1 <!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN"
2 "http://www.w3.org/TR/html4/strict.dtd">
5 <title>LLVM Assembly Language Reference Manual</title>
6 <meta http-equiv="Content-Type" content="text/html; charset=utf-8">
7 <meta name="author" content="Chris Lattner">
8 <meta name="description"
9 content="LLVM Assembly Language Reference Manual.">
10 <link rel="stylesheet" href="llvm.css" type="text/css">
15 <div class="doc_title"> LLVM Language Reference Manual </div>
17 <li><a href="#abstract">Abstract</a></li>
18 <li><a href="#introduction">Introduction</a></li>
19 <li><a href="#identifiers">Identifiers</a></li>
20 <li><a href="#highlevel">High Level Structure</a>
22 <li><a href="#modulestructure">Module Structure</a></li>
23 <li><a href="#linkage">Linkage Types</a>
25 <li><a href="#linkage_private">'<tt>private</tt>' Linkage</a></li>
26 <li><a href="#linkage_linker_private">'<tt>linker_private</tt>' Linkage</a></li>
27 <li><a href="#linkage_internal">'<tt>internal</tt>' Linkage</a></li>
28 <li><a href="#linkage_available_externally">'<tt>available_externally</tt>' Linkage</a></li>
29 <li><a href="#linkage_linkonce">'<tt>linkonce</tt>' Linkage</a></li>
30 <li><a href="#linkage_common">'<tt>common</tt>' Linkage</a></li>
31 <li><a href="#linkage_weak">'<tt>weak</tt>' Linkage</a></li>
32 <li><a href="#linkage_appending">'<tt>appending</tt>' Linkage</a></li>
33 <li><a href="#linkage_externweak">'<tt>extern_weak</tt>' Linkage</a></li>
34 <li><a href="#linkage_linkonce_odr">'<tt>linkonce_odr</tt>' Linkage</a></li>
35 <li><a href="#linkage_weak">'<tt>weak_odr</tt>' Linkage</a></li>
36 <li><a href="#linkage_external">'<tt>externally visible</tt>' Linkage</a></li>
37 <li><a href="#linkage_dllimport">'<tt>dllimport</tt>' Linkage</a></li>
38 <li><a href="#linkage_dllexport">'<tt>dllexport</tt>' Linkage</a></li>
41 <li><a href="#callingconv">Calling Conventions</a></li>
42 <li><a href="#namedtypes">Named Types</a></li>
43 <li><a href="#globalvars">Global Variables</a></li>
44 <li><a href="#functionstructure">Functions</a></li>
45 <li><a href="#aliasstructure">Aliases</a></li>
46 <li><a href="#paramattrs">Parameter Attributes</a></li>
47 <li><a href="#fnattrs">Function Attributes</a></li>
48 <li><a href="#gc">Garbage Collector Names</a></li>
49 <li><a href="#moduleasm">Module-Level Inline Assembly</a></li>
50 <li><a href="#datalayout">Data Layout</a></li>
51 <li><a href="#pointeraliasing">Pointer Aliasing Rules</a></li>
54 <li><a href="#typesystem">Type System</a>
56 <li><a href="#t_classifications">Type Classifications</a></li>
57 <li><a href="#t_primitive">Primitive Types</a>
59 <li><a href="#t_integer">Integer Type</a></li>
60 <li><a href="#t_floating">Floating Point Types</a></li>
61 <li><a href="#t_void">Void Type</a></li>
62 <li><a href="#t_label">Label Type</a></li>
63 <li><a href="#t_metadata">Metadata Type</a></li>
66 <li><a href="#t_derived">Derived Types</a>
68 <li><a href="#t_array">Array Type</a></li>
69 <li><a href="#t_function">Function Type</a></li>
70 <li><a href="#t_pointer">Pointer Type</a></li>
71 <li><a href="#t_struct">Structure Type</a></li>
72 <li><a href="#t_pstruct">Packed Structure Type</a></li>
73 <li><a href="#t_vector">Vector Type</a></li>
74 <li><a href="#t_opaque">Opaque Type</a></li>
77 <li><a href="#t_uprefs">Type Up-references</a></li>
80 <li><a href="#constants">Constants</a>
82 <li><a href="#simpleconstants">Simple Constants</a></li>
83 <li><a href="#complexconstants">Complex Constants</a></li>
84 <li><a href="#globalconstants">Global Variable and Function Addresses</a></li>
85 <li><a href="#undefvalues">Undefined Values</a></li>
86 <li><a href="#constantexprs">Constant Expressions</a></li>
87 <li><a href="#metadata">Embedded Metadata</a></li>
90 <li><a href="#othervalues">Other Values</a>
92 <li><a href="#inlineasm">Inline Assembler Expressions</a></li>
95 <li><a href="#intrinsic_globals">Intrinsic Global Variables</a>
97 <li><a href="#intg_used">The '<tt>llvm.used</tt>' Global Variable</a></li>
98 <li><a href="#intg_compiler_used">The '<tt>llvm.compiler.used</tt>'
99 Global Variable</a></li>
100 <li><a href="#intg_global_ctors">The '<tt>llvm.global_ctors</tt>'
101 Global Variable</a></li>
102 <li><a href="#intg_global_dtors">The '<tt>llvm.global_dtors</tt>'
103 Global Variable</a></li>
106 <li><a href="#instref">Instruction Reference</a>
108 <li><a href="#terminators">Terminator Instructions</a>
110 <li><a href="#i_ret">'<tt>ret</tt>' Instruction</a></li>
111 <li><a href="#i_br">'<tt>br</tt>' Instruction</a></li>
112 <li><a href="#i_switch">'<tt>switch</tt>' Instruction</a></li>
113 <li><a href="#i_invoke">'<tt>invoke</tt>' Instruction</a></li>
114 <li><a href="#i_unwind">'<tt>unwind</tt>' Instruction</a></li>
115 <li><a href="#i_unreachable">'<tt>unreachable</tt>' Instruction</a></li>
118 <li><a href="#binaryops">Binary Operations</a>
120 <li><a href="#i_add">'<tt>add</tt>' Instruction</a></li>
121 <li><a href="#i_fadd">'<tt>fadd</tt>' Instruction</a></li>
122 <li><a href="#i_sub">'<tt>sub</tt>' Instruction</a></li>
123 <li><a href="#i_fsub">'<tt>fsub</tt>' Instruction</a></li>
124 <li><a href="#i_mul">'<tt>mul</tt>' Instruction</a></li>
125 <li><a href="#i_fmul">'<tt>fmul</tt>' Instruction</a></li>
126 <li><a href="#i_udiv">'<tt>udiv</tt>' Instruction</a></li>
127 <li><a href="#i_sdiv">'<tt>sdiv</tt>' Instruction</a></li>
128 <li><a href="#i_fdiv">'<tt>fdiv</tt>' Instruction</a></li>
129 <li><a href="#i_urem">'<tt>urem</tt>' Instruction</a></li>
130 <li><a href="#i_srem">'<tt>srem</tt>' Instruction</a></li>
131 <li><a href="#i_frem">'<tt>frem</tt>' Instruction</a></li>
134 <li><a href="#bitwiseops">Bitwise Binary Operations</a>
136 <li><a href="#i_shl">'<tt>shl</tt>' Instruction</a></li>
137 <li><a href="#i_lshr">'<tt>lshr</tt>' Instruction</a></li>
138 <li><a href="#i_ashr">'<tt>ashr</tt>' Instruction</a></li>
139 <li><a href="#i_and">'<tt>and</tt>' Instruction</a></li>
140 <li><a href="#i_or">'<tt>or</tt>' Instruction</a></li>
141 <li><a href="#i_xor">'<tt>xor</tt>' Instruction</a></li>
144 <li><a href="#vectorops">Vector Operations</a>
146 <li><a href="#i_extractelement">'<tt>extractelement</tt>' Instruction</a></li>
147 <li><a href="#i_insertelement">'<tt>insertelement</tt>' Instruction</a></li>
148 <li><a href="#i_shufflevector">'<tt>shufflevector</tt>' Instruction</a></li>
151 <li><a href="#aggregateops">Aggregate Operations</a>
153 <li><a href="#i_extractvalue">'<tt>extractvalue</tt>' Instruction</a></li>
154 <li><a href="#i_insertvalue">'<tt>insertvalue</tt>' Instruction</a></li>
157 <li><a href="#memoryops">Memory Access and Addressing Operations</a>
159 <li><a href="#i_malloc">'<tt>malloc</tt>' Instruction</a></li>
160 <li><a href="#i_free">'<tt>free</tt>' Instruction</a></li>
161 <li><a href="#i_alloca">'<tt>alloca</tt>' Instruction</a></li>
162 <li><a href="#i_load">'<tt>load</tt>' Instruction</a></li>
163 <li><a href="#i_store">'<tt>store</tt>' Instruction</a></li>
164 <li><a href="#i_getelementptr">'<tt>getelementptr</tt>' Instruction</a></li>
167 <li><a href="#convertops">Conversion Operations</a>
169 <li><a href="#i_trunc">'<tt>trunc .. to</tt>' Instruction</a></li>
170 <li><a href="#i_zext">'<tt>zext .. to</tt>' Instruction</a></li>
171 <li><a href="#i_sext">'<tt>sext .. to</tt>' Instruction</a></li>
172 <li><a href="#i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a></li>
173 <li><a href="#i_fpext">'<tt>fpext .. to</tt>' Instruction</a></li>
174 <li><a href="#i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a></li>
175 <li><a href="#i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a></li>
176 <li><a href="#i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a></li>
177 <li><a href="#i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a></li>
178 <li><a href="#i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a></li>
179 <li><a href="#i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a></li>
180 <li><a href="#i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a></li>
183 <li><a href="#otherops">Other Operations</a>
185 <li><a href="#i_icmp">'<tt>icmp</tt>' Instruction</a></li>
186 <li><a href="#i_fcmp">'<tt>fcmp</tt>' Instruction</a></li>
187 <li><a href="#i_phi">'<tt>phi</tt>' Instruction</a></li>
188 <li><a href="#i_select">'<tt>select</tt>' Instruction</a></li>
189 <li><a href="#i_call">'<tt>call</tt>' Instruction</a></li>
190 <li><a href="#i_va_arg">'<tt>va_arg</tt>' Instruction</a></li>
195 <li><a href="#intrinsics">Intrinsic Functions</a>
197 <li><a href="#int_varargs">Variable Argument Handling Intrinsics</a>
199 <li><a href="#int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a></li>
200 <li><a href="#int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a></li>
201 <li><a href="#int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a></li>
204 <li><a href="#int_gc">Accurate Garbage Collection Intrinsics</a>
206 <li><a href="#int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a></li>
207 <li><a href="#int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a></li>
208 <li><a href="#int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a></li>
211 <li><a href="#int_codegen">Code Generator Intrinsics</a>
213 <li><a href="#int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a></li>
214 <li><a href="#int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a></li>
215 <li><a href="#int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a></li>
216 <li><a href="#int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a></li>
217 <li><a href="#int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a></li>
218 <li><a href="#int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a></li>
219 <li><a href="#int_readcyclecounter"><tt>llvm.readcyclecounter</tt>' Intrinsic</a></li>
222 <li><a href="#int_libc">Standard C Library Intrinsics</a>
224 <li><a href="#int_memcpy">'<tt>llvm.memcpy.*</tt>' Intrinsic</a></li>
225 <li><a href="#int_memmove">'<tt>llvm.memmove.*</tt>' Intrinsic</a></li>
226 <li><a href="#int_memset">'<tt>llvm.memset.*</tt>' Intrinsic</a></li>
227 <li><a href="#int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a></li>
228 <li><a href="#int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a></li>
229 <li><a href="#int_sin">'<tt>llvm.sin.*</tt>' Intrinsic</a></li>
230 <li><a href="#int_cos">'<tt>llvm.cos.*</tt>' Intrinsic</a></li>
231 <li><a href="#int_pow">'<tt>llvm.pow.*</tt>' Intrinsic</a></li>
234 <li><a href="#int_manip">Bit Manipulation Intrinsics</a>
236 <li><a href="#int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a></li>
237 <li><a href="#int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic </a></li>
238 <li><a href="#int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic </a></li>
239 <li><a href="#int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic </a></li>
242 <li><a href="#int_overflow">Arithmetic with Overflow Intrinsics</a>
244 <li><a href="#int_sadd_overflow">'<tt>llvm.sadd.with.overflow.*</tt> Intrinsics</a></li>
245 <li><a href="#int_uadd_overflow">'<tt>llvm.uadd.with.overflow.*</tt> Intrinsics</a></li>
246 <li><a href="#int_ssub_overflow">'<tt>llvm.ssub.with.overflow.*</tt> Intrinsics</a></li>
247 <li><a href="#int_usub_overflow">'<tt>llvm.usub.with.overflow.*</tt> Intrinsics</a></li>
248 <li><a href="#int_smul_overflow">'<tt>llvm.smul.with.overflow.*</tt> Intrinsics</a></li>
249 <li><a href="#int_umul_overflow">'<tt>llvm.umul.with.overflow.*</tt> Intrinsics</a></li>
252 <li><a href="#int_debugger">Debugger intrinsics</a></li>
253 <li><a href="#int_eh">Exception Handling intrinsics</a></li>
254 <li><a href="#int_trampoline">Trampoline Intrinsic</a>
256 <li><a href="#int_it">'<tt>llvm.init.trampoline</tt>' Intrinsic</a></li>
259 <li><a href="#int_atomics">Atomic intrinsics</a>
261 <li><a href="#int_memory_barrier"><tt>llvm.memory_barrier</tt></a></li>
262 <li><a href="#int_atomic_cmp_swap"><tt>llvm.atomic.cmp.swap</tt></a></li>
263 <li><a href="#int_atomic_swap"><tt>llvm.atomic.swap</tt></a></li>
264 <li><a href="#int_atomic_load_add"><tt>llvm.atomic.load.add</tt></a></li>
265 <li><a href="#int_atomic_load_sub"><tt>llvm.atomic.load.sub</tt></a></li>
266 <li><a href="#int_atomic_load_and"><tt>llvm.atomic.load.and</tt></a></li>
267 <li><a href="#int_atomic_load_nand"><tt>llvm.atomic.load.nand</tt></a></li>
268 <li><a href="#int_atomic_load_or"><tt>llvm.atomic.load.or</tt></a></li>
269 <li><a href="#int_atomic_load_xor"><tt>llvm.atomic.load.xor</tt></a></li>
270 <li><a href="#int_atomic_load_max"><tt>llvm.atomic.load.max</tt></a></li>
271 <li><a href="#int_atomic_load_min"><tt>llvm.atomic.load.min</tt></a></li>
272 <li><a href="#int_atomic_load_umax"><tt>llvm.atomic.load.umax</tt></a></li>
273 <li><a href="#int_atomic_load_umin"><tt>llvm.atomic.load.umin</tt></a></li>
276 <li><a href="#int_general">General intrinsics</a>
278 <li><a href="#int_var_annotation">
279 '<tt>llvm.var.annotation</tt>' Intrinsic</a></li>
280 <li><a href="#int_annotation">
281 '<tt>llvm.annotation.*</tt>' Intrinsic</a></li>
282 <li><a href="#int_trap">
283 '<tt>llvm.trap</tt>' Intrinsic</a></li>
284 <li><a href="#int_stackprotector">
285 '<tt>llvm.stackprotector</tt>' Intrinsic</a></li>
292 <div class="doc_author">
293 <p>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a>
294 and <a href="mailto:vadve@cs.uiuc.edu">Vikram Adve</a></p>
297 <!-- *********************************************************************** -->
298 <div class="doc_section"> <a name="abstract">Abstract </a></div>
299 <!-- *********************************************************************** -->
301 <div class="doc_text">
303 <p>This document is a reference manual for the LLVM assembly language. LLVM is
304 a Static Single Assignment (SSA) based representation that provides type
305 safety, low-level operations, flexibility, and the capability of representing
306 'all' high-level languages cleanly. It is the common code representation
307 used throughout all phases of the LLVM compilation strategy.</p>
311 <!-- *********************************************************************** -->
312 <div class="doc_section"> <a name="introduction">Introduction</a> </div>
313 <!-- *********************************************************************** -->
315 <div class="doc_text">
317 <p>The LLVM code representation is designed to be used in three different forms:
318 as an in-memory compiler IR, as an on-disk bitcode representation (suitable
319 for fast loading by a Just-In-Time compiler), and as a human readable
320 assembly language representation. This allows LLVM to provide a powerful
321 intermediate representation for efficient compiler transformations and
322 analysis, while providing a natural means to debug and visualize the
323 transformations. The three different forms of LLVM are all equivalent. This
324 document describes the human readable representation and notation.</p>
326 <p>The LLVM representation aims to be light-weight and low-level while being
327 expressive, typed, and extensible at the same time. It aims to be a
328 "universal IR" of sorts, by being at a low enough level that high-level ideas
329 may be cleanly mapped to it (similar to how microprocessors are "universal
330 IR's", allowing many source languages to be mapped to them). By providing
331 type information, LLVM can be used as the target of optimizations: for
332 example, through pointer analysis, it can be proven that a C automatic
333 variable is never accessed outside of the current function... allowing it to
334 be promoted to a simple SSA value instead of a memory location.</p>
338 <!-- _______________________________________________________________________ -->
339 <div class="doc_subsubsection"> <a name="wellformed">Well-Formedness</a> </div>
341 <div class="doc_text">
343 <p>It is important to note that this document describes 'well formed' LLVM
344 assembly language. There is a difference between what the parser accepts and
345 what is considered 'well formed'. For example, the following instruction is
346 syntactically okay, but not well formed:</p>
348 <div class="doc_code">
350 %x = <a href="#i_add">add</a> i32 1, %x
354 <p>...because the definition of <tt>%x</tt> does not dominate all of its
355 uses. The LLVM infrastructure provides a verification pass that may be used
356 to verify that an LLVM module is well formed. This pass is automatically run
357 by the parser after parsing input assembly and by the optimizer before it
358 outputs bitcode. The violations pointed out by the verifier pass indicate
359 bugs in transformation passes or input to the parser.</p>
363 <!-- Describe the typesetting conventions here. -->
365 <!-- *********************************************************************** -->
366 <div class="doc_section"> <a name="identifiers">Identifiers</a> </div>
367 <!-- *********************************************************************** -->
369 <div class="doc_text">
371 <p>LLVM identifiers come in two basic types: global and local. Global
372 identifiers (functions, global variables) begin with the <tt>'@'</tt>
373 character. Local identifiers (register names, types) begin with
374 the <tt>'%'</tt> character. Additionally, there are three different formats
375 for identifiers, for different purposes:</p>
378 <li>Named values are represented as a string of characters with their prefix.
379 For example, <tt>%foo</tt>, <tt>@DivisionByZero</tt>,
380 <tt>%a.really.long.identifier</tt>. The actual regular expression used is
381 '<tt>[%@][a-zA-Z$._][a-zA-Z$._0-9]*</tt>'. Identifiers which require
382 other characters in their names can be surrounded with quotes. Special
383 characters may be escaped using <tt>"\xx"</tt> where <tt>xx</tt> is the
384 ASCII code for the character in hexadecimal. In this way, any character
385 can be used in a name value, even quotes themselves.</li>
387 <li>Unnamed values are represented as an unsigned numeric value with their
388 prefix. For example, <tt>%12</tt>, <tt>@2</tt>, <tt>%44</tt>.</li>
390 <li>Constants, which are described in a <a href="#constants">section about
391 constants</a>, below.</li>
394 <p>LLVM requires that values start with a prefix for two reasons: Compilers
395 don't need to worry about name clashes with reserved words, and the set of
396 reserved words may be expanded in the future without penalty. Additionally,
397 unnamed identifiers allow a compiler to quickly come up with a temporary
398 variable without having to avoid symbol table conflicts.</p>
400 <p>Reserved words in LLVM are very similar to reserved words in other
401 languages. There are keywords for different opcodes
402 ('<tt><a href="#i_add">add</a></tt>',
403 '<tt><a href="#i_bitcast">bitcast</a></tt>',
404 '<tt><a href="#i_ret">ret</a></tt>', etc...), for primitive type names
405 ('<tt><a href="#t_void">void</a></tt>',
406 '<tt><a href="#t_primitive">i32</a></tt>', etc...), and others. These
407 reserved words cannot conflict with variable names, because none of them
408 start with a prefix character (<tt>'%'</tt> or <tt>'@'</tt>).</p>
410 <p>Here is an example of LLVM code to multiply the integer variable
411 '<tt>%X</tt>' by 8:</p>
415 <div class="doc_code">
417 %result = <a href="#i_mul">mul</a> i32 %X, 8
421 <p>After strength reduction:</p>
423 <div class="doc_code">
425 %result = <a href="#i_shl">shl</a> i32 %X, i8 3
429 <p>And the hard way:</p>
431 <div class="doc_code">
433 <a href="#i_add">add</a> i32 %X, %X <i>; yields {i32}:%0</i>
434 <a href="#i_add">add</a> i32 %0, %0 <i>; yields {i32}:%1</i>
435 %result = <a href="#i_add">add</a> i32 %1, %1
439 <p>This last way of multiplying <tt>%X</tt> by 8 illustrates several important
440 lexical features of LLVM:</p>
443 <li>Comments are delimited with a '<tt>;</tt>' and go until the end of
446 <li>Unnamed temporaries are created when the result of a computation is not
447 assigned to a named value.</li>
449 <li>Unnamed temporaries are numbered sequentially</li>
452 <p>...and it also shows a convention that we follow in this document. When
453 demonstrating instructions, we will follow an instruction with a comment that
454 defines the type and name of value produced. Comments are shown in italic
459 <!-- *********************************************************************** -->
460 <div class="doc_section"> <a name="highlevel">High Level Structure</a> </div>
461 <!-- *********************************************************************** -->
463 <!-- ======================================================================= -->
464 <div class="doc_subsection"> <a name="modulestructure">Module Structure</a>
467 <div class="doc_text">
469 <p>LLVM programs are composed of "Module"s, each of which is a translation unit
470 of the input programs. Each module consists of functions, global variables,
471 and symbol table entries. Modules may be combined together with the LLVM
472 linker, which merges function (and global variable) definitions, resolves
473 forward declarations, and merges symbol table entries. Here is an example of
474 the "hello world" module:</p>
476 <div class="doc_code">
477 <pre><i>; Declare the string constant as a global constant...</i>
478 <a href="#identifiers">@.LC0</a> = <a href="#linkage_internal">internal</a> <a
479 href="#globalvars">constant</a> <a href="#t_array">[13 x i8]</a> c"hello world\0A\00" <i>; [13 x i8]*</i>
481 <i>; External declaration of the puts function</i>
482 <a href="#functionstructure">declare</a> i32 @puts(i8 *) <i>; i32(i8 *)* </i>
484 <i>; Definition of main function</i>
485 define i32 @main() { <i>; i32()* </i>
486 <i>; Convert [13 x i8]* to i8 *...</i>
488 href="#i_getelementptr">getelementptr</a> [13 x i8]* @.LC0, i64 0, i64 0 <i>; i8 *</i>
490 <i>; Call puts function to write out the string to stdout...</i>
492 href="#i_call">call</a> i32 @puts(i8 * %cast210) <i>; i32</i>
494 href="#i_ret">ret</a> i32 0<br>}<br>
498 <p>This example is made up of a <a href="#globalvars">global variable</a> named
499 "<tt>.LC0</tt>", an external declaration of the "<tt>puts</tt>" function, and
500 a <a href="#functionstructure">function definition</a> for
503 <p>In general, a module is made up of a list of global values, where both
504 functions and global variables are global values. Global values are
505 represented by a pointer to a memory location (in this case, a pointer to an
506 array of char, and a pointer to a function), and have one of the
507 following <a href="#linkage">linkage types</a>.</p>
511 <!-- ======================================================================= -->
512 <div class="doc_subsection">
513 <a name="linkage">Linkage Types</a>
516 <div class="doc_text">
518 <p>All Global Variables and Functions have one of the following types of
522 <dt><tt><b><a name="linkage_private">private</a></b></tt>: </dt>
523 <dd>Global values with private linkage are only directly accessible by objects
524 in the current module. In particular, linking code into a module with an
525 private global value may cause the private to be renamed as necessary to
526 avoid collisions. Because the symbol is private to the module, all
527 references can be updated. This doesn't show up in any symbol table in the
530 <dt><tt><b><a name="linkage_linker_private">linker_private</a></b></tt>: </dt>
531 <dd>Similar to private, but the symbol is passed through the assembler and
532 removed by the linker after evaluation. Note that (unlike private
533 symbols) linker_private symbols are subject to coalescing by the linker:
534 weak symbols get merged and redefinitions are rejected. However, unlike
535 normal strong symbols, they are removed by the linker from the final
536 linked image (executable or dynamic library).</dd>
538 <dt><tt><b><a name="linkage_internal">internal</a></b></tt>: </dt>
539 <dd>Similar to private, but the value shows as a local symbol
540 (<tt>STB_LOCAL</tt> in the case of ELF) in the object file. This
541 corresponds to the notion of the '<tt>static</tt>' keyword in C.</dd>
543 <dt><tt><b><a name="linkage_available_externally">available_externally</a></b></tt>: </dt>
544 <dd>Globals with "<tt>available_externally</tt>" linkage are never emitted
545 into the object file corresponding to the LLVM module. They exist to
546 allow inlining and other optimizations to take place given knowledge of
547 the definition of the global, which is known to be somewhere outside the
548 module. Globals with <tt>available_externally</tt> linkage are allowed to
549 be discarded at will, and are otherwise the same as <tt>linkonce_odr</tt>.
550 This linkage type is only allowed on definitions, not declarations.</dd>
552 <dt><tt><b><a name="linkage_linkonce">linkonce</a></b></tt>: </dt>
553 <dd>Globals with "<tt>linkonce</tt>" linkage are merged with other globals of
554 the same name when linkage occurs. This is typically used to implement
555 inline functions, templates, or other code which must be generated in each
556 translation unit that uses it. Unreferenced <tt>linkonce</tt> globals are
557 allowed to be discarded.</dd>
559 <dt><tt><b><a name="linkage_weak">weak</a></b></tt>: </dt>
560 <dd>"<tt>weak</tt>" linkage has the same merging semantics as
561 <tt>linkonce</tt> linkage, except that unreferenced globals with
562 <tt>weak</tt> linkage may not be discarded. This is used for globals that
563 are declared "weak" in C source code.</dd>
565 <dt><tt><b><a name="linkage_common">common</a></b></tt>: </dt>
566 <dd>"<tt>common</tt>" linkage is most similar to "<tt>weak</tt>" linkage, but
567 they are used for tentative definitions in C, such as "<tt>int X;</tt>" at
569 Symbols with "<tt>common</tt>" linkage are merged in the same way as
570 <tt>weak symbols</tt>, and they may not be deleted if unreferenced.
571 <tt>common</tt> symbols may not have an explicit section,
572 must have a zero initializer, and may not be marked '<a
573 href="#globalvars"><tt>constant</tt></a>'. Functions and aliases may not
574 have common linkage.</dd>
577 <dt><tt><b><a name="linkage_appending">appending</a></b></tt>: </dt>
578 <dd>"<tt>appending</tt>" linkage may only be applied to global variables of
579 pointer to array type. When two global variables with appending linkage
580 are linked together, the two global arrays are appended together. This is
581 the LLVM, typesafe, equivalent of having the system linker append together
582 "sections" with identical names when .o files are linked.</dd>
584 <dt><tt><b><a name="linkage_externweak">extern_weak</a></b></tt>: </dt>
585 <dd>The semantics of this linkage follow the ELF object file model: the symbol
586 is weak until linked, if not linked, the symbol becomes null instead of
587 being an undefined reference.</dd>
589 <dt><tt><b><a name="linkage_linkonce_odr">linkonce_odr</a></b></tt>: </dt>
590 <dt><tt><b><a name="linkage_weak_odr">weak_odr</a></b></tt>: </dt>
591 <dd>Some languages allow differing globals to be merged, such as two functions
592 with different semantics. Other languages, such as <tt>C++</tt>, ensure
593 that only equivalent globals are ever merged (the "one definition rule" -
594 "ODR"). Such languages can use the <tt>linkonce_odr</tt>
595 and <tt>weak_odr</tt> linkage types to indicate that the global will only
596 be merged with equivalent globals. These linkage types are otherwise the
597 same as their non-<tt>odr</tt> versions.</dd>
599 <dt><tt><b><a name="linkage_external">externally visible</a></b></tt>:</dt>
600 <dd>If none of the above identifiers are used, the global is externally
601 visible, meaning that it participates in linkage and can be used to
602 resolve external symbol references.</dd>
605 <p>The next two types of linkage are targeted for Microsoft Windows platform
606 only. They are designed to support importing (exporting) symbols from (to)
607 DLLs (Dynamic Link Libraries).</p>
610 <dt><tt><b><a name="linkage_dllimport">dllimport</a></b></tt>: </dt>
611 <dd>"<tt>dllimport</tt>" linkage causes the compiler to reference a function
612 or variable via a global pointer to a pointer that is set up by the DLL
613 exporting the symbol. On Microsoft Windows targets, the pointer name is
614 formed by combining <code>__imp_</code> and the function or variable
617 <dt><tt><b><a name="linkage_dllexport">dllexport</a></b></tt>: </dt>
618 <dd>"<tt>dllexport</tt>" linkage causes the compiler to provide a global
619 pointer to a pointer in a DLL, so that it can be referenced with the
620 <tt>dllimport</tt> attribute. On Microsoft Windows targets, the pointer
621 name is formed by combining <code>__imp_</code> and the function or
625 <p>For example, since the "<tt>.LC0</tt>" variable is defined to be internal, if
626 another module defined a "<tt>.LC0</tt>" variable and was linked with this
627 one, one of the two would be renamed, preventing a collision. Since
628 "<tt>main</tt>" and "<tt>puts</tt>" are external (i.e., lacking any linkage
629 declarations), they are accessible outside of the current module.</p>
631 <p>It is illegal for a function <i>declaration</i> to have any linkage type
632 other than "externally visible", <tt>dllimport</tt>
633 or <tt>extern_weak</tt>.</p>
635 <p>Aliases can have only <tt>external</tt>, <tt>internal</tt>, <tt>weak</tt>
636 or <tt>weak_odr</tt> linkages.</p>
640 <!-- ======================================================================= -->
641 <div class="doc_subsection">
642 <a name="callingconv">Calling Conventions</a>
645 <div class="doc_text">
647 <p>LLVM <a href="#functionstructure">functions</a>, <a href="#i_call">calls</a>
648 and <a href="#i_invoke">invokes</a> can all have an optional calling
649 convention specified for the call. The calling convention of any pair of
650 dynamic caller/callee must match, or the behavior of the program is
651 undefined. The following calling conventions are supported by LLVM, and more
652 may be added in the future:</p>
655 <dt><b>"<tt>ccc</tt>" - The C calling convention</b>:</dt>
656 <dd>This calling convention (the default if no other calling convention is
657 specified) matches the target C calling conventions. This calling
658 convention supports varargs function calls and tolerates some mismatch in
659 the declared prototype and implemented declaration of the function (as
662 <dt><b>"<tt>fastcc</tt>" - The fast calling convention</b>:</dt>
663 <dd>This calling convention attempts to make calls as fast as possible
664 (e.g. by passing things in registers). This calling convention allows the
665 target to use whatever tricks it wants to produce fast code for the
666 target, without having to conform to an externally specified ABI
667 (Application Binary Interface). Implementations of this convention should
668 allow arbitrary <a href="CodeGenerator.html#tailcallopt">tail call
669 optimization</a> to be supported. This calling convention does not
670 support varargs and requires the prototype of all callees to exactly match
671 the prototype of the function definition.</dd>
673 <dt><b>"<tt>coldcc</tt>" - The cold calling convention</b>:</dt>
674 <dd>This calling convention attempts to make code in the caller as efficient
675 as possible under the assumption that the call is not commonly executed.
