> <a name="i_fdiv">'<tt>fdiv</tt>'
3243 Instruction</a> </div>
3245 <div class="doc_text">
3249 <result> = fdiv <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3253 <p>The '<tt>fdiv</tt>' instruction returns the quotient of its two operands.</p>
3256 <p>The two arguments to the '<tt>fdiv</tt>' instruction must be
3257 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3258 floating point values. Both arguments must have identical types.</p>
3261 <p>The value produced is the floating point quotient of the two operands.</p>
3265 <result> = fdiv float 4.0, %var <i>; yields {float}:result = 4.0 / %var</i>
3270 <!-- _______________________________________________________________________ -->
3271 <div class="doc_subsubsection"> <a name="i_urem">'<tt>urem</tt>' Instruction</a>
3274 <div class="doc_text">
3278 <result> = urem <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3282 <p>The '<tt>urem</tt>' instruction returns the remainder from the unsigned
3283 division of its two arguments.</p>
3286 <p>The two arguments to the '<tt>urem</tt>' instruction must be
3287 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3288 values. Both arguments must have identical types.</p>
3291 <p>This instruction returns the unsigned integer <i>remainder</i> of a division.
3292 This instruction always performs an unsigned division to get the
3295 <p>Note that unsigned integer remainder and signed integer remainder are
3296 distinct operations; for signed integer remainder, use '<tt>srem</tt>'.</p>
3298 <p>Taking the remainder of a division by zero leads to undefined behavior.</p>
3302 <result> = urem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
3307 <!-- _______________________________________________________________________ -->
3308 <div class="doc_subsubsection">
3309 <a name="i_srem">'<tt>srem</tt>' Instruction</a>
3312 <div class="doc_text">
3316 <result> = srem <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3320 <p>The '<tt>srem</tt>' instruction returns the remainder from the signed
3321 division of its two operands. This instruction can also take
3322 <a href="#t_vector">vector</a> versions of the values in which case the
3323 elements must be integers.</p>
3326 <p>The two arguments to the '<tt>srem</tt>' instruction must be
3327 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3328 values. Both arguments must have identical types.</p>
3331 <p>This instruction returns the <i>remainder</i> of a division (where the result
3332 has the same sign as the dividend, <tt>op1</tt>), not the <i>modulo</i>
3333 operator (where the result has the same sign as the divisor, <tt>op2</tt>) of
3334 a value. For more information about the difference,
3335 see <a href="http://mathforum.org/dr.math/problems/anne.4.28.99.html">The
3336 Math Forum</a>. For a table of how this is implemented in various languages,
3337 please see <a href="http://en.wikipedia.org/wiki/Modulo_operation">
3338 Wikipedia: modulo operation</a>.</p>
3340 <p>Note that signed integer remainder and unsigned integer remainder are
3341 distinct operations; for unsigned integer remainder, use '<tt>urem</tt>'.</p>
3343 <p>Taking the remainder of a division by zero leads to undefined behavior.
3344 Overflow also leads to undefined behavior; this is a rare case, but can
3345 occur, for example, by taking the remainder of a 32-bit division of
3346 -2147483648 by -1. (The remainder doesn't actually overflow, but this rule
3347 lets srem be implemented using instructions that return both the result of
3348 the division and the remainder.)</p>
3352 <result> = srem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
3357 <!-- _______________________________________________________________________ -->
3358 <div class="doc_subsubsection">
3359 <a name="i_frem">'<tt>frem</tt>' Instruction</a> </div>
3361 <div class="doc_text">
3365 <result> = frem <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3369 <p>The '<tt>frem</tt>' instruction returns the remainder from the division of
3370 its two operands.</p>
3373 <p>The two arguments to the '<tt>frem</tt>' instruction must be
3374 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3375 floating point values. Both arguments must have identical types.</p>
3378 <p>This instruction returns the <i>remainder</i> of a division. The remainder
3379 has the same sign as the dividend.</p>
3383 <result> = frem float 4.0, %var <i>; yields {float}:result = 4.0 % %var</i>
3388 <!-- ======================================================================= -->
3389 <div class="doc_subsection"> <a name="bitwiseops">Bitwise Binary
3390 Operations</a> </div>
3392 <div class="doc_text">
3394 <p>Bitwise binary operators are used to do various forms of bit-twiddling in a
3395 program. They are generally very efficient instructions and can commonly be
3396 strength reduced from other instructions. They require two operands of the
3397 same type, execute an operation on them, and produce a single value. The
3398 resulting value is the same type as its operands.</p>
3402 <!-- _______________________________________________________________________ -->
3403 <div class="doc_subsubsection"> <a name="i_shl">'<tt>shl</tt>'
3404 Instruction</a> </div>
3406 <div class="doc_text">
3410 <result> = shl <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3414 <p>The '<tt>shl</tt>' instruction returns the first operand shifted to the left
3415 a specified number of bits.</p>
3418 <p>Both arguments to the '<tt>shl</tt>' instruction must be the
3419 same <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
3420 integer type. '<tt>op2</tt>' is treated as an unsigned value.</p>
3423 <p>The value produced is <tt>op1</tt> * 2<sup><tt>op2</tt></sup> mod
3424 2<sup>n</sup>, where <tt>n</tt> is the width of the result. If <tt>op2</tt>
3425 is (statically or dynamically) negative or equal to or larger than the number
3426 of bits in <tt>op1</tt>, the result is undefined. If the arguments are
3427 vectors, each vector element of <tt>op1</tt> is shifted by the corresponding
3428 shift amount in <tt>op2</tt>.</p>
3432 <result> = shl i32 4, %var <i>; yields {i32}: 4 << %var</i>
3433 <result> = shl i32 4, 2 <i>; yields {i32}: 16</i>
3434 <result> = shl i32 1, 10 <i>; yields {i32}: 1024</i>
3435 <result> = shl i32 1, 32 <i>; undefined</i>
3436 <result> = shl <2 x i32> < i32 1, i32 1>, < i32 1, i32 2> <i>; yields: result=<2 x i32> < i32 2, i32 4></i>
3441 <!-- _______________________________________________________________________ -->
3442 <div class="doc_subsubsection"> <a name="i_lshr">'<tt>lshr</tt>'
3443 Instruction</a> </div>
3445 <div class="doc_text">
3449 <result> = lshr <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3453 <p>The '<tt>lshr</tt>' instruction (logical shift right) returns the first
3454 operand shifted to the right a specified number of bits with zero fill.</p>
3457 <p>Both arguments to the '<tt>lshr</tt>' instruction must be the same
3458 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3459 type. '<tt>op2</tt>' is treated as an unsigned value.</p>
3462 <p>This instruction always performs a logical shift right operation. The most
3463 significant bits of the result will be filled with zero bits after the shift.
3464 If <tt>op2</tt> is (statically or dynamically) equal to or larger than the
3465 number of bits in <tt>op1</tt>, the result is undefined. If the arguments are
3466 vectors, each vector element of <tt>op1</tt> is shifted by the corresponding
3467 shift amount in <tt>op2</tt>.</p>
3471 <result> = lshr i32 4, 1 <i>; yields {i32}:result = 2</i>
3472 <result> = lshr i32 4, 2 <i>; yields {i32}:result = 1</i>
3473 <result> = lshr i8 4, 3 <i>; yields {i8}:result = 0</i>
3474 <result> = lshr i8 -2, 1 <i>; yields {i8}:result = 0x7FFFFFFF </i>
3475 <result> = lshr i32 1, 32 <i>; undefined</i>
3476 <result> = lshr <2 x i32> < i32 -2, i32 4>, < i32 1, i32 2> <i>; yields: result=<2 x i32> < i32 0x7FFFFFFF, i32 1></i>
3481 <!-- _______________________________________________________________________ -->
3482 <div class="doc_subsubsection"> <a name="i_ashr">'<tt>ashr</tt>'
3483 Instruction</a> </div>
3484 <div class="doc_text">
3488 <result> = ashr <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3492 <p>The '<tt>ashr</tt>' instruction (arithmetic shift right) returns the first
3493 operand shifted to the right a specified number of bits with sign
3497 <p>Both arguments to the '<tt>ashr</tt>' instruction must be the same
3498 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3499 type. '<tt>op2</tt>' is treated as an unsigned value.</p>
3502 <p>This instruction always performs an arithmetic shift right operation, The
3503 most significant bits of the result will be filled with the sign bit
3504 of <tt>op1</tt>. If <tt>op2</tt> is (statically or dynamically) equal to or
3505 larger than the number of bits in <tt>op1</tt>, the result is undefined. If
3506 the arguments are vectors, each vector element of <tt>op1</tt> is shifted by
3507 the corresponding shift amount in <tt>op2</tt>.</p>
3511 <result> = ashr i32 4, 1 <i>; yields {i32}:result = 2</i>
3512 <result> = ashr i32 4, 2 <i>; yields {i32}:result = 1</i>
3513 <result> = ashr i8 4, 3 <i>; yields {i8}:result = 0</i>
3514 <result> = ashr i8 -2, 1 <i>; yields {i8}:result = -1</i>
3515 <result> = ashr i32 1, 32 <i>; undefined</i>
3516 <result> = ashr <2 x i32> < i32 -2, i32 4>, < i32 1, i32 3> <i>; yields: result=<2 x i32> < i32 -1, i32 0></i>
3521 <!-- _______________________________________________________________________ -->
3522 <div class="doc_subsubsection"> <a name="i_and">'<tt>and</tt>'
3523 Instruction</a> </div>
3525 <div class="doc_text">
3529 <result> = and <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3533 <p>The '<tt>and</tt>' instruction returns the bitwise logical and of its two
3537 <p>The two arguments to the '<tt>and</tt>' instruction must be
3538 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3539 values. Both arguments must have identical types.</p>
3542 <p>The truth table used for the '<tt>and</tt>' instruction is:</p>
3544 <table border="1" cellspacing="0" cellpadding="4">
3576 <result> = and i32 4, %var <i>; yields {i32}:result = 4 & %var</i>
3577 <result> = and i32 15, 40 <i>; yields {i32}:result = 8</i>
3578 <result> = and i32 4, 8 <i>; yields {i32}:result = 0</i>
3581 <!-- _______________________________________________________________________ -->
3582 <div class="doc_subsubsection"> <a name="i_or">'<tt>or</tt>' Instruction</a> </div>
3584 <div class="doc_text">
3588 <result> = or <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3592 <p>The '<tt>or</tt>' instruction returns the bitwise logical inclusive or of its
3596 <p>The two arguments to the '<tt>or</tt>' instruction must be
3597 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3598 values. Both arguments must have identical types.</p>
3601 <p>The truth table used for the '<tt>or</tt>' instruction is:</p>
3603 <table border="1" cellspacing="0" cellpadding="4">
3635 <result> = or i32 4, %var <i>; yields {i32}:result = 4 | %var</i>
3636 <result> = or i32 15, 40 <i>; yields {i32}:result = 47</i>
3637 <result> = or i32 4, 8 <i>; yields {i32}:result = 12</i>
3642 <!-- _______________________________________________________________________ -->
3643 <div class="doc_subsubsection"> <a name="i_xor">'<tt>xor</tt>'
3644 Instruction</a> </div>
3646 <div class="doc_text">
3650 <result> = xor <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3654 <p>The '<tt>xor</tt>' instruction returns the bitwise logical exclusive or of
3655 its two operands. The <tt>xor</tt> is used to implement the "one's
3656 complement" operation, which is the "~" operator in C.</p>
3659 <p>The two arguments to the '<tt>xor</tt>' instruction must be
3660 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3661 values. Both arguments must have identical types.</p>
3664 <p>The truth table used for the '<tt>xor</tt>' instruction is:</p>
3666 <table border="1" cellspacing="0" cellpadding="4">
3698 <result> = xor i32 4, %var <i>; yields {i32}:result = 4 ^ %var</i>
3699 <result> = xor i32 15, 40 <i>; yields {i32}:result = 39</i>
3700 <result> = xor i32 4, 8 <i>; yields {i32}:result = 12</i>
3701 <result> = xor i32 %V, -1 <i>; yields {i32}:result = ~%V</i>
3706 <!-- ======================================================================= -->
3707 <div class="doc_subsection">
3708 <a name="vectorops">Vector Operations</a>
3711 <div class="doc_text">
3713 <p>LLVM supports several instructions to represent vector operations in a
3714 target-independent manner. These instructions cover the element-access and
3715 vector-specific operations needed to process vectors effectively. While LLVM
3716 does directly support these vector operations, many sophisticated algorithms
3717 will want to use target-specific intrinsics to take full advantage of a
3718 specific target.</p>
3722 <!-- _______________________________________________________________________ -->
3723 <div class="doc_subsubsection">
3724 <a name="i_extractelement">'<tt>extractelement</tt>' Instruction</a>
3727 <div class="doc_text">
3731 <result> = extractelement <n x <ty>> <val>, i32 <idx> <i>; yields <ty></i>
3735 <p>The '<tt>extractelement</tt>' instruction extracts a single scalar element
3736 from a vector at a specified index.</p>
3740 <p>The first operand of an '<tt>extractelement</tt>' instruction is a value
3741 of <a href="#t_vector">vector</a> type. The second operand is an index
3742 indicating the position from which to extract the element. The index may be
3746 <p>The result is a scalar of the same type as the element type of
3747 <tt>val</tt>. Its value is the value at position <tt>idx</tt> of
3748 <tt>val</tt>. If <tt>idx</tt> exceeds the length of <tt>val</tt>, the
3749 results are undefined.</p>
3753 <result> = extractelement <4 x i32> %vec, i32 0 <i>; yields i32</i>
3758 <!-- _______________________________________________________________________ -->
3759 <div class="doc_subsubsection">
3760 <a name="i_insertelement">'<tt>insertelement</tt>' Instruction</a>
3763 <div class="doc_text">
3767 <result> = insertelement <n x <ty>> <val>, <ty> <elt>, i32 <idx> <i>; yields <n x <ty>></i>
3771 <p>The '<tt>insertelement</tt>' instruction inserts a scalar element into a
3772 vector at a specified index.</p>
3775 <p>The first operand of an '<tt>insertelement</tt>' instruction is a value
3776 of <a href="#t_vector">vector</a> type. The second operand is a scalar value
3777 whose type must equal the element type of the first operand. The third
3778 operand is an index indicating the position at which to insert the value.
3779 The index may be a variable.</p>
3782 <p>The result is a vector of the same type as <tt>val</tt>. Its element values
3783 are those of <tt>val</tt> except at position <tt>idx</tt>, where it gets the
3784 value <tt>elt</tt>. If <tt>idx</tt> exceeds the length of <tt>val</tt>, the
3785 results are undefined.</p>
3789 <result> = insertelement <4 x i32> %vec, i32 1, i32 0 <i>; yields <4 x i32></i>
3794 <!-- _______________________________________________________________________ -->
3795 <div class="doc_subsubsection">
3796 <a name="i_shufflevector">'<tt>shufflevector</tt>' Instruction</a>
3799 <div class="doc_text">
3803 <result> = shufflevector <n x <ty>> <v1>, <n x <ty>> <v2>, <m x i32> <mask> <i>; yields <m x <ty>></i>
3807 <p>The '<tt>shufflevector</tt>' instruction constructs a permutation of elements
3808 from two input vectors, returning a vector with the same element type as the
3809 input and length that is the same as the shuffle mask.</p>
3812 <p>The first two operands of a '<tt>shufflevector</tt>' instruction are vectors
3813 with types that match each other. The third argument is a shuffle mask whose
3814 element type is always 'i32'. The result of the instruction is a vector
3815 whose length is the same as the shuffle mask and whose element type is the
3816 same as the element type of the first two operands.</p>
3818 <p>The shuffle mask operand is required to be a constant vector with either
3819 constant integer or undef values.</p>
3822 <p>The elements of the two input vectors are numbered from left to right across
3823 both of the vectors. The shuffle mask operand specifies, for each element of
3824 the result vector, which element of the two input vectors the result element
3825 gets. The element selector may be undef (meaning "don't care") and the
3826 second operand may be undef if performing a shuffle from only one vector.</p>
3830 <result> = shufflevector <4 x i32> %v1, <4 x i32> %v2,
3831 <4 x i32> <i32 0, i32 4, i32 1, i32 5> <i>; yields <4 x i32></i>
3832 <result> = shufflevector <4 x i32> %v1, <4 x i32> undef,
3833 <4 x i32> <i32 0, i32 1, i32 2, i32 3> <i>; yields <4 x i32></i> - Identity shuffle.
3834 <result> = shufflevector <8 x i32> %v1, <8 x i32> undef,
3835 <4 x i32> <i32 0, i32 1, i32 2, i32 3> <i>; yields <4 x i32></i>
3836 <result> = shufflevector <4 x i32> %v1, <4 x i32> %v2,
3837 <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>
3842 <!-- ======================================================================= -->
3843 <div class="doc_subsection">
3844 <a name="aggregateops">Aggregate Operations</a>
3847 <div class="doc_text">
3849 <p>LLVM supports several instructions for working with aggregate values.</p>
3853 <!-- _______________________________________________________________________ -->
3854 <div class="doc_subsubsection">
3855 <a name="i_extractvalue">'<tt>extractvalue</tt>' Instruction</a>
3858 <div class="doc_text">
3862 <result> = extractvalue <aggregate type> <val>, <idx>{, <idx>}*
3866 <p>The '<tt>extractvalue</tt>' instruction extracts the value of a struct field
3867 or array element from an aggregate value.</p>
3870 <p>The first operand of an '<tt>extractvalue</tt>' instruction is a value
3871 of <a href="#t_struct">struct</a> or <a href="#t_array">array</a> type. The
3872 operands are constant indices to specify which value to extract in a similar
3873 manner as indices in a
3874 '<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.</p>
3877 <p>The result is the value at the position in the aggregate specified by the
3882 <result> = extractvalue {i32, float} %agg, 0 <i>; yields i32</i>
3887 <!-- _______________________________________________________________________ -->
3888 <div class="doc_subsubsection">
3889 <a name="i_insertvalue">'<tt>insertvalue</tt>' Instruction</a>
3892 <div class="doc_text">
3896 <result> = insertvalue <aggregate type> <val>, <ty> <elt>, <idx> <i>; yields <aggregate type></i>
3900 <p>The '<tt>insertvalue</tt>' instruction inserts a value into a struct field or
3901 array element in an aggregate.</p>
3905 <p>The first operand of an '<tt>insertvalue</tt>' instruction is a value
3906 of <a href="#t_struct">struct</a> or <a href="#t_array">array</a> type. The
3907 second operand is a first-class value to insert. The following operands are
3908 constant indices indicating the position at which to insert the value in a
3909 similar manner as indices in a
3910 '<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction. The
3911 value to insert must have the same type as the value identified by the
3915 <p>The result is an aggregate of the same type as <tt>val</tt>. Its value is
3916 that of <tt>val</tt> except that the value at the position specified by the
3917 indices is that of <tt>elt</tt>.</p>
3921 %agg1 = insertvalue {i32, float} undef, i32 1, 0 <i>; yields {i32 1, float undef}</i>
3922 %agg2 = insertvalue {i32, float} %agg1, float %val, 1 <i>; yields {i32 1, float %val}</i>
3928 <!-- ======================================================================= -->
3929 <div class="doc_subsection">
3930 <a name="memoryops">Memory Access and Addressing Operations</a>
3933 <div class="doc_text">
3935 <p>A key design point of an SSA-based representation is how it represents
3936 memory. In LLVM, no memory locations are in SSA form, which makes things
3937 very simple. This section describes how to read, write, and allocate
3942 <!-- _______________________________________________________________________ -->
3943 <div class="doc_subsubsection">
3944 <a name="i_alloca">'<tt>alloca</tt>' Instruction</a>
3947 <div class="doc_text">
3951 <result> = alloca <type>[, i32 <NumElements>][, align <alignment>] <i>; yields {type*}:result</i>
3955 <p>The '<tt>alloca</tt>' instruction allocates memory on the stack frame of the
3956 currently executing function, to be automatically released when this function
3957 returns to its caller. The object is always allocated in the generic address
3958 space (address space zero).</p>
3961 <p>The '<tt>alloca</tt>' instruction
3962 allocates <tt>sizeof(<type>)*NumElements</tt> bytes of memory on the
3963 runtime stack, returning a pointer of the appropriate type to the program.
