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_linker_private_weak">'<tt>linker_private_weak</tt>' Linkage</a></li>
28 <li><a href="#linkage_internal">'<tt>internal</tt>' Linkage</a></li>
29 <li><a href="#linkage_available_externally">'<tt>available_externally</tt>' Linkage</a></li>
30 <li><a href="#linkage_linkonce">'<tt>linkonce</tt>' Linkage</a></li>
31 <li><a href="#linkage_common">'<tt>common</tt>' Linkage</a></li>
32 <li><a href="#linkage_weak">'<tt>weak</tt>' Linkage</a></li>
33 <li><a href="#linkage_appending">'<tt>appending</tt>' Linkage</a></li>
34 <li><a href="#linkage_externweak">'<tt>extern_weak</tt>' Linkage</a></li>
35 <li><a href="#linkage_linkonce_odr">'<tt>linkonce_odr</tt>' Linkage</a></li>
36 <li><a href="#linkage_weak">'<tt>weak_odr</tt>' Linkage</a></li>
37 <li><a href="#linkage_external">'<tt>externally visible</tt>' Linkage</a></li>
38 <li><a href="#linkage_dllimport">'<tt>dllimport</tt>' Linkage</a></li>
39 <li><a href="#linkage_dllexport">'<tt>dllexport</tt>' Linkage</a></li>
42 <li><a href="#callingconv">Calling Conventions</a></li>
43 <li><a href="#namedtypes">Named Types</a></li>
44 <li><a href="#globalvars">Global Variables</a></li>
45 <li><a href="#functionstructure">Functions</a></li>
46 <li><a href="#aliasstructure">Aliases</a></li>
47 <li><a href="#namedmetadatastructure">Named Metadata</a></li>
48 <li><a href="#paramattrs">Parameter Attributes</a></li>
49 <li><a href="#fnattrs">Function Attributes</a></li>
50 <li><a href="#gc">Garbage Collector Names</a></li>
51 <li><a href="#moduleasm">Module-Level Inline Assembly</a></li>
52 <li><a href="#datalayout">Data Layout</a></li>
53 <li><a href="#pointeraliasing">Pointer Aliasing Rules</a></li>
54 <li><a href="#volatile">Volatile Memory Accesses</a></li>
57 <li><a href="#typesystem">Type System</a>
59 <li><a href="#t_classifications">Type Classifications</a></li>
60 <li><a href="#t_primitive">Primitive Types</a>
62 <li><a href="#t_integer">Integer Type</a></li>
63 <li><a href="#t_floating">Floating Point Types</a></li>
64 <li><a href="#t_void">Void Type</a></li>
65 <li><a href="#t_label">Label Type</a></li>
66 <li><a href="#t_metadata">Metadata Type</a></li>
69 <li><a href="#t_derived">Derived Types</a>
71 <li><a href="#t_aggregate">Aggregate Types</a>
73 <li><a href="#t_array">Array Type</a></li>
74 <li><a href="#t_struct">Structure Type</a></li>
75 <li><a href="#t_pstruct">Packed Structure Type</a></li>
76 <li><a href="#t_union">Union Type</a></li>
77 <li><a href="#t_vector">Vector Type</a></li>
80 <li><a href="#t_function">Function Type</a></li>
81 <li><a href="#t_pointer">Pointer Type</a></li>
82 <li><a href="#t_opaque">Opaque Type</a></li>
85 <li><a href="#t_uprefs">Type Up-references</a></li>
88 <li><a href="#constants">Constants</a>
90 <li><a href="#simpleconstants">Simple Constants</a></li>
91 <li><a href="#complexconstants">Complex Constants</a></li>
92 <li><a href="#globalconstants">Global Variable and Function Addresses</a></li>
93 <li><a href="#undefvalues">Undefined Values</a></li>
94 <li><a href="#trapvalues">Trap Values</a></li>
95 <li><a href="#blockaddress">Addresses of Basic Blocks</a></li>
96 <li><a href="#constantexprs">Constant Expressions</a></li>
99 <li><a href="#othervalues">Other Values</a>
101 <li><a href="#inlineasm">Inline Assembler Expressions</a></li>
102 <li><a href="#metadata">Metadata Nodes and Metadata Strings</a></li>
105 <li><a href="#intrinsic_globals">Intrinsic Global Variables</a>
107 <li><a href="#intg_used">The '<tt>llvm.used</tt>' Global Variable</a></li>
108 <li><a href="#intg_compiler_used">The '<tt>llvm.compiler.used</tt>'
109 Global Variable</a></li>
110 <li><a href="#intg_global_ctors">The '<tt>llvm.global_ctors</tt>'
111 Global Variable</a></li>
112 <li><a href="#intg_global_dtors">The '<tt>llvm.global_dtors</tt>'
113 Global Variable</a></li>
116 <li><a href="#instref">Instruction Reference</a>
118 <li><a href="#terminators">Terminator Instructions</a>
120 <li><a href="#i_ret">'<tt>ret</tt>' Instruction</a></li>
121 <li><a href="#i_br">'<tt>br</tt>' Instruction</a></li>
122 <li><a href="#i_switch">'<tt>switch</tt>' Instruction</a></li>
123 <li><a href="#i_indirectbr">'<tt>indirectbr</tt>' Instruction</a></li>
124 <li><a href="#i_invoke">'<tt>invoke</tt>' Instruction</a></li>
125 <li><a href="#i_unwind">'<tt>unwind</tt>' Instruction</a></li>
126 <li><a href="#i_unreachable">'<tt>unreachable</tt>' Instruction</a></li>
129 <li><a href="#binaryops">Binary Operations</a>
131 <li><a href="#i_add">'<tt>add</tt>' Instruction</a></li>
132 <li><a href="#i_fadd">'<tt>fadd</tt>' Instruction</a></li>
133 <li><a href="#i_sub">'<tt>sub</tt>' Instruction</a></li>
134 <li><a href="#i_fsub">'<tt>fsub</tt>' Instruction</a></li>
135 <li><a href="#i_mul">'<tt>mul</tt>' Instruction</a></li>
136 <li><a href="#i_fmul">'<tt>fmul</tt>' Instruction</a></li>
137 <li><a href="#i_udiv">'<tt>udiv</tt>' Instruction</a></li>
138 <li><a href="#i_sdiv">'<tt>sdiv</tt>' Instruction</a></li>
139 <li><a href="#i_fdiv">'<tt>fdiv</tt>' Instruction</a></li>
140 <li><a href="#i_urem">'<tt>urem</tt>' Instruction</a></li>
141 <li><a href="#i_srem">'<tt>srem</tt>' Instruction</a></li>
142 <li><a href="#i_frem">'<tt>frem</tt>' Instruction</a></li>
145 <li><a href="#bitwiseops">Bitwise Binary Operations</a>
147 <li><a href="#i_shl">'<tt>shl</tt>' Instruction</a></li>
148 <li><a href="#i_lshr">'<tt>lshr</tt>' Instruction</a></li>
149 <li><a href="#i_ashr">'<tt>ashr</tt>' Instruction</a></li>
150 <li><a href="#i_and">'<tt>and</tt>' Instruction</a></li>
151 <li><a href="#i_or">'<tt>or</tt>' Instruction</a></li>
152 <li><a href="#i_xor">'<tt>xor</tt>' Instruction</a></li>
155 <li><a href="#vectorops">Vector Operations</a>
157 <li><a href="#i_extractelement">'<tt>extractelement</tt>' Instruction</a></li>
158 <li><a href="#i_insertelement">'<tt>insertelement</tt>' Instruction</a></li>
159 <li><a href="#i_shufflevector">'<tt>shufflevector</tt>' Instruction</a></li>
162 <li><a href="#aggregateops">Aggregate Operations</a>
164 <li><a href="#i_extractvalue">'<tt>extractvalue</tt>' Instruction</a></li>
165 <li><a href="#i_insertvalue">'<tt>insertvalue</tt>' Instruction</a></li>
168 <li><a href="#memoryops">Memory Access and Addressing Operations</a>
170 <li><a href="#i_alloca">'<tt>alloca</tt>' Instruction</a></li>
171 <li><a href="#i_load">'<tt>load</tt>' Instruction</a></li>
172 <li><a href="#i_store">'<tt>store</tt>' Instruction</a></li>
173 <li><a href="#i_getelementptr">'<tt>getelementptr</tt>' Instruction</a></li>
176 <li><a href="#convertops">Conversion Operations</a>
178 <li><a href="#i_trunc">'<tt>trunc .. to</tt>' Instruction</a></li>
179 <li><a href="#i_zext">'<tt>zext .. to</tt>' Instruction</a></li>
180 <li><a href="#i_sext">'<tt>sext .. to</tt>' Instruction</a></li>
181 <li><a href="#i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a></li>
182 <li><a href="#i_fpext">'<tt>fpext .. to</tt>' Instruction</a></li>
183 <li><a href="#i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a></li>
184 <li><a href="#i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a></li>
185 <li><a href="#i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a></li>
186 <li><a href="#i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a></li>
187 <li><a href="#i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a></li>
188 <li><a href="#i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a></li>
189 <li><a href="#i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a></li>
192 <li><a href="#otherops">Other Operations</a>
194 <li><a href="#i_icmp">'<tt>icmp</tt>' Instruction</a></li>
195 <li><a href="#i_fcmp">'<tt>fcmp</tt>' Instruction</a></li>
196 <li><a href="#i_phi">'<tt>phi</tt>' Instruction</a></li>
197 <li><a href="#i_select">'<tt>select</tt>' Instruction</a></li>
198 <li><a href="#i_call">'<tt>call</tt>' Instruction</a></li>
199 <li><a href="#i_va_arg">'<tt>va_arg</tt>' Instruction</a></li>
204 <li><a href="#intrinsics">Intrinsic Functions</a>
206 <li><a href="#int_varargs">Variable Argument Handling Intrinsics</a>
208 <li><a href="#int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a></li>
209 <li><a href="#int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a></li>
210 <li><a href="#int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a></li>
213 <li><a href="#int_gc">Accurate Garbage Collection Intrinsics</a>
215 <li><a href="#int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a></li>
216 <li><a href="#int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a></li>
217 <li><a href="#int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a></li>
220 <li><a href="#int_codegen">Code Generator Intrinsics</a>
222 <li><a href="#int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a></li>
223 <li><a href="#int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a></li>
224 <li><a href="#int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a></li>
225 <li><a href="#int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a></li>
226 <li><a href="#int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a></li>
227 <li><a href="#int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a></li>
228 <li><a href="#int_readcyclecounter">'<tt>llvm.readcyclecounter</tt>' Intrinsic</a></li>
231 <li><a href="#int_libc">Standard C Library Intrinsics</a>
233 <li><a href="#int_memcpy">'<tt>llvm.memcpy.*</tt>' Intrinsic</a></li>
234 <li><a href="#int_memmove">'<tt>llvm.memmove.*</tt>' Intrinsic</a></li>
235 <li><a href="#int_memset">'<tt>llvm.memset.*</tt>' Intrinsic</a></li>
236 <li><a href="#int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a></li>
237 <li><a href="#int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a></li>
238 <li><a href="#int_sin">'<tt>llvm.sin.*</tt>' Intrinsic</a></li>
239 <li><a href="#int_cos">'<tt>llvm.cos.*</tt>' Intrinsic</a></li>
240 <li><a href="#int_pow">'<tt>llvm.pow.*</tt>' Intrinsic</a></li>
243 <li><a href="#int_manip">Bit Manipulation Intrinsics</a>
245 <li><a href="#int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a></li>
246 <li><a href="#int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic </a></li>
247 <li><a href="#int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic </a></li>
248 <li><a href="#int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic </a></li>
251 <li><a href="#int_overflow">Arithmetic with Overflow Intrinsics</a>
253 <li><a href="#int_sadd_overflow">'<tt>llvm.sadd.with.overflow.*</tt> Intrinsics</a></li>
254 <li><a href="#int_uadd_overflow">'<tt>llvm.uadd.with.overflow.*</tt> Intrinsics</a></li>
255 <li><a href="#int_ssub_overflow">'<tt>llvm.ssub.with.overflow.*</tt> Intrinsics</a></li>
256 <li><a href="#int_usub_overflow">'<tt>llvm.usub.with.overflow.*</tt> Intrinsics</a></li>
257 <li><a href="#int_smul_overflow">'<tt>llvm.smul.with.overflow.*</tt> Intrinsics</a></li>
258 <li><a href="#int_umul_overflow">'<tt>llvm.umul.with.overflow.*</tt> Intrinsics</a></li>
261 <li><a href="#int_fp16">Half Precision Floating Point Intrinsics</a>
263 <li><a href="#int_convert_to_fp16">'<tt>llvm.convert.to.fp16</tt>' Intrinsic</a></li>
264 <li><a href="#int_convert_from_fp16">'<tt>llvm.convert.from.fp16</tt>' Intrinsic</a></li>
267 <li><a href="#int_debugger">Debugger intrinsics</a></li>
268 <li><a href="#int_eh">Exception Handling intrinsics</a></li>
269 <li><a href="#int_trampoline">Trampoline Intrinsic</a>
271 <li><a href="#int_it">'<tt>llvm.init.trampoline</tt>' Intrinsic</a></li>
274 <li><a href="#int_atomics">Atomic intrinsics</a>
276 <li><a href="#int_memory_barrier"><tt>llvm.memory_barrier</tt></a></li>
277 <li><a href="#int_atomic_cmp_swap"><tt>llvm.atomic.cmp.swap</tt></a></li>
278 <li><a href="#int_atomic_swap"><tt>llvm.atomic.swap</tt></a></li>
279 <li><a href="#int_atomic_load_add"><tt>llvm.atomic.load.add</tt></a></li>
280 <li><a href="#int_atomic_load_sub"><tt>llvm.atomic.load.sub</tt></a></li>
281 <li><a href="#int_atomic_load_and"><tt>llvm.atomic.load.and</tt></a></li>
282 <li><a href="#int_atomic_load_nand"><tt>llvm.atomic.load.nand</tt></a></li>
283 <li><a href="#int_atomic_load_or"><tt>llvm.atomic.load.or</tt></a></li>
284 <li><a href="#int_atomic_load_xor"><tt>llvm.atomic.load.xor</tt></a></li>
285 <li><a href="#int_atomic_load_max"><tt>llvm.atomic.load.max</tt></a></li>
286 <li><a href="#int_atomic_load_min"><tt>llvm.atomic.load.min</tt></a></li>
287 <li><a href="#int_atomic_load_umax"><tt>llvm.atomic.load.umax</tt></a></li>
288 <li><a href="#int_atomic_load_umin"><tt>llvm.atomic.load.umin</tt></a></li>
291 <li><a href="#int_memorymarkers">Memory Use Markers</a>
293 <li><a href="#int_lifetime_start"><tt>llvm.lifetime.start</tt></a></li>
294 <li><a href="#int_lifetime_end"><tt>llvm.lifetime.end</tt></a></li>
295 <li><a href="#int_invariant_start"><tt>llvm.invariant.start</tt></a></li>
296 <li><a href="#int_invariant_end"><tt>llvm.invariant.end</tt></a></li>
299 <li><a href="#int_general">General intrinsics</a>
301 <li><a href="#int_var_annotation">
302 '<tt>llvm.var.annotation</tt>' Intrinsic</a></li>
303 <li><a href="#int_annotation">
304 '<tt>llvm.annotation.*</tt>' Intrinsic</a></li>
305 <li><a href="#int_trap">
306 '<tt>llvm.trap</tt>' Intrinsic</a></li>
307 <li><a href="#int_stackprotector">
308 '<tt>llvm.stackprotector</tt>' Intrinsic</a></li>
309 <li><a href="#int_objectsize">
310 '<tt>llvm.objectsize</tt>' Intrinsic</a></li>
317 <div class="doc_author">
318 <p>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a>
319 and <a href="mailto:vadve@cs.uiuc.edu">Vikram Adve</a></p>
322 <!-- *********************************************************************** -->
323 <div class="doc_section"> <a name="abstract">Abstract </a></div>
324 <!-- *********************************************************************** -->
326 <div class="doc_text">
328 <p>This document is a reference manual for the LLVM assembly language. LLVM is
329 a Static Single Assignment (SSA) based representation that provides type
330 safety, low-level operations, flexibility, and the capability of representing
331 'all' high-level languages cleanly. It is the common code representation
332 used throughout all phases of the LLVM compilation strategy.</p>
336 <!-- *********************************************************************** -->
337 <div class="doc_section"> <a name="introduction">Introduction</a> </div>
338 <!-- *********************************************************************** -->
340 <div class="doc_text">
342 <p>The LLVM code representation is designed to be used in three different forms:
343 as an in-memory compiler IR, as an on-disk bitcode representation (suitable
344 for fast loading by a Just-In-Time compiler), and as a human readable
345 assembly language representation. This allows LLVM to provide a powerful
346 intermediate representation for efficient compiler transformations and
347 analysis, while providing a natural means to debug and visualize the
348 transformations. The three different forms of LLVM are all equivalent. This
349 document describes the human readable representation and notation.</p>
351 <p>The LLVM representation aims to be light-weight and low-level while being
352 expressive, typed, and extensible at the same time. It aims to be a
353 "universal IR" of sorts, by being at a low enough level that high-level ideas
354 may be cleanly mapped to it (similar to how microprocessors are "universal
355 IR's", allowing many source languages to be mapped to them). By providing
356 type information, LLVM can be used as the target of optimizations: for
357 example, through pointer analysis, it can be proven that a C automatic
358 variable is never accessed outside of the current function, allowing it to
359 be promoted to a simple SSA value instead of a memory location.</p>
363 <!-- _______________________________________________________________________ -->
364 <div class="doc_subsubsection"> <a name="wellformed">Well-Formedness</a> </div>
366 <div class="doc_text">
368 <p>It is important to note that this document describes 'well formed' LLVM
369 assembly language. There is a difference between what the parser accepts and
370 what is considered 'well formed'. For example, the following instruction is
371 syntactically okay, but not well formed:</p>
373 <div class="doc_code">
375 %x = <a href="#i_add">add</a> i32 1, %x
379 <p>because the definition of <tt>%x</tt> does not dominate all of its uses. The
380 LLVM infrastructure provides a verification pass that may be used to verify
381 that an LLVM module is well formed. This pass is automatically run by the
382 parser after parsing input assembly and by the optimizer before it outputs
383 bitcode. The violations pointed out by the verifier pass indicate bugs in
384 transformation passes or input to the parser.</p>
388 <!-- Describe the typesetting conventions here. -->
390 <!-- *********************************************************************** -->
391 <div class="doc_section"> <a name="identifiers">Identifiers</a> </div>
392 <!-- *********************************************************************** -->
394 <div class="doc_text">
396 <p>LLVM identifiers come in two basic types: global and local. Global
397 identifiers (functions, global variables) begin with the <tt>'@'</tt>
398 character. Local identifiers (register names, types) begin with
399 the <tt>'%'</tt> character. Additionally, there are three different formats
400 for identifiers, for different purposes:</p>
403 <li>Named values are represented as a string of characters with their prefix.
404 For example, <tt>%foo</tt>, <tt>@DivisionByZero</tt>,
405 <tt>%a.really.long.identifier</tt>. The actual regular expression used is
406 '<tt>[%@][a-zA-Z$._][a-zA-Z$._0-9]*</tt>'. Identifiers which require
407 other characters in their names can be surrounded with quotes. Special
408 characters may be escaped using <tt>"\xx"</tt> where <tt>xx</tt> is the
409 ASCII code for the character in hexadecimal. In this way, any character
410 can be used in a name value, even quotes themselves.</li>
412 <li>Unnamed values are represented as an unsigned numeric value with their
413 prefix. For example, <tt>%12</tt>, <tt>@2</tt>, <tt>%44</tt>.</li>
415 <li>Constants, which are described in a <a href="#constants">section about
416 constants</a>, below.</li>
419 <p>LLVM requires that values start with a prefix for two reasons: Compilers
420 don't need to worry about name clashes with reserved words, and the set of
421 reserved words may be expanded in the future without penalty. Additionally,
422 unnamed identifiers allow a compiler to quickly come up with a temporary
423 variable without having to avoid symbol table conflicts.</p>
425 <p>Reserved words in LLVM are very similar to reserved words in other
426 languages. There are keywords for different opcodes
427 ('<tt><a href="#i_add">add</a></tt>',
428 '<tt><a href="#i_bitcast">bitcast</a></tt>',
429 '<tt><a href="#i_ret">ret</a></tt>', etc...), for primitive type names
430 ('<tt><a href="#t_void">void</a></tt>',
431 '<tt><a href="#t_primitive">i32</a></tt>', etc...), and others. These
432 reserved words cannot conflict with variable names, because none of them
433 start with a prefix character (<tt>'%'</tt> or <tt>'@'</tt>).</p>
435 <p>Here is an example of LLVM code to multiply the integer variable
436 '<tt>%X</tt>' by 8:</p>
440 <div class="doc_code">
442 %result = <a href="#i_mul">mul</a> i32 %X, 8
446 <p>After strength reduction:</p>
448 <div class="doc_code">
450 %result = <a href="#i_shl">shl</a> i32 %X, i8 3
454 <p>And the hard way:</p>
456 <div class="doc_code">
458 %0 = <a href="#i_add">add</a> i32 %X, %X <i>; yields {i32}:%0</i>
459 %1 = <a href="#i_add">add</a> i32 %0, %0 <i>; yields {i32}:%1</i>
460 %result = <a href="#i_add">add</a> i32 %1, %1
464 <p>This last way of multiplying <tt>%X</tt> by 8 illustrates several important
465 lexical features of LLVM:</p>
468 <li>Comments are delimited with a '<tt>;</tt>' and go until the end of
471 <li>Unnamed temporaries are created when the result of a computation is not
472 assigned to a named value.</li>
474 <li>Unnamed temporaries are numbered sequentially</li>
477 <p>It also shows a convention that we follow in this document. When
478 demonstrating instructions, we will follow an instruction with a comment that
479 defines the type and name of value produced. Comments are shown in italic
484 <!-- *********************************************************************** -->
485 <div class="doc_section"> <a name="highlevel">High Level Structure</a> </div>
486 <!-- *********************************************************************** -->
488 <!-- ======================================================================= -->
489 <div class="doc_subsection"> <a name="modulestructure">Module Structure</a>
492 <div class="doc_text">
494 <p>LLVM programs are composed of "Module"s, each of which is a translation unit
495 of the input programs. Each module consists of functions, global variables,
496 and symbol table entries. Modules may be combined together with the LLVM
497 linker, which merges function (and global variable) definitions, resolves
498 forward declarations, and merges symbol table entries. Here is an example of
499 the "hello world" module:</p>
501 <div class="doc_code">
503 <i>; Declare the string constant as a global constant.</i>
504 <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>
506 <i>; External declaration of the puts function</i>
507 <a href="#functionstructure">declare</a> i32 @puts(i8*) <i>; i32 (i8*)* </i>
509 <i>; Definition of main function</i>
510 define i32 @main() { <i>; i32()* </i>
511 <i>; Convert [13 x i8]* to i8 *...</i>
512 %cast210 = <a href="#i_getelementptr">getelementptr</a> [13 x i8]* @.LC0, i64 0, i64 0 <i>; i8*</i>
514 <i>; Call puts function to write out the string to stdout.</i>
515 <a href="#i_call">call</a> i32 @puts(i8* %cast210) <i>; i32</i>
516 <a href="#i_ret">ret</a> i32 0<br>}
518 <i>; Named metadata</i>
519 !1 = metadata !{i32 41}
524 <p>This example is made up of a <a href="#globalvars">global variable</a> named
525 "<tt>.LC0</tt>", an external declaration of the "<tt>puts</tt>" function,
526 a <a href="#functionstructure">function definition</a> for
527 "<tt>main</tt>" and <a href="#namedmetadatastructure">named metadata</a>
530 <p>In general, a module is made up of a list of global values, where both
531 functions and global variables are global values. Global values are
532 represented by a pointer to a memory location (in this case, a pointer to an
533 array of char, and a pointer to a function), and have one of the
534 following <a href="#linkage">linkage types</a>.</p>
538 <!-- ======================================================================= -->
539 <div class="doc_subsection">
540 <a name="linkage">Linkage Types</a>
543 <div class="doc_text">
545 <p>All Global Variables and Functions have one of the following types of
549 <dt><tt><b><a name="linkage_private">private</a></b></tt></dt>
550 <dd>Global values with "<tt>private</tt>" linkage are only directly accessible
551 by objects in the current module. In particular, linking code into a
552 module with an private global value may cause the private to be renamed as
553 necessary to avoid collisions. Because the symbol is private to the
554 module, all references can be updated. This doesn't show up in any symbol
555 table in the object file.</dd>
557 <dt><tt><b><a name="linkage_linker_private">linker_private</a></b></tt></dt>
558 <dd>Similar to <tt>private</tt>, but the symbol is passed through the
559 assembler and evaluated by the linker. Unlike normal strong symbols, they
560 are removed by the linker from the final linked image (executable or
561 dynamic library).</dd>
563 <dt><tt><b><a name="linkage_linker_private_weak">linker_private_weak</a></b></tt></dt>
564 <dd>Similar to "<tt>linker_private</tt>", but the symbol is weak. Note that
565 <tt>linker_private_weak</tt> symbols are subject to coalescing by the
566 linker. The symbols are removed by the linker from the final linked image
567 (executable or dynamic library).</dd>
569 <dt><tt><b><a name="linkage_internal">internal</a></b></tt></dt>
570 <dd>Similar to private, but the value shows as a local symbol
571 (<tt>STB_LOCAL</tt> in the case of ELF) in the object file. This
572 corresponds to the notion of the '<tt>static</tt>' keyword in C.</dd>
574 <dt><tt><b><a name="linkage_available_externally">available_externally</a></b></tt></dt>
575 <dd>Globals with "<tt>available_externally</tt>" linkage are never emitted
576 into the object file corresponding to the LLVM module. They exist to
577 allow inlining and other optimizations to take place given knowledge of
578 the definition of the global, which is known to be somewhere outside the
579 module. Globals with <tt>available_externally</tt> linkage are allowed to
580 be discarded at will, and are otherwise the same as <tt>linkonce_odr</tt>.
581 This linkage type is only allowed on definitions, not declarations.</dd>
583 <dt><tt><b><a name="linkage_linkonce">linkonce</a></b></tt></dt>
584 <dd>Globals with "<tt>linkonce</tt>" linkage are merged with other globals of
585 the same name when linkage occurs. This can be used to implement
586 some forms of inline functions, templates, or other code which must be
587 generated in each translation unit that uses it, but where the body may
588 be overridden with a more definitive definition later. Unreferenced
589 <tt>linkonce</tt> globals are allowed to be discarded. Note that
590 <tt>linkonce</tt> linkage does not actually allow the optimizer to
591 inline the body of this function into callers because it doesn't know if
592 this definition of the function is the definitive definition within the
593 program or whether it will be overridden by a stronger definition.
594 To enable inlining and other optimizations, use "<tt>linkonce_odr</tt>"
597 <dt><tt><b><a name="linkage_weak">weak</a></b></tt></dt>
598 <dd>"<tt>weak</tt>" linkage has the same merging semantics as
599 <tt>linkonce</tt> linkage, except that unreferenced globals with
600 <tt>weak</tt> linkage may not be discarded. This is used for globals that
601 are declared "weak" in C source code.</dd>
603 <dt><tt><b><a name="linkage_common">common</a></b></tt></dt>
604 <dd>"<tt>common</tt>" linkage is most similar to "<tt>weak</tt>" linkage, but
605 they are used for tentative definitions in C, such as "<tt>int X;</tt>" at
607 Symbols with "<tt>common</tt>" linkage are merged in the same way as
608 <tt>weak symbols</tt>, and they may not be deleted if unreferenced.
609 <tt>common</tt> symbols may not have an explicit section,
610 must have a zero initializer, and may not be marked '<a
611 href="#globalvars"><tt>constant</tt></a>'. Functions and aliases may not
612 have common linkage.</dd>
615 <dt><tt><b><a name="linkage_appending">appending</a></b></tt></dt>
616 <dd>"<tt>appending</tt>" linkage may only be applied to global variables of
617 pointer to array type. When two global variables with appending linkage
618 are linked together, the two global arrays are appended together. This is
619 the LLVM, typesafe, equivalent of having the system linker append together
620 "sections" with identical names when .o files are linked.</dd>
622 <dt><tt><b><a name="linkage_externweak">extern_weak</a></b></tt></dt>
623 <dd>The semantics of this linkage follow the ELF object file model: the symbol
624 is weak until linked, if not linked, the symbol becomes null instead of
625 being an undefined reference.</dd>
627 <dt><tt><b><a name="linkage_linkonce_odr">linkonce_odr</a></b></tt></dt>
628 <dt><tt><b><a name="linkage_weak_odr">weak_odr</a></b></tt></dt>
629 <dd>Some languages allow differing globals to be merged, such as two functions
630 with different semantics. Other languages, such as <tt>C++</tt>, ensure
631 that only equivalent globals are ever merged (the "one definition rule"
632 — "ODR"). Such languages can use the <tt>linkonce_odr</tt>
633 and <tt>weak_odr</tt> linkage types to indicate that the global will only
634 be merged with equivalent globals. These linkage types are otherwise the
635 same as their non-<tt>odr</tt> versions.</dd>
637 <dt><tt><b><a name="linkage_external">externally visible</a></b></tt>:</dt>
638 <dd>If none of the above identifiers are used, the global is externally
639 visible, meaning that it participates in linkage and can be used to
640 resolve external symbol references.</dd>
643 <p>The next two types of linkage are targeted for Microsoft Windows platform
644 only. They are designed to support importing (exporting) symbols from (to)
645 DLLs (Dynamic Link Libraries).</p>
648 <dt><tt><b><a name="linkage_dllimport">dllimport</a></b></tt></dt>
649 <dd>"<tt>dllimport</tt>" linkage causes the compiler to reference a function
650 or variable via a global pointer to a pointer that is set up by the DLL
651 exporting the symbol. On Microsoft Windows targets, the pointer name is
652 formed by combining <code>__imp_</code> and the function or variable
655 <dt><tt><b><a name="linkage_dllexport">dllexport</a></b></tt></dt>
656 <dd>"<tt>dllexport</tt>" linkage causes the compiler to provide a global
657 pointer to a pointer in a DLL, so that it can be referenced with the
658 <tt>dllimport</tt> attribute. On Microsoft Windows targets, the pointer
659 name is formed by combining <code>__imp_</code> and the function or
663 <p>For example, since the "<tt>.LC0</tt>" variable is defined to be internal, if
664 another module defined a "<tt>.LC0</tt>" variable and was linked with this
665 one, one of the two would be renamed, preventing a collision. Since
666 "<tt>main</tt>" and "<tt>puts</tt>" are external (i.e., lacking any linkage
667 declarations), they are accessible outside of the current module.</p>
669 <p>It is illegal for a function <i>declaration</i> to have any linkage type
670 other than "externally visible", <tt>dllimport</tt>
671 or <tt>extern_weak</tt>.</p>
673 <p>Aliases can have only <tt>external</tt>, <tt>internal</tt>, <tt>weak</tt>
674 or <tt>weak_odr</tt> linkages.</p>
678 <!-- ======================================================================= -->
679 <div class="doc_subsection">
680 <a name="callingconv">Calling Conventions</a>
683 <div class="doc_text">
685 <p>LLVM <a href="#functionstructure">functions</a>, <a href="#i_call">calls</a>
686 and <a href="#i_invoke">invokes</a> can all have an optional calling
687 convention specified for the call. The calling convention of any pair of
688 dynamic caller/callee must match, or the behavior of the program is
689 undefined. The following calling conventions are supported by LLVM, and more
690 may be added in the future:</p>
693 <dt><b>"<tt>ccc</tt>" - The C calling convention</b>:</dt>
694 <dd>This calling convention (the default if no other calling convention is
695 specified) matches the target C calling conventions. This calling
696 convention supports varargs function calls and tolerates some mismatch in
697 the declared prototype and implemented declaration of the function (as
700 <dt><b>"<tt>fastcc</tt>" - The fast calling convention</b>:</dt>
701 <dd>This calling convention attempts to make calls as fast as possible
702 (e.g. by passing things in registers). This calling convention allows the
703 target to use whatever tricks it wants to produce fast code for the
704 target, without having to conform to an externally specified ABI
705 (Application Binary Interface).
706 <a href="CodeGenerator.html#tailcallopt">Tail calls can only be optimized
707 when this or the GHC convention is used.</a> This calling convention
708 does not support varargs and requires the prototype of all callees to
709 exactly match the prototype of the function definition.</dd>
711 <dt><b>"<tt>coldcc</tt>" - The cold calling convention</b>:</dt>
712 <dd>This calling convention attempts to make code in the caller as efficient
713 as possible under the assumption that the call is not commonly executed.
714 As such, these calls often preserve all registers so that the call does
715 not break any live ranges in the caller side. This calling convention
716 does not support varargs and requires the prototype of all callees to
717 exactly match the prototype of the function definition.</dd>
719 <dt><b>"<tt>cc <em>10</em></tt>" - GHC convention</b>:</dt>
720 <dd>This calling convention has been implemented specifically for use by the
721 <a href="http://www.haskell.org/ghc">Glasgow Haskell Compiler (GHC)</a>.
722 It passes everything in registers, going to extremes to achieve this by
723 disabling callee save registers. This calling convention should not be
724 used lightly but only for specific situations such as an alternative to
725 the <em>register pinning</em> performance technique often used when
726 implementing functional programming languages.At the moment only X86
727 supports this convention and it has the following limitations:
729 <li>On <em>X86-32</em> only supports up to 4 bit type parameters. No
730 floating point types are supported.</li>
731 <li>On <em>X86-64</em> only supports up to 10 bit type parameters and
732 6 floating point parameters.</li>
734 This calling convention supports
735 <a href="CodeGenerator.html#tailcallopt">tail call optimization</a> but
736 requires both the caller and callee are using it.
739 <dt><b>"<tt>cc <<em>n</em>></tt>" - Numbered convention</b>:</dt>
740 <dd>Any calling convention may be specified by number, allowing
741 target-specific calling conventions to be used. Target specific calling
742 conventions start at 64.</dd>
745 <p>More calling conventions can be added/defined on an as-needed basis, to
746 support Pascal conventions or any other well-known target-independent
751 <!-- ======================================================================= -->
752 <div class="doc_subsection">
753 <a name="visibility">Visibility Styles</a>
756 <div class="doc_text">
758 <p>All Global Variables and Functions have one of the following visibility
762 <dt><b>"<tt>default</tt>" - Default style</b>:</dt>
763 <dd>On targets that use the ELF object file format, default visibility means
764 that the declaration is visible to other modules and, in shared libraries,
765 means that the declared entity may be overridden. On Darwin, default
766 visibility means that the declaration is visible to other modules. Default
767 visibility corresponds to "external linkage" in the language.</dd>
769 <dt><b>"<tt>hidden</tt>" - Hidden style</b>:</dt>
770 <dd>Two declarations of an object with hidden visibility refer to the same
771 object if they are in the same shared object. Usually, hidden visibility
772 indicates that the symbol will not be placed into the dynamic symbol
773 table, so no other module (executable or shared library) can reference it
776 <dt><b>"<tt>protected</tt>" - Protected style</b>:</dt>
777 <dd>On ELF, protected visibility indicates that the symbol will be placed in
778 the dynamic symbol table, but that references within the defining module
779 will bind to the local symbol. That is, the symbol cannot be overridden by
785 <!-- ======================================================================= -->
786 <div class="doc_subsection">
787 <a name="namedtypes">Named Types</a>
790 <div class="doc_text">
792 <p>LLVM IR allows you to specify name aliases for certain types. This can make
793 it easier to read the IR and make the IR more condensed (particularly when
794 recursive types are involved). An example of a name specification is:</p>
796 <div class="doc_code">
798 %mytype = type { %mytype*, i32 }
802 <p>You may give a name to any <a href="#typesystem">type</a> except
803 "<a href="t_void">void</a>". Type name aliases may be used anywhere a type
804 is expected with the syntax "%mytype".</p>
806 <p>Note that type names are aliases for the structural type that they indicate,
807 and that you can therefore specify multiple names for the same type. This
808 often leads to confusing behavior when dumping out a .ll file. Since LLVM IR
809 uses structural typing, the name is not part of the type. When printing out
810 LLVM IR, the printer will pick <em>one name</em> to render all types of a
811 particular shape. This means that if you have code where two different
812 source types end up having the same LLVM type, that the dumper will sometimes
813 print the "wrong" or unexpected type. This is an important design point and
814 isn't going to change.</p>
818 <!-- ======================================================================= -->
819 <div class="doc_subsection">
820 <a name="globalvars">Global Variables</a>
823 <div class="doc_text">
825 <p>Global variables define regions of memory allocated at compilation time
826 instead of run-time. Global variables may optionally be initialized, may
827 have an explicit section to be placed in, and may have an optional explicit
828 alignment specified. A variable may be defined as "thread_local", which
829 means that it will not be shared by threads (each thread will have a
830 separated copy of the variable). A variable may be defined as a global
831 "constant," which indicates that the contents of the variable
832 will <b>never</b> be modified (enabling better optimization, allowing the
833 global data to be placed in the read-only section of an executable, etc).