676 As such, these calls often preserve all registers so that the call does
677 not break any live ranges in the caller side. This calling convention
678 does not support varargs and requires the prototype of all callees to
679 exactly match the prototype of the function definition.</dd>
681 <dt><b>"<tt>cc <<em>n</em>></tt>" - Numbered convention</b>:</dt>
682 <dd>Any calling convention may be specified by number, allowing
683 target-specific calling conventions to be used. Target specific calling
684 conventions start at 64.</dd>
687 <p>More calling conventions can be added/defined on an as-needed basis, to
688 support Pascal conventions or any other well-known target-independent
693 <!-- ======================================================================= -->
694 <div class="doc_subsection">
695 <a name="visibility">Visibility Styles</a>
698 <div class="doc_text">
700 <p>All Global Variables and Functions have one of the following visibility
704 <dt><b>"<tt>default</tt>" - Default style</b>:</dt>
705 <dd>On targets that use the ELF object file format, default visibility means
706 that the declaration is visible to other modules and, in shared libraries,
707 means that the declared entity may be overridden. On Darwin, default
708 visibility means that the declaration is visible to other modules. Default
709 visibility corresponds to "external linkage" in the language.</dd>
711 <dt><b>"<tt>hidden</tt>" - Hidden style</b>:</dt>
712 <dd>Two declarations of an object with hidden visibility refer to the same
713 object if they are in the same shared object. Usually, hidden visibility
714 indicates that the symbol will not be placed into the dynamic symbol
715 table, so no other module (executable or shared library) can reference it
718 <dt><b>"<tt>protected</tt>" - Protected style</b>:</dt>
719 <dd>On ELF, protected visibility indicates that the symbol will be placed in
720 the dynamic symbol table, but that references within the defining module
721 will bind to the local symbol. That is, the symbol cannot be overridden by
727 <!-- ======================================================================= -->
728 <div class="doc_subsection">
729 <a name="namedtypes">Named Types</a>
732 <div class="doc_text">
734 <p>LLVM IR allows you to specify name aliases for certain types. This can make
735 it easier to read the IR and make the IR more condensed (particularly when
736 recursive types are involved). An example of a name specification is:</p>
738 <div class="doc_code">
740 %mytype = type { %mytype*, i32 }
744 <p>You may give a name to any <a href="#typesystem">type</a> except
745 "<a href="t_void">void</a>". Type name aliases may be used anywhere a type
746 is expected with the syntax "%mytype".</p>
748 <p>Note that type names are aliases for the structural type that they indicate,
749 and that you can therefore specify multiple names for the same type. This
750 often leads to confusing behavior when dumping out a .ll file. Since LLVM IR
751 uses structural typing, the name is not part of the type. When printing out
752 LLVM IR, the printer will pick <em>one name</em> to render all types of a
753 particular shape. This means that if you have code where two different
754 source types end up having the same LLVM type, that the dumper will sometimes
755 print the "wrong" or unexpected type. This is an important design point and
756 isn't going to change.</p>
760 <!-- ======================================================================= -->
761 <div class="doc_subsection">
762 <a name="globalvars">Global Variables</a>
765 <div class="doc_text">
767 <p>Global variables define regions of memory allocated at compilation time
768 instead of run-time. Global variables may optionally be initialized, may
769 have an explicit section to be placed in, and may have an optional explicit
770 alignment specified. A variable may be defined as "thread_local", which
771 means that it will not be shared by threads (each thread will have a
772 separated copy of the variable). A variable may be defined as a global
773 "constant," which indicates that the contents of the variable
774 will <b>never</b> be modified (enabling better optimization, allowing the
775 global data to be placed in the read-only section of an executable, etc).
776 Note that variables that need runtime initialization cannot be marked
777 "constant" as there is a store to the variable.</p>
779 <p>LLVM explicitly allows <em>declarations</em> of global variables to be marked
780 constant, even if the final definition of the global is not. This capability
781 can be used to enable slightly better optimization of the program, but
782 requires the language definition to guarantee that optimizations based on the
783 'constantness' are valid for the translation units that do not include the
786 <p>As SSA values, global variables define pointer values that are in scope
787 (i.e. they dominate) all basic blocks in the program. Global variables
788 always define a pointer to their "content" type because they describe a
789 region of memory, and all memory objects in LLVM are accessed through
792 <p>A global variable may be declared to reside in a target-specific numbered
793 address space. For targets that support them, address spaces may affect how
794 optimizations are performed and/or what target instructions are used to
795 access the variable. The default address space is zero. The address space
796 qualifier must precede any other attributes.</p>
798 <p>LLVM allows an explicit section to be specified for globals. If the target
799 supports it, it will emit globals to the section specified.</p>
801 <p>An explicit alignment may be specified for a global. If not present, or if
802 the alignment is set to zero, the alignment of the global is set by the
803 target to whatever it feels convenient. If an explicit alignment is
804 specified, the global is forced to have at least that much alignment. All
805 alignments must be a power of 2.</p>
807 <p>For example, the following defines a global in a numbered address space with
808 an initializer, section, and alignment:</p>
810 <div class="doc_code">
812 @G = addrspace(5) constant float 1.0, section "foo", align 4
819 <!-- ======================================================================= -->
820 <div class="doc_subsection">
821 <a name="functionstructure">Functions</a>
824 <div class="doc_text">
826 <p>LLVM function definitions consist of the "<tt>define</tt>" keyord, an
827 optional <a href="#linkage">linkage type</a>, an optional
828 <a href="#visibility">visibility style</a>, an optional
829 <a href="#callingconv">calling convention</a>, a return type, an optional
830 <a href="#paramattrs">parameter attribute</a> for the return type, a function
831 name, a (possibly empty) argument list (each with optional
832 <a href="#paramattrs">parameter attributes</a>), optional
833 <a href="#fnattrs">function attributes</a>, an optional section, an optional
834 alignment, an optional <a href="#gc">garbage collector name</a>, an opening
835 curly brace, a list of basic blocks, and a closing curly brace.</p>
837 <p>LLVM function declarations consist of the "<tt>declare</tt>" keyword, an
838 optional <a href="#linkage">linkage type</a>, an optional
839 <a href="#visibility">visibility style</a>, an optional
840 <a href="#callingconv">calling convention</a>, a return type, an optional
841 <a href="#paramattrs">parameter attribute</a> for the return type, a function
842 name, a possibly empty list of arguments, an optional alignment, and an
843 optional <a href="#gc">garbage collector name</a>.</p>
845 <p>A function definition contains a list of basic blocks, forming the CFG
846 (Control Flow Graph) for the function. Each basic block may optionally start
847 with a label (giving the basic block a symbol table entry), contains a list
848 of instructions, and ends with a <a href="#terminators">terminator</a>
849 instruction (such as a branch or function return).</p>
851 <p>The first basic block in a function is special in two ways: it is immediately
852 executed on entrance to the function, and it is not allowed to have
853 predecessor basic blocks (i.e. there can not be any branches to the entry
854 block of a function). Because the block can have no predecessors, it also
855 cannot have any <a href="#i_phi">PHI nodes</a>.</p>
857 <p>LLVM allows an explicit section to be specified for functions. If the target
858 supports it, it will emit functions to the section specified.</p>
860 <p>An explicit alignment may be specified for a function. If not present, or if
861 the alignment is set to zero, the alignment of the function is set by the
862 target to whatever it feels convenient. If an explicit alignment is
863 specified, the function is forced to have at least that much alignment. All
864 alignments must be a power of 2.</p>
867 <div class="doc_code">
869 define [<a href="#linkage">linkage</a>] [<a href="#visibility">visibility</a>]
870 [<a href="#callingconv">cconv</a>] [<a href="#paramattrs">ret attrs</a>]
871 <ResultType> @<FunctionName> ([argument list])
872 [<a href="#fnattrs">fn Attrs</a>] [section "name"] [align N]
873 [<a href="#gc">gc</a>] { ... }
879 <!-- ======================================================================= -->
880 <div class="doc_subsection">
881 <a name="aliasstructure">Aliases</a>
884 <div class="doc_text">
886 <p>Aliases act as "second name" for the aliasee value (which can be either
887 function, global variable, another alias or bitcast of global value). Aliases
888 may have an optional <a href="#linkage">linkage type</a>, and an
889 optional <a href="#visibility">visibility style</a>.</p>
892 <div class="doc_code">
894 @<Name> = alias [Linkage] [Visibility] <AliaseeTy> @<Aliasee>
900 <!-- ======================================================================= -->
901 <div class="doc_subsection"><a name="paramattrs">Parameter Attributes</a></div>
903 <div class="doc_text">
905 <p>The return type and each parameter of a function type may have a set of
906 <i>parameter attributes</i> associated with them. Parameter attributes are
907 used to communicate additional information about the result or parameters of
908 a function. Parameter attributes are considered to be part of the function,
909 not of the function type, so functions with different parameter attributes
910 can have the same function type.</p>
912 <p>Parameter attributes are simple keywords that follow the type specified. If
913 multiple parameter attributes are needed, they are space separated. For
916 <div class="doc_code">
918 declare i32 @printf(i8* noalias nocapture, ...)
919 declare i32 @atoi(i8 zeroext)
920 declare signext i8 @returns_signed_char()
924 <p>Note that any attributes for the function result (<tt>nounwind</tt>,
925 <tt>readonly</tt>) come immediately after the argument list.</p>
927 <p>Currently, only the following parameter attributes are defined:</p>
930 <dt><tt>zeroext</tt></dt>
931 <dd>This indicates to the code generator that the parameter or return value
932 should be zero-extended to a 32-bit value by the caller (for a parameter)
933 or the callee (for a return value).</dd>
935 <dt><tt>signext</tt></dt>
936 <dd>This indicates to the code generator that the parameter or return value
937 should be sign-extended to a 32-bit value by the caller (for a parameter)
938 or the callee (for a return value).</dd>
940 <dt><tt>inreg</tt></dt>
941 <dd>This indicates that this parameter or return value should be treated in a
942 special target-dependent fashion during while emitting code for a function
943 call or return (usually, by putting it in a register as opposed to memory,
944 though some targets use it to distinguish between two different kinds of
945 registers). Use of this attribute is target-specific.</dd>
947 <dt><tt><a name="byval">byval</a></tt></dt>
948 <dd>This indicates that the pointer parameter should really be passed by value
949 to the function. The attribute implies that a hidden copy of the pointee
950 is made between the caller and the callee, so the callee is unable to
951 modify the value in the callee. This attribute is only valid on LLVM
952 pointer arguments. It is generally used to pass structs and arrays by
953 value, but is also valid on pointers to scalars. The copy is considered
954 to belong to the caller not the callee (for example,
955 <tt><a href="#readonly">readonly</a></tt> functions should not write to
956 <tt>byval</tt> parameters). This is not a valid attribute for return
957 values. The byval attribute also supports specifying an alignment with
958 the align attribute. This has a target-specific effect on the code
959 generator that usually indicates a desired alignment for the synthesized
962 <dt><tt>sret</tt></dt>
963 <dd>This indicates that the pointer parameter specifies the address of a
964 structure that is the return value of the function in the source program.
965 This pointer must be guaranteed by the caller to be valid: loads and
966 stores to the structure may be assumed by the callee to not to trap. This
967 may only be applied to the first parameter. This is not a valid attribute
968 for return values. </dd>
970 <dt><tt>noalias</tt></dt>
971 <dd>This indicates that the pointer does not alias any global or any other
972 parameter. The caller is responsible for ensuring that this is the
973 case. On a function return value, <tt>noalias</tt> additionally indicates
974 that the pointer does not alias any other pointers visible to the
975 caller. For further details, please see the discussion of the NoAlias
977 <a href="http://llvm.org/docs/AliasAnalysis.html#MustMayNo">alias
980 <dt><tt>nocapture</tt></dt>
981 <dd>This indicates that the callee does not make any copies of the pointer
982 that outlive the callee itself. This is not a valid attribute for return
985 <dt><tt>nest</tt></dt>
986 <dd>This indicates that the pointer parameter can be excised using the
987 <a href="#int_trampoline">trampoline intrinsics</a>. This is not a valid
988 attribute for return values.</dd>
993 <!-- ======================================================================= -->
994 <div class="doc_subsection">
995 <a name="gc">Garbage Collector Names</a>
998 <div class="doc_text">
1000 <p>Each function may specify a garbage collector name, which is simply a
1003 <div class="doc_code">
1005 define void @f() gc "name" { ...
1009 <p>The compiler declares the supported values of <i>name</i>. Specifying a
1010 collector which will cause the compiler to alter its output in order to
1011 support the named garbage collection algorithm.</p>
1015 <!-- ======================================================================= -->
1016 <div class="doc_subsection">
1017 <a name="fnattrs">Function Attributes</a>
1020 <div class="doc_text">
1022 <p>Function attributes are set to communicate additional information about a
1023 function. Function attributes are considered to be part of the function, not
1024 of the function type, so functions with different parameter attributes can
1025 have the same function type.</p>
1027 <p>Function attributes are simple keywords that follow the type specified. If
1028 multiple attributes are needed, they are space separated. For example:</p>
1030 <div class="doc_code">
1032 define void @f() noinline { ... }
1033 define void @f() alwaysinline { ... }
1034 define void @f() alwaysinline optsize { ... }
1035 define void @f() optsize
1040 <dt><tt>alwaysinline</tt></dt>
1041 <dd>This attribute indicates that the inliner should attempt to inline this
1042 function into callers whenever possible, ignoring any active inlining size
1043 threshold for this caller.</dd>
1045 <dt><tt>inlinehint</tt></dt>
1046 <dd>This attribute indicates that the source code contained a hint that inlining
1047 this function is desirable (such as the "inline" keyword in C/C++). It
1048 is just a hint; it imposes no requirements on the inliner.</dd>
1050 <dt><tt>noinline</tt></dt>
1051 <dd>This attribute indicates that the inliner should never inline this
1052 function in any situation. This attribute may not be used together with
1053 the <tt>alwaysinline</tt> attribute.</dd>
1055 <dt><tt>optsize</tt></dt>
1056 <dd>This attribute suggests that optimization passes and code generator passes
1057 make choices that keep the code size of this function low, and otherwise
1058 do optimizations specifically to reduce code size.</dd>
1060 <dt><tt>noreturn</tt></dt>
1061 <dd>This function attribute indicates that the function never returns
1062 normally. This produces undefined behavior at runtime if the function
1063 ever does dynamically return.</dd>
1065 <dt><tt>nounwind</tt></dt>
1066 <dd>This function attribute indicates that the function never returns with an
1067 unwind or exceptional control flow. If the function does unwind, its
1068 runtime behavior is undefined.</dd>
1070 <dt><tt>readnone</tt></dt>
1071 <dd>This attribute indicates that the function computes its result (or decides
1072 to unwind an exception) based strictly on its arguments, without
1073 dereferencing any pointer arguments or otherwise accessing any mutable
1074 state (e.g. memory, control registers, etc) visible to caller functions.
1075 It does not write through any pointer arguments
1076 (including <tt><a href="#byval">byval</a></tt> arguments) and never
1077 changes any state visible to callers. This means that it cannot unwind
1078 exceptions by calling the <tt>C++</tt> exception throwing methods, but
1079 could use the <tt>unwind</tt> instruction.</dd>
1081 <dt><tt><a name="readonly">readonly</a></tt></dt>
1082 <dd>This attribute indicates that the function does not write through any
1083 pointer arguments (including <tt><a href="#byval">byval</a></tt>
1084 arguments) or otherwise modify any state (e.g. memory, control registers,
1085 etc) visible to caller functions. It may dereference pointer arguments
1086 and read state that may be set in the caller. A readonly function always
1087 returns the same value (or unwinds an exception identically) when called
1088 with the same set of arguments and global state. It cannot unwind an
1089 exception by calling the <tt>C++</tt> exception throwing methods, but may
1090 use the <tt>unwind</tt> instruction.</dd>
1092 <dt><tt><a name="ssp">ssp</a></tt></dt>
1093 <dd>This attribute indicates that the function should emit a stack smashing
1094 protector. It is in the form of a "canary"—a random value placed on
1095 the stack before the local variables that's checked upon return from the
1096 function to see if it has been overwritten. A heuristic is used to
1097 determine if a function needs stack protectors or not.<br>
1099 If a function that has an <tt>ssp</tt> attribute is inlined into a
1100 function that doesn't have an <tt>ssp</tt> attribute, then the resulting
1101 function will have an <tt>ssp</tt> attribute.</dd>
1103 <dt><tt>sspreq</tt></dt>
1104 <dd>This attribute indicates that the function should <em>always</em> emit a
1105 stack smashing protector. This overrides
1106 the <tt><a href="#ssp">ssp</a></tt> function attribute.<br>
1108 If a function that has an <tt>sspreq</tt> attribute is inlined into a
1109 function that doesn't have an <tt>sspreq</tt> attribute or which has
1110 an <tt>ssp</tt> attribute, then the resulting function will have
1111 an <tt>sspreq</tt> attribute.</dd>
1113 <dt><tt>noredzone</tt></dt>
1114 <dd>This attribute indicates that the code generator should not use a red
1115 zone, even if the target-specific ABI normally permits it.</dd>
1117 <dt><tt>noimplicitfloat</tt></dt>
1118 <dd>This attributes disables implicit floating point instructions.</dd>
1120 <dt><tt>naked</tt></dt>
1121 <dd>This attribute disables prologue / epilogue emission for the function.
1122 This can have very system-specific consequences.</dd>
1127 <!-- ======================================================================= -->
1128 <div class="doc_subsection">
1129 <a name="moduleasm">Module-Level Inline Assembly</a>
1132 <div class="doc_text">
1134 <p>Modules may contain "module-level inline asm" blocks, which corresponds to
1135 the GCC "file scope inline asm" blocks. These blocks are internally
1136 concatenated by LLVM and treated as a single unit, but may be separated in
1137 the <tt>.ll</tt> file if desired. The syntax is very simple:</p>
1139 <div class="doc_code">
1141 module asm "inline asm code goes here"
1142 module asm "more can go here"
1146 <p>The strings can contain any character by escaping non-printable characters.
1147 The escape sequence used is simply "\xx" where "xx" is the two digit hex code
1150 <p>The inline asm code is simply printed to the machine code .s file when
1151 assembly code is generated.</p>
1155 <!-- ======================================================================= -->
1156 <div class="doc_subsection">
1157 <a name="datalayout">Data Layout</a>
1160 <div class="doc_text">
1162 <p>A module may specify a target specific data layout string that specifies how
1163 data is to be laid out in memory. The syntax for the data layout is
1166 <div class="doc_code">
1168 target datalayout = "<i>layout specification</i>"
1172 <p>The <i>layout specification</i> consists of a list of specifications
1173 separated by the minus sign character ('-'). Each specification starts with
1174 a letter and may include other information after the letter to define some
1175 aspect of the data layout. The specifications accepted are as follows:</p>
1179 <dd>Specifies that the target lays out data in big-endian form. That is, the
1180 bits with the most significance have the lowest address location.</dd>
1183 <dd>Specifies that the target lays out data in little-endian form. That is,
1184 the bits with the least significance have the lowest address
1187 <dt><tt>p:<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1188 <dd>This specifies the <i>size</i> of a pointer and its <i>abi</i> and
1189 <i>preferred</i> alignments. All sizes are in bits. Specifying
1190 the <i>pref</i> alignment is optional. If omitted, the
1191 preceding <tt>:</tt> should be omitted too.</dd>
1193 <dt><tt>i<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1194 <dd>This specifies the alignment for an integer type of a given bit
1195 <i>size</i>. The value of <i>size</i> must be in the range [1,2^23).</dd>
1197 <dt><tt>v<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1198 <dd>This specifies the alignment for a vector type of a given bit
1201 <dt><tt>f<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1202 <dd>This specifies the alignment for a floating point type of a given bit
1203 <i>size</i>. The value of <i>size</i> must be either 32 (float) or 64
1206 <dt><tt>a<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1207 <dd>This specifies the alignment for an aggregate type of a given bit
1210 <dt><tt>s<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1211 <dd>This specifies the alignment for a stack object of a given bit
1215 <p>When constructing the data layout for a given target, LLVM starts with a
1216 default set of specifications which are then (possibly) overriden by the
1217 specifications in the <tt>datalayout</tt> keyword. The default specifications
1218 are given in this list:</p>
1221 <li><tt>E</tt> - big endian</li>
1222 <li><tt>p:32:64:64</tt> - 32-bit pointers with 64-bit alignment</li>
1223 <li><tt>i1:8:8</tt> - i1 is 8-bit (byte) aligned</li>
1224 <li><tt>i8:8:8</tt> - i8 is 8-bit (byte) aligned</li>
1225 <li><tt>i16:16:16</tt> - i16 is 16-bit aligned</li>
1226 <li><tt>i32:32:32</tt> - i32 is 32-bit aligned</li>
1227 <li><tt>i64:32:64</tt> - i64 has ABI alignment of 32-bits but preferred
1228 alignment of 64-bits</li>
1229 <li><tt>f32:32:32</tt> - float is 32-bit aligned</li>
1230 <li><tt>f64:64:64</tt> - double is 64-bit aligned</li>
1231 <li><tt>v64:64:64</tt> - 64-bit vector is 64-bit aligned</li>
1232 <li><tt>v128:128:128</tt> - 128-bit vector is 128-bit aligned</li>
1233 <li><tt>a0:0:1</tt> - aggregates are 8-bit aligned</li>
1234 <li><tt>s0:64:64</tt> - stack objects are 64-bit aligned</li>
1237 <p>When LLVM is determining the alignment for a given type, it uses the
1238 following rules:</p>
1241 <li>If the type sought is an exact match for one of the specifications, that
1242 specification is used.</li>
1244 <li>If no match is found, and the type sought is an integer type, then the
1245 smallest integer type that is larger than the bitwidth of the sought type
1246 is used. If none of the specifications are larger than the bitwidth then
1247 the the largest integer type is used. For example, given the default
1248 specifications above, the i7 type will use the alignment of i8 (next
1249 largest) while both i65 and i256 will use the alignment of i64 (largest
1252 <li>If no match is found, and the type sought is a vector type, then the
1253 largest vector type that is smaller than the sought vector type will be
1254 used as a fall back. This happens because <128 x double> can be
1255 implemented in terms of 64 <2 x double>, for example.</li>
1260 <!-- ======================================================================= -->
1261 <div class="doc_subsection">
1262 <a name="pointeraliasing">Pointer Aliasing Rules</a>
1265 <div class="doc_text">
1267 <p>Any memory access must be done through a pointer value associated
1268 with an address range of the memory access, otherwise the behavior
1269 is undefined. Pointer values are associated with address ranges
1270 according to the following rules:</p>
1273 <li>A pointer value formed from a
1274 <tt><a href="#i_getelementptr">getelementptr</a></tt> instruction
1275 is associated with the addresses associated with the first operand
1276 of the <tt>getelementptr</tt>.</li>
1277 <li>An address of a global variable is associated with the address
1278 range of the variable's storage.</li>
1279 <li>The result value of an allocation instruction is associated with
1280 the address range of the allocated storage.</li>
1281 <li>A null pointer in the default address-space is associated with
1283 <li>A pointer value formed by an
1284 <tt><a href="#i_inttoptr">inttoptr</a></tt> is associated with all
1285 address ranges of all pointer values that contribute (directly or
1286 indirectly) to the computation of the pointer's value.</li>
1287 <li>The result value of a
1288 <tt><a href="#i_bitcast">bitcast</a></tt> is associated with all
1289 addresses associated with the operand of the <tt>bitcast</tt>.</li>
1290 <li>An integer constant other than zero or a pointer value returned
1291 from a function not defined within LLVM may be associated with address
1292 ranges allocated through mechanisms other than those provided by
1293 LLVM. Such ranges shall not overlap with any ranges of addresses
1294 allocated by mechanisms provided by LLVM.</li>
1297 <p>LLVM IR does not associate types with memory. The result type of a
1298 <tt><a href="#i_load">load</a></tt> merely indicates the size and
1299 alignment of the memory from which to load, as well as the
1300 interpretation of the value. The first operand of a
1301 <tt><a href="#i_store">store</a></tt> similarly only indicates the size
1302 and alignment of the store.</p>
1304 <p>Consequently, type-based alias analysis, aka TBAA, aka
1305 <tt>-fstrict-aliasing</tt>, is not applicable to general unadorned
1306 LLVM IR. <a href="#metadata">Metadata</a> may be used to encode
1307 additional information which specialized optimization passes may use
1308 to implement type-based alias analysis.</p>
1312 <!-- *********************************************************************** -->
1313 <div class="doc_section"> <a name="typesystem">Type System</a> </div>
1314 <!-- *********************************************************************** -->
1316 <div class="doc_text">
1318 <p>The LLVM type system is one of the most important features of the
1319 intermediate representation. Being typed enables a number of optimizations
1320 to be performed on the intermediate representation directly, without having
1321 to do extra analyses on the side before the transformation. A strong type
1322 system makes it easier to read the generated code and enables novel analyses
1323 and transformations that are not feasible to perform on normal three address
1324 code representations.</p>
1328 <!-- ======================================================================= -->
1329 <div class="doc_subsection"> <a name="t_classifications">Type
1330 Classifications</a> </div>
1332 <div class="doc_text">
1334 <p>The types fall into a few useful classifications:</p>
1336 <table border="1" cellspacing="0" cellpadding="4">
1338 <tr><th>Classification</th><th>Types</th></tr>
1340 <td><a href="#t_integer">integer</a></td>
1341 <td><tt>i1, i2, i3, ... i8, ... i16, ... i32, ... i64, ... </tt></td>
1344 <td><a href="#t_floating">floating point</a></td>
1345 <td><tt>float, double, x86_fp80, fp128, ppc_fp128</tt></td>
1348 <td><a name="t_firstclass">first class</a></td>
1349 <td><a href="#t_integer">integer</a>,
1350 <a href="#t_floating">floating point</a>,
1351 <a href="#t_pointer">pointer</a>,
1352 <a href="#t_vector">vector</a>,
1353 <a href="#t_struct">structure</a>,
1354 <a href="#t_array">array</a>,
1355 <a href="#t_label">label</a>,
1356 <a href="#t_metadata">metadata</a>.
1360 <td><a href="#t_primitive">primitive</a></td>
1361 <td><a href="#t_label">label</a>,
1362 <a href="#t_void">void</a>,
1363 <a href="#t_floating">floating point</a>,
1364 <a href="#t_metadata">metadata</a>.</td>
1367 <td><a href="#t_derived">derived</a></td>
1368 <td><a href="#t_integer">integer</a>,
1369 <a href="#t_array">array</a>,
1370 <a href="#t_function">function</a>,
1371 <a href="#t_pointer">pointer</a>,
1372 <a href="#t_struct">structure</a>,
1373 <a href="#t_pstruct">packed structure</a>,
1374 <a href="#t_vector">vector</a>,
1375 <a href="#t_opaque">opaque</a>.
1381 <p>The <a href="#t_firstclass">first class</a> types are perhaps the most
1382 important. Values of these types are the only ones which can be produced by
1387 <!-- ======================================================================= -->
1388 <div class="doc_subsection"> <a name="t_primitive">Primitive Types</a> </div>
1390 <div class="doc_text">
1392 <p>The primitive types are the fundamental building blocks of the LLVM
1397 <!-- _______________________________________________________________________ -->
1398 <div class="doc_subsubsection"> <a name="t_integer">Integer Type</a> </div>
1400 <div class="doc_text">
1403 <p>The integer type is a very simple type that simply specifies an arbitrary
1404 bit width for the integer type desired. Any bit width from 1 bit to
1405 2<sup>23</sup>-1 (about 8 million) can be specified.</p>
1412 <p>The number of bits the integer will occupy is specified by the <tt>N</tt>
1416 <table class="layout">
1418 <td class="left"><tt>i1</tt></td>
1419 <td class="left">a single-bit integer.</td>
1422 <td class="left"><tt>i32</tt></td>
1423 <td class="left">a 32-bit integer.</td>
1426 <td class="left"><tt>i1942652</tt></td>
1427 <td class="left">a really big integer of over 1 million bits.</td>
1431 <p>Note that the code generator does not yet support large integer types to be
1432 used as function return types. The specific limit on how large a return type
1433 the code generator can currently handle is target-dependent; currently it's
1434 often 64 bits for 32-bit targets and 128 bits for 64-bit targets.</p>
1438 <!-- _______________________________________________________________________ -->
1439 <div class="doc_subsubsection"> <a name="t_floating">Floating Point Types</a> </div>
1441 <div class="doc_text">
1445 <tr><th>Type</th><th>Description</th></tr>
1446 <tr><td><tt>float</tt></td><td>32-bit floating point value</td></tr>
1447 <tr><td><tt>double</tt></td><td>64-bit floating point value</td></tr>
1448 <tr><td><tt>fp128</tt></td><td>128-bit floating point value (112-bit mantissa)</td></tr>
1449 <tr><td><tt>x86_fp80</tt></td><td>80-bit floating point value (X87)</td></tr>
1450 <tr><td><tt>ppc_fp128</tt></td><td>128-bit floating point value (two 64-bits)</td></tr>
1456 <!-- _______________________________________________________________________ -->
1457 <div class="doc_subsubsection"> <a name="t_void">Void Type</a> </div>
1459 <div class="doc_text">
1462 <p>The void type does not represent any value and has no size.</p>
1471 <!-- _______________________________________________________________________ -->
1472 <div class="doc_subsubsection"> <a name="t_label">Label Type</a> </div>
1474 <div class="doc_text">
1477 <p>The label type represents code labels.</p>
1486 <!-- _______________________________________________________________________ -->
1487 <div class="doc_subsubsection"> <a name="t_metadata">Metadata Type</a> </div>
1489 <div class="doc_text">
1492 <p>The metadata type represents embedded metadata. No derived types may be
1493 created from metadata except for <a href="#t_function">function</a>
1504 <!-- ======================================================================= -->
1505 <div class="doc_subsection"> <a name="t_derived">Derived Types</a> </div>
1507 <div class="doc_text">
1509 <p>The real power in LLVM comes from the derived types in the system. This is
1510 what allows a programmer to represent arrays, functions, pointers, and other
1511 useful types. Each of these types contain one or more element types which
1512 may be a primitive type, or another derived type. For example, it is
1513 possible to have a two dimensional array, using an array as the element type
1514 of another array.</p>
1518 <!-- _______________________________________________________________________ -->
1519 <div class="doc_subsubsection"> <a name="t_array">Array Type</a> </div>
1521 <div class="doc_text">
1524 <p>The array type is a very simple derived type that arranges elements
1525 sequentially in memory. The array type requires a size (number of elements)
1526 and an underlying data type.</p>
1530 [<# elements> x <elementtype>]
1533 <p>The number of elements is a constant integer value; <tt>elementtype</tt> may
1534 be any type with a size.</p>
1537 <table class="layout">
1539 <td class="left"><tt>[40 x i32]</tt></td>
1540 <td class="left">Array of 40 32-bit integer values.</td>
1543 <td class="left"><tt>[41 x i32]</tt></td>
1544 <td class="left">Array of 41 32-bit integer values.</td>
1547 <td class="left"><tt>[4 x i8]</tt></td>
1548 <td class="left">Array of 4 8-bit integer values.</td>
1551 <p>Here are some examples of multidimensional arrays:</p>
1552 <table class="layout">
1554 <td class="left"><tt>[3 x [4 x i32]]</tt></td>
1555 <td class="left">3x4 array of 32-bit integer values.</td>
1558 <td class="left"><tt>[12 x [10 x float]]</tt></td>
1559 <td class="left">12x10 array of single precision floating point values.</td>
1562 <td class="left"><tt>[2 x [3 x [4 x i16]]]</tt></td>
1563 <td class="left">2x3x4 array of 16-bit integer values.</td>
1567 <p>Note that 'variable sized arrays' can be implemented in LLVM with a zero
1568 length array. Normally, accesses past the end of an array are undefined in
1569 LLVM (e.g. it is illegal to access the 5th element of a 3 element array). As
1570 a special case, however, zero length arrays are recognized to be variable
1571 length. This allows implementation of 'pascal style arrays' with the LLVM
1572 type "<tt>{ i32, [0 x float]}</tt>", for example.</p>
1574 <p>Note that the code generator does not yet support large aggregate types to be
1575 used as function return types. The specific limit on how large an aggregate
1576 return type the code generator can currently handle is target-dependent, and
1577 also dependent on the aggregate element types.</p>
1581 <!-- _______________________________________________________________________ -->
1582 <div class="doc_subsubsection"> <a name="t_function">Function Type</a> </div>
1584 <div class="doc_text">
1587 <p>The function type can be thought of as a function signature. It consists of
1588 a return type and a list of formal parameter types. The return type of a
1589 function type is a scalar type, a void type, or a struct type. If the return
1590 type is a struct type then all struct elements must be of first class types,
1591 and the struct must have at least one element.</p>
1595 <returntype> (<parameter list>)
1598 <p>...where '<tt><parameter list></tt>' is a comma-separated list of type
1599 specifiers. Optionally, the parameter list may include a type <tt>...</tt>,
1600 which indicates that the function takes a variable number of arguments.
