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></li>
24 <li><a href="#callingconv">Calling Conventions</a></li>
25 <li><a href="#globalvars">Global Variables</a></li>
26 <li><a href="#functionstructure">Function Structure</a></li>
29 <li><a href="#typesystem">Type System</a>
31 <li><a href="#t_primitive">Primitive Types</a>
33 <li><a href="#t_classifications">Type Classifications</a></li>
36 <li><a href="#t_derived">Derived Types</a>
38 <li><a href="#t_array">Array Type</a></li>
39 <li><a href="#t_function">Function Type</a></li>
40 <li><a href="#t_pointer">Pointer Type</a></li>
41 <li><a href="#t_struct">Structure Type</a></li>
42 <li><a href="#t_packed">Packed Type</a></li>
43 <li><a href="#t_opaque">Opaque Type</a></li>
48 <li><a href="#constants">Constants</a>
50 <li><a href="#simpleconstants">Simple Constants</a>
51 <li><a href="#aggregateconstants">Aggregate Constants</a>
52 <li><a href="#globalconstants">Global Variable and Function Addresses</a>
53 <li><a href="#undefvalues">Undefined Values</a>
54 <li><a href="#constantexprs">Constant Expressions</a>
57 <li><a href="#instref">Instruction Reference</a>
59 <li><a href="#terminators">Terminator Instructions</a>
61 <li><a href="#i_ret">'<tt>ret</tt>' Instruction</a></li>
62 <li><a href="#i_br">'<tt>br</tt>' Instruction</a></li>
63 <li><a href="#i_switch">'<tt>switch</tt>' Instruction</a></li>
64 <li><a href="#i_invoke">'<tt>invoke</tt>' Instruction</a></li>
65 <li><a href="#i_unwind">'<tt>unwind</tt>' Instruction</a></li>
66 <li><a href="#i_unreachable">'<tt>unreachable</tt>' Instruction</a></li>
69 <li><a href="#binaryops">Binary Operations</a>
71 <li><a href="#i_add">'<tt>add</tt>' Instruction</a></li>
72 <li><a href="#i_sub">'<tt>sub</tt>' Instruction</a></li>
73 <li><a href="#i_mul">'<tt>mul</tt>' Instruction</a></li>
74 <li><a href="#i_div">'<tt>div</tt>' Instruction</a></li>
75 <li><a href="#i_rem">'<tt>rem</tt>' Instruction</a></li>
76 <li><a href="#i_setcc">'<tt>set<i>cc</i></tt>' Instructions</a></li>
79 <li><a href="#bitwiseops">Bitwise Binary Operations</a>
81 <li><a href="#i_and">'<tt>and</tt>' Instruction</a></li>
82 <li><a href="#i_or">'<tt>or</tt>' Instruction</a></li>
83 <li><a href="#i_xor">'<tt>xor</tt>' Instruction</a></li>
84 <li><a href="#i_shl">'<tt>shl</tt>' Instruction</a></li>
85 <li><a href="#i_shr">'<tt>shr</tt>' Instruction</a></li>
88 <li><a href="#memoryops">Memory Access Operations</a>
90 <li><a href="#i_malloc">'<tt>malloc</tt>' Instruction</a></li>
91 <li><a href="#i_free">'<tt>free</tt>' Instruction</a></li>
92 <li><a href="#i_alloca">'<tt>alloca</tt>' Instruction</a></li>
93 <li><a href="#i_load">'<tt>load</tt>' Instruction</a></li>
94 <li><a href="#i_store">'<tt>store</tt>' Instruction</a></li>
95 <li><a href="#i_getelementptr">'<tt>getelementptr</tt>' Instruction</a></li>
98 <li><a href="#otherops">Other Operations</a>
100 <li><a href="#i_phi">'<tt>phi</tt>' Instruction</a></li>
101 <li><a href="#i_cast">'<tt>cast .. to</tt>' Instruction</a></li>
102 <li><a href="#i_select">'<tt>select</tt>' Instruction</a></li>
103 <li><a href="#i_call">'<tt>call</tt>' Instruction</a></li>
104 <li><a href="#i_vaarg">'<tt>vaarg</tt>' Instruction</a></li>
109 <li><a href="#intrinsics">Intrinsic Functions</a>
111 <li><a href="#int_varargs">Variable Argument Handling Intrinsics</a>
113 <li><a href="#i_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a></li>
114 <li><a href="#i_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a></li>
115 <li><a href="#i_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a></li>
118 <li><a href="#int_gc">Accurate Garbage Collection Intrinsics</a>
120 <li><a href="#i_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a></li>
121 <li><a href="#i_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a></li>
122 <li><a href="#i_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a></li>
125 <li><a href="#int_codegen">Code Generator Intrinsics</a>
127 <li><a href="#i_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a></li>
128 <li><a href="#i_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a></li>
129 <li><a href="#i_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a></li>
130 <li><a href="#i_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a></li>
133 <li><a href="#int_os">Operating System Intrinsics</a>
135 <li><a href="#i_readport">'<tt>llvm.readport</tt>' Intrinsic</a></li>
136 <li><a href="#i_writeport">'<tt>llvm.writeport</tt>' Intrinsic</a></li>
137 <li><a href="#i_readio">'<tt>llvm.readio</tt>' Intrinsic</a></li>
138 <li><a href="#i_writeio">'<tt>llvm.writeio</tt>' Intrinsic</a></li>
140 <li><a href="#int_libc">Standard C Library Intrinsics</a>
142 <li><a href="#i_memcpy">'<tt>llvm.memcpy</tt>' Intrinsic</a></li>
143 <li><a href="#i_memmove">'<tt>llvm.memmove</tt>' Intrinsic</a></li>
144 <li><a href="#i_memset">'<tt>llvm.memset</tt>' Intrinsic</a></li>
145 <li><a href="#i_isunordered">'<tt>llvm.isunordered</tt>' Intrinsic</a></li>
146 <li><a href="#i_sqrt">'<tt>llvm.sqrt</tt>' Intrinsic</a></li>
150 <li><a href="#int_count">Bit counting Intrinsics</a>
152 <li><a href="#int_ctpop">'<tt>llvm.ctpop</tt>' Intrinsic </a></li>
153 <li><a href="#int_ctlz">'<tt>llvm.ctlz</tt>' Intrinsic </a></li>
154 <li><a href="#int_cttz">'<tt>llvm.cttz</tt>' Intrinsic </a></li>
157 <li><a href="#int_debugger">Debugger intrinsics</a></li>
162 <div class="doc_author">
163 <p>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a>
164 and <a href="mailto:vadve@cs.uiuc.edu">Vikram Adve</a></p>
167 <!-- *********************************************************************** -->
168 <div class="doc_section"> <a name="abstract">Abstract </a></div>
169 <!-- *********************************************************************** -->
171 <div class="doc_text">
172 <p>This document is a reference manual for the LLVM assembly language.
173 LLVM is an SSA based representation that provides type safety,
174 low-level operations, flexibility, and the capability of representing
175 'all' high-level languages cleanly. It is the common code
176 representation used throughout all phases of the LLVM compilation
180 <!-- *********************************************************************** -->
181 <div class="doc_section"> <a name="introduction">Introduction</a> </div>
182 <!-- *********************************************************************** -->
184 <div class="doc_text">
186 <p>The LLVM code representation is designed to be used in three
187 different forms: as an in-memory compiler IR, as an on-disk bytecode
188 representation (suitable for fast loading by a Just-In-Time compiler),
189 and as a human readable assembly language representation. This allows
190 LLVM to provide a powerful intermediate representation for efficient
191 compiler transformations and analysis, while providing a natural means
192 to debug and visualize the transformations. The three different forms
193 of LLVM are all equivalent. This document describes the human readable
194 representation and notation.</p>
196 <p>The LLVM representation aims to be light-weight and low-level
197 while being expressive, typed, and extensible at the same time. It
198 aims to be a "universal IR" of sorts, by being at a low enough level
199 that high-level ideas may be cleanly mapped to it (similar to how
200 microprocessors are "universal IR's", allowing many source languages to
201 be mapped to them). By providing type information, LLVM can be used as
202 the target of optimizations: for example, through pointer analysis, it
203 can be proven that a C automatic variable is never accessed outside of
204 the current function... allowing it to be promoted to a simple SSA
205 value instead of a memory location.</p>
209 <!-- _______________________________________________________________________ -->
210 <div class="doc_subsubsection"> <a name="wellformed">Well-Formedness</a> </div>
212 <div class="doc_text">
214 <p>It is important to note that this document describes 'well formed'
215 LLVM assembly language. There is a difference between what the parser
216 accepts and what is considered 'well formed'. For example, the
217 following instruction is syntactically okay, but not well formed:</p>
220 %x = <a href="#i_add">add</a> int 1, %x
223 <p>...because the definition of <tt>%x</tt> does not dominate all of
224 its uses. The LLVM infrastructure provides a verification pass that may
225 be used to verify that an LLVM module is well formed. This pass is
226 automatically run by the parser after parsing input assembly and by
227 the optimizer before it outputs bytecode. The violations pointed out
228 by the verifier pass indicate bugs in transformation passes or input to
231 <!-- Describe the typesetting conventions here. --> </div>
233 <!-- *********************************************************************** -->
234 <div class="doc_section"> <a name="identifiers">Identifiers</a> </div>
235 <!-- *********************************************************************** -->
237 <div class="doc_text">
239 <p>LLVM uses three different forms of identifiers, for different
243 <li>Named values are represented as a string of characters with a '%' prefix.
244 For example, %foo, %DivisionByZero, %a.really.long.identifier. The actual
245 regular expression used is '<tt>%[a-zA-Z$._][a-zA-Z$._0-9]*</tt>'.
246 Identifiers which require other characters in their names can be surrounded
247 with quotes. In this way, anything except a <tt>"</tt> character can be used
250 <li>Unnamed values are represented as an unsigned numeric value with a '%'
251 prefix. For example, %12, %2, %44.</li>
253 <li>Constants, which are described in a <a href="#constants">section about
254 constants</a>, below.</li>
257 <p>LLVM requires that values start with a '%' sign for two reasons: Compilers
258 don't need to worry about name clashes with reserved words, and the set of
259 reserved words may be expanded in the future without penalty. Additionally,
260 unnamed identifiers allow a compiler to quickly come up with a temporary
261 variable without having to avoid symbol table conflicts.</p>
263 <p>Reserved words in LLVM are very similar to reserved words in other
264 languages. There are keywords for different opcodes ('<tt><a
265 href="#i_add">add</a></tt>', '<tt><a href="#i_cast">cast</a></tt>', '<tt><a
266 href="#i_ret">ret</a></tt>', etc...), for primitive type names ('<tt><a
267 href="#t_void">void</a></tt>', '<tt><a href="#t_uint">uint</a></tt>', etc...),
268 and others. These reserved words cannot conflict with variable names, because
269 none of them start with a '%' character.</p>
271 <p>Here is an example of LLVM code to multiply the integer variable
272 '<tt>%X</tt>' by 8:</p>
277 %result = <a href="#i_mul">mul</a> uint %X, 8
280 <p>After strength reduction:</p>
283 %result = <a href="#i_shl">shl</a> uint %X, ubyte 3
286 <p>And the hard way:</p>
289 <a href="#i_add">add</a> uint %X, %X <i>; yields {uint}:%0</i>
290 <a href="#i_add">add</a> uint %0, %0 <i>; yields {uint}:%1</i>
291 %result = <a href="#i_add">add</a> uint %1, %1
294 <p>This last way of multiplying <tt>%X</tt> by 8 illustrates several
295 important lexical features of LLVM:</p>
299 <li>Comments are delimited with a '<tt>;</tt>' and go until the end of
302 <li>Unnamed temporaries are created when the result of a computation is not
303 assigned to a named value.</li>
305 <li>Unnamed temporaries are numbered sequentially</li>
309 <p>...and it also shows a convention that we follow in this document. When
310 demonstrating instructions, we will follow an instruction with a comment that
311 defines the type and name of value produced. Comments are shown in italic
316 <!-- *********************************************************************** -->
317 <div class="doc_section"> <a name="highlevel">High Level Structure</a> </div>
318 <!-- *********************************************************************** -->
320 <!-- ======================================================================= -->
321 <div class="doc_subsection"> <a name="modulestructure">Module Structure</a>
324 <div class="doc_text">
326 <p>LLVM programs are composed of "Module"s, each of which is a
327 translation unit of the input programs. Each module consists of
328 functions, global variables, and symbol table entries. Modules may be
329 combined together with the LLVM linker, which merges function (and
330 global variable) definitions, resolves forward declarations, and merges
331 symbol table entries. Here is an example of the "hello world" module:</p>
333 <pre><i>; Declare the string constant as a global constant...</i>
334 <a href="#identifiers">%.LC0</a> = <a href="#linkage_internal">internal</a> <a
335 href="#globalvars">constant</a> <a href="#t_array">[13 x sbyte]</a> c"hello world\0A\00" <i>; [13 x sbyte]*</i>
337 <i>; External declaration of the puts function</i>
338 <a href="#functionstructure">declare</a> int %puts(sbyte*) <i>; int(sbyte*)* </i>
340 <i>; Definition of main function</i>
341 int %main() { <i>; int()* </i>
342 <i>; Convert [13x sbyte]* to sbyte *...</i>
344 href="#i_getelementptr">getelementptr</a> [13 x sbyte]* %.LC0, long 0, long 0 <i>; sbyte*</i>
346 <i>; Call puts function to write out the string to stdout...</i>
348 href="#i_call">call</a> int %puts(sbyte* %cast210) <i>; int</i>
350 href="#i_ret">ret</a> int 0<br>}<br></pre>
352 <p>This example is made up of a <a href="#globalvars">global variable</a>
353 named "<tt>.LC0</tt>", an external declaration of the "<tt>puts</tt>"
354 function, and a <a href="#functionstructure">function definition</a>
355 for "<tt>main</tt>".</p>
357 <p>In general, a module is made up of a list of global values,
358 where both functions and global variables are global values. Global values are
359 represented by a pointer to a memory location (in this case, a pointer to an
360 array of char, and a pointer to a function), and have one of the following <a
361 href="#linkage">linkage types</a>.</p>
365 <!-- ======================================================================= -->
366 <div class="doc_subsection">
367 <a name="linkage">Linkage Types</a>
370 <div class="doc_text">
373 All Global Variables and Functions have one of the following types of linkage:
378 <dt><tt><b><a name="linkage_internal">internal</a></b></tt> </dt>
380 <dd>Global values with internal linkage are only directly accessible by
381 objects in the current module. In particular, linking code into a module with
382 an internal global value may cause the internal to be renamed as necessary to
383 avoid collisions. Because the symbol is internal to the module, all
384 references can be updated. This corresponds to the notion of the
385 '<tt>static</tt>' keyword in C, or the idea of "anonymous namespaces" in C++.
