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 <h1>LLVM Language Reference Manual</h1>
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_linker_private_weak_def_auto">'<tt>linker_private_weak_def_auto</tt>' Linkage</a></li>
29 <li><a href="#linkage_internal">'<tt>internal</tt>' Linkage</a></li>
30 <li><a href="#linkage_available_externally">'<tt>available_externally</tt>' Linkage</a></li>
31 <li><a href="#linkage_linkonce">'<tt>linkonce</tt>' Linkage</a></li>
32 <li><a href="#linkage_common">'<tt>common</tt>' Linkage</a></li>
33 <li><a href="#linkage_weak">'<tt>weak</tt>' Linkage</a></li>
34 <li><a href="#linkage_appending">'<tt>appending</tt>' Linkage</a></li>
35 <li><a href="#linkage_externweak">'<tt>extern_weak</tt>' Linkage</a></li>
36 <li><a href="#linkage_linkonce_odr">'<tt>linkonce_odr</tt>' Linkage</a></li>
37 <li><a href="#linkage_weak">'<tt>weak_odr</tt>' Linkage</a></li>
38 <li><a href="#linkage_external">'<tt>external</tt>' Linkage</a></li>
39 <li><a href="#linkage_dllimport">'<tt>dllimport</tt>' Linkage</a></li>
40 <li><a href="#linkage_dllexport">'<tt>dllexport</tt>' Linkage</a></li>
43 <li><a href="#callingconv">Calling Conventions</a></li>
44 <li><a href="#namedtypes">Named Types</a></li>
45 <li><a href="#globalvars">Global Variables</a></li>
46 <li><a href="#functionstructure">Functions</a></li>
47 <li><a href="#aliasstructure">Aliases</a></li>
48 <li><a href="#namedmetadatastructure">Named Metadata</a></li>
49 <li><a href="#paramattrs">Parameter Attributes</a></li>
50 <li><a href="#fnattrs">Function Attributes</a></li>
51 <li><a href="#gc">Garbage Collector Names</a></li>
52 <li><a href="#moduleasm">Module-Level Inline Assembly</a></li>
53 <li><a href="#datalayout">Data Layout</a></li>
54 <li><a href="#pointeraliasing">Pointer Aliasing Rules</a></li>
55 <li><a href="#volatile">Volatile Memory Accesses</a></li>
56 <li><a href="#memmodel">Memory Model for Concurrent Operations</a></li>
57 <li><a href="#ordering">Atomic Memory Ordering Constraints</a></li>
60 <li><a href="#typesystem">Type System</a>
62 <li><a href="#t_classifications">Type Classifications</a></li>
63 <li><a href="#t_primitive">Primitive Types</a>
65 <li><a href="#t_integer">Integer Type</a></li>
66 <li><a href="#t_floating">Floating Point Types</a></li>
67 <li><a href="#t_x86mmx">X86mmx Type</a></li>
68 <li><a href="#t_void">Void Type</a></li>
69 <li><a href="#t_label">Label Type</a></li>
70 <li><a href="#t_metadata">Metadata Type</a></li>
73 <li><a href="#t_derived">Derived Types</a>
75 <li><a href="#t_aggregate">Aggregate Types</a>
77 <li><a href="#t_array">Array Type</a></li>
78 <li><a href="#t_struct">Structure Type</a></li>
79 <li><a href="#t_opaque">Opaque Structure Types</a></li>
80 <li><a href="#t_vector">Vector Type</a></li>
83 <li><a href="#t_function">Function Type</a></li>
84 <li><a href="#t_pointer">Pointer Type</a></li>
89 <li><a href="#constants">Constants</a>
91 <li><a href="#simpleconstants">Simple Constants</a></li>
92 <li><a href="#complexconstants">Complex Constants</a></li>
93 <li><a href="#globalconstants">Global Variable and Function Addresses</a></li>
94 <li><a href="#undefvalues">Undefined Values</a></li>
95 <li><a href="#trapvalues">Trap Values</a></li>
96 <li><a href="#blockaddress">Addresses of Basic Blocks</a></li>
97 <li><a href="#constantexprs">Constant Expressions</a></li>
100 <li><a href="#othervalues">Other Values</a>
102 <li><a href="#inlineasm">Inline Assembler Expressions</a></li>
103 <li><a href="#metadata">Metadata Nodes and Metadata Strings</a></li>
106 <li><a href="#intrinsic_globals">Intrinsic Global Variables</a>
108 <li><a href="#intg_used">The '<tt>llvm.used</tt>' Global Variable</a></li>
109 <li><a href="#intg_compiler_used">The '<tt>llvm.compiler.used</tt>'
110 Global Variable</a></li>
111 <li><a href="#intg_global_ctors">The '<tt>llvm.global_ctors</tt>'
112 Global Variable</a></li>
113 <li><a href="#intg_global_dtors">The '<tt>llvm.global_dtors</tt>'
114 Global Variable</a></li>
117 <li><a href="#instref">Instruction Reference</a>
119 <li><a href="#terminators">Terminator Instructions</a>
121 <li><a href="#i_ret">'<tt>ret</tt>' Instruction</a></li>
122 <li><a href="#i_br">'<tt>br</tt>' Instruction</a></li>
123 <li><a href="#i_switch">'<tt>switch</tt>' Instruction</a></li>
124 <li><a href="#i_indirectbr">'<tt>indirectbr</tt>' Instruction</a></li>
125 <li><a href="#i_invoke">'<tt>invoke</tt>' Instruction</a></li>
126 <li><a href="#i_unwind">'<tt>unwind</tt>' Instruction</a></li>
127 <li><a href="#i_resume">'<tt>resume</tt>' Instruction</a></li>
128 <li><a href="#i_unreachable">'<tt>unreachable</tt>' Instruction</a></li>
131 <li><a href="#binaryops">Binary Operations</a>
133 <li><a href="#i_add">'<tt>add</tt>' Instruction</a></li>
134 <li><a href="#i_fadd">'<tt>fadd</tt>' Instruction</a></li>
135 <li><a href="#i_sub">'<tt>sub</tt>' Instruction</a></li>
136 <li><a href="#i_fsub">'<tt>fsub</tt>' Instruction</a></li>
137 <li><a href="#i_mul">'<tt>mul</tt>' Instruction</a></li>
138 <li><a href="#i_fmul">'<tt>fmul</tt>' Instruction</a></li>
139 <li><a href="#i_udiv">'<tt>udiv</tt>' Instruction</a></li>
140 <li><a href="#i_sdiv">'<tt>sdiv</tt>' Instruction</a></li>
141 <li><a href="#i_fdiv">'<tt>fdiv</tt>' Instruction</a></li>
142 <li><a href="#i_urem">'<tt>urem</tt>' Instruction</a></li>
143 <li><a href="#i_srem">'<tt>srem</tt>' Instruction</a></li>
144 <li><a href="#i_frem">'<tt>frem</tt>' Instruction</a></li>
147 <li><a href="#bitwiseops">Bitwise Binary Operations</a>
149 <li><a href="#i_shl">'<tt>shl</tt>' Instruction</a></li>
150 <li><a href="#i_lshr">'<tt>lshr</tt>' Instruction</a></li>
151 <li><a href="#i_ashr">'<tt>ashr</tt>' Instruction</a></li>
152 <li><a href="#i_and">'<tt>and</tt>' Instruction</a></li>
153 <li><a href="#i_or">'<tt>or</tt>' Instruction</a></li>
154 <li><a href="#i_xor">'<tt>xor</tt>' Instruction</a></li>
157 <li><a href="#vectorops">Vector Operations</a>
159 <li><a href="#i_extractelement">'<tt>extractelement</tt>' Instruction</a></li>
160 <li><a href="#i_insertelement">'<tt>insertelement</tt>' Instruction</a></li>
161 <li><a href="#i_shufflevector">'<tt>shufflevector</tt>' Instruction</a></li>
164 <li><a href="#aggregateops">Aggregate Operations</a>
166 <li><a href="#i_extractvalue">'<tt>extractvalue</tt>' Instruction</a></li>
167 <li><a href="#i_insertvalue">'<tt>insertvalue</tt>' Instruction</a></li>
170 <li><a href="#memoryops">Memory Access and Addressing Operations</a>
172 <li><a href="#i_alloca">'<tt>alloca</tt>' Instruction</a></li>
173 <li><a href="#i_load">'<tt>load</tt>' Instruction</a></li>
174 <li><a href="#i_store">'<tt>store</tt>' Instruction</a></li>
175 <li><a href="#i_fence">'<tt>fence</tt>' Instruction</a></li>
176 <li><a href="#i_cmpxchg">'<tt>cmpxchg</tt>' Instruction</a></li>
177 <li><a href="#i_atomicrmw">'<tt>atomicrmw</tt>' Instruction</a></li>
178 <li><a href="#i_getelementptr">'<tt>getelementptr</tt>' Instruction</a></li>
181 <li><a href="#convertops">Conversion Operations</a>
183 <li><a href="#i_trunc">'<tt>trunc .. to</tt>' Instruction</a></li>
184 <li><a href="#i_zext">'<tt>zext .. to</tt>' Instruction</a></li>
185 <li><a href="#i_sext">'<tt>sext .. to</tt>' Instruction</a></li>
186 <li><a href="#i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a></li>
187 <li><a href="#i_fpext">'<tt>fpext .. to</tt>' Instruction</a></li>
188 <li><a href="#i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a></li>
189 <li><a href="#i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a></li>
190 <li><a href="#i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a></li>
191 <li><a href="#i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a></li>
192 <li><a href="#i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a></li>
193 <li><a href="#i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a></li>
194 <li><a href="#i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a></li>
197 <li><a href="#otherops">Other Operations</a>
199 <li><a href="#i_icmp">'<tt>icmp</tt>' Instruction</a></li>
200 <li><a href="#i_fcmp">'<tt>fcmp</tt>' Instruction</a></li>
201 <li><a href="#i_phi">'<tt>phi</tt>' Instruction</a></li>
202 <li><a href="#i_select">'<tt>select</tt>' Instruction</a></li>
203 <li><a href="#i_call">'<tt>call</tt>' Instruction</a></li>
204 <li><a href="#i_va_arg">'<tt>va_arg</tt>' Instruction</a></li>
205 <li><a href="#i_landingpad">'<tt>landingpad</tt>' Instruction</a></li>
210 <li><a href="#intrinsics">Intrinsic Functions</a>
212 <li><a href="#int_varargs">Variable Argument Handling Intrinsics</a>
214 <li><a href="#int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a></li>
215 <li><a href="#int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a></li>
216 <li><a href="#int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a></li>
219 <li><a href="#int_gc">Accurate Garbage Collection Intrinsics</a>
221 <li><a href="#int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a></li>
222 <li><a href="#int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a></li>
223 <li><a href="#int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a></li>
226 <li><a href="#int_codegen">Code Generator Intrinsics</a>
228 <li><a href="#int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a></li>
229 <li><a href="#int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a></li>
230 <li><a href="#int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a></li>
231 <li><a href="#int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a></li>
232 <li><a href="#int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a></li>
233 <li><a href="#int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a></li>
234 <li><a href="#int_readcyclecounter">'<tt>llvm.readcyclecounter</tt>' Intrinsic</a></li>
237 <li><a href="#int_libc">Standard C Library Intrinsics</a>
239 <li><a href="#int_memcpy">'<tt>llvm.memcpy.*</tt>' Intrinsic</a></li>
240 <li><a href="#int_memmove">'<tt>llvm.memmove.*</tt>' Intrinsic</a></li>
241 <li><a href="#int_memset">'<tt>llvm.memset.*</tt>' Intrinsic</a></li>
242 <li><a href="#int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a></li>
243 <li><a href="#int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a></li>
244 <li><a href="#int_sin">'<tt>llvm.sin.*</tt>' Intrinsic</a></li>
245 <li><a href="#int_cos">'<tt>llvm.cos.*</tt>' Intrinsic</a></li>
246 <li><a href="#int_pow">'<tt>llvm.pow.*</tt>' Intrinsic</a></li>
247 <li><a href="#int_exp">'<tt>llvm.exp.*</tt>' Intrinsic</a></li>
248 <li><a href="#int_log">'<tt>llvm.log.*</tt>' Intrinsic</a></li>
249 <li><a href="#int_fma">'<tt>llvm.fma.*</tt>' Intrinsic</a></li>
252 <li><a href="#int_manip">Bit Manipulation Intrinsics</a>
254 <li><a href="#int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a></li>
255 <li><a href="#int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic </a></li>
256 <li><a href="#int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic </a></li>
257 <li><a href="#int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic </a></li>
260 <li><a href="#int_overflow">Arithmetic with Overflow Intrinsics</a>
262 <li><a href="#int_sadd_overflow">'<tt>llvm.sadd.with.overflow.*</tt> Intrinsics</a></li>
263 <li><a href="#int_uadd_overflow">'<tt>llvm.uadd.with.overflow.*</tt> Intrinsics</a></li>
264 <li><a href="#int_ssub_overflow">'<tt>llvm.ssub.with.overflow.*</tt> Intrinsics</a></li>
265 <li><a href="#int_usub_overflow">'<tt>llvm.usub.with.overflow.*</tt> Intrinsics</a></li>
266 <li><a href="#int_smul_overflow">'<tt>llvm.smul.with.overflow.*</tt> Intrinsics</a></li>
267 <li><a href="#int_umul_overflow">'<tt>llvm.umul.with.overflow.*</tt> Intrinsics</a></li>
270 <li><a href="#int_fp16">Half Precision Floating Point Intrinsics</a>
272 <li><a href="#int_convert_to_fp16">'<tt>llvm.convert.to.fp16</tt>' Intrinsic</a></li>
273 <li><a href="#int_convert_from_fp16">'<tt>llvm.convert.from.fp16</tt>' Intrinsic</a></li>
276 <li><a href="#int_debugger">Debugger intrinsics</a></li>
277 <li><a href="#int_eh">Exception Handling intrinsics</a></li>
278 <li><a href="#int_trampoline">Trampoline Intrinsics</a>
280 <li><a href="#int_it">'<tt>llvm.init.trampoline</tt>' Intrinsic</a></li>
281 <li><a href="#int_at">'<tt>llvm.adjust.trampoline</tt>' Intrinsic</a></li>
284 <li><a href="#int_atomics">Atomic intrinsics</a>
286 <li><a href="#int_memory_barrier"><tt>llvm.memory_barrier</tt></a></li>
287 <li><a href="#int_atomic_cmp_swap"><tt>llvm.atomic.cmp.swap</tt></a></li>
288 <li><a href="#int_atomic_swap"><tt>llvm.atomic.swap</tt></a></li>
289 <li><a href="#int_atomic_load_add"><tt>llvm.atomic.load.add</tt></a></li>
290 <li><a href="#int_atomic_load_sub"><tt>llvm.atomic.load.sub</tt></a></li>
291 <li><a href="#int_atomic_load_and"><tt>llvm.atomic.load.and</tt></a></li>
292 <li><a href="#int_atomic_load_nand"><tt>llvm.atomic.load.nand</tt></a></li>
293 <li><a href="#int_atomic_load_or"><tt>llvm.atomic.load.or</tt></a></li>
294 <li><a href="#int_atomic_load_xor"><tt>llvm.atomic.load.xor</tt></a></li>
295 <li><a href="#int_atomic_load_max"><tt>llvm.atomic.load.max</tt></a></li>
296 <li><a href="#int_atomic_load_min"><tt>llvm.atomic.load.min</tt></a></li>
297 <li><a href="#int_atomic_load_umax"><tt>llvm.atomic.load.umax</tt></a></li>
298 <li><a href="#int_atomic_load_umin"><tt>llvm.atomic.load.umin</tt></a></li>
301 <li><a href="#int_memorymarkers">Memory Use Markers</a>
303 <li><a href="#int_lifetime_start"><tt>llvm.lifetime.start</tt></a></li>
304 <li><a href="#int_lifetime_end"><tt>llvm.lifetime.end</tt></a></li>
305 <li><a href="#int_invariant_start"><tt>llvm.invariant.start</tt></a></li>
306 <li><a href="#int_invariant_end"><tt>llvm.invariant.end</tt></a></li>
309 <li><a href="#int_general">General intrinsics</a>
311 <li><a href="#int_var_annotation">
312 '<tt>llvm.var.annotation</tt>' Intrinsic</a></li>
313 <li><a href="#int_annotation">
314 '<tt>llvm.annotation.*</tt>' Intrinsic</a></li>
315 <li><a href="#int_trap">
316 '<tt>llvm.trap</tt>' Intrinsic</a></li>
317 <li><a href="#int_stackprotector">
318 '<tt>llvm.stackprotector</tt>' Intrinsic</a></li>
319 <li><a href="#int_objectsize">
320 '<tt>llvm.objectsize</tt>' Intrinsic</a></li>
327 <div class="doc_author">
328 <p>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a>
329 and <a href="mailto:vadve@cs.uiuc.edu">Vikram Adve</a></p>
332 <!-- *********************************************************************** -->
333 <h2><a name="abstract">Abstract</a></h2>
334 <!-- *********************************************************************** -->
338 <p>This document is a reference manual for the LLVM assembly language. LLVM is
339 a Static Single Assignment (SSA) based representation that provides type
340 safety, low-level operations, flexibility, and the capability of representing
341 'all' high-level languages cleanly. It is the common code representation
342 used throughout all phases of the LLVM compilation strategy.</p>
346 <!-- *********************************************************************** -->
347 <h2><a name="introduction">Introduction</a></h2>
348 <!-- *********************************************************************** -->
352 <p>The LLVM code representation is designed to be used in three different forms:
353 as an in-memory compiler IR, as an on-disk bitcode representation (suitable
354 for fast loading by a Just-In-Time compiler), and as a human readable
355 assembly language representation. This allows LLVM to provide a powerful
356 intermediate representation for efficient compiler transformations and
357 analysis, while providing a natural means to debug and visualize the
358 transformations. The three different forms of LLVM are all equivalent. This
359 document describes the human readable representation and notation.</p>
361 <p>The LLVM representation aims to be light-weight and low-level while being
362 expressive, typed, and extensible at the same time. It aims to be a
363 "universal IR" of sorts, by being at a low enough level that high-level ideas
364 may be cleanly mapped to it (similar to how microprocessors are "universal
365 IR's", allowing many source languages to be mapped to them). By providing
366 type information, LLVM can be used as the target of optimizations: for
367 example, through pointer analysis, it can be proven that a C automatic
368 variable is never accessed outside of the current function, allowing it to
369 be promoted to a simple SSA value instead of a memory location.</p>
371 <!-- _______________________________________________________________________ -->
373 <a name="wellformed">Well-Formedness</a>
378 <p>It is important to note that this document describes 'well formed' LLVM
379 assembly language. There is a difference between what the parser accepts and
380 what is considered 'well formed'. For example, the following instruction is
381 syntactically okay, but not well formed:</p>
383 <pre class="doc_code">
384 %x = <a href="#i_add">add</a> i32 1, %x
387 <p>because the definition of <tt>%x</tt> does not dominate all of its uses. The
388 LLVM infrastructure provides a verification pass that may be used to verify
389 that an LLVM module is well formed. This pass is automatically run by the
390 parser after parsing input assembly and by the optimizer before it outputs
391 bitcode. The violations pointed out by the verifier pass indicate bugs in
392 transformation passes or input to the parser.</p>
398 <!-- Describe the typesetting conventions here. -->
400 <!-- *********************************************************************** -->
401 <h2><a name="identifiers">Identifiers</a></h2>
402 <!-- *********************************************************************** -->
406 <p>LLVM identifiers come in two basic types: global and local. Global
407 identifiers (functions, global variables) begin with the <tt>'@'</tt>
408 character. Local identifiers (register names, types) begin with
409 the <tt>'%'</tt> character. Additionally, there are three different formats
410 for identifiers, for different purposes:</p>
413 <li>Named values are represented as a string of characters with their prefix.
414 For example, <tt>%foo</tt>, <tt>@DivisionByZero</tt>,
415 <tt>%a.really.long.identifier</tt>. The actual regular expression used is
416 '<tt>[%@][a-zA-Z$._][a-zA-Z$._0-9]*</tt>'. Identifiers which require
417 other characters in their names can be surrounded with quotes. Special
418 characters may be escaped using <tt>"\xx"</tt> where <tt>xx</tt> is the
419 ASCII code for the character in hexadecimal. In this way, any character
420 can be used in a name value, even quotes themselves.</li>
422 <li>Unnamed values are represented as an unsigned numeric value with their
423 prefix. For example, <tt>%12</tt>, <tt>@2</tt>, <tt>%44</tt>.</li>
425 <li>Constants, which are described in a <a href="#constants">section about
426 constants</a>, below.</li>
429 <p>LLVM requires that values start with a prefix for two reasons: Compilers
430 don't need to worry about name clashes with reserved words, and the set of
431 reserved words may be expanded in the future without penalty. Additionally,
432 unnamed identifiers allow a compiler to quickly come up with a temporary
433 variable without having to avoid symbol table conflicts.</p>
435 <p>Reserved words in LLVM are very similar to reserved words in other
436 languages. There are keywords for different opcodes
437 ('<tt><a href="#i_add">add</a></tt>',
438 '<tt><a href="#i_bitcast">bitcast</a></tt>',
439 '<tt><a href="#i_ret">ret</a></tt>', etc...), for primitive type names
440 ('<tt><a href="#t_void">void</a></tt>',
441 '<tt><a href="#t_primitive">i32</a></tt>', etc...), and others. These
442 reserved words cannot conflict with variable names, because none of them
443 start with a prefix character (<tt>'%'</tt> or <tt>'@'</tt>).</p>
445 <p>Here is an example of LLVM code to multiply the integer variable
446 '<tt>%X</tt>' by 8:</p>
450 <pre class="doc_code">
451 %result = <a href="#i_mul">mul</a> i32 %X, 8
454 <p>After strength reduction:</p>
456 <pre class="doc_code">
457 %result = <a href="#i_shl">shl</a> i32 %X, i8 3
460 <p>And the hard way:</p>
462 <pre class="doc_code">
463 %0 = <a href="#i_add">add</a> i32 %X, %X <i>; yields {i32}:%0</i>
464 %1 = <a href="#i_add">add</a> i32 %0, %0 <i>; yields {i32}:%1</i>
465 %result = <a href="#i_add">add</a> i32 %1, %1
468 <p>This last way of multiplying <tt>%X</tt> by 8 illustrates several important
469 lexical features of LLVM:</p>
472 <li>Comments are delimited with a '<tt>;</tt>' and go until the end of
475 <li>Unnamed temporaries are created when the result of a computation is not
476 assigned to a named value.</li>
478 <li>Unnamed temporaries are numbered sequentially</li>
481 <p>It also shows a convention that we follow in this document. When
482 demonstrating instructions, we will follow an instruction with a comment that
483 defines the type and name of value produced. Comments are shown in italic
488 <!-- *********************************************************************** -->
489 <h2><a name="highlevel">High Level Structure</a></h2>
490 <!-- *********************************************************************** -->
492 <!-- ======================================================================= -->
494 <a name="modulestructure">Module Structure</a>
499 <p>LLVM programs are composed of "Module"s, each of which is a translation unit
500 of the input programs. Each module consists of functions, global variables,
501 and symbol table entries. Modules may be combined together with the LLVM
502 linker, which merges function (and global variable) definitions, resolves
503 forward declarations, and merges symbol table entries. Here is an example of
504 the "hello world" module:</p>
506 <pre class="doc_code">
507 <i>; Declare the string constant as a global constant.</i>
508 <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>
510 <i>; External declaration of the puts function</i>
511 <a href="#functionstructure">declare</a> i32 @puts(i8*) <i>; i32 (i8*)* </i>
513 <i>; Definition of main function</i>
514 define i32 @main() { <i>; i32()* </i>
515 <i>; Convert [13 x i8]* to i8 *...</i>
516 %cast210 = <a href="#i_getelementptr">getelementptr</a> [13 x i8]* @.LC0, i64 0, i64 0 <i>; i8*</i>
518 <i>; Call puts function to write out the string to stdout.</i>
519 <a href="#i_call">call</a> i32 @puts(i8* %cast210) <i>; i32</i>
520 <a href="#i_ret">ret</a> i32 0
523 <i>; Named metadata</i>
524 !1 = metadata !{i32 41}
528 <p>This example is made up of a <a href="#globalvars">global variable</a> named
529 "<tt>.LC0</tt>", an external declaration of the "<tt>puts</tt>" function,
530 a <a href="#functionstructure">function definition</a> for
531 "<tt>main</tt>" and <a href="#namedmetadatastructure">named metadata</a>
534 <p>In general, a module is made up of a list of global values, where both
535 functions and global variables are global values. Global values are
536 represented by a pointer to a memory location (in this case, a pointer to an
537 array of char, and a pointer to a function), and have one of the
538 following <a href="#linkage">linkage types</a>.</p>
542 <!-- ======================================================================= -->
544 <a name="linkage">Linkage Types</a>
549 <p>All Global Variables and Functions have one of the following types of
553 <dt><tt><b><a name="linkage_private">private</a></b></tt></dt>
554 <dd>Global values with "<tt>private</tt>" linkage are only directly accessible
555 by objects in the current module. In particular, linking code into a
556 module with an private global value may cause the private to be renamed as
557 necessary to avoid collisions. Because the symbol is private to the
558 module, all references can be updated. This doesn't show up in any symbol
559 table in the object file.</dd>
561 <dt><tt><b><a name="linkage_linker_private">linker_private</a></b></tt></dt>
562 <dd>Similar to <tt>private</tt>, but the symbol is passed through the
563 assembler and evaluated by the linker. Unlike normal strong symbols, they
564 are removed by the linker from the final linked image (executable or
565 dynamic library).</dd>
567 <dt><tt><b><a name="linkage_linker_private_weak">linker_private_weak</a></b></tt></dt>
568 <dd>Similar to "<tt>linker_private</tt>", but the symbol is weak. Note that
569 <tt>linker_private_weak</tt> symbols are subject to coalescing by the
570 linker. The symbols are removed by the linker from the final linked image
571 (executable or dynamic library).</dd>
573 <dt><tt><b><a name="linkage_linker_private_weak_def_auto">linker_private_weak_def_auto</a></b></tt></dt>
574 <dd>Similar to "<tt>linker_private_weak</tt>", but it's known that the address
575 of the object is not taken. For instance, functions that had an inline
576 definition, but the compiler decided not to inline it. Note,
577 unlike <tt>linker_private</tt> and <tt>linker_private_weak</tt>,
578 <tt>linker_private_weak_def_auto</tt> may have only <tt>default</tt>
579 visibility. The symbols are removed by the linker from the final linked
580 image (executable or dynamic library).</dd>
582 <dt><tt><b><a name="linkage_internal">internal</a></b></tt></dt>
583 <dd>Similar to private, but the value shows as a local symbol
584 (<tt>STB_LOCAL</tt> in the case of ELF) in the object file. This
585 corresponds to the notion of the '<tt>static</tt>' keyword in C.</dd>
587 <dt><tt><b><a name="linkage_available_externally">available_externally</a></b></tt></dt>
588 <dd>Globals with "<tt>available_externally</tt>" linkage are never emitted
589 into the object file corresponding to the LLVM module. They exist to
590 allow inlining and other optimizations to take place given knowledge of
591 the definition of the global, which is known to be somewhere outside the
592 module. Globals with <tt>available_externally</tt> linkage are allowed to
593 be discarded at will, and are otherwise the same as <tt>linkonce_odr</tt>.
594 This linkage type is only allowed on definitions, not declarations.</dd>
596 <dt><tt><b><a name="linkage_linkonce">linkonce</a></b></tt></dt>
597 <dd>Globals with "<tt>linkonce</tt>" linkage are merged with other globals of
598 the same name when linkage occurs. This can be used to implement
599 some forms of inline functions, templates, or other code which must be
600 generated in each translation unit that uses it, but where the body may
601 be overridden with a more definitive definition later. Unreferenced
602 <tt>linkonce</tt> globals are allowed to be discarded. Note that
603 <tt>linkonce</tt> linkage does not actually allow the optimizer to
604 inline the body of this function into callers because it doesn't know if
605 this definition of the function is the definitive definition within the
606 program or whether it will be overridden by a stronger definition.
607 To enable inlining and other optimizations, use "<tt>linkonce_odr</tt>"
610 <dt><tt><b><a name="linkage_weak">weak</a></b></tt></dt>
611 <dd>"<tt>weak</tt>" linkage has the same merging semantics as
612 <tt>linkonce</tt> linkage, except that unreferenced globals with
613 <tt>weak</tt> linkage may not be discarded. This is used for globals that
614 are declared "weak" in C source code.</dd>
616 <dt><tt><b><a name="linkage_common">common</a></b></tt></dt>
617 <dd>"<tt>common</tt>" linkage is most similar to "<tt>weak</tt>" linkage, but
618 they are used for tentative definitions in C, such as "<tt>int X;</tt>" at
620 Symbols with "<tt>common</tt>" linkage are merged in the same way as
621 <tt>weak symbols</tt>, and they may not be deleted if unreferenced.
622 <tt>common</tt> symbols may not have an explicit section,
623 must have a zero initializer, and may not be marked '<a
624 href="#globalvars"><tt>constant</tt></a>'. Functions and aliases may not
625 have common linkage.</dd>
628 <dt><tt><b><a name="linkage_appending">appending</a></b></tt></dt>
629 <dd>"<tt>appending</tt>" linkage may only be applied to global variables of
630 pointer to array type. When two global variables with appending linkage
631 are linked together, the two global arrays are appended together. This is
632 the LLVM, typesafe, equivalent of having the system linker append together
633 "sections" with identical names when .o files are linked.</dd>
635 <dt><tt><b><a name="linkage_externweak">extern_weak</a></b></tt></dt>
636 <dd>The semantics of this linkage follow the ELF object file model: the symbol
637 is weak until linked, if not linked, the symbol becomes null instead of
638 being an undefined reference.</dd>
640 <dt><tt><b><a name="linkage_linkonce_odr">linkonce_odr</a></b></tt></dt>
641 <dt><tt><b><a name="linkage_weak_odr">weak_odr</a></b></tt></dt>
642 <dd>Some languages allow differing globals to be merged, such as two functions
643 with different semantics. Other languages, such as <tt>C++</tt>, ensure
644 that only equivalent globals are ever merged (the "one definition rule"
645 — "ODR"). Such languages can use the <tt>linkonce_odr</tt>
646 and <tt>weak_odr</tt> linkage types to indicate that the global will only
647 be merged with equivalent globals. These linkage types are otherwise the
648 same as their non-<tt>odr</tt> versions.</dd>
650 <dt><tt><b><a name="linkage_external">external</a></b></tt>:</dt>
651 <dd>If none of the above identifiers are used, the global is externally
652 visible, meaning that it participates in linkage and can be used to
653 resolve external symbol references.</dd>
656 <p>The next two types of linkage are targeted for Microsoft Windows platform
657 only. They are designed to support importing (exporting) symbols from (to)
658 DLLs (Dynamic Link Libraries).</p>
661 <dt><tt><b><a name="linkage_dllimport">dllimport</a></b></tt></dt>
662 <dd>"<tt>dllimport</tt>" linkage causes the compiler to reference a function
663 or variable via a global pointer to a pointer that is set up by the DLL
664 exporting the symbol. On Microsoft Windows targets, the pointer name is
665 formed by combining <code>__imp_</code> and the function or variable
668 <dt><tt><b><a name="linkage_dllexport">dllexport</a></b></tt></dt>
669 <dd>"<tt>dllexport</tt>" linkage causes the compiler to provide a global
670 pointer to a pointer in a DLL, so that it can be referenced with the
671 <tt>dllimport</tt> attribute. On Microsoft Windows targets, the pointer
672 name is formed by combining <code>__imp_</code> and the function or
676 <p>For example, since the "<tt>.LC0</tt>" variable is defined to be internal, if
677 another module defined a "<tt>.LC0</tt>" variable and was linked with this
678 one, one of the two would be renamed, preventing a collision. Since
679 "<tt>main</tt>" and "<tt>puts</tt>" are external (i.e., lacking any linkage
680 declarations), they are accessible outside of the current module.</p>
682 <p>It is illegal for a function <i>declaration</i> to have any linkage type
683 other than <tt>external</tt>, <tt>dllimport</tt>
684 or <tt>extern_weak</tt>.</p>
686 <p>Aliases can have only <tt>external</tt>, <tt>internal</tt>, <tt>weak</tt>
687 or <tt>weak_odr</tt> linkages.</p>
691 <!-- ======================================================================= -->
693 <a name="callingconv">Calling Conventions</a>
698 <p>LLVM <a href="#functionstructure">functions</a>, <a href="#i_call">calls</a>
699 and <a href="#i_invoke">invokes</a> can all have an optional calling
700 convention specified for the call. The calling convention of any pair of
701 dynamic caller/callee must match, or the behavior of the program is
702 undefined. The following calling conventions are supported by LLVM, and more
703 may be added in the future:</p>
706 <dt><b>"<tt>ccc</tt>" - The C calling convention</b>:</dt>
707 <dd>This calling convention (the default if no other calling convention is
708 specified) matches the target C calling conventions. This calling
709 convention supports varargs function calls and tolerates some mismatch in
710 the declared prototype and implemented declaration of the function (as
713 <dt><b>"<tt>fastcc</tt>" - The fast calling convention</b>:</dt>
714 <dd>This calling convention attempts to make calls as fast as possible
715 (e.g. by passing things in registers). This calling convention allows the
716 target to use whatever tricks it wants to produce fast code for the
717 target, without having to conform to an externally specified ABI
718 (Application Binary Interface).
719 <a href="CodeGenerator.html#tailcallopt">Tail calls can only be optimized
720 when this or the GHC convention is used.</a> This calling convention
721 does not support varargs and requires the prototype of all callees to
722 exactly match the prototype of the function definition.</dd>
724 <dt><b>"<tt>coldcc</tt>" - The cold calling convention</b>:</dt>
725 <dd>This calling convention attempts to make code in the caller as efficient
726 as possible under the assumption that the call is not commonly executed.
727 As such, these calls often preserve all registers so that the call does
728 not break any live ranges in the caller side. This calling convention
729 does not support varargs and requires the prototype of all callees to
730 exactly match the prototype of the function definition.</dd>
732 <dt><b>"<tt>cc <em>10</em></tt>" - GHC convention</b>:</dt>
733 <dd>This calling convention has been implemented specifically for use by the
734 <a href="http://www.haskell.org/ghc">Glasgow Haskell Compiler (GHC)</a>.
735 It passes everything in registers, going to extremes to achieve this by
736 disabling callee save registers. This calling convention should not be
737 used lightly but only for specific situations such as an alternative to
738 the <em>register pinning</em> performance technique often used when
739 implementing functional programming languages.At the moment only X86
740 supports this convention and it has the following limitations:
742 <li>On <em>X86-32</em> only supports up to 4 bit type parameters. No
743 floating point types are supported.</li>
744 <li>On <em>X86-64</em> only supports up to 10 bit type parameters and
745 6 floating point parameters.</li>
747 This calling convention supports
748 <a href="CodeGenerator.html#tailcallopt">tail call optimization</a> but
749 requires both the caller and callee are using it.
752 <dt><b>"<tt>cc <<em>n</em>></tt>" - Numbered convention</b>:</dt>
753 <dd>Any calling convention may be specified by number, allowing
754 target-specific calling conventions to be used. Target specific calling
755 conventions start at 64.</dd>
758 <p>More calling conventions can be added/defined on an as-needed basis, to
759 support Pascal conventions or any other well-known target-independent
764 <!-- ======================================================================= -->
766 <a name="visibility">Visibility Styles</a>
771 <p>All Global Variables and Functions have one of the following visibility
775 <dt><b>"<tt>default</tt>" - Default style</b>:</dt>
776 <dd>On targets that use the ELF object file format, default visibility means
777 that the declaration is visible to other modules and, in shared libraries,
778 means that the declared entity may be overridden. On Darwin, default
779 visibility means that the declaration is visible to other modules. Default
780 visibility corresponds to "external linkage" in the language.</dd>
782 <dt><b>"<tt>hidden</tt>" - Hidden style</b>:</dt>
783 <dd>Two declarations of an object with hidden visibility refer to the same
784 object if they are in the same shared object. Usually, hidden visibility
785 indicates that the symbol will not be placed into the dynamic symbol
786 table, so no other module (executable or shared library) can reference it
789 <dt><b>"<tt>protected</tt>" - Protected style</b>:</dt>
790 <dd>On ELF, protected visibility indicates that the symbol will be placed in
791 the dynamic symbol table, but that references within the defining module
792 will bind to the local symbol. That is, the symbol cannot be overridden by
798 <!-- ======================================================================= -->
800 <a name="namedtypes">Named Types</a>
805 <p>LLVM IR allows you to specify name aliases for certain types. This can make
806 it easier to read the IR and make the IR more condensed (particularly when
807 recursive types are involved). An example of a name specification is:</p>
809 <pre class="doc_code">
810 %mytype = type { %mytype*, i32 }
813 <p>You may give a name to any <a href="#typesystem">type</a> except
814 "<a href="#t_void">void</a>". Type name aliases may be used anywhere a type
815 is expected with the syntax "%mytype".</p>
817 <p>Note that type names are aliases for the structural type that they indicate,
818 and that you can therefore specify multiple names for the same type. This
819 often leads to confusing behavior when dumping out a .ll file. Since LLVM IR
820 uses structural typing, the name is not part of the type. When printing out
821 LLVM IR, the printer will pick <em>one name</em> to render all types of a
822 particular shape. This means that if you have code where two different
823 source types end up having the same LLVM type, that the dumper will sometimes
824 print the "wrong" or unexpected type. This is an important design point and
825 isn't going to change.</p>
829 <!-- ======================================================================= -->
831 <a name="globalvars">Global Variables</a>
836 <p>Global variables define regions of memory allocated at compilation time
837 instead of run-time. Global variables may optionally be initialized, may
838 have an explicit section to be placed in, and may have an optional explicit
839 alignment specified. A variable may be defined as "thread_local", which
840 means that it will not be shared by threads (each thread will have a
841 separated copy of the variable). A variable may be defined as a global
842 "constant," which indicates that the contents of the variable
843 will <b>never</b> be modified (enabling better optimization, allowing the
844 global data to be placed in the read-only section of an executable, etc).