1601 Variable argument functions can access their arguments with
1602 the <a href="#int_varargs">variable argument handling intrinsic</a>
1603 functions. '<tt><returntype></tt>' is a any type except
1604 <a href="#t_label">label</a>.</p>
1607 <table class="layout">
1609 <td class="left"><tt>i32 (i32)</tt></td>
1610 <td class="left">function taking an <tt>i32</tt>, returning an <tt>i32</tt>
1612 </tr><tr class="layout">
1613 <td class="left"><tt>float (i16 signext, i32 *) *
1615 <td class="left"><a href="#t_pointer">Pointer</a> to a function that takes
1616 an <tt>i16</tt> that should be sign extended and a
1617 <a href="#t_pointer">pointer</a> to <tt>i32</tt>, returning
1620 </tr><tr class="layout">
1621 <td class="left"><tt>i32 (i8*, ...)</tt></td>
1622 <td class="left">A vararg function that takes at least one
1623 <a href="#t_pointer">pointer</a> to <tt>i8 </tt> (char in C),
1624 which returns an integer. This is the signature for <tt>printf</tt> in
1627 </tr><tr class="layout">
1628 <td class="left"><tt>{i32, i32} (i32)</tt></td>
1629 <td class="left">A function taking an <tt>i32</tt>, returning a
1630 <a href="#t_struct">structure</a> containing two <tt>i32</tt> values
1637 <!-- _______________________________________________________________________ -->
1638 <div class="doc_subsubsection"> <a name="t_struct">Structure Type</a> </div>
1640 <div class="doc_text">
1643 <p>The structure type is used to represent a collection of data members together
1644 in memory. The packing of the field types is defined to match the ABI of the
1645 underlying processor. The elements of a structure may be any type that has a
1648 <p>Structures are accessed using '<tt><a href="#i_load">load</a></tt> and
1649 '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a field with
1650 the '<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.</p>
1654 { <type list> }
1658 <table class="layout">
1660 <td class="left"><tt>{ i32, i32, i32 }</tt></td>
1661 <td class="left">A triple of three <tt>i32</tt> values</td>
1662 </tr><tr class="layout">
1663 <td class="left"><tt>{ float, i32 (i32) * }</tt></td>
1664 <td class="left">A pair, where the first element is a <tt>float</tt> and the
1665 second element is a <a href="#t_pointer">pointer</a> to a
1666 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
1667 an <tt>i32</tt>.</td>
1671 <p>Note that the code generator does not yet support large aggregate types to be
1672 used as function return types. The specific limit on how large an aggregate
1673 return type the code generator can currently handle is target-dependent, and
1674 also dependent on the aggregate element types.</p>
1678 <!-- _______________________________________________________________________ -->
1679 <div class="doc_subsubsection"> <a name="t_pstruct">Packed Structure Type</a>
1682 <div class="doc_text">
1685 <p>The packed structure type is used to represent a collection of data members
1686 together in memory. There is no padding between fields. Further, the
1687 alignment of a packed structure is 1 byte. The elements of a packed
1688 structure may be any type that has a size.</p>
1690 <p>Structures are accessed using '<tt><a href="#i_load">load</a></tt> and
1691 '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a field with
1692 the '<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.</p>
1696 < { <type list> } >
1700 <table class="layout">
1702 <td class="left"><tt>< { i32, i32, i32 } ></tt></td>
1703 <td class="left">A triple of three <tt>i32</tt> values</td>
1704 </tr><tr class="layout">
1706 <tt>< { float, i32 (i32)* } ></tt></td>
1707 <td class="left">A pair, where the first element is a <tt>float</tt> and the
1708 second element is a <a href="#t_pointer">pointer</a> to a
1709 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
1710 an <tt>i32</tt>.</td>
1716 <!-- _______________________________________________________________________ -->
1717 <div class="doc_subsubsection"> <a name="t_pointer">Pointer Type</a> </div>
1719 <div class="doc_text">
1722 <p>As in many languages, the pointer type represents a pointer or reference to
1723 another object, which must live in memory. Pointer types may have an optional
1724 address space attribute defining the target-specific numbered address space
1725 where the pointed-to object resides. The default address space is zero.</p>
1727 <p>Note that LLVM does not permit pointers to void (<tt>void*</tt>) nor does it
1728 permit pointers to labels (<tt>label*</tt>). Use <tt>i8*</tt> instead.</p>
1736 <table class="layout">
1738 <td class="left"><tt>[4 x i32]*</tt></td>
1739 <td class="left">A <a href="#t_pointer">pointer</a> to <a
1740 href="#t_array">array</a> of four <tt>i32</tt> values.</td>
1743 <td class="left"><tt>i32 (i32 *) *</tt></td>
1744 <td class="left"> A <a href="#t_pointer">pointer</a> to a <a
1745 href="#t_function">function</a> that takes an <tt>i32*</tt>, returning an
1749 <td class="left"><tt>i32 addrspace(5)*</tt></td>
1750 <td class="left">A <a href="#t_pointer">pointer</a> to an <tt>i32</tt> value
1751 that resides in address space #5.</td>
1757 <!-- _______________________________________________________________________ -->
1758 <div class="doc_subsubsection"> <a name="t_vector">Vector Type</a> </div>
1760 <div class="doc_text">
1763 <p>A vector type is a simple derived type that represents a vector of elements.
1764 Vector types are used when multiple primitive data are operated in parallel
1765 using a single instruction (SIMD). A vector type requires a size (number of
1766 elements) and an underlying primitive data type. Vectors must have a power
1767 of two length (1, 2, 4, 8, 16 ...). Vector types are considered
1768 <a href="#t_firstclass">first class</a>.</p>
1772 < <# elements> x <elementtype> >
1775 <p>The number of elements is a constant integer value; elementtype may be any
1776 integer or floating point type.</p>
1779 <table class="layout">
1781 <td class="left"><tt><4 x i32></tt></td>
1782 <td class="left">Vector of 4 32-bit integer values.</td>
1785 <td class="left"><tt><8 x float></tt></td>
1786 <td class="left">Vector of 8 32-bit floating-point values.</td>
1789 <td class="left"><tt><2 x i64></tt></td>
1790 <td class="left">Vector of 2 64-bit integer values.</td>
1794 <p>Note that the code generator does not yet support large vector types to be
1795 used as function return types. The specific limit on how large a vector
1796 return type codegen can currently handle is target-dependent; currently it's
1797 often a few times longer than a hardware vector register.</p>
1801 <!-- _______________________________________________________________________ -->
1802 <div class="doc_subsubsection"> <a name="t_opaque">Opaque Type</a> </div>
1803 <div class="doc_text">
1806 <p>Opaque types are used to represent unknown types in the system. This
1807 corresponds (for example) to the C notion of a forward declared structure
1808 type. In LLVM, opaque types can eventually be resolved to any type (not just
1809 a structure type).</p>
1817 <table class="layout">
1819 <td class="left"><tt>opaque</tt></td>
1820 <td class="left">An opaque type.</td>
1826 <!-- ======================================================================= -->
1827 <div class="doc_subsection">
1828 <a name="t_uprefs">Type Up-references</a>
1831 <div class="doc_text">
1834 <p>An "up reference" allows you to refer to a lexically enclosing type without
1835 requiring it to have a name. For instance, a structure declaration may
1836 contain a pointer to any of the types it is lexically a member of. Example
1837 of up references (with their equivalent as named type declarations)
1841 { \2 * } %x = type { %x* }
1842 { \2 }* %y = type { %y }*
1846 <p>An up reference is needed by the asmprinter for printing out cyclic types
1847 when there is no declared name for a type in the cycle. Because the
1848 asmprinter does not want to print out an infinite type string, it needs a
1849 syntax to handle recursive types that have no names (all names are optional
1857 <p>The level is the count of the lexical type that is being referred to.</p>
1860 <table class="layout">
1862 <td class="left"><tt>\1*</tt></td>
1863 <td class="left">Self-referential pointer.</td>
1866 <td class="left"><tt>{ { \3*, i8 }, i32 }</tt></td>
1867 <td class="left">Recursive structure where the upref refers to the out-most
1874 <!-- *********************************************************************** -->
1875 <div class="doc_section"> <a name="constants">Constants</a> </div>
1876 <!-- *********************************************************************** -->
1878 <div class="doc_text">
1880 <p>LLVM has several different basic types of constants. This section describes
1881 them all and their syntax.</p>
1885 <!-- ======================================================================= -->
1886 <div class="doc_subsection"><a name="simpleconstants">Simple Constants</a></div>
1888 <div class="doc_text">
1891 <dt><b>Boolean constants</b></dt>
1892 <dd>The two strings '<tt>true</tt>' and '<tt>false</tt>' are both valid
1893 constants of the <tt><a href="#t_integer">i1</a></tt> type.</dd>
1895 <dt><b>Integer constants</b></dt>
1896 <dd>Standard integers (such as '4') are constants of
1897 the <a href="#t_integer">integer</a> type. Negative numbers may be used
1898 with integer types.</dd>
1900 <dt><b>Floating point constants</b></dt>
1901 <dd>Floating point constants use standard decimal notation (e.g. 123.421),
1902 exponential notation (e.g. 1.23421e+2), or a more precise hexadecimal
1903 notation (see below). The assembler requires the exact decimal value of a
1904 floating-point constant. For example, the assembler accepts 1.25 but
1905 rejects 1.3 because 1.3 is a repeating decimal in binary. Floating point
1906 constants must have a <a href="#t_floating">floating point</a> type. </dd>
1908 <dt><b>Null pointer constants</b></dt>
1909 <dd>The identifier '<tt>null</tt>' is recognized as a null pointer constant
1910 and must be of <a href="#t_pointer">pointer type</a>.</dd>
1913 <p>The one non-intuitive notation for constants is the hexadecimal form of
1914 floating point constants. For example, the form '<tt>double
1915 0x432ff973cafa8000</tt>' is equivalent to (but harder to read than)
1916 '<tt>double 4.5e+15</tt>'. The only time hexadecimal floating point
1917 constants are required (and the only time that they are generated by the
1918 disassembler) is when a floating point constant must be emitted but it cannot
1919 be represented as a decimal floating point number in a reasonable number of
1920 digits. For example, NaN's, infinities, and other special values are
1921 represented in their IEEE hexadecimal format so that assembly and disassembly
1922 do not cause any bits to change in the constants.</p>
1924 <p>When using the hexadecimal form, constants of types float and double are
1925 represented using the 16-digit form shown above (which matches the IEEE754
1926 representation for double); float values must, however, be exactly
1927 representable as IEE754 single precision. Hexadecimal format is always used
1928 for long double, and there are three forms of long double. The 80-bit format
1929 used by x86 is represented as <tt>0xK</tt> followed by 20 hexadecimal digits.
1930 The 128-bit format used by PowerPC (two adjacent doubles) is represented
1931 by <tt>0xM</tt> followed by 32 hexadecimal digits. The IEEE 128-bit format
1932 is represented by <tt>0xL</tt> followed by 32 hexadecimal digits; no
1933 currently supported target uses this format. Long doubles will only work if
1934 they match the long double format on your target. All hexadecimal formats
1935 are big-endian (sign bit at the left).</p>
1939 <!-- ======================================================================= -->
1940 <div class="doc_subsection">
1941 <a name="aggregateconstants"></a> <!-- old anchor -->
1942 <a name="complexconstants">Complex Constants</a>
1945 <div class="doc_text">
1947 <p>Complex constants are a (potentially recursive) combination of simple
1948 constants and smaller complex constants.</p>
1951 <dt><b>Structure constants</b></dt>
1952 <dd>Structure constants are represented with notation similar to structure
1953 type definitions (a comma separated list of elements, surrounded by braces
1954 (<tt>{}</tt>)). For example: "<tt>{ i32 4, float 17.0, i32* @G }</tt>",
1955 where "<tt>@G</tt>" is declared as "<tt>@G = external global i32</tt>".
1956 Structure constants must have <a href="#t_struct">structure type</a>, and
1957 the number and types of elements must match those specified by the
1960 <dt><b>Array constants</b></dt>
1961 <dd>Array constants are represented with notation similar to array type
1962 definitions (a comma separated list of elements, surrounded by square
1963 brackets (<tt>[]</tt>)). For example: "<tt>[ i32 42, i32 11, i32 74
1964 ]</tt>". Array constants must have <a href="#t_array">array type</a>, and
1965 the number and types of elements must match those specified by the
1968 <dt><b>Vector constants</b></dt>
1969 <dd>Vector constants are represented with notation similar to vector type
1970 definitions (a comma separated list of elements, surrounded by
1971 less-than/greater-than's (<tt><></tt>)). For example: "<tt>< i32
1972 42, i32 11, i32 74, i32 100 ></tt>". Vector constants must
1973 have <a href="#t_vector">vector type</a>, and the number and types of
1974 elements must match those specified by the type.</dd>
1976 <dt><b>Zero initialization</b></dt>
1977 <dd>The string '<tt>zeroinitializer</tt>' can be used to zero initialize a
1978 value to zero of <em>any</em> type, including scalar and aggregate types.
1979 This is often used to avoid having to print large zero initializers
1980 (e.g. for large arrays) and is always exactly equivalent to using explicit
1981 zero initializers.</dd>
1983 <dt><b>Metadata node</b></dt>
1984 <dd>A metadata node is a structure-like constant with
1985 <a href="#t_metadata">metadata type</a>. For example: "<tt>metadata !{
1986 i32 0, metadata !"test" }</tt>". Unlike other constants that are meant to
1987 be interpreted as part of the instruction stream, metadata is a place to
1988 attach additional information such as debug info.</dd>
1993 <!-- ======================================================================= -->
1994 <div class="doc_subsection">
1995 <a name="globalconstants">Global Variable and Function Addresses</a>
1998 <div class="doc_text">
2000 <p>The addresses of <a href="#globalvars">global variables</a>
2001 and <a href="#functionstructure">functions</a> are always implicitly valid
2002 (link-time) constants. These constants are explicitly referenced when
2003 the <a href="#identifiers">identifier for the global</a> is used and always
2004 have <a href="#t_pointer">pointer</a> type. For example, the following is a
2005 legal LLVM file:</p>
2007 <div class="doc_code">
2011 @Z = global [2 x i32*] [ i32* @X, i32* @Y ]
2017 <!-- ======================================================================= -->
2018 <div class="doc_subsection"><a name="undefvalues">Undefined Values</a></div>
2019 <div class="doc_text">
2021 <p>The string '<tt>undef</tt>' can be used anywhere a constant is expected, and
2022 indicates that the user of the value may recieve an unspecified bit-pattern.
2023 Undefined values may be of any type (other than label or void) and be used
2024 anywhere a constant is permitted.</p>
2026 <p>Undefined values are useful because they indicate to the compiler that the
2027 program is well defined no matter what value is used. This gives the
2028 compiler more freedom to optimize. Here are some examples of (potentially
2029 surprising) transformations that are valid (in pseudo IR):</p>
2032 <div class="doc_code">
2044 <p>This is safe because all of the output bits are affected by the undef bits.
2045 Any output bit can have a zero or one depending on the input bits.</p>
2047 <div class="doc_code">
2060 <p>These logical operations have bits that are not always affected by the input.
2061 For example, if "%X" has a zero bit, then the output of the 'and' operation will
2062 always be a zero, no matter what the corresponding bit from the undef is. As
2063 such, it is unsafe to optimize or assume that the result of the and is undef.
2064 However, it is safe to assume that all bits of the undef could be 0, and
2065 optimize the and to 0. Likewise, it is safe to assume that all the bits of
2066 the undef operand to the or could be set, allowing the or to be folded to
2069 <div class="doc_code">
2071 %A = select undef, %X, %Y
2072 %B = select undef, 42, %Y
2073 %C = select %X, %Y, undef
2085 <p>This set of examples show that undefined select (and conditional branch)
2086 conditions can go "either way" but they have to come from one of the two
2087 operands. In the %A example, if %X and %Y were both known to have a clear low
2088 bit, then %A would have to have a cleared low bit. However, in the %C example,
2089 the optimizer is allowed to assume that the undef operand could be the same as
2090 %Y, allowing the whole select to be eliminated.</p>
2093 <div class="doc_code">
2095 %A = xor undef, undef
2114 <p>This example points out that two undef operands are not necessarily the same.
2115 This can be surprising to people (and also matches C semantics) where they
2116 assume that "X^X" is always zero, even if X is undef. This isn't true for a
2117 number of reasons, but the short answer is that an undef "variable" can
2118 arbitrarily change its value over its "live range". This is true because the
2119 "variable" doesn't actually <em>have a live range</em>. Instead, the value is
2120 logically read from arbitrary registers that happen to be around when needed,
2121 so the value is not neccesarily consistent over time. In fact, %A and %C need
2122 to have the same semantics or the core LLVM "replace all uses with" concept
2125 <div class="doc_code">
2135 <p>These examples show the crucial difference between an <em>undefined
2136 value</em> and <em>undefined behavior</em>. An undefined value (like undef) is
2137 allowed to have an arbitrary bit-pattern. This means that the %A operation
2138 can be constant folded to undef because the undef could be an SNaN, and fdiv is
2139 not (currently) defined on SNaN's. However, in the second example, we can make
2140 a more aggressive assumption: because the undef is allowed to be an arbitrary
2141 value, we are allowed to assume that it could be zero. Since a divide by zero
2142 has <em>undefined behavior</em>, we are allowed to assume that the operation
2143 does not execute at all. This allows us to delete the divide and all code after
2144 it: since the undefined operation "can't happen", the optimizer can assume that
2145 it occurs in dead code.
2148 <div class="doc_code">
2150 a: store undef -> %X
2151 b: store %X -> undef
2158 <p>These examples reiterate the fdiv example: a store "of" an undefined value
2159 can be assumed to not have any effect: we can assume that the value is
2160 overwritten with bits that happen to match what was already there. However, a
2161 store "to" an undefined location could clobber arbitrary memory, therefore, it
2162 has undefined behavior.</p>
2166 <!-- ======================================================================= -->
2167 <div class="doc_subsection"><a name="constantexprs">Constant Expressions</a>
2170 <div class="doc_text">
2172 <p>Constant expressions are used to allow expressions involving other constants
2173 to be used as constants. Constant expressions may be of
2174 any <a href="#t_firstclass">first class</a> type and may involve any LLVM
2175 operation that does not have side effects (e.g. load and call are not
2176 supported). The following is the syntax for constant expressions:</p>
2179 <dt><b><tt>trunc ( CST to TYPE )</tt></b></dt>
2180 <dd>Truncate a constant to another type. The bit size of CST must be larger
2181 than the bit size of TYPE. Both types must be integers.</dd>
2183 <dt><b><tt>zext ( CST to TYPE )</tt></b></dt>
2184 <dd>Zero extend a constant to another type. The bit size of CST must be
2185 smaller or equal to the bit size of TYPE. Both types must be
2188 <dt><b><tt>sext ( CST to TYPE )</tt></b></dt>
2189 <dd>Sign extend a constant to another type. The bit size of CST must be
2190 smaller or equal to the bit size of TYPE. Both types must be
2193 <dt><b><tt>fptrunc ( CST to TYPE )</tt></b></dt>
2194 <dd>Truncate a floating point constant to another floating point type. The
2195 size of CST must be larger than the size of TYPE. Both types must be
2196 floating point.</dd>
2198 <dt><b><tt>fpext ( CST to TYPE )</tt></b></dt>
2199 <dd>Floating point extend a constant to another type. The size of CST must be
2200 smaller or equal to the size of TYPE. Both types must be floating
2203 <dt><b><tt>fptoui ( CST to TYPE )</tt></b></dt>
2204 <dd>Convert a floating point constant to the corresponding unsigned integer
2205 constant. TYPE must be a scalar or vector integer type. CST must be of
2206 scalar or vector floating point type. Both CST and TYPE must be scalars,
2207 or vectors of the same number of elements. If the value won't fit in the
2208 integer type, the results are undefined.</dd>
2210 <dt><b><tt>fptosi ( CST to TYPE )</tt></b></dt>
2211 <dd>Convert a floating point constant to the corresponding signed integer
2212 constant. TYPE must be a scalar or vector integer type. CST must be of
2213 scalar or vector floating point type. Both CST and TYPE must be scalars,
2214 or vectors of the same number of elements. If the value won't fit in the
2215 integer type, the results are undefined.</dd>
2217 <dt><b><tt>uitofp ( CST to TYPE )</tt></b></dt>
2218 <dd>Convert an unsigned integer constant to the corresponding floating point
2219 constant. TYPE must be a scalar or vector floating point type. CST must be
2220 of scalar or vector integer type. Both CST and TYPE must be scalars, or
2221 vectors of the same number of elements. If the value won't fit in the
2222 floating point type, the results are undefined.</dd>
2224 <dt><b><tt>sitofp ( CST to TYPE )</tt></b></dt>
2225 <dd>Convert a signed integer constant to the corresponding floating point
2226 constant. TYPE must be a scalar or vector floating point type. CST must be
2227 of scalar or vector integer type. Both CST and TYPE must be scalars, or
2228 vectors of the same number of elements. If the value won't fit in the
2229 floating point type, the results are undefined.</dd>
2231 <dt><b><tt>ptrtoint ( CST to TYPE )</tt></b></dt>
2232 <dd>Convert a pointer typed constant to the corresponding integer constant
2233 <tt>TYPE</tt> must be an integer type. <tt>CST</tt> must be of pointer
2234 type. The <tt>CST</tt> value is zero extended, truncated, or unchanged to
2235 make it fit in <tt>TYPE</tt>.</dd>
2237 <dt><b><tt>inttoptr ( CST to TYPE )</tt></b></dt>
2238 <dd>Convert a integer constant to a pointer constant. TYPE must be a pointer
2239 type. CST must be of integer type. The CST value is zero extended,
2240 truncated, or unchanged to make it fit in a pointer size. This one is
2241 <i>really</i> dangerous!</dd>
2243 <dt><b><tt>bitcast ( CST to TYPE )</tt></b></dt>
2244 <dd>Convert a constant, CST, to another TYPE. The constraints of the operands
2245 are the same as those for the <a href="#i_bitcast">bitcast
2246 instruction</a>.</dd>
2248 <dt><b><tt>getelementptr ( CSTPTR, IDX0, IDX1, ... )</tt></b></dt>
2249 <dt><b><tt>getelementptr inbounds ( CSTPTR, IDX0, IDX1, ... )</tt></b></dt>
2250 <dd>Perform the <a href="#i_getelementptr">getelementptr operation</a> on
2251 constants. As with the <a href="#i_getelementptr">getelementptr</a>
2252 instruction, the index list may have zero or more indexes, which are
2253 required to make sense for the type of "CSTPTR".</dd>
2255 <dt><b><tt>select ( COND, VAL1, VAL2 )</tt></b></dt>
2256 <dd>Perform the <a href="#i_select">select operation</a> on constants.</dd>
2258 <dt><b><tt>icmp COND ( VAL1, VAL2 )</tt></b></dt>
2259 <dd>Performs the <a href="#i_icmp">icmp operation</a> on constants.</dd>
2261 <dt><b><tt>fcmp COND ( VAL1, VAL2 )</tt></b></dt>
2262 <dd>Performs the <a href="#i_fcmp">fcmp operation</a> on constants.</dd>
2264 <dt><b><tt>extractelement ( VAL, IDX )</tt></b></dt>
2265 <dd>Perform the <a href="#i_extractelement">extractelement operation</a> on
2268 <dt><b><tt>insertelement ( VAL, ELT, IDX )</tt></b></dt>
2269 <dd>Perform the <a href="#i_insertelement">insertelement operation</a> on
2272 <dt><b><tt>shufflevector ( VEC1, VEC2, IDXMASK )</tt></b></dt>
2273 <dd>Perform the <a href="#i_shufflevector">shufflevector operation</a> on
2276 <dt><b><tt>OPCODE ( LHS, RHS )</tt></b></dt>
2277 <dd>Perform the specified operation of the LHS and RHS constants. OPCODE may
2278 be any of the <a href="#binaryops">binary</a>
2279 or <a href="#bitwiseops">bitwise binary</a> operations. The constraints
2280 on operands are the same as those for the corresponding instruction
2281 (e.g. no bitwise operations on floating point values are allowed).</dd>
2286 <!-- ======================================================================= -->
2287 <div class="doc_subsection"><a name="metadata">Embedded Metadata</a>
2290 <div class="doc_text">
2292 <p>Embedded metadata provides a way to attach arbitrary data to the instruction
2293 stream without affecting the behaviour of the program. There are two
2294 metadata primitives, strings and nodes. All metadata has the
2295 <tt>metadata</tt> type and is identified in syntax by a preceding exclamation
2296 point ('<tt>!</tt>').</p>
2298 <p>A metadata string is a string surrounded by double quotes. It can contain
2299 any character by escaping non-printable characters with "\xx" where "xx" is
2300 the two digit hex code. For example: "<tt>!"test\00"</tt>".</p>
2302 <p>Metadata nodes are represented with notation similar to structure constants
2303 (a comma separated list of elements, surrounded by braces and preceeded by an
2304 exclamation point). For example: "<tt>!{ metadata !"test\00", i32
2307 <p>A metadata node will attempt to track changes to the values it holds. In the
2308 event that a value is deleted, it will be replaced with a typeless
2309 "<tt>null</tt>", such as "<tt>metadata !{null, i32 10}</tt>".</p>
2311 <p>Optimizations may rely on metadata to provide additional information about
2312 the program that isn't available in the instructions, or that isn't easily
2313 computable. Similarly, the code generator may expect a certain metadata
2314 format to be used to express debugging information.</p>
2318 <!-- *********************************************************************** -->
2319 <div class="doc_section"> <a name="othervalues">Other Values</a> </div>
2320 <!-- *********************************************************************** -->
2322 <!-- ======================================================================= -->
2323 <div class="doc_subsection">
2324 <a name="inlineasm">Inline Assembler Expressions</a>
2327 <div class="doc_text">
2329 <p>LLVM supports inline assembler expressions (as opposed
2330 to <a href="#moduleasm"> Module-Level Inline Assembly</a>) through the use of
2331 a special value. This value represents the inline assembler as a string
2332 (containing the instructions to emit), a list of operand constraints (stored
2333 as a string), and a flag that indicates whether or not the inline asm
2334 expression has side effects. An example inline assembler expression is:</p>
2336 <div class="doc_code">
2338 i32 (i32) asm "bswap $0", "=r,r"
2342 <p>Inline assembler expressions may <b>only</b> be used as the callee operand of
2343 a <a href="#i_call"><tt>call</tt> instruction</a>. Thus, typically we
2346 <div class="doc_code">
2348 %X = call i32 asm "<a href="#int_bswap">bswap</a> $0", "=r,r"(i32 %Y)
2352 <p>Inline asms with side effects not visible in the constraint list must be
2353 marked as having side effects. This is done through the use of the
2354 '<tt>sideeffect</tt>' keyword, like so:</p>
2356 <div class="doc_code">
2358 call void asm sideeffect "eieio", ""()
2362 <p>TODO: The format of the asm and constraints string still need to be
2363 documented here. Constraints on what can be done (e.g. duplication, moving,
2364 etc need to be documented). This is probably best done by reference to
2365 another document that covers inline asm from a holistic perspective.</p>
2370 <!-- *********************************************************************** -->
2371 <div class="doc_section">
2372 <a name="intrinsic_globals">Intrinsic Global Variables</a>
2374 <!-- *********************************************************************** -->
2376 <p>LLVM has a number of "magic" global variables that contain data that affect
2377 code generation or other IR semantics. These are documented here. All globals
2378 of this sort should have a section specified as "<tt>llvm.metadata</tt>". This
2379 section and all globals that start with "<tt>llvm.</tt>" are reserved for use
2382 <!-- ======================================================================= -->
2383 <div class="doc_subsection">
2384 <a name="intg_used">The '<tt>llvm.used</tt>' Global Variable</a>
2387 <div class="doc_text">
2389 <p>The <tt>@llvm.used</tt> global is an array with i8* element type which has <a
2390 href="#linkage_appending">appending linkage</a>. This array contains a list of
2391 pointers to global variables and functions which may optionally have a pointer
2392 cast formed of bitcast or getelementptr. For example, a legal use of it is:</p>
2398 @llvm.used = appending global [2 x i8*] [
2400 i8* bitcast (i32* @Y to i8*)
2401 ], section "llvm.metadata"
2404 <p>If a global variable appears in the <tt>@llvm.used</tt> list, then the
2405 compiler, assembler, and linker are required to treat the symbol as if there is
2406 a reference to the global that it cannot see. For example, if a variable has
2407 internal linkage and no references other than that from the <tt>@llvm.used</tt>
2408 list, it cannot be deleted. This is commonly used to represent references from
2409 inline asms and other things the compiler cannot "see", and corresponds to
2410 "attribute((used))" in GNU C.</p>
2412 <p>On some targets, the code generator must emit a directive to the assembler or
2413 object file to prevent the assembler and linker from molesting the symbol.</p>
2417 <!-- ======================================================================= -->
2418 <div class="doc_subsection">
2419 <a name="intg_compiler_used">The '<tt>llvm.compiler.used</tt>' Global Variable</a>
2422 <div class="doc_text">
2424 <p>The <tt>@llvm.compiler.used</tt> directive is the same as the
2425 <tt>@llvm.used</tt> directive, except that it only prevents the compiler from
2426 touching the symbol. On targets that support it, this allows an intelligent
2427 linker to optimize references to the symbol without being impeded as it would be
2428 by <tt>@llvm.used</tt>.</p>
2430 <p>This is a rare construct that should only be used in rare circumstances, and
2431 should not be exposed to source languages.</p>
2435 <!-- ======================================================================= -->
2436 <div class="doc_subsection">
2437 <a name="intg_global_ctors">The '<tt>llvm.global_ctors</tt>' Global Variable</a>
2440 <div class="doc_text">
2442 <p>TODO: Describe this.</p>
2446 <!-- ======================================================================= -->
2447 <div class="doc_subsection">
2448 <a name="intg_global_dtors">The '<tt>llvm.global_dtors</tt>' Global Variable</a>
2451 <div class="doc_text">
2453 <p>TODO: Describe this.</p>
2458 <!-- *********************************************************************** -->
2459 <div class="doc_section"> <a name="instref">Instruction Reference</a> </div>
2460 <!-- *********************************************************************** -->
2462 <div class="doc_text">
2464 <p>The LLVM instruction set consists of several different classifications of
2465 instructions: <a href="#terminators">terminator
2466 instructions</a>, <a href="#binaryops">binary instructions</a>,
2467 <a href="#bitwiseops">bitwise binary instructions</a>,
2468 <a href="#memoryops">memory instructions</a>, and
2469 <a href="#otherops">other instructions</a>.</p>
2473 <!-- ======================================================================= -->
2474 <div class="doc_subsection"> <a name="terminators">Terminator
2475 Instructions</a> </div>
2477 <div class="doc_text">
2479 <p>As mentioned <a href="#functionstructure">previously</a>, every basic block
2480 in a program ends with a "Terminator" instruction, which indicates which
2481 block should be executed after the current block is finished. These
2482 terminator instructions typically yield a '<tt>void</tt>' value: they produce
2483 control flow, not values (the one exception being the
2484 '<a href="#i_invoke"><tt>invoke</tt></a>' instruction).</p>
2486 <p>There are six different terminator instructions: the
2487 '<a href="#i_ret"><tt>ret</tt></a>' instruction, the
2488 '<a href="#i_br"><tt>br</tt></a>' instruction, the
2489 '<a href="#i_switch"><tt>switch</tt></a>' instruction, the
2490 '<a href="#i_invoke"><tt>invoke</tt></a>' instruction, the
2491 '<a href="#i_unwind"><tt>unwind</tt></a>' instruction, and the
2492 '<a href="#i_unreachable"><tt>unreachable</tt></a>' instruction.</p>
2496 <!-- _______________________________________________________________________ -->
2497 <div class="doc_subsubsection"> <a name="i_ret">'<tt>ret</tt>'
2498 Instruction</a> </div>
2500 <div class="doc_text">
2504 ret <type> <value> <i>; Return a value from a non-void function</i>
2505 ret void <i>; Return from void function</i>
2509 <p>The '<tt>ret</tt>' instruction is used to return control flow (and optionally
2510 a value) from a function back to the caller.</p>
2512 <p>There are two forms of the '<tt>ret</tt>' instruction: one that returns a
2513 value and then causes control flow, and one that just causes control flow to
2517 <p>The '<tt>ret</tt>' instruction optionally accepts a single argument, the
2518 return value. The type of the return value must be a
2519 '<a href="#t_firstclass">first class</a>' type.</p>
2521 <p>A function is not <a href="#wellformed">well formed</a> if it it has a
2522 non-void return type and contains a '<tt>ret</tt>' instruction with no return
2523 value or a return value with a type that does not match its type, or if it
2524 has a void return type and contains a '<tt>ret</tt>' instruction with a
2528 <p>When the '<tt>ret</tt>' instruction is executed, control flow returns back to
2529 the calling function's context. If the caller is a
2530 "<a href="#i_call"><tt>call</tt></a>" instruction, execution continues at the
2531 instruction after the call. If the caller was an
2532 "<a href="#i_invoke"><tt>invoke</tt></a>" instruction, execution continues at
2533 the beginning of the "normal" destination block. If the instruction returns
2534 a value, that value shall set the call or invoke instruction's return
2539 ret i32 5 <i>; Return an integer value of 5</i>
2540 ret void <i>; Return from a void function</i>
2541 ret { i32, i8 } { i32 4, i8 2 } <i>; Return a struct of values 4 and 2</i>
2544 <p>Note that the code generator does not yet fully support large
2545 return values. The specific sizes that are currently supported are
2546 dependent on the target. For integers, on 32-bit targets the limit
2547 is often 64 bits, and on 64-bit targets the limit is often 128 bits.