3964 If "NumElements" is specified, it is the number of elements allocated,
3965 otherwise "NumElements" is defaulted to be one. If a constant alignment is
3966 specified, the value result of the allocation is guaranteed to be aligned to
3967 at least that boundary. If not specified, or if zero, the target can choose
3968 to align the allocation on any convenient boundary compatible with the
3971 <p>'<tt>type</tt>' may be any sized type.</p>
3974 <p>Memory is allocated; a pointer is returned. The operation is undefined if
3975 there is insufficient stack space for the allocation. '<tt>alloca</tt>'d
3976 memory is automatically released when the function returns. The
3977 '<tt>alloca</tt>' instruction is commonly used to represent automatic
3978 variables that must have an address available. When the function returns
3979 (either with the <tt><a href="#i_ret">ret</a></tt>
3980 or <tt><a href="#i_unwind">unwind</a></tt> instructions), the memory is
3981 reclaimed. Allocating zero bytes is legal, but the result is undefined.</p>
3985 %ptr = alloca i32 <i>; yields {i32*}:ptr</i>
3986 %ptr = alloca i32, i32 4 <i>; yields {i32*}:ptr</i>
3987 %ptr = alloca i32, i32 4, align 1024 <i>; yields {i32*}:ptr</i>
3988 %ptr = alloca i32, align 1024 <i>; yields {i32*}:ptr</i>
3993 <!-- _______________________________________________________________________ -->
3994 <div class="doc_subsubsection"> <a name="i_load">'<tt>load</tt>'
3995 Instruction</a> </div>
3997 <div class="doc_text">
4001 <result> = load <ty>* <pointer>[, align <alignment>]
4002 <result> = volatile load <ty>* <pointer>[, align <alignment>]
4006 <p>The '<tt>load</tt>' instruction is used to read from memory.</p>
4009 <p>The argument to the '<tt>load</tt>' instruction specifies the memory address
4010 from which to load. The pointer must point to
4011 a <a href="#t_firstclass">first class</a> type. If the <tt>load</tt> is
4012 marked as <tt>volatile</tt>, then the optimizer is not allowed to modify the
4013 number or order of execution of this <tt>load</tt> with other
4014 volatile <tt>load</tt> and <tt><a href="#i_store">store</a></tt>
4017 <p>The optional constant "align" argument specifies the alignment of the
4018 operation (that is, the alignment of the memory address). A value of 0 or an
4019 omitted "align" argument means that the operation has the preferential
4020 alignment for the target. It is the responsibility of the code emitter to
4021 ensure that the alignment information is correct. Overestimating the
4022 alignment results in an undefined behavior. Underestimating the alignment may
4023 produce less efficient code. An alignment of 1 is always safe.</p>
4026 <p>The location of memory pointed to is loaded. If the value being loaded is of
4027 scalar type then the number of bytes read does not exceed the minimum number
4028 of bytes needed to hold all bits of the type. For example, loading an
4029 <tt>i24</tt> reads at most three bytes. When loading a value of a type like
4030 <tt>i20</tt> with a size that is not an integral number of bytes, the result
4031 is undefined if the value was not originally written using a store of the
4036 %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
4037 <a href="#i_store">store</a> i32 3, i32* %ptr <i>; yields {void}</i>
4038 %val = load i32* %ptr <i>; yields {i32}:val = i32 3</i>
4043 <!-- _______________________________________________________________________ -->
4044 <div class="doc_subsubsection"> <a name="i_store">'<tt>store</tt>'
4045 Instruction</a> </div>
4047 <div class="doc_text">
4051 store <ty> <value>, <ty>* <pointer>[, align <alignment>] <i>; yields {void}</i>
4052 volatile store <ty> <value>, <ty>* <pointer>[, align <alignment>] <i>; yields {void}</i>
4056 <p>The '<tt>store</tt>' instruction is used to write to memory.</p>
4059 <p>There are two arguments to the '<tt>store</tt>' instruction: a value to store
4060 and an address at which to store it. The type of the
4061 '<tt><pointer></tt>' operand must be a pointer to
4062 the <a href="#t_firstclass">first class</a> type of the
4063 '<tt><value></tt>' operand. If the <tt>store</tt> is marked
4064 as <tt>volatile</tt>, then the optimizer is not allowed to modify the number
4065 or order of execution of this <tt>store</tt> with other
4066 volatile <tt>load</tt> and <tt><a href="#i_store">store</a></tt>
4069 <p>The optional constant "align" argument specifies the alignment of the
4070 operation (that is, the alignment of the memory address). A value of 0 or an
4071 omitted "align" argument means that the operation has the preferential
4072 alignment for the target. It is the responsibility of the code emitter to
4073 ensure that the alignment information is correct. Overestimating the
4074 alignment results in an undefined behavior. Underestimating the alignment may
4075 produce less efficient code. An alignment of 1 is always safe.</p>
4078 <p>The contents of memory are updated to contain '<tt><value></tt>' at the
4079 location specified by the '<tt><pointer></tt>' operand. If
4080 '<tt><value></tt>' is of scalar type then the number of bytes written
4081 does not exceed the minimum number of bytes needed to hold all bits of the
4082 type. For example, storing an <tt>i24</tt> writes at most three bytes. When
4083 writing a value of a type like <tt>i20</tt> with a size that is not an
4084 integral number of bytes, it is unspecified what happens to the extra bits
4085 that do not belong to the type, but they will typically be overwritten.</p>
4089 %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
4090 store i32 3, i32* %ptr <i>; yields {void}</i>
4091 %val = <a href="#i_load">load</a> i32* %ptr <i>; yields {i32}:val = i32 3</i>
4096 <!-- _______________________________________________________________________ -->
4097 <div class="doc_subsubsection">
4098 <a name="i_getelementptr">'<tt>getelementptr</tt>' Instruction</a>
4101 <div class="doc_text">
4105 <result> = getelementptr <pty>* <ptrval>{, <ty> <idx>}*
4106 <result> = getelementptr inbounds <pty>* <ptrval>{, <ty> <idx>}*
4110 <p>The '<tt>getelementptr</tt>' instruction is used to get the address of a
4111 subelement of an aggregate data structure. It performs address calculation
4112 only and does not access memory.</p>
4115 <p>The first argument is always a pointer, and forms the basis of the
4116 calculation. The remaining arguments are indices that indicate which of the
4117 elements of the aggregate object are indexed. The interpretation of each
4118 index is dependent on the type being indexed into. The first index always
4119 indexes the pointer value given as the first argument, the second index
4120 indexes a value of the type pointed to (not necessarily the value directly
4121 pointed to, since the first index can be non-zero), etc. The first type
4122 indexed into must be a pointer value, subsequent types can be arrays, vectors
4123 and structs. Note that subsequent types being indexed into can never be
4124 pointers, since that would require loading the pointer before continuing
4127 <p>The type of each index argument depends on the type it is indexing into.
4128 When indexing into a (optionally packed) structure, only <tt>i32</tt> integer
4129 <b>constants</b> are allowed. When indexing into an array, pointer or
4130 vector, integers of any width are allowed, and they are not required to be
4133 <p>For example, let's consider a C code fragment and how it gets compiled to
4136 <div class="doc_code">
4149 int *foo(struct ST *s) {
4150 return &s[1].Z.B[5][13];
4155 <p>The LLVM code generated by the GCC frontend is:</p>
4157 <div class="doc_code">
4159 %RT = <a href="#namedtypes">type</a> { i8 , [10 x [20 x i32]], i8 }
4160 %ST = <a href="#namedtypes">type</a> { i32, double, %RT }
4162 define i32* @foo(%ST* %s) {
4164 %reg = getelementptr %ST* %s, i32 1, i32 2, i32 1, i32 5, i32 13
4171 <p>In the example above, the first index is indexing into the '<tt>%ST*</tt>'
4172 type, which is a pointer, yielding a '<tt>%ST</tt>' = '<tt>{ i32, double, %RT
4173 }</tt>' type, a structure. The second index indexes into the third element
4174 of the structure, yielding a '<tt>%RT</tt>' = '<tt>{ i8 , [10 x [20 x i32]],
4175 i8 }</tt>' type, another structure. The third index indexes into the second
4176 element of the structure, yielding a '<tt>[10 x [20 x i32]]</tt>' type, an
4177 array. The two dimensions of the array are subscripted into, yielding an
4178 '<tt>i32</tt>' type. The '<tt>getelementptr</tt>' instruction returns a
4179 pointer to this element, thus computing a value of '<tt>i32*</tt>' type.</p>
4181 <p>Note that it is perfectly legal to index partially through a structure,
4182 returning a pointer to an inner element. Because of this, the LLVM code for
4183 the given testcase is equivalent to:</p>
4186 define i32* @foo(%ST* %s) {
4187 %t1 = getelementptr %ST* %s, i32 1 <i>; yields %ST*:%t1</i>
4188 %t2 = getelementptr %ST* %t1, i32 0, i32 2 <i>; yields %RT*:%t2</i>
4189 %t3 = getelementptr %RT* %t2, i32 0, i32 1 <i>; yields [10 x [20 x i32]]*:%t3</i>
4190 %t4 = getelementptr [10 x [20 x i32]]* %t3, i32 0, i32 5 <i>; yields [20 x i32]*:%t4</i>
4191 %t5 = getelementptr [20 x i32]* %t4, i32 0, i32 13 <i>; yields i32*:%t5</i>
4196 <p>If the <tt>inbounds</tt> keyword is present, the result value of the
4197 <tt>getelementptr</tt> is undefined if the base pointer is not an
4198 <i>in bounds</i> address of an allocated object, or if any of the addresses
4199 that would be formed by successive addition of the offsets implied by the
4200 indices to the base address with infinitely precise arithmetic are not an
4201 <i>in bounds</i> address of that allocated object.
4202 The <i>in bounds</i> addresses for an allocated object are all the addresses
4203 that point into the object, plus the address one byte past the end.</p>
4205 <p>If the <tt>inbounds</tt> keyword is not present, the offsets are added to
4206 the base address with silently-wrapping two's complement arithmetic, and
4207 the result value of the <tt>getelementptr</tt> may be outside the object
4208 pointed to by the base pointer. The result value may not necessarily be
4209 used to access memory though, even if it happens to point into allocated
4210 storage. See the <a href="#pointeraliasing">Pointer Aliasing Rules</a>
4211 section for more information.</p>
4213 <p>The getelementptr instruction is often confusing. For some more insight into
4214 how it works, see <a href="GetElementPtr.html">the getelementptr FAQ</a>.</p>
4218 <i>; yields [12 x i8]*:aptr</i>
4219 %aptr = getelementptr {i32, [12 x i8]}* %saptr, i64 0, i32 1
4220 <i>; yields i8*:vptr</i>
4221 %vptr = getelementptr {i32, <2 x i8>}* %svptr, i64 0, i32 1, i32 1
4222 <i>; yields i8*:eptr</i>
4223 %eptr = getelementptr [12 x i8]* %aptr, i64 0, i32 1
4224 <i>; yields i32*:iptr</i>
4225 %iptr = getelementptr [10 x i32]* @arr, i16 0, i16 0
4230 <!-- ======================================================================= -->
4231 <div class="doc_subsection"> <a name="convertops">Conversion Operations</a>
4234 <div class="doc_text">
4236 <p>The instructions in this category are the conversion instructions (casting)
4237 which all take a single operand and a type. They perform various bit
4238 conversions on the operand.</p>
4242 <!-- _______________________________________________________________________ -->
4243 <div class="doc_subsubsection">
4244 <a name="i_trunc">'<tt>trunc .. to</tt>' Instruction</a>
4246 <div class="doc_text">
4250 <result> = trunc <ty> <value> to <ty2> <i>; yields ty2</i>
4254 <p>The '<tt>trunc</tt>' instruction truncates its operand to the
4255 type <tt>ty2</tt>.</p>
4258 <p>The '<tt>trunc</tt>' instruction takes a <tt>value</tt> to trunc, which must
4259 be an <a href="#t_integer">integer</a> type, and a type that specifies the
4260 size and type of the result, which must be
4261 an <a href="#t_integer">integer</a> type. The bit size of <tt>value</tt> must
4262 be larger than the bit size of <tt>ty2</tt>. Equal sized types are not
4266 <p>The '<tt>trunc</tt>' instruction truncates the high order bits
4267 in <tt>value</tt> and converts the remaining bits to <tt>ty2</tt>. Since the
4268 source size must be larger than the destination size, <tt>trunc</tt> cannot
4269 be a <i>no-op cast</i>. It will always truncate bits.</p>
4273 %X = trunc i32 257 to i8 <i>; yields i8:1</i>
4274 %Y = trunc i32 123 to i1 <i>; yields i1:true</i>
4275 %Z = trunc i32 122 to i1 <i>; yields i1:false</i>
4280 <!-- _______________________________________________________________________ -->
4281 <div class="doc_subsubsection">
4282 <a name="i_zext">'<tt>zext .. to</tt>' Instruction</a>
4284 <div class="doc_text">
4288 <result> = zext <ty> <value> to <ty2> <i>; yields ty2</i>
4292 <p>The '<tt>zext</tt>' instruction zero extends its operand to type
4297 <p>The '<tt>zext</tt>' instruction takes a value to cast, which must be of
4298 <a href="#t_integer">integer</a> type, and a type to cast it to, which must
4299 also be of <a href="#t_integer">integer</a> type. The bit size of the
4300 <tt>value</tt> must be smaller than the bit size of the destination type,
4304 <p>The <tt>zext</tt> fills the high order bits of the <tt>value</tt> with zero
4305 bits until it reaches the size of the destination type, <tt>ty2</tt>.</p>
4307 <p>When zero extending from i1, the result will always be either 0 or 1.</p>
4311 %X = zext i32 257 to i64 <i>; yields i64:257</i>
4312 %Y = zext i1 true to i32 <i>; yields i32:1</i>
4317 <!-- _______________________________________________________________________ -->
4318 <div class="doc_subsubsection">
4319 <a name="i_sext">'<tt>sext .. to</tt>' Instruction</a>
4321 <div class="doc_text">
4325 <result> = sext <ty> <value> to <ty2> <i>; yields ty2</i>
4329 <p>The '<tt>sext</tt>' sign extends <tt>value</tt> to the type <tt>ty2</tt>.</p>
4332 <p>The '<tt>sext</tt>' instruction takes a value to cast, which must be of
4333 <a href="#t_integer">integer</a> type, and a type to cast it to, which must
4334 also be of <a href="#t_integer">integer</a> type. The bit size of the
4335 <tt>value</tt> must be smaller than the bit size of the destination type,
4339 <p>The '<tt>sext</tt>' instruction performs a sign extension by copying the sign
4340 bit (highest order bit) of the <tt>value</tt> until it reaches the bit size
4341 of the type <tt>ty2</tt>.</p>
4343 <p>When sign extending from i1, the extension always results in -1 or 0.</p>
4347 %X = sext i8 -1 to i16 <i>; yields i16 :65535</i>
4348 %Y = sext i1 true to i32 <i>; yields i32:-1</i>
4353 <!-- _______________________________________________________________________ -->
4354 <div class="doc_subsubsection">
4355 <a name="i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a>
4358 <div class="doc_text">
4362 <result> = fptrunc <ty> <value> to <ty2> <i>; yields ty2</i>
4366 <p>The '<tt>fptrunc</tt>' instruction truncates <tt>value</tt> to type
4370 <p>The '<tt>fptrunc</tt>' instruction takes a <a href="#t_floating">floating
4371 point</a> value to cast and a <a href="#t_floating">floating point</a> type
4372 to cast it to. The size of <tt>value</tt> must be larger than the size of
4373 <tt>ty2</tt>. This implies that <tt>fptrunc</tt> cannot be used to make a
4374 <i>no-op cast</i>.</p>
4377 <p>The '<tt>fptrunc</tt>' instruction truncates a <tt>value</tt> from a larger
4378 <a href="#t_floating">floating point</a> type to a smaller
4379 <a href="#t_floating">floating point</a> type. If the value cannot fit
4380 within the destination type, <tt>ty2</tt>, then the results are
4385 %X = fptrunc double 123.0 to float <i>; yields float:123.0</i>
4386 %Y = fptrunc double 1.0E+300 to float <i>; yields undefined</i>
4391 <!-- _______________________________________________________________________ -->
4392 <div class="doc_subsubsection">
4393 <a name="i_fpext">'<tt>fpext .. to</tt>' Instruction</a>
4395 <div class="doc_text">
4399 <result> = fpext <ty> <value> to <ty2> <i>; yields ty2</i>
4403 <p>The '<tt>fpext</tt>' extends a floating point <tt>value</tt> to a larger
4404 floating point value.</p>
4407 <p>The '<tt>fpext</tt>' instruction takes a
4408 <a href="#t_floating">floating point</a> <tt>value</tt> to cast, and
4409 a <a href="#t_floating">floating point</a> type to cast it to. The source
4410 type must be smaller than the destination type.</p>
4413 <p>The '<tt>fpext</tt>' instruction extends the <tt>value</tt> from a smaller
4414 <a href="#t_floating">floating point</a> type to a larger
4415 <a href="#t_floating">floating point</a> type. The <tt>fpext</tt> cannot be
4416 used to make a <i>no-op cast</i> because it always changes bits. Use
4417 <tt>bitcast</tt> to make a <i>no-op cast</i> for a floating point cast.</p>
4421 %X = fpext float 3.1415 to double <i>; yields double:3.1415</i>
4422 %Y = fpext float 1.0 to float <i>; yields float:1.0 (no-op)</i>
4427 <!-- _______________________________________________________________________ -->
4428 <div class="doc_subsubsection">
4429 <a name="i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a>
4431 <div class="doc_text">
4435 <result> = fptoui <ty> <value> to <ty2> <i>; yields ty2</i>
4439 <p>The '<tt>fptoui</tt>' converts a floating point <tt>value</tt> to its
4440 unsigned integer equivalent of type <tt>ty2</tt>.</p>
4443 <p>The '<tt>fptoui</tt>' instruction takes a value to cast, which must be a
4444 scalar or vector <a href="#t_floating">floating point</a> value, and a type
4445 to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
4446 type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
4447 vector integer type with the same number of elements as <tt>ty</tt></p>