388 <dt><tt><b><a name="linkage_linkonce">linkonce</a></b></tt>: </dt>
390 <dd>"<tt>linkonce</tt>" linkage is similar to <tt>internal</tt> linkage, with
391 the twist that linking together two modules defining the same
392 <tt>linkonce</tt> globals will cause one of the globals to be discarded. This
393 is typically used to implement inline functions. Unreferenced
394 <tt>linkonce</tt> globals are allowed to be discarded.
397 <dt><tt><b><a name="linkage_weak">weak</a></b></tt>: </dt>
399 <dd>"<tt>weak</tt>" linkage is exactly the same as <tt>linkonce</tt> linkage,
400 except that unreferenced <tt>weak</tt> globals may not be discarded. This is
401 used to implement constructs in C such as "<tt>int X;</tt>" at global scope.
404 <dt><tt><b><a name="linkage_appending">appending</a></b></tt>: </dt>
406 <dd>"<tt>appending</tt>" linkage may only be applied to global variables of
407 pointer to array type. When two global variables with appending linkage are
408 linked together, the two global arrays are appended together. This is the
409 LLVM, typesafe, equivalent of having the system linker append together
410 "sections" with identical names when .o files are linked.
413 <dt><tt><b><a name="linkage_external">externally visible</a></b></tt>:</dt>
415 <dd>If none of the above identifiers are used, the global is externally
416 visible, meaning that it participates in linkage and can be used to resolve
417 external symbol references.
421 <p><a name="linkage_external">For example, since the "<tt>.LC0</tt>"
422 variable is defined to be internal, if another module defined a "<tt>.LC0</tt>"
423 variable and was linked with this one, one of the two would be renamed,
424 preventing a collision. Since "<tt>main</tt>" and "<tt>puts</tt>" are
425 external (i.e., lacking any linkage declarations), they are accessible
426 outside of the current module. It is illegal for a function <i>declaration</i>
427 to have any linkage type other than "externally visible".</a></p>
431 <!-- ======================================================================= -->
432 <div class="doc_subsection">
433 <a name="callingconv">Calling Conventions</a>
436 <div class="doc_text">
438 <p>LLVM <a href="#functionstructure">functions</a>, <a href="#i_call">calls</a>
439 and <a href="#i_invoke">invokes</a> can all have an optional calling convention
440 specified for the call. The calling convention of any pair of dynamic
441 caller/callee must match, or the behavior of the program is undefined. The
442 following calling conventions are supported by LLVM, and more may be added in
446 <dt><b>"<tt>ccc</tt>" - The C calling convention</b>:</dt>
448 <dd>This calling convention (the default if no other calling convention is
449 specified) matches the target C calling conventions. This calling convention
450 supports varargs function calls and tolerates some mismatch in the declared
451 prototype and implemented declaration of the function (as does normal C).
454 <dt><b>"<tt>fastcc</tt>" - The fast calling convention</b>:</dt>
456 <dd>This calling convention attempts to make calls as fast as possible
457 (e.g. by passing things in registers). This calling convention allows the
458 target to use whatever tricks it wants to produce fast code for the target,
459 without having to conform to an externally specified ABI. Implementations of
460 this convention should allow arbitrary tail call optimization to be supported.
461 This calling convention does not support varargs and requires the prototype of
462 all callees to exactly match the prototype of the function definition.
465 <dt><b>"<tt>coldcc</tt>" - The cold calling convention</b>:</dt>
467 <dd>This calling convention attempts to make code in the caller as efficient
468 as possible under the assumption that the call is not commonly executed. As
469 such, these calls often preserve all registers so that the call does not break
470 any live ranges in the caller side. This calling convention does not support
471 varargs and requires the prototype of all callees to exactly match the
472 prototype of the function definition.
475 <dt><b>"<tt>cc <<em>n</em>></tt>" - Numbered convention</b>:</dt>
477 <dd>Any calling convention may be specified by number, allowing
478 target-specific calling conventions to be used. Target specific calling
479 conventions start at 64.
483 <p>More calling conventions can be added/defined on an as-needed basis, to
484 support pascal conventions or any other well-known target-independent
489 <!-- ======================================================================= -->
490 <div class="doc_subsection">
491 <a name="globalvars">Global Variables</a>
494 <div class="doc_text">
496 <p>Global variables define regions of memory allocated at compilation time
497 instead of run-time. Global variables may optionally be initialized. A
498 variable may be defined as a global "constant", which indicates that the
499 contents of the variable will <b>never</b> be modified (enabling better
500 optimization, allowing the global data to be placed in the read-only section of
501 an executable, etc). Note that variables that need runtime initialization
502 cannot be marked "constant", as there is a store to the variable.</p>
505 LLVM explicitly allows <em>declarations</em> of global variables to be marked
506 constant, even if the final definition of the global is not. This capability
507 can be used to enable slightly better optimization of the program, but requires
508 the language definition to guarantee that optimizations based on the
509 'constantness' are valid for the translation units that do not include the
513 <p>As SSA values, global variables define pointer values that are in
514 scope (i.e. they dominate) all basic blocks in the program. Global
515 variables always define a pointer to their "content" type because they
516 describe a region of memory, and all memory objects in LLVM are
517 accessed through pointers.</p>
522 <!-- ======================================================================= -->
523 <div class="doc_subsection">
524 <a name="functionstructure">Functions</a>
527 <div class="doc_text">
529 <p>LLVM function definitions consist of an optional <a href="#linkage">linkage
530 type</a>, an optional <a href="#callingconv">calling convention</a>, a return
531 type, a function name, a (possibly empty) argument list, an opening curly brace,
532 a list of basic blocks, and a closing curly brace. LLVM function declarations
533 are defined with the "<tt>declare</tt>" keyword, an optional <a
534 href="#callingconv">calling convention</a>, a return type, a function name, and
535 a possibly empty list of arguments.</p>
537 <p>A function definition contains a list of basic blocks, forming the CFG for
538 the function. Each basic block may optionally start with a label (giving the
539 basic block a symbol table entry), contains a list of instructions, and ends
540 with a <a href="#terminators">terminator</a> instruction (such as a branch or
541 function return).</p>
543 <p>The first basic block in a program is special in two ways: it is immediately
544 executed on entrance to the function, and it is not allowed to have predecessor
545 basic blocks (i.e. there can not be any branches to the entry block of a
546 function). Because the block can have no predecessors, it also cannot have any
547 <a href="#i_phi">PHI nodes</a>.</p>
549 <p>LLVM functions are identified by their name and type signature. Hence, two
550 functions with the same name but different parameter lists or return values are
551 considered different functions, and LLVM will resolve references to each
558 <!-- *********************************************************************** -->
559 <div class="doc_section"> <a name="typesystem">Type System</a> </div>
560 <!-- *********************************************************************** -->
562 <div class="doc_text">
564 <p>The LLVM type system is one of the most important features of the
565 intermediate representation. Being typed enables a number of
566 optimizations to be performed on the IR directly, without having to do
567 extra analyses on the side before the transformation. A strong type
568 system makes it easier to read the generated code and enables novel
569 analyses and transformations that are not feasible to perform on normal
570 three address code representations.</p>
574 <!-- ======================================================================= -->
575 <div class="doc_subsection"> <a name="t_primitive">Primitive Types</a> </div>
576 <div class="doc_text">
577 <p>The primitive types are the fundamental building blocks of the LLVM
578 system. The current set of primitive types is as follows:</p>
580 <table class="layout">
585 <tr><th>Type</th><th>Description</th></tr>
586 <tr><td><tt>void</tt></td><td>No value</td></tr>
587 <tr><td><tt>ubyte</tt></td><td>Unsigned 8-bit value</td></tr>
588 <tr><td><tt>ushort</tt></td><td>Unsigned 16-bit value</td></tr>
589 <tr><td><tt>uint</tt></td><td>Unsigned 32-bit value</td></tr>
590 <tr><td><tt>ulong</tt></td><td>Unsigned 64-bit value</td></tr>
591 <tr><td><tt>float</tt></td><td>32-bit floating point value</td></tr>
592 <tr><td><tt>label</tt></td><td>Branch destination</td></tr>
599 <tr><th>Type</th><th>Description</th></tr>
600 <tr><td><tt>bool</tt></td><td>True or False value</td></tr>
601 <tr><td><tt>sbyte</tt></td><td>Signed 8-bit value</td></tr>
602 <tr><td><tt>short</tt></td><td>Signed 16-bit value</td></tr>
603 <tr><td><tt>int</tt></td><td>Signed 32-bit value</td></tr>
604 <tr><td><tt>long</tt></td><td>Signed 64-bit value</td></tr>
605 <tr><td><tt>double</tt></td><td>64-bit floating point value</td></tr>
613 <!-- _______________________________________________________________________ -->
614 <div class="doc_subsubsection"> <a name="t_classifications">Type
615 Classifications</a> </div>
616 <div class="doc_text">
617 <p>These different primitive types fall into a few useful
620 <table border="1" cellspacing="0" cellpadding="4">
622 <tr><th>Classification</th><th>Types</th></tr>
624 <td><a name="t_signed">signed</a></td>
625 <td><tt>sbyte, short, int, long, float, double</tt></td>
628 <td><a name="t_unsigned">unsigned</a></td>
629 <td><tt>ubyte, ushort, uint, ulong</tt></td>
632 <td><a name="t_integer">integer</a></td>
633 <td><tt>ubyte, sbyte, ushort, short, uint, int, ulong, long</tt></td>
636 <td><a name="t_integral">integral</a></td>
637 <td><tt>bool, ubyte, sbyte, ushort, short, uint, int, ulong, long</tt>
641 <td><a name="t_floating">floating point</a></td>
642 <td><tt>float, double</tt></td>
645 <td><a name="t_firstclass">first class</a></td>
646 <td><tt>bool, ubyte, sbyte, ushort, short, uint, int, ulong, long,<br>
647 float, double, <a href="#t_pointer">pointer</a>,
648 <a href="#t_packed">packed</a></tt></td>
653 <p>The <a href="#t_firstclass">first class</a> types are perhaps the
654 most important. Values of these types are the only ones which can be
655 produced by instructions, passed as arguments, or used as operands to
656 instructions. This means that all structures and arrays must be
657 manipulated either by pointer or by component.</p>
660 <!-- ======================================================================= -->
661 <div class="doc_subsection"> <a name="t_derived">Derived Types</a> </div>
663 <div class="doc_text">
665 <p>The real power in LLVM comes from the derived types in the system.
666 This is what allows a programmer to represent arrays, functions,
667 pointers, and other useful types. Note that these derived types may be
668 recursive: For example, it is possible to have a two dimensional array.</p>
672 <!-- _______________________________________________________________________ -->
673 <div class="doc_subsubsection"> <a name="t_array">Array Type</a> </div>
675 <div class="doc_text">
679 <p>The array type is a very simple derived type that arranges elements
680 sequentially in memory. The array type requires a size (number of
681 elements) and an underlying data type.</p>
686 [<# elements> x <elementtype>]
689 <p>The number of elements is a constant integer value; elementtype may
690 be any type with a size.</p>
693 <table class="layout">
696 <tt>[40 x int ]</tt><br/>
697 <tt>[41 x int ]</tt><br/>
698 <tt>[40 x uint]</tt><br/>
701 Array of 40 integer values.<br/>
702 Array of 41 integer values.<br/>
703 Array of 40 unsigned integer values.<br/>
707 <p>Here are some examples of multidimensional arrays:</p>
708 <table class="layout">
711 <tt>[3 x [4 x int]]</tt><br/>
712 <tt>[12 x [10 x float]]</tt><br/>
713 <tt>[2 x [3 x [4 x uint]]]</tt><br/>
716 3x4 array of integer values.<br/>
717 12x10 array of single precision floating point values.<br/>
718 2x3x4 array of unsigned integer values.<br/>
723 <p>Note that 'variable sized arrays' can be implemented in LLVM With a zero
724 length array. Normally accesses past the end of an array are undefined in
725 LLVM (e.g. it is illegal to access the 5th element of a 3 element array).
726 As a special case, however, zero length arrays are recognized to be variable
727 length. This allows implementation of 'pascal style arrays' with the LLVM
728 type "{ int, [0 x float]}", for example.</p>
732 <!-- _______________________________________________________________________ -->
733 <div class="doc_subsubsection"> <a name="t_function">Function Type</a> </div>
734 <div class="doc_text">
736 <p>The function type can be thought of as a function signature. It
737 consists of a return type and a list of formal parameter types.