845 Note that variables that need runtime initialization cannot be marked
846 "constant" as there is a store to the variable.</p>
848 <p>LLVM explicitly allows <em>declarations</em> of global variables to be marked
849 constant, even if the final definition of the global is not. This capability
850 can be used to enable slightly better optimization of the program, but
851 requires the language definition to guarantee that optimizations based on the
852 'constantness' are valid for the translation units that do not include the
855 <p>As SSA values, global variables define pointer values that are in scope
856 (i.e. they dominate) all basic blocks in the program. Global variables
857 always define a pointer to their "content" type because they describe a
858 region of memory, and all memory objects in LLVM are accessed through
861 <p>Global variables can be marked with <tt>unnamed_addr</tt> which indicates
862 that the address is not significant, only the content. Constants marked
863 like this can be merged with other constants if they have the same
864 initializer. Note that a constant with significant address <em>can</em>
865 be merged with a <tt>unnamed_addr</tt> constant, the result being a
866 constant whose address is significant.</p>
868 <p>A global variable may be declared to reside in a target-specific numbered
869 address space. For targets that support them, address spaces may affect how
870 optimizations are performed and/or what target instructions are used to
871 access the variable. The default address space is zero. The address space
872 qualifier must precede any other attributes.</p>
874 <p>LLVM allows an explicit section to be specified for globals. If the target
875 supports it, it will emit globals to the section specified.</p>
877 <p>An explicit alignment may be specified for a global, which must be a power
878 of 2. If not present, or if the alignment is set to zero, the alignment of
879 the global is set by the target to whatever it feels convenient. If an
880 explicit alignment is specified, the global is forced to have exactly that
881 alignment. Targets and optimizers are not allowed to over-align the global
882 if the global has an assigned section. In this case, the extra alignment
883 could be observable: for example, code could assume that the globals are
884 densely packed in their section and try to iterate over them as an array,
885 alignment padding would break this iteration.</p>
887 <p>For example, the following defines a global in a numbered address space with
888 an initializer, section, and alignment:</p>
890 <pre class="doc_code">
891 @G = addrspace(5) constant float 1.0, section "foo", align 4
897 <!-- ======================================================================= -->
899 <a name="functionstructure">Functions</a>
904 <p>LLVM function definitions consist of the "<tt>define</tt>" keyword, an
905 optional <a href="#linkage">linkage type</a>, an optional
906 <a href="#visibility">visibility style</a>, an optional
907 <a href="#callingconv">calling convention</a>,
908 an optional <tt>unnamed_addr</tt> attribute, a return type, an optional
909 <a href="#paramattrs">parameter attribute</a> for the return type, a function
910 name, a (possibly empty) argument list (each with optional
911 <a href="#paramattrs">parameter attributes</a>), optional
912 <a href="#fnattrs">function attributes</a>, an optional section, an optional
913 alignment, an optional <a href="#gc">garbage collector name</a>, an opening
914 curly brace, a list of basic blocks, and a closing curly brace.</p>
916 <p>LLVM function declarations consist of the "<tt>declare</tt>" keyword, an
917 optional <a href="#linkage">linkage type</a>, an optional
918 <a href="#visibility">visibility style</a>, an optional
919 <a href="#callingconv">calling convention</a>,
920 an optional <tt>unnamed_addr</tt> attribute, a return type, an optional
921 <a href="#paramattrs">parameter attribute</a> for the return type, a function
922 name, a possibly empty list of arguments, an optional alignment, and an
923 optional <a href="#gc">garbage collector name</a>.</p>
925 <p>A function definition contains a list of basic blocks, forming the CFG
926 (Control Flow Graph) for the function. Each basic block may optionally start
927 with a label (giving the basic block a symbol table entry), contains a list
928 of instructions, and ends with a <a href="#terminators">terminator</a>
929 instruction (such as a branch or function return).</p>
931 <p>The first basic block in a function is special in two ways: it is immediately
932 executed on entrance to the function, and it is not allowed to have
933 predecessor basic blocks (i.e. there can not be any branches to the entry
934 block of a function). Because the block can have no predecessors, it also
935 cannot have any <a href="#i_phi">PHI nodes</a>.</p>
937 <p>LLVM allows an explicit section to be specified for functions. If the target
938 supports it, it will emit functions to the section specified.</p>
940 <p>An explicit alignment may be specified for a function. If not present, or if
941 the alignment is set to zero, the alignment of the function is set by the
942 target to whatever it feels convenient. If an explicit alignment is
943 specified, the function is forced to have at least that much alignment. All
944 alignments must be a power of 2.</p>
946 <p>If the <tt>unnamed_addr</tt> attribute is given, the address is know to not
947 be significant and two identical functions can be merged</p>.
950 <pre class="doc_code">
951 define [<a href="#linkage">linkage</a>] [<a href="#visibility">visibility</a>]
952 [<a href="#callingconv">cconv</a>] [<a href="#paramattrs">ret attrs</a>]
953 <ResultType> @<FunctionName> ([argument list])
954 [<a href="#fnattrs">fn Attrs</a>] [section "name"] [align N]
955 [<a href="#gc">gc</a>] { ... }
960 <!-- ======================================================================= -->
962 <a name="aliasstructure">Aliases</a>
967 <p>Aliases act as "second name" for the aliasee value (which can be either
968 function, global variable, another alias or bitcast of global value). Aliases
969 may have an optional <a href="#linkage">linkage type</a>, and an
970 optional <a href="#visibility">visibility style</a>.</p>
973 <pre class="doc_code">
974 @<Name> = alias [Linkage] [Visibility] <AliaseeTy> @<Aliasee>
979 <!-- ======================================================================= -->
981 <a name="namedmetadatastructure">Named Metadata</a>
986 <p>Named metadata is a collection of metadata. <a href="#metadata">Metadata
987 nodes</a> (but not metadata strings) are the only valid operands for
988 a named metadata.</p>
991 <pre class="doc_code">
992 ; Some unnamed metadata nodes, which are referenced by the named metadata.
993 !0 = metadata !{metadata !"zero"}
994 !1 = metadata !{metadata !"one"}
995 !2 = metadata !{metadata !"two"}
997 !name = !{!0, !1, !2}
1002 <!-- ======================================================================= -->
1004 <a name="paramattrs">Parameter Attributes</a>
1009 <p>The return type and each parameter of a function type may have a set of
1010 <i>parameter attributes</i> associated with them. Parameter attributes are
1011 used to communicate additional information about the result or parameters of
1012 a function. Parameter attributes are considered to be part of the function,
1013 not of the function type, so functions with different parameter attributes
1014 can have the same function type.</p>
1016 <p>Parameter attributes are simple keywords that follow the type specified. If
1017 multiple parameter attributes are needed, they are space separated. For
1020 <pre class="doc_code">
1021 declare i32 @printf(i8* noalias nocapture, ...)
1022 declare i32 @atoi(i8 zeroext)
1023 declare signext i8 @returns_signed_char()
1026 <p>Note that any attributes for the function result (<tt>nounwind</tt>,
1027 <tt>readonly</tt>) come immediately after the argument list.</p>
1029 <p>Currently, only the following parameter attributes are defined:</p>
1032 <dt><tt><b>zeroext</b></tt></dt>
1033 <dd>This indicates to the code generator that the parameter or return value
1034 should be zero-extended to the extent required by the target's ABI (which
1035 is usually 32-bits, but is 8-bits for a i1 on x86-64) by the caller (for a
1036 parameter) or the callee (for a return value).</dd>
1038 <dt><tt><b>signext</b></tt></dt>
1039 <dd>This indicates to the code generator that the parameter or return value
1040 should be sign-extended to the extent required by the target's ABI (which
1041 is usually 32-bits) by the caller (for a parameter) or the callee (for a
1044 <dt><tt><b>inreg</b></tt></dt>
1045 <dd>This indicates that this parameter or return value should be treated in a
1046 special target-dependent fashion during while emitting code for a function
1047 call or return (usually, by putting it in a register as opposed to memory,
1048 though some targets use it to distinguish between two different kinds of
1049 registers). Use of this attribute is target-specific.</dd>
1051 <dt><tt><b><a name="byval">byval</a></b></tt></dt>
1052 <dd><p>This indicates that the pointer parameter should really be passed by
1053 value to the function. The attribute implies that a hidden copy of the
1055 is made between the caller and the callee, so the callee is unable to
1056 modify the value in the callee. This attribute is only valid on LLVM
1057 pointer arguments. It is generally used to pass structs and arrays by
1058 value, but is also valid on pointers to scalars. The copy is considered
1059 to belong to the caller not the callee (for example,
1060 <tt><a href="#readonly">readonly</a></tt> functions should not write to
1061 <tt>byval</tt> parameters). This is not a valid attribute for return
1064 <p>The byval attribute also supports specifying an alignment with
1065 the align attribute. It indicates the alignment of the stack slot to
1066 form and the known alignment of the pointer specified to the call site. If
1067 the alignment is not specified, then the code generator makes a
1068 target-specific assumption.</p></dd>
1070 <dt><tt><b><a name="sret">sret</a></b></tt></dt>
1071 <dd>This indicates that the pointer parameter specifies the address of a
1072 structure that is the return value of the function in the source program.
1073 This pointer must be guaranteed by the caller to be valid: loads and
1074 stores to the structure may be assumed by the callee to not to trap. This
1075 may only be applied to the first parameter. This is not a valid attribute
1076 for return values. </dd>
1078 <dt><tt><b><a name="noalias">noalias</a></b></tt></dt>
1079 <dd>This indicates that pointer values
1080 <a href="#pointeraliasing"><i>based</i></a> on the argument or return
1081 value do not alias pointer values which are not <i>based</i> on it,
1082 ignoring certain "irrelevant" dependencies.
1083 For a call to the parent function, dependencies between memory
1084 references from before or after the call and from those during the call
1085 are "irrelevant" to the <tt>noalias</tt> keyword for the arguments and
1086 return value used in that call.
1087 The caller shares the responsibility with the callee for ensuring that
1088 these requirements are met.
1089 For further details, please see the discussion of the NoAlias response in
1090 <a href="AliasAnalysis.html#MustMayNo">alias analysis</a>.<br>
1092 Note that this definition of <tt>noalias</tt> is intentionally
1093 similar to the definition of <tt>restrict</tt> in C99 for function
1094 arguments, though it is slightly weaker.
1096 For function return values, C99's <tt>restrict</tt> is not meaningful,
1097 while LLVM's <tt>noalias</tt> is.
1100 <dt><tt><b><a name="nocapture">nocapture</a></b></tt></dt>
1101 <dd>This indicates that the callee does not make any copies of the pointer
1102 that outlive the callee itself. This is not a valid attribute for return
1105 <dt><tt><b><a name="nest">nest</a></b></tt></dt>
1106 <dd>This indicates that the pointer parameter can be excised using the
1107 <a href="#int_trampoline">trampoline intrinsics</a>. This is not a valid
1108 attribute for return values.</dd>
1113 <!-- ======================================================================= -->
1115 <a name="gc">Garbage Collector Names</a>
1120 <p>Each function may specify a garbage collector name, which is simply a
1123 <pre class="doc_code">
1124 define void @f() gc "name" { ... }
1127 <p>The compiler declares the supported values of <i>name</i>. Specifying a
1128 collector which will cause the compiler to alter its output in order to
1129 support the named garbage collection algorithm.</p>
1133 <!-- ======================================================================= -->
1135 <a name="fnattrs">Function Attributes</a>
1140 <p>Function attributes are set to communicate additional information about a
1141 function. Function attributes are considered to be part of the function, not
1142 of the function type, so functions with different parameter attributes can
1143 have the same function type.</p>
1145 <p>Function attributes are simple keywords that follow the type specified. If
1146 multiple attributes are needed, they are space separated. For example:</p>
1148 <pre class="doc_code">
1149 define void @f() noinline { ... }
1150 define void @f() alwaysinline { ... }
1151 define void @f() alwaysinline optsize { ... }
1152 define void @f() optsize { ... }
1156 <dt><tt><b>alignstack(<<em>n</em>>)</b></tt></dt>
1157 <dd>This attribute indicates that, when emitting the prologue and epilogue,
1158 the backend should forcibly align the stack pointer. Specify the
1159 desired alignment, which must be a power of two, in parentheses.
1161 <dt><tt><b>alwaysinline</b></tt></dt>
1162 <dd>This attribute indicates that the inliner should attempt to inline this
1163 function into callers whenever possible, ignoring any active inlining size
1164 threshold for this caller.</dd>
1166 <dt><tt><b>nonlazybind</b></tt></dt>
1167 <dd>This attribute suppresses lazy symbol binding for the function. This
1168 may make calls to the function faster, at the cost of extra program
1169 startup time if the function is not called during program startup.</dd>
1171 <dt><tt><b>inlinehint</b></tt></dt>
1172 <dd>This attribute indicates that the source code contained a hint that inlining
1173 this function is desirable (such as the "inline" keyword in C/C++). It
1174 is just a hint; it imposes no requirements on the inliner.</dd>
1176 <dt><tt><b>naked</b></tt></dt>
1177 <dd>This attribute disables prologue / epilogue emission for the function.
1178 This can have very system-specific consequences.</dd>
1180 <dt><tt><b>noimplicitfloat</b></tt></dt>
1181 <dd>This attributes disables implicit floating point instructions.</dd>
1183 <dt><tt><b>noinline</b></tt></dt>
1184 <dd>This attribute indicates that the inliner should never inline this
1185 function in any situation. This attribute may not be used together with
1186 the <tt>alwaysinline</tt> attribute.</dd>
1188 <dt><tt><b>noredzone</b></tt></dt>
1189 <dd>This attribute indicates that the code generator should not use a red
1190 zone, even if the target-specific ABI normally permits it.</dd>
1192 <dt><tt><b>noreturn</b></tt></dt>
1193 <dd>This function attribute indicates that the function never returns
1194 normally. This produces undefined behavior at runtime if the function
1195 ever does dynamically return.</dd>
1197 <dt><tt><b>nounwind</b></tt></dt>
1198 <dd>This function attribute indicates that the function never returns with an
1199 unwind or exceptional control flow. If the function does unwind, its
1200 runtime behavior is undefined.</dd>
1202 <dt><tt><b>optsize</b></tt></dt>
1203 <dd>This attribute suggests that optimization passes and code generator passes
1204 make choices that keep the code size of this function low, and otherwise
1205 do optimizations specifically to reduce code size.</dd>
1207 <dt><tt><b>readnone</b></tt></dt>
1208 <dd>This attribute indicates that the function computes its result (or decides
1209 to unwind an exception) based strictly on its arguments, without
1210 dereferencing any pointer arguments or otherwise accessing any mutable
1211 state (e.g. memory, control registers, etc) visible to caller functions.
1212 It does not write through any pointer arguments
1213 (including <tt><a href="#byval">byval</a></tt> arguments) and never
1214 changes any state visible to callers. This means that it cannot unwind
1215 exceptions by calling the <tt>C++</tt> exception throwing methods, but
1216 could use the <tt>unwind</tt> instruction.</dd>
1218 <dt><tt><b><a name="readonly">readonly</a></b></tt></dt>
1219 <dd>This attribute indicates that the function does not write through any
1220 pointer arguments (including <tt><a href="#byval">byval</a></tt>
1221 arguments) or otherwise modify any state (e.g. memory, control registers,
1222 etc) visible to caller functions. It may dereference pointer arguments
1223 and read state that may be set in the caller. A readonly function always
1224 returns the same value (or unwinds an exception identically) when called
1225 with the same set of arguments and global state. It cannot unwind an
1226 exception by calling the <tt>C++</tt> exception throwing methods, but may
1227 use the <tt>unwind</tt> instruction.</dd>
1229 <dt><tt><b><a name="ssp">ssp</a></b></tt></dt>
1230 <dd>This attribute indicates that the function should emit a stack smashing
1231 protector. It is in the form of a "canary"—a random value placed on
1232 the stack before the local variables that's checked upon return from the
1233 function to see if it has been overwritten. A heuristic is used to
1234 determine if a function needs stack protectors or not.<br>
1236 If a function that has an <tt>ssp</tt> attribute is inlined into a
1237 function that doesn't have an <tt>ssp</tt> attribute, then the resulting
1238 function will have an <tt>ssp</tt> attribute.</dd>
1240 <dt><tt><b>sspreq</b></tt></dt>
1241 <dd>This attribute indicates that the function should <em>always</em> emit a
1242 stack smashing protector. This overrides
1243 the <tt><a href="#ssp">ssp</a></tt> function attribute.<br>
1245 If a function that has an <tt>sspreq</tt> attribute is inlined into a
1246 function that doesn't have an <tt>sspreq</tt> attribute or which has
1247 an <tt>ssp</tt> attribute, then the resulting function will have
1248 an <tt>sspreq</tt> attribute.</dd>
1250 <dt><tt><b><a name="uwtable">uwtable</a></b></tt></dt>
1251 <dd>This attribute indicates that the ABI being targeted requires that
1252 an unwind table entry be produce for this function even if we can
1253 show that no exceptions passes by it. This is normally the case for
1254 the ELF x86-64 abi, but it can be disabled for some compilation
1257 <dt><tt><b><a name="returns_twice">returns_twice</a></b></tt></dt>
1258 <dd>This attribute indicates that this function can return
1259 twice. The C <code>setjmp</code> is an example of such a function.
1260 The compiler disables some optimizations (like tail calls) in the caller of
1261 these functions.</dd>
1266 <!-- ======================================================================= -->
1268 <a name="moduleasm">Module-Level Inline Assembly</a>
1273 <p>Modules may contain "module-level inline asm" blocks, which corresponds to
1274 the GCC "file scope inline asm" blocks. These blocks are internally
1275 concatenated by LLVM and treated as a single unit, but may be separated in
1276 the <tt>.ll</tt> file if desired. The syntax is very simple:</p>
1278 <pre class="doc_code">
1279 module asm "inline asm code goes here"
1280 module asm "more can go here"
1283 <p>The strings can contain any character by escaping non-printable characters.
1284 The escape sequence used is simply "\xx" where "xx" is the two digit hex code
1287 <p>The inline asm code is simply printed to the machine code .s file when
1288 assembly code is generated.</p>
1292 <!-- ======================================================================= -->
1294 <a name="datalayout">Data Layout</a>
1299 <p>A module may specify a target specific data layout string that specifies how
1300 data is to be laid out in memory. The syntax for the data layout is
1303 <pre class="doc_code">
1304 target datalayout = "<i>layout specification</i>"
1307 <p>The <i>layout specification</i> consists of a list of specifications
1308 separated by the minus sign character ('-'). Each specification starts with
1309 a letter and may include other information after the letter to define some
1310 aspect of the data layout. The specifications accepted are as follows:</p>
1314 <dd>Specifies that the target lays out data in big-endian form. That is, the
1315 bits with the most significance have the lowest address location.</dd>
1318 <dd>Specifies that the target lays out data in little-endian form. That is,
1319 the bits with the least significance have the lowest address
1322 <dt><tt>S<i>size</i></tt></dt>
1323 <dd>Specifies the natural alignment of the stack in bits. Alignment promotion
1324 of stack variables is limited to the natural stack alignment to avoid
1325 dynamic stack realignment. The stack alignment must be a multiple of
1326 8-bits. If omitted, the natural stack alignment defaults to "unspecified",
1327 which does not prevent any alignment promotions.</dd>
1329 <dt><tt>p:<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1330 <dd>This specifies the <i>size</i> of a pointer and its <i>abi</i> and
1331 <i>preferred</i> alignments. All sizes are in bits. Specifying
1332 the <i>pref</i> alignment is optional. If omitted, the
1333 preceding <tt>:</tt> should be omitted too.</dd>
1335 <dt><tt>i<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1336 <dd>This specifies the alignment for an integer type of a given bit
1337 <i>size</i>. The value of <i>size</i> must be in the range [1,2^23).</dd>
1339 <dt><tt>v<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1340 <dd>This specifies the alignment for a vector type of a given bit
1343 <dt><tt>f<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1344 <dd>This specifies the alignment for a floating point type of a given bit
1345 <i>size</i>. Only values of <i>size</i> that are supported by the target
1346 will work. 32 (float) and 64 (double) are supported on all targets;
1347 80 or 128 (different flavors of long double) are also supported on some
1350 <dt><tt>a<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1351 <dd>This specifies the alignment for an aggregate type of a given bit
1354 <dt><tt>s<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1355 <dd>This specifies the alignment for a stack object of a given bit
1358 <dt><tt>n<i>size1</i>:<i>size2</i>:<i>size3</i>...</tt></dt>
1359 <dd>This specifies a set of native integer widths for the target CPU
1360 in bits. For example, it might contain "n32" for 32-bit PowerPC,
1361 "n32:64" for PowerPC 64, or "n8:16:32:64" for X86-64. Elements of
1362 this set are considered to support most general arithmetic
1363 operations efficiently.</dd>
1366 <p>When constructing the data layout for a given target, LLVM starts with a
1367 default set of specifications which are then (possibly) overridden by the
1368 specifications in the <tt>datalayout</tt> keyword. The default specifications
1369 are given in this list:</p>
1372 <li><tt>E</tt> - big endian</li>
1373 <li><tt>p:64:64:64</tt> - 64-bit pointers with 64-bit alignment</li>
1374 <li><tt>i1:8:8</tt> - i1 is 8-bit (byte) aligned</li>
1375 <li><tt>i8:8:8</tt> - i8 is 8-bit (byte) aligned</li>
1376 <li><tt>i16:16:16</tt> - i16 is 16-bit aligned</li>
1377 <li><tt>i32:32:32</tt> - i32 is 32-bit aligned</li>
1378 <li><tt>i64:32:64</tt> - i64 has ABI alignment of 32-bits but preferred
1379 alignment of 64-bits</li>
1380 <li><tt>f32:32:32</tt> - float is 32-bit aligned</li>
1381 <li><tt>f64:64:64</tt> - double is 64-bit aligned</li>
1382 <li><tt>v64:64:64</tt> - 64-bit vector is 64-bit aligned</li>
1383 <li><tt>v128:128:128</tt> - 128-bit vector is 128-bit aligned</li>
1384 <li><tt>a0:0:1</tt> - aggregates are 8-bit aligned</li>
1385 <li><tt>s0:64:64</tt> - stack objects are 64-bit aligned</li>
1388 <p>When LLVM is determining the alignment for a given type, it uses the
1389 following rules:</p>
1392 <li>If the type sought is an exact match for one of the specifications, that
1393 specification is used.</li>
1395 <li>If no match is found, and the type sought is an integer type, then the
1396 smallest integer type that is larger than the bitwidth of the sought type
1397 is used. If none of the specifications are larger than the bitwidth then
1398 the the largest integer type is used. For example, given the default
1399 specifications above, the i7 type will use the alignment of i8 (next
1400 largest) while both i65 and i256 will use the alignment of i64 (largest
1403 <li>If no match is found, and the type sought is a vector type, then the
1404 largest vector type that is smaller than the sought vector type will be
1405 used as a fall back. This happens because <128 x double> can be
1406 implemented in terms of 64 <2 x double>, for example.</li>
1409 <p>The function of the data layout string may not be what you expect. Notably,
1410 this is not a specification from the frontend of what alignment the code
1411 generator should use.</p>
1413 <p>Instead, if specified, the target data layout is required to match what the
1414 ultimate <em>code generator</em> expects. This string is used by the
1415 mid-level optimizers to
1416 improve code, and this only works if it matches what the ultimate code
1417 generator uses. If you would like to generate IR that does not embed this
1418 target-specific detail into the IR, then you don't have to specify the
1419 string. This will disable some optimizations that require precise layout
1420 information, but this also prevents those optimizations from introducing
1421 target specificity into the IR.</p>
1427 <!-- ======================================================================= -->
1429 <a name="pointeraliasing">Pointer Aliasing Rules</a>
1434 <p>Any memory access must be done through a pointer value associated
1435 with an address range of the memory access, otherwise the behavior
1436 is undefined. Pointer values are associated with address ranges
1437 according to the following rules:</p>
1440 <li>A pointer value is associated with the addresses associated with
1441 any value it is <i>based</i> on.
1442 <li>An address of a global variable is associated with the address
1443 range of the variable's storage.</li>
1444 <li>The result value of an allocation instruction is associated with
1445 the address range of the allocated storage.</li>
1446 <li>A null pointer in the default address-space is associated with
1448 <li>An integer constant other than zero or a pointer value returned
1449 from a function not defined within LLVM may be associated with address
1450 ranges allocated through mechanisms other than those provided by
1451 LLVM. Such ranges shall not overlap with any ranges of addresses
1452 allocated by mechanisms provided by LLVM.</li>
1455 <p>A pointer value is <i>based</i> on another pointer value according
1456 to the following rules:</p>
1459 <li>A pointer value formed from a
1460 <tt><a href="#i_getelementptr">getelementptr</a></tt> operation
1461 is <i>based</i> on the first operand of the <tt>getelementptr</tt>.</li>
1462 <li>The result value of a
1463 <tt><a href="#i_bitcast">bitcast</a></tt> is <i>based</i> on the operand
1464 of the <tt>bitcast</tt>.</li>
1465 <li>A pointer value formed by an
1466 <tt><a href="#i_inttoptr">inttoptr</a></tt> is <i>based</i> on all
1467 pointer values that contribute (directly or indirectly) to the
1468 computation of the pointer's value.</li>
1469 <li>The "<i>based</i> on" relationship is transitive.</li>
1472 <p>Note that this definition of <i>"based"</i> is intentionally
1473 similar to the definition of <i>"based"</i> in C99, though it is
1474 slightly weaker.</p>
1476 <p>LLVM IR does not associate types with memory. The result type of a
1477 <tt><a href="#i_load">load</a></tt> merely indicates the size and
1478 alignment of the memory from which to load, as well as the
1479 interpretation of the value. The first operand type of a
1480 <tt><a href="#i_store">store</a></tt> similarly only indicates the size
1481 and alignment of the store.</p>
1483 <p>Consequently, type-based alias analysis, aka TBAA, aka
1484 <tt>-fstrict-aliasing</tt>, is not applicable to general unadorned
1485 LLVM IR. <a href="#metadata">Metadata</a> may be used to encode
1486 additional information which specialized optimization passes may use
1487 to implement type-based alias analysis.</p>
1491 <!-- ======================================================================= -->
1493 <a name="volatile">Volatile Memory Accesses</a>
1498 <p>Certain memory accesses, such as <a href="#i_load"><tt>load</tt></a>s, <a
1499 href="#i_store"><tt>store</tt></a>s, and <a
1500 href="#int_memcpy"><tt>llvm.memcpy</tt></a>s may be marked <tt>volatile</tt>.
1501 The optimizers must not change the number of volatile operations or change their
1502 order of execution relative to other volatile operations. The optimizers
1503 <i>may</i> change the order of volatile operations relative to non-volatile
1504 operations. This is not Java's "volatile" and has no cross-thread
1505 synchronization behavior.</p>
1509 <!-- ======================================================================= -->
1511 <a name="memmodel">Memory Model for Concurrent Operations</a>
1516 <p>The LLVM IR does not define any way to start parallel threads of execution
1517 or to register signal handlers. Nonetheless, there are platform-specific
1518 ways to create them, and we define LLVM IR's behavior in their presence. This
1519 model is inspired by the C++0x memory model.</p>
1521 <p>For a more informal introduction to this model, see the
1522 <a href="Atomics.html">LLVM Atomic Instructions and Concurrency Guide</a>.
1524 <p>We define a <i>happens-before</i> partial order as the least partial order
1527 <li>Is a superset of single-thread program order, and</li>
1528 <li>When a <i>synchronizes-with</i> <tt>b</tt>, includes an edge from
1529 <tt>a</tt> to <tt>b</tt>. <i>Synchronizes-with</i> pairs are introduced
1530 by platform-specific techniques, like pthread locks, thread
1531 creation, thread joining, etc., and by atomic instructions.
1532 (See also <a href="#ordering">Atomic Memory Ordering Constraints</a>).
1536 <p>Note that program order does not introduce <i>happens-before</i> edges
1537 between a thread and signals executing inside that thread.</p>
1539 <p>Every (defined) read operation (load instructions, memcpy, atomic
1540 loads/read-modify-writes, etc.) <var>R</var> reads a series of bytes written by
1541 (defined) write operations (store instructions, atomic
1542 stores/read-modify-writes, memcpy, etc.). For the purposes of this section,
1543 initialized globals are considered to have a write of the initializer which is
1544 atomic and happens before any other read or write of the memory in question.
1545 For each byte of a read <var>R</var>, <var>R<sub>byte</sub></var> may see
1546 any write to the same byte, except:</p>
1549 <li>If <var>write<sub>1</sub></var> happens before
1550 <var>write<sub>2</sub></var>, and <var>write<sub>2</sub></var> happens
1551 before <var>R<sub>byte</sub></var>, then <var>R<sub>byte</sub></var>
1552 does not see <var>write<sub>1</sub></var>.
1553 <li>If <var>R<sub>byte</sub></var> happens before
1554 <var>write<sub>3</sub></var>, then <var>R<sub>byte</sub></var> does not
1555 see <var>write<sub>3</sub></var>.
1558 <p>Given that definition, <var>R<sub>byte</sub></var> is defined as follows:
1560 <li>If <var>R</var> is volatile, the result is target-dependent. (Volatile
1561 is supposed to give guarantees which can support
1562 <code>sig_atomic_t</code> in C/C++, and may be used for accesses to
1563 addresses which do not behave like normal memory. It does not generally
1564 provide cross-thread synchronization.)
1565 <li>Otherwise, if there is no write to the same byte that happens before
1566 <var>R<sub>byte</sub></var>, <var>R<sub>byte</sub></var> returns
1567 <tt>undef</tt> for that byte.
1568 <li>Otherwise, if <var>R<sub>byte</sub></var> may see exactly one write,
1569 <var>R<sub>byte</sub></var> returns the value written by that
1571 <li>Otherwise, if <var>R</var> is atomic, and all the writes
1572 <var>R<sub>byte</sub></var> may see are atomic, it chooses one of the
1573 values written. See the <a href="#ordering">Atomic Memory Ordering
1574 Constraints</a> section for additional constraints on how the choice
1576 <li>Otherwise <var>R<sub>byte</sub></var> returns <tt>undef</tt>.</li>
1579 <p><var>R</var> returns the value composed of the series of bytes it read.
1580 This implies that some bytes within the value may be <tt>undef</tt>
1581 <b>without</b> the entire value being <tt>undef</tt>. Note that this only
1582 defines the semantics of the operation; it doesn't mean that targets will
1583 emit more than one instruction to read the series of bytes.</p>
1585 <p>Note that in cases where none of the atomic intrinsics are used, this model
1586 places only one restriction on IR transformations on top of what is required
1587 for single-threaded execution: introducing a store to a byte which might not
1588 otherwise be stored is not allowed in general. (Specifically, in the case
1589 where another thread might write to and read from an address, introducing a
1590 store can change a load that may see exactly one write into a load that may
1591 see multiple writes.)</p>
1593 <!-- FIXME: This model assumes all targets where concurrency is relevant have
1594 a byte-size store which doesn't affect adjacent bytes. As far as I can tell,
1595 none of the backends currently in the tree fall into this category; however,
1596 there might be targets which care. If there are, we want a paragraph
1599 Targets may specify that stores narrower than a certain width are not
1600 available; on such a target, for the purposes of this model, treat any
1601 non-atomic write with an alignment or width less than the minimum width
1602 as if it writes to the relevant surrounding bytes.
1607 <!-- ======================================================================= -->
1609 <a name="ordering">Atomic Memory Ordering Constraints</a>
1614 <p>Atomic instructions (<a href="#i_cmpxchg"><code>cmpxchg</code></a>,
1615 <a href="#i_atomicrmw"><code>atomicrmw</code></a>,
1616 <a href="#i_fence"><code>fence</code></a>,
1617 <a href="#i_load"><code>atomic load</code></a>, and
1618 <a href="#i_store"><code>atomic store</code></a>) take an ordering parameter
1619 that determines which other atomic instructions on the same address they
1620 <i>synchronize with</i>. These semantics are borrowed from Java and C++0x,
1621 but are somewhat more colloquial. If these descriptions aren't precise enough,
1622 check those specs (see spec references in the
1623 <a href="Atomic.html#introduction">atomics guide</a>).
1624 <a href="#i_fence"><code>fence</code></a> instructions
1625 treat these orderings somewhat differently since they don't take an address.
1626 See that instruction's documentation for details.</p>
1628 <p>For a simpler introduction to the ordering constraints, see the
1629 <a href="Atomics.html">LLVM Atomic Instructions and Concurrency Guide</a>.</p>
1632 <dt><code>unordered</code></dt>
1633 <dd>The set of values that can be read is governed by the happens-before
1634 partial order. A value cannot be read unless some operation wrote it.
1635 This is intended to provide a guarantee strong enough to model Java's
1636 non-volatile shared variables. This ordering cannot be specified for
1637 read-modify-write operations; it is not strong enough to make them atomic
1638 in any interesting way.</dd>
1639 <dt><code>monotonic</code></dt>
1640 <dd>In addition to the guarantees of <code>unordered</code>, there is a single
1641 total order for modifications by <code>monotonic</code> operations on each
1642 address. All modification orders must be compatible with the happens-before
1643 order. There is no guarantee that the modification orders can be combined to
1644 a global total order for the whole program (and this often will not be
1645 possible). The read in an atomic read-modify-write operation
1646 (<a href="#i_cmpxchg"><code>cmpxchg</code></a> and
1647 <a href="#i_atomicrmw"><code>atomicrmw</code></a>)
1648 reads the value in the modification order immediately before the value it
1649 writes. If one atomic read happens before another atomic read of the same
1650 address, the later read must see the same value or a later value in the
1651 address's modification order. This disallows reordering of
1652 <code>monotonic</code> (or stronger) operations on the same address. If an
1653 address is written <code>monotonic</code>ally by one thread, and other threads
1654 <code>monotonic</code>ally read that address repeatedly, the other threads must
1655 eventually see the write. This corresponds to the C++0x/C1x
1656 <code>memory_order_relaxed</code>.</dd>
1657 <dt><code>acquire</code></dt>
1658 <dd>In addition to the guarantees of <code>monotonic</code>,
1659 a <i>synchronizes-with</i> edge may be formed with a <code>release</code>
1660 operation. This is intended to model C++'s <code>memory_order_acquire</code>.</dd>
1661 <dt><code>release</code></dt>
1662 <dd>In addition to the guarantees of <code>monotonic</code>, if this operation
1663 writes a value which is subsequently read by an <code>acquire</code> operation,
1664 it <i>synchronizes-with</i> that operation. (This isn't a complete
1665 description; see the C++0x definition of a release sequence.) This corresponds
1666 to the C++0x/C1x <code>memory_order_release</code>.</dd>
1667 <dt><code>acq_rel</code> (acquire+release)</dt><dd>Acts as both an
1668 <code>acquire</code> and <code>release</code> operation on its address.
1669 This corresponds to the C++0x/C1x <code>memory_order_acq_rel</code>.</dd>
1670 <dt><code>seq_cst</code> (sequentially consistent)</dt><dd>
1671 <dd>In addition to the guarantees of <code>acq_rel</code>
1672 (<code>acquire</code> for an operation which only reads, <code>release</code>
1673 for an operation which only writes), there is a global total order on all
1674 sequentially-consistent operations on all addresses, which is consistent with
1675 the <i>happens-before</i> partial order and with the modification orders of
1676 all the affected addresses. Each sequentially-consistent read sees the last
1677 preceding write to the same address in this global order. This corresponds
1678 to the C++0x/C1x <code>memory_order_seq_cst</code> and Java volatile.</dd>
1681 <p id="singlethread">If an atomic operation is marked <code>singlethread</code>,
1682 it only <i>synchronizes with</i> or participates in modification and seq_cst
1683 total orderings with other operations running in the same thread (for example,
1684 in signal handlers).</p>
1690 <!-- *********************************************************************** -->
1691 <h2><a name="typesystem">Type System</a></h2>
1692 <!-- *********************************************************************** -->
1696 <p>The LLVM type system is one of the most important features of the
1697 intermediate representation. Being typed enables a number of optimizations
1698 to be performed on the intermediate representation directly, without having
1699 to do extra analyses on the side before the transformation. A strong type
1700 system makes it easier to read the generated code and enables novel analyses
1701 and transformations that are not feasible to perform on normal three address
1702 code representations.</p>
1704 <!-- ======================================================================= -->
1706 <a name="t_classifications">Type Classifications</a>
1711 <p>The types fall into a few useful classifications:</p>
1713 <table border="1" cellspacing="0" cellpadding="4">
1715 <tr><th>Classification</th><th>Types</th></tr>
1717 <td><a href="#t_integer">integer</a></td>
1718 <td><tt>i1, i2, i3, ... i8, ... i16, ... i32, ... i64, ... </tt></td>
1721 <td><a href="#t_floating">floating point</a></td>
1722 <td><tt>float, double, x86_fp80, fp128, ppc_fp128</tt></td>
1725 <td><a name="t_firstclass">first class</a></td>
1726 <td><a href="#t_integer">integer</a>,
1727 <a href="#t_floating">floating point</a>,
1728 <a href="#t_pointer">pointer</a>,
1729 <a href="#t_vector">vector</a>,
1730 <a href="#t_struct">structure</a>,
1731 <a href="#t_array">array</a>,
1732 <a href="#t_label">label</a>,
1733 <a href="#t_metadata">metadata</a>.
1737 <td><a href="#t_primitive">primitive</a></td>
1738 <td><a href="#t_label">label</a>,
1739 <a href="#t_void">void</a>,
1740 <a href="#t_integer">integer</a>,
1741 <a href="#t_floating">floating point</a>,
1742 <a href="#t_x86mmx">x86mmx</a>,
1743 <a href="#t_metadata">metadata</a>.</td>
1746 <td><a href="#t_derived">derived</a></td>
1747 <td><a href="#t_array">array</a>,
1748 <a href="#t_function">function</a>,
1749 <a href="#t_pointer">pointer</a>,
1750 <a href="#t_struct">structure</a>,
1751 <a href="#t_vector">vector</a>,
1752 <a href="#t_opaque">opaque</a>.