4450 <p>The '<tt>fptoui</tt>' instruction converts its
4451 <a href="#t_floating">floating point</a> operand into the nearest (rounding
4452 towards zero) unsigned integer value. If the value cannot fit
4453 in <tt>ty2</tt>, the results are undefined.</p>
4457 %X = fptoui double 123.0 to i32 <i>; yields i32:123</i>
4458 %Y = fptoui float 1.0E+300 to i1 <i>; yields undefined:1</i>
4459 %Z = fptoui float 1.04E+17 to i8 <i>; yields undefined:1</i>
4464 <!-- _______________________________________________________________________ -->
4465 <div class="doc_subsubsection">
4466 <a name="i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a>
4468 <div class="doc_text">
4472 <result> = fptosi <ty> <value> to <ty2> <i>; yields ty2</i>
4476 <p>The '<tt>fptosi</tt>' instruction converts
4477 <a href="#t_floating">floating point</a> <tt>value</tt> to
4478 type <tt>ty2</tt>.</p>
4481 <p>The '<tt>fptosi</tt>' instruction takes a value to cast, which must be a
4482 scalar or vector <a href="#t_floating">floating point</a> value, and a type
4483 to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
4484 type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
4485 vector integer type with the same number of elements as <tt>ty</tt></p>
4488 <p>The '<tt>fptosi</tt>' instruction converts its
4489 <a href="#t_floating">floating point</a> operand into the nearest (rounding
4490 towards zero) signed integer value. If the value cannot fit in <tt>ty2</tt>,
4491 the results are undefined.</p>
4495 %X = fptosi double -123.0 to i32 <i>; yields i32:-123</i>
4496 %Y = fptosi float 1.0E-247 to i1 <i>; yields undefined:1</i>
4497 %Z = fptosi float 1.04E+17 to i8 <i>; yields undefined:1</i>
4502 <!-- _______________________________________________________________________ -->
4503 <div class="doc_subsubsection">
4504 <a name="i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a>
4506 <div class="doc_text">
4510 <result> = uitofp <ty> <value> to <ty2> <i>; yields ty2</i>
4514 <p>The '<tt>uitofp</tt>' instruction regards <tt>value</tt> as an unsigned
4515 integer and converts that value to the <tt>ty2</tt> type.</p>
4518 <p>The '<tt>uitofp</tt>' instruction takes a value to cast, which must be a
4519 scalar or vector <a href="#t_integer">integer</a> value, and a type to cast
4520 it to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
4521 type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
4522 floating point type with the same number of elements as <tt>ty</tt></p>
4525 <p>The '<tt>uitofp</tt>' instruction interprets its operand as an unsigned
4526 integer quantity and converts it to the corresponding floating point
4527 value. If the value cannot fit in the floating point value, the results are
4532 %X = uitofp i32 257 to float <i>; yields float:257.0</i>
4533 %Y = uitofp i8 -1 to double <i>; yields double:255.0</i>
4538 <!-- _______________________________________________________________________ -->
4539 <div class="doc_subsubsection">
4540 <a name="i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a>
4542 <div class="doc_text">
4546 <result> = sitofp <ty> <value> to <ty2> <i>; yields ty2</i>
4550 <p>The '<tt>sitofp</tt>' instruction regards <tt>value</tt> as a signed integer
4551 and converts that value to the <tt>ty2</tt> type.</p>
4554 <p>The '<tt>sitofp</tt>' instruction takes a value to cast, which must be a
4555 scalar or vector <a href="#t_integer">integer</a> value, and a type to cast
4556 it to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
4557 type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
4558 floating point type with the same number of elements as <tt>ty</tt></p>
4561 <p>The '<tt>sitofp</tt>' instruction interprets its operand as a signed integer
4562 quantity and converts it to the corresponding floating point value. If the
4563 value cannot fit in the floating point value, the results are undefined.</p>
4567 %X = sitofp i32 257 to float <i>; yields float:257.0</i>
4568 %Y = sitofp i8 -1 to double <i>; yields double:-1.0</i>
4573 <!-- _______________________________________________________________________ -->
4574 <div class="doc_subsubsection">
4575 <a name="i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a>
4577 <div class="doc_text">
4581 <result> = ptrtoint <ty> <value> to <ty2> <i>; yields ty2</i>
4585 <p>The '<tt>ptrtoint</tt>' instruction converts the pointer <tt>value</tt> to
4586 the integer type <tt>ty2</tt>.</p>
4589 <p>The '<tt>ptrtoint</tt>' instruction takes a <tt>value</tt> to cast, which
4590 must be a <a href="#t_pointer">pointer</a> value, and a type to cast it to
4591 <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a> type.</p>
4594 <p>The '<tt>ptrtoint</tt>' instruction converts <tt>value</tt> to integer type
4595 <tt>ty2</tt> by interpreting the pointer value as an integer and either
4596 truncating or zero extending that value to the size of the integer type. If
4597 <tt>value</tt> is smaller than <tt>ty2</tt> then a zero extension is done. If
4598 <tt>value</tt> is larger than <tt>ty2</tt> then a truncation is done. If they
4599 are the same size, then nothing is done (<i>no-op cast</i>) other than a type
4604 %X = ptrtoint i32* %X to i8 <i>; yields truncation on 32-bit architecture</i>
4605 %Y = ptrtoint i32* %x to i64 <i>; yields zero extension on 32-bit architecture</i>
4610 <!-- _______________________________________________________________________ -->
4611 <div class="doc_subsubsection">
4612 <a name="i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a>
4614 <div class="doc_text">
4618 <result> = inttoptr <ty> <value> to <ty2> <i>; yields ty2</i>
4622 <p>The '<tt>inttoptr</tt>' instruction converts an integer <tt>value</tt> to a
4623 pointer type, <tt>ty2</tt>.</p>
4626 <p>The '<tt>inttoptr</tt>' instruction takes an <a href="#t_integer">integer</a>
4627 value to cast, and a type to cast it to, which must be a
4628 <a href="#t_pointer">pointer</a> type.</p>
4631 <p>The '<tt>inttoptr</tt>' instruction converts <tt>value</tt> to type
4632 <tt>ty2</tt> by applying either a zero extension or a truncation depending on
4633 the size of the integer <tt>value</tt>. If <tt>value</tt> is larger than the
4634 size of a pointer then a truncation is done. If <tt>value</tt> is smaller
4635 than the size of a pointer then a zero extension is done. If they are the
4636 same size, nothing is done (<i>no-op cast</i>).</p>
4640 %X = inttoptr i32 255 to i32* <i>; yields zero extension on 64-bit architecture</i>
4641 %Y = inttoptr i32 255 to i32* <i>; yields no-op on 32-bit architecture</i>
4642 %Z = inttoptr i64 0 to i32* <i>; yields truncation on 32-bit architecture</i>
4647 <!-- _______________________________________________________________________ -->
4648 <div class="doc_subsubsection">
4649 <a name="i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a>
4651 <div class="doc_text">
4655 <result> = bitcast <ty> <value> to <ty2> <i>; yields ty2</i>
4659 <p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
4660 <tt>ty2</tt> without changing any bits.</p>
4663 <p>The '<tt>bitcast</tt>' instruction takes a value to cast, which must be a
4664 non-aggregate first class value, and a type to cast it to, which must also be
4665 a non-aggregate <a href="#t_firstclass">first class</a> type. The bit sizes
4666 of <tt>value</tt> and the destination type, <tt>ty2</tt>, must be
4667 identical. If the source type is a pointer, the destination type must also be
4668 a pointer. This instruction supports bitwise conversion of vectors to
4669 integers and to vectors of other types (as long as they have the same
4673 <p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
4674 <tt>ty2</tt>. It is always a <i>no-op cast</i> because no bits change with
4675 this conversion. The conversion is done as if the <tt>value</tt> had been
4676 stored to memory and read back as type <tt>ty2</tt>. Pointer types may only
4677 be converted to other pointer types with this instruction. To convert
4678 pointers to other types, use the <a href="#i_inttoptr">inttoptr</a> or
4679 <a href="#i_ptrtoint">ptrtoint</a> instructions first.</p>
4683 %X = bitcast i8 255 to i8 <i>; yields i8 :-1</i>
4684 %Y = bitcast i32* %x to sint* <i>; yields sint*:%x</i>
4685 %Z = bitcast <2 x int> %V to i64; <i>; yields i64: %V</i>
4690 <!-- ======================================================================= -->
4691 <div class="doc_subsection"> <a name="otherops">Other Operations</a> </div>
4693 <div class="doc_text">
4695 <p>The instructions in this category are the "miscellaneous" instructions, which
4696 defy better classification.</p>
4700 <!-- _______________________________________________________________________ -->
4701 <div class="doc_subsubsection"><a name="i_icmp">'<tt>icmp</tt>' Instruction</a>
4704 <div class="doc_text">
4708 <result> = icmp <cond> <ty> <op1>, <op2> <i>; yields {i1} or {<N x i1>}:result</i>
4712 <p>The '<tt>icmp</tt>' instruction returns a boolean value or a vector of
4713 boolean values based on comparison of its two integer, integer vector, or
4714 pointer operands.</p>
4717 <p>The '<tt>icmp</tt>' instruction takes three operands. The first operand is
4718 the condition code indicating the kind of comparison to perform. It is not a
4719 value, just a keyword. The possible condition code are:</p>
4722 <li><tt>eq</tt>: equal</li>
4723 <li><tt>ne</tt>: not equal </li>
4724 <li><tt>ugt</tt>: unsigned greater than</li>
4725 <li><tt>uge</tt>: unsigned greater or equal</li>
4726 <li><tt>ult</tt>: unsigned less than</li>
4727 <li><tt>ule</tt>: unsigned less or equal</li>
4728 <li><tt>sgt</tt>: signed greater than</li>
4729 <li><tt>sge</tt>: signed greater or equal</li>
4730 <li><tt>slt</tt>: signed less than</li>
4731 <li><tt>sle</tt>: signed less or equal</li>
4734 <p>The remaining two arguments must be <a href="#t_integer">integer</a> or
4735 <a href="#t_pointer">pointer</a> or integer <a href="#t_vector">vector</a>
4736 typed. They must also be identical types.</p>
4739 <p>The '<tt>icmp</tt>' compares <tt>op1</tt> and <tt>op2</tt> according to the
4740 condition code given as <tt>cond</tt>. The comparison performed always yields
4741 either an <a href="#t_integer"><tt>i1</tt></a> or vector of <tt>i1</tt>
4742 result, as follows:</p>
4745 <li><tt>eq</tt>: yields <tt>true</tt> if the operands are equal,
4746 <tt>false</tt> otherwise. No sign interpretation is necessary or
4749 <li><tt>ne</tt>: yields <tt>true</tt> if the operands are unequal,
4750 <tt>false</tt> otherwise. No sign interpretation is necessary or
4753 <li><tt>ugt</tt>: interprets the operands as unsigned values and yields
4754 <tt>true</tt> if <tt>op1</tt> is greater than <tt>op2</tt>.</li>
4756 <li><tt>uge</tt>: interprets the operands as unsigned values and yields
4757 <tt>true</tt> if <tt>op1</tt> is greater than or equal
4758 to <tt>op2</tt>.</li>
4760 <li><tt>ult</tt>: interprets the operands as unsigned values and yields
4761 <tt>true</tt> if <tt>op1</tt> is less than <tt>op2</tt>.</li>
4763 <li><tt>ule</tt>: interprets the operands as unsigned values and yields
4764 <tt>true</tt> if <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
4766 <li><tt>sgt</tt>: interprets the operands as signed values and yields
4767 <tt>true</tt> if <tt>op1</tt> is greater than <tt>op2</tt>.</li>
4769 <li><tt>sge</tt>: interprets the operands as signed values and yields
4770 <tt>true</tt> if <tt>op1</tt> is greater than or equal
4771 to <tt>op2</tt>.</li>
4773 <li><tt>slt</tt>: interprets the operands as signed values and yields
4774 <tt>true</tt> if <tt>op1</tt> is less than <tt>op2</tt>.</li>
4776 <li><tt>sle</tt>: interprets the operands as signed values and yields
4777 <tt>true</tt> if <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
4780 <p>If the operands are <a href="#t_pointer">pointer</a> typed, the pointer
4781 values are compared as if they were integers.</p>
4783 <p>If the operands are integer vectors, then they are compared element by
4784 element. The result is an <tt>i1</tt> vector with the same number of elements
4785 as the values being compared. Otherwise, the result is an <tt>i1</tt>.</p>
4789 <result> = icmp eq i32 4, 5 <i>; yields: result=false</i>
4790 <result> = icmp ne float* %X, %X <i>; yields: result=false</i>
4791 <result> = icmp ult i16 4, 5 <i>; yields: result=true</i>
4792 <result> = icmp sgt i16 4, 5 <i>; yields: result=false</i>
4793 <result> = icmp ule i16 -4, 5 <i>; yields: result=false</i>
4794 <result> = icmp sge i16 4, 5 <i>; yields: result=false</i>
4797 <p>Note that the code generator does not yet support vector types with
4798 the <tt>icmp</tt> instruction.</p>
4802 <!-- _______________________________________________________________________ -->
4803 <div class="doc_subsubsection"><a name="i_fcmp">'<tt>fcmp</tt>' Instruction</a>
4806 <div class="doc_text">
4810 <result> = fcmp <cond> <ty> <op1>, <op2> <i>; yields {i1} or {<N x i1>}:result</i>
4814 <p>The '<tt>fcmp</tt>' instruction returns a boolean value or vector of boolean
4815 values based on comparison of its operands.</p>
4817 <p>If the operands are floating point scalars, then the result type is a boolean
4818 (<a href="#t_integer"><tt>i1</tt></a>).</p>
4820 <p>If the operands are floating point vectors, then the result type is a vector
4821 of boolean with the same number of elements as the operands being
4825 <p>The '<tt>fcmp</tt>' instruction takes three operands. The first operand is
4826 the condition code indicating the kind of comparison to perform. It is not a
4827 value, just a keyword. The possible condition code are:</p>
4830 <li><tt>false</tt>: no comparison, always returns false</li>
4831 <li><tt>oeq</tt>: ordered and equal</li>
4832 <li><tt>ogt</tt>: ordered and greater than </li>
4833 <li><tt>oge</tt>: ordered and greater than or equal</li>
4834 <li><tt>olt</tt>: ordered and less than </li>
4835 <li><tt>ole</tt>: ordered and less than or equal</li>
4836 <li><tt>one</tt>: ordered and not equal</li>
4837 <li><tt>ord</tt>: ordered (no nans)</li>
4838 <li><tt>ueq</tt>: unordered or equal</li>
4839 <li><tt>ugt</tt>: unordered or greater than </li>
4840 <li><tt>uge</tt>: unordered or greater than or equal</li>
4841 <li><tt>ult</tt>: unordered or less than </li>
4842 <li><tt>ule</tt>: unordered or less than or equal</li>
4843 <li><tt>une</tt>: unordered or not equal</li>
4844 <li><tt>uno</tt>: unordered (either nans)</li>
4845 <li><tt>true</tt>: no comparison, always returns true</li>
4848 <p><i>Ordered</i> means that neither operand is a QNAN while
4849 <i>unordered</i> means that either operand may be a QNAN.</p>
4851 <p>Each of <tt>val1</tt> and <tt>val2</tt> arguments must be either
4852 a <a href="#t_floating">floating point</a> type or
4853 a <a href="#t_vector">vector</a> of floating point type. They must have
4854 identical types.</p>
4857 <p>The '<tt>fcmp</tt>' instruction compares <tt>op1</tt> and <tt>op2</tt>
4858 according to the condition code given as <tt>cond</tt>. If the operands are
4859 vectors, then the vectors are compared element by element. Each comparison
4860 performed always yields an <a href="#t_integer">i1</a> result, as
4864 <li><tt>false</tt>: always yields <tt>false</tt>, regardless of operands.</li>
4866 <li><tt>oeq</tt>: yields <tt>true</tt> if both operands are not a QNAN and
4867 <tt>op1</tt> is equal to <tt>op2</tt>.</li>
4869 <li><tt>ogt</tt>: yields <tt>true</tt> if both operands are not a QNAN and
4870 <tt>op1</tt> is greather than <tt>op2</tt>.</li>
4872 <li><tt>oge</tt>: yields <tt>true</tt> if both operands are not a QNAN and
4873 <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
4875 <li><tt>olt</tt>: yields <tt>true</tt> if both operands are not a QNAN and
4876 <tt>op1</tt> is less than <tt>op2</tt>.</li>
4878 <li><tt>ole</tt>: yields <tt>true</tt> if both operands are not a QNAN and
4879 <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
4881 <li><tt>one</tt>: yields <tt>true</tt> if both operands are not a QNAN and
4882 <tt>op1</tt> is not equal to <tt>op2</tt>.</li>
4884 <li><tt>ord</tt>: yields <tt>true</tt> if both operands are not a QNAN.</li>
4886 <li><tt>ueq</tt>: yields <tt>true</tt> if either operand is a QNAN or
4887 <tt>op1</tt> is equal to <tt>op2</tt>.</li>
4889 <li><tt>ugt</tt>: yields <tt>true</tt> if either operand is a QNAN or
4890 <tt>op1</tt> is greater than <tt>op2</tt>.</li>
4892 <li><tt>uge</tt>: yields <tt>true</tt> if either operand is a QNAN or
4893 <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
4895 <li><tt>ult</tt>: yields <tt>true</tt> if either operand is a QNAN or
4896 <tt>op1</tt> is less than <tt>op2</tt>.</li>
4898 <li><tt>ule</tt>: yields <tt>true</tt> if either operand is a QNAN or
4899 <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
4901 <li><tt>une</tt>: yields <tt>true</tt> if either operand is a QNAN or
4902 <tt>op1</tt> is not equal to <tt>op2</tt>.</li>
4904 <li><tt>uno</tt>: yields <tt>true</tt> if either operand is a QNAN.</li>
4906 <li><tt>true</tt>: always yields <tt>true</tt>, regardless of operands.</li>
4911 <result> = fcmp oeq float 4.0, 5.0 <i>; yields: result=false</i>
4912 <result> = fcmp one float 4.0, 5.0 <i>; yields: result=true</i>
4913 <result> = fcmp olt float 4.0, 5.0 <i>; yields: result=true</i>
4914 <result> = fcmp ueq double 1.0, 2.0 <i>; yields: result=false</i>
4917 <p>Note that the code generator does not yet support vector types with
4918 the <tt>fcmp</tt> instruction.</p>
4922 <!-- _______________________________________________________________________ -->
4923 <div class="doc_subsubsection">
4924 <a name="i_phi">'<tt>phi</tt>' Instruction</a>
4927 <div class="doc_text">
4931 <result> = phi <ty> [ <val0>, <label0>], ...