2548 For aggregate types, the current limits are dependent on the element
2549 types; for example targets are often limited to 2 total integer
2550 elements and 2 total floating-point elements.</p>
2553 <!-- _______________________________________________________________________ -->
2554 <div class="doc_subsubsection"> <a name="i_br">'<tt>br</tt>' Instruction</a> </div>
2556 <div class="doc_text">
2560 br i1 <cond>, label <iftrue>, label <iffalse><br> br label <dest> <i>; Unconditional branch</i>
2564 <p>The '<tt>br</tt>' instruction is used to cause control flow to transfer to a
2565 different basic block in the current function. There are two forms of this
2566 instruction, corresponding to a conditional branch and an unconditional
2570 <p>The conditional branch form of the '<tt>br</tt>' instruction takes a single
2571 '<tt>i1</tt>' value and two '<tt>label</tt>' values. The unconditional form
2572 of the '<tt>br</tt>' instruction takes a single '<tt>label</tt>' value as a
2576 <p>Upon execution of a conditional '<tt>br</tt>' instruction, the '<tt>i1</tt>'
2577 argument is evaluated. If the value is <tt>true</tt>, control flows to the
2578 '<tt>iftrue</tt>' <tt>label</tt> argument. If "cond" is <tt>false</tt>,
2579 control flows to the '<tt>iffalse</tt>' <tt>label</tt> argument.</p>
2584 %cond = <a href="#i_icmp">icmp</a> eq i32 %a, %b
2585 br i1 %cond, label %IfEqual, label %IfUnequal
2587 <a href="#i_ret">ret</a> i32 1
2589 <a href="#i_ret">ret</a> i32 0
2594 <!-- _______________________________________________________________________ -->
2595 <div class="doc_subsubsection">
2596 <a name="i_switch">'<tt>switch</tt>' Instruction</a>
2599 <div class="doc_text">
2603 switch <intty> <value>, label <defaultdest> [ <intty> <val>, label <dest> ... ]
2607 <p>The '<tt>switch</tt>' instruction is used to transfer control flow to one of
2608 several different places. It is a generalization of the '<tt>br</tt>'
2609 instruction, allowing a branch to occur to one of many possible
2613 <p>The '<tt>switch</tt>' instruction uses three parameters: an integer
2614 comparison value '<tt>value</tt>', a default '<tt>label</tt>' destination,
2615 and an array of pairs of comparison value constants and '<tt>label</tt>'s.
2616 The table is not allowed to contain duplicate constant entries.</p>
2619 <p>The <tt>switch</tt> instruction specifies a table of values and
2620 destinations. When the '<tt>switch</tt>' instruction is executed, this table
2621 is searched for the given value. If the value is found, control flow is
2622 transfered to the corresponding destination; otherwise, control flow is
2623 transfered to the default destination.</p>
2625 <h5>Implementation:</h5>
2626 <p>Depending on properties of the target machine and the particular
2627 <tt>switch</tt> instruction, this instruction may be code generated in
2628 different ways. For example, it could be generated as a series of chained
2629 conditional branches or with a lookup table.</p>
2633 <i>; Emulate a conditional br instruction</i>
2634 %Val = <a href="#i_zext">zext</a> i1 %value to i32
2635 switch i32 %Val, label %truedest [ i32 0, label %falsedest ]
2637 <i>; Emulate an unconditional br instruction</i>
2638 switch i32 0, label %dest [ ]
2640 <i>; Implement a jump table:</i>
2641 switch i32 %val, label %otherwise [ i32 0, label %onzero
2643 i32 2, label %ontwo ]
2648 <!-- _______________________________________________________________________ -->
2649 <div class="doc_subsubsection">
2650 <a name="i_invoke">'<tt>invoke</tt>' Instruction</a>
2653 <div class="doc_text">
2657 <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>]
2658 to label <normal label> unwind label <exception label>
2662 <p>The '<tt>invoke</tt>' instruction causes control to transfer to a specified
2663 function, with the possibility of control flow transfer to either the
2664 '<tt>normal</tt>' label or the '<tt>exception</tt>' label. If the callee
2665 function returns with the "<tt><a href="#i_ret">ret</a></tt>" instruction,
2666 control flow will return to the "normal" label. If the callee (or any
2667 indirect callees) returns with the "<a href="#i_unwind"><tt>unwind</tt></a>"
2668 instruction, control is interrupted and continued at the dynamically nearest
2669 "exception" label.</p>
2672 <p>This instruction requires several arguments:</p>
2675 <li>The optional "cconv" marker indicates which <a href="#callingconv">calling
2676 convention</a> the call should use. If none is specified, the call
2677 defaults to using C calling conventions.</li>
2679 <li>The optional <a href="#paramattrs">Parameter Attributes</a> list for
2680 return values. Only '<tt>zeroext</tt>', '<tt>signext</tt>', and
2681 '<tt>inreg</tt>' attributes are valid here.</li>
2683 <li>'<tt>ptr to function ty</tt>': shall be the signature of the pointer to
2684 function value being invoked. In most cases, this is a direct function
2685 invocation, but indirect <tt>invoke</tt>s are just as possible, branching
2686 off an arbitrary pointer to function value.</li>
2688 <li>'<tt>function ptr val</tt>': An LLVM value containing a pointer to a
2689 function to be invoked. </li>
2691 <li>'<tt>function args</tt>': argument list whose types match the function
2692 signature argument types. If the function signature indicates the
2693 function accepts a variable number of arguments, the extra arguments can
2696 <li>'<tt>normal label</tt>': the label reached when the called function
2697 executes a '<tt><a href="#i_ret">ret</a></tt>' instruction. </li>
2699 <li>'<tt>exception label</tt>': the label reached when a callee returns with
2700 the <a href="#i_unwind"><tt>unwind</tt></a> instruction. </li>
2702 <li>The optional <a href="#fnattrs">function attributes</a> list. Only
2703 '<tt>noreturn</tt>', '<tt>nounwind</tt>', '<tt>readonly</tt>' and
2704 '<tt>readnone</tt>' attributes are valid here.</li>
2708 <p>This instruction is designed to operate as a standard
2709 '<tt><a href="#i_call">call</a></tt>' instruction in most regards. The
2710 primary difference is that it establishes an association with a label, which
2711 is used by the runtime library to unwind the stack.</p>
2713 <p>This instruction is used in languages with destructors to ensure that proper
2714 cleanup is performed in the case of either a <tt>longjmp</tt> or a thrown
2715 exception. Additionally, this is important for implementation of
2716 '<tt>catch</tt>' clauses in high-level languages that support them.</p>
2718 <p>For the purposes of the SSA form, the definition of the value returned by the
2719 '<tt>invoke</tt>' instruction is deemed to occur on the edge from the current
2720 block to the "normal" label. If the callee unwinds then no return value is
2725 %retval = invoke i32 @Test(i32 15) to label %Continue
2726 unwind label %TestCleanup <i>; {i32}:retval set</i>
2727 %retval = invoke <a href="#callingconv">coldcc</a> i32 %Testfnptr(i32 15) to label %Continue
2728 unwind label %TestCleanup <i>; {i32}:retval set</i>
2733 <!-- _______________________________________________________________________ -->
2735 <div class="doc_subsubsection"> <a name="i_unwind">'<tt>unwind</tt>'
2736 Instruction</a> </div>
2738 <div class="doc_text">
2746 <p>The '<tt>unwind</tt>' instruction unwinds the stack, continuing control flow
2747 at the first callee in the dynamic call stack which used
2748 an <a href="#i_invoke"><tt>invoke</tt></a> instruction to perform the call.
2749 This is primarily used to implement exception handling.</p>
2752 <p>The '<tt>unwind</tt>' instruction causes execution of the current function to
2753 immediately halt. The dynamic call stack is then searched for the
2754 first <a href="#i_invoke"><tt>invoke</tt></a> instruction on the call stack.
2755 Once found, execution continues at the "exceptional" destination block
2756 specified by the <tt>invoke</tt> instruction. If there is no <tt>invoke</tt>
2757 instruction in the dynamic call chain, undefined behavior results.</p>
2761 <!-- _______________________________________________________________________ -->
2763 <div class="doc_subsubsection"> <a name="i_unreachable">'<tt>unreachable</tt>'
2764 Instruction</a> </div>
2766 <div class="doc_text">
2774 <p>The '<tt>unreachable</tt>' instruction has no defined semantics. This
2775 instruction is used to inform the optimizer that a particular portion of the
2776 code is not reachable. This can be used to indicate that the code after a
2777 no-return function cannot be reached, and other facts.</p>
2780 <p>The '<tt>unreachable</tt>' instruction has no defined semantics.</p>
2784 <!-- ======================================================================= -->
2785 <div class="doc_subsection"> <a name="binaryops">Binary Operations</a> </div>
2787 <div class="doc_text">
2789 <p>Binary operators are used to do most of the computation in a program. They
2790 require two operands of the same type, execute an operation on them, and
2791 produce a single value. The operands might represent multiple data, as is
2792 the case with the <a href="#t_vector">vector</a> data type. The result value
2793 has the same type as its operands.</p>
2795 <p>There are several different binary operators:</p>
2799 <!-- _______________________________________________________________________ -->
2800 <div class="doc_subsubsection">
2801 <a name="i_add">'<tt>add</tt>' Instruction</a>
2804 <div class="doc_text">
2808 <result> = add <ty> <op1>, <op2> <i>; yields {ty}:result</i>
2809 <result> = add nuw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
2810 <result> = add nsw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
2811 <result> = add nuw nsw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
2815 <p>The '<tt>add</tt>' instruction returns the sum of its two operands.</p>
2818 <p>The two arguments to the '<tt>add</tt>' instruction must
2819 be <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
2820 integer values. Both arguments must have identical types.</p>
2823 <p>The value produced is the integer sum of the two operands.</p>
2825 <p>If the sum has unsigned overflow, the result returned is the mathematical
2826 result modulo 2<sup>n</sup>, where n is the bit width of the result.</p>
2828 <p>Because LLVM integers use a two's complement representation, this instruction
2829 is appropriate for both signed and unsigned integers.</p>
2831 <p><tt>nuw</tt> and <tt>nsw</tt> stand for "No Unsigned Wrap"
2832 and "No Signed Wrap", respectively. If the <tt>nuw</tt> and/or
2833 <tt>nsw</tt> keywords are present, the result value of the <tt>add</tt>
2834 is undefined if unsigned and/or signed overflow, respectively, occurs.</p>
2838 <result> = add i32 4, %var <i>; yields {i32}:result = 4 + %var</i>
2843 <!-- _______________________________________________________________________ -->
2844 <div class="doc_subsubsection">
2845 <a name="i_fadd">'<tt>fadd</tt>' Instruction</a>
2848 <div class="doc_text">
2852 <result> = fadd <ty> <op1>, <op2> <i>; yields {ty}:result</i>
2856 <p>The '<tt>fadd</tt>' instruction returns the sum of its two operands.</p>
2859 <p>The two arguments to the '<tt>fadd</tt>' instruction must be
2860 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
2861 floating point values. Both arguments must have identical types.</p>
2864 <p>The value produced is the floating point sum of the two operands.</p>
2868 <result> = fadd float 4.0, %var <i>; yields {float}:result = 4.0 + %var</i>
2873 <!-- _______________________________________________________________________ -->
2874 <div class="doc_subsubsection">
2875 <a name="i_sub">'<tt>sub</tt>' Instruction</a>
2878 <div class="doc_text">
2882 <result> = sub <ty> <op1>, <op2> <i>; yields {ty}:result</i>
2883 <result> = sub nuw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
2884 <result> = sub nsw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
2885 <result> = sub nuw nsw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
2889 <p>The '<tt>sub</tt>' instruction returns the difference of its two
2892 <p>Note that the '<tt>sub</tt>' instruction is used to represent the
2893 '<tt>neg</tt>' instruction present in most other intermediate
2894 representations.</p>
2897 <p>The two arguments to the '<tt>sub</tt>' instruction must
2898 be <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
2899 integer values. Both arguments must have identical types.</p>
2902 <p>The value produced is the integer difference of the two operands.</p>
2904 <p>If the difference has unsigned overflow, the result returned is the
2905 mathematical result modulo 2<sup>n</sup>, where n is the bit width of the
2908 <p>Because LLVM integers use a two's complement representation, this instruction
2909 is appropriate for both signed and unsigned integers.</p>
2911 <p><tt>nuw</tt> and <tt>nsw</tt> stand for "No Unsigned Wrap"
2912 and "No Signed Wrap", respectively. If the <tt>nuw</tt> and/or
2913 <tt>nsw</tt> keywords are present, the result value of the <tt>sub</tt>
2914 is undefined if unsigned and/or signed overflow, respectively, occurs.</p>
2918 <result> = sub i32 4, %var <i>; yields {i32}:result = 4 - %var</i>
2919 <result> = sub i32 0, %val <i>; yields {i32}:result = -%var</i>
2924 <!-- _______________________________________________________________________ -->
2925 <div class="doc_subsubsection">
2926 <a name="i_fsub">'<tt>fsub</tt>' Instruction</a>
2929 <div class="doc_text">
2933 <result> = fsub <ty> <op1>, <op2> <i>; yields {ty}:result</i>
2937 <p>The '<tt>fsub</tt>' instruction returns the difference of its two
2940 <p>Note that the '<tt>fsub</tt>' instruction is used to represent the
2941 '<tt>fneg</tt>' instruction present in most other intermediate
2942 representations.</p>
2945 <p>The two arguments to the '<tt>fsub</tt>' instruction must be
2946 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
2947 floating point values. Both arguments must have identical types.</p>
2950 <p>The value produced is the floating point difference of the two operands.</p>
2954 <result> = fsub float 4.0, %var <i>; yields {float}:result = 4.0 - %var</i>
2955 <result> = fsub float -0.0, %val <i>; yields {float}:result = -%var</i>
2960 <!-- _______________________________________________________________________ -->
2961 <div class="doc_subsubsection">
2962 <a name="i_mul">'<tt>mul</tt>' Instruction</a>
2965 <div class="doc_text">
2969 <result> = mul <ty> <op1>, <op2> <i>; yields {ty}:result</i>
2970 <result> = mul nuw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
2971 <result> = mul nsw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
2972 <result> = mul nuw nsw <ty> <op1>, <op2> <i>; yields {ty}:result</i>
2976 <p>The '<tt>mul</tt>' instruction returns the product of its two operands.</p>
2979 <p>The two arguments to the '<tt>mul</tt>' instruction must
2980 be <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
2981 integer values. Both arguments must have identical types.</p>
2984 <p>The value produced is the integer product of the two operands.</p>
2986 <p>If the result of the multiplication has unsigned overflow, the result
2987 returned is the mathematical result modulo 2<sup>n</sup>, where n is the bit
2988 width of the result.</p>
2990 <p>Because LLVM integers use a two's complement representation, and the result
2991 is the same width as the operands, this instruction returns the correct
2992 result for both signed and unsigned integers. If a full product
2993 (e.g. <tt>i32</tt>x<tt>i32</tt>-><tt>i64</tt>) is needed, the operands should
2994 be sign-extended or zero-extended as appropriate to the width of the full
2997 <p><tt>nuw</tt> and <tt>nsw</tt> stand for "No Unsigned Wrap"
2998 and "No Signed Wrap", respectively. If the <tt>nuw</tt> and/or
2999 <tt>nsw</tt> keywords are present, the result value of the <tt>mul</tt>
3000 is undefined if unsigned and/or signed overflow, respectively, occurs.</p>
3004 <result> = mul i32 4, %var <i>; yields {i32}:result = 4 * %var</i>
3009 <!-- _______________________________________________________________________ -->
3010 <div class="doc_subsubsection">
3011 <a name="i_fmul">'<tt>fmul</tt>' Instruction</a>
3014 <div class="doc_text">
3018 <result> = fmul <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3022 <p>The '<tt>fmul</tt>' instruction returns the product of its two operands.</p>
3025 <p>The two arguments to the '<tt>fmul</tt>' instruction must be
3026 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3027 floating point values. Both arguments must have identical types.</p>
3030 <p>The value produced is the floating point product of the two operands.</p>
3034 <result> = fmul float 4.0, %var <i>; yields {float}:result = 4.0 * %var</i>
3039 <!-- _______________________________________________________________________ -->
3040 <div class="doc_subsubsection"> <a name="i_udiv">'<tt>udiv</tt>' Instruction
3043 <div class="doc_text">
3047 <result> = udiv <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3051 <p>The '<tt>udiv</tt>' instruction returns the quotient of its two operands.</p>
3054 <p>The two arguments to the '<tt>udiv</tt>' instruction must be
3055 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3056 values. Both arguments must have identical types.</p>
3059 <p>The value produced is the unsigned integer quotient of the two operands.</p>
3061 <p>Note that unsigned integer division and signed integer division are distinct
3062 operations; for signed integer division, use '<tt>sdiv</tt>'.</p>
3064 <p>Division by zero leads to undefined behavior.</p>
3068 <result> = udiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
3073 <!-- _______________________________________________________________________ -->
3074 <div class="doc_subsubsection"> <a name="i_sdiv">'<tt>sdiv</tt>' Instruction
3077 <div class="doc_text">
3081 <result> = sdiv <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3082 <result> = sdiv exact <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3086 <p>The '<tt>sdiv</tt>' instruction returns the quotient of its two operands.</p>
3089 <p>The two arguments to the '<tt>sdiv</tt>' instruction must be
3090 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3091 values. Both arguments must have identical types.</p>
3094 <p>The value produced is the signed integer quotient of the two operands rounded
3097 <p>Note that signed integer division and unsigned integer division are distinct
3098 operations; for unsigned integer division, use '<tt>udiv</tt>'.</p>
3100 <p>Division by zero leads to undefined behavior. Overflow also leads to
3101 undefined behavior; this is a rare case, but can occur, for example, by doing
3102 a 32-bit division of -2147483648 by -1.</p>
3104 <p>If the <tt>exact</tt> keyword is present, the result value of the
3105 <tt>sdiv</tt> is undefined if the result would be rounded or if overflow
3110 <result> = sdiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
3115 <!-- _______________________________________________________________________ -->
3116 <div class="doc_subsubsection"> <a name="i_fdiv">'<tt>fdiv</tt>'
3117 Instruction</a> </div>
3119 <div class="doc_text">
3123 <result> = fdiv <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3127 <p>The '<tt>fdiv</tt>' instruction returns the quotient of its two operands.</p>
3130 <p>The two arguments to the '<tt>fdiv</tt>' instruction must be
3131 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3132 floating point values. Both arguments must have identical types.</p>
3135 <p>The value produced is the floating point quotient of the two operands.</p>
3139 <result> = fdiv float 4.0, %var <i>; yields {float}:result = 4.0 / %var</i>
3144 <!-- _______________________________________________________________________ -->
3145 <div class="doc_subsubsection"> <a name="i_urem">'<tt>urem</tt>' Instruction</a>
3148 <div class="doc_text">
3152 <result> = urem <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3156 <p>The '<tt>urem</tt>' instruction returns the remainder from the unsigned
3157 division of its two arguments.</p>
3160 <p>The two arguments to the '<tt>urem</tt>' instruction must be
3161 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3162 values. Both arguments must have identical types.</p>
3165 <p>This instruction returns the unsigned integer <i>remainder</i> of a division.
3166 This instruction always performs an unsigned division to get the
3169 <p>Note that unsigned integer remainder and signed integer remainder are
3170 distinct operations; for signed integer remainder, use '<tt>srem</tt>'.</p>
3172 <p>Taking the remainder of a division by zero leads to undefined behavior.</p>
3176 <result> = urem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
3181 <!-- _______________________________________________________________________ -->
3182 <div class="doc_subsubsection">
3183 <a name="i_srem">'<tt>srem</tt>' Instruction</a>
3186 <div class="doc_text">
3190 <result> = srem <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3194 <p>The '<tt>srem</tt>' instruction returns the remainder from the signed
3195 division of its two operands. This instruction can also take
3196 <a href="#t_vector">vector</a> versions of the values in which case the
3197 elements must be integers.</p>
3200 <p>The two arguments to the '<tt>srem</tt>' instruction must be
3201 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3202 values. Both arguments must have identical types.</p>
3205 <p>This instruction returns the <i>remainder</i> of a division (where the result
3206 has the same sign as the dividend, <tt>op1</tt>), not the <i>modulo</i>
3207 operator (where the result has the same sign as the divisor, <tt>op2</tt>) of
3208 a value. For more information about the difference,
3209 see <a href="http://mathforum.org/dr.math/problems/anne.4.28.99.html">The
3210 Math Forum</a>. For a table of how this is implemented in various languages,
3211 please see <a href="http://en.wikipedia.org/wiki/Modulo_operation">
3212 Wikipedia: modulo operation</a>.</p>
3214 <p>Note that signed integer remainder and unsigned integer remainder are
3215 distinct operations; for unsigned integer remainder, use '<tt>urem</tt>'.</p>
3217 <p>Taking the remainder of a division by zero leads to undefined behavior.
3218 Overflow also leads to undefined behavior; this is a rare case, but can
3219 occur, for example, by taking the remainder of a 32-bit division of
3220 -2147483648 by -1. (The remainder doesn't actually overflow, but this rule
3221 lets srem be implemented using instructions that return both the result of
3222 the division and the remainder.)</p>
3226 <result> = srem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
3231 <!-- _______________________________________________________________________ -->
3232 <div class="doc_subsubsection">
3233 <a name="i_frem">'<tt>frem</tt>' Instruction</a> </div>
3235 <div class="doc_text">
3239 <result> = frem <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3243 <p>The '<tt>frem</tt>' instruction returns the remainder from the division of
3244 its two operands.</p>
3247 <p>The two arguments to the '<tt>frem</tt>' instruction must be
3248 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3249 floating point values. Both arguments must have identical types.</p>
3252 <p>This instruction returns the <i>remainder</i> of a division. The remainder
3253 has the same sign as the dividend.</p>
3257 <result> = frem float 4.0, %var <i>; yields {float}:result = 4.0 % %var</i>
3262 <!-- ======================================================================= -->
3263 <div class="doc_subsection"> <a name="bitwiseops">Bitwise Binary
3264 Operations</a> </div>
3266 <div class="doc_text">
3268 <p>Bitwise binary operators are used to do various forms of bit-twiddling in a
3269 program. They are generally very efficient instructions and can commonly be
3270 strength reduced from other instructions. They require two operands of the
3271 same type, execute an operation on them, and produce a single value. The
3272 resulting value is the same type as its operands.</p>
3276 <!-- _______________________________________________________________________ -->
3277 <div class="doc_subsubsection"> <a name="i_shl">'<tt>shl</tt>'
3278 Instruction</a> </div>
3280 <div class="doc_text">
3284 <result> = shl <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3288 <p>The '<tt>shl</tt>' instruction returns the first operand shifted to the left
3289 a specified number of bits.</p>
3292 <p>Both arguments to the '<tt>shl</tt>' instruction must be the
3293 same <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
3294 integer type. '<tt>op2</tt>' is treated as an unsigned value.</p>
3297 <p>The value produced is <tt>op1</tt> * 2<sup><tt>op2</tt></sup> mod
3298 2<sup>n</sup>, where <tt>n</tt> is the width of the result. If <tt>op2</tt>
3299 is (statically or dynamically) negative or equal to or larger than the number
3300 of bits in <tt>op1</tt>, the result is undefined. If the arguments are
3301 vectors, each vector element of <tt>op1</tt> is shifted by the corresponding
3302 shift amount in <tt>op2</tt>.</p>
3306 <result> = shl i32 4, %var <i>; yields {i32}: 4 << %var</i>
3307 <result> = shl i32 4, 2 <i>; yields {i32}: 16</i>
3308 <result> = shl i32 1, 10 <i>; yields {i32}: 1024</i>
3309 <result> = shl i32 1, 32 <i>; undefined</i>
3310 <result> = shl <2 x i32> < i32 1, i32 1>, < i32 1, i32 2> <i>; yields: result=<2 x i32> < i32 2, i32 4></i>
3315 <!-- _______________________________________________________________________ -->
3316 <div class="doc_subsubsection"> <a name="i_lshr">'<tt>lshr</tt>'
3317 Instruction</a> </div>
3319 <div class="doc_text">
3323 <result> = lshr <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3327 <p>The '<tt>lshr</tt>' instruction (logical shift right) returns the first
3328 operand shifted to the right a specified number of bits with zero fill.</p>
3331 <p>Both arguments to the '<tt>lshr</tt>' instruction must be the same
3332 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3333 type. '<tt>op2</tt>' is treated as an unsigned value.</p>
3336 <p>This instruction always performs a logical shift right operation. The most
3337 significant bits of the result will be filled with zero bits after the shift.
3338 If <tt>op2</tt> is (statically or dynamically) equal to or larger than the
3339 number of bits in <tt>op1</tt>, the result is undefined. If the arguments are
3340 vectors, each vector element of <tt>op1</tt> is shifted by the corresponding
3341 shift amount in <tt>op2</tt>.</p>
3345 <result> = lshr i32 4, 1 <i>; yields {i32}:result = 2</i>
3346 <result> = lshr i32 4, 2 <i>; yields {i32}:result = 1</i>
3347 <result> = lshr i8 4, 3 <i>; yields {i8}:result = 0</i>
3348 <result> = lshr i8 -2, 1 <i>; yields {i8}:result = 0x7FFFFFFF </i>
3349 <result> = lshr i32 1, 32 <i>; undefined</i>
3350 <result> = lshr <2 x i32> < i32 -2, i32 4>, < i32 1, i32 2> <i>; yields: result=<2 x i32> < i32 0x7FFFFFFF, i32 1></i>
3355 <!-- _______________________________________________________________________ -->
3356 <div class="doc_subsubsection"> <a name="i_ashr">'<tt>ashr</tt>'
3357 Instruction</a> </div>
3358 <div class="doc_text">
3362 <result> = ashr <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3366 <p>The '<tt>ashr</tt>' instruction (arithmetic shift right) returns the first
3367 operand shifted to the right a specified number of bits with sign
3371 <p>Both arguments to the '<tt>ashr</tt>' instruction must be the same
3372 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3373 type. '<tt>op2</tt>' is treated as an unsigned value.</p>
3376 <p>This instruction always performs an arithmetic shift right operation, The
3377 most significant bits of the result will be filled with the sign bit
3378 of <tt>op1</tt>. If <tt>op2</tt> is (statically or dynamically) equal to or
3379 larger than the number of bits in <tt>op1</tt>, the result is undefined. If
3380 the arguments are vectors, each vector element of <tt>op1</tt> is shifted by
3381 the corresponding shift amount in <tt>op2</tt>.</p>
3385 <result> = ashr i32 4, 1 <i>; yields {i32}:result = 2</i>
3386 <result> = ashr i32 4, 2 <i>; yields {i32}:result = 1</i>
3387 <result> = ashr i8 4, 3 <i>; yields {i8}:result = 0</i>
3388 <result> = ashr i8 -2, 1 <i>; yields {i8}:result = -1</i>
3389 <result> = ashr i32 1, 32 <i>; undefined</i>
3390 <result> = ashr <2 x i32> < i32 -2, i32 4>, < i32 1, i32 3> <i>; yields: result=<2 x i32> < i32 -1, i32 0></i>
3395 <!-- _______________________________________________________________________ -->
3396 <div class="doc_subsubsection"> <a name="i_and">'<tt>and</tt>'
3397 Instruction</a> </div>
3399 <div class="doc_text">
3403 <result> = and <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3407 <p>The '<tt>and</tt>' instruction returns the bitwise logical and of its two
3411 <p>The two arguments to the '<tt>and</tt>' instruction must be
3412 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3413 values. Both arguments must have identical types.</p>
3416 <p>The truth table used for the '<tt>and</tt>' instruction is:</p>
3418 <table border="1" cellspacing="0" cellpadding="4">
3450 <result> = and i32 4, %var <i>; yields {i32}:result = 4 & %var</i>
3451 <result> = and i32 15, 40 <i>; yields {i32}:result = 8</i>
3452 <result> = and i32 4, 8 <i>; yields {i32}:result = 0</i>
3455 <!-- _______________________________________________________________________ -->
3456 <div class="doc_subsubsection"> <a name="i_or">'<tt>or</tt>' Instruction</a> </div>
3458 <div class="doc_text">
3462 <result> = or <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3466 <p>The '<tt>or</tt>' instruction returns the bitwise logical inclusive or of its
3470 <p>The two arguments to the '<tt>or</tt>' instruction must be
3471 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3472 values. Both arguments must have identical types.</p>
3475 <p>The truth table used for the '<tt>or</tt>' instruction is:</p>
3477 <table border="1" cellspacing="0" cellpadding="4">
3509 <result> = or i32 4, %var <i>; yields {i32}:result = 4 | %var</i>
3510 <result> = or i32 15, 40 <i>; yields {i32}:result = 47</i>
3511 <result> = or i32 4, 8 <i>; yields {i32}:result = 12</i>
3516 <!-- _______________________________________________________________________ -->
3517 <div class="doc_subsubsection"> <a name="i_xor">'<tt>xor</tt>'
3518 Instruction</a> </div>
3520 <div class="doc_text">
3524 <result> = xor <ty> <op1>, <op2> <i>; yields {ty}:result</i>
3528 <p>The '<tt>xor</tt>' instruction returns the bitwise logical exclusive or of
3529 its two operands. The <tt>xor</tt> is used to implement the "one's
3530 complement" operation, which is the "~" operator in C.</p>
3533 <p>The two arguments to the '<tt>xor</tt>' instruction must be
3534 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3535 values. Both arguments must have identical types.</p>
3538 <p>The truth table used for the '<tt>xor</tt>' instruction is:</p>
3540 <table border="1" cellspacing="0" cellpadding="4">
3572 <result> = xor i32 4, %var <i>; yields {i32}:result = 4 ^ %var</i>
3573 <result> = xor i32 15, 40 <i>; yields {i32}:result = 39</i>
3574 <result> = xor i32 4, 8 <i>; yields {i32}:result = 12</i>
3575 <result> = xor i32 %V, -1 <i>; yields {i32}:result = ~%V</i>
3580 <!-- ======================================================================= -->
3581 <div class="doc_subsection">
3582 <a name="vectorops">Vector Operations</a>
3585 <div class="doc_text">
3587 <p>LLVM supports several instructions to represent vector operations in a
3588 target-independent manner. These instructions cover the element-access and
3589 vector-specific operations needed to process vectors effectively. While LLVM
3590 does directly support these vector operations, many sophisticated algorithms
3591 will want to use target-specific intrinsics to take full advantage of a
3592 specific target.</p>
3596 <!-- _______________________________________________________________________ -->
3597 <div class="doc_subsubsection">
3598 <a name="i_extractelement">'<tt>extractelement</tt>' Instruction</a>
3601 <div class="doc_text">
3605 <result> = extractelement <n x <ty>> <val>, i32 <idx> <i>; yields <ty></i>
3609 <p>The '<tt>extractelement</tt>' instruction extracts a single scalar element
3610 from a vector at a specified index.</p>
3614 <p>The first operand of an '<tt>extractelement</tt>' instruction is a value
3615 of <a href="#t_vector">vector</a> type. The second operand is an index
3616 indicating the position from which to extract the element. The index may be
3620 <p>The result is a scalar of the same type as the element type of
3621 <tt>val</tt>. Its value is the value at position <tt>idx</tt> of
3622 <tt>val</tt>. If <tt>idx</tt> exceeds the length of <tt>val</tt>, the
3623 results are undefined.</p>
3627 %result = extractelement <4 x i32> %vec, i32 0 <i>; yields i32</i>
3632 <!-- _______________________________________________________________________ -->
3633 <div class="doc_subsubsection">
3634 <a name="i_insertelement">'<tt>insertelement</tt>' Instruction</a>
3637 <div class="doc_text">
3641 <result> = insertelement <n x <ty>> <val>, <ty> <elt>, i32 <idx> <i>; yields <n x <ty>></i>
3645 <p>The '<tt>insertelement</tt>' instruction inserts a scalar element into a
3646 vector at a specified index.</p>
3649 <p>The first operand of an '<tt>insertelement</tt>' instruction is a value
3650 of <a href="#t_vector">vector</a> type. The second operand is a scalar value
3651 whose type must equal the element type of the first operand. The third
3652 operand is an index indicating the position at which to insert the value.