834 Note that variables that need runtime initialization cannot be marked
835 "constant" as there is a store to the variable.</p>
837 <p>LLVM explicitly allows <em>declarations</em> of global variables to be marked
838 constant, even if the final definition of the global is not. This capability
839 can be used to enable slightly better optimization of the program, but
840 requires the language definition to guarantee that optimizations based on the
841 'constantness' are valid for the translation units that do not include the
844 <p>As SSA values, global variables define pointer values that are in scope
845 (i.e. they dominate) all basic blocks in the program. Global variables
846 always define a pointer to their "content" type because they describe a
847 region of memory, and all memory objects in LLVM are accessed through
850 <p>A global variable may be declared to reside in a target-specific numbered
851 address space. For targets that support them, address spaces may affect how
852 optimizations are performed and/or what target instructions are used to
853 access the variable. The default address space is zero. The address space
854 qualifier must precede any other attributes.</p>
856 <p>LLVM allows an explicit section to be specified for globals. If the target
857 supports it, it will emit globals to the section specified.</p>
859 <p>An explicit alignment may be specified for a global, which must be a power
860 of 2. If not present, or if the alignment is set to zero, the alignment of
861 the global is set by the target to whatever it feels convenient. If an
862 explicit alignment is specified, the global is forced to have exactly that
863 alignment. Targets and optimizers are not allowed to over-align the global
864 if the global has an assigned section. In this case, the extra alignment
865 could be observable: for example, code could assume that the globals are
866 densely packed in their section and try to iterate over them as an array,
867 alignment padding would break this iteration.</p>
869 <p>For example, the following defines a global in a numbered address space with
870 an initializer, section, and alignment:</p>
872 <div class="doc_code">
874 @G = addrspace(5) constant float 1.0, section "foo", align 4
881 <!-- ======================================================================= -->
882 <div class="doc_subsection">
883 <a name="functionstructure">Functions</a>
886 <div class="doc_text">
888 <p>LLVM function definitions consist of the "<tt>define</tt>" keyword, an
889 optional <a href="#linkage">linkage type</a>, an optional
890 <a href="#visibility">visibility style</a>, an optional
891 <a href="#callingconv">calling convention</a>, a return type, an optional
892 <a href="#paramattrs">parameter attribute</a> for the return type, a function
893 name, a (possibly empty) argument list (each with optional
894 <a href="#paramattrs">parameter attributes</a>), optional
895 <a href="#fnattrs">function attributes</a>, an optional section, an optional
896 alignment, an optional <a href="#gc">garbage collector name</a>, an opening
897 curly brace, a list of basic blocks, and a closing curly brace.</p>
899 <p>LLVM function declarations consist of the "<tt>declare</tt>" keyword, an
900 optional <a href="#linkage">linkage type</a>, an optional
901 <a href="#visibility">visibility style</a>, an optional
902 <a href="#callingconv">calling convention</a>, a return type, an optional
903 <a href="#paramattrs">parameter attribute</a> for the return type, a function
904 name, a possibly empty list of arguments, an optional alignment, and an
905 optional <a href="#gc">garbage collector name</a>.</p>
907 <p>A function definition contains a list of basic blocks, forming the CFG
908 (Control Flow Graph) for the function. Each basic block may optionally start
909 with a label (giving the basic block a symbol table entry), contains a list
910 of instructions, and ends with a <a href="#terminators">terminator</a>
911 instruction (such as a branch or function return).</p>
913 <p>The first basic block in a function is special in two ways: it is immediately
914 executed on entrance to the function, and it is not allowed to have
915 predecessor basic blocks (i.e. there can not be any branches to the entry
916 block of a function). Because the block can have no predecessors, it also
917 cannot have any <a href="#i_phi">PHI nodes</a>.</p>
919 <p>LLVM allows an explicit section to be specified for functions. If the target
920 supports it, it will emit functions to the section specified.</p>
922 <p>An explicit alignment may be specified for a function. If not present, or if
923 the alignment is set to zero, the alignment of the function is set by the
924 target to whatever it feels convenient. If an explicit alignment is
925 specified, the function is forced to have at least that much alignment. All
926 alignments must be a power of 2.</p>
929 <div class="doc_code">
931 define [<a href="#linkage">linkage</a>] [<a href="#visibility">visibility</a>]
932 [<a href="#callingconv">cconv</a>] [<a href="#paramattrs">ret attrs</a>]
933 <ResultType> @<FunctionName> ([argument list])
934 [<a href="#fnattrs">fn Attrs</a>] [section "name"] [align N]
935 [<a href="#gc">gc</a>] { ... }
941 <!-- ======================================================================= -->
942 <div class="doc_subsection">
943 <a name="aliasstructure">Aliases</a>
946 <div class="doc_text">
948 <p>Aliases act as "second name" for the aliasee value (which can be either
949 function, global variable, another alias or bitcast of global value). Aliases
950 may have an optional <a href="#linkage">linkage type</a>, and an
951 optional <a href="#visibility">visibility style</a>.</p>
954 <div class="doc_code">
956 @<Name> = alias [Linkage] [Visibility] <AliaseeTy> @<Aliasee>
962 <!-- ======================================================================= -->
963 <div class="doc_subsection">
964 <a name="namedmetadatastructure">Named Metadata</a>
967 <div class="doc_text">
969 <p>Named metadata is a collection of metadata. <a href="#metadata">Metadata
970 nodes</a> (but not metadata strings) and null are the only valid operands for
971 a named metadata.</p>
974 <div class="doc_code">
976 !1 = metadata !{metadata !"one"}
983 <!-- ======================================================================= -->
984 <div class="doc_subsection"><a name="paramattrs">Parameter Attributes</a></div>
986 <div class="doc_text">
988 <p>The return type and each parameter of a function type may have a set of
989 <i>parameter attributes</i> associated with them. Parameter attributes are
990 used to communicate additional information about the result or parameters of
991 a function. Parameter attributes are considered to be part of the function,
992 not of the function type, so functions with different parameter attributes
993 can have the same function type.</p>
995 <p>Parameter attributes are simple keywords that follow the type specified. If
996 multiple parameter attributes are needed, they are space separated. For
999 <div class="doc_code">
1001 declare i32 @printf(i8* noalias nocapture, ...)
1002 declare i32 @atoi(i8 zeroext)
1003 declare signext i8 @returns_signed_char()
1007 <p>Note that any attributes for the function result (<tt>nounwind</tt>,
1008 <tt>readonly</tt>) come immediately after the argument list.</p>
1010 <p>Currently, only the following parameter attributes are defined:</p>
1013 <dt><tt><b>zeroext</b></tt></dt>
1014 <dd>This indicates to the code generator that the parameter or return value
1015 should be zero-extended to a 32-bit value by the caller (for a parameter)
1016 or the callee (for a return value).</dd>
1018 <dt><tt><b>signext</b></tt></dt>
1019 <dd>This indicates to the code generator that the parameter or return value
1020 should be sign-extended to a 32-bit value by the caller (for a parameter)
1021 or the callee (for a return value).</dd>
1023 <dt><tt><b>inreg</b></tt></dt>
1024 <dd>This indicates that this parameter or return value should be treated in a
1025 special target-dependent fashion during while emitting code for a function
1026 call or return (usually, by putting it in a register as opposed to memory,
1027 though some targets use it to distinguish between two different kinds of
1028 registers). Use of this attribute is target-specific.</dd>
1030 <dt><tt><b><a name="byval">byval</a></b></tt></dt>
1031 <dd>This indicates that the pointer parameter should really be passed by value
1032 to the function. The attribute implies that a hidden copy of the pointee
1033 is made between the caller and the callee, so the callee is unable to
1034 modify the value in the callee. This attribute is only valid on LLVM
1035 pointer arguments. It is generally used to pass structs and arrays by
1036 value, but is also valid on pointers to scalars. The copy is considered
1037 to belong to the caller not the callee (for example,
1038 <tt><a href="#readonly">readonly</a></tt> functions should not write to
1039 <tt>byval</tt> parameters). This is not a valid attribute for return
1040 values. The byval attribute also supports specifying an alignment with
1041 the align attribute. This has a target-specific effect on the code
1042 generator that usually indicates a desired alignment for the synthesized
1045 <dt><tt><b><a name="sret">sret</a></b></tt></dt>
1046 <dd>This indicates that the pointer parameter specifies the address of a
1047 structure that is the return value of the function in the source program.
1048 This pointer must be guaranteed by the caller to be valid: loads and
1049 stores to the structure may be assumed by the callee to not to trap. This
1050 may only be applied to the first parameter. This is not a valid attribute
1051 for return values. </dd>
1053 <dt><tt><b><a name="noalias">noalias</a></b></tt></dt>
1054 <dd>This indicates that pointer values
1055 <a href="#pointeraliasing"><i>based</i></a> on the argument or return
1056 value do not alias pointer values which are not <i>based</i> on it,
1057 ignoring certain "irrelevant" dependencies.
1058 For a call to the parent function, dependencies between memory
1059 references from before or after the call and from those during the call
1060 are "irrelevant" to the <tt>noalias</tt> keyword for the arguments and
1061 return value used in that call.
1062 The caller shares the responsibility with the callee for ensuring that
1063 these requirements are met.
1064 For further details, please see the discussion of the NoAlias response in
1065 <a href="AliasAnalysis.html#MustMayNo">alias analysis</a>.</dd>
1067 <dt><tt><b><a name="nocapture">nocapture</a></b></tt></dt>
1068 <dd>This indicates that the callee does not make any copies of the pointer
1069 that outlive the callee itself. This is not a valid attribute for return
1072 <dt><tt><b><a name="nest">nest</a></b></tt></dt>
1073 <dd>This indicates that the pointer parameter can be excised using the
1074 <a href="#int_trampoline">trampoline intrinsics</a>. This is not a valid
1075 attribute for return values.</dd>
1080 <!-- ======================================================================= -->
1081 <div class="doc_subsection">
1082 <a name="gc">Garbage Collector Names</a>
1085 <div class="doc_text">
1087 <p>Each function may specify a garbage collector name, which is simply a
1090 <div class="doc_code">
1092 define void @f() gc "name" { ... }
1096 <p>The compiler declares the supported values of <i>name</i>. Specifying a
1097 collector which will cause the compiler to alter its output in order to
1098 support the named garbage collection algorithm.</p>
1102 <!-- ======================================================================= -->
1103 <div class="doc_subsection">
1104 <a name="fnattrs">Function Attributes</a>
1107 <div class="doc_text">
1109 <p>Function attributes are set to communicate additional information about a
1110 function. Function attributes are considered to be part of the function, not
1111 of the function type, so functions with different parameter attributes can
1112 have the same function type.</p>
1114 <p>Function attributes are simple keywords that follow the type specified. If
1115 multiple attributes are needed, they are space separated. For example:</p>
1117 <div class="doc_code">
1119 define void @f() noinline { ... }
1120 define void @f() alwaysinline { ... }
1121 define void @f() alwaysinline optsize { ... }
1122 define void @f() optsize { ... }
1127 <dt><tt><b>alignstack(<<em>n</em>>)</b></tt></dt>
1128 <dd>This attribute indicates that, when emitting the prologue and epilogue,
1129 the backend should forcibly align the stack pointer. Specify the
1130 desired alignment, which must be a power of two, in parentheses.
1132 <dt><tt><b>alwaysinline</b></tt></dt>
1133 <dd>This attribute indicates that the inliner should attempt to inline this
1134 function into callers whenever possible, ignoring any active inlining size
1135 threshold for this caller.</dd>
1137 <dt><tt><b>inlinehint</b></tt></dt>
1138 <dd>This attribute indicates that the source code contained a hint that inlining
1139 this function is desirable (such as the "inline" keyword in C/C++). It
1140 is just a hint; it imposes no requirements on the inliner.</dd>
1142 <dt><tt><b>noinline</b></tt></dt>
1143 <dd>This attribute indicates that the inliner should never inline this
1144 function in any situation. This attribute may not be used together with
1145 the <tt>alwaysinline</tt> attribute.</dd>
1147 <dt><tt><b>optsize</b></tt></dt>
1148 <dd>This attribute suggests that optimization passes and code generator passes
1149 make choices that keep the code size of this function low, and otherwise
1150 do optimizations specifically to reduce code size.</dd>
1152 <dt><tt><b>noreturn</b></tt></dt>
1153 <dd>This function attribute indicates that the function never returns
1154 normally. This produces undefined behavior at runtime if the function
1155 ever does dynamically return.</dd>
1157 <dt><tt><b>nounwind</b></tt></dt>
1158 <dd>This function attribute indicates that the function never returns with an
1159 unwind or exceptional control flow. If the function does unwind, its
1160 runtime behavior is undefined.</dd>
1162 <dt><tt><b>readnone</b></tt></dt>
1163 <dd>This attribute indicates that the function computes its result (or decides
1164 to unwind an exception) based strictly on its arguments, without
1165 dereferencing any pointer arguments or otherwise accessing any mutable
1166 state (e.g. memory, control registers, etc) visible to caller functions.
1167 It does not write through any pointer arguments
1168 (including <tt><a href="#byval">byval</a></tt> arguments) and never
1169 changes any state visible to callers. This means that it cannot unwind
1170 exceptions by calling the <tt>C++</tt> exception throwing methods, but
1171 could use the <tt>unwind</tt> instruction.</dd>
1173 <dt><tt><b><a name="readonly">readonly</a></b></tt></dt>
1174 <dd>This attribute indicates that the function does not write through any
1175 pointer arguments (including <tt><a href="#byval">byval</a></tt>
1176 arguments) or otherwise modify any state (e.g. memory, control registers,
1177 etc) visible to caller functions. It may dereference pointer arguments
1178 and read state that may be set in the caller. A readonly function always
1179 returns the same value (or unwinds an exception identically) when called
1180 with the same set of arguments and global state. It cannot unwind an
1181 exception by calling the <tt>C++</tt> exception throwing methods, but may
1182 use the <tt>unwind</tt> instruction.</dd>
1184 <dt><tt><b><a name="ssp">ssp</a></b></tt></dt>
1185 <dd>This attribute indicates that the function should emit a stack smashing
1186 protector. It is in the form of a "canary"—a random value placed on
1187 the stack before the local variables that's checked upon return from the
1188 function to see if it has been overwritten. A heuristic is used to
1189 determine if a function needs stack protectors or not.<br>
1191 If a function that has an <tt>ssp</tt> attribute is inlined into a
1192 function that doesn't have an <tt>ssp</tt> attribute, then the resulting
1193 function will have an <tt>ssp</tt> attribute.</dd>
1195 <dt><tt><b>sspreq</b></tt></dt>
1196 <dd>This attribute indicates that the function should <em>always</em> emit a
1197 stack smashing protector. This overrides
1198 the <tt><a href="#ssp">ssp</a></tt> function attribute.<br>
1200 If a function that has an <tt>sspreq</tt> attribute is inlined into a
1201 function that doesn't have an <tt>sspreq</tt> attribute or which has
1202 an <tt>ssp</tt> attribute, then the resulting function will have
1203 an <tt>sspreq</tt> attribute.</dd>
1205 <dt><tt><b>noredzone</b></tt></dt>
1206 <dd>This attribute indicates that the code generator should not use a red
1207 zone, even if the target-specific ABI normally permits it.</dd>
1209 <dt><tt><b>noimplicitfloat</b></tt></dt>
1210 <dd>This attributes disables implicit floating point instructions.</dd>
1212 <dt><tt><b>naked</b></tt></dt>
1213 <dd>This attribute disables prologue / epilogue emission for the function.
1214 This can have very system-specific consequences.</dd>
1219 <!-- ======================================================================= -->
1220 <div class="doc_subsection">
1221 <a name="moduleasm">Module-Level Inline Assembly</a>
1224 <div class="doc_text">
1226 <p>Modules may contain "module-level inline asm" blocks, which corresponds to
1227 the GCC "file scope inline asm" blocks. These blocks are internally
1228 concatenated by LLVM and treated as a single unit, but may be separated in
1229 the <tt>.ll</tt> file if desired. The syntax is very simple:</p>
1231 <div class="doc_code">
1233 module asm "inline asm code goes here"
1234 module asm "more can go here"
1238 <p>The strings can contain any character by escaping non-printable characters.
1239 The escape sequence used is simply "\xx" where "xx" is the two digit hex code
1242 <p>The inline asm code is simply printed to the machine code .s file when
1243 assembly code is generated.</p>
1247 <!-- ======================================================================= -->
1248 <div class="doc_subsection">
1249 <a name="datalayout">Data Layout</a>
1252 <div class="doc_text">
1254 <p>A module may specify a target specific data layout string that specifies how
1255 data is to be laid out in memory. The syntax for the data layout is
1258 <div class="doc_code">
1260 target datalayout = "<i>layout specification</i>"
1264 <p>The <i>layout specification</i> consists of a list of specifications
1265 separated by the minus sign character ('-'). Each specification starts with
1266 a letter and may include other information after the letter to define some
1267 aspect of the data layout. The specifications accepted are as follows:</p>
1271 <dd>Specifies that the target lays out data in big-endian form. That is, the
1272 bits with the most significance have the lowest address location.</dd>
1275 <dd>Specifies that the target lays out data in little-endian form. That is,
1276 the bits with the least significance have the lowest address
1279 <dt><tt>p:<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1280 <dd>This specifies the <i>size</i> of a pointer and its <i>abi</i> and
1281 <i>preferred</i> alignments. All sizes are in bits. Specifying
1282 the <i>pref</i> alignment is optional. If omitted, the
1283 preceding <tt>:</tt> should be omitted too.</dd>
1285 <dt><tt>i<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1286 <dd>This specifies the alignment for an integer type of a given bit
1287 <i>size</i>. The value of <i>size</i> must be in the range [1,2^23).</dd>
1289 <dt><tt>v<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1290 <dd>This specifies the alignment for a vector type of a given bit
1293 <dt><tt>f<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1294 <dd>This specifies the alignment for a floating point type of a given bit
1295 <i>size</i>. Only values of <i>size</i> that are supported by the target
1296 will work. 32 (float) and 64 (double) are supported on all targets;
1297 80 or 128 (different flavors of long double) are also supported on some
1300 <dt><tt>a<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1301 <dd>This specifies the alignment for an aggregate type of a given bit
1304 <dt><tt>s<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1305 <dd>This specifies the alignment for a stack object of a given bit
1308 <dt><tt>n<i>size1</i>:<i>size2</i>:<i>size3</i>...</tt></dt>
1309 <dd>This specifies a set of native integer widths for the target CPU
1310 in bits. For example, it might contain "n32" for 32-bit PowerPC,
1311 "n32:64" for PowerPC 64, or "n8:16:32:64" for X86-64. Elements of
1312 this set are considered to support most general arithmetic
1313 operations efficiently.</dd>
1316 <p>When constructing the data layout for a given target, LLVM starts with a
1317 default set of specifications which are then (possibly) overridden by the
1318 specifications in the <tt>datalayout</tt> keyword. The default specifications
1319 are given in this list:</p>
1322 <li><tt>E</tt> - big endian</li>
1323 <li><tt>p:64:64:64</tt> - 64-bit pointers with 64-bit alignment</li>
1324 <li><tt>i1:8:8</tt> - i1 is 8-bit (byte) aligned</li>
1325 <li><tt>i8:8:8</tt> - i8 is 8-bit (byte) aligned</li>
1326 <li><tt>i16:16:16</tt> - i16 is 16-bit aligned</li>
1327 <li><tt>i32:32:32</tt> - i32 is 32-bit aligned</li>
1328 <li><tt>i64:32:64</tt> - i64 has ABI alignment of 32-bits but preferred
1329 alignment of 64-bits</li>
1330 <li><tt>f32:32:32</tt> - float is 32-bit aligned</li>
1331 <li><tt>f64:64:64</tt> - double is 64-bit aligned</li>
1332 <li><tt>v64:64:64</tt> - 64-bit vector is 64-bit aligned</li>
1333 <li><tt>v128:128:128</tt> - 128-bit vector is 128-bit aligned</li>
1334 <li><tt>a0:0:1</tt> - aggregates are 8-bit aligned</li>
1335 <li><tt>s0:64:64</tt> - stack objects are 64-bit aligned</li>
1338 <p>When LLVM is determining the alignment for a given type, it uses the
1339 following rules:</p>
1342 <li>If the type sought is an exact match for one of the specifications, that
1343 specification is used.</li>
1345 <li>If no match is found, and the type sought is an integer type, then the
1346 smallest integer type that is larger than the bitwidth of the sought type
1347 is used. If none of the specifications are larger than the bitwidth then
1348 the the largest integer type is used. For example, given the default
1349 specifications above, the i7 type will use the alignment of i8 (next
1350 largest) while both i65 and i256 will use the alignment of i64 (largest
1353 <li>If no match is found, and the type sought is a vector type, then the
1354 largest vector type that is smaller than the sought vector type will be
1355 used as a fall back. This happens because <128 x double> can be
1356 implemented in terms of 64 <2 x double>, for example.</li>
1361 <!-- ======================================================================= -->
1362 <div class="doc_subsection">
1363 <a name="pointeraliasing">Pointer Aliasing Rules</a>
1366 <div class="doc_text">
1368 <p>Any memory access must be done through a pointer value associated
1369 with an address range of the memory access, otherwise the behavior
1370 is undefined. Pointer values are associated with address ranges
1371 according to the following rules:</p>
1374 <li>A pointer value is associated with the addresses associated with
1375 any value it is <i>based</i> on.
1376 <li>An address of a global variable is associated with the address
1377 range of the variable's storage.</li>
1378 <li>The result value of an allocation instruction is associated with
1379 the address range of the allocated storage.</li>
1380 <li>A null pointer in the default address-space is associated with
1382 <li>An integer constant other than zero or a pointer value returned
1383 from a function not defined within LLVM may be associated with address
1384 ranges allocated through mechanisms other than those provided by
1385 LLVM. Such ranges shall not overlap with any ranges of addresses
1386 allocated by mechanisms provided by LLVM.</li>
1389 <p>A pointer value is <i>based</i> on another pointer value according
1390 to the following rules:</p>
1393 <li>A pointer value formed from a
1394 <tt><a href="#i_getelementptr">getelementptr</a></tt> operation
1395 is <i>based</i> on the first operand of the <tt>getelementptr</tt>.</li>
1396 <li>The result value of a
1397 <tt><a href="#i_bitcast">bitcast</a></tt> is <i>based</i> on the operand
1398 of the <tt>bitcast</tt>.</li>
1399 <li>A pointer value formed by an
1400 <tt><a href="#i_inttoptr">inttoptr</a></tt> is <i>based</i> on all
1401 pointer values that contribute (directly or indirectly) to the
1402 computation of the pointer's value.</li>
1403 <li>The "<i>based</i> on" relationship is transitive.</li>
1406 <p>Note that this definition of <i>"based"</i> is intentionally
1407 similar to the definition of <i>"based"</i> in C99, though it is
1408 slightly weaker.</p>
1410 <p>LLVM IR does not associate types with memory. The result type of a
1411 <tt><a href="#i_load">load</a></tt> merely indicates the size and
1412 alignment of the memory from which to load, as well as the
1413 interpretation of the value. The first operand type of a
1414 <tt><a href="#i_store">store</a></tt> similarly only indicates the size
1415 and alignment of the store.</p>
1417 <p>Consequently, type-based alias analysis, aka TBAA, aka
1418 <tt>-fstrict-aliasing</tt>, is not applicable to general unadorned
1419 LLVM IR. <a href="#metadata">Metadata</a> may be used to encode
1420 additional information which specialized optimization passes may use
1421 to implement type-based alias analysis.</p>
1425 <!-- ======================================================================= -->
1426 <div class="doc_subsection">
1427 <a name="volatile">Volatile Memory Accesses</a>
1430 <div class="doc_text">
1432 <p>Certain memory accesses, such as <a href="#i_load"><tt>load</tt></a>s, <a
1433 href="#i_store"><tt>store</tt></a>s, and <a
1434 href="#int_memcpy"><tt>llvm.memcpy</tt></a>s may be marked <tt>volatile</tt>.
1435 The optimizers must not change the number of volatile operations or change their
1436 order of execution relative to other volatile operations. The optimizers
1437 <i>may</i> change the order of volatile operations relative to non-volatile
1438 operations. This is not Java's "volatile" and has no cross-thread
1439 synchronization behavior.</p>
1443 <!-- *********************************************************************** -->
1444 <div class="doc_section"> <a name="typesystem">Type System</a> </div>
1445 <!-- *********************************************************************** -->
1447 <div class="doc_text">
1449 <p>The LLVM type system is one of the most important features of the
1450 intermediate representation. Being typed enables a number of optimizations
1451 to be performed on the intermediate representation directly, without having
1452 to do extra analyses on the side before the transformation. A strong type
1453 system makes it easier to read the generated code and enables novel analyses
1454 and transformations that are not feasible to perform on normal three address
1455 code representations.</p>
1459 <!-- ======================================================================= -->
1460 <div class="doc_subsection"> <a name="t_classifications">Type
1461 Classifications</a> </div>
1463 <div class="doc_text">
1465 <p>The types fall into a few useful classifications:</p>
1467 <table border="1" cellspacing="0" cellpadding="4">
1469 <tr><th>Classification</th><th>Types</th></tr>
1471 <td><a href="#t_integer">integer</a></td>
1472 <td><tt>i1, i2, i3, ... i8, ... i16, ... i32, ... i64, ... </tt></td>
1475 <td><a href="#t_floating">floating point</a></td>
1476 <td><tt>float, double, x86_fp80, fp128, ppc_fp128</tt></td>
1479 <td><a name="t_firstclass">first class</a></td>
1480 <td><a href="#t_integer">integer</a>,
1481 <a href="#t_floating">floating point</a>,
1482 <a href="#t_pointer">pointer</a>,
1483 <a href="#t_vector">vector</a>,
1484 <a href="#t_struct">structure</a>,
1485 <a href="#t_union">union</a>,
1486 <a href="#t_array">array</a>,
1487 <a href="#t_label">label</a>,
1488 <a href="#t_metadata">metadata</a>.
1492 <td><a href="#t_primitive">primitive</a></td>
1493 <td><a href="#t_label">label</a>,
1494 <a href="#t_void">void</a>,
1495 <a href="#t_floating">floating point</a>,
1496 <a href="#t_metadata">metadata</a>.</td>
1499 <td><a href="#t_derived">derived</a></td>
1500 <td><a href="#t_array">array</a>,
1501 <a href="#t_function">function</a>,
1502 <a href="#t_pointer">pointer</a>,
1503 <a href="#t_struct">structure</a>,
1504 <a href="#t_pstruct">packed structure</a>,
1505 <a href="#t_union">union</a>,
1506 <a href="#t_vector">vector</a>,
1507 <a href="#t_opaque">opaque</a>.
1513 <p>The <a href="#t_firstclass">first class</a> types are perhaps the most
1514 important. Values of these types are the only ones which can be produced by
1519 <!-- ======================================================================= -->
1520 <div class="doc_subsection"> <a name="t_primitive">Primitive Types</a> </div>
1522 <div class="doc_text">
1524 <p>The primitive types are the fundamental building blocks of the LLVM
1529 <!-- _______________________________________________________________________ -->
1530 <div class="doc_subsubsection"> <a name="t_integer">Integer Type</a> </div>
1532 <div class="doc_text">
1535 <p>The integer type is a very simple type that simply specifies an arbitrary
1536 bit width for the integer type desired. Any bit width from 1 bit to
1537 2<sup>23</sup>-1 (about 8 million) can be specified.</p>
1544 <p>The number of bits the integer will occupy is specified by the <tt>N</tt>
1548 <table class="layout">
1550 <td class="left"><tt>i1</tt></td>
1551 <td class="left">a single-bit integer.</td>
1554 <td class="left"><tt>i32</tt></td>
1555 <td class="left">a 32-bit integer.</td>
1558 <td class="left"><tt>i1942652</tt></td>
1559 <td class="left">a really big integer of over 1 million bits.</td>
1565 <!-- _______________________________________________________________________ -->
1566 <div class="doc_subsubsection"> <a name="t_floating">Floating Point Types</a> </div>
1568 <div class="doc_text">
1572 <tr><th>Type</th><th>Description</th></tr>
1573 <tr><td><tt>float</tt></td><td>32-bit floating point value</td></tr>
1574 <tr><td><tt>double</tt></td><td>64-bit floating point value</td></tr>
1575 <tr><td><tt>fp128</tt></td><td>128-bit floating point value (112-bit mantissa)</td></tr>
1576 <tr><td><tt>x86_fp80</tt></td><td>80-bit floating point value (X87)</td></tr>
1577 <tr><td><tt>ppc_fp128</tt></td><td>128-bit floating point value (two 64-bits)</td></tr>
1583 <!-- _______________________________________________________________________ -->
1584 <div class="doc_subsubsection"> <a name="t_void">Void Type</a> </div>
1586 <div class="doc_text">
1589 <p>The void type does not represent any value and has no size.</p>
1598 <!-- _______________________________________________________________________ -->
1599 <div class="doc_subsubsection"> <a name="t_label">Label Type</a> </div>
1601 <div class="doc_text">
1604 <p>The label type represents code labels.</p>
1613 <!-- _______________________________________________________________________ -->
1614 <div class="doc_subsubsection"> <a name="t_metadata">Metadata Type</a> </div>
1616 <div class="doc_text">
1619 <p>The metadata type represents embedded metadata. No derived types may be
1620 created from metadata except for <a href="#t_function">function</a>
1631 <!-- ======================================================================= -->
1632 <div class="doc_subsection"> <a name="t_derived">Derived Types</a> </div>
1634 <div class="doc_text">
1636 <p>The real power in LLVM comes from the derived types in the system. This is
1637 what allows a programmer to represent arrays, functions, pointers, and other
1638 useful types. Each of these types contain one or more element types which
1639 may be a primitive type, or another derived type. For example, it is
1640 possible to have a two dimensional array, using an array as the element type
1641 of another array.</p>
1646 <!-- _______________________________________________________________________ -->
1647 <div class="doc_subsubsection"> <a name="t_aggregate">Aggregate Types</a> </div>
1649 <div class="doc_text">
1651 <p>Aggregate Types are a subset of derived types that can contain multiple
1652 member types. <a href="#t_array">Arrays</a>,
1653 <a href="#t_struct">structs</a>, <a href="#t_vector">vectors</a> and
1654 <a href="#t_union">unions</a> are aggregate types.</p>
1660 <!-- _______________________________________________________________________ -->
1661 <div class="doc_subsubsection"> <a name="t_array">Array Type</a> </div>
1663 <div class="doc_text">
1666 <p>The array type is a very simple derived type that arranges elements
1667 sequentially in memory. The array type requires a size (number of elements)
1668 and an underlying data type.</p>
1672 [<# elements> x <elementtype>]
1675 <p>The number of elements is a constant integer value; <tt>elementtype</tt> may
1676 be any type with a size.</p>
1679 <table class="layout">
1681 <td class="left"><tt>[40 x i32]</tt></td>
1682 <td class="left">Array of 40 32-bit integer values.</td>
1685 <td class="left"><tt>[41 x i32]</tt></td>
1686 <td class="left">Array of 41 32-bit integer values.</td>
1689 <td class="left"><tt>[4 x i8]</tt></td>
1690 <td class="left">Array of 4 8-bit integer values.</td>
1693 <p>Here are some examples of multidimensional arrays:</p>
1694 <table class="layout">
1696 <td class="left"><tt>[3 x [4 x i32]]</tt></td>
1697 <td class="left">3x4 array of 32-bit integer values.</td>
1700 <td class="left"><tt>[12 x [10 x float]]</tt></td>
1701 <td class="left">12x10 array of single precision floating point values.</td>
1704 <td class="left"><tt>[2 x [3 x [4 x i16]]]</tt></td>
1705 <td class="left">2x3x4 array of 16-bit integer values.</td>
1709 <p>There is no restriction on indexing beyond the end of the array implied by
1710 a static type (though there are restrictions on indexing beyond the bounds
1711 of an allocated object in some cases). This means that single-dimension
1712 'variable sized array' addressing can be implemented in LLVM with a zero
1713 length array type. An implementation of 'pascal style arrays' in LLVM could
1714 use the type "<tt>{ i32, [0 x float]}</tt>", for example.</p>
1718 <!-- _______________________________________________________________________ -->
1719 <div class="doc_subsubsection"> <a name="t_function">Function Type</a> </div>
1721 <div class="doc_text">
1724 <p>The function type can be thought of as a function signature. It consists of
1725 a return type and a list of formal parameter types. The return type of a
1726 function type is a scalar type, a void type, a struct type, or a union
1727 type. If the return type is a struct type then all struct elements must be
1728 of first class types, and the struct must have at least one element.</p>
1732 <returntype> (<parameter list>)
1735 <p>...where '<tt><parameter list></tt>' is a comma-separated list of type
1736 specifiers. Optionally, the parameter list may include a type <tt>...</tt>,
1737 which indicates that the function takes a variable number of arguments.
1738 Variable argument functions can access their arguments with
1739 the <a href="#int_varargs">variable argument handling intrinsic</a>
1740 functions. '<tt><returntype></tt>' is any type except
1741 <a href="#t_label">label</a>.</p>
1744 <table class="layout">
1746 <td class="left"><tt>i32 (i32)</tt></td>
1747 <td class="left">function taking an <tt>i32</tt>, returning an <tt>i32</tt>
1749 </tr><tr class="layout">
1750 <td class="left"><tt>float (i16, i32 *) *
1752 <td class="left"><a href="#t_pointer">Pointer</a> to a function that takes
1753 an <tt>i16</tt> and a <a href="#t_pointer">pointer</a> to <tt>i32</tt>,
1754 returning <tt>float</tt>.