4935 <p>The '<tt>phi</tt>' instruction is used to implement the φ node in the
4936 SSA graph representing the function.</p>
4939 <p>The type of the incoming values is specified with the first type field. After
4940 this, the '<tt>phi</tt>' instruction takes a list of pairs as arguments, with
4941 one pair for each predecessor basic block of the current block. Only values
4942 of <a href="#t_firstclass">first class</a> type may be used as the value
4943 arguments to the PHI node. Only labels may be used as the label
4946 <p>There must be no non-phi instructions between the start of a basic block and
4947 the PHI instructions: i.e. PHI instructions must be first in a basic
4950 <p>For the purposes of the SSA form, the use of each incoming value is deemed to
4951 occur on the edge from the corresponding predecessor block to the current
4952 block (but after any definition of an '<tt>invoke</tt>' instruction's return
4953 value on the same edge).</p>
4956 <p>At runtime, the '<tt>phi</tt>' instruction logically takes on the value
4957 specified by the pair corresponding to the predecessor basic block that
4958 executed just prior to the current block.</p>
4962 Loop: ; Infinite loop that counts from 0 on up...
4963 %indvar = phi i32 [ 0, %LoopHeader ], [ %nextindvar, %Loop ]
4964 %nextindvar = add i32 %indvar, 1
4970 <!-- _______________________________________________________________________ -->
4971 <div class="doc_subsubsection">
4972 <a name="i_select">'<tt>select</tt>' Instruction</a>
4975 <div class="doc_text">
4979 <result> = select <i>selty</i> <cond>, <ty> <val1>, <ty> <val2> <i>; yields ty</i>
4981 <i>selty</i> is either i1 or {<N x i1>}
4985 <p>The '<tt>select</tt>' instruction is used to choose one value based on a
4986 condition, without branching.</p>
4990 <p>The '<tt>select</tt>' instruction requires an 'i1' value or a vector of 'i1'
4991 values indicating the condition, and two values of the
4992 same <a href="#t_firstclass">first class</a> type. If the val1/val2 are
4993 vectors and the condition is a scalar, then entire vectors are selected, not
4994 individual elements.</p>
4997 <p>If the condition is an i1 and it evaluates to 1, the instruction returns the
4998 first value argument; otherwise, it returns the second value argument.</p>
5000 <p>If the condition is a vector of i1, then the value arguments must be vectors
5001 of the same size, and the selection is done element by element.</p>
5005 %X = select i1 true, i8 17, i8 42 <i>; yields i8:17</i>
5008 <p>Note that the code generator does not yet support conditions
5009 with vector type.</p>
5013 <!-- _______________________________________________________________________ -->
5014 <div class="doc_subsubsection">
5015 <a name="i_call">'<tt>call</tt>' Instruction</a>
5018 <div class="doc_text">
5022 <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>]
5026 <p>The '<tt>call</tt>' instruction represents a simple function call.</p>
5029 <p>This instruction requires several arguments:</p>
5032 <li>The optional "tail" marker indicates that the callee function does not
5033 access any allocas or varargs in the caller. Note that calls may be
5034 marked "tail" even if they do not occur before
5035 a <a href="#i_ret"><tt>ret</tt></a> instruction. If the "tail" marker is
5036 present, the function call is eligible for tail call optimization,
5037 but <a href="CodeGenerator.html#tailcallopt">might not in fact be
5038 optimized into a jump</a>. As of this writing, the extra requirements for
5039 a call to actually be optimized are:
5041 <li>Caller and callee both have the calling
5042 convention <tt>fastcc</tt>.</li>
5043 <li>The call is in tail position (ret immediately follows call and ret
5044 uses value of call or is void).</li>
5045 <li>Option <tt>-tailcallopt</tt> is enabled,
5046 or <code>llvm::PerformTailCallOpt</code> is <code>true</code>.</li>
5047 <li><a href="CodeGenerator.html#tailcallopt">Platform specific
5048 constraints are met.</a></li>
5052 <li>The optional "cconv" marker indicates which <a href="#callingconv">calling
5053 convention</a> the call should use. If none is specified, the call
5054 defaults to using C calling conventions. The calling convention of the
5055 call must match the calling convention of the target function, or else the
5056 behavior is undefined.</li>
5058 <li>The optional <a href="#paramattrs">Parameter Attributes</a> list for
5059 return values. Only '<tt>zeroext</tt>', '<tt>signext</tt>', and
5060 '<tt>inreg</tt>' attributes are valid here.</li>
5062 <li>'<tt>ty</tt>': the type of the call instruction itself which is also the
5063 type of the return value. Functions that return no value are marked
5064 <tt><a href="#t_void">void</a></tt>.</li>
5066 <li>'<tt>fnty</tt>': shall be the signature of the pointer to function value
5067 being invoked. The argument types must match the types implied by this
5068 signature. This type can be omitted if the function is not varargs and if
5069 the function type does not return a pointer to a function.</li>
5071 <li>'<tt>fnptrval</tt>': An LLVM value containing a pointer to a function to
5072 be invoked. In most cases, this is a direct function invocation, but
5073 indirect <tt>call</tt>s are just as possible, calling an arbitrary pointer
5074 to function value.</li>
5076 <li>'<tt>function args</tt>': argument list whose types match the function
5077 signature argument types. All arguments must be of
5078 <a href="#t_firstclass">first class</a> type. If the function signature
5079 indicates the function accepts a variable number of arguments, the extra
5080 arguments can be specified.</li>
5082 <li>The optional <a href="#fnattrs">function attributes</a> list. Only
5083 '<tt>noreturn</tt>', '<tt>nounwind</tt>', '<tt>readonly</tt>' and
5084 '<tt>readnone</tt>' attributes are valid here.</li>
5088 <p>The '<tt>call</tt>' instruction is used to cause control flow to transfer to
5089 a specified function, with its incoming arguments bound to the specified
5090 values. Upon a '<tt><a href="#i_ret">ret</a></tt>' instruction in the called
5091 function, control flow continues with the instruction after the function
5092 call, and the return value of the function is bound to the result
5097 %retval = call i32 @test(i32 %argc)
5098 call i32 (i8 *, ...)* @printf(i8 * %msg, i32 12, i8 42) <i>; yields i32</i>
5099 %X = tail call i32 @foo() <i>; yields i32</i>
5100 %Y = tail call <a href="#callingconv">fastcc</a> i32 @foo() <i>; yields i32</i>
5101 call void %foo(i8 97 signext)
5103 %struct.A = type { i32, i8 }
5104 %r = call %struct.A @foo() <i>; yields { 32, i8 }</i>
5105 %gr = extractvalue %struct.A %r, 0 <i>; yields i32</i>
5106 %gr1 = extractvalue %struct.A %r, 1 <i>; yields i8</i>
5107 %Z = call void @foo() noreturn <i>; indicates that %foo never returns normally</i>
5108 %ZZ = call zeroext i32 @bar() <i>; Return value is %zero extended</i>
5111 <p>llvm treats calls to some functions with names and arguments that match the
5112 standard C99 library as being the C99 library functions, and may perform
5113 optimizations or generate code for them under that assumption. This is
5114 something we'd like to change in the future to provide better support for
5115 freestanding environments and non-C-based langauges.</p>
5119 <!-- _______________________________________________________________________ -->
5120 <div class="doc_subsubsection">
5121 <a name="i_va_arg">'<tt>va_arg</tt>' Instruction</a>
5124 <div class="doc_text">
5128 <resultval> = va_arg <va_list*> <arglist>, <argty>
5132 <p>The '<tt>va_arg</tt>' instruction is used to access arguments passed through
5133 the "variable argument" area of a function call. It is used to implement the
5134 <tt>va_arg</tt> macro in C.</p>
5137 <p>This instruction takes a <tt>va_list*</tt> value and the type of the
5138 argument. It returns a value of the specified argument type and increments
5139 the <tt>va_list</tt> to point to the next argument. The actual type
5140 of <tt>va_list</tt> is target specific.</p>
5143 <p>The '<tt>va_arg</tt>' instruction loads an argument of the specified type
5144 from the specified <tt>va_list</tt> and causes the <tt>va_list</tt> to point
5145 to the next argument. For more information, see the variable argument
5146 handling <a href="#int_varargs">Intrinsic Functions</a>.</p>
5148 <p>It is legal for this instruction to be called in a function which does not
5149 take a variable number of arguments, for example, the <tt>vfprintf</tt>
5152 <p><tt>va_arg</tt> is an LLVM instruction instead of
5153 an <a href="#intrinsics">intrinsic function</a> because it takes a type as an
5157 <p>See the <a href="#int_varargs">variable argument processing</a> section.</p>
5159 <p>Note that the code generator does not yet fully support va_arg on many
5160 targets. Also, it does not currently support va_arg with aggregate types on
5165 <!-- *********************************************************************** -->
5166 <div class="doc_section"> <a name="intrinsics">Intrinsic Functions</a> </div>
5167 <!-- *********************************************************************** -->
5169 <div class="doc_text">
5171 <p>LLVM supports the notion of an "intrinsic function". These functions have
5172 well known names and semantics and are required to follow certain
5173 restrictions. Overall, these intrinsics represent an extension mechanism for
5174 the LLVM language that does not require changing all of the transformations
5175 in LLVM when adding to the language (or the bitcode reader/writer, the
5176 parser, etc...).</p>
5178 <p>Intrinsic function names must all start with an "<tt>llvm.</tt>" prefix. This
5179 prefix is reserved in LLVM for intrinsic names; thus, function names may not
5180 begin with this prefix. Intrinsic functions must always be external
5181 functions: you cannot define the body of intrinsic functions. Intrinsic
5182 functions may only be used in call or invoke instructions: it is illegal to
5183 take the address of an intrinsic function. Additionally, because intrinsic
5184 functions are part of the LLVM language, it is required if any are added that
5185 they be documented here.</p>
5187 <p>Some intrinsic functions can be overloaded, i.e., the intrinsic represents a
5188 family of functions that perform the same operation but on different data
5189 types. Because LLVM can represent over 8 million different integer types,
5190 overloading is used commonly to allow an intrinsic function to operate on any
5191 integer type. One or more of the argument types or the result type can be
5192 overloaded to accept any integer type. Argument types may also be defined as
5193 exactly matching a previous argument's type or the result type. This allows
5194 an intrinsic function which accepts multiple arguments, but needs all of them
5195 to be of the same type, to only be overloaded with respect to a single
5196 argument or the result.</p>
5198 <p>Overloaded intrinsics will have the names of its overloaded argument types
5199 encoded into its function name, each preceded by a period. Only those types
5200 which are overloaded result in a name suffix. Arguments whose type is matched
5201 against another type do not. For example, the <tt>llvm.ctpop</tt> function
5202 can take an integer of any width and returns an integer of exactly the same
5203 integer width. This leads to a family of functions such as
5204 <tt>i8 @llvm.ctpop.i8(i8 %val)</tt> and <tt>i29 @llvm.ctpop.i29(i29
5205 %val)</tt>. Only one type, the return type, is overloaded, and only one type
5206 suffix is required. Because the argument's type is matched against the return
5207 type, it does not require its own name suffix.</p>
5209 <p>To learn how to add an intrinsic function, please see the
5210 <a href="ExtendingLLVM.html">Extending LLVM Guide</a>.</p>
5214 <!-- ======================================================================= -->
5215 <div class="doc_subsection">
5216 <a name="int_varargs">Variable Argument Handling Intrinsics</a>
5219 <div class="doc_text">
5221 <p>Variable argument support is defined in LLVM with
5222 the <a href="#i_va_arg"><tt>va_arg</tt></a> instruction and these three
5223 intrinsic functions. These functions are related to the similarly named
5224 macros defined in the <tt><stdarg.h></tt> header file.</p>
5226 <p>All of these functions operate on arguments that use a target-specific value
5227 type "<tt>va_list</tt>". The LLVM assembly language reference manual does
5228 not define what this type is, so all transformations should be prepared to
5229 handle these functions regardless of the type used.</p>
5231 <p>This example shows how the <a href="#i_va_arg"><tt>va_arg</tt></a>
5232 instruction and the variable argument handling intrinsic functions are
5235 <div class="doc_code">
5237 define i32 @test(i32 %X, ...) {
5238 ; Initialize variable argument processing
5240 %ap2 = bitcast i8** %ap to i8*
5241 call void @llvm.va_start(i8* %ap2)
5243 ; Read a single integer argument
5244 %tmp = va_arg i8** %ap, i32
5246 ; Demonstrate usage of llvm.va_copy and llvm.va_end
5248 %aq2 = bitcast i8** %aq to i8*
5249 call void @llvm.va_copy(i8* %aq2, i8* %ap2)
5250 call void @llvm.va_end(i8* %aq2)
5252 ; Stop processing of arguments.
5253 call void @llvm.va_end(i8* %ap2)
5257 declare void @llvm.va_start(i8*)
5258 declare void @llvm.va_copy(i8*, i8*)
5259 declare void @llvm.va_end(i8*)
5265 <!-- _______________________________________________________________________ -->
5266 <div class="doc_subsubsection">
5267 <a name="int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a>
5271 <div class="doc_text">
5275 declare void %llvm.va_start(i8* <arglist>)
5279 <p>The '<tt>llvm.va_start</tt>' intrinsic initializes <tt>*<arglist></tt>
5280 for subsequent use by <tt><a href="#i_va_arg">va_arg</a></tt>.</p>
5283 <p>The argument is a pointer to a <tt>va_list</tt> element to initialize.</p>
5286 <p>The '<tt>llvm.va_start</tt>' intrinsic works just like the <tt>va_start</tt>
5287 macro available in C. In a target-dependent way, it initializes
5288 the <tt>va_list</tt> element to which the argument points, so that the next
5289 call to <tt>va_arg</tt> will produce the first variable argument passed to
5290 the function. Unlike the C <tt>va_start</tt> macro, this intrinsic does not
5291 need to know the last argument of the function as the compiler can figure
5296 <!-- _______________________________________________________________________ -->
5297 <div class="doc_subsubsection">
5298 <a name="int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a>
5301 <div class="doc_text">
5305 declare void @llvm.va_end(i8* <arglist>)
5309 <p>The '<tt>llvm.va_end</tt>' intrinsic destroys <tt>*<arglist></tt>,
5310 which has been initialized previously
5311 with <tt><a href="#int_va_start">llvm.va_start</a></tt>
5312 or <tt><a href="#i_va_copy">llvm.va_copy</a></tt>.</p>
5315 <p>The argument is a pointer to a <tt>va_list</tt> to destroy.</p>
5318 <p>The '<tt>llvm.va_end</tt>' intrinsic works just like the <tt>va_end</tt>
5319 macro available in C. In a target-dependent way, it destroys
5320 the <tt>va_list</tt> element to which the argument points. Calls
5321 to <a href="#int_va_start"><tt>llvm.va_start</tt></a>
5322 and <a href="#int_va_copy"> <tt>llvm.va_copy</tt></a> must be matched exactly
5323 with calls to <tt>llvm.va_end</tt>.</p>
5327 <!-- _______________________________________________________________________ -->
5328 <div class="doc_subsubsection">
5329 <a name="int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a>
5332 <div class="doc_text">
5336 declare void @llvm.va_copy(i8* <destarglist>, i8* <srcarglist>)
5340 <p>The '<tt>llvm.va_copy</tt>' intrinsic copies the current argument position
5341 from the source argument list to the destination argument list.</p>
5344 <p>The first argument is a pointer to a <tt>va_list</tt> element to initialize.