3653 The index may be a variable.</p>
3656 <p>The result is a vector of the same type as <tt>val</tt>. Its element values
3657 are those of <tt>val</tt> except at position <tt>idx</tt>, where it gets the
3658 value <tt>elt</tt>. If <tt>idx</tt> exceeds the length of <tt>val</tt>, the
3659 results are undefined.</p>
3663 %result = insertelement <4 x i32> %vec, i32 1, i32 0 <i>; yields <4 x i32></i>
3668 <!-- _______________________________________________________________________ -->
3669 <div class="doc_subsubsection">
3670 <a name="i_shufflevector">'<tt>shufflevector</tt>' Instruction</a>
3673 <div class="doc_text">
3677 <result> = shufflevector <n x <ty>> <v1>, <n x <ty>> <v2>, <m x i32> <mask> <i>; yields <m x <ty>></i>
3681 <p>The '<tt>shufflevector</tt>' instruction constructs a permutation of elements
3682 from two input vectors, returning a vector with the same element type as the
3683 input and length that is the same as the shuffle mask.</p>
3686 <p>The first two operands of a '<tt>shufflevector</tt>' instruction are vectors
3687 with types that match each other. The third argument is a shuffle mask whose
3688 element type is always 'i32'. The result of the instruction is a vector
3689 whose length is the same as the shuffle mask and whose element type is the
3690 same as the element type of the first two operands.</p>
3692 <p>The shuffle mask operand is required to be a constant vector with either
3693 constant integer or undef values.</p>
3696 <p>The elements of the two input vectors are numbered from left to right across
3697 both of the vectors. The shuffle mask operand specifies, for each element of
3698 the result vector, which element of the two input vectors the result element
3699 gets. The element selector may be undef (meaning "don't care") and the
3700 second operand may be undef if performing a shuffle from only one vector.</p>
3704 %result = shufflevector <4 x i32> %v1, <4 x i32> %v2,
3705 <4 x i32> <i32 0, i32 4, i32 1, i32 5> <i>; yields <4 x i32></i>
3706 %result = shufflevector <4 x i32> %v1, <4 x i32> undef,
3707 <4 x i32> <i32 0, i32 1, i32 2, i32 3> <i>; yields <4 x i32></i> - Identity shuffle.
3708 %result = shufflevector <8 x i32> %v1, <8 x i32> undef,
3709 <4 x i32> <i32 0, i32 1, i32 2, i32 3> <i>; yields <4 x i32></i>
3710 %result = shufflevector <4 x i32> %v1, <4 x i32> %v2,
3711 <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>
3716 <!-- ======================================================================= -->
3717 <div class="doc_subsection">
3718 <a name="aggregateops">Aggregate Operations</a>
3721 <div class="doc_text">
3723 <p>LLVM supports several instructions for working with aggregate values.</p>
3727 <!-- _______________________________________________________________________ -->
3728 <div class="doc_subsubsection">
3729 <a name="i_extractvalue">'<tt>extractvalue</tt>' Instruction</a>
3732 <div class="doc_text">
3736 <result> = extractvalue <aggregate type> <val>, <idx>{, <idx>}*
3740 <p>The '<tt>extractvalue</tt>' instruction extracts the value of a struct field
3741 or array element from an aggregate value.</p>
3744 <p>The first operand of an '<tt>extractvalue</tt>' instruction is a value
3745 of <a href="#t_struct">struct</a> or <a href="#t_array">array</a> type. The
3746 operands are constant indices to specify which value to extract in a similar
3747 manner as indices in a
3748 '<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.</p>
3751 <p>The result is the value at the position in the aggregate specified by the
3756 %result = extractvalue {i32, float} %agg, 0 <i>; yields i32</i>
3761 <!-- _______________________________________________________________________ -->
3762 <div class="doc_subsubsection">
3763 <a name="i_insertvalue">'<tt>insertvalue</tt>' Instruction</a>
3766 <div class="doc_text">
3770 <result> = insertvalue <aggregate type> <val>, <ty> <val>, <idx> <i>; yields <n x <ty>></i>
3774 <p>The '<tt>insertvalue</tt>' instruction inserts a value into a struct field or
3775 array element in an aggregate.</p>
3779 <p>The first operand of an '<tt>insertvalue</tt>' instruction is a value
3780 of <a href="#t_struct">struct</a> or <a href="#t_array">array</a> type. The
3781 second operand is a first-class value to insert. The following operands are
3782 constant indices indicating the position at which to insert the value in a
3783 similar manner as indices in a
3784 '<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction. The
3785 value to insert must have the same type as the value identified by the
3789 <p>The result is an aggregate of the same type as <tt>val</tt>. Its value is
3790 that of <tt>val</tt> except that the value at the position specified by the
3791 indices is that of <tt>elt</tt>.</p>
3795 %result = insertvalue {i32, float} %agg, i32 1, 0 <i>; yields {i32, float}</i>
3801 <!-- ======================================================================= -->
3802 <div class="doc_subsection">
3803 <a name="memoryops">Memory Access and Addressing Operations</a>
3806 <div class="doc_text">
3808 <p>A key design point of an SSA-based representation is how it represents
3809 memory. In LLVM, no memory locations are in SSA form, which makes things
3810 very simple. This section describes how to read, write, allocate, and free
3815 <!-- _______________________________________________________________________ -->
3816 <div class="doc_subsubsection">
3817 <a name="i_malloc">'<tt>malloc</tt>' Instruction</a>
3820 <div class="doc_text">
3824 <result> = malloc <type>[, i32 <NumElements>][, align <alignment>] <i>; yields {type*}:result</i>
3828 <p>The '<tt>malloc</tt>' instruction allocates memory from the system heap and
3829 returns a pointer to it. The object is always allocated in the generic
3830 address space (address space zero).</p>
3833 <p>The '<tt>malloc</tt>' instruction allocates
3834 <tt>sizeof(<type>)*NumElements</tt> bytes of memory from the operating
3835 system and returns a pointer of the appropriate type to the program. If
3836 "NumElements" is specified, it is the number of elements allocated, otherwise
3837 "NumElements" is defaulted to be one. If a constant alignment is specified,
3838 the value result of the allocation is guaranteed to be aligned to at least
3839 that boundary. If not specified, or if zero, the target can choose to align
3840 the allocation on any convenient boundary compatible with the type.</p>
3842 <p>'<tt>type</tt>' must be a sized type.</p>
3845 <p>Memory is allocated using the system "<tt>malloc</tt>" function, and a
3846 pointer is returned. The result of a zero byte allocation is undefined. The
3847 result is null if there is insufficient memory available.</p>
3851 %array = malloc [4 x i8] <i>; yields {[%4 x i8]*}:array</i>
3853 %size = <a href="#i_add">add</a> i32 2, 2 <i>; yields {i32}:size = i32 4</i>
3854 %array1 = malloc i8, i32 4 <i>; yields {i8*}:array1</i>
3855 %array2 = malloc [12 x i8], i32 %size <i>; yields {[12 x i8]*}:array2</i>
3856 %array3 = malloc i32, i32 4, align 1024 <i>; yields {i32*}:array3</i>
3857 %array4 = malloc i32, align 1024 <i>; yields {i32*}:array4</i>
3860 <p>Note that the code generator does not yet respect the alignment value.</p>
3864 <!-- _______________________________________________________________________ -->
3865 <div class="doc_subsubsection">
3866 <a name="i_free">'<tt>free</tt>' Instruction</a>
3869 <div class="doc_text">
3873 free <type> <value> <i>; yields {void}</i>
3877 <p>The '<tt>free</tt>' instruction returns memory back to the unused memory heap
3878 to be reallocated in the future.</p>
3881 <p>'<tt>value</tt>' shall be a pointer value that points to a value that was
3882 allocated with the '<tt><a href="#i_malloc">malloc</a></tt>' instruction.</p>
3885 <p>Access to the memory pointed to by the pointer is no longer defined after
3886 this instruction executes. If the pointer is null, the operation is a
3891 %array = <a href="#i_malloc">malloc</a> [4 x i8] <i>; yields {[4 x i8]*}:array</i>
3892 free [4 x i8]* %array
3897 <!-- _______________________________________________________________________ -->
3898 <div class="doc_subsubsection">
3899 <a name="i_alloca">'<tt>alloca</tt>' Instruction</a>
3902 <div class="doc_text">
3906 <result> = alloca <type>[, i32 <NumElements>][, align <alignment>] <i>; yields {type*}:result</i>
3910 <p>The '<tt>alloca</tt>' instruction allocates memory on the stack frame of the
3911 currently executing function, to be automatically released when this function
3912 returns to its caller. The object is always allocated in the generic address
3913 space (address space zero).</p>
3916 <p>The '<tt>alloca</tt>' instruction
3917 allocates <tt>sizeof(<type>)*NumElements</tt> bytes of memory on the
3918 runtime stack, returning a pointer of the appropriate type to the program.
3919 If "NumElements" is specified, it is the number of elements allocated,
3920 otherwise "NumElements" is defaulted to be one. If a constant alignment is
3921 specified, the value result of the allocation is guaranteed to be aligned to
3922 at least that boundary. If not specified, or if zero, the target can choose
3923 to align the allocation on any convenient boundary compatible with the
3926 <p>'<tt>type</tt>' may be any sized type.</p>
3929 <p>Memory is allocated; a pointer is returned. The operation is undefined if
3930 there is insufficient stack space for the allocation. '<tt>alloca</tt>'d
3931 memory is automatically released when the function returns. The
3932 '<tt>alloca</tt>' instruction is commonly used to represent automatic
3933 variables that must have an address available. When the function returns
3934 (either with the <tt><a href="#i_ret">ret</a></tt>
3935 or <tt><a href="#i_unwind">unwind</a></tt> instructions), the memory is
3936 reclaimed. Allocating zero bytes is legal, but the result is undefined.</p>
3940 %ptr = alloca i32 <i>; yields {i32*}:ptr</i>
3941 %ptr = alloca i32, i32 4 <i>; yields {i32*}:ptr</i>
3942 %ptr = alloca i32, i32 4, align 1024 <i>; yields {i32*}:ptr</i>
3943 %ptr = alloca i32, align 1024 <i>; yields {i32*}:ptr</i>
3948 <!-- _______________________________________________________________________ -->
3949 <div class="doc_subsubsection"> <a name="i_load">'<tt>load</tt>'
3950 Instruction</a> </div>
3952 <div class="doc_text">
3956 <result> = load <ty>* <pointer>[, align <alignment>]
3957 <result> = volatile load <ty>* <pointer>[, align <alignment>]
3961 <p>The '<tt>load</tt>' instruction is used to read from memory.</p>
3964 <p>The argument to the '<tt>load</tt>' instruction specifies the memory address
3965 from which to load. The pointer must point to
3966 a <a href="#t_firstclass">first class</a> type. If the <tt>load</tt> is
3967 marked as <tt>volatile</tt>, then the optimizer is not allowed to modify the
3968 number or order of execution of this <tt>load</tt> with other
3969 volatile <tt>load</tt> and <tt><a href="#i_store">store</a></tt>
3972 <p>The optional constant "align" argument specifies the alignment of the
3973 operation (that is, the alignment of the memory address). A value of 0 or an
3974 omitted "align" argument means that the operation has the preferential
3975 alignment for the target. It is the responsibility of the code emitter to
3976 ensure that the alignment information is correct. Overestimating the
3977 alignment results in an undefined behavior. Underestimating the alignment may
3978 produce less efficient code. An alignment of 1 is always safe.</p>
3981 <p>The location of memory pointed to is loaded. If the value being loaded is of
3982 scalar type then the number of bytes read does not exceed the minimum number
3983 of bytes needed to hold all bits of the type. For example, loading an
3984 <tt>i24</tt> reads at most three bytes. When loading a value of a type like
3985 <tt>i20</tt> with a size that is not an integral number of bytes, the result
3986 is undefined if the value was not originally written using a store of the
3991 %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
3992 <a href="#i_store">store</a> i32 3, i32* %ptr <i>; yields {void}</i>
3993 %val = load i32* %ptr <i>; yields {i32}:val = i32 3</i>
3998 <!-- _______________________________________________________________________ -->
3999 <div class="doc_subsubsection"> <a name="i_store">'<tt>store</tt>'
4000 Instruction</a> </div>
4002 <div class="doc_text">
4006 store <ty> <value>, <ty>* <pointer>[, align <alignment>] <i>; yields {void}</i>
4007 volatile store <ty> <value>, <ty>* <pointer>[, align <alignment>] <i>; yields {void}</i>
4011 <p>The '<tt>store</tt>' instruction is used to write to memory.</p>
4014 <p>There are two arguments to the '<tt>store</tt>' instruction: a value to store
4015 and an address at which to store it. The type of the
4016 '<tt><pointer></tt>' operand must be a pointer to
4017 the <a href="#t_firstclass">first class</a> type of the
4018 '<tt><value></tt>' operand. If the <tt>store</tt> is marked
4019 as <tt>volatile</tt>, then the optimizer is not allowed to modify the number
4020 or order of execution of this <tt>store</tt> with other
4021 volatile <tt>load</tt> and <tt><a href="#i_store">store</a></tt>
4024 <p>The optional constant "align" argument specifies the alignment of the
4025 operation (that is, the alignment of the memory address). A value of 0 or an
4026 omitted "align" argument means that the operation has the preferential
4027 alignment for the target. It is the responsibility of the code emitter to
4028 ensure that the alignment information is correct. Overestimating the
4029 alignment results in an undefined behavior. Underestimating the alignment may
4030 produce less efficient code. An alignment of 1 is always safe.</p>
4033 <p>The contents of memory are updated to contain '<tt><value></tt>' at the
4034 location specified by the '<tt><pointer></tt>' operand. If
4035 '<tt><value></tt>' is of scalar type then the number of bytes written
4036 does not exceed the minimum number of bytes needed to hold all bits of the
4037 type. For example, storing an <tt>i24</tt> writes at most three bytes. When
4038 writing a value of a type like <tt>i20</tt> with a size that is not an
4039 integral number of bytes, it is unspecified what happens to the extra bits
4040 that do not belong to the type, but they will typically be overwritten.</p>
4044 %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
4045 store i32 3, i32* %ptr <i>; yields {void}</i>
4046 %val = <a href="#i_load">load</a> i32* %ptr <i>; yields {i32}:val = i32 3</i>
4051 <!-- _______________________________________________________________________ -->
4052 <div class="doc_subsubsection">
4053 <a name="i_getelementptr">'<tt>getelementptr</tt>' Instruction</a>
4056 <div class="doc_text">
4060 <result> = getelementptr <pty>* <ptrval>{, <ty> <idx>}*
4061 <result> = getelementptr inbounds <pty>* <ptrval>{, <ty> <idx>}*
4065 <p>The '<tt>getelementptr</tt>' instruction is used to get the address of a
4066 subelement of an aggregate data structure. It performs address calculation
4067 only and does not access memory.</p>
4070 <p>The first argument is always a pointer, and forms the basis of the
4071 calculation. The remaining arguments are indices that indicate which of the
4072 elements of the aggregate object are indexed. The interpretation of each
4073 index is dependent on the type being indexed into. The first index always
4074 indexes the pointer value given as the first argument, the second index
4075 indexes a value of the type pointed to (not necessarily the value directly
4076 pointed to, since the first index can be non-zero), etc. The first type
4077 indexed into must be a pointer value, subsequent types can be arrays, vectors
4078 and structs. Note that subsequent types being indexed into can never be
4079 pointers, since that would require loading the pointer before continuing
4082 <p>The type of each index argument depends on the type it is indexing into.
4083 When indexing into a (optionally packed) structure, only <tt>i32</tt> integer
4084 <b>constants</b> are allowed. When indexing into an array, pointer or
4085 vector, integers of any width are allowed, and they are not required to be
4088 <p>For example, let's consider a C code fragment and how it gets compiled to
4091 <div class="doc_code">
4104 int *foo(struct ST *s) {
4105 return &s[1].Z.B[5][13];
4110 <p>The LLVM code generated by the GCC frontend is:</p>
4112 <div class="doc_code">
4114 %RT = <a href="#namedtypes">type</a> { i8 , [10 x [20 x i32]], i8 }
4115 %ST = <a href="#namedtypes">type</a> { i32, double, %RT }
4117 define i32* @foo(%ST* %s) {
4119 %reg = getelementptr %ST* %s, i32 1, i32 2, i32 1, i32 5, i32 13
4126 <p>In the example above, the first index is indexing into the '<tt>%ST*</tt>'
4127 type, which is a pointer, yielding a '<tt>%ST</tt>' = '<tt>{ i32, double, %RT
4128 }</tt>' type, a structure. The second index indexes into the third element
4129 of the structure, yielding a '<tt>%RT</tt>' = '<tt>{ i8 , [10 x [20 x i32]],
4130 i8 }</tt>' type, another structure. The third index indexes into the second
4131 element of the structure, yielding a '<tt>[10 x [20 x i32]]</tt>' type, an
4132 array. The two dimensions of the array are subscripted into, yielding an
4133 '<tt>i32</tt>' type. The '<tt>getelementptr</tt>' instruction returns a
4134 pointer to this element, thus computing a value of '<tt>i32*</tt>' type.</p>
4136 <p>Note that it is perfectly legal to index partially through a structure,
4137 returning a pointer to an inner element. Because of this, the LLVM code for
4138 the given testcase is equivalent to:</p>
4141 define i32* @foo(%ST* %s) {
4142 %t1 = getelementptr %ST* %s, i32 1 <i>; yields %ST*:%t1</i>
4143 %t2 = getelementptr %ST* %t1, i32 0, i32 2 <i>; yields %RT*:%t2</i>
4144 %t3 = getelementptr %RT* %t2, i32 0, i32 1 <i>; yields [10 x [20 x i32]]*:%t3</i>
4145 %t4 = getelementptr [10 x [20 x i32]]* %t3, i32 0, i32 5 <i>; yields [20 x i32]*:%t4</i>
4146 %t5 = getelementptr [20 x i32]* %t4, i32 0, i32 13 <i>; yields i32*:%t5</i>
4151 <p>If the <tt>inbounds</tt> keyword is present, the result value of the
4152 <tt>getelementptr</tt> is undefined if the base pointer is not an
4153 <i>in bounds</i> address of an allocated object, or if any of the addresses
4154 that would be formed by successive addition of the offsets implied by the
4155 indices to the base address with infinitely precise arithmetic are not an
4156 <i>in bounds</i> address of that allocated object.
4157 The <i>in bounds</i> addresses for an allocated object are all the addresses
4158 that point into the object, plus the address one byte past the end.</p>
4160 <p>If the <tt>inbounds</tt> keyword is not present, the offsets are added to
4161 the base address with silently-wrapping two's complement arithmetic, and
4162 the result value of the <tt>getelementptr</tt> may be outside the object
4163 pointed to by the base pointer. The result value may not necessarily be
4164 used to access memory though, even if it happens to point into allocated
4165 storage. See the <a href="#pointeraliasing">Pointer Aliasing Rules</a>
4166 section for more information.</p>
4168 <p>The getelementptr instruction is often confusing. For some more insight into
4169 how it works, see <a href="GetElementPtr.html">the getelementptr FAQ</a>.</p>
4173 <i>; yields [12 x i8]*:aptr</i>
4174 %aptr = getelementptr {i32, [12 x i8]}* %saptr, i64 0, i32 1
4175 <i>; yields i8*:vptr</i>
4176 %vptr = getelementptr {i32, <2 x i8>}* %svptr, i64 0, i32 1, i32 1
4177 <i>; yields i8*:eptr</i>
4178 %eptr = getelementptr [12 x i8]* %aptr, i64 0, i32 1
4179 <i>; yields i32*:iptr</i>
4180 %iptr = getelementptr [10 x i32]* @arr, i16 0, i16 0
4185 <!-- ======================================================================= -->
4186 <div class="doc_subsection"> <a name="convertops">Conversion Operations</a>
4189 <div class="doc_text">
4191 <p>The instructions in this category are the conversion instructions (casting)
4192 which all take a single operand and a type. They perform various bit
4193 conversions on the operand.</p>
4197 <!-- _______________________________________________________________________ -->
4198 <div class="doc_subsubsection">
4199 <a name="i_trunc">'<tt>trunc .. to</tt>' Instruction</a>
4201 <div class="doc_text">
4205 <result> = trunc <ty> <value> to <ty2> <i>; yields ty2</i>
4209 <p>The '<tt>trunc</tt>' instruction truncates its operand to the
4210 type <tt>ty2</tt>.</p>
4213 <p>The '<tt>trunc</tt>' instruction takes a <tt>value</tt> to trunc, which must
4214 be an <a href="#t_integer">integer</a> type, and a type that specifies the
4215 size and type of the result, which must be
4216 an <a href="#t_integer">integer</a> type. The bit size of <tt>value</tt> must
4217 be larger than the bit size of <tt>ty2</tt>. Equal sized types are not
4221 <p>The '<tt>trunc</tt>' instruction truncates the high order bits
4222 in <tt>value</tt> and converts the remaining bits to <tt>ty2</tt>. Since the
4223 source size must be larger than the destination size, <tt>trunc</tt> cannot
4224 be a <i>no-op cast</i>. It will always truncate bits.</p>
4228 %X = trunc i32 257 to i8 <i>; yields i8:1</i>
4229 %Y = trunc i32 123 to i1 <i>; yields i1:true</i>
4230 %Y = trunc i32 122 to i1 <i>; yields i1:false</i>
4235 <!-- _______________________________________________________________________ -->
4236 <div class="doc_subsubsection">
4237 <a name="i_zext">'<tt>zext .. to</tt>' Instruction</a>
4239 <div class="doc_text">
4243 <result> = zext <ty> <value> to <ty2> <i>; yields ty2</i>
4247 <p>The '<tt>zext</tt>' instruction zero extends its operand to type
4252 <p>The '<tt>zext</tt>' instruction takes a value to cast, which must be of
4253 <a href="#t_integer">integer</a> type, and a type to cast it to, which must
4254 also be of <a href="#t_integer">integer</a> type. The bit size of the
4255 <tt>value</tt> must be smaller than the bit size of the destination type,
4259 <p>The <tt>zext</tt> fills the high order bits of the <tt>value</tt> with zero
4260 bits until it reaches the size of the destination type, <tt>ty2</tt>.</p>
4262 <p>When zero extending from i1, the result will always be either 0 or 1.</p>
4266 %X = zext i32 257 to i64 <i>; yields i64:257</i>
4267 %Y = zext i1 true to i32 <i>; yields i32:1</i>
4272 <!-- _______________________________________________________________________ -->
4273 <div class="doc_subsubsection">
4274 <a name="i_sext">'<tt>sext .. to</tt>' Instruction</a>
4276 <div class="doc_text">
4280 <result> = sext <ty> <value> to <ty2> <i>; yields ty2</i>
4284 <p>The '<tt>sext</tt>' sign extends <tt>value</tt> to the type <tt>ty2</tt>.</p>
4287 <p>The '<tt>sext</tt>' instruction takes a value to cast, which must be of
4288 <a href="#t_integer">integer</a> type, and a type to cast it to, which must
4289 also be of <a href="#t_integer">integer</a> type. The bit size of the
4290 <tt>value</tt> must be smaller than the bit size of the destination type,
4294 <p>The '<tt>sext</tt>' instruction performs a sign extension by copying the sign
4295 bit (highest order bit) of the <tt>value</tt> until it reaches the bit size
4296 of the type <tt>ty2</tt>.</p>
4298 <p>When sign extending from i1, the extension always results in -1 or 0.</p>
4302 %X = sext i8 -1 to i16 <i>; yields i16 :65535</i>
4303 %Y = sext i1 true to i32 <i>; yields i32:-1</i>
4308 <!-- _______________________________________________________________________ -->
4309 <div class="doc_subsubsection">
4310 <a name="i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a>
4313 <div class="doc_text">
4317 <result> = fptrunc <ty> <value> to <ty2> <i>; yields ty2</i>
4321 <p>The '<tt>fptrunc</tt>' instruction truncates <tt>value</tt> to type
4325 <p>The '<tt>fptrunc</tt>' instruction takes a <a href="#t_floating">floating
4326 point</a> value to cast and a <a href="#t_floating">floating point</a> type
4327 to cast it to. The size of <tt>value</tt> must be larger than the size of
4328 <tt>ty2</tt>. This implies that <tt>fptrunc</tt> cannot be used to make a
4329 <i>no-op cast</i>.</p>
4332 <p>The '<tt>fptrunc</tt>' instruction truncates a <tt>value</tt> from a larger
4333 <a href="#t_floating">floating point</a> type to a smaller
4334 <a href="#t_floating">floating point</a> type. If the value cannot fit
4335 within the destination type, <tt>ty2</tt>, then the results are
4340 %X = fptrunc double 123.0 to float <i>; yields float:123.0</i>
4341 %Y = fptrunc double 1.0E+300 to float <i>; yields undefined</i>
4346 <!-- _______________________________________________________________________ -->
4347 <div class="doc_subsubsection">
4348 <a name="i_fpext">'<tt>fpext .. to</tt>' Instruction</a>
4350 <div class="doc_text">
4354 <result> = fpext <ty> <value> to <ty2> <i>; yields ty2</i>
4358 <p>The '<tt>fpext</tt>' extends a floating point <tt>value</tt> to a larger
4359 floating point value.</p>
4362 <p>The '<tt>fpext</tt>' instruction takes a
4363 <a href="#t_floating">floating point</a> <tt>value</tt> to cast, and
4364 a <a href="#t_floating">floating point</a> type to cast it to. The source
4365 type must be smaller than the destination type.</p>
4368 <p>The '<tt>fpext</tt>' instruction extends the <tt>value</tt> from a smaller
4369 <a href="#t_floating">floating point</a> type to a larger
4370 <a href="#t_floating">floating point</a> type. The <tt>fpext</tt> cannot be
4371 used to make a <i>no-op cast</i> because it always changes bits. Use
4372 <tt>bitcast</tt> to make a <i>no-op cast</i> for a floating point cast.</p>
4376 %X = fpext float 3.1415 to double <i>; yields double:3.1415</i>
4377 %Y = fpext float 1.0 to float <i>; yields float:1.0 (no-op)</i>
4382 <!-- _______________________________________________________________________ -->
4383 <div class="doc_subsubsection">
4384 <a name="i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a>
4386 <div class="doc_text">
4390 <result> = fptoui <ty> <value> to <ty2> <i>; yields ty2</i>
4394 <p>The '<tt>fptoui</tt>' converts a floating point <tt>value</tt> to its
4395 unsigned integer equivalent of type <tt>ty2</tt>.</p>
4398 <p>The '<tt>fptoui</tt>' instruction takes a value to cast, which must be a
4399 scalar or vector <a href="#t_floating">floating point</a> value, and a type
4400 to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
4401 type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
4402 vector integer type with the same number of elements as <tt>ty</tt></p>
4405 <p>The '<tt>fptoui</tt>' instruction converts its
4406 <a href="#t_floating">floating point</a> operand into the nearest (rounding
4407 towards zero) unsigned integer value. If the value cannot fit
4408 in <tt>ty2</tt>, the results are undefined.</p>
4412 %X = fptoui double 123.0 to i32 <i>; yields i32:123</i>
4413 %Y = fptoui float 1.0E+300 to i1 <i>; yields undefined:1</i>
4414 %X = fptoui float 1.04E+17 to i8 <i>; yields undefined:1</i>
4419 <!-- _______________________________________________________________________ -->
4420 <div class="doc_subsubsection">
4421 <a name="i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a>
4423 <div class="doc_text">
4427 <result> = fptosi <ty> <value> to <ty2> <i>; yields ty2</i>
4431 <p>The '<tt>fptosi</tt>' instruction converts
4432 <a href="#t_floating">floating point</a> <tt>value</tt> to
4433 type <tt>ty2</tt>.</p>
4436 <p>The '<tt>fptosi</tt>' instruction takes a value to cast, which must be a
4437 scalar or vector <a href="#t_floating">floating point</a> value, and a type
4438 to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
4439 type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
4440 vector integer type with the same number of elements as <tt>ty</tt></p>
4443 <p>The '<tt>fptosi</tt>' instruction converts its
4444 <a href="#t_floating">floating point</a> operand into the nearest (rounding
4445 towards zero) signed integer value. If the value cannot fit in <tt>ty2</tt>,
4446 the results are undefined.</p>
4450 %X = fptosi double -123.0 to i32 <i>; yields i32:-123</i>
4451 %Y = fptosi float 1.0E-247 to i1 <i>; yields undefined:1</i>
4452 %X = fptosi float 1.04E+17 to i8 <i>; yields undefined:1</i>
4457 <!-- _______________________________________________________________________ -->
4458 <div class="doc_subsubsection">
4459 <a name="i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a>
4461 <div class="doc_text">
4465 <result> = uitofp <ty> <value> to <ty2> <i>; yields ty2</i>
4469 <p>The '<tt>uitofp</tt>' instruction regards <tt>value</tt> as an unsigned
4470 integer and converts that value to the <tt>ty2</tt> type.</p>
4473 <p>The '<tt>uitofp</tt>' instruction takes a value to cast, which must be a
4474 scalar or vector <a href="#t_integer">integer</a> value, and a type to cast
4475 it to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
4476 type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
4477 floating point type with the same number of elements as <tt>ty</tt></p>
4480 <p>The '<tt>uitofp</tt>' instruction interprets its operand as an unsigned
4481 integer quantity and converts it to the corresponding floating point
4482 value. If the value cannot fit in the floating point value, the results are
4487 %X = uitofp i32 257 to float <i>; yields float:257.0</i>
4488 %Y = uitofp i8 -1 to double <i>; yields double:255.0</i>
4493 <!-- _______________________________________________________________________ -->
4494 <div class="doc_subsubsection">
4495 <a name="i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a>
4497 <div class="doc_text">
4501 <result> = sitofp <ty> <value> to <ty2> <i>; yields ty2</i>
4505 <p>The '<tt>sitofp</tt>' instruction regards <tt>value</tt> as a signed integer
4506 and converts that value to the <tt>ty2</tt> type.</p>
4509 <p>The '<tt>sitofp</tt>' instruction takes a value to cast, which must be a
4510 scalar or vector <a href="#t_integer">integer</a> value, and a type to cast
4511 it to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
4512 type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
4513 floating point type with the same number of elements as <tt>ty</tt></p>
4516 <p>The '<tt>sitofp</tt>' instruction interprets its operand as a signed integer
4517 quantity and converts it to the corresponding floating point value. If the
4518 value cannot fit in the floating point value, the results are undefined.</p>
4522 %X = sitofp i32 257 to float <i>; yields float:257.0</i>
4523 %Y = sitofp i8 -1 to double <i>; yields double:-1.0</i>
4528 <!-- _______________________________________________________________________ -->
4529 <div class="doc_subsubsection">
4530 <a name="i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a>
4532 <div class="doc_text">
4536 <result> = ptrtoint <ty> <value> to <ty2> <i>; yields ty2</i>
4540 <p>The '<tt>ptrtoint</tt>' instruction converts the pointer <tt>value</tt> to
4541 the integer type <tt>ty2</tt>.</p>
4544 <p>The '<tt>ptrtoint</tt>' instruction takes a <tt>value</tt> to cast, which
4545 must be a <a href="#t_pointer">pointer</a> value, and a type to cast it to
4546 <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a> type.</p>
4549 <p>The '<tt>ptrtoint</tt>' instruction converts <tt>value</tt> to integer type
4550 <tt>ty2</tt> by interpreting the pointer value as an integer and either
4551 truncating or zero extending that value to the size of the integer type. If
4552 <tt>value</tt> is smaller than <tt>ty2</tt> then a zero extension is done. If
4553 <tt>value</tt> is larger than <tt>ty2</tt> then a truncation is done. If they
4554 are the same size, then nothing is done (<i>no-op cast</i>) other than a type
4559 %X = ptrtoint i32* %X to i8 <i>; yields truncation on 32-bit architecture</i>
4560 %Y = ptrtoint i32* %x to i64 <i>; yields zero extension on 32-bit architecture</i>
4565 <!-- _______________________________________________________________________ -->
4566 <div class="doc_subsubsection">
4567 <a name="i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a>
4569 <div class="doc_text">
4573 <result> = inttoptr <ty> <value> to <ty2> <i>; yields ty2</i>
4577 <p>The '<tt>inttoptr</tt>' instruction converts an integer <tt>value</tt> to a
4578 pointer type, <tt>ty2</tt>.</p>
4581 <p>The '<tt>inttoptr</tt>' instruction takes an <a href="#t_integer">integer</a>
4582 value to cast, and a type to cast it to, which must be a
4583 <a href="#t_pointer">pointer</a> type.</p>