1756 </tr><tr class="layout">
1757 <td class="left"><tt>i32 (i8*, ...)</tt></td>
1758 <td class="left">A vararg function that takes at least one
1759 <a href="#t_pointer">pointer</a> to <tt>i8 </tt> (char in C),
1760 which returns an integer. This is the signature for <tt>printf</tt> in
1763 </tr><tr class="layout">
1764 <td class="left"><tt>{i32, i32} (i32)</tt></td>
1765 <td class="left">A function taking an <tt>i32</tt>, returning a
1766 <a href="#t_struct">structure</a> containing two <tt>i32</tt> values
1773 <!-- _______________________________________________________________________ -->
1774 <div class="doc_subsubsection"> <a name="t_struct">Structure Type</a> </div>
1776 <div class="doc_text">
1779 <p>The structure type is used to represent a collection of data members together
1780 in memory. The packing of the field types is defined to match the ABI of the
1781 underlying processor. The elements of a structure may be any type that has a
1784 <p>Structures in memory are accessed using '<tt><a href="#i_load">load</a></tt>'
1785 and '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a field
1786 with the '<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.
1787 Structures in registers are accessed using the
1788 '<tt><a href="#i_extractvalue">extractvalue</a></tt>' and
1789 '<tt><a href="#i_insertvalue">insertvalue</a></tt>' instructions.</p>
1792 { <type list> }
1796 <table class="layout">
1798 <td class="left"><tt>{ i32, i32, i32 }</tt></td>
1799 <td class="left">A triple of three <tt>i32</tt> values</td>
1800 </tr><tr class="layout">
1801 <td class="left"><tt>{ float, i32 (i32) * }</tt></td>
1802 <td class="left">A pair, where the first element is a <tt>float</tt> and the
1803 second element is a <a href="#t_pointer">pointer</a> to a
1804 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
1805 an <tt>i32</tt>.</td>
1811 <!-- _______________________________________________________________________ -->
1812 <div class="doc_subsubsection"> <a name="t_pstruct">Packed Structure Type</a>
1815 <div class="doc_text">
1818 <p>The packed structure type is used to represent a collection of data members
1819 together in memory. There is no padding between fields. Further, the
1820 alignment of a packed structure is 1 byte. The elements of a packed
1821 structure may be any type that has a size.</p>
1823 <p>Structures are accessed using '<tt><a href="#i_load">load</a></tt> and
1824 '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a field with
1825 the '<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.</p>
1829 < { <type list> } >
1833 <table class="layout">
1835 <td class="left"><tt>< { i32, i32, i32 } ></tt></td>
1836 <td class="left">A triple of three <tt>i32</tt> values</td>
1837 </tr><tr class="layout">
1839 <tt>< { float, i32 (i32)* } ></tt></td>
1840 <td class="left">A pair, where the first element is a <tt>float</tt> and the
1841 second element is a <a href="#t_pointer">pointer</a> to a
1842 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
1843 an <tt>i32</tt>.</td>
1849 <!-- _______________________________________________________________________ -->
1850 <div class="doc_subsubsection"> <a name="t_union">Union Type</a> </div>
1852 <div class="doc_text">
1855 <p>A union type describes an object with size and alignment suitable for
1856 an object of any one of a given set of types (also known as an "untagged"
1857 union). It is similar in concept and usage to a
1858 <a href="#t_struct">struct</a>, except that all members of the union
1859 have an offset of zero. The elements of a union may be any type that has a
1860 size. Unions must have at least one member - empty unions are not allowed.
1863 <p>The size of the union as a whole will be the size of its largest member,
1864 and the alignment requirements of the union as a whole will be the largest
1865 alignment requirement of any member.</p>
1867 <p>Union members are accessed using '<tt><a href="#i_load">load</a></tt> and
1868 '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a field with
1869 the '<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.
1870 Since all members are at offset zero, the getelementptr instruction does
1871 not affect the address, only the type of the resulting pointer.</p>
1875 union { <type list> }
1879 <table class="layout">
1881 <td class="left"><tt>union { i32, i32*, float }</tt></td>
1882 <td class="left">A union of three types: an <tt>i32</tt>, a pointer to
1883 an <tt>i32</tt>, and a <tt>float</tt>.</td>
1884 </tr><tr class="layout">
1886 <tt>union { float, i32 (i32) * }</tt></td>
1887 <td class="left">A union, where the first element is a <tt>float</tt> and the
1888 second element is a <a href="#t_pointer">pointer</a> to a
1889 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
1890 an <tt>i32</tt>.</td>
1896 <!-- _______________________________________________________________________ -->
1897 <div class="doc_subsubsection"> <a name="t_pointer">Pointer Type</a> </div>
1899 <div class="doc_text">
1902 <p>The pointer type is used to specify memory locations.
1903 Pointers are commonly used to reference objects in memory.</p>
1905 <p>Pointer types may have an optional address space attribute defining the
1906 numbered address space where the pointed-to object resides. The default
1907 address space is number zero. The semantics of non-zero address
1908 spaces are target-specific.</p>
1910 <p>Note that LLVM does not permit pointers to void (<tt>void*</tt>) nor does it
1911 permit pointers to labels (<tt>label*</tt>). Use <tt>i8*</tt> instead.</p>
1919 <table class="layout">
1921 <td class="left"><tt>[4 x i32]*</tt></td>
1922 <td class="left">A <a href="#t_pointer">pointer</a> to <a
1923 href="#t_array">array</a> of four <tt>i32</tt> values.</td>
1926 <td class="left"><tt>i32 (i32*) *</tt></td>
1927 <td class="left"> A <a href="#t_pointer">pointer</a> to a <a
1928 href="#t_function">function</a> that takes an <tt>i32*</tt>, returning an
1932 <td class="left"><tt>i32 addrspace(5)*</tt></td>
1933 <td class="left">A <a href="#t_pointer">pointer</a> to an <tt>i32</tt> value
1934 that resides in address space #5.</td>
1940 <!-- _______________________________________________________________________ -->
1941 <div class="doc_subsubsection"> <a name="t_vector">Vector Type</a> </div>
1943 <div class="doc_text">
1946 <p>A vector type is a simple derived type that represents a vector of elements.
1947 Vector types are used when multiple primitive data are operated in parallel
1948 using a single instruction (SIMD). A vector type requires a size (number of
1949 elements) and an underlying primitive data type. Vector types are considered
1950 <a href="#t_firstclass">first class</a>.</p>
1954 < <# elements> x <elementtype> >
1957 <p>The number of elements is a constant integer value; elementtype may be any
1958 integer or floating point type.</p>
1961 <table class="layout">
1963 <td class="left"><tt><4 x i32></tt></td>
1964 <td class="left">Vector of 4 32-bit integer values.</td>
1967 <td class="left"><tt><8 x float></tt></td>
1968 <td class="left">Vector of 8 32-bit floating-point values.</td>
1971 <td class="left"><tt><2 x i64></tt></td>
1972 <td class="left">Vector of 2 64-bit integer values.</td>
1978 <!-- _______________________________________________________________________ -->
1979 <div class="doc_subsubsection"> <a name="t_opaque">Opaque Type</a> </div>
1980 <div class="doc_text">
1983 <p>Opaque types are used to represent unknown types in the system. This
1984 corresponds (for example) to the C notion of a forward declared structure
1985 type. In LLVM, opaque types can eventually be resolved to any type (not just
1986 a structure type).</p>
1994 <table class="layout">
1996 <td class="left"><tt>opaque</tt></td>
1997 <td class="left">An opaque type.</td>
2003 <!-- ======================================================================= -->
2004 <div class="doc_subsection">
2005 <a name="t_uprefs">Type Up-references</a>
2008 <div class="doc_text">
2011 <p>An "up reference" allows you to refer to a lexically enclosing type without
2012 requiring it to have a name. For instance, a structure declaration may
2013 contain a pointer to any of the types it is lexically a member of. Example
2014 of up references (with their equivalent as named type declarations)
2018 { \2 * } %x = type { %x* }
2019 { \2 }* %y = type { %y }*
2023 <p>An up reference is needed by the asmprinter for printing out cyclic types
2024 when there is no declared name for a type in the cycle. Because the
2025 asmprinter does not want to print out an infinite type string, it needs a
2026 syntax to handle recursive types that have no names (all names are optional
2034 <p>The level is the count of the lexical type that is being referred to.</p>
2037 <table class="layout">
2039 <td class="left"><tt>\1*</tt></td>
2040 <td class="left">Self-referential pointer.</td>
2043 <td class="left"><tt>{ { \3*, i8 }, i32 }</tt></td>
2044 <td class="left">Recursive structure where the upref refers to the out-most
2051 <!-- *********************************************************************** -->
2052 <div class="doc_section"> <a name="constants">Constants</a> </div>
2053 <!-- *********************************************************************** -->
2055 <div class="doc_text">
2057 <p>LLVM has several different basic types of constants. This section describes
2058 them all and their syntax.</p>
2062 <!-- ======================================================================= -->
2063 <div class="doc_subsection"><a name="simpleconstants">Simple Constants</a></div>
2065 <div class="doc_text">
2068 <dt><b>Boolean constants</b></dt>
2069 <dd>The two strings '<tt>true</tt>' and '<tt>false</tt>' are both valid
2070 constants of the <tt><a href="#t_integer">i1</a></tt> type.</dd>
2072 <dt><b>Integer constants</b></dt>
2073 <dd>Standard integers (such as '4') are constants of
2074 the <a href="#t_integer">integer</a> type. Negative numbers may be used
2075 with integer types.</dd>
2077 <dt><b>Floating point constants</b></dt>
2078 <dd>Floating point constants use standard decimal notation (e.g. 123.421),
2079 exponential notation (e.g. 1.23421e+2), or a more precise hexadecimal
2080 notation (see below). The assembler requires the exact decimal value of a
2081 floating-point constant. For example, the assembler accepts 1.25 but
2082 rejects 1.3 because 1.3 is a repeating decimal in binary. Floating point
2083 constants must have a <a href="#t_floating">floating point</a> type. </dd>
2085 <dt><b>Null pointer constants</b></dt>
2086 <dd>The identifier '<tt>null</tt>' is recognized as a null pointer constant
2087 and must be of <a href="#t_pointer">pointer type</a>.</dd>
2090 <p>The one non-intuitive notation for constants is the hexadecimal form of
2091 floating point constants. For example, the form '<tt>double
2092 0x432ff973cafa8000</tt>' is equivalent to (but harder to read than)
2093 '<tt>double 4.5e+15</tt>'. The only time hexadecimal floating point
2094 constants are required (and the only time that they are generated by the
2095 disassembler) is when a floating point constant must be emitted but it cannot
2096 be represented as a decimal floating point number in a reasonable number of
2097 digits. For example, NaN's, infinities, and other special values are
2098 represented in their IEEE hexadecimal format so that assembly and disassembly
2099 do not cause any bits to change in the constants.</p>
2101 <p>When using the hexadecimal form, constants of types float and double are
2102 represented using the 16-digit form shown above (which matches the IEEE754
2103 representation for double); float values must, however, be exactly
2104 representable as IEE754 single precision. Hexadecimal format is always used
2105 for long double, and there are three forms of long double. The 80-bit format
2106 used by x86 is represented as <tt>0xK</tt> followed by 20 hexadecimal digits.
2107 The 128-bit format used by PowerPC (two adjacent doubles) is represented
2108 by <tt>0xM</tt> followed by 32 hexadecimal digits. The IEEE 128-bit format
2109 is represented by <tt>0xL</tt> followed by 32 hexadecimal digits; no
2110 currently supported target uses this format. Long doubles will only work if
2111 they match the long double format on your target. All hexadecimal formats
2112 are big-endian (sign bit at the left).</p>
2116 <!-- ======================================================================= -->
2117 <div class="doc_subsection">
2118 <a name="aggregateconstants"></a> <!-- old anchor -->
2119 <a name="complexconstants">Complex Constants</a>
2122 <div class="doc_text">
2124 <p>Complex constants are a (potentially recursive) combination of simple
2125 constants and smaller complex constants.</p>
2128 <dt><b>Structure constants</b></dt>
2129 <dd>Structure constants are represented with notation similar to structure
2130 type definitions (a comma separated list of elements, surrounded by braces
2131 (<tt>{}</tt>)). For example: "<tt>{ i32 4, float 17.0, i32* @G }</tt>",
2132 where "<tt>@G</tt>" is declared as "<tt>@G = external global i32</tt>".
2133 Structure constants must have <a href="#t_struct">structure type</a>, and
2134 the number and types of elements must match those specified by the
2137 <dt><b>Union constants</b></dt>
2138 <dd>Union constants are represented with notation similar to a structure with
2139 a single element - that is, a single typed element surrounded
2140 by braces (<tt>{}</tt>)). For example: "<tt>{ i32 4 }</tt>". The
2141 <a href="#t_union">union type</a> can be initialized with a single-element
2142 struct as long as the type of the struct element matches the type of
2143 one of the union members.</dd>
2145 <dt><b>Array constants</b></dt>
2146 <dd>Array constants are represented with notation similar to array type
2147 definitions (a comma separated list of elements, surrounded by square
2148 brackets (<tt>[]</tt>)). For example: "<tt>[ i32 42, i32 11, i32 74
2149 ]</tt>". Array constants must have <a href="#t_array">array type</a>, and
2150 the number and types of elements must match those specified by the
2153 <dt><b>Vector constants</b></dt>
2154 <dd>Vector constants are represented with notation similar to vector type
2155 definitions (a comma separated list of elements, surrounded by
2156 less-than/greater-than's (<tt><></tt>)). For example: "<tt>< i32
2157 42, i32 11, i32 74, i32 100 ></tt>". Vector constants must
2158 have <a href="#t_vector">vector type</a>, and the number and types of
2159 elements must match those specified by the type.</dd>
2161 <dt><b>Zero initialization</b></dt>
2162 <dd>The string '<tt>zeroinitializer</tt>' can be used to zero initialize a
2163 value to zero of <em>any</em> type, including scalar and
2164 <a href="#t_aggregate">aggregate</a> types.
2165 This is often used to avoid having to print large zero initializers
2166 (e.g. for large arrays) and is always exactly equivalent to using explicit
2167 zero initializers.</dd>
2169 <dt><b>Metadata node</b></dt>
2170 <dd>A metadata node is a structure-like constant with
2171 <a href="#t_metadata">metadata type</a>. For example: "<tt>metadata !{
2172 i32 0, metadata !"test" }</tt>". Unlike other constants that are meant to
2173 be interpreted as part of the instruction stream, metadata is a place to
2174 attach additional information such as debug info.</dd>
2179 <!-- ======================================================================= -->
2180 <div class="doc_subsection">
2181 <a name="globalconstants">Global Variable and Function Addresses</a>
2184 <div class="doc_text">
2186 <p>The addresses of <a href="#globalvars">global variables</a>
2187 and <a href="#functionstructure">functions</a> are always implicitly valid
2188 (link-time) constants. These constants are explicitly referenced when
2189 the <a href="#identifiers">identifier for the global</a> is used and always
2190 have <a href="#t_pointer">pointer</a> type. For example, the following is a
2191 legal LLVM file:</p>
2193 <div class="doc_code">
2197 @Z = global [2 x i32*] [ i32* @X, i32* @Y ]
2203 <!-- ======================================================================= -->
2204 <div class="doc_subsection"><a name="undefvalues">Undefined Values</a></div>
2205 <div class="doc_text">
2207 <p>The string '<tt>undef</tt>' can be used anywhere a constant is expected, and
2208 indicates that the user of the value may receive an unspecified bit-pattern.
2209 Undefined values may be of any type (other than label or void) and be used
2210 anywhere a constant is permitted.</p>
2212 <p>Undefined values are useful because they indicate to the compiler that the
2213 program is well defined no matter what value is used. This gives the
2214 compiler more freedom to optimize. Here are some examples of (potentially
2215 surprising) transformations that are valid (in pseudo IR):</p>
2218 <div class="doc_code">
2230 <p>This is safe because all of the output bits are affected by the undef bits.
2231 Any output bit can have a zero or one depending on the input bits.</p>
2233 <div class="doc_code">
2246 <p>These logical operations have bits that are not always affected by the input.
2247 For example, if "%X" has a zero bit, then the output of the 'and' operation will
2248 always be a zero, no matter what the corresponding bit from the undef is. As
2249 such, it is unsafe to optimize or assume that the result of the and is undef.
2250 However, it is safe to assume that all bits of the undef could be 0, and
2251 optimize the and to 0. Likewise, it is safe to assume that all the bits of
2252 the undef operand to the or could be set, allowing the or to be folded to
2255 <div class="doc_code">
2257 %A = select undef, %X, %Y
2258 %B = select undef, 42, %Y
2259 %C = select %X, %Y, undef
2271 <p>This set of examples show that undefined select (and conditional branch)
2272 conditions can go "either way" but they have to come from one of the two
2273 operands. In the %A example, if %X and %Y were both known to have a clear low
2274 bit, then %A would have to have a cleared low bit. However, in the %C example,
2275 the optimizer is allowed to assume that the undef operand could be the same as
2276 %Y, allowing the whole select to be eliminated.</p>
2279 <div class="doc_code">
2281 %A = xor undef, undef
2300 <p>This example points out that two undef operands are not necessarily the same.
2301 This can be surprising to people (and also matches C semantics) where they
2302 assume that "X^X" is always zero, even if X is undef. This isn't true for a
2303 number of reasons, but the short answer is that an undef "variable" can
2304 arbitrarily change its value over its "live range". This is true because the
2305 "variable" doesn't actually <em>have a live range</em>. Instead, the value is
2306 logically read from arbitrary registers that happen to be around when needed,
2307 so the value is not necessarily consistent over time. In fact, %A and %C need
2308 to have the same semantics or the core LLVM "replace all uses with" concept
2311 <div class="doc_code">
2321 <p>These examples show the crucial difference between an <em>undefined
2322 value</em> and <em>undefined behavior</em>. An undefined value (like undef) is
2323 allowed to have an arbitrary bit-pattern. This means that the %A operation
2324 can be constant folded to undef because the undef could be an SNaN, and fdiv is
2325 not (currently) defined on SNaN's. However, in the second example, we can make
2326 a more aggressive assumption: because the undef is allowed to be an arbitrary
2327 value, we are allowed to assume that it could be zero. Since a divide by zero
2328 has <em>undefined behavior</em>, we are allowed to assume that the operation
2329 does not execute at all. This allows us to delete the divide and all code after
2330 it: since the undefined operation "can't happen", the optimizer can assume that
2331 it occurs in dead code.
2334 <div class="doc_code">
2336 a: store undef -> %X
2337 b: store %X -> undef
2344 <p>These examples reiterate the fdiv example: a store "of" an undefined value
2345 can be assumed to not have any effect: we can assume that the value is
2346 overwritten with bits that happen to match what was already there. However, a
2347 store "to" an undefined location could clobber arbitrary memory, therefore, it
2348 has undefined behavior.</p>
2352 <!-- ======================================================================= -->
2353 <div class="doc_subsection"><a name="trapvalues">Trap Values</a></div>
2354 <div class="doc_text">
2356 <p>Trap values are similar to <a href="#undefvalues">undef values</a>, however
2357 instead of representing an unspecified bit pattern, they represent the
2358 fact that an instruction or constant expression which cannot evoke side
2359 effects has nevertheless detected a condition which results in undefined
2362 <p>There is currently no way of representing a trap value in the IR; they
2363 only exist when produced by operations such as
2364 <a href="#i_add"><tt>add</tt></a> with the <tt>nsw</tt> flag.</p>
2366 <p>Trap value behavior is defined in terms of value <i>dependence</i>:</p>
2370 <li>Values other than <a href="#i_phi"><tt>phi</tt></a> nodes depend on
2371 their operands.</li>
2373 <li><a href="#i_phi"><tt>Phi</tt></a> nodes depend on the operand corresponding
2374 to their dynamic predecessor basic block.</li>
2376 <li>Function arguments depend on the corresponding actual argument values in
2377 the dynamic callers of their functions.</li>
2379 <li><a href="#i_call"><tt>Call</tt></a> instructions depend on the
2380 <a href="#i_ret"><tt>ret</tt></a> instructions that dynamically transfer
2381 control back to them.</li>
2383 <li><a href="#i_invoke"><tt>Invoke</tt></a> instructions depend on the
2384 <a href="#i_ret"><tt>ret</tt></a>, <a href="#i_unwind"><tt>unwind</tt></a>,
2385 or exception-throwing call instructions that dynamically transfer control
2388 <li>Non-volatile loads and stores depend on the most recent stores to all of the
2389 referenced memory addresses, following the order in the IR
2390 (including loads and stores implied by intrinsics such as
2391 <a href="#int_memcpy"><tt>@llvm.memcpy</tt></a>.)</li>
2393 <!-- TODO: In the case of multiple threads, this only applies if the store
2394 "happens-before" the load or store. -->
2396 <!-- TODO: floating-point exception state -->
2398 <li>An instruction with externally visible side effects depends on the most
2399 recent preceding instruction with externally visible side effects, following
2400 the order in the IR. (This includes volatile loads and stores.)</li>
2402 <li>An instruction <i>control-depends</i> on a
2403 <a href="#terminators">terminator instruction</a>
2404 if the terminator instruction has multiple successors and the instruction
2405 is always executed when control transfers to one of the successors, and
2406 may not be executed when control is transfered to another.</li>
2408 <li>Dependence is transitive.</li>
2413 <p>Whenever a trap value is generated, all values which depend on it evaluate
2414 to trap. If they have side effects, the evoke their side effects as if each
2415 operand with a trap value were undef. If they have externally-visible side
2416 effects, the behavior is undefined.</p>
2418 <p>Here are some examples:</p>
2420 <div class="doc_code">
2423 %trap = sub nuw i32 0, 1 ; Results in a trap value.
2424 %still_trap = and i32 %trap, 0 ; Whereas (and i32 undef, 0) would return 0.
2425 %trap_yet_again = getelementptr i32* @h, i32 %still_trap
2426 store i32 0, i32* %trap_yet_again ; undefined behavior
2428 store i32 %trap, i32* @g ; Trap value conceptually stored to memory.
2429 %trap2 = load i32* @g ; Returns a trap value, not just undef.
2431 volatile store i32 %trap, i32* @g ; External observation; undefined behavior.
2433 %narrowaddr = bitcast i32* @g to i16*
2434 %wideaddr = bitcast i32* @g to i64*
2435 %trap3 = load 16* %narrowaddr ; Returns a trap value.
2436 %trap4 = load i64* %widaddr ; Returns a trap value.
2438 %cmp = icmp i32 slt %trap, 0 ; Returns a trap value.
2439 %br i1 %cmp, %true, %end ; Branch to either destination.
2442 volatile store i32 0, i32* @g ; This is control-dependent on %cmp, so
2443 ; it has undefined behavior.
2447 %p = phi i32 [ 0, %entry ], [ 1, %true ]
2448 ; Both edges into this PHI are
2449 ; control-dependent on %cmp, so this
2450 ; always results in a trap value.
2452 volatile store i32 0, i32* @g ; %end is control-equivalent to %entry
2453 ; so this is defined (ignoring earlier
2454 ; undefined behavior in this example).
2460 <!-- ======================================================================= -->
2461 <div class="doc_subsection"><a name="blockaddress">Addresses of Basic
2463 <div class="doc_text">
2465 <p><b><tt>blockaddress(@function, %block)</tt></b></p>
2467 <p>The '<tt>blockaddress</tt>' constant computes the address of the specified
2468 basic block in the specified function, and always has an i8* type. Taking
2469 the address of the entry block is illegal.</p>
2471 <p>This value only has defined behavior when used as an operand to the
2472 '<a href="#i_indirectbr"><tt>indirectbr</tt></a>' instruction or for comparisons
2473 against null. Pointer equality tests between labels addresses is undefined
2474 behavior - though, again, comparison against null is ok, and no label is
2475 equal to the null pointer. This may also be passed around as an opaque
2476 pointer sized value as long as the bits are not inspected. This allows
2477 <tt>ptrtoint</tt> and arithmetic to be performed on these values so long as
2478 the original value is reconstituted before the <tt>indirectbr</tt>.</p>
2480 <p>Finally, some targets may provide defined semantics when
2481 using the value as the operand to an inline assembly, but that is target
2488 <!-- ======================================================================= -->
2489 <div class="doc_subsection"><a name="constantexprs">Constant Expressions</a>
2492 <div class="doc_text">
2494 <p>Constant expressions are used to allow expressions involving other constants
2495 to be used as constants. Constant expressions may be of
2496 any <a href="#t_firstclass">first class</a> type and may involve any LLVM
2497 operation that does not have side effects (e.g. load and call are not
2498 supported). The following is the syntax for constant expressions:</p>
2501 <dt><b><tt>trunc (CST to TYPE)</tt></b></dt>
2502 <dd>Truncate a constant to another type. The bit size of CST must be larger
2503 than the bit size of TYPE. Both types must be integers.</dd>
2505 <dt><b><tt>zext (CST to TYPE)</tt></b></dt>
2506 <dd>Zero extend a constant to another type. The bit size of CST must be
2507 smaller or equal to the bit size of TYPE. Both types must be
2510 <dt><b><tt>sext (CST to TYPE)</tt></b></dt>
2511 <dd>Sign extend a constant to another type. The bit size of CST must be
2512 smaller or equal to the bit size of TYPE. Both types must be
2515 <dt><b><tt>fptrunc (CST to TYPE)</tt></b></dt>
2516 <dd>Truncate a floating point constant to another floating point type. The
2517 size of CST must be larger than the size of TYPE. Both types must be
2518 floating point.</dd>
2520 <dt><b><tt>fpext (CST to TYPE)</tt></b></dt>
2521 <dd>Floating point extend a constant to another type. The size of CST must be
2522 smaller or equal to the size of TYPE. Both types must be floating
2525 <dt><b><tt>fptoui (CST to TYPE)</tt></b></dt>
2526 <dd>Convert a floating point constant to the corresponding unsigned integer
2527 constant. TYPE must be a scalar or vector integer type. CST must be of
2528 scalar or vector floating point type. Both CST and TYPE must be scalars,
2529 or vectors of the same number of elements. If the value won't fit in the
2530 integer type, the results are undefined.</dd>
2532 <dt><b><tt>fptosi (CST to TYPE)</tt></b></dt>
2533 <dd>Convert a floating point constant to the corresponding signed integer
2534 constant. TYPE must be a scalar or vector integer type. CST must be of
2535 scalar or vector floating point type. Both CST and TYPE must be scalars,
2536 or vectors of the same number of elements. If the value won't fit in the
2537 integer type, the results are undefined.</dd>
2539 <dt><b><tt>uitofp (CST to TYPE)</tt></b></dt>
2540 <dd>Convert an unsigned integer constant to the corresponding floating point
2541 constant. TYPE must be a scalar or vector floating point type. CST must be
2542 of scalar or vector integer type. Both CST and TYPE must be scalars, or
2543 vectors of the same number of elements. If the value won't fit in the
2544 floating point type, the results are undefined.</dd>
2546 <dt><b><tt>sitofp (CST to TYPE)</tt></b></dt>
2547 <dd>Convert a signed integer constant to the corresponding floating point
2548 constant. TYPE must be a scalar or vector floating point type. CST must be
2549 of scalar or vector integer type. Both CST and TYPE must be scalars, or
2550 vectors of the same number of elements. If the value won't fit in the
2551 floating point type, the results are undefined.</dd>
2553 <dt><b><tt>ptrtoint (CST to TYPE)</tt></b></dt>
2554 <dd>Convert a pointer typed constant to the corresponding integer constant
2555 <tt>TYPE</tt> must be an integer type. <tt>CST</tt> must be of pointer
2556 type. The <tt>CST</tt> value is zero extended, truncated, or unchanged to
2557 make it fit in <tt>TYPE</tt>.</dd>
2559 <dt><b><tt>inttoptr (CST to TYPE)</tt></b></dt>
2560 <dd>Convert a integer constant to a pointer constant. TYPE must be a pointer
2561 type. CST must be of integer type. The CST value is zero extended,
2562 truncated, or unchanged to make it fit in a pointer size. This one is
2563 <i>really</i> dangerous!</dd>
2565 <dt><b><tt>bitcast (CST to TYPE)</tt></b></dt>
2566 <dd>Convert a constant, CST, to another TYPE. The constraints of the operands
2567 are the same as those for the <a href="#i_bitcast">bitcast
2568 instruction</a>.</dd>
2570 <dt><b><tt>getelementptr (CSTPTR, IDX0, IDX1, ...)</tt></b></dt>
2571 <dt><b><tt>getelementptr inbounds (CSTPTR, IDX0, IDX1, ...)</tt></b></dt>
2572 <dd>Perform the <a href="#i_getelementptr">getelementptr operation</a> on
2573 constants. As with the <a href="#i_getelementptr">getelementptr</a>
2574 instruction, the index list may have zero or more indexes, which are
2575 required to make sense for the type of "CSTPTR".</dd>
2577 <dt><b><tt>select (COND, VAL1, VAL2)</tt></b></dt>
2578 <dd>Perform the <a href="#i_select">select operation</a> on constants.</dd>
2580 <dt><b><tt>icmp COND (VAL1, VAL2)</tt></b></dt>
2581 <dd>Performs the <a href="#i_icmp">icmp operation</a> on constants.</dd>
2583 <dt><b><tt>fcmp COND (VAL1, VAL2)</tt></b></dt>
2584 <dd>Performs the <a href="#i_fcmp">fcmp operation</a> on constants.</dd>
2586 <dt><b><tt>extractelement (VAL, IDX)</tt></b></dt>
2587 <dd>Perform the <a href="#i_extractelement">extractelement operation</a> on
2590 <dt><b><tt>insertelement (VAL, ELT, IDX)</tt></b></dt>
2591 <dd>Perform the <a href="#i_insertelement">insertelement operation</a> on
2594 <dt><b><tt>shufflevector (VEC1, VEC2, IDXMASK)</tt></b></dt>
2595 <dd>Perform the <a href="#i_shufflevector">shufflevector operation</a> on
2598 <dt><b><tt>extractvalue (VAL, IDX0, IDX1, ...)</tt></b></dt>
2599 <dd>Perform the <a href="#i_extractvalue">extractvalue operation</a> on
2600 constants. The index list is interpreted in a similar manner as indices in
2601 a '<a href="#i_getelementptr">getelementptr</a>' operation. At least one
2602 index value must be specified.</dd>
2604 <dt><b><tt>insertvalue (VAL, ELT, IDX0, IDX1, ...)</tt></b></dt>
2605 <dd>Perform the <a href="#i_insertvalue">insertvalue operation</a> on
2606 constants. The index list is interpreted in a similar manner as indices in
2607 a '<a href="#i_getelementptr">getelementptr</a>' operation. At least one
2608 index value must be specified.</dd>
2610 <dt><b><tt>OPCODE (LHS, RHS)</tt></b></dt>
2611 <dd>Perform the specified operation of the LHS and RHS constants. OPCODE may
2612 be any of the <a href="#binaryops">binary</a>
2613 or <a href="#bitwiseops">bitwise binary</a> operations. The constraints
2614 on operands are the same as those for the corresponding instruction
2615 (e.g. no bitwise operations on floating point values are allowed).</dd>
2620 <!-- *********************************************************************** -->
2621 <div class="doc_section"> <a name="othervalues">Other Values</a> </div>
2622 <!-- *********************************************************************** -->
2624 <!-- ======================================================================= -->
2625 <div class="doc_subsection">
2626 <a name="inlineasm">Inline Assembler Expressions</a>
2629 <div class="doc_text">
2631 <p>LLVM supports inline assembler expressions (as opposed
2632 to <a href="#moduleasm"> Module-Level Inline Assembly</a>) through the use of
2633 a special value. This value represents the inline assembler as a string
2634 (containing the instructions to emit), a list of operand constraints (stored
2635 as a string), a flag that indicates whether or not the inline asm
2636 expression has side effects, and a flag indicating whether the function
2637 containing the asm needs to align its stack conservatively. An example
2638 inline assembler expression is:</p>
2640 <div class="doc_code">
2642 i32 (i32) asm "bswap $0", "=r,r"
2646 <p>Inline assembler expressions may <b>only</b> be used as the callee operand of
2647 a <a href="#i_call"><tt>call</tt> instruction</a>. Thus, typically we
2650 <div class="doc_code">
2652 %X = call i32 asm "<a href="#int_bswap">bswap</a> $0", "=r,r"(i32 %Y)
2656 <p>Inline asms with side effects not visible in the constraint list must be
2657 marked as having side effects. This is done through the use of the
2658 '<tt>sideeffect</tt>' keyword, like so:</p>
2660 <div class="doc_code">
2662 call void asm sideeffect "eieio", ""()
2666 <p>In some cases inline asms will contain code that will not work unless the
2667 stack is aligned in some way, such as calls or SSE instructions on x86,
2668 yet will not contain code that does that alignment within the asm.
2669 The compiler should make conservative assumptions about what the asm might
2670 contain and should generate its usual stack alignment code in the prologue
2671 if the '<tt>alignstack</tt>' keyword is present:</p>
2673 <div class="doc_code">
2675 call void asm alignstack "eieio", ""()
2679 <p>If both keywords appear the '<tt>sideeffect</tt>' keyword must come
2682 <p>TODO: The format of the asm and constraints string still need to be
2683 documented here. Constraints on what can be done (e.g. duplication, moving,
2684 etc need to be documented). This is probably best done by reference to
2685 another document that covers inline asm from a holistic perspective.</p>
2688 <div class="doc_subsubsection">
2689 <a name="inlineasm_md">Inline Asm Metadata</a>
2692 <div class="doc_text">
2694 <p>The call instructions that wrap inline asm nodes may have a "!srcloc" MDNode
2695 attached to it that contains a constant integer. If present, the code
2696 generator will use the integer as the location cookie value when report
2697 errors through the LLVMContext error reporting mechanisms. This allows a
2698 front-end to correlate backend errors that occur with inline asm back to the
2699 source code that produced it. For example:</p>
2701 <div class="doc_code">
2703 call void asm sideeffect "something bad", ""()<b>, !srcloc !42</b>
2705 !42 = !{ i32 1234567 }
2709 <p>It is up to the front-end to make sense of the magic numbers it places in the
2714 <!-- ======================================================================= -->
2715 <div class="doc_subsection"><a name="metadata">Metadata Nodes and Metadata
2719 <div class="doc_text">
2721 <p>LLVM IR allows metadata to be attached to instructions in the program that
2722 can convey extra information about the code to the optimizers and code
2723 generator. One example application of metadata is source-level debug
2724 information. There are two metadata primitives: strings and nodes. All
2725 metadata has the <tt>metadata</tt> type and is identified in syntax by a
2726 preceding exclamation point ('<tt>!</tt>').</p>
2728 <p>A metadata string is a string surrounded by double quotes. It can contain
2729 any character by escaping non-printable characters with "\xx" where "xx" is
2730 the two digit hex code. For example: "<tt>!"test\00"</tt>".</p>
2732 <p>Metadata nodes are represented with notation similar to structure constants
2733 (a comma separated list of elements, surrounded by braces and preceded by an
2734 exclamation point). For example: "<tt>!{ metadata !"test\00", i32
2735 10}</tt>". Metadata nodes can have any values as their operand.</p>
2737 <p>A <a href="#namedmetadatastructure">named metadata</a> is a collection of
2738 metadata nodes, which can be looked up in the module symbol table. For
2739 example: "<tt>!foo = metadata !{!4, !3}</tt>".
2741 <p>Metadata can be used as function arguments. Here <tt>llvm.dbg.value</tt>
2742 function is using two metadata arguments.
2744 <div class="doc_code">
2746 call void @llvm.dbg.value(metadata !24, i64 0, metadata !25)
2750 <p>Metadata can be attached with an instruction. Here metadata <tt>!21</tt> is
2751 attached with <tt>add</tt> instruction using <tt>!dbg</tt> identifier.