5345 The second argument is a pointer to a <tt>va_list</tt> element to copy
5349 <p>The '<tt>llvm.va_copy</tt>' intrinsic works just like the <tt>va_copy</tt>
5350 macro available in C. In a target-dependent way, it copies the
5351 source <tt>va_list</tt> element into the destination <tt>va_list</tt>
5352 element. This intrinsic is necessary because
5353 the <tt><a href="#int_va_start"> llvm.va_start</a></tt> intrinsic may be
5354 arbitrarily complex and require, for example, memory allocation.</p>
5358 <!-- ======================================================================= -->
5359 <div class="doc_subsection">
5360 <a name="int_gc">Accurate Garbage Collection Intrinsics</a>
5363 <div class="doc_text">
5365 <p>LLVM support for <a href="GarbageCollection.html">Accurate Garbage
5366 Collection</a> (GC) requires the implementation and generation of these
5367 intrinsics. These intrinsics allow identification of <a href="#int_gcroot">GC
5368 roots on the stack</a>, as well as garbage collector implementations that
5369 require <a href="#int_gcread">read</a> and <a href="#int_gcwrite">write</a>
5370 barriers. Front-ends for type-safe garbage collected languages should generate
5371 these intrinsics to make use of the LLVM garbage collectors. For more details,
5372 see <a href="GarbageCollection.html">Accurate Garbage Collection with
5375 <p>The garbage collection intrinsics only operate on objects in the generic
5376 address space (address space zero).</p>
5380 <!-- _______________________________________________________________________ -->
5381 <div class="doc_subsubsection">
5382 <a name="int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a>
5385 <div class="doc_text">
5389 declare void @llvm.gcroot(i8** %ptrloc, i8* %metadata)
5393 <p>The '<tt>llvm.gcroot</tt>' intrinsic declares the existence of a GC root to
5394 the code generator, and allows some metadata to be associated with it.</p>
5397 <p>The first argument specifies the address of a stack object that contains the
5398 root pointer. The second pointer (which must be either a constant or a
5399 global value address) contains the meta-data to be associated with the
5403 <p>At runtime, a call to this intrinsic stores a null pointer into the "ptrloc"
5404 location. At compile-time, the code generator generates information to allow
5405 the runtime to find the pointer at GC safe points. The '<tt>llvm.gcroot</tt>'
5406 intrinsic may only be used in a function which <a href="#gc">specifies a GC
5411 <!-- _______________________________________________________________________ -->
5412 <div class="doc_subsubsection">
5413 <a name="int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a>
5416 <div class="doc_text">
5420 declare i8* @llvm.gcread(i8* %ObjPtr, i8** %Ptr)
5424 <p>The '<tt>llvm.gcread</tt>' intrinsic identifies reads of references from heap
5425 locations, allowing garbage collector implementations that require read
5429 <p>The second argument is the address to read from, which should be an address
5430 allocated from the garbage collector. The first object is a pointer to the
5431 start of the referenced object, if needed by the language runtime (otherwise
5435 <p>The '<tt>llvm.gcread</tt>' intrinsic has the same semantics as a load
5436 instruction, but may be replaced with substantially more complex code by the
5437 garbage collector runtime, as needed. The '<tt>llvm.gcread</tt>' intrinsic
5438 may only be used in a function which <a href="#gc">specifies a GC
5443 <!-- _______________________________________________________________________ -->
5444 <div class="doc_subsubsection">
5445 <a name="int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a>
5448 <div class="doc_text">
5452 declare void @llvm.gcwrite(i8* %P1, i8* %Obj, i8** %P2)
5456 <p>The '<tt>llvm.gcwrite</tt>' intrinsic identifies writes of references to heap
5457 locations, allowing garbage collector implementations that require write
5458 barriers (such as generational or reference counting collectors).</p>
5461 <p>The first argument is the reference to store, the second is the start of the
5462 object to store it to, and the third is the address of the field of Obj to
5463 store to. If the runtime does not require a pointer to the object, Obj may
5467 <p>The '<tt>llvm.gcwrite</tt>' intrinsic has the same semantics as a store
5468 instruction, but may be replaced with substantially more complex code by the
5469 garbage collector runtime, as needed. The '<tt>llvm.gcwrite</tt>' intrinsic
5470 may only be used in a function which <a href="#gc">specifies a GC
5475 <!-- ======================================================================= -->
5476 <div class="doc_subsection">
5477 <a name="int_codegen">Code Generator Intrinsics</a>
5480 <div class="doc_text">
5482 <p>These intrinsics are provided by LLVM to expose special features that may
5483 only be implemented with code generator support.</p>
5487 <!-- _______________________________________________________________________ -->
5488 <div class="doc_subsubsection">
5489 <a name="int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a>
5492 <div class="doc_text">
5496 declare i8 *@llvm.returnaddress(i32 <level>)
5500 <p>The '<tt>llvm.returnaddress</tt>' intrinsic attempts to compute a
5501 target-specific value indicating the return address of the current function
5502 or one of its callers.</p>
5505 <p>The argument to this intrinsic indicates which function to return the address
5506 for. Zero indicates the calling function, one indicates its caller, etc.
5507 The argument is <b>required</b> to be a constant integer value.</p>
5510 <p>The '<tt>llvm.returnaddress</tt>' intrinsic either returns a pointer
5511 indicating the return address of the specified call frame, or zero if it
5512 cannot be identified. The value returned by this intrinsic is likely to be
5513 incorrect or 0 for arguments other than zero, so it should only be used for
5514 debugging purposes.</p>
5516 <p>Note that calling this intrinsic does not prevent function inlining or other
5517 aggressive transformations, so the value returned may not be that of the
5518 obvious source-language caller.</p>
5522 <!-- _______________________________________________________________________ -->
5523 <div class="doc_subsubsection">
5524 <a name="int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a>
5527 <div class="doc_text">
5531 declare i8 *@llvm.frameaddress(i32 <level>)
5535 <p>The '<tt>llvm.frameaddress</tt>' intrinsic attempts to return the
5536 target-specific frame pointer value for the specified stack frame.</p>
5539 <p>The argument to this intrinsic indicates which function to return the frame
5540 pointer for. Zero indicates the calling function, one indicates its caller,
5541 etc. The argument is <b>required</b> to be a constant integer value.</p>
5544 <p>The '<tt>llvm.frameaddress</tt>' intrinsic either returns a pointer
5545 indicating the frame address of the specified call frame, or zero if it
5546 cannot be identified. The value returned by this intrinsic is likely to be
5547 incorrect or 0 for arguments other than zero, so it should only be used for
5548 debugging purposes.</p>
5550 <p>Note that calling this intrinsic does not prevent function inlining or other
5551 aggressive transformations, so the value returned may not be that of the
5552 obvious source-language caller.</p>
5556 <!-- _______________________________________________________________________ -->
5557 <div class="doc_subsubsection">
5558 <a name="int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a>
5561 <div class="doc_text">
5565 declare i8 *@llvm.stacksave()
5569 <p>The '<tt>llvm.stacksave</tt>' intrinsic is used to remember the current state
5570 of the function stack, for use
5571 with <a href="#int_stackrestore"> <tt>llvm.stackrestore</tt></a>. This is
5572 useful for implementing language features like scoped automatic variable
5573 sized arrays in C99.</p>
5576 <p>This intrinsic returns a opaque pointer value that can be passed
5577 to <a href="#int_stackrestore"><tt>llvm.stackrestore</tt></a>. When
5578 an <tt>llvm.stackrestore</tt> intrinsic is executed with a value saved
5579 from <tt>llvm.stacksave</tt>, it effectively restores the state of the stack
5580 to the state it was in when the <tt>llvm.stacksave</tt> intrinsic executed.
5581 In practice, this pops any <a href="#i_alloca">alloca</a> blocks from the
5582 stack that were allocated after the <tt>llvm.stacksave</tt> was executed.</p>
5586 <!-- _______________________________________________________________________ -->
5587 <div class="doc_subsubsection">
5588 <a name="int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a>
5591 <div class="doc_text">
5595 declare void @llvm.stackrestore(i8 * %ptr)
5599 <p>The '<tt>llvm.stackrestore</tt>' intrinsic is used to restore the state of
5600 the function stack to the state it was in when the
5601 corresponding <a href="#int_stacksave"><tt>llvm.stacksave</tt></a> intrinsic
5602 executed. This is useful for implementing language features like scoped
5603 automatic variable sized arrays in C99.</p>
5606 <p>See the description
5607 for <a href="#int_stacksave"><tt>llvm.stacksave</tt></a>.</p>
5611 <!-- _______________________________________________________________________ -->
5612 <div class="doc_subsubsection">
5613 <a name="int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a>
5616 <div class="doc_text">
5620 declare void @llvm.prefetch(i8* <address>, i32 <rw>, i32 <locality>)
5624 <p>The '<tt>llvm.prefetch</tt>' intrinsic is a hint to the code generator to
5625 insert a prefetch instruction if supported; otherwise, it is a noop.
5626 Prefetches have no effect on the behavior of the program but can change its
5627 performance characteristics.</p>
5630 <p><tt>address</tt> is the address to be prefetched, <tt>rw</tt> is the
5631 specifier determining if the fetch should be for a read (0) or write (1),
5632 and <tt>locality</tt> is a temporal locality specifier ranging from (0) - no
5633 locality, to (3) - extremely local keep in cache. The <tt>rw</tt>
5634 and <tt>locality</tt> arguments must be constant integers.</p>
5637 <p>This intrinsic does not modify the behavior of the program. In particular,
5638 prefetches cannot trap and do not produce a value. On targets that support
5639 this intrinsic, the prefetch can provide hints to the processor cache for
5640 better performance.</p>
5644 <!-- _______________________________________________________________________ -->
5645 <div class="doc_subsubsection">
5646 <a name="int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a>
5649 <div class="doc_text">
5653 declare void @llvm.pcmarker(i32 <id>)
5657 <p>The '<tt>llvm.pcmarker</tt>' intrinsic is a method to export a Program
5658 Counter (PC) in a region of code to simulators and other tools. The method
5659 is target specific, but it is expected that the marker will use exported
5660 symbols to transmit the PC of the marker. The marker makes no guarantees
5661 that it will remain with any specific instruction after optimizations. It is
5662 possible that the presence of a marker will inhibit optimizations. The
5663 intended use is to be inserted after optimizations to allow correlations of
5664 simulation runs.</p>
5667 <p><tt>id</tt> is a numerical id identifying the marker.</p>
5670 <p>This intrinsic does not modify the behavior of the program. Backends that do
5671 not support this intrinisic may ignore it.</p>
5675 <!-- _______________________________________________________________________ -->
5676 <div class="doc_subsubsection">
5677 <a name="int_readcyclecounter">'<tt>llvm.readcyclecounter</tt>' Intrinsic</a>
5680 <div class="doc_text">
5684 declare i64 @llvm.readcyclecounter( )
5688 <p>The '<tt>llvm.readcyclecounter</tt>' intrinsic provides access to the cycle
5689 counter register (or similar low latency, high accuracy clocks) on those
5690 targets that support it. On X86, it should map to RDTSC. On Alpha, it
5691 should map to RPCC. As the backing counters overflow quickly (on the order
5692 of 9 seconds on alpha), this should only be used for small timings.</p>
5695 <p>When directly supported, reading the cycle counter should not modify any
5696 memory. Implementations are allowed to either return a application specific
5697 value or a system wide value. On backends without support, this is lowered
5698 to a constant 0.</p>
5702 <!-- ======================================================================= -->
5703 <div class="doc_subsection">
5704 <a name="int_libc">Standard C Library Intrinsics</a>
5707 <div class="doc_text">
5709 <p>LLVM provides intrinsics for a few important standard C library functions.
5710 These intrinsics allow source-language front-ends to pass information about
5711 the alignment of the pointer arguments to the code generator, providing
5712 opportunity for more efficient code generation.</p>
5716 <!-- _______________________________________________________________________ -->
5717 <div class="doc_subsubsection">
5718 <a name="int_memcpy">'<tt>llvm.memcpy</tt>' Intrinsic</a>
5721 <div class="doc_text">
5724 <p>This is an overloaded intrinsic. You can use <tt>llvm.memcpy</tt> on any
5725 integer bit width. Not all targets support all bit widths however.</p>
5728 declare void @llvm.memcpy.i8(i8 * <dest>, i8 * <src>,
5729 i8 <len>, i32 <align>)
5730 declare void @llvm.memcpy.i16(i8 * <dest>, i8 * <src>,
5731 i16 <len>, i32 <align>)
5732 declare void @llvm.memcpy.i32(i8 * <dest>, i8 * <src>,
5733 i32 <len>, i32 <align>)
5734 declare void @llvm.memcpy.i64(i8 * <dest>, i8 * <src>,
5735 i64 <len>, i32 <align>)
5739 <p>The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the
5740 source location to the destination location.</p>
5742 <p>Note that, unlike the standard libc function, the <tt>llvm.memcpy.*</tt>
5743 intrinsics do not return a value, and takes an extra alignment argument.</p>
5746 <p>The first argument is a pointer to the destination, the second is a pointer
5747 to the source. The third argument is an integer argument specifying the
5748 number of bytes to copy, and the fourth argument is the alignment of the
5749 source and destination locations.</p>
5751 <p>If the call to this intrinisic has an alignment value that is not 0 or 1,
5752 then the caller guarantees that both the source and destination pointers are
5753 aligned to that boundary.</p>
5756 <p>The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the
5757 source location to the destination location, which are not allowed to
5758 overlap. It copies "len" bytes of memory over. If the argument is known to
5759 be aligned to some boundary, this can be specified as the fourth argument,
5760 otherwise it should be set to 0 or 1.</p>
5764 <!-- _______________________________________________________________________ -->
5765 <div class="doc_subsubsection">
5766 <a name="int_memmove">'<tt>llvm.memmove</tt>' Intrinsic</a>
5769 <div class="doc_text">
5772 <p>This is an overloaded intrinsic. You can use llvm.memmove on any integer bit
5773 width. Not all targets support all bit widths however.</p>
5776 declare void @llvm.memmove.i8(i8 * <dest>, i8 * <src>,
5777 i8 <len>, i32 <align>)
5778 declare void @llvm.memmove.i16(i8 * <dest>, i8 * <src>,
5779 i16 <len>, i32 <align>)
5780 declare void @llvm.memmove.i32(i8 * <dest>, i8 * <src>,
5781 i32 <len>, i32 <align>)
5782 declare void @llvm.memmove.i64(i8 * <dest>, i8 * <src>,
5783 i64 <len>, i32 <align>)
5787 <p>The '<tt>llvm.memmove.*</tt>' intrinsics move a block of memory from the
5788 source location to the destination location. It is similar to the
5789 '<tt>llvm.memcpy</tt>' intrinsic but allows the two memory locations to
5792 <p>Note that, unlike the standard libc function, the <tt>llvm.memmove.*</tt>
5793 intrinsics do not return a value, and takes an extra alignment argument.</p>
5796 <p>The first argument is a pointer to the destination, the second is a pointer
5797 to the source. The third argument is an integer argument specifying the
5798 number of bytes to copy, and the fourth argument is the alignment of the
5799 source and destination locations.</p>
5801 <p>If the call to this intrinisic has an alignment value that is not 0 or 1,
5802 then the caller guarantees that the source and destination pointers are
5803 aligned to that boundary.</p>
5806 <p>The '<tt>llvm.memmove.*</tt>' intrinsics copy a block of memory from the
5807 source location to the destination location, which may overlap. It copies
5808 "len" bytes of memory over. If the argument is known to be aligned to some
5809 boundary, this can be specified as the fourth argument, otherwise it should
5810 be set to 0 or 1.</p>
5814 <!-- _______________________________________________________________________ -->
5815 <div class="doc_subsubsection">
5816 <a name="int_memset">'<tt>llvm.memset.*</tt>' Intrinsics</a>
5819 <div class="doc_text">
5822 <p>This is an overloaded intrinsic. You can use llvm.memset on any integer bit
5823 width. Not all targets support all bit widths however.</p>
5826 declare void @llvm.memset.i8(i8 * <dest>, i8 <val>,
5827 i8 <len>, i32 <align>)
5828 declare void @llvm.memset.i16(i8 * <dest>, i8 <val>,
5829 i16 <len>, i32 <align>)
5830 declare void @llvm.memset.i32(i8 * <dest>, i8 <val>,
5831 i32 <len>, i32 <align>)
5832 declare void @llvm.memset.i64(i8 * <dest>, i8 <val>,
5833 i64 <len>, i32 <align>)
5837 <p>The '<tt>llvm.memset.*</tt>' intrinsics fill a block of memory with a
5838 particular byte value.</p>
5840 <p>Note that, unlike the standard libc function, the <tt>llvm.memset</tt>
5841 intrinsic does not return a value, and takes an extra alignment argument.</p>
5844 <p>The first argument is a pointer to the destination to fill, the second is the
5845 byte value to fill it with, the third argument is an integer argument
5846 specifying the number of bytes to fill, and the fourth argument is the known
5847 alignment of destination location.</p>
5849 <p>If the call to this intrinisic has an alignment value that is not 0 or 1,
5850 then the caller guarantees that the destination pointer is aligned to that
5854 <p>The '<tt>llvm.memset.*</tt>' intrinsics fill "len" bytes of memory starting
5855 at the destination location. If the argument is known to be aligned to some
5856 boundary, this can be specified as the fourth argument, otherwise it should
5857 be set to 0 or 1.</p>
5861 <!-- _______________________________________________________________________ -->
5862 <div class="doc_subsubsection">
5863 <a name="int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a>
5866 <div class="doc_text">
5869 <p>This is an overloaded intrinsic. You can use <tt>llvm.sqrt</tt> on any
5870 floating point or vector of floating point type. Not all targets support all
5874 declare float @llvm.sqrt.f32(float %Val)
5875 declare double @llvm.sqrt.f64(double %Val)
5876 declare x86_fp80 @llvm.sqrt.f80(x86_fp80 %Val)
5877 declare fp128 @llvm.sqrt.f128(fp128 %Val)
5878 declare ppc_fp128 @llvm.sqrt.ppcf128(ppc_fp128 %Val)
5882 <p>The '<tt>llvm.sqrt</tt>' intrinsics return the sqrt of the specified operand,
5883 returning the same value as the libm '<tt>sqrt</tt>' functions would.