1758 <p>The <a href="#t_firstclass">first class</a> types are perhaps the most
1759 important. Values of these types are the only ones which can be produced by
1764 <!-- ======================================================================= -->
1766 <a name="t_primitive">Primitive Types</a>
1771 <p>The primitive types are the fundamental building blocks of the LLVM
1774 <!-- _______________________________________________________________________ -->
1776 <a name="t_integer">Integer Type</a>
1782 <p>The integer type is a very simple type that simply specifies an arbitrary
1783 bit width for the integer type desired. Any bit width from 1 bit to
1784 2<sup>23</sup>-1 (about 8 million) can be specified.</p>
1791 <p>The number of bits the integer will occupy is specified by the <tt>N</tt>
1795 <table class="layout">
1797 <td class="left"><tt>i1</tt></td>
1798 <td class="left">a single-bit integer.</td>
1801 <td class="left"><tt>i32</tt></td>
1802 <td class="left">a 32-bit integer.</td>
1805 <td class="left"><tt>i1942652</tt></td>
1806 <td class="left">a really big integer of over 1 million bits.</td>
1812 <!-- _______________________________________________________________________ -->
1814 <a name="t_floating">Floating Point Types</a>
1821 <tr><th>Type</th><th>Description</th></tr>
1822 <tr><td><tt>float</tt></td><td>32-bit floating point value</td></tr>
1823 <tr><td><tt>double</tt></td><td>64-bit floating point value</td></tr>
1824 <tr><td><tt>fp128</tt></td><td>128-bit floating point value (112-bit mantissa)</td></tr>
1825 <tr><td><tt>x86_fp80</tt></td><td>80-bit floating point value (X87)</td></tr>
1826 <tr><td><tt>ppc_fp128</tt></td><td>128-bit floating point value (two 64-bits)</td></tr>
1832 <!-- _______________________________________________________________________ -->
1834 <a name="t_x86mmx">X86mmx Type</a>
1840 <p>The x86mmx type represents a value held in an MMX register on an x86 machine. The operations allowed on it are quite limited: parameters and return values, load and store, and bitcast. User-specified MMX instructions are represented as intrinsic or asm calls with arguments and/or results of this type. There are no arrays, vectors or constants of this type.</p>
1849 <!-- _______________________________________________________________________ -->
1851 <a name="t_void">Void Type</a>
1857 <p>The void type does not represent any value and has no size.</p>
1866 <!-- _______________________________________________________________________ -->
1868 <a name="t_label">Label Type</a>
1874 <p>The label type represents code labels.</p>
1883 <!-- _______________________________________________________________________ -->
1885 <a name="t_metadata">Metadata Type</a>
1891 <p>The metadata type represents embedded metadata. No derived types may be
1892 created from metadata except for <a href="#t_function">function</a>
1904 <!-- ======================================================================= -->
1906 <a name="t_derived">Derived Types</a>
1911 <p>The real power in LLVM comes from the derived types in the system. This is
1912 what allows a programmer to represent arrays, functions, pointers, and other
1913 useful types. Each of these types contain one or more element types which
1914 may be a primitive type, or another derived type. For example, it is
1915 possible to have a two dimensional array, using an array as the element type
1916 of another array.</p>
1921 <!-- _______________________________________________________________________ -->
1923 <a name="t_aggregate">Aggregate Types</a>
1928 <p>Aggregate Types are a subset of derived types that can contain multiple
1929 member types. <a href="#t_array">Arrays</a>,
1930 <a href="#t_struct">structs</a>, and <a href="#t_vector">vectors</a> are
1931 aggregate types.</p>
1935 <!-- _______________________________________________________________________ -->
1937 <a name="t_array">Array Type</a>
1943 <p>The array type is a very simple derived type that arranges elements
1944 sequentially in memory. The array type requires a size (number of elements)
1945 and an underlying data type.</p>
1949 [<# elements> x <elementtype>]
1952 <p>The number of elements is a constant integer value; <tt>elementtype</tt> may
1953 be any type with a size.</p>
1956 <table class="layout">
1958 <td class="left"><tt>[40 x i32]</tt></td>
1959 <td class="left">Array of 40 32-bit integer values.</td>
1962 <td class="left"><tt>[41 x i32]</tt></td>
1963 <td class="left">Array of 41 32-bit integer values.</td>
1966 <td class="left"><tt>[4 x i8]</tt></td>
1967 <td class="left">Array of 4 8-bit integer values.</td>
1970 <p>Here are some examples of multidimensional arrays:</p>
1971 <table class="layout">
1973 <td class="left"><tt>[3 x [4 x i32]]</tt></td>
1974 <td class="left">3x4 array of 32-bit integer values.</td>
1977 <td class="left"><tt>[12 x [10 x float]]</tt></td>
1978 <td class="left">12x10 array of single precision floating point values.</td>
1981 <td class="left"><tt>[2 x [3 x [4 x i16]]]</tt></td>
1982 <td class="left">2x3x4 array of 16-bit integer values.</td>
1986 <p>There is no restriction on indexing beyond the end of the array implied by
1987 a static type (though there are restrictions on indexing beyond the bounds
1988 of an allocated object in some cases). This means that single-dimension
1989 'variable sized array' addressing can be implemented in LLVM with a zero
1990 length array type. An implementation of 'pascal style arrays' in LLVM could
1991 use the type "<tt>{ i32, [0 x float]}</tt>", for example.</p>
1995 <!-- _______________________________________________________________________ -->
1997 <a name="t_function">Function Type</a>
2003 <p>The function type can be thought of as a function signature. It consists of
2004 a return type and a list of formal parameter types. The return type of a
2005 function type is a first class type or a void type.</p>
2009 <returntype> (<parameter list>)
2012 <p>...where '<tt><parameter list></tt>' is a comma-separated list of type
2013 specifiers. Optionally, the parameter list may include a type <tt>...</tt>,
2014 which indicates that the function takes a variable number of arguments.
2015 Variable argument functions can access their arguments with
2016 the <a href="#int_varargs">variable argument handling intrinsic</a>
2017 functions. '<tt><returntype></tt>' is any type except
2018 <a href="#t_label">label</a>.</p>
2021 <table class="layout">
2023 <td class="left"><tt>i32 (i32)</tt></td>
2024 <td class="left">function taking an <tt>i32</tt>, returning an <tt>i32</tt>
2026 </tr><tr class="layout">
2027 <td class="left"><tt>float (i16, i32 *) *
2029 <td class="left"><a href="#t_pointer">Pointer</a> to a function that takes
2030 an <tt>i16</tt> and a <a href="#t_pointer">pointer</a> to <tt>i32</tt>,
2031 returning <tt>float</tt>.
2033 </tr><tr class="layout">
2034 <td class="left"><tt>i32 (i8*, ...)</tt></td>
2035 <td class="left">A vararg function that takes at least one
2036 <a href="#t_pointer">pointer</a> to <tt>i8 </tt> (char in C),
2037 which returns an integer. This is the signature for <tt>printf</tt> in
2040 </tr><tr class="layout">
2041 <td class="left"><tt>{i32, i32} (i32)</tt></td>
2042 <td class="left">A function taking an <tt>i32</tt>, returning a
2043 <a href="#t_struct">structure</a> containing two <tt>i32</tt> values
2050 <!-- _______________________________________________________________________ -->
2052 <a name="t_struct">Structure Type</a>
2058 <p>The structure type is used to represent a collection of data members together
2059 in memory. The elements of a structure may be any type that has a size.</p>
2061 <p>Structures in memory are accessed using '<tt><a href="#i_load">load</a></tt>'
2062 and '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a field
2063 with the '<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.
2064 Structures in registers are accessed using the
2065 '<tt><a href="#i_extractvalue">extractvalue</a></tt>' and
2066 '<tt><a href="#i_insertvalue">insertvalue</a></tt>' instructions.</p>
2068 <p>Structures may optionally be "packed" structures, which indicate that the
2069 alignment of the struct is one byte, and that there is no padding between
2070 the elements. In non-packed structs, padding between field types is inserted
2071 as defined by the TargetData string in the module, which is required to match
2072 what the underlying processor expects.</p>
2074 <p>Structures can either be "literal" or "identified". A literal structure is
2075 defined inline with other types (e.g. <tt>{i32, i32}*</tt>) whereas identified
2076 types are always defined at the top level with a name. Literal types are
2077 uniqued by their contents and can never be recursive or opaque since there is
2078 no way to write one. Identified types can be recursive, can be opaqued, and are
2084 %T1 = type { <type list> } <i>; Identified normal struct type</i>
2085 %T2 = type <{ <type list> }> <i>; Identified packed struct type</i>
2089 <table class="layout">
2091 <td class="left"><tt>{ i32, i32, i32 }</tt></td>
2092 <td class="left">A triple of three <tt>i32</tt> values</td>
2095 <td class="left"><tt>{ float, i32 (i32) * }</tt></td>
2096 <td class="left">A pair, where the first element is a <tt>float</tt> and the
2097 second element is a <a href="#t_pointer">pointer</a> to a
2098 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
2099 an <tt>i32</tt>.</td>
2102 <td class="left"><tt><{ i8, i32 }></tt></td>
2103 <td class="left">A packed struct known to be 5 bytes in size.</td>
2109 <!-- _______________________________________________________________________ -->
2111 <a name="t_opaque">Opaque Structure Types</a>
2117 <p>Opaque structure types are used to represent named structure types that do
2118 not have a body specified. This corresponds (for example) to the C notion of
2119 a forward declared structure.</p>
2128 <table class="layout">
2130 <td class="left"><tt>opaque</tt></td>
2131 <td class="left">An opaque type.</td>
2139 <!-- _______________________________________________________________________ -->
2141 <a name="t_pointer">Pointer Type</a>
2147 <p>The pointer type is used to specify memory locations.
2148 Pointers are commonly used to reference objects in memory.</p>
2150 <p>Pointer types may have an optional address space attribute defining the
2151 numbered address space where the pointed-to object resides. The default
2152 address space is number zero. The semantics of non-zero address
2153 spaces are target-specific.</p>
2155 <p>Note that LLVM does not permit pointers to void (<tt>void*</tt>) nor does it
2156 permit pointers to labels (<tt>label*</tt>). Use <tt>i8*</tt> instead.</p>
2164 <table class="layout">
2166 <td class="left"><tt>[4 x i32]*</tt></td>
2167 <td class="left">A <a href="#t_pointer">pointer</a> to <a
2168 href="#t_array">array</a> of four <tt>i32</tt> values.</td>
2171 <td class="left"><tt>i32 (i32*) *</tt></td>
2172 <td class="left"> A <a href="#t_pointer">pointer</a> to a <a
2173 href="#t_function">function</a> that takes an <tt>i32*</tt>, returning an
2177 <td class="left"><tt>i32 addrspace(5)*</tt></td>
2178 <td class="left">A <a href="#t_pointer">pointer</a> to an <tt>i32</tt> value
2179 that resides in address space #5.</td>
2185 <!-- _______________________________________________________________________ -->
2187 <a name="t_vector">Vector Type</a>
2193 <p>A vector type is a simple derived type that represents a vector of elements.
2194 Vector types are used when multiple primitive data are operated in parallel
2195 using a single instruction (SIMD). A vector type requires a size (number of
2196 elements) and an underlying primitive data type. Vector types are considered
2197 <a href="#t_firstclass">first class</a>.</p>
2201 < <# elements> x <elementtype> >
2204 <p>The number of elements is a constant integer value larger than 0; elementtype
2205 may be any integer or floating point type. Vectors of size zero are not
2206 allowed, and pointers are not allowed as the element type.</p>
2209 <table class="layout">
2211 <td class="left"><tt><4 x i32></tt></td>
2212 <td class="left">Vector of 4 32-bit integer values.</td>
2215 <td class="left"><tt><8 x float></tt></td>
2216 <td class="left">Vector of 8 32-bit floating-point values.</td>
2219 <td class="left"><tt><2 x i64></tt></td>
2220 <td class="left">Vector of 2 64-bit integer values.</td>
2228 <!-- *********************************************************************** -->
2229 <h2><a name="constants">Constants</a></h2>
2230 <!-- *********************************************************************** -->
2234 <p>LLVM has several different basic types of constants. This section describes
2235 them all and their syntax.</p>
2237 <!-- ======================================================================= -->
2239 <a name="simpleconstants">Simple Constants</a>
2245 <dt><b>Boolean constants</b></dt>
2246 <dd>The two strings '<tt>true</tt>' and '<tt>false</tt>' are both valid
2247 constants of the <tt><a href="#t_integer">i1</a></tt> type.</dd>
2249 <dt><b>Integer constants</b></dt>
2250 <dd>Standard integers (such as '4') are constants of
2251 the <a href="#t_integer">integer</a> type. Negative numbers may be used
2252 with integer types.</dd>
2254 <dt><b>Floating point constants</b></dt>
2255 <dd>Floating point constants use standard decimal notation (e.g. 123.421),
2256 exponential notation (e.g. 1.23421e+2), or a more precise hexadecimal
2257 notation (see below). The assembler requires the exact decimal value of a
2258 floating-point constant. For example, the assembler accepts 1.25 but
2259 rejects 1.3 because 1.3 is a repeating decimal in binary. Floating point
2260 constants must have a <a href="#t_floating">floating point</a> type. </dd>
2262 <dt><b>Null pointer constants</b></dt>
2263 <dd>The identifier '<tt>null</tt>' is recognized as a null pointer constant
2264 and must be of <a href="#t_pointer">pointer type</a>.</dd>
2267 <p>The one non-intuitive notation for constants is the hexadecimal form of
2268 floating point constants. For example, the form '<tt>double
2269 0x432ff973cafa8000</tt>' is equivalent to (but harder to read than)
2270 '<tt>double 4.5e+15</tt>'. The only time hexadecimal floating point
2271 constants are required (and the only time that they are generated by the
2272 disassembler) is when a floating point constant must be emitted but it cannot
2273 be represented as a decimal floating point number in a reasonable number of
2274 digits. For example, NaN's, infinities, and other special values are
2275 represented in their IEEE hexadecimal format so that assembly and disassembly
2276 do not cause any bits to change in the constants.</p>
2278 <p>When using the hexadecimal form, constants of types float and double are
2279 represented using the 16-digit form shown above (which matches the IEEE754
2280 representation for double); float values must, however, be exactly
2281 representable as IEE754 single precision. Hexadecimal format is always used
2282 for long double, and there are three forms of long double. The 80-bit format
2283 used by x86 is represented as <tt>0xK</tt> followed by 20 hexadecimal digits.
2284 The 128-bit format used by PowerPC (two adjacent doubles) is represented
2285 by <tt>0xM</tt> followed by 32 hexadecimal digits. The IEEE 128-bit format
2286 is represented by <tt>0xL</tt> followed by 32 hexadecimal digits; no
2287 currently supported target uses this format. Long doubles will only work if
2288 they match the long double format on your target. All hexadecimal formats
2289 are big-endian (sign bit at the left).</p>
2291 <p>There are no constants of type x86mmx.</p>
2294 <!-- ======================================================================= -->
2296 <a name="aggregateconstants"></a> <!-- old anchor -->
2297 <a name="complexconstants">Complex Constants</a>
2302 <p>Complex constants are a (potentially recursive) combination of simple
2303 constants and smaller complex constants.</p>
2306 <dt><b>Structure constants</b></dt>
2307 <dd>Structure constants are represented with notation similar to structure
2308 type definitions (a comma separated list of elements, surrounded by braces
2309 (<tt>{}</tt>)). For example: "<tt>{ i32 4, float 17.0, i32* @G }</tt>",
2310 where "<tt>@G</tt>" is declared as "<tt>@G = external global i32</tt>".
2311 Structure constants must have <a href="#t_struct">structure type</a>, and
2312 the number and types of elements must match those specified by the
2315 <dt><b>Array constants</b></dt>
2316 <dd>Array constants are represented with notation similar to array type
2317 definitions (a comma separated list of elements, surrounded by square
2318 brackets (<tt>[]</tt>)). For example: "<tt>[ i32 42, i32 11, i32 74
2319 ]</tt>". Array constants must have <a href="#t_array">array type</a>, and
2320 the number and types of elements must match those specified by the
2323 <dt><b>Vector constants</b></dt>
2324 <dd>Vector constants are represented with notation similar to vector type
2325 definitions (a comma separated list of elements, surrounded by
2326 less-than/greater-than's (<tt><></tt>)). For example: "<tt>< i32
2327 42, i32 11, i32 74, i32 100 ></tt>". Vector constants must
2328 have <a href="#t_vector">vector type</a>, and the number and types of
2329 elements must match those specified by the type.</dd>
2331 <dt><b>Zero initialization</b></dt>
2332 <dd>The string '<tt>zeroinitializer</tt>' can be used to zero initialize a
2333 value to zero of <em>any</em> type, including scalar and
2334 <a href="#t_aggregate">aggregate</a> types.
2335 This is often used to avoid having to print large zero initializers
2336 (e.g. for large arrays) and is always exactly equivalent to using explicit
2337 zero initializers.</dd>
2339 <dt><b>Metadata node</b></dt>
2340 <dd>A metadata node is a structure-like constant with
2341 <a href="#t_metadata">metadata type</a>. For example: "<tt>metadata !{
2342 i32 0, metadata !"test" }</tt>". Unlike other constants that are meant to
2343 be interpreted as part of the instruction stream, metadata is a place to
2344 attach additional information such as debug info.</dd>
2349 <!-- ======================================================================= -->
2351 <a name="globalconstants">Global Variable and Function Addresses</a>
2356 <p>The addresses of <a href="#globalvars">global variables</a>
2357 and <a href="#functionstructure">functions</a> are always implicitly valid
2358 (link-time) constants. These constants are explicitly referenced when
2359 the <a href="#identifiers">identifier for the global</a> is used and always
2360 have <a href="#t_pointer">pointer</a> type. For example, the following is a
2361 legal LLVM file:</p>
2363 <pre class="doc_code">
2366 @Z = global [2 x i32*] [ i32* @X, i32* @Y ]
2371 <!-- ======================================================================= -->
2373 <a name="undefvalues">Undefined Values</a>
2378 <p>The string '<tt>undef</tt>' can be used anywhere a constant is expected, and
2379 indicates that the user of the value may receive an unspecified bit-pattern.
2380 Undefined values may be of any type (other than '<tt>label</tt>'
2381 or '<tt>void</tt>') and be used anywhere a constant is permitted.</p>
2383 <p>Undefined values are useful because they indicate to the compiler that the
2384 program is well defined no matter what value is used. This gives the
2385 compiler more freedom to optimize. Here are some examples of (potentially
2386 surprising) transformations that are valid (in pseudo IR):</p>
2389 <pre class="doc_code">
2399 <p>This is safe because all of the output bits are affected by the undef bits.
2400 Any output bit can have a zero or one depending on the input bits.</p>
2402 <pre class="doc_code">
2413 <p>These logical operations have bits that are not always affected by the input.
2414 For example, if <tt>%X</tt> has a zero bit, then the output of the
2415 '<tt>and</tt>' operation will always be a zero for that bit, no matter what
2416 the corresponding bit from the '<tt>undef</tt>' is. As such, it is unsafe to
2417 optimize or assume that the result of the '<tt>and</tt>' is '<tt>undef</tt>'.
2418 However, it is safe to assume that all bits of the '<tt>undef</tt>' could be
2419 0, and optimize the '<tt>and</tt>' to 0. Likewise, it is safe to assume that
2420 all the bits of the '<tt>undef</tt>' operand to the '<tt>or</tt>' could be
2421 set, allowing the '<tt>or</tt>' to be folded to -1.</p>
2423 <pre class="doc_code">
2424 %A = select undef, %X, %Y
2425 %B = select undef, 42, %Y
2426 %C = select %X, %Y, undef
2437 <p>This set of examples shows that undefined '<tt>select</tt>' (and conditional
2438 branch) conditions can go <em>either way</em>, but they have to come from one
2439 of the two operands. In the <tt>%A</tt> example, if <tt>%X</tt> and
2440 <tt>%Y</tt> were both known to have a clear low bit, then <tt>%A</tt> would
2441 have to have a cleared low bit. However, in the <tt>%C</tt> example, the
2442 optimizer is allowed to assume that the '<tt>undef</tt>' operand could be the
2443 same as <tt>%Y</tt>, allowing the whole '<tt>select</tt>' to be
2446 <pre class="doc_code">
2447 %A = xor undef, undef
2465 <p>This example points out that two '<tt>undef</tt>' operands are not
2466 necessarily the same. This can be surprising to people (and also matches C
2467 semantics) where they assume that "<tt>X^X</tt>" is always zero, even
2468 if <tt>X</tt> is undefined. This isn't true for a number of reasons, but the
2469 short answer is that an '<tt>undef</tt>' "variable" can arbitrarily change
2470 its value over its "live range". This is true because the variable doesn't
2471 actually <em>have a live range</em>. Instead, the value is logically read
2472 from arbitrary registers that happen to be around when needed, so the value
2473 is not necessarily consistent over time. In fact, <tt>%A</tt> and <tt>%C</tt>
2474 need to have the same semantics or the core LLVM "replace all uses with"
2475 concept would not hold.</p>
2477 <pre class="doc_code">
2485 <p>These examples show the crucial difference between an <em>undefined
2486 value</em> and <em>undefined behavior</em>. An undefined value (like
2487 '<tt>undef</tt>') is allowed to have an arbitrary bit-pattern. This means that
2488 the <tt>%A</tt> operation can be constant folded to '<tt>undef</tt>', because
2489 the '<tt>undef</tt>' could be an SNaN, and <tt>fdiv</tt> is not (currently)
2490 defined on SNaN's. However, in the second example, we can make a more
2491 aggressive assumption: because the <tt>undef</tt> is allowed to be an
2492 arbitrary value, we are allowed to assume that it could be zero. Since a
2493 divide by zero has <em>undefined behavior</em>, we are allowed to assume that
2494 the operation does not execute at all. This allows us to delete the divide and
2495 all code after it. Because the undefined operation "can't happen", the
2496 optimizer can assume that it occurs in dead code.</p>
2498 <pre class="doc_code">
2499 a: store undef -> %X
2500 b: store %X -> undef
2506 <p>These examples reiterate the <tt>fdiv</tt> example: a store <em>of</em> an
2507 undefined value can be assumed to not have any effect; we can assume that the
2508 value is overwritten with bits that happen to match what was already there.
2509 However, a store <em>to</em> an undefined location could clobber arbitrary
2510 memory, therefore, it has undefined behavior.</p>
2514 <!-- ======================================================================= -->
2516 <a name="trapvalues">Trap Values</a>
2521 <p>Trap values are similar to <a href="#undefvalues">undef values</a>, however
2522 instead of representing an unspecified bit pattern, they represent the
2523 fact that an instruction or constant expression which cannot evoke side
2524 effects has nevertheless detected a condition which results in undefined
2527 <p>There is currently no way of representing a trap value in the IR; they
2528 only exist when produced by operations such as
2529 <a href="#i_add"><tt>add</tt></a> with the <tt>nsw</tt> flag.</p>
2531 <p>Trap value behavior is defined in terms of value <i>dependence</i>:</p>
2534 <li>Values other than <a href="#i_phi"><tt>phi</tt></a> nodes depend on
2535 their operands.</li>
2537 <li><a href="#i_phi"><tt>Phi</tt></a> nodes depend on the operand corresponding
2538 to their dynamic predecessor basic block.</li>
2540 <li>Function arguments depend on the corresponding actual argument values in
2541 the dynamic callers of their functions.</li>
2543 <li><a href="#i_call"><tt>Call</tt></a> instructions depend on the
2544 <a href="#i_ret"><tt>ret</tt></a> instructions that dynamically transfer
2545 control back to them.</li>
2547 <li><a href="#i_invoke"><tt>Invoke</tt></a> instructions depend on the
2548 <a href="#i_ret"><tt>ret</tt></a>, <a href="#i_unwind"><tt>unwind</tt></a>,
2549 or exception-throwing call instructions that dynamically transfer control
2552 <li>Non-volatile loads and stores depend on the most recent stores to all of the
2553 referenced memory addresses, following the order in the IR
2554 (including loads and stores implied by intrinsics such as
2555 <a href="#int_memcpy"><tt>@llvm.memcpy</tt></a>.)</li>
2557 <!-- TODO: In the case of multiple threads, this only applies if the store
2558 "happens-before" the load or store. -->
2560 <!-- TODO: floating-point exception state -->
2562 <li>An instruction with externally visible side effects depends on the most
2563 recent preceding instruction with externally visible side effects, following
2564 the order in the IR. (This includes
2565 <a href="#volatile">volatile operations</a>.)</li>
2567 <li>An instruction <i>control-depends</i> on a
2568 <a href="#terminators">terminator instruction</a>
2569 if the terminator instruction has multiple successors and the instruction
2570 is always executed when control transfers to one of the successors, and
2571 may not be executed when control is transferred to another.</li>
2573 <li>Additionally, an instruction also <i>control-depends</i> on a terminator
2574 instruction if the set of instructions it otherwise depends on would be
2575 different if the terminator had transferred control to a different
2578 <li>Dependence is transitive.</li>
2582 <p>Whenever a trap value is generated, all values which depend on it evaluate
2583 to trap. If they have side effects, the evoke their side effects as if each
2584 operand with a trap value were undef. If they have externally-visible side
2585 effects, the behavior is undefined.</p>
2587 <p>Here are some examples:</p>
2589 <pre class="doc_code">
2591 %trap = sub nuw i32 0, 1 ; Results in a trap value.
2592 %still_trap = and i32 %trap, 0 ; Whereas (and i32 undef, 0) would return 0.
2593 %trap_yet_again = getelementptr i32* @h, i32 %still_trap
2594 store i32 0, i32* %trap_yet_again ; undefined behavior
2596 store i32 %trap, i32* @g ; Trap value conceptually stored to memory.
2597 %trap2 = load i32* @g ; Returns a trap value, not just undef.
2599 volatile store i32 %trap, i32* @g ; External observation; undefined behavior.
2601 %narrowaddr = bitcast i32* @g to i16*
2602 %wideaddr = bitcast i32* @g to i64*
2603 %trap3 = load i16* %narrowaddr ; Returns a trap value.
2604 %trap4 = load i64* %wideaddr ; Returns a trap value.
2606 %cmp = icmp slt i32 %trap, 0 ; Returns a trap value.
2607 br i1 %cmp, label %true, label %end ; Branch to either destination.
2610 volatile store i32 0, i32* @g ; This is control-dependent on %cmp, so
2611 ; it has undefined behavior.
2615 %p = phi i32 [ 0, %entry ], [ 1, %true ]
2616 ; Both edges into this PHI are
2617 ; control-dependent on %cmp, so this
2618 ; always results in a trap value.
2620 volatile store i32 0, i32* @g ; This would depend on the store in %true
2621 ; if %cmp is true, or the store in %entry
2622 ; otherwise, so this is undefined behavior.
2624 br i1 %cmp, label %second_true, label %second_end
2625 ; The same branch again, but this time the
2626 ; true block doesn't have side effects.
2633 volatile store i32 0, i32* @g ; This time, the instruction always depends
2634 ; on the store in %end. Also, it is
2635 ; control-equivalent to %end, so this is
2636 ; well-defined (again, ignoring earlier
2637 ; undefined behavior in this example).
2642 <!-- ======================================================================= -->
2644 <a name="blockaddress">Addresses of Basic Blocks</a>
2649 <p><b><tt>blockaddress(@function, %block)</tt></b></p>
2651 <p>The '<tt>blockaddress</tt>' constant computes the address of the specified
2652 basic block in the specified function, and always has an i8* type. Taking
2653 the address of the entry block is illegal.</p>
2655 <p>This value only has defined behavior when used as an operand to the
2656 '<a href="#i_indirectbr"><tt>indirectbr</tt></a>' instruction, or for
2657 comparisons against null. Pointer equality tests between labels addresses
2658 results in undefined behavior — though, again, comparison against null
2659 is ok, and no label is equal to the null pointer. This may be passed around
2660 as an opaque pointer sized value as long as the bits are not inspected. This
2661 allows <tt>ptrtoint</tt> and arithmetic to be performed on these values so
2662 long as the original value is reconstituted before the <tt>indirectbr</tt>
2665 <p>Finally, some targets may provide defined semantics when using the value as
2666 the operand to an inline assembly, but that is target specific.</p>
2671 <!-- ======================================================================= -->
2673 <a name="constantexprs">Constant Expressions</a>
2678 <p>Constant expressions are used to allow expressions involving other constants
2679 to be used as constants. Constant expressions may be of
2680 any <a href="#t_firstclass">first class</a> type and may involve any LLVM
2681 operation that does not have side effects (e.g. load and call are not
2682 supported). The following is the syntax for constant expressions:</p>
2685 <dt><b><tt>trunc (CST to TYPE)</tt></b></dt>
2686 <dd>Truncate a constant to another type. The bit size of CST must be larger
2687 than the bit size of TYPE. Both types must be integers.</dd>
2689 <dt><b><tt>zext (CST to TYPE)</tt></b></dt>
2690 <dd>Zero extend a constant to another type. The bit size of CST must be
2691 smaller than the bit size of TYPE. Both types must be integers.</dd>
2693 <dt><b><tt>sext (CST to TYPE)</tt></b></dt>
2694 <dd>Sign extend a constant to another type. The bit size of CST must be
2695 smaller than the bit size of TYPE. Both types must be integers.</dd>
2697 <dt><b><tt>fptrunc (CST to TYPE)</tt></b></dt>
2698 <dd>Truncate a floating point constant to another floating point type. The
2699 size of CST must be larger than the size of TYPE. Both types must be
2700 floating point.</dd>
2702 <dt><b><tt>fpext (CST to TYPE)</tt></b></dt>
2703 <dd>Floating point extend a constant to another type. The size of CST must be
2704 smaller or equal to the size of TYPE. Both types must be floating
2707 <dt><b><tt>fptoui (CST to TYPE)</tt></b></dt>
2708 <dd>Convert a floating point constant to the corresponding unsigned integer
2709 constant. TYPE must be a scalar or vector integer type. CST must be of
2710 scalar or vector floating point type. Both CST and TYPE must be scalars,
2711 or vectors of the same number of elements. If the value won't fit in the
2712 integer type, the results are undefined.</dd>
2714 <dt><b><tt>fptosi (CST to TYPE)</tt></b></dt>
2715 <dd>Convert a floating point constant to the corresponding signed integer
2716 constant. TYPE must be a scalar or vector integer type. CST must be of
2717 scalar or vector floating point type. Both CST and TYPE must be scalars,
2718 or vectors of the same number of elements. If the value won't fit in the
2719 integer type, the results are undefined.</dd>
2721 <dt><b><tt>uitofp (CST to TYPE)</tt></b></dt>
2722 <dd>Convert an unsigned integer constant to the corresponding floating point
2723 constant. TYPE must be a scalar or vector floating point type. CST must be
2724 of scalar or vector integer type. Both CST and TYPE must be scalars, or
2725 vectors of the same number of elements. If the value won't fit in the
2726 floating point type, the results are undefined.</dd>
2728 <dt><b><tt>sitofp (CST to TYPE)</tt></b></dt>
2729 <dd>Convert a signed integer constant to the corresponding floating point
2730 constant. TYPE must be a scalar or vector floating point type. CST must be
2731 of scalar or vector integer type. Both CST and TYPE must be scalars, or
2732 vectors of the same number of elements. If the value won't fit in the
2733 floating point type, the results are undefined.</dd>
2735 <dt><b><tt>ptrtoint (CST to TYPE)</tt></b></dt>
2736 <dd>Convert a pointer typed constant to the corresponding integer constant
2737 <tt>TYPE</tt> must be an integer type. <tt>CST</tt> must be of pointer
2738 type. The <tt>CST</tt> value is zero extended, truncated, or unchanged to
2739 make it fit in <tt>TYPE</tt>.</dd>
2741 <dt><b><tt>inttoptr (CST to TYPE)</tt></b></dt>
2742 <dd>Convert a integer constant to a pointer constant. TYPE must be a pointer
2743 type. CST must be of integer type. The CST value is zero extended,
2744 truncated, or unchanged to make it fit in a pointer size. This one is
2745 <i>really</i> dangerous!</dd>
2747 <dt><b><tt>bitcast (CST to TYPE)</tt></b></dt>
2748 <dd>Convert a constant, CST, to another TYPE. The constraints of the operands
2749 are the same as those for the <a href="#i_bitcast">bitcast
2750 instruction</a>.</dd>
2752 <dt><b><tt>getelementptr (CSTPTR, IDX0, IDX1, ...)</tt></b></dt>
2753 <dt><b><tt>getelementptr inbounds (CSTPTR, IDX0, IDX1, ...)</tt></b></dt>
2754 <dd>Perform the <a href="#i_getelementptr">getelementptr operation</a> on
2755 constants. As with the <a href="#i_getelementptr">getelementptr</a>
2756 instruction, the index list may have zero or more indexes, which are
2757 required to make sense for the type of "CSTPTR".</dd>
2759 <dt><b><tt>select (COND, VAL1, VAL2)</tt></b></dt>
2760 <dd>Perform the <a href="#i_select">select operation</a> on constants.</dd>
2762 <dt><b><tt>icmp COND (VAL1, VAL2)</tt></b></dt>
2763 <dd>Performs the <a href="#i_icmp">icmp operation</a> on constants.</dd>
2765 <dt><b><tt>fcmp COND (VAL1, VAL2)</tt></b></dt>
2766 <dd>Performs the <a href="#i_fcmp">fcmp operation</a> on constants.</dd>
2768 <dt><b><tt>extractelement (VAL, IDX)</tt></b></dt>
2769 <dd>Perform the <a href="#i_extractelement">extractelement operation</a> on
2772 <dt><b><tt>insertelement (VAL, ELT, IDX)</tt></b></dt>
2773 <dd>Perform the <a href="#i_insertelement">insertelement operation</a> on
2776 <dt><b><tt>shufflevector (VEC1, VEC2, IDXMASK)</tt></b></dt>
2777 <dd>Perform the <a href="#i_shufflevector">shufflevector operation</a> on
2780 <dt><b><tt>extractvalue (VAL, IDX0, IDX1, ...)</tt></b></dt>
2781 <dd>Perform the <a href="#i_extractvalue">extractvalue operation</a> on
2782 constants. The index list is interpreted in a similar manner as indices in
2783 a '<a href="#i_getelementptr">getelementptr</a>' operation. At least one
2784 index value must be specified.</dd>
2786 <dt><b><tt>insertvalue (VAL, ELT, IDX0, IDX1, ...)</tt></b></dt>
2787 <dd>Perform the <a href="#i_insertvalue">insertvalue operation</a> on
2788 constants. The index list is interpreted in a similar manner as indices in
2789 a '<a href="#i_getelementptr">getelementptr</a>' operation. At least one
2790 index value must be specified.</dd>
2792 <dt><b><tt>OPCODE (LHS, RHS)</tt></b></dt>
2793 <dd>Perform the specified operation of the LHS and RHS constants. OPCODE may
2794 be any of the <a href="#binaryops">binary</a>
2795 or <a href="#bitwiseops">bitwise binary</a> operations. The constraints
2796 on operands are the same as those for the corresponding instruction
2797 (e.g. no bitwise operations on floating point values are allowed).</dd>
2804 <!-- *********************************************************************** -->
2805 <h2><a name="othervalues">Other Values</a></h2>
2806 <!-- *********************************************************************** -->
2808 <!-- ======================================================================= -->
2810 <a name="inlineasm">Inline Assembler Expressions</a>
2815 <p>LLVM supports inline assembler expressions (as opposed
2816 to <a href="#moduleasm"> Module-Level Inline Assembly</a>) through the use of
2817 a special value. This value represents the inline assembler as a string
2818 (containing the instructions to emit), a list of operand constraints (stored
2819 as a string), a flag that indicates whether or not the inline asm
2820 expression has side effects, and a flag indicating whether the function
2821 containing the asm needs to align its stack conservatively. An example
2822 inline assembler expression is:</p>
2824 <pre class="doc_code">
2825 i32 (i32) asm "bswap $0", "=r,r"
2828 <p>Inline assembler expressions may <b>only</b> be used as the callee operand of
2829 a <a href="#i_call"><tt>call</tt> instruction</a>. Thus, typically we
2832 <pre class="doc_code">
2833 %X = call i32 asm "<a href="#int_bswap">bswap</a> $0", "=r,r"(i32 %Y)
2836 <p>Inline asms with side effects not visible in the constraint list must be
2837 marked as having side effects. This is done through the use of the
2838 '<tt>sideeffect</tt>' keyword, like so:</p>
2840 <pre class="doc_code">
2841 call void asm sideeffect "eieio", ""()
2844 <p>In some cases inline asms will contain code that will not work unless the
2845 stack is aligned in some way, such as calls or SSE instructions on x86,
2846 yet will not contain code that does that alignment within the asm.
2847 The compiler should make conservative assumptions about what the asm might
2848 contain and should generate its usual stack alignment code in the prologue
2849 if the '<tt>alignstack</tt>' keyword is present:</p>
2851 <pre class="doc_code">
2852 call void asm alignstack "eieio", ""()
2855 <p>If both keywords appear the '<tt>sideeffect</tt>' keyword must come
2858 <p>TODO: The format of the asm and constraints string still need to be
2859 documented here. Constraints on what can be done (e.g. duplication, moving,
2860 etc need to be documented). This is probably best done by reference to
2861 another document that covers inline asm from a holistic perspective.</p>
2864 <a name="inlineasm_md">Inline Asm Metadata</a>
2869 <p>The call instructions that wrap inline asm nodes may have a "!srcloc" MDNode
2870 attached to it that contains a list of constant integers. If present, the
2871 code generator will use the integer as the location cookie value when report
2872 errors through the LLVMContext error reporting mechanisms. This allows a
2873 front-end to correlate backend errors that occur with inline asm back to the
2874 source code that produced it. For example:</p>
2876 <pre class="doc_code">
2877 call void asm sideeffect "something bad", ""()<b>, !srcloc !42</b>
2879 !42 = !{ i32 1234567 }
2882 <p>It is up to the front-end to make sense of the magic numbers it places in the
2883 IR. If the MDNode contains multiple constants, the code generator will use
2884 the one that corresponds to the line of the asm that the error occurs on.</p>
2890 <!-- ======================================================================= -->
2892 <a name="metadata">Metadata Nodes and Metadata Strings</a>
2897 <p>LLVM IR allows metadata to be attached to instructions in the program that
2898 can convey extra information about the code to the optimizers and code
2899 generator. One example application of metadata is source-level debug
2900 information. There are two metadata primitives: strings and nodes. All
2901 metadata has the <tt>metadata</tt> type and is identified in syntax by a
2902 preceding exclamation point ('<tt>!</tt>').</p>
2904 <p>A metadata string is a string surrounded by double quotes. It can contain
2905 any character by escaping non-printable characters with "\xx" where "xx" is
2906 the two digit hex code. For example: "<tt>!"test\00"</tt>".</p>
2908 <p>Metadata nodes are represented with notation similar to structure constants
2909 (a comma separated list of elements, surrounded by braces and preceded by an
2910 exclamation point). For example: "<tt>!{ metadata !"test\00", i32
2911 10}</tt>". Metadata nodes can have any values as their operand.</p>
2913 <p>A <a href="#namedmetadatastructure">named metadata</a> is a collection of
2914 metadata nodes, which can be looked up in the module symbol table. For
2915 example: "<tt>!foo = metadata !{!4, !3}</tt>".