738 Function types are usually used to build virtual function tables
739 (which are structures of pointers to functions), for indirect function
740 calls, and when defining a function.</p>
742 The return type of a function type cannot be an aggregate type.
745 <pre> <returntype> (<parameter list>)<br></pre>
746 <p>Where '<tt><parameter list></tt>' is a comma-separated list of type
747 specifiers. Optionally, the parameter list may include a type <tt>...</tt>,
748 which indicates that the function takes a variable number of arguments.
749 Variable argument functions can access their arguments with the <a
750 href="#int_varargs">variable argument handling intrinsic</a> functions.</p>
752 <table class="layout">
755 <tt>int (int)</tt> <br/>
756 <tt>float (int, int *) *</tt><br/>
757 <tt>int (sbyte *, ...)</tt><br/>
760 function taking an <tt>int</tt>, returning an <tt>int</tt><br/>
761 <a href="#t_pointer">Pointer</a> to a function that takes an
762 <tt>int</tt> and a <a href="#t_pointer">pointer</a> to <tt>int</tt>,
763 returning <tt>float</tt>.<br/>
764 A vararg function that takes at least one <a href="#t_pointer">pointer</a>
765 to <tt>sbyte</tt> (signed char in C), which returns an integer. This is
766 the signature for <tt>printf</tt> in LLVM.<br/>
772 <!-- _______________________________________________________________________ -->
773 <div class="doc_subsubsection"> <a name="t_struct">Structure Type</a> </div>
774 <div class="doc_text">
776 <p>The structure type is used to represent a collection of data members
777 together in memory. The packing of the field types is defined to match
778 the ABI of the underlying processor. The elements of a structure may
779 be any type that has a size.</p>
780 <p>Structures are accessed using '<tt><a href="#i_load">load</a></tt>
781 and '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a
782 field with the '<tt><a href="#i_getelementptr">getelementptr</a></tt>'
785 <pre> { <type list> }<br></pre>
787 <table class="layout">
790 <tt>{ int, int, int }</tt><br/>
791 <tt>{ float, int (int) * }</tt><br/>
794 a triple of three <tt>int</tt> values<br/>
795 A pair, where the first element is a <tt>float</tt> and the second element
796 is a <a href="#t_pointer">pointer</a> to a <a href="#t_function">function</a>
797 that takes an <tt>int</tt>, returning an <tt>int</tt>.<br/>
803 <!-- _______________________________________________________________________ -->
804 <div class="doc_subsubsection"> <a name="t_pointer">Pointer Type</a> </div>
805 <div class="doc_text">
807 <p>As in many languages, the pointer type represents a pointer or
808 reference to another object, which must live in memory.</p>
810 <pre> <type> *<br></pre>
812 <table class="layout">
815 <tt>[4x int]*</tt><br/>
816 <tt>int (int *) *</tt><br/>
819 A <a href="#t_pointer">pointer</a> to <a href="#t_array">array</a> of
820 four <tt>int</tt> values<br/>
821 A <a href="#t_pointer">pointer</a> to a <a
822 href="#t_function">function</a> that takes an <tt>int*</tt>, returning an
829 <!-- _______________________________________________________________________ -->
830 <div class="doc_subsubsection"> <a name="t_packed">Packed Type</a> </div>
831 <div class="doc_text">
835 <p>A packed type is a simple derived type that represents a vector
836 of elements. Packed types are used when multiple primitive data
837 are operated in parallel using a single instruction (SIMD).
838 A packed type requires a size (number of
839 elements) and an underlying primitive data type. Packed types are
840 considered <a href="#t_firstclass">first class</a>.</p>
845 < <# elements> x <elementtype> >
848 <p>The number of elements is a constant integer value; elementtype may
849 be any integral or floating point type.</p>
853 <table class="layout">
856 <tt><4 x int></tt><br/>
857 <tt><8 x float></tt><br/>
858 <tt><2 x uint></tt><br/>
861 Packed vector of 4 integer values.<br/>
862 Packed vector of 8 floating-point values.<br/>
863 Packed vector of 2 unsigned integer values.<br/>
869 <!-- _______________________________________________________________________ -->
870 <div class="doc_subsubsection"> <a name="t_opaque">Opaque Type</a> </div>
871 <div class="doc_text">
875 <p>Opaque types are used to represent unknown types in the system. This
876 corresponds (for example) to the C notion of a foward declared structure type.
877 In LLVM, opaque types can eventually be resolved to any type (not just a
888 <table class="layout">
901 <!-- *********************************************************************** -->
902 <div class="doc_section"> <a name="constants">Constants</a> </div>
903 <!-- *********************************************************************** -->
905 <div class="doc_text">
907 <p>LLVM has several different basic types of constants. This section describes
908 them all and their syntax.</p>
912 <!-- ======================================================================= -->
913 <div class="doc_subsection"><a name="simpleconstants">Simple Constants</a></div>
915 <div class="doc_text">
918 <dt><b>Boolean constants</b></dt>
920 <dd>The two strings '<tt>true</tt>' and '<tt>false</tt>' are both valid
921 constants of the <tt><a href="#t_primitive">bool</a></tt> type.
924 <dt><b>Integer constants</b></dt>
926 <dd>Standard integers (such as '4') are constants of the <a
927 href="#t_integer">integer</a> type. Negative numbers may be used with signed
931 <dt><b>Floating point constants</b></dt>
933 <dd>Floating point constants use standard decimal notation (e.g. 123.421),
934 exponential notation (e.g. 1.23421e+2), or a more precise hexadecimal
935 notation (see below). Floating point constants must have a <a
936 href="#t_floating">floating point</a> type. </dd>
938 <dt><b>Null pointer constants</b></dt>
940 <dd>The identifier '<tt>null</tt>' is recognized as a null pointer constant
941 and must be of <a href="#t_pointer">pointer type</a>.</dd>
945 <p>The one non-intuitive notation for constants is the optional hexadecimal form
946 of floating point constants. For example, the form '<tt>double
947 0x432ff973cafa8000</tt>' is equivalent to (but harder to read than) '<tt>double
948 4.5e+15</tt>'. The only time hexadecimal floating point constants are required
949 (and the only time that they are generated by the disassembler) is when a
950 floating point constant must be emitted but it cannot be represented as a
951 decimal floating point number. For example, NaN's, infinities, and other
952 special values are represented in their IEEE hexadecimal format so that
953 assembly and disassembly do not cause any bits to change in the constants.</p>
957 <!-- ======================================================================= -->
958 <div class="doc_subsection"><a name="aggregateconstants">Aggregate Constants</a>
961 <div class="doc_text">
962 <p>Aggregate constants arise from aggregation of simple constants
963 and smaller aggregate constants.</p>
966 <dt><b>Structure constants</b></dt>
968 <dd>Structure constants are represented with notation similar to structure
969 type definitions (a comma separated list of elements, surrounded by braces
970 (<tt>{}</tt>)). For example: "<tt>{ int 4, float 17.0, int* %G }</tt>",
971 where "<tt>%G</tt>" is declared as "<tt>%G = external global int</tt>". Structure constants
972 must have <a href="#t_struct">structure type</a>, and the number and
973 types of elements must match those specified by the type.
976 <dt><b>Array constants</b></dt>
978 <dd>Array constants are represented with notation similar to array type
979 definitions (a comma separated list of elements, surrounded by square brackets
980 (<tt>[]</tt>)). For example: "<tt>[ int 42, int 11, int 74 ]</tt>". Array
981 constants must have <a href="#t_array">array type</a>, and the number and
982 types of elements must match those specified by the type.
985 <dt><b>Packed constants</b></dt>
987 <dd>Packed constants are represented with notation similar to packed type
988 definitions (a comma separated list of elements, surrounded by
989 less-than/greater-than's (<tt><></tt>)). For example: "<tt>< int 42,
990 int 11, int 74, int 100 ></tt>". Packed constants must have <a
991 href="#t_packed">packed type</a>, and the number and types of elements must
992 match those specified by the type.
995 <dt><b>Zero initialization</b></dt>
997 <dd>The string '<tt>zeroinitializer</tt>' can be used to zero initialize a
998 value to zero of <em>any</em> type, including scalar and aggregate types.
999 This is often used to avoid having to print large zero initializers (e.g. for
1000 large arrays), and is always exactly equivalent to using explicit zero
1007 <!-- ======================================================================= -->
1008 <div class="doc_subsection">
1009 <a name="globalconstants">Global Variable and Function Addresses</a>
1012 <div class="doc_text">
1014 <p>The addresses of <a href="#globalvars">global variables</a> and <a
1015 href="#functionstructure">functions</a> are always implicitly valid (link-time)
1016 constants. These constants are explicitly referenced when the <a
1017 href="#identifiers">identifier for the global</a> is used and always have <a
1018 href="#t_pointer">pointer</a> type. For example, the following is a legal LLVM
1024 %Z = global [2 x int*] [ int* %X, int* %Y ]
1029 <!-- ======================================================================= -->
1030 <div class="doc_subsection"><a name="undefvalues">Undefined Values</a></div>
1031 <div class="doc_text">
1032 <p>The string '<tt>undef</tt>' is recognized as a type-less constant that has
1033 no specific value. Undefined values may be of any type and be used anywhere
1034 a constant is permitted.</p>
1036 <p>Undefined values indicate to the compiler that the program is well defined
1037 no matter what value is used, giving the compiler more freedom to optimize.
1041 <!-- ======================================================================= -->
1042 <div class="doc_subsection"><a name="constantexprs">Constant Expressions</a>
1045 <div class="doc_text">
1047 <p>Constant expressions are used to allow expressions involving other constants
1048 to be used as constants. Constant expressions may be of any <a
1049 href="#t_firstclass">first class</a> type and may involve any LLVM operation
1050 that does not have side effects (e.g. load and call are not supported). The
1051 following is the syntax for constant expressions:</p>
1054 <dt><b><tt>cast ( CST to TYPE )</tt></b></dt>
1056 <dd>Cast a constant to another type.</dd>
1058 <dt><b><tt>getelementptr ( CSTPTR, IDX0, IDX1, ... )</tt></b></dt>
1060 <dd>Perform the <a href="#i_getelementptr">getelementptr operation</a> on
1061 constants. As with the <a href="#i_getelementptr">getelementptr</a>
1062 instruction, the index list may have zero or more indexes, which are required
1063 to make sense for the type of "CSTPTR".</dd>
1065 <dt><b><tt>OPCODE ( LHS, RHS )</tt></b></dt>
1067 <dd>Perform the specified operation of the LHS and RHS constants. OPCODE may
1068 be any of the <a href="#binaryops">binary</a> or <a href="#bitwiseops">bitwise
1069 binary</a> operations. The constraints on operands are the same as those for
1070 the corresponding instruction (e.g. no bitwise operations on floating point
1071 values are allowed).</dd>
1075 <!-- *********************************************************************** -->
1076 <div class="doc_section"> <a name="instref">Instruction Reference</a> </div>
1077 <!-- *********************************************************************** -->
1079 <div class="doc_text">
1081 <p>The LLVM instruction set consists of several different
1082 classifications of instructions: <a href="#terminators">terminator
1083 instructions</a>, <a href="#binaryops">binary instructions</a>,
1084 <a href="#bitwiseops">bitwise binary instructions</a>, <a
1085 href="#memoryops">memory instructions</a>, and <a href="#otherops">other
1086 instructions</a>.</p>
1090 <!-- ======================================================================= -->
1091 <div class="doc_subsection"> <a name="terminators">Terminator
1092 Instructions</a> </div>
1094 <div class="doc_text">
1096 <p>As mentioned <a href="#functionstructure">previously</a>, every
1097 basic block in a program ends with a "Terminator" instruction, which
1098 indicates which block should be executed after the current block is
1099 finished. These terminator instructions typically yield a '<tt>void</tt>'
1100 value: they produce control flow, not values (the one exception being
1101 the '<a href="#i_invoke"><tt>invoke</tt></a>' instruction).</p>
1102 <p>There are six different terminator instructions: the '<a
1103 href="#i_ret"><tt>ret</tt></a>' instruction, the '<a href="#i_br"><tt>br</tt></a>'
1104 instruction, the '<a href="#i_switch"><tt>switch</tt></a>' instruction,
1105 the '<a href="#i_invoke"><tt>invoke</tt></a>' instruction, the '<a
1106 href="#i_unwind"><tt>unwind</tt></a>' instruction, and the '<a
1107 href="#i_unreachable"><tt>unreachable</tt></a>' instruction.</p>
1111 <!-- _______________________________________________________________________ -->
1112 <div class="doc_subsubsection"> <a name="i_ret">'<tt>ret</tt>'
1113 Instruction</a> </div>
1114 <div class="doc_text">
1116 <pre> ret <type> <value> <i>; Return a value from a non-void function</i>
1117 ret void <i>; Return from void function</i>
1120 <p>The '<tt>ret</tt>' instruction is used to return control flow (and a