2753 <div class="doc_code">
2755 %indvar.next = add i64 %indvar, 1, !dbg !21
2761 <!-- *********************************************************************** -->
2762 <div class="doc_section">
2763 <a name="intrinsic_globals">Intrinsic Global Variables</a>
2765 <!-- *********************************************************************** -->
2767 <p>LLVM has a number of "magic" global variables that contain data that affect
2768 code generation or other IR semantics. These are documented here. All globals
2769 of this sort should have a section specified as "<tt>llvm.metadata</tt>". This
2770 section and all globals that start with "<tt>llvm.</tt>" are reserved for use
2773 <!-- ======================================================================= -->
2774 <div class="doc_subsection">
2775 <a name="intg_used">The '<tt>llvm.used</tt>' Global Variable</a>
2778 <div class="doc_text">
2780 <p>The <tt>@llvm.used</tt> global is an array with i8* element type which has <a
2781 href="#linkage_appending">appending linkage</a>. This array contains a list of
2782 pointers to global variables and functions which may optionally have a pointer
2783 cast formed of bitcast or getelementptr. For example, a legal use of it is:</p>
2789 @llvm.used = appending global [2 x i8*] [
2791 i8* bitcast (i32* @Y to i8*)
2792 ], section "llvm.metadata"
2795 <p>If a global variable appears in the <tt>@llvm.used</tt> list, then the
2796 compiler, assembler, and linker are required to treat the symbol as if there is
2797 a reference to the global that it cannot see. For example, if a variable has
2798 internal linkage and no references other than that from the <tt>@llvm.used</tt>
2799 list, it cannot be deleted. This is commonly used to represent references from
2800 inline asms and other things the compiler cannot "see", and corresponds to
2801 "attribute((used))" in GNU C.</p>
2803 <p>On some targets, the code generator must emit a directive to the assembler or
2804 object file to prevent the assembler and linker from molesting the symbol.</p>
2808 <!-- ======================================================================= -->
2809 <div class="doc_subsection">
2810 <a name="intg_compiler_used">The '<tt>llvm.compiler.used</tt>' Global Variable</a>
2813 <div class="doc_text">
2815 <p>The <tt>@llvm.compiler.used</tt> directive is the same as the
2816 <tt>@llvm.used</tt> directive, except that it only prevents the compiler from
2817 touching the symbol. On targets that support it, this allows an intelligent
2818 linker to optimize references to the symbol without being impeded as it would be
2819 by <tt>@llvm.used</tt>.</p>
2821 <p>This is a rare construct that should only be used in rare circumstances, and
2822 should not be exposed to source languages.</p>
2826 <!-- ======================================================================= -->
2827 <div class="doc_subsection">
2828 <a name="intg_global_ctors">The '<tt>llvm.global_ctors</tt>' Global Variable</a>
2831 <div class="doc_text">
2833 %0 = type { i32, void ()* }
2834 @llvm.global_ctors = appending global [1 x %0] [%0 { i32 65535, void ()* @ctor }]
2836 <p>The <tt>@llvm.global_ctors</tt> array contains a list of constructor functions and associated priorities. The functions referenced by this array will be called in ascending order of priority (i.e. lowest first) when the module is loaded. The order of functions with the same priority is not defined.
2841 <!-- ======================================================================= -->
2842 <div class="doc_subsection">
2843 <a name="intg_global_dtors">The '<tt>llvm.global_dtors</tt>' Global Variable</a>
2846 <div class="doc_text">
2848 %0 = type { i32, void ()* }
2849 @llvm.global_dtors = appending global [1 x %0] [%0 { i32 65535, void ()* @dtor }]
2852 <p>The <tt>@llvm.global_dtors</tt> array contains a list of destructor functions and associated priorities. The functions referenced by this array will be called in descending order of priority (i.e. highest first) when the module is loaded. The order of functions with the same priority is not defined.
2858 <!-- *********************************************************************** -->
2859 <div class="doc_section"> <a name="instref">Instruction Reference</a> </div>
2860 <!-- *********************************************************************** -->
2862 <div class="doc_text">
2864 <p>The LLVM instruction set consists of several different classifications of
2865 instructions: <a href="#terminators">terminator
2866 instructions</a>, <a href="#binaryops">binary instructions</a>,
2867 <a href="#bitwiseops">bitwise binary instructions</a>,
2868 <a href="#memoryops">memory instructions</a>, and
2869 <a href="#otherops">other instructions</a>.</p>
2873 <!-- ======================================================================= -->
2874 <div class="doc_subsection"> <a name="terminators">Terminator
2875 Instructions</a> </div>
2877 <div class="doc_text">
2879 <p>As mentioned <a href="#functionstructure">previously</a>, every basic block
2880 in a program ends with a "Terminator" instruction, which indicates which
2881 block should be executed after the current block is finished. These
2882 terminator instructions typically yield a '<tt>void</tt>' value: they produce
2883 control flow, not values (the one exception being the
2884 '<a href="#i_invoke"><tt>invoke</tt></a>' instruction).</p>
2886 <p>There are seven different terminator instructions: the
2887 '<a href="#i_ret"><tt>ret</tt></a>' instruction, the
2888 '<a href="#i_br"><tt>br</tt></a>' instruction, the
2889 '<a href="#i_switch"><tt>switch</tt></a>' instruction, the
2890 '<a href="#i_indirectbr">'<tt>indirectbr</tt></a>' Instruction, the
2891 '<a href="#i_invoke"><tt>invoke</tt></a>' instruction, the
2892 '<a href="#i_unwind"><tt>unwind</tt></a>' instruction, and the
2893 '<a href="#i_unreachable"><tt>unreachable</tt></a>' instruction.</p>
2897 <!-- _______________________________________________________________________ -->
2898 <div class="doc_subsubsection"> <a name="i_ret">'<tt>ret</tt>'
2899 Instruction</a> </div>
2901 <div class="doc_text">
2905 ret <type> <value> <i>; Return a value from a non-void function</i>
2906 ret void <i>; Return from void function</i>
2910 <p>The '<tt>ret</tt>' instruction is used to return control flow (and optionally
2911 a value) from a function back to the caller.</p>
2913 <p>There are two forms of the '<tt>ret</tt>' instruction: one that returns a
2914 value and then causes control flow, and one that just causes control flow to
2918 <p>The '<tt>ret</tt>' instruction optionally accepts a single argument, the
2919 return value. The type of the return value must be a
2920 '<a href="#t_firstclass">first class</a>' type.</p>
2922 <p>A function is not <a href="#wellformed">well formed</a> if it it has a
2923 non-void return type and contains a '<tt>ret</tt>' instruction with no return
2924 value or a return value with a type that does not match its type, or if it
2925 has a void return type and contains a '<tt>ret</tt>' instruction with a
2929 <p>When the '<tt>ret</tt>' instruction is executed, control flow returns back to
2930 the calling function's context. If the caller is a
2931 "<a href="#i_call"><tt>call</tt></a>" instruction, execution continues at the
2932 instruction after the call. If the caller was an
2933 "<a href="#i_invoke"><tt>invoke</tt></a>" instruction, execution continues at
2934 the beginning of the "normal" destination block. If the instruction returns
2935 a value, that value shall set the call or invoke instruction's return
2940 ret i32 5 <i>; Return an integer value of 5</i>
2941 ret void <i>; Return from a void function</i>
2942 ret { i32, i8 } { i32 4, i8 2 } <i>; Return a struct of values 4 and 2</i>
2946 <!-- _______________________________________________________________________ -->
2947 <div class="doc_subsubsection"> <a name="i_br">'<tt>br</tt>' Instruction</a> </div>
2949 <div class="doc_text">
2953 br i1 <cond>, label <iftrue>, label <iffalse><br> br label <dest> <i>; Unconditional branch</i>
2957 <p>The '<tt>br</tt>' instruction is used to cause control flow to transfer to a
2958 different basic block in the current function. There are two forms of this
2959 instruction, corresponding to a conditional branch and an unconditional
2963 <p>The conditional branch form of the '<tt>br</tt>' instruction takes a single
2964 '<tt>i1</tt>' value and two '<tt>label</tt>' values. The unconditional form
2965 of the '<tt>br</tt>' instruction takes a single '<tt>label</tt>' value as a
2969 <p>Upon execution of a conditional '<tt>br</tt>' instruction, the '<tt>i1</tt>'
2970 argument is evaluated. If the value is <tt>true</tt>, control flows to the
2971 '<tt>iftrue</tt>' <tt>label</tt> argument. If "cond" is <tt>false</tt>,
2972 control flows to the '<tt>iffalse</tt>' <tt>label</tt> argument.</p>
2977 %cond = <a href="#i_icmp">icmp</a> eq i32 %a, %b
2978 br i1 %cond, label %IfEqual, label %IfUnequal
2980 <a href="#i_ret">ret</a> i32 1
2982 <a href="#i_ret">ret</a> i32 0
2987 <!-- _______________________________________________________________________ -->
2988 <div class="doc_subsubsection">
2989 <a name="i_switch">'<tt>switch</tt>' Instruction</a>
2992 <div class="doc_text">
2996 switch <intty> <value>, label <defaultdest> [ <intty> <val>, label <dest> ... ]
3000 <p>The '<tt>switch</tt>' instruction is used to transfer control flow to one of
3001 several different places. It is a generalization of the '<tt>br</tt>'
3002 instruction, allowing a branch to occur to one of many possible
3006 <p>The '<tt>switch</tt>' instruction uses three parameters: an integer
3007 comparison value '<tt>value</tt>', a default '<tt>label</tt>' destination,
3008 and an array of pairs of comparison value constants and '<tt>label</tt>'s.
3009 The table is not allowed to contain duplicate constant entries.</p>
3012 <p>The <tt>switch</tt> instruction specifies a table of values and
3013 destinations. When the '<tt>switch</tt>' instruction is executed, this table
3014 is searched for the given value. If the value is found, control flow is
3015 transferred to the corresponding destination; otherwise, control flow is
3016 transferred to the default destination.</p>
3018 <h5>Implementation:</h5>
3019 <p>Depending on properties of the target machine and the particular
3020 <tt>switch</tt> instruction, this instruction may be code generated in
3021 different ways. For example, it could be generated as a series of chained
3022 conditional branches or with a lookup table.</p>
3026 <i>; Emulate a conditional br instruction</i>
3027 %Val = <a href="#i_zext">zext</a> i1 %value to i32
3028 switch i32 %Val, label %truedest [ i32 0, label %falsedest ]
3030 <i>; Emulate an unconditional br instruction</i>
3031 switch i32 0, label %dest [ ]
3033 <i>; Implement a jump table:</i>
3034 switch i32 %val, label %otherwise [ i32 0, label %onzero
3036 i32 2, label %ontwo ]
3042 <!-- _______________________________________________________________________ -->
3043 <div class="doc_subsubsection">
3044 <a name="i_indirectbr">'<tt>indirectbr</tt>' Instruction</a>
3047 <div class="doc_text">
3051 indirectbr <somety>* <address>, [ label <dest1>, label <dest2>, ... ]
3056 <p>The '<tt>indirectbr</tt>' instruction implements an indirect branch to a label
3057 within the current function, whose address is specified by
3058 "<tt>address</tt>". Address must be derived from a <a
3059 href="#blockaddress">blockaddress</a> constant.</p>
3063 <p>The '<tt>address</tt>' argument is the address of the label to jump to. The
3064 rest of the arguments indicate the full set of possible destinations that the
3065 address may point to. Blocks are allowed to occur multiple times in the
3066 destination list, though this isn't particularly useful.</p>
3068 <p>This destination list is required so that dataflow analysis has an accurate
3069 understanding of the CFG.</p>
3073 <p>Control transfers to the block specified in the address argument. All
3074 possible destination blocks must be listed in the label list, otherwise this
3075 instruction has undefined behavior. This implies that jumps to labels
3076 defined in other functions have undefined behavior as well.</p>
3078 <h5>Implementation:</h5>
3080 <p>This is typically implemented with a jump through a register.</p>
3084 indirectbr i8* %Addr, [ label %bb1, label %bb2, label %bb3 ]
3090 <!-- _______________________________________________________________________ -->
3091 <div class="doc_subsubsection">
3092 <a name="i_invoke">'<tt>invoke</tt>' Instruction</a>
3095 <div class="doc_text">
3099 <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>]
3100 to label <normal label> unwind label <exception label>
3104 <p>The '<tt>invoke</tt>' instruction causes control to transfer to a specified
3105 function, with the possibility of control flow transfer to either the
3106 '<tt>normal</tt>' label or the '<tt>exception</tt>' label. If the callee
3107 function returns with the "<tt><a href="#i_ret">ret</a></tt>" instruction,
3108 control flow will return to the "normal" label. If the callee (or any
3109 indirect callees) returns with the "<a href="#i_unwind"><tt>unwind</tt></a>"
3110 instruction, control is interrupted and continued at the dynamically nearest
3111 "exception" label.</p>
3114 <p>This instruction requires several arguments:</p>
3117 <li>The optional "cconv" marker indicates which <a href="#callingconv">calling
3118 convention</a> the call should use. If none is specified, the call
3119 defaults to using C calling conventions.</li>
3121 <li>The optional <a href="#paramattrs">Parameter Attributes</a> list for
3122 return values. Only '<tt>zeroext</tt>', '<tt>signext</tt>', and
3123 '<tt>inreg</tt>' attributes are valid here.</li>
3125 <li>'<tt>ptr to function ty</tt>': shall be the signature of the pointer to
3126 function value being invoked. In most cases, this is a direct function
3127 invocation, but indirect <tt>invoke</tt>s are just as possible, branching
3128 off an arbitrary pointer to function value.</li>
3130 <li>'<tt>function ptr val</tt>': An LLVM value containing a pointer to a
3131 function to be invoked. </li>
3133 <li>'<tt>function args</tt>': argument list whose types match the function
3134 signature argument types and parameter attributes. All arguments must be
3135 of <a href="#t_firstclass">first class</a> type. If the function
3136 signature indicates the function accepts a variable number of arguments,
3137 the extra arguments can be specified.</li>
3139 <li>'<tt>normal label</tt>': the label reached when the called function
3140 executes a '<tt><a href="#i_ret">ret</a></tt>' instruction. </li>
3142 <li>'<tt>exception label</tt>': the label reached when a callee returns with
3143 the <a href="#i_unwind"><tt>unwind</tt></a> instruction. </li>
3145 <li>The optional <a href="#fnattrs">function attributes</a> list. Only
3146 '<tt>noreturn</tt>', '<tt>nounwind</tt>', '<tt>readonly</tt>' and
3147 '<tt>readnone</tt>' attributes are valid here.</li>
3151 <p>This instruction is designed to operate as a standard
3152 '<tt><a href="#i_call">call</a></tt>' instruction in most regards. The
3153 primary difference is that it establishes an association with a label, which
3154 is used by the runtime library to unwind the stack.</p>
3156 <p>This instruction is used in languages with destructors to ensure that proper
3157 cleanup is performed in the case of either a <tt>longjmp</tt> or a thrown
3158 exception. Additionally, this is important for implementation of
3159 '<tt>catch</tt>' clauses in high-level languages that support them.</p>
3161 <p>For the purposes of the SSA form, the definition of the value returned by the
3162 '<tt>invoke</tt>' instruction is deemed to occur on the edge from the current
3163 block to the "normal" label. If the callee unwinds then no return value is
3166 <p>Note that the code generator does not yet completely support unwind, and
3167 that the invoke/unwind semantics are likely to change in future versions.</p>
3171 %retval = invoke i32 @Test(i32 15) to label %Continue
3172 unwind label %TestCleanup <i>; {i32}:retval set</i>
3173 %retval = invoke <a href="#callingconv">coldcc</a> i32 %Testfnptr(i32 15) to label %Continue
3174 unwind label %TestCleanup <i>; {i32}:retval set</i>
3179 <!-- _______________________________________________________________________ -->
3181 <div class="doc_subsubsection"> <a name="i_unwind">'<tt>unwind</tt>'
3182 Instruction</a> </div>
3184 <div class="doc_text">
3192 <p>The '<tt>unwind</tt>' instruction unwinds the stack, continuing control flow
3193 at the first callee in the dynamic call stack which used
3194 an <a href="#i_invoke"><tt>invoke</tt></a> instruction to perform the call.
3195 This is primarily used to implement exception handling.</p>
3198 <p>The '<tt>unwind</tt>' instruction causes execution of the current function to
3199 immediately halt. The dynamic call stack is then searched for the
3200 first <a href="#i_invoke"><tt>invoke</tt></a> instruction on the call stack.
3201 Once found, execution continues at the "exceptional" destination block
3202 specified by the <tt>invoke</tt> instruction. If there is no <tt>invoke</tt>
3203 instruction in the dynamic call chain, undefined behavior results.</p>
3205 <p>Note that the code generator does not yet completely support unwind, and
3206 that the invoke/unwind semantics are likely to change in future versions.</p>
3210 <!-- _______________________________________________________________________ -->
3212 <div class="doc_subsubsection"> <a name="i_unreachable">'<tt>unreachable</tt>'
3213 Instruction</a> </div>
3215 <div class="doc_text">
3223 <p>The '<tt>unreachable</tt>' instruction has no defined semantics. This
3224 instruction is used to inform the optimizer that a particular portion of the
3225 code is not reachable. This can be used to indicate that the code after a
3226 no-return function cannot be reached, and other facts.</p>
3229 <p>The '<tt>unreachable</tt>' instruction has no defined semantics.</p>
3233 <!-- ======================================================================= -->
3234 <div class="doc_subsection"> <a name="binaryops">Binary Operations</a> </div>
3236 <div class="doc_text">
3238 <p>Binary operators are used to do most of the computation in a program. They
3239 require two operands of the same type, execute an operation on them, and
3240 produce a single value. The operands might represent multiple data, as is
3241 the case with the <a href="#t_vector">vector</a> data type. The result value
3242 has the same type as its operands.</p>
3244 <p>There are several different binary operators:</p>
3248 <!-- _______________________________________________________________________ -->
3249 <div class="doc_subsubsection">
3250 <a name="i_add">'<tt>add</tt>' Instruction</a>
3253 <div class="doc_text">
3257 <result> = add <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3258 <result> = add nuw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3259 <result> = add nsw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3260 <result> = add nuw nsw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3264 <p>The '<tt>add</tt>' instruction returns the sum of its two operands.</p>
3267 <p>The two arguments to the '<tt>add</tt>' instruction must
3268 be <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
3269 integer values. Both arguments must have identical types.</p>
3272 <p>The value produced is the integer sum of the two operands.</p>
3274 <p>If the sum has unsigned overflow, the result returned is the mathematical
3275 result modulo 2<sup>n</sup>, where n is the bit width of the result.</p>
3277 <p>Because LLVM integers use a two's complement representation, this instruction
3278 is appropriate for both signed and unsigned integers.</p>
3280 <p><tt>nuw</tt> and <tt>nsw</tt> stand for "No Unsigned Wrap"
3281 and "No Signed Wrap", respectively. If the <tt>nuw</tt> and/or
3282 <tt>nsw</tt> keywords are present, the result value of the <tt>add</tt>
3283 is a <a href="#trapvalues">trap value</a> if unsigned and/or signed overflow,
3284 respectively, occurs.</p>
3288 <result> = add i32 4, %var <i>; yields {i32}:result = 4 + %var</i>
3293 <!-- _______________________________________________________________________ -->
3294 <div class="doc_subsubsection">
3295 <a name="i_fadd">'<tt>fadd</tt>' Instruction</a>
3298 <div class="doc_text">
3302 <result> = fadd <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3306 <p>The '<tt>fadd</tt>' instruction returns the sum of its two operands.</p>
3309 <p>The two arguments to the '<tt>fadd</tt>' instruction must be
3310 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3311 floating point values. Both arguments must have identical types.</p>
3314 <p>The value produced is the floating point sum of the two operands.</p>
3318 <result> = fadd float 4.0, %var <i>; yields {float}:result = 4.0 + %var</i>
3323 <!-- _______________________________________________________________________ -->
3324 <div class="doc_subsubsection">
3325 <a name="i_sub">'<tt>sub</tt>' Instruction</a>
3328 <div class="doc_text">
3332 <result> = sub <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3333 <result> = sub nuw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3334 <result> = sub nsw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3335 <result> = sub nuw nsw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3339 <p>The '<tt>sub</tt>' instruction returns the difference of its two
3342 <p>Note that the '<tt>sub</tt>' instruction is used to represent the
3343 '<tt>neg</tt>' instruction present in most other intermediate
3344 representations.</p>
3347 <p>The two arguments to the '<tt>sub</tt>' instruction must
3348 be <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
3349 integer values. Both arguments must have identical types.</p>
3352 <p>The value produced is the integer difference of the two operands.</p>
3354 <p>If the difference has unsigned overflow, the result returned is the
3355 mathematical result modulo 2<sup>n</sup>, where n is the bit width of the
3358 <p>Because LLVM integers use a two's complement representation, this instruction
3359 is appropriate for both signed and unsigned integers.</p>
3361 <p><tt>nuw</tt> and <tt>nsw</tt> stand for "No Unsigned Wrap"
3362 and "No Signed Wrap", respectively. If the <tt>nuw</tt> and/or
3363 <tt>nsw</tt> keywords are present, the result value of the <tt>sub</tt>
3364 is a <a href="#trapvalues">trap value</a> if unsigned and/or signed overflow,
3365 respectively, occurs.</p>
3369 <result> = sub i32 4, %var <i>; yields {i32}:result = 4 - %var</i>
3370 <result> = sub i32 0, %val <i>; yields {i32}:result = -%var</i>
3375 <!-- _______________________________________________________________________ -->
3376 <div class="doc_subsubsection">
3377 <a name="i_fsub">'<tt>fsub</tt>' Instruction</a>
3380 <div class="doc_text">
3384 <result> = fsub <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3388 <p>The '<tt>fsub</tt>' instruction returns the difference of its two
3391 <p>Note that the '<tt>fsub</tt>' instruction is used to represent the
3392 '<tt>fneg</tt>' instruction present in most other intermediate
3393 representations.</p>
3396 <p>The two arguments to the '<tt>fsub</tt>' instruction must be
3397 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3398 floating point values. Both arguments must have identical types.</p>
3401 <p>The value produced is the floating point difference of the two operands.</p>
3405 <result> = fsub float 4.0, %var <i>; yields {float}:result = 4.0 - %var</i>
3406 <result> = fsub float -0.0, %val <i>; yields {float}:result = -%var</i>
3411 <!-- _______________________________________________________________________ -->
3412 <div class="doc_subsubsection">
3413 <a name="i_mul">'<tt>mul</tt>' Instruction</a>
3416 <div class="doc_text">
3420 <result> = mul <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3421 <result> = mul nuw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3422 <result> = mul nsw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3423 <result> = mul nuw nsw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3427 <p>The '<tt>mul</tt>' instruction returns the product of its two operands.</p>
3430 <p>The two arguments to the '<tt>mul</tt>' instruction must
3431 be <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
3432 integer values. Both arguments must have identical types.</p>
3435 <p>The value produced is the integer product of the two operands.</p>
3437 <p>If the result of the multiplication has unsigned overflow, the result
3438 returned is the mathematical result modulo 2<sup>n</sup>, where n is the bit
3439 width of the result.</p>
3441 <p>Because LLVM integers use a two's complement representation, and the result
3442 is the same width as the operands, this instruction returns the correct
3443 result for both signed and unsigned integers. If a full product
3444 (e.g. <tt>i32</tt>x<tt>i32</tt>-><tt>i64</tt>) is needed, the operands should
3445 be sign-extended or zero-extended as appropriate to the width of the full
3448 <p><tt>nuw</tt> and <tt>nsw</tt> stand for "No Unsigned Wrap"
3449 and "No Signed Wrap", respectively. If the <tt>nuw</tt> and/or
3450 <tt>nsw</tt> keywords are present, the result value of the <tt>mul</tt>
3451 is a <a href="#trapvalues">trap value</a> if unsigned and/or signed overflow,
3452 respectively, occurs.</p>
3456 <result> = mul i32 4, %var <i>; yields {i32}:result = 4 * %var</i>
3461 <!-- _______________________________________________________________________ -->
3462 <div class="doc_subsubsection">
3463 <a name="i_fmul">'<tt>fmul</tt>' Instruction</a>
3466 <div class="doc_text">
3470 <result> = fmul <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3474 <p>The '<tt>fmul</tt>' instruction returns the product of its two operands.</p>
3477 <p>The two arguments to the '<tt>fmul</tt>' instruction must be
3478 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3479 floating point values. Both arguments must have identical types.</p>
3482 <p>The value produced is the floating point product of the two operands.</p>
3486 <result> = fmul float 4.0, %var <i>; yields {float}:result = 4.0 * %var</i>
3491 <!-- _______________________________________________________________________ -->
3492 <div class="doc_subsubsection"> <a name="i_udiv">'<tt>udiv</tt>' Instruction
3495 <div class="doc_text">
3499 <result> = udiv <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3503 <p>The '<tt>udiv</tt>' instruction returns the quotient of its two operands.</p>
3506 <p>The two arguments to the '<tt>udiv</tt>' instruction must be
3507 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3508 values. Both arguments must have identical types.</p>
3511 <p>The value produced is the unsigned integer quotient of the two operands.</p>
3513 <p>Note that unsigned integer division and signed integer division are distinct
3514 operations; for signed integer division, use '<tt>sdiv</tt>'.</p>
3516 <p>Division by zero leads to undefined behavior.</p>
3520 <result> = udiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
3525 <!-- _______________________________________________________________________ -->
3526 <div class="doc_subsubsection"> <a name="i_sdiv">'<tt>sdiv</tt>' Instruction
3529 <div class="doc_text">
3533 <result> = sdiv <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3534 <result> = sdiv exact <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3538 <p>The '<tt>sdiv</tt>' instruction returns the quotient of its two operands.</p>
3541 <p>The two arguments to the '<tt>sdiv</tt>' instruction must be
3542 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3543 values. Both arguments must have identical types.</p>
3546 <p>The value produced is the signed integer quotient of the two operands rounded
3549 <p>Note that signed integer division and unsigned integer division are distinct
3550 operations; for unsigned integer division, use '<tt>udiv</tt>'.</p>
3552 <p>Division by zero leads to undefined behavior. Overflow also leads to
3553 undefined behavior; this is a rare case, but can occur, for example, by doing
3554 a 32-bit division of -2147483648 by -1.</p>
3556 <p>If the <tt>exact</tt> keyword is present, the result value of the
3557 <tt>sdiv</tt> is a <a href="#trapvalues">trap value</a> if the result would
3558 be rounded or if overflow would occur.</p>
3562 <result> = sdiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
3567 <!-- _______________________________________________________________________ -->
3568 <div class="doc_subsubsection"> <a name="i_fdiv">'<tt>fdiv</tt>'
3569 Instruction</a> </div>
3571 <div class="doc_text">
3575 <result> = fdiv <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3579 <p>The '<tt>fdiv</tt>' instruction returns the quotient of its two operands.</p>
3582 <p>The two arguments to the '<tt>fdiv</tt>' instruction must be
3583 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3584 floating point values. Both arguments must have identical types.</p>
3587 <p>The value produced is the floating point quotient of the two operands.</p>
3591 <result> = fdiv float 4.0, %var <i>; yields {float}:result = 4.0 / %var</i>
3596 <!-- _______________________________________________________________________ -->
3597 <div class="doc_subsubsection"> <a name="i_urem">'<tt>urem</tt>' Instruction</a>
3600 <div class="doc_text">
3604 <result> = urem <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3608 <p>The '<tt>urem</tt>' instruction returns the remainder from the unsigned
3609 division of its two arguments.</p>
3612 <p>The two arguments to the '<tt>urem</tt>' instruction must be
3613 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3614 values. Both arguments must have identical types.</p>
3617 <p>This instruction returns the unsigned integer <i>remainder</i> of a division.
3618 This instruction always performs an unsigned division to get the
3621 <p>Note that unsigned integer remainder and signed integer remainder are
3622 distinct operations; for signed integer remainder, use '<tt>srem</tt>'.</p>
3624 <p>Taking the remainder of a division by zero leads to undefined behavior.</p>
3628 <result> = urem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
3633 <!-- _______________________________________________________________________ -->
3634 <div class="doc_subsubsection">
3635 <a name="i_srem">'<tt>srem</tt>' Instruction</a>
3638 <div class="doc_text">
3642 <result> = srem <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3646 <p>The '<tt>srem</tt>' instruction returns the remainder from the signed
3647 division of its two operands. This instruction can also take
3648 <a href="#t_vector">vector</a> versions of the values in which case the
3649 elements must be integers.</p>
3652 <p>The two arguments to the '<tt>srem</tt>' instruction must be
3653 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3654 values. Both arguments must have identical types.</p>
3657 <p>This instruction returns the <i>remainder</i> of a division (where the result
3658 has the same sign as the dividend, <tt>op1</tt>), not the <i>modulo</i>
3659 operator (where the result has the same sign as the divisor, <tt>op2</tt>) of
3660 a value. For more information about the difference,
3661 see <a href="http://mathforum.org/dr.math/problems/anne.4.28.99.html">The
3662 Math Forum</a>. For a table of how this is implemented in various languages,
3663 please see <a href="http://en.wikipedia.org/wiki/Modulo_operation">
3664 Wikipedia: modulo operation</a>.</p>
3666 <p>Note that signed integer remainder and unsigned integer remainder are
3667 distinct operations; for unsigned integer remainder, use '<tt>urem</tt>'.</p>
3669 <p>Taking the remainder of a division by zero leads to undefined behavior.
3670 Overflow also leads to undefined behavior; this is a rare case, but can
3671 occur, for example, by taking the remainder of a 32-bit division of
3672 -2147483648 by -1. (The remainder doesn't actually overflow, but this rule
3673 lets srem be implemented using instructions that return both the result of
3674 the division and the remainder.)</p>
3678 <result> = srem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
3683 <!-- _______________________________________________________________________ -->
3684 <div class="doc_subsubsection">
3685 <a name="i_frem">'<tt>frem</tt>' Instruction</a> </div>
3687 <div class="doc_text">
3691 <result> = frem <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3695 <p>The '<tt>frem</tt>' instruction returns the remainder from the division of
3696 its two operands.</p>
3699 <p>The two arguments to the '<tt>frem</tt>' instruction must be
3700 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3701 floating point values. Both arguments must have identical types.</p>
3704 <p>This instruction returns the <i>remainder</i> of a division. The remainder
3705 has the same sign as the dividend.</p>
3709 <result> = frem float 4.0, %var <i>; yields {float}:result = 4.0 % %var</i>
3714 <!-- ======================================================================= -->
3715 <div class="doc_subsection"> <a name="bitwiseops">Bitwise Binary
3716 Operations</a> </div>
3718 <div class="doc_text">
3720 <p>Bitwise binary operators are used to do various forms of bit-twiddling in a
3721 program. They are generally very efficient instructions and can commonly be
3722 strength reduced from other instructions. They require two operands of the
3723 same type, execute an operation on them, and produce a single value. The
3724 resulting value is the same type as its operands.</p>
3728 <!-- _______________________________________________________________________ -->
3729 <div class="doc_subsubsection"> <a name="i_shl">'<tt>shl</tt>'
3730 Instruction</a> </div>
3732 <div class="doc_text">
3736 <result> = shl <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3740 <p>The '<tt>shl</tt>' instruction returns the first operand shifted to the left
3741 a specified number of bits.</p>
3744 <p>Both arguments to the '<tt>shl</tt>' instruction must be the
3745 same <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
3746 integer type. '<tt>op2</tt>' is treated as an unsigned value.</p>
3749 <p>The value produced is <tt>op1</tt> * 2<sup><tt>op2</tt></sup> mod
3750 2<sup>n</sup>, where <tt>n</tt> is the width of the result. If <tt>op2</tt>
3751 is (statically or dynamically) negative or equal to or larger than the number
3752 of bits in <tt>op1</tt>, the result is undefined. If the arguments are
3753 vectors, each vector element of <tt>op1</tt> is shifted by the corresponding
3754 shift amount in <tt>op2</tt>.</p>
3758 <result> = shl i32 4, %var <i>; yields {i32}: 4 << %var</i>
3759 <result> = shl i32 4, 2 <i>; yields {i32}: 16</i>
3760 <result> = shl i32 1, 10 <i>; yields {i32}: 1024</i>
3761 <result> = shl i32 1, 32 <i>; undefined</i>
3762 <result> = shl <2 x i32> < i32 1, i32 1>, < i32 1, i32 2> <i>; yields: result=<2 x i32> < i32 2, i32 4></i>
3767 <!-- _______________________________________________________________________ -->
3768 <div class="doc_subsubsection"> <a name="i_lshr">'<tt>lshr</tt>'
3769 Instruction</a> </div>
3771 <div class="doc_text">
3775 <result> = lshr <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3779 <p>The '<tt>lshr</tt>' instruction (logical shift right) returns the first
3780 operand shifted to the right a specified number of bits with zero fill.</p>
3783 <p>Both arguments to the '<tt>lshr</tt>' instruction must be the same
3784 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3785 type. '<tt>op2</tt>' is treated as an unsigned value.</p>
3788 <p>This instruction always performs a logical shift right operation. The most
3789 significant bits of the result will be filled with zero bits after the shift.