4586 <p>The '<tt>inttoptr</tt>' instruction converts <tt>value</tt> to type
4587 <tt>ty2</tt> by applying either a zero extension or a truncation depending on
4588 the size of the integer <tt>value</tt>. If <tt>value</tt> is larger than the
4589 size of a pointer then a truncation is done. If <tt>value</tt> is smaller
4590 than the size of a pointer then a zero extension is done. If they are the
4591 same size, nothing is done (<i>no-op cast</i>).</p>
4595 %X = inttoptr i32 255 to i32* <i>; yields zero extension on 64-bit architecture</i>
4596 %X = inttoptr i32 255 to i32* <i>; yields no-op on 32-bit architecture</i>
4597 %Y = inttoptr i64 0 to i32* <i>; yields truncation on 32-bit architecture</i>
4602 <!-- _______________________________________________________________________ -->
4603 <div class="doc_subsubsection">
4604 <a name="i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a>
4606 <div class="doc_text">
4610 <result> = bitcast <ty> <value> to <ty2> <i>; yields ty2</i>
4614 <p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
4615 <tt>ty2</tt> without changing any bits.</p>
4618 <p>The '<tt>bitcast</tt>' instruction takes a value to cast, which must be a
4619 non-aggregate first class value, and a type to cast it to, which must also be
4620 a non-aggregate <a href="#t_firstclass">first class</a> type. The bit sizes
4621 of <tt>value</tt> and the destination type, <tt>ty2</tt>, must be
4622 identical. If the source type is a pointer, the destination type must also be
4623 a pointer. This instruction supports bitwise conversion of vectors to
4624 integers and to vectors of other types (as long as they have the same
4628 <p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
4629 <tt>ty2</tt>. It is always a <i>no-op cast</i> because no bits change with
4630 this conversion. The conversion is done as if the <tt>value</tt> had been
4631 stored to memory and read back as type <tt>ty2</tt>. Pointer types may only
4632 be converted to other pointer types with this instruction. To convert
4633 pointers to other types, use the <a href="#i_inttoptr">inttoptr</a> or
4634 <a href="#i_ptrtoint">ptrtoint</a> instructions first.</p>
4638 %X = bitcast i8 255 to i8 <i>; yields i8 :-1</i>
4639 %Y = bitcast i32* %x to sint* <i>; yields sint*:%x</i>
4640 %Z = bitcast <2 x int> %V to i64; <i>; yields i64: %V</i>
4645 <!-- ======================================================================= -->
4646 <div class="doc_subsection"> <a name="otherops">Other Operations</a> </div>
4648 <div class="doc_text">
4650 <p>The instructions in this category are the "miscellaneous" instructions, which
4651 defy better classification.</p>
4655 <!-- _______________________________________________________________________ -->
4656 <div class="doc_subsubsection"><a name="i_icmp">'<tt>icmp</tt>' Instruction</a>
4659 <div class="doc_text">
4663 <result> = icmp <cond> <ty> <op1>, <op2> <i>; yields {i1} or {<N x i1>}:result</i>
4667 <p>The '<tt>icmp</tt>' instruction returns a boolean value or a vector of
4668 boolean values based on comparison of its two integer, integer vector, or
4669 pointer operands.</p>
4672 <p>The '<tt>icmp</tt>' instruction takes three operands. The first operand is
4673 the condition code indicating the kind of comparison to perform. It is not a
4674 value, just a keyword. The possible condition code are:</p>
4677 <li><tt>eq</tt>: equal</li>
4678 <li><tt>ne</tt>: not equal </li>
4679 <li><tt>ugt</tt>: unsigned greater than</li>
4680 <li><tt>uge</tt>: unsigned greater or equal</li>
4681 <li><tt>ult</tt>: unsigned less than</li>
4682 <li><tt>ule</tt>: unsigned less or equal</li>
4683 <li><tt>sgt</tt>: signed greater than</li>
4684 <li><tt>sge</tt>: signed greater or equal</li>
4685 <li><tt>slt</tt>: signed less than</li>
4686 <li><tt>sle</tt>: signed less or equal</li>
4689 <p>The remaining two arguments must be <a href="#t_integer">integer</a> or
4690 <a href="#t_pointer">pointer</a> or integer <a href="#t_vector">vector</a>
4691 typed. They must also be identical types.</p>
4694 <p>The '<tt>icmp</tt>' compares <tt>op1</tt> and <tt>op2</tt> according to the
4695 condition code given as <tt>cond</tt>. The comparison performed always yields
4696 either an <a href="#t_integer"><tt>i1</tt></a> or vector of <tt>i1</tt>
4697 result, as follows:</p>
4700 <li><tt>eq</tt>: yields <tt>true</tt> if the operands are equal,
4701 <tt>false</tt> otherwise. No sign interpretation is necessary or
4704 <li><tt>ne</tt>: yields <tt>true</tt> if the operands are unequal,
4705 <tt>false</tt> otherwise. No sign interpretation is necessary or
4708 <li><tt>ugt</tt>: interprets the operands as unsigned values and yields
4709 <tt>true</tt> if <tt>op1</tt> is greater than <tt>op2</tt>.</li>
4711 <li><tt>uge</tt>: interprets the operands as unsigned values and yields
4712 <tt>true</tt> if <tt>op1</tt> is greater than or equal
4713 to <tt>op2</tt>.</li>
4715 <li><tt>ult</tt>: interprets the operands as unsigned values and yields
4716 <tt>true</tt> if <tt>op1</tt> is less than <tt>op2</tt>.</li>
4718 <li><tt>ule</tt>: interprets the operands as unsigned values and yields
4719 <tt>true</tt> if <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
4721 <li><tt>sgt</tt>: interprets the operands as signed values and yields
4722 <tt>true</tt> if <tt>op1</tt> is greater than <tt>op2</tt>.</li>
4724 <li><tt>sge</tt>: interprets the operands as signed values and yields
4725 <tt>true</tt> if <tt>op1</tt> is greater than or equal
4726 to <tt>op2</tt>.</li>
4728 <li><tt>slt</tt>: interprets the operands as signed values and yields
4729 <tt>true</tt> if <tt>op1</tt> is less than <tt>op2</tt>.</li>
4731 <li><tt>sle</tt>: interprets the operands as signed values and yields
4732 <tt>true</tt> if <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
4735 <p>If the operands are <a href="#t_pointer">pointer</a> typed, the pointer
4736 values are compared as if they were integers.</p>
4738 <p>If the operands are integer vectors, then they are compared element by
4739 element. The result is an <tt>i1</tt> vector with the same number of elements
4740 as the values being compared. Otherwise, the result is an <tt>i1</tt>.</p>
4744 <result> = icmp eq i32 4, 5 <i>; yields: result=false</i>
4745 <result> = icmp ne float* %X, %X <i>; yields: result=false</i>
4746 <result> = icmp ult i16 4, 5 <i>; yields: result=true</i>
4747 <result> = icmp sgt i16 4, 5 <i>; yields: result=false</i>
4748 <result> = icmp ule i16 -4, 5 <i>; yields: result=false</i>
4749 <result> = icmp sge i16 4, 5 <i>; yields: result=false</i>
4752 <p>Note that the code generator does not yet support vector types with
4753 the <tt>icmp</tt> instruction.</p>
4757 <!-- _______________________________________________________________________ -->
4758 <div class="doc_subsubsection"><a name="i_fcmp">'<tt>fcmp</tt>' Instruction</a>
4761 <div class="doc_text">
4765 <result> = fcmp <cond> <ty> <op1>, <op2> <i>; yields {i1} or {<N x i1>}:result</i>
4769 <p>The '<tt>fcmp</tt>' instruction returns a boolean value or vector of boolean
4770 values based on comparison of its operands.</p>
4772 <p>If the operands are floating point scalars, then the result type is a boolean
4773 (<a href="#t_integer"><tt>i1</tt></a>).</p>
4775 <p>If the operands are floating point vectors, then the result type is a vector
4776 of boolean with the same number of elements as the operands being
4780 <p>The '<tt>fcmp</tt>' instruction takes three operands. The first operand is
4781 the condition code indicating the kind of comparison to perform. It is not a
4782 value, just a keyword. The possible condition code are:</p>
4785 <li><tt>false</tt>: no comparison, always returns false</li>
4786 <li><tt>oeq</tt>: ordered and equal</li>
4787 <li><tt>ogt</tt>: ordered and greater than </li>
4788 <li><tt>oge</tt>: ordered and greater than or equal</li>
4789 <li><tt>olt</tt>: ordered and less than </li>
4790 <li><tt>ole</tt>: ordered and less than or equal</li>
4791 <li><tt>one</tt>: ordered and not equal</li>
4792 <li><tt>ord</tt>: ordered (no nans)</li>
4793 <li><tt>ueq</tt>: unordered or equal</li>
4794 <li><tt>ugt</tt>: unordered or greater than </li>
4795 <li><tt>uge</tt>: unordered or greater than or equal</li>
4796 <li><tt>ult</tt>: unordered or less than </li>
4797 <li><tt>ule</tt>: unordered or less than or equal</li>
4798 <li><tt>une</tt>: unordered or not equal</li>
4799 <li><tt>uno</tt>: unordered (either nans)</li>
4800 <li><tt>true</tt>: no comparison, always returns true</li>
4803 <p><i>Ordered</i> means that neither operand is a QNAN while
4804 <i>unordered</i> means that either operand may be a QNAN.</p>
4806 <p>Each of <tt>val1</tt> and <tt>val2</tt> arguments must be either
4807 a <a href="#t_floating">floating point</a> type or
4808 a <a href="#t_vector">vector</a> of floating point type. They must have
4809 identical types.</p>
4812 <p>The '<tt>fcmp</tt>' instruction compares <tt>op1</tt> and <tt>op2</tt>
4813 according to the condition code given as <tt>cond</tt>. If the operands are
4814 vectors, then the vectors are compared element by element. Each comparison
4815 performed always yields an <a href="#t_integer">i1</a> result, as
4819 <li><tt>false</tt>: always yields <tt>false</tt>, regardless of operands.</li>
4821 <li><tt>oeq</tt>: yields <tt>true</tt> if both operands are not a QNAN and
4822 <tt>op1</tt> is equal to <tt>op2</tt>.</li>
4824 <li><tt>ogt</tt>: yields <tt>true</tt> if both operands are not a QNAN and
4825 <tt>op1</tt> is greather than <tt>op2</tt>.</li>
4827 <li><tt>oge</tt>: yields <tt>true</tt> if both operands are not a QNAN and
4828 <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
4830 <li><tt>olt</tt>: yields <tt>true</tt> if both operands are not a QNAN and
4831 <tt>op1</tt> is less than <tt>op2</tt>.</li>
4833 <li><tt>ole</tt>: yields <tt>true</tt> if both operands are not a QNAN and
4834 <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
4836 <li><tt>one</tt>: yields <tt>true</tt> if both operands are not a QNAN and
4837 <tt>op1</tt> is not equal to <tt>op2</tt>.</li>
4839 <li><tt>ord</tt>: yields <tt>true</tt> if both operands are not a QNAN.</li>
4841 <li><tt>ueq</tt>: yields <tt>true</tt> if either operand is a QNAN or
4842 <tt>op1</tt> is equal to <tt>op2</tt>.</li>
4844 <li><tt>ugt</tt>: yields <tt>true</tt> if either operand is a QNAN or
4845 <tt>op1</tt> is greater than <tt>op2</tt>.</li>
4847 <li><tt>uge</tt>: yields <tt>true</tt> if either operand is a QNAN or
4848 <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
4850 <li><tt>ult</tt>: yields <tt>true</tt> if either operand is a QNAN or
4851 <tt>op1</tt> is less than <tt>op2</tt>.</li>
4853 <li><tt>ule</tt>: yields <tt>true</tt> if either operand is a QNAN or
4854 <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
4856 <li><tt>une</tt>: yields <tt>true</tt> if either operand is a QNAN or
4857 <tt>op1</tt> is not equal to <tt>op2</tt>.</li>
4859 <li><tt>uno</tt>: yields <tt>true</tt> if either operand is a QNAN.</li>
4861 <li><tt>true</tt>: always yields <tt>true</tt>, regardless of operands.</li>
4866 <result> = fcmp oeq float 4.0, 5.0 <i>; yields: result=false</i>
4867 <result> = fcmp one float 4.0, 5.0 <i>; yields: result=true</i>
4868 <result> = fcmp olt float 4.0, 5.0 <i>; yields: result=true</i>
4869 <result> = fcmp ueq double 1.0, 2.0 <i>; yields: result=false</i>
4872 <p>Note that the code generator does not yet support vector types with
4873 the <tt>fcmp</tt> instruction.</p>
4877 <!-- _______________________________________________________________________ -->
4878 <div class="doc_subsubsection">
4879 <a name="i_phi">'<tt>phi</tt>' Instruction</a>
4882 <div class="doc_text">
4886 <result> = phi <ty> [ <val0>, <label0>], ...
4890 <p>The '<tt>phi</tt>' instruction is used to implement the φ node in the
4891 SSA graph representing the function.</p>
4894 <p>The type of the incoming values is specified with the first type field. After
4895 this, the '<tt>phi</tt>' instruction takes a list of pairs as arguments, with
4896 one pair for each predecessor basic block of the current block. Only values
4897 of <a href="#t_firstclass">first class</a> type may be used as the value
4898 arguments to the PHI node. Only labels may be used as the label
4901 <p>There must be no non-phi instructions between the start of a basic block and
4902 the PHI instructions: i.e. PHI instructions must be first in a basic
4905 <p>For the purposes of the SSA form, the use of each incoming value is deemed to
4906 occur on the edge from the corresponding predecessor block to the current
4907 block (but after any definition of an '<tt>invoke</tt>' instruction's return
4908 value on the same edge).</p>
4911 <p>At runtime, the '<tt>phi</tt>' instruction logically takes on the value
4912 specified by the pair corresponding to the predecessor basic block that
4913 executed just prior to the current block.</p>
4917 Loop: ; Infinite loop that counts from 0 on up...
4918 %indvar = phi i32 [ 0, %LoopHeader ], [ %nextindvar, %Loop ]
4919 %nextindvar = add i32 %indvar, 1
4925 <!-- _______________________________________________________________________ -->
4926 <div class="doc_subsubsection">
4927 <a name="i_select">'<tt>select</tt>' Instruction</a>
4930 <div class="doc_text">
4934 <result> = select <i>selty</i> <cond>, <ty> <val1>, <ty> <val2> <i>; yields ty</i>
4936 <i>selty</i> is either i1 or {<N x i1>}
4940 <p>The '<tt>select</tt>' instruction is used to choose one value based on a
4941 condition, without branching.</p>
4945 <p>The '<tt>select</tt>' instruction requires an 'i1' value or a vector of 'i1'
4946 values indicating the condition, and two values of the
4947 same <a href="#t_firstclass">first class</a> type. If the val1/val2 are
4948 vectors and the condition is a scalar, then entire vectors are selected, not
4949 individual elements.</p>
4952 <p>If the condition is an i1 and it evaluates to 1, the instruction returns the
4953 first value argument; otherwise, it returns the second value argument.</p>
4955 <p>If the condition is a vector of i1, then the value arguments must be vectors
4956 of the same size, and the selection is done element by element.</p>
4960 %X = select i1 true, i8 17, i8 42 <i>; yields i8:17</i>
4963 <p>Note that the code generator does not yet support conditions
4964 with vector type.</p>
4968 <!-- _______________________________________________________________________ -->
4969 <div class="doc_subsubsection">
4970 <a name="i_call">'<tt>call</tt>' Instruction</a>
4973 <div class="doc_text">
4977 <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>]
4981 <p>The '<tt>call</tt>' instruction represents a simple function call.</p>
4984 <p>This instruction requires several arguments:</p>
4987 <li>The optional "tail" marker indicates whether the callee function accesses
4988 any allocas or varargs in the caller. If the "tail" marker is present,
4989 the function call is eligible for tail call optimization. Note that calls
4990 may be marked "tail" even if they do not occur before
4991 a <a href="#i_ret"><tt>ret</tt></a> instruction.</li>
4993 <li>The optional "cconv" marker indicates which <a href="#callingconv">calling
4994 convention</a> the call should use. If none is specified, the call
4995 defaults to using C calling conventions.</li>
4997 <li>The optional <a href="#paramattrs">Parameter Attributes</a> list for
4998 return values. Only '<tt>zeroext</tt>', '<tt>signext</tt>', and
4999 '<tt>inreg</tt>' attributes are valid here.</li>
5001 <li>'<tt>ty</tt>': the type of the call instruction itself which is also the
5002 type of the return value. Functions that return no value are marked
5003 <tt><a href="#t_void">void</a></tt>.</li>
5005 <li>'<tt>fnty</tt>': shall be the signature of the pointer to function value
5006 being invoked. The argument types must match the types implied by this
5007 signature. This type can be omitted if the function is not varargs and if
5008 the function type does not return a pointer to a function.</li>
5010 <li>'<tt>fnptrval</tt>': An LLVM value containing a pointer to a function to
5011 be invoked. In most cases, this is a direct function invocation, but
5012 indirect <tt>call</tt>s are just as possible, calling an arbitrary pointer
5013 to function value.</li>
5015 <li>'<tt>function args</tt>': argument list whose types match the function
5016 signature argument types. All arguments must be of
5017 <a href="#t_firstclass">first class</a> type. If the function signature
5018 indicates the function accepts a variable number of arguments, the extra
5019 arguments can be specified.</li>
5021 <li>The optional <a href="#fnattrs">function attributes</a> list. Only
5022 '<tt>noreturn</tt>', '<tt>nounwind</tt>', '<tt>readonly</tt>' and
5023 '<tt>readnone</tt>' attributes are valid here.</li>
5027 <p>The '<tt>call</tt>' instruction is used to cause control flow to transfer to
5028 a specified function, with its incoming arguments bound to the specified
5029 values. Upon a '<tt><a href="#i_ret">ret</a></tt>' instruction in the called
5030 function, control flow continues with the instruction after the function
5031 call, and the return value of the function is bound to the result
5036 %retval = call i32 @test(i32 %argc)
5037 call i32 (i8 *, ...)* @printf(i8 * %msg, i32 12, i8 42) <i>; yields i32</i>
5038 %X = tail call i32 @foo() <i>; yields i32</i>
5039 %Y = tail call <a href="#callingconv">fastcc</a> i32 @foo() <i>; yields i32</i>
5040 call void %foo(i8 97 signext)
5042 %struct.A = type { i32, i8 }
5043 %r = call %struct.A @foo() <i>; yields { 32, i8 }</i>
5044 %gr = extractvalue %struct.A %r, 0 <i>; yields i32</i>
5045 %gr1 = extractvalue %struct.A %r, 1 <i>; yields i8</i>
5046 %Z = call void @foo() noreturn <i>; indicates that %foo never returns normally</i>
5047 %ZZ = call zeroext i32 @bar() <i>; Return value is %zero extended</i>
5050 <p>llvm treats calls to some functions with names and arguments that match the
5051 standard C99 library as being the C99 library functions, and may perform
5052 optimizations or generate code for them under that assumption. This is
5053 something we'd like to change in the future to provide better support for
5054 freestanding environments and non-C-based langauges.</p>
5058 <!-- _______________________________________________________________________ -->
5059 <div class="doc_subsubsection">
5060 <a name="i_va_arg">'<tt>va_arg</tt>' Instruction</a>
5063 <div class="doc_text">
5067 <resultval> = va_arg <va_list*> <arglist>, <argty>
5071 <p>The '<tt>va_arg</tt>' instruction is used to access arguments passed through
5072 the "variable argument" area of a function call. It is used to implement the
5073 <tt>va_arg</tt> macro in C.</p>
5076 <p>This instruction takes a <tt>va_list*</tt> value and the type of the
5077 argument. It returns a value of the specified argument type and increments
5078 the <tt>va_list</tt> to point to the next argument. The actual type
5079 of <tt>va_list</tt> is target specific.</p>
5082 <p>The '<tt>va_arg</tt>' instruction loads an argument of the specified type
5083 from the specified <tt>va_list</tt> and causes the <tt>va_list</tt> to point
5084 to the next argument. For more information, see the variable argument
5085 handling <a href="#int_varargs">Intrinsic Functions</a>.</p>
5087 <p>It is legal for this instruction to be called in a function which does not
5088 take a variable number of arguments, for example, the <tt>vfprintf</tt>
5091 <p><tt>va_arg</tt> is an LLVM instruction instead of
5092 an <a href="#intrinsics">intrinsic function</a> because it takes a type as an
5096 <p>See the <a href="#int_varargs">variable argument processing</a> section.</p>
5098 <p>Note that the code generator does not yet fully support va_arg on many
5099 targets. Also, it does not currently support va_arg with aggregate types on
5104 <!-- *********************************************************************** -->
5105 <div class="doc_section"> <a name="intrinsics">Intrinsic Functions</a> </div>
5106 <!-- *********************************************************************** -->
5108 <div class="doc_text">
5110 <p>LLVM supports the notion of an "intrinsic function". These functions have
5111 well known names and semantics and are required to follow certain
5112 restrictions. Overall, these intrinsics represent an extension mechanism for
5113 the LLVM language that does not require changing all of the transformations
5114 in LLVM when adding to the language (or the bitcode reader/writer, the
5115 parser, etc...).</p>
5117 <p>Intrinsic function names must all start with an "<tt>llvm.</tt>" prefix. This
5118 prefix is reserved in LLVM for intrinsic names; thus, function names may not
5119 begin with this prefix. Intrinsic functions must always be external
5120 functions: you cannot define the body of intrinsic functions. Intrinsic
5121 functions may only be used in call or invoke instructions: it is illegal to
5122 take the address of an intrinsic function. Additionally, because intrinsic
5123 functions are part of the LLVM language, it is required if any are added that
5124 they be documented here.</p>
5126 <p>Some intrinsic functions can be overloaded, i.e., the intrinsic represents a
5127 family of functions that perform the same operation but on different data
5128 types. Because LLVM can represent over 8 million different integer types,
5129 overloading is used commonly to allow an intrinsic function to operate on any
5130 integer type. One or more of the argument types or the result type can be
5131 overloaded to accept any integer type. Argument types may also be defined as
5132 exactly matching a previous argument's type or the result type. This allows
5133 an intrinsic function which accepts multiple arguments, but needs all of them
5134 to be of the same type, to only be overloaded with respect to a single
5135 argument or the result.</p>
5137 <p>Overloaded intrinsics will have the names of its overloaded argument types
5138 encoded into its function name, each preceded by a period. Only those types
5139 which are overloaded result in a name suffix. Arguments whose type is matched
5140 against another type do not. For example, the <tt>llvm.ctpop</tt> function
5141 can take an integer of any width and returns an integer of exactly the same
5142 integer width. This leads to a family of functions such as
5143 <tt>i8 @llvm.ctpop.i8(i8 %val)</tt> and <tt>i29 @llvm.ctpop.i29(i29
5144 %val)</tt>. Only one type, the return type, is overloaded, and only one type
5145 suffix is required. Because the argument's type is matched against the return
5146 type, it does not require its own name suffix.</p>
5148 <p>To learn how to add an intrinsic function, please see the
5149 <a href="ExtendingLLVM.html">Extending LLVM Guide</a>.</p>
5153 <!-- ======================================================================= -->
5154 <div class="doc_subsection">
5155 <a name="int_varargs">Variable Argument Handling Intrinsics</a>
5158 <div class="doc_text">
5160 <p>Variable argument support is defined in LLVM with
5161 the <a href="#i_va_arg"><tt>va_arg</tt></a> instruction and these three
5162 intrinsic functions. These functions are related to the similarly named
5163 macros defined in the <tt><stdarg.h></tt> header file.</p>
5165 <p>All of these functions operate on arguments that use a target-specific value
5166 type "<tt>va_list</tt>". The LLVM assembly language reference manual does
5167 not define what this type is, so all transformations should be prepared to
5168 handle these functions regardless of the type used.</p>
5170 <p>This example shows how the <a href="#i_va_arg"><tt>va_arg</tt></a>
5171 instruction and the variable argument handling intrinsic functions are
5174 <div class="doc_code">
5176 define i32 @test(i32 %X, ...) {
5177 ; Initialize variable argument processing
5179 %ap2 = bitcast i8** %ap to i8*
5180 call void @llvm.va_start(i8* %ap2)
5182 ; Read a single integer argument
5183 %tmp = va_arg i8** %ap, i32
5185 ; Demonstrate usage of llvm.va_copy and llvm.va_end
5187 %aq2 = bitcast i8** %aq to i8*
5188 call void @llvm.va_copy(i8* %aq2, i8* %ap2)
5189 call void @llvm.va_end(i8* %aq2)
5191 ; Stop processing of arguments.
5192 call void @llvm.va_end(i8* %ap2)
5196 declare void @llvm.va_start(i8*)
5197 declare void @llvm.va_copy(i8*, i8*)
5198 declare void @llvm.va_end(i8*)
5204 <!-- _______________________________________________________________________ -->
5205 <div class="doc_subsubsection">
5206 <a name="int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a>
5210 <div class="doc_text">
5214 declare void %llvm.va_start(i8* <arglist>)
5218 <p>The '<tt>llvm.va_start</tt>' intrinsic initializes <tt>*<arglist></tt>
5219 for subsequent use by <tt><a href="#i_va_arg">va_arg</a></tt>.</p>
5222 <p>The argument is a pointer to a <tt>va_list</tt> element to initialize.</p>
5225 <p>The '<tt>llvm.va_start</tt>' intrinsic works just like the <tt>va_start</tt>
5226 macro available in C. In a target-dependent way, it initializes
5227 the <tt>va_list</tt> element to which the argument points, so that the next
5228 call to <tt>va_arg</tt> will produce the first variable argument passed to
5229 the function. Unlike the C <tt>va_start</tt> macro, this intrinsic does not
5230 need to know the last argument of the function as the compiler can figure
5235 <!-- _______________________________________________________________________ -->
5236 <div class="doc_subsubsection">
5237 <a name="int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a>
5240 <div class="doc_text">
5244 declare void @llvm.va_end(i8* <arglist>)
5248 <p>The '<tt>llvm.va_end</tt>' intrinsic destroys <tt>*<arglist></tt>,
5249 which has been initialized previously
5250 with <tt><a href="#int_va_start">llvm.va_start</a></tt>
5251 or <tt><a href="#i_va_copy">llvm.va_copy</a></tt>.</p>
5254 <p>The argument is a pointer to a <tt>va_list</tt> to destroy.</p>
5257 <p>The '<tt>llvm.va_end</tt>' intrinsic works just like the <tt>va_end</tt>
5258 macro available in C. In a target-dependent way, it destroys
5259 the <tt>va_list</tt> element to which the argument points. Calls
5260 to <a href="#int_va_start"><tt>llvm.va_start</tt></a>
5261 and <a href="#int_va_copy"> <tt>llvm.va_copy</tt></a> must be matched exactly
5262 with calls to <tt>llvm.va_end</tt>.</p>
5266 <!-- _______________________________________________________________________ -->
5267 <div class="doc_subsubsection">
5268 <a name="int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a>
5271 <div class="doc_text">
5275 declare void @llvm.va_copy(i8* <destarglist>, i8* <srcarglist>)
5279 <p>The '<tt>llvm.va_copy</tt>' intrinsic copies the current argument position
5280 from the source argument list to the destination argument list.</p>
5283 <p>The first argument is a pointer to a <tt>va_list</tt> element to initialize.
5284 The second argument is a pointer to a <tt>va_list</tt> element to copy
5288 <p>The '<tt>llvm.va_copy</tt>' intrinsic works just like the <tt>va_copy</tt>
5289 macro available in C. In a target-dependent way, it copies the
5290 source <tt>va_list</tt> element into the destination <tt>va_list</tt>
5291 element. This intrinsic is necessary because
5292 the <tt><a href="#int_va_start"> llvm.va_start</a></tt> intrinsic may be
5293 arbitrarily complex and require, for example, memory allocation.</p>
5297 <!-- ======================================================================= -->
5298 <div class="doc_subsection">
5299 <a name="int_gc">Accurate Garbage Collection Intrinsics</a>
5302 <div class="doc_text">
5304 <p>LLVM support for <a href="GarbageCollection.html">Accurate Garbage
5305 Collection</a> (GC) requires the implementation and generation of these
5306 intrinsics. These intrinsics allow identification of <a href="#int_gcroot">GC
5307 roots on the stack</a>, as well as garbage collector implementations that
5308 require <a href="#int_gcread">read</a> and <a href="#int_gcwrite">write</a>
5309 barriers. Front-ends for type-safe garbage collected languages should generate
5310 these intrinsics to make use of the LLVM garbage collectors. For more details,
5311 see <a href="GarbageCollection.html">Accurate Garbage Collection with
5314 <p>The garbage collection intrinsics only operate on objects in the generic
5315 address space (address space zero).</p>
5319 <!-- _______________________________________________________________________ -->
5320 <div class="doc_subsubsection">
5321 <a name="int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a>
5324 <div class="doc_text">
5328 declare void @llvm.gcroot(i8** %ptrloc, i8* %metadata)
5332 <p>The '<tt>llvm.gcroot</tt>' intrinsic declares the existence of a GC root to
5333 the code generator, and allows some metadata to be associated with it.</p>
5336 <p>The first argument specifies the address of a stack object that contains the
5337 root pointer. The second pointer (which must be either a constant or a
5338 global value address) contains the meta-data to be associated with the
5342 <p>At runtime, a call to this intrinsic stores a null pointer into the "ptrloc"
5343 location. At compile-time, the code generator generates information to allow
5344 the runtime to find the pointer at GC safe points. The '<tt>llvm.gcroot</tt>'
5345 intrinsic may only be used in a function which <a href="#gc">specifies a GC
5350 <!-- _______________________________________________________________________ -->
5351 <div class="doc_subsubsection">
5352 <a name="int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a>
5355 <div class="doc_text">
5359 declare i8* @llvm.gcread(i8* %ObjPtr, i8** %Ptr)
5363 <p>The '<tt>llvm.gcread</tt>' intrinsic identifies reads of references from heap
5364 locations, allowing garbage collector implementations that require read
5368 <p>The second argument is the address to read from, which should be an address
5369 allocated from the garbage collector. The first object is a pointer to the
5370 start of the referenced object, if needed by the language runtime (otherwise
5374 <p>The '<tt>llvm.gcread</tt>' intrinsic has the same semantics as a load
5375 instruction, but may be replaced with substantially more complex code by the
5376 garbage collector runtime, as needed. The '<tt>llvm.gcread</tt>' intrinsic
5377 may only be used in a function which <a href="#gc">specifies a GC
5382 <!-- _______________________________________________________________________ -->
5383 <div class="doc_subsubsection">
5384 <a name="int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a>
5387 <div class="doc_text">
5391 declare void @llvm.gcwrite(i8* %P1, i8* %Obj, i8** %P2)
5395 <p>The '<tt>llvm.gcwrite</tt>' intrinsic identifies writes of references to heap
5396 locations, allowing garbage collector implementations that require write
5397 barriers (such as generational or reference counting collectors).</p>
5400 <p>The first argument is the reference to store, the second is the start of the
5401 object to store it to, and the third is the address of the field of Obj to
5402 store to. If the runtime does not require a pointer to the object, Obj may
5406 <p>The '<tt>llvm.gcwrite</tt>' intrinsic has the same semantics as a store
5407 instruction, but may be replaced with substantially more complex code by the
5408 garbage collector runtime, as needed. The '<tt>llvm.gcwrite</tt>' intrinsic
5409 may only be used in a function which <a href="#gc">specifies a GC
5414 <!-- ======================================================================= -->
5415 <div class="doc_subsection">
5416 <a name="int_codegen">Code Generator Intrinsics</a>
5419 <div class="doc_text">
5421 <p>These intrinsics are provided by LLVM to expose special features that may
5422 only be implemented with code generator support.</p>
5426 <!-- _______________________________________________________________________ -->
5427 <div class="doc_subsubsection">
5428 <a name="int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a>
5431 <div class="doc_text">
5435 declare i8 *@llvm.returnaddress(i32 <level>)
5439 <p>The '<tt>llvm.returnaddress</tt>' intrinsic attempts to compute a
5440 target-specific value indicating the return address of the current function
5441 or one of its callers.</p>
5444 <p>The argument to this intrinsic indicates which function to return the address
5445 for. Zero indicates the calling function, one indicates its caller, etc.