1121 value) from a function back to the caller.</p>
1122 <p>There are two forms of the '<tt>ret</tt>' instruction: one that
1123 returns a value and then causes control flow, and one that just causes
1124 control flow to occur.</p>
1126 <p>The '<tt>ret</tt>' instruction may return any '<a
1127 href="#t_firstclass">first class</a>' type. Notice that a function is
1128 not <a href="#wellformed">well formed</a> if there exists a '<tt>ret</tt>'
1129 instruction inside of the function that returns a value that does not
1130 match the return type of the function.</p>
1132 <p>When the '<tt>ret</tt>' instruction is executed, control flow
1133 returns back to the calling function's context. If the caller is a "<a
1134 href="#i_call"><tt>call</tt></a>" instruction, execution continues at
1135 the instruction after the call. If the caller was an "<a
1136 href="#i_invoke"><tt>invoke</tt></a>" instruction, execution continues
1137 at the beginning of the "normal" destination block. If the instruction
1138 returns a value, that value shall set the call or invoke instruction's
1141 <pre> ret int 5 <i>; Return an integer value of 5</i>
1142 ret void <i>; Return from a void function</i>
1145 <!-- _______________________________________________________________________ -->
1146 <div class="doc_subsubsection"> <a name="i_br">'<tt>br</tt>' Instruction</a> </div>
1147 <div class="doc_text">
1149 <pre> br bool <cond>, label <iftrue>, label <iffalse><br> br label <dest> <i>; Unconditional branch</i>
1152 <p>The '<tt>br</tt>' instruction is used to cause control flow to
1153 transfer to a different basic block in the current function. There are
1154 two forms of this instruction, corresponding to a conditional branch
1155 and an unconditional branch.</p>
1157 <p>The conditional branch form of the '<tt>br</tt>' instruction takes a
1158 single '<tt>bool</tt>' value and two '<tt>label</tt>' values. The
1159 unconditional form of the '<tt>br</tt>' instruction takes a single '<tt>label</tt>'
1160 value as a target.</p>
1162 <p>Upon execution of a conditional '<tt>br</tt>' instruction, the '<tt>bool</tt>'
1163 argument is evaluated. If the value is <tt>true</tt>, control flows
1164 to the '<tt>iftrue</tt>' <tt>label</tt> argument. If "cond" is <tt>false</tt>,
1165 control flows to the '<tt>iffalse</tt>' <tt>label</tt> argument.</p>
1167 <pre>Test:<br> %cond = <a href="#i_setcc">seteq</a> int %a, %b<br> br bool %cond, label %IfEqual, label %IfUnequal<br>IfEqual:<br> <a
1168 href="#i_ret">ret</a> int 1<br>IfUnequal:<br> <a href="#i_ret">ret</a> int 0<br></pre>
1170 <!-- _______________________________________________________________________ -->
1171 <div class="doc_subsubsection">
1172 <a name="i_switch">'<tt>switch</tt>' Instruction</a>
1175 <div class="doc_text">
1179 switch <intty> <value>, label <defaultdest> [ <intty> <val>, label <dest> ... ]
1184 <p>The '<tt>switch</tt>' instruction is used to transfer control flow to one of
1185 several different places. It is a generalization of the '<tt>br</tt>'
1186 instruction, allowing a branch to occur to one of many possible
1192 <p>The '<tt>switch</tt>' instruction uses three parameters: an integer
1193 comparison value '<tt>value</tt>', a default '<tt>label</tt>' destination, and
1194 an array of pairs of comparison value constants and '<tt>label</tt>'s. The
1195 table is not allowed to contain duplicate constant entries.</p>
1199 <p>The <tt>switch</tt> instruction specifies a table of values and
1200 destinations. When the '<tt>switch</tt>' instruction is executed, this
1201 table is searched for the given value. If the value is found, control flow is
1202 transfered to the corresponding destination; otherwise, control flow is
1203 transfered to the default destination.</p>
1205 <h5>Implementation:</h5>
1207 <p>Depending on properties of the target machine and the particular
1208 <tt>switch</tt> instruction, this instruction may be code generated in different
1209 ways. For example, it could be generated as a series of chained conditional
1210 branches or with a lookup table.</p>
1215 <i>; Emulate a conditional br instruction</i>
1216 %Val = <a href="#i_cast">cast</a> bool %value to int
1217 switch int %Val, label %truedest [int 0, label %falsedest ]
1219 <i>; Emulate an unconditional br instruction</i>
1220 switch uint 0, label %dest [ ]
1222 <i>; Implement a jump table:</i>
1223 switch uint %val, label %otherwise [ uint 0, label %onzero
1224 uint 1, label %onone
1225 uint 2, label %ontwo ]
1229 <!-- _______________________________________________________________________ -->
1230 <div class="doc_subsubsection">
1231 <a name="i_invoke">'<tt>invoke</tt>' Instruction</a>
1234 <div class="doc_text">
1239 <result> = invoke [<a href="#callingconv">cconv</a>] <ptr to function ty> %<function ptr val>(<function args>)
1240 to label <normal label> except label <exception label>
1245 <p>The '<tt>invoke</tt>' instruction causes control to transfer to a specified
1246 function, with the possibility of control flow transfer to either the
1247 '<tt>normal</tt>' label or the
1248 '<tt>exception</tt>' label. If the callee function returns with the
1249 "<tt><a href="#i_ret">ret</a></tt>" instruction, control flow will return to the
1250 "normal" label. If the callee (or any indirect callees) returns with the "<a
1251 href="#i_unwind"><tt>unwind</tt></a>" instruction, control is interrupted and
1252 continued at the dynamically nearest "exception" label.</p>
1256 <p>This instruction requires several arguments:</p>
1260 The optional "cconv" marker indicates which <a href="callingconv">calling
1261 convention</a> the call should use. If none is specified, the call defaults
1262 to using C calling conventions.
1264 <li>'<tt>ptr to function ty</tt>': shall be the signature of the pointer to
1265 function value being invoked. In most cases, this is a direct function
1266 invocation, but indirect <tt>invoke</tt>s are just as possible, branching off
1267 an arbitrary pointer to function value.
1270 <li>'<tt>function ptr val</tt>': An LLVM value containing a pointer to a
1271 function to be invoked. </li>
1273 <li>'<tt>function args</tt>': argument list whose types match the function
1274 signature argument types. If the function signature indicates the function
1275 accepts a variable number of arguments, the extra arguments can be
1278 <li>'<tt>normal label</tt>': the label reached when the called function
1279 executes a '<tt><a href="#i_ret">ret</a></tt>' instruction. </li>
1281 <li>'<tt>exception label</tt>': the label reached when a callee returns with
1282 the <a href="#i_unwind"><tt>unwind</tt></a> instruction. </li>
1288 <p>This instruction is designed to operate as a standard '<tt><a
1289 href="#i_call">call</a></tt>' instruction in most regards. The primary
1290 difference is that it establishes an association with a label, which is used by
1291 the runtime library to unwind the stack.</p>
1293 <p>This instruction is used in languages with destructors to ensure that proper
1294 cleanup is performed in the case of either a <tt>longjmp</tt> or a thrown
1295 exception. Additionally, this is important for implementation of
1296 '<tt>catch</tt>' clauses in high-level languages that support them.</p>
1300 %retval = invoke int %Test(int 15) to label %Continue
1301 except label %TestCleanup <i>; {int}:retval set</i>
1302 %retval = invoke <a href="#callingconv">coldcc</a> int %Test(int 15) to label %Continue
1303 except label %TestCleanup <i>; {int}:retval set</i>
1308 <!-- _______________________________________________________________________ -->
1310 <div class="doc_subsubsection"> <a name="i_unwind">'<tt>unwind</tt>'
1311 Instruction</a> </div>
1313 <div class="doc_text">
1322 <p>The '<tt>unwind</tt>' instruction unwinds the stack, continuing control flow
1323 at the first callee in the dynamic call stack which used an <a
1324 href="#i_invoke"><tt>invoke</tt></a> instruction to perform the call. This is
1325 primarily used to implement exception handling.</p>
1329 <p>The '<tt>unwind</tt>' intrinsic causes execution of the current function to
1330 immediately halt. The dynamic call stack is then searched for the first <a
1331 href="#i_invoke"><tt>invoke</tt></a> instruction on the call stack. Once found,
1332 execution continues at the "exceptional" destination block specified by the
1333 <tt>invoke</tt> instruction. If there is no <tt>invoke</tt> instruction in the
1334 dynamic call chain, undefined behavior results.</p>
1337 <!-- _______________________________________________________________________ -->
1339 <div class="doc_subsubsection"> <a name="i_unreachable">'<tt>unreachable</tt>'
1340 Instruction</a> </div>
1342 <div class="doc_text">
1351 <p>The '<tt>unreachable</tt>' instruction has no defined semantics. This
1352 instruction is used to inform the optimizer that a particular portion of the
1353 code is not reachable. This can be used to indicate that the code after a
1354 no-return function cannot be reached, and other facts.</p>
1358 <p>The '<tt>unreachable</tt>' instruction has no defined semantics.</p>
1363 <!-- ======================================================================= -->
1364 <div class="doc_subsection"> <a name="binaryops">Binary Operations</a> </div>
1365 <div class="doc_text">
1366 <p>Binary operators are used to do most of the computation in a
1367 program. They require two operands, execute an operation on them, and
1368 produce a single value. The operands might represent
1369 multiple data, as is the case with the <a href="#t_packed">packed</a> data type.
1370 The result value of a binary operator is not
1371 necessarily the same type as its operands.</p>
1372 <p>There are several different binary operators:</p>
1374 <!-- _______________________________________________________________________ -->
1375 <div class="doc_subsubsection"> <a name="i_add">'<tt>add</tt>'
1376 Instruction</a> </div>
1377 <div class="doc_text">
1379 <pre> <result> = add <ty> <var1>, <var2> <i>; yields {ty}:result</i>
1382 <p>The '<tt>add</tt>' instruction returns the sum of its two operands.</p>
1384 <p>The two arguments to the '<tt>add</tt>' instruction must be either <a
1385 href="#t_integer">integer</a> or <a href="#t_floating">floating point</a> values.
1386 This instruction can also take <a href="#t_packed">packed</a> versions of the values.
1387 Both arguments must have identical types.</p>
1389 <p>The value produced is the integer or floating point sum of the two
1392 <pre> <result> = add int 4, %var <i>; yields {int}:result = 4 + %var</i>
1395 <!-- _______________________________________________________________________ -->
1396 <div class="doc_subsubsection"> <a name="i_sub">'<tt>sub</tt>'
1397 Instruction</a> </div>
1398 <div class="doc_text">
1400 <pre> <result> = sub <ty> <var1>, <var2> <i>; yields {ty}:result</i>
1403 <p>The '<tt>sub</tt>' instruction returns the difference of its two
1405 <p>Note that the '<tt>sub</tt>' instruction is used to represent the '<tt>neg</tt>'
1406 instruction present in most other intermediate representations.</p>
1408 <p>The two arguments to the '<tt>sub</tt>' instruction must be either <a
1409 href="#t_integer">integer</a> or <a href="#t_floating">floating point</a>
1411 This instruction can also take <a href="#t_packed">packed</a> versions of the values.
1412 Both arguments must have identical types.</p>
1414 <p>The value produced is the integer or floating point difference of
1415 the two operands.</p>
1417 <pre> <result> = sub int 4, %var <i>; yields {int}:result = 4 - %var</i>
1418 <result> = sub int 0, %val <i>; yields {int}:result = -%var</i>
1421 <!-- _______________________________________________________________________ -->
1422 <div class="doc_subsubsection"> <a name="i_mul">'<tt>mul</tt>'
1423 Instruction</a> </div>
1424 <div class="doc_text">
1426 <pre> <result> = mul <ty> <var1>, <var2> <i>; yields {ty}:result</i>
1429 <p>The '<tt>mul</tt>' instruction returns the product of its two
1432 <p>The two arguments to the '<tt>mul</tt>' instruction must be either <a
1433 href="#t_integer">integer</a> or <a href="#t_floating">floating point</a>
1435 This instruction can also take <a href="#t_packed">packed</a> versions of the values.
1436 Both arguments must have identical types.</p>
1438 <p>The value produced is the integer or floating point product of the
1440 <p>There is no signed vs unsigned multiplication. The appropriate
1441 action is taken based on the type of the operand.</p>
1443 <pre> <result> = mul int 4, %var <i>; yields {int}:result = 4 * %var</i>
1446 <!-- _______________________________________________________________________ -->
1447 <div class="doc_subsubsection"> <a name="i_div">'<tt>div</tt>'
1448 Instruction</a> </div>
1449 <div class="doc_text">
1451 <pre> <result> = div <ty> <var1>, <var2> <i>; yields {ty}:result</i>
1454 <p>The '<tt>div</tt>' instruction returns the quotient of its two
1457 <p>The two arguments to the '<tt>div</tt>' instruction must be either <a
1458 href="#t_integer">integer</a> or <a href="#t_floating">floating point</a>
1460 This instruction can also take <a href="#t_packed">packed</a> versions of the values.
1461 Both arguments must have identical types.</p>
1463 <p>The value produced is the integer or floating point quotient of the
1466 <pre> <result> = div int 4, %var <i>; yields {int}:result = 4 / %var</i>
1469 <!-- _______________________________________________________________________ -->
1470 <div class="doc_subsubsection"> <a name="i_rem">'<tt>rem</tt>'