2917 <p>Metadata can be used as function arguments. Here <tt>llvm.dbg.value</tt>
2918 function is using two metadata arguments.</p>
2920 <div class="doc_code">
2922 call void @llvm.dbg.value(metadata !24, i64 0, metadata !25)
2926 <p>Metadata can be attached with an instruction. Here metadata <tt>!21</tt> is
2927 attached with <tt>add</tt> instruction using <tt>!dbg</tt> identifier.</p>
2929 <div class="doc_code">
2931 %indvar.next = add i64 %indvar, 1, !dbg !21
2939 <!-- *********************************************************************** -->
2941 <a name="intrinsic_globals">Intrinsic Global Variables</a>
2943 <!-- *********************************************************************** -->
2945 <p>LLVM has a number of "magic" global variables that contain data that affect
2946 code generation or other IR semantics. These are documented here. All globals
2947 of this sort should have a section specified as "<tt>llvm.metadata</tt>". This
2948 section and all globals that start with "<tt>llvm.</tt>" are reserved for use
2951 <!-- ======================================================================= -->
2953 <a name="intg_used">The '<tt>llvm.used</tt>' Global Variable</a>
2958 <p>The <tt>@llvm.used</tt> global is an array with i8* element type which has <a
2959 href="#linkage_appending">appending linkage</a>. This array contains a list of
2960 pointers to global variables and functions which may optionally have a pointer
2961 cast formed of bitcast or getelementptr. For example, a legal use of it is:</p>
2967 @llvm.used = appending global [2 x i8*] [
2969 i8* bitcast (i32* @Y to i8*)
2970 ], section "llvm.metadata"
2973 <p>If a global variable appears in the <tt>@llvm.used</tt> list, then the
2974 compiler, assembler, and linker are required to treat the symbol as if there is
2975 a reference to the global that it cannot see. For example, if a variable has
2976 internal linkage and no references other than that from the <tt>@llvm.used</tt>
2977 list, it cannot be deleted. This is commonly used to represent references from
2978 inline asms and other things the compiler cannot "see", and corresponds to
2979 "attribute((used))" in GNU C.</p>
2981 <p>On some targets, the code generator must emit a directive to the assembler or
2982 object file to prevent the assembler and linker from molesting the symbol.</p>
2986 <!-- ======================================================================= -->
2988 <a name="intg_compiler_used">
2989 The '<tt>llvm.compiler.used</tt>' Global Variable
2995 <p>The <tt>@llvm.compiler.used</tt> directive is the same as the
2996 <tt>@llvm.used</tt> directive, except that it only prevents the compiler from
2997 touching the symbol. On targets that support it, this allows an intelligent
2998 linker to optimize references to the symbol without being impeded as it would be
2999 by <tt>@llvm.used</tt>.</p>
3001 <p>This is a rare construct that should only be used in rare circumstances, and
3002 should not be exposed to source languages.</p>
3006 <!-- ======================================================================= -->
3008 <a name="intg_global_ctors">The '<tt>llvm.global_ctors</tt>' Global Variable</a>
3013 %0 = type { i32, void ()* }
3014 @llvm.global_ctors = appending global [1 x %0] [%0 { i32 65535, void ()* @ctor }]
3016 <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.
3021 <!-- ======================================================================= -->
3023 <a name="intg_global_dtors">The '<tt>llvm.global_dtors</tt>' Global Variable</a>
3028 %0 = type { i32, void ()* }
3029 @llvm.global_dtors = appending global [1 x %0] [%0 { i32 65535, void ()* @dtor }]
3032 <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.
3039 <!-- *********************************************************************** -->
3040 <h2><a name="instref">Instruction Reference</a></h2>
3041 <!-- *********************************************************************** -->
3045 <p>The LLVM instruction set consists of several different classifications of
3046 instructions: <a href="#terminators">terminator
3047 instructions</a>, <a href="#binaryops">binary instructions</a>,
3048 <a href="#bitwiseops">bitwise binary instructions</a>,
3049 <a href="#memoryops">memory instructions</a>, and
3050 <a href="#otherops">other instructions</a>.</p>
3052 <!-- ======================================================================= -->
3054 <a name="terminators">Terminator Instructions</a>
3059 <p>As mentioned <a href="#functionstructure">previously</a>, every basic block
3060 in a program ends with a "Terminator" instruction, which indicates which
3061 block should be executed after the current block is finished. These
3062 terminator instructions typically yield a '<tt>void</tt>' value: they produce
3063 control flow, not values (the one exception being the
3064 '<a href="#i_invoke"><tt>invoke</tt></a>' instruction).</p>
3066 <p>The terminator instructions are:
3067 '<a href="#i_ret"><tt>ret</tt></a>',
3068 '<a href="#i_br"><tt>br</tt></a>',
3069 '<a href="#i_switch"><tt>switch</tt></a>',
3070 '<a href="#i_indirectbr"><tt>indirectbr</tt></a>',
3071 '<a href="#i_invoke"><tt>invoke</tt></a>',
3072 '<a href="#i_unwind"><tt>unwind</tt></a>',
3073 '<a href="#i_resume"><tt>resume</tt></a>', and
3074 '<a href="#i_unreachable"><tt>unreachable</tt></a>'.</p>
3076 <!-- _______________________________________________________________________ -->
3078 <a name="i_ret">'<tt>ret</tt>' Instruction</a>
3085 ret <type> <value> <i>; Return a value from a non-void function</i>
3086 ret void <i>; Return from void function</i>
3090 <p>The '<tt>ret</tt>' instruction is used to return control flow (and optionally
3091 a value) from a function back to the caller.</p>
3093 <p>There are two forms of the '<tt>ret</tt>' instruction: one that returns a
3094 value and then causes control flow, and one that just causes control flow to
3098 <p>The '<tt>ret</tt>' instruction optionally accepts a single argument, the
3099 return value. The type of the return value must be a
3100 '<a href="#t_firstclass">first class</a>' type.</p>
3102 <p>A function is not <a href="#wellformed">well formed</a> if it it has a
3103 non-void return type and contains a '<tt>ret</tt>' instruction with no return
3104 value or a return value with a type that does not match its type, or if it
3105 has a void return type and contains a '<tt>ret</tt>' instruction with a
3109 <p>When the '<tt>ret</tt>' instruction is executed, control flow returns back to
3110 the calling function's context. If the caller is a
3111 "<a href="#i_call"><tt>call</tt></a>" instruction, execution continues at the
3112 instruction after the call. If the caller was an
3113 "<a href="#i_invoke"><tt>invoke</tt></a>" instruction, execution continues at
3114 the beginning of the "normal" destination block. If the instruction returns
3115 a value, that value shall set the call or invoke instruction's return
3120 ret i32 5 <i>; Return an integer value of 5</i>
3121 ret void <i>; Return from a void function</i>
3122 ret { i32, i8 } { i32 4, i8 2 } <i>; Return a struct of values 4 and 2</i>
3126 <!-- _______________________________________________________________________ -->
3128 <a name="i_br">'<tt>br</tt>' Instruction</a>
3135 br i1 <cond>, label <iftrue>, label <iffalse>
3136 br label <dest> <i>; Unconditional branch</i>
3140 <p>The '<tt>br</tt>' instruction is used to cause control flow to transfer to a
3141 different basic block in the current function. There are two forms of this
3142 instruction, corresponding to a conditional branch and an unconditional
3146 <p>The conditional branch form of the '<tt>br</tt>' instruction takes a single
3147 '<tt>i1</tt>' value and two '<tt>label</tt>' values. The unconditional form
3148 of the '<tt>br</tt>' instruction takes a single '<tt>label</tt>' value as a
3152 <p>Upon execution of a conditional '<tt>br</tt>' instruction, the '<tt>i1</tt>'
3153 argument is evaluated. If the value is <tt>true</tt>, control flows to the
3154 '<tt>iftrue</tt>' <tt>label</tt> argument. If "cond" is <tt>false</tt>,
3155 control flows to the '<tt>iffalse</tt>' <tt>label</tt> argument.</p>
3160 %cond = <a href="#i_icmp">icmp</a> eq i32 %a, %b
3161 br i1 %cond, label %IfEqual, label %IfUnequal
3163 <a href="#i_ret">ret</a> i32 1
3165 <a href="#i_ret">ret</a> i32 0
3170 <!-- _______________________________________________________________________ -->
3172 <a name="i_switch">'<tt>switch</tt>' Instruction</a>
3179 switch <intty> <value>, label <defaultdest> [ <intty> <val>, label <dest> ... ]
3183 <p>The '<tt>switch</tt>' instruction is used to transfer control flow to one of
3184 several different places. It is a generalization of the '<tt>br</tt>'
3185 instruction, allowing a branch to occur to one of many possible
3189 <p>The '<tt>switch</tt>' instruction uses three parameters: an integer
3190 comparison value '<tt>value</tt>', a default '<tt>label</tt>' destination,
3191 and an array of pairs of comparison value constants and '<tt>label</tt>'s.
3192 The table is not allowed to contain duplicate constant entries.</p>
3195 <p>The <tt>switch</tt> instruction specifies a table of values and
3196 destinations. When the '<tt>switch</tt>' instruction is executed, this table
3197 is searched for the given value. If the value is found, control flow is
3198 transferred to the corresponding destination; otherwise, control flow is
3199 transferred to the default destination.</p>
3201 <h5>Implementation:</h5>
3202 <p>Depending on properties of the target machine and the particular
3203 <tt>switch</tt> instruction, this instruction may be code generated in
3204 different ways. For example, it could be generated as a series of chained
3205 conditional branches or with a lookup table.</p>
3209 <i>; Emulate a conditional br instruction</i>
3210 %Val = <a href="#i_zext">zext</a> i1 %value to i32
3211 switch i32 %Val, label %truedest [ i32 0, label %falsedest ]
3213 <i>; Emulate an unconditional br instruction</i>
3214 switch i32 0, label %dest [ ]
3216 <i>; Implement a jump table:</i>
3217 switch i32 %val, label %otherwise [ i32 0, label %onzero
3219 i32 2, label %ontwo ]
3225 <!-- _______________________________________________________________________ -->
3227 <a name="i_indirectbr">'<tt>indirectbr</tt>' Instruction</a>
3234 indirectbr <somety>* <address>, [ label <dest1>, label <dest2>, ... ]
3239 <p>The '<tt>indirectbr</tt>' instruction implements an indirect branch to a label
3240 within the current function, whose address is specified by
3241 "<tt>address</tt>". Address must be derived from a <a
3242 href="#blockaddress">blockaddress</a> constant.</p>
3246 <p>The '<tt>address</tt>' argument is the address of the label to jump to. The
3247 rest of the arguments indicate the full set of possible destinations that the
3248 address may point to. Blocks are allowed to occur multiple times in the
3249 destination list, though this isn't particularly useful.</p>
3251 <p>This destination list is required so that dataflow analysis has an accurate
3252 understanding of the CFG.</p>
3256 <p>Control transfers to the block specified in the address argument. All
3257 possible destination blocks must be listed in the label list, otherwise this
3258 instruction has undefined behavior. This implies that jumps to labels
3259 defined in other functions have undefined behavior as well.</p>
3261 <h5>Implementation:</h5>
3263 <p>This is typically implemented with a jump through a register.</p>
3267 indirectbr i8* %Addr, [ label %bb1, label %bb2, label %bb3 ]
3273 <!-- _______________________________________________________________________ -->
3275 <a name="i_invoke">'<tt>invoke</tt>' Instruction</a>
3282 <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>]
3283 to label <normal label> unwind label <exception label>
3287 <p>The '<tt>invoke</tt>' instruction causes control to transfer to a specified
3288 function, with the possibility of control flow transfer to either the
3289 '<tt>normal</tt>' label or the '<tt>exception</tt>' label. If the callee
3290 function returns with the "<tt><a href="#i_ret">ret</a></tt>" instruction,
3291 control flow will return to the "normal" label. If the callee (or any
3292 indirect callees) returns with the "<a href="#i_unwind"><tt>unwind</tt></a>"
3293 instruction, control is interrupted and continued at the dynamically nearest
3294 "exception" label.</p>
3296 <p>The '<tt>exception</tt>' label is a
3297 <i><a href="ExceptionHandling.html#overview">landing pad</a></i> for the
3298 exception. As such, '<tt>exception</tt>' label is required to have the
3299 "<a href="#i_landingpad"><tt>landingpad</tt></a>" instruction, which contains
3300 the information about about the behavior of the program after unwinding
3301 happens, as its first non-PHI instruction. The restrictions on the
3302 "<tt>landingpad</tt>" instruction's tightly couples it to the
3303 "<tt>invoke</tt>" instruction, so that the important information contained
3304 within the "<tt>landingpad</tt>" instruction can't be lost through normal
3308 <p>This instruction requires several arguments:</p>
3311 <li>The optional "cconv" marker indicates which <a href="#callingconv">calling
3312 convention</a> the call should use. If none is specified, the call
3313 defaults to using C calling conventions.</li>
3315 <li>The optional <a href="#paramattrs">Parameter Attributes</a> list for
3316 return values. Only '<tt>zeroext</tt>', '<tt>signext</tt>', and
3317 '<tt>inreg</tt>' attributes are valid here.</li>
3319 <li>'<tt>ptr to function ty</tt>': shall be the signature of the pointer to
3320 function value being invoked. In most cases, this is a direct function
3321 invocation, but indirect <tt>invoke</tt>s are just as possible, branching
3322 off an arbitrary pointer to function value.</li>
3324 <li>'<tt>function ptr val</tt>': An LLVM value containing a pointer to a
3325 function to be invoked. </li>
3327 <li>'<tt>function args</tt>': argument list whose types match the function
3328 signature argument types and parameter attributes. All arguments must be
3329 of <a href="#t_firstclass">first class</a> type. If the function
3330 signature indicates the function accepts a variable number of arguments,
3331 the extra arguments can be specified.</li>
3333 <li>'<tt>normal label</tt>': the label reached when the called function
3334 executes a '<tt><a href="#i_ret">ret</a></tt>' instruction. </li>
3336 <li>'<tt>exception label</tt>': the label reached when a callee returns with
3337 the <a href="#i_unwind"><tt>unwind</tt></a> instruction. </li>
3339 <li>The optional <a href="#fnattrs">function attributes</a> list. Only
3340 '<tt>noreturn</tt>', '<tt>nounwind</tt>', '<tt>readonly</tt>' and
3341 '<tt>readnone</tt>' attributes are valid here.</li>
3345 <p>This instruction is designed to operate as a standard
3346 '<tt><a href="#i_call">call</a></tt>' instruction in most regards. The
3347 primary difference is that it establishes an association with a label, which
3348 is used by the runtime library to unwind the stack.</p>
3350 <p>This instruction is used in languages with destructors to ensure that proper
3351 cleanup is performed in the case of either a <tt>longjmp</tt> or a thrown
3352 exception. Additionally, this is important for implementation of
3353 '<tt>catch</tt>' clauses in high-level languages that support them.</p>
3355 <p>For the purposes of the SSA form, the definition of the value returned by the
3356 '<tt>invoke</tt>' instruction is deemed to occur on the edge from the current
3357 block to the "normal" label. If the callee unwinds then no return value is
3360 <p>Note that the code generator does not yet completely support unwind, and
3361 that the invoke/unwind semantics are likely to change in future versions.</p>
3365 %retval = invoke i32 @Test(i32 15) to label %Continue
3366 unwind label %TestCleanup <i>; {i32}:retval set</i>
3367 %retval = invoke <a href="#callingconv">coldcc</a> i32 %Testfnptr(i32 15) to label %Continue
3368 unwind label %TestCleanup <i>; {i32}:retval set</i>
3373 <!-- _______________________________________________________________________ -->
3376 <a name="i_unwind">'<tt>unwind</tt>' Instruction</a>
3387 <p>The '<tt>unwind</tt>' instruction unwinds the stack, continuing control flow
3388 at the first callee in the dynamic call stack which used
3389 an <a href="#i_invoke"><tt>invoke</tt></a> instruction to perform the call.
3390 This is primarily used to implement exception handling.</p>
3393 <p>The '<tt>unwind</tt>' instruction causes execution of the current function to
3394 immediately halt. The dynamic call stack is then searched for the
3395 first <a href="#i_invoke"><tt>invoke</tt></a> instruction on the call stack.
3396 Once found, execution continues at the "exceptional" destination block
3397 specified by the <tt>invoke</tt> instruction. If there is no <tt>invoke</tt>
3398 instruction in the dynamic call chain, undefined behavior results.</p>
3400 <p>Note that the code generator does not yet completely support unwind, and
3401 that the invoke/unwind semantics are likely to change in future versions.</p>
3405 <!-- _______________________________________________________________________ -->
3408 <a name="i_resume">'<tt>resume</tt>' Instruction</a>
3415 resume <type> <value>
3419 <p>The '<tt>resume</tt>' instruction is a terminator instruction that has no
3423 <p>The '<tt>resume</tt>' instruction requires one argument, which must have the
3424 same type as the result of any '<tt>landingpad</tt>' instruction in the same
3428 <p>The '<tt>resume</tt>' instruction resumes propagation of an existing
3429 (in-flight) exception whose unwinding was interrupted with
3430 a <a href="#i_landingpad"><tt>landingpad</tt></a> instruction.</p>
3434 resume { i8*, i32 } %exn
3439 <!-- _______________________________________________________________________ -->
3442 <a name="i_unreachable">'<tt>unreachable</tt>' Instruction</a>
3453 <p>The '<tt>unreachable</tt>' instruction has no defined semantics. This
3454 instruction is used to inform the optimizer that a particular portion of the
3455 code is not reachable. This can be used to indicate that the code after a
3456 no-return function cannot be reached, and other facts.</p>
3459 <p>The '<tt>unreachable</tt>' instruction has no defined semantics.</p>
3465 <!-- ======================================================================= -->
3467 <a name="binaryops">Binary Operations</a>
3472 <p>Binary operators are used to do most of the computation in a program. They
3473 require two operands of the same type, execute an operation on them, and
3474 produce a single value. The operands might represent multiple data, as is
3475 the case with the <a href="#t_vector">vector</a> data type. The result value
3476 has the same type as its operands.</p>
3478 <p>There are several different binary operators:</p>
3480 <!-- _______________________________________________________________________ -->
3482 <a name="i_add">'<tt>add</tt>' Instruction</a>
3489 <result> = add <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3490 <result> = add nuw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3491 <result> = add nsw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3492 <result> = add nuw nsw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3496 <p>The '<tt>add</tt>' instruction returns the sum of its two operands.</p>
3499 <p>The two arguments to the '<tt>add</tt>' instruction must
3500 be <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
3501 integer values. Both arguments must have identical types.</p>
3504 <p>The value produced is the integer sum of the two operands.</p>
3506 <p>If the sum has unsigned overflow, the result returned is the mathematical
3507 result modulo 2<sup>n</sup>, where n is the bit width of the result.</p>
3509 <p>Because LLVM integers use a two's complement representation, this instruction
3510 is appropriate for both signed and unsigned integers.</p>
3512 <p><tt>nuw</tt> and <tt>nsw</tt> stand for "No Unsigned Wrap"
3513 and "No Signed Wrap", respectively. If the <tt>nuw</tt> and/or
3514 <tt>nsw</tt> keywords are present, the result value of the <tt>add</tt>
3515 is a <a href="#trapvalues">trap value</a> if unsigned and/or signed overflow,
3516 respectively, occurs.</p>
3520 <result> = add i32 4, %var <i>; yields {i32}:result = 4 + %var</i>
3525 <!-- _______________________________________________________________________ -->
3527 <a name="i_fadd">'<tt>fadd</tt>' Instruction</a>
3534 <result> = fadd <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3538 <p>The '<tt>fadd</tt>' instruction returns the sum of its two operands.</p>
3541 <p>The two arguments to the '<tt>fadd</tt>' instruction must be
3542 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3543 floating point values. Both arguments must have identical types.</p>
3546 <p>The value produced is the floating point sum of the two operands.</p>
3550 <result> = fadd float 4.0, %var <i>; yields {float}:result = 4.0 + %var</i>
3555 <!-- _______________________________________________________________________ -->
3557 <a name="i_sub">'<tt>sub</tt>' Instruction</a>
3564 <result> = sub <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3565 <result> = sub nuw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3566 <result> = sub nsw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3567 <result> = sub nuw nsw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3571 <p>The '<tt>sub</tt>' instruction returns the difference of its two
3574 <p>Note that the '<tt>sub</tt>' instruction is used to represent the
3575 '<tt>neg</tt>' instruction present in most other intermediate
3576 representations.</p>
3579 <p>The two arguments to the '<tt>sub</tt>' instruction must
3580 be <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
3581 integer values. Both arguments must have identical types.</p>
3584 <p>The value produced is the integer difference of the two operands.</p>
3586 <p>If the difference has unsigned overflow, the result returned is the
3587 mathematical result modulo 2<sup>n</sup>, where n is the bit width of the
3590 <p>Because LLVM integers use a two's complement representation, this instruction
3591 is appropriate for both signed and unsigned integers.</p>
3593 <p><tt>nuw</tt> and <tt>nsw</tt> stand for "No Unsigned Wrap"
3594 and "No Signed Wrap", respectively. If the <tt>nuw</tt> and/or
3595 <tt>nsw</tt> keywords are present, the result value of the <tt>sub</tt>
3596 is a <a href="#trapvalues">trap value</a> if unsigned and/or signed overflow,
3597 respectively, occurs.</p>
3601 <result> = sub i32 4, %var <i>; yields {i32}:result = 4 - %var</i>
3602 <result> = sub i32 0, %val <i>; yields {i32}:result = -%var</i>
3607 <!-- _______________________________________________________________________ -->
3609 <a name="i_fsub">'<tt>fsub</tt>' Instruction</a>
3616 <result> = fsub <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3620 <p>The '<tt>fsub</tt>' instruction returns the difference of its two
3623 <p>Note that the '<tt>fsub</tt>' instruction is used to represent the
3624 '<tt>fneg</tt>' instruction present in most other intermediate
3625 representations.</p>
3628 <p>The two arguments to the '<tt>fsub</tt>' instruction must be
3629 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3630 floating point values. Both arguments must have identical types.</p>
3633 <p>The value produced is the floating point difference of the two operands.</p>
3637 <result> = fsub float 4.0, %var <i>; yields {float}:result = 4.0 - %var</i>
3638 <result> = fsub float -0.0, %val <i>; yields {float}:result = -%var</i>
3643 <!-- _______________________________________________________________________ -->
3645 <a name="i_mul">'<tt>mul</tt>' Instruction</a>
3652 <result> = mul <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3653 <result> = mul nuw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3654 <result> = mul nsw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3655 <result> = mul nuw nsw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3659 <p>The '<tt>mul</tt>' instruction returns the product of its two operands.</p>
3662 <p>The two arguments to the '<tt>mul</tt>' instruction must
3663 be <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
3664 integer values. Both arguments must have identical types.</p>
3667 <p>The value produced is the integer product of the two operands.</p>
3669 <p>If the result of the multiplication has unsigned overflow, the result
3670 returned is the mathematical result modulo 2<sup>n</sup>, where n is the bit
3671 width of the result.</p>
3673 <p>Because LLVM integers use a two's complement representation, and the result
3674 is the same width as the operands, this instruction returns the correct
3675 result for both signed and unsigned integers. If a full product
3676 (e.g. <tt>i32</tt>x<tt>i32</tt>-><tt>i64</tt>) is needed, the operands should
3677 be sign-extended or zero-extended as appropriate to the width of the full
3680 <p><tt>nuw</tt> and <tt>nsw</tt> stand for "No Unsigned Wrap"
3681 and "No Signed Wrap", respectively. If the <tt>nuw</tt> and/or
3682 <tt>nsw</tt> keywords are present, the result value of the <tt>mul</tt>
3683 is a <a href="#trapvalues">trap value</a> if unsigned and/or signed overflow,
3684 respectively, occurs.</p>
3688 <result> = mul i32 4, %var <i>; yields {i32}:result = 4 * %var</i>
3693 <!-- _______________________________________________________________________ -->
3695 <a name="i_fmul">'<tt>fmul</tt>' Instruction</a>
3702 <result> = fmul <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3706 <p>The '<tt>fmul</tt>' instruction returns the product of its two operands.</p>
3709 <p>The two arguments to the '<tt>fmul</tt>' instruction must be
3710 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3711 floating point values. Both arguments must have identical types.</p>
3714 <p>The value produced is the floating point product of the two operands.</p>
3718 <result> = fmul float 4.0, %var <i>; yields {float}:result = 4.0 * %var</i>
3723 <!-- _______________________________________________________________________ -->
3725 <a name="i_udiv">'<tt>udiv</tt>' Instruction</a>
3732 <result> = udiv <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3733 <result> = udiv exact <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3737 <p>The '<tt>udiv</tt>' instruction returns the quotient of its two operands.</p>
3740 <p>The two arguments to the '<tt>udiv</tt>' instruction must be
3741 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3742 values. Both arguments must have identical types.</p>
3745 <p>The value produced is the unsigned integer quotient of the two operands.</p>
3747 <p>Note that unsigned integer division and signed integer division are distinct
3748 operations; for signed integer division, use '<tt>sdiv</tt>'.</p>
3750 <p>Division by zero leads to undefined behavior.</p>
3752 <p>If the <tt>exact</tt> keyword is present, the result value of the
3753 <tt>udiv</tt> is a <a href="#trapvalues">trap value</a> if %op1 is not a
3754 multiple of %op2 (as such, "((a udiv exact b) mul b) == a").</p>
3759 <result> = udiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
3764 <!-- _______________________________________________________________________ -->
3766 <a name="i_sdiv">'<tt>sdiv</tt>' Instruction</a>
3773 <result> = sdiv <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3774 <result> = sdiv exact <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3778 <p>The '<tt>sdiv</tt>' instruction returns the quotient of its two operands.</p>
3781 <p>The two arguments to the '<tt>sdiv</tt>' instruction must be
3782 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3783 values. Both arguments must have identical types.</p>
3786 <p>The value produced is the signed integer quotient of the two operands rounded
3789 <p>Note that signed integer division and unsigned integer division are distinct
3790 operations; for unsigned integer division, use '<tt>udiv</tt>'.</p>
3792 <p>Division by zero leads to undefined behavior. Overflow also leads to
3793 undefined behavior; this is a rare case, but can occur, for example, by doing
3794 a 32-bit division of -2147483648 by -1.</p>
3796 <p>If the <tt>exact</tt> keyword is present, the result value of the
3797 <tt>sdiv</tt> is a <a href="#trapvalues">trap value</a> if the result would
3802 <result> = sdiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
3807 <!-- _______________________________________________________________________ -->
3809 <a name="i_fdiv">'<tt>fdiv</tt>' Instruction</a>
3816 <result> = fdiv <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3820 <p>The '<tt>fdiv</tt>' instruction returns the quotient of its two operands.</p>
3823 <p>The two arguments to the '<tt>fdiv</tt>' instruction must be
3824 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3825 floating point values. Both arguments must have identical types.</p>
3828 <p>The value produced is the floating point quotient of the two operands.</p>
3832 <result> = fdiv float 4.0, %var <i>; yields {float}:result = 4.0 / %var</i>
3837 <!-- _______________________________________________________________________ -->
3839 <a name="i_urem">'<tt>urem</tt>' Instruction</a>
3846 <result> = urem <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3850 <p>The '<tt>urem</tt>' instruction returns the remainder from the unsigned
3851 division of its two arguments.</p>
3854 <p>The two arguments to the '<tt>urem</tt>' instruction must be
3855 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3856 values. Both arguments must have identical types.</p>
3859 <p>This instruction returns the unsigned integer <i>remainder</i> of a division.
3860 This instruction always performs an unsigned division to get the
3863 <p>Note that unsigned integer remainder and signed integer remainder are
3864 distinct operations; for signed integer remainder, use '<tt>srem</tt>'.</p>
3866 <p>Taking the remainder of a division by zero leads to undefined behavior.</p>
3870 <result> = urem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
3875 <!-- _______________________________________________________________________ -->
3877 <a name="i_srem">'<tt>srem</tt>' Instruction</a>
3884 <result> = srem <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3888 <p>The '<tt>srem</tt>' instruction returns the remainder from the signed
3889 division of its two operands. This instruction can also take
3890 <a href="#t_vector">vector</a> versions of the values in which case the
3891 elements must be integers.</p>
3894 <p>The two arguments to the '<tt>srem</tt>' instruction must be
3895 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3896 values. Both arguments must have identical types.</p>
3899 <p>This instruction returns the <i>remainder</i> of a division (where the result
3900 is either zero or has the same sign as the dividend, <tt>op1</tt>), not the
3901 <i>modulo</i> operator (where the result is either zero or has the same sign
3902 as the divisor, <tt>op2</tt>) of a value.
3903 For more information about the difference,
3904 see <a href="http://mathforum.org/dr.math/problems/anne.4.28.99.html">The
3905 Math Forum</a>. For a table of how this is implemented in various languages,
3906 please see <a href="http://en.wikipedia.org/wiki/Modulo_operation">
3907 Wikipedia: modulo operation</a>.</p>
3909 <p>Note that signed integer remainder and unsigned integer remainder are
3910 distinct operations; for unsigned integer remainder, use '<tt>urem</tt>'.</p>
3912 <p>Taking the remainder of a division by zero leads to undefined behavior.
3913 Overflow also leads to undefined behavior; this is a rare case, but can
3914 occur, for example, by taking the remainder of a 32-bit division of
3915 -2147483648 by -1. (The remainder doesn't actually overflow, but this rule
3916 lets srem be implemented using instructions that return both the result of
3917 the division and the remainder.)</p>
3921 <result> = srem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
3926 <!-- _______________________________________________________________________ -->
3928 <a name="i_frem">'<tt>frem</tt>' Instruction</a>
3935 <result> = frem <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3939 <p>The '<tt>frem</tt>' instruction returns the remainder from the division of
3940 its two operands.</p>
3943 <p>The two arguments to the '<tt>frem</tt>' instruction must be
3944 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3945 floating point values. Both arguments must have identical types.</p>
3948 <p>This instruction returns the <i>remainder</i> of a division. The remainder
3949 has the same sign as the dividend.</p>
3953 <result> = frem float 4.0, %var <i>; yields {float}:result = 4.0 % %var</i>
3960 <!-- ======================================================================= -->
3962 <a name="bitwiseops">Bitwise Binary Operations</a>
3967 <p>Bitwise binary operators are used to do various forms of bit-twiddling in a
3968 program. They are generally very efficient instructions and can commonly be
3969 strength reduced from other instructions. They require two operands of the
3970 same type, execute an operation on them, and produce a single value. The
3971 resulting value is the same type as its operands.</p>
3973 <!-- _______________________________________________________________________ -->
3975 <a name="i_shl">'<tt>shl</tt>' Instruction</a>
3982 <result> = shl <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3983 <result> = shl nuw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3984 <result> = shl nsw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3985 <result> = shl nuw nsw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3989 <p>The '<tt>shl</tt>' instruction returns the first operand shifted to the left
3990 a specified number of bits.</p>
3993 <p>Both arguments to the '<tt>shl</tt>' instruction must be the
3994 same <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
3995 integer type. '<tt>op2</tt>' is treated as an unsigned value.</p>
3998 <p>The value produced is <tt>op1</tt> * 2<sup><tt>op2</tt></sup> mod
3999 2<sup>n</sup>, where <tt>n</tt> is the width of the result. If <tt>op2</tt>
4000 is (statically or dynamically) negative or equal to or larger than the number
4001 of bits in <tt>op1</tt>, the result is undefined. If the arguments are
4002 vectors, each vector element of <tt>op1</tt> is shifted by the corresponding
4003 shift amount in <tt>op2</tt>.</p>
4005 <p>If the <tt>nuw</tt> keyword is present, then the shift produces a
4006 <a href="#trapvalues">trap value</a> if it shifts out any non-zero bits. If
4007 the <tt>nsw</tt> keyword is present, then the shift produces a
4008 <a href="#trapvalues">trap value</a> if it shifts out any bits that disagree
4009 with the resultant sign bit. As such, NUW/NSW have the same semantics as
4010 they would if the shift were expressed as a mul instruction with the same
4011 nsw/nuw bits in (mul %op1, (shl 1, %op2)).</p>
4015 <result> = shl i32 4, %var <i>; yields {i32}: 4 << %var</i>
4016 <result> = shl i32 4, 2 <i>; yields {i32}: 16</i>
4017 <result> = shl i32 1, 10 <i>; yields {i32}: 1024</i>
4018 <result> = shl i32 1, 32 <i>; undefined</i>
4019 <result> = shl <2 x i32> < i32 1, i32 1>, < i32 1, i32 2> <i>; yields: result=<2 x i32> < i32 2, i32 4></i>
4024 <!-- _______________________________________________________________________ -->
4026 <a name="i_lshr">'<tt>lshr</tt>' Instruction</a>
4033 <result> = lshr <ty> <op1>, <op2> <i>; yields {ty}:result</i>
4034 <result> = lshr exact <ty> <op1>, <op2> <i>; yields {ty}:result</i>
4038 <p>The '<tt>lshr</tt>' instruction (logical shift right) returns the first
4039 operand shifted to the right a specified number of bits with zero fill.</p>
4042 <p>Both arguments to the '<tt>lshr</tt>' instruction must be the same
4043 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
4044 type. '<tt>op2</tt>' is treated as an unsigned value.</p>
4047 <p>This instruction always performs a logical shift right operation. The most
4048 significant bits of the result will be filled with zero bits after the shift.
4049 If <tt>op2</tt> is (statically or dynamically) equal to or larger than the
4050 number of bits in <tt>op1</tt>, the result is undefined. If the arguments are
4051 vectors, each vector element of <tt>op1</tt> is shifted by the corresponding
4052 shift amount in <tt>op2</tt>.</p>
4054 <p>If the <tt>exact</tt> keyword is present, the result value of the
4055 <tt>lshr</tt> is a <a href="#trapvalues">trap value</a> if any of the bits
4056 shifted out are non-zero.</p>
4061 <result> = lshr i32 4, 1 <i>; yields {i32}:result = 2</i>
4062 <result> = lshr i32 4, 2 <i>; yields {i32}:result = 1</i>
4063 <result> = lshr i8 4, 3 <i>; yields {i8}:result = 0</i>
4064 <result> = lshr i8 -2, 1 <i>; yields {i8}:result = 0x7FFFFFFF </i>
4065 <result> = lshr i32 1, 32 <i>; undefined</i>
4066 <result> = lshr <2 x i32> < i32 -2, i32 4>, < i32 1, i32 2> <i>; yields: result=<2 x i32> < i32 0x7FFFFFFF, i32 1></i>
4071 <!-- _______________________________________________________________________ -->
4073 <a name="i_ashr">'<tt>ashr</tt>' Instruction</a>
4080 <result> = ashr <ty> <op1>, <op2> <i>; yields {ty}:result</i>
4081 <result> = ashr exact <ty> <op1>, <op2> <i>; yields {ty}:result</i>
4085 <p>The '<tt>ashr</tt>' instruction (arithmetic shift right) returns the first
4086 operand shifted to the right a specified number of bits with sign
4090 <p>Both arguments to the '<tt>ashr</tt>' instruction must be the same
4091 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
4092 type. '<tt>op2</tt>' is treated as an unsigned value.</p>
4095 <p>This instruction always performs an arithmetic shift right operation, The
4096 most significant bits of the result will be filled with the sign bit
4097 of <tt>op1</tt>. If <tt>op2</tt> is (statically or dynamically) equal to or
4098 larger than the number of bits in <tt>op1</tt>, the result is undefined. If
4099 the arguments are vectors, each vector element of <tt>op1</tt> is shifted by
4100 the corresponding shift amount in <tt>op2</tt>.</p>
4102 <p>If the <tt>exact</tt> keyword is present, the result value of the
4103 <tt>ashr</tt> is a <a href="#trapvalues">trap value</a> if any of the bits
4104 shifted out are non-zero.</p>
4108 <result> = ashr i32 4, 1 <i>; yields {i32}:result = 2</i>
4109 <result> = ashr i32 4, 2 <i>; yields {i32}:result = 1</i>
4110 <result> = ashr i8 4, 3 <i>; yields {i8}:result = 0</i>
4111 <result> = ashr i8 -2, 1 <i>; yields {i8}:result = -1</i>
4112 <result> = ashr i32 1, 32 <i>; undefined</i>
4113 <result> = ashr <2 x i32> < i32 -2, i32 4>, < i32 1, i32 3> <i>; yields: result=<2 x i32> < i32 -1, i32 0></i>
4118 <!-- _______________________________________________________________________ -->
4120 <a name="i_and">'<tt>and</tt>' Instruction</a>
4127 <result> = and <ty> <op1>, <op2> <i>; yields {ty}:result</i>
4131 <p>The '<tt>and</tt>' instruction returns the bitwise logical and of its two
4135 <p>The two arguments to the '<tt>and</tt>' instruction must be
4136 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
4137 values. Both arguments must have identical types.</p>
4140 <p>The truth table used for the '<tt>and</tt>' instruction is:</p>
4142 <table border="1" cellspacing="0" cellpadding="4">
4174 <result> = and i32 4, %var <i>; yields {i32}:result = 4 & %var</i>
4175 <result> = and i32 15, 40 <i>; yields {i32}:result = 8</i>
4176 <result> = and i32 4, 8 <i>; yields {i32}:result = 0</i>
4179 <!-- _______________________________________________________________________ -->
4181 <a name="i_or">'<tt>or</tt>' Instruction</a>
4188 <result> = or <ty> <op1>, <op2> <i>; yields {ty}:result</i>
4192 <p>The '<tt>or</tt>' instruction returns the bitwise logical inclusive or of its
4196 <p>The two arguments to the '<tt>or</tt>' instruction must be
4197 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
4198 values. Both arguments must have identical types.</p>
4201 <p>The truth table used for the '<tt>or</tt>' instruction is:</p>
4203 <table border="1" cellspacing="0" cellpadding="4">
4235 <result> = or i32 4, %var <i>; yields {i32}:result = 4 | %var</i>
4236 <result> = or i32 15, 40 <i>; yields {i32}:result = 47</i>
4237 <result> = or i32 4, 8 <i>; yields {i32}:result = 12</i>
4242 <!-- _______________________________________________________________________ -->
4244 <a name="i_xor">'<tt>xor</tt>' Instruction</a>
4251 <result> = xor <ty> <op1>, <op2> <i>; yields {ty}:result</i>
4255 <p>The '<tt>xor</tt>' instruction returns the bitwise logical exclusive or of
4256 its two operands. The <tt>xor</tt> is used to implement the "one's
4257 complement" operation, which is the "~" operator in C.</p>
4260 <p>The two arguments to the '<tt>xor</tt>' instruction must be
4261 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
4262 values. Both arguments must have identical types.</p>
4265 <p>The truth table used for the '<tt>xor</tt>' instruction is:</p>
4267 <table border="1" cellspacing="0" cellpadding="4">
4299 <result> = xor i32 4, %var <i>; yields {i32}:result = 4 ^ %var</i>
4300 <result> = xor i32 15, 40 <i>; yields {i32}:result = 39</i>
4301 <result> = xor i32 4, 8 <i>; yields {i32}:result = 12</i>
4302 <result> = xor i32 %V, -1 <i>; yields {i32}:result = ~%V</i>
4309 <!-- ======================================================================= -->
4311 <a name="vectorops">Vector Operations</a>
4316 <p>LLVM supports several instructions to represent vector operations in a
4317 target-independent manner. These instructions cover the element-access and
4318 vector-specific operations needed to process vectors effectively. While LLVM
4319 does directly support these vector operations, many sophisticated algorithms
4320 will want to use target-specific intrinsics to take full advantage of a
4321 specific target.</p>
4323 <!-- _______________________________________________________________________ -->
4325 <a name="i_extractelement">'<tt>extractelement</tt>' Instruction</a>
4332 <result> = extractelement <n x <ty>> <val>, i32 <idx> <i>; yields <ty></i>
4336 <p>The '<tt>extractelement</tt>' instruction extracts a single scalar element
4337 from a vector at a specified index.</p>
4341 <p>The first operand of an '<tt>extractelement</tt>' instruction is a value
4342 of <a href="#t_vector">vector</a> type. The second operand is an index
4343 indicating the position from which to extract the element. The index may be
4347 <p>The result is a scalar of the same type as the element type of
4348 <tt>val</tt>. Its value is the value at position <tt>idx</tt> of
4349 <tt>val</tt>. If <tt>idx</tt> exceeds the length of <tt>val</tt>, the
4350 results are undefined.</p>
4354 <result> = extractelement <4 x i32> %vec, i32 0 <i>; yields i32</i>
4359 <!-- _______________________________________________________________________ -->
4361 <a name="i_insertelement">'<tt>insertelement</tt>' Instruction</a>
4368 <result> = insertelement <n x <ty>> <val>, <ty> <elt>, i32 <idx> <i>; yields <n x <ty>></i>
4372 <p>The '<tt>insertelement</tt>' instruction inserts a scalar element into a
4373 vector at a specified index.</p>
4376 <p>The first operand of an '<tt>insertelement</tt>' instruction is a value
4377 of <a href="#t_vector">vector</a> type. The second operand is a scalar value
4378 whose type must equal the element type of the first operand. The third
4379 operand is an index indicating the position at which to insert the value.