5446 The argument is <b>required</b> to be a constant integer value.</p>
5449 <p>The '<tt>llvm.returnaddress</tt>' intrinsic either returns a pointer
5450 indicating the return address of the specified call frame, or zero if it
5451 cannot be identified. The value returned by this intrinsic is likely to be
5452 incorrect or 0 for arguments other than zero, so it should only be used for
5453 debugging purposes.</p>
5455 <p>Note that calling this intrinsic does not prevent function inlining or other
5456 aggressive transformations, so the value returned may not be that of the
5457 obvious source-language caller.</p>
5461 <!-- _______________________________________________________________________ -->
5462 <div class="doc_subsubsection">
5463 <a name="int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a>
5466 <div class="doc_text">
5470 declare i8 *@llvm.frameaddress(i32 <level>)
5474 <p>The '<tt>llvm.frameaddress</tt>' intrinsic attempts to return the
5475 target-specific frame pointer value for the specified stack frame.</p>
5478 <p>The argument to this intrinsic indicates which function to return the frame
5479 pointer for. Zero indicates the calling function, one indicates its caller,
5480 etc. The argument is <b>required</b> to be a constant integer value.</p>
5483 <p>The '<tt>llvm.frameaddress</tt>' intrinsic either returns a pointer
5484 indicating the frame address of the specified call frame, or zero if it
5485 cannot be identified. The value returned by this intrinsic is likely to be
5486 incorrect or 0 for arguments other than zero, so it should only be used for
5487 debugging purposes.</p>
5489 <p>Note that calling this intrinsic does not prevent function inlining or other
5490 aggressive transformations, so the value returned may not be that of the
5491 obvious source-language caller.</p>
5495 <!-- _______________________________________________________________________ -->
5496 <div class="doc_subsubsection">
5497 <a name="int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a>
5500 <div class="doc_text">
5504 declare i8 *@llvm.stacksave()
5508 <p>The '<tt>llvm.stacksave</tt>' intrinsic is used to remember the current state
5509 of the function stack, for use
5510 with <a href="#int_stackrestore"> <tt>llvm.stackrestore</tt></a>. This is
5511 useful for implementing language features like scoped automatic variable
5512 sized arrays in C99.</p>
5515 <p>This intrinsic returns a opaque pointer value that can be passed
5516 to <a href="#int_stackrestore"><tt>llvm.stackrestore</tt></a>. When
5517 an <tt>llvm.stackrestore</tt> intrinsic is executed with a value saved
5518 from <tt>llvm.stacksave</tt>, it effectively restores the state of the stack
5519 to the state it was in when the <tt>llvm.stacksave</tt> intrinsic executed.
5520 In practice, this pops any <a href="#i_alloca">alloca</a> blocks from the
5521 stack that were allocated after the <tt>llvm.stacksave</tt> was executed.</p>
5525 <!-- _______________________________________________________________________ -->
5526 <div class="doc_subsubsection">
5527 <a name="int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a>
5530 <div class="doc_text">
5534 declare void @llvm.stackrestore(i8 * %ptr)
5538 <p>The '<tt>llvm.stackrestore</tt>' intrinsic is used to restore the state of
5539 the function stack to the state it was in when the
5540 corresponding <a href="#int_stacksave"><tt>llvm.stacksave</tt></a> intrinsic
5541 executed. This is useful for implementing language features like scoped
5542 automatic variable sized arrays in C99.</p>
5545 <p>See the description
5546 for <a href="#int_stacksave"><tt>llvm.stacksave</tt></a>.</p>
5550 <!-- _______________________________________________________________________ -->
5551 <div class="doc_subsubsection">
5552 <a name="int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a>
5555 <div class="doc_text">
5559 declare void @llvm.prefetch(i8* <address>, i32 <rw>, i32 <locality>)
5563 <p>The '<tt>llvm.prefetch</tt>' intrinsic is a hint to the code generator to
5564 insert a prefetch instruction if supported; otherwise, it is a noop.
5565 Prefetches have no effect on the behavior of the program but can change its
5566 performance characteristics.</p>
5569 <p><tt>address</tt> is the address to be prefetched, <tt>rw</tt> is the
5570 specifier determining if the fetch should be for a read (0) or write (1),
5571 and <tt>locality</tt> is a temporal locality specifier ranging from (0) - no
5572 locality, to (3) - extremely local keep in cache. The <tt>rw</tt>
5573 and <tt>locality</tt> arguments must be constant integers.</p>
5576 <p>This intrinsic does not modify the behavior of the program. In particular,
5577 prefetches cannot trap and do not produce a value. On targets that support
5578 this intrinsic, the prefetch can provide hints to the processor cache for
5579 better performance.</p>
5583 <!-- _______________________________________________________________________ -->
5584 <div class="doc_subsubsection">
5585 <a name="int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a>
5588 <div class="doc_text">
5592 declare void @llvm.pcmarker(i32 <id>)
5596 <p>The '<tt>llvm.pcmarker</tt>' intrinsic is a method to export a Program
5597 Counter (PC) in a region of code to simulators and other tools. The method
5598 is target specific, but it is expected that the marker will use exported
5599 symbols to transmit the PC of the marker. The marker makes no guarantees
5600 that it will remain with any specific instruction after optimizations. It is
5601 possible that the presence of a marker will inhibit optimizations. The
5602 intended use is to be inserted after optimizations to allow correlations of
5603 simulation runs.</p>
5606 <p><tt>id</tt> is a numerical id identifying the marker.</p>
5609 <p>This intrinsic does not modify the behavior of the program. Backends that do
5610 not support this intrinisic may ignore it.</p>
5614 <!-- _______________________________________________________________________ -->
5615 <div class="doc_subsubsection">
5616 <a name="int_readcyclecounter">'<tt>llvm.readcyclecounter</tt>' Intrinsic</a>
5619 <div class="doc_text">
5623 declare i64 @llvm.readcyclecounter( )
5627 <p>The '<tt>llvm.readcyclecounter</tt>' intrinsic provides access to the cycle
5628 counter register (or similar low latency, high accuracy clocks) on those
5629 targets that support it. On X86, it should map to RDTSC. On Alpha, it
5630 should map to RPCC. As the backing counters overflow quickly (on the order
5631 of 9 seconds on alpha), this should only be used for small timings.</p>
5634 <p>When directly supported, reading the cycle counter should not modify any
5635 memory. Implementations are allowed to either return a application specific
5636 value or a system wide value. On backends without support, this is lowered
5637 to a constant 0.</p>
5641 <!-- ======================================================================= -->
5642 <div class="doc_subsection">
5643 <a name="int_libc">Standard C Library Intrinsics</a>
5646 <div class="doc_text">
5648 <p>LLVM provides intrinsics for a few important standard C library functions.
5649 These intrinsics allow source-language front-ends to pass information about
5650 the alignment of the pointer arguments to the code generator, providing
5651 opportunity for more efficient code generation.</p>
5655 <!-- _______________________________________________________________________ -->
5656 <div class="doc_subsubsection">
5657 <a name="int_memcpy">'<tt>llvm.memcpy</tt>' Intrinsic</a>
5660 <div class="doc_text">
5663 <p>This is an overloaded intrinsic. You can use <tt>llvm.memcpy</tt> on any
5664 integer bit width. Not all targets support all bit widths however.</p>
5667 declare void @llvm.memcpy.i8(i8 * <dest>, i8 * <src>,
5668 i8 <len>, i32 <align>)
5669 declare void @llvm.memcpy.i16(i8 * <dest>, i8 * <src>,
5670 i16 <len>, i32 <align>)
5671 declare void @llvm.memcpy.i32(i8 * <dest>, i8 * <src>,
5672 i32 <len>, i32 <align>)
5673 declare void @llvm.memcpy.i64(i8 * <dest>, i8 * <src>,
5674 i64 <len>, i32 <align>)
5678 <p>The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the
5679 source location to the destination location.</p>
5681 <p>Note that, unlike the standard libc function, the <tt>llvm.memcpy.*</tt>
5682 intrinsics do not return a value, and takes an extra alignment argument.</p>
5685 <p>The first argument is a pointer to the destination, the second is a pointer
5686 to the source. The third argument is an integer argument specifying the
5687 number of bytes to copy, and the fourth argument is the alignment of the
5688 source and destination locations.</p>
5690 <p>If the call to this intrinisic has an alignment value that is not 0 or 1,
5691 then the caller guarantees that both the source and destination pointers are
5692 aligned to that boundary.</p>
5695 <p>The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the
5696 source location to the destination location, which are not allowed to
5697 overlap. It copies "len" bytes of memory over. If the argument is known to
5698 be aligned to some boundary, this can be specified as the fourth argument,
5699 otherwise it should be set to 0 or 1.</p>
5703 <!-- _______________________________________________________________________ -->
5704 <div class="doc_subsubsection">
5705 <a name="int_memmove">'<tt>llvm.memmove</tt>' Intrinsic</a>
5708 <div class="doc_text">
5711 <p>This is an overloaded intrinsic. You can use llvm.memmove on any integer bit
5712 width. Not all targets support all bit widths however.</p>
5715 declare void @llvm.memmove.i8(i8 * <dest>, i8 * <src>,
5716 i8 <len>, i32 <align>)
5717 declare void @llvm.memmove.i16(i8 * <dest>, i8 * <src>,
5718 i16 <len>, i32 <align>)
5719 declare void @llvm.memmove.i32(i8 * <dest>, i8 * <src>,
5720 i32 <len>, i32 <align>)
5721 declare void @llvm.memmove.i64(i8 * <dest>, i8 * <src>,
5722 i64 <len>, i32 <align>)
5726 <p>The '<tt>llvm.memmove.*</tt>' intrinsics move a block of memory from the
5727 source location to the destination location. It is similar to the
5728 '<tt>llvm.memcpy</tt>' intrinsic but allows the two memory locations to
5731 <p>Note that, unlike the standard libc function, the <tt>llvm.memmove.*</tt>
5732 intrinsics do not return a value, and takes an extra alignment argument.</p>
5735 <p>The first argument is a pointer to the destination, the second is a pointer
5736 to the source. The third argument is an integer argument specifying the
5737 number of bytes to copy, and the fourth argument is the alignment of the
5738 source and destination locations.</p>
5740 <p>If the call to this intrinisic has an alignment value that is not 0 or 1,
5741 then the caller guarantees that the source and destination pointers are
5742 aligned to that boundary.</p>
5745 <p>The '<tt>llvm.memmove.*</tt>' intrinsics copy a block of memory from the
5746 source location to the destination location, which may overlap. It copies
5747 "len" bytes of memory over. If the argument is known to be aligned to some
5748 boundary, this can be specified as the fourth argument, otherwise it should
5749 be set to 0 or 1.</p>
5753 <!-- _______________________________________________________________________ -->
5754 <div class="doc_subsubsection">
5755 <a name="int_memset">'<tt>llvm.memset.*</tt>' Intrinsics</a>
5758 <div class="doc_text">
5761 <p>This is an overloaded intrinsic. You can use llvm.memset on any integer bit
5762 width. Not all targets support all bit widths however.</p>
5765 declare void @llvm.memset.i8(i8 * <dest>, i8 <val>,
5766 i8 <len>, i32 <align>)
5767 declare void @llvm.memset.i16(i8 * <dest>, i8 <val>,
5768 i16 <len>, i32 <align>)
5769 declare void @llvm.memset.i32(i8 * <dest>, i8 <val>,
5770 i32 <len>, i32 <align>)
5771 declare void @llvm.memset.i64(i8 * <dest>, i8 <val>,
5772 i64 <len>, i32 <align>)
5776 <p>The '<tt>llvm.memset.*</tt>' intrinsics fill a block of memory with a
5777 particular byte value.</p>
5779 <p>Note that, unlike the standard libc function, the <tt>llvm.memset</tt>
5780 intrinsic does not return a value, and takes an extra alignment argument.</p>
5783 <p>The first argument is a pointer to the destination to fill, the second is the
5784 byte value to fill it with, the third argument is an integer argument
5785 specifying the number of bytes to fill, and the fourth argument is the known
5786 alignment of destination location.</p>
5788 <p>If the call to this intrinisic has an alignment value that is not 0 or 1,
5789 then the caller guarantees that the destination pointer is aligned to that
5793 <p>The '<tt>llvm.memset.*</tt>' intrinsics fill "len" bytes of memory starting
5794 at the destination location. If the argument is known to be aligned to some
5795 boundary, this can be specified as the fourth argument, otherwise it should
5796 be set to 0 or 1.</p>
5800 <!-- _______________________________________________________________________ -->
5801 <div class="doc_subsubsection">
5802 <a name="int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a>
5805 <div class="doc_text">
5808 <p>This is an overloaded intrinsic. You can use <tt>llvm.sqrt</tt> on any
5809 floating point or vector of floating point type. Not all targets support all
5813 declare float @llvm.sqrt.f32(float %Val)
5814 declare double @llvm.sqrt.f64(double %Val)
5815 declare x86_fp80 @llvm.sqrt.f80(x86_fp80 %Val)
5816 declare fp128 @llvm.sqrt.f128(fp128 %Val)
5817 declare ppc_fp128 @llvm.sqrt.ppcf128(ppc_fp128 %Val)
5821 <p>The '<tt>llvm.sqrt</tt>' intrinsics return the sqrt of the specified operand,
5822 returning the same value as the libm '<tt>sqrt</tt>' functions would.
5823 Unlike <tt>sqrt</tt> in libm, however, <tt>llvm.sqrt</tt> has undefined
5824 behavior for negative numbers other than -0.0 (which allows for better
5825 optimization, because there is no need to worry about errno being
5826 set). <tt>llvm.sqrt(-0.0)</tt> is defined to return -0.0 like IEEE sqrt.</p>
5829 <p>The argument and return value are floating point numbers of the same
5833 <p>This function returns the sqrt of the specified operand if it is a
5834 nonnegative floating point number.</p>
5838 <!-- _______________________________________________________________________ -->
5839 <div class="doc_subsubsection">
5840 <a name="int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a>
5843 <div class="doc_text">
5846 <p>This is an overloaded intrinsic. You can use <tt>llvm.powi</tt> on any
5847 floating point or vector of floating point type. Not all targets support all
5851 declare float @llvm.powi.f32(float %Val, i32 %power)
5852 declare double @llvm.powi.f64(double %Val, i32 %power)
5853 declare x86_fp80 @llvm.powi.f80(x86_fp80 %Val, i32 %power)
5854 declare fp128 @llvm.powi.f128(fp128 %Val, i32 %power)
5855 declare ppc_fp128 @llvm.powi.ppcf128(ppc_fp128 %Val, i32 %power)
5859 <p>The '<tt>llvm.powi.*</tt>' intrinsics return the first operand raised to the
5860 specified (positive or negative) power. The order of evaluation of
5861 multiplications is not defined. When a vector of floating point type is
5862 used, the second argument remains a scalar integer value.</p>
5865 <p>The second argument is an integer power, and the first is a value to raise to
5869 <p>This function returns the first value raised to the second power with an
5870 unspecified sequence of rounding operations.</p>
5874 <!-- _______________________________________________________________________ -->
5875 <div class="doc_subsubsection">
5876 <a name="int_sin">'<tt>llvm.sin.*</tt>' Intrinsic</a>
5879 <div class="doc_text">
5882 <p>This is an overloaded intrinsic. You can use <tt>llvm.sin</tt> on any
5883 floating point or vector of floating point type. Not all targets support all
5887 declare float @llvm.sin.f32(float %Val)
5888 declare double @llvm.sin.f64(double %Val)
5889 declare x86_fp80 @llvm.sin.f80(x86_fp80 %Val)
5890 declare fp128 @llvm.sin.f128(fp128 %Val)
5891 declare ppc_fp128 @llvm.sin.ppcf128(ppc_fp128 %Val)
5895 <p>The '<tt>llvm.sin.*</tt>' intrinsics return the sine of the operand.</p>
5898 <p>The argument and return value are floating point numbers of the same
5902 <p>This function returns the sine of the specified operand, returning the same
5903 values as the libm <tt>sin</tt> functions would, and handles error conditions
5904 in the same way.</p>
5908 <!-- _______________________________________________________________________ -->
5909 <div class="doc_subsubsection">
5910 <a name="int_cos">'<tt>llvm.cos.*</tt>' Intrinsic</a>
5913 <div class="doc_text">
5916 <p>This is an overloaded intrinsic. You can use <tt>llvm.cos</tt> on any
5917 floating point or vector of floating point type. Not all targets support all
5921 declare float @llvm.cos.f32(float %Val)
5922 declare double @llvm.cos.f64(double %Val)
5923 declare x86_fp80 @llvm.cos.f80(x86_fp80 %Val)
5924 declare fp128 @llvm.cos.f128(fp128 %Val)
5925 declare ppc_fp128 @llvm.cos.ppcf128(ppc_fp128 %Val)
5929 <p>The '<tt>llvm.cos.*</tt>' intrinsics return the cosine of the operand.</p>
5932 <p>The argument and return value are floating point numbers of the same
5936 <p>This function returns the cosine of the specified operand, returning the same
5937 values as the libm <tt>cos</tt> functions would, and handles error conditions
5938 in the same way.</p>
5942 <!-- _______________________________________________________________________ -->
5943 <div class="doc_subsubsection">
5944 <a name="int_pow">'<tt>llvm.pow.*</tt>' Intrinsic</a>
5947 <div class="doc_text">
5950 <p>This is an overloaded intrinsic. You can use <tt>llvm.pow</tt> on any
5951 floating point or vector of floating point type. Not all targets support all
5955 declare float @llvm.pow.f32(float %Val, float %Power)
5956 declare double @llvm.pow.f64(double %Val, double %Power)
5957 declare x86_fp80 @llvm.pow.f80(x86_fp80 %Val, x86_fp80 %Power)
5958 declare fp128 @llvm.pow.f128(fp128 %Val, fp128 %Power)
5959 declare ppc_fp128 @llvm.pow.ppcf128(ppc_fp128 %Val, ppc_fp128 Power)
5963 <p>The '<tt>llvm.pow.*</tt>' intrinsics return the first operand raised to the
5964 specified (positive or negative) power.</p>
5967 <p>The second argument is a floating point power, and the first is a value to
5968 raise to that power.</p>
5971 <p>This function returns the first value raised to the second power, returning
5972 the same values as the libm <tt>pow</tt> functions would, and handles error
5973 conditions in the same way.</p>
5977 <!-- ======================================================================= -->
5978 <div class="doc_subsection">
5979 <a name="int_manip">Bit Manipulation Intrinsics</a>
5982 <div class="doc_text">
5984 <p>LLVM provides intrinsics for a few important bit manipulation operations.
5985 These allow efficient code generation for some algorithms.</p>
5989 <!-- _______________________________________________________________________ -->
5990 <div class="doc_subsubsection">
5991 <a name="int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a>
5994 <div class="doc_text">
5997 <p>This is an overloaded intrinsic function. You can use bswap on any integer
5998 type that is an even number of bytes (i.e. BitWidth % 16 == 0).</p>
6001 declare i16 @llvm.bswap.i16(i16 <id>)
6002 declare i32 @llvm.bswap.i32(i32 <id>)
6003 declare i64 @llvm.bswap.i64(i64 <id>)
6007 <p>The '<tt>llvm.bswap</tt>' family of intrinsics is used to byte swap integer
6008 values with an even number of bytes (positive multiple of 16 bits). These
6009 are useful for performing operations on data that is not in the target's
6010 native byte order.</p>
6013 <p>The <tt>llvm.bswap.i16</tt> intrinsic returns an i16 value that has the high
6014 and low byte of the input i16 swapped. Similarly,
6015 the <tt>llvm.bswap.i32</tt> intrinsic returns an i32 value that has the four
6016 bytes of the input i32 swapped, so that if the input bytes are numbered 0, 1,
6017 2, 3 then the returned i32 will have its bytes in 3, 2, 1, 0 order.
6018 The <tt>llvm.bswap.i48</tt>, <tt>llvm.bswap.i64</tt> and other intrinsics
6019 extend this concept to additional even-byte lengths (6 bytes, 8 bytes and
6020 more, respectively).</p>
6024 <!-- _______________________________________________________________________ -->
6025 <div class="doc_subsubsection">
6026 <a name="int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic</a>
6029 <div class="doc_text">
6032 <p>This is an overloaded intrinsic. You can use llvm.ctpop on any integer bit
6033 width. Not all targets support all bit widths however.</p>
6036 declare i8 @llvm.ctpop.i8(i8 <src>)
6037 declare i16 @llvm.ctpop.i16(i16 <src>)
6038 declare i32 @llvm.ctpop.i32(i32 <src>)
6039 declare i64 @llvm.ctpop.i64(i64 <src>)
6040 declare i256 @llvm.ctpop.i256(i256 <src>)
6044 <p>The '<tt>llvm.ctpop</tt>' family of intrinsics counts the number of bits set
6048 <p>The only argument is the value to be counted. The argument may be of any
6049 integer type. The return type must match the argument type.</p>
6052 <p>The '<tt>llvm.ctpop</tt>' intrinsic counts the 1's in a variable.</p>
6056 <!-- _______________________________________________________________________ -->
6057 <div class="doc_subsubsection">
6058 <a name="int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic</a>
6061 <div class="doc_text">
6064 <p>This is an overloaded intrinsic. You can use <tt>llvm.ctlz</tt> on any
6065 integer bit width. Not all targets support all bit widths however.</p>
6068 declare i8 @llvm.ctlz.i8 (i8 <src>)
6069 declare i16 @llvm.ctlz.i16(i16 <src>)
6070 declare i32 @llvm.ctlz.i32(i32 <src>)
6071 declare i64 @llvm.ctlz.i64(i64 <src>)
6072 declare i256 @llvm.ctlz.i256(i256 <src>)
6076 <p>The '<tt>llvm.ctlz</tt>' family of intrinsic functions counts the number of
6077 leading zeros in a variable.</p>
6080 <p>The only argument is the value to be counted. The argument may be of any
6081 integer type. The return type must match the argument type.</p>
6084 <p>The '<tt>llvm.ctlz</tt>' intrinsic counts the leading (most significant)
6085 zeros in a variable. If the src == 0 then the result is the size in bits of
6086 the type of src. For example, <tt>llvm.ctlz(i32 2) = 30</tt>.</p>
6090 <!-- _______________________________________________________________________ -->
6091 <div class="doc_subsubsection">
6092 <a name="int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic</a>
6095 <div class="doc_text">
6098 <p>This is an overloaded intrinsic. You can use <tt>llvm.cttz</tt> on any
6099 integer bit width. Not all targets support all bit widths however.</p>
6102 declare i8 @llvm.cttz.i8 (i8 <src>)
6103 declare i16 @llvm.cttz.i16(i16 <src>)
6104 declare i32 @llvm.cttz.i32(i32 <src>)
6105 declare i64 @llvm.cttz.i64(i64 <src>)
6106 declare i256 @llvm.cttz.i256(i256 <src>)
6110 <p>The '<tt>llvm.cttz</tt>' family of intrinsic functions counts the number of
6114 <p>The only argument is the value to be counted. The argument may be of any
6115 integer type. The return type must match the argument type.</p>
6118 <p>The '<tt>llvm.cttz</tt>' intrinsic counts the trailing (least significant)
6119 zeros in a variable. If the src == 0 then the result is the size in bits of
6120 the type of src. For example, <tt>llvm.cttz(2) = 1</tt>.</p>
6124 <!-- ======================================================================= -->
6125 <div class="doc_subsection">
6126 <a name="int_overflow">Arithmetic with Overflow Intrinsics</a>
6129 <div class="doc_text">
6131 <p>LLVM provides intrinsics for some arithmetic with overflow operations.</p>
6135 <!-- _______________________________________________________________________ -->
6136 <div class="doc_subsubsection">
6137 <a name="int_sadd_overflow">'<tt>llvm.sadd.with.overflow.*</tt>' Intrinsics</a>
6140 <div class="doc_text">
6143 <p>This is an overloaded intrinsic. You can use <tt>llvm.sadd.with.overflow</tt>
6144 on any integer bit width.</p>
6147 declare {i16, i1} @llvm.sadd.with.overflow.i16(i16 %a, i16 %b)
6148 declare {i32, i1} @llvm.sadd.with.overflow.i32(i32 %a, i32 %b)
6149 declare {i64, i1} @llvm.sadd.with.overflow.i64(i64 %a, i64 %b)
6153 <p>The '<tt>llvm.sadd.with.overflow</tt>' family of intrinsic functions perform
6154 a signed addition of the two arguments, and indicate whether an overflow
6155 occurred during the signed summation.</p>
6158 <p>The arguments (%a and %b) and the first element of the result structure may
6159 be of integer types of any bit width, but they must have the same bit
6160 width. The second element of the result structure must be of
6161 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
6162 undergo signed addition.</p>
6165 <p>The '<tt>llvm.sadd.with.overflow</tt>' family of intrinsic functions perform
6166 a signed addition of the two variables. They return a structure — the
6167 first element of which is the signed summation, and the second element of
6168 which is a bit specifying if the signed summation resulted in an
6173 %res = call {i32, i1} @llvm.sadd.with.overflow.i32(i32 %a, i32 %b)
6174 %sum = extractvalue {i32, i1} %res, 0
6175 %obit = extractvalue {i32, i1} %res, 1
6176 br i1 %obit, label %overflow, label %normal
6181 <!-- _______________________________________________________________________ -->
6182 <div class="doc_subsubsection">
6183 <a name="int_uadd_overflow">'<tt>llvm.uadd.with.overflow.*</tt>' Intrinsics</a>
6186 <div class="doc_text">
6189 <p>This is an overloaded intrinsic. You can use <tt>llvm.uadd.with.overflow</tt>
6190 on any integer bit width.</p>
6193 declare {i16, i1} @llvm.uadd.with.overflow.i16(i16 %a, i16 %b)
6194 declare {i32, i1} @llvm.uadd.with.overflow.i32(i32 %a, i32 %b)
6195 declare {i64, i1} @llvm.uadd.with.overflow.i64(i64 %a, i64 %b)
6199 <p>The '<tt>llvm.uadd.with.overflow</tt>' family of intrinsic functions perform
6200 an unsigned addition of the two arguments, and indicate whether a carry
6201 occurred during the unsigned summation.</p>
6204 <p>The arguments (%a and %b) and the first element of the result structure may
6205 be of integer types of any bit width, but they must have the same bit
6206 width. The second element of the result structure must be of
6207 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
6208 undergo unsigned addition.</p>
6211 <p>The '<tt>llvm.uadd.with.overflow</tt>' family of intrinsic functions perform
6212 an unsigned addition of the two arguments. They return a structure —
6213 the first element of which is the sum, and the second element of which is a
6214 bit specifying if the unsigned summation resulted in a carry.</p>
6218 %res = call {i32, i1} @llvm.uadd.with.overflow.i32(i32 %a, i32 %b)
6219 %sum = extractvalue {i32, i1} %res, 0
6220 %obit = extractvalue {i32, i1} %res, 1
6221 br i1 %obit, label %carry, label %normal
6226 <!-- _______________________________________________________________________ -->
6227 <div class="doc_subsubsection">
6228 <a name="int_ssub_overflow">'<tt>llvm.ssub.with.overflow.*</tt>' Intrinsics</a>
6231 <div class="doc_text">
6234 <p>This is an overloaded intrinsic. You can use <tt>llvm.ssub.with.overflow</tt>
6235 on any integer bit width.</p>
6238 declare {i16, i1} @llvm.ssub.with.overflow.i16(i16 %a, i16 %b)
6239 declare {i32, i1} @llvm.ssub.with.overflow.i32(i32 %a, i32 %b)
6240 declare {i64, i1} @llvm.ssub.with.overflow.i64(i64 %a, i64 %b)
6244 <p>The '<tt>llvm.ssub.with.overflow</tt>' family of intrinsic functions perform
6245 a signed subtraction of the two arguments, and indicate whether an overflow
6246 occurred during the signed subtraction.</p>
6249 <p>The arguments (%a and %b) and the first element of the result structure may
6250 be of integer types of any bit width, but they must have the same bit
6251 width. The second element of the result structure must be of
6252 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
6253 undergo signed subtraction.</p>
6256 <p>The '<tt>llvm.ssub.with.overflow</tt>' family of intrinsic functions perform
6257 a signed subtraction of the two arguments. They return a structure —
6258 the first element of which is the subtraction, and the second element of
6259 which is a bit specifying if the signed subtraction resulted in an
6264 %res = call {i32, i1} @llvm.ssub.with.overflow.i32(i32 %a, i32 %b)
6265 %sum = extractvalue {i32, i1} %res, 0
6266 %obit = extractvalue {i32, i1} %res, 1
6267 br i1 %obit, label %overflow, label %normal
6272 <!-- _______________________________________________________________________ -->
6273 <div class="doc_subsubsection">
6274 <a name="int_usub_overflow">'<tt>llvm.usub.with.overflow.*</tt>' Intrinsics</a>
6277 <div class="doc_text">
6280 <p>This is an overloaded intrinsic. You can use <tt>llvm.usub.with.overflow</tt>
6281 on any integer bit width.</p>
6284 declare {i16, i1} @llvm.usub.with.overflow.i16(i16 %a, i16 %b)
6285 declare {i32, i1} @llvm.usub.with.overflow.i32(i32 %a, i32 %b)
6286 declare {i64, i1} @llvm.usub.with.overflow.i64(i64 %a, i64 %b)
6290 <p>The '<tt>llvm.usub.with.overflow</tt>' family of intrinsic functions perform
6291 an unsigned subtraction of the two arguments, and indicate whether an
6292 overflow occurred during the unsigned subtraction.</p>
6295 <p>The arguments (%a and %b) and the first element of the result structure may
6296 be of integer types of any bit width, but they must have the same bit
6297 width. The second element of the result structure must be of
6298 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
6299 undergo unsigned subtraction.</p>
6302 <p>The '<tt>llvm.usub.with.overflow</tt>' family of intrinsic functions perform
6303 an unsigned subtraction of the two arguments. They return a structure —
6304 the first element of which is the subtraction, and the second element of
6305 which is a bit specifying if the unsigned subtraction resulted in an
6310 %res = call {i32, i1} @llvm.usub.with.overflow.i32(i32 %a, i32 %b)
6311 %sum = extractvalue {i32, i1} %res, 0
6312 %obit = extractvalue {i32, i1} %res, 1
6313 br i1 %obit, label %overflow, label %normal
6318 <!-- _______________________________________________________________________ -->
6319 <div class="doc_subsubsection">
6320 <a name="int_smul_overflow">'<tt>llvm.smul.with.overflow.*</tt>' Intrinsics</a>
6323 <div class="doc_text">
6326 <p>This is an overloaded intrinsic. You can use <tt>llvm.smul.with.overflow</tt>
6327 on any integer bit width.</p>
6330 declare {i16, i1} @llvm.smul.with.overflow.i16(i16 %a, i16 %b)
6331 declare {i32, i1} @llvm.smul.with.overflow.i32(i32 %a, i32 %b)
6332 declare {i64, i1} @llvm.smul.with.overflow.i64(i64 %a, i64 %b)
6337 <p>The '<tt>llvm.smul.with.overflow</tt>' family of intrinsic functions perform
6338 a signed multiplication of the two arguments, and indicate whether an
6339 overflow occurred during the signed multiplication.</p>
6342 <p>The arguments (%a and %b) and the first element of the result structure may
6343 be of integer types of any bit width, but they must have the same bit
6344 width. The second element of the result structure must be of
6345 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
6346 undergo signed multiplication.</p>
6349 <p>The '<tt>llvm.smul.with.overflow</tt>' family of intrinsic functions perform
6350 a signed multiplication of the two arguments. They return a structure —
6351 the first element of which is the multiplication, and the second element of
6352 which is a bit specifying if the signed multiplication resulted in an
6357 %res = call {i32, i1} @llvm.smul.with.overflow.i32(i32 %a, i32 %b)
6358 %sum = extractvalue {i32, i1} %res, 0
6359 %obit = extractvalue {i32, i1} %res, 1
6360 br i1 %obit, label %overflow, label %normal
6365 <!-- _______________________________________________________________________ -->
6366 <div class="doc_subsubsection">
6367 <a name="int_umul_overflow">'<tt>llvm.umul.with.overflow.*</tt>' Intrinsics</a>
6370 <div class="doc_text">
6373 <p>This is an overloaded intrinsic. You can use <tt>llvm.umul.with.overflow</tt>
6374 on any integer bit width.</p>
6377 declare {i16, i1} @llvm.umul.with.overflow.i16(i16 %a, i16 %b)
6378 declare {i32, i1} @llvm.umul.with.overflow.i32(i32 %a, i32 %b)
6379 declare {i64, i1} @llvm.umul.with.overflow.i64(i64 %a, i64 %b)
6383 <p>The '<tt>llvm.umul.with.overflow</tt>' family of intrinsic functions perform
6384 a unsigned multiplication of the two arguments, and indicate whether an
6385 overflow occurred during the unsigned multiplication.</p>
6388 <p>The arguments (%a and %b) and the first element of the result structure may
6389 be of integer types of any bit width, but they must have the same bit
6390 width. The second element of the result structure must be of
6391 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
6392 undergo unsigned multiplication.</p>
6395 <p>The '<tt>llvm.umul.with.overflow</tt>' family of intrinsic functions perform
6396 an unsigned multiplication of the two arguments. They return a structure
6397 — the first element of which is the multiplication, and the second
6398 element of which is a bit specifying if the unsigned multiplication resulted
6403 %res = call {i32, i1} @llvm.umul.with.overflow.i32(i32 %a, i32 %b)
6404 %sum = extractvalue {i32, i1} %res, 0
6405 %obit = extractvalue {i32, i1} %res, 1
6406 br i1 %obit, label %overflow, label %normal
6411 <!-- ======================================================================= -->
6412 <div class="doc_subsection">
6413 <a name="int_debugger">Debugger Intrinsics</a>
6416 <div class="doc_text">
6418 <p>The LLVM debugger intrinsics (which all start with <tt>llvm.dbg.</tt>
6419 prefix), are described in
6420 the <a href="SourceLevelDebugging.html#format_common_intrinsics">LLVM Source
6421 Level Debugging</a> document.</p>
6425 <!-- ======================================================================= -->
6426 <div class="doc_subsection">
6427 <a name="int_eh">Exception Handling Intrinsics</a>
6430 <div class="doc_text">
6432 <p>The LLVM exception handling intrinsics (which all start with
6433 <tt>llvm.eh.</tt> prefix), are described in
6434 the <a href="ExceptionHandling.html#format_common_intrinsics">LLVM Exception
6435 Handling</a> document.</p>
6439 <!-- ======================================================================= -->
6440 <div class="doc_subsection">
6441 <a name="int_trampoline">Trampoline Intrinsic</a>
6444 <div class="doc_text">
6446 <p>This intrinsic makes it possible to excise one parameter, marked with
6447 the <tt>nest</tt> attribute, from a function. The result is a callable
6448 function pointer lacking the nest parameter - the caller does not need to
6449 provide a value for it. Instead, the value to use is stored in advance in a
6450 "trampoline", a block of memory usually allocated on the stack, which also
6451 contains code to splice the nest value into the argument list. This is used
6452 to implement the GCC nested function address extension.</p>
6454 <p>For example, if the function is
6455 <tt>i32 f(i8* nest %c, i32 %x, i32 %y)</tt> then the resulting function
6456 pointer has signature <tt>i32 (i32, i32)*</tt>. It can be created as
6459 <div class="doc_code">
6461 %tramp = alloca [10 x i8], align 4 ; size and alignment only correct for X86
6462 %tramp1 = getelementptr [10 x i8]* %tramp, i32 0, i32 0
6463 %p = call i8* @llvm.init.trampoline( i8* %tramp1, i8* bitcast (i32 (i8* nest , i32, i32)* @f to i8*), i8* %nval )
6464 %fp = bitcast i8* %p to i32 (i32, i32)*
6468 <p>The call <tt>%val = call i32 %fp( i32 %x, i32 %y )</tt> is then equivalent
6469 to <tt>%val = call i32 %f( i8* %nval, i32 %x, i32 %y )</tt>.</p>
6473 <!-- _______________________________________________________________________ -->
6474 <div class="doc_subsubsection">
6475 <a name="int_it">'<tt>llvm.init.trampoline</tt>' Intrinsic</a>
6478 <div class="doc_text">
6482 declare i8* @llvm.init.trampoline(i8* <tramp>, i8* <func>, i8* <nval>)
6486 <p>This fills the memory pointed to by <tt>tramp</tt> with code and returns a
6487 function pointer suitable for executing it.</p>
6490 <p>The <tt>llvm.init.trampoline</tt> intrinsic takes three arguments, all
6491 pointers. The <tt>tramp</tt> argument must point to a sufficiently large and
6492 sufficiently aligned block of memory; this memory is written to by the
6493 intrinsic. Note that the size and the alignment are target-specific - LLVM
6494 currently provides no portable way of determining them, so a front-end that
6495 generates this intrinsic needs to have some target-specific knowledge.