1471 Instruction</a> </div>
1472 <div class="doc_text">
1474 <pre> <result> = rem <ty> <var1>, <var2> <i>; yields {ty}:result</i>
1477 <p>The '<tt>rem</tt>' instruction returns the remainder from the
1478 division of its two operands.</p>
1480 <p>The two arguments to the '<tt>rem</tt>' instruction must be either <a
1481 href="#t_integer">integer</a> or <a href="#t_floating">floating point</a>
1483 This instruction can also take <a href="#t_packed">packed</a> versions of the values.
1484 Both arguments must have identical types.</p>
1486 <p>This returns the <i>remainder</i> of a division (where the result
1487 has the same sign as the divisor), not the <i>modulus</i> (where the
1488 result has the same sign as the dividend) of a value. For more
1489 information about the difference, see: <a
1490 href="http://mathforum.org/dr.math/problems/anne.4.28.99.html">The
1493 <pre> <result> = rem int 4, %var <i>; yields {int}:result = 4 % %var</i>
1496 <!-- _______________________________________________________________________ -->
1497 <div class="doc_subsubsection"> <a name="i_setcc">'<tt>set<i>cc</i></tt>'
1498 Instructions</a> </div>
1499 <div class="doc_text">
1501 <pre> <result> = seteq <ty> <var1>, <var2> <i>; yields {bool}:result</i>
1502 <result> = setne <ty> <var1>, <var2> <i>; yields {bool}:result</i>
1503 <result> = setlt <ty> <var1>, <var2> <i>; yields {bool}:result</i>
1504 <result> = setgt <ty> <var1>, <var2> <i>; yields {bool}:result</i>
1505 <result> = setle <ty> <var1>, <var2> <i>; yields {bool}:result</i>
1506 <result> = setge <ty> <var1>, <var2> <i>; yields {bool}:result</i>
1509 <p>The '<tt>set<i>cc</i></tt>' family of instructions returns a boolean
1510 value based on a comparison of their two operands.</p>
1512 <p>The two arguments to the '<tt>set<i>cc</i></tt>' instructions must
1513 be of <a href="#t_firstclass">first class</a> type (it is not possible
1514 to compare '<tt>label</tt>'s, '<tt>array</tt>'s, '<tt>structure</tt>'
1515 or '<tt>void</tt>' values, etc...). Both arguments must have identical
1518 <p>The '<tt>seteq</tt>' instruction yields a <tt>true</tt> '<tt>bool</tt>'
1519 value if both operands are equal.<br>
1520 The '<tt>setne</tt>' instruction yields a <tt>true</tt> '<tt>bool</tt>'
1521 value if both operands are unequal.<br>
1522 The '<tt>setlt</tt>' instruction yields a <tt>true</tt> '<tt>bool</tt>'
1523 value if the first operand is less than the second operand.<br>
1524 The '<tt>setgt</tt>' instruction yields a <tt>true</tt> '<tt>bool</tt>'
1525 value if the first operand is greater than the second operand.<br>
1526 The '<tt>setle</tt>' instruction yields a <tt>true</tt> '<tt>bool</tt>'
1527 value if the first operand is less than or equal to the second operand.<br>
1528 The '<tt>setge</tt>' instruction yields a <tt>true</tt> '<tt>bool</tt>'
1529 value if the first operand is greater than or equal to the second
1532 <pre> <result> = seteq int 4, 5 <i>; yields {bool}:result = false</i>
1533 <result> = setne float 4, 5 <i>; yields {bool}:result = true</i>
1534 <result> = setlt uint 4, 5 <i>; yields {bool}:result = true</i>
1535 <result> = setgt sbyte 4, 5 <i>; yields {bool}:result = false</i>
1536 <result> = setle sbyte 4, 5 <i>; yields {bool}:result = true</i>
1537 <result> = setge sbyte 4, 5 <i>; yields {bool}:result = false</i>
1540 <!-- ======================================================================= -->
1541 <div class="doc_subsection"> <a name="bitwiseops">Bitwise Binary
1542 Operations</a> </div>
1543 <div class="doc_text">
1544 <p>Bitwise binary operators are used to do various forms of
1545 bit-twiddling in a program. They are generally very efficient
1546 instructions and can commonly be strength reduced from other
1547 instructions. They require two operands, execute an operation on them,
1548 and produce a single value. The resulting value of the bitwise binary
1549 operators is always the same type as its first operand.</p>
1551 <!-- _______________________________________________________________________ -->
1552 <div class="doc_subsubsection"> <a name="i_and">'<tt>and</tt>'
1553 Instruction</a> </div>
1554 <div class="doc_text">
1556 <pre> <result> = and <ty> <var1>, <var2> <i>; yields {ty}:result</i>
1559 <p>The '<tt>and</tt>' instruction returns the bitwise logical and of
1560 its two operands.</p>
1562 <p>The two arguments to the '<tt>and</tt>' instruction must be <a
1563 href="#t_integral">integral</a> values. Both arguments must have
1564 identical types.</p>
1566 <p>The truth table used for the '<tt>and</tt>' instruction is:</p>
1568 <div style="align: center">
1569 <table border="1" cellspacing="0" cellpadding="4">
1600 <pre> <result> = and int 4, %var <i>; yields {int}:result = 4 & %var</i>
1601 <result> = and int 15, 40 <i>; yields {int}:result = 8</i>
1602 <result> = and int 4, 8 <i>; yields {int}:result = 0</i>
1605 <!-- _______________________________________________________________________ -->
1606 <div class="doc_subsubsection"> <a name="i_or">'<tt>or</tt>' Instruction</a> </div>
1607 <div class="doc_text">
1609 <pre> <result> = or <ty> <var1>, <var2> <i>; yields {ty}:result</i>
1612 <p>The '<tt>or</tt>' instruction returns the bitwise logical inclusive
1613 or of its two operands.</p>
1615 <p>The two arguments to the '<tt>or</tt>' instruction must be <a
1616 href="#t_integral">integral</a> values. Both arguments must have
1617 identical types.</p>
1619 <p>The truth table used for the '<tt>or</tt>' instruction is:</p>
1621 <div style="align: center">
1622 <table border="1" cellspacing="0" cellpadding="4">
1653 <pre> <result> = or int 4, %var <i>; yields {int}:result = 4 | %var</i>
1654 <result> = or int 15, 40 <i>; yields {int}:result = 47</i>
1655 <result> = or int 4, 8 <i>; yields {int}:result = 12</i>
1658 <!-- _______________________________________________________________________ -->
1659 <div class="doc_subsubsection"> <a name="i_xor">'<tt>xor</tt>'
1660 Instruction</a> </div>
1661 <div class="doc_text">
1663 <pre> <result> = xor <ty> <var1>, <var2> <i>; yields {ty}:result</i>
1666 <p>The '<tt>xor</tt>' instruction returns the bitwise logical exclusive
1667 or of its two operands. The <tt>xor</tt> is used to implement the
1668 "one's complement" operation, which is the "~" operator in C.</p>
1670 <p>The two arguments to the '<tt>xor</tt>' instruction must be <a
1671 href="#t_integral">integral</a> values. Both arguments must have
1672 identical types.</p>
1674 <p>The truth table used for the '<tt>xor</tt>' instruction is:</p>
1676 <div style="align: center">
1677 <table border="1" cellspacing="0" cellpadding="4">
1709 <pre> <result> = xor int 4, %var <i>; yields {int}:result = 4 ^ %var</i>
1710 <result> = xor int 15, 40 <i>; yields {int}:result = 39</i>
1711 <result> = xor int 4, 8 <i>; yields {int}:result = 12</i>
1712 <result> = xor int %V, -1 <i>; yields {int}:result = ~%V</i>
1715 <!-- _______________________________________________________________________ -->
1716 <div class="doc_subsubsection"> <a name="i_shl">'<tt>shl</tt>'
1717 Instruction</a> </div>
1718 <div class="doc_text">
1720 <pre> <result> = shl <ty> <var1>, ubyte <var2> <i>; yields {ty}:result</i>
1723 <p>The '<tt>shl</tt>' instruction returns the first operand shifted to
1724 the left a specified number of bits.</p>
1726 <p>The first argument to the '<tt>shl</tt>' instruction must be an <a
1727 href="#t_integer">integer</a> type. The second argument must be an '<tt>ubyte</tt>'
1730 <p>The value produced is <tt>var1</tt> * 2<sup><tt>var2</tt></sup>.</p>
1732 <pre> <result> = shl int 4, ubyte %var <i>; yields {int}:result = 4 << %var</i>
1733 <result> = shl int 4, ubyte 2 <i>; yields {int}:result = 16</i>
1734 <result> = shl int 1, ubyte 10 <i>; yields {int}:result = 1024</i>
1737 <!-- _______________________________________________________________________ -->
1738 <div class="doc_subsubsection"> <a name="i_shr">'<tt>shr</tt>'
1739 Instruction</a> </div>
1740 <div class="doc_text">
1742 <pre> <result> = shr <ty> <var1>, ubyte <var2> <i>; yields {ty}:result</i>
1745 <p>The '<tt>shr</tt>' instruction returns the first operand shifted to
1746 the right a specified number of bits.</p>
1748 <p>The first argument to the '<tt>shr</tt>' instruction must be an <a
1749 href="#t_integer">integer</a> type. The second argument must be an '<tt>ubyte</tt>'
1752 <p>If the first argument is a <a href="#t_signed">signed</a> type, the
1753 most significant bit is duplicated in the newly free'd bit positions.
1754 If the first argument is unsigned, zero bits shall fill the empty
1757 <pre> <result> = shr int 4, ubyte %var <i>; yields {int}:result = 4 >> %var</i>
1758 <result> = shr uint 4, ubyte 1 <i>; yields {uint}:result = 2</i>
1759 <result> = shr int 4, ubyte 2 <i>; yields {int}:result = 1</i>
1760 <result> = shr sbyte 4, ubyte 3 <i>; yields {sbyte}:result = 0</i>
1761 <result> = shr sbyte -2, ubyte 1 <i>; yields {sbyte}:result = -1</i>
1764 <!-- ======================================================================= -->
1765 <div class="doc_subsection"> <a name="memoryops">Memory Access
1766 Operations</a></div>
1767 <div class="doc_text">
1768 <p>A key design point of an SSA-based representation is how it
1769 represents memory. In LLVM, no memory locations are in SSA form, which
1770 makes things very simple. This section describes how to read, write,
1771 allocate, and free memory in LLVM.</p>
1773 <!-- _______________________________________________________________________ -->
1774 <div class="doc_subsubsection"> <a name="i_malloc">'<tt>malloc</tt>'
1775 Instruction</a> </div>
1776 <div class="doc_text">
1778 <pre> <result> = malloc <type>, uint <NumElements> <i>; yields {type*}:result</i>
1779 <result> = malloc <type> <i>; yields {type*}:result</i>
1782 <p>The '<tt>malloc</tt>' instruction allocates memory from the system
1783 heap and returns a pointer to it.</p>
1785 <p>The '<tt>malloc</tt>' instruction allocates <tt>sizeof(<type>)*NumElements</tt>
1786 bytes of memory from the operating system and returns a pointer of the
1787 appropriate type to the program. The second form of the instruction is
1788 a shorter version of the first instruction that defaults to allocating
1790 <p>'<tt>type</tt>' must be a sized type.</p>
1792 <p>Memory is allocated using the system "<tt>malloc</tt>" function, and
1793 a pointer is returned.</p>
1795 <pre> %array = malloc [4 x ubyte ] <i>; yields {[%4 x ubyte]*}:array</i>
1798 href="#i_add">add</a> uint 2, 2 <i>; yields {uint}:size = uint 4</i>
1799 %array1 = malloc ubyte, uint 4 <i>; yields {ubyte*}:array1</i>
1800 %array2 = malloc [12 x ubyte], uint %size <i>; yields {[12 x ubyte]*}:array2</i>
1803 <!-- _______________________________________________________________________ -->
1804 <div class="doc_subsubsection"> <a name="i_free">'<tt>free</tt>'
1805 Instruction</a> </div>
1806 <div class="doc_text">
1808 <pre> free <type> <value> <i>; yields {void}</i>
1811 <p>The '<tt>free</tt>' instruction returns memory back to the unused
1812 memory heap to be reallocated in the future.</p>
1815 <p>'<tt>value</tt>' shall be a pointer value that points to a value
1816 that was allocated with the '<tt><a href="#i_malloc">malloc</a></tt>'
1819 <p>Access to the memory pointed to by the pointer is no longer defined
1820 after this instruction executes.</p>
1822 <pre> %array = <a href="#i_malloc">malloc</a> [4 x ubyte] <i>; yields {[4 x ubyte]*}:array</i>
1823 free [4 x ubyte]* %array
1826 <!-- _______________________________________________________________________ -->
1827 <div class="doc_subsubsection"> <a name="i_alloca">'<tt>alloca</tt>'
1828 Instruction</a> </div>
1829 <div class="doc_text">
1831 <pre> <result> = alloca <type>, uint <NumElements> <i>; yields {type*}:result</i>
1832 <result> = alloca <type> <i>; yields {type*}:result</i>
1835 <p>The '<tt>alloca</tt>' instruction allocates memory on the current
1836 stack frame of the procedure that is live until the current function
1837 returns to its caller.</p>
1839 <p>The '<tt>alloca</tt>' instruction allocates <tt>sizeof(<type>)*NumElements</tt>
1840 bytes of memory on the runtime stack, returning a pointer of the
1841 appropriate type to the program. The second form of the instruction is
1842 a shorter version of the first that defaults to allocating one element.</p>
1843 <p>'<tt>type</tt>' may be any sized type.</p>
1845 <p>Memory is allocated; a pointer is returned. '<tt>alloca</tt>'d
1846 memory is automatically released when the function returns. The '<tt>alloca</tt>'
1847 instruction is commonly used to represent automatic variables that must
1848 have an address available. When the function returns (either with the <tt><a
1849 href="#i_ret">ret</a></tt> or <tt><a href="#i_unwind">unwind</a></tt>
1850 instructions), the memory is reclaimed.</p>
1852 <pre> %ptr = alloca int <i>; yields {int*}:ptr</i>
1853 %ptr = alloca int, uint 4 <i>; yields {int*}:ptr</i>
1856 <!-- _______________________________________________________________________ -->
1857 <div class="doc_subsubsection"> <a name="i_load">'<tt>load</tt>'
1858 Instruction</a> </div>
1859 <div class="doc_text">
1861 <pre> <result> = load <ty>* <pointer><br> <result> = volatile load <ty>* <pointer><br></pre>
1863 <p>The '<tt>load</tt>' instruction is used to read from memory.</p>
1865 <p>The argument to the '<tt>load</tt>' instruction specifies the memory
1866 address to load from. The pointer must point to a <a
1867 href="#t_firstclass">first class</a> type. If the <tt>load</tt> is
1868 marked as <tt>volatile</tt> then the optimizer is not allowed to modify
1869 the number or order of execution of this <tt>load</tt> with other
1870 volatile <tt>load</tt> and <tt><a href="#i_store">store</a></tt>
1873 <p>The location of memory pointed to is loaded.</p>
1875 <pre> %ptr = <a href="#i_alloca">alloca</a> int <i>; yields {int*}:ptr</i>
1877 href="#i_store">store</a> int 3, int* %ptr <i>; yields {void}</i>
1878 %val = load int* %ptr <i>; yields {int}:val = int 3</i>
1881 <!-- _______________________________________________________________________ -->