4380 The index may be a variable.</p>
4383 <p>The result is a vector of the same type as <tt>val</tt>. Its element values
4384 are those of <tt>val</tt> except at position <tt>idx</tt>, where it gets the
4385 value <tt>elt</tt>. If <tt>idx</tt> exceeds the length of <tt>val</tt>, the
4386 results are undefined.</p>
4390 <result> = insertelement <4 x i32> %vec, i32 1, i32 0 <i>; yields <4 x i32></i>
4395 <!-- _______________________________________________________________________ -->
4397 <a name="i_shufflevector">'<tt>shufflevector</tt>' Instruction</a>
4404 <result> = shufflevector <n x <ty>> <v1>, <n x <ty>> <v2>, <m x i32> <mask> <i>; yields <m x <ty>></i>
4408 <p>The '<tt>shufflevector</tt>' instruction constructs a permutation of elements
4409 from two input vectors, returning a vector with the same element type as the
4410 input and length that is the same as the shuffle mask.</p>
4413 <p>The first two operands of a '<tt>shufflevector</tt>' instruction are vectors
4414 with types that match each other. The third argument is a shuffle mask whose
4415 element type is always 'i32'. The result of the instruction is a vector
4416 whose length is the same as the shuffle mask and whose element type is the
4417 same as the element type of the first two operands.</p>
4419 <p>The shuffle mask operand is required to be a constant vector with either
4420 constant integer or undef values.</p>
4423 <p>The elements of the two input vectors are numbered from left to right across
4424 both of the vectors. The shuffle mask operand specifies, for each element of
4425 the result vector, which element of the two input vectors the result element
4426 gets. The element selector may be undef (meaning "don't care") and the
4427 second operand may be undef if performing a shuffle from only one vector.</p>
4431 <result> = shufflevector <4 x i32> %v1, <4 x i32> %v2,
4432 <4 x i32> <i32 0, i32 4, i32 1, i32 5> <i>; yields <4 x i32></i>
4433 <result> = shufflevector <4 x i32> %v1, <4 x i32> undef,
4434 <4 x i32> <i32 0, i32 1, i32 2, i32 3> <i>; yields <4 x i32></i> - Identity shuffle.
4435 <result> = shufflevector <8 x i32> %v1, <8 x i32> undef,
4436 <4 x i32> <i32 0, i32 1, i32 2, i32 3> <i>; yields <4 x i32></i>
4437 <result> = shufflevector <4 x i32> %v1, <4 x i32> %v2,
4438 <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>
4445 <!-- ======================================================================= -->
4447 <a name="aggregateops">Aggregate Operations</a>
4452 <p>LLVM supports several instructions for working with
4453 <a href="#t_aggregate">aggregate</a> values.</p>
4455 <!-- _______________________________________________________________________ -->
4457 <a name="i_extractvalue">'<tt>extractvalue</tt>' Instruction</a>
4464 <result> = extractvalue <aggregate type> <val>, <idx>{, <idx>}*
4468 <p>The '<tt>extractvalue</tt>' instruction extracts the value of a member field
4469 from an <a href="#t_aggregate">aggregate</a> value.</p>
4472 <p>The first operand of an '<tt>extractvalue</tt>' instruction is a value
4473 of <a href="#t_struct">struct</a> or
4474 <a href="#t_array">array</a> type. The operands are constant indices to
4475 specify which value to extract in a similar manner as indices in a
4476 '<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.</p>
4477 <p>The major differences to <tt>getelementptr</tt> indexing are:</p>
4479 <li>Since the value being indexed is not a pointer, the first index is
4480 omitted and assumed to be zero.</li>
4481 <li>At least one index must be specified.</li>
4482 <li>Not only struct indices but also array indices must be in
4487 <p>The result is the value at the position in the aggregate specified by the
4492 <result> = extractvalue {i32, float} %agg, 0 <i>; yields i32</i>
4497 <!-- _______________________________________________________________________ -->
4499 <a name="i_insertvalue">'<tt>insertvalue</tt>' Instruction</a>
4506 <result> = insertvalue <aggregate type> <val>, <ty> <elt>, <idx>{, <idx>}* <i>; yields <aggregate type></i>
4510 <p>The '<tt>insertvalue</tt>' instruction inserts a value into a member field
4511 in an <a href="#t_aggregate">aggregate</a> value.</p>
4514 <p>The first operand of an '<tt>insertvalue</tt>' instruction is a value
4515 of <a href="#t_struct">struct</a> or
4516 <a href="#t_array">array</a> type. The second operand is a first-class
4517 value to insert. The following operands are constant indices indicating
4518 the position at which to insert the value in a similar manner as indices in a
4519 '<tt><a href="#i_extractvalue">extractvalue</a></tt>' instruction. The
4520 value to insert must have the same type as the value identified by the
4524 <p>The result is an aggregate of the same type as <tt>val</tt>. Its value is
4525 that of <tt>val</tt> except that the value at the position specified by the
4526 indices is that of <tt>elt</tt>.</p>
4530 %agg1 = insertvalue {i32, float} undef, i32 1, 0 <i>; yields {i32 1, float undef}</i>
4531 %agg2 = insertvalue {i32, float} %agg1, float %val, 1 <i>; yields {i32 1, float %val}</i>
4532 %agg3 = insertvalue {i32, {float}} %agg1, float %val, 1, 0 <i>; yields {i32 1, float %val}</i>
4539 <!-- ======================================================================= -->
4541 <a name="memoryops">Memory Access and Addressing Operations</a>
4546 <p>A key design point of an SSA-based representation is how it represents
4547 memory. In LLVM, no memory locations are in SSA form, which makes things
4548 very simple. This section describes how to read, write, and allocate
4551 <!-- _______________________________________________________________________ -->
4553 <a name="i_alloca">'<tt>alloca</tt>' Instruction</a>
4560 <result> = alloca <type>[, <ty> <NumElements>][, align <alignment>] <i>; yields {type*}:result</i>
4564 <p>The '<tt>alloca</tt>' instruction allocates memory on the stack frame of the
4565 currently executing function, to be automatically released when this function
4566 returns to its caller. The object is always allocated in the generic address
4567 space (address space zero).</p>
4570 <p>The '<tt>alloca</tt>' instruction
4571 allocates <tt>sizeof(<type>)*NumElements</tt> bytes of memory on the
4572 runtime stack, returning a pointer of the appropriate type to the program.
4573 If "NumElements" is specified, it is the number of elements allocated,
4574 otherwise "NumElements" is defaulted to be one. If a constant alignment is
4575 specified, the value result of the allocation is guaranteed to be aligned to
4576 at least that boundary. If not specified, or if zero, the target can choose
4577 to align the allocation on any convenient boundary compatible with the
4580 <p>'<tt>type</tt>' may be any sized type.</p>
4583 <p>Memory is allocated; a pointer is returned. The operation is undefined if
4584 there is insufficient stack space for the allocation. '<tt>alloca</tt>'d
4585 memory is automatically released when the function returns. The
4586 '<tt>alloca</tt>' instruction is commonly used to represent automatic
4587 variables that must have an address available. When the function returns
4588 (either with the <tt><a href="#i_ret">ret</a></tt>
4589 or <tt><a href="#i_unwind">unwind</a></tt> instructions), the memory is
4590 reclaimed. Allocating zero bytes is legal, but the result is undefined.</p>
4594 %ptr = alloca i32 <i>; yields {i32*}:ptr</i>
4595 %ptr = alloca i32, i32 4 <i>; yields {i32*}:ptr</i>
4596 %ptr = alloca i32, i32 4, align 1024 <i>; yields {i32*}:ptr</i>
4597 %ptr = alloca i32, align 1024 <i>; yields {i32*}:ptr</i>
4602 <!-- _______________________________________________________________________ -->
4604 <a name="i_load">'<tt>load</tt>' Instruction</a>
4611 <result> = load [volatile] <ty>* <pointer>[, align <alignment>][, !nontemporal !<index>]
4612 <result> = load atomic [volatile] <ty>* <pointer> [singlethread] <ordering>, align <alignment>
4613 !<index> = !{ i32 1 }
4617 <p>The '<tt>load</tt>' instruction is used to read from memory.</p>
4620 <p>The argument to the '<tt>load</tt>' instruction specifies the memory address
4621 from which to load. The pointer must point to
4622 a <a href="#t_firstclass">first class</a> type. If the <tt>load</tt> is
4623 marked as <tt>volatile</tt>, then the optimizer is not allowed to modify the
4624 number or order of execution of this <tt>load</tt> with other <a
4625 href="#volatile">volatile operations</a>.</p>
4627 <p>If the <code>load</code> is marked as <code>atomic</code>, it takes an extra
4628 <a href="#ordering">ordering</a> and optional <code>singlethread</code>
4629 argument. The <code>release</code> and <code>acq_rel</code> orderings are
4630 not valid on <code>load</code> instructions. Atomic loads produce <a
4631 href="#memorymodel">defined</a> results when they may see multiple atomic
4632 stores. The type of the pointee must be an integer type whose bit width
4633 is a power of two greater than or equal to eight and less than or equal
4634 to a target-specific size limit. <code>align</code> must be explicitly
4635 specified on atomic loads, and the load has undefined behavior if the
4636 alignment is not set to a value which is at least the size in bytes of
4637 the pointee. <code>!nontemporal</code> does not have any defined semantics
4638 for atomic loads.</p>
4640 <p>The optional constant <tt>align</tt> argument specifies the alignment of the
4641 operation (that is, the alignment of the memory address). A value of 0 or an
4642 omitted <tt>align</tt> argument means that the operation has the preferential
4643 alignment for the target. It is the responsibility of the code emitter to
4644 ensure that the alignment information is correct. Overestimating the
4645 alignment results in undefined behavior. Underestimating the alignment may
4646 produce less efficient code. An alignment of 1 is always safe.</p>
4648 <p>The optional <tt>!nontemporal</tt> metadata must reference a single
4649 metatadata name <index> corresponding to a metadata node with
4650 one <tt>i32</tt> entry of value 1. The existence of
4651 the <tt>!nontemporal</tt> metatadata on the instruction tells the optimizer
4652 and code generator that this load is not expected to be reused in the cache.
4653 The code generator may select special instructions to save cache bandwidth,
4654 such as the <tt>MOVNT</tt> instruction on x86.</p>
4657 <p>The location of memory pointed to is loaded. If the value being loaded is of
4658 scalar type then the number of bytes read does not exceed the minimum number
4659 of bytes needed to hold all bits of the type. For example, loading an
4660 <tt>i24</tt> reads at most three bytes. When loading a value of a type like
4661 <tt>i20</tt> with a size that is not an integral number of bytes, the result
4662 is undefined if the value was not originally written using a store of the
4667 %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
4668 <a href="#i_store">store</a> i32 3, i32* %ptr <i>; yields {void}</i>
4669 %val = load i32* %ptr <i>; yields {i32}:val = i32 3</i>
4674 <!-- _______________________________________________________________________ -->
4676 <a name="i_store">'<tt>store</tt>' Instruction</a>
4683 store [volatile] <ty> <value>, <ty>* <pointer>[, align <alignment>][, !nontemporal !<index>] <i>; yields {void}</i>
4684 store atomic [volatile] <ty> <value>, <ty>* <pointer> [singlethread] <ordering>, align <alignment> <i>; yields {void}</i>
4688 <p>The '<tt>store</tt>' instruction is used to write to memory.</p>
4691 <p>There are two arguments to the '<tt>store</tt>' instruction: a value to store
4692 and an address at which to store it. The type of the
4693 '<tt><pointer></tt>' operand must be a pointer to
4694 the <a href="#t_firstclass">first class</a> type of the
4695 '<tt><value></tt>' operand. If the <tt>store</tt> is marked as
4696 <tt>volatile</tt>, then the optimizer is not allowed to modify the number or
4697 order of execution of this <tt>store</tt> with other <a
4698 href="#volatile">volatile operations</a>.</p>
4700 <p>If the <code>store</code> is marked as <code>atomic</code>, it takes an extra
4701 <a href="#ordering">ordering</a> and optional <code>singlethread</code>
4702 argument. The <code>acquire</code> and <code>acq_rel</code> orderings aren't
4703 valid on <code>store</code> instructions. Atomic loads produce <a
4704 href="#memorymodel">defined</a> results when they may see multiple atomic
4705 stores. The type of the pointee must be an integer type whose bit width
4706 is a power of two greater than or equal to eight and less than or equal
4707 to a target-specific size limit. <code>align</code> must be explicitly
4708 specified on atomic stores, and the store has undefined behavior if the
4709 alignment is not set to a value which is at least the size in bytes of
4710 the pointee. <code>!nontemporal</code> does not have any defined semantics
4711 for atomic stores.</p>
4713 <p>The optional constant "align" argument specifies the alignment of the
4714 operation (that is, the alignment of the memory address). A value of 0 or an
4715 omitted "align" argument means that the operation has the preferential
4716 alignment for the target. It is the responsibility of the code emitter to
4717 ensure that the alignment information is correct. Overestimating the
4718 alignment results in an undefined behavior. Underestimating the alignment may
4719 produce less efficient code. An alignment of 1 is always safe.</p>
4721 <p>The optional !nontemporal metadata must reference a single metatadata
4722 name <index> corresponding to a metadata node with one i32 entry of
4723 value 1. The existence of the !nontemporal metatadata on the
4724 instruction tells the optimizer and code generator that this load is
4725 not expected to be reused in the cache. The code generator may
4726 select special instructions to save cache bandwidth, such as the
4727 MOVNT instruction on x86.</p>
4731 <p>The contents of memory are updated to contain '<tt><value></tt>' at the
4732 location specified by the '<tt><pointer></tt>' operand. If
4733 '<tt><value></tt>' is of scalar type then the number of bytes written
4734 does not exceed the minimum number of bytes needed to hold all bits of the
4735 type. For example, storing an <tt>i24</tt> writes at most three bytes. When
4736 writing a value of a type like <tt>i20</tt> with a size that is not an
4737 integral number of bytes, it is unspecified what happens to the extra bits
4738 that do not belong to the type, but they will typically be overwritten.</p>
4742 %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
4743 store i32 3, i32* %ptr <i>; yields {void}</i>
4744 %val = <a href="#i_load">load</a> i32* %ptr <i>; yields {i32}:val = i32 3</i>
4749 <!-- _______________________________________________________________________ -->
4751 <a name="i_fence">'<tt>fence</tt>' Instruction</a>
4758 fence [singlethread] <ordering> <i>; yields {void}</i>
4762 <p>The '<tt>fence</tt>' instruction is used to introduce happens-before edges
4763 between operations.</p>
4765 <h5>Arguments:</h5> <p>'<code>fence</code>' instructions take an <a
4766 href="#ordering">ordering</a> argument which defines what
4767 <i>synchronizes-with</i> edges they add. They can only be given
4768 <code>acquire</code>, <code>release</code>, <code>acq_rel</code>, and
4769 <code>seq_cst</code> orderings.</p>
4772 <p>A fence <var>A</var> which has (at least) <code>release</code> ordering
4773 semantics <i>synchronizes with</i> a fence <var>B</var> with (at least)
4774 <code>acquire</code> ordering semantics if and only if there exist atomic
4775 operations <var>X</var> and <var>Y</var>, both operating on some atomic object
4776 <var>M</var>, such that <var>A</var> is sequenced before <var>X</var>,
4777 <var>X</var> modifies <var>M</var> (either directly or through some side effect
4778 of a sequence headed by <var>X</var>), <var>Y</var> is sequenced before
4779 <var>B</var>, and <var>Y</var> observes <var>M</var>. This provides a
4780 <i>happens-before</i> dependency between <var>A</var> and <var>B</var>. Rather
4781 than an explicit <code>fence</code>, one (but not both) of the atomic operations
4782 <var>X</var> or <var>Y</var> might provide a <code>release</code> or
4783 <code>acquire</code> (resp.) ordering constraint and still
4784 <i>synchronize-with</i> the explicit <code>fence</code> and establish the
4785 <i>happens-before</i> edge.</p>
4787 <p>A <code>fence</code> which has <code>seq_cst</code> ordering, in addition to
4788 having both <code>acquire</code> and <code>release</code> semantics specified
4789 above, participates in the global program order of other <code>seq_cst</code>
4790 operations and/or fences.</p>
4792 <p>The optional "<a href="#singlethread"><code>singlethread</code></a>" argument
4793 specifies that the fence only synchronizes with other fences in the same
4794 thread. (This is useful for interacting with signal handlers.)</p>
4798 fence acquire <i>; yields {void}</i>
4799 fence singlethread seq_cst <i>; yields {void}</i>
4804 <!-- _______________________________________________________________________ -->
4806 <a name="i_cmpxchg">'<tt>cmpxchg</tt>' Instruction</a>
4813 cmpxchg [volatile] <ty>* <pointer>, <ty> <cmp>, <ty> <new> [singlethread] <ordering> <i>; yields {ty}</i>
4817 <p>The '<tt>cmpxchg</tt>' instruction is used to atomically modify memory.
4818 It loads a value in memory and compares it to a given value. If they are
4819 equal, it stores a new value into the memory.</p>
4822 <p>There are three arguments to the '<code>cmpxchg</code>' instruction: an
4823 address to operate on, a value to compare to the value currently be at that
4824 address, and a new value to place at that address if the compared values are
4825 equal. The type of '<var><cmp></var>' must be an integer type whose
4826 bit width is a power of two greater than or equal to eight and less than
4827 or equal to a target-specific size limit. '<var><cmp></var>' and
4828 '<var><new></var>' must have the same type, and the type of
4829 '<var><pointer></var>' must be a pointer to that type. If the
4830 <code>cmpxchg</code> is marked as <code>volatile</code>, then the
4831 optimizer is not allowed to modify the number or order of execution
4832 of this <code>cmpxchg</code> with other <a href="#volatile">volatile
4835 <!-- FIXME: Extend allowed types. -->
4837 <p>The <a href="#ordering"><var>ordering</var></a> argument specifies how this
4838 <code>cmpxchg</code> synchronizes with other atomic operations.</p>
4840 <p>The optional "<code>singlethread</code>" argument declares that the
4841 <code>cmpxchg</code> is only atomic with respect to code (usually signal
4842 handlers) running in the same thread as the <code>cmpxchg</code>. Otherwise the
4843 cmpxchg is atomic with respect to all other code in the system.</p>
4845 <p>The pointer passed into cmpxchg must have alignment greater than or equal to
4846 the size in memory of the operand.
4849 <p>The contents of memory at the location specified by the
4850 '<tt><pointer></tt>' operand is read and compared to
4851 '<tt><cmp></tt>'; if the read value is the equal,
4852 '<tt><new></tt>' is written. The original value at the location
4855 <p>A successful <code>cmpxchg</code> is a read-modify-write instruction for the
4856 purpose of identifying <a href="#release_sequence">release sequences</a>. A
4857 failed <code>cmpxchg</code> is equivalent to an atomic load with an ordering
4858 parameter determined by dropping any <code>release</code> part of the
4859 <code>cmpxchg</code>'s ordering.</p>
4862 FIXME: Is compare_exchange_weak() necessary? (Consider after we've done
4863 optimization work on ARM.)
4865 FIXME: Is a weaker ordering constraint on failure helpful in practice?
4871 %orig = atomic <a href="#i_load">load</a> i32* %ptr unordered <i>; yields {i32}</i>
4872 <a href="#i_br">br</a> label %loop
4875 %cmp = <a href="#i_phi">phi</a> i32 [ %orig, %entry ], [%old, %loop]
4876 %squared = <a href="#i_mul">mul</a> i32 %cmp, %cmp
4877 %old = cmpxchg i32* %ptr, i32 %cmp, i32 %squared <i>; yields {i32}</i>
4878 %success = <a href="#i_icmp">icmp</a> eq i32 %cmp, %old
4879 <a href="#i_br">br</a> i1 %success, label %done, label %loop
4887 <!-- _______________________________________________________________________ -->
4889 <a name="i_atomicrmw">'<tt>atomicrmw</tt>' Instruction</a>
4896 atomicrmw [volatile] <operation> <ty>* <pointer>, <ty> <value> [singlethread] <ordering> <i>; yields {ty}</i>
4900 <p>The '<tt>atomicrmw</tt>' instruction is used to atomically modify memory.</p>
4903 <p>There are three arguments to the '<code>atomicrmw</code>' instruction: an
4904 operation to apply, an address whose value to modify, an argument to the
4905 operation. The operation must be one of the following keywords:</p>
4920 <p>The type of '<var><value></var>' must be an integer type whose
4921 bit width is a power of two greater than or equal to eight and less than
4922 or equal to a target-specific size limit. The type of the
4923 '<code><pointer></code>' operand must be a pointer to that type.
4924 If the <code>atomicrmw</code> is marked as <code>volatile</code>, then the
4925 optimizer is not allowed to modify the number or order of execution of this
4926 <code>atomicrmw</code> with other <a href="#volatile">volatile
4929 <!-- FIXME: Extend allowed types. -->
4932 <p>The contents of memory at the location specified by the
4933 '<tt><pointer></tt>' operand are atomically read, modified, and written
4934 back. The original value at the location is returned. The modification is
4935 specified by the <var>operation</var> argument:</p>
4938 <li>xchg: <code>*ptr = val</code></li>
4939 <li>add: <code>*ptr = *ptr + val</code></li>
4940 <li>sub: <code>*ptr = *ptr - val</code></li>
4941 <li>and: <code>*ptr = *ptr & val</code></li>
4942 <li>nand: <code>*ptr = ~(*ptr & val)</code></li>
4943 <li>or: <code>*ptr = *ptr | val</code></li>
4944 <li>xor: <code>*ptr = *ptr ^ val</code></li>
4945 <li>max: <code>*ptr = *ptr > val ? *ptr : val</code> (using a signed comparison)</li>
4946 <li>min: <code>*ptr = *ptr < val ? *ptr : val</code> (using a signed comparison)</li>
4947 <li>umax: <code>*ptr = *ptr > val ? *ptr : val</code> (using an unsigned comparison)</li>
4948 <li>umin: <code>*ptr = *ptr < val ? *ptr : val</code> (using an unsigned comparison)</li>
4953 %old = atomicrmw add i32* %ptr, i32 1 acquire <i>; yields {i32}</i>
4958 <!-- _______________________________________________________________________ -->
4960 <a name="i_getelementptr">'<tt>getelementptr</tt>' Instruction</a>
4967 <result> = getelementptr <pty>* <ptrval>{, <ty> <idx>}*
4968 <result> = getelementptr inbounds <pty>* <ptrval>{, <ty> <idx>}*
4972 <p>The '<tt>getelementptr</tt>' instruction is used to get the address of a
4973 subelement of an <a href="#t_aggregate">aggregate</a> data structure.
4974 It performs address calculation only and does not access memory.</p>
4977 <p>The first argument is always a pointer, and forms the basis of the
4978 calculation. The remaining arguments are indices that indicate which of the
4979 elements of the aggregate object are indexed. The interpretation of each
4980 index is dependent on the type being indexed into. The first index always
4981 indexes the pointer value given as the first argument, the second index
4982 indexes a value of the type pointed to (not necessarily the value directly
4983 pointed to, since the first index can be non-zero), etc. The first type
4984 indexed into must be a pointer value, subsequent types can be arrays,
4985 vectors, and structs. Note that subsequent types being indexed into
4986 can never be pointers, since that would require loading the pointer before
4987 continuing calculation.</p>
4989 <p>The type of each index argument depends on the type it is indexing into.
4990 When indexing into a (optionally packed) structure, only <tt>i32</tt>
4991 integer <b>constants</b> are allowed. When indexing into an array, pointer
4992 or vector, integers of any width are allowed, and they are not required to be
4993 constant. These integers are treated as signed values where relevant.</p>
4995 <p>For example, let's consider a C code fragment and how it gets compiled to
4998 <pre class="doc_code">
5010 int *foo(struct ST *s) {
5011 return &s[1].Z.B[5][13];
5015 <p>The LLVM code generated by the GCC frontend is:</p>
5017 <pre class="doc_code">
5018 %RT = <a href="#namedtypes">type</a> { i8 , [10 x [20 x i32]], i8 }
5019 %ST = <a href="#namedtypes">type</a> { i32, double, %RT }
5021 define i32* @foo(%ST* %s) {
5023 %reg = getelementptr %ST* %s, i32 1, i32 2, i32 1, i32 5, i32 13
5029 <p>In the example above, the first index is indexing into the '<tt>%ST*</tt>'
5030 type, which is a pointer, yielding a '<tt>%ST</tt>' = '<tt>{ i32, double, %RT
5031 }</tt>' type, a structure. The second index indexes into the third element
5032 of the structure, yielding a '<tt>%RT</tt>' = '<tt>{ i8 , [10 x [20 x i32]],
5033 i8 }</tt>' type, another structure. The third index indexes into the second
5034 element of the structure, yielding a '<tt>[10 x [20 x i32]]</tt>' type, an
5035 array. The two dimensions of the array are subscripted into, yielding an
5036 '<tt>i32</tt>' type. The '<tt>getelementptr</tt>' instruction returns a
5037 pointer to this element, thus computing a value of '<tt>i32*</tt>' type.</p>
5039 <p>Note that it is perfectly legal to index partially through a structure,
5040 returning a pointer to an inner element. Because of this, the LLVM code for
5041 the given testcase is equivalent to:</p>
5044 define i32* @foo(%ST* %s) {
5045 %t1 = getelementptr %ST* %s, i32 1 <i>; yields %ST*:%t1</i>
5046 %t2 = getelementptr %ST* %t1, i32 0, i32 2 <i>; yields %RT*:%t2</i>
5047 %t3 = getelementptr %RT* %t2, i32 0, i32 1 <i>; yields [10 x [20 x i32]]*:%t3</i>
5048 %t4 = getelementptr [10 x [20 x i32]]* %t3, i32 0, i32 5 <i>; yields [20 x i32]*:%t4</i>
5049 %t5 = getelementptr [20 x i32]* %t4, i32 0, i32 13 <i>; yields i32*:%t5</i>
5054 <p>If the <tt>inbounds</tt> keyword is present, the result value of the
5055 <tt>getelementptr</tt> is a <a href="#trapvalues">trap value</a> if the
5056 base pointer is not an <i>in bounds</i> address of an allocated object,
5057 or if any of the addresses that would be formed by successive addition of
5058 the offsets implied by the indices to the base address with infinitely
5059 precise signed arithmetic are not an <i>in bounds</i> address of that
5060 allocated object. The <i>in bounds</i> addresses for an allocated object
5061 are all the addresses that point into the object, plus the address one
5062 byte past the end.</p>
5064 <p>If the <tt>inbounds</tt> keyword is not present, the offsets are added to
5065 the base address with silently-wrapping two's complement arithmetic. If the
5066 offsets have a different width from the pointer, they are sign-extended or
5067 truncated to the width of the pointer. The result value of the
5068 <tt>getelementptr</tt> may be outside the object pointed to by the base
5069 pointer. The result value may not necessarily be used to access memory
5070 though, even if it happens to point into allocated storage. See the
5071 <a href="#pointeraliasing">Pointer Aliasing Rules</a> section for more
5074 <p>The getelementptr instruction is often confusing. For some more insight into
5075 how it works, see <a href="GetElementPtr.html">the getelementptr FAQ</a>.</p>
5079 <i>; yields [12 x i8]*:aptr</i>
5080 %aptr = getelementptr {i32, [12 x i8]}* %saptr, i64 0, i32 1
5081 <i>; yields i8*:vptr</i>
5082 %vptr = getelementptr {i32, <2 x i8>}* %svptr, i64 0, i32 1, i32 1
5083 <i>; yields i8*:eptr</i>
5084 %eptr = getelementptr [12 x i8]* %aptr, i64 0, i32 1
5085 <i>; yields i32*:iptr</i>
5086 %iptr = getelementptr [10 x i32]* @arr, i16 0, i16 0
5093 <!-- ======================================================================= -->
5095 <a name="convertops">Conversion Operations</a>
5100 <p>The instructions in this category are the conversion instructions (casting)
5101 which all take a single operand and a type. They perform various bit
5102 conversions on the operand.</p>
5104 <!-- _______________________________________________________________________ -->
5106 <a name="i_trunc">'<tt>trunc .. to</tt>' Instruction</a>
5113 <result> = trunc <ty> <value> to <ty2> <i>; yields ty2</i>
5117 <p>The '<tt>trunc</tt>' instruction truncates its operand to the
5118 type <tt>ty2</tt>.</p>
5121 <p>The '<tt>trunc</tt>' instruction takes a value to trunc, and a type to trunc it to.
5122 Both types must be of <a href="#t_integer">integer</a> types, or vectors
5123 of the same number of integers.
5124 The bit size of the <tt>value</tt> must be larger than
5125 the bit size of the destination type, <tt>ty2</tt>.
5126 Equal sized types are not allowed.</p>
5129 <p>The '<tt>trunc</tt>' instruction truncates the high order bits
5130 in <tt>value</tt> and converts the remaining bits to <tt>ty2</tt>. Since the
5131 source size must be larger than the destination size, <tt>trunc</tt> cannot
5132 be a <i>no-op cast</i>. It will always truncate bits.</p>
5136 %X = trunc i32 257 to i8 <i>; yields i8:1</i>
5137 %Y = trunc i32 123 to i1 <i>; yields i1:true</i>
5138 %Z = trunc i32 122 to i1 <i>; yields i1:false</i>
5139 %W = trunc <2 x i16> <i16 8, i16 7> to <2 x i8> <i>; yields <i8 8, i8 7></i>
5144 <!-- _______________________________________________________________________ -->
5146 <a name="i_zext">'<tt>zext .. to</tt>' Instruction</a>
5153 <result> = zext <ty> <value> to <ty2> <i>; yields ty2</i>
5157 <p>The '<tt>zext</tt>' instruction zero extends its operand to type
5162 <p>The '<tt>zext</tt>' instruction takes a value to cast, and a type to cast it to.
5163 Both types must be of <a href="#t_integer">integer</a> types, or vectors
5164 of the same number of integers.
5165 The bit size of the <tt>value</tt> must be smaller than
5166 the bit size of the destination type,
5170 <p>The <tt>zext</tt> fills the high order bits of the <tt>value</tt> with zero
5171 bits until it reaches the size of the destination type, <tt>ty2</tt>.</p>
5173 <p>When zero extending from i1, the result will always be either 0 or 1.</p>
5177 %X = zext i32 257 to i64 <i>; yields i64:257</i>
5178 %Y = zext i1 true to i32 <i>; yields i32:1</i>
5179 %Z = zext <2 x i16> <i16 8, i16 7> to <2 x i32> <i>; yields <i32 8, i32 7></i>
5184 <!-- _______________________________________________________________________ -->
5186 <a name="i_sext">'<tt>sext .. to</tt>' Instruction</a>
5193 <result> = sext <ty> <value> to <ty2> <i>; yields ty2</i>
5197 <p>The '<tt>sext</tt>' sign extends <tt>value</tt> to the type <tt>ty2</tt>.</p>
5200 <p>The '<tt>sext</tt>' instruction takes a value to cast, and a type to cast it to.
5201 Both types must be of <a href="#t_integer">integer</a> types, or vectors
5202 of the same number of integers.