5884 Unlike <tt>sqrt</tt> in libm, however, <tt>llvm.sqrt</tt> has undefined
5885 behavior for negative numbers other than -0.0 (which allows for better
5886 optimization, because there is no need to worry about errno being
5887 set). <tt>llvm.sqrt(-0.0)</tt> is defined to return -0.0 like IEEE sqrt.</p>
5890 <p>The argument and return value are floating point numbers of the same
5894 <p>This function returns the sqrt of the specified operand if it is a
5895 nonnegative floating point number.</p>
5899 <!-- _______________________________________________________________________ -->
5900 <div class="doc_subsubsection">
5901 <a name="int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a>
5904 <div class="doc_text">
5907 <p>This is an overloaded intrinsic. You can use <tt>llvm.powi</tt> on any
5908 floating point or vector of floating point type. Not all targets support all
5912 declare float @llvm.powi.f32(float %Val, i32 %power)
5913 declare double @llvm.powi.f64(double %Val, i32 %power)
5914 declare x86_fp80 @llvm.powi.f80(x86_fp80 %Val, i32 %power)
5915 declare fp128 @llvm.powi.f128(fp128 %Val, i32 %power)
5916 declare ppc_fp128 @llvm.powi.ppcf128(ppc_fp128 %Val, i32 %power)
5920 <p>The '<tt>llvm.powi.*</tt>' intrinsics return the first operand raised to the
5921 specified (positive or negative) power. The order of evaluation of
5922 multiplications is not defined. When a vector of floating point type is
5923 used, the second argument remains a scalar integer value.</p>
5926 <p>The second argument is an integer power, and the first is a value to raise to
5930 <p>This function returns the first value raised to the second power with an
5931 unspecified sequence of rounding operations.</p>
5935 <!-- _______________________________________________________________________ -->
5936 <div class="doc_subsubsection">
5937 <a name="int_sin">'<tt>llvm.sin.*</tt>' Intrinsic</a>
5940 <div class="doc_text">
5943 <p>This is an overloaded intrinsic. You can use <tt>llvm.sin</tt> on any
5944 floating point or vector of floating point type. Not all targets support all
5948 declare float @llvm.sin.f32(float %Val)
5949 declare double @llvm.sin.f64(double %Val)
5950 declare x86_fp80 @llvm.sin.f80(x86_fp80 %Val)
5951 declare fp128 @llvm.sin.f128(fp128 %Val)
5952 declare ppc_fp128 @llvm.sin.ppcf128(ppc_fp128 %Val)
5956 <p>The '<tt>llvm.sin.*</tt>' intrinsics return the sine of the operand.</p>
5959 <p>The argument and return value are floating point numbers of the same
5963 <p>This function returns the sine of the specified operand, returning the same
5964 values as the libm <tt>sin</tt> functions would, and handles error conditions
5965 in the same way.</p>
5969 <!-- _______________________________________________________________________ -->
5970 <div class="doc_subsubsection">
5971 <a name="int_cos">'<tt>llvm.cos.*</tt>' Intrinsic</a>
5974 <div class="doc_text">
5977 <p>This is an overloaded intrinsic. You can use <tt>llvm.cos</tt> on any
5978 floating point or vector of floating point type. Not all targets support all
5982 declare float @llvm.cos.f32(float %Val)
5983 declare double @llvm.cos.f64(double %Val)
5984 declare x86_fp80 @llvm.cos.f80(x86_fp80 %Val)
5985 declare fp128 @llvm.cos.f128(fp128 %Val)
5986 declare ppc_fp128 @llvm.cos.ppcf128(ppc_fp128 %Val)
5990 <p>The '<tt>llvm.cos.*</tt>' intrinsics return the cosine of the operand.</p>
5993 <p>The argument and return value are floating point numbers of the same
5997 <p>This function returns the cosine of the specified operand, returning the same
5998 values as the libm <tt>cos</tt> functions would, and handles error conditions
5999 in the same way.</p>
6003 <!-- _______________________________________________________________________ -->
6004 <div class="doc_subsubsection">
6005 <a name="int_pow">'<tt>llvm.pow.*</tt>' Intrinsic</a>
6008 <div class="doc_text">
6011 <p>This is an overloaded intrinsic. You can use <tt>llvm.pow</tt> on any
6012 floating point or vector of floating point type. Not all targets support all
6016 declare float @llvm.pow.f32(float %Val, float %Power)
6017 declare double @llvm.pow.f64(double %Val, double %Power)
6018 declare x86_fp80 @llvm.pow.f80(x86_fp80 %Val, x86_fp80 %Power)
6019 declare fp128 @llvm.pow.f128(fp128 %Val, fp128 %Power)
6020 declare ppc_fp128 @llvm.pow.ppcf128(ppc_fp128 %Val, ppc_fp128 Power)
6024 <p>The '<tt>llvm.pow.*</tt>' intrinsics return the first operand raised to the
6025 specified (positive or negative) power.</p>
6028 <p>The second argument is a floating point power, and the first is a value to
6029 raise to that power.</p>
6032 <p>This function returns the first value raised to the second power, returning
6033 the same values as the libm <tt>pow</tt> functions would, and handles error
6034 conditions in the same way.</p>
6038 <!-- ======================================================================= -->
6039 <div class="doc_subsection">
6040 <a name="int_manip">Bit Manipulation Intrinsics</a>
6043 <div class="doc_text">
6045 <p>LLVM provides intrinsics for a few important bit manipulation operations.
6046 These allow efficient code generation for some algorithms.</p>
6050 <!-- _______________________________________________________________________ -->
6051 <div class="doc_subsubsection">
6052 <a name="int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a>
6055 <div class="doc_text">
6058 <p>This is an overloaded intrinsic function. You can use bswap on any integer
6059 type that is an even number of bytes (i.e. BitWidth % 16 == 0).</p>
6062 declare i16 @llvm.bswap.i16(i16 <id>)
6063 declare i32 @llvm.bswap.i32(i32 <id>)
6064 declare i64 @llvm.bswap.i64(i64 <id>)
6068 <p>The '<tt>llvm.bswap</tt>' family of intrinsics is used to byte swap integer
6069 values with an even number of bytes (positive multiple of 16 bits). These
6070 are useful for performing operations on data that is not in the target's
6071 native byte order.</p>
6074 <p>The <tt>llvm.bswap.i16</tt> intrinsic returns an i16 value that has the high
6075 and low byte of the input i16 swapped. Similarly,
6076 the <tt>llvm.bswap.i32</tt> intrinsic returns an i32 value that has the four
6077 bytes of the input i32 swapped, so that if the input bytes are numbered 0, 1,
6078 2, 3 then the returned i32 will have its bytes in 3, 2, 1, 0 order.
6079 The <tt>llvm.bswap.i48</tt>, <tt>llvm.bswap.i64</tt> and other intrinsics
6080 extend this concept to additional even-byte lengths (6 bytes, 8 bytes and
6081 more, respectively).</p>
6085 <!-- _______________________________________________________________________ -->
6086 <div class="doc_subsubsection">
6087 <a name="int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic</a>
6090 <div class="doc_text">
6093 <p>This is an overloaded intrinsic. You can use llvm.ctpop on any integer bit
6094 width. Not all targets support all bit widths however.</p>
6097 declare i8 @llvm.ctpop.i8(i8 <src>)
6098 declare i16 @llvm.ctpop.i16(i16 <src>)
6099 declare i32 @llvm.ctpop.i32(i32 <src>)
6100 declare i64 @llvm.ctpop.i64(i64 <src>)
6101 declare i256 @llvm.ctpop.i256(i256 <src>)
6105 <p>The '<tt>llvm.ctpop</tt>' family of intrinsics counts the number of bits set
6109 <p>The only argument is the value to be counted. The argument may be of any
6110 integer type. The return type must match the argument type.</p>
6113 <p>The '<tt>llvm.ctpop</tt>' intrinsic counts the 1's in a variable.</p>
6117 <!-- _______________________________________________________________________ -->
6118 <div class="doc_subsubsection">
6119 <a name="int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic</a>
6122 <div class="doc_text">
6125 <p>This is an overloaded intrinsic. You can use <tt>llvm.ctlz</tt> on any
6126 integer bit width. Not all targets support all bit widths however.</p>
6129 declare i8 @llvm.ctlz.i8 (i8 <src>)
6130 declare i16 @llvm.ctlz.i16(i16 <src>)
6131 declare i32 @llvm.ctlz.i32(i32 <src>)
6132 declare i64 @llvm.ctlz.i64(i64 <src>)
6133 declare i256 @llvm.ctlz.i256(i256 <src>)
6137 <p>The '<tt>llvm.ctlz</tt>' family of intrinsic functions counts the number of
6138 leading zeros in a variable.</p>
6141 <p>The only argument is the value to be counted. The argument may be of any
6142 integer type. The return type must match the argument type.</p>
6145 <p>The '<tt>llvm.ctlz</tt>' intrinsic counts the leading (most significant)
6146 zeros in a variable. If the src == 0 then the result is the size in bits of
6147 the type of src. For example, <tt>llvm.ctlz(i32 2) = 30</tt>.</p>
6151 <!-- _______________________________________________________________________ -->
6152 <div class="doc_subsubsection">
6153 <a name="int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic</a>
6156 <div class="doc_text">
6159 <p>This is an overloaded intrinsic. You can use <tt>llvm.cttz</tt> on any
6160 integer bit width. Not all targets support all bit widths however.</p>
6163 declare i8 @llvm.cttz.i8 (i8 <src>)
6164 declare i16 @llvm.cttz.i16(i16 <src>)
6165 declare i32 @llvm.cttz.i32(i32 <src>)
6166 declare i64 @llvm.cttz.i64(i64 <src>)
6167 declare i256 @llvm.cttz.i256(i256 <src>)
6171 <p>The '<tt>llvm.cttz</tt>' family of intrinsic functions counts the number of
6175 <p>The only argument is the value to be counted. The argument may be of any
6176 integer type. The return type must match the argument type.</p>
6179 <p>The '<tt>llvm.cttz</tt>' intrinsic counts the trailing (least significant)
6180 zeros in a variable. If the src == 0 then the result is the size in bits of
6181 the type of src. For example, <tt>llvm.cttz(2) = 1</tt>.</p>
6185 <!-- ======================================================================= -->
6186 <div class="doc_subsection">
6187 <a name="int_overflow">Arithmetic with Overflow Intrinsics</a>
6190 <div class="doc_text">
6192 <p>LLVM provides intrinsics for some arithmetic with overflow operations.</p>
6196 <!-- _______________________________________________________________________ -->
6197 <div class="doc_subsubsection">
6198 <a name="int_sadd_overflow">'<tt>llvm.sadd.with.overflow.*</tt>' Intrinsics</a>
6201 <div class="doc_text">
6204 <p>This is an overloaded intrinsic. You can use <tt>llvm.sadd.with.overflow</tt>
6205 on any integer bit width.</p>
6208 declare {i16, i1} @llvm.sadd.with.overflow.i16(i16 %a, i16 %b)
6209 declare {i32, i1} @llvm.sadd.with.overflow.i32(i32 %a, i32 %b)
6210 declare {i64, i1} @llvm.sadd.with.overflow.i64(i64 %a, i64 %b)
6214 <p>The '<tt>llvm.sadd.with.overflow</tt>' family of intrinsic functions perform
6215 a signed addition of the two arguments, and indicate whether an overflow
6216 occurred during the signed summation.</p>
6219 <p>The arguments (%a and %b) and the first element of the result structure may
6220 be of integer types of any bit width, but they must have the same bit
6221 width. The second element of the result structure must be of
6222 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
6223 undergo signed addition.</p>
6226 <p>The '<tt>llvm.sadd.with.overflow</tt>' family of intrinsic functions perform
6227 a signed addition of the two variables. They return a structure — the
6228 first element of which is the signed summation, and the second element of
6229 which is a bit specifying if the signed summation resulted in an
6234 %res = call {i32, i1} @llvm.sadd.with.overflow.i32(i32 %a, i32 %b)
6235 %sum = extractvalue {i32, i1} %res, 0
6236 %obit = extractvalue {i32, i1} %res, 1
6237 br i1 %obit, label %overflow, label %normal
6242 <!-- _______________________________________________________________________ -->
6243 <div class="doc_subsubsection">
6244 <a name="int_uadd_overflow">'<tt>llvm.uadd.with.overflow.*</tt>' Intrinsics</a>
6247 <div class="doc_text">
6250 <p>This is an overloaded intrinsic. You can use <tt>llvm.uadd.with.overflow</tt>
6251 on any integer bit width.</p>
6254 declare {i16, i1} @llvm.uadd.with.overflow.i16(i16 %a, i16 %b)
6255 declare {i32, i1} @llvm.uadd.with.overflow.i32(i32 %a, i32 %b)
6256 declare {i64, i1} @llvm.uadd.with.overflow.i64(i64 %a, i64 %b)
6260 <p>The '<tt>llvm.uadd.with.overflow</tt>' family of intrinsic functions perform
6261 an unsigned addition of the two arguments, and indicate whether a carry
6262 occurred during the unsigned summation.</p>
6265 <p>The arguments (%a and %b) and the first element of the result structure may
6266 be of integer types of any bit width, but they must have the same bit
6267 width. The second element of the result structure must be of
6268 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
6269 undergo unsigned addition.</p>
6272 <p>The '<tt>llvm.uadd.with.overflow</tt>' family of intrinsic functions perform
6273 an unsigned addition of the two arguments. They return a structure —
6274 the first element of which is the sum, and the second element of which is a
6275 bit specifying if the unsigned summation resulted in a carry.</p>
6279 %res = call {i32, i1} @llvm.uadd.with.overflow.i32(i32 %a, i32 %b)
6280 %sum = extractvalue {i32, i1} %res, 0
6281 %obit = extractvalue {i32, i1} %res, 1
6282 br i1 %obit, label %carry, label %normal
6287 <!-- _______________________________________________________________________ -->
6288 <div class="doc_subsubsection">
6289 <a name="int_ssub_overflow">'<tt>llvm.ssub.with.overflow.*</tt>' Intrinsics</a>
6292 <div class="doc_text">
6295 <p>This is an overloaded intrinsic. You can use <tt>llvm.ssub.with.overflow</tt>
6296 on any integer bit width.</p>
6299 declare {i16, i1} @llvm.ssub.with.overflow.i16(i16 %a, i16 %b)
6300 declare {i32, i1} @llvm.ssub.with.overflow.i32(i32 %a, i32 %b)
6301 declare {i64, i1} @llvm.ssub.with.overflow.i64(i64 %a, i64 %b)
6305 <p>The '<tt>llvm.ssub.with.overflow</tt>' family of intrinsic functions perform
6306 a signed subtraction of the two arguments, and indicate whether an overflow
6307 occurred during the signed subtraction.</p>
6310 <p>The arguments (%a and %b) and the first element of the result structure may
6311 be of integer types of any bit width, but they must have the same bit
6312 width. The second element of the result structure must be of
6313 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
6314 undergo signed subtraction.</p>
6317 <p>The '<tt>llvm.ssub.with.overflow</tt>' family of intrinsic functions perform
6318 a signed subtraction of the two arguments. They return a structure —
6319 the first element of which is the subtraction, and the second element of
6320 which is a bit specifying if the signed subtraction resulted in an
6325 %res = call {i32, i1} @llvm.ssub.with.overflow.i32(i32 %a, i32 %b)
6326 %sum = extractvalue {i32, i1} %res, 0
6327 %obit = extractvalue {i32, i1} %res, 1
6328 br i1 %obit, label %overflow, label %normal
6333 <!-- _______________________________________________________________________ -->
6334 <div class="doc_subsubsection">
6335 <a name="int_usub_overflow">'<tt>llvm.usub.with.overflow.*</tt>' Intrinsics</a>
6338 <div class="doc_text">
6341 <p>This is an overloaded intrinsic. You can use <tt>llvm.usub.with.overflow</tt>
6342 on any integer bit width.</p>
6345 declare {i16, i1} @llvm.usub.with.overflow.i16(i16 %a, i16 %b)
6346 declare {i32, i1} @llvm.usub.with.overflow.i32(i32 %a, i32 %b)
6347 declare {i64, i1} @llvm.usub.with.overflow.i64(i64 %a, i64 %b)
6351 <p>The '<tt>llvm.usub.with.overflow</tt>' family of intrinsic functions perform
6352 an unsigned subtraction of the two arguments, and indicate whether an
6353 overflow occurred during the unsigned subtraction.</p>
6356 <p>The arguments (%a and %b) and the first element of the result structure may
6357 be of integer types of any bit width, but they must have the same bit
6358 width. The second element of the result structure must be of
6359 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
6360 undergo unsigned subtraction.</p>
6363 <p>The '<tt>llvm.usub.with.overflow</tt>' family of intrinsic functions perform
6364 an unsigned subtraction of the two arguments. They return a structure —
6365 the first element of which is the subtraction, and the second element of
6366 which is a bit specifying if the unsigned subtraction resulted in an
6371 %res = call {i32, i1} @llvm.usub.with.overflow.i32(i32 %a, i32 %b)
6372 %sum = extractvalue {i32, i1} %res, 0
6373 %obit = extractvalue {i32, i1} %res, 1
6374 br i1 %obit, label %overflow, label %normal
6379 <!-- _______________________________________________________________________ -->
6380 <div class="doc_subsubsection">
6381 <a name="int_smul_overflow">'<tt>llvm.smul.with.overflow.*</tt>' Intrinsics</a>
6384 <div class="doc_text">
6387 <p>This is an overloaded intrinsic. You can use <tt>llvm.smul.with.overflow</tt>
6388 on any integer bit width.</p>
6391 declare {i16, i1} @llvm.smul.with.overflow.i16(i16 %a, i16 %b)
6392 declare {i32, i1} @llvm.smul.with.overflow.i32(i32 %a, i32 %b)
6393 declare {i64, i1} @llvm.smul.with.overflow.i64(i64 %a, i64 %b)
6398 <p>The '<tt>llvm.smul.with.overflow</tt>' family of intrinsic functions perform
6399 a signed multiplication of the two arguments, and indicate whether an
6400 overflow occurred during the signed multiplication.</p>
6403 <p>The arguments (%a and %b) and the first element of the result structure may
6404 be of integer types of any bit width, but they must have the same bit
6405 width. The second element of the result structure must be of
6406 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
6407 undergo signed multiplication.</p>
6410 <p>The '<tt>llvm.smul.with.overflow</tt>' family of intrinsic functions perform
6411 a signed multiplication of the two arguments. They return a structure —
6412 the first element of which is the multiplication, and the second element of
6413 which is a bit specifying if the signed multiplication resulted in an
6418 %res = call {i32, i1} @llvm.smul.with.overflow.i32(i32 %a, i32 %b)
6419 %sum = extractvalue {i32, i1} %res, 0
6420 %obit = extractvalue {i32, i1} %res, 1
6421 br i1 %obit, label %overflow, label %normal
6426 <!-- _______________________________________________________________________ -->
6427 <div class="doc_subsubsection">
6428 <a name="int_umul_overflow">'<tt>llvm.umul.with.overflow.*</tt>' Intrinsics</a>
6431 <div class="doc_text">
6434 <p>This is an overloaded intrinsic. You can use <tt>llvm.umul.with.overflow</tt>
6435 on any integer bit width.</p>
6438 declare {i16, i1} @llvm.umul.with.overflow.i16(i16 %a, i16 %b)
6439 declare {i32, i1} @llvm.umul.with.overflow.i32(i32 %a, i32 %b)
6440 declare {i64, i1} @llvm.umul.with.overflow.i64(i64 %a, i64 %b)
6444 <p>The '<tt>llvm.umul.with.overflow</tt>' family of intrinsic functions perform
6445 a unsigned multiplication of the two arguments, and indicate whether an
6446 overflow occurred during the unsigned multiplication.</p>
6449 <p>The arguments (%a and %b) and the first element of the result structure may
6450 be of integer types of any bit width, but they must have the same bit
6451 width. The second element of the result structure must be of
6452 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
6453 undergo unsigned multiplication.</p>
6456 <p>The '<tt>llvm.umul.with.overflow</tt>' family of intrinsic functions perform
6457 an unsigned multiplication of the two arguments. They return a structure
6458 — the first element of which is the multiplication, and the second
6459 element of which is a bit specifying if the unsigned multiplication resulted
6464 %res = call {i32, i1} @llvm.umul.with.overflow.i32(i32 %a, i32 %b)
6465 %sum = extractvalue {i32, i1} %res, 0
6466 %obit = extractvalue {i32, i1} %res, 1
6467 br i1 %obit, label %overflow, label %normal
6472 <!-- ======================================================================= -->
6473 <div class="doc_subsection">
6474 <a name="int_debugger">Debugger Intrinsics</a>
6477 <div class="doc_text">
6479 <p>The LLVM debugger intrinsics (which all start with <tt>llvm.dbg.</tt>
6480 prefix), are described in
6481 the <a href="SourceLevelDebugging.html#format_common_intrinsics">LLVM Source
6482 Level Debugging</a> document.</p>
6486 <!-- ======================================================================= -->
6487 <div class="doc_subsection">
6488 <a name="int_eh">Exception Handling Intrinsics</a>
6491 <div class="doc_text">
6493 <p>The LLVM exception handling intrinsics (which all start with
6494 <tt>llvm.eh.</tt> prefix), are described in
6495 the <a href="ExceptionHandling.html#format_common_intrinsics">LLVM Exception
6496 Handling</a> document.</p>
6500 <!-- ======================================================================= -->
6501 <div class="doc_subsection">
6502 <a name="int_trampoline">Trampoline Intrinsic</a>
6505 <div class="doc_text">
6507 <p>This intrinsic makes it possible to excise one parameter, marked with
6508 the <tt>nest</tt> attribute, from a function. The result is a callable
6509 function pointer lacking the nest parameter - the caller does not need to
6510 provide a value for it. Instead, the value to use is stored in advance in a
6511 "trampoline", a block of memory usually allocated on the stack, which also
6512 contains code to splice the nest value into the argument list. This is used
6513 to implement the GCC nested function address extension.</p>
6515 <p>For example, if the function is
6516 <tt>i32 f(i8* nest %c, i32 %x, i32 %y)</tt> then the resulting function
6517 pointer has signature <tt>i32 (i32, i32)*</tt>. It can be created as
6520 <div class="doc_code">
6522 %tramp = alloca [10 x i8], align 4 ; size and alignment only correct for X86
6523 %tramp1 = getelementptr [10 x i8]* %tramp, i32 0, i32 0
6524 %p = call i8* @llvm.init.trampoline( i8* %tramp1, i8* bitcast (i32 (i8* nest , i32, i32)* @f to i8*), i8* %nval )
6525 %fp = bitcast i8* %p to i32 (i32, i32)*
6529 <p>The call <tt>%val = call i32 %fp( i32 %x, i32 %y )</tt> is then equivalent
6530 to <tt>%val = call i32 %f( i8* %nval, i32 %x, i32 %y )</tt>.</p>
6534 <!-- _______________________________________________________________________ -->
6535 <div class="doc_subsubsection">
6536 <a name="int_it">'<tt>llvm.init.trampoline</tt>' Intrinsic</a>
6539 <div class="doc_text">
6543 declare i8* @llvm.init.trampoline(i8* <tramp>, i8* <func>, i8* <nval>)
6547 <p>This fills the memory pointed to by <tt>tramp</tt> with code and returns a
6548 function pointer suitable for executing it.</p>
6551 <p>The <tt>llvm.init.trampoline</tt> intrinsic takes three arguments, all
6552 pointers. The <tt>tramp</tt> argument must point to a sufficiently large and
6553 sufficiently aligned block of memory; this memory is written to by the
6554 intrinsic. Note that the size and the alignment are target-specific - LLVM
6555 currently provides no portable way of determining them, so a front-end that
6556 generates this intrinsic needs to have some target-specific knowledge.