1882 <div class="doc_subsubsection"> <a name="i_store">'<tt>store</tt>'
1883 Instruction</a> </div>
1885 <pre> store <ty> <value>, <ty>* <pointer> <i>; yields {void}</i>
1886 volatile store <ty> <value>, <ty>* <pointer> <i>; yields {void}</i>
1889 <p>The '<tt>store</tt>' instruction is used to write to memory.</p>
1891 <p>There are two arguments to the '<tt>store</tt>' instruction: a value
1892 to store and an address to store it into. The type of the '<tt><pointer></tt>'
1893 operand must be a pointer to the type of the '<tt><value></tt>'
1894 operand. If the <tt>store</tt> is marked as <tt>volatile</tt>, then the
1895 optimizer is not allowed to modify the number or order of execution of
1896 this <tt>store</tt> with other volatile <tt>load</tt> and <tt><a
1897 href="#i_store">store</a></tt> instructions.</p>
1899 <p>The contents of memory are updated to contain '<tt><value></tt>'
1900 at the location specified by the '<tt><pointer></tt>' operand.</p>
1902 <pre> %ptr = <a href="#i_alloca">alloca</a> int <i>; yields {int*}:ptr</i>
1904 href="#i_store">store</a> int 3, int* %ptr <i>; yields {void}</i>
1905 %val = load int* %ptr <i>; yields {int}:val = int 3</i>
1907 <!-- _______________________________________________________________________ -->
1908 <div class="doc_subsubsection">
1909 <a name="i_getelementptr">'<tt>getelementptr</tt>' Instruction</a>
1912 <div class="doc_text">
1915 <result> = getelementptr <ty>* <ptrval>{, <ty> <idx>}*
1921 The '<tt>getelementptr</tt>' instruction is used to get the address of a
1922 subelement of an aggregate data structure.</p>
1926 <p>This instruction takes a list of integer constants that indicate what
1927 elements of the aggregate object to index to. The actual types of the arguments
1928 provided depend on the type of the first pointer argument. The
1929 '<tt>getelementptr</tt>' instruction is used to index down through the type
1930 levels of a structure or to a specific index in an array. When indexing into a
1931 structure, only <tt>uint</tt>
1932 integer constants are allowed. When indexing into an array or pointer,
1933 <tt>int</tt> and <tt>long</tt> indexes are allowed of any sign.</p>
1935 <p>For example, let's consider a C code fragment and how it gets
1936 compiled to LLVM:</p>
1950 int *foo(struct ST *s) {
1951 return &s[1].Z.B[5][13];
1955 <p>The LLVM code generated by the GCC frontend is:</p>
1958 %RT = type { sbyte, [10 x [20 x int]], sbyte }
1959 %ST = type { int, double, %RT }
1963 int* %foo(%ST* %s) {
1965 %reg = getelementptr %ST* %s, int 1, uint 2, uint 1, int 5, int 13
1972 <p>The index types specified for the '<tt>getelementptr</tt>' instruction depend
1973 on the pointer type that is being indexed into. <a href="#t_pointer">Pointer</a>
1974 and <a href="#t_array">array</a> types require <tt>uint</tt>, <tt>int</tt>,
1975 <tt>ulong</tt>, or <tt>long</tt> values, and <a href="#t_struct">structure</a>
1976 types require <tt>uint</tt> <b>constants</b>.</p>
1978 <p>In the example above, the first index is indexing into the '<tt>%ST*</tt>'
1979 type, which is a pointer, yielding a '<tt>%ST</tt>' = '<tt>{ int, double, %RT
1980 }</tt>' type, a structure. The second index indexes into the third element of
1981 the structure, yielding a '<tt>%RT</tt>' = '<tt>{ sbyte, [10 x [20 x int]],
1982 sbyte }</tt>' type, another structure. The third index indexes into the second
1983 element of the structure, yielding a '<tt>[10 x [20 x int]]</tt>' type, an
1984 array. The two dimensions of the array are subscripted into, yielding an
1985 '<tt>int</tt>' type. The '<tt>getelementptr</tt>' instruction returns a pointer
1986 to this element, thus computing a value of '<tt>int*</tt>' type.</p>
1988 <p>Note that it is perfectly legal to index partially through a
1989 structure, returning a pointer to an inner element. Because of this,
1990 the LLVM code for the given testcase is equivalent to:</p>
1993 int* %foo(%ST* %s) {
1994 %t1 = getelementptr %ST* %s, int 1 <i>; yields %ST*:%t1</i>
1995 %t2 = getelementptr %ST* %t1, int 0, uint 2 <i>; yields %RT*:%t2</i>
1996 %t3 = getelementptr %RT* %t2, int 0, uint 1 <i>; yields [10 x [20 x int]]*:%t3</i>
1997 %t4 = getelementptr [10 x [20 x int]]* %t3, int 0, int 5 <i>; yields [20 x int]*:%t4</i>
1998 %t5 = getelementptr [20 x int]* %t4, int 0, int 13 <i>; yields int*:%t5</i>
2003 <p>Note that it is undefined to access an array out of bounds: array and
2004 pointer indexes must always be within the defined bounds of the array type.
2005 The one exception for this rules is zero length arrays. These arrays are
2006 defined to be accessible as variable length arrays, which requires access
2007 beyond the zero'th element.</p>
2012 <i>; yields [12 x ubyte]*:aptr</i>
2013 %aptr = getelementptr {int, [12 x ubyte]}* %sptr, long 0, uint 1
2017 <!-- ======================================================================= -->
2018 <div class="doc_subsection"> <a name="otherops">Other Operations</a> </div>
2019 <div class="doc_text">
2020 <p>The instructions in this category are the "miscellaneous"
2021 instructions, which defy better classification.</p>
2023 <!-- _______________________________________________________________________ -->
2024 <div class="doc_subsubsection"> <a name="i_phi">'<tt>phi</tt>'
2025 Instruction</a> </div>
2026 <div class="doc_text">
2028 <pre> <result> = phi <ty> [ <val0>, <label0>], ...<br></pre>
2030 <p>The '<tt>phi</tt>' instruction is used to implement the φ node in
2031 the SSA graph representing the function.</p>
2033 <p>The type of the incoming values are specified with the first type
2034 field. After this, the '<tt>phi</tt>' instruction takes a list of pairs
2035 as arguments, with one pair for each predecessor basic block of the
2036 current block. Only values of <a href="#t_firstclass">first class</a>
2037 type may be used as the value arguments to the PHI node. Only labels
2038 may be used as the label arguments.</p>
2039 <p>There must be no non-phi instructions between the start of a basic
2040 block and the PHI instructions: i.e. PHI instructions must be first in
2043 <p>At runtime, the '<tt>phi</tt>' instruction logically takes on the
2044 value specified by the parameter, depending on which basic block we
2045 came from in the last <a href="#terminators">terminator</a> instruction.</p>
2047 <pre>Loop: ; Infinite loop that counts from 0 on up...<br> %indvar = phi uint [ 0, %LoopHeader ], [ %nextindvar, %Loop ]<br> %nextindvar = add uint %indvar, 1<br> br label %Loop<br></pre>
2050 <!-- _______________________________________________________________________ -->
2051 <div class="doc_subsubsection">
2052 <a name="i_cast">'<tt>cast .. to</tt>' Instruction</a>
2055 <div class="doc_text">
2060 <result> = cast <ty> <value> to <ty2> <i>; yields ty2</i>
2066 The '<tt>cast</tt>' instruction is used as the primitive means to convert
2067 integers to floating point, change data type sizes, and break type safety (by
2075 The '<tt>cast</tt>' instruction takes a value to cast, which must be a first
2076 class value, and a type to cast it to, which must also be a <a
2077 href="#t_firstclass">first class</a> type.
2083 This instruction follows the C rules for explicit casts when determining how the
2084 data being cast must change to fit in its new container.
2088 When casting to bool, any value that would be considered true in the context of
2089 a C '<tt>if</tt>' condition is converted to the boolean '<tt>true</tt>' values,
2090 all else are '<tt>false</tt>'.
2094 When extending an integral value from a type of one signness to another (for
2095 example '<tt>sbyte</tt>' to '<tt>ulong</tt>'), the value is sign-extended if the
2096 <b>source</b> value is signed, and zero-extended if the source value is
2097 unsigned. <tt>bool</tt> values are always zero extended into either zero or
2104 %X = cast int 257 to ubyte <i>; yields ubyte:1</i>
2105 %Y = cast int 123 to bool <i>; yields bool:true</i>
2109 <!-- _______________________________________________________________________ -->
2110 <div class="doc_subsubsection">
2111 <a name="i_select">'<tt>select</tt>' Instruction</a>
2114 <div class="doc_text">
2119 <result> = select bool <cond>, <ty> <val1>, <ty> <val2> <i>; yields ty</i>
2125 The '<tt>select</tt>' instruction is used to choose one value based on a
2126 condition, without branching.
2133 The '<tt>select</tt>' instruction requires a boolean value indicating the condition, and two values of the same <a href="#t_firstclass">first class</a> type.
2139 If the boolean condition evaluates to true, the instruction returns the first
2140 value argument; otherwise, it returns the second value argument.
2146 %X = select bool true, ubyte 17, ubyte 42 <i>; yields ubyte:17</i>
2154 <!-- _______________________________________________________________________ -->
2155 <div class="doc_subsubsection">
2156 <a name="i_call">'<tt>call</tt>' Instruction</a>
2159 <div class="doc_text">
2163 <result> = [tail] call [<a href="#callingconv">cconv</a>] <ty>* <fnptrval>(<param list>)
2168 <p>The '<tt>call</tt>' instruction represents a simple function call.</p>
2172 <p>This instruction requires several arguments:</p>
2176 <p>The optional "tail" marker indicates whether the callee function accesses
2177 any allocas or varargs in the caller. If the "tail" marker is present, the
2178 function call is eligible for tail call optimization. Note that calls may
2179 be marked "tail" even if they do not occur before a <a
2180 href="#i_ret"><tt>ret</tt></a> instruction.
2183 <p>The optional "cconv" marker indicates which <a href="callingconv">calling
2184 convention</a> the call should use. If none is specified, the call defaults
2185 to using C calling conventions.
2188 <p>'<tt>ty</tt>': shall be the signature of the pointer to function value
2189 being invoked. The argument types must match the types implied by this
2190 signature. This type can be omitted if the function is not varargs and
2191 if the function type does not return a pointer to a function.</p>
2194 <p>'<tt>fnptrval</tt>': An LLVM value containing a pointer to a function to
2195 be invoked. In most cases, this is a direct function invocation, but
2196 indirect <tt>call</tt>s are just as possible, calling an arbitrary pointer
2197 to function value.</p>
2200 <p>'<tt>function args</tt>': argument list whose types match the
2201 function signature argument types. All arguments must be of
2202 <a href="#t_firstclass">first class</a> type. If the function signature
2203 indicates the function accepts a variable number of arguments, the extra
2204 arguments can be specified.</p>
2210 <p>The '<tt>call</tt>' instruction is used to cause control flow to
2211 transfer to a specified function, with its incoming arguments bound to
2212 the specified values. Upon a '<tt><a href="#i_ret">ret</a></tt>'
2213 instruction in the called function, control flow continues with the
2214 instruction after the function call, and the return value of the
2215 function is bound to the result argument. This is a simpler case of
2216 the <a href="#i_invoke">invoke</a> instruction.</p>
2221 %retval = call int %test(int %argc)
2222 call int(sbyte*, ...) *%printf(sbyte* %msg, int 12, sbyte 42);
2223 %X = tail call int %foo()
2224 %Y = tail call <a href="#callingconv">fastcc</a> int %foo()
2229 <!-- _______________________________________________________________________ -->
2230 <div class="doc_subsubsection">
2231 <a name="i_vaarg">'<tt>vaarg</tt>' Instruction</a>
2234 <div class="doc_text">
2239 <resultval> = va_arg <va_list*> <arglist>, <argty>
2244 <p>The '<tt>va_arg</tt>' instruction is used to access arguments passed through
2245 the "variable argument" area of a function call. It is used to implement the
2246 <tt>va_arg</tt> macro in C.</p>
2250 <p>This instruction takes a <tt>va_list*</tt> value and the type of
2251 the argument. It returns a value of the specified argument type and
2252 increments the <tt>va_list</tt> to poin to the next argument. Again, the
2253 actual type of <tt>va_list</tt> is target specific.</p>
2257 <p>The '<tt>va_arg</tt>' instruction loads an argument of the specified
2258 type from the specified <tt>va_list</tt> and causes the
2259 <tt>va_list</tt> to point to the next argument. For more information,
2260 see the variable argument handling <a href="#int_varargs">Intrinsic
2263 <p>It is legal for this instruction to be called in a function which does not
2264 take a variable number of arguments, for example, the <tt>vfprintf</tt>
2267 <p><tt>va_arg</tt> is an LLVM instruction instead of an <a
2268 href="#intrinsics">intrinsic function</a> because it takes a type as an
2273 <p>See the <a href="#int_varargs">variable argument processing</a> section.</p>
2277 <!-- *********************************************************************** -->
2278 <div class="doc_section"> <a name="intrinsics">Intrinsic Functions</a> </div>
2279 <!-- *********************************************************************** -->
2281 <div class="doc_text">
2283 <p>LLVM supports the notion of an "intrinsic function". These functions have
2284 well known names and semantics and are required to follow certain
2285 restrictions. Overall, these instructions represent an extension mechanism for
2286 the LLVM language that does not require changing all of the transformations in
2287 LLVM to add to the language (or the bytecode reader/writer, the parser,
2290 <p>Intrinsic function names must all start with an "<tt>llvm.</tt>" prefix. This
2291 prefix is reserved in LLVM for intrinsic names; thus, functions may not be named
2292 this. Intrinsic functions must always be external functions: you cannot define
2293 the body of intrinsic functions. Intrinsic functions may only be used in call
2294 or invoke instructions: it is illegal to take the address of an intrinsic
2295 function. Additionally, because intrinsic functions are part of the LLVM
2296 language, it is required that they all be documented here if any are added.</p>
2299 <p>To learn how to add an intrinsic function, please see the <a
2300 href="ExtendingLLVM.html">Extending LLVM Guide</a>.