5203 The bit size of the <tt>value</tt> must be smaller than
5204 the bit size of the destination type,
5208 <p>The '<tt>sext</tt>' instruction performs a sign extension by copying the sign
5209 bit (highest order bit) of the <tt>value</tt> until it reaches the bit size
5210 of the type <tt>ty2</tt>.</p>
5212 <p>When sign extending from i1, the extension always results in -1 or 0.</p>
5216 %X = sext i8 -1 to i16 <i>; yields i16 :65535</i>
5217 %Y = sext i1 true to i32 <i>; yields i32:-1</i>
5218 %Z = sext <2 x i16> <i16 8, i16 7> to <2 x i32> <i>; yields <i32 8, i32 7></i>
5223 <!-- _______________________________________________________________________ -->
5225 <a name="i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a>
5232 <result> = fptrunc <ty> <value> to <ty2> <i>; yields ty2</i>
5236 <p>The '<tt>fptrunc</tt>' instruction truncates <tt>value</tt> to type
5240 <p>The '<tt>fptrunc</tt>' instruction takes a <a href="#t_floating">floating
5241 point</a> value to cast and a <a href="#t_floating">floating point</a> type
5242 to cast it to. The size of <tt>value</tt> must be larger than the size of
5243 <tt>ty2</tt>. This implies that <tt>fptrunc</tt> cannot be used to make a
5244 <i>no-op cast</i>.</p>
5247 <p>The '<tt>fptrunc</tt>' instruction truncates a <tt>value</tt> from a larger
5248 <a href="#t_floating">floating point</a> type to a smaller
5249 <a href="#t_floating">floating point</a> type. If the value cannot fit
5250 within the destination type, <tt>ty2</tt>, then the results are
5255 %X = fptrunc double 123.0 to float <i>; yields float:123.0</i>
5256 %Y = fptrunc double 1.0E+300 to float <i>; yields undefined</i>
5261 <!-- _______________________________________________________________________ -->
5263 <a name="i_fpext">'<tt>fpext .. to</tt>' Instruction</a>
5270 <result> = fpext <ty> <value> to <ty2> <i>; yields ty2</i>
5274 <p>The '<tt>fpext</tt>' extends a floating point <tt>value</tt> to a larger
5275 floating point value.</p>
5278 <p>The '<tt>fpext</tt>' instruction takes a
5279 <a href="#t_floating">floating point</a> <tt>value</tt> to cast, and
5280 a <a href="#t_floating">floating point</a> type to cast it to. The source
5281 type must be smaller than the destination type.</p>
5284 <p>The '<tt>fpext</tt>' instruction extends the <tt>value</tt> from a smaller
5285 <a href="#t_floating">floating point</a> type to a larger
5286 <a href="#t_floating">floating point</a> type. The <tt>fpext</tt> cannot be
5287 used to make a <i>no-op cast</i> because it always changes bits. Use
5288 <tt>bitcast</tt> to make a <i>no-op cast</i> for a floating point cast.</p>
5292 %X = fpext float 3.125 to double <i>; yields double:3.125000e+00</i>
5293 %Y = fpext double %X to fp128 <i>; yields fp128:0xL00000000000000004000900000000000</i>
5298 <!-- _______________________________________________________________________ -->
5300 <a name="i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a>
5307 <result> = fptoui <ty> <value> to <ty2> <i>; yields ty2</i>
5311 <p>The '<tt>fptoui</tt>' converts a floating point <tt>value</tt> to its
5312 unsigned integer equivalent of type <tt>ty2</tt>.</p>
5315 <p>The '<tt>fptoui</tt>' instruction takes a value to cast, which must be a
5316 scalar or vector <a href="#t_floating">floating point</a> value, and a type
5317 to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
5318 type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
5319 vector integer type with the same number of elements as <tt>ty</tt></p>
5322 <p>The '<tt>fptoui</tt>' instruction converts its
5323 <a href="#t_floating">floating point</a> operand into the nearest (rounding
5324 towards zero) unsigned integer value. If the value cannot fit
5325 in <tt>ty2</tt>, the results are undefined.</p>
5329 %X = fptoui double 123.0 to i32 <i>; yields i32:123</i>
5330 %Y = fptoui float 1.0E+300 to i1 <i>; yields undefined:1</i>
5331 %Z = fptoui float 1.04E+17 to i8 <i>; yields undefined:1</i>
5336 <!-- _______________________________________________________________________ -->
5338 <a name="i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a>
5345 <result> = fptosi <ty> <value> to <ty2> <i>; yields ty2</i>
5349 <p>The '<tt>fptosi</tt>' instruction converts
5350 <a href="#t_floating">floating point</a> <tt>value</tt> to
5351 type <tt>ty2</tt>.</p>
5354 <p>The '<tt>fptosi</tt>' instruction takes a value to cast, which must be a
5355 scalar or vector <a href="#t_floating">floating point</a> value, and a type
5356 to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
5357 type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
5358 vector integer type with the same number of elements as <tt>ty</tt></p>
5361 <p>The '<tt>fptosi</tt>' instruction converts its
5362 <a href="#t_floating">floating point</a> operand into the nearest (rounding
5363 towards zero) signed integer value. If the value cannot fit in <tt>ty2</tt>,
5364 the results are undefined.</p>
5368 %X = fptosi double -123.0 to i32 <i>; yields i32:-123</i>
5369 %Y = fptosi float 1.0E-247 to i1 <i>; yields undefined:1</i>
5370 %Z = fptosi float 1.04E+17 to i8 <i>; yields undefined:1</i>
5375 <!-- _______________________________________________________________________ -->
5377 <a name="i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a>
5384 <result> = uitofp <ty> <value> to <ty2> <i>; yields ty2</i>
5388 <p>The '<tt>uitofp</tt>' instruction regards <tt>value</tt> as an unsigned
5389 integer and converts that value to the <tt>ty2</tt> type.</p>
5392 <p>The '<tt>uitofp</tt>' instruction takes a value to cast, which must be a
5393 scalar or vector <a href="#t_integer">integer</a> value, and a type to cast
5394 it to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
5395 type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
5396 floating point type with the same number of elements as <tt>ty</tt></p>
5399 <p>The '<tt>uitofp</tt>' instruction interprets its operand as an unsigned
5400 integer quantity and converts it to the corresponding floating point
5401 value. If the value cannot fit in the floating point value, the results are
5406 %X = uitofp i32 257 to float <i>; yields float:257.0</i>
5407 %Y = uitofp i8 -1 to double <i>; yields double:255.0</i>
5412 <!-- _______________________________________________________________________ -->
5414 <a name="i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a>
5421 <result> = sitofp <ty> <value> to <ty2> <i>; yields ty2</i>
5425 <p>The '<tt>sitofp</tt>' instruction regards <tt>value</tt> as a signed integer
5426 and converts that value to the <tt>ty2</tt> type.</p>
5429 <p>The '<tt>sitofp</tt>' instruction takes a value to cast, which must be a
5430 scalar or vector <a href="#t_integer">integer</a> value, and a type to cast
5431 it to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
5432 type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
5433 floating point type with the same number of elements as <tt>ty</tt></p>
5436 <p>The '<tt>sitofp</tt>' instruction interprets its operand as a signed integer
5437 quantity and converts it to the corresponding floating point value. If the
5438 value cannot fit in the floating point value, the results are undefined.</p>
5442 %X = sitofp i32 257 to float <i>; yields float:257.0</i>
5443 %Y = sitofp i8 -1 to double <i>; yields double:-1.0</i>
5448 <!-- _______________________________________________________________________ -->
5450 <a name="i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a>
5457 <result> = ptrtoint <ty> <value> to <ty2> <i>; yields ty2</i>
5461 <p>The '<tt>ptrtoint</tt>' instruction converts the pointer <tt>value</tt> to
5462 the integer type <tt>ty2</tt>.</p>
5465 <p>The '<tt>ptrtoint</tt>' instruction takes a <tt>value</tt> to cast, which
5466 must be a <a href="#t_pointer">pointer</a> value, and a type to cast it to
5467 <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a> type.</p>
5470 <p>The '<tt>ptrtoint</tt>' instruction converts <tt>value</tt> to integer type
5471 <tt>ty2</tt> by interpreting the pointer value as an integer and either
5472 truncating or zero extending that value to the size of the integer type. If
5473 <tt>value</tt> is smaller than <tt>ty2</tt> then a zero extension is done. If
5474 <tt>value</tt> is larger than <tt>ty2</tt> then a truncation is done. If they
5475 are the same size, then nothing is done (<i>no-op cast</i>) other than a type
5480 %X = ptrtoint i32* %X to i8 <i>; yields truncation on 32-bit architecture</i>
5481 %Y = ptrtoint i32* %x to i64 <i>; yields zero extension on 32-bit architecture</i>
5486 <!-- _______________________________________________________________________ -->
5488 <a name="i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a>
5495 <result> = inttoptr <ty> <value> to <ty2> <i>; yields ty2</i>
5499 <p>The '<tt>inttoptr</tt>' instruction converts an integer <tt>value</tt> to a
5500 pointer type, <tt>ty2</tt>.</p>
5503 <p>The '<tt>inttoptr</tt>' instruction takes an <a href="#t_integer">integer</a>
5504 value to cast, and a type to cast it to, which must be a
5505 <a href="#t_pointer">pointer</a> type.</p>
5508 <p>The '<tt>inttoptr</tt>' instruction converts <tt>value</tt> to type
5509 <tt>ty2</tt> by applying either a zero extension or a truncation depending on
5510 the size of the integer <tt>value</tt>. If <tt>value</tt> is larger than the
5511 size of a pointer then a truncation is done. If <tt>value</tt> is smaller
5512 than the size of a pointer then a zero extension is done. If they are the
5513 same size, nothing is done (<i>no-op cast</i>).</p>
5517 %X = inttoptr i32 255 to i32* <i>; yields zero extension on 64-bit architecture</i>
5518 %Y = inttoptr i32 255 to i32* <i>; yields no-op on 32-bit architecture</i>
5519 %Z = inttoptr i64 0 to i32* <i>; yields truncation on 32-bit architecture</i>
5524 <!-- _______________________________________________________________________ -->
5526 <a name="i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a>
5533 <result> = bitcast <ty> <value> to <ty2> <i>; yields ty2</i>
5537 <p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
5538 <tt>ty2</tt> without changing any bits.</p>
5541 <p>The '<tt>bitcast</tt>' instruction takes a value to cast, which must be a
5542 non-aggregate first class value, and a type to cast it to, which must also be
5543 a non-aggregate <a href="#t_firstclass">first class</a> type. The bit sizes
5544 of <tt>value</tt> and the destination type, <tt>ty2</tt>, must be
5545 identical. If the source type is a pointer, the destination type must also be
5546 a pointer. This instruction supports bitwise conversion of vectors to
5547 integers and to vectors of other types (as long as they have the same
5551 <p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
5552 <tt>ty2</tt>. It is always a <i>no-op cast</i> because no bits change with
5553 this conversion. The conversion is done as if the <tt>value</tt> had been
5554 stored to memory and read back as type <tt>ty2</tt>. Pointer types may only
5555 be converted to other pointer types with this instruction. To convert
5556 pointers to other types, use the <a href="#i_inttoptr">inttoptr</a> or
5557 <a href="#i_ptrtoint">ptrtoint</a> instructions first.</p>
5561 %X = bitcast i8 255 to i8 <i>; yields i8 :-1</i>
5562 %Y = bitcast i32* %x to sint* <i>; yields sint*:%x</i>
5563 %Z = bitcast <2 x int> %V to i64; <i>; yields i64: %V</i>
5570 <!-- ======================================================================= -->
5572 <a name="otherops">Other Operations</a>
5577 <p>The instructions in this category are the "miscellaneous" instructions, which
5578 defy better classification.</p>
5580 <!-- _______________________________________________________________________ -->
5582 <a name="i_icmp">'<tt>icmp</tt>' Instruction</a>
5589 <result> = icmp <cond> <ty> <op1>, <op2> <i>; yields {i1} or {<N x i1>}:result</i>
5593 <p>The '<tt>icmp</tt>' instruction returns a boolean value or a vector of
5594 boolean values based on comparison of its two integer, integer vector, or
5595 pointer operands.</p>
5598 <p>The '<tt>icmp</tt>' instruction takes three operands. The first operand is
5599 the condition code indicating the kind of comparison to perform. It is not a
5600 value, just a keyword. The possible condition code are:</p>
5603 <li><tt>eq</tt>: equal</li>
5604 <li><tt>ne</tt>: not equal </li>
5605 <li><tt>ugt</tt>: unsigned greater than</li>
5606 <li><tt>uge</tt>: unsigned greater or equal</li>
5607 <li><tt>ult</tt>: unsigned less than</li>
5608 <li><tt>ule</tt>: unsigned less or equal</li>
5609 <li><tt>sgt</tt>: signed greater than</li>
5610 <li><tt>sge</tt>: signed greater or equal</li>
5611 <li><tt>slt</tt>: signed less than</li>
5612 <li><tt>sle</tt>: signed less or equal</li>
5615 <p>The remaining two arguments must be <a href="#t_integer">integer</a> or
5616 <a href="#t_pointer">pointer</a> or integer <a href="#t_vector">vector</a>
5617 typed. They must also be identical types.</p>
5620 <p>The '<tt>icmp</tt>' compares <tt>op1</tt> and <tt>op2</tt> according to the
5621 condition code given as <tt>cond</tt>. The comparison performed always yields
5622 either an <a href="#t_integer"><tt>i1</tt></a> or vector of <tt>i1</tt>
5623 result, as follows:</p>
5626 <li><tt>eq</tt>: yields <tt>true</tt> if the operands are equal,
5627 <tt>false</tt> otherwise. No sign interpretation is necessary or
5630 <li><tt>ne</tt>: yields <tt>true</tt> if the operands are unequal,
5631 <tt>false</tt> otherwise. No sign interpretation is necessary or
5634 <li><tt>ugt</tt>: interprets the operands as unsigned values and yields
5635 <tt>true</tt> if <tt>op1</tt> is greater than <tt>op2</tt>.</li>
5637 <li><tt>uge</tt>: interprets the operands as unsigned values and yields
5638 <tt>true</tt> if <tt>op1</tt> is greater than or equal
5639 to <tt>op2</tt>.</li>
5641 <li><tt>ult</tt>: interprets the operands as unsigned values and yields
5642 <tt>true</tt> if <tt>op1</tt> is less than <tt>op2</tt>.</li>
5644 <li><tt>ule</tt>: interprets the operands as unsigned values and yields
5645 <tt>true</tt> if <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
5647 <li><tt>sgt</tt>: interprets the operands as signed values and yields
5648 <tt>true</tt> if <tt>op1</tt> is greater than <tt>op2</tt>.</li>
5650 <li><tt>sge</tt>: interprets the operands as signed values and yields
5651 <tt>true</tt> if <tt>op1</tt> is greater than or equal
5652 to <tt>op2</tt>.</li>
5654 <li><tt>slt</tt>: interprets the operands as signed values and yields
5655 <tt>true</tt> if <tt>op1</tt> is less than <tt>op2</tt>.</li>
5657 <li><tt>sle</tt>: interprets the operands as signed values and yields
5658 <tt>true</tt> if <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
5661 <p>If the operands are <a href="#t_pointer">pointer</a> typed, the pointer
5662 values are compared as if they were integers.</p>
5664 <p>If the operands are integer vectors, then they are compared element by
5665 element. The result is an <tt>i1</tt> vector with the same number of elements
5666 as the values being compared. Otherwise, the result is an <tt>i1</tt>.</p>
5670 <result> = icmp eq i32 4, 5 <i>; yields: result=false</i>
5671 <result> = icmp ne float* %X, %X <i>; yields: result=false</i>
5672 <result> = icmp ult i16 4, 5 <i>; yields: result=true</i>
5673 <result> = icmp sgt i16 4, 5 <i>; yields: result=false</i>
5674 <result> = icmp ule i16 -4, 5 <i>; yields: result=false</i>
5675 <result> = icmp sge i16 4, 5 <i>; yields: result=false</i>
5678 <p>Note that the code generator does not yet support vector types with
5679 the <tt>icmp</tt> instruction.</p>
5683 <!-- _______________________________________________________________________ -->
5685 <a name="i_fcmp">'<tt>fcmp</tt>' Instruction</a>
5692 <result> = fcmp <cond> <ty> <op1>, <op2> <i>; yields {i1} or {<N x i1>}:result</i>
5696 <p>The '<tt>fcmp</tt>' instruction returns a boolean value or vector of boolean
5697 values based on comparison of its operands.</p>
5699 <p>If the operands are floating point scalars, then the result type is a boolean
5700 (<a href="#t_integer"><tt>i1</tt></a>).</p>
5702 <p>If the operands are floating point vectors, then the result type is a vector
5703 of boolean with the same number of elements as the operands being
5707 <p>The '<tt>fcmp</tt>' instruction takes three operands. The first operand is
5708 the condition code indicating the kind of comparison to perform. It is not a
5709 value, just a keyword. The possible condition code are:</p>
5712 <li><tt>false</tt>: no comparison, always returns false</li>
5713 <li><tt>oeq</tt>: ordered and equal</li>
5714 <li><tt>ogt</tt>: ordered and greater than </li>
5715 <li><tt>oge</tt>: ordered and greater than or equal</li>
5716 <li><tt>olt</tt>: ordered and less than </li>
5717 <li><tt>ole</tt>: ordered and less than or equal</li>
5718 <li><tt>one</tt>: ordered and not equal</li>
5719 <li><tt>ord</tt>: ordered (no nans)</li>
5720 <li><tt>ueq</tt>: unordered or equal</li>
5721 <li><tt>ugt</tt>: unordered or greater than </li>
5722 <li><tt>uge</tt>: unordered or greater than or equal</li>
5723 <li><tt>ult</tt>: unordered or less than </li>
5724 <li><tt>ule</tt>: unordered or less than or equal</li>
5725 <li><tt>une</tt>: unordered or not equal</li>
5726 <li><tt>uno</tt>: unordered (either nans)</li>
5727 <li><tt>true</tt>: no comparison, always returns true</li>
5730 <p><i>Ordered</i> means that neither operand is a QNAN while
5731 <i>unordered</i> means that either operand may be a QNAN.</p>
5733 <p>Each of <tt>val1</tt> and <tt>val2</tt> arguments must be either
5734 a <a href="#t_floating">floating point</a> type or
5735 a <a href="#t_vector">vector</a> of floating point type. They must have
5736 identical types.</p>
5739 <p>The '<tt>fcmp</tt>' instruction compares <tt>op1</tt> and <tt>op2</tt>
5740 according to the condition code given as <tt>cond</tt>. If the operands are
5741 vectors, then the vectors are compared element by element. Each comparison
5742 performed always yields an <a href="#t_integer">i1</a> result, as
5746 <li><tt>false</tt>: always yields <tt>false</tt>, regardless of operands.</li>
5748 <li><tt>oeq</tt>: yields <tt>true</tt> if both operands are not a QNAN and
5749 <tt>op1</tt> is equal to <tt>op2</tt>.</li>
5751 <li><tt>ogt</tt>: yields <tt>true</tt> if both operands are not a QNAN and
5752 <tt>op1</tt> is greater than <tt>op2</tt>.</li>
5754 <li><tt>oge</tt>: yields <tt>true</tt> if both operands are not a QNAN and
5755 <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
5757 <li><tt>olt</tt>: yields <tt>true</tt> if both operands are not a QNAN and
5758 <tt>op1</tt> is less than <tt>op2</tt>.</li>
5760 <li><tt>ole</tt>: yields <tt>true</tt> if both operands are not a QNAN and
5761 <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
5763 <li><tt>one</tt>: yields <tt>true</tt> if both operands are not a QNAN and
5764 <tt>op1</tt> is not equal to <tt>op2</tt>.</li>
5766 <li><tt>ord</tt>: yields <tt>true</tt> if both operands are not a QNAN.</li>
5768 <li><tt>ueq</tt>: yields <tt>true</tt> if either operand is a QNAN or
5769 <tt>op1</tt> is equal to <tt>op2</tt>.</li>
5771 <li><tt>ugt</tt>: yields <tt>true</tt> if either operand is a QNAN or
5772 <tt>op1</tt> is greater than <tt>op2</tt>.</li>
5774 <li><tt>uge</tt>: yields <tt>true</tt> if either operand is a QNAN or
5775 <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
5777 <li><tt>ult</tt>: yields <tt>true</tt> if either operand is a QNAN or
5778 <tt>op1</tt> is less than <tt>op2</tt>.</li>
5780 <li><tt>ule</tt>: yields <tt>true</tt> if either operand is a QNAN or
5781 <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
5783 <li><tt>une</tt>: yields <tt>true</tt> if either operand is a QNAN or
5784 <tt>op1</tt> is not equal to <tt>op2</tt>.</li>
5786 <li><tt>uno</tt>: yields <tt>true</tt> if either operand is a QNAN.</li>
5788 <li><tt>true</tt>: always yields <tt>true</tt>, regardless of operands.</li>
5793 <result> = fcmp oeq float 4.0, 5.0 <i>; yields: result=false</i>
5794 <result> = fcmp one float 4.0, 5.0 <i>; yields: result=true</i>
5795 <result> = fcmp olt float 4.0, 5.0 <i>; yields: result=true</i>
5796 <result> = fcmp ueq double 1.0, 2.0 <i>; yields: result=false</i>
5799 <p>Note that the code generator does not yet support vector types with
5800 the <tt>fcmp</tt> instruction.</p>
5804 <!-- _______________________________________________________________________ -->
5806 <a name="i_phi">'<tt>phi</tt>' Instruction</a>
5813 <result> = phi <ty> [ <val0>, <label0>], ...
5817 <p>The '<tt>phi</tt>' instruction is used to implement the φ node in the
5818 SSA graph representing the function.</p>
5821 <p>The type of the incoming values is specified with the first type field. After
5822 this, the '<tt>phi</tt>' instruction takes a list of pairs as arguments, with
5823 one pair for each predecessor basic block of the current block. Only values
5824 of <a href="#t_firstclass">first class</a> type may be used as the value
5825 arguments to the PHI node. Only labels may be used as the label
5828 <p>There must be no non-phi instructions between the start of a basic block and
5829 the PHI instructions: i.e. PHI instructions must be first in a basic
5832 <p>For the purposes of the SSA form, the use of each incoming value is deemed to
5833 occur on the edge from the corresponding predecessor block to the current
5834 block (but after any definition of an '<tt>invoke</tt>' instruction's return
5835 value on the same edge).</p>
5838 <p>At runtime, the '<tt>phi</tt>' instruction logically takes on the value
5839 specified by the pair corresponding to the predecessor basic block that
5840 executed just prior to the current block.</p>
5844 Loop: ; Infinite loop that counts from 0 on up...
5845 %indvar = phi i32 [ 0, %LoopHeader ], [ %nextindvar, %Loop ]
5846 %nextindvar = add i32 %indvar, 1
5852 <!-- _______________________________________________________________________ -->
5854 <a name="i_select">'<tt>select</tt>' Instruction</a>
5861 <result> = select <i>selty</i> <cond>, <ty> <val1>, <ty> <val2> <i>; yields ty</i>
5863 <i>selty</i> is either i1 or {<N x i1>}
5867 <p>The '<tt>select</tt>' instruction is used to choose one value based on a
5868 condition, without branching.</p>
5872 <p>The '<tt>select</tt>' instruction requires an 'i1' value or a vector of 'i1'
5873 values indicating the condition, and two values of the
5874 same <a href="#t_firstclass">first class</a> type. If the val1/val2 are
5875 vectors and the condition is a scalar, then entire vectors are selected, not
5876 individual elements.</p>
5879 <p>If the condition is an i1 and it evaluates to 1, the instruction returns the
5880 first value argument; otherwise, it returns the second value argument.</p>
5882 <p>If the condition is a vector of i1, then the value arguments must be vectors
5883 of the same size, and the selection is done element by element.</p>
5887 %X = select i1 true, i8 17, i8 42 <i>; yields i8:17</i>
5890 <p>Note that the code generator does not yet support conditions
5891 with vector type.</p>
5895 <!-- _______________________________________________________________________ -->
5897 <a name="i_call">'<tt>call</tt>' Instruction</a>
5904 <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>]
5908 <p>The '<tt>call</tt>' instruction represents a simple function call.</p>
5911 <p>This instruction requires several arguments:</p>
5914 <li>The optional "tail" marker indicates that the callee function does not
5915 access any allocas or varargs in the caller. Note that calls may be
5916 marked "tail" even if they do not occur before
5917 a <a href="#i_ret"><tt>ret</tt></a> instruction. If the "tail" marker is
5918 present, the function call is eligible for tail call optimization,
5919 but <a href="CodeGenerator.html#tailcallopt">might not in fact be
5920 optimized into a jump</a>. The code generator may optimize calls marked
5921 "tail" with either 1) automatic <a href="CodeGenerator.html#sibcallopt">
5922 sibling call optimization</a> when the caller and callee have
5923 matching signatures, or 2) forced tail call optimization when the
5924 following extra requirements are met:
5926 <li>Caller and callee both have the calling
5927 convention <tt>fastcc</tt>.</li>
5928 <li>The call is in tail position (ret immediately follows call and ret
5929 uses value of call or is void).</li>
5930 <li>Option <tt>-tailcallopt</tt> is enabled,
5931 or <code>llvm::GuaranteedTailCallOpt</code> is <code>true</code>.</li>
5932 <li><a href="CodeGenerator.html#tailcallopt">Platform specific
5933 constraints are met.</a></li>
5937 <li>The optional "cconv" marker indicates which <a href="#callingconv">calling
5938 convention</a> the call should use. If none is specified, the call
5939 defaults to using C calling conventions. The calling convention of the
5940 call must match the calling convention of the target function, or else the
5941 behavior is undefined.</li>
5943 <li>The optional <a href="#paramattrs">Parameter Attributes</a> list for
5944 return values. Only '<tt>zeroext</tt>', '<tt>signext</tt>', and
5945 '<tt>inreg</tt>' attributes are valid here.</li>
5947 <li>'<tt>ty</tt>': the type of the call instruction itself which is also the
5948 type of the return value. Functions that return no value are marked
5949 <tt><a href="#t_void">void</a></tt>.</li>
5951 <li>'<tt>fnty</tt>': shall be the signature of the pointer to function value
5952 being invoked. The argument types must match the types implied by this
5953 signature. This type can be omitted if the function is not varargs and if
5954 the function type does not return a pointer to a function.</li>
5956 <li>'<tt>fnptrval</tt>': An LLVM value containing a pointer to a function to
5957 be invoked. In most cases, this is a direct function invocation, but
5958 indirect <tt>call</tt>s are just as possible, calling an arbitrary pointer
5959 to function value.</li>
5961 <li>'<tt>function args</tt>': argument list whose types match the function
5962 signature argument types and parameter attributes. All arguments must be
5963 of <a href="#t_firstclass">first class</a> type. If the function
5964 signature indicates the function accepts a variable number of arguments,
5965 the extra arguments can be specified.</li>
5967 <li>The optional <a href="#fnattrs">function attributes</a> list. Only
5968 '<tt>noreturn</tt>', '<tt>nounwind</tt>', '<tt>readonly</tt>' and
5969 '<tt>readnone</tt>' attributes are valid here.</li>
5973 <p>The '<tt>call</tt>' instruction is used to cause control flow to transfer to
5974 a specified function, with its incoming arguments bound to the specified
5975 values. Upon a '<tt><a href="#i_ret">ret</a></tt>' instruction in the called
5976 function, control flow continues with the instruction after the function
5977 call, and the return value of the function is bound to the result
5982 %retval = call i32 @test(i32 %argc)
5983 call i32 (i8*, ...)* @printf(i8* %msg, i32 12, i8 42) <i>; yields i32</i>
5984 %X = tail call i32 @foo() <i>; yields i32</i>
5985 %Y = tail call <a href="#callingconv">fastcc</a> i32 @foo() <i>; yields i32</i>
5986 call void %foo(i8 97 signext)
5988 %struct.A = type { i32, i8 }
5989 %r = call %struct.A @foo() <i>; yields { 32, i8 }</i>
5990 %gr = extractvalue %struct.A %r, 0 <i>; yields i32</i>
5991 %gr1 = extractvalue %struct.A %r, 1 <i>; yields i8</i>
5992 %Z = call void @foo() noreturn <i>; indicates that %foo never returns normally</i>
5993 %ZZ = call zeroext i32 @bar() <i>; Return value is %zero extended</i>
5996 <p>llvm treats calls to some functions with names and arguments that match the
5997 standard C99 library as being the C99 library functions, and may perform
5998 optimizations or generate code for them under that assumption. This is
5999 something we'd like to change in the future to provide better support for
6000 freestanding environments and non-C-based languages.</p>
6004 <!-- _______________________________________________________________________ -->
6006 <a name="i_va_arg">'<tt>va_arg</tt>' Instruction</a>
6013 <resultval> = va_arg <va_list*> <arglist>, <argty>
6017 <p>The '<tt>va_arg</tt>' instruction is used to access arguments passed through
6018 the "variable argument" area of a function call. It is used to implement the
6019 <tt>va_arg</tt> macro in C.</p>
6022 <p>This instruction takes a <tt>va_list*</tt> value and the type of the
6023 argument. It returns a value of the specified argument type and increments
6024 the <tt>va_list</tt> to point to the next argument. The actual type
6025 of <tt>va_list</tt> is target specific.</p>
6028 <p>The '<tt>va_arg</tt>' instruction loads an argument of the specified type
6029 from the specified <tt>va_list</tt> and causes the <tt>va_list</tt> to point
6030 to the next argument. For more information, see the variable argument
6031 handling <a href="#int_varargs">Intrinsic Functions</a>.</p>
6033 <p>It is legal for this instruction to be called in a function which does not
6034 take a variable number of arguments, for example, the <tt>vfprintf</tt>
6037 <p><tt>va_arg</tt> is an LLVM instruction instead of
6038 an <a href="#intrinsics">intrinsic function</a> because it takes a type as an
6042 <p>See the <a href="#int_varargs">variable argument processing</a> section.</p>
6044 <p>Note that the code generator does not yet fully support va_arg on many
6045 targets. Also, it does not currently support va_arg with aggregate types on
6050 <!-- _______________________________________________________________________ -->
6052 <a name="i_landingpad">'<tt>landingpad</tt>' Instruction</a>
6059 <resultval> = landingpad <somety> personality <type> <pers_fn> <clause>+
6060 <resultval> = landingpad <somety> personality <type> <pers_fn> cleanup <clause>*
6062 <clause> := catch <type> <value>
6063 <clause> := filter <array constant type> <array constant>
6067 <p>The '<tt>landingpad</tt>' instruction is used by
6068 <a href="ExceptionHandling.html#overview">LLVM's exception handling
6069 system</a> to specify that a basic block is a landing pad — one where
6070 the exception lands, and corresponds to the code found in the
6071 <i><tt>catch</tt></i> portion of a <i><tt>try/catch</tt></i> sequence. It
6072 defines values supplied by the personality function (<tt>pers_fn</tt>) upon
6073 re-entry to the function. The <tt>resultval</tt> has the
6074 type <tt>somety</tt>.</p>
6077 <p>This instruction takes a <tt>pers_fn</tt> value. This is the personality
6078 function associated with the unwinding mechanism. The optional
6079 <tt>cleanup</tt> flag indicates that the landing pad block is a cleanup.</p>
6081 <p>A <tt>clause</tt> begins with the clause type — <tt>catch</tt>
6082 or <tt>filter</tt> — and contains the global variable representing the
6083 "type" that may be caught or filtered respectively. Unlike the
6084 <tt>catch</tt> clause, the <tt>filter</tt> clause takes an array constant as
6085 its argument. Use "<tt>[0 x i8**] undef</tt>" for a filter which cannot
6086 throw. The '<tt>landingpad</tt>' instruction must contain <em>at least</em>
6087 one <tt>clause</tt> or the <tt>cleanup</tt> flag.</p>
6090 <p>The '<tt>landingpad</tt>' instruction defines the values which are set by the
6091 personality function (<tt>pers_fn</tt>) upon re-entry to the function, and
6092 therefore the "result type" of the <tt>landingpad</tt> instruction. As with
6093 calling conventions, how the personality function results are represented in
6094 LLVM IR is target specific.</p>
6096 <p>The clauses are applied in order from top to bottom. If two
6097 <tt>landingpad</tt> instructions are merged together through inlining, the
6098 clauses from the calling function are appended to the list of clauses.</p>
6100 <p>The <tt>landingpad</tt> instruction has several restrictions:</p>
6103 <li>A landing pad block is a basic block which is the unwind destination of an
6104 '<tt>invoke</tt>' instruction.</li>
6105 <li>A landing pad block must have a '<tt>landingpad</tt>' instruction as its
6106 first non-PHI instruction.</li>
6107 <li>There can be only one '<tt>landingpad</tt>' instruction within the landing
6109 <li>A basic block that is not a landing pad block may not include a
6110 '<tt>landingpad</tt>' instruction.</li>
6111 <li>All '<tt>landingpad</tt>' instructions in a function must have the same
6112 personality function.</li>
6117 ;; A landing pad which can catch an integer.
6118 %res = landingpad { i8*, i32 } personality i32 (...)* @__gxx_personality_v0
6120 ;; A landing pad that is a cleanup.
6121 %res = landingpad { i8*, i32 } personality i32 (...)* @__gxx_personality_v0
6123 ;; A landing pad which can catch an integer and can only throw a double.
6124 %res = landingpad { i8*, i32 } personality i32 (...)* @__gxx_personality_v0
6126 filter [1 x i8**] [@_ZTId]
6135 <!-- *********************************************************************** -->
6136 <h2><a name="intrinsics">Intrinsic Functions</a></h2>
6137 <!-- *********************************************************************** -->
6141 <p>LLVM supports the notion of an "intrinsic function". These functions have
6142 well known names and semantics and are required to follow certain
6143 restrictions. Overall, these intrinsics represent an extension mechanism for
6144 the LLVM language that does not require changing all of the transformations
6145 in LLVM when adding to the language (or the bitcode reader/writer, the
6146 parser, etc...).</p>
6148 <p>Intrinsic function names must all start with an "<tt>llvm.</tt>" prefix. This
6149 prefix is reserved in LLVM for intrinsic names; thus, function names may not
6150 begin with this prefix. Intrinsic functions must always be external
6151 functions: you cannot define the body of intrinsic functions. Intrinsic
6152 functions may only be used in call or invoke instructions: it is illegal to
6153 take the address of an intrinsic function. Additionally, because intrinsic
6154 functions are part of the LLVM language, it is required if any are added that
6155 they be documented here.</p>
6157 <p>Some intrinsic functions can be overloaded, i.e., the intrinsic represents a
6158 family of functions that perform the same operation but on different data
6159 types. Because LLVM can represent over 8 million different integer types,
6160 overloading is used commonly to allow an intrinsic function to operate on any
6161 integer type. One or more of the argument types or the result type can be
6162 overloaded to accept any integer type. Argument types may also be defined as
6163 exactly matching a previous argument's type or the result type. This allows
6164 an intrinsic function which accepts multiple arguments, but needs all of them
6165 to be of the same type, to only be overloaded with respect to a single
6166 argument or the result.</p>
6168 <p>Overloaded intrinsics will have the names of its overloaded argument types
6169 encoded into its function name, each preceded by a period. Only those types
6170 which are overloaded result in a name suffix. Arguments whose type is matched
6171 against another type do not. For example, the <tt>llvm.ctpop</tt> function
6172 can take an integer of any width and returns an integer of exactly the same
6173 integer width. This leads to a family of functions such as
6174 <tt>i8 @llvm.ctpop.i8(i8 %val)</tt> and <tt>i29 @llvm.ctpop.i29(i29
6175 %val)</tt>. Only one type, the return type, is overloaded, and only one type
6176 suffix is required. Because the argument's type is matched against the return
6177 type, it does not require its own name suffix.</p>
6179 <p>To learn how to add an intrinsic function, please see the
6180 <a href="ExtendingLLVM.html">Extending LLVM Guide</a>.</p>
6182 <!-- ======================================================================= -->
6184 <a name="int_varargs">Variable Argument Handling Intrinsics</a>
6189 <p>Variable argument support is defined in LLVM with
6190 the <a href="#i_va_arg"><tt>va_arg</tt></a> instruction and these three
6191 intrinsic functions. These functions are related to the similarly named
6192 macros defined in the <tt><stdarg.h></tt> header file.</p>
6194 <p>All of these functions operate on arguments that use a target-specific value
6195 type "<tt>va_list</tt>". The LLVM assembly language reference manual does
6196 not define what this type is, so all transformations should be prepared to
6197 handle these functions regardless of the type used.</p>
6199 <p>This example shows how the <a href="#i_va_arg"><tt>va_arg</tt></a>
6200 instruction and the variable argument handling intrinsic functions are
6203 <pre class="doc_code">
6204 define i32 @test(i32 %X, ...) {
6205 ; Initialize variable argument processing
6207 %ap2 = bitcast i8** %ap to i8*
6208 call void @llvm.va_start(i8* %ap2)
6210 ; Read a single integer argument
6211 %tmp = va_arg i8** %ap, i32
6213 ; Demonstrate usage of llvm.va_copy and llvm.va_end
6215 %aq2 = bitcast i8** %aq to i8*
6216 call void @llvm.va_copy(i8* %aq2, i8* %ap2)
6217 call void @llvm.va_end(i8* %aq2)
6219 ; Stop processing of arguments.
6220 call void @llvm.va_end(i8* %ap2)
6224 declare void @llvm.va_start(i8*)
6225 declare void @llvm.va_copy(i8*, i8*)
6226 declare void @llvm.va_end(i8*)
6229 <!-- _______________________________________________________________________ -->
6231 <a name="int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a>
6239 declare void %llvm.va_start(i8* <arglist>)
6243 <p>The '<tt>llvm.va_start</tt>' intrinsic initializes <tt>*<arglist></tt>
6244 for subsequent use by <tt><a href="#i_va_arg">va_arg</a></tt>.</p>
6247 <p>The argument is a pointer to a <tt>va_list</tt> element to initialize.</p>
6250 <p>The '<tt>llvm.va_start</tt>' intrinsic works just like the <tt>va_start</tt>
6251 macro available in C. In a target-dependent way, it initializes
6252 the <tt>va_list</tt> element to which the argument points, so that the next
6253 call to <tt>va_arg</tt> will produce the first variable argument passed to
6254 the function. Unlike the C <tt>va_start</tt> macro, this intrinsic does not
6255 need to know the last argument of the function as the compiler can figure
6260 <!-- _______________________________________________________________________ -->
6262 <a name="int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a>
6269 declare void @llvm.va_end(i8* <arglist>)
6273 <p>The '<tt>llvm.va_end</tt>' intrinsic destroys <tt>*<arglist></tt>,
6274 which has been initialized previously
6275 with <tt><a href="#int_va_start">llvm.va_start</a></tt>
6276 or <tt><a href="#i_va_copy">llvm.va_copy</a></tt>.</p>
6279 <p>The argument is a pointer to a <tt>va_list</tt> to destroy.</p>
6282 <p>The '<tt>llvm.va_end</tt>' intrinsic works just like the <tt>va_end</tt>
6283 macro available in C. In a target-dependent way, it destroys
6284 the <tt>va_list</tt> element to which the argument points. Calls
6285 to <a href="#int_va_start"><tt>llvm.va_start</tt></a>
6286 and <a href="#int_va_copy"> <tt>llvm.va_copy</tt></a> must be matched exactly
6287 with calls to <tt>llvm.va_end</tt>.</p>
6291 <!-- _______________________________________________________________________ -->
6293 <a name="int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a>
6300 declare void @llvm.va_copy(i8* <destarglist>, i8* <srcarglist>)
6304 <p>The '<tt>llvm.va_copy</tt>' intrinsic copies the current argument position
6305 from the source argument list to the destination argument list.</p>
6308 <p>The first argument is a pointer to a <tt>va_list</tt> element to initialize.