6557 The <tt>func</tt> argument must hold a function bitcast to
6558 an <tt>i8*</tt>.</p>
6561 <p>The block of memory pointed to by <tt>tramp</tt> is filled with target
6562 dependent code, turning it into a function. A pointer to this function is
6563 returned, but needs to be bitcast to an <a href="#int_trampoline">appropriate
6564 function pointer type</a> before being called. The new function's signature
6565 is the same as that of <tt>func</tt> with any arguments marked with
6566 the <tt>nest</tt> attribute removed. At most one such <tt>nest</tt> argument
6567 is allowed, and it must be of pointer type. Calling the new function is
6568 equivalent to calling <tt>func</tt> with the same argument list, but
6569 with <tt>nval</tt> used for the missing <tt>nest</tt> argument. If, after
6570 calling <tt>llvm.init.trampoline</tt>, the memory pointed to
6571 by <tt>tramp</tt> is modified, then the effect of any later call to the
6572 returned function pointer is undefined.</p>
6576 <!-- ======================================================================= -->
6577 <div class="doc_subsection">
6578 <a name="int_atomics">Atomic Operations and Synchronization Intrinsics</a>
6581 <div class="doc_text">
6583 <p>These intrinsic functions expand the "universal IR" of LLVM to represent
6584 hardware constructs for atomic operations and memory synchronization. This
6585 provides an interface to the hardware, not an interface to the programmer. It
6586 is aimed at a low enough level to allow any programming models or APIs
6587 (Application Programming Interfaces) which need atomic behaviors to map
6588 cleanly onto it. It is also modeled primarily on hardware behavior. Just as
6589 hardware provides a "universal IR" for source languages, it also provides a
6590 starting point for developing a "universal" atomic operation and
6591 synchronization IR.</p>
6593 <p>These do <em>not</em> form an API such as high-level threading libraries,
6594 software transaction memory systems, atomic primitives, and intrinsic
6595 functions as found in BSD, GNU libc, atomic_ops, APR, and other system and
6596 application libraries. The hardware interface provided by LLVM should allow
6597 a clean implementation of all of these APIs and parallel programming models.
6598 No one model or paradigm should be selected above others unless the hardware
6599 itself ubiquitously does so.</p>
6603 <!-- _______________________________________________________________________ -->
6604 <div class="doc_subsubsection">
6605 <a name="int_memory_barrier">'<tt>llvm.memory.barrier</tt>' Intrinsic</a>
6607 <div class="doc_text">
6610 declare void @llvm.memory.barrier( i1 <ll>, i1 <ls>, i1 <sl>, i1 <ss>, i1 <device> )
6614 <p>The <tt>llvm.memory.barrier</tt> intrinsic guarantees ordering between
6615 specific pairs of memory access types.</p>
6618 <p>The <tt>llvm.memory.barrier</tt> intrinsic requires five boolean arguments.
6619 The first four arguments enables a specific barrier as listed below. The
6620 fith argument specifies that the barrier applies to io or device or uncached
6624 <li><tt>ll</tt>: load-load barrier</li>
6625 <li><tt>ls</tt>: load-store barrier</li>
6626 <li><tt>sl</tt>: store-load barrier</li>
6627 <li><tt>ss</tt>: store-store barrier</li>
6628 <li><tt>device</tt>: barrier applies to device and uncached memory also.</li>
6632 <p>This intrinsic causes the system to enforce some ordering constraints upon
6633 the loads and stores of the program. This barrier does not
6634 indicate <em>when</em> any events will occur, it only enforces
6635 an <em>order</em> in which they occur. For any of the specified pairs of load
6636 and store operations (f.ex. load-load, or store-load), all of the first
6637 operations preceding the barrier will complete before any of the second
6638 operations succeeding the barrier begin. Specifically the semantics for each
6639 pairing is as follows:</p>
6642 <li><tt>ll</tt>: All loads before the barrier must complete before any load
6643 after the barrier begins.</li>
6644 <li><tt>ls</tt>: All loads before the barrier must complete before any
6645 store after the barrier begins.</li>
6646 <li><tt>ss</tt>: All stores before the barrier must complete before any
6647 store after the barrier begins.</li>
6648 <li><tt>sl</tt>: All stores before the barrier must complete before any
6649 load after the barrier begins.</li>
6652 <p>These semantics are applied with a logical "and" behavior when more than one
6653 is enabled in a single memory barrier intrinsic.</p>
6655 <p>Backends may implement stronger barriers than those requested when they do
6656 not support as fine grained a barrier as requested. Some architectures do
6657 not need all types of barriers and on such architectures, these become
6662 %mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
6663 %ptr = bitcast i8* %mallocP to i32*
6666 %result1 = load i32* %ptr <i>; yields {i32}:result1 = 4</i>
6667 call void @llvm.memory.barrier( i1 false, i1 true, i1 false, i1 false )
6668 <i>; guarantee the above finishes</i>
6669 store i32 8, %ptr <i>; before this begins</i>
6674 <!-- _______________________________________________________________________ -->
6675 <div class="doc_subsubsection">
6676 <a name="int_atomic_cmp_swap">'<tt>llvm.atomic.cmp.swap.*</tt>' Intrinsic</a>
6679 <div class="doc_text">
6682 <p>This is an overloaded intrinsic. You can use <tt>llvm.atomic.cmp.swap</tt> on
6683 any integer bit width and for different address spaces. Not all targets
6684 support all bit widths however.</p>
6687 declare i8 @llvm.atomic.cmp.swap.i8.p0i8( i8* <ptr>, i8 <cmp>, i8 <val> )
6688 declare i16 @llvm.atomic.cmp.swap.i16.p0i16( i16* <ptr>, i16 <cmp>, i16 <val> )
6689 declare i32 @llvm.atomic.cmp.swap.i32.p0i32( i32* <ptr>, i32 <cmp>, i32 <val> )
6690 declare i64 @llvm.atomic.cmp.swap.i64.p0i64( i64* <ptr>, i64 <cmp>, i64 <val> )
6694 <p>This loads a value in memory and compares it to a given value. If they are
6695 equal, it stores a new value into the memory.</p>
6698 <p>The <tt>llvm.atomic.cmp.swap</tt> intrinsic takes three arguments. The result
6699 as well as both <tt>cmp</tt> and <tt>val</tt> must be integer values with the
6700 same bit width. The <tt>ptr</tt> argument must be a pointer to a value of
6701 this integer type. While any bit width integer may be used, targets may only
6702 lower representations they support in hardware.</p>
6705 <p>This entire intrinsic must be executed atomically. It first loads the value
6706 in memory pointed to by <tt>ptr</tt> and compares it with the
6707 value <tt>cmp</tt>. If they are equal, <tt>val</tt> is stored into the
6708 memory. The loaded value is yielded in all cases. This provides the
6709 equivalent of an atomic compare-and-swap operation within the SSA
6714 %mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
6715 %ptr = bitcast i8* %mallocP to i32*
6718 %val1 = add i32 4, 4
6719 %result1 = call i32 @llvm.atomic.cmp.swap.i32.p0i32( i32* %ptr, i32 4, %val1 )
6720 <i>; yields {i32}:result1 = 4</i>
6721 %stored1 = icmp eq i32 %result1, 4 <i>; yields {i1}:stored1 = true</i>
6722 %memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 8</i>
6724 %val2 = add i32 1, 1
6725 %result2 = call i32 @llvm.atomic.cmp.swap.i32.p0i32( i32* %ptr, i32 5, %val2 )
6726 <i>; yields {i32}:result2 = 8</i>
6727 %stored2 = icmp eq i32 %result2, 5 <i>; yields {i1}:stored2 = false</i>
6729 %memval2 = load i32* %ptr <i>; yields {i32}:memval2 = 8</i>
6734 <!-- _______________________________________________________________________ -->
6735 <div class="doc_subsubsection">
6736 <a name="int_atomic_swap">'<tt>llvm.atomic.swap.*</tt>' Intrinsic</a>
6738 <div class="doc_text">
6741 <p>This is an overloaded intrinsic. You can use <tt>llvm.atomic.swap</tt> on any
6742 integer bit width. Not all targets support all bit widths however.</p>
6745 declare i8 @llvm.atomic.swap.i8.p0i8( i8* <ptr>, i8 <val> )
6746 declare i16 @llvm.atomic.swap.i16.p0i16( i16* <ptr>, i16 <val> )
6747 declare i32 @llvm.atomic.swap.i32.p0i32( i32* <ptr>, i32 <val> )
6748 declare i64 @llvm.atomic.swap.i64.p0i64( i64* <ptr>, i64 <val> )
6752 <p>This intrinsic loads the value stored in memory at <tt>ptr</tt> and yields
6753 the value from memory. It then stores the value in <tt>val</tt> in the memory
6754 at <tt>ptr</tt>.</p>
6757 <p>The <tt>llvm.atomic.swap</tt> intrinsic takes two arguments. Both
6758 the <tt>val</tt> argument and the result must be integers of the same bit
6759 width. The first argument, <tt>ptr</tt>, must be a pointer to a value of this
6760 integer type. The targets may only lower integer representations they
6764 <p>This intrinsic loads the value pointed to by <tt>ptr</tt>, yields it, and
6765 stores <tt>val</tt> back into <tt>ptr</tt> atomically. This provides the
6766 equivalent of an atomic swap operation within the SSA framework.</p>
6770 %mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
6771 %ptr = bitcast i8* %mallocP to i32*
6774 %val1 = add i32 4, 4
6775 %result1 = call i32 @llvm.atomic.swap.i32.p0i32( i32* %ptr, i32 %val1 )
6776 <i>; yields {i32}:result1 = 4</i>
6777 %stored1 = icmp eq i32 %result1, 4 <i>; yields {i1}:stored1 = true</i>
6778 %memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 8</i>
6780 %val2 = add i32 1, 1
6781 %result2 = call i32 @llvm.atomic.swap.i32.p0i32( i32* %ptr, i32 %val2 )
6782 <i>; yields {i32}:result2 = 8</i>
6784 %stored2 = icmp eq i32 %result2, 8 <i>; yields {i1}:stored2 = true</i>
6785 %memval2 = load i32* %ptr <i>; yields {i32}:memval2 = 2</i>
6790 <!-- _______________________________________________________________________ -->
6791 <div class="doc_subsubsection">
6792 <a name="int_atomic_load_add">'<tt>llvm.atomic.load.add.*</tt>' Intrinsic</a>
6796 <div class="doc_text">
6799 <p>This is an overloaded intrinsic. You can use <tt>llvm.atomic.load.add</tt> on
6800 any integer bit width. Not all targets support all bit widths however.</p>
6803 declare i8 @llvm.atomic.load.add.i8..p0i8( i8* <ptr>, i8 <delta> )
6804 declare i16 @llvm.atomic.load.add.i16..p0i16( i16* <ptr>, i16 <delta> )
6805 declare i32 @llvm.atomic.load.add.i32..p0i32( i32* <ptr>, i32 <delta> )
6806 declare i64 @llvm.atomic.load.add.i64..p0i64( i64* <ptr>, i64 <delta> )
6810 <p>This intrinsic adds <tt>delta</tt> to the value stored in memory
6811 at <tt>ptr</tt>. It yields the original value at <tt>ptr</tt>.</p>
6814 <p>The intrinsic takes two arguments, the first a pointer to an integer value
6815 and the second an integer value. The result is also an integer value. These
6816 integer types can have any bit width, but they must all have the same bit
6817 width. The targets may only lower integer representations they support.</p>
6820 <p>This intrinsic does a series of operations atomically. It first loads the
6821 value stored at <tt>ptr</tt>. It then adds <tt>delta</tt>, stores the result
6822 to <tt>ptr</tt>. It yields the original value stored at <tt>ptr</tt>.</p>
6826 %mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
6827 %ptr = bitcast i8* %mallocP to i32*
6829 %result1 = call i32 @llvm.atomic.load.add.i32.p0i32( i32* %ptr, i32 4 )
6830 <i>; yields {i32}:result1 = 4</i>
6831 %result2 = call i32 @llvm.atomic.load.add.i32.p0i32( i32* %ptr, i32 2 )
6832 <i>; yields {i32}:result2 = 8</i>
6833 %result3 = call i32 @llvm.atomic.load.add.i32.p0i32( i32* %ptr, i32 5 )
6834 <i>; yields {i32}:result3 = 10</i>
6835 %memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 15</i>
6840 <!-- _______________________________________________________________________ -->
6841 <div class="doc_subsubsection">
6842 <a name="int_atomic_load_sub">'<tt>llvm.atomic.load.sub.*</tt>' Intrinsic</a>
6846 <div class="doc_text">
6849 <p>This is an overloaded intrinsic. You can use <tt>llvm.atomic.load.sub</tt> on
6850 any integer bit width and for different address spaces. Not all targets
6851 support all bit widths however.</p>
6854 declare i8 @llvm.atomic.load.sub.i8.p0i32( i8* <ptr>, i8 <delta> )
6855 declare i16 @llvm.atomic.load.sub.i16.p0i32( i16* <ptr>, i16 <delta> )
6856 declare i32 @llvm.atomic.load.sub.i32.p0i32( i32* <ptr>, i32 <delta> )
6857 declare i64 @llvm.atomic.load.sub.i64.p0i32( i64* <ptr>, i64 <delta> )
6861 <p>This intrinsic subtracts <tt>delta</tt> to the value stored in memory at
6862 <tt>ptr</tt>. It yields the original value at <tt>ptr</tt>.</p>
6865 <p>The intrinsic takes two arguments, the first a pointer to an integer value
6866 and the second an integer value. The result is also an integer value. These
6867 integer types can have any bit width, but they must all have the same bit
6868 width. The targets may only lower integer representations they support.</p>
6871 <p>This intrinsic does a series of operations atomically. It first loads the
6872 value stored at <tt>ptr</tt>. It then subtracts <tt>delta</tt>, stores the
6873 result to <tt>ptr</tt>. It yields the original value stored
6874 at <tt>ptr</tt>.</p>
6878 %mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
6879 %ptr = bitcast i8* %mallocP to i32*
6881 %result1 = call i32 @llvm.atomic.load.sub.i32.p0i32( i32* %ptr, i32 4 )
6882 <i>; yields {i32}:result1 = 8</i>
6883 %result2 = call i32 @llvm.atomic.load.sub.i32.p0i32( i32* %ptr, i32 2 )
6884 <i>; yields {i32}:result2 = 4</i>
6885 %result3 = call i32 @llvm.atomic.load.sub.i32.p0i32( i32* %ptr, i32 5 )
6886 <i>; yields {i32}:result3 = 2</i>
6887 %memval1 = load i32* %ptr <i>; yields {i32}:memval1 = -3</i>
6892 <!-- _______________________________________________________________________ -->
6893 <div class="doc_subsubsection">
6894 <a name="int_atomic_load_and">'<tt>llvm.atomic.load.and.*</tt>' Intrinsic</a><br>
6895 <a name="int_atomic_load_nand">'<tt>llvm.atomic.load.nand.*</tt>' Intrinsic</a><br>
6896 <a name="int_atomic_load_or">'<tt>llvm.atomic.load.or.*</tt>' Intrinsic</a><br>
6897 <a name="int_atomic_load_xor">'<tt>llvm.atomic.load.xor.*</tt>' Intrinsic</a><br>
6900 <div class="doc_text">
6903 <p>These are overloaded intrinsics. You can
6904 use <tt>llvm.atomic.load_and</tt>, <tt>llvm.atomic.load_nand</tt>,
6905 <tt>llvm.atomic.load_or</tt>, and <tt>llvm.atomic.load_xor</tt> on any integer
6906 bit width and for different address spaces. Not all targets support all bit
6910 declare i8 @llvm.atomic.load.and.i8.p0i8( i8* <ptr>, i8 <delta> )
6911 declare i16 @llvm.atomic.load.and.i16.p0i16( i16* <ptr>, i16 <delta> )
6912 declare i32 @llvm.atomic.load.and.i32.p0i32( i32* <ptr>, i32 <delta> )
6913 declare i64 @llvm.atomic.load.and.i64.p0i64( i64* <ptr>, i64 <delta> )
6917 declare i8 @llvm.atomic.load.or.i8.p0i8( i8* <ptr>, i8 <delta> )
6918 declare i16 @llvm.atomic.load.or.i16.p0i16( i16* <ptr>, i16 <delta> )
6919 declare i32 @llvm.atomic.load.or.i32.p0i32( i32* <ptr>, i32 <delta> )
6920 declare i64 @llvm.atomic.load.or.i64.p0i64( i64* <ptr>, i64 <delta> )
6924 declare i8 @llvm.atomic.load.nand.i8.p0i32( i8* <ptr>, i8 <delta> )
6925 declare i16 @llvm.atomic.load.nand.i16.p0i32( i16* <ptr>, i16 <delta> )
6926 declare i32 @llvm.atomic.load.nand.i32.p0i32( i32* <ptr>, i32 <delta> )
6927 declare i64 @llvm.atomic.load.nand.i64.p0i32( i64* <ptr>, i64 <delta> )