3790 If <tt>op2</tt> is (statically or dynamically) equal to or larger than the
3791 number of bits in <tt>op1</tt>, the result is undefined. If the arguments are
3792 vectors, each vector element of <tt>op1</tt> is shifted by the corresponding
3793 shift amount in <tt>op2</tt>.</p>
3797 <result> = lshr i32 4, 1 <i>; yields {i32}:result = 2</i>
3798 <result> = lshr i32 4, 2 <i>; yields {i32}:result = 1</i>
3799 <result> = lshr i8 4, 3 <i>; yields {i8}:result = 0</i>
3800 <result> = lshr i8 -2, 1 <i>; yields {i8}:result = 0x7FFFFFFF </i>
3801 <result> = lshr i32 1, 32 <i>; undefined</i>
3802 <result> = lshr <2 x i32> < i32 -2, i32 4>, < i32 1, i32 2> <i>; yields: result=<2 x i32> < i32 0x7FFFFFFF, i32 1></i>
3807 <!-- _______________________________________________________________________ -->
3808 <div class="doc_subsubsection"> <a name="i_ashr">'<tt>ashr</tt>'
3809 Instruction</a> </div>
3810 <div class="doc_text">
3814 <result> = ashr <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3818 <p>The '<tt>ashr</tt>' instruction (arithmetic shift right) returns the first
3819 operand shifted to the right a specified number of bits with sign
3823 <p>Both arguments to the '<tt>ashr</tt>' instruction must be the same
3824 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3825 type. '<tt>op2</tt>' is treated as an unsigned value.</p>
3828 <p>This instruction always performs an arithmetic shift right operation, The
3829 most significant bits of the result will be filled with the sign bit
3830 of <tt>op1</tt>. If <tt>op2</tt> is (statically or dynamically) equal to or
3831 larger than the number of bits in <tt>op1</tt>, the result is undefined. If
3832 the arguments are vectors, each vector element of <tt>op1</tt> is shifted by
3833 the corresponding shift amount in <tt>op2</tt>.</p>
3837 <result> = ashr i32 4, 1 <i>; yields {i32}:result = 2</i>
3838 <result> = ashr i32 4, 2 <i>; yields {i32}:result = 1</i>
3839 <result> = ashr i8 4, 3 <i>; yields {i8}:result = 0</i>
3840 <result> = ashr i8 -2, 1 <i>; yields {i8}:result = -1</i>
3841 <result> = ashr i32 1, 32 <i>; undefined</i>
3842 <result> = ashr <2 x i32> < i32 -2, i32 4>, < i32 1, i32 3> <i>; yields: result=<2 x i32> < i32 -1, i32 0></i>
3847 <!-- _______________________________________________________________________ -->
3848 <div class="doc_subsubsection"> <a name="i_and">'<tt>and</tt>'
3849 Instruction</a> </div>
3851 <div class="doc_text">
3855 <result> = and <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3859 <p>The '<tt>and</tt>' instruction returns the bitwise logical and of its two
3863 <p>The two arguments to the '<tt>and</tt>' instruction must be
3864 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3865 values. Both arguments must have identical types.</p>
3868 <p>The truth table used for the '<tt>and</tt>' instruction is:</p>
3870 <table border="1" cellspacing="0" cellpadding="4">
3902 <result> = and i32 4, %var <i>; yields {i32}:result = 4 & %var</i>
3903 <result> = and i32 15, 40 <i>; yields {i32}:result = 8</i>
3904 <result> = and i32 4, 8 <i>; yields {i32}:result = 0</i>
3907 <!-- _______________________________________________________________________ -->
3908 <div class="doc_subsubsection"> <a name="i_or">'<tt>or</tt>' Instruction</a> </div>
3910 <div class="doc_text">
3914 <result> = or <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3918 <p>The '<tt>or</tt>' instruction returns the bitwise logical inclusive or of its
3922 <p>The two arguments to the '<tt>or</tt>' instruction must be
3923 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3924 values. Both arguments must have identical types.</p>
3927 <p>The truth table used for the '<tt>or</tt>' instruction is:</p>
3929 <table border="1" cellspacing="0" cellpadding="4">
3961 <result> = or i32 4, %var <i>; yields {i32}:result = 4 | %var</i>
3962 <result> = or i32 15, 40 <i>; yields {i32}:result = 47</i>
3963 <result> = or i32 4, 8 <i>; yields {i32}:result = 12</i>
3968 <!-- _______________________________________________________________________ -->
3969 <div class="doc_subsubsection"> <a name="i_xor">'<tt>xor</tt>'
3970 Instruction</a> </div>
3972 <div class="doc_text">
3976 <result> = xor <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3980 <p>The '<tt>xor</tt>' instruction returns the bitwise logical exclusive or of
3981 its two operands. The <tt>xor</tt> is used to implement the "one's
3982 complement" operation, which is the "~" operator in C.</p>
3985 <p>The two arguments to the '<tt>xor</tt>' instruction must be
3986 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3987 values. Both arguments must have identical types.</p>
3990 <p>The truth table used for the '<tt>xor</tt>' instruction is:</p>
3992 <table border="1" cellspacing="0" cellpadding="4">
4024 <result> = xor i32 4, %var <i>; yields {i32}:result = 4 ^ %var</i>
4025 <result> = xor i32 15, 40 <i>; yields {i32}:result = 39</i>
4026 <result> = xor i32 4, 8 <i>; yields {i32}:result = 12</i>
4027 <result> = xor i32 %V, -1 <i>; yields {i32}:result = ~%V</i>
4032 <!-- ======================================================================= -->
4033 <div class="doc_subsection">
4034 <a name="vectorops">Vector Operations</a>
4037 <div class="doc_text">
4039 <p>LLVM supports several instructions to represent vector operations in a
4040 target-independent manner. These instructions cover the element-access and
4041 vector-specific operations needed to process vectors effectively. While LLVM
4042 does directly support these vector operations, many sophisticated algorithms
4043 will want to use target-specific intrinsics to take full advantage of a
4044 specific target.</p>
4048 <!-- _______________________________________________________________________ -->
4049 <div class="doc_subsubsection">
4050 <a name="i_extractelement">'<tt>extractelement</tt>' Instruction</a>
4053 <div class="doc_text">
4057 <result> = extractelement <n x <ty>> <val>, i32 <idx> <i>; yields <ty></i>
4061 <p>The '<tt>extractelement</tt>' instruction extracts a single scalar element
4062 from a vector at a specified index.</p>
4066 <p>The first operand of an '<tt>extractelement</tt>' instruction is a value
4067 of <a href="#t_vector">vector</a> type. The second operand is an index
4068 indicating the position from which to extract the element. The index may be
4072 <p>The result is a scalar of the same type as the element type of
4073 <tt>val</tt>. Its value is the value at position <tt>idx</tt> of
4074 <tt>val</tt>. If <tt>idx</tt> exceeds the length of <tt>val</tt>, the
4075 results are undefined.</p>
4079 <result> = extractelement <4 x i32> %vec, i32 0 <i>; yields i32</i>
4084 <!-- _______________________________________________________________________ -->
4085 <div class="doc_subsubsection">
4086 <a name="i_insertelement">'<tt>insertelement</tt>' Instruction</a>
4089 <div class="doc_text">
4093 <result> = insertelement <n x <ty>> <val>, <ty> <elt>, i32 <idx> <i>; yields <n x <ty>></i>
4097 <p>The '<tt>insertelement</tt>' instruction inserts a scalar element into a
4098 vector at a specified index.</p>
4101 <p>The first operand of an '<tt>insertelement</tt>' instruction is a value
4102 of <a href="#t_vector">vector</a> type. The second operand is a scalar value
4103 whose type must equal the element type of the first operand. The third
4104 operand is an index indicating the position at which to insert the value.
4105 The index may be a variable.</p>
4108 <p>The result is a vector of the same type as <tt>val</tt>. Its element values
4109 are those of <tt>val</tt> except at position <tt>idx</tt>, where it gets the
4110 value <tt>elt</tt>. If <tt>idx</tt> exceeds the length of <tt>val</tt>, the
4111 results are undefined.</p>
4115 <result> = insertelement <4 x i32> %vec, i32 1, i32 0 <i>; yields <4 x i32></i>
4120 <!-- _______________________________________________________________________ -->
4121 <div class="doc_subsubsection">
4122 <a name="i_shufflevector">'<tt>shufflevector</tt>' Instruction</a>
4125 <div class="doc_text">
4129 <result> = shufflevector <n x <ty>> <v1>, <n x <ty>> <v2>, <m x i32> <mask> <i>; yields <m x <ty>></i>
4133 <p>The '<tt>shufflevector</tt>' instruction constructs a permutation of elements
4134 from two input vectors, returning a vector with the same element type as the
4135 input and length that is the same as the shuffle mask.</p>
4138 <p>The first two operands of a '<tt>shufflevector</tt>' instruction are vectors
4139 with types that match each other. The third argument is a shuffle mask whose
4140 element type is always 'i32'. The result of the instruction is a vector
4141 whose length is the same as the shuffle mask and whose element type is the
4142 same as the element type of the first two operands.</p>
4144 <p>The shuffle mask operand is required to be a constant vector with either
4145 constant integer or undef values.</p>
4148 <p>The elements of the two input vectors are numbered from left to right across
4149 both of the vectors. The shuffle mask operand specifies, for each element of
4150 the result vector, which element of the two input vectors the result element
4151 gets. The element selector may be undef (meaning "don't care") and the
4152 second operand may be undef if performing a shuffle from only one vector.</p>
4156 <result> = shufflevector <4 x i32> %v1, <4 x i32> %v2,
4157 <4 x i32> <i32 0, i32 4, i32 1, i32 5> <i>; yields <4 x i32></i>
4158 <result> = shufflevector <4 x i32> %v1, <4 x i32> undef,
4159 <4 x i32> <i32 0, i32 1, i32 2, i32 3> <i>; yields <4 x i32></i> - Identity shuffle.
4160 <result> = shufflevector <8 x i32> %v1, <8 x i32> undef,
4161 <4 x i32> <i32 0, i32 1, i32 2, i32 3> <i>; yields <4 x i32></i>
4162 <result> = shufflevector <4 x i32> %v1, <4 x i32> %v2,
4163 <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>
4168 <!-- ======================================================================= -->
4169 <div class="doc_subsection">
4170 <a name="aggregateops">Aggregate Operations</a>
4173 <div class="doc_text">
4175 <p>LLVM supports several instructions for working with
4176 <a href="#t_aggregate">aggregate</a> values.</p>
4180 <!-- _______________________________________________________________________ -->
4181 <div class="doc_subsubsection">
4182 <a name="i_extractvalue">'<tt>extractvalue</tt>' Instruction</a>
4185 <div class="doc_text">
4189 <result> = extractvalue <aggregate type> <val>, <idx>{, <idx>}*
4193 <p>The '<tt>extractvalue</tt>' instruction extracts the value of a member field
4194 from an <a href="#t_aggregate">aggregate</a> value.</p>
4197 <p>The first operand of an '<tt>extractvalue</tt>' instruction is a value
4198 of <a href="#t_struct">struct</a>, <a href="#t_union">union</a> or
4199 <a href="#t_array">array</a> type. The operands are constant indices to
4200 specify which value to extract in a similar manner as indices in a
4201 '<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.</p>
4204 <p>The result is the value at the position in the aggregate specified by the
4209 <result> = extractvalue {i32, float} %agg, 0 <i>; yields i32</i>
4214 <!-- _______________________________________________________________________ -->
4215 <div class="doc_subsubsection">
4216 <a name="i_insertvalue">'<tt>insertvalue</tt>' Instruction</a>
4219 <div class="doc_text">
4223 <result> = insertvalue <aggregate type> <val>, <ty> <elt>, <idx> <i>; yields <aggregate type></i>
4227 <p>The '<tt>insertvalue</tt>' instruction inserts a value into a member field
4228 in an <a href="#t_aggregate">aggregate</a> value.</p>
4231 <p>The first operand of an '<tt>insertvalue</tt>' instruction is a value
4232 of <a href="#t_struct">struct</a>, <a href="#t_union">union</a> or
4233 <a href="#t_array">array</a> type. The second operand is a first-class
4234 value to insert. The following operands are constant indices indicating
4235 the position at which to insert the value in a similar manner as indices in a
4236 '<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction. The
4237 value to insert must have the same type as the value identified by the
4241 <p>The result is an aggregate of the same type as <tt>val</tt>. Its value is
4242 that of <tt>val</tt> except that the value at the position specified by the
4243 indices is that of <tt>elt</tt>.</p>
4247 %agg1 = insertvalue {i32, float} undef, i32 1, 0 <i>; yields {i32 1, float undef}</i>
4248 %agg2 = insertvalue {i32, float} %agg1, float %val, 1 <i>; yields {i32 1, float %val}</i>
4254 <!-- ======================================================================= -->
4255 <div class="doc_subsection">
4256 <a name="memoryops">Memory Access and Addressing Operations</a>
4259 <div class="doc_text">
4261 <p>A key design point of an SSA-based representation is how it represents
4262 memory. In LLVM, no memory locations are in SSA form, which makes things
4263 very simple. This section describes how to read, write, and allocate
4268 <!-- _______________________________________________________________________ -->
4269 <div class="doc_subsubsection">
4270 <a name="i_alloca">'<tt>alloca</tt>' Instruction</a>
4273 <div class="doc_text">
4277 <result> = alloca <type>[, <ty> <NumElements>][, align <alignment>] <i>; yields {type*}:result</i>
4281 <p>The '<tt>alloca</tt>' instruction allocates memory on the stack frame of the
4282 currently executing function, to be automatically released when this function
4283 returns to its caller. The object is always allocated in the generic address
4284 space (address space zero).</p>
4287 <p>The '<tt>alloca</tt>' instruction
4288 allocates <tt>sizeof(<type>)*NumElements</tt> bytes of memory on the
4289 runtime stack, returning a pointer of the appropriate type to the program.
4290 If "NumElements" is specified, it is the number of elements allocated,
4291 otherwise "NumElements" is defaulted to be one. If a constant alignment is
4292 specified, the value result of the allocation is guaranteed to be aligned to
4293 at least that boundary. If not specified, or if zero, the target can choose
4294 to align the allocation on any convenient boundary compatible with the
4297 <p>'<tt>type</tt>' may be any sized type.</p>
4300 <p>Memory is allocated; a pointer is returned. The operation is undefined if
4301 there is insufficient stack space for the allocation. '<tt>alloca</tt>'d
4302 memory is automatically released when the function returns. The
4303 '<tt>alloca</tt>' instruction is commonly used to represent automatic
4304 variables that must have an address available. When the function returns
4305 (either with the <tt><a href="#i_ret">ret</a></tt>
4306 or <tt><a href="#i_unwind">unwind</a></tt> instructions), the memory is
4307 reclaimed. Allocating zero bytes is legal, but the result is undefined.</p>
4311 %ptr = alloca i32 <i>; yields {i32*}:ptr</i>
4312 %ptr = alloca i32, i32 4 <i>; yields {i32*}:ptr</i>
4313 %ptr = alloca i32, i32 4, align 1024 <i>; yields {i32*}:ptr</i>
4314 %ptr = alloca i32, align 1024 <i>; yields {i32*}:ptr</i>
4319 <!-- _______________________________________________________________________ -->
4320 <div class="doc_subsubsection"> <a name="i_load">'<tt>load</tt>'
4321 Instruction</a> </div>
4323 <div class="doc_text">
4327 <result> = load <ty>* <pointer>[, align <alignment>][, !nontemporal !<index>]
4328 <result> = volatile load <ty>* <pointer>[, align <alignment>][, !nontemporal !<index>]
4329 !<index> = !{ i32 1 }
4333 <p>The '<tt>load</tt>' instruction is used to read from memory.</p>
4336 <p>The argument to the '<tt>load</tt>' instruction specifies the memory address
4337 from which to load. The pointer must point to
4338 a <a href="#t_firstclass">first class</a> type. If the <tt>load</tt> is
4339 marked as <tt>volatile</tt>, then the optimizer is not allowed to modify the
4340 number or order of execution of this <tt>load</tt> with other <a
4341 href="#volatile">volatile operations</a>.</p>
4343 <p>The optional constant <tt>align</tt> argument specifies the alignment of the
4344 operation (that is, the alignment of the memory address). A value of 0 or an
4345 omitted <tt>align</tt> argument means that the operation has the preferential
4346 alignment for the target. It is the responsibility of the code emitter to
4347 ensure that the alignment information is correct. Overestimating the
4348 alignment results in undefined behavior. Underestimating the alignment may
4349 produce less efficient code. An alignment of 1 is always safe.</p>
4351 <p>The optional <tt>!nontemporal</tt> metadata must reference a single
4352 metatadata name <index> corresponding to a metadata node with
4353 one <tt>i32</tt> entry of value 1. The existence of
4354 the <tt>!nontemporal</tt> metatadata on the instruction tells the optimizer
4355 and code generator that this load is not expected to be reused in the cache.
4356 The code generator may select special instructions to save cache bandwidth,
4357 such as the <tt>MOVNT</tt> instruction on x86.</p>
4360 <p>The location of memory pointed to is loaded. If the value being loaded is of
4361 scalar type then the number of bytes read does not exceed the minimum number
4362 of bytes needed to hold all bits of the type. For example, loading an
4363 <tt>i24</tt> reads at most three bytes. When loading a value of a type like
4364 <tt>i20</tt> with a size that is not an integral number of bytes, the result
4365 is undefined if the value was not originally written using a store of the
4370 %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
4371 <a href="#i_store">store</a> i32 3, i32* %ptr <i>; yields {void}</i>
4372 %val = load i32* %ptr <i>; yields {i32}:val = i32 3</i>
4377 <!-- _______________________________________________________________________ -->
4378 <div class="doc_subsubsection"> <a name="i_store">'<tt>store</tt>'
4379 Instruction</a> </div>
4381 <div class="doc_text">
4385 store <ty> <value>, <ty>* <pointer>[, align <alignment>][, !nontemporal !<index>] <i>; yields {void}</i>
4386 volatile store <ty> <value>, <ty>* <pointer>[, align <alignment>][, !nontemporal !<index>] <i>; yields {void}</i>
4390 <p>The '<tt>store</tt>' instruction is used to write to memory.</p>
4393 <p>There are two arguments to the '<tt>store</tt>' instruction: a value to store
4394 and an address at which to store it. The type of the
4395 '<tt><pointer></tt>' operand must be a pointer to
4396 the <a href="#t_firstclass">first class</a> type of the
4397 '<tt><value></tt>' operand. If the <tt>store</tt> is marked as
4398 <tt>volatile</tt>, then the optimizer is not allowed to modify the number or
4399 order of execution of this <tt>store</tt> with other <a
4400 href="#volatile">volatile operations</a>.</p>
4402 <p>The optional constant "align" argument specifies the alignment of the
4403 operation (that is, the alignment of the memory address). A value of 0 or an
4404 omitted "align" argument means that the operation has the preferential
4405 alignment for the target. It is the responsibility of the code emitter to
4406 ensure that the alignment information is correct. Overestimating the
4407 alignment results in an undefined behavior. Underestimating the alignment may
4408 produce less efficient code. An alignment of 1 is always safe.</p>
4410 <p>The optional !nontemporal metadata must reference a single metatadata
4411 name <index> corresponding to a metadata node with one i32 entry of
4412 value 1. The existence of the !nontemporal metatadata on the
4413 instruction tells the optimizer and code generator that this load is
4414 not expected to be reused in the cache. The code generator may
4415 select special instructions to save cache bandwidth, such as the
4416 MOVNT instruction on x86.</p>
4420 <p>The contents of memory are updated to contain '<tt><value></tt>' at the
4421 location specified by the '<tt><pointer></tt>' operand. If
4422 '<tt><value></tt>' is of scalar type then the number of bytes written
4423 does not exceed the minimum number of bytes needed to hold all bits of the
4424 type. For example, storing an <tt>i24</tt> writes at most three bytes. When
4425 writing a value of a type like <tt>i20</tt> with a size that is not an
4426 integral number of bytes, it is unspecified what happens to the extra bits
4427 that do not belong to the type, but they will typically be overwritten.</p>
4431 %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
4432 store i32 3, i32* %ptr <i>; yields {void}</i>
4433 %val = <a href="#i_load">load</a> i32* %ptr <i>; yields {i32}:val = i32 3</i>
4438 <!-- _______________________________________________________________________ -->
4439 <div class="doc_subsubsection">
4440 <a name="i_getelementptr">'<tt>getelementptr</tt>' Instruction</a>
4443 <div class="doc_text">
4447 <result> = getelementptr <pty>* <ptrval>{, <ty> <idx>}*
4448 <result> = getelementptr inbounds <pty>* <ptrval>{, <ty> <idx>}*
4452 <p>The '<tt>getelementptr</tt>' instruction is used to get the address of a
4453 subelement of an <a href="#t_aggregate">aggregate</a> data structure.
4454 It performs address calculation only and does not access memory.</p>
4457 <p>The first argument is always a pointer, and forms the basis of the
4458 calculation. The remaining arguments are indices that indicate which of the
4459 elements of the aggregate object are indexed. The interpretation of each
4460 index is dependent on the type being indexed into. The first index always
4461 indexes the pointer value given as the first argument, the second index
4462 indexes a value of the type pointed to (not necessarily the value directly
4463 pointed to, since the first index can be non-zero), etc. The first type
4464 indexed into must be a pointer value, subsequent types can be arrays,
4465 vectors, structs and unions. Note that subsequent types being indexed into
4466 can never be pointers, since that would require loading the pointer before
4467 continuing calculation.</p>
4469 <p>The type of each index argument depends on the type it is indexing into.
4470 When indexing into a (optionally packed) structure or union, only <tt>i32</tt>
4471 integer <b>constants</b> are allowed. When indexing into an array, pointer
4472 or vector, integers of any width are allowed, and they are not required to be
4475 <p>For example, let's consider a C code fragment and how it gets compiled to
4478 <div class="doc_code">
4491 int *foo(struct ST *s) {
4492 return &s[1].Z.B[5][13];
4497 <p>The LLVM code generated by the GCC frontend is:</p>
4499 <div class="doc_code">
4501 %RT = <a href="#namedtypes">type</a> { i8 , [10 x [20 x i32]], i8 }
4502 %ST = <a href="#namedtypes">type</a> { i32, double, %RT }
4504 define i32* @foo(%ST* %s) {
4506 %reg = getelementptr %ST* %s, i32 1, i32 2, i32 1, i32 5, i32 13
4513 <p>In the example above, the first index is indexing into the '<tt>%ST*</tt>'
4514 type, which is a pointer, yielding a '<tt>%ST</tt>' = '<tt>{ i32, double, %RT
4515 }</tt>' type, a structure. The second index indexes into the third element
4516 of the structure, yielding a '<tt>%RT</tt>' = '<tt>{ i8 , [10 x [20 x i32]],
4517 i8 }</tt>' type, another structure. The third index indexes into the second
4518 element of the structure, yielding a '<tt>[10 x [20 x i32]]</tt>' type, an
4519 array. The two dimensions of the array are subscripted into, yielding an
4520 '<tt>i32</tt>' type. The '<tt>getelementptr</tt>' instruction returns a
4521 pointer to this element, thus computing a value of '<tt>i32*</tt>' type.</p>
4523 <p>Note that it is perfectly legal to index partially through a structure,
4524 returning a pointer to an inner element. Because of this, the LLVM code for
4525 the given testcase is equivalent to:</p>
4528 define i32* @foo(%ST* %s) {
4529 %t1 = getelementptr %ST* %s, i32 1 <i>; yields %ST*:%t1</i>
4530 %t2 = getelementptr %ST* %t1, i32 0, i32 2 <i>; yields %RT*:%t2</i>
4531 %t3 = getelementptr %RT* %t2, i32 0, i32 1 <i>; yields [10 x [20 x i32]]*:%t3</i>
4532 %t4 = getelementptr [10 x [20 x i32]]* %t3, i32 0, i32 5 <i>; yields [20 x i32]*:%t4</i>
4533 %t5 = getelementptr [20 x i32]* %t4, i32 0, i32 13 <i>; yields i32*:%t5</i>
4538 <p>If the <tt>inbounds</tt> keyword is present, the result value of the
4539 <tt>getelementptr</tt> is a <a href="#trapvalues">trap value</a> if the
4540 base pointer is not an <i>in bounds</i> address of an allocated object,
4541 or if any of the addresses that would be formed by successive addition of
4542 the offsets implied by the indices to the base address with infinitely
4543 precise arithmetic are not an <i>in bounds</i> address of that allocated
4544 object. The <i>in bounds</i> addresses for an allocated object are all
4545 the addresses that point into the object, plus the address one byte past
4548 <p>If the <tt>inbounds</tt> keyword is not present, the offsets are added to
4549 the base address with silently-wrapping two's complement arithmetic, and
4550 the result value of the <tt>getelementptr</tt> may be outside the object
4551 pointed to by the base pointer. The result value may not necessarily be
4552 used to access memory though, even if it happens to point into allocated
4553 storage. See the <a href="#pointeraliasing">Pointer Aliasing Rules</a>
4554 section for more information.</p>
4556 <p>The getelementptr instruction is often confusing. For some more insight into
4557 how it works, see <a href="GetElementPtr.html">the getelementptr FAQ</a>.</p>
4561 <i>; yields [12 x i8]*:aptr</i>
4562 %aptr = getelementptr {i32, [12 x i8]}* %saptr, i64 0, i32 1
4563 <i>; yields i8*:vptr</i>
4564 %vptr = getelementptr {i32, <2 x i8>}* %svptr, i64 0, i32 1, i32 1
4565 <i>; yields i8*:eptr</i>
4566 %eptr = getelementptr [12 x i8]* %aptr, i64 0, i32 1
4567 <i>; yields i32*:iptr</i>
4568 %iptr = getelementptr [10 x i32]* @arr, i16 0, i16 0
4573 <!-- ======================================================================= -->
4574 <div class="doc_subsection"> <a name="convertops">Conversion Operations</a>
4577 <div class="doc_text">
4579 <p>The instructions in this category are the conversion instructions (casting)
4580 which all take a single operand and a type. They perform various bit
4581 conversions on the operand.</p>
4585 <!-- _______________________________________________________________________ -->
4586 <div class="doc_subsubsection">
4587 <a name="i_trunc">'<tt>trunc .. to</tt>' Instruction</a>
4589 <div class="doc_text">
4593 <result> = trunc <ty> <value> to <ty2> <i>; yields ty2</i>
4597 <p>The '<tt>trunc</tt>' instruction truncates its operand to the
4598 type <tt>ty2</tt>.</p>
4601 <p>The '<tt>trunc</tt>' instruction takes a <tt>value</tt> to trunc, which must
4602 be an <a href="#t_integer">integer</a> type, and a type that specifies the
4603 size and type of the result, which must be
4604 an <a href="#t_integer">integer</a> type. The bit size of <tt>value</tt> must
4605 be larger than the bit size of <tt>ty2</tt>. Equal sized types are not
4609 <p>The '<tt>trunc</tt>' instruction truncates the high order bits
4610 in <tt>value</tt> and converts the remaining bits to <tt>ty2</tt>. Since the
4611 source size must be larger than the destination size, <tt>trunc</tt> cannot
4612 be a <i>no-op cast</i>. It will always truncate bits.</p>
4616 %X = trunc i32 257 to i8 <i>; yields i8:1</i>
4617 %Y = trunc i32 123 to i1 <i>; yields i1:true</i>
4618 %Z = trunc i32 122 to i1 <i>; yields i1:false</i>
4623 <!-- _______________________________________________________________________ -->
4624 <div class="doc_subsubsection">
4625 <a name="i_zext">'<tt>zext .. to</tt>' Instruction</a>
4627 <div class="doc_text">
4631 <result> = zext <ty> <value> to <ty2> <i>; yields ty2</i>
4635 <p>The '<tt>zext</tt>' instruction zero extends its operand to type
4640 <p>The '<tt>zext</tt>' instruction takes a value to cast, which must be of
4641 <a href="#t_integer">integer</a> type, and a type to cast it to, which must
4642 also be of <a href="#t_integer">integer</a> type. The bit size of the
4643 <tt>value</tt> must be smaller than the bit size of the destination type,
4647 <p>The <tt>zext</tt> fills the high order bits of the <tt>value</tt> with zero
4648 bits until it reaches the size of the destination type, <tt>ty2</tt>.</p>
4650 <p>When zero extending from i1, the result will always be either 0 or 1.</p>
4654 %X = zext i32 257 to i64 <i>; yields i64:257</i>
4655 %Y = zext i1 true to i32 <i>; yields i32:1</i>
4660 <!-- _______________________________________________________________________ -->
4661 <div class="doc_subsubsection">
4662 <a name="i_sext">'<tt>sext .. to</tt>' Instruction</a>
4664 <div class="doc_text">
4668 <result> = sext <ty> <value> to <ty2> <i>; yields ty2</i>
4672 <p>The '<tt>sext</tt>' sign extends <tt>value</tt> to the type <tt>ty2</tt>.</p>
4675 <p>The '<tt>sext</tt>' instruction takes a value to cast, which must be of
4676 <a href="#t_integer">integer</a> type, and a type to cast it to, which must
4677 also be of <a href="#t_integer">integer</a> type. The bit size of the
4678 <tt>value</tt> must be smaller than the bit size of the destination type,
4682 <p>The '<tt>sext</tt>' instruction performs a sign extension by copying the sign
4683 bit (highest order bit) of the <tt>value</tt> until it reaches the bit size
4684 of the type <tt>ty2</tt>.</p>
4686 <p>When sign extending from i1, the extension always results in -1 or 0.</p>
4690 %X = sext i8 -1 to i16 <i>; yields i16 :65535</i>
4691 %Y = sext i1 true to i32 <i>; yields i32:-1</i>
4696 <!-- _______________________________________________________________________ -->
4697 <div class="doc_subsubsection">
4698 <a name="i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a>
4701 <div class="doc_text">
4705 <result> = fptrunc <ty> <value> to <ty2> <i>; yields ty2</i>
4709 <p>The '<tt>fptrunc</tt>' instruction truncates <tt>value</tt> to type
4713 <p>The '<tt>fptrunc</tt>' instruction takes a <a href="#t_floating">floating
4714 point</a> value to cast and a <a href="#t_floating">floating point</a> type
4715 to cast it to. The size of <tt>value</tt> must be larger than the size of
4716 <tt>ty2</tt>. This implies that <tt>fptrunc</tt> cannot be used to make a
4717 <i>no-op cast</i>.</p>
4720 <p>The '<tt>fptrunc</tt>' instruction truncates a <tt>value</tt> from a larger
4721 <a href="#t_floating">floating point</a> type to a smaller
4722 <a href="#t_floating">floating point</a> type. If the value cannot fit
4723 within the destination type, <tt>ty2</tt>, then the results are
4728 %X = fptrunc double 123.0 to float <i>; yields float:123.0</i>
4729 %Y = fptrunc double 1.0E+300 to float <i>; yields undefined</i>
4734 <!-- _______________________________________________________________________ -->
4735 <div class="doc_subsubsection">
4736 <a name="i_fpext">'<tt>fpext .. to</tt>' Instruction</a>
4738 <div class="doc_text">
4742 <result> = fpext <ty> <value> to <ty2> <i>; yields ty2</i>
4746 <p>The '<tt>fpext</tt>' extends a floating point <tt>value</tt> to a larger
4747 floating point value.</p>
4750 <p>The '<tt>fpext</tt>' instruction takes a
4751 <a href="#t_floating">floating point</a> <tt>value</tt> to cast, and
4752 a <a href="#t_floating">floating point</a> type to cast it to. The source
4753 type must be smaller than the destination type.</p>
4756 <p>The '<tt>fpext</tt>' instruction extends the <tt>value</tt> from a smaller
4757 <a href="#t_floating">floating point</a> type to a larger
4758 <a href="#t_floating">floating point</a> type. The <tt>fpext</tt> cannot be
4759 used to make a <i>no-op cast</i> because it always changes bits. Use
4760 <tt>bitcast</tt> to make a <i>no-op cast</i> for a floating point cast.</p>
4764 %X = fpext float 3.1415 to double <i>; yields double:3.1415</i>
4765 %Y = fpext float 1.0 to float <i>; yields float:1.0 (no-op)</i>
4770 <!-- _______________________________________________________________________ -->
4771 <div class="doc_subsubsection">
4772 <a name="i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a>
4774 <div class="doc_text">
4778 <result> = fptoui <ty> <value> to <ty2> <i>; yields ty2</i>
4782 <p>The '<tt>fptoui</tt>' converts a floating point <tt>value</tt> to its
4783 unsigned integer equivalent of type <tt>ty2</tt>.</p>
4786 <p>The '<tt>fptoui</tt>' instruction takes a value to cast, which must be a
4787 scalar or vector <a href="#t_floating">floating point</a> value, and a type
4788 to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
4789 type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
4790 vector integer type with the same number of elements as <tt>ty</tt></p>
4793 <p>The '<tt>fptoui</tt>' instruction converts its
4794 <a href="#t_floating">floating point</a> operand into the nearest (rounding
4795 towards zero) unsigned integer value. If the value cannot fit
4796 in <tt>ty2</tt>, the results are undefined.</p>
4800 %X = fptoui double 123.0 to i32 <i>; yields i32:123</i>
4801 %Y = fptoui float 1.0E+300 to i1 <i>; yields undefined:1</i>
4802 %Z = fptoui float 1.04E+17 to i8 <i>; yields undefined:1</i>
4807 <!-- _______________________________________________________________________ -->
4808 <div class="doc_subsubsection">
4809 <a name="i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a>
4811 <div class="doc_text">
4815 <result> = fptosi <ty> <value> to <ty2> <i>; yields ty2</i>
4819 <p>The '<tt>fptosi</tt>' instruction converts
4820 <a href="#t_floating">floating point</a> <tt>value</tt> to
4821 type <tt>ty2</tt>.</p>
4824 <p>The '<tt>fptosi</tt>' instruction takes a value to cast, which must be a
4825 scalar or vector <a href="#t_floating">floating point</a> value, and a type
4826 to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
4827 type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
4828 vector integer type with the same number of elements as <tt>ty</tt></p>
4831 <p>The '<tt>fptosi</tt>' instruction converts its
4832 <a href="#t_floating">floating point</a> operand into the nearest (rounding
4833 towards zero) signed integer value. If the value cannot fit in <tt>ty2</tt>,
4834 the results are undefined.</p>
4838 %X = fptosi double -123.0 to i32 <i>; yields i32:-123</i>
4839 %Y = fptosi float 1.0E-247 to i1 <i>; yields undefined:1</i>
4840 %Z = fptosi float 1.04E+17 to i8 <i>; yields undefined:1</i>
4845 <!-- _______________________________________________________________________ -->
4846 <div class="doc_subsubsection">
4847 <a name="i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a>
4849 <div class="doc_text">
4853 <result> = uitofp <ty> <value> to <ty2> <i>; yields ty2</i>
4857 <p>The '<tt>uitofp</tt>' instruction regards <tt>value</tt> as an unsigned
4858 integer and converts that value to the <tt>ty2</tt> type.</p>
4861 <p>The '<tt>uitofp</tt>' instruction takes a value to cast, which must be a
4862 scalar or vector <a href="#t_integer">integer</a> value, and a type to cast
4863 it to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
4864 type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
4865 floating point type with the same number of elements as <tt>ty</tt></p>
4868 <p>The '<tt>uitofp</tt>' instruction interprets its operand as an unsigned
4869 integer quantity and converts it to the corresponding floating point
4870 value. If the value cannot fit in the floating point value, the results are
4875 %X = uitofp i32 257 to float <i>; yields float:257.0</i>
4876 %Y = uitofp i8 -1 to double <i>; yields double:255.0</i>
4881 <!-- _______________________________________________________________________ -->
4882 <div class="doc_subsubsection">
4883 <a name="i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a>
4885 <div class="doc_text">
4889 <result> = sitofp <ty> <value> to <ty2> <i>; yields ty2</i>
4893 <p>The '<tt>sitofp</tt>' instruction regards <tt>value</tt> as a signed integer
4894 and converts that value to the <tt>ty2</tt> type.</p>
4897 <p>The '<tt>sitofp</tt>' instruction takes a value to cast, which must be a
4898 scalar or vector <a href="#t_integer">integer</a> value, and a type to cast
4899 it to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
4900 type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
4901 floating point type with the same number of elements as <tt>ty</tt></p>
4904 <p>The '<tt>sitofp</tt>' instruction interprets its operand as a signed integer
4905 quantity and converts it to the corresponding floating point value. If the
4906 value cannot fit in the floating point value, the results are undefined.</p>
4910 %X = sitofp i32 257 to float <i>; yields float:257.0</i>
4911 %Y = sitofp i8 -1 to double <i>; yields double:-1.0</i>
4916 <!-- _______________________________________________________________________ -->
4917 <div class="doc_subsubsection">
4918 <a name="i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a>
4920 <div class="doc_text">
4924 <result> = ptrtoint <ty> <value> to <ty2> <i>; yields ty2</i>
4928 <p>The '<tt>ptrtoint</tt>' instruction converts the pointer <tt>value</tt> to
4929 the integer type <tt>ty2</tt>.</p>
4932 <p>The '<tt>ptrtoint</tt>' instruction takes a <tt>value</tt> to cast, which
4933 must be a <a href="#t_pointer">pointer</a> value, and a type to cast it to
4934 <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a> type.</p>
4937 <p>The '<tt>ptrtoint</tt>' instruction converts <tt>value</tt> to integer type
4938 <tt>ty2</tt> by interpreting the pointer value as an integer and either