6309 The second argument is a pointer to a <tt>va_list</tt> element to copy
6313 <p>The '<tt>llvm.va_copy</tt>' intrinsic works just like the <tt>va_copy</tt>
6314 macro available in C. In a target-dependent way, it copies the
6315 source <tt>va_list</tt> element into the destination <tt>va_list</tt>
6316 element. This intrinsic is necessary because
6317 the <tt><a href="#int_va_start"> llvm.va_start</a></tt> intrinsic may be
6318 arbitrarily complex and require, for example, memory allocation.</p>
6326 <!-- ======================================================================= -->
6328 <a name="int_gc">Accurate Garbage Collection Intrinsics</a>
6333 <p>LLVM support for <a href="GarbageCollection.html">Accurate Garbage
6334 Collection</a> (GC) requires the implementation and generation of these
6335 intrinsics. These intrinsics allow identification of <a href="#int_gcroot">GC
6336 roots on the stack</a>, as well as garbage collector implementations that
6337 require <a href="#int_gcread">read</a> and <a href="#int_gcwrite">write</a>
6338 barriers. Front-ends for type-safe garbage collected languages should generate
6339 these intrinsics to make use of the LLVM garbage collectors. For more details,
6340 see <a href="GarbageCollection.html">Accurate Garbage Collection with
6343 <p>The garbage collection intrinsics only operate on objects in the generic
6344 address space (address space zero).</p>
6346 <!-- _______________________________________________________________________ -->
6348 <a name="int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a>
6355 declare void @llvm.gcroot(i8** %ptrloc, i8* %metadata)
6359 <p>The '<tt>llvm.gcroot</tt>' intrinsic declares the existence of a GC root to
6360 the code generator, and allows some metadata to be associated with it.</p>
6363 <p>The first argument specifies the address of a stack object that contains the
6364 root pointer. The second pointer (which must be either a constant or a
6365 global value address) contains the meta-data to be associated with the
6369 <p>At runtime, a call to this intrinsic stores a null pointer into the "ptrloc"
6370 location. At compile-time, the code generator generates information to allow
6371 the runtime to find the pointer at GC safe points. The '<tt>llvm.gcroot</tt>'
6372 intrinsic may only be used in a function which <a href="#gc">specifies a GC
6377 <!-- _______________________________________________________________________ -->
6379 <a name="int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a>
6386 declare i8* @llvm.gcread(i8* %ObjPtr, i8** %Ptr)
6390 <p>The '<tt>llvm.gcread</tt>' intrinsic identifies reads of references from heap
6391 locations, allowing garbage collector implementations that require read
6395 <p>The second argument is the address to read from, which should be an address
6396 allocated from the garbage collector. The first object is a pointer to the
6397 start of the referenced object, if needed by the language runtime (otherwise
6401 <p>The '<tt>llvm.gcread</tt>' intrinsic has the same semantics as a load
6402 instruction, but may be replaced with substantially more complex code by the
6403 garbage collector runtime, as needed. The '<tt>llvm.gcread</tt>' intrinsic
6404 may only be used in a function which <a href="#gc">specifies a GC
6409 <!-- _______________________________________________________________________ -->
6411 <a name="int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a>
6418 declare void @llvm.gcwrite(i8* %P1, i8* %Obj, i8** %P2)
6422 <p>The '<tt>llvm.gcwrite</tt>' intrinsic identifies writes of references to heap
6423 locations, allowing garbage collector implementations that require write
6424 barriers (such as generational or reference counting collectors).</p>
6427 <p>The first argument is the reference to store, the second is the start of the
6428 object to store it to, and the third is the address of the field of Obj to
6429 store to. If the runtime does not require a pointer to the object, Obj may
6433 <p>The '<tt>llvm.gcwrite</tt>' intrinsic has the same semantics as a store
6434 instruction, but may be replaced with substantially more complex code by the
6435 garbage collector runtime, as needed. The '<tt>llvm.gcwrite</tt>' intrinsic
6436 may only be used in a function which <a href="#gc">specifies a GC
6443 <!-- ======================================================================= -->
6445 <a name="int_codegen">Code Generator Intrinsics</a>
6450 <p>These intrinsics are provided by LLVM to expose special features that may
6451 only be implemented with code generator support.</p>
6453 <!-- _______________________________________________________________________ -->
6455 <a name="int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a>
6462 declare i8 *@llvm.returnaddress(i32 <level>)
6466 <p>The '<tt>llvm.returnaddress</tt>' intrinsic attempts to compute a
6467 target-specific value indicating the return address of the current function
6468 or one of its callers.</p>
6471 <p>The argument to this intrinsic indicates which function to return the address
6472 for. Zero indicates the calling function, one indicates its caller, etc.
6473 The argument is <b>required</b> to be a constant integer value.</p>
6476 <p>The '<tt>llvm.returnaddress</tt>' intrinsic either returns a pointer
6477 indicating the return address of the specified call frame, or zero if it
6478 cannot be identified. The value returned by this intrinsic is likely to be
6479 incorrect or 0 for arguments other than zero, so it should only be used for
6480 debugging purposes.</p>
6482 <p>Note that calling this intrinsic does not prevent function inlining or other
6483 aggressive transformations, so the value returned may not be that of the
6484 obvious source-language caller.</p>
6488 <!-- _______________________________________________________________________ -->
6490 <a name="int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a>
6497 declare i8* @llvm.frameaddress(i32 <level>)
6501 <p>The '<tt>llvm.frameaddress</tt>' intrinsic attempts to return the
6502 target-specific frame pointer value for the specified stack frame.</p>
6505 <p>The argument to this intrinsic indicates which function to return the frame
6506 pointer for. Zero indicates the calling function, one indicates its caller,
6507 etc. The argument is <b>required</b> to be a constant integer value.</p>
6510 <p>The '<tt>llvm.frameaddress</tt>' intrinsic either returns a pointer
6511 indicating the frame address of the specified call frame, or zero if it
6512 cannot be identified. The value returned by this intrinsic is likely to be
6513 incorrect or 0 for arguments other than zero, so it should only be used for
6514 debugging purposes.</p>
6516 <p>Note that calling this intrinsic does not prevent function inlining or other
6517 aggressive transformations, so the value returned may not be that of the
6518 obvious source-language caller.</p>
6522 <!-- _______________________________________________________________________ -->
6524 <a name="int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a>
6531 declare i8* @llvm.stacksave()
6535 <p>The '<tt>llvm.stacksave</tt>' intrinsic is used to remember the current state
6536 of the function stack, for use
6537 with <a href="#int_stackrestore"> <tt>llvm.stackrestore</tt></a>. This is
6538 useful for implementing language features like scoped automatic variable
6539 sized arrays in C99.</p>
6542 <p>This intrinsic returns a opaque pointer value that can be passed
6543 to <a href="#int_stackrestore"><tt>llvm.stackrestore</tt></a>. When
6544 an <tt>llvm.stackrestore</tt> intrinsic is executed with a value saved
6545 from <tt>llvm.stacksave</tt>, it effectively restores the state of the stack
6546 to the state it was in when the <tt>llvm.stacksave</tt> intrinsic executed.
6547 In practice, this pops any <a href="#i_alloca">alloca</a> blocks from the
6548 stack that were allocated after the <tt>llvm.stacksave</tt> was executed.</p>
6552 <!-- _______________________________________________________________________ -->
6554 <a name="int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a>
6561 declare void @llvm.stackrestore(i8* %ptr)
6565 <p>The '<tt>llvm.stackrestore</tt>' intrinsic is used to restore the state of
6566 the function stack to the state it was in when the
6567 corresponding <a href="#int_stacksave"><tt>llvm.stacksave</tt></a> intrinsic
6568 executed. This is useful for implementing language features like scoped
6569 automatic variable sized arrays in C99.</p>
6572 <p>See the description
6573 for <a href="#int_stacksave"><tt>llvm.stacksave</tt></a>.</p>
6577 <!-- _______________________________________________________________________ -->
6579 <a name="int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a>
6586 declare void @llvm.prefetch(i8* <address>, i32 <rw>, i32 <locality>, i32 <cache type>)
6590 <p>The '<tt>llvm.prefetch</tt>' intrinsic is a hint to the code generator to
6591 insert a prefetch instruction if supported; otherwise, it is a noop.
6592 Prefetches have no effect on the behavior of the program but can change its
6593 performance characteristics.</p>
6596 <p><tt>address</tt> is the address to be prefetched, <tt>rw</tt> is the
6597 specifier determining if the fetch should be for a read (0) or write (1),
6598 and <tt>locality</tt> is a temporal locality specifier ranging from (0) - no
6599 locality, to (3) - extremely local keep in cache. The <tt>cache type</tt>
6600 specifies whether the prefetch is performed on the data (1) or instruction (0)
6601 cache. The <tt>rw</tt>, <tt>locality</tt> and <tt>cache type</tt> arguments
6602 must be constant integers.</p>
6605 <p>This intrinsic does not modify the behavior of the program. In particular,
6606 prefetches cannot trap and do not produce a value. On targets that support
6607 this intrinsic, the prefetch can provide hints to the processor cache for
6608 better performance.</p>
6612 <!-- _______________________________________________________________________ -->
6614 <a name="int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a>
6621 declare void @llvm.pcmarker(i32 <id>)
6625 <p>The '<tt>llvm.pcmarker</tt>' intrinsic is a method to export a Program
6626 Counter (PC) in a region of code to simulators and other tools. The method
6627 is target specific, but it is expected that the marker will use exported
6628 symbols to transmit the PC of the marker. The marker makes no guarantees
6629 that it will remain with any specific instruction after optimizations. It is
6630 possible that the presence of a marker will inhibit optimizations. The
6631 intended use is to be inserted after optimizations to allow correlations of
6632 simulation runs.</p>
6635 <p><tt>id</tt> is a numerical id identifying the marker.</p>
6638 <p>This intrinsic does not modify the behavior of the program. Backends that do
6639 not support this intrinsic may ignore it.</p>
6643 <!-- _______________________________________________________________________ -->
6645 <a name="int_readcyclecounter">'<tt>llvm.readcyclecounter</tt>' Intrinsic</a>
6652 declare i64 @llvm.readcyclecounter()
6656 <p>The '<tt>llvm.readcyclecounter</tt>' intrinsic provides access to the cycle
6657 counter register (or similar low latency, high accuracy clocks) on those
6658 targets that support it. On X86, it should map to RDTSC. On Alpha, it
6659 should map to RPCC. As the backing counters overflow quickly (on the order
6660 of 9 seconds on alpha), this should only be used for small timings.</p>
6663 <p>When directly supported, reading the cycle counter should not modify any
6664 memory. Implementations are allowed to either return a application specific
6665 value or a system wide value. On backends without support, this is lowered
6666 to a constant 0.</p>
6672 <!-- ======================================================================= -->
6674 <a name="int_libc">Standard C Library Intrinsics</a>
6679 <p>LLVM provides intrinsics for a few important standard C library functions.
6680 These intrinsics allow source-language front-ends to pass information about
6681 the alignment of the pointer arguments to the code generator, providing
6682 opportunity for more efficient code generation.</p>
6684 <!-- _______________________________________________________________________ -->
6686 <a name="int_memcpy">'<tt>llvm.memcpy</tt>' Intrinsic</a>
6692 <p>This is an overloaded intrinsic. You can use <tt>llvm.memcpy</tt> on any
6693 integer bit width and for different address spaces. Not all targets support
6694 all bit widths however.</p>
6697 declare void @llvm.memcpy.p0i8.p0i8.i32(i8* <dest>, i8* <src>,
6698 i32 <len>, i32 <align>, i1 <isvolatile>)
6699 declare void @llvm.memcpy.p0i8.p0i8.i64(i8* <dest>, i8* <src>,
6700 i64 <len>, i32 <align>, i1 <isvolatile>)
6704 <p>The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the
6705 source location to the destination location.</p>
6707 <p>Note that, unlike the standard libc function, the <tt>llvm.memcpy.*</tt>
6708 intrinsics do not return a value, takes extra alignment/isvolatile arguments
6709 and the pointers can be in specified address spaces.</p>
6713 <p>The first argument is a pointer to the destination, the second is a pointer
6714 to the source. The third argument is an integer argument specifying the
6715 number of bytes to copy, the fourth argument is the alignment of the
6716 source and destination locations, and the fifth is a boolean indicating a
6717 volatile access.</p>
6719 <p>If the call to this intrinsic has an alignment value that is not 0 or 1,
6720 then the caller guarantees that both the source and destination pointers are
6721 aligned to that boundary.</p>
6723 <p>If the <tt>isvolatile</tt> parameter is <tt>true</tt>, the
6724 <tt>llvm.memcpy</tt> call is a <a href="#volatile">volatile operation</a>.
6725 The detailed access behavior is not very cleanly specified and it is unwise
6726 to depend on it.</p>
6730 <p>The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the
6731 source location to the destination location, which are not allowed to
6732 overlap. It copies "len" bytes of memory over. If the argument is known to
6733 be aligned to some boundary, this can be specified as the fourth argument,
6734 otherwise it should be set to 0 or 1.</p>
6738 <!-- _______________________________________________________________________ -->
6740 <a name="int_memmove">'<tt>llvm.memmove</tt>' Intrinsic</a>
6746 <p>This is an overloaded intrinsic. You can use llvm.memmove on any integer bit
6747 width and for different address space. Not all targets support all bit
6751 declare void @llvm.memmove.p0i8.p0i8.i32(i8* <dest>, i8* <src>,
6752 i32 <len>, i32 <align>, i1 <isvolatile>)
6753 declare void @llvm.memmove.p0i8.p0i8.i64(i8* <dest>, i8* <src>,
6754 i64 <len>, i32 <align>, i1 <isvolatile>)
6758 <p>The '<tt>llvm.memmove.*</tt>' intrinsics move a block of memory from the
6759 source location to the destination location. It is similar to the
6760 '<tt>llvm.memcpy</tt>' intrinsic but allows the two memory locations to
6763 <p>Note that, unlike the standard libc function, the <tt>llvm.memmove.*</tt>
6764 intrinsics do not return a value, takes extra alignment/isvolatile arguments
6765 and the pointers can be in specified address spaces.</p>
6769 <p>The first argument is a pointer to the destination, the second is a pointer
6770 to the source. The third argument is an integer argument specifying the
6771 number of bytes to copy, the fourth argument is the alignment of the
6772 source and destination locations, and the fifth is a boolean indicating a
6773 volatile access.</p>
6775 <p>If the call to this intrinsic has an alignment value that is not 0 or 1,
6776 then the caller guarantees that the source and destination pointers are
6777 aligned to that boundary.</p>
6779 <p>If the <tt>isvolatile</tt> parameter is <tt>true</tt>, the
6780 <tt>llvm.memmove</tt> call is a <a href="#volatile">volatile operation</a>.
6781 The detailed access behavior is not very cleanly specified and it is unwise
6782 to depend on it.</p>
6786 <p>The '<tt>llvm.memmove.*</tt>' intrinsics copy a block of memory from the
6787 source location to the destination location, which may overlap. It copies
6788 "len" bytes of memory over. If the argument is known to be aligned to some
6789 boundary, this can be specified as the fourth argument, otherwise it should
6790 be set to 0 or 1.</p>
6794 <!-- _______________________________________________________________________ -->
6796 <a name="int_memset">'<tt>llvm.memset.*</tt>' Intrinsics</a>
6802 <p>This is an overloaded intrinsic. You can use llvm.memset on any integer bit
6803 width and for different address spaces. However, not all targets support all
6807 declare void @llvm.memset.p0i8.i32(i8* <dest>, i8 <val>,
6808 i32 <len>, i32 <align>, i1 <isvolatile>)
6809 declare void @llvm.memset.p0i8.i64(i8* <dest>, i8 <val>,
6810 i64 <len>, i32 <align>, i1 <isvolatile>)
6814 <p>The '<tt>llvm.memset.*</tt>' intrinsics fill a block of memory with a
6815 particular byte value.</p>
6817 <p>Note that, unlike the standard libc function, the <tt>llvm.memset</tt>
6818 intrinsic does not return a value and takes extra alignment/volatile
6819 arguments. Also, the destination can be in an arbitrary address space.</p>
6822 <p>The first argument is a pointer to the destination to fill, the second is the
6823 byte value with which to fill it, the third argument is an integer argument
6824 specifying the number of bytes to fill, and the fourth argument is the known
6825 alignment of the destination location.</p>
6827 <p>If the call to this intrinsic has an alignment value that is not 0 or 1,
6828 then the caller guarantees that the destination pointer is aligned to that
6831 <p>If the <tt>isvolatile</tt> parameter is <tt>true</tt>, the
6832 <tt>llvm.memset</tt> call is a <a href="#volatile">volatile operation</a>.
6833 The detailed access behavior is not very cleanly specified and it is unwise
6834 to depend on it.</p>
6837 <p>The '<tt>llvm.memset.*</tt>' intrinsics fill "len" bytes of memory starting
6838 at the destination location. If the argument is known to be aligned to some
6839 boundary, this can be specified as the fourth argument, otherwise it should
6840 be set to 0 or 1.</p>
6844 <!-- _______________________________________________________________________ -->
6846 <a name="int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a>
6852 <p>This is an overloaded intrinsic. You can use <tt>llvm.sqrt</tt> on any
6853 floating point or vector of floating point type. Not all targets support all
6857 declare float @llvm.sqrt.f32(float %Val)
6858 declare double @llvm.sqrt.f64(double %Val)
6859 declare x86_fp80 @llvm.sqrt.f80(x86_fp80 %Val)
6860 declare fp128 @llvm.sqrt.f128(fp128 %Val)
6861 declare ppc_fp128 @llvm.sqrt.ppcf128(ppc_fp128 %Val)
6865 <p>The '<tt>llvm.sqrt</tt>' intrinsics return the sqrt of the specified operand,
6866 returning the same value as the libm '<tt>sqrt</tt>' functions would.
6867 Unlike <tt>sqrt</tt> in libm, however, <tt>llvm.sqrt</tt> has undefined
6868 behavior for negative numbers other than -0.0 (which allows for better
6869 optimization, because there is no need to worry about errno being
6870 set). <tt>llvm.sqrt(-0.0)</tt> is defined to return -0.0 like IEEE sqrt.</p>
6873 <p>The argument and return value are floating point numbers of the same
6877 <p>This function returns the sqrt of the specified operand if it is a
6878 nonnegative floating point number.</p>
6882 <!-- _______________________________________________________________________ -->
6884 <a name="int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a>
6890 <p>This is an overloaded intrinsic. You can use <tt>llvm.powi</tt> on any
6891 floating point or vector of floating point type. Not all targets support all
6895 declare float @llvm.powi.f32(float %Val, i32 %power)
6896 declare double @llvm.powi.f64(double %Val, i32 %power)
6897 declare x86_fp80 @llvm.powi.f80(x86_fp80 %Val, i32 %power)
6898 declare fp128 @llvm.powi.f128(fp128 %Val, i32 %power)
6899 declare ppc_fp128 @llvm.powi.ppcf128(ppc_fp128 %Val, i32 %power)
6903 <p>The '<tt>llvm.powi.*</tt>' intrinsics return the first operand raised to the
6904 specified (positive or negative) power. The order of evaluation of
6905 multiplications is not defined. When a vector of floating point type is
6906 used, the second argument remains a scalar integer value.</p>
6909 <p>The second argument is an integer power, and the first is a value to raise to
6913 <p>This function returns the first value raised to the second power with an
6914 unspecified sequence of rounding operations.</p>
6918 <!-- _______________________________________________________________________ -->
6920 <a name="int_sin">'<tt>llvm.sin.*</tt>' Intrinsic</a>
6926 <p>This is an overloaded intrinsic. You can use <tt>llvm.sin</tt> on any
6927 floating point or vector of floating point type. Not all targets support all
6931 declare float @llvm.sin.f32(float %Val)
6932 declare double @llvm.sin.f64(double %Val)
6933 declare x86_fp80 @llvm.sin.f80(x86_fp80 %Val)
6934 declare fp128 @llvm.sin.f128(fp128 %Val)
6935 declare ppc_fp128 @llvm.sin.ppcf128(ppc_fp128 %Val)
6939 <p>The '<tt>llvm.sin.*</tt>' intrinsics return the sine of the operand.</p>
6942 <p>The argument and return value are floating point numbers of the same
6946 <p>This function returns the sine of the specified operand, returning the same
6947 values as the libm <tt>sin</tt> functions would, and handles error conditions
6948 in the same way.</p>
6952 <!-- _______________________________________________________________________ -->
6954 <a name="int_cos">'<tt>llvm.cos.*</tt>' Intrinsic</a>
6960 <p>This is an overloaded intrinsic. You can use <tt>llvm.cos</tt> on any
6961 floating point or vector of floating point type. Not all targets support all
6965 declare float @llvm.cos.f32(float %Val)
6966 declare double @llvm.cos.f64(double %Val)
6967 declare x86_fp80 @llvm.cos.f80(x86_fp80 %Val)
6968 declare fp128 @llvm.cos.f128(fp128 %Val)
6969 declare ppc_fp128 @llvm.cos.ppcf128(ppc_fp128 %Val)
6973 <p>The '<tt>llvm.cos.*</tt>' intrinsics return the cosine of the operand.</p>
6976 <p>The argument and return value are floating point numbers of the same
6980 <p>This function returns the cosine of the specified operand, returning the same
6981 values as the libm <tt>cos</tt> functions would, and handles error conditions
6982 in the same way.</p>
6986 <!-- _______________________________________________________________________ -->
6988 <a name="int_pow">'<tt>llvm.pow.*</tt>' Intrinsic</a>
6994 <p>This is an overloaded intrinsic. You can use <tt>llvm.pow</tt> on any
6995 floating point or vector of floating point type. Not all targets support all
6999 declare float @llvm.pow.f32(float %Val, float %Power)
7000 declare double @llvm.pow.f64(double %Val, double %Power)
7001 declare x86_fp80 @llvm.pow.f80(x86_fp80 %Val, x86_fp80 %Power)
7002 declare fp128 @llvm.pow.f128(fp128 %Val, fp128 %Power)
7003 declare ppc_fp128 @llvm.pow.ppcf128(ppc_fp128 %Val, ppc_fp128 Power)
7007 <p>The '<tt>llvm.pow.*</tt>' intrinsics return the first operand raised to the
7008 specified (positive or negative) power.</p>
7011 <p>The second argument is a floating point power, and the first is a value to
7012 raise to that power.</p>
7015 <p>This function returns the first value raised to the second power, returning
7016 the same values as the libm <tt>pow</tt> functions would, and handles error
7017 conditions in the same way.</p>
7023 <!-- _______________________________________________________________________ -->
7025 <a name="int_exp">'<tt>llvm.exp.*</tt>' Intrinsic</a>
7031 <p>This is an overloaded intrinsic. You can use <tt>llvm.exp</tt> on any
7032 floating point or vector of floating point type. Not all targets support all
7036 declare float @llvm.exp.f32(float %Val)
7037 declare double @llvm.exp.f64(double %Val)
7038 declare x86_fp80 @llvm.exp.f80(x86_fp80 %Val)
7039 declare fp128 @llvm.exp.f128(fp128 %Val)
7040 declare ppc_fp128 @llvm.exp.ppcf128(ppc_fp128 %Val)
7044 <p>The '<tt>llvm.exp.*</tt>' intrinsics perform the exp function.</p>
7047 <p>The argument and return value are floating point numbers of the same
7051 <p>This function returns the same values as the libm <tt>exp</tt> functions
7052 would, and handles error conditions in the same way.</p>
7056 <!-- _______________________________________________________________________ -->
7058 <a name="int_log">'<tt>llvm.log.*</tt>' Intrinsic</a>
7064 <p>This is an overloaded intrinsic. You can use <tt>llvm.log</tt> on any
7065 floating point or vector of floating point type. Not all targets support all
7069 declare float @llvm.log.f32(float %Val)
7070 declare double @llvm.log.f64(double %Val)
7071 declare x86_fp80 @llvm.log.f80(x86_fp80 %Val)
7072 declare fp128 @llvm.log.f128(fp128 %Val)
7073 declare ppc_fp128 @llvm.log.ppcf128(ppc_fp128 %Val)
7077 <p>The '<tt>llvm.log.*</tt>' intrinsics perform the log function.</p>
7080 <p>The argument and return value are floating point numbers of the same
7084 <p>This function returns the same values as the libm <tt>log</tt> functions
7085 would, and handles error conditions in the same way.</p>
7088 <a name="int_fma">'<tt>llvm.fma.*</tt>' Intrinsic</a>
7094 <p>This is an overloaded intrinsic. You can use <tt>llvm.fma</tt> on any
7095 floating point or vector of floating point type. Not all targets support all
7099 declare float @llvm.fma.f32(float %a, float %b, float %c)
7100 declare double @llvm.fma.f64(double %a, double %b, double %c)
7101 declare x86_fp80 @llvm.fma.f80(x86_fp80 %a, x86_fp80 %b, x86_fp80 %c)
7102 declare fp128 @llvm.fma.f128(fp128 %a, fp128 %b, fp128 %c)
7103 declare ppc_fp128 @llvm.fma.ppcf128(ppc_fp128 %a, ppc_fp128 %b, ppc_fp128 %c)
7107 <p>The '<tt>llvm.fma.*</tt>' intrinsics perform the fused multiply-add
7111 <p>The argument and return value are floating point numbers of the same
7115 <p>This function returns the same values as the libm <tt>fma</tt> functions
7120 <!-- ======================================================================= -->
7122 <a name="int_manip">Bit Manipulation Intrinsics</a>
7127 <p>LLVM provides intrinsics for a few important bit manipulation operations.
7128 These allow efficient code generation for some algorithms.</p>
7130 <!-- _______________________________________________________________________ -->
7132 <a name="int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a>
7138 <p>This is an overloaded intrinsic function. You can use bswap on any integer
7139 type that is an even number of bytes (i.e. BitWidth % 16 == 0).</p>
7142 declare i16 @llvm.bswap.i16(i16 <id>)
7143 declare i32 @llvm.bswap.i32(i32 <id>)
7144 declare i64 @llvm.bswap.i64(i64 <id>)
7148 <p>The '<tt>llvm.bswap</tt>' family of intrinsics is used to byte swap integer
7149 values with an even number of bytes (positive multiple of 16 bits). These
7150 are useful for performing operations on data that is not in the target's
7151 native byte order.</p>
7154 <p>The <tt>llvm.bswap.i16</tt> intrinsic returns an i16 value that has the high
7155 and low byte of the input i16 swapped. Similarly,
7156 the <tt>llvm.bswap.i32</tt> intrinsic returns an i32 value that has the four
7157 bytes of the input i32 swapped, so that if the input bytes are numbered 0, 1,
7158 2, 3 then the returned i32 will have its bytes in 3, 2, 1, 0 order.
7159 The <tt>llvm.bswap.i48</tt>, <tt>llvm.bswap.i64</tt> and other intrinsics
7160 extend this concept to additional even-byte lengths (6 bytes, 8 bytes and
7161 more, respectively).</p>
7165 <!-- _______________________________________________________________________ -->
7167 <a name="int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic</a>
7173 <p>This is an overloaded intrinsic. You can use llvm.ctpop on any integer bit
7174 width, or on any vector with integer elements. Not all targets support all
7175 bit widths or vector types, however.</p>
7178 declare i8 @llvm.ctpop.i8(i8 <src>)
7179 declare i16 @llvm.ctpop.i16(i16 <src>)
7180 declare i32 @llvm.ctpop.i32(i32 <src>)
7181 declare i64 @llvm.ctpop.i64(i64 <src>)
7182 declare i256 @llvm.ctpop.i256(i256 <src>)
7183 declare <2 x i32> @llvm.ctpop.v2i32(<2 x i32> <src>)
7187 <p>The '<tt>llvm.ctpop</tt>' family of intrinsics counts the number of bits set
7191 <p>The only argument is the value to be counted. The argument may be of any
7192 integer type, or a vector with integer elements.
7193 The return type must match the argument type.</p>
7196 <p>The '<tt>llvm.ctpop</tt>' intrinsic counts the 1's in a variable, or within each
7197 element of a vector.</p>
7201 <!-- _______________________________________________________________________ -->
7203 <a name="int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic</a>
7209 <p>This is an overloaded intrinsic. You can use <tt>llvm.ctlz</tt> on any
7210 integer bit width, or any vector whose elements are integers. Not all
7211 targets support all bit widths or vector types, however.</p>
7214 declare i8 @llvm.ctlz.i8 (i8 <src>)
7215 declare i16 @llvm.ctlz.i16(i16 <src>)
7216 declare i32 @llvm.ctlz.i32(i32 <src>)
7217 declare i64 @llvm.ctlz.i64(i64 <src>)
7218 declare i256 @llvm.ctlz.i256(i256 <src>)
7219 declare <2 x i32> @llvm.ctlz.v2i32(<2 x i32> <src;gt)
7223 <p>The '<tt>llvm.ctlz</tt>' family of intrinsic functions counts the number of
7224 leading zeros in a variable.</p>
7227 <p>The only argument is the value to be counted. The argument may be of any
7228 integer type, or any vector type with integer element type.
7229 The return type must match the argument type.</p>
7232 <p>The '<tt>llvm.ctlz</tt>' intrinsic counts the leading (most significant)
7233 zeros in a variable, or within each element of the vector if the operation
7234 is of vector type. If the src == 0 then the result is the size in bits of
7235 the type of src. For example, <tt>llvm.ctlz(i32 2) = 30</tt>.</p>
7239 <!-- _______________________________________________________________________ -->
7241 <a name="int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic</a>
7247 <p>This is an overloaded intrinsic. You can use <tt>llvm.cttz</tt> on any
7248 integer bit width, or any vector of integer elements. Not all targets
7249 support all bit widths or vector types, however.</p>
7252 declare i8 @llvm.cttz.i8 (i8 <src>)
7253 declare i16 @llvm.cttz.i16(i16 <src>)
7254 declare i32 @llvm.cttz.i32(i32 <src>)
7255 declare i64 @llvm.cttz.i64(i64 <src>)
7256 declare i256 @llvm.cttz.i256(i256 <src>)
7257 declase <2 x i32> @llvm.cttz.v2i32(<2 x i32> <src>)
7261 <p>The '<tt>llvm.cttz</tt>' family of intrinsic functions counts the number of
7265 <p>The only argument is the value to be counted. The argument may be of any
7266 integer type, or a vectory with integer element type.. The return type
7267 must match the argument type.</p>
7270 <p>The '<tt>llvm.cttz</tt>' intrinsic counts the trailing (least significant)
7271 zeros in a variable, or within each element of a vector.
7272 If the src == 0 then the result is the size in bits of
7273 the type of src. For example, <tt>llvm.cttz(2) = 1</tt>.</p>
7279 <!-- ======================================================================= -->
7281 <a name="int_overflow">Arithmetic with Overflow Intrinsics</a>
7286 <p>LLVM provides intrinsics for some arithmetic with overflow operations.</p>
7288 <!-- _______________________________________________________________________ -->
7290 <a name="int_sadd_overflow">
7291 '<tt>llvm.sadd.with.overflow.*</tt>' Intrinsics
7298 <p>This is an overloaded intrinsic. You can use <tt>llvm.sadd.with.overflow</tt>
7299 on any integer bit width.</p>
7302 declare {i16, i1} @llvm.sadd.with.overflow.i16(i16 %a, i16 %b)
7303 declare {i32, i1} @llvm.sadd.with.overflow.i32(i32 %a, i32 %b)
7304 declare {i64, i1} @llvm.sadd.with.overflow.i64(i64 %a, i64 %b)
7308 <p>The '<tt>llvm.sadd.with.overflow</tt>' family of intrinsic functions perform
7309 a signed addition of the two arguments, and indicate whether an overflow
7310 occurred during the signed summation.</p>
7313 <p>The arguments (%a and %b) and the first element of the result structure may
7314 be of integer types of any bit width, but they must have the same bit
7315 width. The second element of the result structure must be of
7316 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
7317 undergo signed addition.</p>
7320 <p>The '<tt>llvm.sadd.with.overflow</tt>' family of intrinsic functions perform
7321 a signed addition of the two variables. They return a structure — the
7322 first element of which is the signed summation, and the second element of
7323 which is a bit specifying if the signed summation resulted in an
7328 %res = call {i32, i1} @llvm.sadd.with.overflow.i32(i32 %a, i32 %b)
7329 %sum = extractvalue {i32, i1} %res, 0
7330 %obit = extractvalue {i32, i1} %res, 1
7331 br i1 %obit, label %overflow, label %normal
7336 <!-- _______________________________________________________________________ -->
7338 <a name="int_uadd_overflow">
7339 '<tt>llvm.uadd.with.overflow.*</tt>' Intrinsics
7346 <p>This is an overloaded intrinsic. You can use <tt>llvm.uadd.with.overflow</tt>
7347 on any integer bit width.</p>
7350 declare {i16, i1} @llvm.uadd.with.overflow.i16(i16 %a, i16 %b)
7351 declare {i32, i1} @llvm.uadd.with.overflow.i32(i32 %a, i32 %b)
7352 declare {i64, i1} @llvm.uadd.with.overflow.i64(i64 %a, i64 %b)
7356 <p>The '<tt>llvm.uadd.with.overflow</tt>' family of intrinsic functions perform
7357 an unsigned addition of the two arguments, and indicate whether a carry
7358 occurred during the unsigned summation.</p>
7361 <p>The arguments (%a and %b) and the first element of the result structure may
7362 be of integer types of any bit width, but they must have the same bit
7363 width. The second element of the result structure must be of
7364 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
7365 undergo unsigned addition.</p>
7368 <p>The '<tt>llvm.uadd.with.overflow</tt>' family of intrinsic functions perform
7369 an unsigned addition of the two arguments. They return a structure —
7370 the first element of which is the sum, and the second element of which is a
7371 bit specifying if the unsigned summation resulted in a carry.</p>
7375 %res = call {i32, i1} @llvm.uadd.with.overflow.i32(i32 %a, i32 %b)
7376 %sum = extractvalue {i32, i1} %res, 0
7377 %obit = extractvalue {i32, i1} %res, 1
7378 br i1 %obit, label %carry, label %normal
7383 <!-- _______________________________________________________________________ -->
7385 <a name="int_ssub_overflow">
7386 '<tt>llvm.ssub.with.overflow.*</tt>' Intrinsics
7393 <p>This is an overloaded intrinsic. You can use <tt>llvm.ssub.with.overflow</tt>
7394 on any integer bit width.</p>
7397 declare {i16, i1} @llvm.ssub.with.overflow.i16(i16 %a, i16 %b)
7398 declare {i32, i1} @llvm.ssub.with.overflow.i32(i32 %a, i32 %b)
7399 declare {i64, i1} @llvm.ssub.with.overflow.i64(i64 %a, i64 %b)
7403 <p>The '<tt>llvm.ssub.with.overflow</tt>' family of intrinsic functions perform
7404 a signed subtraction of the two arguments, and indicate whether an overflow
7405 occurred during the signed subtraction.</p>
7408 <p>The arguments (%a and %b) and the first element of the result structure may
7409 be of integer types of any bit width, but they must have the same bit
7410 width. The second element of the result structure must be of
7411 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
7412 undergo signed subtraction.</p>
7415 <p>The '<tt>llvm.ssub.with.overflow</tt>' family of intrinsic functions perform
7416 a signed subtraction of the two arguments. They return a structure —
7417 the first element of which is the subtraction, and the second element of
7418 which is a bit specifying if the signed subtraction resulted in an
7423 %res = call {i32, i1} @llvm.ssub.with.overflow.i32(i32 %a, i32 %b)
7424 %sum = extractvalue {i32, i1} %res, 0
7425 %obit = extractvalue {i32, i1} %res, 1
7426 br i1 %obit, label %overflow, label %normal
7431 <!-- _______________________________________________________________________ -->
7433 <a name="int_usub_overflow">
7434 '<tt>llvm.usub.with.overflow.*</tt>' Intrinsics
7441 <p>This is an overloaded intrinsic. You can use <tt>llvm.usub.with.overflow</tt>
7442 on any integer bit width.</p>
7445 declare {i16, i1} @llvm.usub.with.overflow.i16(i16 %a, i16 %b)
7446 declare {i32, i1} @llvm.usub.with.overflow.i32(i32 %a, i32 %b)
7447 declare {i64, i1} @llvm.usub.with.overflow.i64(i64 %a, i64 %b)
7451 <p>The '<tt>llvm.usub.with.overflow</tt>' family of intrinsic functions perform
7452 an unsigned subtraction of the two arguments, and indicate whether an
7453 overflow occurred during the unsigned subtraction.</p>
7456 <p>The arguments (%a and %b) and the first element of the result structure may
7457 be of integer types of any bit width, but they must have the same bit
7458 width. The second element of the result structure must be of
7459 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
7460 undergo unsigned subtraction.</p>
7463 <p>The '<tt>llvm.usub.with.overflow</tt>' family of intrinsic functions perform
7464 an unsigned subtraction of the two arguments. They return a structure —
7465 the first element of which is the subtraction, and the second element of
7466 which is a bit specifying if the unsigned subtraction resulted in an
7471 %res = call {i32, i1} @llvm.usub.with.overflow.i32(i32 %a, i32 %b)
7472 %sum = extractvalue {i32, i1} %res, 0
7473 %obit = extractvalue {i32, i1} %res, 1
7474 br i1 %obit, label %overflow, label %normal
7479 <!-- _______________________________________________________________________ -->
7481 <a name="int_smul_overflow">
7482 '<tt>llvm.smul.with.overflow.*</tt>' Intrinsics
7489 <p>This is an overloaded intrinsic. You can use <tt>llvm.smul.with.overflow</tt>
7490 on any integer bit width.</p>
7493 declare {i16, i1} @llvm.smul.with.overflow.i16(i16 %a, i16 %b)
7494 declare {i32, i1} @llvm.smul.with.overflow.i32(i32 %a, i32 %b)
7495 declare {i64, i1} @llvm.smul.with.overflow.i64(i64 %a, i64 %b)
7500 <p>The '<tt>llvm.smul.with.overflow</tt>' family of intrinsic functions perform
7501 a signed multiplication of the two arguments, and indicate whether an
7502 overflow occurred during the signed multiplication.</p>
7505 <p>The arguments (%a and %b) and the first element of the result structure may
7506 be of integer types of any bit width, but they must have the same bit
7507 width. The second element of the result structure must be of
7508 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
7509 undergo signed multiplication.</p>
7512 <p>The '<tt>llvm.smul.with.overflow</tt>' family of intrinsic functions perform
7513 a signed multiplication of the two arguments. They return a structure —
7514 the first element of which is the multiplication, and the second element of
7515 which is a bit specifying if the signed multiplication resulted in an
7520 %res = call {i32, i1} @llvm.smul.with.overflow.i32(i32 %a, i32 %b)
7521 %sum = extractvalue {i32, i1} %res, 0
7522 %obit = extractvalue {i32, i1} %res, 1
7523 br i1 %obit, label %overflow, label %normal
7528 <!-- _______________________________________________________________________ -->
7530 <a name="int_umul_overflow">
7531 '<tt>llvm.umul.with.overflow.*</tt>' Intrinsics
7538 <p>This is an overloaded intrinsic. You can use <tt>llvm.umul.with.overflow</tt>
7539 on any integer bit width.</p>
7542 declare {i16, i1} @llvm.umul.with.overflow.i16(i16 %a, i16 %b)
7543 declare {i32, i1} @llvm.umul.with.overflow.i32(i32 %a, i32 %b)
7544 declare {i64, i1} @llvm.umul.with.overflow.i64(i64 %a, i64 %b)
7548 <p>The '<tt>llvm.umul.with.overflow</tt>' family of intrinsic functions perform
7549 a unsigned multiplication of the two arguments, and indicate whether an
7550 overflow occurred during the unsigned multiplication.</p>
7553 <p>The arguments (%a and %b) and the first element of the result structure may
7554 be of integer types of any bit width, but they must have the same bit
7555 width. The second element of the result structure must be of
7556 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
7557 undergo unsigned multiplication.</p>
7560 <p>The '<tt>llvm.umul.with.overflow</tt>' family of intrinsic functions perform
7561 an unsigned multiplication of the two arguments. They return a structure
7562 — the first element of which is the multiplication, and the second
7563 element of which is a bit specifying if the unsigned multiplication resulted
7568 %res = call {i32, i1} @llvm.umul.with.overflow.i32(i32 %a, i32 %b)
7569 %sum = extractvalue {i32, i1} %res, 0
7570 %obit = extractvalue {i32, i1} %res, 1
7571 br i1 %obit, label %overflow, label %normal
7578 <!-- ======================================================================= -->
7580 <a name="int_fp16">Half Precision Floating Point Intrinsics</a>
7585 <p>Half precision floating point is a storage-only format. This means that it is
7586 a dense encoding (in memory) but does not support computation in the
7589 <p>This means that code must first load the half-precision floating point
7590 value as an i16, then convert it to float with <a
7591 href="#int_convert_from_fp16"><tt>llvm.convert.from.fp16</tt></a>.