6496 The <tt>func</tt> argument must hold a function bitcast to
6497 an <tt>i8*</tt>.</p>
6500 <p>The block of memory pointed to by <tt>tramp</tt> is filled with target
6501 dependent code, turning it into a function. A pointer to this function is
6502 returned, but needs to be bitcast to an <a href="#int_trampoline">appropriate
6503 function pointer type</a> before being called. The new function's signature
6504 is the same as that of <tt>func</tt> with any arguments marked with
6505 the <tt>nest</tt> attribute removed. At most one such <tt>nest</tt> argument
6506 is allowed, and it must be of pointer type. Calling the new function is
6507 equivalent to calling <tt>func</tt> with the same argument list, but
6508 with <tt>nval</tt> used for the missing <tt>nest</tt> argument. If, after
6509 calling <tt>llvm.init.trampoline</tt>, the memory pointed to
6510 by <tt>tramp</tt> is modified, then the effect of any later call to the
6511 returned function pointer is undefined.</p>
6515 <!-- ======================================================================= -->
6516 <div class="doc_subsection">
6517 <a name="int_atomics">Atomic Operations and Synchronization Intrinsics</a>
6520 <div class="doc_text">
6522 <p>These intrinsic functions expand the "universal IR" of LLVM to represent
6523 hardware constructs for atomic operations and memory synchronization. This
6524 provides an interface to the hardware, not an interface to the programmer. It
6525 is aimed at a low enough level to allow any programming models or APIs
6526 (Application Programming Interfaces) which need atomic behaviors to map
6527 cleanly onto it. It is also modeled primarily on hardware behavior. Just as
6528 hardware provides a "universal IR" for source languages, it also provides a
6529 starting point for developing a "universal" atomic operation and
6530 synchronization IR.</p>
6532 <p>These do <em>not</em> form an API such as high-level threading libraries,
6533 software transaction memory systems, atomic primitives, and intrinsic
6534 functions as found in BSD, GNU libc, atomic_ops, APR, and other system and
6535 application libraries. The hardware interface provided by LLVM should allow
6536 a clean implementation of all of these APIs and parallel programming models.
6537 No one model or paradigm should be selected above others unless the hardware
6538 itself ubiquitously does so.</p>
6542 <!-- _______________________________________________________________________ -->
6543 <div class="doc_subsubsection">
6544 <a name="int_memory_barrier">'<tt>llvm.memory.barrier</tt>' Intrinsic</a>
6546 <div class="doc_text">
6549 declare void @llvm.memory.barrier( i1 <ll>, i1 <ls>, i1 <sl>, i1 <ss>, i1 <device> )
6553 <p>The <tt>llvm.memory.barrier</tt> intrinsic guarantees ordering between
6554 specific pairs of memory access types.</p>
6557 <p>The <tt>llvm.memory.barrier</tt> intrinsic requires five boolean arguments.
6558 The first four arguments enables a specific barrier as listed below. The
6559 fith argument specifies that the barrier applies to io or device or uncached
6563 <li><tt>ll</tt>: load-load barrier</li>
6564 <li><tt>ls</tt>: load-store barrier</li>
6565 <li><tt>sl</tt>: store-load barrier</li>
6566 <li><tt>ss</tt>: store-store barrier</li>
6567 <li><tt>device</tt>: barrier applies to device and uncached memory also.</li>
6571 <p>This intrinsic causes the system to enforce some ordering constraints upon
6572 the loads and stores of the program. This barrier does not
6573 indicate <em>when</em> any events will occur, it only enforces
6574 an <em>order</em> in which they occur. For any of the specified pairs of load
6575 and store operations (f.ex. load-load, or store-load), all of the first
6576 operations preceding the barrier will complete before any of the second
6577 operations succeeding the barrier begin. Specifically the semantics for each
6578 pairing is as follows:</p>
6581 <li><tt>ll</tt>: All loads before the barrier must complete before any load
6582 after the barrier begins.</li>
6583 <li><tt>ls</tt>: All loads before the barrier must complete before any
6584 store after the barrier begins.</li>
6585 <li><tt>ss</tt>: All stores before the barrier must complete before any
6586 store after the barrier begins.</li>
6587 <li><tt>sl</tt>: All stores before the barrier must complete before any
6588 load after the barrier begins.</li>
6591 <p>These semantics are applied with a logical "and" behavior when more than one
6592 is enabled in a single memory barrier intrinsic.</p>
6594 <p>Backends may implement stronger barriers than those requested when they do
6595 not support as fine grained a barrier as requested. Some architectures do
6596 not need all types of barriers and on such architectures, these become
6604 %result1 = load i32* %ptr <i>; yields {i32}:result1 = 4</i>
6605 call void @llvm.memory.barrier( i1 false, i1 true, i1 false, i1 false )
6606 <i>; guarantee the above finishes</i>
6607 store i32 8, %ptr <i>; before this begins</i>
6612 <!-- _______________________________________________________________________ -->
6613 <div class="doc_subsubsection">
6614 <a name="int_atomic_cmp_swap">'<tt>llvm.atomic.cmp.swap.*</tt>' Intrinsic</a>
6617 <div class="doc_text">
6620 <p>This is an overloaded intrinsic. You can use <tt>llvm.atomic.cmp.swap</tt> on
6621 any integer bit width and for different address spaces. Not all targets
6622 support all bit widths however.</p>
6625 declare i8 @llvm.atomic.cmp.swap.i8.p0i8( i8* <ptr>, i8 <cmp>, i8 <val> )
6626 declare i16 @llvm.atomic.cmp.swap.i16.p0i16( i16* <ptr>, i16 <cmp>, i16 <val> )
6627 declare i32 @llvm.atomic.cmp.swap.i32.p0i32( i32* <ptr>, i32 <cmp>, i32 <val> )
6628 declare i64 @llvm.atomic.cmp.swap.i64.p0i64( i64* <ptr>, i64 <cmp>, i64 <val> )
6632 <p>This loads a value in memory and compares it to a given value. If they are
6633 equal, it stores a new value into the memory.</p>
6636 <p>The <tt>llvm.atomic.cmp.swap</tt> intrinsic takes three arguments. The result
6637 as well as both <tt>cmp</tt> and <tt>val</tt> must be integer values with the
6638 same bit width. The <tt>ptr</tt> argument must be a pointer to a value of
6639 this integer type. While any bit width integer may be used, targets may only
6640 lower representations they support in hardware.</p>
6643 <p>This entire intrinsic must be executed atomically. It first loads the value
6644 in memory pointed to by <tt>ptr</tt> and compares it with the
6645 value <tt>cmp</tt>. If they are equal, <tt>val</tt> is stored into the
6646 memory. The loaded value is yielded in all cases. This provides the
6647 equivalent of an atomic compare-and-swap operation within the SSA
6655 %val1 = add i32 4, 4
6656 %result1 = call i32 @llvm.atomic.cmp.swap.i32.p0i32( i32* %ptr, i32 4, %val1 )
6657 <i>; yields {i32}:result1 = 4</i>
6658 %stored1 = icmp eq i32 %result1, 4 <i>; yields {i1}:stored1 = true</i>
6659 %memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 8</i>
6661 %val2 = add i32 1, 1
6662 %result2 = call i32 @llvm.atomic.cmp.swap.i32.p0i32( i32* %ptr, i32 5, %val2 )
6663 <i>; yields {i32}:result2 = 8</i>
6664 %stored2 = icmp eq i32 %result2, 5 <i>; yields {i1}:stored2 = false</i>
6666 %memval2 = load i32* %ptr <i>; yields {i32}:memval2 = 8</i>
6671 <!-- _______________________________________________________________________ -->
6672 <div class="doc_subsubsection">
6673 <a name="int_atomic_swap">'<tt>llvm.atomic.swap.*</tt>' Intrinsic</a>
6675 <div class="doc_text">
6678 <p>This is an overloaded intrinsic. You can use <tt>llvm.atomic.swap</tt> on any
6679 integer bit width. Not all targets support all bit widths however.</p>
6682 declare i8 @llvm.atomic.swap.i8.p0i8( i8* <ptr>, i8 <val> )
6683 declare i16 @llvm.atomic.swap.i16.p0i16( i16* <ptr>, i16 <val> )
6684 declare i32 @llvm.atomic.swap.i32.p0i32( i32* <ptr>, i32 <val> )
6685 declare i64 @llvm.atomic.swap.i64.p0i64( i64* <ptr>, i64 <val> )
6689 <p>This intrinsic loads the value stored in memory at <tt>ptr</tt> and yields
6690 the value from memory. It then stores the value in <tt>val</tt> in the memory
6691 at <tt>ptr</tt>.</p>
6694 <p>The <tt>llvm.atomic.swap</tt> intrinsic takes two arguments. Both
6695 the <tt>val</tt> argument and the result must be integers of the same bit
6696 width. The first argument, <tt>ptr</tt>, must be a pointer to a value of this
6697 integer type. The targets may only lower integer representations they
6701 <p>This intrinsic loads the value pointed to by <tt>ptr</tt>, yields it, and
6702 stores <tt>val</tt> back into <tt>ptr</tt> atomically. This provides the
6703 equivalent of an atomic swap operation within the SSA framework.</p>
6710 %val1 = add i32 4, 4
6711 %result1 = call i32 @llvm.atomic.swap.i32.p0i32( i32* %ptr, i32 %val1 )
6712 <i>; yields {i32}:result1 = 4</i>
6713 %stored1 = icmp eq i32 %result1, 4 <i>; yields {i1}:stored1 = true</i>
6714 %memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 8</i>
6716 %val2 = add i32 1, 1
6717 %result2 = call i32 @llvm.atomic.swap.i32.p0i32( i32* %ptr, i32 %val2 )
6718 <i>; yields {i32}:result2 = 8</i>
6720 %stored2 = icmp eq i32 %result2, 8 <i>; yields {i1}:stored2 = true</i>
6721 %memval2 = load i32* %ptr <i>; yields {i32}:memval2 = 2</i>
6726 <!-- _______________________________________________________________________ -->
6727 <div class="doc_subsubsection">
6728 <a name="int_atomic_load_add">'<tt>llvm.atomic.load.add.*</tt>' Intrinsic</a>
6732 <div class="doc_text">
6735 <p>This is an overloaded intrinsic. You can use <tt>llvm.atomic.load.add</tt> on
6736 any integer bit width. Not all targets support all bit widths however.</p>
6739 declare i8 @llvm.atomic.load.add.i8..p0i8( i8* <ptr>, i8 <delta> )
6740 declare i16 @llvm.atomic.load.add.i16..p0i16( i16* <ptr>, i16 <delta> )
6741 declare i32 @llvm.atomic.load.add.i32..p0i32( i32* <ptr>, i32 <delta> )
6742 declare i64 @llvm.atomic.load.add.i64..p0i64( i64* <ptr>, i64 <delta> )
6746 <p>This intrinsic adds <tt>delta</tt> to the value stored in memory
6747 at <tt>ptr</tt>. It yields the original value at <tt>ptr</tt>.</p>
6750 <p>The intrinsic takes two arguments, the first a pointer to an integer value
6751 and the second an integer value. The result is also an integer value. These
6752 integer types can have any bit width, but they must all have the same bit
6753 width. The targets may only lower integer representations they support.</p>
6756 <p>This intrinsic does a series of operations atomically. It first loads the
6757 value stored at <tt>ptr</tt>. It then adds <tt>delta</tt>, stores the result
6758 to <tt>ptr</tt>. It yields the original value stored at <tt>ptr</tt>.</p>
6764 %result1 = call i32 @llvm.atomic.load.add.i32.p0i32( i32* %ptr, i32 4 )
6765 <i>; yields {i32}:result1 = 4</i>
6766 %result2 = call i32 @llvm.atomic.load.add.i32.p0i32( i32* %ptr, i32 2 )
6767 <i>; yields {i32}:result2 = 8</i>
6768 %result3 = call i32 @llvm.atomic.load.add.i32.p0i32( i32* %ptr, i32 5 )
6769 <i>; yields {i32}:result3 = 10</i>
6770 %memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 15</i>
6775 <!-- _______________________________________________________________________ -->
6776 <div class="doc_subsubsection">
6777 <a name="int_atomic_load_sub">'<tt>llvm.atomic.load.sub.*</tt>' Intrinsic</a>
6781 <div class="doc_text">
6784 <p>This is an overloaded intrinsic. You can use <tt>llvm.atomic.load.sub</tt> on
6785 any integer bit width and for different address spaces. Not all targets
6786 support all bit widths however.</p>
6789 declare i8 @llvm.atomic.load.sub.i8.p0i32( i8* <ptr>, i8 <delta> )
6790 declare i16 @llvm.atomic.load.sub.i16.p0i32( i16* <ptr>, i16 <delta> )
6791 declare i32 @llvm.atomic.load.sub.i32.p0i32( i32* <ptr>, i32 <delta> )
6792 declare i64 @llvm.atomic.load.sub.i64.p0i32( i64* <ptr>, i64 <delta> )
6796 <p>This intrinsic subtracts <tt>delta</tt> to the value stored in memory at
6797 <tt>ptr</tt>. It yields the original value at <tt>ptr</tt>.</p>
6800 <p>The intrinsic takes two arguments, the first a pointer to an integer value
6801 and the second an integer value. The result is also an integer value. These
6802 integer types can have any bit width, but they must all have the same bit
6803 width. The targets may only lower integer representations they support.</p>
6806 <p>This intrinsic does a series of operations atomically. It first loads the
6807 value stored at <tt>ptr</tt>. It then subtracts <tt>delta</tt>, stores the
6808 result to <tt>ptr</tt>. It yields the original value stored
6809 at <tt>ptr</tt>.</p>
6815 %result1 = call i32 @llvm.atomic.load.sub.i32.p0i32( i32* %ptr, i32 4 )
6816 <i>; yields {i32}:result1 = 8</i>
6817 %result2 = call i32 @llvm.atomic.load.sub.i32.p0i32( i32* %ptr, i32 2 )
6818 <i>; yields {i32}:result2 = 4</i>
6819 %result3 = call i32 @llvm.atomic.load.sub.i32.p0i32( i32* %ptr, i32 5 )
6820 <i>; yields {i32}:result3 = 2</i>
6821 %memval1 = load i32* %ptr <i>; yields {i32}:memval1 = -3</i>
6826 <!-- _______________________________________________________________________ -->
6827 <div class="doc_subsubsection">
6828 <a name="int_atomic_load_and">'<tt>llvm.atomic.load.and.*</tt>' Intrinsic</a><br>
6829 <a name="int_atomic_load_nand">'<tt>llvm.atomic.load.nand.*</tt>' Intrinsic</a><br>
6830 <a name="int_atomic_load_or">'<tt>llvm.atomic.load.or.*</tt>' Intrinsic</a><br>
6831 <a name="int_atomic_load_xor">'<tt>llvm.atomic.load.xor.*</tt>' Intrinsic</a><br>
6834 <div class="doc_text">
6837 <p>These are overloaded intrinsics. You can
6838 use <tt>llvm.atomic.load_and</tt>, <tt>llvm.atomic.load_nand</tt>,
6839 <tt>llvm.atomic.load_or</tt>, and <tt>llvm.atomic.load_xor</tt> on any integer
6840 bit width and for different address spaces. Not all targets support all bit
6844 declare i8 @llvm.atomic.load.and.i8.p0i8( i8* <ptr>, i8 <delta> )
6845 declare i16 @llvm.atomic.load.and.i16.p0i16( i16* <ptr>, i16 <delta> )
6846 declare i32 @llvm.atomic.load.and.i32.p0i32( i32* <ptr>, i32 <delta> )
6847 declare i64 @llvm.atomic.load.and.i64.p0i64( i64* <ptr>, i64 <delta> )
6851 declare i8 @llvm.atomic.load.or.i8.p0i8( i8* <ptr>, i8 <delta> )
6852 declare i16 @llvm.atomic.load.or.i16.p0i16( i16* <ptr>, i16 <delta> )
6853 declare i32 @llvm.atomic.load.or.i32.p0i32( i32* <ptr>, i32 <delta> )
6854 declare i64 @llvm.atomic.load.or.i64.p0i64( i64* <ptr>, i64 <delta> )
6858 declare i8 @llvm.atomic.load.nand.i8.p0i32( i8* <ptr>, i8 <delta> )
6859 declare i16 @llvm.atomic.load.nand.i16.p0i32( i16* <ptr>, i16 <delta> )
6860 declare i32 @llvm.atomic.load.nand.i32.p0i32( i32* <ptr>, i32 <delta> )
6861 declare i64 @llvm.atomic.load.nand.i64.p0i32( i64* <ptr>, i64 <delta> )
6865 declare i8 @llvm.atomic.load.xor.i8.p0i32( i8* <ptr>, i8 <delta> )
6866 declare i16 @llvm.atomic.load.xor.i16.p0i32( i16* <ptr>, i16 <delta> )
6867 declare i32 @llvm.atomic.load.xor.i32.p0i32( i32* <ptr>, i32 <delta> )
6868 declare i64 @llvm.atomic.load.xor.i64.p0i32( i64* <ptr>, i64 <delta> )
6872 <p>These intrinsics bitwise the operation (and, nand, or, xor) <tt>delta</tt> to
6873 the value stored in memory at <tt>ptr</tt>. It yields the original value
6874 at <tt>ptr</tt>.</p>
6877 <p>These intrinsics take two arguments, the first a pointer to an integer value
6878 and the second an integer value. The result is also an integer value. These
6879 integer types can have any bit width, but they must all have the same bit
6880 width. The targets may only lower integer representations they support.</p>
6883 <p>These intrinsics does a series of operations atomically. They first load the
6884 value stored at <tt>ptr</tt>. They then do the bitwise
6885 operation <tt>delta</tt>, store the result to <tt>ptr</tt>. They yield the
6886 original value stored at <tt>ptr</tt>.</p>
6891 store i32 0x0F0F, %ptr
6892 %result0 = call i32 @llvm.atomic.load.nand.i32.p0i32( i32* %ptr, i32 0xFF )
6893 <i>; yields {i32}:result0 = 0x0F0F</i>
6894 %result1 = call i32 @llvm.atomic.load.and.i32.p0i32( i32* %ptr, i32 0xFF )
6895 <i>; yields {i32}:result1 = 0xFFFFFFF0</i>
6896 %result2 = call i32 @llvm.atomic.load.or.i32.p0i32( i32* %ptr, i32 0F )
6897 <i>; yields {i32}:result2 = 0xF0</i>
6898 %result3 = call i32 @llvm.atomic.load.xor.i32.p0i32( i32* %ptr, i32 0F )
6899 <i>; yields {i32}:result3 = FF</i>
6900 %memval1 = load i32* %ptr <i>; yields {i32}:memval1 = F0</i>
6905 <!-- _______________________________________________________________________ -->
6906 <div class="doc_subsubsection">
6907 <a name="int_atomic_load_max">'<tt>llvm.atomic.load.max.*</tt>' Intrinsic</a><br>
6908 <a name="int_atomic_load_min">'<tt>llvm.atomic.load.min.*</tt>' Intrinsic</a><br>
6909 <a name="int_atomic_load_umax">'<tt>llvm.atomic.load.umax.*</tt>' Intrinsic</a><br>
6910 <a name="int_atomic_load_umin">'<tt>llvm.atomic.load.umin.*</tt>' Intrinsic</a><br>
6913 <div class="doc_text">
6916 <p>These are overloaded intrinsics. You can use <tt>llvm.atomic.load_max</tt>,
6917 <tt>llvm.atomic.load_min</tt>, <tt>llvm.atomic.load_umax</tt>, and
6918 <tt>llvm.atomic.load_umin</tt> on any integer bit width and for different
6919 address spaces. Not all targets support all bit widths however.</p>
6922 declare i8 @llvm.atomic.load.max.i8.p0i8( i8* <ptr>, i8 <delta> )
6923 declare i16 @llvm.atomic.load.max.i16.p0i16( i16* <ptr>, i16 <delta> )
6924 declare i32 @llvm.atomic.load.max.i32.p0i32( i32* <ptr>, i32 <delta> )
6925 declare i64 @llvm.atomic.load.max.i64.p0i64( i64* <ptr>, i64 <delta> )
6929 declare i8 @llvm.atomic.load.min.i8.p0i8( i8* <ptr>, i8 <delta> )
6930 declare i16 @llvm.atomic.load.min.i16.p0i16( i16* <ptr>, i16 <delta> )
6931 declare i32 @llvm.atomic.load.min.i32..p0i32( i32* <ptr>, i32 <delta> )
6932 declare i64 @llvm.atomic.load.min.i64..p0i64( i64* <ptr>, i64 <delta> )
6936 declare i8 @llvm.atomic.load.umax.i8.p0i8( i8* <ptr>, i8 <delta> )
6937 declare i16 @llvm.atomic.load.umax.i16.p0i16( i16* <ptr>, i16 <delta> )
6938 declare i32 @llvm.atomic.load.umax.i32.p0i32( i32* <ptr>, i32 <delta> )
6939 declare i64 @llvm.atomic.load.umax.i64.p0i64( i64* <ptr>, i64 <delta> )
6943 declare i8 @llvm.atomic.load.umin.i8..p0i8( i8* <ptr>, i8 <delta> )
6944 declare i16 @llvm.atomic.load.umin.i16.p0i16( i16* <ptr>, i16 <delta> )
6945 declare i32 @llvm.atomic.load.umin.i32..p0i32( i32* <ptr>, i32 <delta> )
6946 declare i64 @llvm.atomic.load.umin.i64..p0i64( i64* <ptr>, i64 <delta> )
6950 <p>These intrinsics takes the signed or unsigned minimum or maximum of
6951 <tt>delta</tt> and the value stored in memory at <tt>ptr</tt>. It yields the
6952 original value at <tt>ptr</tt>.</p>
6955 <p>These intrinsics take two arguments, the first a pointer to an integer value
6956 and the second an integer value. The result is also an integer value. These
6957 integer types can have any bit width, but they must all have the same bit
6958 width. The targets may only lower integer representations they support.</p>
6961 <p>These intrinsics does a series of operations atomically. They first load the
6962 value stored at <tt>ptr</tt>. They then do the signed or unsigned min or
6963 max <tt>delta</tt> and the value, store the result to <tt>ptr</tt>. They
6964 yield the original value stored at <tt>ptr</tt>.</p>
6970 %result0 = call i32 @llvm.atomic.load.min.i32.p0i32( i32* %ptr, i32 -2 )
6971 <i>; yields {i32}:result0 = 7</i>
6972 %result1 = call i32 @llvm.atomic.load.max.i32.p0i32( i32* %ptr, i32 8 )
6973 <i>; yields {i32}:result1 = -2</i>
6974 %result2 = call i32 @llvm.atomic.load.umin.i32.p0i32( i32* %ptr, i32 10 )
6975 <i>; yields {i32}:result2 = 8</i>
6976 %result3 = call i32 @llvm.atomic.load.umax.i32.p0i32( i32* %ptr, i32 30 )
6977 <i>; yields {i32}:result3 = 8</i>
6978 %memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 30</i>
6983 <!-- ======================================================================= -->
6984 <div class="doc_subsection">
6985 <a name="int_general">General Intrinsics</a>
6988 <div class="doc_text">
6990 <p>This class of intrinsics is designed to be generic and has no specific
6995 <!-- _______________________________________________________________________ -->
6996 <div class="doc_subsubsection">
6997 <a name="int_var_annotation">'<tt>llvm.var.annotation</tt>' Intrinsic</a>
7000 <div class="doc_text">
7004 declare void @llvm.var.annotation(i8* <val>, i8* <str>, i8* <str>, i32 <int> )
7008 <p>The '<tt>llvm.var.annotation</tt>' intrinsic.</p>
7011 <p>The first argument is a pointer to a value, the second is a pointer to a
7012 global string, the third is a pointer to a global string which is the source
7013 file name, and the last argument is the line number.</p>
7016 <p>This intrinsic allows annotation of local variables with arbitrary strings.
7017 This can be useful for special purpose optimizations that want to look for
7018 these annotations. These have no other defined use, they are ignored by code
7019 generation and optimization.</p>
7023 <!-- _______________________________________________________________________ -->
7024 <div class="doc_subsubsection">
7025 <a name="int_annotation">'<tt>llvm.annotation.*</tt>' Intrinsic</a>
7028 <div class="doc_text">
7031 <p>This is an overloaded intrinsic. You can use '<tt>llvm.annotation</tt>' on
7032 any integer bit width.</p>
7035 declare i8 @llvm.annotation.i8(i8 <val>, i8* <str>, i8* <str>, i32 <int> )
7036 declare i16 @llvm.annotation.i16(i16 <val>, i8* <str>, i8* <str>, i32 <int> )
7037 declare i32 @llvm.annotation.i32(i32 <val>, i8* <str>, i8* <str>, i32 <int> )
7038 declare i64 @llvm.annotation.i64(i64 <val>, i8* <str>, i8* <str>, i32 <int> )
7039 declare i256 @llvm.annotation.i256(i256 <val>, i8* <str>, i8* <str>, i32 <int> )
7043 <p>The '<tt>llvm.annotation</tt>' intrinsic.</p>
7046 <p>The first argument is an integer value (result of some expression), the
7047 second is a pointer to a global string, the third is a pointer to a global
7048 string which is the source file name, and the last argument is the line
7049 number. It returns the value of the first argument.</p>
7052 <p>This intrinsic allows annotations to be put on arbitrary expressions with
7053 arbitrary strings. This can be useful for special purpose optimizations that
7054 want to look for these annotations. These have no other defined use, they
7055 are ignored by code generation and optimization.</p>
7059 <!-- _______________________________________________________________________ -->
7060 <div class="doc_subsubsection">
7061 <a name="int_trap">'<tt>llvm.trap</tt>' Intrinsic</a>
7064 <div class="doc_text">
7068 declare void @llvm.trap()
7072 <p>The '<tt>llvm.trap</tt>' intrinsic.</p>
7078 <p>This intrinsics is lowered to the target dependent trap instruction. If the
7079 target does not have a trap instruction, this intrinsic will be lowered to
7080 the call of the <tt>abort()</tt> function.</p>
7084 <!-- _______________________________________________________________________ -->
7085 <div class="doc_subsubsection">
7086 <a name="int_stackprotector">'<tt>llvm.stackprotector</tt>' Intrinsic</a>
7089 <div class="doc_text">
7093 declare void @llvm.stackprotector( i8* <guard>, i8** <slot> )
7097 <p>The <tt>llvm.stackprotector</tt> intrinsic takes the <tt>guard</tt> and
7098 stores it onto the stack at <tt>slot</tt>. The stack slot is adjusted to
7099 ensure that it is placed on the stack before local variables.</p>
7102 <p>The <tt>llvm.stackprotector</tt> intrinsic requires two pointer
7103 arguments. The first argument is the value loaded from the stack
7104 guard <tt>@__stack_chk_guard</tt>. The second variable is an <tt>alloca</tt>
7105 that has enough space to hold the value of the guard.</p>
7108 <p>This intrinsic causes the prologue/epilogue inserter to force the position of
7109 the <tt>AllocaInst</tt> stack slot to be before local variables on the
7110 stack. This is to ensure that if a local variable on the stack is
7111 overwritten, it will destroy the value of the guard. When the function exits,
7112 the guard on the stack is checked against the original guard. If they're
7113 different, then the program aborts by calling the <tt>__stack_chk_fail()</tt>
7118 <!-- *********************************************************************** -->
7121 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
7122 src="http://jigsaw.w3.org/css-validator/images/vcss-blue" alt="Valid CSS"></a>
7123 <a href="http://validator.w3.org/check/referer"><img
7124 src="http://www.w3.org/Icons/valid-html401-blue" alt="Valid HTML 4.01"></a>
7126 <a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
7127 <a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br>
7128 Last modified: $Date$