6931 declare i8 @llvm.atomic.load.xor.i8.p0i32( i8* <ptr>, i8 <delta> )
6932 declare i16 @llvm.atomic.load.xor.i16.p0i32( i16* <ptr>, i16 <delta> )
6933 declare i32 @llvm.atomic.load.xor.i32.p0i32( i32* <ptr>, i32 <delta> )
6934 declare i64 @llvm.atomic.load.xor.i64.p0i32( i64* <ptr>, i64 <delta> )
6938 <p>These intrinsics bitwise the operation (and, nand, or, xor) <tt>delta</tt> to
6939 the value stored in memory at <tt>ptr</tt>. It yields the original value
6940 at <tt>ptr</tt>.</p>
6943 <p>These intrinsics take two arguments, the first a pointer to an integer value
6944 and the second an integer value. The result is also an integer value. These
6945 integer types can have any bit width, but they must all have the same bit
6946 width. The targets may only lower integer representations they support.</p>
6949 <p>These intrinsics does a series of operations atomically. They first load the
6950 value stored at <tt>ptr</tt>. They then do the bitwise
6951 operation <tt>delta</tt>, store the result to <tt>ptr</tt>. They yield the
6952 original value stored at <tt>ptr</tt>.</p>
6956 %mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
6957 %ptr = bitcast i8* %mallocP to i32*
6958 store i32 0x0F0F, %ptr
6959 %result0 = call i32 @llvm.atomic.load.nand.i32.p0i32( i32* %ptr, i32 0xFF )
6960 <i>; yields {i32}:result0 = 0x0F0F</i>
6961 %result1 = call i32 @llvm.atomic.load.and.i32.p0i32( i32* %ptr, i32 0xFF )
6962 <i>; yields {i32}:result1 = 0xFFFFFFF0</i>
6963 %result2 = call i32 @llvm.atomic.load.or.i32.p0i32( i32* %ptr, i32 0F )
6964 <i>; yields {i32}:result2 = 0xF0</i>
6965 %result3 = call i32 @llvm.atomic.load.xor.i32.p0i32( i32* %ptr, i32 0F )
6966 <i>; yields {i32}:result3 = FF</i>
6967 %memval1 = load i32* %ptr <i>; yields {i32}:memval1 = F0</i>
6972 <!-- _______________________________________________________________________ -->
6973 <div class="doc_subsubsection">
6974 <a name="int_atomic_load_max">'<tt>llvm.atomic.load.max.*</tt>' Intrinsic</a><br>
6975 <a name="int_atomic_load_min">'<tt>llvm.atomic.load.min.*</tt>' Intrinsic</a><br>
6976 <a name="int_atomic_load_umax">'<tt>llvm.atomic.load.umax.*</tt>' Intrinsic</a><br>
6977 <a name="int_atomic_load_umin">'<tt>llvm.atomic.load.umin.*</tt>' Intrinsic</a><br>
6980 <div class="doc_text">
6983 <p>These are overloaded intrinsics. You can use <tt>llvm.atomic.load_max</tt>,
6984 <tt>llvm.atomic.load_min</tt>, <tt>llvm.atomic.load_umax</tt>, and
6985 <tt>llvm.atomic.load_umin</tt> on any integer bit width and for different
6986 address spaces. Not all targets support all bit widths however.</p>
6989 declare i8 @llvm.atomic.load.max.i8.p0i8( i8* <ptr>, i8 <delta> )
6990 declare i16 @llvm.atomic.load.max.i16.p0i16( i16* <ptr>, i16 <delta> )
6991 declare i32 @llvm.atomic.load.max.i32.p0i32( i32* <ptr>, i32 <delta> )
6992 declare i64 @llvm.atomic.load.max.i64.p0i64( i64* <ptr>, i64 <delta> )
6996 declare i8 @llvm.atomic.load.min.i8.p0i8( i8* <ptr>, i8 <delta> )
6997 declare i16 @llvm.atomic.load.min.i16.p0i16( i16* <ptr>, i16 <delta> )
6998 declare i32 @llvm.atomic.load.min.i32..p0i32( i32* <ptr>, i32 <delta> )
6999 declare i64 @llvm.atomic.load.min.i64..p0i64( i64* <ptr>, i64 <delta> )
7003 declare i8 @llvm.atomic.load.umax.i8.p0i8( i8* <ptr>, i8 <delta> )
7004 declare i16 @llvm.atomic.load.umax.i16.p0i16( i16* <ptr>, i16 <delta> )
7005 declare i32 @llvm.atomic.load.umax.i32.p0i32( i32* <ptr>, i32 <delta> )
7006 declare i64 @llvm.atomic.load.umax.i64.p0i64( i64* <ptr>, i64 <delta> )
7010 declare i8 @llvm.atomic.load.umin.i8..p0i8( i8* <ptr>, i8 <delta> )
7011 declare i16 @llvm.atomic.load.umin.i16.p0i16( i16* <ptr>, i16 <delta> )
7012 declare i32 @llvm.atomic.load.umin.i32..p0i32( i32* <ptr>, i32 <delta> )
7013 declare i64 @llvm.atomic.load.umin.i64..p0i64( i64* <ptr>, i64 <delta> )
7017 <p>These intrinsics takes the signed or unsigned minimum or maximum of
7018 <tt>delta</tt> and the value stored in memory at <tt>ptr</tt>. It yields the
7019 original value at <tt>ptr</tt>.</p>
7022 <p>These intrinsics take two arguments, the first a pointer to an integer value
7023 and the second an integer value. The result is also an integer value. These
7024 integer types can have any bit width, but they must all have the same bit
7025 width. The targets may only lower integer representations they support.</p>
7028 <p>These intrinsics does a series of operations atomically. They first load the
7029 value stored at <tt>ptr</tt>. They then do the signed or unsigned min or
7030 max <tt>delta</tt> and the value, store the result to <tt>ptr</tt>. They
7031 yield the original value stored at <tt>ptr</tt>.</p>
7035 %mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
7036 %ptr = bitcast i8* %mallocP to i32*
7038 %result0 = call i32 @llvm.atomic.load.min.i32.p0i32( i32* %ptr, i32 -2 )
7039 <i>; yields {i32}:result0 = 7</i>
7040 %result1 = call i32 @llvm.atomic.load.max.i32.p0i32( i32* %ptr, i32 8 )
7041 <i>; yields {i32}:result1 = -2</i>
7042 %result2 = call i32 @llvm.atomic.load.umin.i32.p0i32( i32* %ptr, i32 10 )
7043 <i>; yields {i32}:result2 = 8</i>
7044 %result3 = call i32 @llvm.atomic.load.umax.i32.p0i32( i32* %ptr, i32 30 )
7045 <i>; yields {i32}:result3 = 8</i>
7046 %memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 30</i>
7052 <!-- ======================================================================= -->
7053 <div class="doc_subsection">
7054 <a name="int_memorymarkers">Memory Use Markers</a>
7057 <div class="doc_text">
7059 <p>This class of intrinsics exists to information about the lifetime of memory
7060 objects and ranges where variables are immutable.</p>
7064 <!-- _______________________________________________________________________ -->
7065 <div class="doc_subsubsection">
7066 <a name="int_lifetime_start">'<tt>llvm.lifetime.start</tt>' Intrinsic</a>
7069 <div class="doc_text">
7073 declare void @llvm.lifetime.start(i64 <size>, i8* nocapture <ptr>)
7077 <p>The '<tt>llvm.lifetime.start</tt>' intrinsic specifies the start of a memory
7078 object's lifetime.</p>
7081 <p>The first argument is a constant integer representing the size of the
7082 object, or -1 if it is variable sized. The second argument is a pointer to
7086 <p>This intrinsic indicates that before this point in the code, the value of the
7087 memory pointed to by <tt>ptr</tt> is dead. This means that it is known to
7088 never be used and has an undefined value. A load from the pointer that
7089 precedes this intrinsic can be replaced with
7090 <tt>'<a href="#undefvalues">undef</a>'</tt>.</p>
7094 <!-- _______________________________________________________________________ -->
7095 <div class="doc_subsubsection">
7096 <a name="int_lifetime_end">'<tt>llvm.lifetime.end</tt>' Intrinsic</a>
7099 <div class="doc_text">
7103 declare void @llvm.lifetime.end(i64 <size>, i8* nocapture <ptr>)
7107 <p>The '<tt>llvm.lifetime.end</tt>' intrinsic specifies the end of a memory
7108 object's lifetime.</p>
7111 <p>The first argument is a constant integer representing the size of the
7112 object, or -1 if it is variable sized. The second argument is a pointer to
7116 <p>This intrinsic indicates that after this point in the code, the value of the
7117 memory pointed to by <tt>ptr</tt> is dead. This means that it is known to
7118 never be used and has an undefined value. Any stores into the memory object
7119 following this intrinsic may be removed as dead.
7123 <!-- _______________________________________________________________________ -->
7124 <div class="doc_subsubsection">
7125 <a name="int_invariant_start">'<tt>llvm.invariant.start</tt>' Intrinsic</a>
7128 <div class="doc_text">
7132 declare {}* @llvm.invariant.start(i64 <size>, i8* nocapture <ptr>) readonly
7136 <p>The '<tt>llvm.invariant.start</tt>' intrinsic specifies that the contents of
7137 a memory object will not change.</p>
7140 <p>The first argument is a constant integer representing the size of the
7141 object, or -1 if it is variable sized. The second argument is a pointer to
7145 <p>This intrinsic indicates that until an <tt>llvm.invariant.end</tt> that uses
7146 the return value, the referenced memory location is constant and
7151 <!-- _______________________________________________________________________ -->
7152 <div class="doc_subsubsection">
7153 <a name="int_invariant_end">'<tt>llvm.invariant.end</tt>' Intrinsic</a>
7156 <div class="doc_text">
7160 declare void @llvm.invariant.end({}* <start>, i64 <size>, i8* nocapture <ptr>)
7164 <p>The '<tt>llvm.invariant.end</tt>' intrinsic specifies that the contents of
7165 a memory object are mutable.</p>
7168 <p>The first argument is the matching <tt>llvm.invariant.start</tt> intrinsic.
7169 The second argument is a constant integer representing the size of the
7170 object, or -1 if it is variable sized and the third argument is a pointer
7174 <p>This intrinsic indicates that the memory is mutable again.</p>
7178 <!-- ======================================================================= -->
7179 <div class="doc_subsection">
7180 <a name="int_general">General Intrinsics</a>
7183 <div class="doc_text">
7185 <p>This class of intrinsics is designed to be generic and has no specific
7190 <!-- _______________________________________________________________________ -->
7191 <div class="doc_subsubsection">
7192 <a name="int_var_annotation">'<tt>llvm.var.annotation</tt>' Intrinsic</a>
7195 <div class="doc_text">
7199 declare void @llvm.var.annotation(i8* <val>, i8* <str>, i8* <str>, i32 <int> )
7203 <p>The '<tt>llvm.var.annotation</tt>' intrinsic.</p>
7206 <p>The first argument is a pointer to a value, the second is a pointer to a
7207 global string, the third is a pointer to a global string which is the source
7208 file name, and the last argument is the line number.</p>
7211 <p>This intrinsic allows annotation of local variables with arbitrary strings.
7212 This can be useful for special purpose optimizations that want to look for
7213 these annotations. These have no other defined use, they are ignored by code
7214 generation and optimization.</p>
7218 <!-- _______________________________________________________________________ -->
7219 <div class="doc_subsubsection">
7220 <a name="int_annotation">'<tt>llvm.annotation.*</tt>' Intrinsic</a>
7223 <div class="doc_text">
7226 <p>This is an overloaded intrinsic. You can use '<tt>llvm.annotation</tt>' on
7227 any integer bit width.</p>
7230 declare i8 @llvm.annotation.i8(i8 <val>, i8* <str>, i8* <str>, i32 <int> )
7231 declare i16 @llvm.annotation.i16(i16 <val>, i8* <str>, i8* <str>, i32 <int> )
7232 declare i32 @llvm.annotation.i32(i32 <val>, i8* <str>, i8* <str>, i32 <int> )
7233 declare i64 @llvm.annotation.i64(i64 <val>, i8* <str>, i8* <str>, i32 <int> )
7234 declare i256 @llvm.annotation.i256(i256 <val>, i8* <str>, i8* <str>, i32 <int> )
7238 <p>The '<tt>llvm.annotation</tt>' intrinsic.</p>
7241 <p>The first argument is an integer value (result of some expression), the
7242 second is a pointer to a global string, the third is a pointer to a global
7243 string which is the source file name, and the last argument is the line
7244 number. It returns the value of the first argument.</p>
7247 <p>This intrinsic allows annotations to be put on arbitrary expressions with
7248 arbitrary strings. This can be useful for special purpose optimizations that
7249 want to look for these annotations. These have no other defined use, they
7250 are ignored by code generation and optimization.</p>
7254 <!-- _______________________________________________________________________ -->
7255 <div class="doc_subsubsection">
7256 <a name="int_trap">'<tt>llvm.trap</tt>' Intrinsic</a>
7259 <div class="doc_text">
7263 declare void @llvm.trap()
7267 <p>The '<tt>llvm.trap</tt>' intrinsic.</p>
7273 <p>This intrinsics is lowered to the target dependent trap instruction. If the
7274 target does not have a trap instruction, this intrinsic will be lowered to
7275 the call of the <tt>abort()</tt> function.</p>
7279 <!-- _______________________________________________________________________ -->
7280 <div class="doc_subsubsection">
7281 <a name="int_stackprotector">'<tt>llvm.stackprotector</tt>' Intrinsic</a>
7284 <div class="doc_text">
7288 declare void @llvm.stackprotector( i8* <guard>, i8** <slot> )
7292 <p>The <tt>llvm.stackprotector</tt> intrinsic takes the <tt>guard</tt> and
7293 stores it onto the stack at <tt>slot</tt>. The stack slot is adjusted to
7294 ensure that it is placed on the stack before local variables.</p>
7297 <p>The <tt>llvm.stackprotector</tt> intrinsic requires two pointer
7298 arguments. The first argument is the value loaded from the stack
7299 guard <tt>@__stack_chk_guard</tt>. The second variable is an <tt>alloca</tt>
7300 that has enough space to hold the value of the guard.</p>
7303 <p>This intrinsic causes the prologue/epilogue inserter to force the position of
7304 the <tt>AllocaInst</tt> stack slot to be before local variables on the
7305 stack. This is to ensure that if a local variable on the stack is
7306 overwritten, it will destroy the value of the guard. When the function exits,
7307 the guard on the stack is checked against the original guard. If they're
7308 different, then the program aborts by calling the <tt>__stack_chk_fail()</tt>
7313 <!-- _______________________________________________________________________ -->
7314 <div class="doc_subsubsection">
7315 <a name="int_objectsize">'<tt>llvm.objectsize</tt>' Intrinsic</a>
7318 <div class="doc_text">
7322 declare i32 @llvm.objectsize.i32( i8* <object>, i1 <type> )
7323 declare i64 @llvm.objectsize.i64( i8* <object>, i1 <type> )
7327 <p>The <tt>llvm.objectsize</tt> intrinsic is designed to provide information
7328 to the optimizers to discover at compile time either a) when an
7329 operation like memcpy will either overflow a buffer that corresponds to
7330 an object, or b) to determine that a runtime check for overflow isn't
7331 necessary. An object in this context means an allocation of a
7332 specific class, structure, array, or other object.</p>
7335 <p>The <tt>llvm.objectsize</tt> intrinsic takes two arguments. The first
7336 argument is a pointer to or into the <tt>object</tt>. The second argument
7337 is a boolean 0 or 1. This argument determines whether you want the
7338 maximum (0) or minimum (1) bytes remaining. This needs to be a literal 0 or
7339 1, variables are not allowed.</p>
7342 <p>The <tt>llvm.objectsize</tt> intrinsic is lowered to either a constant
7343 representing the size of the object concerned or <tt>i32/i64 -1 or 0</tt>
7344 (depending on the <tt>type</tt> argument if the size cannot be determined
7345 at compile time.</p>
7349 <!-- *********************************************************************** -->
7352 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
7353 src="http://jigsaw.w3.org/css-validator/images/vcss-blue" alt="Valid CSS"></a>
7354 <a href="http://validator.w3.org/check/referer"><img
7355 src="http://www.w3.org/Icons/valid-html401-blue" alt="Valid HTML 4.01"></a>
7357 <a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
7358 <a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br>
7359 Last modified: $Date$