4939 truncating or zero extending that value to the size of the integer type. If
4940 <tt>value</tt> is smaller than <tt>ty2</tt> then a zero extension is done. If
4941 <tt>value</tt> is larger than <tt>ty2</tt> then a truncation is done. If they
4942 are the same size, then nothing is done (<i>no-op cast</i>) other than a type
4947 %X = ptrtoint i32* %X to i8 <i>; yields truncation on 32-bit architecture</i>
4948 %Y = ptrtoint i32* %x to i64 <i>; yields zero extension on 32-bit architecture</i>
4953 <!-- _______________________________________________________________________ -->
4954 <div class="doc_subsubsection">
4955 <a name="i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a>
4957 <div class="doc_text">
4961 <result> = inttoptr <ty> <value> to <ty2> <i>; yields ty2</i>
4965 <p>The '<tt>inttoptr</tt>' instruction converts an integer <tt>value</tt> to a
4966 pointer type, <tt>ty2</tt>.</p>
4969 <p>The '<tt>inttoptr</tt>' instruction takes an <a href="#t_integer">integer</a>
4970 value to cast, and a type to cast it to, which must be a
4971 <a href="#t_pointer">pointer</a> type.</p>
4974 <p>The '<tt>inttoptr</tt>' instruction converts <tt>value</tt> to type
4975 <tt>ty2</tt> by applying either a zero extension or a truncation depending on
4976 the size of the integer <tt>value</tt>. If <tt>value</tt> is larger than the
4977 size of a pointer then a truncation is done. If <tt>value</tt> is smaller
4978 than the size of a pointer then a zero extension is done. If they are the
4979 same size, nothing is done (<i>no-op cast</i>).</p>
4983 %X = inttoptr i32 255 to i32* <i>; yields zero extension on 64-bit architecture</i>
4984 %Y = inttoptr i32 255 to i32* <i>; yields no-op on 32-bit architecture</i>
4985 %Z = inttoptr i64 0 to i32* <i>; yields truncation on 32-bit architecture</i>
4990 <!-- _______________________________________________________________________ -->
4991 <div class="doc_subsubsection">
4992 <a name="i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a>
4994 <div class="doc_text">
4998 <result> = bitcast <ty> <value> to <ty2> <i>; yields ty2</i>
5002 <p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
5003 <tt>ty2</tt> without changing any bits.</p>
5006 <p>The '<tt>bitcast</tt>' instruction takes a value to cast, which must be a
5007 non-aggregate first class value, and a type to cast it to, which must also be
5008 a non-aggregate <a href="#t_firstclass">first class</a> type. The bit sizes
5009 of <tt>value</tt> and the destination type, <tt>ty2</tt>, must be
5010 identical. If the source type is a pointer, the destination type must also be
5011 a pointer. This instruction supports bitwise conversion of vectors to
5012 integers and to vectors of other types (as long as they have the same
5016 <p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
5017 <tt>ty2</tt>. It is always a <i>no-op cast</i> because no bits change with
5018 this conversion. The conversion is done as if the <tt>value</tt> had been
5019 stored to memory and read back as type <tt>ty2</tt>. Pointer types may only
5020 be converted to other pointer types with this instruction. To convert
5021 pointers to other types, use the <a href="#i_inttoptr">inttoptr</a> or
5022 <a href="#i_ptrtoint">ptrtoint</a> instructions first.</p>
5026 %X = bitcast i8 255 to i8 <i>; yields i8 :-1</i>
5027 %Y = bitcast i32* %x to sint* <i>; yields sint*:%x</i>
5028 %Z = bitcast <2 x int> %V to i64; <i>; yields i64: %V</i>
5033 <!-- ======================================================================= -->
5034 <div class="doc_subsection"> <a name="otherops">Other Operations</a> </div>
5036 <div class="doc_text">
5038 <p>The instructions in this category are the "miscellaneous" instructions, which
5039 defy better classification.</p>
5043 <!-- _______________________________________________________________________ -->
5044 <div class="doc_subsubsection"><a name="i_icmp">'<tt>icmp</tt>' Instruction</a>
5047 <div class="doc_text">
5051 <result> = icmp <cond> <ty> <op1>, <op2> <i>; yields {i1} or {<N x i1>}:result</i>
5055 <p>The '<tt>icmp</tt>' instruction returns a boolean value or a vector of
5056 boolean values based on comparison of its two integer, integer vector, or
5057 pointer operands.</p>
5060 <p>The '<tt>icmp</tt>' instruction takes three operands. The first operand is
5061 the condition code indicating the kind of comparison to perform. It is not a
5062 value, just a keyword. The possible condition code are:</p>
5065 <li><tt>eq</tt>: equal</li>
5066 <li><tt>ne</tt>: not equal </li>
5067 <li><tt>ugt</tt>: unsigned greater than</li>
5068 <li><tt>uge</tt>: unsigned greater or equal</li>
5069 <li><tt>ult</tt>: unsigned less than</li>
5070 <li><tt>ule</tt>: unsigned less or equal</li>
5071 <li><tt>sgt</tt>: signed greater than</li>
5072 <li><tt>sge</tt>: signed greater or equal</li>
5073 <li><tt>slt</tt>: signed less than</li>
5074 <li><tt>sle</tt>: signed less or equal</li>
5077 <p>The remaining two arguments must be <a href="#t_integer">integer</a> or
5078 <a href="#t_pointer">pointer</a> or integer <a href="#t_vector">vector</a>
5079 typed. They must also be identical types.</p>
5082 <p>The '<tt>icmp</tt>' compares <tt>op1</tt> and <tt>op2</tt> according to the
5083 condition code given as <tt>cond</tt>. The comparison performed always yields
5084 either an <a href="#t_integer"><tt>i1</tt></a> or vector of <tt>i1</tt>
5085 result, as follows:</p>
5088 <li><tt>eq</tt>: yields <tt>true</tt> if the operands are equal,
5089 <tt>false</tt> otherwise. No sign interpretation is necessary or
5092 <li><tt>ne</tt>: yields <tt>true</tt> if the operands are unequal,
5093 <tt>false</tt> otherwise. No sign interpretation is necessary or
5096 <li><tt>ugt</tt>: interprets the operands as unsigned values and yields
5097 <tt>true</tt> if <tt>op1</tt> is greater than <tt>op2</tt>.</li>
5099 <li><tt>uge</tt>: interprets the operands as unsigned values and yields
5100 <tt>true</tt> if <tt>op1</tt> is greater than or equal
5101 to <tt>op2</tt>.</li>
5103 <li><tt>ult</tt>: interprets the operands as unsigned values and yields
5104 <tt>true</tt> if <tt>op1</tt> is less than <tt>op2</tt>.</li>
5106 <li><tt>ule</tt>: interprets the operands as unsigned values and yields
5107 <tt>true</tt> if <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
5109 <li><tt>sgt</tt>: interprets the operands as signed values and yields
5110 <tt>true</tt> if <tt>op1</tt> is greater than <tt>op2</tt>.</li>
5112 <li><tt>sge</tt>: interprets the operands as signed values and yields
5113 <tt>true</tt> if <tt>op1</tt> is greater than or equal
5114 to <tt>op2</tt>.</li>
5116 <li><tt>slt</tt>: interprets the operands as signed values and yields
5117 <tt>true</tt> if <tt>op1</tt> is less than <tt>op2</tt>.</li>
5119 <li><tt>sle</tt>: interprets the operands as signed values and yields
5120 <tt>true</tt> if <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
5123 <p>If the operands are <a href="#t_pointer">pointer</a> typed, the pointer
5124 values are compared as if they were integers.</p>
5126 <p>If the operands are integer vectors, then they are compared element by
5127 element. The result is an <tt>i1</tt> vector with the same number of elements
5128 as the values being compared. Otherwise, the result is an <tt>i1</tt>.</p>
5132 <result> = icmp eq i32 4, 5 <i>; yields: result=false</i>
5133 <result> = icmp ne float* %X, %X <i>; yields: result=false</i>
5134 <result> = icmp ult i16 4, 5 <i>; yields: result=true</i>
5135 <result> = icmp sgt i16 4, 5 <i>; yields: result=false</i>
5136 <result> = icmp ule i16 -4, 5 <i>; yields: result=false</i>
5137 <result> = icmp sge i16 4, 5 <i>; yields: result=false</i>
5140 <p>Note that the code generator does not yet support vector types with
5141 the <tt>icmp</tt> instruction.</p>
5145 <!-- _______________________________________________________________________ -->
5146 <div class="doc_subsubsection"><a name="i_fcmp">'<tt>fcmp</tt>' Instruction</a>
5149 <div class="doc_text">
5153 <result> = fcmp <cond> <ty> <op1>, <op2> <i>; yields {i1} or {<N x i1>}:result</i>
5157 <p>The '<tt>fcmp</tt>' instruction returns a boolean value or vector of boolean
5158 values based on comparison of its operands.</p>
5160 <p>If the operands are floating point scalars, then the result type is a boolean
5161 (<a href="#t_integer"><tt>i1</tt></a>).</p>
5163 <p>If the operands are floating point vectors, then the result type is a vector
5164 of boolean with the same number of elements as the operands being
5168 <p>The '<tt>fcmp</tt>' instruction takes three operands. The first operand is
5169 the condition code indicating the kind of comparison to perform. It is not a
5170 value, just a keyword. The possible condition code are:</p>
5173 <li><tt>false</tt>: no comparison, always returns false</li>
5174 <li><tt>oeq</tt>: ordered and equal</li>
5175 <li><tt>ogt</tt>: ordered and greater than </li>
5176 <li><tt>oge</tt>: ordered and greater than or equal</li>
5177 <li><tt>olt</tt>: ordered and less than </li>
5178 <li><tt>ole</tt>: ordered and less than or equal</li>
5179 <li><tt>one</tt>: ordered and not equal</li>
5180 <li><tt>ord</tt>: ordered (no nans)</li>
5181 <li><tt>ueq</tt>: unordered or equal</li>
5182 <li><tt>ugt</tt>: unordered or greater than </li>
5183 <li><tt>uge</tt>: unordered or greater than or equal</li>
5184 <li><tt>ult</tt>: unordered or less than </li>
5185 <li><tt>ule</tt>: unordered or less than or equal</li>
5186 <li><tt>une</tt>: unordered or not equal</li>
5187 <li><tt>uno</tt>: unordered (either nans)</li>
5188 <li><tt>true</tt>: no comparison, always returns true</li>
5191 <p><i>Ordered</i> means that neither operand is a QNAN while
5192 <i>unordered</i> means that either operand may be a QNAN.</p>
5194 <p>Each of <tt>val1</tt> and <tt>val2</tt> arguments must be either
5195 a <a href="#t_floating">floating point</a> type or
5196 a <a href="#t_vector">vector</a> of floating point type. They must have
5197 identical types.</p>
5200 <p>The '<tt>fcmp</tt>' instruction compares <tt>op1</tt> and <tt>op2</tt>
5201 according to the condition code given as <tt>cond</tt>. If the operands are
5202 vectors, then the vectors are compared element by element. Each comparison
5203 performed always yields an <a href="#t_integer">i1</a> result, as
5207 <li><tt>false</tt>: always yields <tt>false</tt>, regardless of operands.</li>
5209 <li><tt>oeq</tt>: yields <tt>true</tt> if both operands are not a QNAN and
5210 <tt>op1</tt> is equal to <tt>op2</tt>.</li>
5212 <li><tt>ogt</tt>: yields <tt>true</tt> if both operands are not a QNAN and
5213 <tt>op1</tt> is greater than <tt>op2</tt>.</li>
5215 <li><tt>oge</tt>: yields <tt>true</tt> if both operands are not a QNAN and
5216 <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
5218 <li><tt>olt</tt>: yields <tt>true</tt> if both operands are not a QNAN and
5219 <tt>op1</tt> is less than <tt>op2</tt>.</li>
5221 <li><tt>ole</tt>: yields <tt>true</tt> if both operands are not a QNAN and
5222 <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
5224 <li><tt>one</tt>: yields <tt>true</tt> if both operands are not a QNAN and
5225 <tt>op1</tt> is not equal to <tt>op2</tt>.</li>
5227 <li><tt>ord</tt>: yields <tt>true</tt> if both operands are not a QNAN.</li>
5229 <li><tt>ueq</tt>: yields <tt>true</tt> if either operand is a QNAN or
5230 <tt>op1</tt> is equal to <tt>op2</tt>.</li>
5232 <li><tt>ugt</tt>: yields <tt>true</tt> if either operand is a QNAN or
5233 <tt>op1</tt> is greater than <tt>op2</tt>.</li>
5235 <li><tt>uge</tt>: yields <tt>true</tt> if either operand is a QNAN or
5236 <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
5238 <li><tt>ult</tt>: yields <tt>true</tt> if either operand is a QNAN or
5239 <tt>op1</tt> is less than <tt>op2</tt>.</li>
5241 <li><tt>ule</tt>: yields <tt>true</tt> if either operand is a QNAN or
5242 <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
5244 <li><tt>une</tt>: yields <tt>true</tt> if either operand is a QNAN or
5245 <tt>op1</tt> is not equal to <tt>op2</tt>.</li>
5247 <li><tt>uno</tt>: yields <tt>true</tt> if either operand is a QNAN.</li>
5249 <li><tt>true</tt>: always yields <tt>true</tt>, regardless of operands.</li>
5254 <result> = fcmp oeq float 4.0, 5.0 <i>; yields: result=false</i>
5255 <result> = fcmp one float 4.0, 5.0 <i>; yields: result=true</i>
5256 <result> = fcmp olt float 4.0, 5.0 <i>; yields: result=true</i>
5257 <result> = fcmp ueq double 1.0, 2.0 <i>; yields: result=false</i>
5260 <p>Note that the code generator does not yet support vector types with
5261 the <tt>fcmp</tt> instruction.</p>
5265 <!-- _______________________________________________________________________ -->
5266 <div class="doc_subsubsection">
5267 <a name="i_phi">'<tt>phi</tt>' Instruction</a>
5270 <div class="doc_text">
5274 <result> = phi <ty> [ <val0>, <label0>], ...
5278 <p>The '<tt>phi</tt>' instruction is used to implement the φ node in the
5279 SSA graph representing the function.</p>
5282 <p>The type of the incoming values is specified with the first type field. After
5283 this, the '<tt>phi</tt>' instruction takes a list of pairs as arguments, with
5284 one pair for each predecessor basic block of the current block. Only values
5285 of <a href="#t_firstclass">first class</a> type may be used as the value
5286 arguments to the PHI node. Only labels may be used as the label
5289 <p>There must be no non-phi instructions between the start of a basic block and
5290 the PHI instructions: i.e. PHI instructions must be first in a basic
5293 <p>For the purposes of the SSA form, the use of each incoming value is deemed to
5294 occur on the edge from the corresponding predecessor block to the current
5295 block (but after any definition of an '<tt>invoke</tt>' instruction's return
5296 value on the same edge).</p>
5299 <p>At runtime, the '<tt>phi</tt>' instruction logically takes on the value
5300 specified by the pair corresponding to the predecessor basic block that
5301 executed just prior to the current block.</p>
5305 Loop: ; Infinite loop that counts from 0 on up...
5306 %indvar = phi i32 [ 0, %LoopHeader ], [ %nextindvar, %Loop ]
5307 %nextindvar = add i32 %indvar, 1
5313 <!-- _______________________________________________________________________ -->
5314 <div class="doc_subsubsection">
5315 <a name="i_select">'<tt>select</tt>' Instruction</a>
5318 <div class="doc_text">
5322 <result> = select <i>selty</i> <cond>, <ty> <val1>, <ty> <val2> <i>; yields ty</i>
5324 <i>selty</i> is either i1 or {<N x i1>}
5328 <p>The '<tt>select</tt>' instruction is used to choose one value based on a
5329 condition, without branching.</p>
5333 <p>The '<tt>select</tt>' instruction requires an 'i1' value or a vector of 'i1'
5334 values indicating the condition, and two values of the
5335 same <a href="#t_firstclass">first class</a> type. If the val1/val2 are
5336 vectors and the condition is a scalar, then entire vectors are selected, not
5337 individual elements.</p>
5340 <p>If the condition is an i1 and it evaluates to 1, the instruction returns the
5341 first value argument; otherwise, it returns the second value argument.</p>
5343 <p>If the condition is a vector of i1, then the value arguments must be vectors
5344 of the same size, and the selection is done element by element.</p>
5348 %X = select i1 true, i8 17, i8 42 <i>; yields i8:17</i>
5351 <p>Note that the code generator does not yet support conditions
5352 with vector type.</p>
5356 <!-- _______________________________________________________________________ -->
5357 <div class="doc_subsubsection">
5358 <a name="i_call">'<tt>call</tt>' Instruction</a>
5361 <div class="doc_text">
5365 <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>]
5369 <p>The '<tt>call</tt>' instruction represents a simple function call.</p>
5372 <p>This instruction requires several arguments:</p>
5375 <li>The optional "tail" marker indicates that the callee function does not
5376 access any allocas or varargs in the caller. Note that calls may be
5377 marked "tail" even if they do not occur before
5378 a <a href="#i_ret"><tt>ret</tt></a> instruction. If the "tail" marker is
5379 present, the function call is eligible for tail call optimization,
5380 but <a href="CodeGenerator.html#tailcallopt">might not in fact be
5381 optimized into a jump</a>. The code generator may optimize calls marked
5382 "tail" with either 1) automatic <a href="CodeGenerator.html#sibcallopt">
5383 sibling call optimization</a> when the caller and callee have
5384 matching signatures, or 2) forced tail call optimization when the
5385 following extra requirements are met:
5387 <li>Caller and callee both have the calling
5388 convention <tt>fastcc</tt>.</li>
5389 <li>The call is in tail position (ret immediately follows call and ret
5390 uses value of call or is void).</li>
5391 <li>Option <tt>-tailcallopt</tt> is enabled,
5392 or <code>llvm::GuaranteedTailCallOpt</code> is <code>true</code>.</li>
5393 <li><a href="CodeGenerator.html#tailcallopt">Platform specific
5394 constraints are met.</a></li>
5398 <li>The optional "cconv" marker indicates which <a href="#callingconv">calling
5399 convention</a> the call should use. If none is specified, the call
5400 defaults to using C calling conventions. The calling convention of the
5401 call must match the calling convention of the target function, or else the
5402 behavior is undefined.</li>
5404 <li>The optional <a href="#paramattrs">Parameter Attributes</a> list for
5405 return values. Only '<tt>zeroext</tt>', '<tt>signext</tt>', and
5406 '<tt>inreg</tt>' attributes are valid here.</li>
5408 <li>'<tt>ty</tt>': the type of the call instruction itself which is also the
5409 type of the return value. Functions that return no value are marked
5410 <tt><a href="#t_void">void</a></tt>.</li>
5412 <li>'<tt>fnty</tt>': shall be the signature of the pointer to function value
5413 being invoked. The argument types must match the types implied by this
5414 signature. This type can be omitted if the function is not varargs and if
5415 the function type does not return a pointer to a function.</li>
5417 <li>'<tt>fnptrval</tt>': An LLVM value containing a pointer to a function to
5418 be invoked. In most cases, this is a direct function invocation, but
5419 indirect <tt>call</tt>s are just as possible, calling an arbitrary pointer
5420 to function value.</li>
5422 <li>'<tt>function args</tt>': argument list whose types match the function
5423 signature argument types and parameter attributes. All arguments must be
5424 of <a href="#t_firstclass">first class</a> type. If the function
5425 signature indicates the function accepts a variable number of arguments,
5426 the extra arguments can be specified.</li>
5428 <li>The optional <a href="#fnattrs">function attributes</a> list. Only
5429 '<tt>noreturn</tt>', '<tt>nounwind</tt>', '<tt>readonly</tt>' and
5430 '<tt>readnone</tt>' attributes are valid here.</li>
5434 <p>The '<tt>call</tt>' instruction is used to cause control flow to transfer to
5435 a specified function, with its incoming arguments bound to the specified
5436 values. Upon a '<tt><a href="#i_ret">ret</a></tt>' instruction in the called
5437 function, control flow continues with the instruction after the function
5438 call, and the return value of the function is bound to the result
5443 %retval = call i32 @test(i32 %argc)
5444 call i32 (i8*, ...)* @printf(i8* %msg, i32 12, i8 42) <i>; yields i32</i>
5445 %X = tail call i32 @foo() <i>; yields i32</i>
5446 %Y = tail call <a href="#callingconv">fastcc</a> i32 @foo() <i>; yields i32</i>
5447 call void %foo(i8 97 signext)
5449 %struct.A = type { i32, i8 }
5450 %r = call %struct.A @foo() <i>; yields { 32, i8 }</i>
5451 %gr = extractvalue %struct.A %r, 0 <i>; yields i32</i>
5452 %gr1 = extractvalue %struct.A %r, 1 <i>; yields i8</i>
5453 %Z = call void @foo() noreturn <i>; indicates that %foo never returns normally</i>
5454 %ZZ = call zeroext i32 @bar() <i>; Return value is %zero extended</i>
5457 <p>llvm treats calls to some functions with names and arguments that match the
5458 standard C99 library as being the C99 library functions, and may perform
5459 optimizations or generate code for them under that assumption. This is
5460 something we'd like to change in the future to provide better support for
5461 freestanding environments and non-C-based languages.</p>
5465 <!-- _______________________________________________________________________ -->
5466 <div class="doc_subsubsection">
5467 <a name="i_va_arg">'<tt>va_arg</tt>' Instruction</a>
5470 <div class="doc_text">
5474 <resultval> = va_arg <va_list*> <arglist>, <argty>
5478 <p>The '<tt>va_arg</tt>' instruction is used to access arguments passed through
5479 the "variable argument" area of a function call. It is used to implement the
5480 <tt>va_arg</tt> macro in C.</p>
5483 <p>This instruction takes a <tt>va_list*</tt> value and the type of the
5484 argument. It returns a value of the specified argument type and increments
5485 the <tt>va_list</tt> to point to the next argument. The actual type
5486 of <tt>va_list</tt> is target specific.</p>
5489 <p>The '<tt>va_arg</tt>' instruction loads an argument of the specified type
5490 from the specified <tt>va_list</tt> and causes the <tt>va_list</tt> to point
5491 to the next argument. For more information, see the variable argument
5492 handling <a href="#int_varargs">Intrinsic Functions</a>.</p>
5494 <p>It is legal for this instruction to be called in a function which does not
5495 take a variable number of arguments, for example, the <tt>vfprintf</tt>
5498 <p><tt>va_arg</tt> is an LLVM instruction instead of
5499 an <a href="#intrinsics">intrinsic function</a> because it takes a type as an
5503 <p>See the <a href="#int_varargs">variable argument processing</a> section.</p>
5505 <p>Note that the code generator does not yet fully support va_arg on many
5506 targets. Also, it does not currently support va_arg with aggregate types on
5511 <!-- *********************************************************************** -->
5512 <div class="doc_section"> <a name="intrinsics">Intrinsic Functions</a> </div>
5513 <!-- *********************************************************************** -->
5515 <div class="doc_text">
5517 <p>LLVM supports the notion of an "intrinsic function". These functions have
5518 well known names and semantics and are required to follow certain
5519 restrictions. Overall, these intrinsics represent an extension mechanism for
5520 the LLVM language that does not require changing all of the transformations
5521 in LLVM when adding to the language (or the bitcode reader/writer, the
5522 parser, etc...).</p>
5524 <p>Intrinsic function names must all start with an "<tt>llvm.</tt>" prefix. This
5525 prefix is reserved in LLVM for intrinsic names; thus, function names may not
5526 begin with this prefix. Intrinsic functions must always be external
5527 functions: you cannot define the body of intrinsic functions. Intrinsic
5528 functions may only be used in call or invoke instructions: it is illegal to
5529 take the address of an intrinsic function. Additionally, because intrinsic
5530 functions are part of the LLVM language, it is required if any are added that
5531 they be documented here.</p>
5533 <p>Some intrinsic functions can be overloaded, i.e., the intrinsic represents a
5534 family of functions that perform the same operation but on different data
5535 types. Because LLVM can represent over 8 million different integer types,
5536 overloading is used commonly to allow an intrinsic function to operate on any
5537 integer type. One or more of the argument types or the result type can be
5538 overloaded to accept any integer type. Argument types may also be defined as
5539 exactly matching a previous argument's type or the result type. This allows
5540 an intrinsic function which accepts multiple arguments, but needs all of them
5541 to be of the same type, to only be overloaded with respect to a single
5542 argument or the result.</p>
5544 <p>Overloaded intrinsics will have the names of its overloaded argument types
5545 encoded into its function name, each preceded by a period. Only those types
5546 which are overloaded result in a name suffix. Arguments whose type is matched
5547 against another type do not. For example, the <tt>llvm.ctpop</tt> function
5548 can take an integer of any width and returns an integer of exactly the same
5549 integer width. This leads to a family of functions such as
5550 <tt>i8 @llvm.ctpop.i8(i8 %val)</tt> and <tt>i29 @llvm.ctpop.i29(i29
5551 %val)</tt>. Only one type, the return type, is overloaded, and only one type
5552 suffix is required. Because the argument's type is matched against the return
5553 type, it does not require its own name suffix.</p>
5555 <p>To learn how to add an intrinsic function, please see the
5556 <a href="ExtendingLLVM.html">Extending LLVM Guide</a>.</p>
5560 <!-- ======================================================================= -->
5561 <div class="doc_subsection">
5562 <a name="int_varargs">Variable Argument Handling Intrinsics</a>
5565 <div class="doc_text">
5567 <p>Variable argument support is defined in LLVM with
5568 the <a href="#i_va_arg"><tt>va_arg</tt></a> instruction and these three
5569 intrinsic functions. These functions are related to the similarly named
5570 macros defined in the <tt><stdarg.h></tt> header file.</p>
5572 <p>All of these functions operate on arguments that use a target-specific value
5573 type "<tt>va_list</tt>". The LLVM assembly language reference manual does
5574 not define what this type is, so all transformations should be prepared to
5575 handle these functions regardless of the type used.</p>
5577 <p>This example shows how the <a href="#i_va_arg"><tt>va_arg</tt></a>
5578 instruction and the variable argument handling intrinsic functions are
5581 <div class="doc_code">
5583 define i32 @test(i32 %X, ...) {
5584 ; Initialize variable argument processing
5586 %ap2 = bitcast i8** %ap to i8*
5587 call void @llvm.va_start(i8* %ap2)
5589 ; Read a single integer argument
5590 %tmp = va_arg i8** %ap, i32
5592 ; Demonstrate usage of llvm.va_copy and llvm.va_end
5594 %aq2 = bitcast i8** %aq to i8*
5595 call void @llvm.va_copy(i8* %aq2, i8* %ap2)
5596 call void @llvm.va_end(i8* %aq2)
5598 ; Stop processing of arguments.
5599 call void @llvm.va_end(i8* %ap2)
5603 declare void @llvm.va_start(i8*)
5604 declare void @llvm.va_copy(i8*, i8*)
5605 declare void @llvm.va_end(i8*)
5611 <!-- _______________________________________________________________________ -->
5612 <div class="doc_subsubsection">
5613 <a name="int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a>
5617 <div class="doc_text">
5621 declare void %llvm.va_start(i8* <arglist>)
5625 <p>The '<tt>llvm.va_start</tt>' intrinsic initializes <tt>*<arglist></tt>
5626 for subsequent use by <tt><a href="#i_va_arg">va_arg</a></tt>.</p>
5629 <p>The argument is a pointer to a <tt>va_list</tt> element to initialize.</p>
5632 <p>The '<tt>llvm.va_start</tt>' intrinsic works just like the <tt>va_start</tt>
5633 macro available in C. In a target-dependent way, it initializes
5634 the <tt>va_list</tt> element to which the argument points, so that the next
5635 call to <tt>va_arg</tt> will produce the first variable argument passed to
5636 the function. Unlike the C <tt>va_start</tt> macro, this intrinsic does not
5637 need to know the last argument of the function as the compiler can figure
5642 <!-- _______________________________________________________________________ -->
5643 <div class="doc_subsubsection">
5644 <a name="int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a>
5647 <div class="doc_text">
5651 declare void @llvm.va_end(i8* <arglist>)
5655 <p>The '<tt>llvm.va_end</tt>' intrinsic destroys <tt>*<arglist></tt>,
5656 which has been initialized previously
5657 with <tt><a href="#int_va_start">llvm.va_start</a></tt>
5658 or <tt><a href="#i_va_copy">llvm.va_copy</a></tt>.</p>
5661 <p>The argument is a pointer to a <tt>va_list</tt> to destroy.</p>
5664 <p>The '<tt>llvm.va_end</tt>' intrinsic works just like the <tt>va_end</tt>
5665 macro available in C. In a target-dependent way, it destroys
5666 the <tt>va_list</tt> element to which the argument points. Calls
5667 to <a href="#int_va_start"><tt>llvm.va_start</tt></a>
5668 and <a href="#int_va_copy"> <tt>llvm.va_copy</tt></a> must be matched exactly
5669 with calls to <tt>llvm.va_end</tt>.</p>
5673 <!-- _______________________________________________________________________ -->
5674 <div class="doc_subsubsection">
5675 <a name="int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a>
5678 <div class="doc_text">
5682 declare void @llvm.va_copy(i8* <destarglist>, i8* <srcarglist>)
5686 <p>The '<tt>llvm.va_copy</tt>' intrinsic copies the current argument position
5687 from the source argument list to the destination argument list.</p>
5690 <p>The first argument is a pointer to a <tt>va_list</tt> element to initialize.
5691 The second argument is a pointer to a <tt>va_list</tt> element to copy
5695 <p>The '<tt>llvm.va_copy</tt>' intrinsic works just like the <tt>va_copy</tt>
5696 macro available in C. In a target-dependent way, it copies the
5697 source <tt>va_list</tt> element into the destination <tt>va_list</tt>
5698 element. This intrinsic is necessary because
5699 the <tt><a href="#int_va_start"> llvm.va_start</a></tt> intrinsic may be
5700 arbitrarily complex and require, for example, memory allocation.</p>
5704 <!-- ======================================================================= -->
5705 <div class="doc_subsection">
5706 <a name="int_gc">Accurate Garbage Collection Intrinsics</a>
5709 <div class="doc_text">
5711 <p>LLVM support for <a href="GarbageCollection.html">Accurate Garbage
5712 Collection</a> (GC) requires the implementation and generation of these
5713 intrinsics. These intrinsics allow identification of <a href="#int_gcroot">GC
5714 roots on the stack</a>, as well as garbage collector implementations that
5715 require <a href="#int_gcread">read</a> and <a href="#int_gcwrite">write</a>
5716 barriers. Front-ends for type-safe garbage collected languages should generate
5717 these intrinsics to make use of the LLVM garbage collectors. For more details,
5718 see <a href="GarbageCollection.html">Accurate Garbage Collection with
5721 <p>The garbage collection intrinsics only operate on objects in the generic
5722 address space (address space zero).</p>
5726 <!-- _______________________________________________________________________ -->
5727 <div class="doc_subsubsection">
5728 <a name="int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a>
5731 <div class="doc_text">
5735 declare void @llvm.gcroot(i8** %ptrloc, i8* %metadata)
5739 <p>The '<tt>llvm.gcroot</tt>' intrinsic declares the existence of a GC root to
5740 the code generator, and allows some metadata to be associated with it.</p>
5743 <p>The first argument specifies the address of a stack object that contains the
5744 root pointer. The second pointer (which must be either a constant or a
5745 global value address) contains the meta-data to be associated with the
5749 <p>At runtime, a call to this intrinsic stores a null pointer into the "ptrloc"
5750 location. At compile-time, the code generator generates information to allow
5751 the runtime to find the pointer at GC safe points. The '<tt>llvm.gcroot</tt>'
5752 intrinsic may only be used in a function which <a href="#gc">specifies a GC
5757 <!-- _______________________________________________________________________ -->
5758 <div class="doc_subsubsection">
5759 <a name="int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a>
5762 <div class="doc_text">
5766 declare i8* @llvm.gcread(i8* %ObjPtr, i8** %Ptr)
5770 <p>The '<tt>llvm.gcread</tt>' intrinsic identifies reads of references from heap
5771 locations, allowing garbage collector implementations that require read
5775 <p>The second argument is the address to read from, which should be an address
5776 allocated from the garbage collector. The first object is a pointer to the
5777 start of the referenced object, if needed by the language runtime (otherwise
5781 <p>The '<tt>llvm.gcread</tt>' intrinsic has the same semantics as a load
5782 instruction, but may be replaced with substantially more complex code by the
5783 garbage collector runtime, as needed. The '<tt>llvm.gcread</tt>' intrinsic
5784 may only be used in a function which <a href="#gc">specifies a GC
5789 <!-- _______________________________________________________________________ -->
5790 <div class="doc_subsubsection">
5791 <a name="int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a>
5794 <div class="doc_text">
5798 declare void @llvm.gcwrite(i8* %P1, i8* %Obj, i8** %P2)
5802 <p>The '<tt>llvm.gcwrite</tt>' intrinsic identifies writes of references to heap
5803 locations, allowing garbage collector implementations that require write
5804 barriers (such as generational or reference counting collectors).</p>
5807 <p>The first argument is the reference to store, the second is the start of the
5808 object to store it to, and the third is the address of the field of Obj to
5809 store to. If the runtime does not require a pointer to the object, Obj may
5813 <p>The '<tt>llvm.gcwrite</tt>' intrinsic has the same semantics as a store
5814 instruction, but may be replaced with substantially more complex code by the
5815 garbage collector runtime, as needed. The '<tt>llvm.gcwrite</tt>' intrinsic
5816 may only be used in a function which <a href="#gc">specifies a GC
5821 <!-- ======================================================================= -->
5822 <div class="doc_subsection">
5823 <a name="int_codegen">Code Generator Intrinsics</a>
5826 <div class="doc_text">
5828 <p>These intrinsics are provided by LLVM to expose special features that may
5829 only be implemented with code generator support.</p>
5833 <!-- _______________________________________________________________________ -->
5834 <div class="doc_subsubsection">
5835 <a name="int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a>
5838 <div class="doc_text">
5842 declare i8 *@llvm.returnaddress(i32 <level>)
5846 <p>The '<tt>llvm.returnaddress</tt>' intrinsic attempts to compute a
5847 target-specific value indicating the return address of the current function
5848 or one of its callers.</p>
5851 <p>The argument to this intrinsic indicates which function to return the address
5852 for. Zero indicates the calling function, one indicates its caller, etc.
5853 The argument is <b>required</b> to be a constant integer value.</p>
5856 <p>The '<tt>llvm.returnaddress</tt>' intrinsic either returns a pointer
5857 indicating the return address of the specified call frame, or zero if it
5858 cannot be identified. The value returned by this intrinsic is likely to be
5859 incorrect or 0 for arguments other than zero, so it should only be used for
5860 debugging purposes.</p>
5862 <p>Note that calling this intrinsic does not prevent function inlining or other
5863 aggressive transformations, so the value returned may not be that of the
5864 obvious source-language caller.</p>
5868 <!-- _______________________________________________________________________ -->
5869 <div class="doc_subsubsection">
5870 <a name="int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a>
5873 <div class="doc_text">
5877 declare i8* @llvm.frameaddress(i32 <level>)
5881 <p>The '<tt>llvm.frameaddress</tt>' intrinsic attempts to return the
5882 target-specific frame pointer value for the specified stack frame.</p>
5885 <p>The argument to this intrinsic indicates which function to return the frame
5886 pointer for. Zero indicates the calling function, one indicates its caller,
5887 etc. The argument is <b>required</b> to be a constant integer value.</p>
5890 <p>The '<tt>llvm.frameaddress</tt>' intrinsic either returns a pointer
5891 indicating the frame address of the specified call frame, or zero if it
5892 cannot be identified. The value returned by this intrinsic is likely to be
5893 incorrect or 0 for arguments other than zero, so it should only be used for
5894 debugging purposes.</p>
5896 <p>Note that calling this intrinsic does not prevent function inlining or other
5897 aggressive transformations, so the value returned may not be that of the
5898 obvious source-language caller.</p>
5902 <!-- _______________________________________________________________________ -->
5903 <div class="doc_subsubsection">
5904 <a name="int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a>
5907 <div class="doc_text">
5911 declare i8* @llvm.stacksave()
5915 <p>The '<tt>llvm.stacksave</tt>' intrinsic is used to remember the current state
5916 of the function stack, for use
5917 with <a href="#int_stackrestore"> <tt>llvm.stackrestore</tt></a>. This is
5918 useful for implementing language features like scoped automatic variable
5919 sized arrays in C99.</p>
5922 <p>This intrinsic returns a opaque pointer value that can be passed
5923 to <a href="#int_stackrestore"><tt>llvm.stackrestore</tt></a>. When
5924 an <tt>llvm.stackrestore</tt> intrinsic is executed with a value saved
5925 from <tt>llvm.stacksave</tt>, it effectively restores the state of the stack
5926 to the state it was in when the <tt>llvm.stacksave</tt> intrinsic executed.