2305 <!-- ======================================================================= -->
2306 <div class="doc_subsection">
2307 <a name="int_varargs">Variable Argument Handling Intrinsics</a>
2310 <div class="doc_text">
2312 <p>Variable argument support is defined in LLVM with the <a
2313 href="#i_vanext"><tt>vanext</tt></a> instruction and these three
2314 intrinsic functions. These functions are related to the similarly
2315 named macros defined in the <tt><stdarg.h></tt> header file.</p>
2317 <p>All of these functions operate on arguments that use a
2318 target-specific value type "<tt>va_list</tt>". The LLVM assembly
2319 language reference manual does not define what this type is, so all
2320 transformations should be prepared to handle intrinsics with any type
2323 <p>This example shows how the <a href="#i_vanext"><tt>vanext</tt></a>
2324 instruction and the variable argument handling intrinsic functions are
2328 int %test(int %X, ...) {
2329 ; Initialize variable argument processing
2331 call void %<a href="#i_va_start">llvm.va_start</a>(sbyte** %ap)
2333 ; Read a single integer argument
2334 %tmp = va_arg sbyte** %ap, int
2336 ; Demonstrate usage of llvm.va_copy and llvm.va_end
2338 call void %<a href="#i_va_copy">llvm.va_copy</a>(sbyte** %aq, sbyte** %ap)
2339 call void %<a href="#i_va_end">llvm.va_end</a>(sbyte** %aq)
2341 ; Stop processing of arguments.
2342 call void %<a href="#i_va_end">llvm.va_end</a>(sbyte** %ap)
2348 <!-- _______________________________________________________________________ -->
2349 <div class="doc_subsubsection">
2350 <a name="i_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a>
2354 <div class="doc_text">
2356 <pre> declare void %llvm.va_start(<va_list>* <arglist>)<br></pre>
2358 <P>The '<tt>llvm.va_start</tt>' intrinsic initializes
2359 <tt>*<arglist></tt> for subsequent use by <tt><a
2360 href="#i_va_arg">va_arg</a></tt>.</p>
2364 <P>The argument is a pointer to a <tt>va_list</tt> element to initialize.</p>
2368 <P>The '<tt>llvm.va_start</tt>' intrinsic works just like the <tt>va_start</tt>
2369 macro available in C. In a target-dependent way, it initializes the
2370 <tt>va_list</tt> element the argument points to, so that the next call to
2371 <tt>va_arg</tt> will produce the first variable argument passed to the function.
2372 Unlike the C <tt>va_start</tt> macro, this intrinsic does not need to know the
2373 last argument of the function, the compiler can figure that out.</p>
2377 <!-- _______________________________________________________________________ -->
2378 <div class="doc_subsubsection">
2379 <a name="i_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a>
2382 <div class="doc_text">
2384 <pre> declare void %llvm.va_end(<va_list*> <arglist>)<br></pre>
2386 <p>The '<tt>llvm.va_end</tt>' intrinsic destroys <tt><arglist></tt>
2387 which has been initialized previously with <tt><a href="#i_va_start">llvm.va_start</a></tt>
2388 or <tt><a href="#i_va_copy">llvm.va_copy</a></tt>.</p>
2390 <p>The argument is a <tt>va_list</tt> to destroy.</p>
2392 <p>The '<tt>llvm.va_end</tt>' intrinsic works just like the <tt>va_end</tt>
2393 macro available in C. In a target-dependent way, it destroys the <tt>va_list</tt>.
2394 Calls to <a href="#i_va_start"><tt>llvm.va_start</tt></a> and <a
2395 href="#i_va_copy"><tt>llvm.va_copy</tt></a> must be matched exactly
2396 with calls to <tt>llvm.va_end</tt>.</p>
2399 <!-- _______________________________________________________________________ -->
2400 <div class="doc_subsubsection">
2401 <a name="i_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a>
2404 <div class="doc_text">
2409 declare void %llvm.va_copy(<va_list>* <destarglist>,
2410 <va_list>* <srcarglist>)
2415 <p>The '<tt>llvm.va_copy</tt>' intrinsic copies the current argument position from
2416 the source argument list to the destination argument list.</p>
2420 <p>The first argument is a pointer to a <tt>va_list</tt> element to initialize.
2421 The second argument is a pointer to a <tt>va_list</tt> element to copy from.</p>
2426 <p>The '<tt>llvm.va_copy</tt>' intrinsic works just like the <tt>va_copy</tt> macro
2427 available in C. In a target-dependent way, it copies the source
2428 <tt>va_list</tt> element into the destination list. This intrinsic is necessary
2429 because the <tt><a href="i_va_begin">llvm.va_begin</a></tt> intrinsic may be
2430 arbitrarily complex and require memory allocation, for example.</p>
2434 <!-- ======================================================================= -->
2435 <div class="doc_subsection">
2436 <a name="int_gc">Accurate Garbage Collection Intrinsics</a>
2439 <div class="doc_text">
2442 LLVM support for <a href="GarbageCollection.html">Accurate Garbage
2443 Collection</a> requires the implementation and generation of these intrinsics.
2444 These intrinsics allow identification of <a href="#i_gcroot">GC roots on the
2445 stack</a>, as well as garbage collector implementations that require <a
2446 href="#i_gcread">read</a> and <a href="#i_gcwrite">write</a> barriers.
2447 Front-ends for type-safe garbage collected languages should generate these
2448 intrinsics to make use of the LLVM garbage collectors. For more details, see <a
2449 href="GarbageCollection.html">Accurate Garbage Collection with LLVM</a>.
2453 <!-- _______________________________________________________________________ -->
2454 <div class="doc_subsubsection">
2455 <a name="i_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a>
2458 <div class="doc_text">
2463 declare void %llvm.gcroot(<ty>** %ptrloc, <ty2>* %metadata)
2468 <p>The '<tt>llvm.gcroot</tt>' intrinsic declares the existence of a GC root to
2469 the code generator, and allows some metadata to be associated with it.</p>
2473 <p>The first argument specifies the address of a stack object that contains the
2474 root pointer. The second pointer (which must be either a constant or a global
2475 value address) contains the meta-data to be associated with the root.</p>
2479 <p>At runtime, a call to this intrinsics stores a null pointer into the "ptrloc"
2480 location. At compile-time, the code generator generates information to allow
2481 the runtime to find the pointer at GC safe points.
2487 <!-- _______________________________________________________________________ -->
2488 <div class="doc_subsubsection">
2489 <a name="i_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a>
2492 <div class="doc_text">
2497 declare sbyte* %llvm.gcread(sbyte** %Ptr)
2502 <p>The '<tt>llvm.gcread</tt>' intrinsic identifies reads of references from heap
2503 locations, allowing garbage collector implementations that require read
2508 <p>The argument is the address to read from, which should be an address
2509 allocated from the garbage collector.</p>
2513 <p>The '<tt>llvm.gcread</tt>' intrinsic has the same semantics as a load
2514 instruction, but may be replaced with substantially more complex code by the
2515 garbage collector runtime, as needed.</p>
2520 <!-- _______________________________________________________________________ -->
2521 <div class="doc_subsubsection">
2522 <a name="i_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a>
2525 <div class="doc_text">
2530 declare void %llvm.gcwrite(sbyte* %P1, sbyte** %P2)
2535 <p>The '<tt>llvm.gcwrite</tt>' intrinsic identifies writes of references to heap
2536 locations, allowing garbage collector implementations that require write
2537 barriers (such as generational or reference counting collectors).</p>
2541 <p>The first argument is the reference to store, and the second is the heap
2542 location to store to.</p>
2546 <p>The '<tt>llvm.gcwrite</tt>' intrinsic has the same semantics as a store
2547 instruction, but may be replaced with substantially more complex code by the
2548 garbage collector runtime, as needed.</p>
2554 <!-- ======================================================================= -->
2555 <div class="doc_subsection">
2556 <a name="int_codegen">Code Generator Intrinsics</a>
2559 <div class="doc_text">
2561 These intrinsics are provided by LLVM to expose special features that may only
2562 be implemented with code generator support.
2567 <!-- _______________________________________________________________________ -->
2568 <div class="doc_subsubsection">
2569 <a name="i_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a>
2572 <div class="doc_text">
2576 declare void* %llvm.returnaddress(uint <level>)
2582 The '<tt>llvm.returnaddress</tt>' intrinsic returns a target-specific value
2583 indicating the return address of the current function or one of its callers.
2589 The argument to this intrinsic indicates which function to return the address
2590 for. Zero indicates the calling function, one indicates its caller, etc. The
2591 argument is <b>required</b> to be a constant integer value.
2597 The '<tt>llvm.returnaddress</tt>' intrinsic either returns a pointer indicating
2598 the return address of the specified call frame, or zero if it cannot be
2599 identified. The value returned by this intrinsic is likely to be incorrect or 0
2600 for arguments other than zero, so it should only be used for debugging purposes.
2604 Note that calling this intrinsic does not prevent function inlining or other
2605 aggressive transformations, so the value returned may not be that of the obvious
2606 source-language caller.
2611 <!-- _______________________________________________________________________ -->
2612 <div class="doc_subsubsection">
2613 <a name="i_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a>
2616 <div class="doc_text">
2620 declare void* %llvm.frameaddress(uint <level>)
2626 The '<tt>llvm.frameaddress</tt>' intrinsic returns the target-specific frame
2627 pointer value for the specified stack frame.
2633 The argument to this intrinsic indicates which function to return the frame
2634 pointer for. Zero indicates the calling function, one indicates its caller,
2635 etc. The argument is <b>required</b> to be a constant integer value.
2641 The '<tt>llvm.frameaddress</tt>' intrinsic either returns a pointer indicating
2642 the frame address of the specified call frame, or zero if it cannot be
2643 identified. The value returned by this intrinsic is likely to be incorrect or 0
2644 for arguments other than zero, so it should only be used for debugging purposes.
2648 Note that calling this intrinsic does not prevent function inlining or other
2649 aggressive transformations, so the value returned may not be that of the obvious
2650 source-language caller.
2654 <!-- _______________________________________________________________________ -->
2655 <div class="doc_subsubsection">
2656 <a name="i_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a>
2659 <div class="doc_text">
2663 declare void %llvm.prefetch(sbyte * <address>,
2664 uint <rw>, uint <locality>)
2671 The '<tt>llvm.prefetch</tt>' intrinsic is a hint to the code generator to insert
2672 a prefetch instruction if supported; otherwise, it is a noop. Prefetches have
2674 effect on the behavior of the program but can change its performance
2681 <tt>address</tt> is the address to be prefetched, <tt>rw</tt> is the specifier
2682 determining if the fetch should be for a read (0) or write (1), and
2683 <tt>locality</tt> is a temporal locality specifier ranging from (0) - no
2684 locality, to (3) - extremely local keep in cache. The <tt>rw</tt> and
2685 <tt>locality</tt> arguments must be constant integers.
2691 This intrinsic does not modify the behavior of the program. In particular,
2692 prefetches cannot trap and do not produce a value. On targets that support this
2693 intrinsic, the prefetch can provide hints to the processor cache for better
2699 <!-- _______________________________________________________________________ -->
2700 <div class="doc_subsubsection">
2701 <a name="i_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a>
2704 <div class="doc_text">
2708 declare void %llvm.pcmarker( uint <id> )
2715 The '<tt>llvm.pcmarker</tt>' intrinsic is a method to export a Program Counter
2717 code to simulators and other tools. The method is target specific, but it is
2718 expected that the marker will use exported symbols to transmit the PC of the marker.
2719 The marker makes no guaranties that it will remain with any specific instruction
2720 after optimizations. It is possible that the presense of a marker will inhibit
2721 optimizations. The intended use is to be inserted after optmizations to allow
2722 correlations of simulation runs.
2728 <tt>id</tt> is a numerical id identifying the marker.
2734 This intrinsic does not modify the behavior of the program. Backends that do not
2735 support this intrinisic may ignore it.
2741 <!-- ======================================================================= -->
2742 <div class="doc_subsection">
2743 <a name="int_os">Operating System Intrinsics</a>
2746 <div class="doc_text">
2748 These intrinsics are provided by LLVM to support the implementation of
2749 operating system level code.