7592 Computation can then be performed on the float value (including extending to
7593 double etc). To store the value back to memory, it is first converted to
7594 float if needed, then converted to i16 with
7595 <a href="#int_convert_to_fp16"><tt>llvm.convert.to.fp16</tt></a>, then
7596 storing as an i16 value.</p>
7598 <!-- _______________________________________________________________________ -->
7600 <a name="int_convert_to_fp16">
7601 '<tt>llvm.convert.to.fp16</tt>' Intrinsic
7609 declare i16 @llvm.convert.to.fp16(f32 %a)
7613 <p>The '<tt>llvm.convert.to.fp16</tt>' intrinsic function performs
7614 a conversion from single precision floating point format to half precision
7615 floating point format.</p>
7618 <p>The intrinsic function contains single argument - the value to be
7622 <p>The '<tt>llvm.convert.to.fp16</tt>' intrinsic function performs
7623 a conversion from single precision floating point format to half precision
7624 floating point format. The return value is an <tt>i16</tt> which
7625 contains the converted number.</p>
7629 %res = call i16 @llvm.convert.to.fp16(f32 %a)
7630 store i16 %res, i16* @x, align 2
7635 <!-- _______________________________________________________________________ -->
7637 <a name="int_convert_from_fp16">
7638 '<tt>llvm.convert.from.fp16</tt>' Intrinsic
7646 declare f32 @llvm.convert.from.fp16(i16 %a)
7650 <p>The '<tt>llvm.convert.from.fp16</tt>' intrinsic function performs
7651 a conversion from half precision floating point format to single precision
7652 floating point format.</p>
7655 <p>The intrinsic function contains single argument - the value to be
7659 <p>The '<tt>llvm.convert.from.fp16</tt>' intrinsic function performs a
7660 conversion from half single precision floating point format to single
7661 precision floating point format. The input half-float value is represented by
7662 an <tt>i16</tt> value.</p>
7666 %a = load i16* @x, align 2
7667 %res = call f32 @llvm.convert.from.fp16(i16 %a)
7674 <!-- ======================================================================= -->
7676 <a name="int_debugger">Debugger Intrinsics</a>
7681 <p>The LLVM debugger intrinsics (which all start with <tt>llvm.dbg.</tt>
7682 prefix), are described in
7683 the <a href="SourceLevelDebugging.html#format_common_intrinsics">LLVM Source
7684 Level Debugging</a> document.</p>
7688 <!-- ======================================================================= -->
7690 <a name="int_eh">Exception Handling Intrinsics</a>
7695 <p>The LLVM exception handling intrinsics (which all start with
7696 <tt>llvm.eh.</tt> prefix), are described in
7697 the <a href="ExceptionHandling.html#format_common_intrinsics">LLVM Exception
7698 Handling</a> document.</p>
7702 <!-- ======================================================================= -->
7704 <a name="int_trampoline">Trampoline Intrinsics</a>
7709 <p>These intrinsics make it possible to excise one parameter, marked with
7710 the <a href="#nest"><tt>nest</tt></a> attribute, from a function.
7711 The result is a callable
7712 function pointer lacking the nest parameter - the caller does not need to
7713 provide a value for it. Instead, the value to use is stored in advance in a
7714 "trampoline", a block of memory usually allocated on the stack, which also
7715 contains code to splice the nest value into the argument list. This is used
7716 to implement the GCC nested function address extension.</p>
7718 <p>For example, if the function is
7719 <tt>i32 f(i8* nest %c, i32 %x, i32 %y)</tt> then the resulting function
7720 pointer has signature <tt>i32 (i32, i32)*</tt>. It can be created as
7723 <pre class="doc_code">
7724 %tramp = alloca [10 x i8], align 4 ; size and alignment only correct for X86
7725 %tramp1 = getelementptr [10 x i8]* %tramp, i32 0, i32 0
7726 call i8* @llvm.init.trampoline(i8* %tramp1, i8* bitcast (i32 (i8*, i32, i32)* @f to i8*), i8* %nval)
7727 %p = call i8* @llvm.adjust.trampoline(i8* %tramp1)
7728 %fp = bitcast i8* %p to i32 (i32, i32)*
7731 <p>The call <tt>%val = call i32 %fp(i32 %x, i32 %y)</tt> is then equivalent
7732 to <tt>%val = call i32 %f(i8* %nval, i32 %x, i32 %y)</tt>.</p>
7734 <!-- _______________________________________________________________________ -->
7737 '<tt>llvm.init.trampoline</tt>' Intrinsic
7745 declare void @llvm.init.trampoline(i8* <tramp>, i8* <func>, i8* <nval>)
7749 <p>This fills the memory pointed to by <tt>tramp</tt> with executable code,
7750 turning it into a trampoline.</p>
7753 <p>The <tt>llvm.init.trampoline</tt> intrinsic takes three arguments, all
7754 pointers. The <tt>tramp</tt> argument must point to a sufficiently large and
7755 sufficiently aligned block of memory; this memory is written to by the
7756 intrinsic. Note that the size and the alignment are target-specific - LLVM
7757 currently provides no portable way of determining them, so a front-end that
7758 generates this intrinsic needs to have some target-specific knowledge.
7759 The <tt>func</tt> argument must hold a function bitcast to
7760 an <tt>i8*</tt>.</p>
7763 <p>The block of memory pointed to by <tt>tramp</tt> is filled with target
7764 dependent code, turning it into a function. Then <tt>tramp</tt> needs to be
7765 passed to <a href="#int_at">llvm.adjust.trampoline</a> to get a pointer
7766 which can be <a href="#int_trampoline">bitcast (to a new function) and
7767 called</a>. The new function's signature is the same as that of
7768 <tt>func</tt> with any arguments marked with the <tt>nest</tt> attribute
7769 removed. At most one such <tt>nest</tt> argument is allowed, and it must be of
7770 pointer type. Calling the new function is equivalent to calling <tt>func</tt>
7771 with the same argument list, but with <tt>nval</tt> used for the missing
7772 <tt>nest</tt> argument. If, after calling <tt>llvm.init.trampoline</tt>, the
7773 memory pointed to by <tt>tramp</tt> is modified, then the effect of any later call
7774 to the returned function pointer is undefined.</p>
7777 <!-- _______________________________________________________________________ -->
7780 '<tt>llvm.adjust.trampoline</tt>' Intrinsic
7788 declare i8* @llvm.adjust.trampoline(i8* <tramp>)
7792 <p>This performs any required machine-specific adjustment to the address of a
7793 trampoline (passed as <tt>tramp</tt>).</p>
7796 <p><tt>tramp</tt> must point to a block of memory which already has trampoline code
7797 filled in by a previous call to <a href="#int_it"><tt>llvm.init.trampoline</tt>
7801 <p>On some architectures the address of the code to be executed needs to be
7802 different to the address where the trampoline is actually stored. This
7803 intrinsic returns the executable address corresponding to <tt>tramp</tt>
7804 after performing the required machine specific adjustments.
7805 The pointer returned can then be <a href="#int_trampoline"> bitcast and
7813 <!-- ======================================================================= -->
7815 <a name="int_atomics">Atomic Operations and Synchronization Intrinsics</a>
7820 <p>These intrinsic functions expand the "universal IR" of LLVM to represent
7821 hardware constructs for atomic operations and memory synchronization. This
7822 provides an interface to the hardware, not an interface to the programmer. It
7823 is aimed at a low enough level to allow any programming models or APIs
7824 (Application Programming Interfaces) which need atomic behaviors to map
7825 cleanly onto it. It is also modeled primarily on hardware behavior. Just as
7826 hardware provides a "universal IR" for source languages, it also provides a
7827 starting point for developing a "universal" atomic operation and
7828 synchronization IR.</p>
7830 <p>These do <em>not</em> form an API such as high-level threading libraries,
7831 software transaction memory systems, atomic primitives, and intrinsic
7832 functions as found in BSD, GNU libc, atomic_ops, APR, and other system and
7833 application libraries. The hardware interface provided by LLVM should allow
7834 a clean implementation of all of these APIs and parallel programming models.
7835 No one model or paradigm should be selected above others unless the hardware
7836 itself ubiquitously does so.</p>
7838 <!-- _______________________________________________________________________ -->
7840 <a name="int_memory_barrier">'<tt>llvm.memory.barrier</tt>' Intrinsic</a>
7846 declare void @llvm.memory.barrier(i1 <ll>, i1 <ls>, i1 <sl>, i1 <ss>, i1 <device>)
7850 <p>The <tt>llvm.memory.barrier</tt> intrinsic guarantees ordering between
7851 specific pairs of memory access types.</p>
7854 <p>The <tt>llvm.memory.barrier</tt> intrinsic requires five boolean arguments.
7855 The first four arguments enables a specific barrier as listed below. The
7856 fifth argument specifies that the barrier applies to io or device or uncached
7860 <li><tt>ll</tt>: load-load barrier</li>
7861 <li><tt>ls</tt>: load-store barrier</li>
7862 <li><tt>sl</tt>: store-load barrier</li>
7863 <li><tt>ss</tt>: store-store barrier</li>
7864 <li><tt>device</tt>: barrier applies to device and uncached memory also.</li>
7868 <p>This intrinsic causes the system to enforce some ordering constraints upon
7869 the loads and stores of the program. This barrier does not
7870 indicate <em>when</em> any events will occur, it only enforces
7871 an <em>order</em> in which they occur. For any of the specified pairs of load
7872 and store operations (f.ex. load-load, or store-load), all of the first
7873 operations preceding the barrier will complete before any of the second
7874 operations succeeding the barrier begin. Specifically the semantics for each
7875 pairing is as follows:</p>
7878 <li><tt>ll</tt>: All loads before the barrier must complete before any load
7879 after the barrier begins.</li>
7880 <li><tt>ls</tt>: All loads before the barrier must complete before any
7881 store after the barrier begins.</li>
7882 <li><tt>ss</tt>: All stores before the barrier must complete before any
7883 store after the barrier begins.</li>
7884 <li><tt>sl</tt>: All stores before the barrier must complete before any
7885 load after the barrier begins.</li>
7888 <p>These semantics are applied with a logical "and" behavior when more than one
7889 is enabled in a single memory barrier intrinsic.</p>
7891 <p>Backends may implement stronger barriers than those requested when they do
7892 not support as fine grained a barrier as requested. Some architectures do
7893 not need all types of barriers and on such architectures, these become
7898 %mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
7899 %ptr = bitcast i8* %mallocP to i32*
7902 %result1 = load i32* %ptr <i>; yields {i32}:result1 = 4</i>
7903 call void @llvm.memory.barrier(i1 false, i1 true, i1 false, i1 false, i1 true)
7904 <i>; guarantee the above finishes</i>
7905 store i32 8, %ptr <i>; before this begins</i>
7910 <!-- _______________________________________________________________________ -->
7912 <a name="int_atomic_cmp_swap">'<tt>llvm.atomic.cmp.swap.*</tt>' Intrinsic</a>
7918 <p>This is an overloaded intrinsic. You can use <tt>llvm.atomic.cmp.swap</tt> on
7919 any integer bit width and for different address spaces. Not all targets
7920 support all bit widths however.</p>
7923 declare i8 @llvm.atomic.cmp.swap.i8.p0i8(i8* <ptr>, i8 <cmp>, i8 <val>)
7924 declare i16 @llvm.atomic.cmp.swap.i16.p0i16(i16* <ptr>, i16 <cmp>, i16 <val>)
7925 declare i32 @llvm.atomic.cmp.swap.i32.p0i32(i32* <ptr>, i32 <cmp>, i32 <val>)
7926 declare i64 @llvm.atomic.cmp.swap.i64.p0i64(i64* <ptr>, i64 <cmp>, i64 <val>)
7930 <p>This loads a value in memory and compares it to a given value. If they are
7931 equal, it stores a new value into the memory.</p>
7934 <p>The <tt>llvm.atomic.cmp.swap</tt> intrinsic takes three arguments. The result
7935 as well as both <tt>cmp</tt> and <tt>val</tt> must be integer values with the
7936 same bit width. The <tt>ptr</tt> argument must be a pointer to a value of
7937 this integer type. While any bit width integer may be used, targets may only
7938 lower representations they support in hardware.</p>
7941 <p>This entire intrinsic must be executed atomically. It first loads the value
7942 in memory pointed to by <tt>ptr</tt> and compares it with the
7943 value <tt>cmp</tt>. If they are equal, <tt>val</tt> is stored into the
7944 memory. The loaded value is yielded in all cases. This provides the
7945 equivalent of an atomic compare-and-swap operation within the SSA
7950 %mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
7951 %ptr = bitcast i8* %mallocP to i32*
7954 %val1 = add i32 4, 4
7955 %result1 = call i32 @llvm.atomic.cmp.swap.i32.p0i32(i32* %ptr, i32 4, %val1)
7956 <i>; yields {i32}:result1 = 4</i>
7957 %stored1 = icmp eq i32 %result1, 4 <i>; yields {i1}:stored1 = true</i>
7958 %memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 8</i>
7960 %val2 = add i32 1, 1
7961 %result2 = call i32 @llvm.atomic.cmp.swap.i32.p0i32(i32* %ptr, i32 5, %val2)
7962 <i>; yields {i32}:result2 = 8</i>
7963 %stored2 = icmp eq i32 %result2, 5 <i>; yields {i1}:stored2 = false</i>
7965 %memval2 = load i32* %ptr <i>; yields {i32}:memval2 = 8</i>
7970 <!-- _______________________________________________________________________ -->
7972 <a name="int_atomic_swap">'<tt>llvm.atomic.swap.*</tt>' Intrinsic</a>
7978 <p>This is an overloaded intrinsic. You can use <tt>llvm.atomic.swap</tt> on any
7979 integer bit width. Not all targets support all bit widths however.</p>
7982 declare i8 @llvm.atomic.swap.i8.p0i8(i8* <ptr>, i8 <val>)
7983 declare i16 @llvm.atomic.swap.i16.p0i16(i16* <ptr>, i16 <val>)
7984 declare i32 @llvm.atomic.swap.i32.p0i32(i32* <ptr>, i32 <val>)
7985 declare i64 @llvm.atomic.swap.i64.p0i64(i64* <ptr>, i64 <val>)
7989 <p>This intrinsic loads the value stored in memory at <tt>ptr</tt> and yields
7990 the value from memory. It then stores the value in <tt>val</tt> in the memory
7991 at <tt>ptr</tt>.</p>
7994 <p>The <tt>llvm.atomic.swap</tt> intrinsic takes two arguments. Both
7995 the <tt>val</tt> argument and the result must be integers of the same bit
7996 width. The first argument, <tt>ptr</tt>, must be a pointer to a value of this
7997 integer type. The targets may only lower integer representations they
8001 <p>This intrinsic loads the value pointed to by <tt>ptr</tt>, yields it, and
8002 stores <tt>val</tt> back into <tt>ptr</tt> atomically. This provides the
8003 equivalent of an atomic swap operation within the SSA framework.</p>
8007 %mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
8008 %ptr = bitcast i8* %mallocP to i32*
8011 %val1 = add i32 4, 4
8012 %result1 = call i32 @llvm.atomic.swap.i32.p0i32(i32* %ptr, i32 %val1)
8013 <i>; yields {i32}:result1 = 4</i>
8014 %stored1 = icmp eq i32 %result1, 4 <i>; yields {i1}:stored1 = true</i>
8015 %memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 8</i>
8017 %val2 = add i32 1, 1
8018 %result2 = call i32 @llvm.atomic.swap.i32.p0i32(i32* %ptr, i32 %val2)
8019 <i>; yields {i32}:result2 = 8</i>
8021 %stored2 = icmp eq i32 %result2, 8 <i>; yields {i1}:stored2 = true</i>
8022 %memval2 = load i32* %ptr <i>; yields {i32}:memval2 = 2</i>
8027 <!-- _______________________________________________________________________ -->
8029 <a name="int_atomic_load_add">'<tt>llvm.atomic.load.add.*</tt>' Intrinsic</a>
8035 <p>This is an overloaded intrinsic. You can use <tt>llvm.atomic.load.add</tt> on
8036 any integer bit width. Not all targets support all bit widths however.</p>
8039 declare i8 @llvm.atomic.load.add.i8.p0i8(i8* <ptr>, i8 <delta>)
8040 declare i16 @llvm.atomic.load.add.i16.p0i16(i16* <ptr>, i16 <delta>)
8041 declare i32 @llvm.atomic.load.add.i32.p0i32(i32* <ptr>, i32 <delta>)
8042 declare i64 @llvm.atomic.load.add.i64.p0i64(i64* <ptr>, i64 <delta>)
8046 <p>This intrinsic adds <tt>delta</tt> to the value stored in memory
8047 at <tt>ptr</tt>. It yields the original value at <tt>ptr</tt>.</p>
8050 <p>The intrinsic takes two arguments, the first a pointer to an integer value
8051 and the second an integer value. The result is also an integer value. These
8052 integer types can have any bit width, but they must all have the same bit
8053 width. The targets may only lower integer representations they support.</p>
8056 <p>This intrinsic does a series of operations atomically. It first loads the
8057 value stored at <tt>ptr</tt>. It then adds <tt>delta</tt>, stores the result
8058 to <tt>ptr</tt>. It yields the original value stored at <tt>ptr</tt>.</p>
8062 %mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
8063 %ptr = bitcast i8* %mallocP to i32*
8065 %result1 = call i32 @llvm.atomic.load.add.i32.p0i32(i32* %ptr, i32 4)
8066 <i>; yields {i32}:result1 = 4</i>
8067 %result2 = call i32 @llvm.atomic.load.add.i32.p0i32(i32* %ptr, i32 2)
8068 <i>; yields {i32}:result2 = 8</i>
8069 %result3 = call i32 @llvm.atomic.load.add.i32.p0i32(i32* %ptr, i32 5)
8070 <i>; yields {i32}:result3 = 10</i>
8071 %memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 15</i>
8076 <!-- _______________________________________________________________________ -->
8078 <a name="int_atomic_load_sub">'<tt>llvm.atomic.load.sub.*</tt>' Intrinsic</a>
8084 <p>This is an overloaded intrinsic. You can use <tt>llvm.atomic.load.sub</tt> on
8085 any integer bit width and for different address spaces. Not all targets
8086 support all bit widths however.</p>
8089 declare i8 @llvm.atomic.load.sub.i8.p0i32(i8* <ptr>, i8 <delta>)
8090 declare i16 @llvm.atomic.load.sub.i16.p0i32(i16* <ptr>, i16 <delta>)
8091 declare i32 @llvm.atomic.load.sub.i32.p0i32(i32* <ptr>, i32 <delta>)
8092 declare i64 @llvm.atomic.load.sub.i64.p0i32(i64* <ptr>, i64 <delta>)
8096 <p>This intrinsic subtracts <tt>delta</tt> to the value stored in memory at
8097 <tt>ptr</tt>. It yields the original value at <tt>ptr</tt>.</p>
8100 <p>The intrinsic takes two arguments, the first a pointer to an integer value
8101 and the second an integer value. The result is also an integer value. These
8102 integer types can have any bit width, but they must all have the same bit
8103 width. The targets may only lower integer representations they support.</p>
8106 <p>This intrinsic does a series of operations atomically. It first loads the
8107 value stored at <tt>ptr</tt>. It then subtracts <tt>delta</tt>, stores the
8108 result to <tt>ptr</tt>. It yields the original value stored
8109 at <tt>ptr</tt>.</p>
8113 %mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
8114 %ptr = bitcast i8* %mallocP to i32*
8116 %result1 = call i32 @llvm.atomic.load.sub.i32.p0i32(i32* %ptr, i32 4)
8117 <i>; yields {i32}:result1 = 8</i>
8118 %result2 = call i32 @llvm.atomic.load.sub.i32.p0i32(i32* %ptr, i32 2)
8119 <i>; yields {i32}:result2 = 4</i>
8120 %result3 = call i32 @llvm.atomic.load.sub.i32.p0i32(i32* %ptr, i32 5)
8121 <i>; yields {i32}:result3 = 2</i>
8122 %memval1 = load i32* %ptr <i>; yields {i32}:memval1 = -3</i>
8127 <!-- _______________________________________________________________________ -->
8129 <a name="int_atomic_load_and">
8130 '<tt>llvm.atomic.load.and.*</tt>' Intrinsic
8133 <a name="int_atomic_load_nand">
8134 '<tt>llvm.atomic.load.nand.*</tt>' Intrinsic
8137 <a name="int_atomic_load_or">
8138 '<tt>llvm.atomic.load.or.*</tt>' Intrinsic
8141 <a name="int_atomic_load_xor">
8142 '<tt>llvm.atomic.load.xor.*</tt>' Intrinsic
8149 <p>These are overloaded intrinsics. You can
8150 use <tt>llvm.atomic.load_and</tt>, <tt>llvm.atomic.load_nand</tt>,
8151 <tt>llvm.atomic.load_or</tt>, and <tt>llvm.atomic.load_xor</tt> on any integer
8152 bit width and for different address spaces. Not all targets support all bit
8156 declare i8 @llvm.atomic.load.and.i8.p0i8(i8* <ptr>, i8 <delta>)
8157 declare i16 @llvm.atomic.load.and.i16.p0i16(i16* <ptr>, i16 <delta>)
8158 declare i32 @llvm.atomic.load.and.i32.p0i32(i32* <ptr>, i32 <delta>)
8159 declare i64 @llvm.atomic.load.and.i64.p0i64(i64* <ptr>, i64 <delta>)
8163 declare i8 @llvm.atomic.load.or.i8.p0i8(i8* <ptr>, i8 <delta>)
8164 declare i16 @llvm.atomic.load.or.i16.p0i16(i16* <ptr>, i16 <delta>)
8165 declare i32 @llvm.atomic.load.or.i32.p0i32(i32* <ptr>, i32 <delta>)
8166 declare i64 @llvm.atomic.load.or.i64.p0i64(i64* <ptr>, i64 <delta>)
8170 declare i8 @llvm.atomic.load.nand.i8.p0i32(i8* <ptr>, i8 <delta>)
8171 declare i16 @llvm.atomic.load.nand.i16.p0i32(i16* <ptr>, i16 <delta>)
8172 declare i32 @llvm.atomic.load.nand.i32.p0i32(i32* <ptr>, i32 <delta>)
8173 declare i64 @llvm.atomic.load.nand.i64.p0i32(i64* <ptr>, i64 <delta>)
8177 declare i8 @llvm.atomic.load.xor.i8.p0i32(i8* <ptr>, i8 <delta>)
8178 declare i16 @llvm.atomic.load.xor.i16.p0i32(i16* <ptr>, i16 <delta>)
8179 declare i32 @llvm.atomic.load.xor.i32.p0i32(i32* <ptr>, i32 <delta>)
8180 declare i64 @llvm.atomic.load.xor.i64.p0i32(i64* <ptr>, i64 <delta>)
8184 <p>These intrinsics bitwise the operation (and, nand, or, xor) <tt>delta</tt> to
8185 the value stored in memory at <tt>ptr</tt>. It yields the original value
8186 at <tt>ptr</tt>.</p>
8189 <p>These intrinsics take two arguments, the first a pointer to an integer value
8190 and the second an integer value. The result is also an integer value. These
8191 integer types can have any bit width, but they must all have the same bit
8192 width. The targets may only lower integer representations they support.</p>
8195 <p>These intrinsics does a series of operations atomically. They first load the
8196 value stored at <tt>ptr</tt>. They then do the bitwise
8197 operation <tt>delta</tt>, store the result to <tt>ptr</tt>. They yield the
8198 original value stored at <tt>ptr</tt>.</p>
8202 %mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
8203 %ptr = bitcast i8* %mallocP to i32*
8204 store i32 0x0F0F, %ptr
8205 %result0 = call i32 @llvm.atomic.load.nand.i32.p0i32(i32* %ptr, i32 0xFF)
8206 <i>; yields {i32}:result0 = 0x0F0F</i>
8207 %result1 = call i32 @llvm.atomic.load.and.i32.p0i32(i32* %ptr, i32 0xFF)
8208 <i>; yields {i32}:result1 = 0xFFFFFFF0</i>
8209 %result2 = call i32 @llvm.atomic.load.or.i32.p0i32(i32* %ptr, i32 0F)
8210 <i>; yields {i32}:result2 = 0xF0</i>
8211 %result3 = call i32 @llvm.atomic.load.xor.i32.p0i32(i32* %ptr, i32 0F)
8212 <i>; yields {i32}:result3 = FF</i>
8213 %memval1 = load i32* %ptr <i>; yields {i32}:memval1 = F0</i>
8218 <!-- _______________________________________________________________________ -->
8220 <a name="int_atomic_load_max">
8221 '<tt>llvm.atomic.load.max.*</tt>' Intrinsic
8224 <a name="int_atomic_load_min">
8225 '<tt>llvm.atomic.load.min.*</tt>' Intrinsic
8228 <a name="int_atomic_load_umax">
8229 '<tt>llvm.atomic.load.umax.*</tt>' Intrinsic
8232 <a name="int_atomic_load_umin">
8233 '<tt>llvm.atomic.load.umin.*</tt>' Intrinsic
8240 <p>These are overloaded intrinsics. You can use <tt>llvm.atomic.load_max</tt>,
8241 <tt>llvm.atomic.load_min</tt>, <tt>llvm.atomic.load_umax</tt>, and
8242 <tt>llvm.atomic.load_umin</tt> on any integer bit width and for different
8243 address spaces. Not all targets support all bit widths however.</p>
8246 declare i8 @llvm.atomic.load.max.i8.p0i8(i8* <ptr>, i8 <delta>)
8247 declare i16 @llvm.atomic.load.max.i16.p0i16(i16* <ptr>, i16 <delta>)
8248 declare i32 @llvm.atomic.load.max.i32.p0i32(i32* <ptr>, i32 <delta>)
8249 declare i64 @llvm.atomic.load.max.i64.p0i64(i64* <ptr>, i64 <delta>)
8253 declare i8 @llvm.atomic.load.min.i8.p0i8(i8* <ptr>, i8 <delta>)
8254 declare i16 @llvm.atomic.load.min.i16.p0i16(i16* <ptr>, i16 <delta>)
8255 declare i32 @llvm.atomic.load.min.i32.p0i32(i32* <ptr>, i32 <delta>)
8256 declare i64 @llvm.atomic.load.min.i64.p0i64(i64* <ptr>, i64 <delta>)
8260 declare i8 @llvm.atomic.load.umax.i8.p0i8(i8* <ptr>, i8 <delta>)
8261 declare i16 @llvm.atomic.load.umax.i16.p0i16(i16* <ptr>, i16 <delta>)
8262 declare i32 @llvm.atomic.load.umax.i32.p0i32(i32* <ptr>, i32 <delta>)
8263 declare i64 @llvm.atomic.load.umax.i64.p0i64(i64* <ptr>, i64 <delta>)
8267 declare i8 @llvm.atomic.load.umin.i8.p0i8(i8* <ptr>, i8 <delta>)
8268 declare i16 @llvm.atomic.load.umin.i16.p0i16(i16* <ptr>, i16 <delta>)
8269 declare i32 @llvm.atomic.load.umin.i32.p0i32(i32* <ptr>, i32 <delta>)
8270 declare i64 @llvm.atomic.load.umin.i64.p0i64(i64* <ptr>, i64 <delta>)
8274 <p>These intrinsics takes the signed or unsigned minimum or maximum of
8275 <tt>delta</tt> and the value stored in memory at <tt>ptr</tt>. It yields the
8276 original value at <tt>ptr</tt>.</p>
8279 <p>These intrinsics take two arguments, the first a pointer to an integer value
8280 and the second an integer value. The result is also an integer value. These
8281 integer types can have any bit width, but they must all have the same bit
8282 width. The targets may only lower integer representations they support.</p>
8285 <p>These intrinsics does a series of operations atomically. They first load the
8286 value stored at <tt>ptr</tt>. They then do the signed or unsigned min or
8287 max <tt>delta</tt> and the value, store the result to <tt>ptr</tt>. They
8288 yield the original value stored at <tt>ptr</tt>.</p>
8292 %mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
8293 %ptr = bitcast i8* %mallocP to i32*
8295 %result0 = call i32 @llvm.atomic.load.min.i32.p0i32(i32* %ptr, i32 -2)
8296 <i>; yields {i32}:result0 = 7</i>
8297 %result1 = call i32 @llvm.atomic.load.max.i32.p0i32(i32* %ptr, i32 8)
8298 <i>; yields {i32}:result1 = -2</i>
8299 %result2 = call i32 @llvm.atomic.load.umin.i32.p0i32(i32* %ptr, i32 10)
8300 <i>; yields {i32}:result2 = 8</i>
8301 %result3 = call i32 @llvm.atomic.load.umax.i32.p0i32(i32* %ptr, i32 30)
8302 <i>; yields {i32}:result3 = 8</i>
8303 %memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 30</i>
8310 <!-- ======================================================================= -->
8312 <a name="int_memorymarkers">Memory Use Markers</a>
8317 <p>This class of intrinsics exists to information about the lifetime of memory
8318 objects and ranges where variables are immutable.</p>
8320 <!-- _______________________________________________________________________ -->
8322 <a name="int_lifetime_start">'<tt>llvm.lifetime.start</tt>' Intrinsic</a>
8329 declare void @llvm.lifetime.start(i64 <size>, i8* nocapture <ptr>)
8333 <p>The '<tt>llvm.lifetime.start</tt>' intrinsic specifies the start of a memory
8334 object's lifetime.</p>
8337 <p>The first argument is a constant integer representing the size of the
8338 object, or -1 if it is variable sized. The second argument is a pointer to
8342 <p>This intrinsic indicates that before this point in the code, the value of the
8343 memory pointed to by <tt>ptr</tt> is dead. This means that it is known to
8344 never be used and has an undefined value. A load from the pointer that
8345 precedes this intrinsic can be replaced with
8346 <tt>'<a href="#undefvalues">undef</a>'</tt>.</p>
8350 <!-- _______________________________________________________________________ -->
8352 <a name="int_lifetime_end">'<tt>llvm.lifetime.end</tt>' Intrinsic</a>
8359 declare void @llvm.lifetime.end(i64 <size>, i8* nocapture <ptr>)
8363 <p>The '<tt>llvm.lifetime.end</tt>' intrinsic specifies the end of a memory
8364 object's lifetime.</p>
8367 <p>The first argument is a constant integer representing the size of the
8368 object, or -1 if it is variable sized. The second argument is a pointer to
8372 <p>This intrinsic indicates that after this point in the code, the value of the
8373 memory pointed to by <tt>ptr</tt> is dead. This means that it is known to
8374 never be used and has an undefined value. Any stores into the memory object
8375 following this intrinsic may be removed as dead.
8379 <!-- _______________________________________________________________________ -->
8381 <a name="int_invariant_start">'<tt>llvm.invariant.start</tt>' Intrinsic</a>
8388 declare {}* @llvm.invariant.start(i64 <size>, i8* nocapture <ptr>)
8392 <p>The '<tt>llvm.invariant.start</tt>' intrinsic specifies that the contents of
8393 a memory object will not change.</p>
8396 <p>The first argument is a constant integer representing the size of the
8397 object, or -1 if it is variable sized. The second argument is a pointer to
8401 <p>This intrinsic indicates that until an <tt>llvm.invariant.end</tt> that uses
8402 the return value, the referenced memory location is constant and
8407 <!-- _______________________________________________________________________ -->
8409 <a name="int_invariant_end">'<tt>llvm.invariant.end</tt>' Intrinsic</a>
8416 declare void @llvm.invariant.end({}* <start>, i64 <size>, i8* nocapture <ptr>)
8420 <p>The '<tt>llvm.invariant.end</tt>' intrinsic specifies that the contents of
8421 a memory object are mutable.</p>
8424 <p>The first argument is the matching <tt>llvm.invariant.start</tt> intrinsic.
8425 The second argument is a constant integer representing the size of the
8426 object, or -1 if it is variable sized and the third argument is a pointer
8430 <p>This intrinsic indicates that the memory is mutable again.</p>
8436 <!-- ======================================================================= -->
8438 <a name="int_general">General Intrinsics</a>
8443 <p>This class of intrinsics is designed to be generic and has no specific
8446 <!-- _______________________________________________________________________ -->
8448 <a name="int_var_annotation">'<tt>llvm.var.annotation</tt>' Intrinsic</a>
8455 declare void @llvm.var.annotation(i8* <val>, i8* <str>, i8* <str>, i32 <int>)
8459 <p>The '<tt>llvm.var.annotation</tt>' intrinsic.</p>
8462 <p>The first argument is a pointer to a value, the second is a pointer to a
8463 global string, the third is a pointer to a global string which is the source
8464 file name, and the last argument is the line number.</p>
8467 <p>This intrinsic allows annotation of local variables with arbitrary strings.
8468 This can be useful for special purpose optimizations that want to look for
8469 these annotations. These have no other defined use; they are ignored by code
8470 generation and optimization.</p>
8474 <!-- _______________________________________________________________________ -->
8476 <a name="int_annotation">'<tt>llvm.annotation.*</tt>' Intrinsic</a>
8482 <p>This is an overloaded intrinsic. You can use '<tt>llvm.annotation</tt>' on
8483 any integer bit width.</p>
8486 declare i8 @llvm.annotation.i8(i8 <val>, i8* <str>, i8* <str>, i32 <int>)
8487 declare i16 @llvm.annotation.i16(i16 <val>, i8* <str>, i8* <str>, i32 <int>)
8488 declare i32 @llvm.annotation.i32(i32 <val>, i8* <str>, i8* <str>, i32 <int>)
8489 declare i64 @llvm.annotation.i64(i64 <val>, i8* <str>, i8* <str>, i32 <int>)
8490 declare i256 @llvm.annotation.i256(i256 <val>, i8* <str>, i8* <str>, i32 <int>)
8494 <p>The '<tt>llvm.annotation</tt>' intrinsic.</p>
8497 <p>The first argument is an integer value (result of some expression), the
8498 second is a pointer to a global string, the third is a pointer to a global
8499 string which is the source file name, and the last argument is the line
8500 number. It returns the value of the first argument.</p>
8503 <p>This intrinsic allows annotations to be put on arbitrary expressions with
8504 arbitrary strings. This can be useful for special purpose optimizations that
8505 want to look for these annotations. These have no other defined use; they
8506 are ignored by code generation and optimization.</p>
8510 <!-- _______________________________________________________________________ -->
8512 <a name="int_trap">'<tt>llvm.trap</tt>' Intrinsic</a>
8519 declare void @llvm.trap()
8523 <p>The '<tt>llvm.trap</tt>' intrinsic.</p>
8529 <p>This intrinsics is lowered to the target dependent trap instruction. If the
8530 target does not have a trap instruction, this intrinsic will be lowered to
8531 the call of the <tt>abort()</tt> function.</p>
8535 <!-- _______________________________________________________________________ -->
8537 <a name="int_stackprotector">'<tt>llvm.stackprotector</tt>' Intrinsic</a>
8544 declare void @llvm.stackprotector(i8* <guard>, i8** <slot>)
8548 <p>The <tt>llvm.stackprotector</tt> intrinsic takes the <tt>guard</tt> and
8549 stores it onto the stack at <tt>slot</tt>. The stack slot is adjusted to
8550 ensure that it is placed on the stack before local variables.</p>
8553 <p>The <tt>llvm.stackprotector</tt> intrinsic requires two pointer
8554 arguments. The first argument is the value loaded from the stack
8555 guard <tt>@__stack_chk_guard</tt>. The second variable is an <tt>alloca</tt>
8556 that has enough space to hold the value of the guard.</p>
8559 <p>This intrinsic causes the prologue/epilogue inserter to force the position of
8560 the <tt>AllocaInst</tt> stack slot to be before local variables on the
8561 stack. This is to ensure that if a local variable on the stack is
8562 overwritten, it will destroy the value of the guard. When the function exits,
8563 the guard on the stack is checked against the original guard. If they are
8564 different, then the program aborts by calling the <tt>__stack_chk_fail()</tt>
8569 <!-- _______________________________________________________________________ -->
8571 <a name="int_objectsize">'<tt>llvm.objectsize</tt>' Intrinsic</a>
8578 declare i32 @llvm.objectsize.i32(i8* <object>, i1 <type>)
8579 declare i64 @llvm.objectsize.i64(i8* <object>, i1 <type>)
8583 <p>The <tt>llvm.objectsize</tt> intrinsic is designed to provide information to
8584 the optimizers to determine at compile time whether a) an operation (like
8585 memcpy) will overflow a buffer that corresponds to an object, or b) that a
8586 runtime check for overflow isn't necessary. An object in this context means
8587 an allocation of a specific class, structure, array, or other object.</p>
8590 <p>The <tt>llvm.objectsize</tt> intrinsic takes two arguments. The first
8591 argument is a pointer to or into the <tt>object</tt>. The second argument
8592 is a boolean 0 or 1. This argument determines whether you want the
8593 maximum (0) or minimum (1) bytes remaining. This needs to be a literal 0 or
8594 1, variables are not allowed.</p>
8597 <p>The <tt>llvm.objectsize</tt> intrinsic is lowered to either a constant
8598 representing the size of the object concerned, or <tt>i32/i64 -1 or 0</tt>,
8599 depending on the <tt>type</tt> argument, if the size cannot be determined at
8608 <!-- *********************************************************************** -->
8611 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
8612 src="http://jigsaw.w3.org/css-validator/images/vcss-blue" alt="Valid CSS"></a>
8613 <a href="http://validator.w3.org/check/referer"><img
8614 src="http://www.w3.org/Icons/valid-html401-blue" alt="Valid HTML 4.01"></a>
8616 <a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
8617 <a href="http://llvm.org/">The LLVM Compiler Infrastructure</a><br>
8618 Last modified: $Date$
projects / oota-llvm.git / blob