5927 In practice, this pops any <a href="#i_alloca">alloca</a> blocks from the
5928 stack that were allocated after the <tt>llvm.stacksave</tt> was executed.</p>
5932 <!-- _______________________________________________________________________ -->
5933 <div class="doc_subsubsection">
5934 <a name="int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a>
5937 <div class="doc_text">
5941 declare void @llvm.stackrestore(i8* %ptr)
5945 <p>The '<tt>llvm.stackrestore</tt>' intrinsic is used to restore the state of
5946 the function stack to the state it was in when the
5947 corresponding <a href="#int_stacksave"><tt>llvm.stacksave</tt></a> intrinsic
5948 executed. This is useful for implementing language features like scoped
5949 automatic variable sized arrays in C99.</p>
5952 <p>See the description
5953 for <a href="#int_stacksave"><tt>llvm.stacksave</tt></a>.</p>
5957 <!-- _______________________________________________________________________ -->
5958 <div class="doc_subsubsection">
5959 <a name="int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a>
5962 <div class="doc_text">
5966 declare void @llvm.prefetch(i8* <address>, i32 <rw>, i32 <locality>)
5970 <p>The '<tt>llvm.prefetch</tt>' intrinsic is a hint to the code generator to
5971 insert a prefetch instruction if supported; otherwise, it is a noop.
5972 Prefetches have no effect on the behavior of the program but can change its
5973 performance characteristics.</p>
5976 <p><tt>address</tt> is the address to be prefetched, <tt>rw</tt> is the
5977 specifier determining if the fetch should be for a read (0) or write (1),
5978 and <tt>locality</tt> is a temporal locality specifier ranging from (0) - no
5979 locality, to (3) - extremely local keep in cache. The <tt>rw</tt>
5980 and <tt>locality</tt> arguments must be constant integers.</p>
5983 <p>This intrinsic does not modify the behavior of the program. In particular,
5984 prefetches cannot trap and do not produce a value. On targets that support
5985 this intrinsic, the prefetch can provide hints to the processor cache for
5986 better performance.</p>
5990 <!-- _______________________________________________________________________ -->
5991 <div class="doc_subsubsection">
5992 <a name="int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a>
5995 <div class="doc_text">
5999 declare void @llvm.pcmarker(i32 <id>)
6003 <p>The '<tt>llvm.pcmarker</tt>' intrinsic is a method to export a Program
6004 Counter (PC) in a region of code to simulators and other tools. The method
6005 is target specific, but it is expected that the marker will use exported
6006 symbols to transmit the PC of the marker. The marker makes no guarantees
6007 that it will remain with any specific instruction after optimizations. It is
6008 possible that the presence of a marker will inhibit optimizations. The
6009 intended use is to be inserted after optimizations to allow correlations of
6010 simulation runs.</p>
6013 <p><tt>id</tt> is a numerical id identifying the marker.</p>
6016 <p>This intrinsic does not modify the behavior of the program. Backends that do
6017 not support this intrinsic may ignore it.</p>
6021 <!-- _______________________________________________________________________ -->
6022 <div class="doc_subsubsection">
6023 <a name="int_readcyclecounter">'<tt>llvm.readcyclecounter</tt>' Intrinsic</a>
6026 <div class="doc_text">
6030 declare i64 @llvm.readcyclecounter()
6034 <p>The '<tt>llvm.readcyclecounter</tt>' intrinsic provides access to the cycle
6035 counter register (or similar low latency, high accuracy clocks) on those
6036 targets that support it. On X86, it should map to RDTSC. On Alpha, it
6037 should map to RPCC. As the backing counters overflow quickly (on the order
6038 of 9 seconds on alpha), this should only be used for small timings.</p>
6041 <p>When directly supported, reading the cycle counter should not modify any
6042 memory. Implementations are allowed to either return a application specific
6043 value or a system wide value. On backends without support, this is lowered
6044 to a constant 0.</p>
6048 <!-- ======================================================================= -->
6049 <div class="doc_subsection">
6050 <a name="int_libc">Standard C Library Intrinsics</a>
6053 <div class="doc_text">
6055 <p>LLVM provides intrinsics for a few important standard C library functions.
6056 These intrinsics allow source-language front-ends to pass information about
6057 the alignment of the pointer arguments to the code generator, providing
6058 opportunity for more efficient code generation.</p>
6062 <!-- _______________________________________________________________________ -->
6063 <div class="doc_subsubsection">
6064 <a name="int_memcpy">'<tt>llvm.memcpy</tt>' Intrinsic</a>
6067 <div class="doc_text">
6070 <p>This is an overloaded intrinsic. You can use <tt>llvm.memcpy</tt> on any
6071 integer bit width and for different address spaces. Not all targets support
6072 all bit widths however.</p>
6075 declare void @llvm.memcpy.p0i8.p0i8.i32(i8* <dest>, i8* <src>,
6076 i32 <len>, i32 <align>, i1 <isvolatile>)
6077 declare void @llvm.memcpy.p0i8.p0i8.i64(i8* <dest>, i8* <src>,
6078 i64 <len>, i32 <align>, i1 <isvolatile>)
6082 <p>The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the
6083 source location to the destination location.</p>
6085 <p>Note that, unlike the standard libc function, the <tt>llvm.memcpy.*</tt>
6086 intrinsics do not return a value, takes extra alignment/isvolatile arguments
6087 and the pointers can be in specified address spaces.</p>
6091 <p>The first argument is a pointer to the destination, the second is a pointer
6092 to the source. The third argument is an integer argument specifying the
6093 number of bytes to copy, the fourth argument is the alignment of the
6094 source and destination locations, and the fifth is a boolean indicating a
6095 volatile access.</p>
6097 <p>If the call to this intrinsic has an alignment value that is not 0 or 1,
6098 then the caller guarantees that both the source and destination pointers are
6099 aligned to that boundary.</p>
6101 <p>If the <tt>isvolatile</tt> parameter is <tt>true</tt>, the
6102 <tt>llvm.memcpy</tt> call is a <a href="#volatile">volatile operation</a>.
6103 The detailed access behavior is not very cleanly specified and it is unwise
6104 to depend on it.</p>
6108 <p>The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the
6109 source location to the destination location, which are not allowed to
6110 overlap. It copies "len" bytes of memory over. If the argument is known to
6111 be aligned to some boundary, this can be specified as the fourth argument,
6112 otherwise it should be set to 0 or 1.</p>
6116 <!-- _______________________________________________________________________ -->
6117 <div class="doc_subsubsection">
6118 <a name="int_memmove">'<tt>llvm.memmove</tt>' Intrinsic</a>
6121 <div class="doc_text">
6124 <p>This is an overloaded intrinsic. You can use llvm.memmove on any integer bit
6125 width and for different address space. Not all targets support all bit
6129 declare void @llvm.memmove.p0i8.p0i8.i32(i8* <dest>, i8* <src>,
6130 i32 <len>, i32 <align>, i1 <isvolatile>)
6131 declare void @llvm.memmove.p0i8.p0i8.i64(i8* <dest>, i8* <src>,
6132 i64 <len>, i32 <align>, i1 <isvolatile>)
6136 <p>The '<tt>llvm.memmove.*</tt>' intrinsics move a block of memory from the
6137 source location to the destination location. It is similar to the
6138 '<tt>llvm.memcpy</tt>' intrinsic but allows the two memory locations to
6141 <p>Note that, unlike the standard libc function, the <tt>llvm.memmove.*</tt>
6142 intrinsics do not return a value, takes extra alignment/isvolatile arguments
6143 and the pointers can be in specified address spaces.</p>
6147 <p>The first argument is a pointer to the destination, the second is a pointer
6148 to the source. The third argument is an integer argument specifying the
6149 number of bytes to copy, the fourth argument is the alignment of the
6150 source and destination locations, and the fifth is a boolean indicating a
6151 volatile access.</p>
6153 <p>If the call to this intrinsic has an alignment value that is not 0 or 1,
6154 then the caller guarantees that the source and destination pointers are
6155 aligned to that boundary.</p>
6157 <p>If the <tt>isvolatile</tt> parameter is <tt>true</tt>, the
6158 <tt>llvm.memmove</tt> call is a <a href="#volatile">volatile operation</a>.
6159 The detailed access behavior is not very cleanly specified and it is unwise
6160 to depend on it.</p>
6164 <p>The '<tt>llvm.memmove.*</tt>' intrinsics copy a block of memory from the
6165 source location to the destination location, which may overlap. It copies
6166 "len" bytes of memory over. If the argument is known to be aligned to some
6167 boundary, this can be specified as the fourth argument, otherwise it should
6168 be set to 0 or 1.</p>
6172 <!-- _______________________________________________________________________ -->
6173 <div class="doc_subsubsection">
6174 <a name="int_memset">'<tt>llvm.memset.*</tt>' Intrinsics</a>
6177 <div class="doc_text">
6180 <p>This is an overloaded intrinsic. You can use llvm.memset on any integer bit
6181 width and for different address spaces. Not all targets support all bit
6185 declare void @llvm.memset.p0i8.i32(i8* <dest>, i8 <val>,
6186 i32 <len>, i32 <align>, i1 <isvolatile>)
6187 declare void @llvm.memset.p0i8.i64(i8* <dest>, i8 <val>,
6188 i64 <len>, i32 <align>, i1 <isvolatile>)
6192 <p>The '<tt>llvm.memset.*</tt>' intrinsics fill a block of memory with a
6193 particular byte value.</p>
6195 <p>Note that, unlike the standard libc function, the <tt>llvm.memset</tt>
6196 intrinsic does not return a value, takes extra alignment/volatile arguments,
6197 and the destination can be in an arbitrary address space.</p>
6200 <p>The first argument is a pointer to the destination to fill, the second is the
6201 byte value to fill it with, the third argument is an integer argument
6202 specifying the number of bytes to fill, and the fourth argument is the known
6203 alignment of destination location.</p>
6205 <p>If the call to this intrinsic has an alignment value that is not 0 or 1,
6206 then the caller guarantees that the destination pointer is aligned to that
6209 <p>If the <tt>isvolatile</tt> parameter is <tt>true</tt>, the
6210 <tt>llvm.memset</tt> call is a <a href="#volatile">volatile operation</a>.
6211 The detailed access behavior is not very cleanly specified and it is unwise
6212 to depend on it.</p>
6215 <p>The '<tt>llvm.memset.*</tt>' intrinsics fill "len" bytes of memory starting
6216 at the destination location. If the argument is known to be aligned to some
6217 boundary, this can be specified as the fourth argument, otherwise it should
6218 be set to 0 or 1.</p>
6222 <!-- _______________________________________________________________________ -->
6223 <div class="doc_subsubsection">
6224 <a name="int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a>
6227 <div class="doc_text">
6230 <p>This is an overloaded intrinsic. You can use <tt>llvm.sqrt</tt> on any
6231 floating point or vector of floating point type. Not all targets support all
6235 declare float @llvm.sqrt.f32(float %Val)
6236 declare double @llvm.sqrt.f64(double %Val)
6237 declare x86_fp80 @llvm.sqrt.f80(x86_fp80 %Val)
6238 declare fp128 @llvm.sqrt.f128(fp128 %Val)
6239 declare ppc_fp128 @llvm.sqrt.ppcf128(ppc_fp128 %Val)
6243 <p>The '<tt>llvm.sqrt</tt>' intrinsics return the sqrt of the specified operand,
6244 returning the same value as the libm '<tt>sqrt</tt>' functions would.
6245 Unlike <tt>sqrt</tt> in libm, however, <tt>llvm.sqrt</tt> has undefined
6246 behavior for negative numbers other than -0.0 (which allows for better
6247 optimization, because there is no need to worry about errno being
6248 set). <tt>llvm.sqrt(-0.0)</tt> is defined to return -0.0 like IEEE sqrt.</p>
6251 <p>The argument and return value are floating point numbers of the same
6255 <p>This function returns the sqrt of the specified operand if it is a
6256 nonnegative floating point number.</p>
6260 <!-- _______________________________________________________________________ -->
6261 <div class="doc_subsubsection">
6262 <a name="int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a>
6265 <div class="doc_text">
6268 <p>This is an overloaded intrinsic. You can use <tt>llvm.powi</tt> on any
6269 floating point or vector of floating point type. Not all targets support all
6273 declare float @llvm.powi.f32(float %Val, i32 %power)
6274 declare double @llvm.powi.f64(double %Val, i32 %power)
6275 declare x86_fp80 @llvm.powi.f80(x86_fp80 %Val, i32 %power)
6276 declare fp128 @llvm.powi.f128(fp128 %Val, i32 %power)
6277 declare ppc_fp128 @llvm.powi.ppcf128(ppc_fp128 %Val, i32 %power)
6281 <p>The '<tt>llvm.powi.*</tt>' intrinsics return the first operand raised to the
6282 specified (positive or negative) power. The order of evaluation of
6283 multiplications is not defined. When a vector of floating point type is
6284 used, the second argument remains a scalar integer value.</p>
6287 <p>The second argument is an integer power, and the first is a value to raise to
6291 <p>This function returns the first value raised to the second power with an
6292 unspecified sequence of rounding operations.</p>
6296 <!-- _______________________________________________________________________ -->
6297 <div class="doc_subsubsection">
6298 <a name="int_sin">'<tt>llvm.sin.*</tt>' Intrinsic</a>
6301 <div class="doc_text">
6304 <p>This is an overloaded intrinsic. You can use <tt>llvm.sin</tt> on any
6305 floating point or vector of floating point type. Not all targets support all
6309 declare float @llvm.sin.f32(float %Val)
6310 declare double @llvm.sin.f64(double %Val)
6311 declare x86_fp80 @llvm.sin.f80(x86_fp80 %Val)
6312 declare fp128 @llvm.sin.f128(fp128 %Val)
6313 declare ppc_fp128 @llvm.sin.ppcf128(ppc_fp128 %Val)
6317 <p>The '<tt>llvm.sin.*</tt>' intrinsics return the sine of the operand.</p>
6320 <p>The argument and return value are floating point numbers of the same
6324 <p>This function returns the sine of the specified operand, returning the same
6325 values as the libm <tt>sin</tt> functions would, and handles error conditions
6326 in the same way.</p>
6330 <!-- _______________________________________________________________________ -->
6331 <div class="doc_subsubsection">
6332 <a name="int_cos">'<tt>llvm.cos.*</tt>' Intrinsic</a>
6335 <div class="doc_text">
6338 <p>This is an overloaded intrinsic. You can use <tt>llvm.cos</tt> on any
6339 floating point or vector of floating point type. Not all targets support all
6343 declare float @llvm.cos.f32(float %Val)
6344 declare double @llvm.cos.f64(double %Val)
6345 declare x86_fp80 @llvm.cos.f80(x86_fp80 %Val)
6346 declare fp128 @llvm.cos.f128(fp128 %Val)
6347 declare ppc_fp128 @llvm.cos.ppcf128(ppc_fp128 %Val)
6351 <p>The '<tt>llvm.cos.*</tt>' intrinsics return the cosine of the operand.</p>
6354 <p>The argument and return value are floating point numbers of the same
6358 <p>This function returns the cosine of the specified operand, returning the same
6359 values as the libm <tt>cos</tt> functions would, and handles error conditions
6360 in the same way.</p>
6364 <!-- _______________________________________________________________________ -->
6365 <div class="doc_subsubsection">
6366 <a name="int_pow">'<tt>llvm.pow.*</tt>' Intrinsic</a>
6369 <div class="doc_text">
6372 <p>This is an overloaded intrinsic. You can use <tt>llvm.pow</tt> on any
6373 floating point or vector of floating point type. Not all targets support all
6377 declare float @llvm.pow.f32(float %Val, float %Power)
6378 declare double @llvm.pow.f64(double %Val, double %Power)
6379 declare x86_fp80 @llvm.pow.f80(x86_fp80 %Val, x86_fp80 %Power)
6380 declare fp128 @llvm.pow.f128(fp128 %Val, fp128 %Power)
6381 declare ppc_fp128 @llvm.pow.ppcf128(ppc_fp128 %Val, ppc_fp128 Power)
6385 <p>The '<tt>llvm.pow.*</tt>' intrinsics return the first operand raised to the
6386 specified (positive or negative) power.</p>
6389 <p>The second argument is a floating point power, and the first is a value to
6390 raise to that power.</p>
6393 <p>This function returns the first value raised to the second power, returning
6394 the same values as the libm <tt>pow</tt> functions would, and handles error
6395 conditions in the same way.</p>
6399 <!-- ======================================================================= -->
6400 <div class="doc_subsection">
6401 <a name="int_manip">Bit Manipulation Intrinsics</a>
6404 <div class="doc_text">
6406 <p>LLVM provides intrinsics for a few important bit manipulation operations.
6407 These allow efficient code generation for some algorithms.</p>
6411 <!-- _______________________________________________________________________ -->
6412 <div class="doc_subsubsection">
6413 <a name="int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a>
6416 <div class="doc_text">
6419 <p>This is an overloaded intrinsic function. You can use bswap on any integer
6420 type that is an even number of bytes (i.e. BitWidth % 16 == 0).</p>
6423 declare i16 @llvm.bswap.i16(i16 <id>)
6424 declare i32 @llvm.bswap.i32(i32 <id>)
6425 declare i64 @llvm.bswap.i64(i64 <id>)
6429 <p>The '<tt>llvm.bswap</tt>' family of intrinsics is used to byte swap integer
6430 values with an even number of bytes (positive multiple of 16 bits). These
6431 are useful for performing operations on data that is not in the target's
6432 native byte order.</p>
6435 <p>The <tt>llvm.bswap.i16</tt> intrinsic returns an i16 value that has the high
6436 and low byte of the input i16 swapped. Similarly,
6437 the <tt>llvm.bswap.i32</tt> intrinsic returns an i32 value that has the four
6438 bytes of the input i32 swapped, so that if the input bytes are numbered 0, 1,
6439 2, 3 then the returned i32 will have its bytes in 3, 2, 1, 0 order.
6440 The <tt>llvm.bswap.i48</tt>, <tt>llvm.bswap.i64</tt> and other intrinsics
6441 extend this concept to additional even-byte lengths (6 bytes, 8 bytes and
6442 more, respectively).</p>
6446 <!-- _______________________________________________________________________ -->
6447 <div class="doc_subsubsection">
6448 <a name="int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic</a>
6451 <div class="doc_text">
6454 <p>This is an overloaded intrinsic. You can use llvm.ctpop on any integer bit
6455 width. Not all targets support all bit widths however.</p>
6458 declare i8 @llvm.ctpop.i8(i8 <src>)
6459 declare i16 @llvm.ctpop.i16(i16 <src>)
6460 declare i32 @llvm.ctpop.i32(i32 <src>)
6461 declare i64 @llvm.ctpop.i64(i64 <src>)
6462 declare i256 @llvm.ctpop.i256(i256 <src>)
6466 <p>The '<tt>llvm.ctpop</tt>' family of intrinsics counts the number of bits set
6470 <p>The only argument is the value to be counted. The argument may be of any
6471 integer type. The return type must match the argument type.</p>
6474 <p>The '<tt>llvm.ctpop</tt>' intrinsic counts the 1's in a variable.</p>
6478 <!-- _______________________________________________________________________ -->
6479 <div class="doc_subsubsection">
6480 <a name="int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic</a>
6483 <div class="doc_text">
6486 <p>This is an overloaded intrinsic. You can use <tt>llvm.ctlz</tt> on any
6487 integer bit width. Not all targets support all bit widths however.</p>
6490 declare i8 @llvm.ctlz.i8 (i8 <src>)
6491 declare i16 @llvm.ctlz.i16(i16 <src>)
6492 declare i32 @llvm.ctlz.i32(i32 <src>)
6493 declare i64 @llvm.ctlz.i64(i64 <src>)
6494 declare i256 @llvm.ctlz.i256(i256 <src>)
6498 <p>The '<tt>llvm.ctlz</tt>' family of intrinsic functions counts the number of
6499 leading zeros in a variable.</p>
6502 <p>The only argument is the value to be counted. The argument may be of any
6503 integer type. The return type must match the argument type.</p>
6506 <p>The '<tt>llvm.ctlz</tt>' intrinsic counts the leading (most significant)
6507 zeros in a variable. If the src == 0 then the result is the size in bits of
6508 the type of src. For example, <tt>llvm.ctlz(i32 2) = 30</tt>.</p>
6512 <!-- _______________________________________________________________________ -->
6513 <div class="doc_subsubsection">
6514 <a name="int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic</a>
6517 <div class="doc_text">
6520 <p>This is an overloaded intrinsic. You can use <tt>llvm.cttz</tt> on any
6521 integer bit width. Not all targets support all bit widths however.</p>
6524 declare i8 @llvm.cttz.i8 (i8 <src>)
6525 declare i16 @llvm.cttz.i16(i16 <src>)
6526 declare i32 @llvm.cttz.i32(i32 <src>)
6527 declare i64 @llvm.cttz.i64(i64 <src>)
6528 declare i256 @llvm.cttz.i256(i256 <src>)
6532 <p>The '<tt>llvm.cttz</tt>' family of intrinsic functions counts the number of
6536 <p>The only argument is the value to be counted. The argument may be of any
6537 integer type. The return type must match the argument type.</p>
6540 <p>The '<tt>llvm.cttz</tt>' intrinsic counts the trailing (least significant)
6541 zeros in a variable. If the src == 0 then the result is the size in bits of
6542 the type of src. For example, <tt>llvm.cttz(2) = 1</tt>.</p>
6546 <!-- ======================================================================= -->
6547 <div class="doc_subsection">
6548 <a name="int_overflow">Arithmetic with Overflow Intrinsics</a>
6551 <div class="doc_text">
6553 <p>LLVM provides intrinsics for some arithmetic with overflow operations.</p>
6557 <!-- _______________________________________________________________________ -->
6558 <div class="doc_subsubsection">
6559 <a name="int_sadd_overflow">'<tt>llvm.sadd.with.overflow.*</tt>' Intrinsics</a>
6562 <div class="doc_text">
6565 <p>This is an overloaded intrinsic. You can use <tt>llvm.sadd.with.overflow</tt>
6566 on any integer bit width.</p>
6569 declare {i16, i1} @llvm.sadd.with.overflow.i16(i16 %a, i16 %b)
6570 declare {i32, i1} @llvm.sadd.with.overflow.i32(i32 %a, i32 %b)
6571 declare {i64, i1} @llvm.sadd.with.overflow.i64(i64 %a, i64 %b)
6575 <p>The '<tt>llvm.sadd.with.overflow</tt>' family of intrinsic functions perform
6576 a signed addition of the two arguments, and indicate whether an overflow
6577 occurred during the signed summation.</p>
6580 <p>The arguments (%a and %b) and the first element of the result structure may
6581 be of integer types of any bit width, but they must have the same bit
6582 width. The second element of the result structure must be of
6583 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
6584 undergo signed addition.</p>
6587 <p>The '<tt>llvm.sadd.with.overflow</tt>' family of intrinsic functions perform
6588 a signed addition of the two variables. They return a structure — the
6589 first element of which is the signed summation, and the second element of
6590 which is a bit specifying if the signed summation resulted in an
6595 %res = call {i32, i1} @llvm.sadd.with.overflow.i32(i32 %a, i32 %b)
6596 %sum = extractvalue {i32, i1} %res, 0
6597 %obit = extractvalue {i32, i1} %res, 1
6598 br i1 %obit, label %overflow, label %normal
6603 <!-- _______________________________________________________________________ -->
6604 <div class="doc_subsubsection">
6605 <a name="int_uadd_overflow">'<tt>llvm.uadd.with.overflow.*</tt>' Intrinsics</a>
6608 <div class="doc_text">
6611 <p>This is an overloaded intrinsic. You can use <tt>llvm.uadd.with.overflow</tt>
6612 on any integer bit width.</p>
6615 declare {i16, i1} @llvm.uadd.with.overflow.i16(i16 %a, i16 %b)
6616 declare {i32, i1} @llvm.uadd.with.overflow.i32(i32 %a, i32 %b)
6617 declare {i64, i1} @llvm.uadd.with.overflow.i64(i64 %a, i64 %b)
6621 <p>The '<tt>llvm.uadd.with.overflow</tt>' family of intrinsic functions perform
6622 an unsigned addition of the two arguments, and indicate whether a carry
6623 occurred during the unsigned summation.</p>
6626 <p>The arguments (%a and %b) and the first element of the result structure may
6627 be of integer types of any bit width, but they must have the same bit
6628 width. The second element of the result structure must be of
6629 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
6630 undergo unsigned addition.</p>
6633 <p>The '<tt>llvm.uadd.with.overflow</tt>' family of intrinsic functions perform
6634 an unsigned addition of the two arguments. They return a structure —
6635 the first element of which is the sum, and the second element of which is a
6636 bit specifying if the unsigned summation resulted in a carry.</p>
6640 %res = call {i32, i1} @llvm.uadd.with.overflow.i32(i32 %a, i32 %b)
6641 %sum = extractvalue {i32, i1} %res, 0
6642 %obit = extractvalue {i32, i1} %res, 1
6643 br i1 %obit, label %carry, label %normal
6648 <!-- _______________________________________________________________________ -->
6649 <div class="doc_subsubsection">
6650 <a name="int_ssub_overflow">'<tt>llvm.ssub.with.overflow.*</tt>' Intrinsics</a>
6653 <div class="doc_text">
6656 <p>This is an overloaded intrinsic. You can use <tt>llvm.ssub.with.overflow</tt>
6657 on any integer bit width.</p>
6660 declare {i16, i1} @llvm.ssub.with.overflow.i16(i16 %a, i16 %b)
6661 declare {i32, i1} @llvm.ssub.with.overflow.i32(i32 %a, i32 %b)
6662 declare {i64, i1} @llvm.ssub.with.overflow.i64(i64 %a, i64 %b)
6666 <p>The '<tt>llvm.ssub.with.overflow</tt>' family of intrinsic functions perform
6667 a signed subtraction of the two arguments, and indicate whether an overflow
6668 occurred during the signed subtraction.</p>
6671 <p>The arguments (%a and %b) and the first element of the result structure may
6672 be of integer types of any bit width, but they must have the same bit
6673 width. The second element of the result structure must be of
6674 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
6675 undergo signed subtraction.</p>
6678 <p>The '<tt>llvm.ssub.with.overflow</tt>' family of intrinsic functions perform
6679 a signed subtraction of the two arguments. They return a structure —
6680 the first element of which is the subtraction, and the second element of
6681 which is a bit specifying if the signed subtraction resulted in an
6686 %res = call {i32, i1} @llvm.ssub.with.overflow.i32(i32 %a, i32 %b)
6687 %sum = extractvalue {i32, i1} %res, 0
6688 %obit = extractvalue {i32, i1} %res, 1
6689 br i1 %obit, label %overflow, label %normal
6694 <!-- _______________________________________________________________________ -->
6695 <div class="doc_subsubsection">
6696 <a name="int_usub_overflow">'<tt>llvm.usub.with.overflow.*</tt>' Intrinsics</a>
6699 <div class="doc_text">
6702 <p>This is an overloaded intrinsic. You can use <tt>llvm.usub.with.overflow</tt>
6703 on any integer bit width.</p>
6706 declare {i16, i1} @llvm.usub.with.overflow.i16(i16 %a, i16 %b)
6707 declare {i32, i1} @llvm.usub.with.overflow.i32(i32 %a, i32 %b)
6708 declare {i64, i1} @llvm.usub.with.overflow.i64(i64 %a, i64 %b)
6712 <p>The '<tt>llvm.usub.with.overflow</tt>' family of intrinsic functions perform
6713 an unsigned subtraction of the two arguments, and indicate whether an
6714 overflow occurred during the unsigned subtraction.</p>
6717 <p>The arguments (%a and %b) and the first element of the result structure may
6718 be of integer types of any bit width, but they must have the same bit
6719 width. The second element of the result structure must be of
6720 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
6721 undergo unsigned subtraction.</p>
6724 <p>The '<tt>llvm.usub.with.overflow</tt>' family of intrinsic functions perform
6725 an unsigned subtraction of the two arguments. They return a structure —
6726 the first element of which is the subtraction, and the second element of
6727 which is a bit specifying if the unsigned subtraction resulted in an
6732 %res = call {i32, i1} @llvm.usub.with.overflow.i32(i32 %a, i32 %b)
6733 %sum = extractvalue {i32, i1} %res, 0
6734 %obit = extractvalue {i32, i1} %res, 1
6735 br i1 %obit, label %overflow, label %normal
6740 <!-- _______________________________________________________________________ -->
6741 <div class="doc_subsubsection">
6742 <a name="int_smul_overflow">'<tt>llvm.smul.with.overflow.*</tt>' Intrinsics</a>
6745 <div class="doc_text">
6748 <p>This is an overloaded intrinsic. You can use <tt>llvm.smul.with.overflow</tt>
6749 on any integer bit width.</p>
6752 declare {i16, i1} @llvm.smul.with.overflow.i16(i16 %a, i16 %b)
6753 declare {i32, i1} @llvm.smul.with.overflow.i32(i32 %a, i32 %b)
6754 declare {i64, i1} @llvm.smul.with.overflow.i64(i64 %a, i64 %b)
6759 <p>The '<tt>llvm.smul.with.overflow</tt>' family of intrinsic functions perform
6760 a signed multiplication of the two arguments, and indicate whether an
6761 overflow occurred during the signed multiplication.</p>
6764 <p>The arguments (%a and %b) and the first element of the result structure may
6765 be of integer types of any bit width, but they must have the same bit
6766 width. The second element of the result structure must be of
6767 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
6768 undergo signed multiplication.</p>
6771 <p>The '<tt>llvm.smul.with.overflow</tt>' family of intrinsic functions perform
6772 a signed multiplication of the two arguments. They return a structure —
6773 the first element of which is the multiplication, and the second element of
6774 which is a bit specifying if the signed multiplication resulted in an
6779 %res = call {i32, i1} @llvm.smul.with.overflow.i32(i32 %a, i32 %b)
6780 %sum = extractvalue {i32, i1} %res, 0
6781 %obit = extractvalue {i32, i1} %res, 1
6782 br i1 %obit, label %overflow, label %normal
6787 <!-- _______________________________________________________________________ -->
6788 <div class="doc_subsubsection">
6789 <a name="int_umul_overflow">'<tt>llvm.umul.with.overflow.*</tt>' Intrinsics</a>
6792 <div class="doc_text">
6795 <p>This is an overloaded intrinsic. You can use <tt>llvm.umul.with.overflow</tt>
6796 on any integer bit width.</p>
6799 declare {i16, i1} @llvm.umul.with.overflow.i16(i16 %a, i16 %b)
6800 declare {i32, i1} @llvm.umul.with.overflow.i32(i32 %a, i32 %b)
6801 declare {i64, i1} @llvm.umul.with.overflow.i64(i64 %a, i64 %b)
6805 <p>The '<tt>llvm.umul.with.overflow</tt>' family of intrinsic functions perform
6806 a unsigned multiplication of the two arguments, and indicate whether an
6807 overflow occurred during the unsigned multiplication.</p>
6810 <p>The arguments (%a and %b) and the first element of the result structure may
6811 be of integer types of any bit width, but they must have the same bit
6812 width. The second element of the result structure must be of
6813 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
6814 undergo unsigned multiplication.</p>
6817 <p>The '<tt>llvm.umul.with.overflow</tt>' family of intrinsic functions perform
6818 an unsigned multiplication of the two arguments. They return a structure
6819 — the first element of which is the multiplication, and the second
6820 element of which is a bit specifying if the unsigned multiplication resulted
6825 %res = call {i32, i1} @llvm.umul.with.overflow.i32(i32 %a, i32 %b)
6826 %sum = extractvalue {i32, i1} %res, 0
6827 %obit = extractvalue {i32, i1} %res, 1
6828 br i1 %obit, label %overflow, label %normal
6833 <!-- ======================================================================= -->
6834 <div class="doc_subsection">
6835 <a name="int_fp16">Half Precision Floating Point Intrinsics</a>
6838 <div class="doc_text">
6840 <p>Half precision floating point is a storage-only format. This means that it is
6841 a dense encoding (in memory) but does not support computation in the
6844 <p>This means that code must first load the half-precision floating point
6845 value as an i16, then convert it to float with <a
6846 href="#int_convert_from_fp16"><tt>llvm.convert.from.fp16</tt></a>.
6847 Computation can then be performed on the float value (including extending to
6848 double etc). To store the value back to memory, it is first converted to
6849 float if needed, then converted to i16 with
6850 <a href="#int_convert_to_fp16"><tt>llvm.convert.to.fp16</tt></a>, then
6851 storing as an i16 value.</p>
6854 <!-- _______________________________________________________________________ -->
6855 <div class="doc_subsubsection">
6856 <a name="int_convert_to_fp16">'<tt>llvm.convert.to.fp16</tt>' Intrinsic</a>
6859 <div class="doc_text">
6863 declare i16 @llvm.convert.to.fp16(f32 %a)
6867 <p>The '<tt>llvm.convert.to.fp16</tt>' intrinsic function performs
6868 a conversion from single precision floating point format to half precision
6869 floating point format.</p>
6872 <p>The intrinsic function contains single argument - the value to be
6876 <p>The '<tt>llvm.convert.to.fp16</tt>' intrinsic function performs
6877 a conversion from single precision floating point format to half precision
6878 floating point format. The return value is an <tt>i16</tt> which
6879 contains the converted number.</p>
6883 %res = call i16 @llvm.convert.to.fp16(f32 %a)
6884 store i16 %res, i16* @x, align 2
6889 <!-- _______________________________________________________________________ -->
6890 <div class="doc_subsubsection">
6891 <a name="int_convert_from_fp16">'<tt>llvm.convert.from.fp16</tt>' Intrinsic</a>
6894 <div class="doc_text">
6898 declare f32 @llvm.convert.from.fp16(i16 %a)
6902 <p>The '<tt>llvm.convert.from.fp16</tt>' intrinsic function performs
6903 a conversion from half precision floating point format to single precision
6904 floating point format.</p>
6907 <p>The intrinsic function contains single argument - the value to be
6911 <p>The '<tt>llvm.convert.from.fp16</tt>' intrinsic function performs a
6912 conversion from half single precision floating point format to single
6913 precision floating point format. The input half-float value is represented by
6914 an <tt>i16</tt> value.</p>
6918 %a = load i16* @x, align 2
6919 %res = call f32 @llvm.convert.from.fp16(i16 %a)
6924 <!-- ======================================================================= -->
6925 <div class="doc_subsection">
6926 <a name="int_debugger">Debugger Intrinsics</a>
6929 <div class="doc_text">
6931 <p>The LLVM debugger intrinsics (which all start with <tt>llvm.dbg.</tt>
6932 prefix), are described in
6933 the <a href="SourceLevelDebugging.html#format_common_intrinsics">LLVM Source
6934 Level Debugging</a> document.</p>
6938 <!-- ======================================================================= -->
6939 <div class="doc_subsection">
6940 <a name="int_eh">Exception Handling Intrinsics</a>
6943 <div class="doc_text">
6945 <p>The LLVM exception handling intrinsics (which all start with
6946 <tt>llvm.eh.</tt> prefix), are described in
6947 the <a href="ExceptionHandling.html#format_common_intrinsics">LLVM Exception
6948 Handling</a> document.</p>
6952 <!-- ======================================================================= -->
6953 <div class="doc_subsection">
6954 <a name="int_trampoline">Trampoline Intrinsic</a>
6957 <div class="doc_text">
6959 <p>This intrinsic makes it possible to excise one parameter, marked with
6960 the <a href="#nest"><tt>nest</tt></a> attribute, from a function.