2754 <!-- _______________________________________________________________________ -->
2755 <div class="doc_subsubsection">
2756 <a name="i_readport">'<tt>llvm.readport</tt>' Intrinsic</a>
2759 <div class="doc_text">
2763 declare <integer type> %llvm.readport (<integer type> <address>)
2769 The '<tt>llvm.readport</tt>' intrinsic reads data from the specified hardware
2776 The argument to this intrinsic indicates the hardware I/O address from which
2777 to read the data. The address is in the hardware I/O address namespace (as
2778 opposed to being a memory location for memory mapped I/O).
2784 The '<tt>llvm.readport</tt>' intrinsic reads data from the hardware I/O port
2785 specified by <i>address</i> and returns the value. The address and return
2786 value must be integers, but the size is dependent upon the platform upon which
2787 the program is code generated. For example, on x86, the address must be an
2788 unsigned 16-bit value, and the return value must be 8, 16, or 32 bits.
2793 <!-- _______________________________________________________________________ -->
2794 <div class="doc_subsubsection">
2795 <a name="i_writeport">'<tt>llvm.writeport</tt>' Intrinsic</a>
2798 <div class="doc_text">
2802 call void (<integer type>, <integer type>)*
2803 %llvm.writeport (<integer type> <value>,
2804 <integer type> <address>)
2810 The '<tt>llvm.writeport</tt>' intrinsic writes data to the specified hardware
2817 The first argument is the value to write to the I/O port.
2821 The second argument indicates the hardware I/O address to which data should be
2822 written. The address is in the hardware I/O address namespace (as opposed to
2823 being a memory location for memory mapped I/O).
2829 The '<tt>llvm.writeport</tt>' intrinsic writes <i>value</i> to the I/O port
2830 specified by <i>address</i>. The address and value must be integers, but the
2831 size is dependent upon the platform upon which the program is code generated.
2832 For example, on x86, the address must be an unsigned 16-bit value, and the
2833 value written must be 8, 16, or 32 bits in length.
2838 <!-- _______________________________________________________________________ -->
2839 <div class="doc_subsubsection">
2840 <a name="i_readio">'<tt>llvm.readio</tt>' Intrinsic</a>
2843 <div class="doc_text">
2847 declare <result> %llvm.readio (<ty> * <pointer>)
2853 The '<tt>llvm.readio</tt>' intrinsic reads data from a memory mapped I/O
2860 The argument to this intrinsic is a pointer indicating the memory address from
2861 which to read the data. The data must be a
2862 <a href="#t_firstclass">first class</a> type.
2868 The '<tt>llvm.readio</tt>' intrinsic reads data from a memory mapped I/O
2869 location specified by <i>pointer</i> and returns the value. The argument must
2870 be a pointer, and the return value must be a
2871 <a href="#t_firstclass">first class</a> type. However, certain architectures
2872 may not support I/O on all first class types. For example, 32-bit processors
2873 may only support I/O on data types that are 32 bits or less.
2877 This intrinsic enforces an in-order memory model for llvm.readio and
2878 llvm.writeio calls on machines that use dynamic scheduling. Dynamically
2879 scheduled processors may execute loads and stores out of order, re-ordering at
2880 run time accesses to memory mapped I/O registers. Using these intrinsics
2881 ensures that accesses to memory mapped I/O registers occur in program order.
2886 <!-- _______________________________________________________________________ -->
2887 <div class="doc_subsubsection">
2888 <a name="i_writeio">'<tt>llvm.writeio</tt>' Intrinsic</a>
2891 <div class="doc_text">
2895 declare void %llvm.writeio (<ty1> <value>, <ty2> * <pointer>)
2901 The '<tt>llvm.writeio</tt>' intrinsic writes data to the specified memory
2908 The first argument is the value to write to the memory mapped I/O location.
2909 The second argument is a pointer indicating the memory address to which the
2910 data should be written.
2916 The '<tt>llvm.writeio</tt>' intrinsic writes <i>value</i> to the memory mapped
2917 I/O address specified by <i>pointer</i>. The value must be a
2918 <a href="#t_firstclass">first class</a> type. However, certain architectures
2919 may not support I/O on all first class types. For example, 32-bit processors
2920 may only support I/O on data types that are 32 bits or less.
2924 This intrinsic enforces an in-order memory model for llvm.readio and
2925 llvm.writeio calls on machines that use dynamic scheduling. Dynamically
2926 scheduled processors may execute loads and stores out of order, re-ordering at
2927 run time accesses to memory mapped I/O registers. Using these intrinsics
2928 ensures that accesses to memory mapped I/O registers occur in program order.
2933 <!-- ======================================================================= -->
2934 <div class="doc_subsection">
2935 <a name="int_libc">Standard C Library Intrinsics</a>
2938 <div class="doc_text">
2940 LLVM provides intrinsics for a few important standard C library functions.
2941 These intrinsics allow source-language front-ends to pass information about the
2942 alignment of the pointer arguments to the code generator, providing opportunity
2943 for more efficient code generation.
2948 <!-- _______________________________________________________________________ -->
2949 <div class="doc_subsubsection">
2950 <a name="i_memcpy">'<tt>llvm.memcpy</tt>' Intrinsic</a>
2953 <div class="doc_text">
2957 declare void %llvm.memcpy(sbyte* <dest>, sbyte* <src>,
2958 uint <len>, uint <align>)
2964 The '<tt>llvm.memcpy</tt>' intrinsic copies a block of memory from the source
2965 location to the destination location.
2969 Note that, unlike the standard libc function, the <tt>llvm.memcpy</tt> intrinsic
2970 does not return a value, and takes an extra alignment argument.
2976 The first argument is a pointer to the destination, the second is a pointer to
2977 the source. The third argument is an (arbitrarily sized) integer argument
2978 specifying the number of bytes to copy, and the fourth argument is the alignment
2979 of the source and destination locations.
2983 If the call to this intrinisic has an alignment value that is not 0 or 1, then
2984 the caller guarantees that the size of the copy is a multiple of the alignment
2985 and that both the source and destination pointers are aligned to that boundary.
2991 The '<tt>llvm.memcpy</tt>' intrinsic copies a block of memory from the source
2992 location to the destination location, which are not allowed to overlap. It
2993 copies "len" bytes of memory over. If the argument is known to be aligned to
2994 some boundary, this can be specified as the fourth argument, otherwise it should
3000 <!-- _______________________________________________________________________ -->
3001 <div class="doc_subsubsection">
3002 <a name="i_memmove">'<tt>llvm.memmove</tt>' Intrinsic</a>
3005 <div class="doc_text">
3009 declare void %llvm.memmove(sbyte* <dest>, sbyte* <src>,
3010 uint <len>, uint <align>)
3016 The '<tt>llvm.memmove</tt>' intrinsic moves a block of memory from the source
3017 location to the destination location. It is similar to the '<tt>llvm.memcpy</tt>'
3018 intrinsic but allows the two memory locations to overlap.
3022 Note that, unlike the standard libc function, the <tt>llvm.memmove</tt> intrinsic
3023 does not return a value, and takes an extra alignment argument.
3029 The first argument is a pointer to the destination, the second is a pointer to
3030 the source. The third argument is an (arbitrarily sized) integer argument
3031 specifying the number of bytes to copy, and the fourth argument is the alignment
3032 of the source and destination locations.
3036 If the call to this intrinisic has an alignment value that is not 0 or 1, then
3037 the caller guarantees that the size of the copy is a multiple of the alignment
3038 and that both the source and destination pointers are aligned to that boundary.
3044 The '<tt>llvm.memmove</tt>' intrinsic copies a block of memory from the source
3045 location to the destination location, which may overlap. It
3046 copies "len" bytes of memory over. If the argument is known to be aligned to
3047 some boundary, this can be specified as the fourth argument, otherwise it should
3053 <!-- _______________________________________________________________________ -->
3054 <div class="doc_subsubsection">
3055 <a name="i_memset">'<tt>llvm.memset</tt>' Intrinsic</a>
3058 <div class="doc_text">
3062 declare void %llvm.memset(sbyte* <dest>, ubyte <val>,
3063 uint <len>, uint <align>)
3069 The '<tt>llvm.memset</tt>' intrinsic fills a block of memory with a particular
3074 Note that, unlike the standard libc function, the <tt>llvm.memset</tt> intrinsic
3075 does not return a value, and takes an extra alignment argument.
3081 The first argument is a pointer to the destination to fill, the second is the
3082 byte value to fill it with, the third argument is an (arbitrarily sized) integer
3083 argument specifying the number of bytes to fill, and the fourth argument is the
3084 known alignment of destination location.
3088 If the call to this intrinisic has an alignment value that is not 0 or 1, then
3089 the caller guarantees that the size of the copy is a multiple of the alignment
3090 and that the destination pointer is aligned to that boundary.
3096 The '<tt>llvm.memset</tt>' intrinsic fills "len" bytes of memory starting at the
3097 destination location. If the argument is known to be aligned to some boundary,
3098 this can be specified as the fourth argument, otherwise it should be set to 0 or
3104 <!-- _______________________________________________________________________ -->
3105 <div class="doc_subsubsection">
3106 <a name="i_isunordered">'<tt>llvm.isunordered</tt>' Intrinsic</a>
3109 <div class="doc_text">
3113 declare bool %llvm.isunordered(<float or double> Val1, <float or double> Val2)
3119 The '<tt>llvm.isunordered</tt>' intrinsic returns true if either or both of the
3120 specified floating point values is a NAN.
3126 The arguments are floating point numbers of the same type.
3132 If either or both of the arguments is a SNAN or QNAN, it returns true, otherwise
3138 <!-- _______________________________________________________________________ -->
3139 <div class="doc_subsubsection">
3140 <a name="i_sqrt">'<tt>llvm.sqrt</tt>' Intrinsic</a>
3143 <div class="doc_text">
3147 declare <float or double> %llvm.sqrt(<float or double> Val)
3153 The '<tt>llvm.sqrt</tt>' intrinsic returns the sqrt of the specified operand,
3154 returning the same value as the libm '<tt>sqrt</tt>' function would. Unlike
3155 <tt>sqrt</tt> in libm, however, <tt>llvm.sqrt</tt> has undefined behavior for
3156 negative numbers (which allows for better optimization).
3162 The argument and return value are floating point numbers of the same type.
3168 This function returns the sqrt of the specified operand if it is a positive
3169 floating point number.
3173 <!-- ======================================================================= -->
3174 <div class="doc_subsection">
3175 <a name="int_count">Bit Counting Intrinsics</a>
3178 <div class="doc_text">
3180 LLVM provides intrinsics for a few important bit counting operations.
3181 These allow efficient code generation for some algorithms.
3186 <!-- _______________________________________________________________________ -->
3187 <div class="doc_subsubsection">
3188 <a name="int_ctpop">'<tt>llvm.ctpop</tt>' Intrinsic</a>
3191 <div class="doc_text">
3195 declare int %llvm.ctpop(int <src>)
3202 The '<tt>llvm.ctpop</tt>' intrinsic counts the number of ones in a variable.
3208 The only argument is the value to be counted. The argument may be of any
3209 integer type. The return type must match the argument type.
3215 The '<tt>llvm.ctpop</tt>' intrinsic counts the 1's in a variable.
3219 <!-- _______________________________________________________________________ -->
3220 <div class="doc_subsubsection">
3221 <a name="int_ctlz">'<tt>llvm.ctlz</tt>' Intrinsic</a>
3224 <div class="doc_text">
3228 declare int %llvm.ctlz(int <src>)
3235 The '<tt>llvm.ctlz</tt>' intrinsic counts the number of leading zeros in a
3242 The only argument is the value to be counted. The argument may be of any
3243 integer type. The return type must match the argument type.
3249 The '<tt>llvm.ctlz</tt>' intrinsic counts the leading (most significant) zeros
3250 in a variable. If the src == 0 then the result is the size in bits of the type
3251 of src. For example, <tt>llvm.cttz(int 2) = 30</tt>.
3257 <!-- _______________________________________________________________________ -->
3258 <div class="doc_subsubsection">
3259 <a name="int_cttz">'<tt>llvm.cttz</tt>' Intrinsic</a>
3262 <div class="doc_text">
3266 declare int %llvm.cttz(int <src>)
3273 The '<tt>llvm.cttz</tt>' intrinsic counts the number of trailing zeros.
3279 The only argument is the value to be counted. The argument may be of any
3280 integer type. The return type must match the argument type.
3286 The '<tt>llvm.cttz</tt>' intrinsic counts the trailing (least significant) zeros
3287 in a variable. If the src == 0 then the result is the size in bits of the type
3288 of src. For example, <tt>llvm.cttz(2) = 1</tt>.
3292 <!-- ======================================================================= -->
3293 <div class="doc_subsection">
3294 <a name="int_debugger">Debugger Intrinsics</a>
3297 <div class="doc_text">
3299 The LLVM debugger intrinsics (which all start with <tt>llvm.dbg.</tt> prefix),
3300 are described in the <a
3301 href="SourceLevelDebugging.html#format_common_intrinsics">LLVM Source Level
3302 Debugging</a> document.
3307 <!-- *********************************************************************** -->
3310 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
3311 src="http://jigsaw.w3.org/css-validator/images/vcss" alt="Valid CSS!"></a>
3312 <a href="http://validator.w3.org/check/referer"><img
3313 src="http://www.w3.org/Icons/valid-html401" alt="Valid HTML 4.01!" /></a>
3315 <a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
3316 <a href="http://llvm.cs.uiuc.edu">The LLVM Compiler Infrastructure</a><br>
3317 Last modified: $Date$