6961 The result is a callable
6962 function pointer lacking the nest parameter - the caller does not need to
6963 provide a value for it. Instead, the value to use is stored in advance in a
6964 "trampoline", a block of memory usually allocated on the stack, which also
6965 contains code to splice the nest value into the argument list. This is used
6966 to implement the GCC nested function address extension.</p>
6968 <p>For example, if the function is
6969 <tt>i32 f(i8* nest %c, i32 %x, i32 %y)</tt> then the resulting function
6970 pointer has signature <tt>i32 (i32, i32)*</tt>. It can be created as
6973 <div class="doc_code">
6975 %tramp = alloca [10 x i8], align 4 ; size and alignment only correct for X86
6976 %tramp1 = getelementptr [10 x i8]* %tramp, i32 0, i32 0
6977 %p = call i8* @llvm.init.trampoline(i8* %tramp1, i8* bitcast (i32 (i8* nest , i32, i32)* @f to i8*), i8* %nval)
6978 %fp = bitcast i8* %p to i32 (i32, i32)*
6982 <p>The call <tt>%val = call i32 %fp(i32 %x, i32 %y)</tt> is then equivalent
6983 to <tt>%val = call i32 %f(i8* %nval, i32 %x, i32 %y)</tt>.</p>
6987 <!-- _______________________________________________________________________ -->
6988 <div class="doc_subsubsection">
6989 <a name="int_it">'<tt>llvm.init.trampoline</tt>' Intrinsic</a>
6992 <div class="doc_text">
6996 declare i8* @llvm.init.trampoline(i8* <tramp>, i8* <func>, i8* <nval>)
7000 <p>This fills the memory pointed to by <tt>tramp</tt> with code and returns a
7001 function pointer suitable for executing it.</p>
7004 <p>The <tt>llvm.init.trampoline</tt> intrinsic takes three arguments, all
7005 pointers. The <tt>tramp</tt> argument must point to a sufficiently large and
7006 sufficiently aligned block of memory; this memory is written to by the
7007 intrinsic. Note that the size and the alignment are target-specific - LLVM
7008 currently provides no portable way of determining them, so a front-end that
7009 generates this intrinsic needs to have some target-specific knowledge.
7010 The <tt>func</tt> argument must hold a function bitcast to
7011 an <tt>i8*</tt>.</p>
7014 <p>The block of memory pointed to by <tt>tramp</tt> is filled with target
7015 dependent code, turning it into a function. A pointer to this function is
7016 returned, but needs to be bitcast to an <a href="#int_trampoline">appropriate
7017 function pointer type</a> before being called. The new function's signature
7018 is the same as that of <tt>func</tt> with any arguments marked with
7019 the <tt>nest</tt> attribute removed. At most one such <tt>nest</tt> argument
7020 is allowed, and it must be of pointer type. Calling the new function is
7021 equivalent to calling <tt>func</tt> with the same argument list, but
7022 with <tt>nval</tt> used for the missing <tt>nest</tt> argument. If, after
7023 calling <tt>llvm.init.trampoline</tt>, the memory pointed to
7024 by <tt>tramp</tt> is modified, then the effect of any later call to the
7025 returned function pointer is undefined.</p>
7029 <!-- ======================================================================= -->
7030 <div class="doc_subsection">
7031 <a name="int_atomics">Atomic Operations and Synchronization Intrinsics</a>
7034 <div class="doc_text">
7036 <p>These intrinsic functions expand the "universal IR" of LLVM to represent
7037 hardware constructs for atomic operations and memory synchronization. This
7038 provides an interface to the hardware, not an interface to the programmer. It
7039 is aimed at a low enough level to allow any programming models or APIs
7040 (Application Programming Interfaces) which need atomic behaviors to map
7041 cleanly onto it. It is also modeled primarily on hardware behavior. Just as
7042 hardware provides a "universal IR" for source languages, it also provides a
7043 starting point for developing a "universal" atomic operation and
7044 synchronization IR.</p>
7046 <p>These do <em>not</em> form an API such as high-level threading libraries,
7047 software transaction memory systems, atomic primitives, and intrinsic
7048 functions as found in BSD, GNU libc, atomic_ops, APR, and other system and
7049 application libraries. The hardware interface provided by LLVM should allow
7050 a clean implementation of all of these APIs and parallel programming models.
7051 No one model or paradigm should be selected above others unless the hardware
7052 itself ubiquitously does so.</p>
7056 <!-- _______________________________________________________________________ -->
7057 <div class="doc_subsubsection">
7058 <a name="int_memory_barrier">'<tt>llvm.memory.barrier</tt>' Intrinsic</a>
7060 <div class="doc_text">
7063 declare void @llvm.memory.barrier(i1 <ll>, i1 <ls>, i1 <sl>, i1 <ss>, i1 <device>)
7067 <p>The <tt>llvm.memory.barrier</tt> intrinsic guarantees ordering between
7068 specific pairs of memory access types.</p>
7071 <p>The <tt>llvm.memory.barrier</tt> intrinsic requires five boolean arguments.
7072 The first four arguments enables a specific barrier as listed below. The
7073 fifth argument specifies that the barrier applies to io or device or uncached
7077 <li><tt>ll</tt>: load-load barrier</li>
7078 <li><tt>ls</tt>: load-store barrier</li>
7079 <li><tt>sl</tt>: store-load barrier</li>
7080 <li><tt>ss</tt>: store-store barrier</li>
7081 <li><tt>device</tt>: barrier applies to device and uncached memory also.</li>
7085 <p>This intrinsic causes the system to enforce some ordering constraints upon
7086 the loads and stores of the program. This barrier does not
7087 indicate <em>when</em> any events will occur, it only enforces
7088 an <em>order</em> in which they occur. For any of the specified pairs of load
7089 and store operations (f.ex. load-load, or store-load), all of the first
7090 operations preceding the barrier will complete before any of the second
7091 operations succeeding the barrier begin. Specifically the semantics for each
7092 pairing is as follows:</p>
7095 <li><tt>ll</tt>: All loads before the barrier must complete before any load
7096 after the barrier begins.</li>
7097 <li><tt>ls</tt>: All loads before the barrier must complete before any
7098 store after the barrier begins.</li>
7099 <li><tt>ss</tt>: All stores before the barrier must complete before any
7100 store after the barrier begins.</li>
7101 <li><tt>sl</tt>: All stores before the barrier must complete before any
7102 load after the barrier begins.</li>
7105 <p>These semantics are applied with a logical "and" behavior when more than one
7106 is enabled in a single memory barrier intrinsic.</p>
7108 <p>Backends may implement stronger barriers than those requested when they do
7109 not support as fine grained a barrier as requested. Some architectures do
7110 not need all types of barriers and on such architectures, these become
7115 %mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
7116 %ptr = bitcast i8* %mallocP to i32*
7119 %result1 = load i32* %ptr <i>; yields {i32}:result1 = 4</i>
7120 call void @llvm.memory.barrier(i1 false, i1 true, i1 false, i1 false)
7121 <i>; guarantee the above finishes</i>
7122 store i32 8, %ptr <i>; before this begins</i>
7127 <!-- _______________________________________________________________________ -->
7128 <div class="doc_subsubsection">
7129 <a name="int_atomic_cmp_swap">'<tt>llvm.atomic.cmp.swap.*</tt>' Intrinsic</a>
7132 <div class="doc_text">
7135 <p>This is an overloaded intrinsic. You can use <tt>llvm.atomic.cmp.swap</tt> on
7136 any integer bit width and for different address spaces. Not all targets
7137 support all bit widths however.</p>
7140 declare i8 @llvm.atomic.cmp.swap.i8.p0i8(i8* <ptr>, i8 <cmp>, i8 <val>)
7141 declare i16 @llvm.atomic.cmp.swap.i16.p0i16(i16* <ptr>, i16 <cmp>, i16 <val>)
7142 declare i32 @llvm.atomic.cmp.swap.i32.p0i32(i32* <ptr>, i32 <cmp>, i32 <val>)
7143 declare i64 @llvm.atomic.cmp.swap.i64.p0i64(i64* <ptr>, i64 <cmp>, i64 <val>)
7147 <p>This loads a value in memory and compares it to a given value. If they are
7148 equal, it stores a new value into the memory.</p>
7151 <p>The <tt>llvm.atomic.cmp.swap</tt> intrinsic takes three arguments. The result
7152 as well as both <tt>cmp</tt> and <tt>val</tt> must be integer values with the
7153 same bit width. The <tt>ptr</tt> argument must be a pointer to a value of
7154 this integer type. While any bit width integer may be used, targets may only
7155 lower representations they support in hardware.</p>
7158 <p>This entire intrinsic must be executed atomically. It first loads the value
7159 in memory pointed to by <tt>ptr</tt> and compares it with the
7160 value <tt>cmp</tt>. If they are equal, <tt>val</tt> is stored into the
7161 memory. The loaded value is yielded in all cases. This provides the
7162 equivalent of an atomic compare-and-swap operation within the SSA
7167 %mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
7168 %ptr = bitcast i8* %mallocP to i32*
7171 %val1 = add i32 4, 4
7172 %result1 = call i32 @llvm.atomic.cmp.swap.i32.p0i32(i32* %ptr, i32 4, %val1)
7173 <i>; yields {i32}:result1 = 4</i>
7174 %stored1 = icmp eq i32 %result1, 4 <i>; yields {i1}:stored1 = true</i>
7175 %memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 8</i>
7177 %val2 = add i32 1, 1
7178 %result2 = call i32 @llvm.atomic.cmp.swap.i32.p0i32(i32* %ptr, i32 5, %val2)
7179 <i>; yields {i32}:result2 = 8</i>
7180 %stored2 = icmp eq i32 %result2, 5 <i>; yields {i1}:stored2 = false</i>
7182 %memval2 = load i32* %ptr <i>; yields {i32}:memval2 = 8</i>
7187 <!-- _______________________________________________________________________ -->
7188 <div class="doc_subsubsection">
7189 <a name="int_atomic_swap">'<tt>llvm.atomic.swap.*</tt>' Intrinsic</a>
7191 <div class="doc_text">
7194 <p>This is an overloaded intrinsic. You can use <tt>llvm.atomic.swap</tt> on any
7195 integer bit width. Not all targets support all bit widths however.</p>
7198 declare i8 @llvm.atomic.swap.i8.p0i8(i8* <ptr>, i8 <val>)
7199 declare i16 @llvm.atomic.swap.i16.p0i16(i16* <ptr>, i16 <val>)
7200 declare i32 @llvm.atomic.swap.i32.p0i32(i32* <ptr>, i32 <val>)
7201 declare i64 @llvm.atomic.swap.i64.p0i64(i64* <ptr>, i64 <val>)
7205 <p>This intrinsic loads the value stored in memory at <tt>ptr</tt> and yields
7206 the value from memory. It then stores the value in <tt>val</tt> in the memory
7207 at <tt>ptr</tt>.</p>
7210 <p>The <tt>llvm.atomic.swap</tt> intrinsic takes two arguments. Both
7211 the <tt>val</tt> argument and the result must be integers of the same bit
7212 width. The first argument, <tt>ptr</tt>, must be a pointer to a value of this
7213 integer type. The targets may only lower integer representations they
7217 <p>This intrinsic loads the value pointed to by <tt>ptr</tt>, yields it, and
7218 stores <tt>val</tt> back into <tt>ptr</tt> atomically. This provides the
7219 equivalent of an atomic swap operation within the SSA framework.</p>
7223 %mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
7224 %ptr = bitcast i8* %mallocP to i32*
7227 %val1 = add i32 4, 4
7228 %result1 = call i32 @llvm.atomic.swap.i32.p0i32(i32* %ptr, i32 %val1)
7229 <i>; yields {i32}:result1 = 4</i>
7230 %stored1 = icmp eq i32 %result1, 4 <i>; yields {i1}:stored1 = true</i>
7231 %memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 8</i>
7233 %val2 = add i32 1, 1
7234 %result2 = call i32 @llvm.atomic.swap.i32.p0i32(i32* %ptr, i32 %val2)
7235 <i>; yields {i32}:result2 = 8</i>
7237 %stored2 = icmp eq i32 %result2, 8 <i>; yields {i1}:stored2 = true</i>
7238 %memval2 = load i32* %ptr <i>; yields {i32}:memval2 = 2</i>
7243 <!-- _______________________________________________________________________ -->
7244 <div class="doc_subsubsection">
7245 <a name="int_atomic_load_add">'<tt>llvm.atomic.load.add.*</tt>' Intrinsic</a>
7249 <div class="doc_text">
7252 <p>This is an overloaded intrinsic. You can use <tt>llvm.atomic.load.add</tt> on
7253 any integer bit width. Not all targets support all bit widths however.</p>
7256 declare i8 @llvm.atomic.load.add.i8.p0i8(i8* <ptr>, i8 <delta>)
7257 declare i16 @llvm.atomic.load.add.i16.p0i16(i16* <ptr>, i16 <delta>)
7258 declare i32 @llvm.atomic.load.add.i32.p0i32(i32* <ptr>, i32 <delta>)
7259 declare i64 @llvm.atomic.load.add.i64.p0i64(i64* <ptr>, i64 <delta>)
7263 <p>This intrinsic adds <tt>delta</tt> to the value stored in memory
7264 at <tt>ptr</tt>. It yields the original value at <tt>ptr</tt>.</p>
7267 <p>The intrinsic takes two arguments, the first a pointer to an integer value
7268 and the second an integer value. The result is also an integer value. These
7269 integer types can have any bit width, but they must all have the same bit
7270 width. The targets may only lower integer representations they support.</p>
7273 <p>This intrinsic does a series of operations atomically. It first loads the
7274 value stored at <tt>ptr</tt>. It then adds <tt>delta</tt>, stores the result
7275 to <tt>ptr</tt>. It yields the original value stored at <tt>ptr</tt>.</p>
7279 %mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
7280 %ptr = bitcast i8* %mallocP to i32*
7282 %result1 = call i32 @llvm.atomic.load.add.i32.p0i32(i32* %ptr, i32 4)
7283 <i>; yields {i32}:result1 = 4</i>
7284 %result2 = call i32 @llvm.atomic.load.add.i32.p0i32(i32* %ptr, i32 2)
7285 <i>; yields {i32}:result2 = 8</i>
7286 %result3 = call i32 @llvm.atomic.load.add.i32.p0i32(i32* %ptr, i32 5)
7287 <i>; yields {i32}:result3 = 10</i>
7288 %memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 15</i>
7293 <!-- _______________________________________________________________________ -->
7294 <div class="doc_subsubsection">
7295 <a name="int_atomic_load_sub">'<tt>llvm.atomic.load.sub.*</tt>' Intrinsic</a>
7299 <div class="doc_text">
7302 <p>This is an overloaded intrinsic. You can use <tt>llvm.atomic.load.sub</tt> on
7303 any integer bit width and for different address spaces. Not all targets
7304 support all bit widths however.</p>
7307 declare i8 @llvm.atomic.load.sub.i8.p0i32(i8* <ptr>, i8 <delta>)
7308 declare i16 @llvm.atomic.load.sub.i16.p0i32(i16* <ptr>, i16 <delta>)
7309 declare i32 @llvm.atomic.load.sub.i32.p0i32(i32* <ptr>, i32 <delta>)
7310 declare i64 @llvm.atomic.load.sub.i64.p0i32(i64* <ptr>, i64 <delta>)
7314 <p>This intrinsic subtracts <tt>delta</tt> to the value stored in memory at
7315 <tt>ptr</tt>. It yields the original value at <tt>ptr</tt>.</p>
7318 <p>The intrinsic takes two arguments, the first a pointer to an integer value
7319 and the second an integer value. The result is also an integer value. These
7320 integer types can have any bit width, but they must all have the same bit
7321 width. The targets may only lower integer representations they support.</p>
7324 <p>This intrinsic does a series of operations atomically. It first loads the
7325 value stored at <tt>ptr</tt>. It then subtracts <tt>delta</tt>, stores the
7326 result to <tt>ptr</tt>. It yields the original value stored
7327 at <tt>ptr</tt>.</p>
7331 %mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
7332 %ptr = bitcast i8* %mallocP to i32*
7334 %result1 = call i32 @llvm.atomic.load.sub.i32.p0i32(i32* %ptr, i32 4)
7335 <i>; yields {i32}:result1 = 8</i>
7336 %result2 = call i32 @llvm.atomic.load.sub.i32.p0i32(i32* %ptr, i32 2)
7337 <i>; yields {i32}:result2 = 4</i>
7338 %result3 = call i32 @llvm.atomic.load.sub.i32.p0i32(i32* %ptr, i32 5)
7339 <i>; yields {i32}:result3 = 2</i>
7340 %memval1 = load i32* %ptr <i>; yields {i32}:memval1 = -3</i>
7345 <!-- _______________________________________________________________________ -->
7346 <div class="doc_subsubsection">
7347 <a name="int_atomic_load_and">'<tt>llvm.atomic.load.and.*</tt>' Intrinsic</a><br>
7348 <a name="int_atomic_load_nand">'<tt>llvm.atomic.load.nand.*</tt>' Intrinsic</a><br>
7349 <a name="int_atomic_load_or">'<tt>llvm.atomic.load.or.*</tt>' Intrinsic</a><br>
7350 <a name="int_atomic_load_xor">'<tt>llvm.atomic.load.xor.*</tt>' Intrinsic</a><br>
7353 <div class="doc_text">
7356 <p>These are overloaded intrinsics. You can
7357 use <tt>llvm.atomic.load_and</tt>, <tt>llvm.atomic.load_nand</tt>,
7358 <tt>llvm.atomic.load_or</tt>, and <tt>llvm.atomic.load_xor</tt> on any integer
7359 bit width and for different address spaces. Not all targets support all bit
7363 declare i8 @llvm.atomic.load.and.i8.p0i8(i8* <ptr>, i8 <delta>)
7364 declare i16 @llvm.atomic.load.and.i16.p0i16(i16* <ptr>, i16 <delta>)
7365 declare i32 @llvm.atomic.load.and.i32.p0i32(i32* <ptr>, i32 <delta>)
7366 declare i64 @llvm.atomic.load.and.i64.p0i64(i64* <ptr>, i64 <delta>)
7370 declare i8 @llvm.atomic.load.or.i8.p0i8(i8* <ptr>, i8 <delta>)
7371 declare i16 @llvm.atomic.load.or.i16.p0i16(i16* <ptr>, i16 <delta>)
7372 declare i32 @llvm.atomic.load.or.i32.p0i32(i32* <ptr>, i32 <delta>)
7373 declare i64 @llvm.atomic.load.or.i64.p0i64(i64* <ptr>, i64 <delta>)
7377 declare i8 @llvm.atomic.load.nand.i8.p0i32(i8* <ptr>, i8 <delta>)
7378 declare i16 @llvm.atomic.load.nand.i16.p0i32(i16* <ptr>, i16 <delta>)
7379 declare i32 @llvm.atomic.load.nand.i32.p0i32(i32* <ptr>, i32 <delta>)
7380 declare i64 @llvm.atomic.load.nand.i64.p0i32(i64* <ptr>, i64 <delta>)
7384 declare i8 @llvm.atomic.load.xor.i8.p0i32(i8* <ptr>, i8 <delta>)
7385 declare i16 @llvm.atomic.load.xor.i16.p0i32(i16* <ptr>, i16 <delta>)
7386 declare i32 @llvm.atomic.load.xor.i32.p0i32(i32* <ptr>, i32 <delta>)
7387 declare i64 @llvm.atomic.load.xor.i64.p0i32(i64* <ptr>, i64 <delta>)
7391 <p>These intrinsics bitwise the operation (and, nand, or, xor) <tt>delta</tt> to
7392 the value stored in memory at <tt>ptr</tt>. It yields the original value
7393 at <tt>ptr</tt>.</p>
7396 <p>These intrinsics take two arguments, the first a pointer to an integer value
7397 and the second an integer value. The result is also an integer value. These
7398 integer types can have any bit width, but they must all have the same bit
7399 width. The targets may only lower integer representations they support.</p>
7402 <p>These intrinsics does a series of operations atomically. They first load the
7403 value stored at <tt>ptr</tt>. They then do the bitwise
7404 operation <tt>delta</tt>, store the result to <tt>ptr</tt>. They yield the
7405 original value stored at <tt>ptr</tt>.</p>
7409 %mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
7410 %ptr = bitcast i8* %mallocP to i32*
7411 store i32 0x0F0F, %ptr
7412 %result0 = call i32 @llvm.atomic.load.nand.i32.p0i32(i32* %ptr, i32 0xFF)
7413 <i>; yields {i32}:result0 = 0x0F0F</i>
7414 %result1 = call i32 @llvm.atomic.load.and.i32.p0i32(i32* %ptr, i32 0xFF)
7415 <i>; yields {i32}:result1 = 0xFFFFFFF0</i>
7416 %result2 = call i32 @llvm.atomic.load.or.i32.p0i32(i32* %ptr, i32 0F)
7417 <i>; yields {i32}:result2 = 0xF0</i>
7418 %result3 = call i32 @llvm.atomic.load.xor.i32.p0i32(i32* %ptr, i32 0F)
7419 <i>; yields {i32}:result3 = FF</i>
7420 %memval1 = load i32* %ptr <i>; yields {i32}:memval1 = F0</i>
7425 <!-- _______________________________________________________________________ -->
7426 <div class="doc_subsubsection">
7427 <a name="int_atomic_load_max">'<tt>llvm.atomic.load.max.*</tt>' Intrinsic</a><br>
7428 <a name="int_atomic_load_min">'<tt>llvm.atomic.load.min.*</tt>' Intrinsic</a><br>
7429 <a name="int_atomic_load_umax">'<tt>llvm.atomic.load.umax.*</tt>' Intrinsic</a><br>
7430 <a name="int_atomic_load_umin">'<tt>llvm.atomic.load.umin.*</tt>' Intrinsic</a><br>
7433 <div class="doc_text">
7436 <p>These are overloaded intrinsics. You can use <tt>llvm.atomic.load_max</tt>,
7437 <tt>llvm.atomic.load_min</tt>, <tt>llvm.atomic.load_umax</tt>, and
7438 <tt>llvm.atomic.load_umin</tt> on any integer bit width and for different
7439 address spaces. Not all targets support all bit widths however.</p>
7442 declare i8 @llvm.atomic.load.max.i8.p0i8(i8* <ptr>, i8 <delta>)
7443 declare i16 @llvm.atomic.load.max.i16.p0i16(i16* <ptr>, i16 <delta>)
7444 declare i32 @llvm.atomic.load.max.i32.p0i32(i32* <ptr>, i32 <delta>)
7445 declare i64 @llvm.atomic.load.max.i64.p0i64(i64* <ptr>, i64 <delta>)
7449 declare i8 @llvm.atomic.load.min.i8.p0i8(i8* <ptr>, i8 <delta>)
7450 declare i16 @llvm.atomic.load.min.i16.p0i16(i16* <ptr>, i16 <delta>)
7451 declare i32 @llvm.atomic.load.min.i32.p0i32(i32* <ptr>, i32 <delta>)
7452 declare i64 @llvm.atomic.load.min.i64.p0i64(i64* <ptr>, i64 <delta>)
7456 declare i8 @llvm.atomic.load.umax.i8.p0i8(i8* <ptr>, i8 <delta>)
7457 declare i16 @llvm.atomic.load.umax.i16.p0i16(i16* <ptr>, i16 <delta>)
7458 declare i32 @llvm.atomic.load.umax.i32.p0i32(i32* <ptr>, i32 <delta>)
7459 declare i64 @llvm.atomic.load.umax.i64.p0i64(i64* <ptr>, i64 <delta>)
7463 declare i8 @llvm.atomic.load.umin.i8.p0i8(i8* <ptr>, i8 <delta>)
7464 declare i16 @llvm.atomic.load.umin.i16.p0i16(i16* <ptr>, i16 <delta>)
7465 declare i32 @llvm.atomic.load.umin.i32.p0i32(i32* <ptr>, i32 <delta>)
7466 declare i64 @llvm.atomic.load.umin.i64.p0i64(i64* <ptr>, i64 <delta>)
7470 <p>These intrinsics takes the signed or unsigned minimum or maximum of
7471 <tt>delta</tt> and the value stored in memory at <tt>ptr</tt>. It yields the
7472 original value at <tt>ptr</tt>.</p>
7475 <p>These intrinsics take two arguments, the first a pointer to an integer value
7476 and the second an integer value. The result is also an integer value. These
7477 integer types can have any bit width, but they must all have the same bit
7478 width. The targets may only lower integer representations they support.</p>
7481 <p>These intrinsics does a series of operations atomically. They first load the
7482 value stored at <tt>ptr</tt>. They then do the signed or unsigned min or
7483 max <tt>delta</tt> and the value, store the result to <tt>ptr</tt>. They
7484 yield the original value stored at <tt>ptr</tt>.</p>
7488 %mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
7489 %ptr = bitcast i8* %mallocP to i32*
7491 %result0 = call i32 @llvm.atomic.load.min.i32.p0i32(i32* %ptr, i32 -2)
7492 <i>; yields {i32}:result0 = 7</i>
7493 %result1 = call i32 @llvm.atomic.load.max.i32.p0i32(i32* %ptr, i32 8)
7494 <i>; yields {i32}:result1 = -2</i>
7495 %result2 = call i32 @llvm.atomic.load.umin.i32.p0i32(i32* %ptr, i32 10)
7496 <i>; yields {i32}:result2 = 8</i>
7497 %result3 = call i32 @llvm.atomic.load.umax.i32.p0i32(i32* %ptr, i32 30)
7498 <i>; yields {i32}:result3 = 8</i>
7499 %memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 30</i>
7505 <!-- ======================================================================= -->
7506 <div class="doc_subsection">
7507 <a name="int_memorymarkers">Memory Use Markers</a>
7510 <div class="doc_text">
7512 <p>This class of intrinsics exists to information about the lifetime of memory
7513 objects and ranges where variables are immutable.</p>
7517 <!-- _______________________________________________________________________ -->
7518 <div class="doc_subsubsection">
7519 <a name="int_lifetime_start">'<tt>llvm.lifetime.start</tt>' Intrinsic</a>
7522 <div class="doc_text">
7526 declare void @llvm.lifetime.start(i64 <size>, i8* nocapture <ptr>)
7530 <p>The '<tt>llvm.lifetime.start</tt>' intrinsic specifies the start of a memory
7531 object's lifetime.</p>
7534 <p>The first argument is a constant integer representing the size of the
7535 object, or -1 if it is variable sized. The second argument is a pointer to
7539 <p>This intrinsic indicates that before this point in the code, the value of the
7540 memory pointed to by <tt>ptr</tt> is dead. This means that it is known to
7541 never be used and has an undefined value. A load from the pointer that
7542 precedes this intrinsic can be replaced with
7543 <tt>'<a href="#undefvalues">undef</a>'</tt>.</p>
7547 <!-- _______________________________________________________________________ -->
7548 <div class="doc_subsubsection">
7549 <a name="int_lifetime_end">'<tt>llvm.lifetime.end</tt>' Intrinsic</a>
7552 <div class="doc_text">
7556 declare void @llvm.lifetime.end(i64 <size>, i8* nocapture <ptr>)
7560 <p>The '<tt>llvm.lifetime.end</tt>' intrinsic specifies the end of a memory
7561 object's lifetime.</p>
7564 <p>The first argument is a constant integer representing the size of the
7565 object, or -1 if it is variable sized. The second argument is a pointer to
7569 <p>This intrinsic indicates that after this point in the code, the value of the
7570 memory pointed to by <tt>ptr</tt> is dead. This means that it is known to
7571 never be used and has an undefined value. Any stores into the memory object
7572 following this intrinsic may be removed as dead.
7576 <!-- _______________________________________________________________________ -->
7577 <div class="doc_subsubsection">
7578 <a name="int_invariant_start">'<tt>llvm.invariant.start</tt>' Intrinsic</a>
7581 <div class="doc_text">
7585 declare {}* @llvm.invariant.start(i64 <size>, i8* nocapture <ptr>) readonly
7589 <p>The '<tt>llvm.invariant.start</tt>' intrinsic specifies that the contents of
7590 a memory object will not change.</p>
7593 <p>The first argument is a constant integer representing the size of the
7594 object, or -1 if it is variable sized. The second argument is a pointer to
7598 <p>This intrinsic indicates that until an <tt>llvm.invariant.end</tt> that uses
7599 the return value, the referenced memory location is constant and
7604 <!-- _______________________________________________________________________ -->
7605 <div class="doc_subsubsection">
7606 <a name="int_invariant_end">'<tt>llvm.invariant.end</tt>' Intrinsic</a>
7609 <div class="doc_text">
7613 declare void @llvm.invariant.end({}* <start>, i64 <size>, i8* nocapture <ptr>)
7617 <p>The '<tt>llvm.invariant.end</tt>' intrinsic specifies that the contents of
7618 a memory object are mutable.</p>
7621 <p>The first argument is the matching <tt>llvm.invariant.start</tt> intrinsic.
7622 The second argument is a constant integer representing the size of the
7623 object, or -1 if it is variable sized and the third argument is a pointer
7627 <p>This intrinsic indicates that the memory is mutable again.</p>
7631 <!-- ======================================================================= -->
7632 <div class="doc_subsection">
7633 <a name="int_general">General Intrinsics</a>
7636 <div class="doc_text">
7638 <p>This class of intrinsics is designed to be generic and has no specific
7643 <!-- _______________________________________________________________________ -->
7644 <div class="doc_subsubsection">
7645 <a name="int_var_annotation">'<tt>llvm.var.annotation</tt>' Intrinsic</a>
7648 <div class="doc_text">
7652 declare void @llvm.var.annotation(i8* <val>, i8* <str>, i8* <str>, i32 <int>)
7656 <p>The '<tt>llvm.var.annotation</tt>' intrinsic.</p>
7659 <p>The first argument is a pointer to a value, the second is a pointer to a
7660 global string, the third is a pointer to a global string which is the source
7661 file name, and the last argument is the line number.</p>
7664 <p>This intrinsic allows annotation of local variables with arbitrary strings.
7665 This can be useful for special purpose optimizations that want to look for
7666 these annotations. These have no other defined use, they are ignored by code
7667 generation and optimization.</p>
7671 <!-- _______________________________________________________________________ -->
7672 <div class="doc_subsubsection">
7673 <a name="int_annotation">'<tt>llvm.annotation.*</tt>' Intrinsic</a>
7676 <div class="doc_text">
7679 <p>This is an overloaded intrinsic. You can use '<tt>llvm.annotation</tt>' on
7680 any integer bit width.</p>
7683 declare i8 @llvm.annotation.i8(i8 <val>, i8* <str>, i8* <str>, i32 <int>)
7684 declare i16 @llvm.annotation.i16(i16 <val>, i8* <str>, i8* <str>, i32 <int>)
7685 declare i32 @llvm.annotation.i32(i32 <val>, i8* <str>, i8* <str>, i32 <int>)
7686 declare i64 @llvm.annotation.i64(i64 <val>, i8* <str>, i8* <str>, i32 <int>)
7687 declare i256 @llvm.annotation.i256(i256 <val>, i8* <str>, i8* <str>, i32 <int>)
7691 <p>The '<tt>llvm.annotation</tt>' intrinsic.</p>
7694 <p>The first argument is an integer value (result of some expression), the
7695 second is a pointer to a global string, the third is a pointer to a global
7696 string which is the source file name, and the last argument is the line
7697 number. It returns the value of the first argument.</p>
7700 <p>This intrinsic allows annotations to be put on arbitrary expressions with
7701 arbitrary strings. This can be useful for special purpose optimizations that
7702 want to look for these annotations. These have no other defined use, they
7703 are ignored by code generation and optimization.</p>
7707 <!-- _______________________________________________________________________ -->
7708 <div class="doc_subsubsection">
7709 <a name="int_trap">'<tt>llvm.trap</tt>' Intrinsic</a>
7712 <div class="doc_text">
7716 declare void @llvm.trap()
7720 <p>The '<tt>llvm.trap</tt>' intrinsic.</p>
7726 <p>This intrinsics is lowered to the target dependent trap instruction. If the
7727 target does not have a trap instruction, this intrinsic will be lowered to
7728 the call of the <tt>abort()</tt> function.</p>
7732 <!-- _______________________________________________________________________ -->
7733 <div class="doc_subsubsection">
7734 <a name="int_stackprotector">'<tt>llvm.stackprotector</tt>' Intrinsic</a>
7737 <div class="doc_text">
7741 declare void @llvm.stackprotector(i8* <guard>, i8** <slot>)
7745 <p>The <tt>llvm.stackprotector</tt> intrinsic takes the <tt>guard</tt> and
7746 stores it onto the stack at <tt>slot</tt>. The stack slot is adjusted to
7747 ensure that it is placed on the stack before local variables.</p>
7750 <p>The <tt>llvm.stackprotector</tt> intrinsic requires two pointer
7751 arguments. The first argument is the value loaded from the stack
7752 guard <tt>@__stack_chk_guard</tt>. The second variable is an <tt>alloca</tt>
7753 that has enough space to hold the value of the guard.</p>
7756 <p>This intrinsic causes the prologue/epilogue inserter to force the position of
7757 the <tt>AllocaInst</tt> stack slot to be before local variables on the
7758 stack. This is to ensure that if a local variable on the stack is
7759 overwritten, it will destroy the value of the guard. When the function exits,
7760 the guard on the stack is checked against the original guard. If they're
7761 different, then the program aborts by calling the <tt>__stack_chk_fail()</tt>
7766 <!-- _______________________________________________________________________ -->
7767 <div class="doc_subsubsection">
7768 <a name="int_objectsize">'<tt>llvm.objectsize</tt>' Intrinsic</a>
7771 <div class="doc_text">
7775 declare i32 @llvm.objectsize.i32(i8* <object>, i1 <type>)
7776 declare i64 @llvm.objectsize.i64(i8* <object>, i1 <type>)
7780 <p>The <tt>llvm.objectsize</tt> intrinsic is designed to provide information
7781 to the optimizers to discover at compile time either a) when an
7782 operation like memcpy will either overflow a buffer that corresponds to
7783 an object, or b) to determine that a runtime check for overflow isn't
7784 necessary. An object in this context means an allocation of a
7785 specific class, structure, array, or other object.</p>
7788 <p>The <tt>llvm.objectsize</tt> intrinsic takes two arguments. The first
7789 argument is a pointer to or into the <tt>object</tt>. The second argument
7790 is a boolean 0 or 1. This argument determines whether you want the
7791 maximum (0) or minimum (1) bytes remaining. This needs to be a literal 0 or
7792 1, variables are not allowed.</p>
7795 <p>The <tt>llvm.objectsize</tt> intrinsic is lowered to either a constant
7796 representing the size of the object concerned or <tt>i32/i64 -1 or 0</tt>
7797 (depending on the <tt>type</tt> argument if the size cannot be determined
7798 at compile time.</p>
7802 <!-- *********************************************************************** -->
7805 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
7806 src="http://jigsaw.w3.org/css-validator/images/vcss-blue" alt="Valid CSS"></a>
7807 <a href="http://validator.w3.org/check/referer"><img
7808 src="http://www.w3.org/Icons/valid-html401-blue" alt="Valid HTML 4.01"></a>
7810 <a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
7811 <a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br>
7812 Last modified: $Date$